Datasets:

TomTBT

/

pmc_open_access_xml

like 4

[ "<title>Background</title>", "<p>Health policy is based on the ideal that all inhabitants should have equal access to health care. US studies have found ethnic differences in the use of health care with lower consumption rates for people from ethnic minorities [##REF##16257951##1##,##REF##11748426##2##]. European studies on ethnic differences in the use of health care have reported mixed outcomes. Some studies reported higher rates by various ethnic minority groups in comparison with the majority populations with respect to the general practitioner (GP) services in combination with lower rates for outpatient services [##REF##9774288##3##, ####REF##12217992##4##, ##REF##9232742##5####9232742##5##]. Other studies reported either no differences in outpatient care use [##REF##11553652##6##,##REF##9764280##7##] or higher rates among ethnic minorities or immigrant populations [##UREF##0##8##, ####REF##9631643##9##, ##REF##10848128##10##, ##REF##12927473##11##, ##REF##9198408##12####9198408##12##].</p>", "<p>Differences in consumption rates could be based upon differences in the incidence of diseases. For some diseases, like diabetes, it has indeed been shown that there are ethnic differences in incidence [##REF##15238806##13##, ####REF##16123507##14##, ##REF##9686553##15####9686553##15##]. However, ethnic differences in consumption rates could also have other reasons. For instance, in case of referral of patients by a GP to an outpatient clinic, patients ethnicity might influence the physicians' beliefs about and expectations of patients, and consequently the physicians' actions [##UREF##1##16##]. There are also indications that, as a result of less effective and satisfying doctor-patient relationship [##REF##12767585##17##], physicians that treat ethnic minority patients are more uncertain in the process of care [##UREF##2##18##]. Especially in case of language problems, which are common among immigrant populations, the latter might be the case. This could clearly have implications for the referral pattern and the care physicians give. A possible result could be that people from ethnic minorities are more often referred on the basis of vague symptoms, and might therefore less often receive a medical diagnosis. Other possible explanations could be that immigrant patients seek professional medical help more often, not only because they actually do have more health problems, but they also tend to report physical symptoms more often and more expressive. It is suggested that this might be due to the fact that they have a more positive attitude towards care-seeking [##UREF##3##19##,##REF##7951325##20##].</p>", "<p>Since the number of immigrants in the industrialized countries is growing, it is also of growing importance to obtain data on ethnic differences in the use of health care, referral patterns en diagnoses. E.g., in the Netherlands the proportion of non-western immigrants increased from about 3% in 1980, to 6% in 1990 and more than 10% in 2005. A limitation of previous studies is their reliance on self-reports of health care utilization. Although self-reports have been shown to be a valid estimate of health care utilization across socio-economic strata [##REF##11821355##21##], there is less evidence for cross-cultural validity, especially among some of the larger immigrant groups in the western European countries, Turkish and Moroccan people [##REF##10760636##22##]. Ethnic differences in recall bias, non-response and tendency for giving socially desirable answers, could undermine the validity of self-reported measures [##REF##9250628##23##]. For instance, illiteracy and limited proficiency in the native language, both more prevalent among immigrants, will increase non-response rates. Therefore we have chosen for the use of hospital registration data in order to examine ethnic differences in the use of health care.</p>", "<p>The ethnic minority populations in the western European countries mainly exist of immigrants who entered the country in the period between 1955–1985 when there was a severe shortage of people to do the unschooled jobs in these countries (first-generation immigrants). In the case that ethnic differences are found, it would also be worthwhile to know whether these differences also persist in the younger generations or whether the consumption rates in younger generations will be more alike those in the majority population.</p>", "<p>Using hospital registration data for an outpatient clinic for internal medicine we investigated the following research questions:</p>", "<p>(1) Are there differences between ethnic groups in the use of outpatient health services?</p>", "<p>(2) Are there differences in reasons for referral and diagnosis between ethnic groups?</p>" ]

[ "<title>Methods</title>", "<title>Population</title>", "<p>From March 2002 to March 2003 the ethnicity of all <italic>new </italic>patients that visited the outpatient clinic of the department of Internal Medicine, Erasmus Medical Center (Erasmus MC), a university hospital in Rotterdam, the Netherlands, was registered (N = 3985). Definition of <italic>new </italic>patients: no referrals in the previous 14 months and referral for medical signs and symptoms that were not examined or treated in an earlier stage. Age of all patients was 15 year and over. This is because children under 15 year do not visit Erasmus MC, they go to a specialised hospital for children.</p>", "<p>Based on the next inclusion criteria the study population was selected out of the 3985 patients:</p>", "<p>1. Age under 71 year, as the older age groups contained very few immigrants. (N = 3270)</p>", "<p>2. The living area, based on zipp codes, is restricted to the municipality of Rotterdam. (N = 1592)</p>", "<p>3. Patients ethnicity: Six ethnicities were included: Surinamese, Turkish, Moroccan, Aruban/Antillean, Cape Verdean and Dutch. (N = 1332)</p>", "<title>Data</title>", "<p>The research proposal, including the plan of data collection, was authorised by the research ethics committee of the Erasmus MC. We used the country of birth of the patient and both parents to assign ethnicity. We applied the standard definition of ethnicity of Statistics Netherlands and considered a person to be Non-Dutch if at least one parent was born abroad [##UREF##4##24##]. During the year all new patients were asked for their and their parents' country of birth. Immigrants who were born in the Netherlands and had at least one parent that was born abroad, were considered second-generation immigrants. If a person was born abroad and at least one of the parents did, we defined the person as first-generation immigrant. A six-digit zip code was used for ascribing socio-economic status (based on standardized household income) and based on quintiles determined for Rotterdam [##UREF##5##25##]. Information about the composition of the population of Rotterdam was obtained from Statistics Rotterdam. A data file was created for the observation period. Persons were allowed to enter the study throughout the study period (open cohort design).</p>", "<p>The patient population was divided into two groups: residents from the referral area of the Erasmus MC and patients living in the municipality of Rotterdam. The referral area is part of the municipality of Rotterdam as a whole; the inhabitants of the referral area constituted about 12% of all inhabitants living in the municipality of Rotterdam. The referral area consists of the neighbourhoods surrounding the Erasmus MC, which is for the greater part a deprived area, for which the hospital has a local community service.</p>", "<p>Medical reports sent to the general practitioners, when the diagnostic analysis was completed, were scrutinized in order to collect the reason for referral and diagnosis. Both paper medical record and electronic records were used complementary in order to collect the data. Complete data were available for 1070 of 1332 patients. Absence of reports was equally distributed over all ethnic groups. After looking into more detail, 82 patients were not new patients. In total, referral reasons and diagnosis were collected for 988 new patients. Referral reasons were coded according to the International Classification of Primary Care (ICPC) and diagnosis according to the Diagnosis Treatment Combination (DTC). The latter is a system used to finance hospitals in the Netherlands. It is based on formation of groups of patients that have a homogeneous health care use profile. We designed meaningful categories by aggregating ICPC and DTC codes, in order to obtain groups of sufficient size for the analyses. In the appendix the original codes and structure as well as the aggregated categories for ICPC and DTC are presented.</p>", "<title>Contextual information</title>", "<p>In the Netherlands general practitioners are the gatekeepers to most other health services. Almost all Dutch inhabitants have a health insurance, at least in the period of this study. That means that there are no financial barriers for seeking professional health care help. In Rotterdam there are, besides Erasmus MC, four more hospitals that offer health care services. However, inhabitants of the referral area mainly visits the Erasmus MC, because of it's local community service for this area.</p>", "<title>Analysis</title>", "<p>With regard to research question 1 it was examined whether ethnic minority groups had a higher or lower use of the outpatient clinic than could be expected <italic>from their relative distribution in the population</italic>. In order to estimate rate ratios (Relative Risks) and 95% confidence intervals (CI) of health care use by ethnicity, in research question 1, Poisson regression analyses were carried out. Ethnicity was the independent and use of health care the dependent variable, while adjusting for age, sex and socio-economic status. For the Poisson distribution, the patients constituted the numbers of observed events (numerator). A base group (as a reference) represents the rate (denominator) at which these events occur. The population of the municipality of Rotterdam, including the ethnic distribution of it, constituted this group. For analyses restricted to the referral area, the population of the referral area was the base group. Both base groups were exactly defined grounded on six-digit zip codes. The composition (concerning age, sex and socio-economic status) of the base groups was obtained from Statistics Rotterdam. We used the multiplicative (relative) risk, which is the standard Poisson regression model. The statistical package used was EGRET (version 2.0.1).</p>", "<p>With regard to research question 2, we examined whether there were ethnic differences in reasons for referral and diagnosis <italic>within the study population of patients</italic>. For the research questions about ethnic differences in referrals and diagnosis, the reference consisted of the patient group. We did not have the data to estimate odds ratios for the population of Rotterdam, that is why we restricted these analyses to the patient population. For these questions logistic regression was used in SPSS (version 11). Ethnicity was the independent variable and reason for referral respectively diagnosis was the dependent variable (yes/no referred for gastro-intestinal signs and symptoms; yes/no gastro-intestinal diagnosis). We adjusted for age, sex and socio-economic status.</p>", "<p>The analyses of research question 1, concerning the population of Rotterdam, were restricted to people aged 15–70, as the older age groups contained very few immigrants. In the models we adjusted for sex, age (10-year age categories), and socio-economic status (SES; quintiles). The analyses concerning differences in generation were restricted to people 15–45, as the second-generation immigrants contained very few people above 45 years.</p>" ]

[ "<title>Results</title>", "<p>In total, 4438 new patients visited the outpatient clinic. From these 4438 patients, ethnicity was registered for 3985 patients (90%). Only residents of the municipality of Rotterdam were included (40%). Six ethnicities were included: Surinamese, Turkish, Moroccan, Aruban/Antillean, Cape Verdean and Dutch (33%). Among the ethnic minorities Surinamese was the largest, and Antillean/Aruban accounted for the smallest group of patients. All patients were referred by their GP, because of the gatekeeper role of the GP in the Netherlands. In table ##TAB##0##1## characteristics of the research population are presented. In total 1332 patients remained.</p>", "<p>For the referral area of the Erasmus MC, immigrant people have an increased use of the outpatient clinic compared to Dutch people, adjusted for sex and age. The increased use was expressed by relative risks of consultations, which ranged from 1.29 in the Cape Verdean group to 1.82 in the Turkish group. The difference was statistically significant only for Surinamese, Turkish and Moroccan people.</p>", "<p>For the municipality of Rotterdam, all immigrant groups included in this study, had a significantly increased use of outpatient care, adjusted for sex and age. Again Turkish immigrants had the highest rates; relative risks ranged from 1.65 in the Antillean/Aruban group to 1.97 in the Turkish group. In table ##TAB##1##2A## relative risks are presented for all ethnic minorities compared to the native Dutch.</p>", "<p>Additional adjustment for socio-economic status hardly changed the estimates (table ##TAB##1##2A##). The largest decrease in relative risk was observed among Cape Verdeans in the analyses for the municipality of Rotterdam, from 1.99 to 1.88. In analyses in which the first and second immigrant generations were distinguished (table ##TAB##1##2B##), no difference in the use of health care were observed between the second-generation and the native Dutch citizens. In both areas the increased use can be predominantly ascribed to the first-generation immigrants.</p>", "<p>In table ##TAB##2##3## odds ratios are represented for ethnic differences in referral reasons. Compared to Dutch patients, immigrant patients are less likely to be referred to the outpatient care of the Erasmus MC because of reasons in the category 'indefinite, ambiguous signs'. Further analysis showed that the difference in this referral reason is mainly based on two underlying categories; general weakness/tiredness and memory disorder, which both occurred more frequently among Dutch patients (data not shown). Immigrant patients are more likely to be referred because of reasons in the category 'signs &amp; symptoms gastro-intestinal'. Underlying codes in these are generalized/diffuse abdominal pain/cramps, localized abdominal pain and viral hepatitis, of which all three conditions had a higher incidence among immigrant patients. The only exception in this category is rectal bleeding which had a lower incidence among immigrant patients (data not shown). In the patient population were no ethnic differences in the likelihood to be referred because of reasons in the category 'risk for vascular diseases' or the category with remaining referral reasons.</p>", "<p>After adjusting for socio-economic status ethnic differences only decreased slightly, indicating that ethnic differences in socio-economic status hardly explained the differences in referral reasons for patients that were referred to the Erasmus MC.</p>", "<p>In table ##TAB##3##4## odds ratios are represented for the categories of the diagnosis, as made by the internist. With regard to diagnosis, immigrant patients have an increased risk to be diagnosed with 'liver' diseases and they have a lower risk for the category 'analysis without diagnosis'. The dominant code in the category liver is hepatitis B/C. The category 'analysis without diagnosis' constituted a set of underlying codes which all have in common that extensive medical examination took place without giving a pathological diagnosis. The underlying codes discriminate between different complaints, from which general weakness/tiredness and a collection of residue complaints (e.g. impairment of visual acuity, sickness, amnesia) occurred more often among Dutch patients. Analysis of abdominal pain without resulting in a pathological diagnosis on the contrary, occurred more often among immigrant patients.</p>", "<p>Ethnic differences in risk for 'liver diseases' are partly explained by differences in socio-economic status; after adjusting for socio-economic status the differences in risk became smaller. For 'liver diseases' the risk decreased from 1.96 to 1.75, but retained a borderline significance. For 'analysis without diagnosis' the risk decreased slightly after adjustment for socio-economic status (from 0.62 to 0.67), it retained statistical significance.</p>", "<p>Finally, we also analysed ethnic differences in the risk of getting a certain diagnosis given the referral reason and looked for ethnic differences in this relationship. There appeared to be no differences between the ethnic groups under study, except for the category 'gastro-intestinal signs &amp; symptoms', in which immigrant patients were more likely to receive a diagnosis in the category 'liver' (data not shown).</p>" ]

[ "<title>Discussion</title>", "<p>There is a higher number of new patient referral visits of Surinamese, Turkish and Moroccan immigrants, living in the referral area of the Erasmus MC, compared to native Dutch people than could be expected from their relative distribution in the population. In Rotterdam municipality the five largest ethnic minority groups all demonstrate a higher use of the outpatient care facilities. This increased use can be predominantly ascribed to the first-generation immigrants; second-generation immigrants do not appear to have an increased use of health care services. Immigrant patients who visited the outpatient clinic were more likely to be referred because of 'gastro-intestinal signs &amp; symptoms' and less likely to be referred because of 'diffuse and ambiguous signs'. Regarding ethnic differences in diagnosis, we noted an increased risk of 'liver related diagnosis' and a decreased risk of 'analysis without diagnosis' for immigrant patients.</p>", "<p>We have to consider a few limitations of the current study. Although over 4000 new patients were registered in the hospital, numbers for those eligible for the study were small for some ethnic groups, and especially for second-generation immigrants. Therefore, not all research questions could be examined for the ethnic groups separately, nor could the first-generation be distinguished from second-generation immigrant for all research questions. For our second research question it was necessary to aggregate all ethnic groups to one 'immigrant' group. The aggregation was justified by the outcomes of table ##TAB##1##2##, in which all ethnic groups show a deviated use of health care in the same range and direction. A similar limitation concerns the aggregating of codes of <italic>referral reasons </italic>and <italic>diagnosis</italic>. In the results of research question 2 no ethnic differences were found for the <italic>referral reason </italic>'risk factor vascular disease'. However, looking in more detail shows large differences between the ethnic minority groups for more specific referral reasons. Surinamese and Cape Verdean patients often are referred with the most prevalent underlying risk of vascular diseases, namely hypertension. The same holds with regard to ethnic differences in <italic>diagnosis</italic>: we found no different risk of <italic>diagnosis </italic>'risk factor vascular diseases' regarding ethnicity. But the underlying codes showed that diabetes mellitus was significantly more prevalent among referred immigrant patients and dyslipidemia more common among Dutch patients. Odds ratios have to be interpreted in a relative sense, because they were calculated for the closed group of patients that visit the outpatient clinic of Erasmus MC. An apparent lower odds ratio might be the result of higher rates in other groups of diagnosis.</p>", "<p>In the second place ethnicity is based on countries of birth. Although this is a well-accepted definition [##REF##11553652##6##,##REF##10220022##26##, ####REF##8272647##27##, ##UREF##6##28####6##28##] we were unable to address ethnic variations within immigrant groups. Differences in the use of health care may have been more differentiated within certain ethnic minority groups, especially for the ethnically diverse Surinamese and Antillean/Aruban population.</p>", "<p>We approximated socioeconomic status at the individual level by making use of mean neighbourhood incomes, a variable at the ecological level. This measure may not be equally good for all ethnic groups. In some groups, the place of residence is determined by the mean socioeconomic status of a neighbourhood, whereas in others it is predominantly determined by the ethnic composition of a neighbourhood. In that case, neighbourhood income may be a less valid indicator of socioeconomic status. For Antilleans this does not seem to be the case, however for Turks, Moroccans and Surinamese a somewhat larger proportion (5 to 15%) of the population belonged to the lowest income quintile according to the measure at the ecological level than according to the measure at the individual level. This means that the place of residence of Turks, Moroccan and Surinamese may be more strongly determined by factors other than neighbourhood income. As the discrepancy was fairly small, the influence of the differential validity on the outcomes of this study would be limited [##UREF##5##25##].</p>", "<p>Besides Erasmus MC, there are four more hospitals in Rotterdam that offer health care services. Differences in preference for Erasmus MC could have introduced the differences in health care use. For at least the referral area, this seems hardly the case. A survey among general practitioners in the referral area reported a slightly different referral pattern among different ethnic groups to Erasmus MC and other hospitals in Rotterdam (unpublished data). General practitioners send immigrant patients more often than Dutch patients to the internal medicine outpatient's care of Erasmus MC. The difference is (at most) 5% and cannot explain the increased use of 80% by immigrant patients. Additional support for our assumption that potential differences in referral patterns (due to preferences or the reputation of the Erasmus MC) between ethnic groups in Rotterdam municipality, have had little influence on the outcomes of our study comes from the analysis of the ethnic differences <italic>in referral reasons </italic>for both areas separately (data not shown). The findings at least indicate that there are no ethnic differences in referral reasons between the referral area and Rotterdam as a whole. Herewith a correct inference for the population of Rotterdam municipality is deduced, since the assumption for representativeness of the patient sample seems to be supported.</p>", "<p>Remarkable is that the ethnic differences in likelihood of being referred are higher when focussing on Rotterdam municipality than when focussing on the referral area. It is uncertain whether this can be attributed to the prevalence of certain diseases, which require special care. Erasmus MC is also a university hospital and delivers tertiary medical care.</p>", "<p>The results of this study regarding the <italic>use of health care </italic>differ from the results of Stronks et al., who reported no differences in the use of outpatient care [##REF##11553652##6##], likewise using registration data. An explanation could be that they addressed outpatient care clinics comprising of several types of specialists, while we made a restriction to internal medicine. Immigrant patients are known to have a higher incidence of several diseases and syndromes, which are referred to the internal medicine clinic (i.e., diabetes, liver diseases and gastro-intestinal complaints). Diseases referred to other outpatient care clinics probably are more equally distributed among different ethnic groups[##UREF##0##8##]</p>", "<p>The results of our study are in agreement with the results of some other studies. Both Manna [##REF##11004953##29##] and Weide and Foets [##UREF##3##19##] reported an increased risk for immigrant patients for referral with 'signs &amp; symptoms gastro-intestinal'. Some of our results however, differ from the results of other studies. Other studies have reported that reasons for medical consultation among immigrants patient's are more often misunderstood or perceived as not being appropriate by the physician, and that the diagnostic process among immigrant patients might be more complicated because of language barriers, other concepts of disease, and other expressions of pain or other symptoms [##UREF##7##30##, ####REF##12796227##31##, ##REF##12653378##32##, ##UREF##8##33####8##33##] Possibly this could lead to more referrals for indefinite or ambiguous signs and immigrant patients would be more likely to end up in the category 'analyses without diagnosis', but we found the opposite: less immigrant patients came to the outpatient clinic with 'indefinite ambiguous signs' and compared to Dutch patients they have a lower risk for the category 'analysis without diagnosis'. Differences in domains of health care under study may explain the dissimilarity of their results with ours, as these other investigations mainly focussed on general practitioners or on health care in general. Given the Dutch system, where general practitioners are the gatekeepers to most other health services, including the outpatient services, health complaints perceived as inappropriate might have been filtered out by the general practitioner effectively.</p>", "<p>A possible explanation for the higher use of outpatient care among immigrants might be a direct reflection of a higher incidence and prevalence of certain diseases. We did not have information about health status, but previous studies have reported a higher incidence of infectious diseases [##REF##15647725##34##], hypertension [##REF##15647724##35##,##REF##7629455##36##], circulatory diseases [##UREF##9##37##, ####REF##11759851##38##, ##REF##11120707##39####11120707##39##], diabetes [##REF##10661656##40##, ####REF##11473073##41##, ##REF##12324981##42####12324981##42##], and worse health status in general [##UREF##0##8##,##REF##15647722##43##] among immigrant groups. Despite these higher incidences, we cannot rule out that referral rates for immigrant groups are inappropriately low or inequitable. Another explanation for higher use could be different styles/patterns in referring immigrant patients and Dutch patients to the outpatient care. Uitewaal [##UREF##10##44##] reported that more diabetes patients from Turkish descent than native Dutch diabetes patients were referred to the outpatient care. Moreover, immigrant patients asked more for referrals to outpatient clinics, instead of analysis or treatment by the general practitioner[##UREF##11##45##] It is known that immigrant patients seek professional medical help more often, not only because they actually do have more health problems, but they also tend to report physical symptoms more often. It is suggested that this might be due to the fact that they have a more positive attitude towards care-seeking [##REF##7951325##20##,##UREF##3##19##] and they have different beliefs concerning health and illness. [##UREF##12##46##] However we did not find evidence for ethnic differences in mismatch between referral and diagnosis, general practitioners can cause differences in referrals between immigrant and Dutch patients, when communication with immigrant patients is less effective than in consultations with Dutch patients, there is more misunderstanding and also more non-compliance. [##REF##14572938##47##] These explanations could also contribute to the interpretation of the finding that the increased use of health care services predominantly can be ascribed to the first-generation immigrants. Compared to the first-generation, immigrants of the second-generation generally have a higher education, better language skills and have better control of their lives [##REF##7951325##20##]. Thus, second-generation immigrants could become more alike to Dutch patients and their health care use will become more similar. While first-generation immigrants directly benefited from the more favourable socio-economic, public health and health-care conditions in the Netherlands compared with their country of origin, they are not yet affected by the higher risks of diseases associated with prosperity. [##UREF##5##25##] In the future, next generations immigrants, will be exposed to new risks similar to the risks of the native Dutch. Old risks, like higher risk for infections, will be substituted for risks more comparable to the native Dutch. Concerning first and second generation immigrants, duration of residence might explain the difference in health care use. However, there is no different mean time of residence between first- and second generation of the ethnic groups under study. Most first generation immigrants came to the Netherlands decades ago as labor workers.</p>", "<p>Besides the differences in health care use between native Dutch and ethnic minority groups, there also appear to be differences among the ethnic minority groups themselves. Additional analyses showed that Cape Verdean immigrants have a statistically significant lower use of health care than Surinam, Turkish and Moroccan immigrants. Further research is needed to explore why Cape Verdean immigrants are more similar to the native Dutch population regarding health care use.</p>", "<p>Because our data are limited to one particular outpatient care unit and moreover to a university hospital, we must be cautious in generalizing the results and the hypotheses generated by this study require further study.</p>" ]

[ "<title>Conclusion</title>", "<p>We conclude that especially first-generation immigrants make significantly more use of the outpatients' care in internal medicine. Ethnic differences might decrease as the share of first-generation immigrants decreases. Concerning this point, it is warranted to monitor the risks of diseases associated with prosperity in the future among immigrant groups. Ethnic differences in referral reasons and diagnosis might be based on a higher prevalence of diseases. Further study is needed to establish this. We found no evidence that the increased use is based on referrals for non-medical reasons, on the contrary. As long as the increased use of outpatient health care is related to ethnic background and the generation of the immigrants rather than to socio-economic status, health professionals have to take ethnicity into account in their daily medical practice. Moreover, they should take the main differences in prevalence of diseases among immigrants into account during the consultations.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>As in other Western countries, the number of immigrants in the Netherlands is growing rapidly. In 1980 non-western immigrants constituted about 3% of the population, in 1990 it was 6% and currently it is more than 10%. Nearly half of the migrant population lives in the four major cities. In the municipality of Rotterdam 34% of the inhabitants are migrants. Health policy is based on the ideal that all inhabitants should have equal access to health care and this requires an efficient planning of health care resources, like staff and required time per patient. The aim of this study is to examine ethnic differences in the use of internal medicine outpatient care, specifically to examine ethnic differences in the reason for referral and diagnosis.</p>", "<title>Methods</title>", "<p>We conducted a study with an open cohort design. We registered the ethnicity, sex, age, referral reasons, diagnosis and living area of all new patients that visited the internal medicine outpatient clinic of the Erasmus Medical Centre in Rotterdam (Erasmus MC) for one year (March 2002–2003). Additionally, we coded referrals according to the International Classification of Primary Care (ICPC) and categorised diagnosis according to the Diagnosis Treatment Combination (DTC). We analysed data by using Poisson regression and logistic regression.</p>", "<title>Results</title>", "<p>All ethnic minority groups (Surinam, Turkish, Moroccan, Antillean/Aruban and Cape Verdean immigrants) living in Rotterdam municipality, make significantly more use of the outpatient clinic than native Dutch people (relative risk versus native Dutch people was 1.83, 1.97, 1.79, 1.65 and 1.88, respectively).</p>", "<p>Immigrant patients are more likely to be referred for analysis and treatment of 'gastro-intestinal signs &amp; symptoms' and were less often referred for 'indefinite, general signs'. Ethnic minorities were more frequently diagnosed with 'Liver diseases', and less often with 'Analysis without diagnosis'. The increased use of the outpatient facilities seems to be restricted to first-generation immigrants, and is mainly based on a higher risk of being referred with 'gastro-intestinal signs &amp; symptoms'.</p>", "<title>Conclusion</title>", "<p>These findings demonstrate substantial ethnic differences in the use of the outpatient care facilities. Ethnic differences may decrease in the future when the proportion of first-generation immigrants decreases. The increased use of outpatient health care seems to be related to ethnic background and the generation of the immigrants rather than to socio-economic status. Further study is needed to establish this.</p>" ]

[ "<title>Authors' contributions</title>", "<p>LCL drafted the manuscript. and performed the statistical analysis. AHB participated in the design of the study. SWJL and JPM both conceived of the study, and participated in its design. IJ participated in the conceive of the study, participated in its coordination and helped to draft the manuscript. All authors read and approved the final manuscript.</p>", "<title>Appendix</title>", "<title>Aggregated categories of ICPC with underlying codes</title>", "<title>Referral indefinite, general signs</title>", "<p>Weakness, tiredness general, feeling ill, pain general/multiple sites, nausea, feeling anxious/nerves/tense, feeling depressed, feeling/behaving irritable/angry, sleep disturbance, memory disorder</p>", "<title>Referral signs &amp; symptoms gastro-enterology</title>", "<p>Abdominal pain/cramps general, abdominal pain epigastric, heartburn, rectal/anal pain, perianal itching, abdominal pain localized other, dyspepsia/indigestion, flatulence/gas/belching, vomiting, diarrhoea, constipation, haematemesis/vomiting blood, maelena, rectal bleeding, incontinence of bowel, change in faeces/bowel movements, abdominal mass nos, abdominal distension, viral hepatitis, injury digestive system other, congenital anomaly digestive system, oesophagus disease, duodenal ulcer, peptic ulcer other, stomach function disorder, appendicitis, hiatus hernia, abdominal hernia other, diverticular disease, irritable bowel syndrome, chronic enteritis/ulcerative colitis, anal fissure/perianal abscess, liver disease nos, cholecystisis/cholelithiasis, disease digestive system other.</p>", "<title>Referral risk factor vascular disease</title>", "<p>Elevated bloodpressure, hypertension uncomplicated, hypertension complicated, lipid disorder, diabetes insulin dependent, diabetes non-insulin dependent, ischaemic heart disease with angina, acute myocardial infarction, ischaemic heart disease without angina, stroke/cerebrovascular accident, cerebrovascular disease, artherosclerosis/peripheral vascular disease, pulmonary embolism, heart failure.</p>", "<title>Remaining referrals</title>", "<p>All rest codes occurring at the outpatient department of internal medicine.</p>", "<title>Aggregated categories of DTC with underlying codes</title>", "<title>Diagnosis cardio vascular diseases and risk factor cardio vascular disease, including diabetes</title>", "<p>Hypertension, stroke (not specified as haemorrhage or infarction), embolism and thrombosis of arteries, aneurysmas, atherosclerosis peripheral, other arterial disorders, post thrombosis syndrome, ischaemic heart diseases, unstable angina, myocardial infarction, heart failure, dyslipidaemia, riskfactors vascular disease, thrombophilia, diabetes.</p>", "<title>Diagnosis liver</title>", "<p>Diseases of liver: Hepatitis B/C, alcoholic hepatitis, livercirrhosis, liver tumours.</p>", "<title>Diagnosis gastro-enterology</title>", "<p>Gastroenterology</p>", "<title>Signs and symptoms without diagnosis</title>", "<p>Diagnostic procedures generated no diagnosis. All diagnostic procedures in the beginning were based on signs (i.e. pain) or symptoms (e.g. Fever, deviant laboratory results)</p>", "<title>Diagnosis endocrinology without diabetes mellitus</title>", "<p>Endocrine System Diseases, without diabetes mellitus.</p>", "<title>Remaining diagnosis</title>", "<p>White rule remaining diagnosis</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2458/8/287/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>This project was funded by Erasmus MC 'Doelmatigheid Zorg' (internal fund for efficiency of care.)</p>", "<p>The authors would like to thank P. Burger of the Centre for Research and Statistics (COS) Rotterdam for his assistance in obtaining the necessary data. We thank Katrina Giskes for her helpful comments on a former version of this paper.</p>" ]

[]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Population by living area, ethnicity, mean age, sex, generation and socio-economic status. N = 1332.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><underline>Referral area</underline></td><td align=\"center\">Dutch</td><td align=\"center\">Surinamese</td><td align=\"center\">Turkish</td><td align=\"center\">Moroccan</td><td align=\"center\">Antillean Aruban</td><td align=\"center\">Cape Verdean</td><td align=\"center\">S, T, M, A/A, C Together*</td></tr><tr><td><underline>Erasmus MC</underline></td><td colspan=\"7\"/></tr></thead><tbody><tr><td align=\"left\">N = 320</td><td align=\"center\">124</td><td align=\"center\">62</td><td align=\"center\">57</td><td align=\"center\">36</td><td align=\"center\">11</td><td align=\"center\">30</td><td align=\"center\">196</td></tr><tr><td align=\"left\">Mean age</td><td align=\"center\">55.0</td><td align=\"center\">46.5</td><td align=\"center\">41.8</td><td align=\"center\">41.4</td><td align=\"center\">39.6</td><td align=\"center\">50.4</td><td align=\"center\">44.4</td></tr><tr><td align=\"left\">% men</td><td align=\"center\">43.5</td><td align=\"center\">32.3</td><td align=\"center\">31.6</td><td align=\"center\">38.9</td><td align=\"center\">63.6</td><td align=\"center\">53.3</td><td align=\"center\">38.3</td></tr><tr><td align=\"left\">% 2^ndgeneration</td><td align=\"center\">-</td><td align=\"center\">6.5</td><td align=\"center\">5.3</td><td align=\"center\">5.6</td><td align=\"center\">-</td><td align=\"center\">-</td><td align=\"center\">4.6</td></tr><tr><td/><td/><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"><underline>Municipality </underline></td><td align=\"center\">Dutch</td><td align=\"center\">Surinamese</td><td align=\"center\">Turkish</td><td align=\"center\">Moroccan</td><td align=\"center\">Antillean/Aruban</td><td align=\"center\">Cape Verdean</td><td align=\"center\">S, T, M, A/A, C Together*</td></tr><tr><td align=\"left\"><underline>Rotterdam</underline></td><td colspan=\"7\"/></tr><tr><td colspan=\"8\"><hr/></td></tr><tr><td align=\"left\">N = 1332</td><td align=\"center\">852</td><td align=\"center\">174</td><td align=\"center\">126</td><td align=\"center\">79</td><td align=\"center\">50</td><td align=\"center\">51</td><td align=\"center\">480</td></tr><tr><td align=\"left\">Mean age</td><td align=\"center\">56.1</td><td align=\"center\">45.0</td><td align=\"center\">41.4</td><td align=\"center\">42.1</td><td align=\"center\">41.3</td><td align=\"center\">47.3</td><td align=\"center\">43.4</td></tr><tr><td align=\"left\">% men</td><td align=\"center\">40.1</td><td align=\"center\">33.9</td><td align=\"center\">38.1</td><td align=\"center\">44.3</td><td align=\"center\">44.0</td><td align=\"center\">51.0</td><td align=\"center\">39.6</td></tr><tr><td align=\"left\">% 2^ndgeneration</td><td align=\"center\">-</td><td align=\"center\">7.5</td><td align=\"center\">8.7</td><td align=\"center\">5.1</td><td align=\"center\">2.0</td><td align=\"center\">3.9</td><td align=\"center\">6.5</td></tr><tr><td align=\"left\">% lowest SES level</td><td align=\"center\">33</td><td align=\"center\">62.1</td><td align=\"center\">80.2</td><td align=\"center\">83.5</td><td align=\"center\">54.0</td><td align=\"center\">76.5</td><td align=\"center\">71.0</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Relative risks are presented for all ethnic minorities compared to the native Dutch.\n</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\" colspan=\"7\">A Referrals to Internal Medicine outpatient care by Ethnicity (Relative risks (CI 95%) with Dutch as reference; age 15–70).</td></tr></thead><tbody><tr><td/><td align=\"center\">Surinamese</td><td align=\"center\">Turkish</td><td align=\"center\">Moroccan</td><td align=\"center\">Antillean/Aruban</td><td align=\"center\">Cape Verdean</td><td align=\"center\">p-value*</td></tr><tr><td colspan=\"1\"><hr/></td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"center\">Referral area Erasmus MC N = 320</td><td/><td/><td/><td/><td/><td/></tr><tr><td colspan=\"1\"><hr/></td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"center\">Adjusted for sex and age</td><td align=\"center\">1.47 (1.05–2.06)</td><td align=\"center\">1.82 (1.29–2.56)</td><td align=\"center\">1.49 (1.00–2.21)</td><td align=\"center\">1.46 (0.78–2.75)</td><td align=\"center\">1.29 (0.85–1.97)</td><td align=\"center\">0.02</td></tr><tr><td align=\"center\">Additional adjustment for socio-economic status</td><td align=\"center\">1.49 (1.06–2.07)</td><td align=\"center\">1.84 (1.31–2.59)</td><td align=\"center\">1.50 (1.01–2.24)</td><td align=\"center\">1.49 (0.79–2.80)</td><td align=\"center\">1.30 (0.86–1.99)</td><td align=\"center\">0.02</td></tr><tr><td colspan=\"1\"><hr/></td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"center\">Municipality of Rotterdam N = 1332</td><td/><td/><td/><td/><td/><td/></tr><tr><td colspan=\"1\"><hr/></td><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"center\">Adjusted for sex and age</td><td align=\"center\">1.88 (1.58–2.24)</td><td align=\"center\">2.05 (1.68–2.50)</td><td align=\"center\">1.88 (1.48–2.39)</td><td align=\"center\">1.67 (1.24–2.26)</td><td align=\"center\">1.99 (1.49–2.67)</td><td align=\"center\">&lt;0.001</td></tr><tr><td align=\"center\">Additional adjustment for socio-economic status</td><td align=\"center\">1.83 (1.53–2.19)</td><td align=\"center\">1.97 (1.59–2.42)</td><td align=\"center\">1.79 (1.40–2.29)</td><td align=\"center\">1.65 (1.22–2.24)</td><td align=\"center\">1.88 (1.40–2.54)</td><td align=\"center\">&lt;0.001</td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"left\" colspan=\"7\">B</td></tr><tr><td align=\"left\" colspan=\"7\">Relative risks (CI 95%) for the use of outpatient care</td></tr><tr><td align=\"left\" colspan=\"7\">N = 385</td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"left\" colspan=\"4\">Comparison</td><td align=\"center\" colspan=\"3\">Relative risk^a</td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"left\" colspan=\"4\">1^stgeneration immigrants versus Dutch</td><td align=\"center\" colspan=\"3\">1.85 (1.51–2.25)</td></tr><tr><td align=\"left\" colspan=\"4\">2^ndgeneration immigrants versus Dutch</td><td align=\"center\" colspan=\"3\">1.08 (0.72–1.63)</td></tr><tr><td align=\"left\" colspan=\"4\">2^ndgeneration versus 1^stgeneration immigrants</td><td align=\"center\" colspan=\"3\">0.59 (0.39–0.88)</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Ethnic differences in referral reasons. N = 988</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\">N total</td><td align=\"center\">Dutch</td><td align=\"center\">Immigrants</td><td align=\"center\">Odds ratios ^abc</td></tr></thead><tbody><tr><td align=\"left\">Indefinite, ambiguous signs</td><td align=\"center\">144</td><td align=\"center\">119</td><td align=\"center\">25</td><td align=\"center\">0.46* (0.27–0.77)</td></tr><tr><td align=\"left\">Signs &amp; symptoms gastro-intestinal</td><td align=\"center\">298</td><td align=\"center\">160</td><td align=\"center\">138</td><td align=\"center\">1.45* (1.05–2.00)</td></tr><tr><td align=\"left\">Risk for vascular diseases</td><td align=\"center\">139</td><td align=\"center\">88</td><td align=\"center\">51</td><td align=\"center\">1.11 (0.71–1.71)</td></tr><tr><td align=\"left\">Remaining category</td><td align=\"center\">407</td><td align=\"center\">257</td><td align=\"center\">150</td><td align=\"center\">0.90 (0.66–1.23)</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Ethnic differences in diagnosis. N = 988</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\">N total</td><td align=\"center\">Dutch</td><td align=\"center\">Immigrants</td><td align=\"center\">Odds ratios ^abc</td></tr></thead><tbody><tr><td align=\"left\">Diagnose category 'risk vascular diseases, including diabetes mellitus'</td><td align=\"center\">143</td><td align=\"center\">93</td><td align=\"center\">52</td><td align=\"center\">1.12 (0.72–1.72)</td></tr><tr><td align=\"left\">Diagnose category 'Liver diseases'</td><td align=\"center\">75</td><td align=\"center\">32</td><td align=\"center\">43</td><td align=\"center\">1.75* (1.00–3.07)</td></tr><tr><td align=\"left\">Diagnose category 'Gastro-intestinal'</td><td align=\"center\">200</td><td align=\"center\">118</td><td align=\"center\">82</td><td align=\"center\">1.07 (0.74–1.55)</td></tr><tr><td align=\"left\">Diagnose category 'Analysis without diagnosis'</td><td align=\"center\">278</td><td align=\"center\">194</td><td align=\"center\">84</td><td align=\"center\">0.68* (0.48–0.95)</td></tr><tr><td align=\"left\">Diagnose category 'Endocrinology without diabetes mellitus'</td><td align=\"center\">108</td><td align=\"center\">60</td><td align=\"center\">48</td><td align=\"center\">0.90 (0.56–1.44)</td></tr><tr><td align=\"left\">Remaining category</td><td align=\"center\">184</td><td align=\"center\">127</td><td align=\"center\">55</td><td align=\"center\">1.17 (0.76–1.81)</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>* Surinamese, Turkish, Moroccan, Antillean, Aruban and Caper Verdean together</p></table-wrap-foot>", "<table-wrap-foot><p>*p-value of the overall ethnic differences (Wald test).</p><p>^aage 15–45, adjusted for sex, age and socio-economic status</p></table-wrap-foot>", "<table-wrap-foot><p>^aAdjusted for sex, age and SES.</p><p>^bConfidence Interval 95%</p><p>^cImmigrants versus Dutch as reference.</p><p>*p &lt; 0.05</p></table-wrap-foot>", "<table-wrap-foot><p>^aAdjusted for sex, age and SES.</p><p>^bConfidence Interval 95%.</p><p>^cImmigrants versus Dutch as reference.</p><p>*p &lt; 0.05.</p></table-wrap-foot>" ]

[]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Meal replacement shakes represent an important strategy in combating the worldwide epidemic of obesity due to their simplicity and convenience [##UREF##0##1##]. Meal replacement shakes have been studied extensively for both medical and public health efforts to combat obesity [##REF##7706595##2##, ####REF##17592648##3##, ##REF##15284372##4####15284372##4##].</p>", "<p>A number of studies have suggested that protein is the most important macronutrient mediating satiety and leads to increased weight loss with retention of lean body mass. Single meals with increased protein to carbohydrate ratios have also been shown to increase satiety and decrease food intake [##REF##10447984##5##,##REF##9272665##6##], resulting in both improved weight loss and improved maintenance of weight loss [##REF##11737954##7##, ####REF##10375057##8##, ##REF##14710168##9####14710168##9##]. Meal replacement simplifies the weight loss regimens by replacing one or two meals a day with a product of defined nutrient and calorie content. MR leads to increased weight losses over twelve weeks compared to simply restricting favorite food intakes, and weight losses have been maintained for up to five years using MR [##REF##10968732##10##]</p>", "<p>An increase in dietary protein content has been proposed to be effective for body weight regulation through effects on satiety, thermogenesis and substrate partitioning. Protein has specific effects on satiety hormones, including PYY 3–36 [##REF##16950139##11##]. When protein replaces carbohydrate within a low-fat diet, reduced insulinemic and glycemic responses have been observed resulting in increased fat oxidation [##REF##7990700##12##]</p>", "<p>The present study was designed to test the hypothesis that simply increasing the protein content of a meal replacement (MR) within a high protein diet without the knowledge of the participant would result in increased weight loss and improved retention of lean body mass in the absence of a resistance exercise program by comparison to standard MR within a standard protein diet. To test the hypothesis, a soy and whey protein powder was used to enrich a standard MR shake in one arm compared to a carbohydrate \"placebo\" powder added to the same MR shake in the other arm. This novel approach has not been tested previously to our knowledge. To minimize variations based on body composition, the diets were also adjusted so that each subject was instructed to follow a diet which provided either 2.2 gm/kg lean body mass protein in the high protein (HP) group or 1.1 gm protein/kg lean body mass in the standard protein (SP) group. Patients received dietary instruction at baseline, and met with the dietitian at weeks 2, 4 and 8 to assess general compliance and to provide additional supplies of the MR products. Therefore, this study examines the effectiveness of protein enrichment of MR in a realistic outpatient setting on weight loss and retention of lean body mass.</p>" ]

[ "<title>Methods</title>", "<p>Subjects were recruited by public advertisement. Subjects over 30 years of age with a body mass index (BMI) between 27 to 40 kg/m², and in good health by history, physical examination, and basic laboratory screening (complete blood count, serum chemistries, liver panel, and lipid panel) were selected for study. Subjects with type 2 diabetes or glucose intolerance were excluded as were individuals who regularly drank more than one alcoholic beverage daily,</p>", "<p>One hundred men and women who met the selection criteria were randomly assigned to either the HP or SP treatment. This was a single-blinded study. The protein powder jars were labeled as either A or B, depending on their protein content. Subjects were randomized in a 1:1 manner to either HP or SP diet for 12 weeks using a computerized random proportion model. Diet plans were individualized per subject. Caloric intake to achieve weight loss was based on a 500 Kcal deficit of the participants' estimated resting metabolic rate as determined by body composition analysis by bioelectrical impedance.</p>", "<p>Participants in the HP group received a diet plan that provided 2.2 grams of protein per kg of LBM while the diet for the SP group provided 1.1 grams of protein per kg of LBM. The meal energy macronutrient composition in the HP group was approximately 30% protein, 30% fat, and 40% carbohydrate. The macronutrient composition in the SP diet was approximately 15% protein, 30% fat, and 55% carbohydrate. Both groups received the same isocaloric MR (Formula 1, Herbalife Intl., Los Angeles) with either a protein supplement for the HP group (Performance Protein Powder, Herbalife Intl., Los Angeles) or with a similar tasting carbohydrate placebo for SP group. Two MR and two meals were eaten daily.</p>", "<p>Instructions were provided for preparation of the MR and subjects were advised to consume one MR as a meal and the other as snack. All subjects were given individualized menu plans that incorporated the two MR (one meal and one snack) and included two all-food meals. All participants met individually with a registered dietitian at baseline for dietary instruction, and at 2, 4, and 8 weeks to assess compliance.</p>", "<p>Participants were weighed and protein powder meal replacement products were dispensed at each visit to ensure compliance. Subjects were given general advice for increasing their activity level with a goal of 30 minutes of aerobic exercise per day, but no heavy resistance exercise.</p>", "<title>Body weight and composition</title>", "<p>Subjects were weighed at each visit (Detecto-Medic; Deteco-Scales; Brooklyn, NY) while wearing no shoes and after an overnight fast. Height was measured with a stadiometer (Detecto-Medic; Deteco-Scales; Brooklyn, NY) at week 0. BMI was calculated as weight (kg)/height squared (m). Body composition was determined by bioelectrical impedance analysis (BIA) (310e Bioimpedance analyzer; Biodynamics; Seattle, WA) and was performed at 0 and 12 weeks.</p>", "<title>Biochemistry</title>", "<p>Fasting blood samples were collected at weeks 0, 4, 8, and 12 for measurement of lipid profiles, blood glucose and liver function tests.</p>", "<title>Statistical analysis</title>", "<p>Weight loss was the primary outcome and the data were analyzed according to intention to treat allocation utilizing SAS version 9 (Cary, North Carolina) in the Department of Biostatistics.</p>", "<p>Patient characteristics and baseline measurements of the two study groups were compared using t-test (for numerical variables) or Chi-square test (for categorical variables) to evaluate quality of the randomization.</p>", "<p>Standard t-tests were used to compare weight losses between the two arms. In addition, to assess weight loss within each treatment arm, paired t-tests were conducted comparing baseline and 12 week weight for each subject. All data except baseline characteristics are presented as means +/- standard error of the mean (SEM). A univariate analysis of variance was used to assess differences between treatment and outcome variables. Since the distributions of change in fat weight and percentage change in fat weight were not normal, signed rank test was used for testing change from baseline within each group. The Wilcoxon rank sum test was used for comparing the change between the two groups. Multivariate analysis was performed to compare the difference between the two diet groups using general linear model. Square root transformation was applied before the multivariate analysis was performed.</p>" ]

[ "<title>Results</title>", "<p>100 obese men and women were randomly assigned to either a HP or SP MR diet plan. Fifteen subjects withdrew from the study within the first week after randomization due to noncompliance with the meal plan (6 in the HP group and 9 in the SP group). All other subjects completed the 12-week study. Subject characteristics in the two treatment arms at baseline were not significantly different (Table ##TAB##0##1##). Mean age was 49.4 ± 1.1 years. Mean BMI at baseline was 33.8 ± 0.53 for HP group and 32.6 ± 0.58 kg/m2 for SP group.</p>", "<title>Weight loss</title>", "<p>Subjects were weighed at baseline, and at 2, 4, 8 and 12 weeks. Baseline body weight was not significantly different between these two groups. Both groups lost significant amount of weight at 12 weeks (-4.19 ± 0.5 kg for HP group and -3.72 ± 0.7 kg for SP group, p &lt; 0.0001 for both groups). (Figure ##FIG##0##1##) After controlling for baseline weight, gender, and time period, there was no significant difference between the two treatment groups. For both dietary groups, BMI was significantly lower at 12 weeks (HP = -1.50 ± 0.58; SP = -1.13 ± 0.24). There were no significant differences in BMI changes between the two dietary groups (Table ##TAB##1##2##).</p>", "<title>Waist circumference</title>", "<p>Change in waist circumference (cm) at 12 weeks was significant in both treatment groups (HP = -6.7 ± 1.1; SP = -5.1 ± 0.8 p &lt; 0.0001). No significant differences in change in waist circumference at any time period were observed between diets (table ##TAB##1##2##).</p>", "<title>Fat mass by BIA</title>", "<p>Subjects in the HP group lost a significant amount of fat at 12 weeks (from 35.2 ± 1.0 kg to 33.6 ± 1.2 kg, p &lt; 0.0001) but not the SP group (32.3 ± 1.3 kg to 31.7 ± 1.0 kg, p &gt; 0.05). Subjects in the HP group lost significantly more fat weight than the SP group (HP = -1.65 ± 0.63; SP = -0.64 ± 0.79 kg p = 0.05) (Figure ##FIG##1##2##, table ##TAB##1##2##).</p>", "<title>Fat-free mass by BIA</title>", "<p>At 12 weeks, the two dietary groups had significantly decreased lean body weight (kg) (HP = -2.78.1 ± 0.62; SP = -4.06 ± 1.74, p &lt; 0.0001). No significant differences were observed between the dietary groups (table ##TAB##1##2##).</p>", "<title>Cholesterol, HDL, LDL, triglyceride, and glucose</title>", "<p>At 12 weeks, there were significant reductions in cholesterol and LDL levels (mg/dL) for the HP group (cholesterol -13.2 ± 5.3, p &lt; 0.05; LDL -7.47 ± 3.38, p &lt; 0.05) but not for the SP group (cholesterol -7.02 ± 4.3 p &gt; 0.05; -9.17 + 5.65, p &gt; 0.05). The difference between the two groups was not significant. There were no significant changes from baseline, nor between dietary groups in serum HDL and triglyceride levels. Similarly, fasting blood glucose levels did not change significantly from baseline for either group at 12 weeks. (Table ##TAB##2##3##)</p>" ]

[ "<title>Discussion</title>", "<p>Protein-enriched meal replacements within a higher protein diet resulted in no greater overall weight loss than the standard protein MR plan over 12 weeks. In this trial, the amounts of weight lost were typical for meal replacement studies done previously [##REF##10968732##10##]. However, the expected effects on increased weight loss resulting from a high protein diet were not seen in this study. There are two possible reasons for the observed similarities in overall weight loss. First, the subjects in the SP group may have eaten foods outside their recommended meal plans which increased protein intake enough to compensate for the difference in protein contents of the MR. Second, the use of MR may have been the major influence on the weight loss by simplifying their weight loss efforts so that the power of the MR intervention may have obscured the difference between the weight loss of subjects using protein-enriched MR shakes by comparison to standard MR [##UREF##1##13##]. The purpose of the study was to test the real world impact of simply enriching MR with more protein. Based on our results, it appears that compliance is a much more important factor in the MR regimen than protein content.</p>", "<p>Protein enrichment of MR did appear to lead to increased retention of lean body mass based on bioelectrical impedance analysis. In this study, greater retention of lean body mass was suggested by the observation of increased fat loss at similar weight losses. Fat loss is determined by subtracting the lean body mass determined based on body water content from the total body weight at baseline compared to 12 weeks. The fat loss was significant both between groups and in individuals over time. Loss of lean mass was not different since the variability in fat free mass between subjects increased the variability of this measurement, reducing our power to see any difference. The observation we made at 12 weeks using bioelectrical impedance will require confirmation in longer-term studies where changes in body composition are more marked and in which additional methods for determining body composition are used. A recent meta-analysis of 87 short-term diet studies where protein and carbohydrate content was varied found that a protein intake of greater than 1.05 g/kg of actual body weight was associated with 0.6 kg additional fat-free mass retention compared with diets with protein intakes ≤1.05 g/kg [##REF##16469983##14##]. Both meal plans in this study had protein greater than this cut point and the effects seen may be blunted by the relatively high protein in the SP group.</p>", "<p>In future studies, it may also be desirable to combine protein enrichment of MR with resistance exercise to demonstrate significant differences in the retention of lean body mass during weight loss due to protein enrichment of MR. Evans and co-workers have shown that healthy free-living elderly men and women accommodate to the Recommended Dietary Allowance (RDA) for protein of 0.8 grams/kilogram/day with a continued decrease in urinary nitrogen excretion and reduced muscle mass. Increased dietary protein intake (up to 1.6 g protein/kg/day) may also enhance the hypertrophic response to resistance exercise that would enhance weight loss maintenance [##REF##15640513##15##].</p>" ]

[ "<title>Conclusion</title>", "<p>In summary, both the HP and SP diets resulted in the expected weight loss typical of an MR diet plan at 12 weeks. Both diets were well tolerated, sustainable, and did not result in any adverse effects. While typical results for outpatient trials of MR were observed in both arms, greater compliance with the MR diet plan may be necessary to obtain an improved sense of the contribution of protein enrichment of MR to lean body mass retention during weight loss. Finally, future studies may be more successful if they include a comparison of standard MR and protein-enriched of MR weight reduction regimens combined with heavy resistance exercise to maintain or increase lean body mass.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>While high protein diets have been shown to improve satiety and retention of lean body mass (LBM), this study was designed to determine effects of a protein-enriched meal replacement (MR) on weight loss and LBM retention by comparison to an isocaloric carbohydrate-enriched MR within customized diet plans utilizing MR to achieve high protein or standard protein intakes.</p>", "<title>Methods</title>", "<p>Single blind, placebo-controlled, randomized outpatient weight loss trial in 100 obese men and women comparing two isocaloric meal plans utilizing a standard MR to which was added supplementary protein or carbohydrate powder. MR was used twice daily (one meal, one snack). One additional meal was included in the meal plan designed to achieve individualized protein intakes of either 1) 2.2 g protein/kg of LBM per day [high protein diet (HP)] or 2) 1.1 g protein/kg LBM/day standard protein diet (SP). LBM was determined using bioelectrical impedance analysis (BIA). Body weight, body composition, and lipid profiles were measured at baseline and 12 weeks.</p>", "<title>Results</title>", "<p>Eighty-five subjects completed the study. Both HP and SP MR were well tolerated, with no adverse effects. There were no differences in weight loss at 12 weeks (-4.19 ± 0.5 kg for HP group and -3.72 ± 0.7 kg for SP group, p &gt; 0.1). Subjects in the HP group lost significantly more fat weight than the SP group (HP = -1.65 ± 0.63 kg; SP = -0.64 ± 0.79 kg, P = 0.05) as estimated by BIA. There were no significant differences in lipids nor fasting blood glucose between groups, but within the HP group a significant decrease in cholesterol and LDL cholesterol was noted at 12 weeks. This was not seen in the SP group.</p>", "<title>Conclusion</title>", "<p>Higher protein MR within a higher protein diet resulted in similar overall weight loss as the standard protein MR plan over 12 weeks. However, there was significantly more fat loss in the HP group but no significant difference in lean body mass. In this trial, subject compliance with both the standard and protein-enriched MR strategy for weight loss may have obscured any effect of increased protein on weight loss demonstrated in prior weight loss studies using whole food diets.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>All authors read and approved the final manuscript. LT participated in the conduct of the study, the analysis of the data, and drafted the manuscript. SC participated in the conduct of the study. KH participated in the conduct of the study. EY participated in the conduct of the study. CLC participated in the study design and statistical analysis. GT participated in the study coordination. HW participated in the statistical analysis. RE participated in the statistical analysis. ZL conceived of the study, participated in its design and coordination, and drafted the manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>This study was supported through an unrestricted gift from Herbalife International, Los Angeles, California. LT was supported by NIH Training Grant No. DK0718033. SC, KH, and EY were supported by NIH Training Grant No.T32 DK 07688.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Weight change from baseline at 12 weeks. </bold>* p &lt; 0.05 compared with base line body weight. Blank bar represents the high protein group, the shaded bar represents the standard protein group.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Change of fat mass by BIA from baseline at 12 weeks.</bold> ** p &lt; 0.001 compared with base line fat mass. Blank bar represents the high protein group and the shaded bar represents the standard protein group.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Patient characteristics at baseline</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\">HP (N = 45)</td><td align=\"center\">SP (N = 42)</td><td align=\"center\">Total (N = 87)</td><td align=\"center\">HP vs. SP</td></tr></thead><tbody><tr><td align=\"left\">Gender</td><td/><td/><td/><td/></tr><tr><td align=\"left\"> F</td><td align=\"center\">34 (76%)</td><td align=\"center\">27 (64%)</td><td align=\"center\">61 (70%)</td><td align=\"center\">NS</td></tr><tr><td align=\"left\"> M</td><td align=\"center\">11 (24%)</td><td align=\"center\">15 (36%)</td><td align=\"center\">26 (30%)</td><td/></tr><tr><td align=\"left\">Age</td><td/><td/><td/><td/></tr><tr><td align=\"left\"> Mean ± SE</td><td align=\"center\">49.2 ± 1.8</td><td align=\"center\">49.7 ± 1.4</td><td align=\"center\">49.4 ± 1.1</td><td align=\"center\">NS</td></tr><tr><td align=\"left\"> Median, range</td><td align=\"center\">47.0, 28–69</td><td align=\"center\">49.5, 30–65</td><td align=\"center\">49.0, 28–69</td><td/></tr><tr><td align=\"left\">Race</td><td/><td/><td/><td/></tr><tr><td align=\"left\"> Asian</td><td align=\"center\">4 (9%)</td><td align=\"center\">1 (2%)</td><td align=\"center\">5 (6%)</td><td align=\"center\">NS</td></tr><tr><td align=\"left\"> Black</td><td align=\"center\">9 (20%)</td><td align=\"center\">7 (17%)</td><td align=\"center\">16 (18%)</td><td/></tr><tr><td align=\"left\"> Caucasian</td><td align=\"center\">25 (55%)</td><td align=\"center\">30 (72%)</td><td align=\"center\">55 (63%)</td><td/></tr><tr><td align=\"left\"> Hispanic</td><td align=\"center\">4 (9%)</td><td align=\"center\">2 (5%)</td><td align=\"center\">6 (7%)</td><td/></tr><tr><td align=\"left\"> Other</td><td align=\"center\">0</td><td align=\"center\">1 (2%)</td><td align=\"center\">1 (1%)</td><td/></tr><tr><td align=\"left\"> Unknown</td><td align=\"center\">3(7%)</td><td align=\"center\">1 (2%)</td><td align=\"center\">4(5%)</td><td/></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Change of BMI, waist circumference, fat mass, and fat fee mass at 12 weeks</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"2\">BMI ((kg/m2)</td><td align=\"center\" colspan=\"2\">Waist Circumference (cm)</td><td align=\"center\" colspan=\"2\">Fat Mass (kg)</td><td align=\"center\" colspan=\"2\">Fat Free Mass (kg)</td></tr></thead><tbody><tr><td/><td align=\"center\">HP</td><td align=\"center\">SP</td><td align=\"center\">HP</td><td align=\"center\">SP</td><td align=\"center\">HP</td><td align=\"center\">SP</td><td align=\"center\">HP</td><td align=\"center\">SP</td></tr><tr><td colspan=\"9\"><hr/></td></tr><tr><td align=\"left\">Baseline</td><td align=\"center\">33.77 ± 0.53</td><td align=\"center\">32.66 ± 0.58</td><td align=\"center\">104.2 ± 1.8</td><td align=\"center\">101.7 ± 2.0</td><td align=\"center\">35.2 ± 1.0</td><td align=\"center\">32.3 ± 1.3</td><td align=\"center\">58.3 ± 1.6</td><td align=\"center\">60.0 ± 1.9</td></tr><tr><td align=\"left\">12 weeks</td><td align=\"center\">32.13 ± 0.54</td><td align=\"center\">31.11 ± 0.56</td><td align=\"center\">98.8 ± 1.6</td><td align=\"center\">97.3 ± 2.0</td><td align=\"center\">33.6 ± 1.2*</td><td align=\"center\">31.7 ± 1.0</td><td align=\"center\">55.6 ± 1.4</td><td align=\"center\">55.9 ± 1.7</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Change of lipid profiles and blood glucose</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"2\">Cholesterol (mg/dL)</td><td align=\"center\" colspan=\"2\">LDL (mg/dL)</td><td align=\"center\" colspan=\"2\">HDL (mg/dL)</td><td align=\"center\" colspan=\"2\">Triglycerides (mg/dL)</td><td align=\"center\" colspan=\"2\">Glucose (mg/dL)</td></tr></thead><tbody><tr><td/><td align=\"center\">HP</td><td align=\"center\">SP</td><td align=\"center\">HP</td><td align=\"center\">SP</td><td align=\"center\">HP</td><td align=\"center\">SP</td><td align=\"center\">HP</td><td align=\"center\">SP</td><td align=\"center\">HP</td><td align=\"center\">SP</td></tr><tr><td colspan=\"11\"><hr/></td></tr><tr><td align=\"left\">Baseline</td><td align=\"center\">199.82 ± 6.91</td><td align=\"center\">204.00 ± 6.07</td><td align=\"center\">118.16 ± 5.41</td><td align=\"center\">128.07 ± 5.69</td><td align=\"center\">54.45 ± 2.05</td><td align=\"center\">52.43 ± 1.69</td><td align=\"center\">122.16 ± 10.06</td><td align=\"center\">117.71 ± 8.44</td><td align=\"center\">89.25 ± 15.62</td><td align=\"center\">84.84 ± 11.40</td></tr><tr><td align=\"left\">12 weeks</td><td align=\"center\">186.69 ± 5.78*</td><td align=\"center\">195.49 ± 5.49</td><td align=\"center\">109.76 ± 4.68*</td><td align=\"center\">117.49 ± 5.65</td><td align=\"center\">55.04 ± 2.23</td><td align=\"center\">55.15 ± 2.75</td><td align=\"center\">107.44 ± 8.12</td><td align=\"center\">114.51 ± 9.92</td><td align=\"center\">85.35 ± 17.19</td><td align=\"center\">84.12 ± 12.90</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>*p &lt; 0.0001 compare 12 weeks vs. baseline</p></table-wrap-foot>", "<table-wrap-foot><p>*p &lt; 0.05 compare 12 weeks vs. baseline</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1475-2891-7-23-1\"/>", "<graphic xlink:href=\"1475-2891-7-23-2\"/>" ]

[]

{ "acronym": [], "definition": [] }

[]

[]

[]

[]

[]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<p>Malarial anaemia is an enormous public health problem in endemic areas and occurs predominantly in children in the first 3 years of life. Anaemia is due to both a great increase in clearance of uninfected cells and a failure of an adequate bone marrow response. Odhiambo, Stoute and colleagues show how the age distribution of malarial anaemia and the haemolysis of red blood cells may be linked by an age-dependent increase in the capacity of red blood cells to inactivate complement components absorbed or deposited directly on to the surface of the red blood cell. In this commentary, we discuss what has been established about the role of complement deposition on the surface of red blood cells in the pathology of malarial anaemia, how genetic polymorphisms of the complement control proteins influence the outcome of malaria infection and how the findings of Odhiambo, Stoute and colleagues and others shed light on the puzzling age distribution of different syndromes of severe malaria.</p>" ]

[ "<title>Commentary</title>", "<p>In the accompanying article, Odhiambo, Stoute and colleagues show how the age distribution of malarial anaemia and the haemolysis of red blood cells (RBCs) may be linked by an age-dependent increase in the capacity of RBCs to inactivate complement components absorbed or deposited directly on to the surface of the RBC [##REF##18717995##1##]. The work raises not only a number of important new lines of research but also challenges malaria researchers to apply this basic work to develop new ways to prevent and treat malaria.</p>", "<p>Malaria remains an enormous problem in public health around the world [##REF##15759000##2##]. Over 2 billion people live in malaria-endemic areas and up to 1 million children die each year of malaria. Severe falciparum malaria may present a variety of syndromes, but presents most frequently in childhood with severe malarial anaemia or coma. The difference in age distributions of children presenting with these syndromes is as striking as it is puzzling; the median age of children presenting with severe malarial anaemia is typically 1 to 3 years old, while the median age of children presenting with coma is significantly and consistently older, typically 3 to 5 years old [##REF##9186382##3##].</p>", "<p>Furthermore, there remain major unsolved problems about the fundamental pathophysiology of all syndromes of severe malaria. The rapid drop in haemoglobin during acute infection and the slower decline in chronic infection appear to be due to increased extravascular haemolysis of RBCs with a concomitant failure of the bone marrow to increase red cell production to compensate for these losses [##REF##17341664##4##].</p>", "<p>The increased clearance of infected cells is readily explained by the rupture of cells after completion of the parasite's intra-erythrocytic life cycle and opsonisation and clearance of intact infected RBCs. Rather less obvious is why and how uninfected cells are also cleared. It has been estimated that approximately 10 uninfected cells are cleared from the circulation for every infected cell and so the clearance of uninfected cells is of crucial importance for the development of malarial anaemia [##REF##10466119##5##].</p>", "<p>Why are uninfected RBCs cleared in such large numbers? Certainly the number and activation of splenic and other macrophages for phagocytosis of red cells is greatly increased during malarial infection [##REF##2119922##6##, ####REF##7821411##7##, ##REF##12437669##8##, ##REF##16765392##9####16765392##9##]. The increased clearance of uninfected erythrocytes is also due to extrinsic and intrinsic changes to the RBCs that enhance their recognition and phagocytosis.</p>", "<p>Uninfected RBCs have a reduced deformability leading to enhanced clearance in the spleen and a severe reduction in red cell deformability is also a strong predictor for mortality measured on admission, both in adults and children with severe malaria [##REF##9392587##10##,##REF##12174780##11##]. Second, the deposition of immunoglobulin and complement on uninfected RBCs may enhance receptor-mediated uptake by macrophages.</p>", "<p>The role of immunoglobulin and complement in marking uninfected RBCs for clearance by phagocytes was first studied by Facer and colleagues [##REF##371880##12##,##REF##6993068##13##] in The Gambia in the 1970s. It soon became clear that the Direct Coombs' Test (DCT) for immunoglobulins and/or complement deposited on the surface of RBCs was frequently positive in children with malaria [##REF##371880##12##]. The antibodies giving rise to the positive DCT were not autoimmune but were directed against malarial antigens [##REF##6993068##13##] (and our unpublished observations) and may include complexes of immunoglobulin G (IgG) with malarial antigens including ring stage protein 2 [##REF##16046531##14##].</p>", "<p>The story of how absorbed immune complexes may contribute to increased clearance of uninfected RBCs lay dormant for 20 years when Waitumbi, Stoute and colleagues based in Western Kenya began to study how immune complexes caused haemolysis [##REF##10666228##15##]. Appreciation of this work requires some knowledge of the control of complement deposition on the surface of RBCs.</p>", "<p>Here, a number of proteins are involved in the control of complement activation. Complement receptor 1 (CR1 or CD35), decay accelerating factor (DAF or CD55) and the membrane inhibitor of reactive lysis (MIRL or CD59) may enhance binding of C3b in immune complexes (CR1), enhance inactivation of C3 convertases (CR1 and CD55) and interfere with the assembly of the terminal components of complement that form the membrane attack complex (CD59) [##REF##1726550##16##]. Immune complexes and CR1 may then be removed by splenic macrophages and the RBCs depleted of immune complexes CR1 (CD35) and CD55 and returned to the circulation.</p>", "<p>In previous papers, Waitumbi, Stoute and colleagues have shown that the amount of red cell surface IgG is increased but red cell surface CR1 and CD55 reduced in children with severe malaria compared with asymptomatic and symptomatic controls [##REF##10666228##15##]. The difference in surface IgG levels appeared to be functionally significant as RBCs from children with severe anaemia were more susceptible to phagocytosis <italic>in vitro </italic>than RBCs from controls. Decline in CD35 or CR1 expression and increases in immune complexes bound on uninfected RBCs were associated with anaemia but these declines in CD35 (CR1) and CD55 expression were only transiently associated with malaria infection and levels returned to normal after infection had been cleared [##REF##12552440##17##]. These data are all consistent with the ability of CD35 and CD55 to inactivate the formation of C3a and reduction in CD35 and CD55 expression when bound immune complexes are cleared by phagocytes.</p>", "<p>How does this relate to the age-dependent incidence of malarial anaemia? Population studies in Europe and Africa showed the CR1 expression was strongly age-dependent and increases of both CR1 and CD55 were seen after 4 years of age and low levels of CR1 and CD55 expression were seen in a cases of severe malarial anaemia compared with slightly older children with cerebral malaria [##REF##15319870##18##]. Odhimbo, Stoute and their colleagues now show that haemoglobin is inversely associated with increased age and percentage C3b-positive uninfected RBCs are inversely and directly associated with CR1 levels measured shortly after infection [##REF##18717995##1##]. Therefore, reduction in these proteins (CR1, CD55 and CD59) may increase the susceptibility of children to malarial anaemia.</p>", "<p>How do these findings help explain the age distribution of syndromes of severe malaria? If clearance of the immune complexes absorbed on to the surface of uninfected cells is reduced in younger children expressing lower levels of CD35 and CR1, then clearance of these cells would be increased, leading to more anaemia in these younger age groups.</p>", "<p>Genetic polymorphisms also affect the expression levels, sequence and domain structure of CR1 in Africans and other populations [##REF##10528197##19##,##REF##14694201##20##]. Moreover, CR1 is a ligand for the variant antigens expressed at the surface of infected RBCs allowing the formation of rosettes of infected and uninfected RBCs [##REF##9230440##21##,##REF##2654325##22##]. In Melanesian populations, low levels of CR1 expression have been associated with reduced rosette formation and protection from severe malaria [##REF##14694201##20##].</p>", "<p>It is possible therefore, that age-related and genetically determined reduction of levels of CR1 expressed on RBCs are associated with an increased susceptibility to anaemia but protection from other forms of severe malaria and may provide an example of how innate resistance to one syndrome of malaria may be at the expense of susceptibility to other pathophysiological pathways involved in malaria infection.</p>", "<p>These hypotheses should ideally be tested in longitudinal studies using genetically and phenotypically well-characterised children to ascertain the levels of these age-dependent, complement regulatory proteins prior to infection and determine their association with haemoglobin levels during acute infection and with the incidence of severe disease through childhood. Such studies can only be done in a few sites in Africa where large populations can be followed before and after infection and presentation to a health facility using a demographic surveillance system.</p>", "<p>In assessing the role of these age-related changes to the complement regulatory proteins in the differential presentation of anaemia and coma, one would also have to consider the recently described findings of increased levels of erythropoietin (EPO) seen in younger children with anaemia [##REF##17517643##23##] and malaria [##REF##16804108##24##] and the increased protective effect of EPO on survival after coma, particularly in a younger age group [##REF##18263734##25##]. These alternative explanations for susceptibility to anaemia in younger age groups on one hand and increased protection from cerebral malaria or coma on the other are not mutually exclusive and it is possible, indeed likely, that many factors underpin such a well-defined difference in the clinical epidemiology of these malarial syndromes.</p>", "<p>Finally, we would ask where these studies will lead in the quest for new methods to prevent or treat malarial infection. The findings of Odhiambo, Stoute and colleagues presented in the accompanying article in <italic>BMC Medicine </italic>provide us with a clearer understanding of the causes of anaemia in children with malaria. Translating this new understanding of pathology into fruitful avenues to investigate prevention and treatment of malaria is perhaps the greatest challenge of clinico-pathological studies in this and other disciplines.</p>", "<title>List of abbreviations</title>", "<p>CR1: complement receptor 1; DAF: decay accelerating factor; DCT: Direct Coombs' Test; EPO: erythropoietin; MIRL: membrane inhibitor of reactive lysis; RBC: red blood cell.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1741-7015/6/24/prepub\"/></p>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>In clinical research and public health, it is frequently necessary to combine findings from multiple interventional or observational studies in order to address important safety and efficacy questions. A single study rarely provides a definitive answer because of limited sample size and the specific attributes of particular study populations. The challenges of combining data from heterogeneous studies are well described in the meta-analysis literature. In the majority of meta-analysis reports, the outcome of interest is a comparative risk estimate such as the odds ratio, relative risk, or risk difference [##UREF##0##1##]. Absolute risks, however, such as the proportion of clinical events among a cohort of patients or the response rate among patients receiving a certain treatment regimen, are important measures for helping to guide clinical and public health decisions. In the correct epidemiology and statistical terminology, these so-called rates are really proportions, but we will treat rates and proportions as equivalent in this paper as this term is commonly used in medical product safety research. Relevant methods to pool the absolute risks are especially important in safety evaluation of medical products as the risks for serious adverse outcomes are often rare, and precise estimates of the probability of these outcomes are crucial in the risk-benefit evaluation.</p>", "<p>In this report we describe the implementation of the beta-binomial method to pool the absolute risks from overdispersed data. This method estimates a summary probability of adverse events and is applicable in medical product safety evaluation as it takes into account the heterogeneity of studies. The application of the beta-binomial method in drug safety settings was previously described by Chuang-Stein in 1993 [##UREF##1##2##]. Here we aim to provide a detailed description of the method and to update and broaden its applications.</p>", "<p>The general setting is that of a clinical trial or cohort study of a specific exposure, such as: drug A with a sample size of <italic>n </italic>resulted in <italic>x </italic>number of adverse events (e.g. liver injury). Within each individual study the probability of encountering <italic>x </italic>number of adverse events out of a sample size of <italic>n </italic>is characterized by the binomial distribution. To summarize multiple studies of the same exposure, we need to account for their heterogeneity of the studies, for they could differ in their sample sizes, clinical settings, investigators, protocols, and prevalence of comorbidity among study subjects. The assumption of one binomial distribution that can describe the proportions of adverse event from all the studies is not always valid. Numerous factors, including ethnic difference, disease severity, comorbid conditions, and concomitant medications can contribute to the variation of the probability of interest, thus requiring additional assumptions beyond the binomial model. This phenomenon is often referred to as overdispersion [##UREF##2##3##,##UREF##3##4##]. Ignoring overdispersion when pooling overdispersed data that are binomial in nature could result in erroneous estimates of the probability of interest and its confidence interval.</p>", "<p>In the clinical trial literature, Chuang-Stein [##UREF##1##2##] proposed using the beta-binomial model to combine binomial event rates across multiple studies in an article titled \"An application of the beta-binomial model to combine and monitor medical event rates in clinical trials.\" Despite its sound statistical basis, this method has not been widely used in clinical and public health research articles during the years since its publication. Meanwhile, the application of the beta-binomial model in other fields is becoming more prevalent as it has been applied in fields as distant as sensory analysis [##UREF##4##5##] and computational linguistics [##UREF##5##6##]. We utilized this method to estimate the risk of liver toxicity among users of oral antifungal treatments [##REF##17765049##7##] and believe that it can be used more widely to help address similar questions. In the rest of this article we describe the statistical assumptions for the beta-binomial model, the process of estimating the probability of interest, methods to test for over-dispersion, and an example of its application.</p>" ]

[ "<title>Methods</title>", "<title>The Beta-Binomial distribution</title>", "<p>Both Chuang-Stein [##UREF##1##2##] and Ennis [##UREF##4##5##] provide excellent references for those who are interested in the history of the beta-binomial model. Recall the definition of the binomial distribution:</p>", "<p></p>", "<p>where <italic>x </italic>is the number of successes in a sequence of <italic>n </italic>independent success/failure experiments, each of which has probability <italic>p </italic>for success.</p>", "<p>Let probability <italic>p </italic>follow a beta distribution (<italic>p</italic>|<italic>α, β</italic>), then</p>", "<p></p>", "<p>where <italic>Γ </italic>is the gamma function over the domain [0, 1]; <italic>α </italic>and <italic>β </italic>are two positive parameters. The beta distribution was selected in the past because of its flexibility (capable of a wide range of shapes, see Figure ##FIG##0##1##) and its ability to provide good approximations. As Skellam [##UREF##6##8##] stated as early as 1948, \"in practice we could, at least in most cases, take this form of distribution as a convenient approximation.\" As a result, we arrive at a combination of the binomial distribution with a beta density function:</p>", "<p></p>", "<p>where <italic>x </italic>takes on the values 0, 1, 2... <italic>n</italic>, and <italic>α </italic>and <italic>β </italic>are positive. Note in equation (3) that <italic>n </italic>is the total number of study subjects, and <italic>x </italic>is the total number of subjects with a certain adverse event, although what most investigators are interested in is the proportion <italic>p </italic>that varies between 0 and 1 and has the appearance of a continuous distribution.</p>", "<p>So let <italic>p</italic>_{<italic>i </italic>}= <italic>x</italic>_{<italic>i</italic>}/<italic>n</italic>_{<italic>i</italic>}, <italic>i </italic>= 1,2, ... <italic>k</italic>, where <italic>i </italic>indexes the different studies, <italic>x</italic>_{<italic>i </italic>}is the number of events in the <italic>i</italic>th study and <italic>n</italic>_{<italic>i </italic>}is the sample size of the study. To reiterate, within the context of multiple studies where each study with sample size <italic>n</italic>_{<italic>i </italic>}and binomial probability <italic>p</italic>_{<italic>i </italic>}(e.g. for adverse events), one binomial distribution cannot adequately describe the additional variation when <italic>p</italic>_{<italic>i </italic>}varies and thus the data are fitted with a beta distribution with parameters (<italic>α, β</italic>), with <italic>α </italic>&gt; 0 and <italic>β </italic>&gt; 0. Let <italic>μ = α/(α+β), θ = </italic>1/<italic>(α+β)</italic>, where <italic>μ </italic>is the mean event rate (i.e., the expected value of a variable binomial parameter <italic>p</italic>) and <italic>θ </italic>is a measure of the variation in <italic>p</italic>. In short, we have constructed a two-stage model:</p>", "<p></p>", "<p></p>", "<p>The mean and variance of <italic>X </italic>are <italic>nμ </italic>and <italic>nμ(</italic>1-<italic>μ){θ/(</italic>1+<italic>θ)} </italic>[##REF##4785230##9##]. One can view the term {<italic>θ/(</italic>1+<italic>θ)} </italic>as a multiplier of the binomial variance. In other words, it models the overdispersion. Some authors (e.g. Kleinman [##UREF##7##10##]) prefer the term <italic>γ </italic>where <italic>γ = θ/(</italic>1+<italic>θ) = </italic>1/<italic>(α+β+1)</italic>. Then the variance is <italic>nμ(</italic>1 <italic>- μ) γ</italic>. In essence, one can derive the same information from <italic>θ </italic>and <italic>γ </italic>about the beta-binomial distribution, so it is beneficial to know both and employ whichever is more convenient for computation.</p>", "<title>Estimation of Parameters</title>", "<p>Two main methods, one involving moments and the other involving maximum likelihood, are often used to estimate the parameters <italic>μ </italic>and <italic>θ</italic>.</p>", "<title>The Moment Estimates Method</title>", "<p>In terms of actual data observed from different studies, let <italic>p</italic>_{<italic>i </italic>}= <italic>x</italic>_{<italic>i</italic>}/<italic>n</italic>_{<italic>i</italic>}, <italic>i </italic>= 1,2, ... <italic>k</italic>, where <italic>i </italic>indexes the different studies, <italic>x</italic>_{<italic>i </italic>}is the number of events in the <italic>i</italic>th study and <italic>n</italic>_{<italic>i </italic>}is the sample size of the study. The <italic>n</italic>_{<italic>i</italic>}<italic>'s </italic>here are almost always unequal in clinical studies.</p>", "<p>Let</p>", "<p></p>", "<p>where {<italic>w</italic>_{<italic>i</italic>}} represents a set of weights and <italic>w </italic>is the sum of all the weights [##UREF##7##10##].</p>", "<p>Let also </p>", "<p>then the moment estimates of <italic>μ </italic>and <italic>γ </italic>are:</p>", "<p> and</p>", "<p></p>", "<p>where . To derive <italic>θ</italic>, we can simply perform the following conversion:</p>", "<p></p>", "<p>Providing the proper set of weights is challenging because {<italic>w</italic>_{<italic>i</italic>}} is a function of the unknown parameter <italic>γ</italic>. Kleinman [##UREF##7##10##] first offered an empirical weighting procedure and suggested to set <italic>w</italic>_{<italic>i </italic>}= <italic>n</italic>_{<italic>i </italic>}or <italic>w</italic>_{<italic>i </italic>}= 1 to obtain an initial approximation of estimates of <italic>μ </italic>and <italic>γ </italic>using equation (4). Using this estimation of <italic>γ </italic>to compute {<italic>w</italic>_{<italic>i</italic>}}, one then can use these \"empirical\" weights to arrive at a new estimate of <italic>μ</italic>. In cases where <italic>γ </italic>estimates are negative, they are to be set to zero. Chuang-Stein [##UREF##1##2##] proposed an improvement on Kleinman's procedure by suggesting that the iteration be carried further until the differences between two consecutive sets of estimates for <italic>μ </italic>and <italic>γ </italic>are both smaller than some predetermined value. The example that was given in the paper [##UREF##1##2##] was 10^-6.</p>", "<p>Notations are simpler in cases where all <italic>n</italic>_{<italic>i</italic>}<italic>'s </italic>are equal, then</p>", "<p></p>", "<p>The moment estimates of <italic>μ </italic>and <italic>γ </italic>are</p>", "<p> and</p>", "<p></p>", "<title>The Maximum Likelihood Estimates Method</title>", "<p>As is written above, let <italic>p</italic>_{<italic>i </italic>}= <italic>x</italic>_{<italic>i</italic>}/<italic>n</italic>_{<italic>i</italic>}, <italic>i </italic>= 1,2, ... <italic>k</italic>, where <italic>i </italic>indexes the different studies, <italic>x</italic>_{<italic>i </italic>}is the number of events in the <italic>i</italic>th study and <italic>n</italic>_{<italic>i </italic>}is the sample size of the study. The maximum likelihood (ML) function involving <italic>α </italic>and <italic>β </italic>can be written as</p>", "<p></p>", "<p>where is the beta function of <italic>α </italic>and <italic>β </italic>and is used here to simplify equation (3). The log likelihood function is then</p>", "<p></p>", "<p>where c is a constant. Next we will need to take the partial derivative of the log likelihood function with respect to <italic>α </italic>and <italic>β</italic>. The ML equations involving <italic>α </italic>and <italic>β </italic>are</p>", "<p></p>", "<p>where</p>", "<p></p>", "<p>The second derivatives of ln<italic>L </italic>are:</p>", "<p></p>", "<p></p>", "<p></p>", "<p>where</p>", "<p></p>", "<p>These second derivatives of the log likelihood function can be used to form the Hessian matrix which, in turn, can be used to derive the standard errors for the parameters. An example will be given in a following section. Most often <italic>μ </italic>is the main parameter of interest, and therefore we present a direct estimation of it rather than proceeding through <italic>α </italic>and <italic>β</italic>.</p>", "<p>Define <italic>f</italic>_{<italic>x</italic>}<italic>(x) (x = 0,1,2, ..., k) </italic>as the observed frequencies of events from <italic>k </italic>trials. Then the likelihood of beta-binomial can be also written as</p>", "<p></p>", "<p>Where <italic>P(x) </italic>has already been stated in (3). Let so that is the total sample size of all the individual trials combined.</p>", "<p>The log likelihood function in terms of <italic>μ </italic>and <italic>θ </italic>is</p>", "<p></p>", "<p>where <italic>c </italic>is a constant and the ML estimators of and are solutions of</p>", "<p></p>", "<p></p>", "<p>These equations can be solved iteratively using the Newton-Raphson method [##UREF##8##11##].</p>", "<p>Again, the second partial derivatives of the log likelihood function can be used to form the Hessian matrix (H) at the ML solution</p>", "<p></p>", "<p>which, after being inverted, can be used to derive the covariance matrix and the standard errors for the parameters:</p>", "<p></p>", "<p>And the confidence intervals for and can be obtained by</p>", "<p></p>", "<p></p>", "<p>where <italic>Z</italic>_{1-<italic>α</italic>/2}is the 1-<italic>α</italic>/2 percentile of a standard normal distribution function.</p>", "<p>Once and are estimated, one can also derive and from the relationships that <italic>μ = α/(α+β), θ = 1/(α+β)</italic>. It can easily be shown that the estimate of is / and the estimate of is (1 - )/. If we substitute these estimates for <italic>α </italic>and <italic>β </italic>in the beta-binomial model (3), then the cumulative distribution can be calculated.</p>", "<p>As we have shown above, either method can be used to estimate the parameters of the beta-binomial distribution. Readers who are interested in more details should consult Griffiths [##REF##4785230##9##] and Kleinman [##UREF##7##10##]. Researchers have implemented the maximum likelihood estimation (MLE) method in two popular commercial statistical software packages. In addition, free statistical software, such as R and WinBUGS, have methods for fitting the beta-binomial model, but they require some programming.</p>", "<p>One of those two popular commercial statistical software packages is SAS (SAS Institute Inc., Cary, NC, USA). The macro BETABIN written by Ian Wakeling [##UREF##9##12##] is freely available. It borrows the existing SAS procedure NLMIXED to provide a maximum likelihood estimation of <italic>μ </italic>and <italic>θ</italic>. It provides not only the standard beta-binomial model, but also Brockhoff's [##UREF##10##13##] corrected beta-binomial model. Interested readers can also experiment directly with Proc NLMIXED to fit the beta-binomial model as others have done [##UREF##11##14##].</p>", "<p>The other software is Stata (College Station, Texas). Guimarães provided the necessary computer commands for beta-binomial estimations using the Stata command xtnbreg with conditional maximum likelihood [##UREF##12##15##]. In addition, Guimarães emphasized the common knowledge that the beta-binomial distribution was a special case of the more general Dirichlet-multinomial (DM) distribution – with two parameters in this case. In the general Dirichlet-multinomial distribution there are <italic>m </italic>parameters, allowing far more than two (<italic>α </italic>and <italic>β</italic>) in the beta-binomial distribution. In situations where one is indeed concerned with multiple types of adverse events associated with the same exposure, expanding to the Dirichlet-multinomial distribution is a logical solution. Technical details of the multinomial model have been given by others [##UREF##12##15##, ####UREF##13##16##, ##UREF##14##17####14##17##].</p>", "<title>Test of overdispersion</title>", "<p>Using the binomial model when the variability in the data exceeds what the binomial model can accommodate could result in an underestimation of the standard error of the pooled event rate and thus increase the chance of a Type I error. Ennis and Bi [##UREF##4##5##] described an experiment with 10,000 sets of simulated overdispersed binomial data where they found that the Type I error was 0.44 and not the false assumption of 0.05. It is precisely because the binomial model is unable to fit overdispersed binomial data that the application of the beta-binomial is necessary. So before one adopts the beta-binomial for the analysis of certain datasets, one must first examine whether the data are overdispersed to the extent that the beta-binomial model would be a better fit than the simple binomial model. There are several ways to examine overdispersion. We know that</p>", "<p></p>", "<p>where <italic>γ </italic>= 1/(1 + <italic>α </italic>+ <italic>β</italic>). If we are able to estimate <italic>γ</italic>, we can test whether <italic>γ </italic>is zero. If it is close to zero, then there is no significant overdispersion, and the binomial model will adequately describe the data. This test, however, has been found to be less sensitive in detecting departure from the binomial model because boundary problems arise as we test whether a positive-valued parameter is greater than 0 (recall that <italic>α </italic>and <italic>β </italic>are positive parameters, and consequently so are <italic>θ </italic>and <italic>γ</italic>) [##UREF##4##5##].</p>", "<p>As one would expect, a likelihood ratio test can also be used to test for overdispersion, but the same boundary problem applies [##REF##10985242##18##,##REF##7115867##19##]. The null hypothesis is that the underlying distribution is binomial while the alternative hypothesis is that the distribution is beta-binomial. The log-likelihood for the binomial model (interpreted to be pooling the data from all studies without weighting) is</p>", "<p></p>", "<p>The likelihood ratio test is</p>", "<p></p>", "<p>where <italic>L</italic>_{<italic>BB </italic>}is the log-likelihood value for the beta-binomial model (9) and <italic>L</italic>_{<italic>B </italic>}is log-likelihood value for the binomial model (15).</p>", "<p>Although a solution for the boundary problem has been offered [##UREF##15##20##], there is no consensus on the optimal solution [##REF##2720053##21##]. To avoid the boundary problem, we can use the alternative – Tarone's Z statistic [##UREF##16##22##] – to test for overdispersion. This has been shown to be more sensitive than the parameter test (e.g. test for <italic>γ </italic>being zero) and the log-likelihood ratio test [##UREF##4##5##]:</p>", "<p></p>", "<p>where</p>", "<p></p>", "<p>This statistic <italic>Z </italic>has an asymptotic standard normal distribution under the null hypothesis of a binomial distribution. In short, we recommend caution in using the likelihood ratio test. It is better to combine it with Tarone's <italic>Z </italic>statistics. The <italic>Z </italic>statistics can also be used as a goodness-of-fit test. It has been shown to be superior to other goodness-of-fit measures [##REF##2720053##21##]. We will be calculating Tarone's <italic>Z </italic>in our application example.</p>", "<title>The Bayesian Approach</title>", "<p>In the preceding sections we describe the beta-binomial model within the frequentist framework of statistics. Interestingly, in the Bayesian statistics field, the beta-binomial model is commonly described in Bayesian statistics textbooks as an example [##UREF##17##23##,##UREF##18##24##]. Since Bayesian statistical methods are now increasingly used in clinical and public health research, we hereby briefly describe the derivation of the beta-binomial model in the Bayesian framework. Some have noted that the Bayesian approach can provide more accurate estimates for small samples [##UREF##19##25##,##UREF##20##26##].</p>", "<p>Recall that the binomial distribution (in equation 1) is the following:</p>", "<p></p>", "<p>Let the conjugate prior <italic>π</italic>(<italic>p</italic>|<italic>α, β</italic>) be a beta distribution (i.e., if <italic>p </italic>in equation 1 follows the beta distribution)</p>", "<p></p>", "<p>where <italic>Γ </italic>is the gamma function. The beta priors are selected because they are very flexible on (0, 1) and can represent a wide range of prior beliefs. These are similar to the reasons for selecting the beta distribution in the frequentist framework. In addition, by starting with the beta distribution as the conjugate prior, we ensure that the posterior distribution is always a beta distribution, and thus mathematically tractable for estimating the parameters.</p>", "<p>For notational convenience, let <italic>μ = α/(α+β), M = α+β </italic>(<italic>i.e. M </italic>= 1/<italic>θ</italic>), so that</p>", "<p></p>", "<p>In short, we again have a two-stage model:</p>", "<p></p>", "<p></p>", "<p>In the Bayesian terminology, the beta prior distribution, when updated with binomial data, gives a beta posterior distribution. The Bayesian estimator can then be chosen as the mean, median, or the mode of this marginal posterior. In many situations, as long as the sample sizes are reasonably large (n = 50 or more), our previous methods of moment estimation and maximum likelihood are still preferred in the Bayesian framework for the estimations of mean and variance. There are other detailed mathematical equations involved in Bayesian estimation of the beta-binomial model for specific cases. Interested readers could consult Lee and Sabavala [##UREF##19##25##] as well as Lee and Lio [##UREF##20##26##].</p>" ]

[ "<title>Results</title>", "<p>We will illustrate the application of the beta-binomial method using an analysis that examined the adverse effects of oral anti-fungal agents. Oral anti-fungal agents, including terbinafine, itraconazole, and fluconazole, have become the treatment of choice for onychomycosis and dermatophytosis not responding to topical therapy. In order to study the safety profiles of these agents, we reviewed data from randomized and non-randomized controlled trials, case series, and cohort studies that enrolled patients having superficial dermatophytosis (tinea pedis, tinea mannus, tinea copora, and tinea cruris) or onychomycosis, aged 18 or above, receiving oral antifungal therapy for two or more weeks. One outcome of interest was the cumulative incidence of patients who withdrew from the study because of adverse reactions [##REF##17765049##7##]. Data for 41 treatment arms of terbinafine from 37 studies (Table ##TAB##0##1## and Appendix) are used as an example.</p>", "<p>Event rates from different studies varied from 0 % to 13.89%. We apply the beta-binomial model with the maximum likelihood method to estimate the pooled event rates using SAS and SAS macro BETABIN. From all the eligible studies, we combine the data and obtain the summary estimate of risks and its 95% confidence intervals (CI).</p>", "<p>The ML estimates for parameters <italic>μ </italic>and <italic>θ </italic>are = 0.0344 and = 0.0278. The estimate of the covariance matrix for and is</p>", "<p></p>", "<p>In Table ##TAB##1##2##, we present different estimations of a pooled proportion (event rates) using the binomial model and the beta-binomial model. Using the binomial model, we compute a binomial probability and variance as if all the data were from a single study with a sample size of over 3,000. The pooled estimate is 3.70%, 8% higher than the beta-binomial estimate of 3.44%. The standard error from the collapsed data is 0.34%, misleadingly smaller than that of the beta-binomial estimation of 0.59%.</p>", "<p>The important issue naturally is the test of overdispersion since that is the basis for preferring the beta-binomial model in these situations. Results from different methods to evaluate overdispersion are presented in Table ##TAB##1##2##. As discussed in previous sections, <italic>θ </italic>and <italic>γ </italic>are indicators of overdispersion. They are significantly greater than zero in this case (p &lt; 0.05), indicating the presence of overdispersion. We also conduct a likelihood-ratio test between the beta-binomial and the binomial, and again the test shows that there is significant overdispersion (p &lt; 0.001). Finally, we calculate Tarone's <italic>Z </italic>statistic, and the result is consistent with other tests. It shows that the beta-binomial has better goodness-of-fit than the binomial (p &lt; 0.001). The fit that the beta-binomial model gives for our example is also graphically presented in Figure ##FIG##1##2##.</p>", "<p>As we have shown above, under the beta-binomial model the summary event rate is 3.44% with an estimated standard error of 0.59%. The <italic>θ </italic>is estimated to be 2.78% (Table ##TAB##1##2##), which gives an <italic>α </italic>estimate of 1.24 and a <italic>β </italic>estimate of 34.72. Once these parameters are estimated, we can use the estimated beta-binomial model to examine the probability of observing, for example, 105 or more adverse events in a new study of 1,000 subjects. Using equation 3, that probability is 5% under our estimated beta-binomial model.</p>" ]

[ "<title>Discussion</title>", "<p>Along with the development of drugs, vaccines, and medical products for unmet medical needs, more robust analytic methods are needed to quantify the risks associated with the use of these agents, so that regulators and clinicians can rigorously assess the risk-benefit profiles of medical products. While randomized controlled trials have been established as the gold standard for efficacy evaluation, comprehensive safety assessment requires a collection of different methods. As any single trial is rarely large enough to estimate precisely the probability of serious adverse events, large observational datasets or aggregations of clinical trial results are necessary. A recent high profile example [##REF##17517853##27##] illustrated the need to combine results from multiple studies to unearth safety signals that may not be apparent in individual studies. Developing on prior work by Chuang-Stein [##UREF##1##2##], we provide a more comprehensive background of the beta-binomial model, a model that could have wider application in clinical and public health research. In order to show new developments in the beta-binomial field over the past decade, we explain and demonstrate that the beta-binomial method can be used for the combination of heterogeneous studies to estimate event rates.</p>", "<p>Estimating the correct summary event rate based on heterogeneous binomial data is so far the main reason for adopting the beta-binomial distribution. Once this is accomplished, one might wish to examine whether specific attributes of the studies will have any meaningful impact. The beta-binomial model can incorporate these attributes into a regression model as covariates. For example, the main purpose of the study might be to evaluate the proportion of adverse events from all clinical trials involving drug A. Different studies might have different proportions of female subjects, and one may link the covariate, the proportion of female subjects, to the <italic>α </italic>parameter. In addition, different studies might include or exclude certain comorbid conditions. The comorbidity, defined as a binary variable, could also be included as a covariate. One can then evaluate the likelihood of the comorbidity increasing a specific side effect. As current meta-regression methods are mainly applied to comparative measures like relative risks, the advantage of the beta-binomial model is that it can assess the correlation between study attributes and absolute risks of events.</p>", "<p>Traditional meta-analysis can also combine event rates from heterogeneous sources by using the DerSimonian and Laird method [##REF##3802833##28##]. We applied this method to the same dataset and placed the summary rate in Table ##TAB##1##2##, with an estimate of 3.90% with a standard error of 0.51%. This is in good agreement with our estimation using the beta-binomial model. In medical product safety assessment, however, being able to derive a clear probability distribution offers advantages that traditional meta-analysis cannot, because the distributions allow the computation of absolute risks or probabilities involved in decision analysis. In the Bayesian framework, the beta-binomial model also enables better incorporation of prior knowledge and its associated uncertainty. In other words, even though traditional meta-analysis can also combine event rates, the adoption of the beta-binomial model can serve multiple purposes.</p>" ]

[ "<title>Conclusion</title>", "<p>In the process of pooling event rates from multiple studies, one must consider the existence of overdispersion and the adequacy of the binomial model. In the example that we have presented, we estimated the pooled proportion of adverse events using the beta-binomial model. While we mainly discussed the application in safety assessment, the same method can be applied to assessment of efficacy of treatment response [##UREF##21##29##].</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>The beta-binomial model is one of the methods that can be used to validly combine event rates from overdispersed binomial data. Our objective is to provide a full description of this method and to update and broaden its applications in clinical and public health research.</p>", "<title>Methods</title>", "<p>We describe the statistical theories behind the beta-binomial model and the associated estimation methods. We supply information about statistical software that can provide beta-binomial estimations. Using a published example, we illustrate the application of the beta-binomial model when pooling overdispersed binomial data.</p>", "<title>Results</title>", "<p>In an example regarding the safety of oral antifungal treatments, we had 41 treatment arms with event rates varying from 0% to 13.89%. Using the beta-binomial model, we obtained a summary event rate of 3.44% with a standard error of 0.59%. The parameters of the beta-binomial model took the values of 1.24 for alpha and 34.73 for beta.</p>", "<title>Conclusion</title>", "<p>The beta-binomial model can provide a robust estimate for the summary event rate by pooling overdispersed binomial data from different studies. The explanation of the method and the demonstration of its applications should help researchers incorporate the beta-binomial method as they aggregate probabilities of events from heterogeneous studies.</p>" ]

[ "<title>Appendix</title>", "<p>Studies Included in Table ##TAB##0##1##</p>", "<p>1. Alpsoy E, Yilmaz E, Basaran E. Intermittent therapy with terbinafine for dermatophyte toe-onychomycosis: a new approach. J Dermatol. 1996:23:259–262.</p>", "<p>2. Arca E, Taştan HB, Akar A, Kurumlu Z, Gür AR. An open, randomized, comparative study of oral fluconazole, itraconazole and terbinafine therapy in onychomycosis. J Dermatolog Treat. 2002:13:3–9.</p>", "<p>3. Arenas R, Dominguez-Cherit J, Fernandez LM. Open randomized comparison of itraconazole versus terbinafine in onychomycosis. Int J Dermatol. 1995:34:138–143.</p>", "<p>4. Avner S, Nir N, Henri T. Combination of oral terbinafine and topical ciclopirox compared to oral terbinafine for the treatment of onychomycosis. J Dermatolog Treat. 2005;16:327–330.</p>", "<p>5. Baldari U, Righini MG, Raccagni AA, et al. Comparative double blind, double dummy study on the efficacy and safety of fluconazole 100 mg/day versus terbinafine 250 mg/day in the treatment of dermatomycoses. G Ital Dermatol Venereol. 2000;135:229–235.</p>", "<p>6. Baran R, Belaich S, Beylot C, et al. Comparative multicentre doubleblind study of terbinafine (250 mg per day) versus griseofulvin (1 g per day) in the treatment of dermatophyte onychomycosis. J Dermatolog Treat. 1997;8:93–97.</p>", "<p>7. Baran R, Feuilhade M, Combernale P, et al. A randomized trial of amorolfine 5% solution nail lacquer combined with oral terbinafine compared with terbinafine alone in the treatment of dermatophytic toenail onychomycoses affecting the matrix region. Br J Dermatol. 2000;142:1177–1183.</p>", "<p>8. Brautigam M, Nolting S, Schopf RE, Weidinger G. Randomised double blind comparison of terbinafine and itraconazole for treatment of toenail tinea infection. Seventh Lamisil German Onychomycosis Study Group. BMJ. 1995;311:919–922.</p>", "<p>9. De Backer M, De Vroey C, Lesaffre E, et al. Twelve weeks of continuous oral therapy for toenail onychomycosis caused by dermatophytes: a double-blind comparative trial of terbinafine 250 mg/day versus itraconazole 200 mg/day. J Am Acad Dermatol. 1998;38 (5 Pt 3):S57–S63.</p>", "<p>10. De Keyser P, De Backer M, Massart DL, Westelinck KJ. Two-week oral treatment of tinea pedis, comparing terbinafine (250 mg/day) with itraconazole (100 mg/day): a double-blind, multicentre study. Br J Dermatol. 1994;130(Suppl 43):22–25.</p>", "<p>11. Degreef H, del Palacio A, Mygind S, et al. Randomized double-blind comparison of short-term itraconazole and terbinafine therapy for toenail onychomycosis. Acta Derm Venereol. 1999;79:221–223.</p>", "<p>12. del Palacio Hernandez A, Lopez Gomez S, Gonzalez Lastra F, et al. A comparative double-blind study of terbinafine (Lamisil) and griseofulvin in tinea corporis and tinea cruris. Clin Exp Dermatol. 1990;15:210–216.</p>", "<p>13. Drake LA, Shear NH, Arlette JP, et al. Oral terbinafine in the treatment of toenail onychomycosis: North American multicenter trial. J Am Acad Dermatol. 1997;37(5 Pt 1):740–745.</p>", "<p>14. Evans EG, Sigurgeirsson B. Double blind, randomised study of continuous terbinafine compared with intermittent itraconazole in treatment of toenail onychomycosis. The LION Study Group. BMJ. 1999;318:1031–1035.</p>", "<p>15. Faergemann J, Anderson C, Hersle K, et al. Double-blind, paralle-lgroup comparison of terbinafine and griseofulvin in the treatment of toenail onychomycosis. J Am Acad Dermatol. 1995;32(5 Pt 1):750–753.</p>", "<p>16. Goodfield MJ, Andrew L, Evans EG. Short term treatment of dermatophyte onychomycosis with terbinafine. BMJ. 1992;304:1151–1154.</p>", "<p>17. Goodfield MJ, Rowell NR, Forster RA, et al. Treatment of dermatophyte infection of the finger- and toe-nails with terbinafine (SF86-327, Lamisil), an orally active fungicidal agent. Br J Dermatol. 1989;121:753–757.</p>", "<p>18. Gupta AK, Gregurek-Novak T. Efficacy of itraconazole, terbinafine, fluconazole, griseofulvin and ketoconazole in the treatment of Scopulariopsis brevicaulis causing onychomycosis of the toes. Dermatology. 2001;202:235–238.</p>", "<p>19. Gupta AK, Konnikov N, Lynde CW, et al. Single-blind, randomized, prospective study on terbinafine and itraconazole for treatment of dermatophyte toenail onychomycosis in the elderly. J Am Acad Dermatol 2001; 44: 479–484.</p>", "<p>20. Haneke E, Tausch I, Brautigam M, et al. Short-duration treatment of fingernail dermatophytosis: a randomized, double-blind study with terbinafine and griseofulvin. LAGOS III Study Group. J Am Acad Dermatol. 1995;32:72–77.</p>", "<p>21. Havu V, Heikkila H, Kuokkanen K, et al. A double-blind, randomized study to compare the efficacy and safety of terbinafine (Lamisil) with fluconazole (Diflucan) in the treatment of onychomycosis. Br J Dermatol. 2000;142:97–102.</p>", "<p>22. Hay RJ, McGregor JM, Wuite J, et al. A comparison of 2 weeks of terbinafine 250 mg/day with 4 weeks of itraconazole 100 mg/day in plantar-type tinea pedis. Br J Dermatol. 1995;132:604–608.</p>", "<p>23. Hofmann H, Brautigam M, Weidinger G, Zaun H. Treatment of toenail onychomycosis. A randomized, double-blind study with terbinafine and griseofulvin. LAGOS II Study Group. Arch Dermatol. 1995;131:919–922.</p>", "<p>24. Honeyman JF, Talarico FS, Arruda LHF, et al. Itraconazole versus terbinafine (LAMISIL(registered trademark)): which is better for the treatment of onychomycosis? J Eur Acad Dermatol Venereol. 1997; 9:215–221.</p>", "<p>25. Kim JH, Yoon KB. Single-blind randomized study of terbinafine vs itraconazole in tinea pedis (two weeks vs four weeks). Terbinafine in the treatment of superficial fungal infections, edited by S. Shuster and M. H. Jafary, 1993; p17-20 Royal Society of Medicine Services International Congress and Symposium Series No. 205, published by Royal Society of Medicine Services Limited.</p>", "<p>26. Savin R. Successful treatment of chronic tinea pedis (moccasin type) with terbinafine (Lamisil). Clin Exp Dermatol. 1989;14:116–119.</p>", "<p>27. Savin RC, Zaias N. Treatment of chronic moccasin-type tinea pedis with terbinafine: a double-blind, placebo-controlled trial. J Am Acad Dermatol. 1990;23(4 Pt 2):804–807.</p>", "<p>28. Svejgaard EL, Brandrup F, Kragballe K, et al. Oral terbinafine in toenail dermatophytosis. A double-blind, placebo-controlled multicenter study with 12 months' follow-up. Acta Derm Venereol. 1997; 77:66–69.</p>", "<p>29. Tausch I, Brautigam M, Weidinger G, Jones TC. Evaluation of 6 weeks treatment of terbinafine in tinea unguium in a double-blind trial comparing 6 and 12 weeks therapy. The Lagos V Study Group. Br J Dermatol. 1997;136:737–742.</p>", "<p>30. Tausch I, Decroix J, Gwiezdzinski Z, et al. Short-term itraconazole versus terbinafine in the treatment of tinea pedis or manus. Int J Dermatol. 1998;37:140–142.</p>", "<p>31. Tosti A, Piraccini BM, Stinchi C, et al. Treatment of dermatophyte nail infections: an open randomized study comparing intermittent terbinafine therapy with continuous terbinafine treatment and intermittent itraconazole therapy. J Am Acad Dermatol. 1996;34:595–600.</p>", "<p>32. van der Schroeff JG, Cirkel PK, Crijns MB, et al. A randomized treatment duration-finding study of terbinafine in onychomycosis. Br J Dermatol. 1992;126(Suppl 39):36–39.</p>", "<p>33. Voravutinon V. Oral treatment of tinea corporis and tinea cruris with terbinafine and griseofulvin: a randomized double blind comparative study. J Med Assoc Thai. 1993;76:388–393.</p>", "<p>34. Warshaw, E.M., D.D. Fett, H.E. Bloomfield, J.P. Grill, D.B. Nelson, V. Quintero, S.M. Carver, G.R. Zielke and F.A. Lederle. Pulse versus continuous terbinafine for onychomycosis: a randomized, double-blind, controlled trial. J Am Acad Dermatol. 2005;53:578–584.</p>", "<p>35. Watson A, Marley J, Ellis D, Williams T. Terbinafine in onychomycosis of the toenail: a novel treatment protocol. J Am Acad Dermatol. 1995;33(5 Pt 1):775–779.</p>", "<p>36. Widyanto BU, Kuswadji KB. A randomized, double blind comparative study of terbinafine vs griseofulvin in tinea pedis. Terbinafine in the treatment of superficial fungal infections, edited by S. Shuster and M. H. Jafary, 1993, p21-24; Royal Society of Medicine Services International Congress and Symposium Series No. 205, published by Royal Society of Medicine Services Limited.</p>", "<p>37. Won YH, Kim SJ, Lee HW, Chun IK. Clinical comparative study of terbinafine and itraconazole in the treatment of tinea pedis. Terbinafine in the treatment of superficial fungal infections, edited by S. Shuster and M. H. Jafary, 1993, p7-10; Royal Society of Medicine Services International Congress and Symposium Series No. 205, published by Royal Society of Medicine Services Limited.</p>", "<title>Competing interests</title>", "<p>Both authors work part-time for for-profit companies (YYX for EpiPatterns and KAC for i3 Drug Safety). KAC received support from the Harvard Pharmacoepidemiology program, which has received unrestricted funds from pharmaceutical companies.</p>", "<title>Authors' contributions</title>", "<p>YYX and KAC conceived of the study. YYX performed the statistical analysis and wrote the manuscript. KAC participated in the analysis of the study and the writing of the manuscript. Both authors read and approved the final manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2288/8/58/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>We thank Dr. Chia-Hsuin Chang for acquisition of data. The study was funded by the Research and Teaching account of K. Arnold Chan at Harvard School of Public Health and the Harvard Pharmacoepidemiology Program. The Harvard Pharmacoepidemiology Program, which has received unrestricted funds from pharmaceutical companies, had no influence on study design or planning; on collection, analysis, or interpretation of data; on the writing of the manuscript; or on the decision to submit the manuscript for publication.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Variety of shapes for beta distributions.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>Beta distribution for the binomial proportions based on example.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Treatment arms of terbinafine included in pooled estimates</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\">Treatment <break/>Arm*<break/></td><td align=\"center\">Sample Size <break/>(No. of Patients – n_i)<break/></td><td align=\"center\">No. of Treatment <break/>Termination Due to<break/>Adverse Effect (x_i)</td><td align=\"center\">Proportion of Treatment <break/>Termination (p_i= x_i/n_i)<break/></td></tr></thead><tbody><tr><td align=\"right\">1</td><td align=\"right\">184</td><td align=\"right\">7</td><td align=\"right\">3.80%</td></tr><tr><td align=\"right\">2</td><td align=\"right\">65</td><td align=\"right\">1</td><td align=\"right\">1.54%</td></tr><tr><td align=\"right\">3</td><td align=\"right\">33</td><td align=\"right\">1</td><td align=\"right\">3.03%</td></tr><tr><td align=\"right\">4</td><td align=\"right\">151</td><td align=\"right\">4</td><td align=\"right\">2.65%</td></tr><tr><td align=\"right\">5</td><td align=\"right\">24</td><td align=\"right\">0</td><td align=\"right\">0.00%</td></tr><tr><td align=\"right\">6</td><td align=\"right\">30</td><td align=\"right\">0</td><td align=\"right\">0.00%</td></tr><tr><td align=\"right\">7</td><td align=\"right\">20</td><td align=\"right\">0</td><td align=\"right\">0.00%</td></tr><tr><td align=\"right\">8</td><td align=\"right\">22</td><td align=\"right\">0</td><td align=\"right\">0.00%</td></tr><tr><td align=\"right\">9</td><td align=\"right\">50</td><td align=\"right\">4</td><td align=\"right\">8.00%</td></tr><tr><td align=\"right\">10</td><td align=\"right\">50</td><td align=\"right\">5</td><td align=\"right\">10.00%</td></tr><tr><td align=\"right\">11</td><td align=\"right\">18</td><td align=\"right\">0</td><td align=\"right\">0.00%</td></tr><tr><td align=\"right\">12</td><td align=\"right\">26</td><td align=\"right\">0</td><td align=\"right\">0.00%</td></tr><tr><td align=\"right\">13</td><td align=\"right\">72</td><td align=\"right\">0</td><td align=\"right\">0.00%</td></tr><tr><td align=\"right\">14</td><td align=\"right\">30</td><td align=\"right\">1</td><td align=\"right\">3.33%</td></tr><tr><td align=\"right\">15</td><td align=\"right\">16</td><td align=\"right\">0</td><td align=\"right\">0.00%</td></tr><tr><td align=\"right\">16</td><td align=\"right\">26</td><td align=\"right\">2</td><td align=\"right\">7.69%</td></tr><tr><td align=\"right\">17</td><td align=\"right\">95</td><td align=\"right\">8</td><td align=\"right\">8.42%</td></tr><tr><td align=\"right\">18</td><td align=\"right\">95</td><td align=\"right\">3</td><td align=\"right\">3.16%</td></tr><tr><td align=\"right\">19</td><td align=\"right\">186</td><td align=\"right\">0</td><td align=\"right\">0.00%</td></tr><tr><td align=\"right\">20</td><td align=\"right\">146</td><td align=\"right\">11</td><td align=\"right\">7.53%</td></tr><tr><td align=\"right\">21</td><td align=\"right\">142</td><td align=\"right\">2</td><td align=\"right\">1.41%</td></tr><tr><td align=\"right\">22</td><td align=\"right\">124</td><td align=\"right\">8</td><td align=\"right\">6.45%</td></tr><tr><td align=\"right\">23</td><td align=\"right\">56</td><td align=\"right\">1</td><td align=\"right\">1.79%</td></tr><tr><td align=\"right\">24</td><td align=\"right\">12</td><td align=\"right\">0</td><td align=\"right\">0.00%</td></tr><tr><td align=\"right\">25</td><td align=\"right\">50</td><td align=\"right\">0</td><td align=\"right\">0.00%</td></tr><tr><td align=\"right\">26</td><td align=\"right\">88</td><td align=\"right\">3</td><td align=\"right\">3.41%</td></tr><tr><td align=\"right\">27</td><td align=\"right\">48</td><td align=\"right\">0</td><td align=\"right\">0.00%</td></tr><tr><td align=\"right\">28</td><td align=\"right\">75</td><td align=\"right\">4</td><td align=\"right\">5.33%</td></tr><tr><td align=\"right\">29</td><td align=\"right\">76</td><td align=\"right\">0</td><td align=\"right\">0.00%</td></tr><tr><td align=\"right\">30</td><td align=\"right\">56</td><td align=\"right\">1</td><td align=\"right\">1.79%</td></tr><tr><td align=\"right\">31</td><td align=\"right\">153</td><td align=\"right\">9</td><td align=\"right\">5.88%</td></tr><tr><td align=\"right\">32</td><td align=\"right\">68</td><td align=\"right\">1</td><td align=\"right\">1.47%</td></tr><tr><td align=\"right\">33</td><td align=\"right\">120</td><td align=\"right\">13</td><td align=\"right\">10.83%</td></tr><tr><td align=\"right\">34</td><td align=\"right\">44</td><td align=\"right\">0</td><td align=\"right\">0.00%</td></tr><tr><td align=\"right\">35</td><td align=\"right\">84</td><td align=\"right\">0</td><td align=\"right\">0.00%</td></tr><tr><td align=\"right\">36</td><td align=\"right\">21</td><td align=\"right\">0</td><td align=\"right\">0.00%</td></tr><tr><td align=\"right\">37</td><td align=\"right\">145</td><td align=\"right\">3</td><td align=\"right\">2.07%</td></tr><tr><td align=\"right\">38</td><td align=\"right\">83</td><td align=\"right\">10</td><td align=\"right\">12.05%</td></tr><tr><td align=\"right\">39</td><td align=\"right\">68</td><td align=\"right\">3</td><td align=\"right\">4.41%</td></tr><tr><td align=\"right\">40</td><td align=\"right\">30</td><td align=\"right\">3</td><td align=\"right\">10.00%</td></tr><tr><td align=\"right\">41</td><td align=\"right\">120</td><td align=\"right\">3</td><td align=\"right\">2.50%</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Estimation of proportion and tests of overdispersion</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\">Methods</td><td align=\"center\">Estimate</td><td align=\"center\">Standard Error</td><td align=\"center\">Lower 95% CI</td><td align=\"center\">Upper 95% CI</td></tr></thead><tbody><tr><td align=\"left\">Simple collapsed binomial</td><td align=\"right\">3.70%</td><td align=\"right\">0.34%</td><td align=\"right\">3.03%</td><td align=\"right\">4.34%</td></tr><tr><td align=\"left\">Beta-Binomial</td><td align=\"right\">3.44%</td><td align=\"right\">0.59%</td><td align=\"right\">2.28%</td><td align=\"right\">4.61%</td></tr><tr><td align=\"left\">Meta-analysis¹</td><td align=\"right\">3.90%</td><td align=\"right\">0.61%</td><td align=\"right\">2.70%</td><td align=\"right\">5.09%</td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td align=\"center\" colspan=\"5\">Test of Overdispersion</td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td/><td align=\"center\">Estimate</td><td align=\"center\">Standard Error</td><td align=\"center\">Statistic</td><td align=\"center\">p-value</td></tr><tr><td align=\"left\">Alpha</td><td align=\"right\">1.24</td><td align=\"right\">0.52</td><td align=\"right\">z = 2.40</td><td align=\"right\">0.02</td></tr><tr><td align=\"left\">Beta</td><td align=\"right\">34.7</td><td align=\"right\">15.08</td><td align=\"right\">z = 2.30</td><td align=\"right\">0.02</td></tr><tr><td align=\"left\">Theta</td><td align=\"right\">2.78%</td><td align=\"right\">1.20%</td><td align=\"right\">z = 2.31</td><td align=\"right\">0.02</td></tr><tr><td align=\"left\">Gamma</td><td align=\"right\">2.71%</td><td align=\"right\">1.14%</td><td align=\"right\">z = 2.38</td><td align=\"right\">0.02</td></tr><tr><td align=\"left\">Likelihood Ratio²</td><td/><td/><td align=\"right\">X₁²= 129.91</td><td align=\"right\">&lt; 0.001</td></tr><tr><td align=\"left\">Tarone's Z</td><td/><td/><td align=\"right\">z = 7.95</td><td align=\"right\">&lt; 0.001</td></tr></tbody></table></table-wrap>" ]

[ "<disp-formula id=\"bmcM1\"><label>(1)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M1\" name=\"1471-2288-8-58-i1\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mrow><mml:mtext>Prob</mml:mtext><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>X</mml:mi><mml:mo>=</mml:mo><mml:mi>x</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mrow><mml:mo>(</mml:mo><mml:mtable columnalign=\"left\"><mml:mtr><mml:mtd><mml:mi>n</mml:mi></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mi>x</mml:mi></mml:mtd></mml:mtr></mml:mtable><mml:mo>)</mml:mo></mml:mrow><mml:mo> </mml:mo><mml:msup><mml:mi>p</mml:mi><mml:mi>x</mml:mi></mml:msup><mml:msup><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>p</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mrow><mml:mi>n</mml:mi><mml:mo>−</mml:mo><mml:mi>x</mml:mi></mml:mrow></mml:msup><mml:mo>,</mml:mo></mml:mrow></mml:mtd><mml:mtd><mml:mrow><mml:mn>0</mml:mn><mml:mo>≤</mml:mo><mml:mi>p</mml:mi><mml:mo>≤</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:mtd></mml:mtr></mml:mtable></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM2\"><label>(2)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M2\" name=\"1471-2288-8-58-i2\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mtext>Beta</mml:mtext><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>p</mml:mi><mml:mo>|</mml:mo><mml:mi>α</mml:mi><mml:mo>,</mml:mo><mml:mi>β</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mi>Γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>+</mml:mo><mml:mi>β</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mrow><mml:mi>Γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mi>Γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>β</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mfrac><mml:msup><mml:mi>p</mml:mi><mml:mrow><mml:mi>α</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:msup><mml:msup><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>p</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mrow><mml:mi>β</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:msup></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM3\"><label>(3)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M3\" name=\"1471-2288-8-58-i3\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mtext>Prob</mml:mtext><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>X</mml:mi><mml:mo>=</mml:mo><mml:mi>x</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mrow><mml:mo>(</mml:mo><mml:mtable columnalign=\"left\"><mml:mtr><mml:mtd><mml:mi>n</mml:mi></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mi>x</mml:mi></mml:mtd></mml:mtr></mml:mtable><mml:mo>)</mml:mo></mml:mrow><mml:mfrac><mml:mrow><mml:mi>Γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>+</mml:mo><mml:mi>β</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mi>Γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>+</mml:mo><mml:mi>x</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mi>Γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>β</mml:mi><mml:mo>+</mml:mo><mml:mi>n</mml:mi><mml:mo>−</mml:mo><mml:mi>x</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mrow><mml:mi>Γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mi>Γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>β</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mi>Γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>+</mml:mo><mml:mi>β</mml:mi><mml:mo>+</mml:mo><mml:mi>n</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula><italic>X</italic>_{<italic>i </italic>}| <italic>p</italic>_{<italic>i </italic>}~<italic>Bin</italic>(<italic>n</italic>_{<italic>i</italic>}, <italic>p</italic>_{<italic>i</italic>})</disp-formula>", "<disp-formula><italic>p</italic>_{<italic>i</italic>}~<italic>Beta </italic>(<italic>μ, θ</italic>),<italic>i.i.d</italic></disp-formula>", "<disp-formula id=\"bmcM4\"><label>(4)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M4\" name=\"1471-2288-8-58-i4\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mrow><mml:mover accent=\"true\"><mml:mi>p</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mn>1</mml:mn><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>w</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:msub><mml:mover accent=\"true\"><mml:mi>p</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mi>i</mml:mi></mml:msub></mml:mrow></mml:mstyle></mml:mrow><mml:mi>w</mml:mi></mml:mfrac><mml:mo>,</mml:mo><mml:mi>w</mml:mi><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mn>1</mml:mn><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>w</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow></mml:mstyle></mml:mrow></mml:mtd><mml:mtd><mml:mrow><mml:mtext>and</mml:mtext></mml:mrow></mml:mtd><mml:mtd><mml:mrow><mml:msub><mml:mi>w</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:mn>1</mml:mn><mml:mo>+</mml:mo><mml:mi>γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:mn>1</mml:mn><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mfrac></mml:mrow></mml:mtd></mml:mtr></mml:mtable></mml:mrow></mml:semantics></mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M5\" name=\"1471-2288-8-58-i5\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>S</mml:mi><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>w</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo stretchy=\"false\">(</mml:mo><mml:msub><mml:mi>p</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow></mml:mstyle><mml:mo>−</mml:mo><mml:mover accent=\"true\"><mml:mi>p</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:msup><mml:mo stretchy=\"false\">)</mml:mo><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M6\" name=\"1471-2288-8-58-i6\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo>=</mml:mo><mml:mover accent=\"true\"><mml:mi>p</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:mrow></mml:semantics></mml:math></inline-formula>", "<disp-formula id=\"bmcM5\"><label>(5)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M7\" name=\"1471-2288-8-58-i7\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mover accent=\"true\"><mml:mi>γ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mi>S</mml:mi><mml:mo>−</mml:mo><mml:mover accent=\"true\"><mml:mi>p</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mover accent=\"true\"><mml:mi>q</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mrow><mml:mo>[</mml:mo><mml:mrow><mml:mstyle displaystyle=\"true\"><mml:mo>∑</mml:mo><mml:mrow><mml:mfrac><mml:mrow><mml:msub><mml:mi>w</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow></mml:mfrac></mml:mrow></mml:mstyle><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>w</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow><mml:mi>w</mml:mi></mml:mfrac><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mo>]</mml:mo></mml:mrow></mml:mrow><mml:mrow><mml:mover accent=\"true\"><mml:mi>p</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mover accent=\"true\"><mml:mi>q</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mrow><mml:mo>[</mml:mo><mml:mrow><mml:mstyle displaystyle=\"true\"><mml:mo>∑</mml:mo><mml:mrow><mml:mi>w</mml:mi><mml:mi>i</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>w</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow><mml:mi>w</mml:mi></mml:mfrac><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>−</mml:mo><mml:mstyle displaystyle=\"true\"><mml:mo>∑</mml:mo><mml:mrow><mml:mfrac><mml:mrow><mml:msub><mml:mi>w</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow></mml:mfrac><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>w</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow><mml:mi>w</mml:mi></mml:mfrac><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mstyle></mml:mrow></mml:mstyle></mml:mrow><mml:mo>]</mml:mo></mml:mrow></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M8\" name=\"1471-2288-8-58-i8\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mover accent=\"true\"><mml:mi>q</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo>=</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mover accent=\"true\"><mml:mi>p</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:mrow></mml:semantics></mml:math></inline-formula>", "<disp-formula><italic>θ </italic>= <italic>γ</italic>/(1 - <italic>γ</italic>)</disp-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M9\" name=\"1471-2288-8-58-i9\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mrow><mml:mover accent=\"true\"><mml:mi>p</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo>=</mml:mo><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>p</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow></mml:mstyle></mml:mrow><mml:mo>)</mml:mo></mml:mrow><mml:mo>/</mml:mo><mml:mi>k</mml:mi></mml:mrow></mml:mtd><mml:mtd><mml:mrow><mml:mtext>and</mml:mtext></mml:mrow></mml:mtd><mml:mtd><mml:mrow><mml:mi>S</mml:mi><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:msub><mml:mi>p</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow></mml:mstyle><mml:mo>−</mml:mo><mml:mover accent=\"true\"><mml:mi>p</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:msup><mml:mo stretchy=\"false\">)</mml:mo><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:mtd></mml:mtr></mml:mtable></mml:mrow></mml:semantics></mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M10\" name=\"1471-2288-8-58-i6\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo>=</mml:mo><mml:mover accent=\"true\"><mml:mi>p</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:mrow></mml:semantics></mml:math></inline-formula>", "<disp-formula id=\"bmcM6\"><label>(6)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M11\" name=\"1471-2288-8-58-i10\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mover accent=\"true\"><mml:mi>γ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mi>n</mml:mi><mml:mi>S</mml:mi></mml:mrow><mml:mrow><mml:mover accent=\"true\"><mml:mi>p</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mover accent=\"true\"><mml:mi>p</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo stretchy=\"false\">)</mml:mo><mml:mi>k</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>n</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mfrac><mml:mo>−</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mrow><mml:mi>n</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:mfrac><mml:mo>.</mml:mo></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M12\" name=\"1471-2288-8-58-i11\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>L</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>,</mml:mo><mml:mi>β</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∏</mml:mo><mml:mrow><mml:mi>x</mml:mi><mml:mo>=</mml:mo><mml:mn>0</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mtable columnalign=\"left\"><mml:mtr><mml:mtd><mml:mi>n</mml:mi></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mi>x</mml:mi></mml:mtd></mml:mtr></mml:mtable><mml:mo>)</mml:mo></mml:mrow><mml:mfrac><mml:mrow><mml:mi>B</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>+</mml:mo><mml:mi>x</mml:mi><mml:mo>,</mml:mo><mml:mi>β</mml:mi><mml:mo>+</mml:mo><mml:mi>n</mml:mi><mml:mo>−</mml:mo><mml:mi>x</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mrow><mml:mi>B</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>,</mml:mo><mml:mi>β</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mfrac></mml:mrow></mml:mstyle></mml:mrow></mml:semantics></mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M13\" name=\"1471-2288-8-58-i12\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>B</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>,</mml:mo><mml:mi>β</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mi>Γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mi>Γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>β</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mrow><mml:mi>Γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>+</mml:mo><mml:mi>β</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></inline-formula>", "<disp-formula id=\"bmcM7\"><label>(7)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M14\" name=\"1471-2288-8-58-i13\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>c</mml:mi><mml:mo>−</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>B</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>,</mml:mo><mml:mi>β</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>+</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>0</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>B</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>+</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>,</mml:mo><mml:mi>β</mml:mi><mml:mo>+</mml:mo><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo stretchy=\"false\">)</mml:mo><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mstyle></mml:mrow></mml:mstyle></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM8\"><label>(8)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M15\" name=\"1471-2288-8-58-i14\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mtable columnalign=\"left\"><mml:mtr columnalign=\"left\"><mml:mtd columnalign=\"left\"><mml:mrow><mml:mn>0</mml:mn><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mo>∂</mml:mo><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mi>L</mml:mi></mml:mrow><mml:mrow><mml:mo>∂</mml:mo><mml:mi>α</mml:mi></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>Δ</mml:mi><mml:mn>1</mml:mn></mml:msub><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>,</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>−</mml:mo></mml:mrow></mml:mstyle><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>Δ</mml:mi><mml:mn>1</mml:mn></mml:msub><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>+</mml:mo><mml:mi>β</mml:mi><mml:mo>,</mml:mo><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mstyle></mml:mrow></mml:mtd></mml:mtr><mml:mtr columnalign=\"left\"><mml:mtd columnalign=\"left\"><mml:mrow><mml:mn>0</mml:mn><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mo>∂</mml:mo><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mi>L</mml:mi></mml:mrow><mml:mrow><mml:mo>∂</mml:mo><mml:mi>β</mml:mi></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>Δ</mml:mi><mml:mn>1</mml:mn></mml:msub><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>β</mml:mi><mml:mo>,</mml:mo><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>−</mml:mo></mml:mrow></mml:mstyle><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>Δ</mml:mi><mml:mn>1</mml:mn></mml:msub><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>+</mml:mo><mml:mi>β</mml:mi><mml:mo>,</mml:mo><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mstyle></mml:mrow></mml:mtd></mml:mtr></mml:mtable></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M16\" name=\"1471-2288-8-58-i15\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msub><mml:mi>Δ</mml:mi><mml:mn>1</mml:mn></mml:msub><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>m</mml:mi><mml:mo>,</mml:mo><mml:mi>n</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mrow><mml:mi>m</mml:mi><mml:mo>+</mml:mo><mml:mi>n</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:mfrac><mml:mo>+</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mrow><mml:mi>m</mml:mi><mml:mo>+</mml:mo><mml:mi>n</mml:mi><mml:mo>−</mml:mo><mml:mn>2</mml:mn></mml:mrow></mml:mfrac><mml:mo>+</mml:mo><mml:mn>...</mml:mn><mml:mo>+</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mi>m</mml:mi></mml:mfrac></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M17\" name=\"1471-2288-8-58-i16\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mfrac><mml:mrow><mml:msup><mml:mo>∂</mml:mo><mml:mn>2</mml:mn></mml:msup><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mi>L</mml:mi></mml:mrow><mml:mrow><mml:mo>∂</mml:mo><mml:msup><mml:mi>α</mml:mi><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:mo>−</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>Δ</mml:mi><mml:mn>2</mml:mn></mml:msub><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>,</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>+</mml:mo></mml:mrow></mml:mstyle><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>Δ</mml:mi><mml:mn>2</mml:mn></mml:msub><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>+</mml:mo><mml:mi>β</mml:mi><mml:mo>,</mml:mo><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mstyle></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M18\" name=\"1471-2288-8-58-i17\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mfrac><mml:mrow><mml:msup><mml:mo>∂</mml:mo><mml:mn>2</mml:mn></mml:msup><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mi>L</mml:mi></mml:mrow><mml:mrow><mml:mo>∂</mml:mo><mml:msup><mml:mi>β</mml:mi><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:mo>−</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>Δ</mml:mi><mml:mn>2</mml:mn></mml:msub><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>β</mml:mi><mml:mo>,</mml:mo><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>+</mml:mo></mml:mrow></mml:mstyle><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>Δ</mml:mi><mml:mn>2</mml:mn></mml:msub><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>+</mml:mo><mml:mi>β</mml:mi><mml:mo>,</mml:mo><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mstyle></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M19\" name=\"1471-2288-8-58-i18\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mfrac><mml:mrow><mml:msup><mml:mo>∂</mml:mo><mml:mn>2</mml:mn></mml:msup><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mi>L</mml:mi></mml:mrow><mml:mrow><mml:mo>∂</mml:mo><mml:mi>α</mml:mi><mml:mo>∂</mml:mo><mml:mi>β</mml:mi></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>Δ</mml:mi><mml:mn>2</mml:mn></mml:msub><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>+</mml:mo><mml:mi>β</mml:mi><mml:mo>,</mml:mo><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mstyle></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M20\" name=\"1471-2288-8-58-i19\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msub><mml:mi>Δ</mml:mi><mml:mn>2</mml:mn></mml:msub><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>m</mml:mi><mml:mo>,</mml:mo><mml:mi>n</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mrow><mml:msup><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>m</mml:mi><mml:mo>+</mml:mo><mml:mi>n</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:mfrac><mml:mo>+</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mrow><mml:msup><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>m</mml:mi><mml:mo>+</mml:mo><mml:mi>n</mml:mi><mml:mo>−</mml:mo><mml:mn>2</mml:mn><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:mfrac><mml:mo>+</mml:mo><mml:mn>...</mml:mn><mml:mo>+</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mrow><mml:msup><mml:mi>m</mml:mi><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M21\" name=\"1471-2288-8-58-i20\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>L</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>,</mml:mo><mml:mi>β</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∏</mml:mo><mml:mrow><mml:mi>x</mml:mi><mml:mo>=</mml:mo><mml:mn>0</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:msup><mml:mrow><mml:mo stretchy=\"false\">[</mml:mo><mml:mi>P</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>x</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo stretchy=\"false\">]</mml:mo></mml:mrow><mml:mrow><mml:msub><mml:mi>f</mml:mi><mml:mi>x</mml:mi></mml:msub></mml:mrow></mml:msup></mml:mrow></mml:mstyle><mml:mo>.</mml:mo></mml:mrow></mml:semantics></mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M22\" name=\"1471-2288-8-58-i21\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msub><mml:mi>S</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>x</mml:mi><mml:mo>=</mml:mo><mml:mn>0</mml:mn></mml:mrow><mml:mi>i</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>f</mml:mi><mml:mi>x</mml:mi></mml:msub><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>0</mml:mn><mml:mo>,</mml:mo><mml:mn>1</mml:mn><mml:mo>,</mml:mo><mml:mn>2</mml:mn><mml:mo>,</mml:mo><mml:mn>...</mml:mn><mml:mo>,</mml:mo><mml:mi>n</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mstyle></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M23\" name=\"1471-2288-8-58-i22\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msub><mml:mi>S</mml:mi><mml:mi>k</mml:mi></mml:msub><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow></mml:mstyle></mml:mrow></mml:semantics></mml:math></inline-formula>", "<disp-formula id=\"bmcM9\"><label>(9)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M24\" name=\"1471-2288-8-58-i23\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>c</mml:mi><mml:mo>−</mml:mo><mml:msub><mml:mi>S</mml:mi><mml:mi>n</mml:mi></mml:msub><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mrow><mml:mi>n</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:munderover><mml:mrow><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>+</mml:mo><mml:mi>i</mml:mi><mml:mi>θ</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>+</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>0</mml:mn></mml:mrow><mml:mrow><mml:mi>n</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:munderover><mml:mrow><mml:mo>{</mml:mo><mml:msub><mml:mi>S</mml:mi><mml:mi>n</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:msub><mml:mi>S</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo stretchy=\"false\">)</mml:mo><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>μ</mml:mi><mml:mo>+</mml:mo><mml:mi>i</mml:mi><mml:mi>θ</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>+</mml:mo><mml:msub><mml:mi>S</mml:mi><mml:mrow><mml:mi>n</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>i</mml:mi></mml:mrow></mml:msub><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>μ</mml:mi><mml:mo>+</mml:mo><mml:mi>i</mml:mi><mml:mi>θ</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>}</mml:mo></mml:mrow></mml:mstyle></mml:mrow></mml:mstyle></mml:mrow></mml:semantics></mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M25\" name=\"1471-2288-8-58-i24\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M26\" name=\"1471-2288-8-58-i25\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M27\" name=\"1471-2288-8-58-i26\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mn>0</mml:mn><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mo>∂</mml:mo><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mi>L</mml:mi></mml:mrow><mml:mrow><mml:mo>∂</mml:mo><mml:mi>μ</mml:mi></mml:mrow></mml:mfrac><mml:mrow><mml:mo>|</mml:mo><mml:mrow><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo>,</mml:mo><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:mrow></mml:mrow><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>0</mml:mn></mml:mrow><mml:mrow><mml:mi>k</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:munderover><mml:mrow><mml:mo>{</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>S</mml:mi><mml:mi>k</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:msub><mml:mi>S</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:mi>μ</mml:mi><mml:mo>+</mml:mo><mml:mi>i</mml:mi><mml:mi>θ</mml:mi></mml:mrow></mml:mfrac><mml:mo>−</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>S</mml:mi><mml:mrow><mml:mi>k</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>i</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mrow><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>μ</mml:mi><mml:mo>+</mml:mo><mml:mi>i</mml:mi><mml:mi>θ</mml:mi></mml:mrow></mml:mfrac><mml:mo>}</mml:mo></mml:mrow></mml:mstyle></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM10\"><label>(10)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M28\" name=\"1471-2288-8-58-i27\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mn>0</mml:mn><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mo>∂</mml:mo><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mi>L</mml:mi></mml:mrow><mml:mrow><mml:mo>∂</mml:mo><mml:mi>μ</mml:mi></mml:mrow></mml:mfrac><mml:mrow><mml:mo>|</mml:mo><mml:mrow><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo>,</mml:mo><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:mrow></mml:mrow><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>0</mml:mn></mml:mrow><mml:mrow><mml:mi>k</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:munderover><mml:mrow><mml:mi>i</mml:mi><mml:mo>{</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>S</mml:mi><mml:mi>k</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:msub><mml:mi>S</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:mi>μ</mml:mi><mml:mo>+</mml:mo><mml:mi>i</mml:mi><mml:mi>θ</mml:mi></mml:mrow></mml:mfrac><mml:mo>+</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>S</mml:mi><mml:mrow><mml:mi>k</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>i</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mrow><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>μ</mml:mi><mml:mo>+</mml:mo><mml:mi>i</mml:mi><mml:mi>θ</mml:mi></mml:mrow></mml:mfrac><mml:mo>−</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>S</mml:mi><mml:mrow><mml:mi>k</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>i</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mrow><mml:mn>1</mml:mn><mml:mo>+</mml:mo><mml:mi>i</mml:mi><mml:mi>θ</mml:mi></mml:mrow></mml:mfrac><mml:mo>}</mml:mo></mml:mrow></mml:mstyle></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M29\" name=\"1471-2288-8-58-i28\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>H</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo>,</mml:mo><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mrow><mml:mo>[</mml:mo><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mrow><mml:mfrac><mml:mrow><mml:msup><mml:mo>∂</mml:mo><mml:mn>2</mml:mn></mml:msup><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mi>L</mml:mi></mml:mrow><mml:mrow><mml:mo>∂</mml:mo><mml:msup><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:mfrac></mml:mrow></mml:mtd><mml:mtd><mml:mrow><mml:mfrac><mml:mrow><mml:msup><mml:mo>∂</mml:mo><mml:mn>2</mml:mn></mml:msup><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mi>L</mml:mi></mml:mrow><mml:mrow><mml:mo>∂</mml:mo><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo>∂</mml:mo><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:mrow></mml:mfrac></mml:mrow></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mrow><mml:mfrac><mml:mrow><mml:msup><mml:mo>∂</mml:mo><mml:mn>2</mml:mn></mml:msup><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mi>L</mml:mi></mml:mrow><mml:mrow><mml:mo>∂</mml:mo><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo>∂</mml:mo><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:mrow></mml:mfrac></mml:mrow></mml:mtd><mml:mtd><mml:mrow><mml:mfrac><mml:mrow><mml:msup><mml:mo>∂</mml:mo><mml:mn>2</mml:mn></mml:msup><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mi>L</mml:mi></mml:mrow><mml:mrow><mml:mo>∂</mml:mo><mml:msup><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:mfrac></mml:mrow></mml:mtd></mml:mtr></mml:mtable></mml:mrow><mml:mo>]</mml:mo></mml:mrow></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M30\" name=\"1471-2288-8-58-i29\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>C</mml:mi><mml:mi>o</mml:mi><mml:mi>v</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo>,</mml:mo><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mrow><mml:mo>[</mml:mo><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mrow><mml:msubsup><mml:mover accent=\"true\"><mml:mi>σ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:mtd><mml:mtd><mml:mrow><mml:mi>ρ</mml:mi><mml:msub><mml:mover accent=\"true\"><mml:mi>σ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:msub><mml:msub><mml:mover accent=\"true\"><mml:mi>σ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:msub></mml:mrow></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mrow><mml:mi>ρ</mml:mi><mml:msub><mml:mover accent=\"true\"><mml:mi>σ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:msub><mml:msub><mml:mover accent=\"true\"><mml:mi>σ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:msub></mml:mrow></mml:mtd><mml:mtd><mml:mrow><mml:msubsup><mml:mover accent=\"true\"><mml:mi>σ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:mtd></mml:mtr></mml:mtable></mml:mrow><mml:mo>]</mml:mo></mml:mrow></mml:mrow></mml:semantics></mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M31\" name=\"1471-2288-8-58-i24\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M32\" name=\"1471-2288-8-58-i25\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<disp-formula id=\"bmcM11\"><label>(11)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M33\" name=\"1471-2288-8-58-i30\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo>±</mml:mo><mml:msub><mml:mi>Z</mml:mi><mml:mrow><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>α</mml:mi><mml:mo>/</mml:mo><mml:mn>2</mml:mn></mml:mrow></mml:msub><mml:msub><mml:mover accent=\"true\"><mml:mi>σ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:msub></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM12\"><label>(12)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M34\" name=\"1471-2288-8-58-i31\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo>±</mml:mo><mml:msub><mml:mi>Z</mml:mi><mml:mrow><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>α</mml:mi><mml:mo>/</mml:mo><mml:mn>2</mml:mn></mml:mrow></mml:msub><mml:msub><mml:mover accent=\"true\"><mml:mi>σ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:msub></mml:mrow></mml:semantics></mml:math></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M35\" name=\"1471-2288-8-58-i24\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M36\" name=\"1471-2288-8-58-i25\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M37\" name=\"1471-2288-8-58-i32\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>α</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M38\" name=\"1471-2288-8-58-i33\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>β</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M39\" name=\"1471-2288-8-58-i32\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>α</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M40\" name=\"1471-2288-8-58-i24\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M41\" name=\"1471-2288-8-58-i25\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M42\" name=\"1471-2288-8-58-i33\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>β</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M43\" name=\"1471-2288-8-58-i24\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M44\" name=\"1471-2288-8-58-i25\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<disp-formula id=\"bmcM13\"><label>(13)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M45\" name=\"1471-2288-8-58-i34\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>E</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:msub><mml:mi>p</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mi>μ</mml:mi><mml:mo>=</mml:mo><mml:mfrac><mml:mi>α</mml:mi><mml:mrow><mml:mi>α</mml:mi><mml:mo>+</mml:mo><mml:mi>β</mml:mi></mml:mrow></mml:mfrac><mml:mo>,</mml:mo><mml:mi>V</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:msub><mml:mi>p</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mi>μ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>μ</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mi>γ</mml:mi></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM15\"><label>(15)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M46\" name=\"1471-2288-8-58-i35\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mi>L</mml:mi><mml:mo>=</mml:mo><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mrow><mml:mo>(</mml:mo><mml:mtable columnalign=\"left\"><mml:mtr><mml:mtd><mml:mi>n</mml:mi></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mi>x</mml:mi></mml:mtd></mml:mtr></mml:mtable><mml:mo>)</mml:mo></mml:mrow><mml:mo>+</mml:mo><mml:mi>y</mml:mi><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>p</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>+</mml:mo><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>n</mml:mi><mml:mo>−</mml:mo><mml:mi>x</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>p</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>.</mml:mo></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM16\"><label>(16)</label><italic>χ</italic>₁²= 2 (<italic>L</italic>_{<italic>BB </italic>}- <italic>L</italic>_{<italic>B</italic>})</disp-formula>", "<disp-formula id=\"bmcM14\"><label>(14)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M47\" name=\"1471-2288-8-58-i36\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>Z</mml:mi><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mi>E</mml:mi><mml:mo>−</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow></mml:mstyle></mml:mrow><mml:mrow><mml:msqrt><mml:mrow><mml:mn>2</mml:mn><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow></mml:mstyle><mml:mo stretchy=\"false\">(</mml:mo><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:mn>1</mml:mn><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:msqrt></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M48\" name=\"1471-2288-8-58-i37\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mrow><mml:mi>E</mml:mi><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:mfrac><mml:mrow><mml:msup><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:msub><mml:mi>n</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mover accent=\"true\"><mml:mi>p</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mn>2</mml:mn></mml:msup></mml:mrow><mml:mrow><mml:mover accent=\"true\"><mml:mi>p</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mover accent=\"true\"><mml:mi>p</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mfrac></mml:mrow></mml:mstyle></mml:mrow></mml:mtd><mml:mtd><mml:mrow><mml:mtext>and</mml:mtext></mml:mrow></mml:mtd><mml:mtd><mml:mrow><mml:mover accent=\"true\"><mml:mi>p</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:munderover><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>k</mml:mi></mml:munderover><mml:mrow><mml:mfrac><mml:mrow><mml:msub><mml:mi>x</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mi>n</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mrow></mml:mfrac></mml:mrow></mml:mstyle><mml:mo>.</mml:mo></mml:mrow></mml:mtd></mml:mtr></mml:mtable></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M49\" name=\"1471-2288-8-58-i38\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mtext>Prob</mml:mtext><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>X</mml:mi><mml:mo>=</mml:mo><mml:mi>x</mml:mi><mml:mo>|</mml:mo><mml:mi>p</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mrow><mml:mo>(</mml:mo><mml:mtable columnalign=\"left\"><mml:mtr><mml:mtd><mml:mi>n</mml:mi></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mi>x</mml:mi></mml:mtd></mml:mtr></mml:mtable><mml:mo>)</mml:mo></mml:mrow><mml:mo> </mml:mo><mml:msup><mml:mi>p</mml:mi><mml:mi>x</mml:mi></mml:msup><mml:msup><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>p</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mrow><mml:mi>n</mml:mi><mml:mo>−</mml:mo><mml:mi>x</mml:mi></mml:mrow></mml:msup></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM17\"><label>(17)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M50\" name=\"1471-2288-8-58-i39\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mtext>Beta</mml:mtext><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>p</mml:mi><mml:mo>|</mml:mo><mml:mi>α</mml:mi><mml:mo>,</mml:mo><mml:mi>β</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mi>Γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo>+</mml:mo><mml:mi>β</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mrow><mml:mi>Γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>α</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mi>Γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>β</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mfrac><mml:msup><mml:mi>p</mml:mi><mml:mrow><mml:mi>α</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:msup><mml:msup><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>p</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mrow><mml:mi>β</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:msup></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M51\" name=\"1471-2288-8-58-i40\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mtext>Beta</mml:mtext><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>p</mml:mi><mml:mo>|</mml:mo><mml:mi>α</mml:mi><mml:mo>,</mml:mo><mml:mi>β</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mi>Γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>M</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mrow><mml:mi>Γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>μ</mml:mi><mml:mi>M</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mi>Γ</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mi>M</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>μ</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow></mml:mfrac><mml:msup><mml:mi>p</mml:mi><mml:mrow><mml:mi>M</mml:mi><mml:mi>μ</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:msup><mml:msup><mml:mrow><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>p</mml:mi><mml:mo stretchy=\"false\">)</mml:mo></mml:mrow><mml:mrow><mml:mi>M</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>μ</mml:mi><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:msup></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula><italic>X</italic>_{<italic>i</italic>}|<italic>p</italic>_{<italic>i</italic>}~<italic>Bin</italic>(<italic>n</italic>_{<italic>i</italic>}, <italic>p</italic>_{<italic>i</italic>})</disp-formula>", "<disp-formula><italic>p</italic>_{<italic>i</italic>}~<italic>Beta </italic>(<italic>μ, M</italic>), <italic>i.i.d</italic></disp-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M52\" name=\"1471-2288-8-58-i24\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M53\" name=\"1471-2288-8-58-i25\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M54\" name=\"1471-2288-8-58-i24\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M55\" name=\"1471-2288-8-58-i25\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<disp-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M56\" name=\"1471-2288-8-58-i41\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>C</mml:mi><mml:mi>o</mml:mi><mml:mi>v</mml:mi><mml:mo stretchy=\"false\">(</mml:mo><mml:mover accent=\"true\"><mml:mi>μ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo>,</mml:mo><mml:mover accent=\"true\"><mml:mi>θ</mml:mi><mml:mo>^</mml:mo></mml:mover><mml:mo stretchy=\"false\">)</mml:mo><mml:mo>=</mml:mo><mml:mrow><mml:mo>[</mml:mo><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mrow><mml:mn>0.00004</mml:mn></mml:mrow></mml:mtd><mml:mtd><mml:mrow><mml:mn>0.00002</mml:mn></mml:mrow></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mrow><mml:mn>0.00002</mml:mn></mml:mrow></mml:mtd><mml:mtd><mml:mrow><mml:mn>0.00013</mml:mn></mml:mrow></mml:mtd></mml:mtr></mml:mtable></mml:mrow><mml:mo>]</mml:mo></mml:mrow><mml:mo>.</mml:mo></mml:mrow></mml:semantics></mml:math></disp-formula>" ]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>*Referenced separately in appendix</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1471-2288-8-58-1\"/>", "<graphic xlink:href=\"1471-2288-8-58-2\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>'What happens to research-based findings after they are completed and published?' This is a question heard more often with the qualitative and quantitative development of research. In the 2004 World Health Organization report on 'knowledge for better health', 'linking research to action' was emphasized, and countries were asked to take serious steps in transferring research-based knowledge [##UREF##0##1##]. Knowledge transfer methods have been classified into active and passive strategies from researchers' perspective [##UREF##1##2##]. In passive strategies, the aim is diffusion and basically changing the awareness of the target audience. Normally, these activities are of importance in the academic environment, and are indicated by the publication of articles in peer-reviewed journals. Conversely, active strategies are based on interaction with the users of research results, and the possibility of behavior change is higher in these cases [##REF##8192299##3##].</p>", "<p>Iran's health systems infrastructure is what makes its medical research unique among other countries. In 1985, Iranian medical schools were integrated into the Ministry of Health, and the Ministry of Health and Medical Education (MOHME) was created. Under this infrastructure, education, research, and service delivery were unified [##REF##8736181##4##], and it was expected that knowledge transfer would take place more effectively. In addition, in the past two decades the number of scientific publications in Iran has considerably increased [##UREF##2##5##], and the number of articles published in ISI journals with medical science content has doubled from 1997 to 2001 [##UREF##3##6##]. Tehran University of Medical Sciences (TUMS) has 1,250 academic members, or 12% of the country's medical academic members. Also, TUMS-affiliated researchers publish more than 30% of Iran's medical scientific articles in international databases.</p>", "<p>The first objective of this study was to determine the frequency of various knowledge transfer activities applied by researchers at TUMS, and the second objective was to find the determining factors leading to the type of strategy ('active' or 'passive'). The findings of this study build a foundation upon which interventions in knowledge utilization can be studied in the future.</p>" ]

[ "<title>Methods</title>", "<title>Data-gathering tools</title>", "<p>The tools for data-gathering consisted of two sections: the data-gathering form (checklist), which was filled by the research team using research proposals and final reports [see Additional File ##SUPPL##0##1##], and the researcher's questionnaire (self-administered) which was sent to the principle investigators (a maximum of three times at one month intervals) [see Additional File ##SUPPL##1##2##].</p>", "<p>The content validity of the questionnaire was approved after literature review and peer review. Pre-testing was done to assess feasibility; face validity, and reliability. A pilot study was performed on 10 data-gathering forms by studying 10 files and creating necessary changes. Also, 20 researchers completed the questionnaire twice at two week intervals to assess repeatability and internal consistency of the questions. The intra-class correlation indicator, which was considered the repeatability indicator, was 0.69 and 0.72 for the domains under study (active and passive strategies domains). The internal consistency (Cronbach's alpha) of these domains was 0.63 and 0.76. The questionnaire included the following variables: the percentage of time the participants allocated to research activities, the 'reasons for choosing the research topic', and the researchers' performances in knowledge transfer activities.</p>", "<p>In order to study their role in knowledge transfer activities, researchers were asked to mark all the activities they had carried out in the field of knowledge transfer (including active and passive strategies) from a list that was presented to them. We also left an open-ended question for the activities that were not listed in the above-mentioned questions. A score of zero was given if the activity was not carried out; a score of one if it was performed once, and a score of two if it was done more than once. The total score then was summed for each research activity. The following activities were considered 'passive' strategies of knowledge transfer: delivery of the project report or its summary to users; preparing articles and publishing reports in domestic and international peer-reviewed journals; displaying results on a website; posting or e-mailing articles or reports and/or their summaries for stakeholders without their request; and presenting the results in domestic or international conferences and seminars, and/or publishing research results in newspapers. The activities that were considered 'active' for knowledge transfer were as follows: preparation and delivery of content in plain language; holding briefings with stakeholders for presentation of research results; and presenting results to the media and participation in interviews. Also, we asked researchers to note the percentage of time, or 'percent effort' they allocated to each activity, including research, education, clinical service delivery, executive responsibilities, and others. Researchers were then asked to estimate their percent effort in a way that the sum would be equal to 100 (Question 6, Additional File ##SUPPL##1##2##).</p>", "<title>Population under study</title>", "<p>All TUMS research projects that received grants from inside and outside the university in 2004 and were completed by the time this study was performed (the second half of 2006) were studied. The number of research projects that met the inclusion criteria of this study was 315, out of which the data-gathering forms were completed for 301 projects (95.6%). Fourteen projects were not entered into the study due to unavailability of files. The researcher questionnaire was then sent to the principle investigators of these projects, and 208 questionnaires were collected. Non-responders included 32 researchers who were unavailable and 75 who did not respond after three requests, giving a final response rate of 74%. In order to assess whether a significant difference existed between those researchers who responded to the questionnaire and those who did not, their project proposal forms were compared. This was carried out by reviewing the 'problem statement' of the research proposals. We observed that 24% of the individuals who did not respond to the questionnaire mentioned choosing their topics on the basis of needs assessment. This proportion was 17% for those who responded to the questionnaire. The difference between these two groups was not statistically significant (p = 0.17).</p>", "<title>Data analysis</title>", "<p>Apart from the usual descriptive statistics for data analysis, multi-variable linear regression was used to control the effect of the potential confounders, including gender, number of years working as a professional, and tenure status (half-time or full-time). For these purposes, the data were analyzed with SPSS/version 11.5 statistical software.</p>", "<title>Ethical considerations</title>", "<p>This study was approved by the TUMS ethics review board as part of the reviewing process of TUMS research projects.</p>" ]

[ "<title>Results</title>", "<title>Population under study</title>", "<p>A total of 208 researchers participated, 130 of whom were male (62.5%). The age range was 25 to 72 years, and the mean age was 45.6 years (SD = 9.4). Regarding academic rank, 15% of researchers were non-academic members, 7% were instructors, and 33%, 26%, and 19% were assistant, associate, or full professors, respectively. Employment status included 181 (87%) full-time employees and 10 (4.8%) part-time employees. The remaining respondents did not answer this question. Number of years working as a professional ranged from one to 43 years, and the mean number of years working in the university was 14.3 (SD = 8.5). Aside from education and research, 123 individuals had executive responsibilities such as management of a hospital, school, department or ward, research deputy of the school, and/or research center, <italic>etc</italic>. Seventy-two individuals (34.6%) were involved solely in education and/or research.</p>", "<p>The research projects were divided into three groups according to proposal type. There are two formats of proposals at TUMS. One is health system research, in which the end-users are policy makers, managers, and health system experts. The other format is for clinical and basic studies, where the researcher chooses which category the proposal most addresses. Nevertheless we confirmed the validity of their choice by checking whether the targets of research were clinical practitioners, basic researchers, or health system researchers. (<italic>e.g.</italic>, a study that is carried out to better understand a topic and has no immediate clinical application is a basic study, a study whose results are directly used by the clinician is a clinical study, and a study whose results are used by managers and policy makers is a health system research study). The researchers were then divided into basic sciences (46 cases), clinical studies (101 cases), and health system research (61 cases). Comparing the duration of time allocated to research in these three groups showed that the mean percentage of time allocated to research in the basic sciences group was 41% (SD = 22), and a significant difference (p &lt; 0.001) was observed between this group and the clinical (27%, SD = 16) and health system research (30%, SD = 19)groups, respectively. Researchers were asked about their reasons for choosing the research topic. Thirty-one participants (14.9%) stated 'personal interest or repeating others research'. This proportion was 23.9% for the basic sciences, 7.9% for clinical studies, and 19.7% in health system researchers (p = 0.02), whereas the remainder mentioned choosing their topics based on 'other organizations request or needs assessment'.</p>", "<title>The knowledge transfer status (First objective)</title>", "<title>Information gathered from the self-administered questionnaire</title>", "<p>Table ##TAB##0##1## shows researcher behavior with respect to passive strategies of knowledge transfer. The first four rows of this table (publishing articles in peer-reviewed journals and presentations at conferences) are criteria which are valued in the assessment of academic staff members, whereas the other criteria are of no value. In all types of research, the researchers stated that publishing in peer-reviewed journals had the greatest impact in disseminating research results. Most basic science research was sent to international journals (71.7%), and most clinical and health system research was sent to domestic journals (74.3% and 57.3% respectively). The last row of this table shows that the least effort made by researchers is for publishing research results in newspapers, which was found in only eight out of 208 cases (4%).</p>", "<p>Table ##TAB##1##2## shows the active strategies of knowledge transfer. In all three fields of basic, clinical, and health system research, the step taken most often was 'preparing and delivering text in plain language'. 'Holding briefings with stakeholders for presentation of research results was also frequently cited for health system research, but presenting results in the media was of little significance.</p>", "<title>Information gathered from files (research proposals and final reports)</title>", "<p>A review of 301 research proposals showed that the total budget of the projects under study was a little less than US$1,290,000: US$324,280 for health system research, US$488,030 for clinical research and US$471,380 for basic research. The total expense considered for knowledge transfer for 301 projects was approximately US$13,200: US$12,790 for health system research, US$376 for clinical research, and none for basic research. This amount was spent on only seven cases (2.3%), of which five were health system research and two were clinical research. In this analysis, a significant difference was found to exist between the groups in this regard, and in the <italic>post hoc </italic>analysis this difference was insignificant among the clinical and basic research groups alone, but the cost for knowledge transfer activities in health system research was significantly higher than that for clinical and basic sciences.</p>", "<p>A review of the project final reports showed that in 142 final reports and/or project summaries (47.2%) the target audience had been identified. In this case, a significant difference did not exist between the three groups (basic, clinical, and health system research) (p = 0.28). In 150 project reports (49.8%), a clear suggestion had been made to the target audience. Even here a significant difference did not exist between the groups (p = 0.11). Of all 150 final reports examined, 87.3% of these suggestions had somehow pointed to the manner of the measure to be taken, but in 37.3% it had been made clear as to who had to take what measure.</p>", "<title>Determinant factors of knowledge transfer (Second objective)</title>", "<p>In the 'passive' strategies section, the maximum score attainable was 18. The mean score for researchers' performance was 4.00 (SD = 3.03) that formed 22% of the total score. The maximum score attainable in the active strategies was six and the mean score of the researchers' performance in these strategies was 0.54 (SD = 1.02), which consisted of only 9% of the total score. Table ##TAB##2##3## and ##TAB##3##4## show the results of a linear regression analysis with the 'Enter' method. As shown in tables ##TAB##2##3## and ##TAB##3##4##, the dependent variables in these regressions are the scores of passive and active strategies, respectively. These scores were obtained from the number of activities the researchers claimed to have carried out, whereas the independent variables included gender, number of years working as a professional, tenure status (half-time or full-time), reasons for choosing the research topic, and type of research (basic sciences were taken as reference with respect to clinical and health system research). Controlling the confounding variables, regression coefficients show the effect of each of these variables on passive and active strategy scores. In table ##TAB##2##3##, the number of years working as a professional and health system research (as compared to basic research) have a significant inverse relationship with the passive strategy scores, whereas choice of the research topic based on other organizations' request or needs assessment increases the score significantly. According to the results of the linear regression analysis in table ##TAB##3##4##, health system research and executive responsibilities had a significant effect on this score.</p>" ]

[ "<title>Discussion</title>", "<p>This study shows that passive strategies hold a greater share of knowledge transfer activities as compared to active ones in TUMS. While TUMS researchers have gained 22% of the total score for passive strategies of knowledge transfer (including preparation of articles for publication in domestic and international peer-reviewed journals, presenting research results at conferences and seminars, <italic>etc</italic>), when it comes to active strategies of knowledge transfer (preparation and delivery of texts suitable to the users, presenting results to mass media, and holding briefings with stakeholders) this percentage amounts to 9% of the total score. The result is that the score obtained for passive strategies of knowledge transfer is 2.44 times greater than the scores obtained for active strategies.</p>", "<p>Regarding publication of results in journals, according to the research regulations of TUMS at the time of this study, sending at least one article for publication from each project was one of the requirements for completing the project. This is why publication of articles in peer-reviewed journals is the most common knowledge transfer activity. According to table ##TAB##0##1##, basic science research studies are published more in international journals than in domestic journals as compared to health system research. This may be because basic science research is less dependent on the location of research. On the other hand, health system research studies that are more dependent on cultural, social, economic, and other contextual factors target domestic journals more than international ones.</p>", "<p>When examining other passive strategies of knowledge transfer we observed that less than 19% of the researchers have displayed the results of their research on websites. The other point worth mentioning is that less than 4% of research results were published in newspapers. Newspapers and websites are important because they have broad geographical coverage and transcend time barriers, even though the evidence should be considered before presenting it to the media; not every research result can be disseminated. Tables ##TAB##0##1## and ##TAB##1##2## show the performance of TUMS' researchers is in accordance with the requirements of the academic promotion criteria. This emphasizes that incentive policies (recruitment, academic members' promotion, and granting financial rewards for publishing articles) are effective. On the contrary, other matters that can lead to implementation of research findings have not received similar attention. In fact, the current state of knowledge production dominant in this university (like most universities in the world) is passive, and for strengthening the connection of 'linking knowledge to action', basic changes are needed.</p>", "<p>Valuing scientific productions (such as publishing articles in peer-reviewed journals and presentation of material at scientific conferences) are among the known factors affecting the knowledge transfer activities of academics [##UREF##4##7##, ####REF##9455024##8##, ##REF##8870150##9##, ##REF##1462179##10##, ##UREF##5##11##, ##UREF##6##12####6##12##]. The known methods of valuing are employment and promotion [##REF##9455024##8##,##UREF##7##13##,##REF##10144898##14##]. When matters such as professional progress are solely dependent on publishing in specialized frameworks, people are not motivated enough in transferring knowledge, and guaranteeing its utilization. For the sake of meeting communities' needs, current efforts are being made to revise the promotion and employment criteria from a new perspective [##UREF##8##15##, ####UREF##9##16##, ##UREF##10##17####10##17##]. On the other hand, intrinsic motivations such as researchers' perceptions, values, and beliefs are influential in this field; how these beliefs are shaped and to what extent they are influenced by education are matters which demand deeper qualitative approaches [##REF##12550152##18##].</p>", "<p>Regarding tables ##TAB##2##3## and ##TAB##3##4##, we note that the method of summing up the scores of knowledge transfer activities as equal weight for various cases is a simple and optional approach. Linear regression analysis was done by entering all variables into the model. This type of analysis was chosen because, compared to other models that try to keep fewer variables in the final model, it has an exploratory aspect, and from the authors' point of view a better understanding of the variables in this field is necessary.</p>", "<p>However, the result of the linear regression analysis showed that the scores of passive strategies of knowledge transfer decreased with the number of years working as a professional. That is, considering that the other variables are constant, with every one-year increase in number of years working as a professional, this score decreases by 0.08. The relationship between choosing the research topic (choice based on other organizations' request or need assessment versus personal interest or repeating others research) and the passive strategy score is positive. The passive strategy scores increase by 1.68 as a result of change of 'reasons for choosing the research topic' from 'personal interest or repeating others researches' to 'choice based on other organizations' request or need assessment'. The health system researchers also registered a lower score as compared to the basic science researchers, which leads to a 1.55 reduction in the passive strategy score.</p>", "<p>Where active strategies are concerned, two variables were significant: First, executive responsibilities can significantly reduce the active strategies score by 0.36. This can be explained by the shortage of time this group is faced with. Second, as compared to basic science research, health system research increased the active strategy score by 0.51.</p>", "<p>As shown in the tables, health system research registered lower scores in the passive strategies of knowledge transfer as compared to basic sciences, whereas in the active field of strategies the reverse was true. The scores registered by health system research were higher than basic sciences.</p>", "<p>Studies of researchers from other countries have shown differences in knowledge transfer activities among various specialties. In a study done on researchers in Canada it was seen that applied science researchers use plain and engaged dissemination measures more than basic science researchers. Apart from the field of research (applied or basic) the researchers' working locations (medical school and others) have also been taken into consideration. Comparing the various methods of knowledge transfer, both these variables were shown to be significantly effective. Their interaction has also shown to be effective in the number of publications in this study [##REF##17204143##19##].</p>", "<p>After studying the final project reports, it was shown that almost 50% of them had proposed a suggestion for utilization of results (although a formal compulsory framework does not exist for writing the final report and having an actionable message). This shows that researchers need to pay more attention to knowledge transfer and that by valuing activities in this field, results can be properly utilized. Also, the target audiences of these messages were clear in 47.2% of cases, even though there is no compulsion for mentioning the target audiences. This shows that if researchers receive basic training for increasing their communication skills we will achieve more satisfactory results. This matter has been mentioned in other references and also been advised [##REF##12841049##20##].</p>", "<p>Review of the research proposals showed that in only 2.3% of the 301 cases under study, expenses for knowledge transfer activities had been considered, amounting to 1% of the funds requested. There are two reasons for this observation: Some researchers fail to consider knowledge transfer to be a part of research at all, and those who evaluate the cost of research (proposal reviewers at TUMS) find these costs unacceptable.</p>", "<p>No doubt knowledge transfer activities require financial resources, be it in the form of cash paid for direct costs (such as preparation and handing out pamphlets or the cost of setting up meetings), or indirect costs (such as purchasing knowledge transfer services). Many authors have stated the lack of these facilities and funds to be potential barriers to the knowledge transfer process [##REF##9455024##8##,##UREF##5##11##,##UREF##7##13##,##UREF##11##21##,##UREF##12##22##].</p>", "<p>Because many of the study's data are based on the self-administered questionnaire, it is possible that responders may have overestimated their knowledge transfer activities. This may be due to the social undesirability of the answers that point to lack of knowledge transfer activity. Therefore this study may be prone to information bias in describing knowledge transfer activities, despite the fact that the questionnaire had been evaluated for repeatability and internal consistency prior to the study. This information bias can affect the first descriptive objective but we do not assume the second objective, <italic>i.e.</italic>, study of determinant factors, to be biased as a result of this.</p>" ]

[ "<title>Conclusion</title>", "<p>This study was carried out in one of the universities of a Middle Eastern developing country. Here we observe that, like many other universities in the world, many academicians still do not give priority to active strategies. Even though previous studies have shown that many factors affect the facilitation of knowledge transfer in the university [##UREF##13##23##], but the matter of giving priority to knowledge transfer largely depends on academic priorities which are shown in its policies. Therefore if knowledge transfer is to be a priority, it is necessary to introduce considerable changes in academic procedures and incentive policies (<italic>e.g.</italic>, employment qualifications and promotion criteria). The universities also need to show commitment to knowledge transfer. This means that apart from creating the necessary motivation in researchers, support mechanisms should also be provided.</p>", "<p>As previously mentioned, the main feature of Iran's medical research is that research and service delivery are under a common stewardship, which is an aftermath of integration of medical universities into the ministry of health. Therefore, it will be interesting to study the impact of integration on knowledge transfer in the future.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>In the past two decades, scientific publications in Iran have considerably increased their medical science content, and the number of articles published in ISI journals has doubled between 1997 and 2001. The aim of the present study was to determine how frequently knowledge transfer strategies were applied in Tehran University of Medical Sciences (TUMS). We were also interested in studying the determining factors leading to the type of strategy selected.</p>", "<title>Methodology</title>", "<p>All TUMS research projects that had received grants from inside and outside the university in 2004, and were completed by the end of 2006, were included in the study. In total, 301 projects were examined, and data on each of the projects were collected by the research team using a standardized questionnaire. The projects' principle investigators filled out a second questionnaire. In all, 208 questionnaires were collected.</p>", "<title>Results</title>", "<p>Researchers stated being more engaged in the passive strategies of knowledge transfer, especially those publishing in peer-reviewed journals. The mean score for the researchers' performance in passive and active strategies were 22% and 9% of the total score, respectively. Linear regression analysis showed that the passive strategy score decreased with the increase in the number of years working as a professional (p = 0.01) and personal interest as the only reason for choosing the research topic (p = 0.01). Regarding the active strategies of knowledge transfer, health system research studies significantly raised the score (p = 0.02) and 'executive responsibility' significantly lowered it (p = 0.03).</p>", "<title>Conclusion</title>", "<p>As a study carried out in a Middle Eastern developing country, we see that, like many other universities in the world, many academicians still do not give priority to active strategies of knowledge transfer. Therefore, if 'linking knowledge to action' is necessary, it may also be necessary to introduce considerable changes in academic procedures and encouragement policies (<italic>e.g.</italic>, employment and promotion criteria of academic members).</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>SN and RM participated in the design, statistical analysis, and manuscript writing. JG designed and conducted the study. SN, MG, MS, and MA gathered the data. KM assisted in interpreting the statistical analysis and manuscript writing. All authors approved the final manuscript.</p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>This study had been sponsored by the deputy of research in TUMS through contract no. 85-03-74-4418. The authors appreciate Mr. Ramavandi's efforts in collecting the questionnaires of the study. We would like to thank Mandi Newton and Jacqueline Tetroe, the referees, for their valuable comments in reviewing the manuscript.</p>" ]

[]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>'Passive' knowledge transfer strategies of TUMS researchers, based on the type of research.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Strategy</td><td align=\"center\" colspan=\"2\">Basic</td><td align=\"center\" colspan=\"2\">Clinical</td><td align=\"center\" colspan=\"2\">Health system</td><td align=\"center\" colspan=\"2\">Total</td></tr><tr><td/><td colspan=\"8\"><hr/></td></tr><tr><td/><td align=\"center\">Number<break/>n = 46</td><td align=\"center\">Percent<break/>22.1</td><td align=\"center\">Number<break/>n = 101</td><td align=\"center\">Percent<break/>48.6</td><td align=\"center\">Number<break/>n = 61</td><td align=\"center\">Percent<break/>29.3</td><td align=\"center\">Number<break/>n = 208</td><td align=\"center\">Percent<break/>100</td></tr></thead><tbody><tr><td align=\"left\">Publicaation of articles in domestic journals</td><td align=\"center\">20</td><td align=\"center\">43.5</td><td align=\"center\">75</td><td align=\"center\">74.3</td><td align=\"center\">35</td><td align=\"center\">57.4</td><td align=\"center\">130</td><td align=\"center\">62.5</td></tr><tr><td align=\"left\">Publication of articles in international journals</td><td align=\"center\">33</td><td align=\"center\">71.7</td><td align=\"center\">55</td><td align=\"center\">54.5</td><td align=\"center\">13</td><td align=\"center\">21.3</td><td align=\"center\">101</td><td align=\"center\">48.6</td></tr><tr><td align=\"left\">Presenting research results in conferences, seminars, and domestic meetings</td><td align=\"center\">20</td><td align=\"center\">43.5</td><td align=\"center\">55</td><td align=\"center\">41.0</td><td align=\"center\">25</td><td align=\"center\">41.0</td><td align=\"center\">100</td><td align=\"center\">48.1</td></tr><tr><td align=\"left\">Presenting research results in conferences, seminars, and international meetings</td><td align=\"center\">22</td><td align=\"center\">47.8</td><td align=\"center\">39</td><td align=\"center\">38.6</td><td align=\"center\">10</td><td align=\"center\">16.4</td><td align=\"center\">71</td><td align=\"center\">34.1</td></tr><tr><td align=\"left\">Sending the complete report of the research project to users</td><td align=\"center\">21</td><td align=\"center\">45.7</td><td align=\"center\">40</td><td align=\"center\">39.6</td><td align=\"center\">32</td><td align=\"center\">52.5</td><td align=\"center\">93</td><td align=\"center\">44.7</td></tr><tr><td align=\"left\">Sending a summary report of the project to users</td><td align=\"center\">19</td><td align=\"center\">41.3</td><td align=\"center\">45</td><td align=\"center\">44.6</td><td align=\"center\">29</td><td align=\"center\">47.5</td><td align=\"center\">93</td><td align=\"center\">44.7</td></tr><tr><td align=\"left\">Displaying the results on the web site</td><td align=\"center\">13</td><td align=\"center\">28.3</td><td align=\"center\">11</td><td align=\"center\">10.9</td><td align=\"center\">15</td><td align=\"center\">24.6</td><td align=\"center\">39</td><td align=\"center\">18.8</td></tr><tr><td align=\"left\">Mailing or emailing articles, reports, or summaries for stakeholders without their request</td><td align=\"center\">4</td><td align=\"center\">8.7</td><td align=\"center\">4</td><td align=\"center\">4.0</td><td align=\"center\">7</td><td align=\"center\">11.5</td><td align=\"center\">15</td><td align=\"center\">7.2</td></tr><tr><td align=\"left\">Publishing research results in newspapers (in which the general public is interested)</td><td align=\"center\">1</td><td align=\"center\">2.2</td><td align=\"center\">4</td><td align=\"center\">4.0</td><td align=\"center\">3</td><td align=\"center\">4.9</td><td align=\"center\">8</td><td align=\"center\">3.8</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>'Active' knowledge transfer strategies of TUMS researchers, based on the type of research.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Strategy</td><td align=\"center\" colspan=\"2\">Basic</td><td align=\"center\" colspan=\"2\">Clinical</td><td align=\"center\" colspan=\"2\">Health</td><td align=\"center\" colspan=\"2\">Total</td></tr><tr><td/><td colspan=\"8\"><hr/></td></tr><tr><td/><td align=\"center\">Number<break/>n = 46</td><td align=\"center\">Percent<break/>22.1</td><td align=\"center\">Number<break/>n = 101</td><td align=\"center\">Percent<break/>48.6</td><td align=\"center\">Number<break/>n = 61</td><td align=\"center\">Percent<break/>29.3</td><td align=\"center\">Number<break/>n = 208</td><td align=\"center\">Percent<break/>100</td></tr></thead><tbody><tr><td align=\"left\">Preparation and delivery of texts suitable to the users (such as plain writings for patients, special texts for managers, practical reports for clinical and lab colleagues, special reports for industrial managers or academics)</td><td align=\"center\">7</td><td align=\"center\">15.2</td><td align=\"center\">11</td><td align=\"center\">10.9</td><td align=\"center\">14</td><td align=\"center\">23.0</td><td align=\"center\">32</td><td align=\"center\">15.4</td></tr><tr><td align=\"left\">Presenting results to reporters, radio and TV for dissemination in the media and participation in interviews</td><td align=\"center\">2</td><td align=\"center\">4.3</td><td align=\"center\">8</td><td align=\"center\">7.9</td><td align=\"center\">6</td><td align=\"center\">9.8</td><td align=\"center\">16</td><td align=\"center\">7.7</td></tr><tr><td align=\"left\">Holding briefings with stakeholders for presentation of research results</td><td align=\"center\">2</td><td align=\"center\">4.3</td><td align=\"center\">6</td><td align=\"center\">5.9</td><td align=\"center\">13</td><td align=\"center\">21.3</td><td align=\"center\">21</td><td align=\"center\">10.1</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>The relation of independent variables on the score obtained on 'passive' strategies of knowledge transfer in the linear regression analysis.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\">Regression coefficient*</td><td align=\"center\">Standard error</td><td align=\"center\">P-value</td></tr></thead><tbody><tr><td align=\"left\">Sex (male/female)</td><td align=\"center\">0.00</td><td align=\"center\">0.46</td><td align=\"center\">0.99</td></tr><tr><td align=\"left\">Associate professor (in comparison to an assistant professor)</td><td align=\"center\">-0.28</td><td align=\"center\">0.57</td><td align=\"center\">0.62</td></tr><tr><td align=\"left\">Professor (in comparison to an assistant professor)</td><td align=\"center\">0.71</td><td align=\"center\">0.68</td><td align=\"center\">0.30</td></tr><tr><td align=\"left\">Instructor (in comparison to an assistant professor)</td><td align=\"center\">-1.09</td><td align=\"center\">0.91</td><td align=\"center\">0.23</td></tr><tr><td align=\"left\">Non-academic member (in comparison to an assistant professor)</td><td align=\"center\">0.61</td><td align=\"center\">0.91</td><td align=\"center\">0.50</td></tr><tr><td align=\"left\">Tenure status (full time/half time)</td><td align=\"center\">-1.02</td><td align=\"center\">1.18</td><td align=\"center\">0.39</td></tr><tr><td align=\"left\">Number of years working as a professional</td><td align=\"center\">-0.08</td><td align=\"center\">0.03</td><td align=\"center\">0.01</td></tr><tr><td align=\"left\">Executive responsibility (has/hasn't)</td><td align=\"center\">-0.65</td><td align=\"center\">0.47</td><td align=\"center\">0.17</td></tr><tr><td align=\"left\">Time allocated to research (percentage of total time)</td><td align=\"center\">0.01</td><td align=\"center\">0.01</td><td align=\"center\">0.39</td></tr><tr><td align=\"left\">Reasons for choosing the research topic (choice based on other organizations' request or need assessment vs. personal interest or repeating others research)</td><td align=\"center\">1.68</td><td align=\"center\">0.63</td><td align=\"center\">0.01</td></tr><tr><td align=\"left\">Clinical researches (in comparison to basic science researches)</td><td align=\"center\">-0.74</td><td align=\"center\">0.65</td><td align=\"center\">0.39</td></tr><tr><td align=\"left\">Health researches (in comparison to basic science researches)</td><td align=\"center\">-1.55</td><td align=\"center\">0.68</td><td align=\"center\">0.02</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>The relation of independent variables on the score obtained on 'active' strategies of knowledge transfer in the linear regression analysis.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\">Regression coefficient*</td><td align=\"center\">Standard error</td><td align=\"center\">P-value</td></tr></thead><tbody><tr><td align=\"left\">Sex (male/female)</td><td align=\"center\">-0.09</td><td align=\"center\">0.16</td><td align=\"center\">0.59</td></tr><tr><td align=\"left\">Associate professor (in comparison to an assistant professor)</td><td align=\"center\">0.09</td><td align=\"center\">0.20</td><td align=\"center\">0.67</td></tr><tr><td align=\"left\">Professor (in comparison to an assistant professor)</td><td align=\"center\">0.31</td><td align=\"center\">0.24</td><td align=\"center\">0.18</td></tr><tr><td align=\"left\">Instructor (in comparison to an assistant professor)</td><td align=\"center\">0.13</td><td align=\"center\">0.31</td><td align=\"center\">0.68</td></tr><tr><td align=\"left\">Non-academic member (in comparison to an assistant professor)</td><td align=\"center\">0.12</td><td align=\"center\">0.31</td><td align=\"center\">0.70</td></tr><tr><td align=\"left\">Tenure status (full time/half time)</td><td align=\"center\">-0.18</td><td align=\"center\">0.41</td><td align=\"center\">0.66</td></tr><tr><td align=\"left\">Number of years working as a professional</td><td align=\"center\">-0.02</td><td align=\"center\">0.01</td><td align=\"center\">0.08</td></tr><tr><td align=\"left\">Executive responsibility (has/hasn't)</td><td align=\"center\">-0.36</td><td align=\"center\">0.16</td><td align=\"center\">0.03</td></tr><tr><td align=\"left\">Time allocated to research (percentage of total time)</td><td align=\"center\">0.01</td><td align=\"center\">0.01</td><td align=\"center\">0.33</td></tr><tr><td align=\"left\">Reasons for choosing the research topic (choice based on other organizations' request or need assessment vs. personal interest or repeating others research)</td><td align=\"center\">0.19</td><td align=\"center\">0.22</td><td align=\"center\">0.39</td></tr><tr><td align=\"left\">Clinical researches (in comparison to basic science researches)</td><td align=\"center\">-0.04</td><td align=\"center\">0.22</td><td align=\"center\">0.87</td></tr><tr><td align=\"left\">Health researches (in comparison to basic science researches)</td><td align=\"center\">0.51</td><td align=\"center\">0.23</td><td align=\"center\">0.02</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional file 1</title><p>The Research Questionnaire (checklist)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S2\"><caption><title>Additional file 2</title><p>Researcher's Questionnaire</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><p>*These coefficients represent the change in the total score of passive strategies, where the maximum score attainable is 18.</p></table-wrap-foot>", "<table-wrap-foot><p>*These coefficients represent the change in the total score of active strategies, where the maximum score attainable is 6.</p></table-wrap-foot>" ]

[]

[ "<media xlink:href=\"1748-5908-3-39-S1.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1748-5908-3-39-S2.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Until recently, research on correlates of physical activity was dominated by studies of individual demographic and psychosocial characteristics [##REF##12471307##1##]. This reflected an emphasis on promoting sport, recreation or health-directed exercise using techniques to encourage individual behaviour change [##REF##10373178##2##]. However, there is little evidence that such approaches are effective in increasing physical activity in the medium-to-long term [##REF##15674903##3##]. If habitual patterns of behaviour are environmentally cued, sustained change is likely to require a supportive environment in which people can be active [##REF##11064848##4##,##REF##15212778##5##]. There is therefore increasing interest in the influence of the social and physical environment on physical activity.</p>", "<p>With respect to the physical (natural or built) environment, a growing body of evidence suggests that certain environmental characteristics may be associated with patterns of physical activity in general or with particular types of physical activity such as walking or cycling as modes of transport [##REF##11064848##4##, ####REF##15212778##5##, ##REF##12704009##6##, ##UREF##0##7##, ##UREF##1##8##, ##UREF##2##9##, ##REF##16138933##10####16138933##10##]. Among the correlates most frequently identified in such reviews – some ascertained using 'objective' measures, and others in terms of people's perceptions – are the aesthetic quality of the surroundings, the presence of pavements (sidewalks), the convenience of facilities for being active, the availability of green space, access to amenities (destinations) within walking or cycling distance, safety from traffic and personal attack, and the lack of heavy traffic. Some of these local characteristics reflect higher-order aspects of urban design and spatial policy such as population density, connectivity and mixed land use [##REF##12704009##6##,##UREF##1##8##]. Importantly, different characteristics may be associated with different types of physical activity; for example, Owen and colleagues found that the aesthetic quality of the surroundings was associated with walking for exercise or recreation and with walking in general, but not with walking for transport, whereas perceptions of traffic were associated with walking for transport and walking in general, but not with walking for exercise or recreation [##REF##15212778##5##].</p>", "<p>Despite the growing volume of published studies in this field, many authors remain circumspect in their interpretation of the available evidence. Giles-Corti and Donovan have described access to a supportive physical environment as a necessary, but insufficient, condition for an increase in physical activity in the population [##REF##12113436##11##], while Handy found 'convincing' evidence of an association between physical activity and the built environment in general but 'less convincing' evidence as to which specific environmental characteristics were most strongly associated [##UREF##0##7##]. One limitation of the available evidence is that most research has been conducted in North America and Australia [##UREF##2##9##,##REF##17152319##12##], and it is not clear whether associations observed in those countries are generalisable to other settings with different aggregate socioeconomic characteristics (e.g. wealth or access to private cars) or environmental characteristics (e.g. climate, patterns of land use, or availability of public transport). For example, North American researchers are often interested in the presence or absence of pavements (sidewalks), but it is unusual for streets in the United Kingdom (UK) not to have a pavement or footpath beside them. Hypotheses about putative environmental correlates of physical activity therefore need to be tested in a wider range of settings.</p>", "<p>A more profound limitation of the available evidence is that identifying a relationship between, for example, urban form and walking for transport is not the same thing as showing that <italic>changing </italic>the built environment will lead to a change in behaviour [##UREF##3##13##]. Few researchers have taken up the opportunity (or challenge) presented by 'natural experiments' to investigate the effects of environmental interventions on physical activity [##REF##15965130##14##]. We therefore established a longitudinal study to examine changes associated with the opening of a new urban section of the M74 motorway (freeway) currently under construction in Glasgow, Scotland. The rationale and design for this study have been described previously [##REF##16829328##15##]. It is claimed that the new motorway, which will mostly pass through or close to densely-populated urban neighbourhoods, will contribute to the regeneration of a region which includes some of the most deprived and least healthy working-class communities in Europe [##UREF##4##16##]. It is also claimed that the new motorway will divert traffic from local streets, reduce traffic noise and bring new local employment opportunities, thereby improving characteristics of the local environment held to be associated with active travel. Others claim that the new motorway will encourage car use, degrade the aesthetic quality of the surroundings and reduce the safety and attractiveness of routes for pedestrians and cyclists across the line of the motorway – all changes which may be expected to discourage active travel [##REF##16829328##15##]. The eventual aim of the M74 study will be to assess the effects of this major modification to the urban built environment and transport infrastructure on perceptions of the local environment and on population health and health-related behaviour, the primary outcome of interest being a change in the quantity of 'active travel' (walking and cycling for transport).</p>", "<p>In this paper, we report findings from the cross-sectional (baseline) phase of the study which contribute evidence on the environmental correlates of physical activity in this comparatively deprived urban population. We focus on two specific hypotheses: first, that levels of active travel and overall physical activity vary with demographic and socioeconomic characteristics, but not necessarily in the same way; second, that these relationships may be partly explained by the perceived characteristics of the local environment in which people live and by their objectively-assessed proximity to motorway and major road infrastructure.</p>" ]

[ "<title>Methods</title>", "<title>Delineation of study areas</title>", "<p>We used spatially referenced census and transport infrastructure data held and analysed in a geographical information system (GIS), combined with field visits, to delineate three study areas in Glasgow with similar aggregate socioeconomic characteristics and broadly similar topographical characteristics apart from their proximity to urban motorway infrastructure (Table ##TAB##0##1##, Figure ##FIG##0##1##). All three study areas extended from inner mixed-use districts close to the city centre to residential suburbs, contained major arterial roads other than motorways, and contained a mixture of housing stock including traditional high-density tenements, high-rise flats and new housing developments (Figure ##FIG##1##2##).</p>", "<title>Sampling and survey administration</title>", "<p>We used the Royal Mail Postcode Address File (PAF) (version 2005.3) to identify all residential addresses whose unit postcode (zip code) was within one of the study areas (total n = 35601) and drew a random sample of 3000 households from each area. Unit postcodes (e.g. G12 8RZ) are the smallest available unit of postal geography in the UK; residential unit postcodes cover about 15 addresses on average. We sent the survey to all households (total n = 9000) between 28 September and 4 October 2005 and resent the survey to all non-responding households between 26 and 31 October 2005. We alerted households to the survey by means of a postcard sent a few days in advance, used coloured paper for some of the survey materials, and posted survey packs in white envelopes printed with the university crest; these techniques have been shown in a meta-analysis to be associated with increased response rates to postal surveys [##REF##12016181##17##]. We asked householders to ensure that the questionnaire was completed by a resident aged 16 or over; if more than one resident was eligible, we asked householders to select the person with the most recent birthday. Respondents who consented to follow-up were entered into a prize draw to win a £50 (€63; US$92) gift voucher. Responses received more than three months after the first mailing wave were disregarded in analysis.</p>", "<title>Data collection</title>", "<p>The questionnaire included items on demographic and socioeconomic characteristics, health and wellbeing (including the the SF-8 scale), perceptions of the local environment, travel behaviour and the short form of the International Physical Activity Questionnaire (IPAQ) (Additional file ##SUPPL##0##1##). We developed a new 'neighbourhood scale' to assess perceptions of relevant characteristics of the local environment (aesthetics, green space, access to amenities, convenience of routes, traffic, road safety and personal safety). The development, principal components analysis and reliability of the items in this scale and the derivation and reliability of summary variables are reported in an accompanying paper [##REF##18513430##18##].</p>", "<title>Data cleaning and derivation of variables</title>", "<title>Demographic and socioeconomic characteristics</title>", "<p>We excluded from analysis all respondents who failed to enter their age or sex. We then examined the distributions of all raw variables and carried out range and consistency checks to identify any anomalous values or variables with a high proportion of missing responses. As a consequence, we collapsed responses on distance to place of work or study, housing tenure, car access and working situation into fewer categories by merging categories with small numbers of responses; we also disregarded household composition and working situation of spouse or partner in analysis because of the large numbers of missing values for these variables.</p>", "<title>Health and wellbeing</title>", "<p>We calculated body mass index (BMI) by converting, where necessary, self-reported heights and weights from imperial to metric units and dividing the height in metres by the square of the weight in kilograms; we also categorised respondents into quintiles of BMI. We calculated physical (PCS-8) and mental (MCS-8) health summary scores from the SF-8 data and scaled these to population norms using the method and coefficients given in the SF-8 manual [##UREF##5##19##].</p>", "<title>Objective environmental characteristics</title>", "<p>We linked each record to the unit postcode of residence. We then constructed concentric buffers at 100-metre intervals up to 500 metres around the routes and access points of existing and planned motorways and around the network of other major (A- and B- class) roads, and assigned each respondent to a category of proximity to each type of road infrastructure (within 100 metres, 101–200 metres, etc.) based on the location of the centroid of their unit postcode.</p>", "<title>Travel behaviour</title>", "<p>For travel time analysis we included travel diaries which recorded no travel at all, but we disregarded travel data from respondents who had not been at home on the day of the travel diary, whose questionnaire had been misprinted such that the travel diary pages were unusable, who had recorded journeys without reporting valid quantitative data on the durations of those journeys, or whose completed travel diary appeared implausible. We also disregarded journeys whose purpose was not stated or was beyond the scope of the travel diary (Additional file ##SUPPL##0##1##, page 8). We summed the reported travel time for each mode of transport, calculated a total travel time by active modes (walking plus cycling) and by all modes combined, and calculated the proportion of total travel time contributed by each mode of transport.</p>", "<title>Physical activity</title>", "<p>We cleaned and analysed IPAQ data in accordance with the IPAQ scoring protocol <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.ipaq.ki.se\"/>. We therefore disregarded physical activity data from respondents who had reported more than 16 hours of physical activity per day or who had missing or internally inconsistent data on the frequency or duration of any of the three categories of physical activity (walking, moderate-intensity activity or vigorous activity). We also recoded reported durations of activity of less than ten minutes to zero, and of greater than 180 minutes to 180 minutes. We calculated the estimated total physical activity energy expenditure for each respondent (MET-min/week) and used a combination of frequency, duration and total energy expenditure to assign each respondent to a 'high', 'moderate' or 'low' category of overall physical activity in accordance with the prescribed IPAQ algorithm. The 'high' category corresponds to a sufficient level of physical activity to meet current public health recommendations for adults [##UREF##6##20##].</p>", "<title>Analysis</title>", "<p>We considered it unlikely that the statistical assumptions required for linear regression could be met because the distributions of time spent walking and cycling and of estimated total physical activity energy expenditure were both strongly positively skewed and dominated by a large number of zero values which meant that the data were not amenable to log-transformation. We therefore modelled the correlates of active travel and physical activity using multivariate logistic regression. We defined 'active travel' as a binary condition achieved by any respondent who had reported at least 30 minutes of travel by walking, cycling or both in their travel diary, reflecting the current recommendation that adults should accumulate at least 30 minutes of moderate-intensity physical activity on most days of the week [##UREF##6##20##], and we defined 'physical activity' as a binary condition achieved by any respondent whose overall physical activity was categorised as 'high' using IPAQ. We then built separate multivariate models for active travel and physical activity following the method of Hosmer and Lemeshow [##UREF##7##21##], first including only 'personal' (individual or household) variables and then adding 'environmental' variables (Additional file ##SUPPL##1##2##).</p>" ]

[ "<title>Results</title>", "<title>Response</title>", "<p>We received 1345 completed questionnaires. After subtracting from the numerator 23 completed questionnaires with missing critical demographic data (age or sex), and after subtracting from the denominator 676 addresses from which survey packs were returned as undeliverable, this left 1322 valid responses to be entered into analysis – a response rate of 1322/(9000-676) = 15.9%.</p>", "<title>Characteristics of study participants</title>", "<title>Demographic and socioeconomic characteristics</title>", "<p>Respondents were aged between 16 and 89 years (median age 48 years). 804 (61%) were women. Only 136 (26%) of the men and 145 (18%) of the women reported having access to a bicycle. For those who usually travelled to a place of work or study, the median reported distance was 3.5 miles (about 5.5 kilometres). Other characteristics of study participants are summarised in Table ##TAB##1##2##.</p>", "<title>Health and wellbeing</title>", "<p>25% of respondents reported difficulty walking for a quarter of a mile, 39% reported a long-term health problem or disability, and 50% were overweight (median BMI 25.1 kg/m²). The median mental health summary score (MCS-8) was significantly lower (i.e. poorer) than the population norm (median 47.3, 95% CI 46.4 to 48.1); the median physical health summary score (PCS-8) was not significantly different from the population norm (median 50.9, 95% CI 49.6 to 51.7).</p>", "<title>Descriptive data on travel behaviour and physical activity</title>", "<title>Travel behaviour</title>", "<p>1099 travel diaries were suitable for travel time analysis. Men and women were equally likely to have returned usable travel time data, but respondents who were older, retired, or living in social rented accommodation or who did not have access to a car were less likely to have returned usable data. On average, respondents recorded about an hour's travel per day (mean 61.5 minutes, median 50.0 minutes), of which a minority was spent using active modes of transport (walking or cycling: mean 20.0 minutes, median 10.0 minutes) (Table ##TAB##2##3##). 304 respondents (28%) recorded at least 30 minutes of active travel, of whom 294 (97%) recorded at least 30 minutes of walking.</p>", "<title>Physical activity</title>", "<p>833 respondents returned complete physical activity data suitable for analysis. Women and respondents who were older, retired, or living in social rented accommodation or who did not have access to a car were less likely to have returned usable data. Respondents reported a mean of 318 minutes' walking per week and a mean estimated total physical activity energy expenditure of 3000 MET-minutes per week (Table ##TAB##3##4##). Only 316 respondents (38%) were categorised as having achieved a 'high' (i.e. sufficient) level of physical activity.</p>", "<title>Correlates of active travel</title>", "<p>Active travel was significantly associated with being younger, living in owner-occupied accommodation, not having to travel more than four miles to work, having access to a bicycle, not having access to a car, and the absence of any difficulty walking. The final best model of the 'personal' correlates of active travel provided satisfactory goodness-of-fit (Hosmer and Lemeshow test: χ²= 13.04, df = 8; P = 0.11) and explained nearly one-fifth of the total variance in active travel (Nagelkerke's R²= 18.7%) (Table ##TAB##4##5##). Adding 'environmental' variables to the model showed an additional significant positive association between active travel and perceived proximity to shops, and an additional significant negative association between active travel and perceived road safety for cyclists. The final best model of the personal and environmental correlates of active travel also provided satisfactory goodness-of-fit (Hosmer and Lemeshow test: χ²= 10.61, df = 8; P = 0.23) and explained slightly more of the total variance in active travel than did the personal model alone (Nagelkerke's R²= 20.1%) (Figure ##FIG##2##3##).</p>", "<p>In order to aid interpretation, we also partitioned the dataset into two strata ('No car available' and 'Car available') and refitted the final model separately to each stratum of the dataset (Table ##TAB##5##6##). This showed that the subset of respondents with no access to a car accounted for the significant overall relationship between active travel and access to a bicycle, whereas those with access to a car accounted for the significant overall relationships with distance to place of work or study and perceptions of the local environment. The relationship with difficulty walking was also stronger in this group than in those without access to a car.</p>", "<title>Correlates of physical activity</title>", "<p>Physical activity was significantly associated with living in social-rented accommodation, not being overweight, and the absence of any difficulty walking. The final best model of the 'personal' correlates of physical activity provided satisfactory goodness-of-fit (Hosmer and Lemeshow test: χ²= 3.89, df = 7; P = 0.89) and explained about one-sixth of the total variance in physical activity (Nagelkerke's R²= 15.9%) (Table ##TAB##6##7##). Adding 'environmental' variables to the model showed an additional significant negative association between physical activity and perception of traffic volume (i.e. respondents who perceived there to be a higher volume of traffic were more likely to report physical activity). The final best model of the personal and environmental correlates of physical activity also provided satisfactory goodness-of-fit (Hosmer and Lemeshow test: χ²= 3.86, df = 8; P = 0.87) and explained slightly more of the total variance in physical activity than did the personal model alone (Nagelkerke's 16.6%) (Figure ##FIG##2##3##).</p>" ]

[ "<title>Discussion</title>", "<title>Principal findings</title>", "<p>In this deprived urban population, the likelihood of reporting active travel was associated with being younger, living in owner-occupied accommodation, not having to travel a long distance to work and not having access to a car, whereas overall physical activity was associated with living in social-rented accommodation and not being overweight. After adjusting for individual and household characteristics, neither perceptions of the local environment nor the objective proximity of respondents' homes to motorway or major road infrastructure appeared to explain much of the variance in active travel or overall physical activity, although we did find a significant positive association between active travel and perceived proximity to shops.</p>", "<title>Representativeness and completeness of survey data</title>", "<p>Our difficulty in obtaining a representative sample of the resident population is not unique to our study. Although our final response rate was low, it was almost identical to that achieved in a recent population-based intervention study elsewhere in Glasgow [##REF##16286490##22##]. Some of the challenges of recruiting research participants in areas of deprivation have been described elsewhere [##UREF##8##23##]; these are superimposed on a downward trend in participation in even the best-resourced national population surveys [##UREF##9##24##] and an upward (and socially biased) trend in opt-outs from the main alternative sampling frame, the edited electoral register [##UREF##10##25##]. Although our achieved sample contained a higher proportion of respondents from owner-occupied and car-owning households than predicted from 2001 census data for the same census output areas, these differences may be partly accounted for by an upward background trend in owner occupation and car access between 2001 and 2005. Our achieved sample is still clearly disadvantaged overall, in terms of socioeconomic and health status, compared with the country as a whole. It also contains sufficient heterogeneity to enable us to examine, in time, how the effects of the intervention are distributed between socioeconomic groups. We therefore consider our achieved sample fit for purpose.</p>", "<p>We had to disregard a substantial proportion of cases in analysis because respondents had returned unusable travel time data or had returned physical activity data that were incomplete, internally inconsistent or included a 'Don't know' response and were therefore unacceptable according to the IPAQ scoring protocol. Most published studies using the same, short form of IPAQ have either not reported the distribution of the continuous summary measures or have not reported data for the UK separately from those for other countries where higher levels of physical activity are reported. Despite the high proportion of missing physical activity data in our dataset, however, the aggregate continuous data we obtained were broadly comparable to those reported in Rütten and colleagues' study of a random sample of UK adults [##REF##12795825##26##]. We could have included more cases in physical activity analysis by, for example, imputing missing values, but the results would not have been comparable with others' owing to the substantial deviations from the scoring protocol which would have been required. The frequency of unusable responses was not reported in the international multi-centre study which originally established the validity and reliability of IPAQ [##REF##12900694##27##]. It is possible that offering a 'Don't know' option in the self-completed IPAQ questionnaire encourages respondents to select this rather than to enter what may be a reasonably precise estimate of the actual time spent in physical activity; the respondent has no way of knowing that a single 'Don't know' response will result in all of their physical activity data being disregarded in analysis. This should be considered in any future revision of the IPAQ questionnaire and scoring protocol.</p>", "<title>Contribution of active travel to overall physical activity</title>", "<p>The explanatory variables that were significantly associated with active travel but not with physical activity (distance to place of work or study, access to a bicycle, access to a car, perceived proximity to shops, and perceived road safety for cyclists) all have an obvious intuitive relationship with the use of walking or cycling as modes of transport. That they were not significantly associated with overall physical activity suggests either that active travel contributes only a minority of respondents' overall physical activity or that other factors not measured in this study are more important correlates of overall physical activity than those which determine active travel. A crude comparision of the quantity of active travel reported in the one-day travel diaries with the quantities of physical activity reported using IPAQ suggests that on average, active travel may indeed make only a small (~15%) contribution to overall physical activity in this study population. However, the real contribution may be substantially greater than this if, as has been shown previously, respondents tend to over-report their physical activity using IPAQ [##REF##12740079##28##]. There can be little doubt that active travel makes a substantial contribution to the total quantity of <italic>walking </italic>reported in this study population. Irrespective of the true contribution of active travel to overall physical activity, however, it remains likely that other unmeasured personal and social factors beyond the scope of this study may be more important correlates of overall physical activity.</p>", "<title>Socio-spatial patterning of active travel and overall physical activity</title>", "<p>Respondents living in owner-occupied households were more likely to report active travel than those living in social-rented accommodation, but less likely to report sufficient overall physical activity. Since neither working situation nor perceived financial situation emerged as significantly associated with active travel or overall physical activity, housing tenure and car access are the remaining explanatory variables in this dataset which can be interpreted as markers of socioeconomic status. Although having access to a car clearly reflects the possession of a material asset, it has been argued that this is a less direct marker of socioeconomic status than some other markers because, in Scotland at least, access to a car is a more-or-less essential requirement for living in many rural areas, whereas it is possible to live in a dense urban settlement such as Glasgow without using a car. In the final models in this study, therefore, housing tenure may be regarded as the primary marker of socioeconomic status. The findings consequently suggest conflicting socioeconomic gradients in prevalence: more advantaged respondents were more likely to report active travel, but more disadvantaged respondents were more likely to report sufficient overall physical activity. The higher prevalence of sufficient overall physical activity among the more disadvantaged despite their lower propensity for active travel is likely to reflect higher quantities of physical activity in other domains, particularly occupational and domestic activities, since leisure-time physical activity tends to be higher among more advantaged groups [##UREF##11##29##].</p>", "<title>Environmental characteristics: paradoxical, unmeasured, or irrelevant?</title>", "<p>The two environmental variables that emerged as significantly associated with active travel, particularly among those without access to a car, were perceived proximity to shops and perceived road safety for cyclists. The positive association with perceived proximity to shops suggests that for active travel to be undertaken in this population, it may be more important that people live close to the amenities they need than that they live in an environment with more favourable subjective or discretionary considerations such as attractiveness or noise. This would be consistent with an understanding that walking as a mode of transport is primarily a way of undertaking journeys which have to be made anyway, as opposed to more discretionary (recreational) forms of walking which may be more susceptible to the influence of less-structural characteristics.</p>", "<p>Although the negative association with perceived road safety for cyclists appears counter-intuitive, similar 'paradoxical inverse relationships' have been reported elsewhere, for example by Titze and colleagues in a study of the correlates of cycling among students [##REF##17489008##30##] and by Humpel and colleagues in a study of correlates of walking for pleasure [##REF##14751322##31##]. Titze and colleagues suggest that respondents who cycle regularly are more likely to be aware of, and report, the danger posed by traffic than non-cyclists or infrequent cyclists. A similar phenomenon could explain the negative association between physical activity and perception of traffic volume.</p>", "<p>Overall, the influence of the putative environmental characteristics examined in this study on active travel and physical activity appeared small compared with that of the personal characteristics found to be significant, and including environmental characteristics in the models did not substantially modify the influence of personal characteristics.</p>", "<p>On the one hand, this could reflect an artefact of the research methods (a false negative error), which could have arisen in various ways. In particular, the 'wrong' environmental exposure may have been measured, in that the environmental characteristics examined were those of the immediate surroundings of respondents' homes, whereas the propensity to choose active modes of transport may be more strongly influenced by the characteristics of the environment elsewhere on their routes [##REF##17489008##30##], for example the perceived danger of cycling in the city centre – an association which may be absent, or at least diluted, when the 'exposure' examined is limited to the residential environment. It could also be argued that the apparently weak influence of environmental characteristics in this study reflects a reliance on respondents' perceptions which have not been objectively verified and may therefore be a weak proxy for the 'true' objectively-measured characteristics of their surroundings. However, as recent reviews have pointed out, the current weight of evidence for objective environmental correlates of walking is no greater than that for subjective environmental correlates [##REF##15212778##5##] and it is entirely plausible that people's perceptions of their environment may be at least as important as their objective conditions in influencing their behaviour [##REF##12704009##6##].</p>", "<p>On the other hand, we may have demonstrated a real absence of any major association. Although at first sight this appears at odds with the growing body of review-level evidence for environmental correlates of physical activity, Wendel-Vos and colleagues noted that of all the environmental factors examined in all the studies included in their review, analysis showed a 'null association' in 76% of cases [##UREF##2##9##], and our finding that personal factors account for a much larger proportion of the variance in active travel or physical activity than is accounted for by environmental factors is consistent with those of some other European studies [##REF##16277805##32##,##REF##17947248##33##]. In the particular context of this study, residents may simply have adapted to adverse conditions in their local environment in the ways identified by Hedges in a qualitative study of people living close to new roads built in the UK in the 1970s [##UREF##12##34##] – particularly by attitudinal adaptation, which Hedges characterises as developing an attitude that it is futile to resist. One can imagine that in the most deprived areas of Glasgow, people may have become resigned to the nature of their surroundings, seeing them as inevitable and not amenable to change either through environmental improvement or through their moving to another area.</p>" ]

[ "<title>Conclusion</title>", "<p>After demographic and socioeconomic characteristics were taken into account, neither perceptions of the local environment nor objective proximity to major road infrastructure appeared to explain much of the variance in active travel or overall physical activity in this study. Our study population may be both objectively constrained by their socioeconomic circumstances (including comparatively limited access to private cars) and adapted to living in conditions which others would consider to pose a barrier to active travel. Under these circumstances, environmental characteristics which have been found to influence discretionary active travel in studies in other, more affluent populations may simply be irrelevant in a population which is more captive in its travel choices. Environmental correlates of active travel should not be assumed to be generalisable between populations; researchers should continue to test hypotheses about putative environmental correlates in different settings, and policymakers should recognise that the effects of interventions to change the environment are likely to vary between populations and between socioeconomic groups within populations.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Environmental characteristics may be associated with patterns of physical activity in general or with particular types of physical activity such as active travel (walking or cycling for transport). However, most studies in this field have been conducted in North America and Australia, and hypotheses about putative correlates should be tested in a wider range of sociospatial contexts. We therefore examined the contribution of putative personal and environmental correlates of active travel and overall physical activity in deprived urban neighbourhoods in Glasgow, Scotland as part of the baseline for a longitudinal study of the effects of opening a new urban motorway (freeway).</p>", "<title>Methods</title>", "<p>We conducted a postal survey of a random sample of residents (n = 1322), collecting data on socioeconomic status, perceptions of the local environment, travel behaviour, physical activity and general health and wellbeing using a new 14-item neighbourhood rating scale, a travel diary, the short form of the International Physical Activity Questionnaire (IPAQ) and the SF-8. We analysed the correlates of active travel and overall physical activity using multivariate logistic regression, first building models using personal (individual and household) explanatory variables and then adding environmental variables.</p>", "<title>Results</title>", "<p>Active travel was associated with being younger, living in owner-occupied accommodation, not having to travel a long distance to work and not having access to a car, whereas overall physical activity was associated with living in social rented accommodation and not being overweight. After adjusting for personal characteristics, neither perceptions of the local environment nor the objective proximity of respondents' homes to motorway or major road infrastructure explained much of the variance in active travel or overall physical activity, although we did identify a significant positive association between active travel and perceived proximity to shops.</p>", "<title>Conclusion</title>", "<p>Apart from access to local amenities, environmental characteristics may have limited influence on active travel in deprived urban populations characterised by a low level of car ownership, in which people may have less capacity for making discretionary travel choices than the populations studied in most published research on the environmental correlates of physical activity.</p>" ]

[ "<title>Competing interests</title>", "<p>This paper is based on material contained in the first author's PhD thesis.</p>", "<title>Authors' contributions</title>", "<p>DO had the original idea for the study, designed the study and the survey materials, applied for ethical approval, cleaned and coded the survey data, carried out all the geographical and statistical analyses and wrote the paper. MP was DO's PhD supervisor. RM, NM, MP and SP constituted the steering group for the study, contributed to and advised on the design of the study and the interpretation of the emerging findings, and contributed to the critical revision of the paper. All authors read and approved the final manuscript.</p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>DO was funded by a Medical Research Council (MRC) special training fellowship in health of the public research (award reference no. G106/1203 67290). MP, RM and SP were funded by the Chief Scientist Office of the Scottish Executive Health Department (now of the Scottish Government Public Health and Wellbeing Directorate). The opinions expressed here are those of the authors and not necessarily those of the funders. The study was approved by the University of Glasgow Medical Faculty Ethics Committee (project reference no. FM01304). The SF-8 health survey was used under licence from QualityMetric Incorporated (licence agreement no. R1-061005-22740). We thank the survey team at the MRC Social and Public Health Sciences Unit for carrying out the survey and the local residents who took part.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Boundaries of local study areas defined in terms of census output areas</bold>. Data and raster image ^©Crown Copyright/database right 2005. An Ordnance Survey/EDINA supplied service.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Examples of scenes in and around the local study areas</bold>. All images ^©David Ogilvie.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p>Estimated proportions of variance in active travel and physical activity explained by personal and environmental characteristics.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Definitions of study areas</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Study area</bold></td><td align=\"left\"><bold>Definition</bold></td></tr></thead><tbody><tr><td align=\"left\">South</td><td align=\"left\">A set of census output areas (the smallest spatial unit for which aggregate census data are available) encroaching within 500 metres of the proposed route of the new M74 motorway</td></tr><tr><td align=\"left\">East</td><td align=\"left\">A set of census output areas encroaching within 500 metres of the routes of the existing M8 and M80 motorways</td></tr><tr><td align=\"left\">North</td><td align=\"left\">A set of census output areas not encroaching within 500 metres of the route of any existing or proposed motorway</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Socioeconomic characteristics of study participants</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Category</bold></td><td align=\"right\"><bold>Frequency (%)</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Working situation</bold></td><td/></tr><tr><td align=\"left\">Employed</td><td align=\"right\">616 (47.2)</td></tr><tr><td align=\"left\">Retired</td><td align=\"right\">333 (25.5)</td></tr><tr><td align=\"left\">Other*</td><td align=\"right\">357 (27.3)</td></tr><tr><td align=\"left\">Missing</td><td align=\"right\">16</td></tr><tr><td align=\"left\"><bold>Financial situation</bold></td><td/></tr><tr><td align=\"left\">Find it a strain to get by from week to week</td><td align=\"right\">233 (17.9)</td></tr><tr><td align=\"left\">Have to be careful about money</td><td align=\"right\">680 (52.2)</td></tr><tr><td align=\"left\">Able to manage without much difficulty</td><td align=\"right\">299 (23.0)</td></tr><tr><td align=\"left\">Quite comfortably off</td><td align=\"right\">90 (6.9)</td></tr><tr><td align=\"left\">Missing</td><td align=\"right\">20</td></tr><tr><td align=\"left\"><bold>Housing tenure</bold></td><td/></tr><tr><td align=\"left\">Owner-occupied</td><td align=\"right\">678 (51.6)</td></tr><tr><td align=\"left\">Social rented</td><td align=\"right\">543 (41.3)</td></tr><tr><td align=\"left\">Other^†</td><td align=\"right\">93 (7.1)</td></tr><tr><td align=\"left\">Missing</td><td align=\"right\">8</td></tr><tr><td align=\"left\"><bold>Cars or vans available to household</bold></td><td/></tr><tr><td align=\"left\">None</td><td align=\"right\">629 (48.4)</td></tr><tr><td align=\"left\">One</td><td align=\"right\">525 (40.4)</td></tr><tr><td align=\"left\">Two or more</td><td align=\"right\">146 (11.2)</td></tr><tr><td align=\"left\">Missing</td><td align=\"right\">22</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Daily travel time by mode recorded in travel diaries</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"3\"><bold>All respondents reporting valid travel time data</bold></td></tr><tr><td align=\"left\"><bold>Mode</bold></td><td colspan=\"3\"><hr/></td></tr><tr><td/><td align=\"center\"><bold>Mean (sd)</bold></td><td align=\"center\"><bold>Median (IQR) (range)</bold></td><td align=\"center\"><bold>Proportion of total</bold></td></tr></thead><tbody><tr><td align=\"left\">Car</td><td align=\"right\">24.4 (40.8)</td><td align=\"right\">0.0 (40.0) (0–510)</td><td align=\"right\">39.7%</td></tr><tr><td align=\"left\">Walking</td><td align=\"right\">19.2 (27.8)</td><td align=\"right\">10.0 (30.0) (0–205)</td><td align=\"right\">31.2%</td></tr><tr><td align=\"left\">Bus</td><td align=\"right\">14.6 (30.8)</td><td align=\"right\">0.0 (20.0) (0–210)</td><td align=\"right\">23.7%</td></tr><tr><td align=\"left\">Rail</td><td align=\"right\">1.8 (10.0)</td><td align=\"right\">0.0 (0.0) (0–165)</td><td align=\"right\">2.9%</td></tr><tr><td align=\"left\">Cycling</td><td align=\"right\">0.7 (7.3)</td><td align=\"right\">0.0 (0.0) (0–130)</td><td align=\"right\">1.1%</td></tr><tr><td align=\"left\">Motorcycle</td><td align=\"right\">0.1 (2.0)</td><td align=\"right\">0.0 (0.0) (0–50)</td><td align=\"right\">0.2%</td></tr><tr><td align=\"left\">Other</td><td align=\"right\">0.6 (9.4)</td><td align=\"right\">0.0 (0.0) (0–240)</td><td align=\"right\">1.0%</td></tr><tr><td align=\"left\">Active modes*</td><td align=\"right\">20.0 (28.5)</td><td align=\"right\">10.0 (30.0) (0–205)</td><td align=\"right\">32.4%</td></tr><tr><td align=\"left\">All modes combined</td><td align=\"right\">61.5 (53.2)</td><td align=\"right\">50.0 (63.0) (0–510)</td><td align=\"right\">100.0%</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Average time spent walking and total physical activity</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Summary measure</bold></td><td align=\"center\"><bold>Mean (standard deviation)</bold></td><td align=\"center\"><bold>Median (interquartile range) (range)</bold></td></tr></thead><tbody><tr><td align=\"left\">Walking (min/week)</td><td align=\"center\">318.4 (366.1)</td><td align=\"center\">180.0 (375.0) (0–1260)</td></tr><tr><td align=\"left\">Total activity (MET-min/week)</td><td align=\"center\">3000.1 (3323.1)</td><td align=\"center\">1935.0 (3645.0) (0–18438)</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T5\"><label>Table 5</label><caption><p>Multivariate logistic regression models of correlates of active travel</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"2\"><bold>Model including personal correlates</bold></td><td align=\"center\" colspan=\"2\"><bold>Model including personal and environmental correlates</bold></td></tr><tr><td align=\"left\"><bold>Variable</bold></td><td align=\"center\"><bold>OR (95% CI)</bold></td><td align=\"center\"><bold>P</bold></td><td align=\"center\"><bold>OR (95% CI)</bold></td><td align=\"center\"><bold>P</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Age</bold></td><td align=\"center\">0.98 (0.97, 0.99)</td><td align=\"center\">&lt;0.001</td><td align=\"center\">0.98 (0.97, 0.99)</td><td align=\"center\">0.001</td></tr><tr><td align=\"left\"><bold>Housing tenure (reference: social renter)</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Owner-occupier</td><td align=\"center\">1.79 (1.19, 2.69)</td><td align=\"center\">0.005</td><td align=\"center\">1.70 (1.13, 2.58)</td><td align=\"center\">0.012</td></tr><tr><td align=\"left\">Other</td><td align=\"center\">1.64 (0.83, 3.24)</td><td align=\"center\">0.159</td><td align=\"center\">1.62 (0.81, 3.23)</td><td align=\"center\">0.17</td></tr><tr><td align=\"left\"><bold>Distance to place of work or study (reference: four miles or more)</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Less than four miles</td><td align=\"center\">1.76 (1.16, 2.68)</td><td align=\"center\">0.008</td><td align=\"center\">1.81 (1.18, 2.76)</td><td align=\"center\">0.006</td></tr><tr><td align=\"left\">Not applicable*</td><td align=\"center\">2.12 (1.27, 3.54)</td><td align=\"center\">0.004</td><td align=\"center\">2.15 (1.28, 3.61)</td><td align=\"center\">0.004</td></tr><tr><td align=\"left\"><bold>Access to bicycle (reference: no)</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Yes</td><td align=\"center\">1.59 (1.07, 2.35)</td><td align=\"center\">0.021</td><td align=\"center\">1.57 (1.06, 2.33)</td><td align=\"center\">0.025</td></tr><tr><td align=\"left\"><bold>Composite variable (reference: access to car and difficulty walking)</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Car, no difficulty</td><td align=\"center\">4.21 (1.43, 12,43)</td><td align=\"center\">0.009</td><td align=\"center\">3.77 (1.27, 11.23)</td><td align=\"center\">0.017</td></tr><tr><td align=\"left\">No car, difficulty</td><td align=\"center\">4.65 (1.48, 14.54)</td><td align=\"center\">0.008</td><td align=\"center\">4.42 (1.40, 13.92)</td><td align=\"center\">0.011</td></tr><tr><td align=\"left\">No car, no difficulty</td><td align=\"center\">14.06 (4.84, 40.80)</td><td align=\"center\">&lt;0.001</td><td align=\"center\">12.88 (4.41, 37.67)</td><td align=\"center\">&lt;0.001</td></tr><tr><td align=\"left\"><bold>Individual items in neighbourhood scale</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Proximity to shops</td><td/><td/><td align=\"center\">1.20 (1.02, 1.41)</td><td align=\"center\">0.031</td></tr><tr><td align=\"left\">Road safety for cyclists</td><td/><td/><td align=\"center\">0.83 (0.70, 0.98)</td><td align=\"center\">0.024</td></tr><tr><td align=\"left\"><bold>Day of travel diary (reference: weekend)</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Weekday</td><td align=\"center\">1.96 (1.32, 3.00)</td><td align=\"center\">0.001</td><td align=\"center\">1.91 (1.26, 2.89)</td><td align=\"center\">0.002</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T6\"><label>Table 6</label><caption><p>Multivariate logistic regression model of personal and environmental correlates of active travel stratified by availability of a car</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"2\"><bold>No car available</bold></td><td align=\"center\" colspan=\"2\"><bold>Car available</bold></td></tr><tr><td align=\"left\"><bold>Variable</bold></td><td align=\"center\"><bold>OR (95% CI)</bold></td><td align=\"center\"><bold>P</bold></td><td align=\"center\"><bold>OR (95% CI)</bold></td><td align=\"center\"><bold>P</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Age</bold></td><td align=\"center\">0.98 (0.97, 1.00)</td><td align=\"center\">0.029</td><td align=\"center\">0.97 (0.95, 0.99)</td><td align=\"center\">0.008</td></tr><tr><td align=\"left\"><bold>Housing tenure (reference: social renter)</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Owner-occupier</td><td align=\"center\">1.57 (0.94, 2.65)</td><td align=\"center\">0.087</td><td align=\"center\">1.77 (0.86, 3.64)</td><td align=\"center\">0.12</td></tr><tr><td align=\"left\">Other</td><td align=\"center\">1.49 (0.61, 3.62)</td><td align=\"center\">0.38</td><td align=\"center\">1.64 (0.51, 5.30)</td><td align=\"center\">0.41</td></tr><tr><td align=\"left\"><bold>Distance to place of work or study (reference: four miles or more)</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Less than four miles</td><td align=\"center\">1.20 (0.57, 2.53)</td><td align=\"center\">0.63</td><td align=\"center\">1.96 (1.14, 3.37)</td><td align=\"center\">0.015</td></tr><tr><td align=\"left\">Not applicable*</td><td align=\"center\">1.00 (0.48, 2.11)</td><td align=\"center\">1.00</td><td align=\"center\">4.84 (2.20, 10.66)</td><td align=\"center\">&lt;0.001</td></tr><tr><td align=\"left\"><bold>Access to bicycle (reference: no)</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Yes</td><td align=\"center\">2.17 (1.10, 4.29)</td><td align=\"center\">0.026</td><td align=\"center\">1.43 (0.86, 2.38)</td><td align=\"center\">0.17</td></tr><tr><td align=\"left\"><bold>Difficulty walking (reference: yes)</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">No</td><td align=\"center\">2.49 (1.35, 4.57)</td><td align=\"center\">0.003</td><td align=\"center\">5.60 (1.74, 17.98)</td><td align=\"center\">0.004</td></tr><tr><td align=\"left\"><bold>Individual items in neighbourhood scale</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Proximity to shops</td><td align=\"center\">1.10 (0.88, 1.37)</td><td align=\"center\">0.39</td><td align=\"center\">1.34 (1.03, 1.74)</td><td align=\"center\">0.032</td></tr><tr><td align=\"left\">Road safety for cyclists</td><td align=\"center\">0.89 (0.71, 1.12)</td><td align=\"center\">0.31</td><td align=\"center\">0.77 (0.06, 0.99)</td><td align=\"center\">0.038</td></tr><tr><td align=\"left\"><bold>Day of travel diary (reference: weekend)</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Weekday</td><td align=\"center\">1.22 (0.71, 2.11)</td><td align=\"center\">0.47</td><td align=\"center\">3.32 (1.62, 6.82)</td><td align=\"center\">0.001</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T7\"><label>Table 7</label><caption><p>Multivariate logistic regression models of correlates of physical activity</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"2\"><bold>Model including personal correlates</bold></td><td align=\"center\" colspan=\"2\"><bold>Model including personal and environmental correlates</bold></td></tr><tr><td align=\"left\"><bold>Variable</bold></td><td align=\"center\"><bold>OR (95% CI)</bold></td><td align=\"center\"><bold>P</bold></td><td align=\"center\"><bold>OR (95% CI)</bold></td><td align=\"center\"><bold>P</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Housing tenure (reference: social renter)</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Owner-occupier</td><td align=\"center\">0.67 (0.46, 0.96)</td><td align=\"center\">0.028</td><td align=\"center\">0.66 (0.46, 0.95)</td><td align=\"center\">0.026</td></tr><tr><td align=\"left\">Other</td><td align=\"center\">1.41 (0.72, 2.79)</td><td align=\"center\">0.32</td><td align=\"center\">1.45 (0.73, 2.87)</td><td align=\"center\">0.29</td></tr><tr><td align=\"left\"><bold>Composite variable (reference: BMI≥25 and difficulty walking)</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">BMI&lt;25, no difficulty</td><td align=\"center\">5.49 (2.97, 10.16)</td><td align=\"center\">&lt;0.001</td><td align=\"center\">5.55 (3.00, 10.28)</td><td align=\"center\">&lt;0.001</td></tr><tr><td align=\"left\">BMI&lt;25, difficulty</td><td align=\"center\">0.32 (0.10, 1.01)</td><td align=\"center\">0.053</td><td align=\"center\">0.31 (0.10, 0.98)</td><td align=\"center\">0.047</td></tr><tr><td align=\"left\">BMI≥25, no difficulty</td><td align=\"center\">3.93 (2.11, 7.32)</td><td align=\"center\">&lt;0.001</td><td align=\"center\">3.92 (2.10, 7.31)</td><td align=\"center\">&lt;0.001</td></tr><tr><td align=\"left\"><bold>Individual items in neighbourhood scale</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Traffic volume</td><td/><td/><td align=\"center\">0.84 (0.70, 1.00)</td><td align=\"center\">0.050</td></tr><tr><td align=\"left\"><bold>Day of travel diary (reference: weekend)</bold></td><td/><td/><td/><td/></tr><tr><td align=\"left\">Weekday</td><td align=\"center\">0.64 (0.44, 0.93)</td><td align=\"center\">0.019</td><td align=\"center\">0.62 (0.43, 0.91)</td><td align=\"center\">0.015</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional file 1</title><p>Survey questionnaire</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S2\"><caption><title>Additional file 2</title><p>Further details of multivariate logistic regression modelling</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><p>n = 1322. * On a government training scheme, full-time student, unemployed, disabled, invalid or permanently sick, or caring for home and family or dependants. ^†Rented in the private sector, part-owned and part-rented, or other form of tenure.</p></table-wrap-foot>", "<table-wrap-foot><p>n = 1099. sd: standard deviation. IQR: interquartile range. * Walking and cycling combined.</p></table-wrap-foot>", "<table-wrap-foot><p>n = 833</p></table-wrap-foot>", "<table-wrap-foot><p>n = 831. * Does not work or study or usually works at home or from home. OR: Exponent of estimated regression coefficient, i.e. estimated odds ratio. 95% CI: 95% confidence interval for estimated odds ratio.</p></table-wrap-foot>", "<table-wrap-foot><p>n = 831. * Does not work or study or usually works at home or from home. OR: Exponent of estimated regression coefficient, i.e. estimated odds ratio. 95% CI: 95% confidence interval for estimated odds ratio.</p></table-wrap-foot>", "<table-wrap-foot><p>n = 684. OR: Exponent of estimated regression coefficient, i.e. estimated odds ratio. 95% CI: 95% confidence interval for estimated odds ratio.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1479-5868-5-43-1\"/>", "<graphic xlink:href=\"1479-5868-5-43-2\"/>", "<graphic xlink:href=\"1479-5868-5-43-3\"/>" ]

[ "<media xlink:href=\"1479-5868-5-43-S1.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1479-5868-5-43-S2.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Mucinous carcinoma of the ovary accounts for 5–10% of all primary epithelial ovarian cancer [##UREF##0##1##]. Patients with mucinous ovarian cancer generally undergo the same first- and second-line treatment as patients with other histological subtypes [##REF##16239238##2##]. However, very few reports in the literature have been published on this topic and activity of chemotherapy has been described in a limited number of patients and only in the first-line setting [##REF##15020606##3##, ####REF##16101169##4##, ##UREF##1##5##, ##REF##16139343##6####16139343##6##]. It has recently been shown in two different series of 27 and 45 patients, that advanced mucinous ovarian carcinoma have a poor response to first line chemotherapy [##REF##15020606##3##,##REF##16139343##6##]. Thus, resistance to chemotherapy has been claimed as one of the main cause of the worse prognosis of mucinous ovarian cancer [##REF##15020606##3##].</p>", "<p>The SOCRATES (<bold>S</bold>tudy of an <bold>O</bold>varian <bold>C</bold>ancer cohort <bold>R</bold>ecurred <bold>A</bold>fter first-line <bold>T</bold>reament: a r<bold>E</bold>strospectivy <bold>S</bold>urvey) study was planned to retrospectively assess the pattern of care of patients with recurrent platinum-sensitive ovarian cancer observed in Italy in the years 2000–2002 [##REF##17878745##7##]. Using this cohort of patients we evaluated the response of mucinous cancer to chemotherapy in the recurrent setting.</p>" ]

[ "<title>Methods</title>", "<p>Patients with recurrent advanced ovarian cancer and a recurrence free interval (RFI) longer than 6 months were considered eligible for the study. The patients were observed in the years 2000–2002 in 37 Italian centres. Data were collected between April and September 2005. Four-hundred-ninety-three patient files were screened and 408 were considered eligible and analyzed in the present study.</p>", "<p>The descriptive analysis of the data has been performed in 2 different subgroups identified according to histology: mucinous cancer and non-mucinous cancer. No central pathology assessment of the cancer samples was done.</p>", "<p>Clinical, pathological and treatment characteristics at initial diagnosis, as well as at recurrence, including surgical and medical treatment (up to 6 lines of chemotherapy) of the recurrence were considered. Response rate was calculated considering RECIST [##REF##10655437##8##] or Ca 125 criteria [##REF##8622070##9##].</p>", "<p>Overall survival was defined as the time elapsed between recurrence diagnosis and the date of death or the date of last follow-up information for live patients. Time to progression and overall survival were described y the Kaplan-Meier product limit method [##UREF##2##10##].</p>", "<p>Differences among baseline variables were analyzed by the Student t test and Wilcoxon rank test for quantitative variables, and by the Mantel Haenszel test and the Chi-square method for the qualitative variables. Differences were considered statistically significant when <italic>p </italic>&lt; 0.05.</p>", "<p>All analysis was done using SAS^®(SAS Institute Inc., Cary, NC, USA-version 9.1.3) statistical software.</p>" ]

[ "<title>Results</title>", "<p>Mucinous tumors were diagnosed in 20 patients, as compared with 388 patients with other histological subtypes (table ##TAB##0##1##). Median age, performance status, results of primary surgery were similar between the two groups. In mucinous ovarian cancer, the grading of the tumors was lower than in the other subtypes (<italic>p </italic>= 0.0056) and stage at diagnosis was less advanced (p = 0.025)</p>", "<p>The main characteristics of the patients at time of recurrence are shown in table ##TAB##1##2##. A statistically significant difference was found in performance status, that was worse in the mucinous group (p = 0.024), while no differences were found in the number of disease sites, age and recurrence free interval.</p>", "<p>About 20% of patients underwent secondary cytoreduction in both groups, with a lower number of patients optimally debulked (no residual disease) in the group of patients with mucinous cancer (p = 0.03). The majority of patients with mucinous tumours had increased CA 125 levels at recurrence (85%).</p>", "<p>Details on second-line chemotherapy are shown in the table ##TAB##2##3##. Patients with mucinous cancer received as second line therapy more frequently single agent platinum (42.1%) than platinum-combination therapy (31.6%) or other non-platinum chemotherapy (26.3%) (p = 0.026). The response rate (CR + PR) to the second line chemotherapy was lower in mucinous cancer than in non-mucinous one (36.4% vs 62.6%, respectively, p = 0.04). Moreover, patients with mucinous cancer received a lower number of lines of chemotherapy as compared to the other histotypes (p = 0.0023). Median progression free survival was 4.5 months in the mucinous and 8 months in non-mucinous group (p = 0.0292). Overall median survival from recurrence was 17.9 months in the mucinous and 28.8 months in non-mucinous group (p = 0.0028) (Figure ##FIG##0##1##).</p>", "<p>In the mucinous cancer group responses were obtained with carboplatin, cisplatin, and carboplatin/paclitaxel (2 responses in patients with 6–12 months and 2 responses in patients with &gt; 12 recurrence free interval). Among patients treated with non platinum-agents, no response was observed at second line, while responses were achieved in third-fourth line with paclitaxel (1/2 patients), topotecan (1/4 patients) and cyclophosphamide (1/1); no activity was observed with liposomal doxorubicin (0/4 patients) and gemcitabine (0/1 patient).</p>" ]

[ "<title>Discussion</title>", "<p>This retrospective study indicates that recurrent mucinous ovarian cancer has a lower response rate to chemotherapy and a worst prognosis compared to non-mucinous subtypes. Moreover, patients receive less chemotherapy lines for recurrence as compared to other histotypes and when undergo secondary cytoreduction, this is less effective. At our knowledge, this analysis describes for the first time the response rate to second line chemotherapy in patients with platinum sensistive mucinous ovarian cancer. At baseline, the only main characteristic differentiating mucinous from non-mucinous tumour was the lower grade of the cancer, according to what previously observed [##REF##16101169##4##]. Although we have not enough data to state that the poor response to chemotherapy is related to the lower grade of the tumours, it is possible to speculate that recurrent low grade cancer may benefit from a more aggressive attempt of cytoreduction before medical therapy. Unfortunately, in our series the patients that underwent secondary cytoreduction did not achieve the goal of obtaining an absence of residual disease; of course, the small number of patients does not allow to reach a definitive conclusion regarding the role of surgery in the treatment of recurrent mucinous ovarian cancer. No other disease related characteristics differed between mucinous and non-mucinous patients at recurrence.</p>", "<p>Mucinous carcinomas of the ovary includes 5–10% of ovarian carcinomas, although recent refinements in the interpretation of the histological features of noninvasive and metastatic mucinous carcinomas suggest that this may be an overestimate [##UREF##0##1##,##REF##16546243##11##]. Clinical stage is the most important predictor of survival in mucinous ovarian carcinoma. The early stages confer a better overall prognosis for survival [##REF##16546243##11##,##REF##11855877##12##], while the advanced disease has been associated with a poorer survival compared to the other histological subgroups [##REF##16546243##11##, ####REF##11855877##12##, ##REF##1904477##13####1904477##13##].</p>", "<p>The rarity of the disease is the main reason of the paucity of literature data regarding the activity of chemotherapy in this entity. Cloven [##REF##14751152##14##] have shown, <italic>\"in vitro\"</italic>, that the frequency of extreme drug resistance to chemotherapeutic agents differs significantly among histological subtypes of epithelial ovarian cancer. These authors demonstrated that mucinous ovarian cancer cells are more frequently resistant to cisplatin, but less frequently resistant to topotecan and doxorubicin compared to papillary serous tumors [##REF##14751152##14##], however clinical data are lacking.</p>", "<p>In a case-controlled study Hess [##REF##15020606##3##] showed, on 27 mucinous and 54 other histological types,, that patients with advanced mucinous ovarian cancer have a poorer response to platinum-based first-line chemotherapy compared with patients with other histological subtypes, along with a worse survival. In this series, only 37% of the patients were treated with carboplatin/paclitaxel combination as first-line treatment, while the remainder received carboplatin alone or platinum plus anthracyclines. The overall response rate was 26% in first-line chemotherapy, while the response rate in second- and third-line chemotherapy was not reported [##REF##15020606##3##]. A poor response to first line chemotherapy has been described by the Hellenic Cooperative Group [##REF##16139343##6##]. In a previous study, we also showed in 21 consecutive patients with mucinous ovarian cancer treated in a single institution that the response rate to first line chemotherapy was significantly lower than that found in the other histological subgroups, with paclitaxel being the only drug showing activity in second line [##REF##16101169##4##].</p>", "<p>Platinum-sensitive recurrent ovarian cancer is usually treated with carboplatin/paclitaxel or carboplatin/gemcitabine, based on the trials showing superiority of combination chemotherapy versus single agent carboplatin [##REF##12826431##15##,##REF##16966687##16##]. In our study an higher proportion of patients with mucinous cancer was treated at recurrence with single agent platinum than with platinum based combination therapy or other non platinum agents. Data clearly indicate that patients with recurrent mucinous ovarian cancer with a recurrence free interval higher than 6 months can respond to a platinum re-treatment, although the response rate is lower than that observed in non-mucinous cancer. Overall, recurrent mucinous cancer patients receive less chemotherapy lines than the others, probably also due to the lack of data in the literature showing activity for the chemotherapy agents more frequently used in this disease.</p>", "<p>Here we report for the first time some responses to paclitaxel, topotecan and cyclophosphamide, while no response was observed with liposomal doxorubicin and gemcitabine. Overall the response rate to non-platinum agents was quite poor.</p>", "<p>A possible limitation of our report is the retrospective nature of the analysis: therefore, survival data should be interpreted with caution. Another weakness of the study may be the lack of a central pathology review, to confirm these were mucinous ovarian cancers versus metastatic malignancies of gastrointestinal origin. However, the differential diagnosis between gastrointestinal and ovarian cancer is a major problem at time of initial diagnosis. In fact, in the case of our series of recurrent ovarian cancer this limitation may be less important since it is likely that during the disease free interval the potential presence of a primary gastrointestinal cancer would have been diagnosed. Moreover, a worse performance status was found in patients with mucinous tumors: however, due to the small number of patients, no definite conclusions can be drawn regarding the potential effect of performance status on the poor survival of patients with mucinous tumors.</p>", "<p>Conventional parameters used to predict the clinical behaviour of advanced ovarian cancer may not adequately correlate with prognosis in mucinous carcinoma. Several studies have shown that mucinous ovarian cancer has a different pattern of expression of some molecular factors compared to the other subtypes. It is possible that a better understanding of tumour biology may help in determining which patients with mucinous ovarian cancer would benefit from traditional chemotherapy or should receive alternative chemotherapy agents. Several studies have shown that RAS mutations (specifically at KRAS codon 12) are prevalent in ovarian cancers of mucinous histology but not in tumors of non-mucinous histologies [##UREF##3##17##,##REF##9118042##19##]. On the contrary, mutation of p53, which is considered important in defining sensitivity to paclitaxel, is less frequent in mucinous tumors [##REF##11595686##20##]. Again, some studies have found that the expression of COX-2 was much less frequent in mucinous cancer than in serous and endometroid ovarian cancers [##REF##11408945##21##,##REF##11891188##22##]. Chemotherapy decisions tailored to the biology of mucinous ovarian cancer should be investigated in the future. The rarity of the disease should not discourage the assessment, in clinical trials, of the activity of different drugs, choosing first among those active in gastrointestinal cancer. Furthermore, <italic>\"in vitro\" </italic>drug response assays could be very useful to select patients that are likely to be resistant to traditional chemotherapy for whom to suggest an alternative, experimental treatment.</p>" ]

[ "<title>Conclusion</title>", "<p>In conclusion, we showed that mucinous ovarian cancer has a poor response to chemotherapy in the recurrence setting along with a worst prognosis. Responses to platinum re-treatment are less frequent than in non-mucinous cancer, while anecdotal responses occur with non-platinum agents. Studies with alternative chemotherapy combinations are mandatory in this histological subgroup.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Mucinous ovarian carcinoma have a poorer prognosis compared with other histological subtypes. The aim of this study was to evaluate, retrospectively, the activity of chemotherapy in patients with platinum sensitive recurrent mucinous ovarian cancer.</p>", "<title>Methods</title>", "<p>The SOCRATES study retrospectively assessed the pattern of care of a cohort of patients with recurrent platinum-sensitive ovarian cancer observed in the years 2000–2002 in 37 Italian centres. Data were collected between April and September 2005. Patients with recurrent ovarian cancer with &gt; 6 months of platinum free interval were considered eligible.</p>", "<title>Results</title>", "<p>Twenty patients with mucinous histotype and 388 patients with other histotypes were analyzed. At baseline, mucinous tumours differed from the others for an higher number of patients with lower tumor grading (p = 0.0056) and less advanced FIGO stage (p = 0.025). At time of recurrence, a statistically significant difference was found in performance status (worse in mucinous, p = 0.024). About 20% of patients underwent secondary cytoreduction in both groups, but a lower number of patients were optimally debulked in the mucinous group (p = 0.03). Patients with mucinous cancer received more frequently single agent platinum than platinum based-combination therapy or other non-platinum schedules as second line therapy (p = 0.026), with a response rate lower than in non-mucinous group (36.4% vs 62.6%, respectively, p = 0.04). Median time to progression and overall survival were worse for mucinous ovarian cancer. Finally, mucinous cancer received a lower number of chemotherapy lines (p = 0.0023).</p>", "<title>Conclusion</title>", "<p>This analysis shows that platinum sensitive mucinous ovarian cancer has a poor response to chemotherapy. Studies dedicated to this histological subgroup are needed.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>SP, GF, GS, PS participated in the design of the study; GM performed the statistical analysis. SP conceived of the study, and participated in its design and coordination. FO, GC, DK, AV, LM, FG, LM, RL, EB, DGA, MB, AVL, RS, GM, DP, AM significantly contributed to data collection. All authors read and approved the final manuscript. Additional co-authors and participating institution are listed in the additional file ##SUPPL##0##1##.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2407/8/252/prepub\"/></p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>The SOCRATES project was supported by an unrestricted grant from GSK Italy. The MITO group is supported by a project grant from Associazione Italiana per la Ricerca sul Cancro. The authors remember Giovanni Favalli, recently deceased, that was a member of the SOCRATES board.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Overall survival from recurrence in patients with mucinous (◆) compared to other histotypes (■).</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Characteristics of the patients with recurrent mucinous ovarian cancer compared to other histological subtypes at the time of initial diagnosis of ovarian cancer.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><bold>Mucinous</bold></td><td align=\"center\"><bold>Other histotypes</bold></td><td align=\"center\"><bold>p</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Number of patients</bold></td><td align=\"center\">20</td><td align=\"center\">388</td><td/></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\"><bold>Age (years)</bold></td><td align=\"center\">20</td><td align=\"center\">384</td><td align=\"center\">n.s.*</td></tr><tr><td align=\"left\"> Mean (± s.d.)</td><td align=\"center\">54.9 ± 12.5</td><td align=\"center\">57.7 ± 10.8</td><td/></tr><tr><td align=\"left\"> median</td><td align=\"center\">55</td><td align=\"center\">57</td><td/></tr><tr><td align=\"left\"> range</td><td align=\"center\">25–71</td><td align=\"center\">31–94</td><td/></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\"><bold>FIGO stage at diagnosis</bold></td><td align=\"center\">20</td><td align=\"center\">383</td><td/></tr><tr><td align=\"left\"> I</td><td align=\"center\">20.0%</td><td align=\"center\">6.5%</td><td align=\"center\"><bold>0.0258**</bold></td></tr><tr><td align=\"left\"> II</td><td align=\"center\">10.0%</td><td align=\"center\">7.3%</td><td/></tr><tr><td align=\"left\"> III</td><td align=\"center\">65.0%</td><td align=\"center\">77.0%</td><td/></tr><tr><td align=\"left\"> IV</td><td align=\"center\">5.0%</td><td align=\"center\">9.1%</td><td/></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\"><bold>Grading</bold></td><td align=\"center\">12</td><td align=\"center\">348</td><td/></tr><tr><td align=\"left\"> 1</td><td align=\"center\">25.0%</td><td align=\"center\">3.4%</td><td align=\"center\"><bold>0.0056**</bold></td></tr><tr><td align=\"left\"> 2</td><td align=\"center\">16.7%</td><td align=\"center\">24.7%</td><td/></tr><tr><td align=\"left\"> 3</td><td align=\"center\">58.3%</td><td align=\"center\">71.8%</td><td/></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\"><bold>Result of cytoreductive surgery</bold></td><td align=\"center\">15</td><td align=\"center\">319</td><td/></tr><tr><td align=\"left\"> No residual disease</td><td align=\"center\">46.7%</td><td align=\"center\">26.0%</td><td align=\"center\"><bold>n.s.**</bold></td></tr><tr><td align=\"left\"> Optimal (≤ 1 cm residual disease)</td><td align=\"center\">20.0%</td><td align=\"center\">29.5%</td><td/></tr><tr><td align=\"left\"> Suboptimal (&gt; 1 cm residual disease)</td><td align=\"center\">33.3%</td><td align=\"center\">44.5%</td><td/></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\"><bold>ECOG performance status</bold></td><td align=\"center\">15</td><td align=\"center\">347</td><td/></tr><tr><td align=\"left\"> 0</td><td align=\"center\">60.0%</td><td align=\"center\">68.3%</td><td align=\"center\"><bold>n.s.**</bold></td></tr><tr><td align=\"left\"> 1</td><td align=\"center\">33.3%</td><td align=\"center\">26.5%</td><td/></tr><tr><td align=\"left\"> 2</td><td align=\"center\">6.7%</td><td align=\"center\">4.3%</td><td/></tr><tr><td align=\"left\"> 3</td><td align=\"center\">.</td><td align=\"center\">0.9%</td><td/></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\"><bold>Type of first line chemotherapy</bold></td><td align=\"center\">20</td><td align=\"center\">338</td><td align=\"center\"><bold>n.s.***</bold></td></tr><tr><td align=\"left\"> Platinum single agent</td><td align=\"center\">15.0%</td><td align=\"center\">11.1%</td><td/></tr><tr><td align=\"left\"> Platinum based combinations</td><td align=\"center\">85.0%</td><td align=\"center\">88.9%</td><td/></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Characteristics of patients with mucinous ovarian cancer with a recurrence free interval &gt; 6 months compared to other histological subtypes at the time of the diagnosis of recurrence.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><bold>Mucinous</bold></td><td align=\"center\"><bold>Other histotypes</bold></td><td align=\"center\"><bold>p</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Number of patients</bold></td><td align=\"center\">20</td><td align=\"center\">388</td><td/></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\"><bold>Age (years)</bold></td><td align=\"center\">20</td><td align=\"center\">380</td><td/></tr><tr><td align=\"left\"> Mean (± s.d.)</td><td align=\"center\">57.6 ± 12.8</td><td align=\"center\">59.8 ± 10.8</td><td align=\"center\">n.s.*</td></tr><tr><td align=\"left\"> Median</td><td align=\"center\">58</td><td align=\"center\">60.5</td><td/></tr><tr><td align=\"left\"> Range</td><td align=\"center\">28–78</td><td align=\"center\">33–97</td><td/></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\"><bold>PS Ecog</bold></td><td align=\"center\">20</td><td align=\"center\">360</td><td/></tr><tr><td align=\"left\"> 0</td><td align=\"center\">45.0%</td><td align=\"center\">62.8%</td><td align=\"center\"><bold>0.0241**</bold></td></tr><tr><td align=\"left\"> 1</td><td align=\"center\">40.0%</td><td align=\"center\">33.9%</td><td/></tr><tr><td align=\"left\"> 2</td><td align=\"center\">15.0%</td><td align=\"center\">3.3%</td><td/></tr><tr><td align=\"left\"> 3</td><td align=\"center\">.</td><td align=\"center\">.</td><td align=\"center\">.</td></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\"><bold>Recurrence free interval </bold>(N)</td><td align=\"center\">17</td><td align=\"center\">378</td><td/></tr><tr><td align=\"left\"> 6–12 months</td><td align=\"center\">58.8%</td><td align=\"center\">38.9%</td><td align=\"center\"><bold>n.s.**</bold></td></tr><tr><td align=\"left\"> &gt; 12 months</td><td align=\"center\">41.2%</td><td align=\"center\">61.1%</td><td/></tr><tr><td align=\"left\"> <bold>median (range) </bold>– <italic>months</italic><bold>:</bold></td><td align=\"center\">10.6 (5–141)</td><td align=\"center\">15.3 (5–160)</td><td align=\"center\"><bold>n.s.****</bold></td></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\"><bold>Number of disease sites</bold></td><td align=\"center\">20</td><td align=\"center\">335</td><td/></tr><tr><td align=\"left\"> 1</td><td align=\"center\">25.0%</td><td align=\"center\">44.8%</td><td align=\"center\"><bold>n.s.**</bold></td></tr><tr><td align=\"left\"> &gt; 1</td><td align=\"center\">75.0%</td><td align=\"center\">55.2%</td><td/></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\"><bold>Surgery</bold></td><td align=\"center\">19</td><td align=\"center\">375</td><td/></tr><tr><td align=\"left\"> Yes</td><td align=\"center\">26.3%</td><td align=\"center\">20.5%</td><td align=\"center\"><bold>n.s.***</bold></td></tr><tr><td align=\"left\"> No</td><td align=\"center\">73.7%</td><td align=\"center\">79.5%</td><td/></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\"><bold>Result of cytoreductive surgery</bold></td><td align=\"center\">5</td><td align=\"center\">64</td><td/></tr><tr><td align=\"left\"> No residual</td><td align=\"center\">.</td><td align=\"center\">50.0%</td><td align=\"center\"><bold>0.0308***</bold></td></tr><tr><td align=\"left\"> Residual desease</td><td align=\"center\">100.0%</td><td align=\"center\">50.0%</td><td/></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\"><bold>Ca 125</bold></td><td align=\"center\">14</td><td align=\"center\">247</td><td align=\"center\"><bold>n.s.**</bold></td></tr><tr><td align=\"left\"> Normal: ≤ 35</td><td align=\"center\">14.3%</td><td align=\"center\">12.1%</td><td/></tr><tr><td align=\"left\"> &gt; 35 U/ml</td><td align=\"center\">85.7%</td><td align=\"center\">87.9%</td><td/></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Response to second line chemotherapy in patients with mucinous ovarian cancer with a recurrence free interval &gt; 6 months compared to other histological subtypes</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><bold>Mucinous</bold></td><td align=\"center\"><bold>Other histotypes</bold></td><td align=\"center\"><bold>p</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>N pts.</bold></td><td align=\"center\">20</td><td align=\"center\">388</td><td/></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\"><bold>Type of second line chemotherapy</bold></td><td align=\"center\">19</td><td align=\"center\">384</td><td/></tr><tr><td align=\"left\"> Platinum single agent</td><td align=\"center\">(8) 42.1%</td><td align=\"center\">(67) 17.4%</td><td align=\"center\"><bold>0.0259*</bold></td></tr><tr><td align=\"left\"> Platinum based combination</td><td align=\"center\">(6) 31.6%</td><td align=\"center\">(184) 47.9%</td><td/></tr><tr><td align=\"left\"> No platinum</td><td align=\"center\">(5) 26.3%</td><td align=\"center\">(133) 34.6%</td><td/></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\"><bold>Response rate to second line</bold></td><td/><td/><td/></tr><tr><td align=\"right\"><bold>All evaluable patients</bold></td><td align=\"center\">11</td><td align=\"center\">340</td><td/></tr><tr><td align=\"left\">  CR + PR</td><td align=\"center\">(4) 36.4%</td><td align=\"center\">(213) 62.6%</td><td align=\"center\"><bold>0.0407*</bold></td></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"right\"><bold>Platinum</bold></td><td align=\"center\">7</td><td align=\"center\">227</td><td/></tr><tr><td align=\"left\">  CR + PR</td><td align=\"center\">(4) 57.1%</td><td align=\"center\">(170) 74.9%</td><td align=\"center\"><bold>n.s.*</bold></td></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"right\"><bold>Non-platinum</bold></td><td align=\"center\">4</td><td align=\"center\">113</td><td/></tr><tr><td align=\"left\">  CR + PR</td><td align=\"center\">(-) 0%</td><td align=\"center\">(43) 38.1%</td><td align=\"center\"><bold>n.s.*</bold></td></tr><tr><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\"><bold>Number of chemotherapy lines received for recurrence</bold></td><td/><td/><td/></tr><tr><td align=\"left\"> Mean (± s.d.)</td><td align=\"center\">1.9 ± 1.1</td><td align=\"center\">2.8 + 1.3</td><td align=\"center\">0.0023**</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional file 1</title><p>participating institutions and co-authors.</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><p>• t-test; ** χ²Mantel-Haenszel; *** χ²</p></table-wrap-foot>", "<table-wrap-foot><p>* t-test; ** χ²Mantel-Haenszel; *** χ²; ****Wilcoxon</p></table-wrap-foot>", "<table-wrap-foot><p>* χ²; **t-test</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1471-2407-8-252-1\"/>" ]

[ "<media xlink:href=\"1471-2407-8-252-S1.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Human tetratricopeptide repeat domain 9 (TTC9) was first reported as a hypothetical protein KIAA0227 by Nagase et al, based on the sequence analysis of a cDNA clone isolated from a brain cDNA library [##REF##9039502##1##]. It was later identified as a steroid hormone-regulated gene in various breast cancer cells [##REF##16678794##2##]. It seems that there is a family of TTC9 protein. The MGC program at the National Institutes also identified cDNA sequences named as TTC9B and TTC9C, which share 46% and 35% homology with TTC9A in amino acids sequence, respectively [##REF##12477932##3##]. However, TTC9B and TTC9C have not been identified at the protein level. Nonetheless, to keep up with the information in the NCBI database, TTC9 is now referred to as TTC9A in this article.</p>", "<p>TTC9 family belongs to a large family of tetratricopeptide repeat (TPR)- containing proteins. The TPR domain is a 34 amino acids (aa) consensus motif that is found in tandem repeats of varying number in different proteins [##REF##10517866##4##, ####REF##7667876##5##, ##REF##15497498##6####15497498##6##]. Circular dichroism (CD) studies indicate that TPR motifs are approximately 50% α-helical structures with little or no β-sheet formation [##REF##2297790##7##]. Crystallographic structure analysis of TPR-containing proteins revealed that TPR motif generally forms an antiparallel α-helical hairpin [##REF##9482716##8##,##REF##10786835##9##]. Clustering of these hairpins in tandem generates a domain with a grooved surface and dimension that can conveniently grasp another polypeptide. Generally, by generating a flexible, mutable domain that can facilitate specific protein-protein interactions, the TPR motif presents an elegant evolutionary solution contributing to the fundamental biological importance of coordinating interactions among gene products [##REF##15497498##6##]. The functions of TPR containing protein include cell cycle control [##REF##1819514##10##], transcription and splicing events [##REF##2201901##11##], protein transport especially protein import [##REF##8265627##12##], regulatory phosphate turnover [##REF##7925273##13##], and protein folding [##REF##8999928##14##]. TTC9A contains three TPR domains at its carboxyl-terminus, at amino acid positions 57–90, 128–161 and 164–197.</p>", "<p>In previous studies, we have identified the open reading frame of TTC9 gene and confirmed the protein size of TTC9A to be 222 aa. Using mouse polyclonal antibody generated against TTC9A protein, TTC9A was shown to be ubiquitously expressed in human tissues. In breast cancer cells, TTC9A was predominantly concentrated to the endoplasmic reticulum and was regulated by a number of factors, such as growth factors, serum factors and steroid hormones [##REF##16678794##2##]. Although existing results suggest that TTC9A could be an important protein ubiquitously expressed in all cell and tissue types, the exact role of TTC9A remains unclear.</p>", "<p>In this study, we found that TTC9A mRNA was significantly over-expressed in breast cancer tissues compared with the adjacent normal breast tissues, which suggested TTC9A could be an important gene involved in hormone signaling and breast cancer development. By yeast-two-hybrid assay, we identified one of TTC9A interacting proteins, TM5/TM30nm, which is also referred to as Tm5NM-1 and is a non-muscle tropomyosin encoded by γ-tropomyosin gene. The tropomyosins are a group of actin-binding proteins found in skeletal muscle, smooth muscle and non-muscle tissues. They are either hetero- or homo-dimeric proteins with a rod-shaped, α-helical coiled-coil structure. Usually the dimers form a head-to-tail polymer running along the major groove in the actin filament [##REF##9002235##15##,##REF##450047##16##]. Mammalian and avian tropomyosins are encoded by four genes, i.e. α, β, γ, δ [##REF##8167032##17##]. Historically, the tropomyosin proteins have been divided into two classes, high-molecular-weight (HMW) and low-molecular- weight (LMW), which are ~284 aa and 247 aa in length, respectively. This size difference is generated by the use of alternative promoters [##REF##8167032##17##,##REF##11368517##18##], alternative splicing of mRNA [##REF##15953552##19##], and different 3' UTR processing [##REF##9660825##20##]. These mechanisms give rise to over 40 tropomyosin isoforms.</p>", "<p>Tropomyosins are believed to be involved in the stabilization of actin filaments [##REF##12176375##21##]. Actin filaments which lack tropomyosins tend to be rapidly undergoing assembly and disassembly process, such as those associated with neuronal growth cone filopodia and the leading edge of mammary adenocarcinoma cells [##REF##15953552##19##,##REF##12415009##22##, ####REF##7876361##23##, ##REF##9143561##24####9143561##24##]. In skeletal muscle, tropomyosins serve to mediate the effect of Ca²⁺on the actin-myosin interaction [##REF##488353##25##,##REF##156624##26##]. Instead of binding Ca²⁺directly, they perform this function by acting as bifunctional molecules, binding to actin on one hand, and providing specific sites for the binding of the troponin complex of regulatory proteins on the other hand [##REF##7096357##27##, ####REF##7310890##28##, ##REF##7107628##29##, ##REF##6822572##30####6822572##30##]. Though the precise function of non-muscle tropomyosin is less understood, some in vitro studies have shown that non-muscle tropomyosins are able to differentially protect actin from the severing action of gelsolin [##REF##2540194##31##] and can regulate the Mg-ATPase activity of myosins to varying degrees [##REF##7820856##32##]. Furthermore, it seems that non-muscle tropomyosins play an important role in tumor development. For example, down-regulation of tropomyosin 2 was essential in ras-mediated malignant transformation of fibroblasts [##REF##12588996##33##,##REF##11884615##34##]. Tropomyosin 3 isoform 2 (also termed as TM5/TM30nm or Tm5NM-1) was expressed at a higher level in highly metastatic B16 mouse cell line than in mouse cell line exhibiting a lower metastasis rate. Similar result was also obtained in rat cells [##REF##8753779##35##,##REF##9222299##36##].</p>", "<p>This study also identified specific domains in TTC9A that is crucial in the interaction with Tm5NM-1. The significance of the interaction between TTC9A and Tm5NM-1 is yet to be elucidated. On the other hand, our previous studies have shown that in breast cancer cells, the up-regulation of TTC9A expression by progesterone was associated with increased cell focal adhesion and motility. All these findings suggest a possible role of TTC9A in cell-matrix adhesion and in tropomyosin-mediated stabilization of actin microfilament.</p>" ]

[ "<title>Methods</title>", "<title>Cell line and reagents</title>", "<p>COS-7 cells were obtained from Dr. Koh Cheng Gee, School of Biological Sciences, Nanyang Technological University, Singapore.</p>", "<p>COS-7 cells were routinely maintained in phenol red- and high D-glucose- containing Dulbecco's Modified Eagle Medium (DMEM) supplemented with 7.5% fetal calf serum (FCS), 2 mM glutamine and 40 mg/L gentamicin.</p>", "<p>All cell culture reagents were bought from Invitrogen (Carlsbad, California). Fetal calf sera were from Hyclone (Logan, UT) or PromoCell GmbH (Heidelberg, Germany). All cell culture plastic wares were purchased from Falcon (Becton Dickinson, San Jose, CA), NUNC (Nalge Nunc International, Rochester, NY), or Corning (Corning, NY).</p>", "<title>TTC9A expression in breast cancer tissues</title>", "<p>Human tissue samples were obtained from the Tissue Repository at the National Cancer Centre (NCC), Singapore. Tissue samples were harvested at the time of mastectomy or breast conserving surgery with prior signed informed consent from the patients. Matched pairs of malignant tissue and the adjacent normal breast tissue were harvested and confirmed histologically by a pathologist and were snap frozen in liquid nitrogen. The cases utilized in this study were collected between January 2002 and December 2003. Clinicopathological data such as tumor size, nuclear grade and hormone receptor status were obtained from a prospective database. This study was approved by the ethics committee at NCC.</p>", "<p>25 matched pairs of breast tissues were mashed using a mortar and pestle. Total RNA was extracted using TRIzol reagent (Life Technologies Inc.) according to the manufacturer's instructions. 5 μg of total RNA from each sample was reverse transcribed using Superscript II reverse transcriptase (Invitrogen). 1 μl cDNA produced from each RT reaction was amplified by PCR. The primers used here were 5'-CACAT GTCTATAACGATTTCC-3' (forward) and 5'-TGCAGGAAACAGGGG ACTCTC-3' (reverse). 10 μl PCR products corresponding to individual breast tissue sample were separated on 1% agarose gel and transferred to nylon Hybond-N membrane (Amersham Biosciences). ³²P-labeled TTC9A were generated by random priming reaction (Amersham Biosciences) using the same PCR product of TTC9A. The band intensities were analyzed using Bio-Rad Molecular Image Analyzer. As internal controls, 36B4 and GAPDH genes were also included for normalization. The primers used to amplify 36B4 gene were 5'-GATTGGCTACCCAACTGTTGCA-3' (forward) and 5'-CAGGGGCAG CAGCCACAAAGGC-3' (reverse). The primers for GAPDH were 5'-TGCACCACCA ACTGCTTAG-3' (forward) and 5'-GAGGCAGGGATGATG TTC-3' (reverse).</p>", "<title>Yeast-two-hybrid assay screen</title>", "<p>Yeast-two-hybrid assay screen was carried out using MATCHMAKER LexA Two-Hybrid System from Clontech Laboratories Inc (Mountain View, CA) according to the manufacturer's instructions. Yeast <italic>Saccharomyces cerevisiae </italic>MATα strain EGY48 was transformed with lacZ reporter gene, bait plasmid containing TTC9A coding sequence (TTC9A-pLexA) and cDNA libraries of human breast cancer cell line MCF-7 (OriGene Technologies, Inc., Rockville, MD) using the lithium acetate method [##REF##7785336##37##]. Transformed EGY48 was plated onto SD/Gal/Raf/-His/-Trp/-Ura/-Leu+X-Gal plates. An interaction was considered positive when two reporter genes, LEU2 and <italic>lac</italic>Z, were activated. The interactions were further verified by co-immunoprecipitation or GST pull-down assay.</p>", "<title>Transfection of COS-7 cells</title>", "<p>COS-7 cells were seeded the day before transfection. The confluence of cells at the time of transfection was 40%–50%. Transfection was carried out with FuGENE 6 Transfection Reagent (Roche Diagnostics, Basel, Switzerland) according to the manufacturer's instructions. Cells were harvested at 48 hours post-transfection. The expression of protein was examined by Western blotting.</p>", "<title>Cell lysates preparation and protein concentration quantification</title>", "<p>Cells were lysed on ice with cold lysis buffer containing 100 mM NaF, 50 mM HEPES (pH7.5), 150 mM NaCl, 1% Triton X-100, 1 mM PMSF and the cocktail of proteinase inhibitors (5 μg/ml pepstatin A, 5 μg/ml leupeptin, 2 μg/ml aprotinin and 1 mM Na3VO4). The cell debris was discarded by centrifuging at 14,000 rpm for 12 min at 4°C. The supernatants were immediately frozen down in liquid nitrogen and were stored at -80°C for future use. Protein concentrations were determined by BCA protein assay kit (Pierce, Rockford, IL).</p>", "<title>Purification of GST-TTC9A</title>", "<p>The 669 bp TTC9A coding sequence was cloned into pGEX-5X-3 (Amersham Biosciences) for the expression of GST-TTC9A protein. GST or GST-TTC9A protein was purified with Glutathione Sepharose 4B (Amersham Biosciences) according to the manufacturer's instructions. Eluted protein was pooled and dialyzed in PBS for future experiment.</p>", "<title>GST pull-down assay</title>", "<p>COS-7 cells were transfected with Tm5NM-1-(His)₆expression vector or control vector (pcDNA3.1/myc-His(-) B) respectively and total cell lysates were collected at 48 h post-transfection. 60 μg GST-TTC9A protein was immobilized onto 12 μl Glutathione Sepharose 4B gel (Amersham Biosciences) by gentle rotation at 4°C for 2 h. 300 μg total cell lysates collected were then added and the total reaction volume was brought up to 1 ml by PBS. The reactions were incubated overnight at 4°C with gentle rotation. Nonspecific binding proteins were removed by washing in cold washing buffer containing 100 mM NaF, 50 mM HEPES (pH 7.5) and 150 mM NaCl for four times followed by one more wash in PBS. Proteins bound to the beads were eluted with 2 × SDS-PAGE sample buffer and were separated on an SDS-PAGE gel. Tm5NM-1-(His)₆protein was detected using anti-His antibody (Amersham Biosciences). GST protein expressed by empty pGEX-5X-3 vector was included as a negative control.</p>", "<title>Co-immunoprecipitation with anti-flag M2 Affinity Gel</title>", "<p>Anti-flag M2 Affinity Gel was bought from Sigma-Aldrich (St. Louis, MO). Co-immunoprecipitation was carried out according to the manufacturer's instructions. Briefly, COS-7 cells were transfected with flag-TTC9A or flag-TTC9A fragments and Tm5NM-1-(His)₆expression vectors using FuGENE 6 Transfection Reagent. The amount of Tm5NM-1-(His)₆vector was two times more than those of flag-TTC9A or flag-TTC9A fragments vectors. 400 μg of total protein lysates collected were mixed with 15 μl anti-flag M2 Affinity Gel and the total reaction volume was brought up to 1 ml by lysis buffer. After incubating with the cell lysates overnight at 4°C with gentle rotation, the affinity gel was washed three times with 0.5 ml TBS. Proteins bound were eluted with 2 × SDS-PAGE sample buffer and were loaded onto an SDS-PAGE gel. Tm5NM-1-(His)₆protein was detected using anti-His antibody (Amersham Biosciences). COS-7 cells transfected with empty pXL-flag vector and Tm5NM-1-(His)₆expression vector were included as a negative control.</p>", "<title>Co-immunoprecipitation of endogenous Tm5NM-1 and TTC9A-flag</title>", "<p>COS-7 cells were transfected with TTC9A-flag expression vector using FuGENE 6 Transfection Reagent according to the manufacturer's instructions. Cell lysates were collected at 48 h post-transfection. 500 μg of total protein lysates collected were mixed with 5 μl anti-Tm5NM-1 antibody (Chemicon International Inc., Temecula, CA) or goat pre-immune serum for 4 h at 4°C with gentle rotation. 40 μl protein A/G plus-agarose beads were then added and the mixture was incubated overnight with gentle rotation at 4°C. The agarose beads were washed four times with washing buffer containing 100 mM NaF, 50 mM HEPES (pH 7.5), 150 mM NaCl, 0.01% Triton X-100. Proteins bound were eluted with 2 × SDS-PAGE sample buffer and were loaded onto an SDS-PAGE gel. TTC9A-flag protein was detected using anti-flag antibody (Sigma-Aldrich).</p>", "<title>Statistical analysis</title>", "<p>The experiment for TTC9A expression in human breast cancer tissues and adjacent normal tissues were analyzed by the Mann-Whitney nonparametric test using the SPSS program for Windows, version 11.5. Difference between the expression of TTC9A in normal and cancer tissues was considered as significant when the P value is less than 0.05.</p>" ]

[ "<title>Results and Discussion</title>", "<title>TTC9A is over-expressed in breast cancer tissues compared with the adjacent normal tissues</title>", "<p>Previous results have shown that TTC9A is a hormonally regulated gene <italic>in vitro </italic>[##REF##16678794##2##]. As breast cancer is well-known to be a hormone-dependent malignancy, it is noteworthy to know whether TTC9A is over-expressed in breast cancer tissues and if its expression is correlated with hormone receptor status. 25 matched pairs of human breast cancer tissue and the adjacent normal tissue were analyzed for TTC9A mRNA expression. The results presented in Fig. ##FIG##0##1## were obtained by RT-PCR as illustrated in the \"materials and methods\" and the PCR products were quantitated by Southern blotting analysis. 36B4, which codes for human acidic ribosomal phosphoprotein P0, was used as a control for cDNA input (Fig. ##FIG##0##1A##). Since the expression of 36B4 varied among samples, GAPDH was also included as a normalization standard for cDNA input. It turned out that the expression levels of these housekeeping genes were always higher in tumor tissues compared with the corresponding normal tissues. This variation has been reported before. For example, GAPDH expression was 3.3-fold higher in seminoma compared with normal testis [##REF##10897012##38##]. Similarly, GAPDH transcription was significantly greater in both colonic adenomas and cancers than in normal mucosa [##REF##15823405##39##]. Nonetheless, the results revealed that the relative expression of TTC9A mRNA, when normalized to either 36B4 or GAPDH, was significantly higher in breast cancer tissue compared with its adjacent normal tissue (Fig. ##FIG##0##1##) (<italic>P </italic>&lt; 0.00001). It is also notable that nearly all the tumor tissues expressed higher level of TTC9A mRNA compared with its adjacent controls. However, we found no correlation of TTC9A expression with other clinic pathological data such as tumor size, nuclear grade, axillary's lymph node status or hormone receptor expression.</p>", "<title>Yeast-two-hybrid assay identified Tm5NM-1 as one of TTC9A interacting proteins</title>", "<p>The observation that TTC9A was over-expressed in breast cancer tissues suggests potential relevance of the protein in breast cancer biology. In an initial attempt to investigate the relationship between TTC9A function and cancer development, we searched for binding partners of the protein. To accomplish this, we performed a yeast-two-hybrid screening of a cDNA library from breast cancer cell line MCF-7, using full-length TTC9A as the bait. The coding region of the 222 aa TTC9A protein was cloned into the 'bait' vector, pLex A, which contains the DNA-binding domain. MCF-7 cDNA libraries were fused to the activation domain in the 'pray' vector p42AD. The bait vector and the cDNA libraries were transformed into yeast strain EGY48 together with a lacZ reporter gene. 30 positive clones, in which the coding regions of the library plasmids were in frame with the activation domain according to the sequencing results, were obtained after two rounds of specificity test. Among them, 5 clones contained genes coding for human tropomyosin Tm5NM-1, which is encoded by γ-tropomyosin gene of the tropomyosin family.</p>", "<title>TTC9A protein can interact with cellular expressed Tm5NM-1</title>", "<p>The binding of TTC9A to Tm5NM-1 was further examined by GST-pull down assay. In Fig. ##FIG##1##2A##, Tm5NM-1-(His)₆expression vector was transfected into COS-7 cells and GST-TTC9A was used as a \"bait\" to pull down the cellular expressed Tm5NM-1 protein. As shown in the figure, Tm5NM-1 was pulled down by GST-TTC9A, but not by the GST-tag, suggesting that GST-TTC9A did interact with cellular expressed Tm5NM-1. The specificity of the interaction was further confirmed by GST pull-down assay with different amount of bait protein. Fig. ##FIG##1##2B## showed that the protein amount of Tm5NM-1 pulled down by GST-TTC9A increased proportionally to the amount of bait protein used in the assay. However, bands other than the expected Tm5NM-1 were also observed in Fig. ##FIG##1##2B## and these non-specific bands were not observed in sample with GST protein. These unexpected bands could be due to non-specific pull-down or degraded GST-TTC9A.</p>", "<title>TTC9A interacts with Tm5NM-1 in mammalian cells</title>", "<p>The interaction between Tm5NM-1 and TTC9A was further verified by co-immunoprecipitation in mammalian cells. As shown in Fig. ##FIG##1##2C##, cellular expressed TTC9A-flag pulled down prominent amount of Tm5NM-1-(His)₆protein in COS-7 cells. To further confirm the interaction, TTC9A-flag expression vector was transfected into COS-7 cells and the interaction between endogenous Tm5NM-1 and cellular expressed TTC9A was examined by co-immunoprecipitation with anti-Tm5NM-1 antibody. Fig. ##FIG##1##2D## revealed that cellular expressed flag-TTC9A could also bind to endogenous Tm5NM-1.</p>", "<title>Identification of the regions/domains in TTC9A which are important in the interaction with Tm5NM-1</title>", "<p>To specify the domains that interact with Tm5NM-1, different truncations of TTC9A protein shown in Fig. ##FIG##2##3## were constructed and tested in COS-7 cells. As is shown in Fig. ##FIG##3##4A##, all truncations were expressed at the expected protein sizes. It is to be noted that the whitish streak in the centre of the bands for TTC9A (1–161), TTC9A (1–197), full length TTC9A and TTC9A (51–222) were due to the over-saturation of the signal. This means that the amounts of protein expressed by these constructs were not necessarily less than those by other smaller truncations which showed broader bands. The reason may be that in 12% gel, the smaller proteins tend to be more diffuse resulting in a broader band, whereas higher molecular weight proteins tend to be more compact in migration.</p>", "<p>Fig. ##FIG##3##4B## showed that TTC9A (1–50) and TTC9A (1–70) did not pull-down Tm5NM-1 visibly as compared with vector-transfected control, regardless of the very high expression level of TTC9A (1–70). TTC9A (1–100), which contains the first TPR domain, interacted with Tm5NM-1 to some extent but the interaction was weaker than TTC9A (1–115). Since the first TPR domain lies in residues 57–90, it is plausible that the first TPR domain is required for the interaction. To confirm this postulation, more truncations of TTC9A protein were tested. Fig. ##FIG##3##4C## revealed that TTC9A (1–95), TTC9A (1–105) and TTC9A (1–110), which include the first TPR domain, showed obvious interaction with Tm5NM-1. In addition, the linker peptide (aa 91–127) between the first and the second TPR may facilitate the binding between these two proteins, as TTC9A (1–128) pulled down more Tm5NM-1 than TTC9A (1–95), TTC9A (1–105) and TTC9A (1–110) did (Fig. ##FIG##3##4C##).</p>", "<p>The observation that full-length TTC9A showed weaker interaction to Tm5NM-1 than TTC9A (1–115) and TTC9A (1–128) (Fig. ##FIG##3##4B## and ##FIG##3##4C##) suggested that the C-terminal part of TTC9A protein could have some inhibitory effect on the interaction between these two proteins (Fig. ##FIG##3##4B## and ##FIG##3##4C##). This notion is further supported by the observation that TTC9A (1–161) and TTC9A (1–197) pulled-down less Tm5NM-1 than TTC9A (1–128) did (Fig. ##FIG##3##4C##).</p>", "<p>To further verify the importance of TTC9A (1–50), the linker region and the second and the third TPR domains in the interaction with Tm5NM-1, an experiment was performed to take account of the amount of TTC9A and the truncated proteins pulled-down (Fig. ##FIG##3##4D##). It shows that TTC9A (1–50) did not interact with Tm5NM-1 but TTC9A (51–115) did, even though similar amount of TTC9A (1–50) and TTC9A (51–115) were pulled-down. This suggests that the first TPR played some role in the interaction with Tm5NM-1. Secondly, although more TTC9A (51–222) (the saturated whitish band pointed by arrow) were pulled-down than TTC9A (51–115), it did not interact with Tm5NM-1 but TTC9A (51–115) did. We speculate that either aa.1–50 is required for the full length TTC9A to interact with Tm5NM-1 or that the second and the third TPR domain adversely affected the interaction. Although there appeared to be more TTC9A (1–115) pulled-down than TTC9A, the relative amount of Tm5NM-1 pulled-down by TTC9A (1–115) appears to be more than that by TTC9A, and this lends support to the speculation that the second and the third TPR domains are inhibitory to the interaction with Tm5NM-1. This takes account of the fact that the signal for the TTC9A band is saturated (Fig. ##FIG##3##4D##, lane two in the lower panel), so the amount of TTC9A pulled-down is more than it appears to be.</p>", "<p>It is very important to verify the interaction between endogenous TTC9A and Tm5NM-1. We have tried several times for the endogenous pull-down using either TTC9A or Tm5NM-1 polyclonal antibodies. Unfortunately, we obtained very weak pull-down of high background. We suspect that the binding of polyclonal antibodies to multiple epitopes of the endogenous TTC9A or Tm5NM-1 interfered with their interaction with the target protein. Regrettably, monoclonal antibodies to TTC9A or to Tm5NM-1 are not available at this point in time.</p>", "<p>Non-muscle tropomyosins generally help to stabilize actin filament. Over- expression of tropomyosin-1 in breast cancer cells MDA-MB-231 was found to promote the assembly of stress fibers [##REF##16170368##40##]. The interaction of TTC9A with Tm5NM-1, together with the observation that in breast cancer cell line ABC28, TTC9A was up-regulated by progesterone, accompanied with a drastic increase in focal adhesion and in F-actin formation [##REF##10707953##41##], led us to hypothesize that TTC9A may be involved in cell cytoskeleton organization and cell adhesion. However, knock-down of TTC9A expression by 70 – 80% did not abolish progesterone-induced increase of F-actin (data not shown). This suggested that either TTC9A was not essential in the formation of focal adhesion and stress fibers, or that other TTC9 family proteins were able to compensate for the lost function of TTC9A.</p>", "<p>Tm5NM-1 and other tropomyosin family members are well-known for their association with the cytoskeleton system. An elevated level of Tm5NM-1 has been found in high-metastatic mouse melanoma cells and transformed rat fibroblastic cells, which suggested a function of Tm5NM-1 in inhibiting the polymerization and/or the formation of the bundles of actin microfilaments [##REF##8753779##35##,##REF##9222299##36##]. Studies have also revealed that the multiple isoforms of non-muscle tropomyosin might play a role in modulating the organization of microfilaments in cells by regulating the interaction between actin and other actin-binding proteins, such as filamin, spectrin, caldesmon, gelsolin and DNase I [##REF##711696##42##, ####REF##3830166##43##, ##REF##152647##44####152647##44##]. Thus, it is possible that TTC9A participates in the complex cytoskeleton regulation through its interaction with Tm5NM-1, or with other tropomyosins.</p>" ]

[ "<title>Results and Discussion</title>", "<title>TTC9A is over-expressed in breast cancer tissues compared with the adjacent normal tissues</title>", "<p>Previous results have shown that TTC9A is a hormonally regulated gene <italic>in vitro </italic>[##REF##16678794##2##]. As breast cancer is well-known to be a hormone-dependent malignancy, it is noteworthy to know whether TTC9A is over-expressed in breast cancer tissues and if its expression is correlated with hormone receptor status. 25 matched pairs of human breast cancer tissue and the adjacent normal tissue were analyzed for TTC9A mRNA expression. The results presented in Fig. ##FIG##0##1## were obtained by RT-PCR as illustrated in the \"materials and methods\" and the PCR products were quantitated by Southern blotting analysis. 36B4, which codes for human acidic ribosomal phosphoprotein P0, was used as a control for cDNA input (Fig. ##FIG##0##1A##). Since the expression of 36B4 varied among samples, GAPDH was also included as a normalization standard for cDNA input. It turned out that the expression levels of these housekeeping genes were always higher in tumor tissues compared with the corresponding normal tissues. This variation has been reported before. For example, GAPDH expression was 3.3-fold higher in seminoma compared with normal testis [##REF##10897012##38##]. Similarly, GAPDH transcription was significantly greater in both colonic adenomas and cancers than in normal mucosa [##REF##15823405##39##]. Nonetheless, the results revealed that the relative expression of TTC9A mRNA, when normalized to either 36B4 or GAPDH, was significantly higher in breast cancer tissue compared with its adjacent normal tissue (Fig. ##FIG##0##1##) (<italic>P </italic>&lt; 0.00001). It is also notable that nearly all the tumor tissues expressed higher level of TTC9A mRNA compared with its adjacent controls. However, we found no correlation of TTC9A expression with other clinic pathological data such as tumor size, nuclear grade, axillary's lymph node status or hormone receptor expression.</p>", "<title>Yeast-two-hybrid assay identified Tm5NM-1 as one of TTC9A interacting proteins</title>", "<p>The observation that TTC9A was over-expressed in breast cancer tissues suggests potential relevance of the protein in breast cancer biology. In an initial attempt to investigate the relationship between TTC9A function and cancer development, we searched for binding partners of the protein. To accomplish this, we performed a yeast-two-hybrid screening of a cDNA library from breast cancer cell line MCF-7, using full-length TTC9A as the bait. The coding region of the 222 aa TTC9A protein was cloned into the 'bait' vector, pLex A, which contains the DNA-binding domain. MCF-7 cDNA libraries were fused to the activation domain in the 'pray' vector p42AD. The bait vector and the cDNA libraries were transformed into yeast strain EGY48 together with a lacZ reporter gene. 30 positive clones, in which the coding regions of the library plasmids were in frame with the activation domain according to the sequencing results, were obtained after two rounds of specificity test. Among them, 5 clones contained genes coding for human tropomyosin Tm5NM-1, which is encoded by γ-tropomyosin gene of the tropomyosin family.</p>", "<title>TTC9A protein can interact with cellular expressed Tm5NM-1</title>", "<p>The binding of TTC9A to Tm5NM-1 was further examined by GST-pull down assay. In Fig. ##FIG##1##2A##, Tm5NM-1-(His)₆expression vector was transfected into COS-7 cells and GST-TTC9A was used as a \"bait\" to pull down the cellular expressed Tm5NM-1 protein. As shown in the figure, Tm5NM-1 was pulled down by GST-TTC9A, but not by the GST-tag, suggesting that GST-TTC9A did interact with cellular expressed Tm5NM-1. The specificity of the interaction was further confirmed by GST pull-down assay with different amount of bait protein. Fig. ##FIG##1##2B## showed that the protein amount of Tm5NM-1 pulled down by GST-TTC9A increased proportionally to the amount of bait protein used in the assay. However, bands other than the expected Tm5NM-1 were also observed in Fig. ##FIG##1##2B## and these non-specific bands were not observed in sample with GST protein. These unexpected bands could be due to non-specific pull-down or degraded GST-TTC9A.</p>", "<title>TTC9A interacts with Tm5NM-1 in mammalian cells</title>", "<p>The interaction between Tm5NM-1 and TTC9A was further verified by co-immunoprecipitation in mammalian cells. As shown in Fig. ##FIG##1##2C##, cellular expressed TTC9A-flag pulled down prominent amount of Tm5NM-1-(His)₆protein in COS-7 cells. To further confirm the interaction, TTC9A-flag expression vector was transfected into COS-7 cells and the interaction between endogenous Tm5NM-1 and cellular expressed TTC9A was examined by co-immunoprecipitation with anti-Tm5NM-1 antibody. Fig. ##FIG##1##2D## revealed that cellular expressed flag-TTC9A could also bind to endogenous Tm5NM-1.</p>", "<title>Identification of the regions/domains in TTC9A which are important in the interaction with Tm5NM-1</title>", "<p>To specify the domains that interact with Tm5NM-1, different truncations of TTC9A protein shown in Fig. ##FIG##2##3## were constructed and tested in COS-7 cells. As is shown in Fig. ##FIG##3##4A##, all truncations were expressed at the expected protein sizes. It is to be noted that the whitish streak in the centre of the bands for TTC9A (1–161), TTC9A (1–197), full length TTC9A and TTC9A (51–222) were due to the over-saturation of the signal. This means that the amounts of protein expressed by these constructs were not necessarily less than those by other smaller truncations which showed broader bands. The reason may be that in 12% gel, the smaller proteins tend to be more diffuse resulting in a broader band, whereas higher molecular weight proteins tend to be more compact in migration.</p>", "<p>Fig. ##FIG##3##4B## showed that TTC9A (1–50) and TTC9A (1–70) did not pull-down Tm5NM-1 visibly as compared with vector-transfected control, regardless of the very high expression level of TTC9A (1–70). TTC9A (1–100), which contains the first TPR domain, interacted with Tm5NM-1 to some extent but the interaction was weaker than TTC9A (1–115). Since the first TPR domain lies in residues 57–90, it is plausible that the first TPR domain is required for the interaction. To confirm this postulation, more truncations of TTC9A protein were tested. Fig. ##FIG##3##4C## revealed that TTC9A (1–95), TTC9A (1–105) and TTC9A (1–110), which include the first TPR domain, showed obvious interaction with Tm5NM-1. In addition, the linker peptide (aa 91–127) between the first and the second TPR may facilitate the binding between these two proteins, as TTC9A (1–128) pulled down more Tm5NM-1 than TTC9A (1–95), TTC9A (1–105) and TTC9A (1–110) did (Fig. ##FIG##3##4C##).</p>", "<p>The observation that full-length TTC9A showed weaker interaction to Tm5NM-1 than TTC9A (1–115) and TTC9A (1–128) (Fig. ##FIG##3##4B## and ##FIG##3##4C##) suggested that the C-terminal part of TTC9A protein could have some inhibitory effect on the interaction between these two proteins (Fig. ##FIG##3##4B## and ##FIG##3##4C##). This notion is further supported by the observation that TTC9A (1–161) and TTC9A (1–197) pulled-down less Tm5NM-1 than TTC9A (1–128) did (Fig. ##FIG##3##4C##).</p>", "<p>To further verify the importance of TTC9A (1–50), the linker region and the second and the third TPR domains in the interaction with Tm5NM-1, an experiment was performed to take account of the amount of TTC9A and the truncated proteins pulled-down (Fig. ##FIG##3##4D##). It shows that TTC9A (1–50) did not interact with Tm5NM-1 but TTC9A (51–115) did, even though similar amount of TTC9A (1–50) and TTC9A (51–115) were pulled-down. This suggests that the first TPR played some role in the interaction with Tm5NM-1. Secondly, although more TTC9A (51–222) (the saturated whitish band pointed by arrow) were pulled-down than TTC9A (51–115), it did not interact with Tm5NM-1 but TTC9A (51–115) did. We speculate that either aa.1–50 is required for the full length TTC9A to interact with Tm5NM-1 or that the second and the third TPR domain adversely affected the interaction. Although there appeared to be more TTC9A (1–115) pulled-down than TTC9A, the relative amount of Tm5NM-1 pulled-down by TTC9A (1–115) appears to be more than that by TTC9A, and this lends support to the speculation that the second and the third TPR domains are inhibitory to the interaction with Tm5NM-1. This takes account of the fact that the signal for the TTC9A band is saturated (Fig. ##FIG##3##4D##, lane two in the lower panel), so the amount of TTC9A pulled-down is more than it appears to be.</p>", "<p>It is very important to verify the interaction between endogenous TTC9A and Tm5NM-1. We have tried several times for the endogenous pull-down using either TTC9A or Tm5NM-1 polyclonal antibodies. Unfortunately, we obtained very weak pull-down of high background. We suspect that the binding of polyclonal antibodies to multiple epitopes of the endogenous TTC9A or Tm5NM-1 interfered with their interaction with the target protein. Regrettably, monoclonal antibodies to TTC9A or to Tm5NM-1 are not available at this point in time.</p>", "<p>Non-muscle tropomyosins generally help to stabilize actin filament. Over- expression of tropomyosin-1 in breast cancer cells MDA-MB-231 was found to promote the assembly of stress fibers [##REF##16170368##40##]. The interaction of TTC9A with Tm5NM-1, together with the observation that in breast cancer cell line ABC28, TTC9A was up-regulated by progesterone, accompanied with a drastic increase in focal adhesion and in F-actin formation [##REF##10707953##41##], led us to hypothesize that TTC9A may be involved in cell cytoskeleton organization and cell adhesion. However, knock-down of TTC9A expression by 70 – 80% did not abolish progesterone-induced increase of F-actin (data not shown). This suggested that either TTC9A was not essential in the formation of focal adhesion and stress fibers, or that other TTC9 family proteins were able to compensate for the lost function of TTC9A.</p>", "<p>Tm5NM-1 and other tropomyosin family members are well-known for their association with the cytoskeleton system. An elevated level of Tm5NM-1 has been found in high-metastatic mouse melanoma cells and transformed rat fibroblastic cells, which suggested a function of Tm5NM-1 in inhibiting the polymerization and/or the formation of the bundles of actin microfilaments [##REF##8753779##35##,##REF##9222299##36##]. Studies have also revealed that the multiple isoforms of non-muscle tropomyosin might play a role in modulating the organization of microfilaments in cells by regulating the interaction between actin and other actin-binding proteins, such as filamin, spectrin, caldesmon, gelsolin and DNase I [##REF##711696##42##, ####REF##3830166##43##, ##REF##152647##44####152647##44##]. Thus, it is possible that TTC9A participates in the complex cytoskeleton regulation through its interaction with Tm5NM-1, or with other tropomyosins.</p>" ]

[ "<title>Conclusion</title>", "<p>In summary, this study revealed that TTC9A was over-expressed in breast cancer tissues compared with the adjacent normal tissues, suggesting that TTC9A might be an important gene involved in the breast cancer development process. We have identified Tm5NM-1, a tropomyosin family protein, as one of the TTC9A-interacting proteins. The results also suggest that the first 50 aa of TTC9A was required for the interaction with Tm5NM-1, although the segment alone did not bind to Tm5NM-1. Furthermore, the first TPR domain and the linker segment between the first two TPR domains may play an important role for the binding of TTC9A to Tm5NM-1, while the last two TPR motifs may be inhibitory on the interaction. The exact function of TTC9A remains unknown at current stage. The interaction with Tm5NM-1 suggests that TTC9A might act as a chaperone protein in the organization of cell cytoskeleton. Currently, TTC9A gene knockout study in mice is underway to define the physiological role of the gene <italic>in vivo</italic>.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Tetratricopeptide repeat domain 9A (TTC9A) protein is a recently identified protein which contains three tetratricopeptide repeats (TPRs) on its C-terminus. In our previous studies, we have shown that TTC9A was a hormonally-regulated gene in breast cancer cells. In this study, we found that TTC9A was over-expressed in breast cancer tissues compared with the adjacent controls (P &lt; 0.00001), suggesting it might be involved in the breast cancer development process. The aim of the current study was to further elucidate the function of TTC9A.</p>", "<title>Methods</title>", "<p>Breast samples from 25 patients including the malignant breast tissues and the adjacent normal tissues were processed for Southern blot analysis. Yeast-two-hybrid assay, GST pull-down assay and co-immunoprecipitation were used to identify and verify the interaction between TTC9A and other proteins.</p>", "<title>Results</title>", "<p>Tropomyosin Tm5NM-1 was identified as one of the TTC9A partner proteins. The interaction between TTC9A and Tm5NM-1 was further confirmed by GST pull-down assay and co-immunoprecipitation in mammalian cells. TTC9A domains required for the interaction were also characterized in this study. The results suggested that the first TPR domain and the linker fragment between the first two TPR domains of TTC9A were important for the interaction with Tm5NM-1 and the second and the third TPR might play an inhibitory role.</p>", "<title>Conclusion</title>", "<p>Since the primary function of tropomyosin is to stabilize actin filament, its interaction with TTC9A may play a role in cell shape and motility. In our previous results, we have found that progesterone-induced TTC9A expression was associated with increased cell motility and cell spreading. We speculate that TTC9A acts as a chaperone protein to facilitate the function of tropomyosins in stabilizing microfilament and it may play a role in cancer cell invasion and metastasis.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>SLC performed most of the experiments, analyzed and interpreted data and participated in the manuscript writing. GHH provided the tissue samples and conducted clinical analysis of the data. VC–LL designed the experiments, acquired financial support and participated in manuscript writing. All authors have contributed to this work, read and approved the final manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2407/8/231/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>We would like to thank Dr. Takahiro Nagase, Kazusa DNA Research Institute Foundation, Japan, for providing the TTC9A cDNA.</p>", "<p>This work was supported by the A*STAR Biomedical Research Council, Republic of Singapore (06/1/221/19/455) to Valerie C-L. Lin.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>The expression level of TTC9A mRNA was significantly higher in breast cancer tissues than that in the adjacent normal breast tissues.</bold> Total RNA was extracted from human breast cancer tissues and the matched adjacent normal breast tissues. Equal amount of RNA from each sample was subjected to reverse transcription and cDNA produced was amplified by PCR using TTC9A, 36B4 or GAPDH primers. 10 μl PCR products were separated on an agarose gel and analyzed by Southern blotting. Band intensity was analyzed by Bio-Rad Molecular Image Analyzer. The figure shows the expression levels of TTC9A in 25 pairs of normal and tumor tissue samples after normalizing to those of 36B4 (A) or GAPDH (B). Each pair of bars represents samples from one patient. The primers used for TTC9A were 5'-CACATGTCTATAACGATTT CC-3' (forward) and 5'-TGCAGGAAACAGGGG ACTCTC-3' (reverse). The primers used to amplify 36B4 gene were 5'-GATTGGCTACCCAACTGTTGCA-3' (forward) and 5'-CAGGGGCAGCAGCCACAAAGGC-3' (reverse). The primers for GAPDH were 5'-TGCACCACCAACTGCTTAG-3' (forward) and 5'-GAGGCAGGGATGATG TTC-3' (reverse).</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>TTC9A binds to Tm5NM-1.</bold> (A) GST-TTC9A binds to Tm5NM-1-(His)₆. COS-7 cells were transfected with Tm5NM-1-(His)₆expression vector or control vector and total cell lysates were collected at 48 h post-transfection. 60 μg GST-TTC9A protein was immobilized onto Glutathione Sepharose 4B gel (Amersham Biosciences) and 300 μg total cell lysates were used for Tm5NM-1-(His)₆pull-down. The proteins bound to the beads were eluted with 2 × SDS-PAGE sample buffer and were separated on an SDS-PAGE gel. Tm5NM-1-(His)₆was detected using anti-His antibody (Amersham Biosciences). GST protein expressed by empty pGEX-5X-3 vector was included as a negative control. 15 μg total cell lysates (5% of input) were loaded in the first lane to indicate the position of Tm5NM-1-(His)₆band. (B) Tm5NM-1-(His)₆pull-down by GST-TTC9A is concentration-dependent. GST-pull down assay was carried out with 2 μg or 20 μg GST-TTC9A as bait protein. The amount of Tm5NM-1-(His)₆pulled down was proportional to the amount of bait protein used. (C) TTC9A-flag interacted with Tm5NM-1-(His)₆. Expression vectors for TTC9A-flag and Tm5NM-1-(His)₆were co-transfected into COS-7 cells. Co-immunoprecipitation was carried out with anti-flag agarose beads (Sigma-Aldrich) and Tm5NM-1 was detected by anti-His antibody (Amersham Biosciences). Upper panel: Tm5NM-1-(His)₆was expressed at similar level in control vector and TTC9A-flag transfected COS-7 cells; lower panel: Tm5NM-1-(His)₆was pulled down by TTC9A-flag. (D) TTC9A-flag interacted with endogenous Tm5NM-1-(His)₆. Expression vector for TTC9A-flag was transfected into COS-7 cells. Co-immunoprecipitation was carried out with anti-Tm5NM-1/2 (Chemicon) antibody and TTC9A was detected by anti-flag antibody (Sigma-Aldrich). Co-immunoprecipitation with goat pre-immune serum was included as a negative control.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p>Truncation constructs of TTC9A protein and their relative binding to Tm5NM-1 based on co-immunoprecipitation experiment.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p><bold>Domains involved in TTC9A and Tm5NM-1 interaction.</bold> (A) Western blot analysis of the expression TTC9A truncates in the cell lysates using anti-flag antibody from Sigma-Aldrich. All truncations expressed proteins of predicted sizes. The hollow streak in some of the bands indicates over-saturation of the signal. (B and C) Test of Tm5NM-1-(His)₆pull-down by different TTC9A truncations in pXL-Flag vectors. The TTC9 truncation constructs were transfected into COS-7 cells together with Tm5NM-1-(His)₆expression vector. The interaction between flag-TTC9A truncates and Tm5NM-1-(His)₆was analyzed by co-immunoprecipitation with anti-flag agarose beads (Sigma-Aldrich). The upper panels in B and C are Western blotting analysis of Tm5NM-1 expression in the cell lysates using anti-His antibody (Amersham Biosciences), and the lower panels are the Tm5NM-1-(His)₆co-immunoprecipitated with flag-TTC9A and the truncated proteins. (D) The upper panel is the analysis of Tm5NM-1 expression in the cell lysates; the middle panel is the co-immunoprecipitated Tm5NM-1 with flag-TTC9A and the truncated proteins; the lower panel represents the amount of TTC9A and its truncates pulled down by the anti-flag agarose beads (Sigma-Aldrich) in the co-immunoprecipitation assay.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1471-2407-8-231-1\"/>", "<graphic xlink:href=\"1471-2407-8-231-2\"/>", "<graphic xlink:href=\"1471-2407-8-231-3\"/>", "<graphic xlink:href=\"1471-2407-8-231-4\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Initially isolated from macrophages as a potent interferon-γ (IFN-γ)-inducing factor ##REF##7477296##[1]##, ##REF##8666798##[2]##, Interleukin-18 (IL-18) is a multifunctional cytokine affecting both innate and adaptive immune responses ##REF##11244043##[3]##, ##REF##12554798##[4]##. IL-18 is produced as an inactive precursor (pro-IL-18), which is activated by proteolytic cleavage, primarily by caspase-1 ##REF##8999548##[5]##, ##REF##9121587##[6]##. Pro-IL-18 is produced by multiple cell types, including macrophages ##REF##7477296##[1]##, dendritic cells ##REF##9808192##[7]##, vascular endothelial cells ##REF##11805151##[8]##, and intestinal epithelial cells ##REF##9232828##[9]##.</p>", "<p>Numerous effector functions have been attributed to IL-18. Perhaps the best characterized of these is IL-18's context-dependent induction of Th1 or Th2 CD4 T cell polarization ##REF##11244043##[3]##. In the presence of IL-12, IL-18 drives the Th1 polarization of activated CD4⁺ T cells. However, in the absence of IL-12 (or in the presence of IL-2 or IL-4), IL-18 promotes IgE expression and Th2 differentiation. More recently, IL-18 has been demonstrated, in synergy with IL-23, to drive TH17 cell polarization ##REF##16670261##[10]##, ##REF##17404271##[11]##. IL-18 also exerts multiple effects on NK cells, including increased cytoxicity ##REF##8892629##[12]##, and in synergy with IL-2, promotion of NK cell expansion and IFNγ production ##REF##11221875##[13]##. Given its widespread immune-augmenting functions, it is not surprising that IL-18 has been evaluated in numerous animal models of cancer. For example, IL-18 enhances the efficacy of DNA vaccines directed against prostate-specific antigen ##REF##16135392##[14]## or Fos-related antigen ##REF##15833877##[15]##. Additionally, IL-18 augments the efficacy of DC-based vaccines ##REF##12384548##[16]##, ##REF##12378411##[17]## as well as whole-cell tumor vaccines ##REF##17495947##[18]##. However, IL-18 has also been reported to enhance tumor progression (reviewed in ##REF##17001512##[19]##). For example, tumor-produced IL-18 induces Fas ligand expression in melanoma cells, possibly resulting in escape from NK cell-mediated immune surveillance ##REF##10825144##[20]##. Furthermore, IL-18 has been shown to increase the invasiveness of myeloid leukemia lines ##REF##14630085##[21]##.</p>", "<p>IL-18 has also been implicated in multiple autoimmune-associated pathologies (reviewed in ##REF##17353157##[22]##). For example, high levels of IL-18 are found in the synovial fluid of rheumatoid arthritis patients ##REF##10562301##[23]##, and alleviation of rheumatoid arthritis symptoms is associated with a decrease in IL-18 levels ##REF##12117680##[24]##. Paradoxically, given its' proinflammatory properties, IL-18 is well tolerated and safe in humans ##REF##16857801##[25]##. In contrast, IL-12 is toxic at doses three orders of magnitude lower ##REF##9326219##[26]##.</p>", "<p>The rationale for the present study was to determine if IL-18 might present a less toxic alternative to IL-12 as an adjunct for cancer adoptive transfer immunotherapy. In the study described herein, we demonstrate that IL-18 administration resulted in increased engraftment of CD8⁺ human T cells. Concurrently, IL-18 administration resulted in a decrease in human CD4⁺CD25⁺FoxP3⁺ regulatory T cells (Tregs). Furthermore, we find that IL-18 augmented xenogeneic GVHD, and overrode the suppressive effect of Tregs <italic>in vivo</italic>. Our findings indicate that by simultaneously affecting both CD8⁺ T cells and Tregs, IL-18 may alter the set point of an immune response, underscoring the potential utility of IL-18 as an adjuvant in cancer therapy.</p>" ]

[ "<title>Methods</title>", "<title>Mice</title>", "<p>All animal experiments were approved by the University of Pennsylvania Institutional Animal Care and Use Committee. NOD/<italic>scid</italic>/β2microglobulin<italic>^null</italic> (NOD/<italic>scid</italic>/β2<italic>m^null</italic>) ##REF##9103418##[27]## and NOD/<italic>scid</italic>/IL-2rγ<italic>^null</italic> (NOG) ##REF##15879151##[28]## mice were purchased from the Jackson Laboratory. Animals were bred in the Animal Services Unit of the University of Pennsylvania. The mice were housed under specific pathogen-free conditions in microisolator cages and given unrestricted access to autoclaved food and acidified water. Animals of both sexes were used for experiments at 6–9 weeks of age.</p>", "<title>Cells and Animal Injections</title>", "<p>Peripheral blood mononuclear cells (PBMCs) were obtained by leukapheresis of healthy volunteer donors by the University of Pennsylvania Human Immunology Core. All specimens were collected under a University Institutional Review Board-approved protocol, and informed consent was obtained from each donor. Cells were injected intraperitoneally into non-irradiated host animals as described in the individual experiments.</p>", "<p>Recombinant human IL-18 (SB-485232) and pegylated human IL-18 (GSK-189720) were prepared as previously described at GlaxoSmithKline ##REF##12597888##[29]##. Subcutaneous injections of 15 µg/animal were performed daily for the recombinant product and twice weekly for the pegylated product, taking into consideration the increased persistence of the pegylated IL-18.</p>", "<title>Animal Necropsy</title>", "<p>Animals were sacrificed by CO₂ asphyxiation. Peripheral blood was obtained by cardiac puncture, and the spleen, liver, lung, and gut were isolated and portions were fixed in 4% paraformaldehyde. Liver leukocytes were separated by Ficoll density gradient centrifugation and splenocytes were prepared by mechanical disassociation. Cells were retrieved from the peritoneal cavity by PBS lavage.</p>", "<title>Immunohistochemistry</title>", "<p>Mouse tissues were fixed in 4% paraformaldehyde and embedded in paraffin. 5–6 µm sections were cut for immunohistochemical staining. Following deparaffinization and high temperature antigen unmasking procedures, sections were incubated with murine monoclonal antibodies to human CD25 (Clone 4C9 [Vector], 1∶100), or Foxp3 (Clone236A/E7 [AbCam], 1∶40). Sections were then incubated with biotinylated secondary antibody (goat anti-mouse IgG [Vector]) and signal was localized using 3,3′-diaminobenzidine tetrahydrochloride (DAB, [Vector]) as the chromogen. Hematoxylin was used for counterstaining. Positively stained cells were quantified with Metamorph software (Universal Imaging Corporation) using the Manual Count option. Quantification was performed in a blinded manner. For each organ, at least 8 fields per animal were tabulated.</p>", "<title>Flow cytometry</title>", "<p>After euthanasia, peripheral blood, peritoneal cavity washings, spleen, and liver were obtained from euthanized animals. Single cell suspensions were stained with conjugated anti-human lymphocyte surface markers (CD45, clones H130 or 2D1, CD4, clone SK3, CD8, clone SK1 or 3B5, and CD25, clone M-A251, all from BD Pharmingen) and Foxp3 (Clone 206D, Biolegend). Cell populations were analyzed using either LSRII or FACSCalibur flow cytometers (Becton Dickinson) and Flowjo software (Tree Star). Human lymphocytes were initially identified by gating on the human CD45⁺ population within the live gate, and human lymphocyte subsets were further defined within this population. The absolute number of human cells was determined using TruCount tubes (BD Biosciences) or a Multisizer 3 (Beckman Coulter).</p>", "<title>Adoptive Transfer of Ex Vivo Expanded Tregs</title>", "<p>Human CD4⁺CD25^high cells were isolated from PBMCs using magnetic beads (Miltenyi Biotech) as described ##UREF##0##[30]##. Gamma-irradiated (100 Gy) K562-based antigen presenting cells modified to express CD64, CD86 and OX40L ##REF##17375070##[31]## were used to stimulate purified populations of CD4⁺CD25^high cells, which were expanded ex vivo in the presence of 300 U/ml IL-2 (Chiron) and 100 ng/ml Rapamycin (Calbiochem) ##UREF##0##[30]##. After one growth cycle (approximately 14–21 days), the cells were harvested and used for injection as described in the text.</p>", "<title>\n<italic>In Vitro</italic> Suppression Assay</title>", "<p>Following harvest of expanded T cells (Tregs and control CD4 cells), varying numbers of Tregs were plated in 100 µl of X-VIVO 15+10% human AB serum in round bottom 96 well plates (Corning Incorporated, Corning, NY). Frozen autologous PBMCs were thawed, CFSE-labeled, and resuspended in culture medium at 1×10⁷/ml. Anti-CD3 beads (Invitrogen) were added at a ratio of 1 bead per cell. 100 µl of PBMC cell suspension (1×10⁶ cells) was added to individual wells in a 96 well plate. CFSE-labeled PBMCs without CD3 beads (and lacking Tregs) were used as negative controls; CFSE-labeled PBMC stimulated with anti-CD3 beads (no Tregs) were used as positive controls. Where indicated, the suppression assay was performed in the presence of 500 ng/ml recombinant human IL-18. The cultures were harvested four days later and stained with APC-conjugated anti-CD8 (BD Pharmingen). Data were acquired on a FACSCalibur flow cytometer using Cell Quest Pro software and analyzed using FlowJo software. Quantitative analysis of Treg suppression capacity was performed by gating on CD8⁺ CFSE⁺ cells ##UREF##0##[30]##.</p>", "<title>Statistical analysis</title>", "<p>All results were expressed as means±standard deviation. Unless otherwise indicated, the statistical significance of differences between groups was evaluated by two-sample t tests assuming equal variance or the Mann-Whitney rank sum test. Survival data were analyzed by lifetable methods using log rank analysis performed using SysStat (Systat Software, Inc.).</p>" ]

[ "<title>Results</title>", "<p>In preliminary experiments we evaluated recombinant human IL-18 at doses from 1 to 100 µg/mouse/day for anti-tumor effects in xenografted mice (data not shown), and 15 µg was chosen as a daily dose based on tolerability and efficacy, and because this dose is less than the equivalent maximum tolerated dose that has been established in monkeys and humans ##REF##16857801##[25]##, ##REF##12441145##[32]##. The effects of IL-18 were tested on human T cell subsets using two humanized mouse models ##REF##9103418##[27]##, ##REF##15879151##[28]##. Initially, we determined if IL-18 directly affected the engraftment of human T cells in NOD/<italic>scid</italic>/β2<italic>m^null</italic> mice. These mice, in addition to lacking mature B and T cells, have reduced levels of NK cells compared to NOD/<italic>scid</italic> mice ##REF##9103418##[27]##. When NOD/<italic>scid</italic>/β2<italic>m^null</italic> mice were injected intraperitoneally with PBMC, human CD45 positive cells were recovered from peripheral blood, liver and spleen. Daily subcutaneous injections of IL-18 caused a substantial increase in the number of human leukocytes recovered from the blood, liver and spleens of the mice (##FIG##0##\nFig. 1A\n##). However, the leukocyte promoting effect of IL-18 was entirely attributable to increased engraftment of CD8⁺ T cells, as the number of CD4⁺ T cells was not increased by IL-18 (##FIG##0##\nFig. 1B and 1C\n##). No NK cells or B cells were observed at this time point (data not shown).</p>", "<p>As is shown below, we observed that the IL-18 treated mice developed xenogeneic GVHD more rapidly than mice engrafted with PBMC only. We first considered the hypothesis that the IL-18-mediated increase in human CD8⁺ cells could account for this. However, an alternative hypothesis was also considered, that decreased engraftment with regulatory T cells could also account for this, because others have shown that regulatory T cell numbers and function can influence the onset and severity of xenogeneic GVHD ##REF##16394018##[33]##, ##REF##17000688##[34]##. Although IL-18 did not influence the total number of CD4 cells (##FIG##0##\nFig. 1B\n##), there were significant decreases in regulatory T cells in the IL-18 treated mice, as judged by CD25 and FoxP3 expression (##FIG##1##\nFig. 2\n##). The percentage and absolute number of CD4⁺CD25⁺ cells was decreased in all organs and tissues examined, i.e. blood, spleen, liver, and peritoneal cavity (##FIG##1##\nFig. 2A and 2B\n##).</p>", "<p>It was possible that the distinct effects of IL-18 on human lymphocyte subsets were due to relative efficiencies of engraftment for effector and regulatory subsets in the NOD/<italic>scid</italic>/β2<italic>m^null</italic> mice. To test this possibility, PBMCs were transferred into NOD/<italic>scid</italic>/IL-2rγ<italic>^null</italic> (NOG) mice. This strain is highly permissive to human hematopoietic stem cell engraftment ##REF##15879151##[28]##. In NOG mice injected with PMBC, the mice treated with IL-18 had consistently elevated numbers of human CD45⁺ cells in the peripheral blood, and this increase was entirely accounted for by increased numbers of CD8⁺ cells (##FIG##2##\nFig. 3A\n##) as was the case for NOD/<italic>scid</italic>/β2M<italic>^null</italic> mice. The NOG mice also had reciprocal decreases in CD4⁺CD25⁺Foxp3⁺ cells in the blood (##FIG##2##\nFig. 3B\n##) and all tissues examined (data not shown). The IL-18 mediated decrease in CD4⁺CD25⁺FoxP3⁺ cells was not due to decreased recovery of the cells, as tissue sections examined by immunohistochemistry also documented consistent decreases in cells that stained for Foxp3 (data not shown). Furthermore, as shown in ##FIG##2##\nFig. 3C\n##, the CD8:Treg ratios in the peripheral blood of individual animals were markedly increased by IL-18 treatment.</p>", "<p>To assess the effects of IL-18 treatment on human T cell differentiation in NOG mice after IL-18 treatment, peripheral blood and spleen were examined by flow cytometry (##FIG##3##\nFig. 4\n##). There were only minor effects of IL-18 on the fraction of T cells with a central memory (CD27⁺CD28⁺) phenotype, and the modest effects were different in blood and spleen. There was only a slight effect of IL-18 on the appearance of T cells with a senescent or effector memory phenotype (CD28⁺CD57⁺ or CD27⁺CD57⁺).</p>", "<p>As was mentioned above, we observed that both the NOD/<italic>scid</italic>/β2M<italic>^null</italic> and NOG mouse strains had earlier onset and more severe xenogeneic GVHD after PBMC engraftment and human IL-18 treatment (##FIG##4##\nFig. 5\n##). Clinical appearance and examination of weight curves revealed that IL-18 treated mice had more significant manifestations of GVHD as indicated by lethargy, hunched posture, generalized erythema, and a progressive reduction in mean weights from pretransplant levels (data not shown). Pathologic examination revealed that the IL-18 treated mice had more severe tissue infiltration and inflammation of the lung, liver and gastrointestinal tracts (##SUPPL##0##\nSupplemental Fig. S1\n##\n<bold> and </bold>\n##SUPPL##1##\nS2\n##).</p>", "<p>To address the mechanisms for these observations, we used an adoptive transfer model of effector and regulatory T cells that we and others have developed ##REF##15090447##[35]##–##REF##11986199##[38]##. PBMCs were subjected to immunoaffinity bead separation to obtain CD4⁺25^high cells and PBMC depleted of CD4⁺25^high cells. The CD4⁺25^high cells at baseline were &gt;70% Foxp3⁺, and these cells were expanded using K562 cells expressing CD64, CD86, and OX40L in the presence of IL-2 and rapamycin as described ##REF##15031211##[37]##. NOG mice were injected with PBMC plus enriched CD4⁺CD25⁻ effector cells, or with PBMC plus <italic>ex vivo</italic>-expanded CD4⁺25⁺ Tregs. IL-18 treatment <italic>in vivo</italic> accelerated GVHD lethality in mice engrafted with PBMC plus enriched CD4⁺CD25⁻ effector cells, or with PBMC plus <italic>ex vivo</italic> expanded CD4⁺25⁺ cells (##FIG##5##\nFig. 6A and B\n##). In contrast, most mice engrafted with PBMC and <italic>ex vivo</italic> expanded Tregs had long term survival. The mice were monitored for engraftment of CD8⁺ T cells by periodic measurements of absolute CD8⁺ T cell numbers in peripheral blood using the Trucount assay (##FIG##5##\nFig. 6C–E\n##). Mice engrafted with PBMC and CD4⁺ effector cells had a progressive increase in the number of circulating CD8⁺ T cells. Previous studies have shown that human Tregs prevent the expansion of the presumably xenoreactive CD8⁺ T cells in mice ##REF##17000688##[34]##. We found that <italic>ex vivo</italic> expanded Tregs also prevented the expansion of CD8⁺ T cells in the mice. However, IL-18 treatment resulted in a striking increase in the numbers of CD8⁺ T cells in mice engrafted with either PBMCs and CD4⁺ effectors or with PBMCs and Tregs. In fact, all IL-18-treated animals were sacrificed early due to acute GVHD symptoms (##FIG##5##\nFig. 6E\n## and data not shown). Thus, in the presence of IL-18, Tregs failed to control CD8+ T cell proliferation. However, the Kaplan-Meier curves show a significant difference between each of the 4 groups, indicating that Tregs have an effect even in the face of IL18 treatment.</p>", "<p>Because IL-18 was able to override some of the effects of the Tregs as indicated by the findings of increased CD8⁺ T cells in mice given adoptively transferred Tregs, we next evaluated the effects of IL-18 on Treg function <italic>in vitro</italic>. We found that the addition of IL-18 to culture medium did not affect the population doubling rate of Tregs <italic>in vitro</italic> after anti-CD3/CD28 stimulation, in the presence or absence of IL-2 (data not shown). Furthermore, IL-18 did not affect the suppressive function of <italic>ex vivo</italic> expanded Tregs (##FIG##6##\nFig. 7\n##) or the expression of Foxp3 (data not shown). Finally, the upregulation of IL-18Rα did not differ between CD4⁺25⁻ or CD4⁺25⁺ T cells in culture after activation using a variety of stimulation conditions. However, it is notable that the expression of IL-18Rα was not sustained on the regulatory T cells after day 3 in culture, in contrast to bulk CD4 and CD8 T cell subsets (##FIG##7##\nFig. 8\n##).</p>" ]

[ "<title>Discussion</title>", "<p>This study demonstrates that in immune-deficient mice engrafted with human PBMCs, human IL-18 administration exerts opposite effects upon two human cell populations: CD8⁺ T cell numbers are increased, while CD4⁺CD25⁺FoxP3⁺ Treg cell numbers are markedly decreased. The altered ratio of effector to Treg engraftment was functionally important as judged by the accelerated onset and severity of xenogeneic GVHD in the recipient mice. These effects were unexpected because we are not aware of studies with mouse IL-18 that have revealed differential effects of IL-18 on regulatory and effector T cells ##REF##9893178##[39]##, ##REF##16507221##[40]##.</p>", "<p>Our studies revealed several potential mechanisms for the effects of IL-18. IL-18 can cause the expansion of mouse NK cells ##REF##11221875##[13]##, but has not been reported to directly mediate the expansion of T cells. Consistent with this, we were unable to demonstrate a direct effect of IL-18 on the expansion of human effector or regulatory T cells <italic>in vitro</italic>. However, we found that the expression of the IL-18Rα subunit was more prolonged on effector T cells than Tregs, providing a potential mechanism to explain the reciprocal effects that we observed <italic>in vivo</italic> on the effects of IL-18 on effector T cells and Tregs. There may be important species specific differences in the biology of IL-18 on human and mouse T cells ##REF##16458015##[41]##. In the mouse, IL-18 has a protective effect on CD4(+)-mediated acute GVHD ##REF##18022572##[42]##, and blockade of IL-18 accelerates acute GVHD-related mortality in mice ##REF##11714750##[43]##. Recent studies in the mouse indicate that IL18 has a role in promoting an IFN-γ dependent positive regulation of memory CD8 T cell proliferation ##REF##18545704##[44]##. Recent studies with human cells indicate that the interaction of IL-18 treated NK cells and DCs induces maturation of DCs that subsequently promote Th1 and IFN-γ secretion ##UREF##1##[45]##. Together, the above studies are consistent with our findings, and suggest a mechanism that IL-18 promotes engraftment of CD8 cells in the mice at the expense of regulatory T cells, leading to increased memory CD8 cells and increased xenogeneic GVHD.</p>", "<p>A complete mechanistic understanding of the effects mediated by IL-18 is not yet available from the present findings. It is still not clear which T cell subsets are preferentially expanded during IL-18 therapy in the mice, although we found some selective effects on subsets of CD8 T cells in ##FIG##3##Figure 4##. We do not yet know if CD8 T cells need to be activated in order to be regulated by IL-18. Similarly, we do not yet know if CD4 T cells need to be present for the IL-18 effects on CD8 cells. Our data in ##FIG##5##Figure 6## suggest that the effects on CD8 T cells and Tregs are independent.</p>", "<p>IL-18 has been proposed to have a role in a number of inflammatory and autoimmune disorders in mice and man ##REF##17336692##[46]##. In mice, IL-18 may have a role in lupus nephritis ##REF##16214093##[47]##. In humans, IL-18 over-expression has been reported in rheumatoid arthritis, sarcoidosis, adult-onset Still's disease, vasculitis, lupus, urticaria and histiocytosis ##REF##10562301##[23]##, ##REF##17502360##[48]##–##REF##10444185##[54]##. Tregs have been reported to be decreased or to have decreased functional activity in a number of autoimmune disorders ##REF##15067033##[55]##, and it is possible that an imbalance of Tregs and effector T cells due to differential IL-18 signaling contributes to the loss of tolerance.</p>", "<p>Depletion of Tregs in tumor bearing mice enhances the response to immunotherapy ##REF##11560997##[56]##. A high ratio of effector CD8 T cells to Tregs in the tumor microenvironment has been shown to be a favorable prognostic feature in patients with ovarian cancer ##REF##12529460##[57]##, ##REF##16344461##[58]##. An increase in the ratio of CD8⁺ T effector to regulatory cells in syngeneic mouse tumor models and in humans with cancer correlates with responses to immunotherapies ##REF##18287062##[59]##, ##REF##16778987##[60]##. Thus, the ability of IL-18 to increase the ratio of effector CD8⁺ T cells to Tregs may have important implications for therapeutic vaccines and adoptive transfer strategies ##REF##17476350##[61]##.</p>", "<p>The ability of IL-18 to increase the ratio of effector CD8⁺ T cells to Tregs may have important implications for therapeutic vaccines and cancer therapy. Depletion of Tregs in tumor bearing mice enhances the response to immunotherapy ##REF##11560997##[56]##. A high ratio of effector CD8 T cells to Tregs in the tumor microenvironment has been shown to be a favorable prognostic feature in patients with ovarian cancer ##REF##12529460##[57]##, ##REF##16344461##[58]##. Furthermore, an increase in the ratio of CD8⁺ T effector to regulatory cells in syngeneic mouse tumor models and in humans with cancer correlates with responses to immunotherapies ##REF##18287062##[59]##, ##REF##16778987##[60]##.</p>", "<p>Our findings may have important implications for cancer therapy due to the immunosuppressive tumor microenvironment. Cancer patients have increased tumor infiltrating Tregs ##REF##11406550##[62]##, and in many cases, CD8⁺ CTLs from the tumor environment are dysfunctional ##REF##16972901##[63]##, ##REF##15776005##[64]##. Recently, it has become more widely appreciated that successful immunotherapy requires neutralization of the immunosuppressive features of the tumor microenvironment, in particular Tregs ##REF##17476346##[65]##. Thus, by inhibiting Tregs and promoting conventional T cells, human IL-18 may have potential to restore immunocompetence in cancer patients, and thereby augment the effects of cytotoxic and biologic therapies, in order to provoke and maintain an effective anti-tumor immune response. Finally, IL-18 may have potential to augment the anti-tumor effects of donor leukocyte infusions by promoting effector T cells at the expense of regulatory T cell expansion.</p>" ]

[]

[ "<p>Conceived and designed the experiments: RGC SMA GC JLR CHJ. Performed the experiments: RGC CC XS GDD RL TG SJ. Analyzed the data: RGC CC XS GDD TG JLR ZLJ SJ. Contributed reagents/materials/analysis tools: ZLJ. Wrote the paper: RGC CHJ.</p>", "<p>IL-18 has pleotropic effects on the activation of T cells during antigen presentation. We investigated the effects of human IL-18 on the engraftment and function of human T cell subsets in xenograft mouse models. IL-18 enhanced the engraftment of human CD8⁺ effector T cells and promoted the development of xenogeneic graft versus host disease (GVHD). In marked contrast, IL-18 had reciprocal effects on the engraftment of CD4⁺CD25⁺Foxp3⁺ regulatory T cells (Tregs) in the xenografted mice. Adoptive transfer experiments indicated that IL-18 prevented the suppressive effects of Tregs on the development of xenogeneic GVHD. The IL-18 results were robust as they were observed in two different mouse strains. In addition, the effects of IL-18 were systemic as IL-18 promoted engraftment and persistence of human effector T cells and decreased Tregs in peripheral blood, peritoneal cavity, spleen and liver. In vitro experiments indicated that the expression of the IL-18Rα was induced on both CD4 and CD8 effector T cells and Tregs, and that the duration of expression was less sustained on Tregs. These preclinical data suggest that human IL-18 may have use as an adjuvant for immune reconstitution after cytotoxic therapies, and to augment adoptive immunotherapy, donor leukocyte infusions, and vaccine strategies.</p>" ]

[ "<title>Supporting Information</title>" ]

[ "<p>We thank Ella Ofori, Laura Pell and Michelle Kanther for expert technical assistance.</p>" ]

[ "<fig id=\"pone-0003289-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003289.g001</object-id><label>Figure 1</label><caption><title>Human IL-18 Promotes Systemic Increases in Human CD8 T Cell Numbers in NOD/<italic>scid/</italic>β2<italic>m</italic>\n^null mice injected with human PBMCs.</title><p>NOD/<italic>scid</italic>/β2<italic>m^null</italic> mice were injected intraperitoneally with 5×10⁷ human PBMCs and 24 hours later, a three week course of daily subcutaneous injections of 15 µg recombinant human IL-18 was initiated. At the end of the injection time course, the animals were sacrificed, peripheral blood was collected, and liver and spleen homogenates were prepared. Human cell populations were detected by flow cytometry. Absolute cell numbers in the peripheral blood were determined using Tru-Count tubes; absolute cell numbers from spleen and liver were determined using a Coulter Multisizer 3. The mean±s.d. is shown for (A) human CD45⁺ cells, (B) human CD4⁺ cells, and (C) human CD8⁺ cells. Open bars represent values from mock-treated animals; filled bars represent values from IL-18 treated animals. These data are a representative single experiment of seven different experiments. Asterisks indicate significant differences between mock- and IL-18-treated animals (n = 6 to 10 mice/group; p&lt;0.05).</p></caption></fig>", "<fig id=\"pone-0003289-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003289.g002</object-id><label>Figure 2</label><caption><title>Human IL-18 Mediates a Decrease in Tregs in NOD/<italic>scid/</italic>β2<italic>m^null</italic> mice Injected with Human PBMCs.</title><p>5×10⁷ PBMCs were transferred into NOD/<italic>scid</italic> /β2M<italic>^null</italic> mice. The day following cell injection, a three week course of subcutaneous huIL-18 injections was initiated. Animals in the experiment depicted in panel A received twice-weekly doses of 15 µg pegylated human IL-18, while the animals in the experiments depicted in panel B received daily doses of 15 µg recombinant human IL-18. (A) At the end of IL-18 treatment, cells isolated from the peripheral blood, peritoneal cavity, spleen and liver were stained with antibodies to human CD45, CD4, and CD25 and analyzed by flow cytometry. The mean percentage of CD⁺CD25⁺ cells from each tissue is depicted. Filled bars represent values obtained from IL-18-treated animals; open bars represent values obtained from mock-treated animals. (B) Paraffin sections from the spleens of mock- or IL-18-treated animals were reacted with antibodies specific for human CD25 or Foxp3 . Antibody-reactive cells per 400X field were enumerated blindly, and the mean±s.d. depicted. Solid bars depict sections from IL-18-treated animals, while open bars represent sections from mock-treated animals. The experiment in panel A was performed 2 times, and the experiment in panel B was performed 6 times. Asterisks indicate significant differences between mock- and IL-18-treated animals (n = 8/group; p&lt;0.05).</p></caption></fig>", "<fig id=\"pone-0003289-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003289.g003</object-id><label>Figure 3</label><caption><title>The Human IL-18-Mediated Decrease in Tregs is Not Restricted to NOD/<italic>scid/</italic>β2<italic>m^null</italic> Mice.</title><p>NOD/<italic>scid</italic>/IL-2rγ<italic>^null</italic> (NOG) mice were injected with 5×10⁶ human PBMCs, and 24 hours later, a three week course of daily subcutaneous injections of 15 µg recombinant human IL-18 was initiated. The absolute numbers of human CD8⁺ cells (A) and CD4⁺CD25⁺Foxp3⁺ cells (B) in the peripheral blood were determined using TruCount tubes. Values from individual mice are depicted. (C) The ratio of CD8 cell number to CD4⁺25⁺Foxp3⁺ cell number for each animal in panels A and B is plotted on a log scale. In panels A and B, asterisks (*) indicate significant differences between mock- and IL-18-treated animals (n = 6 to 8/group; p&lt;0.05). In panel C, asterisks (**) indicate p&lt;0.001.</p></caption></fig>", "<fig id=\"pone-0003289-g004\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003289.g004</object-id><label>Figure 4</label><caption><title>Effects of Human IL-18 on CD8 T Cell Differentiation in NOG mice.</title><p>NOG mice were injected intraperitoneally with 5×10⁶ human PBMCs, and beginning 1 day later, were injected with 15 µg recombinant human IL-18 daily for three weeks. Spleen tissue and peripheral blood were collected and analyzed for the presence of different CD8⁺ T cell subsets by flow cytometry. Cell populations were initially gated on human CD45 and human CD8. Filled bars represent values mean±s.d. obtained from IL-18-treated animals; open bars represent values obtained from mock-treated animals. The data in this figure represent results from one of two independent experiments. Asterisks indicate significant differences between mock- and IL-18-treated animals (p&lt;0.05).</p></caption></fig>", "<fig id=\"pone-0003289-g005\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003289.g005</object-id><label>Figure 5</label><caption><title>Human IL-18 Accelerates Xenogeneic GVHD in NOG mice.</title><p>5×10⁶ PBMC were transferred into NOG mice on day 0, and IL-18 or PBS injected daily starting on day 1. On the x-axis are days after transfer of cells. On the y-axis is the proportion of recipients surviving (n = 8/group; P&lt;.02).</p></caption></fig>", "<fig id=\"pone-0003289-g006\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003289.g006</object-id><label>Figure 6</label><caption><title>Effect of Human IL-18 on Treg-Induced Delay of Xenogeneic GVHD in NOG Mice.</title><p>NOG mice were injected with 2×10⁷ Treg-depleted human PBMCs supplemented with either 4 million autologous CD4⁺CD25⁻ T cells (“CD4 T cells”) or 4 million autologous, ex vivo-expanded CD4⁺CD25⁺ cells (“Tregs”). One cohort of each group received daily injections of 15 µg recombinant human IL-18 for three weeks. Human cell engraftment in the peripheral blood was measured 11, 22, and 30 days post-injection. The animals were followed until the onset of xenogeneic GVHD. (A) Experimental overview. (B) Kaplan-Meier Survival Analysis (Log-Rank) of the indicated cohorts of mice. The Holm-Sidak method for multiple comparisons (significance level 0.05) was performed, and showed significant differences between all groups. (C–E) Peripheral blood levels of human CD8⁺ T cells were measured at Day 11 (C), Day 22 (D), and Day 30 (E) post-injection. Note that all IL-18-treated animals are not represented in panel E due to their death from acute GVHD.</p></caption></fig>", "<fig id=\"pone-0003289-g007\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003289.g007</object-id><label>Figure 7</label><caption><title>Human IL-18 Does Not Prevent Treg Suppressive Function in Vitro.</title><p>Purified Tregs were expanded ex vivo for 18 days, and then tested in an in vitro suppression assay. Autologous PBMCs were labeled with CFSE and stimulated using anti-CD3 Ab coated beads and mixed with either no Tregs (top panel) or expanded Tregs at the indicated ratio (Treg:PBMC) in the presence or absence of recombinant human IL-18 (500 ng/ml). Histograms show the expansion of CD8⁺ cells on day 4 of culture.</p></caption></fig>", "<fig id=\"pone-0003289-g008\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003289.g008</object-id><label>Figure 8</label><caption><title>Rapid Down Regulation of IL18-Rα on Tregs Following Cell Activation.</title><p>Purified populations of Tregs, as well as bulk CD4 and CD8 cells, were activated with either αCD3/αCD28 beads, PHA (5 µg/ml), or PHA+recombinant human IL-18 (500 ng/ml). IL-18Rα expression was measured by flow cytometry. IL-18Rα MFI on effector CD4 cells (A), Tregs (B) and effector CD8 cells (C) is plotted before stimulation and 1, 2, and 3 days following stimulation.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"pone.0003289.s001\"><label>Figure S1</label><caption><p>Photomicrographs of lung and liver tissue sections taken from animals after xenogeneic GVHD induction. 10 million PBMC were transferred into NOG recipient mice followed by daily injections of IL-18 (15 µg) or PBS. Animals were euthanized on day 20 when recipients of PBMC and IL-18 were moribund. Sections were stained with hematoxylin and eosin. Left: Lung sections taken from PBMC or PBMC+IL-18 recipients, demonstrating severe inflammation and lymphocytic infiltration in IL-18 recipients. Right: Liver sections demonstrating marked periportal lymphocytic infiltrates in IL-18 recipients.</p><p>(5.49 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003289.s002\"><label>Figure S2</label><caption><p>Photomicrographs of gut tissue sections taken from animals after xenogeneic GVHD induction. Sections were obtained from the animals described above in Suppl. ##SUPPL##0##Fig. S1##. Left: low magnification. Middle: higher magnification. Right: highest magnification.</p><p>(5.14 MB TIF)</p></caption></supplementary-material>" ]

[ "<fn-group><fn fn-type=\"COI-statement\"><p><bold>Competing Interests: </bold>Z.L.J. is an employee of GlaxoSmithKline; the other authors have no competing financial interests to declare.</p></fn><fn fn-type=\"financial-disclosure\"><p><bold>Funding: </bold>This study was supported by research funding from GlaxoSmithKline to CHJ. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"pone.0003289.g001\"/>", "<graphic xlink:href=\"pone.0003289.g002\"/>", "<graphic xlink:href=\"pone.0003289.g003\"/>", "<graphic xlink:href=\"pone.0003289.g004\"/>", "<graphic xlink:href=\"pone.0003289.g005\"/>", "<graphic xlink:href=\"pone.0003289.g006\"/>", "<graphic xlink:href=\"pone.0003289.g007\"/>", "<graphic xlink:href=\"pone.0003289.g008\"/>" ]

[ "<media xlink:href=\"pone.0003289.s001.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003289.s002.tif\"><caption><p>Click here for additional data file.</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>The systematic mapping of genetic interactions in biological systems has the potential to provide a better understanding of how genes function as networks to carry out and regulate cellular processes. In particular, recent advances in the experimental technologies which allow for the large-scale screening of the effects of combinatorial gene deletions are providing an exciting glimpse into the organization of complex genetic networks in terms of revealing novel interacting cellular components and compensatory pathways involved in many cell functions. Comprehensive maps of genetic interactions in model organisms, such as yeast, may also provide a valuable template for understanding the basic principles underlying the relationships between genotype and phenotype in other populations ##REF##17510664##[1]##. In humans, genetic interactions are involved in many complex phenotypes and they contribute to most genetic disorders, but the organization of the underlying networks is largely unknown ##REF##12360236##[2]##, ##REF##17449820##[3]##. Due to their combinatorial nature, the mapping of genetic interactions is highly labor-intensive even in genetically amenable organisms. Efficient computational frameworks are therefore required to underpin the full potential of these experiments.</p>", "<p>Several large-scale studies, especially in yeast <italic>Saccharomyces cerevisiae</italic>, have already identified a number of synthetic lethal interactions, in which a combination of two individually non-lethal mutations results in lethality ##REF##17510664##[1]##, ##REF##16309778##[4]##. Genome-wide screening strategies for synthetic sick or lethal interactions, such as those based on synthetic genetic arrays (SGA) or the diploid synthetic lethality analysis by microarray (dSLAM), have successfully been used for providing insights into the nature of genetic robustness ##REF##11232561##[5]##, and for identifying functional relationships among the genes and pathways ##REF##14764870##[6]##. In addition to this rather limited spectrum of observed phenotypes (synthetic sick/lethal <italic>vs.</italic> non-interacting pairs), quantitative phenotypes, such as the relative growth rate of yeast colonies, have recently been explored systematically using high-throughput screening approaches like epistatic miniarray profiling (E-MAP) and genetic interaction mapping (GIM) ##REF##16269340##[7]##, ##REF##18408161##[8]##. The importance of measuring a broader spectrum of genetic interactions when identifying functionally-related genes and pathway organizations has been demonstrated in theoretical and experimental studies ##REF##15592468##[9]##–##REF##17314980##[11]##. To provide reliable information on genetic interactions, customized data handling and pre-processing pipelines have been developed for the different screening approaches ##REF##16859555##[12]##–##UREF##0##[13]##.</p>", "<p>Regardless of the experimental technology used, the screening strategies aim to quantify the extent to which a mutation in one gene modulates the phenotype (or fitness) associated with altering the second gene, either by explicitly measuring or analytically comparing the observed fitness of double-mutants to those of single-mutants. More formally, a genetic interaction between mutants <italic>i</italic> and <italic>j</italic> can be defined by the deviation (<italic>ε_ij</italic>) of an observed double-mutant phenotype (<italic>P_ij</italic>) from the expected neutral phenotype of an organism's fitness (<italic>E_ij</italic>) under the hypothesis that it carries two non-interacting mutations (the null hypothesis). If the fitness <italic>P_ij</italic> is evaluated in terms of the growth rate of double-mutant <italic>w_ij</italic>, relative to the wild-type growth rate, and <italic>E_ij</italic> is a function <italic>g</italic>(<italic>w_i</italic>, <italic>w_j</italic>) of the relative single-mutant fitness values <italic>w_i</italic> and <italic>w_j</italic>, this definition can be formulated as:When testing the null hypothesis, a large absolute deviation |<italic>ε_ij</italic>| provides evidence for genetic interaction, while deviations close to zero indicate non-interacting gene pairs. Significant genetic interactions can further be classified into so-called synergistic interactions (<italic>ε_ij</italic>&lt;0) and alleviating interactions (<italic>ε_ij</italic>&gt;0). Synergistic interactions occur when a double-mutant has a more extreme effect on the fitness than would be expected from independent single mutants alone, and can therefore identify e.g. complementary pathways, with synthetic lethality being the extreme case (<italic>w_ij</italic> = 0). Alleviating interactions, in which the double-mutant phenotype is less severe than expected, can occur, for example, when the first mutation already impairs the function of a whole pathway and thereby masks the effect of the second mutation in the same pathway.</p>", "<p>Recently, Mani et al. ##REF##18305163##[14]## demonstrated that the product function <italic>g</italic>(<italic>w_i</italic>, <italic>w_j</italic>) = <italic>w_iw_j</italic> provides a convenient null model in the sense that it yields a distribution with location close to zero and low dispersion over all of the measured deviations. The comparison was based on the principle that, as the vast majority of gene pairs should be non-interacting, the rare gene pairs sharing a specific function should be distinguishable from this background distribution as outlying cases with extreme absolute deviations. Accordingly, it was shown that the observed deviations based on the multiplicative null model were indeed most accurate at identifying functional relationships between the genes ##REF##18305163##[14]##. In the present study, we asked two follow-up questions: (1) whether the observed deviations could be estimated directly from the double-mutant phenotypes under the multiplicative model; and (2) whether the prediction of specific functional links could be made at a similar accuracy without utilizing the measurements of single-mutant phenotypes. Under the assumptions that significant genetic interactions are rare and that the multiplicative model is a reasonable approximation in the case of no interaction, we developed a sequential approach that enables a multi-resolution approximation of the double-mutant fitness matrix to address these particular questions, and more generally, to provide a computational framework for exploring genetic interaction datasets.</p>" ]

[ "<title>Materials and Methods</title>", "<title>Approximating the double-mutant fitness matrix</title>", "<p>For a given subset of mutation pairs, we calculated the matrix approximation of Eq. 2 using the decomposition algorithm by De Leeuw ##UREF##2##[16]##. Formally, we solved the following weighted least-squares optimization problem at each step <italic>l</italic> of the procedurewhere both the given weight matrix <bold>C</bold>\n^{(<italic>l</italic>)} and the double-mutant fitness matrix <bold>W</bold> are symmetric of order <italic>n</italic>. After normalizing for the subset size, the square-root of the objective function obtained as the solution of the optimization of Eq. 3 is referred to as <italic>approximation error</italic> at phase <italic>l</italic>. Using this formulation, the solution <bold>ŵ</bold>\n^{(<italic>l</italic>)} minimizes the sum of the squared <italic>residual errors</italic>\nover all of the <italic>k</italic> gene pairs (<italic>i</italic>, <italic>j</italic>) with which were involved in the approximation process at the <italic>l</italic>\n^th step of the procedure. Although the general formulation in Eq. 3 allows for an element of <bold>C</bold>\n^{(<italic>l</italic>)} to be any non-negative number, we used binary weights only; in particular, we set the diagonal entries for all <italic>i</italic>, and for the missing data points, at each step <italic>l</italic>. Throughout the operation of the approximation procedure, we ensured that the weight matrix remained symmetric.</p>", "<p>In the dataset of St Onge et al. ##REF##17206143##[10]##, there were two missing values because of genetic linkage between the gene pairs involved in the double-deletion strains rad57Δrad61Δ and rad55Δshu2Δ.</p>", "<title>Selecting subsets of mutation pairs for estimation</title>", "<p>The subset of mutation pairs used at a particular step of the approximation process was encoded in the binary weight matrix <bold>C</bold>\n^{(<italic>l</italic>)}, that is, if and only if the mutation pair (<italic>i</italic>, <italic>j</italic>) is used at step <italic>l</italic>. To select increasingly large subsets of non-missing and non-diagonal mutation pairs from the double-mutant fitness matrix <bold>W</bold>, we adapted the floating search method of Pudil et al. ##UREF##3##[17]##. The sequential subset search method is characterized by a dynamically changing number of features included or eliminated at each step. In our implementation, the residual errors were used as the criterion function for adding or deleting mutation pairs. The operation of the forward-type subset selection scheme was organized through the following steps:</p>", "<p>Set <italic>l</italic>←0 and initialize the weight matrix <bold>C</bold>\n⁽⁰⁾ (see the next subsection)</p>", "<p>Estimate <bold>ŵ</bold>\n^{(<italic>l</italic>)} using the decomposition algorithm (previous subsection)</p>", "<p>While there are non-diagonal and non-missing entries with \n</p>", "<p>select pair (<italic>i</italic>, <italic>j</italic>) with having the smallest residual error (Eq. 4)</p>", "<p>set , make a new estimate of <bold>ŵ</bold>\n^{(<italic>l</italic>)} and re-calculate the residuals \n</p>", "<p>if there exists a pair (<italic>i</italic>, <italic>j</italic>) with and , then set \n</p>", "<p>set <italic>l</italic>←<italic>l</italic>+1 and repeat step 3</p>", "<p>We modified the general subset search method by making the deletion of pairs an adaptive process over the evolution of the subset search. More specifically, the threshold <italic>t</italic>\n^{(<italic>l</italic>)} used in the conditional exclusion step 3c is multiplied by 1.5 each time the pair selected for deletion is the same as that added in step 3a. This modification enabled the forward-type algorithm to recover from poor starting tolerance values (we used <italic>t</italic>\n⁽⁰⁾ = 2×10⁻⁵ in the present work) or from a poor initialization of the weight matrix (see the next subsection for details), and it also made it possible to increase the size <italic>k</italic> of the subsets up to the maximal size <italic>K</italic>. ##FIG##0##Figure 1## shows the evolution of the subset selection algorithm from the initial subset configuration to the full set of <italic>K</italic> = 323 mutation pairs in the dataset of St Onge et al. ##REF##17206143##[10]##.</p>", "<title>Initialization of the weight matrix from the data</title>", "<p>The approximation algorithm requires the weight matrix <bold>C</bold>\n^{(<italic>l</italic>)} to be non-singular at each stage of the procedure. Therefore, we cannot start the sequential approximation procedure from the empty subset of mutation pairs, but instead must adjust the initialization <bold>C</bold>\n⁽⁰⁾ = 0 for the dataset under analysis. After calculating the initial residuals from the whole double-mutant matrix, three types of adjustments were made in the present study: (<italic>i</italic>) for each <italic>i</italic>, we set for the pair (<italic>i</italic>, <italic>j</italic>) with the smallest residual to make all of the rows in <bold>C</bold>\n⁽⁰⁾ non-zero; (<italic>ii</italic>) we made an additional setting of for each identical row pair to reduce the linear dependence among the rows of <bold>C</bold>\n⁽⁰⁾; and finally (<italic>iii</italic>) we set pairs (<italic>i</italic>, <italic>j</italic>) with to 1 in the increasing order of their residuals until the determinant of <bold>C</bold>\n⁽⁰⁾ became non-zero. Every time a pair (<italic>i</italic>, <italic>j</italic>) was added in the initialization steps (<italic>i</italic>)–(<italic>iii</italic>), we also set its transpose entry to one, that is, , to keep the weight matrix symmetric. In the dataset of St Onge et al. ##REF##17206143##[10]##, this resulted in an initial weight matrix <bold>C</bold>\n⁽⁰⁾ with 28 entries of ones.</p>", "<title>Sequential estimation of the pairwise deviations</title>", "<p>The <italic>measured deviations ε_ij</italic> = <italic>w_ij</italic>−<italic>w_iw_j</italic> were estimated in two different ways. The sequential approximation procedure gives as its by-product a surrogate for the deviations in the form of the residual errors of Eq. 4. More specifically, we defined the <italic>ranked deviation</italic>\n of a mutation pair (<italic>i</italic>, <italic>j</italic>) as its residual error at the last step <italic>l</italic> during which the pair was included into the approximation subset. In this way, even when there are multiple inclusions and deletions of a particular pair during the procedure, we obtained an unambiguous ranking of the pairs according to their values. This ranking and the corresponding ranked deviations are shown in ##FIG##0##Figure 1## for all gene pairs of the dataset of St Onge et al. ##REF##17206143##[10]##, except for the 28 initial mutation pairs.</p>", "<p>The second set of estimates was obtained by stopping the sequential approximation procedure at a given subset size, <italic>k</italic>, and by using the estimate of Eq. 3 in place of the measured single-mutant fitness vector <bold>w</bold> in the definition of the deviation in Eq. 1. For a mutation pair (<italic>i</italic>, <italic>j</italic>), this resulted in a sequence of <italic>estimated deviations</italic>\n for step <italic>l</italic> at which the size of the approximation subset equals <italic>k</italic>. Note that since the residual errors in Eq. 4 are updated each time that a new pair is added, the estimated deviations can vary considerably as a function of <italic>k</italic>. These estimates are not generally congruent with the ranked deviations. The distributions of the estimated deviations with different subset sizes <italic>k</italic> are illustrated in ##FIG##3##Figure 4## in the dataset of St Onge et al. ##REF##17206143##[10]##.</p>", "<title>Quantitative genetic interaction measurements</title>", "<p>To evaluate the performance of the sequential matrix approximation procedure in practice, we applied it to a recent high-resolution genetic interaction study of St Onge et al. ##REF##17206143##[10]##. This particular study was chosen because it contains quantitative growth-rate measurements of both single- and double-mutant cell populations for a targeted set of 26 genes related to DNA repair in yeast <italic>S. cerevisiae</italic>. The detailed time course fitness measurements were performed in the presence and absence of the DNA-damaging agent methyl methaneusulfonate (MMS). The results of the present study were based on the growth measurements in the absence of MMS. The prediction of functionally related gene pairs was in fact more challenging in this case than in the data measured in the presence of MMS. The measured and estimated single-mutant fitness values and the double-deletion deviations in the dataset are shown in ##SUPPL##0##Figure S1## and ##FIG##4##Figure 5##, respectively.</p>", "<title>Defining gene pairs sharing a specific function</title>", "<p>Functional links among the 26 genes were defined using the same approach as in many previous genetic interaction studies ##REF##14764870##[6]##, ##REF##17206143##[10]##, ##REF##18305163##[14]##. Briefly, a term in the Biological Process branch of the Gene Ontology was considered specific if it was associated with fewer than 30 yeast genes, and two genes were considered to have a specific functional relationship if they shared any of those specific terms ##REF##18305163##[14]##. This resulted in the set of 35 specific functional links in the dataset of St Onge et al. ##REF##17206143##[10]##.</p>", "<title>Statistical evaluation of the predictive power</title>", "<p>Statistical enrichment of the specific functional links among a set of mutation pairs selected by the sequential approximation procedure was assessed using the standard hypergeometric test. Briefly, if <italic>t</italic> is the number of top mutation pairs selected according to their residual errors, and <italic>M</italic> is the total number of the functionally-related links, then the probability of obtaining at least <italic>m</italic> functionally-related pairs when selecting pairs at random from the set of <italic>K</italic> mutation pairs can be calculated using the cumulative distribution function:\n</p>", "<p>The enrichment for the <italic>M</italic> = 35 functionally-linked pairs among the <italic>t</italic> = 50 mutation pairs selected on the basis of either small (<italic>m</italic> = 1) or large (<italic>m</italic> = 22) residual errors is shown in ##FIG##1##Figure 2##. The dotted line shows the expected rate of the functional links when selecting mutation pairs at random, that is, <italic>M</italic>/<italic>K</italic> = 35/323 = 0.108.</p>", "<p>The predictive power of the measured and estimated deviations was assessed using the receiver operating characteristic (ROC) curves that characterize the relative trade-off between the true positive rate (sensitivity) and the false positive rate (1 - specificity). The overall predictive performance was summarized using the area under the ROC curve (AUC). The statistical significance of the difference in the AUC values between two genetic interaction measures was assessed using a custom written algorithm based on the method of DeLong et al. ##REF##3203132##[30]##. This nonparametric method uses the theory of generalized <italic>U</italic>-statistics to calculate an estimated covariance matrix and hence it can also take into account the correlated nature of the data. The ROC curves and the corresponding AUC values for the prediction of the 35 functional links in the dataset of St Onge et al. ##REF##17206143##[10]## are shown in ##FIG##2##Figure 3##.</p>" ]

[ "<title>Results</title>", "<title>Estimating single-mutant fitness values</title>", "<p>As an initial study objective, we sought to assess the accuracy to which the single-mutant fitness vector could be estimated directly from the double-mutant fitness matrix . In the quantitative genetic interaction dataset of St Onge et al. ##REF##17206143##[10]##, which was used in the following results, <bold>w</bold> is a 26-dimensional column vector and <bold>W</bold> is a 26×26-dimensional symmetric matrix. Under the multiplicative null model, Eq. 1 takes the form:where is the tensor product (or outer-product) of the vector <bold>w</bold> with itself, and <bold>E</bold> is the <italic>n</italic>×<italic>n</italic>-matrix comprising the <italic>ε_ij</italic> values as its elements for each gene pair <italic>i</italic>, <italic>j</italic> = 1,2,…,<italic>n</italic>. In the ideal case, when there are no measurement inaccuracies, the approximation problem of Eq. 2 could easily be solved using the well-established machinery of linear algebra. More precisely, using the spectral decomposition theorem, one can represent any symmetric real matrix as , where <italic>λ</italic> is the largest eigenvalue of <bold>W</bold> and <bold>e</bold> is the corresponding eigenvector ##UREF##1##[15]##. Under the unrealistic assumption that there are no genetic interactions among any of the gene pairs, the approximation would in fact be exact, that is, the residual error <bold>E</bold> equals zero. However, as the genetic interaction screens are bound to present with experimental variation, missing data, and hopefully also with significant genetic interactions, the estimation problem must in practice be solved by numerical means.</p>", "<p>In the present work, we developed a sequential matrix approximation procedure, which uses increasingly larger subsets of mutation pairs in <bold>W</bold> to provide a series of estimates for <bold>w</bold> as solutions of the weighted least squares problem ##UREF##2##[16]##. During the first rounds, the procedure solves the approximation problem of Eq. 2 using only those mutation pairs that best fit the multiplicative model, and then gradually extends the subset to also include pairs with larger residual errors (see <xref ref-type=\"sec\" rid=\"s4\">Methods</xref> for details). Already when using all but the diagonal and missing entries of the double-mutant fitness matrix in the dataset of St Onge et al. ##REF##17206143##[10]##, we obtained an estimate relatively close to the actual measured single-mutant fitness vector, as compared to the conventional median estimate (##FIG##0##Figure 1A##). The estimation accuracy could be markedly improved by excluding those pairs with the largest residual errors from the approximation subset. The pairs having the greatest impact on the approximation error, in fact, corresponded to the five confirmed synthetic lethal pairs (##FIG##0##Figure 1B##). When the sequential procedure omitted those pairs from the approximation process, an accurate estimate was obtained for each of the single-mutant fitness measurements (see ##SUPPL##0##Figure S1##). As will be seen in the following subsections, however, the subset of mutation pairs which gives the most accurate estimates of the single-mutant fitness values does not necessarily lead to the best predictive power when identifying functionally related genes.</p>", "<title>Predicting specific functional relationships</title>", "<p>Beyond the dynamic variability in the estimates of the single-mutant fitness values, the behavior of the approximation procedure with different subsets of mutation pairs revealed another interesting observation: the order in which the mutation pairs were added into the approximation process reflects on average the relative order of their actual measured deviations, even if the measurements of single-mutant fitness were not employed (##FIG##0##Figures 1C and D##). In fact, the measured deviations <italic>ε_ij</italic> and the ranked deviations obtained from the approximation procedure were highly correlated (Pearson correlation equals 0.964, see ##SUPPL##1##Figure S2##). This led us to investigate whether such procedure-ranked deviations could be used instead of the measured deviations when predicting functional links between the mutations. To this end, we took the same set of gene pairs which were found to have a highly specific shared function in the previous study by Mani et al. ##REF##18305163##[14]## (see <xref ref-type=\"sec\" rid=\"s4\">Methods</xref> for their definition). Interestingly, the majority of the mutation pairs selected towards the end of the approximation procedure shared a specific function (##FIG##1##Figure 2##). The rate of the functional enrichment observed among the 50 pairs with the largest values was significantly higher than expected (<italic>p</italic>&lt;10⁻¹¹), whereas the functional enrichment was exceptionally low among the 50 pairs with the lowest values (<italic>p</italic> = 0.998). These results show that the sequential matrix approximation procedure gives as its by-product a ranking of the mutation pairs that is in good agreement with their likelihood of sharing a specific function.</p>", "<p>To test more systematically whether the prediction of the functional relationships could be made to an accuracy similar to that obtained when using the measured single-mutant fitness values, we assessed how well the ranking based on values can discriminate between gene pairs with or without a specific functional link. Similarly as in the original study of St Onge et al. ##REF##17206143##[10]##, the prediction capability was evaluated using the receiver operating characteristic (ROC) curves that show the relative trade-off between the sensitivity and specificity of the predictions at multiple decision thresholds (##FIG##2##Figure 3##). Surprisingly, the ranked deviations gave even a better prediction accuracy than the measured deviations according to the area under the ROC curve (AUC) values (AUC = 0.780 vs. AUC = 0.662, <italic>p</italic> = 0.003). The prediction capability was improved systematically at each specificity-level, demonstrating that the procedure could distinguish with high accuracy the functionally related gene pairs over the whole spectrum of the exceptional deviations (##FIG##2##Figure 3##). The order in which the mutation pairs were included into the approximation process further improved the relative classification power (AUC = 0.789, <italic>p</italic> = 0.002). The prediction accuracy of the double-mutant fitness values alone was similar to that of a random classifier (AUC = 0.506), demonstrating that the normalization by the measured or estimated single-mutant fitness values can, in any case, improve the prediction of the functional links. The estimates achieved using different subsets of mutation pairs can also lead to different degrees of prediction accuracy (##TAB##0##Table 1##).</p>", "<title>Distribution of the estimated deviations</title>", "<p>Finally, we investigated how the estimated deviations obtained from the approximation procedure at different subset sizes, <italic>k</italic>, are distributed relative to the true deviations <italic>ε_ij</italic> obtained when using the measured single-mutant fitness values. When comparing the different distributions, we used the same interpretation rule as in the earlier comparison by Mani et al. ##REF##18305163##[14]##: an ideal definition of genetic interaction should result in a tight distribution (indicating low variability) that is centered at zero (indicating low bias) for the bulk of the measured interactions (reflecting the background distribution of non-interacting genes). The subset of mutation pairs being used in the approximation process had a marked effect on the distribution of the estimated deviations (##FIG##3##Figure 4##). Moderate subset sizes generated distributions with a lower bias and variability than those obtained using smaller subset sizes or all of the mutations (##TAB##0##Table 1##). Surprisingly, the cut-off point, <italic>k</italic> = 317, which gave the most accurate estimates for the single-mutant fitness values resulted in a relatively weak prediction accuracy for the functional links (AUC = 0.679). Although the different measures of location and dispersion gave relatively similar results, the trimmed mean and interquartile range identified the distribution with least background bias and variability at <italic>k</italic> = 136 (##FIG##3##Figure 4##). The estimated deviations obtained using this particular subset of mutation pairs were also most successful in predicting the functional relationships (AUC = 0.810). These results indicate that the distribution characteristics of the estimated deviations could serve as a guide to choosing the optimal subset of gene pairs for defining genetic interactions in a given dataset. Both the optimal subset size and the overall performance of the method is likely to depend on the properties of the dataset being analysed, including the number of gene pairs and and the degree of their functional homogeneity.</p>", "<p>To investigate how the differences in the distributions are visible in the conventional heat map visualizations, we displayed the color-coded deviations on a two-dimensional grid spanned by the individual mutations. Here, a special emphasis was placed on analyzing the estimated deviations , which provided the most ideal definition of genetic interactions in terms of both distribution characteristics and predictive power, relative to the measured deviations <italic>ε_ij</italic> (##FIG##4##Figure 5##). In general, the interaction patterns were relatively similar between the measured and estimated deviations. However, the transformation of the double-mutation fitness measurements through the approximation process seemed to emphasize the mutation pairs with exceptionally large absolute deviations (putative genetic interactions), and pointed out especially those pairs having positive deviations (alleviating interactions), while it diminished certain subsets of mutation pairs with negative deviations (synergistic interactions). For instance, a considerable number of double-deletion strains involving either hpr5 or sgs1 mutations showed marked evidence for alleviating interactions in the color map of the estimated deviations (##FIG##4##Figure 5A##), while these pairs were not identifiable from the original map (##FIG##4##Figure 5B##). At the same time, the approximation algorithm blotted out certain moderate degree synergistic interactions in the hpr5 deletion strain, including the effects of additional mutations of mms4, mph1, or mus81. Although these and other changes contributed positively to the prediction of functional relationships in the dataset of St Onge et al., further evaluation how well these findings can be generalized beyond this relatively small set of functionally related genes is required on independent datasets.</p>" ]

[ "<title>Discussion</title>", "<p>The growing availability of large-scale genetic interaction datasets is enabling computational methods to systematically explore how genes interact to produce phenotypes on a global network-level. While these datasets yield an unprecedented insight into the organization and function of complex genetic networks, their analysis also poses many challenging computational problems. Using a high-resolution screen of genetic interactions in yeast as an example dataset of St Onge et al. ##REF##17206143##[10]##, we have demonstrated that the computational approach based on sequential matrix approximations facilitates extraction of pertinent information from the background variation. A key finding of the present work is that the double-mutant fitness matrix alone carries enough information for accurate estimation of the single-mutant fitness values and for prediction of functional relationships among the genes. This makes it possible to avoid performing the single-mutant growth experiments, without compromising - if not even improving – the functional prediction power encoded in the double-mutant measurements. Surprisingly, the subset of mutation pairs which gave the most accurate estimates for the single-mutant fitness values did not lead to the most accurate predictions of the functional links. This may be due to experimental variability, such as differences in growth or screening conditions when measuring the strains carrying either single or double mutations, which may be beyond the capacity of the standard data pre-processing but can be normalized by the sequential approximation procedure. Other possible explanations for this surprising observation include biases in the definition of the gene pairs sharing a specific function or in the targeted set of genes pairs chosen for the particular interaction screen. Further study is therefore needed to confirm whether similar results can be obtained also in larger genetic interaction datasets, in which genes with much wider variety of functional categories are studied.</p>", "<title>Limitations and extensions of the procedure</title>", "<p>Perhaps the biggest technical limitation of the present work concerns the heuristic way in which the subsets of mutation pairs were selected for the approximation of the double-mutant fitness matrix. The greedy subset selection scheme was motivated by a similar approach successfully being used in many feature selection problems ##UREF##3##[17]##. An adaptive version of such a forward floating selection method was applied here because of its low computational complexity and because it was capable of excluding the most prominent outliers during the sequential approximation process (##FIG##0##Fig. 1##). Similarly, despite the weighted least squares matrix approximation algorithm being based on a rather straightforward decomposition method, it was able to reduce some degree of background variation in the data (##FIG##3##Fig. 4##). However, more sophisticated search and approximation schemes based on e.g. genetic algorithms or simulated annealing should lead to even better estimation and prediction results, or at least reduce the computational complexity. Additional modifications to enhance the present framework either in biological and/or computational terms include using deviations from the expected fitness as weights in the least squares approximation and using the sign of the deviations when including or excluding a mutation pair over the course of the sequential approximation process. While the present results were based on the rank-one approximation only, which enabled the partitioning of pairs of genes into two categories (interacting or non-interacting), utilizing the higher order ranks could allow us to classify the quantitative measurements into several categories, for instance, synergistic and alleviating interactions, or even more fine-grained classification of interactions that can occur between genes ##REF##15833125##[18]##. This could also help us also to distinguish those biological modules in which the distribution of genetic interactions does not follow the ideal tight and zero-centered distribution that has been used traditionally ##REF##15592468##[9]##, ##REF##18305163##[14]##.</p>", "<title>Future applications and research directions</title>", "<p>In spite of the above mentioned technical limitations, the present results support the feasibility of the approximation framework for systematic exploration of genetic interaction data, and warrant its applications to larger-scale datasets, such as those generated with the E-MAP or GIM screens, to confirm whether similar findings can be extrapolated to the genetic interactions data derived from high-throughput technologies. Other quantitative phenotypes or experimental techniques for defining or measuring genetic interactions could, in principle, also be used, although certain modifications will be needed to adapt the procedure to the specific characteristics of each genetic interaction screen. In the larger-scale screens, the gene pairs under analysis can be selected more randomly among a wider range of functions, thus increasing the expected proportion of non-interacting pairs. Accordingly, the more interactions there are being measured, the better the assumptions behind the multiplicative model will be justified, provided that significant genetic interactions remain relatively rare. The many missing values typically occurring in the large-scale screens should not pose a major problem for our approach either, due to its sequential nature being able to adapt to those subsets of double-mutant measurements with the best approximation power. Hence, the approximation approach is likely to yield even better results with larger and unbiased datasets. Similarly, even if the assumption of low frequency of significant interactions may become compromised in more targeted studies, such as the one of St Onge et al, this should not have a major effect on the results as the strongest genetic interaction pairs are effectively filtered out in the sequential estimation process. For such smaller-scale and more targeted genetic interaction studies, a further increase in the performance could be obtained by modifying the null model for non-interacting pairs to take into account the multitude of single-mutants affecting the particular double-mutant fitness value. This is one of the modeling challenges which we aim to tackle in the future.</p>", "<title>Integrative analysis together with other data sources</title>", "<p>More generally, the computational approach based on the sequential matrix approximation can provide a principled framework for exploring and classifying genetic networks and interactions using a wide spectrum of global data sources, including the localization, mRNA or protein expression, physical interaction and functional annotation of the proteins encoded by the genes ##REF##15496468##[19]##. It has previously been demonstrated that physical protein-protein interactions, in particular, provide useful information that is, by and large, complementary to that obtained from the functional genetic interactions ##REF##14764870##[6]##, ##REF##17703239##[20]##. To reveal the modular structure of the underlying networks and functional organization the multitude of pathways reflected in such large-scale data types, various network partitioning methods have been used to detect either densely- or similarly-connected clusters as well as significantly-repeated motifs in the individual or integrated interaction networks ##REF##12525261##[21]##–##REF##18421374##[24]##. However, many open questions still remain about the integrative analysis strategy of these datasets and the most meaningful interpretation of their results. For instance, the extent to which the genetic interaction could be explained by the other information sources, such as protein-protein, protein-DNA, metabolic network and protein structure data ##REF##15877074##[25]##–##REF##18671848##[28]##, and how these should be efficiently employed when scoring genetic interactions using measures such as the pairwise deviations, the S- and COP-scores, or the correlation and congruence between the interaction patterns ##REF##15592468##[9]##–##REF##16859555##[12]##, ##REF##18628749##[27]##–##UREF##4##[29]##. Finally, the success of any computational approach for constructing genetic interaction networks is likely to be driven by parallel improvements in the experimental technologies, such as enabling measurement of phenotypic effects in response to the mutation of more than two genes in combination.</p>" ]

[]

[ "<p>Conceived and designed the experiments: APJ LLE TA. Performed the experiments: APJ TA. Analyzed the data: APJ JH. Wrote the paper: APJ JH LLE TA.</p>", "<p>Despite the emerging experimental techniques for perturbing multiple genes and measuring their quantitative phenotypic effects, genetic interactions have remained extremely difficult to predict on a large scale. Using a recent high-resolution screen of genetic interactions in yeast as a case study, we investigated whether the extraction of pertinent information encoded in the quantitative phenotypic measurements could be improved by computational means. By taking advantage of the observation that most gene pairs in the genetic interaction screens have no significant interactions with each other, we developed a sequential approximation procedure which ranks the mutation pairs in order of evidence for a genetic interaction. The sequential approximations can efficiently remove background variation in the double-mutation screens and give increasingly accurate estimates of the single-mutant fitness measurements. Interestingly, these estimates not only provide predictions for genetic interactions which are consistent with those obtained using the measured fitness, but they can even significantly improve the accuracy with which one can distinguish functionally-related gene pairs from the non-interacting pairs. The computational approach, in general, enables an efficient exploration and classification of genetic interactions in other studies and systems as well.</p>" ]

[ "<title>Supporting Information</title>" ]

[ "<p>The authors thank Dr. Arho Virkki for modifying the R-code of the decomposition method for the purposes of the present study and Dr. Milla Kibble for the language review of the manuscript.</p>" ]

[ "<fig id=\"pone-0003284-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003284.g001</object-id><label>Figure 1</label><caption><title>Dynamic behavior of the sequential procedure as a function of mutation pairs.</title><p>(A) The mean absolute error between the measured single-mutant fitness vector and its estimate when using the selected mutation pairs. The dotted horizontal line indicates the estimation error obtained when using the median over the rows/columns of the double-mutant fitness matrix (average absolute error equals 0.0797). In case significant genetic interactions are rare, median of the colony sizes over all of the double mutants arising from the same single deletion strain provides an estimate of the effect of the particular single-mutant on the growth rate ##REF##16859555##[12]##. (B) The approximation error when using the selected mutations pairs to approximate the double-mutant fitness matrix (see Eq. 3 in <xref ref-type=\"sec\" rid=\"s4\">Methods</xref>). The horizontal dotted line indicates the point of steepest increase in the approximation error, <italic>k</italic> = 317, which also gives on average the most accurate estimates of the single-mutant fitness values (average absolute error equals 0.0168). In each panel, the five spots after that line identify the synthetic lethal mutation pairs with double-mutant fitness value of zero. (C) The ranked deviations of the mutation pairs (<italic>i</italic>, <italic>j</italic>) defined according to their residual errors (see Eq. 4 in <xref ref-type=\"sec\" rid=\"s4\">methods</xref>). (D) The measured deviations |<italic>ε_ij</italic>| of the selected mutation pairs obtained using the actual measurements of the single-mutant growth effects.</p></caption></fig>", "<fig id=\"pone-0003284-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003284.g002</object-id><label>Figure 2</label><caption><title>Relationship between the ranking of the mutation pairs and their functional links.</title><p>The proportion of shared specific functional links among the top mutation pairs, when these pairs were selected in the increasing (red) or decreasing (blue) order according to their residual errors during the approximation process. The dotted line indicates the expected rate of the functional links when selecting mutation pairs at random (the expected proportion equals 0.108).</p></caption></fig>", "<fig id=\"pone-0003284-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003284.g003</object-id><label>Figure 3</label><caption><title>Predicting shared functional links using different measures of genetic interactions.</title><p>The accuracy of the prediction is evaluated using the receiver operating characteristic (ROC) curves for each measure: ranked deviations (red), measured deviations (blue), and the double-mutant fitness values (black). The true positive rate (TPR, or sensitivity) is the fraction of gene pairs correctly predicted to have functional links, and the false positive rate (FPR, or 1 - specificity) is the fraction of non-functionally linked gene pairs incorrectly predicted to have functional links. The overall prediction performance is summarized using the area under the ROC curve (AUC). For an ideal classifier, TPR = 1, FPR = 0 and AUC = 1, whereas a random classifier has on average AUC = 0.5 (the dotted diagonal line).</p></caption></fig>", "<fig id=\"pone-0003284-g004\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003284.g004</object-id><label>Figure 4</label><caption><title>Distributions of the measured and estimated deviations over all mutation pairs.</title><p>Two distinct distributions are shown for the mutation pairs with specific functional links (red) and for the background pairs not sharing specific functional links (black). The measured deviations <italic>ε_ij</italic> generate bimodal distribution for the pairs with specific functional links (the upper panel). The estimated deviations generate background distributions with sharper peaks depending upon the size <italic>k</italic> of the subsets of mutations used in the approximation (the lower panels): <italic>k</italic> = 28 (initial subset of mutations), <italic>k</italic> = 136 (good prediction capability), <italic>k</italic> = 317 (smallest estimation error), and <italic>k</italic> = 323 (all mutation pairs). The five smallest deviation values in each distribution plot correspond to the five synthetic lethal mutation pairs. ##TAB##0##Table 1## lists the shape parameters of these distributions calculated over all of the mutation pairs. The two distributions in each individual plot are scaled according to their total number of pairs. The non-scaled versions of the same distributions are provided as ##SUPPL##2##Figure S3##, which allows for better visual comparison between the functionally-linked and the functionally non-linked pairs.</p></caption></fig>", "<fig id=\"pone-0003284-g005\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003284.g005</object-id><label>Figure 5</label><caption><title>Estimated pairwise deviations <italic>vs.</italic> the actual measured deviations.</title><p>The color-coded heat map shows the estimated deviations (A) and the measured deviations (B) on a 26×26 grid. The estimated deviations were obtained at the cut-off point , which generates the most ideal distribution and provides the best discrimination between the functionally-linked and the non-functionally linked pairs (see ##TAB##0##Table 1##). While the five confirmed synthetic lethal pairs (sgs1Δ mus81Δ, sgs1Δmms4Δ, sgs1Δslx4Δ, sgs1Δhpr5Δ and rad54Δhpr5Δ) are clearly visible in both of the maps, there are marked differences in the more subtle interaction scores at many places of the matrix between the estimated deviations and the measured deviations <italic>ε_ij</italic>. Red color corresponds to synergistic interaction scores and blue to alleviating interactions. The grey boxes indicate missing data points.</p></caption></fig>" ]

[ "<table-wrap id=\"pone-0003284-t001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003284.t001</object-id><label>Table 1</label><caption><title>Distribution characteristics for the measured and estimated deviations.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Subset size, <italic>k</italic>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Median deviation</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Trimmed mean</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Median absolute deviation</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Interquartile range</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">AUC value</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Measured</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−0.0141</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−0.0152</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0267</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0169</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.662</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">28 (Initial set)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0000217</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.00107</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0209</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0147</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.772</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>136</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>0.000687</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>0.000388</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>0.0202</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>0.00958</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>0.810</bold>\n</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">208</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.00134</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.00118</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0141</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0108</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.773</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">317</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−0.000634</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−0.000816</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0173</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0120</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.679</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">323 (All pairs)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−0.000634</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.00142</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0236</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0169</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.730</td></tr></tbody></table></alternatives></table-wrap>" ]

[ "<disp-formula><label>(1)</label></disp-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<disp-formula><label>(2)</label></disp-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<disp-formula><label>(3)</label></disp-formula>", "<disp-formula><label>(4)</label></disp-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<disp-formula></disp-formula>" ]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"pone.0003284.s001\"><label>Figure S1</label><caption><p>Estimated vs. measured single-mutant fitness values. The comparison is shown both as histogram and scatter-plot. The two fitness values were highly correlated (Pearson correlation equals 0.952 and the offset and slope of the best fit line are 0.0429 and 0.960, respectively). The estimated values were calculated at the cut-off point k = 317, in which the approximation procedure used all but the diagonal and missing entries of the double-mutant fitness matrix and also omitted those six pairs with the most extreme residual errors (the five synthetic lethal mutations and one plausible synergistic mutation pairs). This point can approximately be identified from the sharp increase in the trace of approximation error (##FIG##0##Figure 1B##, the vertical dotted line).</p><p>(0.01 MB PDF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003284.s002\"><label>Figure S2</label><caption><p>Scatter-plot between the measured and ranked deviations. The ranked deviations were highly correlated with the true measured deviations over all of the mutation pairs (Pearson correlation equals 0.964). The inset shows the five synthetic lethal mutation pairs. The dotted diagonal line corresponds to the one-to-one correspondence between the two deviations.</p><p>(0.01 MB PDF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003284.s003\"><label>Figure S3</label><caption><p>Distributions of the measured and estimated deviations. The non-scaled version of the ##FIG##3##Figure 4##, which can better show the discrimination between the distributions of functionally-linked (red) and functionally non-linked pairs (black).</p><p>(0.02 MB PDF)</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><fn id=\"nt101\"><p>The rows correspond to distributions of the estimated deviations with different sizes of subsets of those gene pairs used in the approximation process (see ##FIG##3##Figure 4##). The distribution of the measured deviations <italic>ε_ij</italic> is used as a references value for the different parameters (the first row). As robust measures of location (bias) and dispersion (variability), we calculated the trimmed mean and interquartile range, respectively, in addition to the median and median absolute deviation that were used in the previous comparative study ##REF##18305163##[14]##. The bold type indicates the subset size which provided the most accurate prediction of the functional links in terms of the area under the receiver operating characteristic curve (AUC).</p></fn></table-wrap-foot>", "<fn-group><fn fn-type=\"COI-statement\"><p><bold>Competing Interests: </bold>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p><bold>Funding: </bold>The Academy of Finland (grant 120569) and the Graduate School in Computational Biology, Bioinformatics and Biometry (ComBi). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"pone.0003284.e001.jpg\" mimetype=\"image\" position=\"float\"/>", "<inline-graphic xlink:href=\"pone.0003284.e002.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e003.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pone.0003284.e004.jpg\" mimetype=\"image\" position=\"float\"/>", "<inline-graphic xlink:href=\"pone.0003284.e005.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e006.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e007.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pone.0003284.g001\"/>", "<inline-graphic xlink:href=\"pone.0003284.e008.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e009.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e010.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pone.0003284.g002\"/>", "<inline-graphic xlink:href=\"pone.0003284.e011.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pone.0003284.g003\"/>", "<graphic id=\"pone-0003284-t001-1\" xlink:href=\"pone.0003284.t001\"/>", "<inline-graphic xlink:href=\"pone.0003284.e012.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e013.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e014.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pone.0003284.g004\"/>", "<inline-graphic xlink:href=\"pone.0003284.e015.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e016.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pone.0003284.g005\"/>", "<graphic xlink:href=\"pone.0003284.e017.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pone.0003284.e018.jpg\" mimetype=\"image\" position=\"float\"/>", "<inline-graphic xlink:href=\"pone.0003284.e019.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e020.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e021.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e022.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e023.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e024.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e025.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e026.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e027.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e028.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e029.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e030.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e031.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e032.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e033.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e034.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e035.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e036.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e037.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e038.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e039.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003284.e040.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pone.0003284.e041.jpg\" mimetype=\"image\" position=\"float\"/>" ]

[ "<media xlink:href=\"pone.0003284.s001.pdf\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003284.s002.pdf\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003284.s003.pdf\"><caption><p>Click here for additional data file.</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Interest in global health research and training is rapidly increasing, especially among medical students. ##REF##17327707##[1]## As more U.S. medical centers strive to provide international research and clinical experiences, the requirement for certified international institutional review boards (IRB) in the research arena has also increased. In addition, large pharmaceutical companies, conduct 29% to 70% of their clinical trials in developing countries, although clear figures are unavailable. ##UREF##0##[2]## Transjurisdictional research requires the approval of both originating country and the country of operation. IRBs or Ethics Committees in developed countries often have little grasp of the conditions for ethical review in other and particularly developing countries. Additionally, there is concern in developed countries that research, particularly industry sponsored, is migrating ‘off shore’ due to lower costs, but more particularly, less burdensome regulatory environments.</p>", "<p>To ensure international research protects the rights and welfare of human subjects, the Office of Human Research Protection (OHRP) of the U.S. Health and Human Services requires all federally-sponsored research conducted on human subjects at international sites have approval by an IRB holding an OHRP Federal wide Assurance (FWA).</p>", "<p>Each institution with a FWA is responsible for ensuring investigators conducting HHS-supported human subjects research understand and act in accordance with the requirements of the HHS regulations for the protection of human subjects. Therefore, as stated in the Terms of the FWA, OHRP strongly recommends that institutions and their designated IRBs establish training and oversight mechanisms to ensure investigators maintain continuing knowledge of, and comply with relevant ethical principles and federal regulations, written IRB procedures, OHRP guidance, state, local laws and international laws, and institutional policies for the protection of human subjects. In addition, OHRP recommends investigators complete institutional educational training before conducting human subjects research; in some instances, such as for the National Institute of Health, training is mandatory for all key personnel conducting NIH-sponsored human subjects research. In addition to the ethical aspects of clinical research, other areas of equal importance include requirements for authorship and dissemination of research results. One of the conclusions of the Ethics of Research Related to Healthcare in the Developing Countries, specifies that ‘there is an urgent need to further education and training to ensure that those [researchers] in developing countries are able to discuss ethical issues effectively with external sponsors and others and to have mechanisms in place to deal with issues that arise. ##UREF##1##[3]##\n</p>", "<p>Strengthening bioethics training for both young and seasoned researchers in Latin America is a vital need, particularly as foreign-funded research conducted in this part of the world continues to increase. Training resources in human research protection are available over the internet and several of them are free of charge. The Collaborative Institutional Training Initiative (CITI) modules offer one of the most complete programs, ranging from Basic Aspects of Human Subject Protection, Good Clinical Practices, Responsible Conduct of Research among many others. Many of their modules have been translated to Spanish and adapted to local practices. The Collaborative Institutional Training Initiative (CITI) modules were first introduced in Peru at the 2007 Conference in Lima and were very well received by the audience. Between May and August 2007, 804 individuals had requested access to the Spanish version of the Basic course module, 90% of whom were Peruvian, The Office of Human Research Protection (OHRP), the National Cancer Institute (NCI), and Family Health International (FHI) also have human subject protection training modules geared to investigators and/or IRB members.. These resources are especially useful to existing research programs in the United States, but may not be as relevant for scientific communities in the developing world in the absence of structured institutional Human Research Protection Programs and lack the one-on-one approach. The Helsinki Declaration issued by the World Medical Association and the International Ethical Guidelines for Biomedical Research published by the International Organizations of Medical Sciences (CIOMS) are essential reference documents for the IRB community, as well, and are discussed fully in several of these training events.</p>", "<p>Although there is general agreement among investigators that training in ethical aspects of research is essential to conducting good and ethical studies, traditionally, clinical research grants have not provided funds for human research protection training. In addition, government and academic institutions at international sites always have limited discretionary funds for this type of training. Live courses and workshops led by experts in the field of research ethics are often prohibitively costly or available in limited geographic areas, but are especially valuable for encouraging one-on-one interaction with other investigators and opportunities for trainees to learn from case-based discussions or clarify areas of uncertainty–which are common in international biomedical ethics.</p>", "<p>Recently international and national agencies such as the World Health Organization, the European Commission, the US National Institutes of Health, and the Wellcome Trust have shown an interest in addressing this concern and have funded initiatives for training programs and capacity building regional and national workshops. Here we will report on our experience in developing a series of live courses and workshops that could provide useful information for these newly developed programs.</p>" ]

[ "<title>Materials and Methods</title>", "<title>The Training Model</title>", "<p>The US Naval Medical Research Center Detachment Peru (NMRCD-Peru) is one of five overseas US military infectious diseases laboratories and the only one located in the Americas. The central geographic location in Peru has made it an easy-to-access country for South American colleagues who desire to participate in training activities. Through collaboration with the University of Washington, the NIH Department of Clinical Bioethics, the Peruvian National Institute of Health, the Peruvian Institutional Review Board Network and local Peruvian universities, NMRCD became the center for bioethics research training for participants from across Latin America from 2004 through 2007. Although universities provide the ideal location and resources for training activities, good networking, the support of domestic and international government and private institutions and a strong commitment from the Peruvian IRB Network made it possible for a series of training efforts to be provided by NMRCD in Peru.</p>", "<p>The training model followed by NMRCD combined distance learning, interactive teaching and high level expert teaching in workshops, courses, conferences, webminars and videoconferences as the key element for success. Courses and conferences involved didactic sessions and mock IRB discussions conducted by experts from the U.S, and from Peru to provide a more relevant approach. Topics included conducting ethical research, informed consent, placebo versus standard of care, research with children, authorship, feedback to research subjects, repository and tissue sample banks and international collaborative research. Each participant completed a test at the end of the event and received a certificate of attendance. Participants included IRB Chairs and members, researchers, persons directly involved in clinical research and professionals directing offices in academic or research organizations. All courses and conferences were co-sponsored by the Peruvian Medical Board Association and local Universities and continuing medical education credits were awarded to participants who completed the post-test.</p>", "<p>The most popular training courses have been the Conferences on Ethical Aspects of International Collaborative Research held in Lima, and Iquitos, a city in the Peruvian Amazon region; and the Conference in Ethics in Collaborative International Research: Practical Issues and Constructive Tools for Latin American Research Teams held in Lima and Arequipa, a city in the Southern Andes. The satellite conferences gathered approximately 80 students each, while the Lima-based conferences had nearly 200 participants each. The lectures presented various topics of crucial importance to ethics in research, such as coercion, undue inducement, exploitation, informed consent, research with vulnerable populations, placebo versus standard of care, research with children, criteria for authorship, feedback to the research subjects, repository and tissue sample banks and international collaborative research. To better illustrate topics and make them more relevant to South America, faculty incorporated results and case-studies from research conducted in Peru and other parts of the world. Faculty members were well-published speakers from the U.S. OHRP and NIH Department of Clinical Bioethics, the Universities of Washington and Texas and Peruvian academic and regulatory entities and IRBs. The events were organized in collaboration with Peruvian organizations, such as the Universidad Peruana Cayetano Heredia, Universidad Nacional Mayor de San Marcos, the Peruvian Medical Association, the Peruvian Ministry of Health, the Peruvian National Institute of Health. The University of Washington provided funding and coordination for the Ethics in Research Courses conducted in Lima and Iquitos in 2005 and the Lima Conference on Ethical Aspects of International Collaborative Research and course in Arequipa in the 2007. The U.S. NIH Department of Clinical Bioethics played a crucial role in providing training through the 2004–2007 video conference course entitled NIH Ethics and Regulatory Practices in Clinical Research, the three Latin American Conferences on Ethical Aspects of International Collaborative Research held in Lima in 2005, 2006 and 2007, and the Ethics in Research course held in Iquitos in 2006. The videoconference course includes 7 sessions and is simulcast to multiple sites throughout the U.S., Peru, Mexico and Puerto Rico.</p>", "<p>Since 2004, other training activities have included: webcast broadcasting on ethics in international clinical trials and the Anniversary of the Belmont Report; and two workshops for IRB Administrators (##FIG##0##Figure 1##).</p>" ]

[ "<title>Results</title>", "<p>A total of 927 (258 of whom were IRB members) from 12 different countries in the Americas participated in training courses between 2004 and 2007 (##FIG##1##Figure 2##). Of the 927 participants, 836 were Peruvian and 510 of these were Peruvian MoH staff. Forty-nine percent (49%) of the participants were women. Suggestions received from the students encouraged the organization of more courses and post-test results demonstrated recognition of basic concepts of ethics in research.</p>", "<p>A total of 137 of the 804 Peruvians who registered to take the CITI online modules completed the CITI course entitled Basic Aspects in the Protection of Human Research Subjects within the first 10 months.</p>", "<p>The NIH Ethics and Regulatory Practices in Clinical Research video conference course has been so successful in Peru that since 2005, two additional sites in Lima were registered directly with the NIH to broadcast the course sessions from their own facilities. Participants in the videoconference series maintained a 70–80% attendance rate and anonymous surveys showed high satisfaction with the overall course (##FIG##2##Figure 3##). A total of 95 Peruvians received NIH certificates of participation from 2004 through 2007.</p>" ]

[ "<title>Discussion</title>", "<p>The developing world needs both, bioethics training and IRB capacity reinforcement to ensure research conducted in each country is compliant with international standards, while at the same time, sensitive to the needs of the local populations. Our collaborative bioethics conferences co-hosted by Peruvian academic and governmental organizations, U.S. NMRCD, and U.S. based government and academic institutions provided a unique opportunity for fostering adherence to ethical standards of research in this region. In Peru, this partnership has been extremely fruitful, with the participation of over 927 professionals from 12 different Latin American countries.</p>", "<p>We believe inclusion of expert speakers, a diverse curriculum and the investment and commitment of local partners made our conferences successful. Involvement of Peruvian members in both the presentations and mock IRBs promoted the inclusion of topics relevant to the developing world and fostered greater understanding between investigators and IRB members from developing and developed countries. Although conference training lacked an applied, practical component, many of the participants had extensive experience already and benefited from reinforcement of theoretical concepts and examples from research conducted in other parts of the world. This training model can be easily reproducible by other developing world countries.</p>", "<p>Training courses on bioethics are essential for encouraging acknowledgement and understanding of the importance of ethical conduct among persons conducting clinical research in the developing world. In addition, these courses strengthen the capabilities of IRB members and encourage better functioning of existing IRBs and the creation of new ones. Our courses and conferences were perceived as very useful by the Latin American scientific community, with a-growing number of attendees registering for these events and requesting additional training opportunities. We believe the next step is to target more advanced individuals, such as IRB chairs and members and develop intensive site evaluations to assist with setting up systems for record-keeping, IRB activity monitoring, tracking modifications and continual renewals, as well as encouraging local hosting of additional training courses.</p>", "<p>Although our courses and conferences have received positive evaluations from the participants, we recognize that such evaluations provide limited evidence of the usefulness of the training program. Ideally, we would like to know if the training program contributes to better research ethics review and ultimately to better protection of research participants. We are not aware of any formal evaluations of training programs using such criteria, and it would be almost impossible to do such an evaluation in a rigorous manner. One could, however, measure the level of knowledge and understanding of ethical principles and human subjects regulations before and after a series of training courses. Again, there are no standardized instruments for such evaluations available right now. Given the increasing interest in funding training workshops by international agencies, we believe that the development of such an instrument should be of high priority.</p>" ]

[]

[ "<p>Conceived and designed the experiments: RL AGL. Performed the experiments: RL ZM RL JRZ. Analyzed the data: RL DLB ZM RL GM AGL. Contributed reagents/materials/analysis tools: DLB SMMT CN ER RL GM JRZ. Wrote the paper: RL DLB GM AGL JRZ.</p>", "<p>With the rapidly increasing number of health care professionals seeking international research experience, comes an urgent need for enhanced capacity of host country institutional review boards (IRB) to review research proposals and ensure research activities are both ethical and relevant to the host country customs and needs. A successful combination of distance learning, interactive courses and expert course instructors has been applied in Peru since 2004 through collaborations between the U.S. Naval Medical Research Center Detachment, the University of Washington and the Department of Clinical Bioethics of the National Institutes of Health to provide training in ethical conduct of research to IRB members and researchers from Peru and other Latin American countries. All training activities were conducted under the auspices of the Peruvian National Institute of Health (INS), Ministry of Health. To date, 927 people from 12 different Latin American countries have participated in several of these training activities. In this article we describe our training model.</p>" ]

[]

[ "<p>Disclaimer:</p>", "<p>The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of the Department of the Navy, Department of Defense, nor the U.S. Government.</p>", "<p>Copyright statement</p>", "<p>Some of the authors are military service members and employees of the U.S. Government. This work was prepared as part of their official duties. Title 17 U.S.C.</p>", "<p>§ 105 provides that ‘Copyright protection under this title is not available for any work of the United States Government’. Title 17 U.S.C. § 101 defines a U.S. Government work as a work prepared by a military service member or employee of the U.S. Government as part of that person's official duties.</p>" ]

[ "<fig id=\"pone-0003274-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003274.g001</object-id><label>Figure 1</label><caption><title>Types of training events held in Peru, 2004–2007.</title></caption></fig>", "<fig id=\"pone-0003274-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003274.g002</object-id><label>Figure 2</label><caption><title>Courses and participants, 2004–2007.</title></caption></fig>", "<fig id=\"pone-0003274-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003274.g003</object-id><label>Figure 3</label><caption><title>Training activities conducted.</title></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[ "<fn-group><fn fn-type=\"COI-statement\"><p><bold>Competing Interests: </bold>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p><bold>Funding: </bold>This work was partially supported by work unit number No. 847705 82000 25GB B0016 and by Grants number R21NS048838, AI27757 and 1D43TW007393-01, Title: Fogarty International Center Grant, AIDS Research and (CFAR) Grant and Fogarty Global Infectious Diseases Training.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"pone.0003274.g001\"/>", "<graphic xlink:href=\"pone.0003274.g002\"/>", "<graphic xlink:href=\"pone.0003274.g003\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>\n<italic>Campylobacter</italic> is the most commonly identified cause of bacterial gastro-enteritis in the developed world ##UREF##0##[1]##,##REF##9396710##[2]##,##REF##15757549##[3]##. Infection can lead to serious sequelae such as Guillain-Barré syndrome and reactive arthritis ##REF##12730525##[4]##. Of the species pathogenic to humans, 90% of disease is caused by <italic>C. jejuni</italic> and most of the rest by <italic>C. coli</italic>\n##REF##12194770##[5]##. Both species are zoonotic pathogens with wide host ranges including farm animals (cattle, sheep, poultry, pigs) and wild animals (birds and mammals) ##UREF##0##[1]##,##REF##17479890##[6]##,##UREF##1##[7]##. The bacterium thrives at 37–42C in the mammalian and avian gut, but survives longest <italic>ex vivo</italic> in cold, dark, moist environments. <italic>Campylobacter</italic> is routinely isolated from fresh and marine water sources, and sewage ##UREF##2##[8]##.</p>", "<p>Epidemiological studies have demonstrated a link with exposure to contaminated food. Handling and eating raw and undercooked poultry have consistently been shown to be important risk factors. Case-control studies show that red meat and seafood are risk factors, as are eating at restaurants and barbecues, and drinking raw milk ##REF##12882945##[9]##,##REF##15095201##[10]##. However, food is not the only danger, and some studies have shown that regularly eating poultry and red meat in the home actually has a protective effect ##REF##15095201##[10]##. Water, particularly when untreated, can present a threat. Incidence of campylobacteriosis is typically sporadic, but outbreaks do occur that can often be traced to contamination of the water supply ##REF##12825731##[11]##–##REF##17469655##[13]##. Some authors have suggested that the strong seasonal variation in sporadic disease, which rises sharply in spring and peaks in summer, bears the hallmark of water-borne diseases such as cryptosporidiosis ##UREF##2##[8]##,##REF##12882945##[9]##,##UREF##3##[14]##.</p>", "<p>DNA-based methods of typing <italic>C. jejuni</italic> have the potential to resolve the controversy surrounding the origin of infection, but have thus far failed to do so. <italic>C. jejuni</italic> isolated from humans show considerable genetic overlap with meat and poultry isolates ##REF##16316490##[15]##,##UREF##4##[16]##,##REF##18045395##[17]##. However, a model-based approach that includes disparate sources is needed. Although <italic>C. jejuni</italic> genotypes do show some host association, the population is not strongly structured into differentiated clusters, so predicting host from genotype is challenging ##REF##17479890##[6]##. Phylogenetic approaches to tracing the source of infection have suggested that human isolates are more closely related to <italic>C. jejuni</italic> found in non-livestock than livestock ##UREF##5##[18]##. But recombination is frequent in <italic>C. jejuni</italic>\n##REF##16044246##[19]##, which means that a single phylogenetic tree is not an appropriate way to represent the ancestral history of a collection of <italic>C. jejuni</italic> genomes ##REF##16701339##[20]##.</p>", "<p>Here we report a systematic study of 1,231 cases of <italic>C. jejuni</italic> infection in Lancashire, England, which we have DNA-sequenced using multi-locus sequence typing ##REF##11136741##[21]## (MLST). We infer the source of infection of each patient by comparison to 1,145 animal and environmental <italic>C. jejuni</italic> isolates collated from previous studies in livestock, wild animals and the environment ##UREF##1##[7]##, ##UREF##4##[16]##, ##REF##11136741##[21]##–##REF##14602588##[28]##, using a novel population genetics approach that models DNA sequence evolution and zoonotic transmission. We treat the animal and environmental reservoirs of <italic>C. jejuni</italic> as populations between which there may be gene flow (migration). Within these populations the bacteria evolve through <italic>de novo</italic> mutation and horizontal gene transfer (recombination). We estimate the amount of mutation, migration and recombination, and use these estimates to assign probabilistically each human case to one of the source populations. From these population assignments we estimate the total amount of human disease attributable to each source.</p>" ]

[ "<title>Materials and Methods</title>", "<title>Human Isolates</title>", "<p>Stool samples were collected from 1,549 patients diagnosed with campylobacteriosis and notified through general practitioners and hospitals to the Preston Microbiology Services Laboratory in the Preston postcode district between January 1^st 2000 and December 31^st 2002. This covers an area of 968 km², comprising 403,000 people at the 2001 census, consisting of both urban (Preston, Leyland, Chorley, Garstang) and rural (Ribble estuary and Ribble valley) districts. As is the norm with campylobacteriosis, the cases we studied were sporadic in nature; there was no evidence for outbreaks. We followed previously published methods for multilocus sequence typing <italic>C. jejuni</italic>\n##REF##11136741##[21]##,##REF##15872261##[38]##. We obtained culturable, uncontaminated isolates of <italic>Campylobacter</italic> species from 1,353 samples, of which we identified 1,255 <italic>C. jejuni</italic>, 86 <italic>C. coli</italic> and 11 other species. One isolate tested positive for both <italic>C. jejuni</italic> and <italic>C. coli</italic> using the hippurate hydrolysis test and PCR. We fully sequenced all seven MLST loci (3,309 nucleotides in total) in 1,231 <italic>C. jejuni</italic> isolates, a sequencing success rate of 98%.</p>", "<title>Animal and Environmental Isolates</title>", "<p>We collated 1,145 <italic>C. jejuni</italic> isolates of animal and environmental origin from ten previously published studies ##UREF##1##[7]##, ##UREF##4##[16]##, ##REF##11136741##[21]##–##REF##14602588##[28]##. Where the sampling date was available, we excluded isolates sampled prior to 1990. We grouped the isolates by host or environmental origin as follows: chicken (515 isolates), cattle (282), sheep (160), pig (30), wild bird (44), wild rabbit (20), bathing beach (71) and environmental water sources (23). ##SUPPL##4##Table S1## gives a detailed breakdown of groups by source type and publication.</p>", "<title>Analysis of Molecular Variance</title>", "<p>To analyze the genetic heterogeneity within each group, we estimated Φ-statistics using analyses of molecular variance (AMOVA ##REF##1644282##[29]##). Genetic distance between a pair of isolates was defined as the number of loci, out of seven, at which they differed. We defined sub-groups using detailed sampling information from each publication (##SUPPL##4##Table S1##). <italic>E.g.</italic> we defined isolates sampled from calves versus cows milk as separate sub-groups within the cattle group. Isolates sampled from the same source type in different studies were also defined as separate sub-groups. Significance was assessed by permutation test, using 999 permutations. To analyze genetic differentiation between groups, over and above within-group differentiation, we performed pairwise nested AMOVA. Significance was assessed in the same fashion.</p>", "<title>Source Attribution</title>", "<p>The parameter of primary interest was the proportion, <italic>F_j</italic>, of human cases attributable to source population <italic>j</italic> (<italic>j</italic> = 1…<italic>n_g</italic>) where <italic>n_g</italic> = 8 was the number of putative source populations, and . If we knew the source of each case, we could estimate <italic>F</italic> directly using the multinomial likelihoodwhere <italic>N</italic> = 1,231 was the number of cases and <italic>G_i</italic> was the source of origin for case <italic>i</italic>. Our approach was Bayesian, so the posterior probability distribution for <italic>F</italic>, upon which inference is based, would bewhere <italic>p</italic>(<italic>F</italic>) is a prior probability distribution on the source attribution probabilities. We used a symmetric Dirichlet(1) prior on <italic>F</italic> in which all sources are considered equally likely <italic>a priori</italic>.</p>", "<p>Of course we did not know <italic>G</italic>, so we used a genetic model of DNA sequence evolution to co-estimate the probable source of human isolates based on their genotypes, <italic>H</italic>, as follows,where <italic>p</italic>(<italic>H</italic>|<italic>G</italic>) is the likelihood of the source assignments <italic>G</italic> under our evolutionary model.</p>", "<p>In our evolutionary model, we envisage the population of <italic>C. jejuni</italic> as a number of discrete islands: each source corresponds to a different island. Within each island the population is homogeneously mixing, and between islands there is migration. Migration rates may be higher between some islands than others, resulting in different levels of genetic differentiation. This is known as the migration matrix model ##REF##5708302##[39]##, a generalization of Wright's island model. We modeled the generation of new alleles within each MLST locus using the infinite alleles model ##REF##14156929##[40]##, and investigated two models of recombination between loci. In the first, the loci were assumed to be <italic>unlinked</italic> (inherited independently, or in linkage equilibrium), which is a computationally convenient but biologically unrealistic assumption. In the second, the loci were treated as <italic>linked</italic> (in linkage disequilibrium) using a model of recombination suitable for bacteria ##REF##10790416##[41]##.</p>", "<p>Human isolates were treated as a direct draw from one of the source populations. Therefore we assumed that the genotype of a human isolate would be representative of genotypes in the source population from which it was acquired. As a consequence, source attribution relies on the calculation of sampling probabilities; the likelihood that human isolate <italic>i</italic> was sampled from source population <italic>j</italic>. Unfortunately the complexity of the evolutionary model, in particular the <italic>linked</italic> model, renders direct calculation of the joint sampling probabilities <italic>p</italic>(<italic>H</italic>|<italic>G</italic>) impracticable, so we developed an approximation; full details of the approximation and the Markov Chain Monte Carlo sampler are provided in the Supplementary Methods (##SUPPL##7##Text S1##). To summarize, we used the animal and environmental isolates to estimate mutation, recombination and migration parameters. We then used these estimates together with all the genetic data (human and non-human genotypes) to jointly estimate the source attribution probabilities <italic>F</italic> and the source of human cases <italic>G</italic>. Except where stated otherwise, we used the mean of the posterior distribution for point estimates, and the (2.5%, 97.5%) quantiles of the posterior distribution for 95% credible intervals.</p>", "<title>Empirical Cross-Validation</title>", "<p>We employed empirical cross-validation to assess various of aspects of our approach: (i) the adequacy of the approximations made in order to perform inference (ii) the robustness to violations of the modeling assumptions, such as genetic heterogeneity within groups, and (iii) the sensitivity to sample size differences between groups. During each iteration of the empirical cross-validation, we artificially removed the population of origin of half the 1,145 animal and environmental isolates at random. We then used the other half to infer their origin, and evaluated the performance of the two models. This procedure was repeated over 100 iterations.</p>", "<p>We calculated several indicators of performance. The predicted proportion of isolates correctly assigned was calculated aswhere there were <italic>M</italic> = 100 simulation, <italic>N</italic> = 572 pseudo-human cases, <italic>n_g</italic> = 8 putative source populations, and was the posterior probability that population <italic>k</italic> is the source of pseudo-human case <italic>j</italic> in simulation <italic>i</italic>. The bias in the estimate of the proportion of pseudo-human cases attributable to population <italic>j</italic> was calculated aswhere was the actual proportion attributable to population <italic>j</italic> in simulation <italic>i</italic>, and was the point estimate, <italic>i.e.</italic> the mean of the posterior distribution. The root mean squared error, which measures the variance of the point estimate, was calculated as\n</p>", "<title>Analysis of Robustness</title>", "<p>As an additional test of robustness to the potentially confounding effects of genetic heterogeneity within the putative source populations, we repeated the source attribution analysis using subsets of the animal and environmental isolates, using the <italic>linked</italic> model only. The idea was to study the effect of removing whole sub-groups of isolates that were derived from the same source type or publication, as defined in ##SUPPL##4##Table S1##. We conducted 100 simulations in order to generate samples of the non-human isolates in which 20% or more of the isolates were excluded, a whole sub-group at a time. The simulations were conducted as follows:</p>", "<p>For source population <italic>j</italic>, sort the sub-groups (defined by ##SUPPL##4##Table S1##) into descending order. Let <italic>n</italic> be the total number of isolates in population <italic>j</italic>, let <italic>x</italic>: = <italic>n</italic>/5 be the minimum number to exclude and let <italic>i</italic>: = 1 index the sub-groups.</p>", "<p>With probability <italic>x</italic>/<italic>n</italic>, or if <italic>n</italic>−<italic>n_i</italic>&lt;<italic>x</italic>, remove all isolates of sub-group <italic>i</italic> and let <italic>x</italic>: = <italic>x</italic>−<italic>n_i</italic>.</p>", "<p>Let <italic>i</italic>: = <italic>i</italic>−1 and <italic>n</italic>: = <italic>n</italic>−<italic>n_i</italic>. Repeat from (2) until <italic>x</italic>≤0.</p>", "<p>On average, this procedure generated subsets in which 24.5% of isolates were excluded. Each of the 100 simulated subsets of the non-human isolates was used to infer the proportion of human cases attributable to each source. ##SUPPL##1##Figure S2## illustrates the difference in the simulation schemes between the empirical cross-validation and the analysis of robustness.</p>", "<title>Genotypes Unique to Humans</title>", "<p>We performed two re-sampling procedures to compare the number of unique genotypes in human isolates to the number in other groups. The aim was to scrutinize two modeling assumptions: (i) that human isolates are merely a sample of <italic>C. jejuni</italic> isolates found in the putative source populations, and (ii) that the major source populations have been identified. In the first procedure, we removed one non-human group, <italic>e.g.</italic> chicken, from the “pool” of non-human isolates and calculated the number of unique genotypes by reference to the pool. We sampled a subset of human isolates, equal in size to the number of chicken isolates, and calculated the number of unique genotypes by reference to the same pool. We repeated the sampling of human isolates 100 times to generate a distribution for the number of genotypes unique to humans, which we compared to the number of genotypes unique to chicken. Because of assumption (i) we expect humans to exhibit fewer unique genotypes.</p>", "<p>In the second re-sampling procedure, we removed half of the isolates belonging to a non-human group, <italic>e.g.</italic> chicken, leaving the rest in the pool in order to emulate the status of human isolates, which we assumed are merely a sample of isolates found in the non-human source populations. We sampled a subset of human isolates equal in number, and calculated the number of genotypes unique to chicken and humans, by reference to the same pool. We repeated the procedure 100 times to generate a distribution of the number of genotypes unique to chickens and humans. Violation of assumptions (i) or (ii) could lead to an excess of genotypes unique to humans.</p>", "<title>Data Deposition</title>", "<p>All newly-sequenced multi-locus sequence types are available for download from pubMLST.org/campylobacter.</p>" ]

[ "<title>Results</title>", "<title>Diversity and Differentiation in <italic>C. jejuni</italic> Populations</title>", "<p>We observed 256 distinct genotypes (or sequence types, STs) in the 1,231 human isolates. The frequency of genotypes was highly skewed, with 20% of STs accounting for 80% of cases. The two most frequent genotypes (STs 21 and 257) made up a quarter of cases alone, while 182 genotypes were observed once only. There were 375 distinct STs in the 1,145 animal and environmental isolates, and overlap with the human genotypes was extensive. Six STs featured in both the human and non-human lists of ten most common genotypes (STs 19, 21, 45, 50, 53, 61). However, nearly a quarter of human cases (278) exhibited genotypes exclusive to humans (189 STs), most of those at low frequency. The most abundant human-specific genotypes were ST 572 (14 cases) and ST 584 (19 cases). Over a third of non-human isolates (440) possessed genotypes absent from our human sample (308 STs).</p>", "<p>Certain genotypes common in non-human isolates were host-restricted to varying degrees. For example, ST 403 was the most prevalent in pigs (5/30 isolates), but absent from other non-human groups. ST 61 is common in ruminants (cattle and sheep) but rare in other groups, while ST 45 was frequent in all the non-human reservoirs except pig and sand. At the level of individual loci, many alleles were frequently observed in a range of animal and environmental sources. Because of the large overlap in genetic variation between <italic>C. jejuni</italic> reservoirs, our method utilizes differences in gene frequency, rather than allele presence or absence <italic>per se</italic>. By pooling samples of <italic>C. jejuni</italic> from similar sources (<italic>e.g.</italic> chicks and chicken meat/offal) and several studies, we intended to improve inference by increasing sample size. See ##SUPPL##4##Table S1## for details. However, by combining samples this way, we implicitly assumed that within each group (chicken, cattle, sheep, pig, bird, rabbit, sand and water) gene frequencies are consistent across sources and across studies. To test this assumption, and to quantify genetic differentiation between groups, we used analyses of molecular variance (AMOVA ##REF##1644282##[29]##).</p>", "<p>AMOVA quantifies genetic differentiation within and between groups using Φ-statistics, which measure the correlation in gene frequencies within sub-populations relative to the total population. A smaller value of Φ indicates lower genetic differentiation between the populations. ##TAB##0##Table 1## shows Φ<italic>_SG</italic>, the genetic differentiation within each group (<italic>e.g.</italic> chicken) between isolates of a different sub-group: <italic>i.e.</italic> source type (<italic>e.g.</italic> chick <italic>vs</italic> chicken meat/offal) or published study. Except for the sand group, there was significant heterogeneity within the groups that comprised more than one source type or study. Genetic differentiation between sub-groups ranged from 2.4% (cattle) to 23.2% (pig). This suggests that gene frequencies vary significantly between similar sources and between different studies of the same source.</p>", "<p>In order to assign human cases to source populations with any degree of accuracy, there must be genetic differentiation between the groups, <italic>over and above</italic> within-group heterogeneity. We estimated that quantity, Φ<italic>_GT</italic>, using nested AMOVA between pairs of groups. ##TAB##0##Table 1## shows the results. For each pair of groups, Φ<italic>_GT</italic> is displayed below the diagonal and the associated <italic>p</italic>-value above the diagonal. All groups were significantly differentiated from at least one other group in this way, with Φ<italic>_GT</italic> ranging from 4.4% (chicken <italic>vs</italic> cattle) to 26.2% (sheep <italic>vs</italic> pig). However, there were some major groups that were not significantly differentiated. Notably, Φ<italic>_GT</italic> was 0.1% for cattle <italic>vs</italic> sheep (<italic>p</italic> = 0.182), which suggested it would be difficult to distinguish human cases attributable to these two groups.</p>", "<p>The preliminary analyses of the animal and environmental <italic>C. jejuni</italic> isolates presented several potential concerns. There was significant variation in gene frequencies within groups, probably caused by the heterogeneous nature of the studies from which the non-human isolates were drawn, and the inherently stochastic nature of the epidemic process. This could distort the gene frequency information upon which source assignment relies, and cause higher than expected linkage disequilibrium between loci. AMOVA also showed that genetic differentiation between groups was weak in some cases. Within-group heterogeneity could therefore obscure or potentially distort the signal of differentiation between groups. Another concern was that the large differences in sample size between non-human groups, which reflect a tendency among researchers to preferentially sample certain hosts, could bias the source assignment. To investigate the sensitivity of our method to these effects, and to test its robustness to violating the assumption of homogeneous mixing within groups, we performed empirical cross-validation.</p>", "<title>Empirical Cross-Validation</title>", "<p>Traditional methods that can assign large numbers of individuals to populations based on their genotype tend to assume that loci provide independent sources of information ##REF##12586722##[30]##,##REF##12930761##[31]##. In other words, they assume that gene frequencies between loci are uncorrelated in the source populations. While this simplifying assumption is computationally convenient, it may not be appropriate for <italic>C. jejuni</italic> because of appreciable linkage disequilibrium between MLST loci ##REF##16044246##[19]##. Therefore we developed two models, one in which loci were assumed to be <italic>unlinked</italic> (<italic>i.e.</italic> independent, or in linkage equilibrium) and another in which loci were <italic>linked</italic> (<italic>i.e.</italic> in linkage disequilibrium); see <xref ref-type=\"sec\" rid=\"s4\">Materials and Methods</xref>. We used empirical cross-validation to scrutinize both.</p>", "<p>In each of 100 simulations, we removed the source information from half the non-human isolates, chosen at random. These we termed the pseudo-human cases. We used our <italic>unlinked</italic> and <italic>linked</italic> models to assign the source of the pseudo-human cases using the other non-human isolates. ##TAB##1##Table 2## shows that the two models differed considerably in performance. On average, the <italic>unlinked</italic> model correctly assigned 52% of the pseudo-human cases (using the rule that each case is assigned to its most probable source), whereas the <italic>linked</italic> model correctly assigned 64%. The <italic>linked</italic> model was well-calibrated in the sense that its estimated success rate was 64% on average, whereas the <italic>unlinked</italic> model grossly over-estimated its success rate (82% on average). We used a number of performance indicators to measure the ability of each model to correctly estimate the total proportion of pseudo-human cases attributable to a given source (see ##TAB##1##Table 2##). The parameter estimates obtained by using the <italic>linked</italic> model generally exhibited lower bias and smaller variance (measured by root mean squared error, RMSE) than those obtained using the <italic>unlinked</italic> model. The <italic>linked</italic> model also out-performed the <italic>unlinked</italic> model in coverage, which we defined as the number of simulations, out of 100, in which the 95% credible interval for the proportion of cases attributable to a given source enveloped the true value. For seven out of eight groups, the <italic>linked</italic> model obtained the target coverage of 95 or above. Coverage was 93 for chicken; the small negative bias suggests this may have been caused by slightly under-estimating the proportion of pseudo-human cases attributable to chicken.</p>", "<p>In the empirical cross-validation the <italic>linked</italic> model performs well despite the potential concerns due to heterogeneity within the animal and environmental groups, and differences in sample size. Most importantly, it is well-calibrated in assigning isolates to source populations, and estimating the overall proportion of cases attributable to each source. In contrast, the <italic>unlinked</italic> model assigns fewer isolates to source populations correctly, and is very poorly calibrated. This underlines the importance of adequately modeling recombination in the study of pathogen evolution. Clearly the computational efficiency gains made by assuming independent inheritance among loci in the <italic>unlinked</italic> model are out-weighed by its poor performance. Therefore we use the <italic>linked</italic> model for our analysis proper.</p>", "<title>Tracing the Source of Human Cases</title>", "<p>We applied our novel method to the 1,231 newly-sequenced human isolates from Lancashire, England. For every case, the assignment probability was calculated for each source population (chicken, cattle, sheep, pig, bird, rabbit, sand, water), and the total proportion of cases attributable to each source was estimated. We found that the vast majority (96.6%) of human cases are attributable to populations of <italic>C. jejuni</italic> carried by livestock (95% credible interval 92.7–98.8%) as opposed to wild animals (2.3%) or environmental isolates (1.1%). ##FIG##0##Figure 1## shows a breakdown of attribution by source; errors bars indicate the 95% credible intervals. We estimated that chicken is the source of infection in the majority (56.5%) of cases (95% C.I. 51.1–61.8%, see ##SUPPL##5##Table S2##), followed by cattle (35.0%) and sheep (4.3%). The 95% credible intervals were wider for cattle (20.8–43.2%) and sheep (0.1–17.5%) than other groups, which reflects the greater difficulty in distinguishing these populations of <italic>C. jejuni</italic> from one another. We found that pig is unlikely to be the source of <italic>C. jejuni</italic> infection in humans (0.8% of cases).</p>", "<p>Of the two groups of wild animals we studied, bird and rabbits, there was somewhat more support for a wild bird origin of human <italic>C. jejuni</italic> (1.7%) than rabbit (0.6%), although the credible intervals (0.1–5.5% and 0.0–3.7% respectively) were largely overlapping. There was very little support for an environmental origin of human infections. Even so, the results suggested that infection with <italic>C. jejuni</italic> found in environmental water sources was more likely (0.9%) than infection with <italic>C. jejuni</italic> isolated from bathing beaches (0.2%), which was the least likely of all sources. Overall, the analysis reported that with 98.3% probability, chicken is the primary, and cattle the secondary source of human infections in our study.</p>", "<p>The posterior probability of source of infection was estimated for each patient in our study; ##FIG##1##Figure 2## illustrates the results. The source populations are color-coded as in ##FIG##0##Figure 1##. Cases are arranged horizontally, and the vertical column space occupied by each color represents the posterior probability of infection from that source. The dominant color in any column indicates the most likely source for a particular case. The principal distinction in human cases is between those attributed to chicken versus ruminants (cattle and sheep). Most cases lie on a continuum between assignment to ruminants and to chicken. The existence of this continuum, as opposed to a clear separation, emphasizes the overlap in genotypes between these source populations, and the advantage of a probabilistic approach to assignment. Some common genotypes were strongly assigned to ruminants (<italic>e.g.</italic> ST 48, 86 cases, posterior probability [<italic>Pr</italic>] = 0.91) and others to chicken (ST 104, 64 cases, <italic>Pr</italic> = 0.93). But within ruminants, it is harder to distinguish cattle from sheep sources. This is borne out by the strong correlation among cases between cattle and sheep assignment probabilities (<italic>ρ</italic> = 0.80).</p>", "<p>In some cases, there is moderate or strong support for a source that is generally found to be rare. For example, there were six cases of ST 403, with a moderately high assignment probability to pig of 0.37. Except for the human isolates, we observed ST 403 only in the pig population. However, because the evidence overall suggests that pig is an unlikely source of infection for humans, and because of the genetic similarity to cattle genotypes (<italic>e.g.</italic> ST 933), it is marginally more likely under the model that cattle is the source of these cases (<italic>Pr</italic> = 0.46). Although it is most probable, on a case-by-case basis, that the source of infection was cattle, when considered together we would expect the source of infection to have been cattle in 2.7 of those cases, pig in 2.2 cases and chicken in 0.6 cases. Another example of this phenomenon is found in birds. There are 28 cases, of which ST 508 was the most common genotype, with an assignment probability to birds greater than 10%, but a larger assignment probability to another source, usually chicken. On an individual basis none of these cases would be assigned to birds, but taken together we estimate that the source of infection was birds in 5.6 of them, chicken in 10.6, cattle in 5.5 and water in 3.8. Overall, the source probabilities in ##FIG##0##Figure 1## and ##SUPPL##5##Table S2## suggest that of the 1,231 human cases, the source of infection was chicken in 696.6 cases, cattle in 432.1, sheep in 53.5, bird in 20.5, water in 10.9, pig in 10.3, rabbit in 7.9 and sand in 2.2.</p>", "<p>Sometimes it is useful to assign a case to a single source, in which case the optimal strategy is to attribute it to the source with highest assignment probability <italic>a posteriori</italic>. We estimate that 76.5% of human cases would be correctly assigned by this procedure. Earlier we showed that this quantity, which is the average maximum source attribution probability per case, was well-calibrated during empirical cross-validation. When cases are assigned to sources in this fashion, most are assigned to chicken (722) or cattle (503). None are assigned to sheep, because ruminant-associated isolates are assigned preferentially to cattle. A small number are assigned to pig, bird and water (three in each case). For example, STs 1286, 1927 and 2973 were the genotypes most strongly assigned to environmental water, pig and wild bird respectively (<italic>Pr</italic> = 0.58, 0.65, 0.87). Interestingly, all three genotypes were human-specific, and each was found in a single patient only. In the case of ST 1286, there was also considerable support for a wild bird origin (<italic>Pr</italic> = 0.35), an observation that may reflect the low genetic differentiation detected between these sources (##TAB##0##Table 1##). ##SUPPL##6##Table S3## gives a detailed breakdown of source attribution probabilities by sequence type.</p>", "<title>Robustness to Within-Group Heterogeneity</title>", "<p>Our collection of animal and environmental isolates which we collated from previously-published studies ##UREF##1##[7]##, ##UREF##4##[16]##, ##REF##11136741##[21]##–##REF##14602588##[28]## were non-ideal in several respects. AMOVA revealed significant variation between isolates from the same group that originated in different sub-groups – <italic>i.e.</italic> different source types or studies. Such genetic structuring will cause higher than expected linkage disequilibrium within groups, and may distort the gene frequencies upon which source attribution relies. Although empirical cross-validation showed that the <italic>linked</italic> model was robust to these effects, the full extent of the difficulty caused by within-group heterogeneity may have been masked because individual isolates were assigned to the pseudo-human class independently, and without reference to their sub-group. Therefore we performed additional simulations in which whole sub-groups of isolates were removed, and the human isolates re-analyzed based on the reduced set of animal and environmental isolates. In each simulation, we removed at least 20% of the animal and environmental isolates, 24.5% on average. ##SUPPL##1##Figure S2## illustrates the simulation scheme and contrasts it to the simulations used in empirical cross-validation.</p>", "<p>Our main conclusions are robust to genetic heterogeneity within the source populations. ##SUPPL##2##Figure S3## summarizes the analysis of robustness by plotting the point estimate and 95% credible interval of various parameters based on the 100 simulations and the full data. In all of the 100 simulated datasets analyzed, chicken was found to be the primary source of human infections. ##SUPPL##2##Figure S3A## shows that in the majority of simulations, chicken accounted for more than 50% of human disease. The conclusion that ruminants are the second most important source of human infection was also supported by the analysis (##SUPPL##2##Figure S3B##). Despite the low genetic differentiation between cattle and sheep, as witnessed by the AMOVA results, the finding that cattle account for considerably more disease than sheep is surprisingly robust to re-sampling the non-human isolates. In ##SUPPL##2##Figure S3C##, the posterior median (rather than the mean) is used to illustrate that in 87 simulations, a greater proportion of human cases were attributed to cattle than to sheep.</p>", "<p>The greatest effect of the re-sampling of non-human isolates was seen in the proportion of human cases attributed to the bird group. ##SUPPL##2##Figure S3D## shows that in a minority of simulations (16 out of 100), the proportion of cases attributed to birds leapt ten-fold to around 20%, promoting it to the second or third most important source, compared to fourth in the analysis of the full data. Of these 16 simulations, there was a significantly lower number of bird isolates (<italic>p</italic> = 0.005) and a significantly higher number of chicken isolates (<italic>p</italic> = 0.040) compared to the other simulations, the relevance of which is that the chicken and bird groups were shown by AMOVA to exhibit extremely low genetic differentiation (##TAB##0##Table 1##). While these results demonstrate that more intense sampling of the smaller groups, particularly birds, is highly desirable, our main conclusions are supported by the vast majority of re-sampled datasets, indicating a satisfactory level of robustness to within-group heterogeneity.</p>", "<title>Existence of Undiscovered Source Populations</title>", "<p>A tacit assumption in our study, and in the ongoing sampling of <italic>C. jejuni</italic> populations, is that the major reservoirs have been identified. However, if a major source of human disease were undiscovered, we would expect to see an excess of genotypes unique to humans. In our study we observed 189 genotypes unique to humans. Of the 1,231 human cases, 278 possessed genotypes absent in the non-human isolates, but most of these (238 cases) were re-assortments of alleles or allele fragments that were present in the non-human isolates. In 254 cases, they differed at three loci or fewer to a non-human isolate. Out of 531 single nucleotide polymorphisms in humans, 40 were absent from the non-human samples. Of those, all were rare except an adenosine at nucleotide 448 in the <italic>glnA</italic> locus (12 copies), and a cytosine at nucleotide 93 in the <italic>tkt</italic> locus (13 copies). Two human-specific STs (572 and 584) had appreciable frequency (14 and 19 cases respectively).</p>", "<p>It is difficult to quantify exactly what would constitute an excess of genotypes unique to humans. We employed a re-sampling procedure to compare the number of unique genotypes in human isolates compared to other groups, controlling for sample size. When sets of human isolates were drawn, equal in size to the number of chicken isolates (515), we observed fewer unique genotypes on average among human isolates (104.4) than among chicken isolates (153), where uniqueness was determined by reference to the “pool” of other non-human and non-chicken isolates. The same pattern was observed when comparing humans to cattle and birds, but not sheep (##SUPPL##3##Figure S4A##). Sheep isolates are genetically similar to cattle, which may explain why humans exhibit no more unique genotypes than do sheep. The observation that cattle isolates appear to exhibit relatively more unique genotypes than sheep suggests there might be an effect of sample size, or that sheep isolates are a subset of cattle isolates.</p>", "<p>A second re-sampling procedure was designed to emulate the status of humans as a sample of isolates drawn from the putative source populations. Taking each non-human group in turn, half the isolates were removed, leaving the other half in the pool, and the number of genotypes unique to the removed isolates was calculated. A set of human isolates was drawn of equal number, and the number of unique genotypes calculated relative to the same pool. The whole procedure was repeated 100 times. If major source populations remained to be discovered, or if humans acted as a reservoir of <italic>C. jejuni</italic> rather than a terminus in the transmission chain, then an excess of genotypes unique to humans would be expected. However, in these simulations the distribution of the number of genotypes unique to humans and the non-human groups overlapped to a great extent (##SUPPL##3##Figure S4B##). Therefore while the more abundant STs and SNPs unique to humans deserve further attention, on the whole there is little indication that another major, genetically distinct, reservoir of human infection remains undiscovered.</p>" ]

[ "<title>Discussion</title>", "<p>Our results show that livestock are the principal source of <italic>C. jejuni</italic> infection in Lancashire, England. The vast majority of those human infections can be attributed to populations of <italic>C. jejuni</italic> found in chicken and cattle. These findings immediately lend weight to the suggestion that the incidence of campylobacteriosis in humans could be significantly reduced by intervention strategies targeted at livestock ##UREF##7##[32]##,##REF##17145967##[33]##, chiefly the strict enforcement of on-farm biosecurity measures including disinfecting farm premises and water supplies, restricting access to livestock to essential personnel, minimizing the use of invasive practices such as thinning in chickens, securing premises from wild birds and mammals, and protecting food supplies from bacterial contamination.</p>", "<p>Moreover, our results are informative about the likely mode of transmission of <italic>C. jejuni</italic> to patients in our study. The genetic analysis identifies the source of infection, rather than the transmission route. The importance of livestock as a reservoir for human disease is consistent with food-borne transmission, but alternative pathways, such as ingestion of animal feces or contamination of water by human or animal waste, must also be considered. Our findings show that, while we can detect cases of human infection with isolates of an environmental or wild animal origin, such cases are rare, and this is surprising if pathways other than food-borne transmission are important. Therefore the dual observations that (i) livestock are a frequent source of human disease isolates and (ii) wild animals and the environment are not, strongly support the notion that preparation or consumption of infected meat and poultry is the dominant transmission route.</p>", "<p>Transmission through the food chain can be controlled in a number of ways. Preventing cross-contamination of carcasses during processing is an effective measure ##REF##17368847##[34]## that can be achieved, for example, by minimizing meat contamination with animal feces, treating carcasses with antimicrobial agents, sterilizing equipment, and careful management of animals or flocks known to be infected. Meat products can be treated directly, for example by freezing or irradiation ##REF##10636787##[35]##. Promoting better standards of food hygiene during preparation and cooking is also an effective measure ##UREF##7##[32]##,##REF##17368847##[34]##.</p>", "<p>Our results pertain to sporadic disease; we know that contamination of drinking water occasionally causes outbreaks ##REF##12825731##[11]##–##REF##17469655##[13]##. The lack of evidence for pigs as a source of <italic>C. jejuni</italic> infection is consistent with their greater susceptibility to <italic>C. coli</italic>\n##REF##16272508##[36]##. Since <italic>C. coli</italic> causes less than 10% of sporadic campylobacteriosis, pigs must be a less important source of infection than chicken and cattle.</p>", "<p>We found considerable variation in the genetic make-up of <italic>C. jejuni</italic> populations sampled from similar sources (<italic>e.g.</italic> chicks <italic>vs</italic> chicken meat/offal) and between different populations from the same source type. This variation may reflect functional differences between <italic>C. jejuni</italic> even from closely related sources, or it may reflect stochastic differences in gene frequency over time or space. The epidemic process may increase variation in gene frequencies because hosts sampled locally are infected from the same source, causing non-independence within samples. We found our method was robust to this heterogeneity, but it is reasonable to think that inference would be improved by sensibly modeling the phenomenon. How to do so is unclear: one option is to split heterogeneous groups into further sub-categories, but that increases the number of parameters in the model and may reduce statistical efficiency or lead to over-fitting. Comprehensive sampling of putative source populations in parallel to human sampling is most desirable, and such studies are on-going by groups in Scotland, New Zealand and the USA.</p>", "<p>Assigning the source of human isolates based on genotype has been attempted before in <italic>C. jejuni</italic>. Our results are in contrast to those of Champion <italic>et al.</italic>\n##UREF##5##[18]## who, using a Bayesian phylogenetic approach applied to comparative genomic hybridization data, found that <italic>C. jejuni</italic> isolates can be divided into livestock and non-livestock clades, with 55.7% of human isolates falling into the non-livestock clade. The existence of these clades was supported by high posterior probabilities, close to <italic>Pr</italic> = 1. The implications of such findings would be dramatic, however there are difficulties with the approach. The principal problem is that <italic>C. jejuni</italic> is known to be highly recombining which means that different genes, or even different parts of the same gene, will have different phylogenetic histories. Inferring a single phylogenetic tree for the whole genome is therefore a case of gross model mis-specification, and the resulting phylogeny is difficult to interpret in any meaningful way ##REF##11014833##[37]##.</p>", "<p>Many pathogens exist as weakly differentiated, genetically overlapping populations or strains between which there is frequent gene flow and within which there is frequent recombination. Such strains may be epidemiologically relevant, but it will be difficult to find stable, well-differentiated genetic markers, the standard tools of molecular epidemiology, to type them unambiguously. In this paper the method we developed addressed the problem in <italic>C. jejuni</italic> by assigning isolates to source populations probabilistically. We used a simple epidemiological model, in which we inferred the probability of infection with each source, to efficiently combine information over cases. That model could be readily extended in the general linear model framework to employ covariates, such as age, sex or host genotype, were they available.</p>", "<p>In conclusion, we have used a novel population genetics approach to identify the source of infection of the zoonotic pathogen <italic>Campylobacter jejuni</italic>. We found that cases of human infection in our study were overwhelmingly attributable to bacteria characteristic of those colonizing animals farmed for meat and poultry, based on genetic similarity. We hope that demonstrating the importance of livestock as reservoirs of <italic>Campylobacter</italic> infectious to humans will add impetus to initiatives aimed at controlling food-borne pathogens.</p>" ]

[]

[ "<p><bold>¤a:</bold> Current address: Department of Human Genetics, University of Chicago, Chicago, Illinois, United States of America</p>", "<p><bold>¤b:</bold> Current address: Université d'Avignon, IUT STID, Site Agroparc, Avignon, France</p>", "<p>Conceived and designed the experiments: JC SG EB AF CAH. Performed the experiments: AJHL. Analyzed the data: DJW EG AJHL PF PJD. Contributed reagents/materials/analysis tools: DJW EG PF PJD. Wrote the paper: DJW EG AJHL JC SG EB AF PF CAH PJD.</p>", "<p>\n<italic>Campylobacter jejuni</italic> is the leading cause of bacterial gastro-enteritis in the developed world. It is thought to infect 2–3 million people a year in the US alone, at a cost to the economy in excess of US $4 billion. <italic>C. jejuni</italic> is a widespread zoonotic pathogen that is carried by animals farmed for meat and poultry. A connection with contaminated food is recognized, but <italic>C. jejuni</italic> is also commonly found in wild animals and water sources. Phylogenetic studies have suggested that genotypes pathogenic to humans bear greatest resemblance to non-livestock isolates. Moreover, seasonal variation in campylobacteriosis bears the hallmarks of water-borne disease, and certain outbreaks have been attributed to contamination of drinking water. As a result, the relative importance of these reservoirs to human disease is controversial. We use multilocus sequence typing to genotype 1,231 cases of <italic>C. jejuni</italic> isolated from patients in Lancashire, England. By modeling the DNA sequence evolution and zoonotic transmission of <italic>C. jejuni</italic> between host species and the environment, we assign human cases probabilistically to source populations. Our novel population genetics approach reveals that the vast majority (97%) of sporadic disease can be attributed to animals farmed for meat and poultry. Chicken and cattle are the principal sources of <italic>C. jejuni</italic> pathogenic to humans, whereas wild animal and environmental sources are responsible for just 3% of disease. Our results imply that the primary transmission route is through the food chain, and suggest that incidence could be dramatically reduced by enhanced on-farm biosecurity or preventing food-borne transmission.</p>", "<title>Author Summary</title>", "<p>\n<italic>C. jejuni</italic> is a bacterium commonly found in the guts of birds and mammals. In humans, it is responsible for causing more gastro-enteritis than any other identified bacterial species. Humans may contract campylobacter from a variety of sources. Eating raw or undercooked meat or poultry, and poor food hygiene that leads to cross-contamination of uncooked food, can cause human disease. However, humans may be exposed to the feces of infected wild animals, and campylobacter can survive in water. Contamination of drinking water can lead to outbreaks, and previous genetic studies have suggested that livestock are not the principal source of human infection. We extracted campylobacter DNA from patients and compared it to campylobacter DNA found in livestock, wild animals, and the environment. We developed a new evolutionary model to identify the most probable source populations. In 97% of cases, we identified chicken, cattle, or sheep as the source of infection. Very few cases were attributable to campylobacter found in wild animals or the environment. Our results imply that the primary transmission route is the food chain and also add new impetus to measures that reduce infection in livestock and prevent food-borne transmission.</p>" ]

[ "<title>Supporting Information</title>" ]

[ "<p>We would like to thank Bev Abram, Daniel Falush, Nigel French, Bill Hanage, Martin Maiden, Noel McCarthy, Gil McVean, Jonathan Pritchard, Sam Sheppard and Brian Spratt for useful advice and stimulating discussion on various aspects of this work.</p>" ]

[ "<fig id=\"pgen-1000203-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pgen.1000203.g001</object-id><label>Figure 1</label><caption><title>Estimated proportion of human cases attributable to animal and environmental sources.</title><p>Error bars indicate the 95% credible interval for each source.</p></caption></fig>", "<fig id=\"pgen-1000203-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pgen.1000203.g002</object-id><label>Figure 2</label><caption><title>Probability of source for human cases.</title><p>The source probability for 1,231 human cases (vertical columns) is depicted for Chicken (yellow), Cattle (red), Sheep (blue), Pig (pink), Bird (green), Rabbit (purple), Sand (beige) and Water (cyan). The isolates have been ordered horizontally to aid visualization.</p></caption></fig>" ]

[ "<table-wrap id=\"pgen-1000203-t001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pgen.1000203.t001</object-id><label>Table 1</label><caption><title>Genetic differentiation within and between groups.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td colspan=\"9\" align=\"left\" rowspan=\"1\">Genetic differentiation (Φ<italic>_SG</italic>) within groups</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">CHICKEN</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CATTLE</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SHEEP</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">PIG</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">BIRD</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">RABBIT</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SAND</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">WATER</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Φ<italic>_SG</italic>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>11.8%</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>2.4%</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>11.2%</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>23.2%</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>9.8%</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>p</italic>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.001</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.003</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.001</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.003</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.003</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.969</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr></tbody></table><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td colspan=\"9\" align=\"left\" rowspan=\"1\">Genetic differentiation (Φ<italic>_GT</italic>) between groups, over and above within-group differentiation</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Φ<italic>_GT</italic>\\<italic>p</italic>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CHICKEN</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CATTLE</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SHEEP</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">PIG</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">BIRD</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">RABBIT</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">SAND</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">WATER</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">CHICKEN</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.001</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.003</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.001</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.319</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.807</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.150</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.270</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">CATTLE</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>4.4%</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.182</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.001</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.049</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.170</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.053</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.001</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">SHEEP</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>4.4%</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.1%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.006</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.145</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.414</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.102</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.544</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">PIG</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>18.6%</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>24.8%</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>26.2%</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.108</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.164</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.089</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.001</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">BIRD</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>9.6%</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8.1%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">13.7%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.756</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.001</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.293</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">RABBIT</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.1%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.9%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">11.8%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.001</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.353</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">SAND</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.0%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8.3%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.6%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">22.2%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>4.7%</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>7.4%</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.001</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">WATER</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>12.7%</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">11.6%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>13.2%</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.3%</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>11.3%</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pgen-1000203-t002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pgen.1000203.t002</object-id><label>Table 2</label><caption><title>Performance of the models during empirical cross-validation.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Unlinked model</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Linked model</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>Proportion of isolates correctly assigned</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Actual</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.52</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.64</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Predicted</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.82</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.64</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>Bias</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Chicken</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−0.10</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−0.03</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Cattle</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−0.13</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.00</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Sheep</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.20</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.03</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Pig</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.01</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.00</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Bird</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.00</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−0.01</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Rabbit</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.00</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.01</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Sand</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−0.01</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.00</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Water</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.03</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.00</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>RMSE</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Chicken</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.11</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.04</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Cattle</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.14</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.07</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Sheep</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.21</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.07</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Pig</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.01</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.01</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Bird</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.03</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.02</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Rabbit</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.02</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.02</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Sand</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.01</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.01</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Water</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.05</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.01</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>Coverage</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Chicken</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">12</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">93</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Cattle</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">19</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">97</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Sheep</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">97</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Pig</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">76</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">99</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Bird</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">86</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">99</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Rabbit</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">73</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Sand</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">93</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">99</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Water</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">84</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">99</td></tr></tbody></table></alternatives></table-wrap>" ]

[ "<inline-formula></inline-formula>", "<disp-formula></disp-formula>", "<disp-formula></disp-formula>", "<disp-formula></disp-formula>", "<disp-formula></disp-formula>", "<inline-formula></inline-formula>", "<disp-formula></disp-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<disp-formula></disp-formula>" ]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"pgen.1000203.s001\"><label>Figure S1</label><caption><p>Migration and mutation probabilities in the animal and environmental samples. For each <italic>C. jejuni</italic> reservoir, the pie chart shows the predictive probability that a newly-sampled allele is a novel mutant (black segment) or identical to one already observed in the same or another population (colored segment: Chicken-yellow, Cattle-red, Sheep-blue, Pig-pink, Bird-green, Rabbit-purple, Sand-beige, Water-cyan). The estimated probability of recombination in each reservoir sample was 0.057, 0.048, 0.046, 0.15, 0.10, 0.061, 0.12 and 0.054 respectively.</p><p>(0.78 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pgen.1000203.s002\"><label>Figure S2</label><caption><p>The simulation schemes used for (A) empirical cross-validation and (B) analysis of robustness. Each row represents a non-human isolate; isolates are ordered vertically by source and sub-group (as defined by ##SUPPL##4##Table S1##), and colored by group. In (A) blank spaces represent isolates assigned to the pseudo-human group. Their source was inferred from the remaining non-human isolates. In (B) blank spaces represent isolates that were excluded, whole sub-groups at a time, from inferring the source of human isolates.</p><p>(2.69 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pgen.1000203.s003\"><label>Figure S3</label><caption><p>Analysis of robustness. For each parameter (proportion of cases attributable to (A) chicken (B) cattle+sheep (C) cattle <italic>vs.</italic> sheep (D) bird (E) water (F) pig (G) rabbit (H) sand), the point estimate and the 95% credible interval is plotted for the analysis of 100 simulations and the full data. The results are ordered vertically by the point estimate, for which the posterior mean was used except in (C) where the posterior median was used. The red dot indicates the analysis of the full data.</p><p>(1.80 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pgen.1000203.s004\"><label>Figure S4</label><caption><p>The number of genotypes unique to humans. Two re-sampling procedures were performed (see <xref ref-type=\"sec\" rid=\"s4\">Materials and Methods</xref>) to compare the number of genotypes unique to humans and other groups, controlling for sample size. The distribution of the number of unique genotypes is represented with box-and-whisker plots. (A) Humans exhibit fewer unique genotypes than non-human groups. (B) Humans exhibit no more unique genotypes than non-human groups that are partially represented in the pool of other non-human isolates.</p><p>(0.65 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pgen.1000203.s005\"><label>Table S1</label><caption><p>Source of animal and environmental isolates.</p><p>(0.11 MB DOC)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pgen.1000203.s006\"><label>Table S2</label><caption><p>Proportion of cases attributable to each source: summary of the posterior distribution of <italic>F</italic>.</p><p>(0.04 MB DOC)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pgen.1000203.s007\"><label>Table S3</label><caption><p>Posterior assignment probabilities by sequence type.</p><p>(0.57 MB DOC)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pgen.1000203.s008\"><label>Text S1</label><caption><p>Supplementary Methods.</p><p>(0.16 MB DOC)</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><fn id=\"nt101\"><p>Total genetic differentiation between isolates from two different groups, Φ<italic>_ST</italic>, equals approximately 1−(1−Φ<italic>_SG</italic>)(1−Φ<italic>_GT</italic>) where Φ<italic>_SG</italic> represents an average for the two groups. Significant Φ-statistics are printed in bold.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt102\"><p>The unlinked and linked models are defined in the <xref ref-type=\"sec\" rid=\"s4\">Methods</xref>. The predicted proportion of isolates correctly assigned assumes that isolates are assigned to their most probable source <italic>a posteriori</italic>. Bias, RMSE (root mean squared error) and coverage are reported for the proportion of isolates estimated to originate from each source. Coverage was defined as the number of simulations, out of 100, in which the true proportion fell inside the 95% credible interval.</p></fn></table-wrap-foot>", "<fn-group><fn fn-type=\"COI-statement\"><p>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p>This work was funded by the Higher Education Funding Council of England and the UK Department for Environment, Food and Rural Affairs as part of the Veterinary Training Research Initiative, and the EPSRC.</p></fn></fn-group>" ]

[ "<graphic id=\"pgen-1000203-t001-1\" xlink:href=\"pgen.1000203.t001\"/>", "<graphic id=\"pgen-1000203-t002-2\" xlink:href=\"pgen.1000203.t002\"/>", "<graphic xlink:href=\"pgen.1000203.g001\"/>", "<graphic xlink:href=\"pgen.1000203.g002\"/>", "<inline-graphic xlink:href=\"pgen.1000203.e001.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pgen.1000203.e002.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pgen.1000203.e003.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pgen.1000203.e004.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pgen.1000203.e005.jpg\" mimetype=\"image\" position=\"float\"/>", "<inline-graphic xlink:href=\"pgen.1000203.e006.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pgen.1000203.e007.jpg\" mimetype=\"image\" position=\"float\"/>", "<inline-graphic xlink:href=\"pgen.1000203.e008.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pgen.1000203.e009.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pgen.1000203.e010.jpg\" mimetype=\"image\" position=\"float\"/>" ]

[ "<media xlink:href=\"pgen.1000203.s001.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pgen.1000203.s002.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pgen.1000203.s003.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pgen.1000203.s004.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pgen.1000203.s005.doc\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pgen.1000203.s006.doc\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pgen.1000203.s007.doc\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pgen.1000203.s008.doc\"><caption><p>Click here for additional data file.</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Mother-to-child transmission (MTCT) of human immunodeficiency virus type 1 (HIV-1) is the primary mode of infection in children. In year 2007, an estimated 420.000 new infections occurred in children aged less than 15 years, most living in Sub-Saharan Africa ##UREF##0##[1]##.</p>", "<p>Several maternal parameters, including advanced clinical stage, low CD4+ T cell counts, high plasma viral load were associated with an increased risk of MTCT of HIV-1 (reviewed in ##REF##15332429##[2]##). There are controversial data concerning the role of viral phenotype in transmission. Although viruses using CXCR4 as coreceptor (X4 phenotype) can be transmitted when present in the mother, CCR5-using viruses (R5 phenotype) are the most frequently detected in newborns ##REF##10985308##[3]##, ##REF##11461678##[4]##.</p>", "<p>Evolution of HIV-1 coreceptor use during disease progression has been demonstrated in adults as well as children ##REF##9311827##[5]##, ##REF##9359702##[6]##. The evolution usually involves change from CCR5 use (R5 phenotype) to CXCR4 use alone (X4) or in combination with CCR5 (R5X4) and/or other minor coreceptors (multitropic viruses). CXCR4-using viruses can be isolated prior to or during progression to AIDS, however only from about one-half of patients with overt AIDS ##REF##7908672##[7]##, ##REF##8096374##[8]##, thus suggesting that R5 viruses obtained during clinical progression may differ in phenotypic characteristics from those obtained during the early stages of infection.</p>", "<p>Phenotypic variation characterizes R5 viruses, as demonstrated by their varying capacity to infect macrophages ##REF##16775320##[9]##, ##REF##1985204##[10]##, ##REF##11897037##[11]##, ##UREF##1##[12]## or their differential susceptibility to inhibition by CC-chemokines ##REF##9359702##[6]##, ##REF##8986820##[13]##, ##REF##12964118##[14]##. Studies on the entry of R5 viruses into cells expressing CCR5/CXCR4 chimeric receptors ##REF##15479822##[15]##, ##REF##14685050##[16]## showed that the differential susceptibility to inhibition by CC-chemokines depends on the mode of CCR5 use. In particular, it has been shown that during disease progression R5 viruses evolve to multiple chimeric receptor usage (called R5^broad), which in turn correlated with CD4+ T cell decline in the patient ##REF##15479822##[15]##. Evolution of the R5^broad phenotype was associated with decreased sensitivity to inhibition by the CC-chemokine RANTES ##REF##15479822##[15]##. The ability of a viral isolates to use one or more chimeric receptors is most probably a reflection of a more efficient usage of the CCR5 molecule, as suggested by our previous results demonstrating a higher infectivity of the wild-type CCR5 expressing cells with R5^broad compared to R5^narrow isolates ##REF##15479822##[15]##. In this study we asked the question whether phenotypic variation, implying different mode of CCR5 use in pregnant women could play a role in MTCT of HIV-1 and in turn, also in pediatric disease progression.</p>" ]

[ "<title>Materials and Methods</title>", "<title>Patients and virus isolation</title>", "<p>Our study population included a total of 59 HIV-1 seropositive women (24 transmitting and 35 non transmitting) and 28 infected children. Viral isolates from mother-child pairs were available in 21 cases. Additional seven children were included in the study, but the maternal samples were not available. Samples from non transmitting mothers and mother-child pairs were collected within the framework of two separate cohort studies in Northern Italy. One cohort consisted of 15 infected mother-child pairs and 17 non transmitting mothers who took part in a HIV-1 MTCT multicenter study from 1989 to 1994 ##REF##11461678##[4]##, ##REF##9359702##[6]##, ##REF##8249285##[17]##, ##REF##1786144##[18]##. The second cohort consisted of 9 infected mother-child pairs and 18 non transmitting mothers collected in the framework of MTCT multicenter study beginning in 1986 ##REF##10985308##[3]##, ##REF##2883489##[19]##. All samples were from Italian subjects and collected before the introduction of any MTCT preventive antiretroviral therapy. In 14 out of 16 children tested the infection was possibly acquired during the intrapartum period, since negative results were obtained by polymerase chain reaction (PCR) within the first week of life. Postnatal transmission was excluded in all cases, as none of the children was breast-fed. The Ethical Committee approved the use of samples according to national laws (in 1986, 1991 and 2008). An informed oral consent was obtained from the pregnant women and, in the case of the children, from the parents.</p>", "<p>The clinical stage of the women and their children (available for 21 transmitting and 24 non transmitting mothers and for 26 out of 28 children respectively), was defined according to the guidelines of the Centers for Disease Control (CDC) ##UREF##2##[20]##, ##UREF##3##[21]##. Most of the mothers were asymptomatic at the time of virus isolation, with the exception of three mothers: one non transmitting (clinical stage B, A206) and two transmitting mothers, one with moderate symptoms (clinical stage B, A225) and one with AIDS (clinical stage C, A130). CD4+ T cell counts were provided with the samples in many cases (for 17 transmitting and 28 non transmitting mothers) ##REF##10985308##[3]##, ##REF##8249285##[17]##. HIV-1 p24 antigenemia and plasma viral load were available for a limited number of women (33 and 17 mothers, respectively) and thus, were not included in the study analysis.</p>", "<p>HIV-1 infection of the children was determined by virus isolation and PCR ##REF##1786144##[18]##, ##REF##2059357##[22]##. Clinical and immunological data of the children were obtained throughout follow-up, i.e. until death or for at least 10 years (##TAB##0##Table 1##). Children were treated only with mono or dual antiretroviral therapy (ART). Highly active ART (HAART) was not administered to children in groups 1, 2 and 3, whereas children in group 4 were treated only after entering CDC category 3 or C.</p>", "<p>Virus isolation was performed from patient's PBMCs as previously described ##REF##1786144##[18]##, ##REF##2059357##[22]##. HIV-1 p24 antigen (Ag) positive culture supernatants, collected during first or second culture passage, were used to prepare virus stocks. Maternal viral isolates were collected during pregnancy, at delivery or within 5 months after delivery (##TAB##1##Table 2##). Virus was isolated from infected children at an age range between birth and 9 months.</p>", "<title>Infection of U87.CD4 cell lines expressing wild type and CCR5/CXCR4 chimeric receptors</title>", "<p>Virus stocks were used to infect human glioma U87.CD4 cells stably expressing the wild type chemokine receptors CCR5 or CXCR4, or the six CCR5/CXCR4 chimeric receptors as previously described ##REF##9359702##[6]##, ##REF##14685050##[16]##. Chimeric receptors were obtained by replacing, beginning from the N-terminal, successively larger parts of CCR5 with corresponding parts of CXCR4 ##REF##14685051##[23]##. In the resulting chimeras CXCR4 comprised gradually larger parts: the N-terminal tail only (FC-1), including the first transmembrane portion (FC-2), the first (FC-4b), second (FC-5 and FC-6) and third (FC-7) extracellular loops (##FIG##0##Figure 1##). Parental U87.CD4 cells, engineered to express CD4 but no chemokine receptor, were used as negative control.</p>", "<p>Cells were infected in duplicate with each virus stock containing at least 2 ng/ml HIV-1 p24 Ag. The cultures were kept for 7 days and inspection for syncytia formation was performed at days 1, 3 and 7. Supernatant was collected on day 1, after washing, and at the last day of infection, and tested for the presence of p24 Ag by an in-house ELISA assay ##REF##2251501##[24]## (<ext-link ext-link-type=\"uri\" xlink:href=\"http://www.aaltobioreagents.ie\">www.aaltobioreagents.ie</ext-link>). Cultures were only considered for evaluation if HIV-1 p24 value at day 1 was below the lower detection limit of the assay. Viral antigen production was considered positive when the absorbance at day 7 exceeded 0.2 Optical Density (O.D.).</p>", "<p>Viruses able to use only the wild type CCR5 as coreceptor were defined as R5^narrow, whereas R5 viruses able to use the chimeric receptors, FC-1, FC-2, and FC-4b singularly or in different combinations were defined as R5^broad, according to the previously published classification ##REF##15479822##[15]##. R5X4 viruses used the six chimeric receptors in different combinations, and were not further classified.</p>", "<title>Statistical analysis</title>", "<p>Correlation between R5 phenotype and immunological CDC stage of infected children were analyzed by Pearson's chi-square test. Comparison of the frequencies of the wild type chemokine receptor and chimeric receptor usage in the two groups of mothers was done by Fisher's Exact test and Pearson's chi-square test. The Mann-Whitney test and Pearson's chi-square test were used to compare the level of CD4+ T cell counts in the two groups of mothers. Analysis of variance (ANOVA) was performed to demonstrate the association between CD4+ T cell values and viral phenotype. Pearson's chi-square test was used to determine the correlation between clinical stage and the transmission status of mothers. Values below 0.05 were regarded as statistically significant.</p>" ]

[ "<title>Results</title>", "<title>HIV-1 with R5^broad phenotype can be transmitted</title>", "<p>Virus isolates obtained close to birth from 28 newborns were analyzed for phenotypic variability in a cell line expressing wild type or chimeric chemokine receptors. CXCR4 using virus was isolated from one newborn only, confirming previous observations that CCR5 is preferentially used by HIV-1 early after infection ##REF##10985308##[3]##. Twenty out of 27 children (74.07%) with an R5 virus carried a virus able to exclusively use wild type CCR5 (R5^narrow), whereas the remaining 7 children (25.93%) harboured virus with broad use of chimeric receptors (R5^broad) (##TAB##0##Table 1##). These results indicate that transmission of R5^broad virus occurred in a significant proportion of children even if the majority of viruses replicating at a time point close to infection are restricted to the use of wild type CCR5.</p>", "<title>The R5^broad phenotype is predictive of early immunological failure in children</title>", "<p>Clinical and immunological data obtained during follow-up were available for 25 out of 27 children carrying an R5 virus close to birth and for the one child with the R5X4 virus (##TAB##0##Table 1##). The presence of viruses with R5^broad phenotype in the infected newborns was accompanied by a faster progression to immunological failure. Indeed, only children who experienced a severe decline of the CD4+ T cell counts as fast as within 2 or 3 years of age or died within 1 year (groups 1, 2 and 3) carried R5^broad viruses close to birth (p = 0.026 or p = 0.0218, Pearson's chi-square, including groups 1 and 2, or groups 1, 2 and 3, respectively). Whereas none of the children, who were classified as CDC category 3 after 36 months of age or did not enter this category during the follow up (groups 4 or 5), had a virus with R5^broad phenotype close to birth. The only child with a R5X4 virus showed also a fast decline of the CD4+ T cell count (classified CDC 3 within 6 months of age) and died at 44 months. Specifically, all but one R5^broad isolate of the newborns used FC-4b either alone (n = 2) or in combination with FC-2 (n = 4), whereas only one virus isolate used FC-1 and FC-2 (##TAB##1##Table 2##).</p>", "<title>Maternal viral phenotype is predictive of the newborn's virus</title>", "<p>With the aim to understand if selective processes operate during transmission, and thus could be predictive for disease progression, we compared the phenotype of the virus isolates available from 21 mothers close to delivery with that of their newborn child (##TAB##1##Table 2##). All ten mothers harbouring an R5^narrow virus had children whose virus displayed the same phenotype. Interestingly, the six mothers carrying R5^broad viruses transmitted in all but one case a virus with R5^broad phenotype (identical or similar to the mother's virus); in the exceptional mother-child pair, virus with R5^narrow phenotype was present in the child. The five mothers with an R5X4 virus transmitted in one case R5X4, in two cases R5^broad and in another two cases R5^narrow virus. Looking at the details, coreceptor use of child's virus was narrower than the corresponding mother's virus in 7 out of 11 cases, identical in two cases, and a more flexible use of the CCR5 was observed in one child (pair 252). Thus, a virus with R5^broad phenotype was transmitted from the majority of mothers carrying such viruses (5 out of 6; 83.3%), indicating that the maternal viral phenotype is generally preserved during transmission and can possibly be predictive of the phenotype of the newborn's viral variant.</p>", "<title>Maternal viral R5 phenotype is not predictive of transmission</title>", "<p>To identify factors predictive of transmission the viral phenotype of 59 mothers (24 transmitting and 35 non transmitting) was investigated. Transmitting mothers carried more often viruses able to use both CCR5 and CXCR4 as coreceptors than non transmitting mothers, 29% (7 out of 24) and 8% (3 out of 35), respectively, but this difference did not reach statistical significance (p = 0.074, Fisher's Exact Test) (##FIG##1##Figure 2##). No mother carried a monotropic X4 virus.</p>", "<p>The analysis of the chimeric receptor usage showed that the frequency of the R5^narrow phenotype was similar between the two groups of mothers, 53% in non transmitting and 65% in transmitting mothers (17 out of 32 and 11 out of 17, respectively) (##FIG##1##Figure 2##). Furthermore, among the R5^broad isolates (15 and 6 in non transmitting and transmitting mothers, respectively), no difference was observed in the use of FC-1 and FC-2 receptors between the 2 groups of mothers, while viruses from non transmitting mothers utilized FC-4b more frequently than viruses from transmitting mothers (13 out of 15 <italic>vs.</italic> 2 out of 6, respectively) with a trend towards significance (p = 0.056, Perason's chi-square test). This suggests that R5^broad viruses from non transmitting mothers can use CCR5 at least as flexibly as viruses from transmitting mothers.</p>", "<p>Analyses of other factors potentially correlated with risk of transmission disclosed that the two groups of mothers transmitting and non transmitting, could not be distinguished by the clinical stage (p = n.s., Pearson's chi-square test) and the CD4+ T cell counts (mean 478 and 426 cells/mm³; p = n.s., Mann-Whitney test). Furthermore CD4+ T cell values of mothers carrying viruses with R5^narrow or R5^broad phenotype did not differ (p = n.s., ANOVA), but were significantly different from those of mothers carrying CXCR4 using viruses (p = 0.0014, ANOVA). Thus the clinical and the immunological stage of the mothers did not seem to influence the risk of transmission in our cohort.</p>" ]

[ "<title>Discussion</title>", "<p>Biological characteristics of HIV-1 that are critical for the risk of MTCT continue to be subject to discussion. R5 is the predominant virus phenotype early in HIV-1 infection, both in adults and in children born to HIV-1-infected mothers. Since tools to test the mode of CCR5 use ##REF##15479822##[15]##, ##REF##14685050##[16]## have been developed in the past years, we asked the question whether flexible use of CCR5, here dissected into narrow and broad phenotype, would influence transmission of HIV-1 and pediatric disease progression. In the present study the infected newborns harboured mostly viruses of R5 phenotype, a significant proportion (25.9%) of these was able to use chimeric receptors, suggesting that viral variants with a more flexible and efficient use of CCR5 (the R5^broad phenotype) can exist close to infection in the child.</p>", "<p>Of utmost relevance are our data, which show that the R5^broad phenotype detected in the newborn shortly after birth was predictive of a fast and severe immunological failure. Thus, R5^broad viruses seem to determine detrimental effects similar to those known for CXCR4 using viruses. These data support the finding by Casper et al, who suggested that the immunological deterioration in HIV-1 infected children precedes the viral phenotypic switch to CXCR4 usage ##REF##11897036##[25]##. We suggest that pre-existing R5^broad viruses may have caused the worsening of the disease. Interestingly, in our study all but one newborn's R5^broad virus were capable of FC-4b usage, which indeed was previously shown to be linked to evolution to CXCR4 use in adults ##REF##15479822##[15]##. Further studies with sequential follow up samples from children will clarify if the association with CXCR4 switch occurs also in pediatric HIV-1 infection.</p>", "<p>In neonates the memory CD4+ T cells, which express high levels of CCR5, are 6–7 times less represented than the naïve CD4+ cells ##REF##15117454##[26]##. The latter predominantly express CXCR4 ##REF##15117454##[26]##, ##REF##8931781##[27]##, ##REF##9050881##[28]##, ##REF##9591715##[29]##, and are primarily infected by CXCR4 using viruses ##REF##10655520##[30]##. It is likely, that the R5^broad compared to R5^narrow viruses may infect CD4+ naïve cells in addition to memory cells despite the limited expression of the CCR5 molecule, due to their more efficient usage of the coreceptor ##REF##15479822##[15]##. Interestingly, Sleasman et al. ##REF##8931781##[27]## described that although HIV-1 infected neonates in general contained 10 to 100 fold greater number of infected CD4+ memory cells than naïve cells, those children who rapidly progressed in the disease had high proviral load in the CD4+ naïve cells. Thus, infection of naïve cells by R5^broad viruses may interfere with CD4+ T cells production, and thus account for the rapid disease progression observed in children harbouring these viruses. It remains to be solved why CXCR4 using viruses are not preferentially maintained during transmission, despite the high prevalence of CXCR4+ naïve cells in neonates.</p>", "<p>The detailed analysis of the mother's viruses in comparison to those of the corresponding child allowed us once again to pinpoint that the R5 phenotype, either narrow or broad, was usually maintained during the transmission event. On the contrary, the R5X4 phenotype was predominantly lost during transmission. Specifically, mothers with an R5X4 virus transmitted virus with a whole array of phenotypes, i.e. R5^narrow, R5^broad or R5X4. These data lend further support to the lack of restriction in transmission of R5^broad viruses and favour the possibility that the maternal viral R5 phenotype is predictive of the transmitted variant.</p>", "<p>Our data suggest that mothers carrying R5X4 viruses have phenotypically highly heterogenous populations. Whether a selective process or simply a random event governs transmission remains, however, a topic of discussion. ##REF##8249285##[17]##, ##REF##9765386##[31]##, ##REF##1546316##[32]##, ##REF##7639967##[33]##.</p>", "<p>We showed that mothers harbouring R5^broad viruses were not at higher risk of transmission and the presence of HIV-1 with R5^broad phenotype was not predictive for MTCT of HIV-1. The use of chimeric receptors suggests a more flexible usage of CCR5, which in turn may indicate increased viral resistance to inhibition by CC-chemokines. In this respect, the trend of the non transmitting mothers to more often carry viruses with the specific usage of the FC-4b chimeric receptor than transmitting mothers, is intriguing. Recently, Meddows-Taylor et al. described that non transmitting mothers have significantly higher levels of CCL3 in the plasma than transmitting mothers ##REF##16760409##[34]##. It could be envisaged that high levels of CCL3 may be needed to efficiently inhibit R5^broad viruses and, as a consequence, to prevent MTCT of HIV-1. In this respect, it will be relevant to compare the level of CC-chemokines in mothers harbouring R5^broad and R5^narrow viruses as well as to study if R5^broad viruses have a lower sensitivity to the new CCR5 inhibiting drugs than R5^narrow viruses.</p>", "<p>In summary our data show that approximately one forth of the newborn's R5 viruses have the capacity to use CCR5/CXCR4 chimeric receptors indicating that phenotypes with increased flexibility of co-receptor use are not hampered during transmission. Conversely these viral variants are significantly linked with severe immunological failure within the first years of age of the infected children. These data may have important implications for timely and appropriate therapeutic choice in pediatric HIV-1 infection.</p>" ]

[]

[ "<p>Conceived and designed the experiments: GS. Performed the experiments: MC IK. Analyzed the data: MC GS. Wrote the paper: MC EMF ADR GS. Performed testing with the U87 cell lines of some of the children early isolates and contributed with her expertise on the U87 cell lines expressing chimeric receptors: IK. Performed the blood sample preparation and virus isolation of part of the samples: MZ. Has produced the U87 cells expressing the chimeric receptors and provided the cells as well as the know-how to the study: LA. Provided part of the samples and clinical data of mothers and children: AP CG. Provided part of the samples: ADR.</p>", "<title>Background</title>", "<p>HIV-1 R5 viruses are characterized by a large phenotypic variation, that is reflected by the mode of coreceptor use. The ability of R5 HIV-1 to infect target cells expressing chimeric receptors between CCR5 and CXCR4 (R5^broad viruses), was shown to correlate with disease stage in HIV-1 infected adults. Here, we ask the question whether phenotypic variation of R5 viruses could play a role also in mother-to-child transmission (MTCT) of HIV-1 and pediatric disease progression.</p>", "<title>Methodology/Principal Findings</title>", "<p>Viral isolates obtained from a total of 59 HIV-1 seropositive women (24 transmitting and 35 non transmitting) and 28 infected newborn children, were used to infect U87.CD4 cells expressing wild type or six different CCR5/CXCR4 chimeric receptors. HIV-1 isolates obtained from newborn infants had predominantly R5^narrow phenotype (n = 20), but R5^broad and R5X4 viruses were also found in seven and one case, respectively. The presence of R5^broad and R5X4 phenotypes correlated significantly with a severe decline of the CD4+ T cells (CDC stage 3) or death within 2 years of age. Forty-three percent of the maternal R5 isolates displayed an R5^broad phenotype, however, the presence of the R5^broad virus was not predictive for MTCT of HIV-1. Of interest, while only 1 of 5 mothers with an R5X4 virus transmitted the dualtropic virus, 5 of 6 mothers carrying R5^broad viruses transmitted viruses with a similar broad chimeric coreceptor usage. Thus, the maternal R5^broad phenotype was largely preserved during transmission and could be predictive of the phenotype of the newborn's viral variant.</p>", "<title>Conclusions/Significance</title>", "<p>Our results show that R5^broad viruses are not hampered in transmission. When transmitted, immunological failure occurs earlier than in children infected with HIV-1 of R5^narrow phenotype. We believe that this finding is of utmost relevance for therapeutic interventions in pediatric HIV-1 infection.</p>" ]

[]

[ "<p>The samples were provided by the National Care System for HIV-1 infected mothers, the Seventh Department of Gynecology and Obstetrics, and the First and Fourth Department of Pediatrics of the University of Milan, Italy and by the AIDS Reference Center, Unit of Viral Oncology, University of Padova, Italy. The authors thank the Italian Register for HIV Infection in Children for providing clinical data of the children.</p>", "<p>The statistical analysis was kindly performed by Clelia Di Serio and Alessandro Ambrosi from the University Centre of Statistics for Biomedical Sciences, Vita Salute San Raffaele University, Milan, Italy.</p>" ]

[ "<fig id=\"pone-0003292-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003292.g001</object-id><label>Figure 1</label><caption><title>Schematic representation of the chemokine receptors CCR5 and CXCR4 and the chimeric CCR5/CXCR4 receptors.</title><p>Chimeric receptors FC-1, FC-2, FC-4b, FC-5, FC-6 and FC-7 were obtained by replacing successively larger parts of CCR5 with corresponding regions of CXCR4.</p></caption></fig>", "<fig id=\"pone-0003292-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003292.g002</object-id><label>Figure 2</label><caption><title>Distribution of the viral phenotype of transmitting and non transmitting mothers.</title><p>Distribution of R5 (black) <italic>vs.</italic> R5X4 (white) viruses within all virus phenotypes (n = 59 viruses; p = n.s., Fisher's Exact Test) and of R5^narrow (dark gray) <italic>vs.</italic> R5^broad (light gray) within the R5 phenotype (n = 49 viruses; p = n.s., Fisher's Exact Test).</p></caption></fig>" ]

[ "<table-wrap id=\"pone-0003292-t001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003292.t001</object-id><label>Table 1</label><caption><title>Clinical and viral characteristics of HIV-1 infected children<xref ref-type=\"table-fn\" rid=\"nt101\">(a)</xref>.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Category of progression</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Child code</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Phenotype<xref ref-type=\"table-fn\" rid=\"nt102\">(b)</xref> close to birth</td><td colspan=\"3\" align=\"left\" rowspan=\"1\">Age at category diagnosed<xref ref-type=\"table-fn\" rid=\"nt104\">(c)</xref>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Death</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Therapy<xref ref-type=\"table-fn\" rid=\"nt105\">(d)</xref>\n</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">CDC 3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CDC B</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">CDC C</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/></tr></thead><tbody><tr><td colspan=\"8\" align=\"left\" rowspan=\"1\">\n<bold>GROUP 1</bold>\n</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>CDC 3 or early death &lt;12 mos</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">B111p</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Broad (1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">60</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B117p</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (5)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">11</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B183</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (1.5)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B193</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Broad (4)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">12</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">28</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">12</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B196</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5X4 (1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">34</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">44</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B201</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B204</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Broad (0)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">38</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B224</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">44</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B314</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Broad (0.5)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B380</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (6)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">15</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td></tr><tr><td colspan=\"8\" align=\"left\" rowspan=\"1\">\n<bold>GROUP 2</bold>\n</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>CDC3 13–24 mos</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">B252</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Broad (1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">14</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">14</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">30</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">14</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B256</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (1.5)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">17</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">46</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7</td></tr><tr><td colspan=\"8\" align=\"left\" rowspan=\"1\">\n<bold>GROUP 3</bold>\n</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>CDC3 25–36 mos</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">B32</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">28</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">28</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">96</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">28</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B34p</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Broad (3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">28</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">70</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B130</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">30</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">38</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">47</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">22</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B199</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (2)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">27</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">24</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">60</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">27</td></tr><tr><td colspan=\"8\" align=\"left\" rowspan=\"1\">\n<bold>GROUP 4</bold>\n</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>CDC3 36–60 mos</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">B3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (6)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">65</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">65</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">54</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B115</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">58</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">58</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">59</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B136</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (3)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">60</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">64</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B145</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">48</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">48</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">98</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">51</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B255</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (3.5)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">46</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">12</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">12</td></tr><tr><td colspan=\"8\" align=\"left\" rowspan=\"1\">\n<bold>GROUP 5</bold>\n</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>CDC3 &gt;60 mos or never</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">B31p</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (0)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">114</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B115p</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (2.5)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">52</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B190</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">77</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B225</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">40</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B306</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (9)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">28</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">72</td></tr><tr><td colspan=\"8\" align=\"left\" rowspan=\"1\">\n<bold>GROUP 6</bold>\n</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>Lost at follow-up</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">B107p</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Narrow (1)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n.a.</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n.a.</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n.a.</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n.a.</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n.a.</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">B139p</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Broad (4)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n.a.</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n.a.</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n.a.</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n.a.</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">n.a.</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pone-0003292-t002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003292.t002</object-id><label>Table 2</label><caption><title>Comparison of viral phenotypes of 21 mother-child pairs.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Patient code<xref ref-type=\"table-fn\" rid=\"nt106\">(a)</xref>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Wild type coreceptor use</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Chimeric receptor use</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Patient code<xref ref-type=\"table-fn\" rid=\"nt106\">(a)</xref>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Wild type coreceptor use</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Chimeric receptor use</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A107p (+1)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A34p (+3)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">FC 1-2-4b</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B107p (+1)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B34p (+3)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">FC 4b</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A115 (+3)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A111p (+1)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">FC 1-2</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B115 (+3)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B111p (+1)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">FC 1-2</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A115p (+2.5)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A139p (+4)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">FC 2</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B115p (+2,5)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B139p (+4)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">FC 4b</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A117p (+5)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A145 (+1)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">FC 2</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B117p (+5)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B145 (+1)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A130 (0)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A252 (−0.5)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">FC 2</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B130 (+3)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B252 (+1)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">FC 2-4b</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A136 (+1)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A314 (0)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">FC 2-4b</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B 136 (+3)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B314 (+0.5)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">FC 2-4b</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A224 (0)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B224 (+1)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A225 (0)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A31p (0)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5X4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">FC 1-2-4b-5-6-7</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B225 (+1)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B31p (0)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A255 (+0.5)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A183 (0)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5X4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">FC 2-4b-5-6-7</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B255 (+3.5)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B183 (1.5)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A256 (+1.5)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A193 (+4)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5X4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">FC 4b-6-7</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B256 (+1.5)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">-</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B193 (+4)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">FC 2-4b</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A196 (0)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5X4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">FC 4b-5-6-7</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B196 (+1)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5X4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">FC 4b-7</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>A204 (0)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5X4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">FC 4b-7</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>B204 (0)</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">R5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">FC 2-4b</td></tr></tbody></table></alternatives></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><fn id=\"nt101\"><label>(a)</label><p>symbol - means that the event has not occurred. n.a. = not available. Mos means months of age. Age of appearance of the different conditions is always indicated in months.</p></fn><fn id=\"nt102\"><label>(b)</label><p>Narrow and broad refer to viruses with R5 phenotype. Viruses able to exclusively use wild type CCR5 receptor are defined narrow, whereas those using chimeric receptors besides the wild type CCR5 are defined broad. In parenthesis is indicated the age in months of the virus phenotype determination.</p></fn><fn id=\"nt103\"><p>Statistical analysis: Influence of virus R5^broad phenotype on disease progression including children of group 1 and 2, or children of group 1, 2 and 3: p = 0.0260 and p = 0.0218 (Pearson's chi Square), respectively.</p></fn><fn id=\"nt104\"><label>(c)</label><p>Age of entry into clinical or immunological category. Categories are defined according to the Centers for Disease Controls ##UREF##3##[21]##: CDC 3 = severe immune suppression; CDC B = moderate clinical manifestations; CDC C = severe clinical manifestations.</p></fn><fn id=\"nt105\"><label>(d)</label><p>Age of start of mono or dual antiretroviral therapy, not HAART.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt106\"><label>(a)</label><p>Isolates from corresponding mother-child pairs are indicated by the same number preceded by a letter: A for mothers and B for children. Time of sampling is indicated in parenthesis as months before (-), after (+) or within 1 week from (0) delivery/birth. </p></fn><fn id=\"nt107\"><p>Pairs were grouped according to the mother's virus phenotype, i.e. first those carrying an R5^narrow, than an R5^broad and last an R5X4 virus.</p></fn><fn id=\"nt108\"><p>Samples were used to infect U87.CD4 cells expressing wild type CCR5 or CXCR4, or one of the chimeric receptors FC-1, FC-2, FC-4b, FC-5, FC-6 or FC-7. Experiments were repeated twice. –, means no chimeric receptor is used.</p></fn></table-wrap-foot>", "<fn-group><fn fn-type=\"COI-statement\"><p><bold>Competing Interests: </bold>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p><bold>Funding: </bold>The financial support was received by the Swedish Research Council and by the Istituto Superiore di Sanità within the framework of the National program for AIDS research, Grant no.40G.21, 45G.12 and 40G.56. Only financial support.</p></fn></fn-group>" ]

[ "<graphic id=\"pone-0003292-t001-1\" xlink:href=\"pone.0003292.t001\"/>", "<graphic id=\"pone-0003292-t002-2\" xlink:href=\"pone.0003292.t002\"/>", "<graphic xlink:href=\"pone.0003292.g001\"/>", "<graphic xlink:href=\"pone.0003292.g002\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Genomewide proteomics studies, primarily yeast two-hybrid assays ##REF##10688190##[1]##,##REF##11283351##[2]## and high-throughput mass spectrometry ##REF##11805826##[3]##,##REF##11805837##[4]##, provide a growing list of putative protein–protein interactions, and demonstrate that most if not all proteins have interacting partners in the cell. Elucidating the atomic details of these complexes requires further biochemical and structural information. While the most complete structural characterization of a complex is provided by X-ray crystallography, protein–protein hetero-complexes constitute less than 2% of protein structures in the Protein Data Bank (PDB) ##REF##10592235##[5]##, and their number increases at a slow rate. In fact, many biologically important interactions occur in weak, transient complexes that will not be amenable to direct experimental analysis, even when both proteins can be isolated and their structures determined. Thus, there is substantial need for computational docking methods that can determine the structure of a complex from the separately solved structures of two component proteins.</p>", "<p>Based on the thermodynamic hypothesis, at fixed temperature and pressure the Gibbs free energy of the macromolecule-solvent system reaches its global minimum at the native state of the complex. Thus, docking requires a computationally feasible free energy evaluation model and an effective minimization algorithm. It is expected that docking methods can utilize the rich set of modeling tools developed for predicting the structures of folded proteins. It has been established over the last two decades that the energy landscape of a foldable protein resembles a many-dimensional funnel with a free energy gradient toward the native structure ##REF##1528885##[6]##–##UREF##0##[9]##. A number of papers suggest that the landscape theory also applies to protein–protein association ##REF##9631300##[10]##–##REF##11468354##[12]##. The size of the funnel is determined by the length scales of the long-range electrostatic and hydrophobic interactions and the geometry of the proteins, and hence the funnel is restricted to a neighborhood of the native complex ##REF##11599022##[13]##. There is a free energy gradient toward the native state, but the funnel is rough, giving rise to many local minima ##REF##9653131##[14]## that correspond to encounter complexes, some of which may be visited along a particular association pathway ##REF##10049302##[15]##,##REF##10692300##[16]##.</p>", "<p>While homology modeling approaches play an important role in protein structure prediction, most current docking methods are based on direct optimization, and attempt to find the global minimum of a function approximating the free energy of the complex. According to the results of CAPRI (Comparative Assessment of PRotein Interactions), a community-wide experiment devoted to protein–protein docking ##REF##12784368##[17]##–##REF##15036860##[20]##, the optimization involves either the systematic sampling of the discretized rotational/translational space using Fast Fourier Transforms ##REF##12784371##[21]##,##REF##16933295##[22]## or geometric hashing ##REF##7739053##[23]##, or it relies on Monte Carlo (or Monte Carlo minimization) algorithms ##REF##12875852##[24]##,##REF##8289329##[25]##. Both optimization methods are generic, i.e., they do not rely on any assumption about the specific shape of the energy function to be minimized.</p>", "<p>The use of special optimization methods that account for the funnel-like shape of the free energy function offers two potential advantages. First, being designed for minimizing funnel-like functions, such algorithms can be more efficient than generic approaches. Second, the success of such algorithms will be a stringent test of how well the funnel assumption describes the binding free energy landscape. This second point is particularly interesting, because protein–protein association occurs in the six-dimensional (6D) space of translations and rotations, at least for the classes of proteins whose backbones remain essentially unchanged upon association (e.g., many enzymes interacting with their inhibitors). Although the association is accompanied by conformational changes, these can be considered auxiliary, and the shape of the funnel can be studied over the entire conformational space. In contrast, the free energy of protein folding is defined in a substantially higher-dimensional space, and hence funnels can be generally studied only along some reaction coordinates ##REF##1528885##[6]##–##UREF##0##[9]##.</p>", "<title>Minimization by Underestimation</title>", "<p>A minimization approach which is specific to funnel-like functions can be based on the concept of <italic>underestimation</italic>. The existence of a funnel implies that the free energy can be locally underestimated by a convex function (##FIG##0##Figure 1##). The original free energy function is extremely rugged with a huge number of local minima even in a small region of conformational space. Yet its convex underestimator is much smoother and still captures the overall funnel-like landscape, which provides a handle to free energy minimization. The quality of minimization through underestimation depends on the choice of underestimator functions, the way they are constructed and utilized to locate the global minimum, as well as how structured the free energy funnels are in conformational space. The Convex Global Underestimation (CGU) method ##UREF##1##[26]## employed canonical quadratic functions as underestimators without any cross-terms. In that case the underestimator, based on a set of local minima, can be constructed by solving a Linear Programming (LP) problem. Uniformly distributed samples in the neighborhood of the underestimator's global minimum were then used to bias further sampling. The process was iterated with the set of local minima being updated, and the search region being reduced until certain convergence criteria are satisfied. CGU has been a very promising method with various applications in molecular structure prediction, including protein folding ##REF##9278057##[27]## and docking small molecules to proteins ##UREF##2##[28]##. However, its restriction of using canonical quadratic functions limits its success in some cases ##REF##11908495##[29]##, since the principal axes of the free energy surface are not necessarily aligned with the canonical coordinates. The performance further deteriorates as the dimensionality of the search space increases. We have used theoretical analysis to show and simple test problems to demonstrate that this restriction can lead to incorrect convergence ##REF##19759849##[30]##.</p>", "<p>Motivated by the potential advantages of underestimation and practical shortcomings of the CGU algorithm, we have recently extended the method using general quadratic underestimators and introduced biased sampling guided by the underestimator ##REF##19759849##[30]##,##UREF##3##[31]##. Since the tightest underestimator in this class is obtained by solving a semi-definite programming problem, the method is termed SDU (Semi-Definite programming-based Underestimation). Semi-definite programming is computationally more demanding than the linear programming (still solvable in polynomial-time though) used in the CGU method. However, SDU typically requires fewer iterations and substantially improves optimization performance.</p>", "<p>The SDU method starts from a set of <italic>K</italic> local minima <bold>x</bold>\n¹, …, <bold>x</bold>\n<italic>^K</italic> of a funnel-like function <italic>f</italic>(<bold>x</bold>): within a given region of the search space. Throughout the course of the algorithm we maintain a set of local minima in the search region; initially . To capture the global funnel-like structure of <italic>f</italic>(<bold>x</bold>) within we construct a smooth (convex) quadratic function <italic>U</italic>(<bold>x</bold>) = <bold>x</bold>′<bold>Qx</bold>+<bold>b</bold>′<bold>x</bold>+<italic>c</italic>, where <bold>Q</bold> is a positive semi-definite matrix, , <italic>c</italic> is a scalar, and prime denotes transpose, such that <italic>U</italic>(<bold>x</bold>) underestimates <italic>f</italic>(<bold>x</bold>) at all local minima in , i.e., <italic>U</italic>(<bold>x</bold>\n<italic>ⁱ</italic>)≤<italic>f</italic>(<bold>x</bold>\n<italic>ⁱ</italic>) for all <italic>i</italic> = {1, 2, …, <italic>K</italic>}. The tightest possible underestimator (with an <italic>L</italic>\n₁ norm metric) can be found by solving a semi-definite programming problem ##REF##19759849##[30]##. <italic>U</italic>(<bold>x</bold>) is a general convex quadratic function.</p>", "<p>The underestimator <italic>U</italic> is used to guide further sampling. The minimum of <italic>U</italic>, denoted by <bold>x</bold>\n<italic>^P</italic> and referred to as the <italic>predictive conformation</italic>, is in an energetically favorable region, and hence a new conformation can be generated by local minimization starting from <bold>x</bold>\n<italic>^P</italic>. Additional conformations are obtained by local minimization with randomly generated starting points such that points in the vicinity of <bold>x</bold>\n<italic>^P</italic> have a higher probability of being selected than points further away. To that end, we simply sample within using a density function shaped as −<italic>U</italic>. The set is being updated by adding these newly obtained conformations while removing unfavorable (i.e., higher energy) conformations, and the search area is being reduced to a neighborhood of <bold>x</bold>\n<italic>^P</italic>. Using the updated conformations in we repeat the underestimation step and the whole process is being iterated until a convergence criterion is met.</p>", "<p>The SDU algorithm has a number of favorable properties when applied to funnel-shaped functions. Theoretical analysis shows ##REF##19759849##[30]## that <bold>x</bold>\n<italic>^P</italic> converges in probability to the global minimum <bold>x</bold>* of the funnel-like function <italic>f</italic> as the number of samples <italic>K</italic> grows. When applied to test functions resembling the funnel-like free energy functions, SDU has been shown outperforming CGU and a simulated annealing algorithm which adaptively tunes its parameters, with much less required function evaluations and much higher success rates ##REF##19759849##[30]##.</p>", "<title>Docking by Semi-Definite Underestimation</title>", "<p>The main goal of this paper is to develop and test docking methods that use the SDU algorithm in order to find the global minimum of a funnel-like function approximating the free energy over regions of the conformational space. Over the last few years we have developed a multistage docking method that starts with rigid body search based on Fast Fourier Transform (FFT), selects and clusters 1000 to 2000 low energy docked structures, and retains the 10 to 30 largest clusters for further processing ##REF##14693807##[32]##. The conformational space is decomposed into separate regions defined by the clusters, where free energy attraction basins are believed to exist and the free energy landscape is assumed to be funnel-like. Then SDU is called upon to locate the global minimum within each region, by utilizing such funnel-like behavior.</p>", "<p>As will be shown, in spite of the success of SDU as an optimizer for functions with funnel-shaped basins, its application to docking turned out to be far from straightforward. Although the method yields meaningful moves in either translational or rotational subspaces ##REF##19759849##[30]##, minimization in the full 6D space of rigid body motions poses a challenge. This difficulty is well known in the robotics literature ##UREF##4##[33]##. The Euclidean group <italic>SE</italic>(3), which is the semidirect product of (translations) and the special orthogonal Euclidean group <italic>SO</italic>(3) (rotations), is a nonlinear manifold and its parametrization is critical to any optimization procedure ##UREF##4##[33]##. In particular, we will show that the funnel-like shape of the free energy surface is affected by the parameterization of the search space, and that underestimation tends to fail unless appropriate parameters are selected.</p>", "<p>We will describe two implementations of the SDU method: (<italic>i</italic>) SDU1, a cyclic coordinate descent strategy where rotational and translational moves alternate; and (<italic>ii</italic>) SDU2, a 5D strategy in which the distance of the two proteins is separately optimized and the SDU-driven search is performed in a lower dimensional space defined by 5 angular coordinates. Since the most energetically favorable distances occur when the two proteins are in contact but do not overlap, this strategy explores the free energy surface spanned by encounter complexes ##REF##10049302##[15]##. Results will show that the methods discover broad energy funnels, generate high quality docking predictions, and produce a substantial efficiency gain compared to Monte Carlo methods.</p>", "<p>The comparison with Monte Carlo methods is based on our earlier preliminary work ##UREF##5##[34]## where we have tested a method similar to SDU1 against a 10-protein set using reduced Gō-type potentials. Even with these relatively smooth potentials, underestimation in the full 6D space has not been effective, and we had to rely on optimizing in each (rotational or translational) subspace separately. In the present paper, we show that the SDU1 strategy is effective against a larger benchmark set using much more refined energy potentials. However, our main contribution is the introduction of SDU2 which is more effective and improves efficiency by a factor of 10 compared to SDU1. In addition, SDU2 provides interesting biophysical insights with its resemblance to docking by repeated micro-collisions.</p>" ]

[ "<title>Methods</title>", "<title>Protein Docking Benchmark</title>", "<p>The SDU1 and SDU2 algorithms were tested on complexes from the protein docking benchmark sets ##REF##12784372##[38]##,##REF##15981264##[39]##. These sets contain enzyme–inhibitor, antigen–antibody, and “other” types of complexes. As described, the algorithms were used to refine the separate clusters generated by the rigid body docking program PIPER. For enzyme–inhibitor complexes we refined only the 10 largest clusters and hence restricted consideration to complexes from the two benchmark sets for which these clusters included at least one near-native conformation. In addition, we disregarded four enzyme–inhibitor pairs that form oligomeric rather than binary complexes, resulting in the 25 test problems. The refinement algorithms were also applied to 11 antigen–antibody and 8 “other” type of complexes from the benchmark set 1 ##REF##12784372##[38]## that had at least one near-native structure in the 30 largest clusters. We emphasize that tests for most complexes involved separately determined protein structures as given in the benchmark sets ##REF##12784372##[38]##,##REF##15981264##[39]##. The exceptions are a few of the antigen–antibody complexes in which the antigen was separately solved but the antibody structure was taken from the complex.</p>", "<title>Rigid Body Docking</title>", "<p>The rigid body docking program PIPER ##REF##16933295##[22]##, based on the FFT correlation approach, systematically samples billions of docked conformations on a grid. Compared with other FFT-based approaches that use only shape complementarity and electrostatics for scoring, the scoring function in PIPER also includes the statistical pairwise potential DARS (Decoy As Reference States) ##REF##16933295##[22]##. Since the potential is represented as the sum of a few correlation functions through the eigenvalue-eigenvector decomposition of the matrix of the DARS interactions energy coefficients, the energy can be very efficiently evaluated using Fourier transforms. In conjunction with this higher accuracy scoring function, the PIPER program significantly enriches the hit rates among top ranked predictions for the benchmark sets described above.</p>", "<p>For each complex, we retained the 1000 lowest energy predictions and clustered them using pairwise RMSD as the distance metric ##REF##15908573##[45]##. The resulting clusters were ranked based on their size, reflecting a preference for local minima with broad regions of attraction ##REF##14693807##[32]##. We have retained at most 30 clusters for each complex, each one being roughly 10 Å RMSD wide.</p>", "<title>The Semi-Definite Programming-Based Underestimation (SDU) Algorithm</title>", "<title>Constructing an underestimator</title>", "<p>We start with a set of K locally minimized structures (, where <italic>n</italic> = 3 for SDU1 and <italic>n</italic> = 5 for SDU2) within each region defined by a cluster and the corresponding free energy values Δ<italic>G</italic>. (The detailed free energy models would be described in the next part.) In this study, the set is initially chosen by the cluster center and (<italic>K</italic>−1) structures with the lowest PIPER scores within each cluster. When no cluster information other than its representative is available, this can be simply a set of locally perturbed structures around the cluster center. We are interested in constructing an underestimator which underestimates the free energy surface at those samples in set and captures the general funnel-like landscape. The family of underestimators used here are convex general quadratic functions , where , and <italic>c</italic> is a scalar.</p>", "<p>Using an L1 norm as a distance metric, the problem of finding the tightest possible such underestimator <italic>U</italic> can be formulated as follows:where the decision variables are <bold>Q</bold>, <bold>b</bold>, and <italic>c</italic>, and denotes positive semi-definiteness. This problem can be reformulated as a <italic>Semi-Definite Programming</italic> (SDP) problem ##REF##19759849##[30]##, an important class of convex programming problems ##UREF##9##[46]## which finds many applications in various subjects recently. SDP problems aim at minimizing a linear function subject to the constraints of linear matrix inequalities. Such constraints are nonlinear but convex. Efficient polynomial-time algorithms, such as interior-point algorithms, exist for solving SDP problems. General-purpose solvers are also readily available ##UREF##10##[47]##,##UREF##11##[48]##. We use the callable library of SDPA v6.20 ##UREF##11##[48]## which solves the SDP problem efficiently with a primal-dual interior-point method and exploits the sparsity of the problem. When canonical quadratic underestimators are used as in CGU ##UREF##1##[26]##, <bold>Q</bold> is restricted to diagonal matrices without off-diagonal elements. The problem then becomes a Linear Programming (LP) one, which can be regarded as a special case of SDP problems.</p>", "<p>There are coefficients for <italic>U</italic>, which suggests that the number of samples <italic>K</italic>≥<italic>K</italic>\n₀. <italic>K</italic>\n₀ equals 10 and 21, respectively, for SDU1 and SDU2. <italic>K</italic> is set to 40 for both SDU methods.</p>", "<title>Biased sampling</title>", "<p>The derived optimal underestimator <italic>U</italic> is exploited to bias further sampling in . The global minimum of <italic>U</italic> is denoted by <bold>x</bold>\n<italic>^P</italic> and referred to as the <italic>predictive conformation</italic>. Since <italic>U</italic> reflects the general structure of the free energy landscape, at least based on the discrete sampling of , our strategy is to sample in the area around <bold>x</bold>\n<italic>^P</italic> such that conformations close to <bold>x</bold>\n<italic>^P</italic> are more likely to be selected. This can be achieved by using an acceptance/rejection scheme and the following probability density function (pdf) in :where .</p>", "<title>Iteration techniques</title>", "<p>The processes of underestimation and biased sampling are iterated with the set of local minima being updated and the search region being gradually shrunk to the neighborhood of predictive conformation <bold>x</bold>\n<italic>^P</italic>. Previous samples in which are energetically unfavorable or too far from <bold>x</bold>\n<italic>^P</italic> are be discarded, namely those structures with . (<italic>ζ</italic> = 0.7 in this work). Newly sampled structures are added to with local minimization starting from <bold>x</bold>\n<italic>^P</italic> and additional biased samples.</p>", "<p>We set the convergence criterion based on the proximity of the predictive conformation <bold>x</bold>\n<italic>^P</italic> and the current lowest-energy structure <bold>x</bold>\n<italic>_min</italic> in . At most 5 iterations are carried on unless convergence is observed earlier, i.e., ∥<bold>x</bold>\n<italic>^P</italic>−<bold>x</bold>\n<italic>_min</italic>∥≤<italic>ε</italic>. <italic>ε</italic> is set at 1 Å in translations and 0.087 radian (5 degrees) in rotations for SDU1. To approximate equal convergence definition, <italic>ε</italic> is set at 0.1 radian for SDU2. The value of <bold>x</bold>\n<italic>_min</italic> at the final iteration provides the final solution for either SDU method.</p>", "<title>Free Energy Evaluation Models</title>", "<p>Docking by the SDU algorithms involves the use of two different free energy models. In the rigid body global search the scoring function iswhere the desolvation free energy Δ<italic>G_des</italic> is estimated by the Atomic Contact Potential ##REF##9126848##[49]##, an atom-level extension of the Miyazawa-Jernigan potential ##UREF##12##[50]##, the electrostatic energy Δ<italic>E_elec</italic> is based on the Coulombic formula with distance-dependent dielectrics ε = 4<italic>r</italic>, and the Van der Waals term Δ<italic>E_vdw</italic> is adopted from the Charmm potential ##UREF##13##[51]##. The scaling factor <italic>λ</italic>∈[0,1] is dynamically adjusted during the course of the refinement to improve the quality of the underestimator by dampening the effect of the van der Waals term Δ<italic>E_vdw</italic> and thereby smoothing Δ<italic>G</italic>. Specifically, whenever a “flat” underestimator <italic>U</italic> (this can be determined when the minimum eigenvalue of <bold>Q</bold> is close to zero) is being computed with <italic>λ</italic> = 1, we gradually reduce <italic>λ</italic> with a stepsize of 0.1 until we obtain a more informative underestimator or <italic>λ</italic> reaches 0.</p>", "<p>In the flexible local minimization we use the Charmm potential with ε = 4<italic>r</italic>, including the internal energy terms, and perform 100 steps of adopted base Newton-Raphson (ABNR) minimization allowing for side-chain flexibility in the interface ##UREF##13##[51]##. The distance between two components is separately optimized. Specifically, the ligand is pushed towards or pulled apart from the receptor at a stepsize of 0.5 Å along the line segment connecting the two centers of mass. The maximum shift in distance (which defines locality) is 2 Å in this study. Each position is followed by a Charmm minimization described above and energy evaluation. Only non-clashing structures are accepted, judged by the condition Δ<italic>E_vdw</italic>&lt;0. In fact, we found that PIPER had a higher tolerance toward clashing structures and pushing component proteins closer generally resulted in increased positive Δ<italic>E_vdw</italic>. To reduce calls to Charmm minimization, in practice we only pull them apart if necessary, i.e., when Δ<italic>E_vdw</italic>&gt;0 is found. For simplicity we will call the work involved in evaluating Δ<italic>G</italic>*(<bold>x</bold>) for each conformation <bold>x</bold> a <italic>function evaluation</italic>, although it generally involves several evaluations of Δ<italic>G</italic>(<bold>x</bold>,<bold>s</bold>).</p>", "<title>Quality Measures</title>", "<p>As a measure of prediction quality, we select the ligand <italic>C_α</italic> atoms in the binding site, and calculate the RMSD between their predicted and observed positions. A ligand residue is considered to be in the binding site if any of its atom is within 10 Å of an atom on the receptor. We refer to a structure as near-native (or a “hit”) if its ligand binding site <italic>C_α</italic> RMSD is less than 10 Å. Although such structures are not really close to the native complex, by rigid body docking it is generally difficult to obtain better results. In fact, since the near-native binding region is selected by cluster size using a clustering radius on the order of 10 Å, the goal of this first step is to generate as many such 10 Å RMSD structures as possible. To show the improvements due to the SDU method, we list the rank of the first cluster that includes a hit, as well as the RMSD between the native structure and the center of the cluster. However, as a more appropriate overall performance measure of the refinement, we also note the number of complexes that have at least one prediction with less than 5 Å RMSD from the native structure in the top 5 clusters.</p>" ]

[ "<title>Results/Discussion</title>", "<title>Development of SDU-Based Docking Algorithms</title>", "<p>To define the docking problem we fix the position and orientation of the first (receptor) protein. The six-dimensional vector <bold>x</bold> specifies the position and orientation of the second (ligand) protein. The variables <bold>s</bold> account for the side chain conformations in both proteins, and Δ<italic>G</italic>(<bold>x</bold>,<bold>s</bold>) denotes the free energy function. The minimization of Δ<italic>G</italic>(<bold>x</bold>,<bold>s</bold>) with respect to <bold>s</bold> is restricted to the side chains in the interface and is carried out by local methods. We define Δ<italic>G</italic>*(<bold>x</bold>) = Δ<italic>G</italic>(<bold>x</bold>,<bold>s</bold>*), where <bold>s</bold>* is an optimal solution; then the protein docking problem is finding the lowest minimum of Δ<italic>G</italic>*(<bold>x</bold>) in the region defined by the cluster, where <bold>x</bold> belongs to the space of rigid body motions, i.e., the Euclidean group <italic>SE</italic>(3). As described before, the parameterization of <italic>SE</italic>(3) is critical for optimization purposes, and hence we first describe our results concerning the parameterization of the search space.</p>", "<p>The space <italic>SE</italic>(3) is the semidirect product of (translations) and <italic>SO</italic>(3) (rotations). The rotation group <italic>SO</italic>(3) is a 3-dimensional Lie group consisting of rotation matrices, i.e, , <bold>RR</bold>′ = <bold>I</bold>, det(<bold>R</bold>) = 1. The Lie algebra of <italic>SO</italic>(3), denoted by <italic>so</italic>(3), may be represented by the real skew-symmetric matriceswhere . It is well-known that the one-parameter subgroups of <italic>SO</italic>(3), i.e., are geodesics ##UREF##4##[33]##,##UREF##6##[35]##,##UREF##7##[36]##, i.e., the shortest paths between two points on <italic>SO</italic>(3). Moreover, for <bold>R</bold>\n₀, <bold>R</bold>\n₁∈<italic>SO</italic>(3), Ω = log(<bold>R</bold>\n₀′<bold>R</bold>\n₁)∈<italic>so</italic>(3) and the distance between <bold>R</bold>\n₀ and <bold>R</bold>\n₁ can be defined by <italic>ρ</italic>(<bold>R</bold>\n₀,<bold>R</bold>\n₁) = ∥log(<bold>R</bold>\n₀′<bold>R</bold>\n₁)∥ = ∥<italic>ω</italic>\n₁−<italic>ω</italic>\n₀∥. This distance is a natural Riemannian metric on <italic>SO</italic>(3), i.e., it is bi-invariant with respect to the actions of the group (rotations). The exponential map from <italic>so</italic>(3) to <italic>SO</italic>(3) defined by Ω → <bold>R</bold>\n₀\n<italic>e</italic>\n^Ω is a local diffeomorphism, i.e., there exist an open neighborhood of 0∈<italic>so</italic>(3) and an open neighborhood of <bold>R</bold>\n₀∈<italic>SO</italic>(3) and an invertible, surjective, and smooth map from one neighborhood to the other whose inverse is also smooth. The local diffeomorphism induces a coordinate chart in a neighborhood of <bold>R</bold>\n₀ that is known as exponential coordinate system. Given the definition of <italic>so</italic>(3) this coordinate system can be parameterized by .</p>", "<p>Given the favorable properties of <italic>SO</italic>(3) (more generally <italic>SO</italic>(<italic>n</italic>)) such as the existence of a natural bi-invariant metric and, in particular, the simplicity of determining the geodesics of the manifold, the standard gradient based optimization algorithms on can be generalized for optimization on <italic>SO</italic>(<italic>n</italic>) ##UREF##6##[35]##. However, generalization to the entire <italic>SE</italic>(3) is more difficult ##UREF##4##[33]##,##UREF##7##[36]##. Although <italic>SE</italic>(3) is also Lie group, its one-parameter subgroups are no longer geodesics. Moreover, <italic>SE</italic>(3) does not admit a natural (bi-invariant) Riemannian metric while its subspaces <italic>SO</italic>(3) and do ##UREF##4##[33]##,##UREF##7##[36]##. Attempting to implement SDU to <italic>SE</italic>(3) by simultaneous translational and rotational optimization, we have generally failed to construct useful underestimators. We have, however, found two strategies that were able to overcome this problem.</p>", "<title>SDU1: Cyclic coordinate descent</title>", "<p>A natural search strategy in <italic>SE</italic>(3) is to alternate between optimizing the free energy in <italic>SO</italic>(3) and in by a series of rotational and translational adjustments. As will be further discussed, the major disadvantage of this approach is that samples in one subspace cannot be reused in the other subspace, resulting in inefficient search.</p>", "<title>SDU2: 5D search in the space of encounter complexes</title>", "<p>The distance between the two proteins is separately optimized along the center-to-center vector, and the SDU search space is reduced to <italic>S</italic>\n²×<italic>SO</italic>(3), where <italic>S</italic>\n² denotes the surface of the unit sphere in . The receptor is fixed, with its center of mass placed at the origin of the coordinate system, and denotes the position of the center of mass of the ligand. In spherical coordinates <bold>y</bold> can be represented by (<italic>r</italic>,<italic>θ</italic>,<italic>ϕ</italic>), where <italic>r</italic> = ∥<bold>y</bold>∥, <italic>θ</italic> is the azimuth angle between the projection of <bold>y</bold> on the <italic>xy</italic>-plane and the <italic>x</italic>-axis (longitude, 0≤<italic>θ</italic>&lt;2<italic>π</italic>), and <italic>ϕ</italic> is the zenith angle between the <italic>z</italic>-axis and the vector <bold>y</bold> (colatitude, 0≤<italic>ϕ</italic>≤<italic>π</italic>). The corresponding exponential coordinates are <bold>σ</bold> = (−sin <italic>θ</italic>·<italic>ϕ</italic>, cos <italic>θ</italic>·<italic>ϕ</italic>). The rotation of the ligand is described by the exponential coordinates in <italic>SO</italic>(3).</p>", "<p>In this coordinate system the free energy function is Δ<italic>G</italic>(<italic>r</italic>,<bold>σ</bold>,<italic>ω</italic>,<bold>s</bold>), where <bold>s</bold> describes the side chain conformations. Since <bold>s</bold> and <italic>r</italic>, respectively, are determined by local minimization and by a line search along the vectors connecting the centers of mass, by SDU we minimize the functionin the (<bold>σ</bold>,<italic>ω</italic>)-space. Thus, the SDU2 algorithm uses separate independent variables for side-chain optimization, center-to-center distance of the two proteins, and five angular descriptors of the relative orientations of the molecules. As will be shown, the removal of the center-to-center distance turns out to vastly improve the efficiency of the search, since the 5-dimensional space now exhibits a well-behaved energy surface suitable for underestimation. In addition, successive underestimators obtained during the course of the SDU2 algorithm can reuse the local minima obtained in the earlier steps, thereby reducing the number of required function evaluations. Moreover, SDU2 can use all conformations contained in the cluster to be refined, while SDU1 may use very few (or none) of these points as they may not lie in the subspaces explored. In fact, we have tested SDU1 for a 10-protein set with reduced energy potentials and compared its performance with a standard Monte Carlo method in ##UREF##5##[34]##. SDU1 showed a modest speed-up factor of 2 compared with the Monte Carlo method, partly due to the issues mentioned above.</p>", "<p>It is important that for any rotational state the energy is separately minimized along the vector connecting the centers of the two molecules. Since the lowest energy is generally attained at a distance that eliminates all atomic overlaps but retains some of the favorable van der Waals interactions, after minimization the two proteins are in contact with each other. Based on this property, the SDU2 algorithm essentially samples encounter complexes ##REF##10049302##[15]##, resulting in meaningful energy values and efficient sampling. Simple arguments show that this parameterization is more natural than the sampling in the space of translations and rotations. In fact, the SDU2 strategy shows strong similarity to the model of macromolecular association in which translational diffusion brings the two proteins to a collision. Unless the enthalpy change of favorable interactions compensates for the free energy increase due to the loss of entropy, the proteins separate, rotate, and collide again. Thus, the conformational search proceeds in a series of “micro-collisions”, each resulting in an encounter complex ##REF##3890878##[37]##.</p>", "<title>Test Results for SDU-Based Docking Algorithms</title>", "<p>The SDU1 and SDU2 algorithms were tested on the protein pairs given in protein docking benchmark sets ##REF##12784372##[38]##,##REF##15981264##[39]## that contain enzyme–inhibitor, antigen–antibody, and “other” types of complexes, using the independently determined (unbound) structures of the component proteins. The algorithms were used to refine the top (most populated) clusters of docked structures generated by the rigid body docking program PIPER ##REF##16933295##[22]##. We consider the 10 largest clusters for enzyme–inhibitor complexes, and 30 clusters for antigen–antibody and “other” complexes. ##TAB##0##Table 1## shows results both from the rigid body docking and the SDU-based refinement procedures for all three types of complexes, each defined by its Protein Data Bank (PDB) code ##REF##10592235##[5]## in Column 1 of ##TAB##0##Table 1##. We emphasize that for most complexes we docked the unbound (separately crystallized) protein structures rather than their bound conformations obtained by separating the complex. The exceptions are mostly a few antigen–antibody complexes for which no separate antibody structures were available and hence were taken from the complex. However, even in these cases we used the separately crystallized structure of the antigen as given in the benchmark sets ##REF##12784372##[38]##,##REF##15981264##[39]##. PDB codes for these “semi-bound” complexes are shown in bold italic fonts in ##TAB##0##Table 1##.</p>", "<p>Columns 2–4 of ##TAB##0##Table 1## describe the docked complex structures generated by the rigid body docking ##REF##16933295##[22]## before any refinement. As described in the <xref ref-type=\"sec\" rid=\"s3\">Methods</xref>, the PIPER docking program evaluates the energy for billions of docked conformations. We retain the 1000 best scoring structures, and cluster them using the pairwise RMSD as the distance measure and an optimally selected clustering radius. We have observed that the near-native structures tend to be in one of the largest clusters, and hence rank the clusters on the basis of their size. In fact, calculated for the rigid protein structures, the energy function is approximate, and better discrimination of the native structures can be achieved by focusing on the large clusters. The properties shown in ##TAB##0##Table 1## are the number of near-native conformations (or “hits” with less than 10 Å ligand interface <italic>C_α</italic> RMSD) among the 1000 best scoring structures retained from the PIPER results, the Root Mean Square deviation (RMSD) between the native ligand structure and the docked structure at the center of the first cluster that includes a near-native conformation, and the rank of the particular cluster based on cluster size. Results are fairly good for enzyme–inhibitor complexes, as the 1000 structures, on the average, include over 200 hits, and the average RMSD is less than 5 Å between the native structure and the center of one of the three largest clusters. Although PIPER yields almost 200 hits for the “other” types of complexes, discrimination by cluster size is more difficult, and retaining the top 6 clusters results in 6.16 Å average RMSD. For antibody-antigen complexes PIPER generates much fewer hits, and on the average we have to retain 15 clusters to have a near-native structure in them.</p>", "<p>The next 6 columns in ##TAB##0##Table 1## show the results for the SDU1 and SDU2 algorithms, in each case listing the RMSD for the lowest energy conformation in the first cluster that includes a near-native structure, the rank of the cluster, and the average number <italic>n̅</italic> of function evaluations. It is important to note that after the SDU refinement the clusters are ranked based on the energies of the SDU solutions rather than on cluster size. For comparison we also generated 2000 conformations (based on the original protocol) for each complex using version 2.3 of the rigid body docking program ZDOCK ##REF##12784371##[21]##, and refined the structures using RDOCK ##REF##14579354##[40]##, which performs local energy minimization. The last two columns of ##TAB##0##Table 1## show the RMSD of the first near-native structure found and its rank based on the RDOCK energy function.</p>", "<p>For enzyme–inhibitor complexes both SDU1 and SDU2 give excellent results. On the average, the RMSD is reduced by almost 1 Å and the average rank is around 2. As the overall performance measure of the refinement we consider the number of complexes that have at least one prediction with less than 5 Å ligand interface <italic>C_α</italic> RMSD from the native structure in the top 5 clusters. Such predictions would be termed “5 Å models” later for simplicity. As noted, the clusters of PIPER-generated structures are ranked based on their size (i.e., the number of structures). However, after the refinement the clusters are ranked based on the energy of their lowest energy structures. Among the PIPER-generated structures the top 5 clusters include less than 5 Å ligand interface RMSD predictions for only 11 of the 25 enzyme–inhibitor complexes. As shown in ##TAB##0##Table 1##, both SDU1 and SDU2 increase this number to 17, i.e., a more than 50% improvement relative to the PIPER results. Notice that the 5 lowest energy predictions obtained by the ZDOCK/RDOCK procedure include 5 Å models only for 7 complexes. However, from the ZDOCK/RDOCK runs we retain low energy models, whereas the results provided by SDU1 and SDU2 are low energy clusters, and ranking clusters rather than individual structures generally removes some false positives, i.e., conformations that have low energy but are far from the native.</p>", "<p>It is well known that antigen–antibody and “other” complexes are more difficult to predict than enzyme–inhibitor complexes ##REF##15036860##[20]##. For antibody-antigen pairs only the refinement by SDU2 improves the PIPER results. Although both the RMSD and the average rank of the first near-native cluster are reduced, at 12.7 the latter remains high (##TAB##0##Table 1##). Similarly, for the complexes in the “other” category only SDU2 improves both the RMSD and the rank. Even SDU2 yields only a total of seven 5 Å models in the top five clusters for antibody–antigen and “other” types of complexes. This result is somewhat disappointing, but note that docking by ZDOCK and refining by RDOCK leads to 5 Å models only for three complexes, and no near-native solution is found in four cases.</p>", "<title>Calculated Free Energy Surfaces</title>", "<p>\n##FIG##1##Figure 2## shows the RMSD vs. calculated free energy for the 25 enzyme–inhibitor complexes. Again we show PDB codes for the “semi-bound” complexes in bold italic fonts. Each point represents a structure sampled in the process of refining the 10 largest clusters using the SDU2 algorithm. The encircled blue asterisk indicates the native structure and the first hit is shown as a red square. In spite of the one-dimensional representation of the energy function defined in the 6D rotational and translational space, the figure demonstrates the multi-funnel behavior over a relatively broad region (within 20 Å RMSD) of the native state. For most complexes the figure shows a well defined deep funnel within 5 Å RMSD from the native structure. According to ##TAB##0##Table 1##, for 12 of the 25 complexes (including 5 “semi-bound” cases), this funnel is deepest among the 10 clusters sampled. For example, for 4HTC the near-native cluster is the 7th largest, but it is energetically the most favorable after refinement. For the remaining 13 complexes, clusters farther from the native structure yield the deepest funnels, resulting in false positive predictions. One of the worst behaviors can be observed for 2PTC with the first near-native cluster ranked 4, and ##FIG##1##Figure 2## shows a number of deep non-native funnels. However, on the average, one of the 3 deepest funnels is near-native (##TAB##0##Table 1##). This also shows the power of heuristic combination of entropic and enthalpic measurement, i.e., cluster size as a filter and refined cluster depth as ranking parameter.</p>", "<p>\n##FIG##2##Figure 3## shows the RMSD vs. calculated free energy for the antigen–antibody and “other” complexes, sampled in the refinement of the 30 largest clusters by SDU2. All but 1SPB and 2BTF have multiple funnels, and the funnels are further from the native state than for enzyme–inhibitor pairs, demonstrating the well-known difficulty of estimating free energy using simple models, particularly for antigen–antibody complexes. Adjusting the conformation of interface side chains only by local minimization and keeping the backbone rigid also limits the accuracy of free energy calculation.</p>", "<title>The Effect of Space Selection on Search Efficiency</title>", "<p>As shown in ##TAB##0##Table 1##, the SDU2 strategy is much more efficient than SDU1, and also provides substantially better results for the more difficult problem of docking antigen–antibody and “other” complexes. These differences are demonstrated in ##FIG##3##Figure 4A and 4B## that show, for the near-native cluster of the complex 4HTC, the conformations sampled and minimized by each algorithm. The horizontal and vertical axes, respectively, represent translational and rotational distances between each sampled ligand conformation and the one in the native structure, the latter placed at the origin of this coordinate system. The rotational distance here is defined as the length of the minimum geodesic (i.e., the minimum rotation in radian) between two rotations ##UREF##7##[36]##. As discussed earlier, this distance between <bold>R</bold>\n₀ and <bold>R</bold>\n₁ can be defined by <italic>ρ</italic>(<bold>R</bold>\n₀,<bold>R</bold>\n₁) = ∥log(<bold>R</bold>\n₀′<bold>R</bold>\n₁)∥ = ∥<italic>ω</italic>\n₁−<italic>ω</italic>\n₀∥. The points are color-coded according to their energies, from low (blue) to high (red) energy structures.</p>", "<p>As shown in ##FIG##3##Figure 4A##, the separate treatment of translational and rotational subspaces in the SDU1 algorithm is highly inefficient, resulting in the sampling of many conformations with relatively high energies. Although all energy values shown are obtained by local minimization, the latter is unable to reduce the energy if the two proteins are too far from each other, which frequently occurs with the SDU1 algorithm. In addition, as indicated by the parallel lines, very similar translational regions are re-sampled at slightly different rotational coordinates, and vice versa. In contrast, as shown in ##FIG##3##Figure 4B##, the SDU2 algorithm smoothly and efficiently descends toward the bottom of the free energy funnel. Based on the ranges of sampled free energy values, SDU2 sampled much lower energy regions compared to SDU1. Since the search is restricted to biophysically meaningful encounter complexes, the more consistent energy values facilitate the construction of better underestimators during the search. According to ##FIG##3##Figure 4B##, for SDU2 there is a clear trend that structures closer to the native complex generally have lower energies, resulting in a deep and broad free energy funnel in this space. The existence of such free energy funnels is much less obvious, even in each separate subspace, when sampled by the SDU1 algorithm (##FIG##3##Figure 4A##).</p>", "<p>In test docking problems even the slower SDU1 algorithm outperformed a standard Monte Carlo method by reducing the number of function evaluations by a factor of two ##UREF##5##[34]##. As shown in ##TAB##0##Table 1##, SDU2 further reduces the computational costs by <italic>a factor of</italic> 11 <italic>to</italic> 15, depending on the type of complex. Since most of the computational time is spent in energy evaluations, the computational gain of SDU2 over SDU1 is more than a factor of 10, and we estimate that SDU2 achieves more than 20-fold efficiency gain compared to Monte Carlo methods. We tested our algorithms on a 128-node biowulf cluster (IBM eServer×Series). Each node contains dual 1 GHz PIII processors with 2 GB memory. A typical refinement by SDU2 for each PIPER-generated cluster would take 2 to 6 processor-hours. The running time varies with the protein complex size (especially the interface size) and the number of iterations before convergence. Notice that on the average SDU2 samples only about 100 encounter structures for each cluster. No particular efforts have been made to accelerate either the interface side-chain search or the line search to determine the center-to-center distance. The CPU times can be compared to those reported for a server ##REF##18442991##[41]## based on the RosettaDock algorithm ##REF##12875852##[42]##. The server performs 1000 independent Monte Carlo simulations within 30 Å <italic>C_α</italic> RMSD of a starting structure, as described in the RosettaDock protocol ##REF##12875852##[42]##. A typical run requires about 65 processor-hours.</p>", "<title>Repeatability of the SDU-Based Docking Algorithm</title>", "<p>SDU based docking algorithms are stochastic in nature. Although we gave a theoretical guarantee of probabilistic convergence to the global minimum of funnel-like functions under some fairly general conditions ##REF##19759849##[30]##, practical protein docking problems do not necessarily satisfy all these conditions. To check the variations in our results we have repeatedly run SDU2 to refine the near-native cluster from PIPER (columns 3 and 4 in ##TAB##0##Table 1##) for 10 randomly selected complexes, including 4 enzyme–inhibitor, 4 antigen–antibody and 2 “other” complexes. SDU2 was run 10 times independently for each complex with the same set of parameters. The results (ligand interface <italic>C_α</italic> RMSD versus corresponding free energy values) are shown in ##FIG##4##Figure 5##, where black circles represent the rigid body predictions from PIPER before refinement and blue asterisks represent the independent SDU2 predictions. Some of the latter overlap, resulting in less than 10 distinct solutions. SDU2 is able to lower the free energy values in all cases and to improve RMSD in most of them. ##TAB##1##Table 2## shows the mean (indicated by overline) and the standard deviation (indicated by <italic>σ</italic>) for both the ligand interface RMSD and the free energy for each complex. Although the average standard deviation is relatively large (about 4.4 Kcal/mol) for free energies, it is less than 0.5 Å for RMSDs. This result indicates a good level of robustness, considering that our free energy model does not discriminate among structures within 1 Å RMSD from each other. Note that the SDU-based algorithms rely more on the collective distribution of a set of encounter structures in the free energy funnel, rather than on a single low energy structure, which reduces the sensitivity of the results to the variations in the starting structures.</p>", "<title>Conclusions</title>", "<p>The successful application of the Semi-Definite programming-based Underestimation (SDU) search algorithm to protein–protein docking further validates the assumption that the free energy landscape of the complex is a funnel in some neighborhood of the native state. However, the direct application of SDU in the space <italic>SE</italic>(3) of rotations and translations fails to yield useful underestimators. Alternating searches in rotational and translational subspaces yields a feasible but inefficient algorithm. We have substantially improved performance by separately optimizing the center-to-center distance and describing <italic>SE</italic>(3) in terms of five angles. It is potentially important that this strategy samples encounter complexes, and hence it is reminiscent of the model of molecular association through a series of micro-collisions ##REF##3890878##[37]##. Results emphasize that the funnel-like shape of the free energy surface seen in this parameterization of <italic>SE</italic>(3) is largely lost when changing to the straightforward description of the space in terms of rotational and translational coordinates.</p>", "<p>The underestimation approach has been used in the latest rounds of CAPRI with considerable success ##REF##17918726##[19]##,##REF##17853451##[43]##, and it provides a promising platform for improving docking methods. We note that Marcia et al. ##UREF##8##[44]## recently reported the application of SDU to the docking problem using the general quadratic underestimation method we have earlier developed ##REF##19759849##[30]##,##UREF##3##[31]##. However, the central problem of parameterizing the search space was not discussed and the method was applied only to five bound docking problems using co-crystallized structures, which is much easier than docking two separately crystallized proteins. In fact, we believe that the separate minimization along the center-to-center vector and the use of five angular descriptors can improve the performance of any minimization algorithm used for docking.</p>" ]

[ "<title>Results/Discussion</title>", "<title>Development of SDU-Based Docking Algorithms</title>", "<p>To define the docking problem we fix the position and orientation of the first (receptor) protein. The six-dimensional vector <bold>x</bold> specifies the position and orientation of the second (ligand) protein. The variables <bold>s</bold> account for the side chain conformations in both proteins, and Δ<italic>G</italic>(<bold>x</bold>,<bold>s</bold>) denotes the free energy function. The minimization of Δ<italic>G</italic>(<bold>x</bold>,<bold>s</bold>) with respect to <bold>s</bold> is restricted to the side chains in the interface and is carried out by local methods. We define Δ<italic>G</italic>*(<bold>x</bold>) = Δ<italic>G</italic>(<bold>x</bold>,<bold>s</bold>*), where <bold>s</bold>* is an optimal solution; then the protein docking problem is finding the lowest minimum of Δ<italic>G</italic>*(<bold>x</bold>) in the region defined by the cluster, where <bold>x</bold> belongs to the space of rigid body motions, i.e., the Euclidean group <italic>SE</italic>(3). As described before, the parameterization of <italic>SE</italic>(3) is critical for optimization purposes, and hence we first describe our results concerning the parameterization of the search space.</p>", "<p>The space <italic>SE</italic>(3) is the semidirect product of (translations) and <italic>SO</italic>(3) (rotations). The rotation group <italic>SO</italic>(3) is a 3-dimensional Lie group consisting of rotation matrices, i.e, , <bold>RR</bold>′ = <bold>I</bold>, det(<bold>R</bold>) = 1. The Lie algebra of <italic>SO</italic>(3), denoted by <italic>so</italic>(3), may be represented by the real skew-symmetric matriceswhere . It is well-known that the one-parameter subgroups of <italic>SO</italic>(3), i.e., are geodesics ##UREF##4##[33]##,##UREF##6##[35]##,##UREF##7##[36]##, i.e., the shortest paths between two points on <italic>SO</italic>(3). Moreover, for <bold>R</bold>\n₀, <bold>R</bold>\n₁∈<italic>SO</italic>(3), Ω = log(<bold>R</bold>\n₀′<bold>R</bold>\n₁)∈<italic>so</italic>(3) and the distance between <bold>R</bold>\n₀ and <bold>R</bold>\n₁ can be defined by <italic>ρ</italic>(<bold>R</bold>\n₀,<bold>R</bold>\n₁) = ∥log(<bold>R</bold>\n₀′<bold>R</bold>\n₁)∥ = ∥<italic>ω</italic>\n₁−<italic>ω</italic>\n₀∥. This distance is a natural Riemannian metric on <italic>SO</italic>(3), i.e., it is bi-invariant with respect to the actions of the group (rotations). The exponential map from <italic>so</italic>(3) to <italic>SO</italic>(3) defined by Ω → <bold>R</bold>\n₀\n<italic>e</italic>\n^Ω is a local diffeomorphism, i.e., there exist an open neighborhood of 0∈<italic>so</italic>(3) and an open neighborhood of <bold>R</bold>\n₀∈<italic>SO</italic>(3) and an invertible, surjective, and smooth map from one neighborhood to the other whose inverse is also smooth. The local diffeomorphism induces a coordinate chart in a neighborhood of <bold>R</bold>\n₀ that is known as exponential coordinate system. Given the definition of <italic>so</italic>(3) this coordinate system can be parameterized by .</p>", "<p>Given the favorable properties of <italic>SO</italic>(3) (more generally <italic>SO</italic>(<italic>n</italic>)) such as the existence of a natural bi-invariant metric and, in particular, the simplicity of determining the geodesics of the manifold, the standard gradient based optimization algorithms on can be generalized for optimization on <italic>SO</italic>(<italic>n</italic>) ##UREF##6##[35]##. However, generalization to the entire <italic>SE</italic>(3) is more difficult ##UREF##4##[33]##,##UREF##7##[36]##. Although <italic>SE</italic>(3) is also Lie group, its one-parameter subgroups are no longer geodesics. Moreover, <italic>SE</italic>(3) does not admit a natural (bi-invariant) Riemannian metric while its subspaces <italic>SO</italic>(3) and do ##UREF##4##[33]##,##UREF##7##[36]##. Attempting to implement SDU to <italic>SE</italic>(3) by simultaneous translational and rotational optimization, we have generally failed to construct useful underestimators. We have, however, found two strategies that were able to overcome this problem.</p>", "<title>SDU1: Cyclic coordinate descent</title>", "<p>A natural search strategy in <italic>SE</italic>(3) is to alternate between optimizing the free energy in <italic>SO</italic>(3) and in by a series of rotational and translational adjustments. As will be further discussed, the major disadvantage of this approach is that samples in one subspace cannot be reused in the other subspace, resulting in inefficient search.</p>", "<title>SDU2: 5D search in the space of encounter complexes</title>", "<p>The distance between the two proteins is separately optimized along the center-to-center vector, and the SDU search space is reduced to <italic>S</italic>\n²×<italic>SO</italic>(3), where <italic>S</italic>\n² denotes the surface of the unit sphere in . The receptor is fixed, with its center of mass placed at the origin of the coordinate system, and denotes the position of the center of mass of the ligand. In spherical coordinates <bold>y</bold> can be represented by (<italic>r</italic>,<italic>θ</italic>,<italic>ϕ</italic>), where <italic>r</italic> = ∥<bold>y</bold>∥, <italic>θ</italic> is the azimuth angle between the projection of <bold>y</bold> on the <italic>xy</italic>-plane and the <italic>x</italic>-axis (longitude, 0≤<italic>θ</italic>&lt;2<italic>π</italic>), and <italic>ϕ</italic> is the zenith angle between the <italic>z</italic>-axis and the vector <bold>y</bold> (colatitude, 0≤<italic>ϕ</italic>≤<italic>π</italic>). The corresponding exponential coordinates are <bold>σ</bold> = (−sin <italic>θ</italic>·<italic>ϕ</italic>, cos <italic>θ</italic>·<italic>ϕ</italic>). The rotation of the ligand is described by the exponential coordinates in <italic>SO</italic>(3).</p>", "<p>In this coordinate system the free energy function is Δ<italic>G</italic>(<italic>r</italic>,<bold>σ</bold>,<italic>ω</italic>,<bold>s</bold>), where <bold>s</bold> describes the side chain conformations. Since <bold>s</bold> and <italic>r</italic>, respectively, are determined by local minimization and by a line search along the vectors connecting the centers of mass, by SDU we minimize the functionin the (<bold>σ</bold>,<italic>ω</italic>)-space. Thus, the SDU2 algorithm uses separate independent variables for side-chain optimization, center-to-center distance of the two proteins, and five angular descriptors of the relative orientations of the molecules. As will be shown, the removal of the center-to-center distance turns out to vastly improve the efficiency of the search, since the 5-dimensional space now exhibits a well-behaved energy surface suitable for underestimation. In addition, successive underestimators obtained during the course of the SDU2 algorithm can reuse the local minima obtained in the earlier steps, thereby reducing the number of required function evaluations. Moreover, SDU2 can use all conformations contained in the cluster to be refined, while SDU1 may use very few (or none) of these points as they may not lie in the subspaces explored. In fact, we have tested SDU1 for a 10-protein set with reduced energy potentials and compared its performance with a standard Monte Carlo method in ##UREF##5##[34]##. SDU1 showed a modest speed-up factor of 2 compared with the Monte Carlo method, partly due to the issues mentioned above.</p>", "<p>It is important that for any rotational state the energy is separately minimized along the vector connecting the centers of the two molecules. Since the lowest energy is generally attained at a distance that eliminates all atomic overlaps but retains some of the favorable van der Waals interactions, after minimization the two proteins are in contact with each other. Based on this property, the SDU2 algorithm essentially samples encounter complexes ##REF##10049302##[15]##, resulting in meaningful energy values and efficient sampling. Simple arguments show that this parameterization is more natural than the sampling in the space of translations and rotations. In fact, the SDU2 strategy shows strong similarity to the model of macromolecular association in which translational diffusion brings the two proteins to a collision. Unless the enthalpy change of favorable interactions compensates for the free energy increase due to the loss of entropy, the proteins separate, rotate, and collide again. Thus, the conformational search proceeds in a series of “micro-collisions”, each resulting in an encounter complex ##REF##3890878##[37]##.</p>", "<title>Test Results for SDU-Based Docking Algorithms</title>", "<p>The SDU1 and SDU2 algorithms were tested on the protein pairs given in protein docking benchmark sets ##REF##12784372##[38]##,##REF##15981264##[39]## that contain enzyme–inhibitor, antigen–antibody, and “other” types of complexes, using the independently determined (unbound) structures of the component proteins. The algorithms were used to refine the top (most populated) clusters of docked structures generated by the rigid body docking program PIPER ##REF##16933295##[22]##. We consider the 10 largest clusters for enzyme–inhibitor complexes, and 30 clusters for antigen–antibody and “other” complexes. ##TAB##0##Table 1## shows results both from the rigid body docking and the SDU-based refinement procedures for all three types of complexes, each defined by its Protein Data Bank (PDB) code ##REF##10592235##[5]## in Column 1 of ##TAB##0##Table 1##. We emphasize that for most complexes we docked the unbound (separately crystallized) protein structures rather than their bound conformations obtained by separating the complex. The exceptions are mostly a few antigen–antibody complexes for which no separate antibody structures were available and hence were taken from the complex. However, even in these cases we used the separately crystallized structure of the antigen as given in the benchmark sets ##REF##12784372##[38]##,##REF##15981264##[39]##. PDB codes for these “semi-bound” complexes are shown in bold italic fonts in ##TAB##0##Table 1##.</p>", "<p>Columns 2–4 of ##TAB##0##Table 1## describe the docked complex structures generated by the rigid body docking ##REF##16933295##[22]## before any refinement. As described in the <xref ref-type=\"sec\" rid=\"s3\">Methods</xref>, the PIPER docking program evaluates the energy for billions of docked conformations. We retain the 1000 best scoring structures, and cluster them using the pairwise RMSD as the distance measure and an optimally selected clustering radius. We have observed that the near-native structures tend to be in one of the largest clusters, and hence rank the clusters on the basis of their size. In fact, calculated for the rigid protein structures, the energy function is approximate, and better discrimination of the native structures can be achieved by focusing on the large clusters. The properties shown in ##TAB##0##Table 1## are the number of near-native conformations (or “hits” with less than 10 Å ligand interface <italic>C_α</italic> RMSD) among the 1000 best scoring structures retained from the PIPER results, the Root Mean Square deviation (RMSD) between the native ligand structure and the docked structure at the center of the first cluster that includes a near-native conformation, and the rank of the particular cluster based on cluster size. Results are fairly good for enzyme–inhibitor complexes, as the 1000 structures, on the average, include over 200 hits, and the average RMSD is less than 5 Å between the native structure and the center of one of the three largest clusters. Although PIPER yields almost 200 hits for the “other” types of complexes, discrimination by cluster size is more difficult, and retaining the top 6 clusters results in 6.16 Å average RMSD. For antibody-antigen complexes PIPER generates much fewer hits, and on the average we have to retain 15 clusters to have a near-native structure in them.</p>", "<p>The next 6 columns in ##TAB##0##Table 1## show the results for the SDU1 and SDU2 algorithms, in each case listing the RMSD for the lowest energy conformation in the first cluster that includes a near-native structure, the rank of the cluster, and the average number <italic>n̅</italic> of function evaluations. It is important to note that after the SDU refinement the clusters are ranked based on the energies of the SDU solutions rather than on cluster size. For comparison we also generated 2000 conformations (based on the original protocol) for each complex using version 2.3 of the rigid body docking program ZDOCK ##REF##12784371##[21]##, and refined the structures using RDOCK ##REF##14579354##[40]##, which performs local energy minimization. The last two columns of ##TAB##0##Table 1## show the RMSD of the first near-native structure found and its rank based on the RDOCK energy function.</p>", "<p>For enzyme–inhibitor complexes both SDU1 and SDU2 give excellent results. On the average, the RMSD is reduced by almost 1 Å and the average rank is around 2. As the overall performance measure of the refinement we consider the number of complexes that have at least one prediction with less than 5 Å ligand interface <italic>C_α</italic> RMSD from the native structure in the top 5 clusters. Such predictions would be termed “5 Å models” later for simplicity. As noted, the clusters of PIPER-generated structures are ranked based on their size (i.e., the number of structures). However, after the refinement the clusters are ranked based on the energy of their lowest energy structures. Among the PIPER-generated structures the top 5 clusters include less than 5 Å ligand interface RMSD predictions for only 11 of the 25 enzyme–inhibitor complexes. As shown in ##TAB##0##Table 1##, both SDU1 and SDU2 increase this number to 17, i.e., a more than 50% improvement relative to the PIPER results. Notice that the 5 lowest energy predictions obtained by the ZDOCK/RDOCK procedure include 5 Å models only for 7 complexes. However, from the ZDOCK/RDOCK runs we retain low energy models, whereas the results provided by SDU1 and SDU2 are low energy clusters, and ranking clusters rather than individual structures generally removes some false positives, i.e., conformations that have low energy but are far from the native.</p>", "<p>It is well known that antigen–antibody and “other” complexes are more difficult to predict than enzyme–inhibitor complexes ##REF##15036860##[20]##. For antibody-antigen pairs only the refinement by SDU2 improves the PIPER results. Although both the RMSD and the average rank of the first near-native cluster are reduced, at 12.7 the latter remains high (##TAB##0##Table 1##). Similarly, for the complexes in the “other” category only SDU2 improves both the RMSD and the rank. Even SDU2 yields only a total of seven 5 Å models in the top five clusters for antibody–antigen and “other” types of complexes. This result is somewhat disappointing, but note that docking by ZDOCK and refining by RDOCK leads to 5 Å models only for three complexes, and no near-native solution is found in four cases.</p>", "<title>Calculated Free Energy Surfaces</title>", "<p>\n##FIG##1##Figure 2## shows the RMSD vs. calculated free energy for the 25 enzyme–inhibitor complexes. Again we show PDB codes for the “semi-bound” complexes in bold italic fonts. Each point represents a structure sampled in the process of refining the 10 largest clusters using the SDU2 algorithm. The encircled blue asterisk indicates the native structure and the first hit is shown as a red square. In spite of the one-dimensional representation of the energy function defined in the 6D rotational and translational space, the figure demonstrates the multi-funnel behavior over a relatively broad region (within 20 Å RMSD) of the native state. For most complexes the figure shows a well defined deep funnel within 5 Å RMSD from the native structure. According to ##TAB##0##Table 1##, for 12 of the 25 complexes (including 5 “semi-bound” cases), this funnel is deepest among the 10 clusters sampled. For example, for 4HTC the near-native cluster is the 7th largest, but it is energetically the most favorable after refinement. For the remaining 13 complexes, clusters farther from the native structure yield the deepest funnels, resulting in false positive predictions. One of the worst behaviors can be observed for 2PTC with the first near-native cluster ranked 4, and ##FIG##1##Figure 2## shows a number of deep non-native funnels. However, on the average, one of the 3 deepest funnels is near-native (##TAB##0##Table 1##). This also shows the power of heuristic combination of entropic and enthalpic measurement, i.e., cluster size as a filter and refined cluster depth as ranking parameter.</p>", "<p>\n##FIG##2##Figure 3## shows the RMSD vs. calculated free energy for the antigen–antibody and “other” complexes, sampled in the refinement of the 30 largest clusters by SDU2. All but 1SPB and 2BTF have multiple funnels, and the funnels are further from the native state than for enzyme–inhibitor pairs, demonstrating the well-known difficulty of estimating free energy using simple models, particularly for antigen–antibody complexes. Adjusting the conformation of interface side chains only by local minimization and keeping the backbone rigid also limits the accuracy of free energy calculation.</p>", "<title>The Effect of Space Selection on Search Efficiency</title>", "<p>As shown in ##TAB##0##Table 1##, the SDU2 strategy is much more efficient than SDU1, and also provides substantially better results for the more difficult problem of docking antigen–antibody and “other” complexes. These differences are demonstrated in ##FIG##3##Figure 4A and 4B## that show, for the near-native cluster of the complex 4HTC, the conformations sampled and minimized by each algorithm. The horizontal and vertical axes, respectively, represent translational and rotational distances between each sampled ligand conformation and the one in the native structure, the latter placed at the origin of this coordinate system. The rotational distance here is defined as the length of the minimum geodesic (i.e., the minimum rotation in radian) between two rotations ##UREF##7##[36]##. As discussed earlier, this distance between <bold>R</bold>\n₀ and <bold>R</bold>\n₁ can be defined by <italic>ρ</italic>(<bold>R</bold>\n₀,<bold>R</bold>\n₁) = ∥log(<bold>R</bold>\n₀′<bold>R</bold>\n₁)∥ = ∥<italic>ω</italic>\n₁−<italic>ω</italic>\n₀∥. The points are color-coded according to their energies, from low (blue) to high (red) energy structures.</p>", "<p>As shown in ##FIG##3##Figure 4A##, the separate treatment of translational and rotational subspaces in the SDU1 algorithm is highly inefficient, resulting in the sampling of many conformations with relatively high energies. Although all energy values shown are obtained by local minimization, the latter is unable to reduce the energy if the two proteins are too far from each other, which frequently occurs with the SDU1 algorithm. In addition, as indicated by the parallel lines, very similar translational regions are re-sampled at slightly different rotational coordinates, and vice versa. In contrast, as shown in ##FIG##3##Figure 4B##, the SDU2 algorithm smoothly and efficiently descends toward the bottom of the free energy funnel. Based on the ranges of sampled free energy values, SDU2 sampled much lower energy regions compared to SDU1. Since the search is restricted to biophysically meaningful encounter complexes, the more consistent energy values facilitate the construction of better underestimators during the search. According to ##FIG##3##Figure 4B##, for SDU2 there is a clear trend that structures closer to the native complex generally have lower energies, resulting in a deep and broad free energy funnel in this space. The existence of such free energy funnels is much less obvious, even in each separate subspace, when sampled by the SDU1 algorithm (##FIG##3##Figure 4A##).</p>", "<p>In test docking problems even the slower SDU1 algorithm outperformed a standard Monte Carlo method by reducing the number of function evaluations by a factor of two ##UREF##5##[34]##. As shown in ##TAB##0##Table 1##, SDU2 further reduces the computational costs by <italic>a factor of</italic> 11 <italic>to</italic> 15, depending on the type of complex. Since most of the computational time is spent in energy evaluations, the computational gain of SDU2 over SDU1 is more than a factor of 10, and we estimate that SDU2 achieves more than 20-fold efficiency gain compared to Monte Carlo methods. We tested our algorithms on a 128-node biowulf cluster (IBM eServer×Series). Each node contains dual 1 GHz PIII processors with 2 GB memory. A typical refinement by SDU2 for each PIPER-generated cluster would take 2 to 6 processor-hours. The running time varies with the protein complex size (especially the interface size) and the number of iterations before convergence. Notice that on the average SDU2 samples only about 100 encounter structures for each cluster. No particular efforts have been made to accelerate either the interface side-chain search or the line search to determine the center-to-center distance. The CPU times can be compared to those reported for a server ##REF##18442991##[41]## based on the RosettaDock algorithm ##REF##12875852##[42]##. The server performs 1000 independent Monte Carlo simulations within 30 Å <italic>C_α</italic> RMSD of a starting structure, as described in the RosettaDock protocol ##REF##12875852##[42]##. A typical run requires about 65 processor-hours.</p>", "<title>Repeatability of the SDU-Based Docking Algorithm</title>", "<p>SDU based docking algorithms are stochastic in nature. Although we gave a theoretical guarantee of probabilistic convergence to the global minimum of funnel-like functions under some fairly general conditions ##REF##19759849##[30]##, practical protein docking problems do not necessarily satisfy all these conditions. To check the variations in our results we have repeatedly run SDU2 to refine the near-native cluster from PIPER (columns 3 and 4 in ##TAB##0##Table 1##) for 10 randomly selected complexes, including 4 enzyme–inhibitor, 4 antigen–antibody and 2 “other” complexes. SDU2 was run 10 times independently for each complex with the same set of parameters. The results (ligand interface <italic>C_α</italic> RMSD versus corresponding free energy values) are shown in ##FIG##4##Figure 5##, where black circles represent the rigid body predictions from PIPER before refinement and blue asterisks represent the independent SDU2 predictions. Some of the latter overlap, resulting in less than 10 distinct solutions. SDU2 is able to lower the free energy values in all cases and to improve RMSD in most of them. ##TAB##1##Table 2## shows the mean (indicated by overline) and the standard deviation (indicated by <italic>σ</italic>) for both the ligand interface RMSD and the free energy for each complex. Although the average standard deviation is relatively large (about 4.4 Kcal/mol) for free energies, it is less than 0.5 Å for RMSDs. This result indicates a good level of robustness, considering that our free energy model does not discriminate among structures within 1 Å RMSD from each other. Note that the SDU-based algorithms rely more on the collective distribution of a set of encounter structures in the free energy funnel, rather than on a single low energy structure, which reduces the sensitivity of the results to the variations in the starting structures.</p>", "<title>Conclusions</title>", "<p>The successful application of the Semi-Definite programming-based Underestimation (SDU) search algorithm to protein–protein docking further validates the assumption that the free energy landscape of the complex is a funnel in some neighborhood of the native state. However, the direct application of SDU in the space <italic>SE</italic>(3) of rotations and translations fails to yield useful underestimators. Alternating searches in rotational and translational subspaces yields a feasible but inefficient algorithm. We have substantially improved performance by separately optimizing the center-to-center distance and describing <italic>SE</italic>(3) in terms of five angles. It is potentially important that this strategy samples encounter complexes, and hence it is reminiscent of the model of molecular association through a series of micro-collisions ##REF##3890878##[37]##. Results emphasize that the funnel-like shape of the free energy surface seen in this parameterization of <italic>SE</italic>(3) is largely lost when changing to the straightforward description of the space in terms of rotational and translational coordinates.</p>", "<p>The underestimation approach has been used in the latest rounds of CAPRI with considerable success ##REF##17918726##[19]##,##REF##17853451##[43]##, and it provides a promising platform for improving docking methods. We note that Marcia et al. ##UREF##8##[44]## recently reported the application of SDU to the docking problem using the general quadratic underestimation method we have earlier developed ##REF##19759849##[30]##,##UREF##3##[31]##. However, the central problem of parameterizing the search space was not discussed and the method was applied only to five bound docking problems using co-crystallized structures, which is much easier than docking two separately crystallized proteins. In fact, we believe that the separate minimization along the center-to-center vector and the use of five angular descriptors can improve the performance of any minimization algorithm used for docking.</p>" ]

[ "<title>Conclusions</title>", "<p>The successful application of the Semi-Definite programming-based Underestimation (SDU) search algorithm to protein–protein docking further validates the assumption that the free energy landscape of the complex is a funnel in some neighborhood of the native state. However, the direct application of SDU in the space <italic>SE</italic>(3) of rotations and translations fails to yield useful underestimators. Alternating searches in rotational and translational subspaces yields a feasible but inefficient algorithm. We have substantially improved performance by separately optimizing the center-to-center distance and describing <italic>SE</italic>(3) in terms of five angles. It is potentially important that this strategy samples encounter complexes, and hence it is reminiscent of the model of molecular association through a series of micro-collisions ##REF##3890878##[37]##. Results emphasize that the funnel-like shape of the free energy surface seen in this parameterization of <italic>SE</italic>(3) is largely lost when changing to the straightforward description of the space in terms of rotational and translational coordinates.</p>", "<p>The underestimation approach has been used in the latest rounds of CAPRI with considerable success ##REF##17918726##[19]##,##REF##17853451##[43]##, and it provides a promising platform for improving docking methods. We note that Marcia et al. ##UREF##8##[44]## recently reported the application of SDU to the docking problem using the general quadratic underestimation method we have earlier developed ##REF##19759849##[30]##,##UREF##3##[31]##. However, the central problem of parameterizing the search space was not discussed and the method was applied only to five bound docking problems using co-crystallized structures, which is much easier than docking two separately crystallized proteins. In fact, we believe that the separate minimization along the center-to-center vector and the use of five angular descriptors can improve the performance of any minimization algorithm used for docking.</p>" ]

[ "<p>Conceived and designed the experiments: YS ICP PV SV. Performed the experiments: YS. Analyzed the data: YS. Wrote the paper: YS ICP PV SV.</p>", "<p>Similarly to protein folding, the association of two proteins is driven by a free energy funnel, determined by favorable interactions in some neighborhood of the native state. We describe a docking method based on stochastic global minimization of funnel-shaped energy functions in the space of rigid body motions (<italic>SE</italic>(3)) while accounting for flexibility of the interface side chains. The method, called semi-definite programming-based underestimation (SDU), employs a general quadratic function to underestimate a set of local energy minima and uses the resulting underestimator to bias further sampling. While SDU effectively minimizes functions with funnel-shaped basins, its application to docking in the rotational and translational space <italic>SE</italic>(3) is not straightforward due to the geometry of that space. We introduce a strategy that uses separate independent variables for side-chain optimization, center-to-center distance of the two proteins, and five angular descriptors of the relative orientations of the molecules. The removal of the center-to-center distance turns out to vastly improve the efficiency of the search, because the five-dimensional space now exhibits a well-behaved energy surface suitable for underestimation. This algorithm explores the free energy surface spanned by encounter complexes that correspond to local free energy minima and shows similarity to the model of macromolecular association that proceeds through a series of collisions. Results for standard protein docking benchmarks establish that in this space the free energy landscape is a funnel in a reasonably broad neighborhood of the native state and that the SDU strategy can generate docking predictions with less than 5 Å ligand interface <italic>C_α</italic> root-mean-square deviation while achieving an approximately 20-fold efficiency gain compared to Monte Carlo methods.</p>", "<title>Author Summary</title>", "<p>Protein–protein interactions play a central role in various aspects of the structural and functional organization of the cell, and their elucidation is crucial for a better understanding of processes such as metabolic control, signal transduction, and gene regulation. Genomewide proteomics studies, primarily yeast two-hybrid assays, will provide an increasing list of interacting proteins, but only a small fraction of the potential complexes will be amenable to direct experimental analysis. Thus, it is important to develop computational docking methods that can elucidate the details of specific interactions at the atomic level. Protein–protein docking generally starts with a rigid body search that generates a large number of docked conformations with good shape, electrostatic, and chemical complementarity. The conformations are clustered to obtain a manageable number of models, but the current methods are unable to select the most likely structure among these models. Here we describe a refinement algorithm that, applied to the individual clusters, improves the quality of the models. The better models are suitable for higher-accuracy energy calculation, thereby increasing the chances that near-native structures can be identified, and thus the refinement increases the reliability of the entire docking algorithm.</p>" ]

[]

[]

[ "<fig id=\"pcbi-1000191-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pcbi.1000191.g001</object-id><label>Figure 1</label><caption><title>Funnel-like function and underestimator at a set of local minima indicated by small squares.</title></caption></fig>", "<fig id=\"pcbi-1000191-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pcbi.1000191.g002</object-id><label>Figure 2</label><caption><title>RMSD vs. energy plots for SDU2 sampled structures in the near-native funnel for enzyme–inhibitor complexes.</title><p>The native structure is indicated by a blue circled asterisk and the SDU2 prediction by a red square.</p></caption></fig>", "<fig id=\"pcbi-1000191-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pcbi.1000191.g003</object-id><label>Figure 3</label><caption><title>RMSD vs. energy plots for SDU2 sampled structures in the near-native funnel for antigen–antibody or other complexes.</title><p>The native structure is indicated by a blue circled asterisk and the SDU2 prediction by a red square.</p></caption></fig>", "<fig id=\"pcbi-1000191-g004\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pcbi.1000191.g004</object-id><label>Figure 4</label><caption><title>SDU1 (A) and SDU2 (B) trajectories in the near-native cluster for 4HTC.</title><p>Translational and rotational distances plotted in the horizontal and vertical axes, are defined as ∥<bold>r</bold>\n¹−<bold>r</bold>\n⁰∥ and ∥<italic>ω</italic>\n¹−<italic>ω</italic>\n⁰∥, respectively, where (<bold>r</bold>\n¹,<italic>ω</italic>\n¹) denotes the position and orientation of the ligand and (<bold>r</bold>\n⁰,<italic>ω</italic>\n⁰) are the corresponding values in the native state. The points are color-coded depending on their energy (in Kcal/mol as shown on the right), and the points in the squares show the best local minimum in each iteration.</p></caption></fig>", "<fig id=\"pcbi-1000191-g005\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pcbi.1000191.g005</object-id><label>Figure 5</label><caption><title>RMSD vs. energy plots for predictions from 10 independent SDU2 runs for the first near-native clusters of 10 randomly chosen protein complexes.</title><p>The rigid body prediction before SDU2 refinement is indicated by a black circle and the SDU2 predictions by blue asterisks. (Note that some of the latter points overlap.)</p></caption></fig>" ]

[ "<table-wrap id=\"pcbi-1000191-t001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pcbi.1000191.t001</object-id><label>Table 1</label><caption><title>Docking and refinement results.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Complex<xref ref-type=\"table-fn\" rid=\"nt101\">a</xref>\n</td><td colspan=\"3\" align=\"left\" rowspan=\"1\">PIPER</td><td colspan=\"3\" align=\"left\" rowspan=\"1\">SDU1</td><td colspan=\"3\" align=\"left\" rowspan=\"1\">SDU2</td><td colspan=\"2\" align=\"left\" rowspan=\"1\">ZDOCK+RDOCK</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Hits</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">RMSD</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Rank</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">RMSD</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Rank</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>n̅</italic>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">RMSD</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Rank</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>n̅</italic>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">RMSD</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Rank</td></tr></thead><tbody><tr><td colspan=\"12\" align=\"left\" rowspan=\"1\">\n<bold>Enzyme–Inhibitor Complexes</bold>\n</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1ACB</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">632</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.11</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.86</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1425</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.78</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">93.8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.66</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1AVW</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">81</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.87</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.31</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1381</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.54</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">90.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.50</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1AVX</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">193</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.27</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.57</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1393</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.18</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">91.2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.62</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">39</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1BRC</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">375</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8.60</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9.97</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1431</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.50</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">98.5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8.92</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1BVN</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">443</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.65</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.95</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1498</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.11</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">81.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.78</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1CGI</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">477</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8.54</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.77</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1425</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.40</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9.03</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">16</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1CHO</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">510</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.10</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.08</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1475</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.64</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">91.4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.25</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1CSE</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">45</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.15</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.03</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1340</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.99</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">97.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.83</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1DFJ</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">30</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.59</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.80</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1378</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.20</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">97.6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.46</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1E6E</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">55</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.88</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.85</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1506</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.12</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">84.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.39</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1EAW</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">114</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.02</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.03</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1442</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.96</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">97.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.21</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">30</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1MAH</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">171</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.80</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.37</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1413</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.44</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">91.2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.48</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold><italic>1PPE</italic></bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">605</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.24</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.42</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1435</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.34</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">96.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.58</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold><italic>1STF</italic></bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">35</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.12</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.31</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1303</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.80</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">92.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.70</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1TGS</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">365</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.01</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.37</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1343</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.02</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">98.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.18</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">158</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1TMQ</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">25</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.03</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.01</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1420</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.09</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">89.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.99</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">13</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold><italic>1UDI</italic></bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">217</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.37</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.73</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1401</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.79</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100.8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.42</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">18</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1UGH</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">97</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.75</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.86</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1386</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.75</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">92.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.63</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">2MTA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">161</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.30</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.47</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1478</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.25</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">86.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.54</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">428</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">2PTC</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">322</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.49</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.37</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1306</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.88</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">98.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.72</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">68</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">2SIC</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">76</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.76</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.93</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1368</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.62</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">94.5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.99</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">2SNI</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">103</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8.56</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8.71</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1386</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9.33</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">99.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.08</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold><italic>2TEC</italic></bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">235</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.89</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.56</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1375</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.69</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">91.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.85</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold><italic>4HTC</italic></bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">73</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.21</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.82</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1436</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.79</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">85.8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.41</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">7CEI</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">178</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.68</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.61</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1345</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.62</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">97.5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.15</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Average</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">224.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.96</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.11</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.76</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1403</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.99</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.12</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">93.5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.37</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">32.8</td></tr><tr><td colspan=\"12\" align=\"left\" rowspan=\"1\">\n<bold>Antigen–Antibody Complexes</bold>\n</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1AHW</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">131</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.94</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.16</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">13</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1163</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.83</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">98.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.08</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1BVK</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">41</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.66</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">28</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9.84</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1249</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.80</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">29</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">96.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.34</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">435</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold><italic>1EO8</italic></bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">25</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.15</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">29</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">{11.60}</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">30</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1194</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.50</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">97.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">N/A</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">N/A</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold><italic>1FBI</italic></bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">32</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.49</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">20</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9.47</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">14</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1212</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.43</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">17</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">92.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.97</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">266</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold><italic>1IAI</italic></bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">35</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.11</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.97</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1210</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.20</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">96.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.90</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">212</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold><italic>1MEL</italic></bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">112</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.90</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.34</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1239</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.65</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">100.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.52</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">103</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1MLC</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">16</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.86</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">26</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8.96</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">22</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1242</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.57</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">19</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">93.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9.50</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">47</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold><italic>1QFU</italic></bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">54</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.20</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.32</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">24</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1091</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.54</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">22</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">90.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.97</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">96</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1WEJ</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">26</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.65</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">19</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.45</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">19</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1257</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.89</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">90.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.41</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold><italic>2JEL</italic></bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">136</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.74</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.40</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1246</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.74</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">10</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">99.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.84</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">142</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold><italic>2VIR</italic></bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">14</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8.90</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">25</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8.50</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">28</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1223</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">{11.12}</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">26</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">95.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">N/A</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">N/A</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Average</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">56.6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.05</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">14.8</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.82</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">15.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1211</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.66</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">12.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">93.4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.61</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">145.4</td></tr><tr><td colspan=\"12\" align=\"left\" rowspan=\"1\">\n<bold>Other Complexes</bold>\n</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold><italic>1A0O</italic></bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">193</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.01</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.91</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1230</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8.45</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">89.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.27</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">50</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold><italic>1ATN</italic></bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">272</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.57</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.38</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1046</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.87</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">83.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.72</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">16</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold><italic>1GLA</italic></bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">396</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8.38</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.65</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1366</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.03</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">84.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.29</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">28</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1MDA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">60</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.59</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">15</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">{12.00}</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">13</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1179</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8.19</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">22</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">92.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">N/A</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">N/A</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold><italic>1SPB</italic></bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">252</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.41</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.80</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1162</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.40</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">90.7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.31</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1WQ1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">143</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8.44</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.99</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1164</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">10.00</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">93.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.88</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">656</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold><italic>2BTF</italic></bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">140</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.81</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.91</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1165</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.68</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">86.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.50</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">2PCC</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">59</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8.12</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">16</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.78</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">26</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1168</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9.96</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">93.6</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">N/A</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">N/A</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Average</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">189.3</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.16</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.4</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.80</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1185</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.44</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">89.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.62</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">125.3</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pcbi-1000191-t002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pcbi.1000191.t002</object-id><label>Table 2</label><caption><title>Repeatability test for SDU2.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\" width=\"759\"><colgroup span=\"1\"><col width=\"9214*\" align=\"left\" span=\"1\"/><col width=\"9214*\" align=\"center\" span=\"1\"/><col width=\"9214*\" align=\"center\" span=\"1\"/><col width=\"10823*\" align=\"center\" span=\"1\"/><col width=\"8979*\" align=\"center\" span=\"1\"/><col width=\"9427*\" align=\"center\" span=\"1\"/><col width=\"12713*\" align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Complex</td><td colspan=\"2\" align=\"left\" rowspan=\"1\">Rigid Body Prediction</td><td colspan=\"4\" align=\"left\" rowspan=\"1\">SDU2 Predictions</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">RMSD</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Energy</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>σ_RMSD</italic>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>σ_Energy</italic>\n</td></tr></thead><tbody><tr><td colspan=\"7\" align=\"left\" rowspan=\"1\">\n<bold>Enzyme–Inhibitor Complexes</bold>\n</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1ACB</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.63</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−21.09</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.17</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.28</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−51.15</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.50</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1MAH</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.80</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−50.22</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.60</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.43</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−74.57</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.74</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1UDI</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.66</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−34.37</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.60</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.34</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−71.58</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.19</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">4HTC</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.32</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−10.22</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.84</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.34</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−86.87</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.72</td></tr><tr><td colspan=\"7\" align=\"left\" rowspan=\"1\">\n<bold>Antigen–Antibody Complexes</bold>\n</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1MEL</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.04</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−9.53</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.67</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.65</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−44.67</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.88</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1QFU</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.02</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−40.35</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.18</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.28</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−43.03</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.06</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1WEJ</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.65</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−36.08</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.04</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.32</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−69.02</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.41</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">2JEL</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">7.96</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−11.61</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">6.08</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.26</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−36.09</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.42</td></tr><tr><td colspan=\"7\" align=\"left\" rowspan=\"1\">\n<bold>Other Complexes</bold>\n</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1GLA</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">8.27</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−22.08</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.10</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.23</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−58.23</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">5.33</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">1SPB</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.41</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−26.31</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">1.51</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.03</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−88.63</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.00</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Average</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.42</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">4.42</td></tr></tbody></table></alternatives></table-wrap>" ]

[ "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<disp-formula></disp-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<disp-formula></disp-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<disp-formula><label>(1)</label></disp-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<disp-formula></disp-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<disp-formula></disp-formula>" ]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><fn id=\"nt101\"><label>a</label><p>PDB codes for the bound-unbound docking problems are shown in italics. All others are unbound-unbound cases.</p></fn></table-wrap-foot>", "<fn-group><fn fn-type=\"COI-statement\"><p>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p>This research was supported by National Institutes of Health/National Institute of General Medical Sciences grants R21-GM079396 and R01-GM061867.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"pcbi.1000191.g001\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e001.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e002.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e003.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e004.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e005.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e006.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e007.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e008.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e009.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e010.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e011.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e012.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e013.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e014.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pcbi.1000191.e015.jpg\" mimetype=\"image\" position=\"float\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e016.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e017.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e018.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e019.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e020.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e021.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e022.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e023.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e024.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pcbi.1000191.e025.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic id=\"pcbi-1000191-t001-1\" xlink:href=\"pcbi.1000191.t001\"/>", "<graphic xlink:href=\"pcbi.1000191.g002\"/>", "<graphic xlink:href=\"pcbi.1000191.g003\"/>", "<graphic xlink:href=\"pcbi.1000191.g004\"/>", "<graphic xlink:href=\"pcbi.1000191.g005\"/>", "<graphic id=\"pcbi-1000191-t002-2\" xlink:href=\"pcbi.1000191.t002\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e026.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e027.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e028.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e029.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e030.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e031.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e032.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e033.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e034.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pcbi.1000191.e035.jpg\" mimetype=\"image\" position=\"float\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e049.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e036.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e037.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e038.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e039.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pcbi.1000191.e040.jpg\" mimetype=\"image\" position=\"float\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e041.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e042.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e043.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e044.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e045.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e046.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000191.e047.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pcbi.1000191.e048.jpg\" mimetype=\"image\" position=\"float\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Actin polymerization at the plasma membrane results in the formation of cellular protrusions known as lamellipodia or filopodia, which mediate cell migration ##REF##9013669##[1]##–##REF##15306794##[3]##. The distinct organization and generation of filaments in each structure uses a different mechanism to produce mechanical force. In the lamellipodia, the actin filaments organize into a flat 2D branched network ##REF##12600310##[2]##, ##REF##10352018##[4]## whereas in the filopodia they are assembled into long, parallel, closely packed bundles ##REF##12566431##[5]##–##REF##1447299##[7]##. Different proteins control the assembly of these structures; in the lamellipodia, the branched nucleation is driven by activation of the Arp2/3 complex ##REF##10801131##[8]## by Wiskott-Aldrich syndrome protein family ##REF##9889097##[9]##, ##REF##11329366##[10]## (WASP), followed by filament elongation and barbed-end capping by capping proteins (CP) ##REF##10679358##[11]##. Formin and Ena/VASP proteins are concentrated at the tips of filopodia ##REF##17475772##[12]##–##REF##15908944##[15]## promoting the continuous elongation of filaments ##REF##15306794##[3]##, ##REF##15294161##[16]## and their successive binding by fascin, which crosslinks actin filaments of the same polarity into the bundles constituting the “body” of filopodia ##REF##11948621##[17]##. In both structures, the barbed (i.e., the fast growing) ends of actin filaments point towards the plasma membrane ##REF##10477762##[18]##.</p>", "<p>Most cultured animal cells assemble both lamellipodia and filopodia. Some cells, like dendritic cells, are dominated by filopodia ##REF##15294161##[16]## while keratocytes grow exclusively lamellipodia ##REF##8968574##[19]##. It is still not fully understood what determines the <italic>in vivo</italic> preference to lamellipodia vs. filopodia, and why certain cells are dominated by only one type of structure. So far, two mechanisms were suggested to explain the formation of filopodia ##REF##17712139##[20]##. In the “<italic>de novo</italic> filopodia nucleation” model, actin filament nucleation and elongation is mediated by Dia2 proteins ##REF##15908944##[15]##, ##REF##16740473##[21]##. According to this mechanism bundles do not emanate from lamellipodia and Arp2/3 complex is not essential for filopodia formation. In the second mechanism, known as the “convergent elongation” model, filopodia emerge from the Arp2/3 lamellipodial network ##REF##12642617##[22]##–##REF##18256280##[25]##. During formation of filopodia, lamellipodial filaments associate at their barbed ends with mDia2 or VASP, elongate continuously and gradually converge into filopodia bundles, by the cross-link of fascin. In contrast to the first mechanism, here filopodia are anchored to lamellipodia. Both systems can be reconstituted <italic>in vitro</italic>. This work primarily focuses on the “convergent elongation” model.</p>", "<p>Notwithstanding recent advances in understanding the dynamic organization of lamellipodia and filopodia protrusions, it is still not fully understood how cells control the transition between these structures, what directs the localization of filopodia formation along the cell leading edge, and how their thickness is regulated. It is expected that the emergence of filopodia from lamellipodia would be strongly affected by the properties (e.g., the density and length of filaments) of the branched lamellipodial network from which they emanate. The concentration of Arp2/3, which strongly affects these structural properties, was recently observed to have a dramatic influence on the formation of filopodia <italic>in vivo</italic>\n##REF##18256280##[25]##. The abundance of fascin, which is the driving force for bundling is also expected to affect bundles formation. Yet, it is still not clear how the structure of lamellipodia and the fascin concentration control filopodia formation.</p>", "<p>To resolve these issues, it is essential to understand the mechanisms underlying the formation of lamellipodia and filopodia and evaluate the factors controlling their structure and dynamics. To this end, in the present work we study a system containing the major proteins participating in the assembly of lamellipodia and filopodia: a) the constitutively active VCA ##REF##12015607##[26]## (or WA ##REF##10801131##[8]##) domain of WASp; b) Arp2/3 complex; c) fascin, and d) actin monomers. This model system enables us to control the concentrations of the various proteins, and to observe actin organization in bulk. Recently, we have demonstrated the feasibility of using such a simple reconstituted system for studying the formation of lamellipodia and filopodia-like structures <italic>in vitro</italic>\n##REF##16549794##[27]##. In this early work, the roles of Arp2/3 complex and fascin in the transition from the Arp2/3 branched network to filopodia-like bundles was addressed only briefly. Also, no attempt was made to model this transition nor to resolve the primary factors controlling the bundling process.</p>", "<p>Our goal in the present work is to study in detail the self-assembly characteristics of actin in the presence of variable amounts of Arp2/3 complex and fascin. We will show that in the absence of fascin, actin organizes into dense 3D, aster-like, structures composed of branched networks of actin filaments ##REF##16549794##[27]##. The growth of the network advances with the barbed ends of the peripheral actin filaments pointing on average outward, as also demonstrated in the simulations ##REF##16549794##[27]##. Addition of fascin leads to the formation of star-like structures, where actin/fascin bundles are nucleated and emanate radially from the branched network core. The polarity of the actin filaments during the transition is preserved, as in the transition from lamellipodia to filopodia ##REF##16549794##[27]##. Our experiments show that in the early stages of actin polymerization, Arp2/3 mediates the formation of the dense and highly branched aster-like network, whereas fascin is rather passive. Fascin becomes dominant at a later stage of actin polymerization, when the (sparsely branched) filaments in the periphery of the aster become long enough, so that the energy gained due to actin-fascin-actin links between neighboring filaments overcomes the unfavorable bending energy required to bring them into contact. As time evolves the initially thin bundles elongate, thus lowering their bending energy thereby enabling their association into thicker bundles, and so on until no more actin monomers are left.</p>", "<p>This picture of stepwise bundle elongation and thickening underlies our structural-energetic model of bundle formation and growth. Using reasonable approximations for the bending energies of actin filaments and bundles, and the cohesive energy due to fasin-actin bonds we can explain the dependence of the thickness and length of the mature bundles upon [Arp2/3]. More explicitly, using Kinetic Monte Carlo (KMC) simulations we can model the nucleation and growth of asters and derive their structure as a function of time thus obtaining two central structural parameters: the length, <italic>L</italic>, of filament tips at the aster periphery, and the spacing, <italic>b</italic>, between their origins (i.e., their pointed ends, anchored at Arp2/3 branch points). These parameters, which determine the initial conditions at the onset of bundling, depend sensitively on [Arp2/3]. As we shall see, our theoretical scheme, which combines the KMC and the stepwise bundle assembly model, accounts adequately for the experimentally observed dependence of bundle structure on the initial concentration of Arp2/3 in the system.</p>" ]

[ "<title>Materials and Methods</title>", "<title>Protein purification</title>", "<p>Actin was purified from rabbit skeletal muscle acetone powder ##REF##4254541##[33]##. Recombinant WASp-VCA ##REF##12015607##[26]## and fascin (recombinant fascin was prepared by a modification of the method of ##REF##8999969##[34]##) were expressed in <italic>E. coli</italic> as GST fusion proteins. Arp2/3 complex was purchased from Cytoskeleton Co. Actin was labeled on Cys374 with Alexa fluor 568 or 488 (Invitrogen, Co.).</p>", "<title>Motility assay</title>", "<p>The motility medium contained 10 mM HEPES, pH = 7.7, 1.7 mM Mg-ATP, 5.5 mM DTT, 0.12 mM Dabco [an anti-bleaching reagent], 0.1 M KCl, 1 mM MgCl₂, 1% BSA and various concentrations of G-actin, Arp2/3 complex, VCA, and fascin.</p>", "<title>Microscopy</title>", "<p>Actin assembly was monitored for about an hour by fluorescence with an Olympus IX-71 microscope. The labeled actin fraction was 1/40 and the temperature was 22°C. Time-lapse images were acquired using an Andor DV887 EMCCD camera (Andor Co., England). Data acquisition and analysis was performed using METAMORPH (Universal Imaging Co.).</p>" ]

[ "<title>Results</title>", "<p>The first part of this section describes the three types of structures, i.e., asters, stars and network of bundles (a phase of actin-fascin bundles) that were experimentally observed and the transition between them. We also present a phase diagram of the system to illustrate the regions of existence of the three types of structures as a function of the initial [Arp2/3] and [fascin]. In the second part, we focus specifically on the transition from asters to stars, as it mimics the transition from lamellipodia to filopodia in cells. Special attention is given to evaluating the roles of [Arp2/3] and [fascin] on filopodia formation and dimensions. In the last part of this section we describe our theoretical model for bundle assembly and apply it to explain our experimental results.</p>", "<title>Fascin mediates structural transition of actin self-assembly</title>", "<p>First, we aimed at studying the self–assembly of actin in the presence of increasing amounts of fascin, at a given [Arp2/3]. At low Arp2/3 complex concentration of 6.25 nM (the [VCA]/[Arp2/3] ratio was kept equal to two in all experiments) two structural transitions are observed, the first, from ‘asters’ to ‘stars’, and the second, from ‘stars’ to ‘network of bundles’ (##FIG##0##Fig. 1##). At a very low fascin concentration (3 nM) the structure of actin network is dominated by Arp2/3 nucleation and branching activity–the system organizes into diffuse highly branched 3D aster-like structures (##FIG##0##Fig. 1a##), similar to those formed in the absence of fascin ##REF##16549794##[27]##. Surprisingly, a minute concentration of fascin ([fascin] = 4 nM) was sufficient to mediate the nucleation and growth of bundles emanating from branched asters' cores. ##FIG##0##Fig. 1b## shows such a star-like structure after reaching final dimensions. The bundles grow with their barbed ends pointing outward, as previously demonstrated ##REF##16549794##[27]##. The time span of this process ranges between few to ten minutes, depending on [Arp2/3] and [fascin] (data not shown). In this work all the experimental data analyses were performed for systems that have reached their final state.</p>", "<p>Gradual addition of fascin, from 0.005 to 0.2 µM, increases the number of bundles emanating from the asters core as well as their length (##FIG##0##Figs. 1c–1f##). Above 0.2 µM of fascin, the system is dominated by the bundling activity of fascin. Most of the actin appears to be located in actin/fascin bundles residing in the bulk solution (see image background, ##FIG##0##Fig. 1g##; [fascin = 0.5 µM]). The stars that still exist are composed of fewer and shorter bundles. Eventually, at high enough [fascin] (3 µM) stars no longer form; instead, a network of actin/fascin bundles is generated (##FIG##0##Fig. 1h##).</p>", "<title>[Arp2/3] controls fascin ability to induce structural changes</title>", "<p>Structural transformations mediated by fascin seem to be facilitated when the initial [Arp2/3] was reduced. The top and bottom panels in ##FIG##1##Fig. 2## distinguish between fluorescence microscopy images corresponding to mixtures containing low ([Arp2/3] = 12.5 nM) and high ([Arp2/3] = 100 nM) initial Arp2/3 contents, respectively. In both cases [fascin] varies in the same manner. Initially, at a very low [fascin] of 3 nM (##FIG##1##Figs. 2a and 2e##), actin organization is fully dominated by Arp2/3 complex activity; diffuse asters are formed, whose number increases with the initial [Arp2/3]. These asters are smaller and their branched actin cores are denser (##FIG##1##Fig. 2e##) compared to those formed at the lower [Arp2/3] (##FIG##1##Fig. 2a##). Increasing the concentration of fascin to 6 nM (blue arrow) induced a transition from asters to stars when [Arp2/3] = 12.5 nM (##FIG##1##Fig. 2b##; ##SUPPL##1##Figs. S1a and S1b## show that the transition occurs between 5 nM and 6 nM of fascin). On the other hand, 6 nM fascin is not sufficient to induce bundle formation in the [Arp2/3] = 100 nM system (##FIG##1##Fig. 2f##). In this latter case, more fascin (7 nM) is required to nucleate bundles from the branched network core (see also ##SUPPL##1##Figs. S1c and S1d##).</p>", "<p>Further addition of fascin induces the formation of additional and longer bundles, resulting in the appearance of fully developed stars, as shown in ##FIG##1##Figs. 2c and 2g##. The total number of bundles nucleated per star in the low [Arp2/3] (12.5 nM) case appears to be larger than in the high [Arp2/3] (100 nM) case. Moreover, the bundles in the low [Arp2/3] system look both thicker (i.e., they comprise a larger number of actin filaments) and longer. Clearly then, the initial concentration of Arp2/3 in solution plays a major role in determining the conditions for the fascin-mediated assembly of actin bundles, thus (albeit indirectly) controlling bundles thickness and length.</p>", "<p>Finally, when the amount of fascin was further elevated to 3 µM (red arrow), a transition from ‘stars’ to ‘a network of bundles’ was observed in the 12.5 nM Arp2/3 system (##FIG##1##Fig. 2d##). In contrast, the system with 100 nM Arp2/3 remained unchanged (##FIG##1##Fig. 2h##) and did not show any structural transition up to 6 µM fascin (data not shown). Actually ‘networks of bundles’ did not appear in any of the systems containing more than 25 nM of Arp2/3 complex, regardless of [fascin].</p>", "<p>In ##FIG##2##Fig. 3## we present a phase diagram of the system, depicting the regions of existence of the three types of experimentally observed structures, as a function of the initial [Arp2/3] and [fascin]. The data shown here corresponds to experiments conducted at [G-actin] = 7 µM; a similar qualitative behavior was observed for the other concentrations of actin we have tested, ranging between 1 µM and 5 µM (data not shown). Careful inspection shows that below 3 nM of fascin only aster-like structures form (circles), regardless of [Arp2/3]. Increasing [fascin], for a given amount of Arp2/3 complex, induces structural transitions into stars (inverted triangles). The amount of fascin required to mediate this transition increases nonlinearly with [Arp2/3], saturating above 40 nM Arp2/3, (the blue curve marks the approximate boundary between the two regions). While only few nano-molars of fascin are sufficient to induce the transition from asters to stars, the subsequent transition to ‘networks’ (squares) requires much larger amounts of fascin (several µM; the red curve marks the approximate boundary between the two regions). We note that a transition to ‘networks’ is observed only for sufficiently low [Arp2/3] (&lt;25 nM), within this region, the amount of fascin required for the nucleation of bundles increases monotonically with Arp2/3 content.</p>", "<title>Aster to star transition–Experimental results</title>", "<title>The densities of asters and stars depend on [Arp2/3] but not on [fascin]</title>", "<p>To test our assumption that each star originates from a preformed aster, we have measured the number of “aggregates” (i.e., asters or stars) per unit area (surface density), within the 1 µm section of the bulk sample observed. The surface density of aggregates <italic>σ</italic> = <italic>d</italic>\n⁻² was evaluated by measuring the average distance, <italic>d</italic>, between adjacent aggregates. Keeping a constant ratio [VCA]/[Arp2/3] = 2, we have measured <italic>d</italic>\n⁻² for several values of the initial [Arp2/3] and [fascin] (##FIG##3##Fig. 4##). For clarity we present only two concentrations of fascin: 5 nM (triangles) and 0.2 µM (circles). Data points taken at 5 nM fascin correspond to density of asters, while those taken at 0.2 µM fascin are associated with density of stars. In both cases, <italic>d</italic>\n⁻² increases monotonically with [Arp2/3]. In contrast, [fascin] does not appear to affect <italic>d</italic>\n⁻² (see inset; conditions: [Arp2/3] = 12.5 nM), indicating that the density of aggregates is dictated primarily by the amount of Arp2/3 complex in the system. These results support our previous findings that Arp2/3 complexes are aster nucleators and that stars develop from preformed asters ##REF##16549794##[27]##. This is consistent with our present conclusion that, at early times, actin polymerization is governed by Arp2/3 nucleation and branching. Fascin comes into play and becomes dominant in star formation only later, when the structural properties of the aster enable the onset of filament bundling.</p>", "<title>The length of bundles in stars depends on Arp2/3 and fascin concentrations</title>", "<p>For all the concentrations of Arp2/3 that we have tested, the mean final length of bundles, <italic>L_Fin</italic>, was found to increase with [fascin], reaching asymptotically a maximal ([Arp2/3] dependent) value, (##FIG##4##Fig. 5a##). The system appears more sensitive to fascin addition when [Arp2/3] is lower (as reflected by the steeper slopes in ##FIG##4##Fig. 5a##). Consequently, the amount of fascin required to reach decreases with the decrease in [Arp2/3]. In contrast, for a fixed concentration of fascin, we found that <italic>L_Fin</italic> decreases monotonously with [Arp2/3], (as seen for [fascin] = 100 nM in ##FIG##4##Fig. 5b##). The inverse correlation between <italic>L_Fin</italic> and [Arp2/3] results from the reduced amount of actin monomers available for bundle growth when [Arp2/3] is increased, as discussed in more detail in the theory section below.</p>", "<title>Theoretical model and analysis</title>", "<title>Bundle formation: A simple model and Kinetic Monte Carlo Simulations</title>", "<p>Our experiments show that upon mixing actin monomers with Arp2/3 complex, VCA and fascin molecules, the first structures to appear in solution are dense and highly branched aster-like networks of actin filaments ##REF##16549794##[27]##. The asters grow radially outwards, and since Arp2/3 joins both old and newly formed filaments, the density of branching points and hence also of polymerized actin is highest in the center of the aster, decreasing gradually towards its periphery, where the average spacing between adjacent branches is relatively large ##REF##16549794##[27]##. To associate into bundles, nearby filaments are generally required to bend towards each other. Inside the aster core, owing primarily to excluded volume interactions between the rigid, dense and highly branched network, filament bending involves a prohibitively large energetic penalty and bundle nucleation is thus highly unlikely.</p>", "<p>Bending the filaments emanating from the surface of the aster is easier, yet two conditions must be fulfilled to enable the onset of bundle formation. First, <italic>L</italic>, the average length of the filament tips in the aster periphery, should be larger than the distance, <italic>b</italic>, between their pointed ends (which mark their origins). The second requirement is that the energy gained due to fascin links between neighboring filaments should exceed the bending energy associated with bringing them into contact. Both conditions are met at some time <italic>t_i</italic>, in the late stages of aster growth, owing mainly to the decrease in [Arp2/3] and hence the increase in <italic>L</italic>. Qualitatively, this explains why in the early stages of actin polymerization the (dense aster-like) structure of the actin network is dominated by Arp2/3, with fascin playing a rather passive role. However, once the peripheral filaments are long enough, fascin comes into play–linking the filaments into bundles, whereas branching essentially ceases; especially along and within the bundles, because the (already largely depleted) Arp2/3 complexes cannot penetrate the dense bundle.</p>", "<p>Our goal in this section is to explain the experimental observation that bundles are thicker and longer when the initial [Arp2/3] is low (see ##FIG##1##Figs. 2## and ##FIG##4##5##). Qualitatively, this behavior is understood based on the fact that since Arp2/3 is an aster nucleator, so that for a given concentration of G-actin, higher [Arp2/3] results in the appearance of more asters in solution. Thus, on average, fewer actin monomers participate in the growth of each aster. Also, since the asters formed are denser, fewer monomers are later available for bundle growth, resulting in relatively short bundles, as observed in our experiments (##FIG##1##Figs. 2## and ##FIG##4##5##). It should be noted that in living cells, the plasma membrane also affects the length of filopodia, as was previously analyzed theoretically ##REF##15879474##[23]##, ##REF##16214866##[28]##. Furthermore, in our system, owing to the large energy penalty associated with the bending of short bundles towards each other, and because the bending energy increases rapidly with bundle thickness (see below), when [Arp2/3] is high, association of bundles is energetically costly, and the mature bundles are both shorter and thinner than in the case of low [Arp2/3].</p>", "<p>The structural properties of the asters at the onset of bundle formation, primarily the distances <italic>b</italic> and <italic>L</italic> defined above, enter our model as input parameters. Numerical estimates of these quantities were derived using our 3D KMC simulations of aster growth. To derive the aster structural characteristics relevant to the experiments reported here, we have used a limited version of the simulation, which includes only actin polymerization and Arp2/3 mediated branching ##REF##16549794##[27]##, (see also ##SUPPL##0##Appendix S1##, in “Supporting information”). A simulation snapshot of an aster is shown in ##FIG##5##Fig. 6a##. It should be noted that (at least presently) these simulations do not account for filament bending, and thus cannot describe bundle formation; they are used here only to provide the initial <italic>L</italic> and <italic>b</italic>.</p>", "<p>At the onset of the aster-to-star transition the radius of the aster is generally much larger than both <italic>b</italic> and <italic>L</italic>. Since bundles are formed by the association of neighboring filaments whose origins are just a few <italic>b</italic>'s apart, we ignore the curvature of the aster's surface and treat it as being planar. It may be noted, however, that once bundles become much larger than the aster core, surface curvature correction should (and can) be taken into account.</p>", "<p>Assuming that all the <italic>M</italic> filaments participating initially in bundle formation are of the same length <italic>L</italic>, and organize into <italic>M/N</italic> identical bundles, each of which consists of <italic>N</italic> filaments, the total energy of the system is given by <italic>E_tot</italic> = (<italic>M</italic>/<italic>N</italic>)<italic>E_bundle</italic>(<italic>N</italic>;<italic>L</italic>,<italic>b</italic>). The bundle energy <italic>E_bundle</italic>(<italic>N</italic>;<italic>L</italic>,<italic>b</italic>) includes the cohesive energy due to the fascin-mediated linking of filaments, the bending energy cost of bringing the filaments into contact, and the (unfavorable) surface energy of a finite size bundle. All these contributions depend on <italic>N</italic>, and parametrically on <italic>L</italic> and <italic>b</italic> (Additional contributions, such as the entropy loss experienced by the filaments upon bundling can be neglected). If bundle elongation were slow compared to filament (and bundle) binding-unbinding events, then the optimal bundle size <italic>N</italic>\n^* could be derived by minimizing <italic>E_tot</italic> with respect to <italic>N</italic>. Using reasonable models for the bending, cohesion and surface energies one can show that the resulting <italic>N</italic>\n^* is indeed larger when [Arp2/3] is lower. It must be noted, however, that in this scheme the optimal size and length of bundles is determined by the principles of equilibrium thermodynamics. However, kinetic estimates of the rates of bundle dissociation (based on calculations of filaments dimerization time, ##REF##16565053##[29]##, ##REF##17930069##[30]## (data not shown) indicate that once bundles are formed, their life time is significantly longer than the time scale of actin polymerization, so that filament and bundle associations are practically irreversible on the time scale of our present experiments.</p>", "<p>Based on this notion we propose here an alternative, stepwise, mechanism whereby single filaments first associate into thin bundles (##FIG##5##Fig. 6b##). The nascent thin bundles are hard to bend, but after elongating their bending and hence fascin-mediated association into thicker bundles becomes easier. This scenario may continue, leading to stepwise bundle thickening. It should be noted that <italic>in vivo</italic> observations of bundles roots also indicate that bundles initiate in a gradual process of elongation and thickening ##REF##12566431##[5]##. In our <italic>in vitro</italic> experiments, this process terminates when there are essentially no more actin monomers in the system.</p>", "<p>Suppose for simplicity that all possible bundles have a circular cross section consisting of a central straight filament surrounded by <italic>s</italic> concentric shells of filaments; <italic>s</italic> = <italic>R</italic>/<italic>d</italic> where <italic>R</italic> is the radius of the bundle, and <italic>d</italic> is the average distance between adjacent filaments. ##FIG##5##Fig. 6c## shows Δ<italic>E</italic>\n_dim(<italic>s</italic>), the energy change corresponding to the dimerization of two bundles of the same size <italic>s</italic> containing <italic>N</italic>≈<italic>πs</italic>\n² filaments, into a thicker bundle comprising 2<italic>N</italic> filaments. This simplified calculation ignores many association processes other than bundle doubling (e.g., monomer-dimer, monomer-tetramer binding, etc.) Yet, it captures, at least qualitatively, the kinetic-energetic picture of bundle growth.</p>", "<p>Results are shown for two of the initial [Arp2/3] studied experimentally (12.5 and 100 nM), and several representative values of <italic>L</italic> which, according to our aster growth simulations, represent relevant filament lengths for the time and concentration scales of the experiments. More specifically, the range of <italic>L</italic> considered is <italic>L_min</italic>([<italic>Arp2</italic>/<italic>3</italic>])&lt;<italic>L</italic>&lt;<italic>L_max</italic>([<italic>Arp2</italic>/<italic>3</italic>]), where <italic>L_min</italic>([<italic>Arp2</italic>/<italic>3</italic>]) is the length of filaments at the time <italic>t_i</italic>, marking the onset of bundling, and <italic>L_max</italic>([<italic>Arp2</italic>/<italic>3</italic>]) is the length of bundles when all G-actin was consumed. Our simulations show that there are about 10⁶ free monomers per aster at <italic>t_i</italic> (see <xref ref-type=\"sec\" rid=\"s4\">methods</xref>) when [Arp2/3] = 100 nM, compared to ∼10⁸ monomers for 12.5 nM. In both cases, the number of filaments in the aster periphery is about 5000 at <italic>t_i</italic>. Thus <italic>L</italic> ranges between 0.8–1.2 µm for 100 nM Arp2/3. In the 12.5 nM Arp2/3 <italic>L</italic> is much larger: 1.5–30 µm.</p>", "<p>Association of two adjacent bundles of thickness <italic>s</italic> to form a thicker one is energetically favorable when Δ<italic>E</italic>\n_dim(<italic>s</italic>) is negative. ##FIG##5##Fig. 6c## shows, for instance, that for [Arp2/3] = 12.5 nM, bundles of length 9 µm that are composed of 5 shells of filaments will not associate with each other. However, once their length increases to, say, <italic>L</italic> = 12 µm, they tend to stick to each other since Δ<italic>E</italic>\n_dim becomes negative. From ##FIG##5##Fig. 6c## we note that for both values of [Arp2/3] considered, as <italic>L</italic> increases thicker bundles become energetically favorable; the optimal bundle thickness (corresponding to the minimum of Δ<italic>E</italic>\n_dim(<italic>s</italic>)) increases as well. For example, in the case [Arp2/3] = 12.5 nM, when <italic>L</italic> = 9 µm dimerization is favorable for bundles comprising no more than <italic>s</italic> = 4 shells, and the highest energetic gain is obtained upon dimerization of bundles comprising of <italic>s</italic> = 2 shells. On the other hand, when <italic>L</italic> = 12 µm, the maximum number of shells in bundles capable of dimerization to is <italic>s</italic> = 5, and the optimal thickness is obtained for doubling <italic>s</italic> = 3 bundles. ##FIG##5##Fig. 6c## also reveals the important role of [Arp2/3] on the length and thickness of the bundles. When [Arp2/3] = 100 nM a larger fraction of actin monomers (as compared to the case [Arp2/3] = 12.5 nM) are consumed during aster's assembly, leaving less monomers for bundle growth. This fact, as well as the initial <italic>b</italic> and <italic>L</italic>, are responsible for the shorter and thinner bundles corresponding to the high [Arp2/3] system, (see magnified box). In this case, the typical bundle consists of a small (s∼1) number of shells.</p>", "<p>To derive the results shown in ##FIG##5##Fig. 6c## we expressed the dimerization energy as a sum of bending and fascin linking energies: Δ<italic>E_dim</italic> = <italic>E_bend</italic>+Δ<italic>E_fascin</italic>, both depending on <italic>s</italic>. To calculate the bending energy of a bundle upon joining another one we have used a simple geometrical model, whereby its bent portion, of length <italic>l</italic>, is represented as composed of two oppositely (and moderately) bent, but otherwise identical arcs, joining smoothly each other to an s-like shape (as may be seen in ##FIG##5##Fig. 6b##). It can be shown that this implies <italic>E_bend</italic> = <italic>γ a</italic>\n²/<italic>l</italic>\n³ (details will be published elsewhere). Here <italic>a</italic> is the distance between the origins of the bundles (<italic>a</italic> = <italic>bs</italic> for two neighboring bundles of size <italic>s</italic>), <italic>γ</italic> = 32<italic>ξkT</italic> where <italic>ξ</italic> is the persistence length of the bundle, <italic>k</italic> is Boltzmann's constant and <italic>T</italic> is the absolute temperature. Since fascin is abundant in the system, the actin filaments within bundles are tightly bound, and the bundles thus behave as semiflexible rods of thickness <italic>s</italic>, and persistence length <italic>ξ_s</italic> = <italic>π</italic>\n²\n<italic>s</italic>\n⁴\n<italic>ξ</italic>\n₁, where <italic>ξ</italic>\n₁≈10 <italic>µm</italic> is the persistence length of a single actin filament.</p>", "<p>The energy change upon linking two bundles is proportional to the length, <italic>L-l</italic>, of their straight and parallel portions (##FIG##5##Fig. 6b##), i.e., Δ<italic>E_fascin</italic> = −(<italic>L</italic>−<italic>l</italic>)<italic>ε</italic>, where <italic>ε</italic> is the gain, per unit length, in the cohesive energy due to the fascin-F-actin bonds. Assuming that all possible actin-fascin contacts are saturated, and that actin filaments are hexagonally packed within a bundle, we find that for two cylindrical bundles which associate into a thicker cylindrical bundle <italic>ε</italic> = 0.2<italic>πsε</italic>\n₁, where <italic>ε</italic>\n₁ = 833<italic>kT</italic>/<italic>µm</italic> is the binding energy of fascin to a unit length of a single F-actin. Minimizing Δ<italic>E_dim</italic> = <italic>γb</italic>\n²\n<italic>s</italic>\n²/<italic>l</italic>\n³−(<italic>L</italic>−<italic>l</italic>)<italic>ε</italic> with respect to <italic>l</italic>, we find the optimal length of the bent (‘root-like”, ##FIG##5##Fig. 6b##) portion of the bundle: <italic>l_s</italic> = <italic>αb</italic>\n^1/2\n<italic>s</italic>\n^1/2, where <italic>α</italic> = (3<italic>γ</italic>/<italic>ε</italic>)^1/4. Substituting <italic>l_s</italic> into Δ<italic>E</italic>\n_dim(<italic>s</italic>) we get Δ<italic>E_dim</italic>(<italic>s</italic>) = <italic>C_As</italic>\n^9/4−<italic>C_BLs</italic> where and <italic>C_B</italic> = 0.2<italic>πε</italic>\n₁.</p>", "<p>We conclude this analysis by noting that from some point in time onwards, the bundles observed experimentally appear to continue elongating without changing their thickness. Based on our model we may conclude that this happens when the time scale of bundle bending fluctuations which lead to bundle-bundle association becomes long compared to the experimental time scale, or simply long relative to the time it takes to all actin monomers to join the growing bundles.</p>" ]

[ "<title>Discussion</title>", "<p>In this work we studied in detail the steady state structures formed in a simple <italic>in vitro</italic> system, in which Arp2/3 complex and fascin compete on their binding to actin filaments. Varying mixtures of actin, Arp2/3, and fascin were analyzed in order to elucidate their delicate interplay in determining which of the structures–asters, stars, or networks of bundles–appear in a given concentration regime.</p>", "<p>We found that in the absence or at very low concentration of fascin the system is dominated by Arp2/3 complex nucleation and branching activities, resulting in the appearance of dense 3D aster-like networks of actin. Increasing fascin concentration induces phase changes, first to stars and then to network of bundles. A star is composed of a dense aster core with actin/fascin bundles emanating radially from its surface. Our experiments show that in the early stages of actin polymerization fascin is passive, while Arp2/3 mediates dense aster-like structures of actin, whose structure is very similar to the one observed in the absence of fascin. Fascin comes into play when actin filaments in the periphery of the aster get long enough, and can thus bend and associate with each other into bundles of parallel filaments, held together by fascin linkers. This is in accordance with <italic>in vivo</italic> experiments that emphasize the importance of filaments length in the creation of filopodia ##REF##15294161##[16]##. The aster-star transition appeared for all Arp2/3 concentrations that we have tested. The second transition, from stars to network of bundles, has only been observed at sufficiently low [Arp2/3]. In conclusion, we find that the system is more sensitive to phase changes when [Arp2/3] is low, and the concentration of fascin necessary to induce structural transformations is lower.</p>", "<p>The competition between Arp2/3 and fascin is critical in determining the actin structures formed. This is because these two proteins are nucleators of different actin geometries; Arp2/3 complex promotes the formation of branched actin seeds ##REF##10801131##[8]## while fascin initiates unbranched bundle seeds ##UREF##0##[31]##. The ability of fascin to nucleate actin bundles <italic>in vitro</italic>, was recently demonstrated in experiments showing that fascin enhances actin polymerization by promoting the formation of stable disc-like bundle nuclei. These nuclei are composed of short filaments (a few monomers long), which serve as seeds for subsequent actin polymerization ##UREF##0##[31]##. In the presence of Arp2/3, this process competes with branch nucleation of actin. The rates of these two processes are expected to be regulated by the abundance of Arp2/3 complex and fascin.</p>", "<p>In formulating our model for bundle formation and growth we assumed that only filaments originating at the surface of the aster take part in this process (similar to <italic>in vivo</italic> systems where filopodia originate in the vicinity of the plasma membrane). Filaments from the inner core of the aster cannot significantly contribute to this process, owing to severe excluded volume interactions on the bending of these highly branched, dense, and entangled filaments. We assumed that bundles begin to form when filaments in the aster surface are long enough so that their bending energy penalty is compensated by the gain of actin-fascin linking energy. Since the unbinding time of bundles is significantly longer than the polymerization time of filaments, their association is essentially irreversible, and thus cannot be adequately described based on thermodynamic equilibrium consideration. We have thus proposed a step-wise mechanism of irreversible assembly of single filaments/bundles into thicker bundles, followed by their elongation. The newly formed bundles elongate until their bending energy decreases, and their net association energy into a thicker bundle becomes energetically favorable, etc. This model can also explain the gradual thickening of newly formed bundles <italic>in vivo</italic>, forming Λ-precursors shaped roots ##REF##12566431##[5]##. In our experimental system, the elongation-thickening sequence terminates when all actin monomers are consumed. Based on this picture we could explain why at low [Arp2/3] the bundles emerging from the aster's core are both longer and thicker as compared to those observed in systems with a higher initial [Arp2/3].</p>", "<p>The evolution of asters to stars is of direct biological relevance because of its similarity to lamellipodia to filopodia transition in cells. Recall that two structural parameters, <italic>L</italic> and <italic>b</italic>, characterize the periphery of aster's network. In our <italic>in vitro</italic> system, the length, <italic>L</italic>, of the (unbranched sections of the) surface filaments increases in time, while the spacing between their origins decreases. In living cells, analogous structural parameters may determine the conditions favoring the formation of filopodia. In the cellular system, filaments length (<italic>L</italic>) and the density of the lamellipodial network (characterized by <italic>b</italic>) is controlled by specific regulatory proteins, such as Arp2/3 complex, VASP and Dia2. In particular, Dia2 was recently shown to be recruited to lamellipodia, where it induces the formation of many long, unbranched filaments (thus locally increasing <italic>L</italic> and reducing <italic>b</italic>), which then gradually converged into filopodial bundles ##REF##18044991##[24]##, ##REF##12086607##[32]##. This behavior is qualitatively consistent with our model, according to which an increasing <italic>L</italic> or decreasing <italic>b</italic>, favors the onset of bundle formation. By recruiting proteins like Dia2 to specific sites along lamellipodia, such a mechanism may enable the cell to control the localization of filopodia along its leading edge.</p>", "<p>To conclude, using a minimal model system containing only Arp2/3 complex, actin and fascin we were able to mimic complex events which take place at the leading edge of cells. Compared to a cell this system is simple and easily controlled, making it possible to characterize the roles of a limited set of proteins in the higher order assembly of actin filaments. In the future, it will be interesting to measure the size of the actin/fascin bundle using high resolution electron microscopy, and correlate it with the properties of the aster network structure (e.g., density of the actin network, filament length and orientation at the aster periphery). It is worthwhile to test the effects of mDia2 and capping proteins concentrations on bundle formation and dimensions. Finally, quantitative information from <italic>in vivo</italic> studies will be of great value, especially regarding the correlation between filopodia thicknesses and structural properties of the lamellipodial network.</p>" ]

[]

[ "<p>Conceived and designed the experiments: LH ABS AB. Performed the experiments: YI YBK. Analyzed the data: YI YBK. Wrote the paper: YBK ABS AB.</p>", "<p>During cellular migration, regulated actin assembly takes place at the cell leading edge, with continuous disassembly deeper in the cell interior. Actin polymerization at the plasma membrane results in the extension of cellular protrusions in the form of lamellipodia and filopodia. To understand how cells regulate the transformation of lamellipodia into filopodia, and to determine the major factors that control their transition, we studied actin self-assembly in the presence of Arp2/3 complex, WASp-VCA and fascin, the major proteins participating in the assembly of lamellipodia and filopodia. We show that in the early stages of actin polymerization fascin is passive while Arp2/3 mediates the formation of dense and highly branched aster-like networks of actin. Once filaments in the periphery of an aster get long enough, fascin becomes active, linking the filaments into bundles which emanate radially from the aster's surface, resulting in the formation of star-like structures. We show that the number of bundles nucleated per star, as well as their thickness and length, is controlled by the initial concentration of Arp2/3 complex ([Arp2/3]). Specifically, we tested several values of [Arp2/3] and found that for given initial concentrations of actin and fascin, the number of bundles per star, as well as their length and thickness are larger when [Arp2/3] is lower. Our experimental findings can be interpreted and explained using a theoretical scheme which combines Kinetic Monte Carlo simulations for aster growth, with a simple mechanistic model for bundles' formation and growth. According to this model, bundles emerge from the aster's (sparsely branched) surface layer. Bundles begin to form when the bending energy associated with bringing two filaments into contact is compensated by the energetic gain resulting from their fascin linking energy. As time evolves the initially thin and short bundles elongate, thus reducing their bending energy and allowing them to further associate and create thicker bundles, until all actin monomers are consumed. This process is essentially irreversible on the time scale of actin polymerization. Two structural parameters, <italic>L</italic>, which is proportional to the length of filament tips at the aster periphery and <italic>b</italic>, the spacing between their origins, dictate the onset of bundling; both depending on [Arp2/3]. Cells may use a similar mechanism to regulate filopodia formation along the cell leading edge. Such a mechanism may allow cells to have control over the localization of filopodia by recruiting specific proteins that regulate filaments length (e.g., Dia2) to specific sites along lamellipodia.</p>" ]

[ "<title>Supporting Information</title>" ]

[ "<p>We would like to thank Nir Gov for useful discussions.</p>" ]

[ "<fig id=\"pone-0003297-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003297.g001</object-id><label>Figure 1</label><caption><title>Phase transition as a function of fascin concentration.</title><p>Conditions are 7 µM G-actin, 6.25 nM Arp2/3, and 12.5 nM GST-VCA. Fascin concentration is: (a) 3 nM, (b) 4 nM, (c) 5 nM, (d) 7 nM, (e) 50 nM, (f) 200 nM, (g) 500 nM, and (h) 3 µM. At very low concentrations of fascin asters are formed. Transition to star-like structures occurs above a fascin concentration of 4 nM (b). The density and the length of the bundles emanating from the star core increase with fascin concentration (b–f). Above a certain fascin concentration of 0.5 µM (g) the size and the number of stars decreases; the stars coexist with an entangled network of actin/fascin bundles, seen in the image background. At 3 µM stars do not form anymore; the system is composed solely of entangled network of actin bundles (h). Bar is 10 µm.</p></caption></fig>", "<fig id=\"pone-0003297-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003297.g002</object-id><label>Figure 2</label><caption><title>Transition from aster to star-like to network structures induced by fascin depends on Arp2/3 concentration.</title><p>Conditions are: 7 µM G-actin, upper line (a–d) 12.5 nM Arp2/3 complex, and 25 nM GST-VCA and bottom line (e–h) 100 nM Arp2/3, and 200 nM GST-VCA. Transition from aster to stars and from stars to network: occurs between 3 to 6 nM and between 0.2 to 3 µM at 12.5 nM [Arp2/3] (blue and red arrows, respectively). At 100 nM [Arp2/3] only the transition from aster to star is visible; transition to a network structure is not visible in these fascin concentration ranges. Bars are 10 µm (a–h).</p></caption></fig>", "<fig id=\"pone-0003297-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003297.g003</object-id><label>Figure 3</label><caption><title>Phase diagram for actin-fascin-arp2/3 complex.</title><p>Condition is: 7 µM G-actin, Arp2/3 complex and fascin concentration was changed according to the graph. Condition for images is: 3 nM, 200 nM and 3 µM fascin; 6.25 nM, 12.5 nM and 6.25 nM Arp2/3 complex (a, b and c, respectively). Experimental points demonstrate different type of structures formed; these are represented by squares - entangled network, triangles - stars and circles - asters. Phase changes from asters to stars (blue line) and from stars to entangled bundle network (red line). Lines of phase separation are plotted as averages of two points of different structures. For low concentration of Arp2/3 complex phase transition is more abrupt for the same change in fascin concentration. No transition to entangled network was visible for high concentrations of Arp2/3 complex.</p></caption></fig>", "<fig id=\"pone-0003297-g004\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003297.g004</object-id><label>Figure 4</label><caption><title>The surface density of objects (asters or stars) is controlled solely by Arp2/3 and not by fascin.</title><p>Conditions are: 7 µM G-actin; fascin concentration of 5 nM (triangles) and 200 nM (circles); and variable amount of Arp2/3 complex: 6.25, 12.5, 25, 40, and 100 nM. The [VCA]/[Arp2/3 complex] = 2 in all experiments. We observe a monotonic growth in density of objects with Arp2/3 complex concentration. The density of objects doesn't show a dependence on fascin concentration (inset, a half log plot is given in order to see clearly all experimental data points). Error bars represent standard deviation from average values.</p></caption></fig>", "<fig id=\"pone-0003297-g005\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003297.g005</object-id><label>Figure 5</label><caption><title>Mean star-like bundles' length, <italic>L_Fin</italic>, dependence on the [Arp2/3] and [fascin].</title><p>(a) Conditions are: [G-actin] = 7 µM; [Arp2/3] = 6.25 nM (green dots), 25 nM (blue dots), and 100 nM (red dots). For all three cases, the mean bundles' length, <italic>L_Fin</italic>, increases with [fascin]. (b) Conditions are: 7 µM G-actin; 100 nM fascin and variable amounts of Arp2/3 complex. The data shows that the average of bundles length, <italic>L_Fin</italic>, decrease with [Arp2/3] monotonically (error bars are ±SD).</p></caption></fig>", "<fig id=\"pone-0003297-g006\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003297.g006</object-id><label>Figure 6</label><caption><title>Analytic Model and KMC simulations emphasize the role of Arp2/3 in determining the thickness of bundles.</title><p>(a). Left: snapshot of an aster taken from the simulations. Filaments originating in the periphery shell of the aster might bend towards each other and link into bundles. Right: Values of <italic>L_min</italic>, <italic>L_max</italic> and <italic>b</italic> that were taken from the simulations at <italic>t_i</italic>, for the two [Arp2/3] tested. (b). Bundles formation from the branched network occurs by the elongation–association model. The length <italic>l</italic> represents the bent portion of a filament/bundle. (c) The change in energy caused by bending and linking of two adjacent bundles, as a function of their number of shells s (which also corresponds to the bundles' thickness), for several values of <italic>L</italic>. For [Arp2/3] = 12.5 nM: <italic>L</italic> = 6 µm (red), 9 µm (blue), 12 µm (light green), and 15 µm (pink), and <italic>b</italic> = 0.3 µm. For [Arp2/3] = 100 nM (magnified in the black box): <italic>L</italic> = 0.8 µm (light blue), 1.0 µm (brown), and 1.2 µm (green), and <italic>b</italic> = 0.11 µm. In both cases [G-actin] = 7.5 µM.</p></caption></fig>" ]

[]

[ "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>" ]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"pone.0003297.s001\"><label>Appendix S1</label><caption><p>(0.38 MB DOC)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003297.s002\"><label>Figure S1</label><caption><p>Conditions are: 7 µM G-actin, (a–b) 12.5 nM Arp2/3 complex, and 25 nM GST-VCA and (c–d) 100 nM Arp2/3, and 200 nM GST-VCA. The amount of fascin required inducing transition from aster (a and c) to stars (b and d) ranges between 5 to 6 nM and between 6.5 to 7 nM at 12.5 nM and 100 nM Arp2/3 complex, respectively. The insets in b and d show zoom-in images of the bundles emanating from the stars at each [Arp2/3]. At the transition, the bundles emanating from the stars at lower [Arp2/3] (b) are thicker than those originating from the aster core at 100 nM Arp2/3 (d). Bars are 10 µm.</p><p>(2.78 MB TIF)</p></caption></supplementary-material>" ]

[ "<fn-group><fn fn-type=\"COI-statement\"><p><bold>Competing Interests: </bold>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p><bold>Funding: </bold>A.B.G. wishes to thank the Joseph and May Winston Foundation Career Development Chair in Chemical Engineering, the Israel Cancer Association (grant No. 20070020B) and the Israel Science Foundation (grant No. 551/04) for financial support. A.B.S. thanks the US-Israel Binational Science Foundation (BSF grant 2006-401), and the Israel Science Foundation (ISF grant 659/06) for financial support. The Fritz Haber research center, where the calculations were carried out, is supported by the Minerva foundation, Germany.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"pone.0003297.g001\"/>", "<graphic xlink:href=\"pone.0003297.g002\"/>", "<graphic xlink:href=\"pone.0003297.g003\"/>", "<graphic xlink:href=\"pone.0003297.g004\"/>", "<inline-graphic xlink:href=\"pone.0003297.e001.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pone.0003297.e002.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pone.0003297.g005\"/>", "<graphic xlink:href=\"pone.0003297.g006\"/>", "<inline-graphic xlink:href=\"pone.0003297.e003.jpg\" mimetype=\"image\"/>" ]

[ "<media xlink:href=\"pone.0003297.s001.doc\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003297.s002.tif\"><caption><p>Click here for additional data file.</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Many studies have demonstrated large differences between males and females in the forces of evolution, i.e., mutation, recombination, selection, gene flow, and genetic drift. For example, mutation rates are often higher in males while females tend to have higher rates of recombination ##REF##17976181##[1]##. While the effects of sex-biased mutation and recombination have been directly estimated through genetic studies, we know very little about the extent to which sex-specific differences in gene flow and genetic drift have shaped patterns of variation at the level of the genome. For mammals, it is well known that females and males do not exhibit symmetrical behavior with respect to mating and dispersal practices. For instance, the typical mammalian system is characterized by polygyny (a mating practice in which a minority of males sire offspring with multiple females) and female philopatry (the tendency for females to breed at or near their place of origin) ##UREF##0##[2]##. The development of sex-specific markers in humans has been instrumental in providing insights into the effects of sex-specific demographic processes. Contrasting patterns of diversity on the mitochondrial DNA (mtDNA) and non-recombining portion of the Y chromosome (NRY) have been interpreted to reflect sex-specificity in the rate and scale of migration and in effective population size ##REF##18093995##[3]##–##REF##15317874##[5]##. However, these patterns could also reflect different molecular properties of these two haploid systems, differential selection, or stochasticity in the evolutionary process ##REF##15317874##[5]##. Unlike mtDNA and the NRY, the autosomes and X chromosome undergo recombination and contain numerous evolutionarily independent loci. Additionally, selection only affects those loci that are closely linked to selected sites. Consequently, different patterns of neutral polymorphism associated with the X chromosome and autosomes may be more directly ascribed to demographic differences between females and males.</p>", "<p>Under standard models of DNA sequence evolution ##REF##5162686##[6]##, the level of neutral polymorphism expected at equilibrium is governed by the product of <italic>N_e</italic> (the effective population size) and the mutation rate. Since males carry only one X chromosome, the ratio of the X chromosome effective population size (<italic>N_x</italic>) to the autosomal effective population size (<italic>N_a</italic>) is expected to be ∼0.75 in simple models of a randomly mating population with equal numbers of breeding males and females (i.e., neutral models). Equivalently, if we correct for any differences in mutation rates across chromosomes, the X chromosome should have roughly 75% of the genetic diversity of the autosomes. However, under more complicated models the ratio of X to autosomal diversity levels can vary considerably ##REF##7713404##[7]##. For example, in populations with a female-biased sex-ratio, X-linked diversity will be higher than 75% of autosomal diversity ##UREF##2##[8]##, while in populations that have undergone recent population bottlenecks X-linked diversity will generally be less than 75% of autosomal diversity ##REF##10406117##[9]##,##REF##12242234##[10]##. In addition, if directional selection typically operates on mutations that are at least partly recessive, standard theory predicts that levels of diversity at linked neutral sites will be differentially affected depending on the chromosomal mode of inheritance. For advantageous recessive mutations, hemizygosity in males leads to a higher fixation rate on the X chromosome relative to the autosomes. This in turn will lead to less variability on the X chromosome relative to the autosomes due to the increased prevalence of genetic ‘hitchhiking’ ##UREF##3##[11]##–##REF##4407212##[13]##. In contrast, widespread purifying or background selection should reduce diversity on the autosomes more so than on the X chromosome ##UREF##3##[11]##.</p>", "<p>In this paper, we analyze DNA sequence data that were collected by Wall et al. ##REF##18493019##[14]## for the purpose of testing models of human demographic history. In particular, we analyze data from the X chromosome and autosomes to examine the role that sex-specific processes have played in shaping genomic patterns of variability. We consider several alternative models that could lead to a skew in the ratio of X chromosome to autosomal diversity. Our sequence database includes 40 intergenic regions (20 on the X chromosome and 20 on the autosomes), each of which encompasses ∼20 kb of DNA (##FIG##0##Figure 1##). The sequenced regions were chosen from intergenic/non-coding (i.e., putatively non-functional) regions of medium or high recombination (r≥0.9 cM / Mb) to minimize any potential confounding effects of natural selection (see ##REF##18493019##[14]## for details). These data are also well-suited for testing the role of demographic processes in influencing patterns of diversity because all sites are resequenced in each individual, and multiple diverse human populations are represented in our survey (i.e., Biaka from Central African Republic, Mandenka from Senegal, San from Namibia, French Basque, Han Chinese and Melanesians from Papua New Guinea). We also utilize the recently available orangutan genome to obtain more accurate estimates of the underlying mutation rate for each of the regions studied.</p>" ]

[ "<title>Methods</title>", "<title>Population Samples and Genomic Regions Sequenced</title>", "<p>The DNA samples used in this study come from the CEPH Human Genome Diversity Panel ##REF##11954565##[41]##, the YCC collection ##REF##11827954##[42]##, and established collections in the Hammer lab (see ##REF##18493019##[14]## for details). The regions used for sequencing were selected to minimize any potential confounding effects of natural selection. Specifically, we identified 40 different intergenic (i.e., putatively non-functional) regions of ∼20 Kb in length with to medium to high recombination (r≥0.9 cM/Mb) ##REF##12053178##[43]##. While the genome-wide average recombination rate (mean±SE) for autosomes and the X chromosome are ∼1.29±0.018 cM/Mb and ∼1.25±0.091 cM/Mb, respectively ##REF##12053178##[43]##, the average recombination rate for our autosomal and X-linked loci are 2.18±0.16 and 2.29±0.23, respectively. Each region was at least 50 Kb (100 Kb for the autosomes) away from the nearest gene; within each region, we gathered ∼4–6 Kb of sequence data from 3 or 4 discrete subsections that spanned most of the distance of each region (<italic>locus trio</italic>). For more details on the sequenced regions and the sequencing strategy, see Wall et al. ##REF##18493019##[14]##. See ##SUPPL##5##Table S5## for the number of alleles sequenced at each locus.</p>", "<title>Estimation of Effective Population Sizes of the X Chromosome and Autosomes (N_X / N_A)</title>", "<p>We used a maximum-likelihood framework for estimating the effective population size for the X chromosome (denoted by <italic>N_x</italic>) and for the autosomes (<italic>N_a</italic>). We did this separately for each of the six study populations. We tabulated the number of segregating sites and the number of fixed differences (between human and orangutan) for each locus, and then used coalescent simulations ##REF##11847089##[44]## to estimate the probability of these observations as a function of the population size and the mutation rate. Similar results were obtained when the chimpanzee was used as an outgroup.</p>", "<p>We assumed an average generation time of <italic>g</italic> = 25 years for all human generations since the human most recent common ancestor (MRCA) and an average generation time of <italic>g</italic> = 20 years for all generations between the human MRCA and the orangutan sequence. We fixed the human–orangutan split time at 15 million years ago and assumed an ancestral human–orangutan population size of 40,000 for the autosomes and 30,000 for the X chromosome. The mutation rate was assumed to be constant per base pair, but different for the X (<italic>μ_X</italic>) and the autosomes (<italic>μ_A</italic>).</p>", "<p>For a specific population, let <italic>S_Ai</italic> denote the number of segregating sites in the <italic>i</italic>-th autosomal locus and let <italic>S_Xj</italic> denote the number of segregating sites in the <italic>j</italic>-th X-linked locus. Similarly, let <italic>D_Ai</italic> and <italic>D_Xj</italic> denote the number of fixed differences between human and orangutan at the <italic>i</italic>-th autosomal and the <italic>j</italic>-th X-linked locus respectively. We correct <italic>D_Ai</italic> and <italic>D_Xj</italic> for multiple hits using the Jukes-Cantor model ##UREF##9##[45]##. Then, the likelihood we are interested in isOur basic strategy is to consider a grid of <italic>μ_A</italic>, <italic>μ_X</italic>, <italic>N_A</italic>, and (<italic>N_X</italic> / <italic>N_A</italic>) values and to use Monte Carlo coalescent simulations to estimate (1) for each grid point. In particular, <italic>μ_A</italic> and <italic>μ_X</italic> are incremented in units of 0.1×10⁻⁸/bp per generation, <italic>N_a</italic> is incremented in units of 500, and (<italic>N_x</italic> / <italic>N_a</italic>) is incremented in units of 0.05. At each locus we generate 10⁵ ancestral recombination graphs (ARGs) of <italic>n</italic> = 9–34 human sequences (corresponding to the sample size for the actual data) and one orangutan sequence, reproducing both the actual lengths sequenced and the gaps between the sequenced segments. These ARGs have a recombination rate that is constant per base pair, with the rate estimated from the deCODE map ##REF##12053178##[43]##, assuming an effective population size of 12,500. Next, we tabulated the total branch lengths of branches that would lead to segregating sites or fixed differences. For any particular set of parameter values {<italic>μ_A</italic>, <italic>μ_X</italic>, <italic>N_a</italic>, and (<italic>N_x</italic> / <italic>N_a</italic>)}, it is straightforward to calculate the expected number of segregating sites and fixed differences under the infinite-sites model. Denote these by <italic>ES</italic> and <italic>ED</italic> respectively. Then, the probabilities in (1) follow from the Poisson distribution, andfor the autosomal loci orfor the X-linked loci.</p>", "<p>Note that the same set of simulations is used to estimate probabilities for a locus over all grid points simultaneously. This added computational efficiency comes at the cost of assuming ρ / bp (for the ARGs) is the same across all different values of <italic>N_a</italic> and <italic>N_x</italic>. Simulations assume a constant population size and no population structure for each human population. The results are somewhat robust to specific demographic assumptions (see below).</p>", "<title>Estimating the Ratio of Female to Male Effective Population Size (Breeding Sex Ratio)</title>", "<p>Denote the female effective population size by <italic>N_f</italic> and the male effective population size by <italic>N_m</italic>. We use two separate approaches for estimating the breeding sex ratio <italic>α</italic> = <italic>N_f</italic> / <italic>N_m</italic>. First, we use a method of moments approach to obtain point estimates of <italic>α</italic>. Define <italic>ß</italic> = <italic>N_x</italic> / <italic>N_a</italic>. From standard population genetics ##UREF##2##[8]##,Substituting and rearranging terms leads toWe then substitute the point estimates for <italic>ß</italic> obtained above to generate point estimates for <italic>α</italic>.</p>", "<p>The second method to estimate the breeding sex ratio is a likelihood-based approach similar to the method for estimating <italic>N_x</italic> / <italic>N_a</italic> described above. As before, we use maximum-likelihood to obtain a point estimate (of <italic>α</italic>) and likelihood-ratio tests to estimate 95% confidence intervals, separately for each of the six populations. In this approach, we assume no recombination within loci, free recombination between loci, and no variation in coalescence times of lines in the ancestral human-orangutan population across the genome. Unlike the previous method, we assume that the mutation rates are not constant across loci. Denote the mutation rates at the i-th autosomal locus and the j-th X-linked locus by <italic>μ_Ai</italic> and <italic>μ_Xj</italic>, respectively. Using the same notation as before, the desired likelihood isSince each locus is independent, we can simply maximize the likelihood over <italic>μ</italic>, <italic>N_a</italic> and <italic>α</italic> separately for each locus.</p>", "<p>For the divergence terms the probability is Poisson distributed:where <italic>t</italic> is the number of generations since the human-orangutan split and <italic>N_o</italic> is the effective population size of the ancestral human-orangutan population for the locus in question. For the polymorphism terms, we utilize an exact expression that is available for the standard coalescent without recombination ##REF##6505980##[46]##:where <italic>N_e</italic> is the effective population size of the locus and <italic>k</italic> is the sample size. Note that no simulations are necessary for calculating likelihoods. Point estimates for <italic>α</italic> (as well as 95% CI) are shown in ##SUPPL##1##Table S1##. Results on the performance and robustness of the various estimation methods used are described in ##SUPPL##6##Text S1##.</p>", "<title>Sensitivity of Estimates of <italic>N_x</italic> / <italic>N_a</italic> to Demographic Assumptions</title>", "<p>To test whether alternative demographic models might influence the observed ratio <italic>N_x</italic> / <italic>N_a</italic>, we considered simple models that incorporated a population bottleneck and/or recent population growth. These simulations assumed <italic>N_x</italic> = 7,500, <italic>N_a</italic> = 10,000, <italic>g</italic> = 25 years, <italic>θ</italic> = ρ = 0.001 / bp in the ancestral population, <italic>n</italic> = 32 for the autosomes and <italic>n</italic> = 16 for the X chromosome. Our growth only model assumed that a population of constant size began growing exponentially at various times in the past (i.e, 10, 15, and 20 kya), expanding to a size 100-fold larger than the ancestral population. Our bottleneck only model assumed that an ancestral population underwent a 100-fold decrease in size at various times in the past (i.e., 10, 20, 30, and 40 kya) before instantaneously recovering its original size. In all cases the bottleneck lasted for 40 generations. We also considered a modification of the bottleneck model where the population grows exponentially at various times (i.e., 10, 15, and 20 kya) after recovering from the bottleneck described above (for cases where the onset of the bottleneck was 20 and 40 kya). For each parameter combination, we simulated 10⁴ replicates of a 5 Kb region, and tabulated <italic>θ̅</italic>\n<italic>_x</italic>/<italic>θ̅</italic>\n<italic>_a</italic>.</p>", "<p>We then considered simple two-deme island models to test the effects of sex-biased migration rates on <italic>θ̅</italic>\n<italic>_x</italic>/<italic>θ̅</italic>\n<italic>_a</italic>. Each population experiences a per-generation migration rate of 3–9×10⁻⁵ and an effective population size of 10⁴. We test symmetric migration models in which females and males migrate equally between demes and in which only females or only males migrate between demes, as well as asymmetric models in which females and males migrate in opposite directions between demes. We performed 10,000 simulations for each model (##SUPPL##2##Table S2##).</p>" ]

[ "<title>Results</title>", "<p>We analyze a total of ∼210 kb of DNA sequence representing 40 loci from the X chromosome and autosomes from each of 90 humans and three great apes, or a total of ∼18.9 Mb ##REF##18493019##[14]##. ##TAB##0##Table 1## provides basic summary statistics for nucleotide diversity in six human populations, as well as the ratio of diversity to human-orangutan sequence divergence. We also use levels of divergence between humans and orangutan (see <xref ref-type=\"sec\" rid=\"s4\">Methods</xref>) to estimate mutation rates for each region (##SUPPL##1##Table S1##), and then estimate relative effective population sizes of the X chromosome and autosomes (<italic>N_x</italic> / <italic>N_a</italic>) based on observed levels of diversity (<italic>θ_W</italic>) ##REF##18493019##[14]##. We find that this ratio is higher than expected in all six populations, ranging from 0.85 in the San to 1.08 in the Basque (##FIG##1##Figure 2##). When we use levels of divergence between humans and chimpanzees to estimate mutation rates for the autosomal and X-linked regions, we obtain similar results. For instance, X/A diversity ratios (e.g., π/D in ##TAB##0##Table 1##) using chimpanzee and orangutan divergence are highly correlated for the six human populations (r² = 0.95, P = 0.001) (data not shown). We also obtain similar π/D values when we subsample the human dataset to standardize the number of autosomes and X chromosomes (##SUPPL##2##Table S2##).</p>", "<p>To test whether the observed ratios are significantly different from 0.75, we employ a maximum-likelihood method to estimate confidence intervals. Our method uses a population genetic model (i.e., the coalescent) to account for the inherent uncertainty in estimating diversity and divergence rates from sequence data. ##FIG##1##Figure 2## shows 95% confidence intervals for <italic>N_x</italic> / <italic>N_a</italic>. For three out of six populations (Basque, Melanesians and Mandenka), the 95% confidence intervals for the ratio of X-linked and autosomal effective population sizes does not include 0.75 (p = 0.001, 0.005 and 0.030 for the Basque, Melanesians and Mandenka, respectively). One interpretation of these results is that there is strong evidence for an unequal female and male <italic>N_e</italic> in at least three of our six populations, with estimates of the breeding sex ratio (i.e., the effective size of females to males) ranging from 2.1 in the San to 12.5 in the Basque. If the observed differences in nucleotide variability on the X chromosome and autosomes are caused by long-term (demographic) processes, then the estimates of <italic>N_x</italic> / <italic>N_a</italic> presented in ##FIG##1##Figure 2## will be highly correlated due to shared population history. When we use the intersection of all six confidence intervals (0.87–1.02) to estimate the range of <italic>N_x</italic> / <italic>N_a</italic> values that are consistent with the data from all six populations, we estimate the range of the breeding sex ratio to be 2.4–8.7. We also note that even with a conservative Bonferroni correction, a 1∶1 breeding sex ratio is rejected in two out of six populations.</p>", "<p>We also employ a separate method for estimating the breeding sex ratio in each population that does not allow for intra-locus recombination but does permit independent mutation rates across loci (see <xref ref-type=\"sec\" rid=\"s4\">Methods</xref>). This method produces similar results to those described above, with estimates of the ratio of female to male effective population size ranging from 1.8 in the San to 14.0 in the Basque (##SUPPL##3##Table S3##). We interpret this as additional evidence that the unusual patterns observed in our data are real and require explanation.</p>" ]

[ "<title>Discussion</title>", "<p>Our findings of high levels of diversity on the X chromosome relative to the autosomes are in marked contrast to results of previous studies in a wide range of species including humans ##REF##12082133##[15]##,##REF##17971168##[16]##, house mice ##REF##17287527##[17]##, flycatcher ##REF##16313454##[18]##, chicken ##REF##15166162##[19]##, and Drosophila ##REF##10823947##[12]##,##REF##11805060##[20]## (but see ##REF##17961244##[21]##). Indeed, many evolutionary models, such as recent population bottlenecks ##REF##10406117##[9]##,##REF##12242234##[10]## and recurrent selective sweeps ##UREF##3##[11]## are expected to <italic>reduce</italic> the relative levels of X-linked diversity, contrary to what we find.</p>", "<p>Could our results be due to sequencing error? Since most of the samples used for sequencing came from males, the X chromosome data are essentially haploid while autosomal data are necessarily diploid. In principle, this could lead to a systematic bias in estimates of genetic diversity across different chromosomes. In particular, if diploid sequencing tends to miss rare variants, we might expect the estimates of autosomal diversity to be too low, leading to overestimates of <italic>N_x</italic> / <italic>N_a</italic>. Given that the levels of diversity for our autosomal loci ##REF##18493019##[14]## are higher than those found in other large-scale studies of human sequence data (e.g., ##REF##16124863##[22]##–##REF##16352722##[24]##), we find this possibility to be highly unlikely. Moreover, there is no evidence that we have preferentially missed rare variants on the autosomes as mean Tajima's D values for our autosomal loci are comparable to those in other studies of non-genic regions ##REF##16352722##[24]## and more negative than those for our X-linked loci in five out of the six populations in ##TAB##0##Table 1##.</p>", "<p>The only other multi-locus study we are aware of that allows for a similar comparison of X versus autosomal diversity in thoroughly sampled, non-admixed human populations is the NIEHS SNPs study ##REF##15364900##[25]##. Since this study focuses on genic regions, it is not clear whether analyses of their data are directly comparable to the results described here. For example, genic regions in non-African populations of <italic>Drosophila melanogaster</italic> show reduced X-linked versus autosomal nucleotide diversity while intergenic regions do not (<italic>i.e.</italic>, X-linked and autosomal diversity levels are similar) ##REF##17961244##[21]##. Nonetheless, application of our methods to genes in the NIEHS study with recombination rates similar to those for our intergenic regions (i.e., r&gt;0.9 cM/Mb) yields estimates of <italic>N_x</italic> / <italic>N_a</italic> ranging from 0.87 in the Yoruba to 1.08 in the CEPH (results not shown), similar to the point estimates shown in ##FIG##1##Figure 2##. One interpretation of these results is that the long-term male effective population size is substantially smaller than the long-term female effective population size ##REF##7713404##[7]##,##REF##11355571##[26]##; however, other evolutionary processes may account for our observations. In the following sections we discuss four alternative models to explain the higher observed levels of X-linked to autosomal diversity than expected under neutral models.</p>", "<title>Background Selection</title>", "<p>While directional selection on recessive beneficial mutations is expected to lead to more frequent hitchhiking and lower diversity on the X chromosome compared with the autosomes, linked negative selection on the X chromosome and autosomes (background selection) predicts the opposite pattern ##UREF##3##[11]##,##UREF##4##[27]##. Because recessive deleterious mutants are maintained at lower frequency and removed from populations more quickly on X chromosomes than on autosomes, neutral alleles on X chromosomes are less likely to be linked to a deleterious mutant compared with neutral alleles on autosomes. Thus, all else being equal, background selection should leave X chromosomes more polymorphic than autosomes at linked, neutral sites after correcting for expected differences in population size between X chromosomes and autosomes ##REF##10823947##[12]##. Because the effects of background selection are expected to be stronger (i.e., reduce local <italic>N_e</italic>) in chromosomal regions with lower rates of recombination, we did not <italic>a priori</italic> believe that background selection would be a significant factor because our experimental design focuses on intergenic DNA in regions of moderate to high recombination ##REF##18493019##[14]##. To further explore the potential effects of background selection we assume an average number of deleterious mutations per generation of 4 ##REF##10978293##[28]## and use equation 15 in Hudson and Kaplan ##UREF##5##[29]## to estimate the ratio of observed to expected polymorphism. We find this ratio to be 0.934, which suggests that background selection is unlikely to reduce autosomal diversity by more than 6.6% relative to X-linked diversity. We note that this estimate is conservative in that it ignores the effects of background selection on the X chromosome ##REF##8082838##[30]##,##REF##8801188##[31]##. Thus, it seems unlikely that background selection alone can explain our results. We also point out that alternative selection-based models involving the greater accumulation of sex-antagonistic polymorphisms on the sex chromosomes ##REF##11886642##[32]## may be viable.</p>", "<title>Demographic Processes Affecting the Entire Population</title>", "<p>Historical changes in population size (such as founder effects and bottlenecks) also might have differential effects on loci with different modes of inheritance ##REF##10406117##[9]##,##REF##12242234##[10]##,##REF##16582448##[33]##. Using a simulation approach, we test three plausible models of recent demographic history that incorporate a recent population bottleneck and/or recent population growth. For example, we test a model incorporating 100-fold exponential growth from a constant effective population size of 10⁴, a bottleneck model with a 100-fold reduction in size followed by instantaneous recovery to an ancestral effective size of 10⁴, and a model incorporating the aforementioned bottleneck followed by 100-fold exponential growth (see <xref ref-type=\"sec\" rid=\"s4\">Methods</xref> for details). For all parameters tested, the effects on expected relative levels of diversity are minor and in the direction towards reduced X-linked polymorphism (##TAB##1##Table 2##).</p>", "<p>Recently, Pool and Nielsen ##REF##17971168##[16]## used an analytical approach to examine the effect of changing population sizes on the expected coalescence time for a pair of sequences with different effective population sizes. They showed that population size reductions can lead to particularly low X-linked diversity, whereas population growth can elevate X-linked relative to autosomal diversity. We employ Pool and Nielsen's ##REF##17971168##[16]## model (which is similar to the bottleneck model described above), substituting parameters that are reasonable for human demographic history (see ##SUPPL##6##Text S1## for details). When we examine the effects of such a bottleneck over a range of times in the past, we do not find that the expected X/A diversity ratio shifts much above 0.75 (##SUPPL##0##Figure S1A##). When we search for combinations of parameters that yield X/A diversity ratios and levels of nucleotide diversity similar to those that we observe, we find that ancient bottlenecks (i.e., older than ∼100 kya) coupled with population growth, can indeed produce expected X/A diversity ratios as high as 0.85 (##SUPPL##0##Figure S1B##). However, computer simulations using these same bottleneck and growth parameters yield summaries of the site frequency spectrum that are inconsistent with those that we observe; i.e., Tajima's D values that are much more negative than those in ##TAB##0##Table 1## for both the X chromosome and autosomes (−1.55 and −1.91, respectively).</p>", "<title>Sex-Biased Forces</title>", "<p>There are a number of sex-biased evolutionary forces acting within human populations that are known to have differential effects on loci with different modes of inheritance. A demographic process that may lead to a skew in X-linked versus autosomal diversity is differential migration rates for males and females in a structured population. To explore the effects of sex-biased migration on ratios of X/A diversity, we simulate a two-deme island model with different rates of male and female migration. First, we simulate a symmetric model with only a single sex (females) migrating. We assume effective population sizes and migration rates that produce F_ST values that are similar to those observed in human populations (i.e., autosomal F_ST ∼0.12; ##REF##18493019##[14]##). Because females are exchanging demes at the same rate, it is not surprising that this model yields X/A diversity ratios that are close to those expected under panmixia (i.e., 0.75) (##SUPPL##4##Table S4##). Second, we simulate a model in which one deme sends out females and the other deme sends out males at the same rate. The results indicate that when the X-linked diversity exceeds the value expected under panmixia in one deme, the other deme always shows a deficit of X-linked diversity (##SUPPL##4##Table S4##). If we assume that the six populations that we sampled here evolve independently according to this two-deme model, the probability of observing excess polymorphism on the X chromosome for all six populations would be at most 1/64 (P&lt;0.016). These results are consistent with Laporte and Charlesworth's ##REF##12242257##[34]## simulations showing that sex-biased migration only weakly skews levels of X/A diversity unless populations are strongly subdivided. Therefore, we believe that population structure is unlikely to generate a bias towards increased diversity on the X chromosome in all populations, but could contribute to differential bias among populations.</p>", "<p>A higher variance in male reproductive success over that in females due to sexual selection is also expected to inflate the ratio of X-linked to autosomal polymorphism. In populations with age structure, an additional contribution to the variance in net reproductive success can be caused by the stochastic nature of survival during the reproductive phase and by differences in fertility among individuals in different age classes ##REF##11355571##[26]##. However, demographic factors of this kind (e.g., lower male survival during adult life or delayed male versus female age of maturity) are unlikely to have a major effect on the relative effective population sizes of X-linked and autosomal loci ##REF##11355571##[26]##. In contrast, an excess variance of male reproductive success over Poisson expectations can have large effects: With an extremely high variance in male fertility relative to female fertility, the ratio <italic>N_x</italic> / <italic>N_A</italic> approaches 1.125 ##REF##7713404##[7]##,##REF##11355571##[26]##,##REF##12242257##[34]##.</p>", "<title>Conclusion</title>", "<p>A number of evolutionary forces may be responsible for increasing the effective population size of X-linked versus autosomal loci. Under reasonable parameters for human populations, our results suggest that background selection, changes in population size, and sex-specific migration in a structured population may each have a minor effect in increasing the ratio of X-linked to autosomal polymorphism over that expected under neutral models. While it is possible that multiple processes acting together might lead to a major effect (i.e., on the order of what is observed here), we hypothesize that a higher variance in male versus female reproductive success can by itself explain most of the observed increase in effective population size of the X chromosome. The human mating system is considered to be moderately polygynous, based on both surveys of world populations ##UREF##6##[35]##,##UREF##7##[36]## and on characteristics of human reproductive physiology ##REF##11932733##[37]##–##REF##7266658##[39]##. The practice of polygyny, in both the traditional sense and via ‘effective polygyny’ (whereby males tend to father children with more females than females do with males—a common practice in many contemporary western cultures ##UREF##8##[40]##), would tend to increase the variance in reproductive success among males. In other words, when more men than women in any generation fail to have any children, and more men than women have very large numbers of children, autosomal <italic>N_e</italic> is reduced relative to that of the X chromosome. While polygyny may be the most important factor influencing the ratio of X-linked to autosomal diversity, we point out that this process by itself is unlikely to account for all the patterns of nucleotide polymorphism observed here (e.g., the frequency spectrum as summarized by Tajima's D in ##TAB##0##Table 1##). Future theoretical work examining the joint effects of multiple demographic processes (e.g., sex-biased bottlenecks in which populations are founded by more females than males (e.g., ##REF##17961244##[21]##) and experimental research (e.g., aimed at refining estimates of the ratio of X-linked to autosomal neutral polymorphism in additional populations) will increase our understanding of how the different forces of evolution influence variation on the autosomes and X chromosome.</p>" ]

[ "<title>Conclusion</title>", "<p>A number of evolutionary forces may be responsible for increasing the effective population size of X-linked versus autosomal loci. Under reasonable parameters for human populations, our results suggest that background selection, changes in population size, and sex-specific migration in a structured population may each have a minor effect in increasing the ratio of X-linked to autosomal polymorphism over that expected under neutral models. While it is possible that multiple processes acting together might lead to a major effect (i.e., on the order of what is observed here), we hypothesize that a higher variance in male versus female reproductive success can by itself explain most of the observed increase in effective population size of the X chromosome. The human mating system is considered to be moderately polygynous, based on both surveys of world populations ##UREF##6##[35]##,##UREF##7##[36]## and on characteristics of human reproductive physiology ##REF##11932733##[37]##–##REF##7266658##[39]##. The practice of polygyny, in both the traditional sense and via ‘effective polygyny’ (whereby males tend to father children with more females than females do with males—a common practice in many contemporary western cultures ##UREF##8##[40]##), would tend to increase the variance in reproductive success among males. In other words, when more men than women in any generation fail to have any children, and more men than women have very large numbers of children, autosomal <italic>N_e</italic> is reduced relative to that of the X chromosome. While polygyny may be the most important factor influencing the ratio of X-linked to autosomal diversity, we point out that this process by itself is unlikely to account for all the patterns of nucleotide polymorphism observed here (e.g., the frequency spectrum as summarized by Tajima's D in ##TAB##0##Table 1##). Future theoretical work examining the joint effects of multiple demographic processes (e.g., sex-biased bottlenecks in which populations are founded by more females than males (e.g., ##REF##17961244##[21]##) and experimental research (e.g., aimed at refining estimates of the ratio of X-linked to autosomal neutral polymorphism in additional populations) will increase our understanding of how the different forces of evolution influence variation on the autosomes and X chromosome.</p>" ]

[ "<p>Conceived and designed the experiments: MFH JDW. Performed the experiments: MPC AEW. Analyzed the data: MFH FLM MPC AEW JDW. Wrote the paper: MFH FLM JDW.</p>", "<p>Comparisons of levels of variability on the autosomes and X chromosome can be used to test hypotheses about factors influencing patterns of genomic variation. While a tremendous amount of nucleotide sequence data from across the genome is now available for multiple human populations, there has been no systematic effort to examine relative levels of neutral polymorphism on the X chromosome versus autosomes. We analyzed ∼210 kb of DNA sequencing data representing 40 independent noncoding regions on the autosomes and X chromosome from each of 90 humans from six geographically diverse populations. We correct for differences in mutation rates between males and females by considering the ratio of within-human diversity to human-orangutan divergence. We find that relative levels of genetic variation are higher than expected on the X chromosome in all six human populations. We test a number of alternative hypotheses to explain the excess polymorphism on the X chromosome, including models of background selection, changes in population size, and sex-specific migration in a structured population. While each of these processes may have a small effect on the relative ratio of X-linked to autosomal diversity, our results point to a systematic difference between the sexes in the variance in reproductive success; namely, the widespread effects of polygyny in human populations. We conclude that factors leading to a lower male versus female effective population size must be considered as important demographic variables in efforts to construct models of human demographic history and for understanding the forces shaping patterns of human genomic variability.</p>", "<title>Author Summary</title>", "<p>Like many primate species, the mating system of humans is considered to be moderately polygynous (i.e., males exhibit a higher variance in reproductive success than females). As a consequence, males are expected to have a lower effective population size (<italic>N_e</italic>) than females, and the proportion of neutral genetic variation on the X chromosome (relative to the autosomes) should be higher than expected under the assumption of strict neutrality and an equal breeding sex ratio. We test for the effects of polygyny by measuring levels of neutral polymorphism at 40 independent loci on the X chromosome and autosomes in six human populations. To correct for mutation rate heterogeneity among loci, we divide our diversity estimates within human populations by divergence with orangutan at each locus. Consistent with expectations under a model of polygyny, we find elevated levels of X-linked versus autosomal diversity. While it is possible that multiple demographic processes may contribute to the observed patterns of genomic diversity (i.e., background selection, changes in population size, and sex-specific migration), we conclude that an historical excess of breeding females over the number of breeding males can by itself explain most of the observed increase in effective population size of the X chromosome.</p>" ]

[ "<title>Supporting Information</title>" ]

[ "<p>We thank Olga Savina for excellent computational support. We also thank Ryan Sprissler and Laurel Johnstone of the Genomic Analysis and Technology Core at the University of Arizona for aid in development of the DNA sequencing pipeline, and John D. Morelli, Brittany Tamarkin, and Kimiko Della Croce for their dedicated computational assistance.</p>" ]

[ "<fig id=\"pgen-1000202-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pgen.1000202.g001</object-id><label>Figure 1</label><caption><title>Loci under study.</title><p>(A) Approximate chromosomal positions of 20 autosomal and 20 X-linked loci (red horizontal line). Each region encompasses ∼20 kb of single-copy non-coding (i.e., putatively non-functional) DNA in regions of medium or high recombination (r≥0.9 cM/Mb). (B) Sequencing strategy. Within each region, ∼4–6 Kb of sequence data were gathered from 3 or 4 discrete subsections (filled blocks) that spanned most of the distance of each region (see ##REF##18493019##[14]## for details).</p></caption></fig>", "<fig id=\"pgen-1000202-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pgen.1000202.g002</object-id><label>Figure 2</label><caption><title>Ratio of effective population sizes for the X chromosome (<italic>N_x</italic>) and autosomes (<italic>N_a</italic>) for each population.</title><p>The diamonds represent the point estimate, while the vertical bar shows the estimated 95% confidence interval. The dotted line represents the expected ratio (0.75) under a neutral model with breeding sex ratio of 1. Three letter population codes are as follows: Melanesians (Mel), Basque (Bas), Han Chinese (Han), Mandenka (Man), Biaka (Bia), San (San).</p></caption></fig>" ]

[ "<table-wrap id=\"pgen-1000202-t001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pgen.1000202.t001</object-id><label>Table 1</label><caption><title>Summaries of nucleotide diversity<xref ref-type=\"table-fn\" rid=\"nt101\">a</xref> and divergence.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Population</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Sample Size<xref ref-type=\"table-fn\" rid=\"nt102\">b</xref>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Segregating Sites</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">θ̅ (%)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">π̅ (%)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">D̅\n(%)<xref ref-type=\"table-fn\" rid=\"nt103\">c</xref>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Tajima's D̅\n</td></tr></thead><tbody><tr><td colspan=\"8\" align=\"left\" rowspan=\"1\">Autosomes</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Mandenka</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">28.2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">536</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.124</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.119</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.429</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.035</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−0.140</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Biaka</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">28.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">574</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.133</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.121</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.429</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.036</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−0.343</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">San</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">19.5</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">500</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.133</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.125</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.421</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.037</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">−0.242</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Han</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">32.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">352</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.079</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.080</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.425</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.023</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.045</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Basque</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">32.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">336</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.076</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.086</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.426</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.025</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.509</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Melanesians</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">18.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">281</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.073</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.078</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">3.426</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.022</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.327</td></tr><tr><td colspan=\"8\" align=\"left\" rowspan=\"1\">X chromosome</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Mandenka</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">16.1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">281</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.090</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.099</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.579</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.038</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.311</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Biaka</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">14.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">281</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.093</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.095</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.586</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.036</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.093</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">San</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">9.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">220</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.083</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.085</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.584</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.033</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.147</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Han</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">16.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">174</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.055</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.058</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.592</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.022</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.089</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Basque</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">16.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">200</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.064</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.071</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.592</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.029</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.535</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Melanesians</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">15.0</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">183</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.059</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.066</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2.588</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.026</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.524</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pgen-1000202-t002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pgen.1000202.t002</object-id><label>Table 2</label><caption><title>Effect of demographic models on X versus autosome diversity. See <xref ref-type=\"sec\" rid=\"s4\">Methods</xref> for details.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Time of growth (Kya)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Time of Bottleneck (Kya)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>θ̅</italic>\n<italic>_x</italic>/<italic>θ̅</italic>\n<italic>_a</italic>\n</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">10</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">—</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.738</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">15</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">—</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.732</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">20</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">—</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.737</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">—</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">10</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.744</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">—</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">20</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.740</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">—</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">30</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.737</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">—</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">40</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.726</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">10</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">20</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.730</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">15</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">20</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.726</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">10</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">40</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.732</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">20</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">40</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.724</td></tr></tbody></table></alternatives></table-wrap>" ]

[ "<inline-formula></inline-formula>", "<disp-formula><label>(1)</label></disp-formula>", "<disp-formula></disp-formula>", "<disp-formula></disp-formula>", "<disp-formula></disp-formula>", "<disp-formula></disp-formula>", "<disp-formula></disp-formula>", "<disp-formula></disp-formula>", "<disp-formula></disp-formula>" ]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"pgen.1000202.s001\"><label>Figure S1</label><caption><p>Predicted ratio of X chromosome to autosome diversity for two bottleneck models.</p><p>(0.28 MB DOC)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pgen.1000202.s002\"><label>Table S1</label><caption><p>Mutation rates at 40 loci assuming a 15-million year human-orangutan divergence time.</p><p>(0.04 MB DOC)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pgen.1000202.s003\"><label>Table S2</label><caption><p>Polymorphism and divergence for autosomal and X-linked loci.</p><p>(0.04 MB DOC)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pgen.1000202.s004\"><label>Table S3</label><caption><p>Breeding sex ratio.</p><p>(0.03 MB DOC)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pgen.1000202.s005\"><label>Table S4</label><caption><p>Migration rates and simulation results for two-deme migration model.</p><p>(0.04 MB DOC)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pgen.1000202.s006\"><label>Table S5</label><caption><p>Sample sizes (number of alleles sequenced) for each locus.</p><p>(0.06 MB DOC)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pgen.1000202.s007\"><label>Text S1</label><caption><p>Supplementary material.</p><p>(0.04 MB DOC)</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><fn id=\"nt101\"><label>a</label><p>Mean nucleotide diversity for 20 autosomal and 20 X-linked loci.</p></fn><fn id=\"nt102\"><label>b</label><p>Mean number of alleles sequenced per locus per population.</p></fn><fn id=\"nt103\"><label>c</label><p>D = human-orangutan sequence divergence.</p></fn><fn id=\"nt104\"><p>Note—slight differences with values in ##REF##18493019##[14]## are due to alignment with different outgroup.</p></fn></table-wrap-foot>", "<fn-group><fn fn-type=\"COI-statement\"><p>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p>This research was supported by a HOMINID grant from the National Science Foundation to MFH and JDW (BCS-0423670 and BCS-0423123).</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"pgen.1000202.g001\"/>", "<graphic xlink:href=\"pgen.1000202.g002\"/>", "<graphic id=\"pgen-1000202-t001-1\" xlink:href=\"pgen.1000202.t001\"/>", "<inline-graphic xlink:href=\"pgen.1000202.e001.jpg\" mimetype=\"image\"/>", "<graphic id=\"pgen-1000202-t002-2\" xlink:href=\"pgen.1000202.t002\"/>", "<graphic xlink:href=\"pgen.1000202.e002.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pgen.1000202.e003.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pgen.1000202.e004.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pgen.1000202.e005.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pgen.1000202.e006.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pgen.1000202.e007.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pgen.1000202.e008.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pgen.1000202.e009.jpg\" mimetype=\"image\" position=\"float\"/>" ]

[ "<media xlink:href=\"pgen.1000202.s001.doc\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pgen.1000202.s002.doc\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pgen.1000202.s003.doc\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pgen.1000202.s004.doc\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pgen.1000202.s005.doc\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pgen.1000202.s006.doc\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pgen.1000202.s007.doc\"><caption><p>Click here for additional data file.</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>The regulation of gene activity is essential for the proper functioning of cells,\nwhich employ a variety of molecular mechanisms to control gene expression. Despite\nthis, there is considerable variation in the precise number and timing of protein\nmolecules that are produced for a given gene under any particular set of\ncircumstances. This is because gene expression is fundamentally a\n“noisy” process, subject to a number of sources of randomness.\nSome of these are <italic>intrinsic</italic> to the biochemical reactions that\ncomprise the transcription and translation of a particular gene ##REF##12183631##[1]##,##REF##16179466##[2]##. Several\nof the reactions involve very small numbers of molecules. There are only one or two\ncopies of the DNA for the gene, and in its vicinity inside the cell there are likely\nto be only a few copies of the relevant transcription factors and of RNA polymerase.\nSimilarly, for each mRNA molecule, the processes of ribosome binding and of mRNA\ndegradation are typically highly stochastic.</p>", "<p>Recent advances in experimental technology have shown that such single molecule\neffects can lead to protein production occurring in bursts of varying size, each due\nto a single transcription factor binding event ##REF##9023339##[3]##,##REF##11438714##[4]##. Other sources of\nvariability are <italic>extrinsic</italic> to the specific reactions, and include\nfluctuations in relevant metabolites, polymerases, ribosomes, etc. ##REF##12183631##[1]##,##REF##16179466##[2]##. These\nwill not be considered further here.</p>", "<p>It is of considerable interest to determine the various contributions of such\ndifferent sources of variability. Within the last few years, experimental techniques\nfor addressing this question have increasingly become available. Elowitz et al.\n##REF##12183631##[1]##\nobserved fluctuations in the expression level of genes tagged both with cyan and\nyellow fluorescent proteins in monoclonal <italic>Escherichia coli</italic> cells\nunder identical environmental conditions. Similar work was carried out by Raser and\nO'Shea ##REF##15166317##[5]## in the eukaryote <italic>Saccharomyces\ncerevisiae</italic>. Such dual-reporter experiments are able to distinguish between\nintrinsic and extrinsic sources of stochasticity. More recently, single molecule\ndata has become available ##REF##16543458##[6]##,##REF##16541077##[7]##, which monitors the expression of a gene a single\nprotein at a time and provides the distribution of the sizes of bursts. It had been\nhoped that data of this kind would answer many of the remaining questions about the\norigin of noise in gene expression and in particular distinguish between the\ndifferent contributions of transcription and translation to intrinsic noise.</p>", "<p>Intuitively, one might expect that randomness due to transcription would play the\nmore significant role than translation, since typically there will be more than one\nmRNA molecule, and the fluctuations due to translation from each of these might to\nsome extent average out. To test this hypothesis and to put it on a quantitative\nbasis, it is necessary to employ mathematical models of gene expression. These also\nprovide a valuable tool for the analysis of experimental data, and in particular of\nthe burst size distributions reported in the literature, e.g., ##REF##16543458##[6]##,##REF##16541077##[7]##.</p>", "<p>A great deal of work has gone into modelling gene expression in both prokaryotic and\neukaryotic systems, with some of the earliest papers predicting fluctuations in mRNA\nand protein levels published 30 years ago ##REF##607033##[8]##,##REF##96307##[9]##. McAdams and Arkin ##REF##9023339##[3]##\nprovided the first model of bursting at the translation level. They showed that the\nnumber of protein molecules produced by a single mRNA transcript is described well\nby a model which considers whether the next event is the production of a further\nprotein, or the degradation of the mRNA molecule. Such competitive binding between\nribosomes and RNase results in a geometric distribution for the protein number. Such\nan analysis can also be applied to transcription following the binding of a\ntranscription factor to a gene and also results in a geometric distribution. The\njoint analysis of these two stochastic processes forms the basis of the present\npaper.</p>", "<p>The integration of simple stochastic (Markov) models of transcription factor, RNA\npolymerase, ribosome and RNase binding leads to what is now widely regarded as the\nstandard model of gene expression for prokaryotes ##REF##11438714##[4]##. The analysis of this\nmodel using a master equation allows the determination of the moments of the\ndistribution of the number of protein molecules when the system is in steady state.\nFurther analysis of this equilibrium distribution was carried out by Paulsson ##REF##14749823##[10]##–##REF##10852944##[12]## who used the master\nequation and the fluctuation–dissipation theorem to obtain predictions\nabout the mean and variances of molecule numbers and lifetimes and the contribution\nmade by transcriptional and translational bursting. Other studies have been carried\nout by Höfer ##REF##16362909##[13]## who used a rapid-equilibrium approximation to\ncompare mRNA levels for genes with one and two active alleles, and by Friedman et\nal. ##REF##17155441##[14]##. The drawback of these approaches is that the master\nequation that describes the temporal evolution of the probability distribution of\nprotein (and mRNA) numbers is too complex to be solved analytically. Furthermore,\nthe burst size distribution necessary for comparison with recent experimental data\n##REF##16543458##[6]##,##REF##16541077##[7]## cannot be\nobtained directly from the master equation. Such difficulties with master equation\nbased approaches are exacerbated in the case of more complex models of gene\nexpression such as multi-step models that include intermediate stages such as the\nformation of DNA–RNA polymerase complexes, phosphorylation events, and\nmRNA–ribosome binding. Both deterministic and stochastic simulation\nstudies of these models have been performed, e.g., ##REF##12087127##[15]## and ##REF##15721042##[16]##, but none of these\napproaches have been useful for the analysis of burst size data.</p>", "<p>In the present work we avoid the problems associated with the master equation\napproach, which are at least in part due to the explicit incorporation of time\nevolution. Instead, we ignore time and directly derive an expression for the burst\nsize distribution by extending the analysis of ##REF##9023339##[3]##. In many ways this\napproach is similar to that used for the analysis of multi-stage queues ##UREF##1##[17]##. The\ndistribution of the number of mRNA molecules produced in a single burst is geometric\nand the distribution of the number of protein molecules produced by a single mRNA is\nalso geometric ##REF##9023339##[3]##. The overall burst size distribution is therefore\ngiven by the compound distribution of two geometric distributions ##UREF##1##[17]##. This\ncan be readily computed using generating functions ##UREF##1##[17]## and is itself\n<italic>not</italic> geometric. However, experimentally it is not possible to\ndetect bursts that produce no protein molecules at all, and therefore the published\ndata ##REF##16543458##[6]##,##REF##16541077##[7]## are in fact\nthe relevant conditional distributions, assuming at least one protein molecule is\nproduced in a burst. Surprisingly, it turns out that when we condition the compound\ndistribution in this way, we again obtain a geometric distribution. This is\ndetermined by a single parameter, which we can derive in terms of physically\nmeaningful constants such as binding and unbinding rates. This shows that different\ncombinations of noise levels in the translation and transcription parts of the\nprocess can give the same overall burst size distribution. Mathematically, this\nmeans that the standard model of gene expression (described in detail below) is\n<italic>nonidentifiable</italic>\n##UREF##2##[18]##,##UREF##3##[19]## from\nburst size data alone. This in turn implies that it is not possible to identify the\nrelative contributions of translation and transcription to the burst size\ndistribution of protein numbers only using this data.</p>", "<p>We also show that our approach is applicable to a variety of more detailed models\nthat incorporate additional steps to provide more realistic descriptions of\nexpression ##REF##15721042##[16]##. These still yield a single parameter geometric\nconditional distribution. This shows that within the context of a very large class\nof models, experimental burst size data on its own cannot identify the relative\ncontributions of different reactions to the overall noise level. However, by\nsimulating the equilibrium distribution of protein numbers for different parameter\ncombinations giving the same burst size distribution we demonstrate that a\ncombination of burst size distribution and equilibrium distribution can discern\ndifferent sources of noise. The difficulty with such an approach is that the\ndetermination of the equilibrium distribution requires the knowledge of two\nadditional kinetic parameters: the transcription factor binding rate and the protein\ndegradation rate. Estimates of these are not easy to obtain independently, so that\nwe now have to estimate six unknown parameters from the combined burst size and\nequilibrium distribution data. Initial simulations (not shown here) suggested that\nit is difficult to do this reliably.</p>", "<p>It is possible however, by using independent estimates of one of the parameters to\nreduce the parameter space from six to five dimensions. Using the relationship\nbetween the remaining parameters determined from the burst size distribution allows\nthe elimination of a further parameter, leaving four kinetic parameters to be\nestimated from the equilibrium distribution. We show below that by using the\nNelder-Mead algorithm to maximize the empirical likelihood, useful estimates of the\nfour remaining parameters can be obtained. We carry out this process twice, first\nusing independent measurements of the mRNA degradation rate and then of the protein\nhalf-life. In the first case we obtain unrealistically short estimates of the\nprotein half-life, and in the second a considerably faster mRNA degradation. This\nsuggests that when in the repressed state, mRNA may be degraded at a faster rate\nthan when the gene is active.</p>", "<p>In principle, this method can be applied to any gene where burst size and equilibrium\ndistributions are available, providing a new approach to the estimation of\nparameters estimates for the ever more sophisticated models increasingly being used\nin computational biology.</p>" ]

[ "<title>Methods</title>", "<title>The Standard Model of Gene Expression</title>", "<p>In the so called “standard model” of gene expression, ##FIG##0##Figure 1##, an inactive gene can\nbe activated by a promoter or transcription factor. This allows molecules of RNA\npolymerase to bind and produce mRNA. This in turn can bind to ribosomes leading\nto the production of protein molecules. Eventually the transcription factor\nunbinds, terminating the production of mRNA, and each mRNA molecule is degraded,\nwhich stops protein production.</p>", "<p>Each of these processes is modelled as a transition in a continuous time Markov\nchain with a particular rate. Such a rate is interpreted as the probability of\nan event occurring in a unit time interval. Thus, if we denote the rate of\ntranscription factor binding by <italic>α</italic>\n₀ then the\nprobability of this occurring in an interval of length\n<italic>δt</italic>, assuming that the transcription factor is not bound\nat the start of the interval, is\n<italic>α</italic>\n₀\n<italic>δt</italic>.\nIntegrating over time, this means that the probability of the event having\nhappened by time <italic>t</italic>, is , whilst the average time for the event to happen is\n1/<italic>α</italic>\n₀. The same holds for the other\ntransitions in the model, with the rate of transcription factor unbinding\ndenoted by <italic>β</italic>\n₁. Whilst the transcription\nfactor is bound, RNA polymerase binds at a rate\n<italic>α</italic>\n₁, and each such binding event is assumed\nto produce one molecule of mRNA. More detailed models that allow the polymerase\nto unbind before it has produced mRNA are considered later and will have no\neffect on our overall conclusions.</p>", "<p>Each mRNA molecule binds to a ribosome at rate\n<italic>α</italic>\n₂ and is degraded at rate\n<italic>β</italic>\n₂. When the last mRNA has decayed no\nmore protein will be produced. We define the number of proteins produced between\nthe transcription factor binding and the last mRNA decaying as a\n“burst”. Note that since a burst begins once the\ntranscription factor has bound, we expect the distribution of burst sizes to be\nindependent of the transcription factor binding rate\n<italic>α</italic>\n₀. This is confirmed by the rigorous\nderivation below.</p>", "<p>Mathematically, the Standard Model of Gene Expression is a continuous time Markov\nchain model. Each particular combination of number of mRNA molecules, number of\nprotein molecules and state of binding of the transcription factor constitutes a\nsingle state of the model. It is possible to derive an (infinite) set of coupled\nordinary differential equations (called the Kolmogorov forward equations or\nmaster equation) that govern the probability at any given time of the system\nbeing in any given state. However, the analysis of a such a complex set of\nequations is difficult. On the other hand, using the same approach as for\nmulti-stage queues, it is relatively easy to derive the distribution of protein\nburst sizes.</p>", "<title>The Component mRNA and Protein Distributions</title>", "<p>We begin with the analysis of McAdams and Arkin ##REF##9023339##[3]## for the\ndistribution of the number of proteins produced by a single mRNA molecule. If a\ncertain number (possibly 0) of protein molecules has been produced, the\nprobability that the next event in which the mRNA molecule participates is the\nproduction of another protein molecule is\n<italic>p</italic> = <italic>α</italic>\n₂/<italic>α</italic>\n₂+<italic>β</italic>\n₂)\n(see ##SUPPL##0##Text\nS1## for derivation). Conversely, the probability that the next event is\nthe degradation of the mRNA molecule is\n1−<italic>p</italic> = <italic>β</italic>\n₂/(<italic>α</italic>\n₂+<italic>β</italic>\n₂).\nIn order to produce precisely <italic>n</italic> molecules of protein, we need\n<italic>n</italic> events of the first type to occur, followed by a final\ndegradation event. The probability of this happening is\n<italic>pⁿ</italic>(1−<italic>p</italic>), giving the\ndistribution <italic>Q</italic>(<italic>n</italic>) of the number of protein\nmolecules produced by a single mRNA molecule\n</p>", "<p>Here\n<italic>A</italic>\n₂ = <italic>α</italic>\n₂/<italic>β</italic>\n₂\nis the expectation of <italic>Q</italic>. Contrasting this with ##REF##9023339##[3]##,\nthe parameter <italic>A</italic>\n₂ defining the distribution is now\nexpressed in terms of physically measurable rate constants. Exactly the same\nargument applies to the distribution of the number of RNA molecules produced\nbetween the successive binding and unbinding of the transcription factor. In\nparticular, the probability of producing one more mRNA molecule before the\ntranscription factor unbinds is\n<italic>α</italic>\n₁/(<italic>α</italic>\n₁+<italic>β</italic>\n₁)\nand the probability of the transcription factor unbinding is\n<italic>β</italic>\n₁/(<italic>α</italic>\n₁+<italic>β</italic>\n₁).\nIn order to produce precisely <italic>m</italic> mRNA molecules before the\ntranscription factor unbinds we need <italic>m</italic> independent production\nevents with probability\n<italic>α</italic>\n₁/(<italic>α</italic>\n₁+<italic>β</italic>\n₁),\nfollowed by the unbinding event with probability\n<italic>β</italic>\n₁/(<italic>α</italic>\n₁+<italic>β</italic>\n₁).</p>", "<p>Thus the probability distribution, <italic>R</italic>(<italic>m</italic>), of the\nnumber of mRNA molecules produced in one burst iswhere\n<italic>A</italic>\n₁ = <italic>α</italic>\n₁/<italic>β</italic>\n₁\nis the expectation of <italic>R</italic>(<italic>m</italic>). In order to derive\nthe overall protein burst size distribution for the Standard Model in ##FIG##0##Figure 1## we need the\nprobability generating functions ##UREF##1##[17]## of the\ndistributions <italic>Q</italic>(<italic>n</italic>) and\n<italic>R</italic>(<italic>m</italic>) which we denote as\n<italic>Q</italic>*(<italic>z</italic>) and\n<italic>R</italic>*(<italic>z</italic>), respectively. These are\nsimply obtained by summing the relevant geometric seriesand\n</p>" ]

[ "<title>Results</title>", "<title>The Compound Protein Burst Size Distribution</title>", "<p>The distribution <italic>P</italic>(<italic>n</italic>) of the total number of\nproteins produced in a single burst is simply the compound distribution of\n<italic>R</italic> and <italic>Q</italic>\n##UREF##1##[17]##.\nThis is easily computed using probability generating functions (see below), and\nis not a geometric distribution. However, it is of relatively little interest\nsince it includes the possibility that the transcription factor unbinds before\nany proteins have been produced (either because no mRNA is produced, or because\nthis mRNA is degraded before binding to a ribosome). Such events cannot be\nobserved in the experimental protocol used in ##REF##16543458##[6]##,##REF##16541077##[7]##, and hence\n<italic>P</italic>(<italic>n</italic>) cannot be directly compared to the\ndata in these papers. However, we can re-scale\n<italic>P</italic>(<italic>n</italic>) to give the probability distribution\n<italic>Pˆ</italic>(<italic>n</italic>) = <italic>P</italic>(<italic>n</italic>)/(1−<italic>P</italic>(0))\nof protein numbers conditional on at least one protein being produced. An\napproximate calculation of this distribution was given in the supplementary\nmaterial of ##REF##16541077##[7]##. This replaced the discrete geometric distribution\n<italic>Q</italic>(<italic>n</italic>) by a continuous exponential\ndistribution of the same mean and then used the Laplace transform to obtain the\n(continuous approximation to the) compound distribution. Here we present an\nexact derivation for the discrete distribution using generating functions (which\nare closely related to the Laplace transform). Furthermore we relate the\nparameter of the final burst size distribution to the original kinetic\nparameters <italic>α</italic>\n₁,\n<italic>α</italic>\n₂,\n<italic>β</italic>\n₁, and\n<italic>β</italic>\n₂.</p>", "<p>Thus, let <italic>X</italic>\n^{(<italic>i</italic>)} be the random\nvariable, with distribution <italic>Q</italic>(<italic>n</italic>), giving the\nnumber of proteins produced by the <italic>i</italic>th mRNA transcript and let\n<italic>Y</italic> be a random variable, with distribution\n<italic>R</italic>(<italic>n</italic>) giving the number of mRNA molecules\nproduced. Then the random variablegives the total number of proteins in a burst. Denote the\ndistribution of <italic>X</italic> by <italic>P</italic>(<italic>n</italic>),\nwith generating function <italic>P</italic>*(<italic>z</italic>). Then a\nstandard result on generating functions of compound distributions ##UREF##1##[17]## givesTo obtain the distribution conditional on at least one protein\nmolecule being produced, we subtract <italic>P</italic>*(0) and\nnormalise (divide) by 1−<italic>P</italic>*(0) to giveThis is the generating function of a conditional geometric\ndistribution with (dimensionless) parameter\n<italic>Â</italic>\n₂ = <italic>A</italic>\n₂(1+<italic>A</italic>\n₁),\nso that <italic>Pˆ</italic>(<italic>n</italic>) has the distributionwhere the parameter <italic>Â</italic>\n₂ can\nbe expressed in terms of the mean number <italic>A</italic>\n₁ of mRNA\nmolecules produced and the mean number <italic>A</italic>\n₂ of protein\nmolecules produced from a single mRNA molecule as\nWe thus see that the burst size distribution is determined by a\nsingle parameter, and that many different combinations of the parameters\n<italic>α</italic>\n₁,\n<italic>α</italic>\n₂, <italic>β</italic>\n₁,\nand <italic>β</italic>\n₂ will lead to the same burst size\ndistribution. In mathematical language this says that the Standard Model with\nparameters <italic>α</italic>\n₁,\n<italic>α</italic>\n₂,\n<italic>β</italic>\n₁, and\n<italic>β</italic>\n₂ is <italic>nonidentifiable</italic> from\nburst size data. In fact we can only estimate a single parameter (or a single\nlinear combination) and the three remaining parameters can be arbitrarily\nchosen.</p>", "<title>Burst Distributions for Extensions of the Standard Model</title>", "<p>It might be hoped that such nonidentifiability is a particular pathology of the\nStandard Model. We thus next consider a number of generalisations of this model,\nwhich provide a more detailed description of the process of gene expression. We\nfind that for a wide range of generalisations we can still derive the burst size\ndistribution in a similar manner the above. It turns out to be geometric in each\ncase and hence all such models are also nonidentifiable.</p>", "<p>One common extension is to include an additional step in the model of the\ntranscription process ##REF##16362909##[13]##, as shown in ##FIG##1##Figure 2##. This accounts for the fact that\nafter the transcription factor has bound, one still requires the RNA polymerase\nto bind to the transcription initiation complex, and this may not always happen\nsuccessfully. A similar modification could be made to the translation loop to\ndescribe the binding of the mRNA transcript to the ribosome in more detail. Both\nof these additions can be considered individually, or in combination.</p>", "<p>Doing this results in distributions <italic>R</italic> and <italic>Q</italic>\nwhich are still geometric, but with the parameters\n<italic>A</italic>\n₁ and <italic>A</italic>\n₂ given by more\ncomplex combinations of the individual rates. We illustrate this for the\ntranscription loop, where we find that in order to produce exactly\n<italic>m</italic> mRNA molecules, the system can pass through state\n<italic>G</italic>* any number\n<italic>i</italic>≥<italic>m</italic> times. On\n<italic>i</italic>−<italic>m</italic> of these occasions the\npolymerase unbinds before an mRNA molecule is produced, returning to\n<italic>G</italic> with rate <italic>δ</italic>\n₁, and on the\nremaining <italic>m</italic> occasions an mRNA molecule is produced, with rate\n<italic>α</italic>\n₁. The <italic>m</italic> productive\nsteps can be interspersed in any order amongst the <italic>i</italic> visits,\ngiving possible choices. The probability of producing\n<italic>m</italic> mRNA molecules is thuswith <italic>A</italic>\n₁ now given by\n<italic>A</italic>\n₁ = <italic>α</italic>\n₁\n<italic>γ</italic>\n₁/<italic>β</italic>\n₁(<italic>α</italic>\n₁+<italic>δ</italic>\n₁).\nA similar derivation holds for the translation loop. We see that carrying out\neither or both of these modifications still results in a geometric distribution\nin the form of Equation 4 for <italic>Pˆ</italic>(<italic>n</italic>),\nwith\n<italic>Â</italic>\n₂ = <italic>A</italic>\n₂(1+<italic>A</italic>\n₁),\nbut <italic>A</italic>\n₁ and <italic>A</italic>\n₂ now given\nby\n<italic>A</italic>\n₁ = <italic>α</italic>\n₁\n<italic>γ</italic>\n₁/<italic>β</italic>\n₁(<italic>α</italic>\n₁+<italic>δ</italic>\n₁)\nand\n<italic>A</italic>\n₂ = <italic>α</italic>\n₂\n<italic>γ</italic>\n₂/<italic>β</italic>\n₂(<italic>α</italic>\n₂+<italic>δ</italic>\n₂).\nAs a consequence the overall conditional protein size distribution,\n<italic>Pˆ</italic>(<italic>n</italic>), will still be given by\nEquation 4, with the parameter\n<italic>Â</italic>\n₂ = <italic>A</italic>\n₂\n<italic>A</italic>\n₁+<italic>A</italic>\n₂\nas before.</p>", "<p>An alternative generalisation is to add additional loops with the same structure\nas the current transcription and translation loops. We prove in the Supporting\nInformation (##SUPPL##0##Text S1##) that if we have <italic>k</italic>−1 such loops,\nthe final conditional protein size distribution <italic>Pˆ</italic>\n<italic>_k</italic>(<italic>n</italic>) will still be geometric.</p>", "<p>We thus conclude that all of these models yield the same geometric protein burst\nsize conditional distribution, determined by a single parameter. In particular,\nmodels which include additional steps to account for DNA–RNAP complex\nformation and mRNA-ribosome complex formation give distributions that are\nmathematically indistinguishable from those from the Standard Model. It is thus\nimpossible to differentiate between these models using experimentally observed\nburst size distributions. Similarly we cannot use such data to differentiate\nbetween the contributions to noisy gene expression from transcriptional versus\ntranslational bursting.</p>", "<title>Comparison with Burst Size Data</title>", "<p>We can compare the probability distribution derived above directly with\nexperimental data. We consider recently published data of burst sizes for two\nfluorescently tagged proteins in the bacterium <italic>Escherichia coli</italic>\n##REF##16543458##[6]##,##REF##16541077##[7]##. In\n##REF##16543458##[6]##, a\nnovel fluorescent imaging technique is used to determine the distribution of\nprotein molecules per transcription factor binding event in live <italic>E.\ncoli</italic> cells. The specific protein studied was a fusion of a yellow\nfluorescent protein variant (Venus) with the membrane protein Tsr. The\n<italic>tsr-venus</italic> gene is incorporated into the <italic>E.\ncoli</italic> chromosome, replacing the <italic>lacZ</italic> gene. This\nmodified gene is then under the control of the <italic>lac</italic> promoter. In\na second publication ##REF##16541077##[7]##, the same group used a different imaging\ntechnique to determine the distribution of protein molecules per transcription\nfactor binding event of <italic>β</italic>-gal in live <italic>E.\ncoli</italic> cells.</p>", "<p>Such experimental data can be compared to the predicted distribution\n<italic>Pˆ</italic>(<italic>n</italic>) in two ways. One\npossibility is to use maximum likelihood estimation to find the value of\n<italic>Â</italic>\n₂ for which\n<italic>Pˆ</italic>(<italic>n</italic>) best fits the data. This\nis illustrated in ##FIG##2##Figure 3##,\nwhich shows that it is possible to obtain excellent agreement between the\ntheoretical and experimental distributions. The estimated value of\n<italic>Â</italic>\n₂ for Tsr-Venus is\n<italic>Â</italic>\n₂ = 3.57,\nwhilst for <italic>β</italic>-gal,\n<italic>Â</italic>\n₂ = 20.96.\nThe difference in magnitude between these two estimates may be partially due to\nthe fact that <italic>β</italic>-gal is only active as a tetramer. Thus,\neach burst of activation measured experimentally (and thus available for\nfitting) corresponds to the production of 4 monomers. The disadvantage of\nfitting the model in this way is it can only provide an estimate of the single\nparameter <italic>Â</italic>\n₂, but not of the underlying\nkinetic parameters <italic>α</italic>\n₁,\n<italic>α</italic>\n₂,\n<italic>β</italic>\n₁, and\n<italic>β</italic>\n₂.</p>", "<p>An alternative approach to verifying the model would be to obtain independent\nestimates of the model parameters from which we can calculate\n<italic>Â</italic>\n₂ using Equation 6. The resulting\ngeometric distribution can then be compared to the observed burst size data.\nUnfortunately, as is common for most models in cell and molecular biology,\ndirect experimental measurements of many of these rates are not available. For\nthe <italic>β</italic>-gal data, <italic>β</italic>\n₂\ncan be obtained from the reported mRNA half life ##REF##16541077##[7]##,##UREF##4##[20]##, but the other\nthree parameters corresponding to the off-rate of the transcription factor and\nto the binding rates of RNA polymerase to DNA and of mRNA to ribosome\nrespectively are not available.</p>", "<title>Application to Experimental Data</title>", "<title>Incorporating steady state distribution data</title>", "<p>We thus conclude that we can neither estimate all the kinetic parameters\n<italic>α</italic>\n₁,\n<italic>α</italic>\n₂,\n<italic>β</italic>\n₁, and\n<italic>β</italic>\n₂ from the burst size data, nor measure\nthem by other means. However, experience suggests that by supplementing the\nburst size distribution with other experimental data it may be possible to\novercome the nonidentifiability of these parameters. This is reinforced by\nthe observation that parameter combinations that lead to the same\n<italic>Â</italic>\n₂ and hence the same burst size\ndistribution can yield quite different steady-state distributions, as shown\nfor example in ##FIG##3##Figure 4##.\nThe two steady state distributions shown have different choices of\n<italic>α</italic>\n₁,\n<italic>α</italic>\n₂,\n<italic>β</italic>\n₁, and\n<italic>β</italic>\n₂ that yield the same value for\n<italic>Â</italic>\n₂, and hence the same burst size\ndistribution. However, the two steady state distributions are clearly\ndifferent. This shows that steady state distribution data should allow us to\ndistinguish between different combinations of parameters with the same\n<italic>Â</italic>\n₂, and hence potentially identify\nsome or all of these parameters.</p>", "<title>Empirical likelihood estimation</title>", "<p>The main difficulty with such an approach is the lack of analytic expressions\nfor the steady state distribution, making it impossible to derive an\nexplicit formula for the likelihood. Instead one has to compute an estimate\nof the equilibrium distribution using simulations of the reaction network\n##UREF##4##[20]## and then use these to derive an empirical\nlikelihood by comparing to the experimental data. This can then be maximized\nin the usual way.</p>", "<p>We applied this approach to the data from ##REF##16541077##[7]##, which presents\nboth burst size and steady state distributions for the same experimental\nsystem. In order to fully specify the steady state distribution, we need two\nadditional parameters: the rate of transcription factor binding\n<italic>α</italic>\n₀ and the rate of protein decay\n<italic>β</italic>\n₃. These do not enter into the\nexpressions for the burst size distribution, and were assumed to be known\n(and fixed) for the simulations shown in ##FIG##3##Figure 4##. In the absence of independent\nestimates of these parameters for the <italic>β</italic>-gal system,\nwe explored the possibility of estimating these from the data in ##REF##16541077##[7]##\ndirectly by computing an empirical likelihood using simulation of the model\n(see below). We attempted both to maximize this empirical likelihood\ndirectly, and to obtain its distribution using Markov Chain Monte Carlo\nsampling. Neither of these approaches were successful with the full six\nparameter model (results not shown).</p>", "<p>We can however, make use of independent estimates of parameters in the model\nto reduce the dimensionality of the parameter space. In effect this\nconstrains the orginal optimization to a lower dimensional sub-space. We\napplied this approach with two different choices of parameter: the mRNA\ndegradation rate <italic>β</italic>\n₂ and the protein\ndegradation rate <italic>β</italic>\n₃.</p>", "<title>Constraining on the mRNA degradation rate</title>", "<p>We chose first to make use of the wide availability of estimates of the value\nof <italic>β</italic>\n₂, the rate of mRNA degradation.\nSince we also have the burst size data, we first estimate\n<italic>Â</italic>\n₂ and then use Equation 6 to\nobtain an expression for <italic>α</italic>\n₂ in terms of\n<italic>α</italic>\n₁,\n<italic>β</italic>\n₁, and\n<italic>β</italic>\n₂. We are left with the four\ndimensional parameter space <italic>α</italic>\n₀,\n<italic>α</italic>\n₁,\n<italic>β</italic>\n₁, and\n<italic>β</italic>\n₃. At each point in this space, we\nsimulate the model using the Gillespie algorithm to given an empirical\nestimate of the probability\n<italic>P_n</italic>(<italic>α</italic>\n₀,<italic>α</italic>\n₁,<italic>β</italic>\n₁,<italic>β</italic>\n₃)\nof observing <italic>n</italic> proteins at equilibrium. This gives the\nempirical log-likelihoodwhere is the number of times that <italic>n</italic> proteins\nare observed in the experimentally data .</p>", "<p>This empirical likelihood function can be maximized using any suitable\noptimization algorithm. Because it is computed using stochastic simulations\nany particular realization of the function is not smooth, making algoritms\nthat use gradient (or Hessian) information unsuitable. We therefore opted to\nuse the Nelder-Mead (simplex) method ##REF##16677373##[21]## with the\nresults shown in ##FIG##4##Figure\n5##.</p>", "<p>We can see that the likelihood has a local maximum at\n<italic>L</italic>≈−1800, and the simplex method frequently\ngets stuck in this region. However the majority of runs (73.4%)\nconverge to the presumed global maximum. The means and standard deviations\nof the estimated parameter values are shown in ##TAB##0##Table 1##. The value obtained for\n<italic>β</italic>\n₃ is 0.0297\ns⁻¹, which corresponds to a half-life of 23 seconds.\nThis appears to be unrealistically short, since\n<italic>β</italic>-galactosidase is a stable protein with a reported\nlifetime of hours.</p>", "<p>One possible explanation of this discrepancy is that the reported mRNA\ndegradation rate\n<italic>β</italic>\n₂ = 7.2×10⁻³\nis always measured in experimental conditions where the gene is active. On\nthe other hand, the burst size and equilibrium distributions in ##REF##16541077##[7]## are\nobtained under conditions where the gene is suppressed. It is possible that\nthe mRNA degradation rates are significantly different in the two cases. To\nexplore this hypothesis, we approached the problem from an alternative\ndirection, fixing the protein degradation rate\n<italic>β</italic>\n₃ to correspond to a half-life of 1\nhour, and estimating the remaining five parameters, including mRNA\ndegradation rate <italic>β</italic>\n₂.</p>", "<title>Constraining on the protein degradation rate</title>", "<p>We therefore fixed <italic>β</italic>\n₃ to\n1.92×10⁻⁴ s⁻¹,\ncorresponding to a protein half-life of one hour, and then used same method\nas described above to estimate the other parameters\n<italic>α</italic>\n₀,\n<italic>α</italic>\n₁,\n<italic>α</italic>\n₂,\n<italic>β</italic>\n₁, and\n<italic>β</italic>\n₂. We ran 10,000 simulations, as a\nrelatively low number of runs converged (23.37%), with the others\nbecoming trapped in a region with physically unrealistic (negative) reaction\nrates, and a log-likelihood of <italic>L</italic>≈−2100. Of\nthe runs which converged, 2057 (88%) converged to a local maximum\nat <italic>L</italic>≈−9150, while 279 (12%)\nconverged to the presumed global maximum at\n<italic>L</italic>≈−1100. The results for the runs which\nconverged can be seen in ##FIG##5##Figure\n6##, whilst summary statistics for the runs which converged to the\npresumed global maximum are presented in ##TAB##1##Table 2##.</p>", "<title>Comparison of the estimates</title>", "<p>The transcription factor binding rate <italic>α</italic>\n₀\nis almost unchanged under both assumptions. When we fix the protein\ndegradation rate in the second set of estimates to a value much lower than\nestimated in the first set we find that the transcription rate (i.e., rate\nof RNA polymerase binding) <italic>α</italic>\n₁, is\napproximately one third of the previous value, decreasing from 0.0017\ns⁻¹ to 0.0006 s⁻¹, whilst the\ntranslation rate, <italic>α</italic>\n₂ shows an\napproximate two-fold increase, from 0.1538 s⁻¹ to\n0.3352 s⁻¹. It is intuitively reasonable that such a\ncombination of decreasing transcription and increasing translation leads to\nthe same overall level of protein expression. The parameter\n<italic>β</italic>\n₁, the rate at which the\ntranscription factor unbinds increases slightly, leading to shorter bursts.\nThe increase in the mRNA degradation rate,\n<italic>β</italic>\n₂ from the original assumption of 0.007\ns⁻¹ to the estimate of 0.0161\ns⁻¹ (corresponding to an mRNA half life of\napproximately 43 seconds) suggests that when expression of the gene is being\nstrongly repressed as in this situation, there may well be active\ndegradation of the mRNA. It would be interesting to experimentally\ninvestigate this biologically significant prediction.</p>" ]

[ "<title>Discussion</title>", "<p>We have shown that it is possible to use results from queuing theory to derive the\nburst size distribution of protein molecules produced by a single transcription\nfactor binding event in terms of physically measurable kinetic rate constants for\nboth the simplest model of gene expression, the so-called Standard Model, and for a\nnumber of natural extensions.</p>", "<p>Furthermore, we have shown that the mathematical form of these models is\nnonidentifiable, and all such burst size distributions are actually determined by a\nsingle parameter. This implies that it is impossible to use burst size data alone to\ndetermine the relative contributions of transcription and translation to the\nvariability in gene expression.</p>", "<p>One possible way of overcoming this limitation is to use a combination of burst size\ndata and steady-state data. However, this requires estimates of a further two\nparameters (which are not needed when using burst-size data alone). We were unable\nto estimate all six parameters directly from the combined data. However, using\nindependent estimates of either the mRNA lifetime or the protein lifetime reduces\nthe number of parameters by one, and enables successfully estimation of the\nremaining five parameters by maximizing an empirical likelihood using the\nNelder-Mead simplex algorithm. Although this suffers from the common problem of\noccasional convergence to a local maximum, by using computing repeated estimates it\nwas possible to identify and exclude such cases and hence obtain good estimates of\nthe desired five kinetic parameters under the different constraints.</p>" ]

[]

[ "<p>Conceived and designed the experiments: PJI MPHS JS. Wrote the paper: PJI JS.\nPerformed the analysis: PJI.</p>", "<p>Over the last few years, experimental data on the fluctuations in gene activity\nbetween individual cells and within the same cell over time have confirmed that\ngene expression is a “noisy” process. This variation is in\npart due to the small number of molecules taking part in some of the key\nreactions that are involved in gene expression. One of the consequences of this\nis that protein production often occurs in bursts, each due to a single promoter\nor transcription factor binding event. Recently, the distribution of the number\nof proteins produced in such bursts has been experimentally measured, offering a\nunique opportunity to study the relative importance of different sources of\nnoise in gene expression. Here, we provide a derivation of the theoretical\nprobability distribution of these bursts for a wide variety of different models\nof gene expression. We show that there is a good fit between our theoretical\ndistribution and that obtained from two different published experimental\ndatasets. We then prove that, irrespective of the details of the model, the\nburst size distribution is always geometric and hence determined by a single\nparameter. Many different combinations of the biochemical rates for the\nconstituent reactions of both transcription and translation will therefore lead\nto the same experimentally observed burst size distribution. It is thus\nimpossible to identify different sources of fluctuations purely from protein\nburst size data or to use such data to estimate all of the model parameters. We\nexplore methods of inferring these values when additional types of experimental\ndata are available.</p>", "<title>Author Summary</title>", "<p>Recent experimental data showing fluctuations in gene activity between individual\ncells and within the same cell over time confirm that gene expression is a\n“noisy” process. This variation is partly due to the small\nnumber of molecules involved in gene expression. One consequence is that protein\nproduction often occurs in bursts, each due to the binding of a single\ntranscription factor. Recently, the distribution of the number of proteins\nproduced in such bursts has been experimentally measured, offering a unique\nopportunity to study the relative importance of different sources of noise in\ngene expression. We derive the theoretical probability distribution of these\nbursts for a wide variety of gene expression models. We show a good fit between\nour theoretical distribution and experimental data and prove that, irrespective\nof the model details, the burst size distribution always has the same shape,\ndetermined by a single parameter. As different combinations of the reaction\nrates lead to the same observed distribution, it is impossible to estimate all\nkinetic parameters from protein burst size data. When additional data, such as\nprotein equilibrium distributions, are available, these can be used to infer\nadditional parameters. We present one approach to this, demonstrating its\napplication to published data.</p>" ]

[ "<title>Supporting Information</title>" ]

[]

[ "<fig id=\"pcbi-1000192-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pcbi.1000192.g001</object-id><label>Figure 1</label><caption><title>The standard gene expression model.</title><p>An inactive sequence of DNA and a transcription factor bind to produce an\nactive gene <italic>G</italic>. This produces mRNA, denoted by\n<italic>M</italic> at a rate <italic>α</italic>\n₁, and\nin turn the mRNA produces protein at rate\n<italic>α</italic>\n₂. Eventually, the\ntranscription factor will unbind (at rate\n<italic>β</italic>\n₁), and the gene will become\ninactive again. Each copy of mRNA produced will also be degraded (at\nrate <italic>β</italic>\n₂).</p></caption></fig>", "<fig id=\"pcbi-1000192-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pcbi.1000192.g002</object-id><label>Figure 2</label><caption><title>Diagram of the generalised situation in which intermediate,\nreversible stages are introduced.</title><p>Here, G represents an active gene, G* an active gene with a bound\nRNA polymerase, M an mRNA molecule, M* an mRNA molecule bound to\na ribosome, P a protein, and <italic>S</italic>\n₀ states which\ncorrespond to transcription factor unbinding and mRNA transcript\ndecay.</p></caption></fig>", "<fig id=\"pcbi-1000192-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pcbi.1000192.g003</object-id><label>Figure 3</label><caption><title>Comparison of the distribution of experimentally measured burst sizes\nfor the proteins <italic>Tsr-Venus</italic> (A) ##REF##16543458##[6]## and for the\n<italic>β-gal</italic> (B) ##REF##16541077##[7]## with the\nstandard model of gene expression.</title><p>In both cases the blue line shows the best fit of the model to the data,\nobtained using the method of maximum likelihood giving\n<italic>Â</italic>\n₂ = 3.57\nfor <italic>Tsr-Venus</italic> and\n<italic>Â</italic>\n₂ = 20.96\nfor <italic>β-gal</italic>. The error bars show the upper and\nlower bounds of the 95% confidence interval for the fitted\nparameter.</p></caption></fig>", "<fig id=\"pcbi-1000192-g004\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pcbi.1000192.g004</object-id><label>Figure 4</label><caption><title>Simulations of steady-state protein expression levels.</title><p>For A we have\n<italic>α</italic>\n₁ = 0.018\nand\n<italic>β</italic>\n₁ = 0.086\nand for B we have\n<italic>α</italic>\n₁ = 0.009\nand\n<italic>β</italic>\n₁ = 0.043,\nresulting in the same <italic>Â</italic>\n₂ and\nhence identical burst size distributions. Other parameters were\n<italic>α</italic>\n₀ = 0.012,\n<italic>α</italic>\n₂ = 0.013,\n<italic>β</italic>\n₂ = 0.0039,\nand\n<italic>β</italic>\n₃ = 0.0007,\nbased on previous simulation studies ##REF##11062240##[22]##. The\ndistributions shown are for a run of 10,000 seconds using the Stocks\nimplementation of Gillespie's method ##REF##4604615##[23]##, after an initial transient of 10,000\nseconds. Previous studies have indicated that the steady state is in\nfact attained in under 1000 seconds.</p></caption></fig>", "<fig id=\"pcbi-1000192-g005\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pcbi.1000192.g005</object-id><label>Figure 5</label><caption><title>Parameter estimation results with fixed mRNA degradation rate.</title><p>The results of 1000 runs of the Nelder-Mead maximisation of the\nlog-likelihood for the parameters\n<italic>α</italic>\n₀,\n<italic>α</italic>\n₁,\n<italic>β</italic>\n₁, and\n<italic>β</italic>\n₃, with\n<italic>α</italic>\n₂ determined by the\nrelationship in Equation 6, and with the mRNA degradation rate\n<italic>β</italic>\n₂ set to\n7.2×10⁻³, corresponding to a\nhalf-life for <italic>β</italic>-gal mRNA of 1.6 mins ##UREF##5##[24]##. The panels in column A show the\nestimates of the values of the parameters and the percentage of\ntimes the Nelder-Mead algorithm converged to those values. The\npanels in column B are scattergrams of the values of the parameter\nestimates against the value of the log-likelihood. Each simulation\nis run 10,000 times to simulate a population of 10,000 cells, and\neach simulation is run for 5000 reaction steps. The starting values\nfor the optimisation routine are:\n<italic>α</italic>\n₀ = 0.01\ns⁻¹,\n<italic>α</italic>\n₁ = 0.02\ns⁻¹,\n<italic>β</italic>\n₁ = 0.1\ns⁻¹, and\n<italic>β</italic>\n₃ = 0.0007\ns⁻¹, and are based on previous simulation\nstudies ##REF##15721042##[16]##.</p></caption></fig>", "<fig id=\"pcbi-1000192-g006\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pcbi.1000192.g006</object-id><label>Figure 6</label><caption><title>Parameter estimation results with fixed protein degradation rate.</title><p>The results of 10,000 runs of the Nelder-Mead maximisation of the\nlog-likelihood for the parameters\n<italic>α</italic>\n₀,\n<italic>α</italic>\n₁,\n<italic>β</italic>\n₁, and\n<italic>β</italic>\n₂, with\n<italic>α</italic>\n₂ determined by the\nrelationship in Equation 6, and with the protein degradation rate\nset, <italic>β</italic>\n₃ set to\n2.77×10⁻⁴, consistent with a\nhalf-life for <italic>β</italic>-gal 60 mins. The panels in\ncolumn A show the estimates of the values of the parameters and the\npercentage of times the Nelder-Mead algorithm converged to those\nvalues. The panels in column B are scattergrams of the values of the\nparameter estimates against the value of the log-likelihood. Each\nsimulation is run 10,000 times to simulate a population of 10,000\ncells, and each simulation is run for 5000 reaction steps. The\nstarting values for the optimisation routine are:\n<italic>α</italic>\n₀ = 0.01\ns⁻¹,\n<italic>α</italic>\n₁ = 0.02\ns⁻¹,\n<italic>β</italic>\n₁ = 0.1\ns⁻¹, and\n<italic>β</italic>\n₂ = 0.007\ns⁻¹, and are based on previous simulation\nstudies ##REF##15721042##[16]##.</p></caption></fig>" ]

[ "<table-wrap id=\"pcbi-1000192-t001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pcbi.1000192.t001</object-id><label>Table 1</label><caption><title>Means and standard deviations of estimates of the other\nparameters when the mRNA degradation rate\n<italic>β</italic>\n₂ set to\n7.2×10⁻³.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>α</italic>\n₀\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>α</italic>\n₁\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>α</italic>\n₂\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>β</italic>\n₁\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>β</italic>\n₃\n</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>μ</italic>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0049</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0017</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.1538</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.1210\n(<italic>τ</italic>\n_1/2 = 5.7\ns)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0297\n(<italic>τ</italic>\n_1/2 = 23.3\ns)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>σ</italic>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0052</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0003</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0088</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0098</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0098</td></tr></tbody></table></alternatives></table-wrap>", "<table-wrap id=\"pcbi-1000192-t002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pcbi.1000192.t002</object-id><label>Table 2</label><caption><title>Means and standard deviations of the parameter estimates, when\nthe protein degradation rate <italic>β</italic>\n₃\nset to 1.92×10⁻⁴\n(<italic>τ</italic>\n_1/2 = 3600\ns).</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>α</italic>\n₀\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>α</italic>\n₁\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>α</italic>\n₂\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>β</italic>\n₁\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>β</italic>\n₂\n</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>μ</italic>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0048</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0006</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.3352</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.1314\n(<italic>τ</italic>\n_1/2 = 5.3\ns)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0161\n(<italic>τ</italic>\n_1/2 = 43.1\ns)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<italic>σ</italic>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0033</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0004</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.1476</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0078</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">0.0071</td></tr></tbody></table></alternatives></table-wrap>" ]

[ "<inline-formula></inline-formula>", "<disp-formula><label>(1)</label></disp-formula>", "<disp-formula><label>(2)</label></disp-formula>", "<disp-formula></disp-formula>", "<disp-formula></disp-formula>", "<disp-formula></disp-formula>", "<disp-formula><label>(3)</label></disp-formula>", "<disp-formula></disp-formula>", "<disp-formula><label>(4)</label></disp-formula>", "<disp-formula><label>(5)</label></disp-formula>", "<disp-formula><label>(6)</label></disp-formula>", "<inline-formula></inline-formula>", "<disp-formula></disp-formula>", "<disp-formula></disp-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>" ]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"pcbi.1000192.s001\"><label>Text S1</label><caption><p>Derivation of probabilities.</p><p>(0.10 MB PDF)</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><fn id=\"nt101\"><p>These statistics are based on those runs that approached the\nglobal maximum (73.4% of all runs). These were\nselected by imposing a threshold at\n<italic>L</italic> = −1350\nand only considering those runs converging to a larger (more\npositive) likelihood. All reaction rates have units of\ns⁻¹.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"nt102\"><p>These statistics are based on those runs which approached the\nglobal maximum (12% of converging runs). These were\nselected by imposing a threshold at\n<italic>L</italic> = −2500\nand only considering those runs converging to a larger (more\npositive) likelihood. All reaction rates have units of\ns⁻¹.</p></fn></table-wrap-foot>", "<fn-group><fn fn-type=\"COI-statement\"><p>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p>PJI and MPHS acknowledge financial support from the Wellcome Trust. MPHS is\ngrateful to the Carlsberg Foundation and the Royal Society, UK, for their\ngenerous support. MPHS receives further support through an EMBO Young\nInvestigator Award. JS is supported by the UK Biotechnology and Biological\nSciences Research Council via the Centre for Integrative Systems Biology at\nImperial College, BB/C519670/1.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"pcbi.1000192.g001\"/>", "<inline-graphic xlink:href=\"pcbi.1000192.e001.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pcbi.1000192.e002.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pcbi.1000192.e003.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pcbi.1000192.e004.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pcbi.1000192.e005.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pcbi.1000192.e006.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pcbi.1000192.e007.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pcbi.1000192.e008.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pcbi.1000192.e009.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pcbi.1000192.e010.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pcbi.1000192.e011.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pcbi.1000192.g002\"/>", "<inline-graphic xlink:href=\"pcbi.1000192.e012.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pcbi.1000192.e013.jpg\" mimetype=\"image\" position=\"float\"/>", "<graphic xlink:href=\"pcbi.1000192.g003\"/>", "<graphic xlink:href=\"pcbi.1000192.g004\"/>", "<graphic xlink:href=\"pcbi.1000192.e014.jpg\" mimetype=\"image\" position=\"float\"/>", "<inline-graphic xlink:href=\"pcbi.1000192.e015.jpg\" mimetype=\"image\"/>", "<inline-graphic xlink:href=\"pcbi.1000192.e016.jpg\" mimetype=\"image\"/>", "<graphic xlink:href=\"pcbi.1000192.g005\"/>", "<graphic id=\"pcbi-1000192-t001-1\" xlink:href=\"pcbi.1000192.t001\"/>", "<graphic xlink:href=\"pcbi.1000192.g006\"/>", "<graphic id=\"pcbi-1000192-t002-2\" xlink:href=\"pcbi.1000192.t002\"/>" ]

[ "<media xlink:href=\"pcbi.1000192.s001.pdf\"><caption><p>Click here for additional data file.</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>The neuropilin-1 (np1) and the neuropilin-2 (np2) receptors were originally characterized as receptors for axon guidance factors of the class-3 semaphorin (sema3) family ##REF##11796239##[1]##. It was subsequently realized that neuropilins are also expressed by endothelial cells and by many types of cancer cells ##REF##18580951##[2]##. Neuropilins function in addition as receptors for several angiogenic factors including heparin binding forms of VEGF and hepatocyte growth factor/scatter factor (HGF/SF) and enhance their pro-angiogenic activity ##REF##8621410##[3]##–##REF##18065694##[6]##. These studies indicate that neuropilins may be targets for anti-angiogenic therapy. Indeed, antibodies directed against np1 and np2 were recently found to inhibit tumor progression ##REF##17222790##[7]##, ##REF##18394556##[8]##.</p>", "<p>Most of the sema3s, with the exception of sema3E which binds to Plexin-D1 (plexD1) ##REF##15550623##[9]##, bind to one of the two neuropilins or to both. Neuropilins form functional semaphorin receptors by associating with members of the plexin receptor family in which neuropilins bind semaphorins and the plexins function as the signal transducing elements ##REF##10520994##[10]##, ##REF##10520995##[11]##. The four type-A plexins as well as plexD1 were found to participate in neuropilin mediated signal transduction ##REF##10520994##[10]##–##REF##15157527##[12]##. The semaphorins sema3B and sema3F were characterized as tumor suppressor genes indicating that additional semaphorins may also possess anti-tumorigenic properties ##REF##11717452##[13]##, ##REF##11980661##[14]##. The identification of neuropilins in endothelial cells suggested that class-3 semaphorins may also regulate angiogenesis. Indeed, the np2 agonist sema3F functions as a repellant of endothelial cells. It also induces apoptosis of endothelial cells, and inhibits tumor angiogenesis and tumor progression ##REF##14871832##[15]##, ##REF##15520858##[16]##. The np1 agonist sema3A also inhibits angiogenesis ##REF##18180379##[17]## but it is as yet unknown whether it can inhibit tumor angiogenesis and tumor progression. Likewise, the plexD1 agonist sema3E was found to inhibit the invasion of blood vessels into somites during embryonic development ##REF##15550623##[9]## suggesting that sema3E too may function as an anti-angiogenic agent.</p>", "<p>The expression of neuropilins and plexins by many types of tumor cells indicates that semaphorins may be able to affect tumor cells directly. Indeed, sema3F and sema3B have been found to inhibit the adhesion, migration and proliferation of several types of lung cancer derived tumor cells ##REF##11717452##[13]##, ##REF##11980661##[14]##, ##REF##15520858##[16]##, ##REF##15802023##[18]##. It follows that semaphorins such as sema3F probably inhibit angiogenesis and tumor cell proliferation simultaneously and may also affect in addition the behavior of additional types of stromal cells. However, it is unclear which of these mechanisms is the primary mechanism used by semaphorins such as sema3F to inhibit tumor development. It is also unclear whether additional sema3s possess anti-angiogenic and anti-tumorigenic properties. We report that four additional class-3 semaphorins which have not yet been found to possess anti-tumorigenic properties, sema3A, sema3D, sema3E, and sema3G possess anti-tumorigenic properties. Furthermore, all these semaphorins with the exception of sema3E strongly reduce the density of blood vessels in tumors. However, we find that inhibition of tumor development by class-3 semaphorins is strongly correlated with the expression of appropriate semaphorin receptors by the tumor cells and that there is a much poorer correlation between their ability to inhibit angiogenesis and their effects on tumor development.</p>" ]

[ "<title>Materials and Methods</title>", "<title>Materials</title>", "<p>Antibodies against β-actin and myc and FLAG epitope tags, as well as chemicals were from Sigma (St. Louis, MI). Media and sera for cell culture were from Biological-Industries Inc. (Kibbutz Beth-Haemek, Israel). Fugene-6 was from Roche Ltd (Switzerland). Antibodies against np1 and np2 were purchased from Santa-Cruz inc. (San-Diego, CA). The cDNAs encoding different semaphorins were subcloned into the NSPI-CMV-MCS-myc-His lentiviral expression vector containing SV40 promoter driving Puromycin selection marker. This vector was kindly given to us by Dr. Aaronson (Mount Sinai Hospital, NY). Partial cDNAs encoding sema3E were kindly given to us by Dr. Claus Christensen (Institute of cancer biology, Copenhagen, Denmark). Antibodies against CD-31 were from BD biosciences Pharmingen. The cDNA's encoding sema3F and sema3A were donated by Dr. Mark Tessier-Lavigne (University of California, San Francisco, CA) and by Drs. David Ginty and Alex Kolodkin (Johns Hopkins University, Baltimore, MD). The PerfectPure RNA reverse PCR kit was from 5-Prime (Gaithersburg, MD).</p>", "<title>Primers</title>", "<p>The following specific primers were used to follow the expression of different plexins and semaphorins in the cancer cell lines. Plexin-A1: <named-content content-type=\"gene\">5′-ctgctggtcatcgtggctgtgct 5′-gggcccttctccatctgctgcttga</named-content>. Plexin-A2: <named-content content-type=\"gene\">5′-gtgcccaccaactgtgcctgtcctg 5′-tcagcgatgatgtattcccctggga</named-content>. Plexin-A3: <named-content content-type=\"gene\">5′-tcttgctctcgaggttcttct 5′-acatgccaagtgatcaacgac</named-content>. Plexin-A4: <named-content content-type=\"gene\">5′-acggtccatcccaacaatatc 5′-ccacgccagcaaccttgacat</named-content>. Plexin-D1: <named-content content-type=\"gene\">5′-gtccatctaccagggcttct 5′-ctggatgtaggactcggtga</named-content>. Sema3A: <named-content content-type=\"gene\">5′-aacgggggcttttcatcc 5′-cccttctcacatcactcatgct</named-content>. Sema3D: <named-content content-type=\"gene\">5′-ggctgctgaggatcgaaggac 5′-atgtgtgtggaactggagca</named-content>. Sema3E: <named-content content-type=\"gene\">5′-gggttacttactggagctttgg 5′-gtcatgctcagtgcggatatg</named-content>. Sema3F: <named-content content-type=\"gene\">5′-gtgctgcccaaggatgacca 5′-cttgttggcattggagttgaacc</named-content>. Sema3G: 5′-aacgcagagctggccgagga 5′-ccggacccacctgcta. Actin: <named-content content-type=\"gene\">5′-tgacggggtcacccacactgtgcccatcta 5′-ctagaagcattgcggtggacgatggaggg</named-content>.</p>", "<title>Expression plasmids</title>", "<p>All the class-3 semaphorin cDNAs were sub-cloned into the NSPI-CMV-myc-his lentiviral expression vector. The sema3G cDNA was cloned from HUVEC mRNA using RT-PCR. The sema3D cDNA was cloned using RT-PCR from HUVEC cells treated with 30 ng/ml of VEGF for 6 hours. cDNA's containing the myc epitope tag were added in frame upstream to the stop codon of sema3D, sema3E, sema3F and sema3G. A FLAG epitope tag was added upstream to the stop codon of sema3A as described ##REF##17569671##[19]##.</p>", "<title>Generation of recombinant lentiviruses and letiviral mediated infection of cells</title>", "<p>HEK293-T cells were seeded in 100 mm tissue culture dishes (2.5×10⁶ cells/dish). A day after seeding, the cells were co-transfected with the appropriate lentiviral expression plasmid (8µg), with the packaging vector pCMVdR8.91 (5 µg), and with a plasmid encoding the vesicular stomatitis virus coat envelope pMD2-VSVG (2 µg) using Fugene-6 according to the instructions of the vendor. Conditioned medium containing infective lentiviral particles was collected 48 hours and 72 hours post transfection. Following addition of polybrene (8 µg/ml) to the conditioned medium it was incubated 8 hours with target cells.</p>", "<title>Cell lines</title>", "<p>Mycoplasma free MDA-MB-231, MDA-MB-435, MDA-MB-468 and MCF7 cancer cells were obtained from the ATCC. The cells were cultured in DMEM containing 4.5 mg/ml glucose supplemented with 10% FCS and antibiotics. HUVEC, PAE, HEK293 and HEK293-T cells were cultured as previously described ##REF##14871832##[15]##. HUVEC were used between passages 3–7.</p>", "<title>Animal experiments</title>", "<p>All the animal experiments were approved by the institutional committee for animal studies according to the NIH guidelines (license IL-095-10-2007).</p>", "<title>\n<italic>In-vivo</italic> tumor formation assays</title>", "<p>Cells expressing semaphorins or control cells infected with empty lentiviral vectors were implanted (5×10⁶/mouse) into the mammary fat pads of 4–6 week old balb\\c nu/nu female mice (Harlan laboratories). In most experiments we used groups of 9 animals/experiment. The tumors were measured twice a week using a caliper. The tumor volume (V) was determined using the formula, V = 0.52×A²×B in which A is the short diameter and B the long. When MDA-MB-231 tumors reached an average volume of 200–300 mm³, they were excised and weighted. Each experiment was repeated at least twice. Estrogen pellets were used in experiments in which the development of tumors from MCF-7 cells was determined as previously described ##REF##12670920##[20]##.</p>", "<title>Immunohistochemistry</title>", "<p>Tumors were embedded in OCT and frozen in 2-methylbutane cooled by liquid nitrogen. They were then sectioned into 30 µm thick sections using a cryostat. Sections were blocked with cold acetone, reacted with an antibody directed against the endothelial marker CD-31, counterstained with hematoxilin and photographed. Eight different microscopic fields derived from different sections of three different tumors were photographed. These photographs were taken from areas in which the density of blood vessels was highest (hot spot method) ##REF##9059336##[21]##, ##REF##17502947##[22]##. The area of the blood vessels in fields of equal area was quantified using the Image Pro Plus software.</p>", "<title>Western Blots</title>", "<p>Cell lysates were prepared and the concentration of protein determined as previously described ##REF##17569671##[19]##. To determine the concentration of secreted sema3s in conditioned mediums of the various cell lines, cells were seeded in 12 well dishes at a concentration of 2×10⁵ cells/well. The cells were incubated for 48 hours in 0.4 ml of serum-free medium. Aliquots of equal volume were examined using western blot analysis for the presence of sema3s using antibodies directed against the appropriate tags as previously described ##REF##17569671##[19]## and the densitometry analysis was preformed using MultiGauge software (FUJIFILM). The expressed semaphorins did not affect the proliferation rate or the survival of the different semaphorin producing cells (data not shown).</p>", "<title>Proliferation assays</title>", "<p>Tumor cells (10⁴ cells/well) were seeded in triplicate in 24 well dishes. Adherent cells were trypsinized and counted every 24 hours for 4 days, using a coulter counter.</p>", "<title>Adhesion assays</title>", "<p>In cell adhesion experiments we used uncoated 12 well cell culture dishes as well as non-adhesive 12 well dishes coated with fibronectin (5 µl/ml). Tumor cells (10⁵ cells/well) were seeded in triplicates in growth media. The cells were washed twice with PBS, trypsinized to release adherent cells, and counted with a coulter counter. The cells were counted 5, 10, 20 and 45 minutes after they were seeded. The percentage of adherent cells relative to the number of seeded cells was then calculated and plotted. The time required for the adherence of 50% of the seeded cells was used as a measure to compare the adhesive properties of control and semaphorin expressing cells.</p>", "<title>Endothelial cells repulsion assay</title>", "<p>Cell repulsion assays were performed essentially as previously described ##REF##17569671##[19]##.</p>", "<title>Soft-agar colony formation assay</title>", "<p>A layer of agar containing 2 ml of 0.5% low melting agar (Bio-Rad) dissolved in growth media was poured into wells of a 6 well cell culture dish and allowed to set at 4°C for 20 minutes. A second layer (1ml) containing 0.3% of low melting agar dissolved in growth media containing cells (3×10³ cells/ml) was placed on top of the first layer and allowed to set at 4°C for 20 minutes. Growth medium (2 ml) was added on top of the second layer and the cells were incubated in a humidified incubator at 37°C for 21 days. Medium was changed twice a week. At the end of the experiment, colonies were stained for 1 hr with 0.005% crystal violet, and incubated with PBS overnight to remove excess crystal violet. The colonies were photographed and colonies with at least one diameter of 150 µm within photographic fields were chosen. The Image-pro morphometric software was then used to measure the area of each of these colonies. Their average area and statistics were then performed using the Microsoft excel software.</p>", "<title>Statistical analysis</title>", "<p>Statistical analysis was performed using the unpaired data with unequal variance student's T-test. Error bars represent the standard error of the mean. Statistical significance is presented in the following manner: *p&lt;0.05, **p&lt;0.01 and ***p&lt;0.001.</p>" ]

[ "<title>Results</title>", "<title>Expression patterns of class-3 semaphorins and class-3 semaphorin receptors in different tumorigenic cell lines</title>", "<p>Semaphorins may affect the development of tumors by directly affecting the behavior of tumor cells or indirectly by affecting angiogenesis or the behavior of stromal cells. To find out if the class-3 semaphorins sema3A, sema3D, sema3E, sema3F and sema3G can inhibit the formation of tumors from cancer cells by directly influencing tumor cell behavior, we first determined the expression patterns of known sema3 receptors in different cancer cell lines. We found that MDA-MB-231 breast cancer derived cells express predominantly np1 ##REF##9529250##[4]##, a receptor for sema3A and sema3D, but very little np2 if at all. MDA-MB-435 melanoma cells express predominantly np2, a receptor for sema3F and sema3G and very little if any np1 (##FIG##0##Fig. 1A##). MCF-7 breast cancer cells express np1 but not np2. The concentration of np1 in MCF-7 cells is about three folds lower as compared to MDA-MB-231 cells (##FIG##0##Fig. 1, A–B##). MDA-MB-468 breast cancer cells differ from the other cell lines since they do not express neuropilins (##FIG##0##Fig. 1A##).</p>", "<p>Because of their short intracellular domains neuropilins do not transduce sema3 signals independently but form complexes with plexins in which the plexins serve as the signal transducing elements ##REF##10520994##[10]##, ##REF##10520995##[11]##. Good antibodies to plexins are not yet readily available so we compared the expression of mRNA encoding various plexins known to transduce class-3 semaphorins signals qualitatively using RT-PCR as an indication for possible protein expression. All four cell lines expressed the plexA1 mRNA and all but the MDA-MB-468 cells also expressed the plexA2 mRNA. None of these cell lines expressed the plexA4 mRNA and only the MCF-7 cells expressed the plexA3 mRNA (##FIG##0##Fig. 1, B–C##). The mRNA encoding the sema3E receptor PlexD1 was expressed in MDA-MB-231 and MCF-7 cells while MDA-MB-435 cells seem to express a lower concentration of PlexD1 mRNA (##FIG##0##Fig. 1C##) and MDA-MB-468 cells did not express plexD1 mRNA at all (##FIG##0##Fig. 1, C–D##). The expression of the neuropilins and of the mRNA encoding the sema3E receptor PlexD1 in the various tumor cells was not altered significantly as a result of the expression of the various recombinant semaphorins ( ##SUPPL##0##Figs. S1## and ##SUPPL##1##S2##).</p>", "<title>The effects of different sema3s on the development of tumors from cancer cells</title>", "<p>To find out if sema3A, sema3D, sema3E, sema3F or sema3G over-expression can affect the development of tumors from different tumorigenic cell lines, we expressed the full length cDNAs encoding the five semaphorins or an empty control vector in the tumor cells using a lentiviral expression vector that confers resistance to puromycin. Pools of infected cells were examined for semaphorin expression following puromycin selection using antibodies directed against epitope tags incorporated into the recombinant semaphorins. The expression levels of the recombinant semaphorins seemed to differ in correlation with the type of the recombinant semaphorin and much less so in correlation with the cell type in which they were expressed. Thus, the concentration of recombinant sema3D found in the conditioned medium of either MDA-MB-231 breast cancer cells or in MDA-MB-435 melanoma cells was significantly lower than the concentrations of sema3F or sema3G (##FIG##1##Fig. 2, A–B##). It was not possible to effectively compare the concentrations of the endogenous semaphorins produced by the tumor cells with the concentrations of the recombinant semaphorins expressed in each of the cell types due to the lack of suitable highly specific antibodies directed against the various semaphorins. However, from reverse transcription followed by PCR (RT-PCR) experiments it is clear that the various tumor cells we studied also express various endogenous mRNA's encoding class-3 semaphorins suggesting that these cells may produce combinations of endogenous semaphorins. Thus, MDA-MB-435 cells express mRNA encoding sema3D while MDA-MB-231 cells express sema3A and sema3E mRNA while MCF-7 and MDA-MB-468 cells express mRNA encoding sema3F (##FIG##1##Fig. 2C##).</p>", "<p>Sema3s contain conserved cleavage sites for furin like pro-protein convertases##REF##9321387##[23]##. In the case of sema3E the cleaved product was reported to possess pro-metastatic properties ##REF##16024618##[24]##. However, the degree of cleavage of the recombinant semaphorins produced by MDA-MB-231 cells or by MDA-MB-435 cells did not exceed 15% of the total amount of semaphorin found in the conditioned medium and in the case of sema3E was almost undetectable (data not shown). The cells were subsequently implanted in mammary fat pads of immune deficient mice, and allowed to form tumors. All these semaphorins were relatively efficiently expressed in MDA-MB-231 cells although there were substantial differences in the expression levels obtained with different semaphorins (##FIG##1##Fig. 2A##). Expression of the np1 agonist sema3A ##REF##9288754##[25]## inhibited almost completely the development of tumors from these cells (##FIG##2##Fig. 3, A–B## and ##SUPPL##2##Fig. S3##, A). Sema3D, an agonist for both np1 and np2 ##REF##15456815##[26]##, inhibited tumor formation completely in one experiment (data not shown) and in another experiment inhibited strongly though not completely tumor development even though it was not as highly expressed as the other semaphorins (##FIG##2##Fig. 3, E–F##). In contrast, the np2 agonist sema3G ##REF##15882579##[27]## was unable to inhibit tumor development from these cells, although it was highly expressed as compared to sema3D (##FIG##2##Fig. 3, G–H##). The np2 agonist sema3F on the other hand, inhibited significantly the development of tumors despite the lack of np2 receptors in MDA-MB-231 cells (##FIG##2##Fig. 3, A–B##). The tumors that developed from sema3F expressing MDA-MB-231 cells (##FIG##2##Fig. 3, A–B##) looked less bloody than the control tumors suggesting that sema3F may inhibit tumor angiogenesis (##SUPPL##2##Fig. S3##, A). Expression of the PlexD1 agonist sema3E ##REF##15550623##[9]## also inhibited significantly the development of tumors from these cells but the resulting tumors did not look starved of blood vessels (##FIG##2##Fig. 3, C–D##, and ##SUPPL##2##Fig. S3##, B).</p>", "<p>A different picture emerged when the effects of these sema3s on the development of tumors from MDA-MB-435 cells were examined. Control cells developed into small tumors that slowed when they reached an average volume of 50–100 mm³ (##FIG##2##Fig. 3, A–B##). Expression of the np2 agonists sema3F and sema3G strongly inhibited the development of tumors from these cells (##FIG##3##Fig. 4, A–D##). In contrast, expression of sema3A did not inhibit tumor development (##FIG##3##Fig. 4, A–B##), while sema3D which binds to both neuropilins ##REF##15456815##[26]##, significantly inhibited tumor development from the MDA-MB-435 cells though less potently than sema3G (##FIG##3##Fig. 4, C–D##) which may be due to the lower expression levels obtained with sema3D in these cells (##FIG##1##Fig. 2B##). MDA-MB-435 cells also express the PlexD1 mRNA, although at lower levels than MDA-MB-231 cells (##FIG##0##Fig. 1C##). Expression of sema3E did not inhibit the formation of tumors from the MDA-MB-435 cells (##FIG##3##Fig. 4 C–D##). This was probably not due to cleavage by furin like pro-protein convertases since less than 5% of the sema3E found in the conditioned medium of these cells was cleaved (data not shown).</p>", "<p>We also determined whether expression of sema3A and sema3F, the best studied np1 and np2 agonists respectively, inhibits tumor development from the non-metastatic, estrogen dependent, np1 expressing MCF-7 cells. Expression of sema3A inhibited significantly tumor development while expression of sema3F did not (##FIG##4##Fig. 5, B–C##). Taken together, these results suggested that the sema3s ability to inhibit tumor formation from a given cancer cell type depends primarily on the identity of the semaphorin receptors expressed by the tumor cells, suggesting that sema3s should not be able to inhibit the formation of tumors from cancer cells that do not express sema3 receptors (##TAB##0##Table 1##).</p>", "<p>To put this prediction to the test we expressed sema3A and sema3F in MDA-MB-468 breast cancer cells, which do not express np1, np2 or PlexD1 (##FIG##0##Fig. 1D##). These cells form slowly growing tumors in mammary fat pads of nu/nu balb/c mice. In agreement with our prediction, neither the expression of recombinant sema3A nor expression of sema3F significantly inhibited the formation of tumors from these cells (##FIG##4##Fig. 5, E–F##).</p>", "<title>The effects of different sema3s on tumor angiogenesis</title>", "<p>Sema3F was characterized in several studies as an inhibitor of tumor angiogenesis and as a repulsive factor for endothelial cells ##REF##14871832##[15]##, ##REF##15520858##[16]##, ##REF##17308083##[28]## and sema3A was also found to function as an inhibitor of VEGF induced angiogenesis and as a repulsive factor for endothelial cells although not as an inhibitor of tumor angiogenesis ##REF##18180379##[17]##, ##REF##17569671##[19]##, ##REF##12879061##[29]##, ##REF##12618135##[30]##. To compare the repulsive properties of different sema3s we seeded HEK293 cells secreting different semaphorins on top of monolayers of human umbilical vein endothelial cells (HUVEC) at clonal densities. Control cells infected with empty vector did not repel endothelial cells while sema3A, sema3D and sema3E expressing cells repelled endothelial cells efficiently (##FIG##5##Fig. 6A##). However, the np2 agonists sema3F and in particular sema3G repelled HUVEC much less potently than the np1 agonists or the PlexD1 agonist sema3E, possibly because these cells contain about 3 fold less np2 as compared to np1 ##REF##10748121##[5]## (data not shown). We therefore seeded HEK293 cells expressing either sema3F or sema3G on top of porcine aortic endothelial (PAE) cells engineered to co-express recombinant np2 and plexA1 ##REF##14871832##[15]##. As expected, these cells were repelled very strongly by sema3F. Surprisingly they were repelled much less potently by sema3G suggesting that plexA1 may not be able to transduce sema3G signals (##FIG##5##Fig. 6B##).</p>", "<p>To find out if the various sema3s inhibit tumor angiogenesis, we determined the concentration of blood vessels in tumors that developed from control or from sema3s expressing cells. Since sema3A inhibited tumor formation in MDA-MB-231 cell almost completely we could not determine the concentration of blood vessels in this case. However, expression of sema3D in MDA-MB-231 cells resulted in the formation of tumors containing a 40% lower density of blood vessels as compared to tumors that developed from control cells (##FIG##5##Fig. 6C##). The reduction in the concentration of tumor associated blood vessels was not correlated with the types of semaphorin receptors expressed by the cancer cells since expression of the np2 agonists sema3F and sema3G also reduced the concentration of blood vessels in tumors derived from MDA-MB-231 cells by about 40% (##FIG##5##Fig. 6C##) even though sema3G did not inhibit the development of tumors from these cells (##FIG##2##Fig. 3, G–H##). In-contrast, even though sema3E expression in MDA-MB-231 cells significantly inhibited the development of tumors (##FIG##2##Fig. 3D##) and even though sema3E expressing cells repulse HUVEC efficiently (##FIG##5##Fig. 6A##) the expression of sema3E in these cells did not reduce significantly the concentration of blood vessels in resulting tumors (##FIG##5##Fig. 6C##).</p>", "<p>We also examined the effects of sema3A and sema3F expression on the concentration of tumor associated blood vessels in MCF-7 cells. These tumors develop in the mammary fat pads of the mice only in the presence of slow estrogen release pellets. Expression of sema3A in these cells significantly reduced the concentration of tumor associated blood vessels. However, expression of sema3F did not (##FIG##5##Fig. 6D##). The lack of inhibition in this case may perhaps be explained by findings suggesting that estrogen is an inhibitor of np2 expression ##REF##11159832##[31]##.</p>", "<p>In the case of tumors that develop from np2 expressing MDA-MB-435 cells, we found that the expression of sema3A and sema3D reduced the concentration of blood vessels in resulting tumors by 65% (##FIG##5##Fig. 6E##) even though tumor development from these cells was not inhibited at all by sema3A (##FIG##2##Fig. 3B##). It was not possible to determine the blood vessel concentration in tumors that developed from cells expressing sema3F or sema3G since the resulting tumors were too small or non-existent as in the case of sema3G. Expression of sema3E in MDA-MB-435 cells produced a small but significant 28% decrease in the concentration of blood vessels in resulting tumors even though tumor formation from these cells was not inhibited by sema3E (##FIG##3##Fig. 4D##).</p>", "<p>Taken together, these experiments indicate that although most of the semaphorins are able to inhibit tumor angiogenesis as manifested by the reduction in the concentration of blood vessels in tumors in response to the expression of recombinant semaphorins. However, even though the inhibition may contribute to the inhibition of tumor progression, there was generally no correlation between the effects of the sema3s on tumor angiogenesis and their effect on tumor development (##TAB##0##Table 1##).</p>", "<title>The effects of different sema3s on the behavior of the tumor cells in-vitro</title>", "<p>The experiments described in the previous sections suggest that semaphorin expression should modulate the behavior of tumor cells. Indeed, sema3s such as sema3F and sema3B were reported to inhibit the adhesion, spreading and proliferation of various types of tumor cells ##REF##11717452##[13]##, ##REF##15520858##[16]##, ##REF##15802023##[18]##. However, the proliferation of the tumor cell types which we have used in the present study was not inhibited as a result of the expression of the various semaphorins when the tumor cells were grown in tissue culture dishes (data not shown). We also examined the effect of the expression of the different semaphorins on the adhesion of the various tumor cells to plastic or to fibronectin. However, none of these semaphorins affected the adhesion of the tumor cells regardless of whether the substrate was plastic or fibronectin (data not shown).</p>", "<p>The ability to form colonies in soft-agar is a hallmark that differentiates many types of cancer cells from their normal counterparts ##REF##3553356##[32]##–##REF##12242156##[34]##. We therefore determined if the expression of different class-3 semaphorins in MDA-MB-231 or MDA-MB-435 cells affects their ability to form colonies in soft-agar. None of the semaphorins inhibited completely the formation of colonies by MDA-MB-231 cells. However, the expression of sema3A and sema3D, semaphorins that strongly inhibited tumor formation from these cells (##FIG##2##Fig. 3##), also inhibited significantly the formation of large colonies in soft agar (##FIG##6##Fig. 7, A–B##). Surprisingly, expression of sema3F also inhibited significantly the formation of large colonies in soft agar despite the absence of np2 receptors on these cells. However, the expression levels of sema3F were the highest of all the semaphorins we examined (##FIG##1##Fig. 2A##) and sema3F was also able to inhibit the formation of tumors from these cells (##FIG##2##Fig. 3B##). Sema3F binds to np1, albeit with a 10 fold lower affinity as compared to its binding affinity to np2 ##REF##9331348##[35]## and there is one additional report suggesting that it may also utilize np1 for signal transduction ##REF##12659673##[36]##. It is therefore possible that this inhibitory effect is mediated by np1. Another np2 agonist, sema3G, which in contrast with sema3F does not inhibit the development of tumors from MDA-MB-231 cells (##FIG##2##Fig. 3H##) and does not bind to np1 ##REF##16098142##[37]##, had no effect on the development of colonies from these cells (##FIG##6##Fig. 7A##). MDA-MB-231 cells also express the sema3E receptor PlexD1 and expression of sema3E inhibits the formation of tumors from these cells (##FIG##2##Fig. 3D##). However, sema3E failed to inhibit the formation of colonies from MDA-MB-231 cells (##FIG##6##Fig. 7, A–B##).</p>", "<p>We also determined whether sema3D, sema3F and sema3G, which inhibit tumor formation from MDA-MB-435 cells (##FIG##3##Fig. 4##), also inhibit the anchorage free growth of these cells. Both sema3D and sema3G inhibited colony formation efficiently as expected. However, to our surprise we found that sema3F did not, even though it did inhibit almost completely the formation of tumors from these cells (##FIG##6##Fig. 7, C–D##). Another unexpected observation was that sema3E, which did not inhibit the formation of tumors from these cells was able to inhibit colony formation (##FIG##6##Fig. 7, C–D##). Lastly, we expected that sema3A will not affect colony formation since its receptor is not expressed by MDA-MB-435 cells (##FIG##0##Fig. 1##). Surprisingly, we found that not only was colony formation not inhibited but it was even significantly enhanced (##FIG##6##Fig. 7, C–D##). Taken together our results suggest that despite a number of exceptions, in most cases inhibition of tumor growth by sema3s is correlated with their ability to inhibit the formation of soft agar colonies from the tumor cells (##TAB##0##Table 1##).</p>", "<title>Enhancement of tumor development from MDA-MB-435 cells by np1 expression is inhibited by co-expression of sema3A</title>", "<p>The experiments described above suggest that the expression of specific sema3s receptors by tumor cells is probably the most important factor that determines whether a given sema3 will function as an inhibitor of tumor development. To test this hypothesis further we expressed np1 in MDA-MB-435 cells in order to determine whether this would render tumors that develop from these cells sensitive to sema3A. The tumors that developed from MDA-MB-435 cells expressing np1 grew very rapidly to a much larger size than tumors derived from empty vector infected MDA-MB-435 cells following an initial lag (##FIG##7##Fig. 8, A–C##). Interestingly, the density of blood vessels within these tumors was not significantly different from that of control tumors (##FIG##7##Fig. 8D##). When the np1 agonist sema3A was co-expressed in these cells along with np1, the cells that expressed both genes reverted to the behavior exhibited by the control cells and formed slowly developing tumors thereby eliminating the growth advantage conferred by the presence of np1 (##FIG##7##Fig. 8, A–C##), but not that conferred by the presence of np2 which can be further inhibited by np2 agonists such as sema3F or sema3G (##FIG##3##Fig. 4##). Interestingly, the density of blood vessels in tumors that developed from MDA-MB-435 cells expressing sema3A or sema3A/np1 was similar and about 50% lower than the concentration of blood vessels in tumors that developed from control cells (##FIG##7##Fig. 8D##). These results also suggest independently that inhibition of angiogenesis may represent part of the mechanism by which semaphorins modulate tumor progression, but that it may not always be sufficient to inhibit tumor growth.</p>" ]

[ "<title>Discussion</title>", "<p>The identification of sema3B and sema3F as tumor suppressors ##REF##11717452##[13]##, ##REF##11980661##[14]##, ##REF##11809707##[38]##, and the identification of sema3F and sema3A as inhibitors of angiogenesis ##REF##14871832##[15]##, ##REF##18391935##[39]##, suggested that additional sema3s may also possess anti-tumorigenic and anti-angiogenic properties. We show here for the first time that sema3A, sema3D, sema3E and sema3G display anti-tumorigenic properties. Furthermore, we show that all of these class-3 semaphorins can also inhibit tumor angiogenesis.</p>", "<p>Many tumorigenic cell lines, including the cell lines we used, express different combinations of neuropilins and plexins ##REF##16731741##[40]##. Neuropilins as well as several types of plexins are also expressed in endothelial cells ##REF##17569671##[19]## and in some bone marrow derived cells ##REF##18483621##[41]## which are frequently recruited into the tumor microenvironment. Complexes formed between the neuropilins and several types of plexins mediate sema3s induced signal transduction with the exception of sema3E which signal through PlexD1 independently of neuropilins ##REF##18580951##[2]##. Our initial experiments suggested that the expression of given semaphorins in tumor cells sometimes inhibited and sometimes did not inhibit the development of tumors from different types of tumor cells.</p>", "<p>To better understand these seemingly conflicting results we compared the effects of the expression of several semaphorins on the development of tumors from tumorigenic cell lines differing in their expression pattern of semaphorin receptors so as to find out if we can identify a property that can predict whether a given semaphorin will be able to inhibit the development of tumors from a given type of tumor cell. The expression of the different recombinant semaphorins in the tumor cells did not change significantly the expression of the primary semaphorin binding receptors np1, np2 and PlexD1 in the tumor cells. Therefore, observed differences in responses to the expression of different semaphorins in the tumor cells were not due to semaphorin induced changes in the expression of direct semaphorin receptors. Semaphorin signaling through these receptors may also be modulated by changes in the expression levels of endogenous semaphorins and by changes in the expression and activity of neuropilin associated receptors such as additional types of plexins and adhesion receptors known to modulate semaphorin signal transduction ##REF##18580951##[2]## which are present on tumor cells as well as on tumor associated stromal cells such as endothelial cells. We have not tested these parameters systematically due to the lack of appropriate specific antibodies directed against these different proteins and because of the large volume of assays required to monitor such changes systematically.</p>", "<p>Regardless of these possible modulating influences, we have found that the expression of a semaphorin receptor able to bind the specific recombinant semaphorin we expressed in the tumor cells was the property that correlated best with the ability to successfully inhibit tumor development (##TAB##0##Table 1##). There were only two cases in which this correlation did not successfully predict whether a specific semaphorin will be able to inhibit tumor development from a given type of tumor cell. In the first such example Sema3E was not able to inhibit tumor development from MDA-MB-435 cells. However, the expression levels of the mRNA encoding the sema3E receptor plexD1 in the MDA-MB-435 cells are lower than their levels in MDA-MB-231 cells which may perhaps account for the discrepancy. The second example was the successful inhibition by sema3F of tumor formation from MDA-MB-231 cells which express np1 but not np2 and thus should not have been inhibited by sema3F. However, the concentration of np1 in these cells is relatively high ##REF##9529250##[4]##. Sema3F is known to bind to np1 with a 10 fold lower affinity as compared to np2. Even though sema3F is usually viewed a pure np2 agonist there is nevertheless some evidence suggesting that sema3F may be able to transduce signals using np1 ##REF##12659673##[36]##. It is thus possible that the inhibition of tumor development from MDA-MB-231 cells by sema3F is mediated by np1 and that it is augmented by the relatively high expression level obtained with sema3F and by the anti-angiogenic effect displayed by sema3F in tumors derived from this cell type. The lack of an anti-tumorigenic effect of sema3F in tumors developing from sema3F expressing MCF-7 cells which also express np1 but no np2 may be explained by the lower concentration of np1 receptors in these cells and by the absence of a sema3F induced anti-angiogenic effect which is probably the result of np2 down regulation in endothelial cells of blood vessels due to the effects of prolonged estrogen administration ##REF##11159832##[31]##.</p>", "<p>In order to determine whether there is a correlation between the anti-angiogenic activity of specific semaphorins and their ability to inhibit tumor development we measured the effects of the expression of several semaphorins in several tumor cell types on the concentration to tumor associated blood vessels. We have presented here for the first time evidence indicating that sema3D, sema3G sema3E and sema3A can significantly reduce the concentration of microvessels in tumors that develop from tumor cells that express these semaphorins. Surprisingly, we found that reduction in the concentration of tumor associated blood vessels was frequently not correlated with the anti-tumorigenic effect of given semaphorins. For example, even though sema3G and sema3A expression did not inhibit at all the development of tumors from MDA-MB-231 and MDA-MB-435 cells respectively, they nevertheless strongly reduced the concentration of tumor associated blood vessels. These observations indicates that inhibition of tumor angiogenesis by the sema3s we examined was probably not sufficiently effective to affect tumor growth. Anti-VEGF antibodies were reported to reduce the concentration of blood vessels in MDA-MB-231 derived tumors by as much as 70% ##REF##17597103##[42]##, ##REF##12549858##[43]## while individual sema3s reduced blood vessel densities in such tumors by up to 40%. It is of course rather difficult to compare two different studies in which the methods used to evaluate blood vessel density were not identical. Nevertheless, it is likely that the anti-angiogenic effects of individual sema3s were not sufficiently strong so as to enable inhibition of tumor development in the case of the cancer cells that we examined. It is however likely that the anti-angiogenic effects of the semaphorins will assume more importance in the case of rapidly growing tumors that may be more dependent on efficient angiogenesis than slowly growing tumors. We have previously shown that combinations of sema3s that interact with different semaphorin receptors can inhibit the proliferation of endothelial cells more effectively than individual sema3s ##REF##17569671##[19]##. Our results suggest that combinations of such sema3s may perhaps be able to increase the anti-angiogenic effects to the point at which they may affect tumor development more effectively.</p>", "<p>Another parameter we have examined as a possible predictor for the effectiveness of class-3 semaphorins as anti-tumorigenic agents was effects of class-3 semaphorin expression in tumor cells on the behavior of the tumor cells <italic>in-vitro</italic>. Contrary to previous reports which observed sema3F and sema3A induced changes in adhesion of tumor cells to fibronectin coated dishes ##REF##15520858##[16]##, ##REF##15802023##[18]##, ##REF##17390026##[44]##, we could not see any effects of any of the semaphorins we tested on the adhesion to fibronectin of any of the tumor cells used here. We also could not detect any effects of semaphorin expression in the various tumor cells on the proliferation of the tumor cells in regular 2D cell culture. This observation contrasts with the strong effects observed when using cultured endothelial cells ##REF##17569671##[19]##. However, some of the semaphorins we used inhibited the growth of colonies of tumor cells in soft agar. The correlation between the ability to inhibit the growth of colonies in soft-agar and the anti-tumorigenic effects of the semaphorins was better than the correlation with the anti-angiogenic effects of the semaphorins, but it was nevertheless a less reliable predictor for the effectiveness of the semaphorins as anti-tumorigenic agents as compared with the presence of appropriate semaphorin receptors in the tumor cells. For example, sema3F did not inhibit the formation of colonies from MDA-MB-435 cells despite the presence of np2 receptors in these cells even though it inhibited strongly tumor formation. In contrast, sema3E inhibited strongly the formation of soft-agar colonies from MDA-MB-435 cells but did not inhibit the formation of tumors. Nevertheless, in most cases the ability to inhibit the formation of colonies in soft agar was correlated with the ability to inhibit the development of tumors <italic>in-vivo</italic>.</p>", "<p>The presence of appropriate signaling semaphorin receptors on tumor cells does not necessarily imply that the anti-tumorigenic effects observed are due to direct effects on the tumor cells, even though in some cases that may be the case. There is some evidence indicating that neuropilins may be able to associate with receptors present on adjacent cells “in-trans” ##REF##11948691##[45]## and it is also possible that the final outcome <italic>in-vivo</italic> will depend on the effects of secondary effectors synthesized in response to semaphorins in responsive tumor cells, resulting in different responses <italic>in-vivo</italic> as compared to <italic>in-vitro</italic> experiments in which the only cell type is the tumorigenic cell. An example for such a modulator of semaphorin function is provided by the furin like pro-protein convertases, which are strongly up-regulated in cancer cells ##REF##16167351##[46]##. The furins cleave class-3 semaphorins at a conserved site and the cleavage results in inactive products in the case of sema3A and sema3B ##REF##9321387##[23]##, ##REF##18757406##[47]##. In the case of sema3E the cleavage generates a pro-metastatic product that affects primarily endothelial cells rather than tumor cells ##REF##16024618##[24]## and is thus an example for an effect that will be seen only <italic>in-vivo</italic> but will not affect <italic>in-vitro</italic> assays such as the soft-agar colony formation assay. It should be noted that in our experiments the maximal amount of cleavage by pro-protein convertases never exceeded 15% of the total amount of sema3s found in the conditioned medium of producing tumor cells, and in the case of sema3E the cleavage was almost undetectable in both MDA-MB-231 and MDA-MB-435 cells, suggesting that the inhibitory effects that we observed are due to the effects of full length sema3E.</p>", "<p>In conclusion, we have found for the first time that the semaphorins sema3A, sema3D, sema3E and sema3G possess anti-tumorigenic and anti-angiogenic properties similar to those displayed by the previously identified tumor suppressor sema3F. However, the anti-angiogenic effects are probably not sufficiently potent so as to enable inhibition of tumor development. The anti-tumorigenic effect of sema3s seems to be associated with the expression of appropriate sema3s receptors by the tumor cells although it is not clear if all the anti-tumorigenic effects are due to direct effects on the tumor cells. Our results argue that for maximal effectiveness, the selection of specific semaphorins or semaphorin combinations will have to take into account the identity of the semaphorin receptors expressed by the tumorigenic cells within target tumors.</p>" ]

[]

[ "<p>Conceived and designed the experiments: BK AV GN. Performed the experiments: BK. Analyzed the data: BK AV OK GN. Contributed reagents/materials/analysis tools: BK OK GN. Wrote the paper: BK OK GN.</p>", "<p>The class-3 semaphorins (sema3s) include seven family members. Six of them bind to neuropilin-1 (np1) or neuropilin-2 (np2) receptors or to both, while the seventh, sema3E, binds to the plexin-D1 receptor. Sema3B and sema3F were previously characterized as tumor suppressors and as inhibitors of tumor angiogenesis. To determine if additional class-3 semaphorins such as sema3A, sema3D, sema3E and sema3G possess anti-angiogenic and anti-tumorigenic properties, we expressed the recombinant full length semaphorins in four different tumorigenic cell lines expressing different combinations of class-3 semaphorin receptors. We show for the first time that sema3A, sema3D, sema3E and sema3G can function as potent anti-tumorigenic agents. All the semaphorins we examined were also able to reduce the concentration of tumor associated blood vessels although the potencies of the anti-angiogenic effects varied depending on the tumor cell type. Surprisingly, there was little correlation between the ability to inhibit tumor angiogenesis and their anti-tumorigenic activity. None of the semaphorins inhibited the adhesion of the tumor cells to plastic or fibronectin nor did they modulate the proliferation of tumor cells cultured in cell culture dishes. However, various semaphorins were able to inhibit the formation of soft agar colonies from tumor cells expressing appropriate semaphorin receptors, although in this case too the inhibitory effect was not always correlated with the anti-tumorigenic effect. In contrast, the anti-tumorigenic effect of each of the semaphorins correlated very well with tumor cell expression of specific signal transducing receptors for particular semaphorins. This correlation was not broken even in cases in which the tumor cells expressed significant concentrations of endogenous semaphorins. Our results suggest that combinations of different class-3 semaphorins may be more effective than single semaphorins in cases in which tumor cells express more than one type of semaphorin receptors.</p>" ]

[ "<title>Supporting Information</title>" ]

[ "<p>We thank Dr. Gal Akiri and Dr. Stuart Aaronson (Mount Sinai School of Medicine, New York, NY) for their gift of the NSPI-CMV-myc-his lentiviral expression vector.</p>" ]

[ "<fig id=\"pone-0003287-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003287.g001</object-id><label>Figure 1</label><caption><title>Expression of semaphorin receptors in different tumorigenic cell lines.</title><p>\n<underline>(A)</underline> Cells were grown to 80% confluence and lysed. Equal amounts of protein were loaded and separated on SDS/PAGE gels and subsequently blotted on nitrocellulose filters. Western blot analysis of np1 and np2 was performed as described in <xref ref-type=\"sec\" rid=\"s2\">materials and methods</xref>. <underline>(B)</underline> Densitometric analysis of three independent experiments showing the relative expression levels of np1 in MDA-MB-231 and MCF-7 cells was preformed using the MultiGauge software. The average expression level of np1 in the MCF-7 cells was taken as 100% and the average expression level of np1 in the MDA-MB-231 cells was compared to the average expression level in MCF-7 cells. <underline>(C, D)</underline> Reverse PCR analysis of the expression of mRNA's encoding plexA1-A4 and plexD1 was performed according to the instruction of the PerfectPure kit using primer pairs specific to the different plexins as described.</p></caption></fig>", "<fig id=\"pone-0003287-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003287.g002</object-id><label>Figure 2</label><caption><title>Determination of the relative concentrations of recombinant class-3 semaphorins secreted into the conditioned medium of MDA-MB-231 and MDA-MB-435 cells.</title><p>Lentiviruses containing the full length cDNAs encoding five semaphorins and a puromycin resistance cassette or empty control lentiviruses were used to infect MDA-MB-231 and MDA-MB-435 cells. Sema3A has a flag epitope tag, whereas the rest of the semaphorins were labeled with a myc epitope tag. <underline>(A, B)</underline> Western blot analysis of equal aliquots of conditioned medium derived from equal numbers of MDA-MB-231 and MDA-MB-435 cells expressing the different sema3s. The expression levels of all the myc tagged semaphorins was quantified as described in <xref ref-type=\"sec\" rid=\"s2\">materials and methods</xref>. <underline>(C)</underline> Reverse PCR analysis of endogenous mRNA's encoding sema3A, sema3D, sema3E, sema3F and sema3G expression was performed according to the instruction of the PerfectPure kit using primer pairs specific to the different semaphorins as described. MDA-MB-231 cells over-expressing the different recombinant semaphorins were used as positive controls.</p></caption></fig>", "<fig id=\"pone-0003287-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003287.g003</object-id><label>Figure 3</label><caption><title>The effect of the expression of different sema3s on the development of tumors from MDA-MB-231 cells.</title><p>MDA-MB-231 cells infected with control lentivirus or infected with lentiviruses directing expression of five different semaphorins were implanted in the mammary fat pads of balb\\c nu/nu mice as described. <underline>(A, C, E, G)</underline> The average volume of the developing tumors was measured as a function of time after implantation as described in <xref ref-type=\"sec\" rid=\"s2\">materials and methods</xref>. <underline>(B, D, E, F)</underline> The average weight of the tumors at the end of the experiment was determined as described in <xref ref-type=\"sec\" rid=\"s2\">materials and methods</xref>.</p></caption></fig>", "<fig id=\"pone-0003287-g004\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003287.g004</object-id><label>Figure 4</label><caption><title>The effect of the expression of different sema3s on the development of tumors from MDA-MB-435 cells.</title><p>MDA-MB-435 cells infected with control lentivirus or infected with lentiviruses directing expression of five different semaphorins were implanted in the mammary fat pads of balb\\c nu/nu mice as described. <underline>(A, C)</underline> The average volume of the developing tumors was measured as described in <xref ref-type=\"sec\" rid=\"s2\">materials and methods</xref>. <underline>(B, D)</underline> The average weight of the tumors at the end of the experiment was determined as described in <xref ref-type=\"sec\" rid=\"s2\">materials and methods</xref>.</p></caption></fig>", "<fig id=\"pone-0003287-g005\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003287.g005</object-id><label>Figure 5</label><caption><title>The effect of the expression of different sema3s on the development of tumors from MCF-7 and MDA-MB-468 cells.</title><p>MCF-7 and MDA-MB-468 cells infected with control lentivirus or infected with lentiviruses directing expression of five different semaphorins were implanted in the mammary fat pads of balb\\c nu/nu mice as described. <underline>(A, D)</underline> Western blot analysis of aliquots of conditioned medium derived from equal numbers of cells infected with empty lentiviruses (C) or with different sema3s (s3X) as indicated. <underline>(B, E)</underline> The average volume of the developing tumors was measured as described in <xref ref-type=\"sec\" rid=\"s2\">materials and methods</xref>. <underline>(C, F)</underline> The average weight of the tumors at the end of the experiment was determined as described in <xref ref-type=\"sec\" rid=\"s2\">materials and methods</xref>.</p></caption></fig>", "<fig id=\"pone-0003287-g006\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003287.g006</object-id><label>Figure 6</label><caption><title>Different sema3s repel endothelial cells <italic>in-vitro</italic> and reduce the density of tumor associated blood vessels <italic>in-vivo</italic>.</title><p>\n<underline>(A)</underline> Control HEK293 cells infected with an empty lentiviral vector or HEK293 cells expressing sema3A, sema3D or sema3E were seeded on top of a monolayer of HUVEC cells as described in <xref ref-type=\"sec\" rid=\"s2\">materials and methods</xref>. The HEK293 cells were labeled with the fluorescent vital dye DIasp prior to seeding. Shown are composite pictures taken by phase and fluorescent microscopy. <underline>(B)</underline> Control HEK293 cells infected with an empty lentiviral vector or HEK293 cells expressing sema3F or sema3G were seeded on a monolayer of PAE cells expressing np2 and plexA1 as described in <xref ref-type=\"sec\" rid=\"s2\">materials and methods</xref>. The HEK293 cells were stained with DIasp and photographed as described in <xref ref-type=\"sec\" rid=\"s2\">material and methods</xref>\n<underline>(C)</underline> The average area of blood vessels per microscopic field was determined in sections derived from tumors that developed from MDA-MB-231 cells infected with empty lentiviruses or from MDA-MB-231 cells expressing different sema3s as described in <xref ref-type=\"sec\" rid=\"s2\">materials and methods</xref>. Since the tumors that did develop from sema3A expressing cells were extremely small (##FIG##2##Fig. 3##) we could not determine the density of blood vessels in them. <underline>(D)</underline> The average area of blood vessels per microscopic field was determined in tumors derived from control MCF-7 cells or from MCF-7 cells expressing sema3A or sema3F as described in <xref ref-type=\"sec\" rid=\"s2\">material and methods</xref>. <underline>(E)</underline> The average area of blood vessels per microscopic field was determined in tumors that developed from control MDA-MB-435 cells or from MDA-MB-435 cells expressing different sema3s as described in <xref ref-type=\"sec\" rid=\"s2\">material and methods</xref>. No tumors developed from sema3G and sema3F expressing MDA-MB-435 cells (##FIG##3##Fig. 4##).</p></caption></fig>", "<fig id=\"pone-0003287-g007\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003287.g007</object-id><label>Figure 7</label><caption><title>Different sema3s inhibit the formation of soft agar colonies from MDA-MB-231 or MDA-MB-435 cells.</title><p>\n<underline>(A)</underline> Single cell suspensions of control MDA-MB-231 cells or MDA-MB-231 cells expressing different sema3s were seeded in soft agar as described in <xref ref-type=\"sec\" rid=\"s2\">materials and methods</xref>. Colonies were allowed to form for 21 days. The colonies were then stained with crystal violet and microscopic fields photographed. The average colony area of colonies with a diameter exceeding 150 µm was then determined as described in <xref ref-type=\"sec\" rid=\"s2\">materials and methods</xref>. <underline>(B)</underline> Photographs of representative microscopic fields containing crystal violet stained colonies that developed in soft agar from control MDA-MB-231 cells or from MDA-MB-231 cells expressing the indicated sema3s. <underline>(C)</underline> The formation of colonies in soft agar from control MDA-MB-435 cells or from MDA-MB-435 cells expressing the indicated sema3s was determined as described in <xref ref-type=\"sec\" rid=\"s2\">materials and methods</xref>. <underline>(D)</underline> Photographs of representative microscopic fields containing crystal violet stained colonies that developed in soft agar from control MDA-MB-435 cells or from sema3s expressing MDA-MB-435 cells.</p></caption></fig>", "<fig id=\"pone-0003287-g008\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003287.g008</object-id><label>Figure 8</label><caption><title>Expression of np1 in MDA-MB-435 cells enhances the growth of resulting tumors and sema3A abrogate the enhancing effect.</title><p>\n<underline>(A)</underline> A western blot comparing the expression of np1 in MDA-MB-435 cells infected with lentivirus directing expression of np1, sema3A or an empty vector is shown at the top. The average volumes of the tumors that developed following the implantation of these cells in mammary fat pads of nude mice as a function of time after implantation is shown in the lower part. <underline>(B)</underline> Photographs of tumors excised at the end of the experiment. <underline>(C)</underline> The average weight of the tumors at the end of the experiment was determined as described in <xref ref-type=\"sec\" rid=\"s2\">materials and methods</xref>. <underline>(D)</underline> The average area of blood vessels/microscopic field in tumor sections was determined as described in <xref ref-type=\"sec\" rid=\"s2\">materials and methods</xref>.</p></caption></fig>" ]

[ "<table-wrap id=\"pone-0003287-t001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003287.t001</object-id><label>Table 1</label><caption><title>Summary of the results.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Cell line</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Expression of recombinant semaphorin</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Known binding receptors</td><td colspan=\"3\" align=\"left\" rowspan=\"1\">Expression of semaphorin receptors in tumor cells</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Inhibition of tumor development</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Inhibition of angiogenesis</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Inhibition of soft agar colony formation</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">NP1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NP2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">PlexD1<xref ref-type=\"table-fn\" rid=\"nt103\">**</xref>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">MDA-MB-231</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">S3A</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NP1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">+++</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">+++</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">N.D.<xref ref-type=\"table-fn\" rid=\"nt102\">*</xref>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">S3D</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NP1, NP2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">S3E</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">PlexD1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NO</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NO</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">S3F</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NP1, NP2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<xref ref-type=\"table-fn\" rid=\"nt104\">***</xref>✓</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">S3G</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NP2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">NO</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NO</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">MDA-MB-435</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">S3A</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NP1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">+++</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">+</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NO</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NO</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">S3D</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NP1, NP2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">S3E</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">PlexD1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">NO</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">S3F</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NP1, NP2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">N.D.<xref ref-type=\"table-fn\" rid=\"nt102\">*</xref>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NO</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">S3G</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NP2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">N.D.<xref ref-type=\"table-fn\" rid=\"nt102\">*</xref>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">MCF-7</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">S3A</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NP1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">++</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">+++</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">✓</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">N.D.</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">S3F</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NP1, NP2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">NO</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NO</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">N.D.</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">MDA-MB-468</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">S3A</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NP1</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">NO</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NO</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">N.D.</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">S3F</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NP1, NP2</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">NO</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">NO</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">N.D.</td></tr></tbody></table></alternatives></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"pone.0003287.s001\"><label>Figure S1</label><caption><p>The endogenous expression levels of NP-1, NP-2 and PlexD1 in MDA-MB-231 and MDA-MB-435 cells infected with lentiviruses directing the expression of different class-3 semaphorins. Cell lysates were prepared from MDA-MB-231 (panel A) and MDA-MB-435 (panel-B) cells infected with lentiviruses directing the expression of the indicated class-3 semaphorins or an empty lentiviral expression vector. The expression levels of NP-1 (Aa, Ba) and NP-2 (Ab, Bb) in the respective cell types were compared using western blot analysis as described in <xref ref-type=\"sec\" rid=\"s2\">materials and methods</xref>. The expression of Plex-D1 (Ac, Bc) was detected by RT-PCR as described in ##FIG##0##Fig. 1##.</p><p>(6.76 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003287.s002\"><label>Figure S2</label><caption><p>The endogenous expression of NP-1, NP-2 and Plexin-D1 in MCF-7 and MDA-MB-468 infected with lentiviruses directing expression of sema3A or sema3F. Cell lysates were prepared from MCF-7 (panel-A) or MDA-MB-468 (panel-B) cells infected with control lentiviruses or lentiviruses directing expression of sema3A or sema3F. The expression of NP-1 (Aa, Ba) and NP-2 (Ab, Ba) was detected using western blot analysis as described in <xref ref-type=\"sec\" rid=\"s2\">materials and methods</xref>. The expression levels of the Plex-D1 mRNA in the two cell types (Ac, Bb) was detected by RT-PCR as described in <xref ref-type=\"sec\" rid=\"s2\">materials and methods</xref>.</p><p>(4.79 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003287.s003\"><label>Figure S3</label><caption><p>Photographs of excised tumors derived from MDA-MB-231 cells expressing sema3A, sema3F, sema3E and an empty expression vector. Control MDA-MB-231 cells infected with empty lentiviruses (C) or MDA-MB-231 cells expressing recombinant sema3A (S3A) sema3F (S3F) or sema3E (S3E) were implanted in the mammary fat pads of balb\\c nu/nu mice as described. At the end of the experiment tumors were excised and photographed.</p><p>(4.93 MB TIF)</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><fn id=\"nt101\"><p>The effects of the expression of the semaphorins in the four tumor cell lines on tumor development, tumor angiogenesis and on the anchorage free growth of the cells are summarized. The relative expression levels of the relevant semaphorin binding receptors in each of the cell lines are shown as well (High level expression: +++, Low level expression: +). N.D., not determined.</p></fn><fn id=\"nt102\"><label>*</label><p>MDA-MB-231 derived tumors expressing sema3A and tumors derived from MDA-MB-435 expressing sema3F and sema3G did not develop and it was therefore not possible to measure effects on tumor angiogenesis.</p></fn><fn id=\"nt103\"><label>**</label><p>Estimation of the relative expression levels of plexD1 is based on estimation of mRNA levels.</p></fn><fn id=\"nt104\"><label>***</label><p>Sema3F binds to np1 with a 10 fold lower affinity as compared to its affinity for np2 but it is unclear whether np1 can transduce sema3F signals.</p></fn></table-wrap-foot>", "<fn-group><fn fn-type=\"COI-statement\"><p><bold>Competing Interests: </bold>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p><bold>Funding: </bold>This work was supported by grants from the Israel Science Foundation (ISF), by the Komen breast cancer foundation, by the International Union against Cancer (AICR), and by the by the Rappaport Family Institute for Research in the Medical Sciences of the Faculty of Medicine at the Technion, Israel Institute of Technology (to G. N.). The funding organizations did not take part or affect the design or the conduct of this study.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"pone.0003287.g001\"/>", "<graphic xlink:href=\"pone.0003287.g002\"/>", "<graphic xlink:href=\"pone.0003287.g003\"/>", "<graphic xlink:href=\"pone.0003287.g004\"/>", "<graphic xlink:href=\"pone.0003287.g005\"/>", "<graphic id=\"pone-0003287-t001-1\" xlink:href=\"pone.0003287.t001\"/>", "<graphic xlink:href=\"pone.0003287.g006\"/>", "<graphic xlink:href=\"pone.0003287.g007\"/>", "<graphic xlink:href=\"pone.0003287.g008\"/>" ]

[ "<media xlink:href=\"pone.0003287.s001.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003287.s002.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003287.s003.tif\"><caption><p>Click here for additional data file.</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Natural killer (NK) cells express a variety of inhibitory receptors that recognize MHC class I molecules and block NK cell–mediated cytotoxicity ##REF##15771571##[1]##, ##REF##18439809##[2]##. In human NK cells, these receptors include the Killer cell Ig-like Receptor (KIR) family, the Leukocyte Immunoglobulin-like Receptor (LILR) family, NKR-P1, and the family of CD94/NKG2 lectin-like receptors. Phosphorylated immunoreceptor tyrosine–based inhibition motifs (ITIM) in the cytoplasmic tails of such inhibitory receptors recruit the tyrosine phosphatases SHP-1 and SHP-2 ##REF##8574854##[3]##–##REF##9916713##[5]##. Inhibition occurs through SHP-mediated dephosphorylation of key components in the signaling pathway for activation, such as Vav1 ##REF##12917349##[6]##. Inhibition by KIR blocks NK cell activation at a very proximal step, which precedes actin-dependent processes ##REF##18759921##[7]##. For instance, binding of inhibitory KIR to MHC class I on target cells prevents the tyrosine phosphorylation of activation receptors 2B4 and NKG2D, as well as their recruitment to detergent-resistant membrane microdomains ##REF##12515815##[8]##, ##REF##17442943##[9]##. Engagement of ITIM-containing inhibitory receptors blocks the accumulation of F-actin at T cell and NK cell immune synapses ##REF##11160312##[10]##–##REF##15494512##[12]##, and prevents the actin-dependent accumulation of glycosphingolipid-enriched domains at inhibitory synapses in YTS cells ##REF##11724921##[13]## and NK clones ##REF##17442943##[9]##, ##REF##15265884##[14]##.</p>", "<p>Reorganization of the actin cytoskeleton is essential for the cytotoxic activity of T cells and NK cells. Inhibitors of actin polymerization prevent cytolytic activity, hinder accumulation of receptors at activating immune synapses ##REF##14612578##[15]##, and block phosphorylation of NK cell activation receptors ##REF##12515815##[8]##, ##REF##17442943##[9]##. Given that actin cytoskeleton rearrangement is inhibited by ITIM-containing receptors, it is generally assumed that KIR engagement at an inhibitory synapse prevents the delivery of activation signals by blocking the cytoskeleton–dependent movement of activating receptors. To test this hypothesis, we visualized the distribution of activation receptors 2B4 and CD2 in activating and inhibitory NK cell immune synapses, using primary human NK cells. We report the surprising finding that KIR engagement at inhibitory synapses promotes the accumulation of activation receptors 2B4 and CD2.</p>" ]

[ "<title>Materials and Methods</title>", "<title>Cells and Abs</title>", "<p>Human NK cells were isolated from peripheral blood of healthy donors using an NK cell isolation kit (Miltenyi Biotech, Auburn, CA). NK cell purity was assessed by FACS and the cells were over 98% CD3⁻CD56⁺CD16⁺. IL-2 activated NK cells were cultured as previously described ##REF##15356110##[27]##. IL-2 activated NK cells were used for assays between 2 and 4 weeks after isolation. The 721.221 EBV-transformed cell line, as well as the transfectants 721.221-HLA-Cw15 and 721.221-HLA-Cw3 were maintained in Iscove's media supplemented with 10% fetal calf serum and L-glutamine. Drosophila Schneider Cell 2 (S2) cells were maintained as previously described ##REF##15356110##[27]##. Expression of transfected proteins was induced with 1 mM copper sulfate for 48 hours. For expression of HLA-Cw4 on S2 cells, either a peptide permissive for KIR2DL1 binding (peptide #1: QYDDAVYKL) or a non-permissive (K8E: QYDDAVYEL) was added at 1 µM during the last 20 hours of the induction. The following PE-conjugated antibodies were used for S2 cell staining: anti-HLA ABC (clone G46-2.6), anti-LFA-3 (1C3), anti-CD48 (HM48-1), and anti-CD54 (HA58) (Pharmingen, San Diego CA). For confocal imaging, the following primary antibodies were used (all from Pharmingen): anti-CD2 (RPA-2.10), anti-CD11a (HI111), and anti-2B4 (2–69), For detection of KIR2DL1 and KIR2DL2,a rabbit antiserum raised against the C–terminal sequence of the cytoplasmic tail of KIR2DL1 and KIR2DL2 (cyt42/43) was used.</p>", "<title>Conjugate Formation</title>", "<p>IL-2 activated NK cells were resuspended with 10⁶ target cells at a 1∶1 ratio. The cells were gently vortexed, centrifuged at 300 RPM for 3 minutes at 4°C, and placed at 37°C for 10 minutes to allow conjugate formation. In the experiment shown in ##FIG##2##Figure 3B##, cells were additionally allowed to form conjugates for only 1 minute. The cell pellet was directly chilled on ice for 10 minutes, and then cells were gently resuspended and allowed to settle on poly-L-lysine (Sigma, St. Louis, MO) coated coverslips for 15 minutes at 4°C. Cells were fixed in 4% paraformaldehyde. Cells were permeabilized in 0.5% Triton-X100 and 10% normal donkey serum (NDS) in PBS for 30 minutes at room temperature. Cells were stained with combinations of the following primary antibodies: 5 µg/ml anti-CD2, 0.1 µg/ml anti-CD11a, 12.5 µg/ml anti-2B4, and 5 µg/ml cyt42/43. Primary antibodies were diluted in 0.5% Triton-X100 and 3% NDS in PBS and incubated with the cells for 1 hour at room temperature. After three washes in PBS, cells were incubated for 1 hour at room temperature with the appropriate secondary antibodies in 0.5% Triton-X100, 3% NDS in PBS. Secondary antibodies used were Alexa 488–conjugated goat anti–mouse and goat anti–rabbit IgG (1∶2000 dilution), and Alexa 568–conjugated goat-anti mouse and goat-anti rabbit IgG (1∶1000 dilution) (Molecular Probes, Eugene OR). After three washes in PBS, coverslips were mounted to slides using the Prolong anti-fade kit (Molecular Probes).</p>", "<title>Confocal Microscopy</title>", "<p>Images of stained conjugates were collected on a Zeiss LSM510 Meta confocal microscope system using a plan apochromat 63x/1.4 oil immersion objective. Excitation wavelengths used were 488nm (argon/krypton) and 543 (helium/neon). Differential interference contrast (DIC) images were collected simultaneously with the fluorescent images. Multitrack acquisition mode was used to avoid crosstalk between the different fluorophores. For 3D reconstructions, about 25 <italic>z</italic> sections were collected at 0.3 µm <italic>z</italic> intervals. Reconstructions were performed using Imaris 3.0.6 analysis software (Bitplane AG, Zurich Switzerland). The <italic>en face</italic> view of the immune synapse was obtained by an <italic>x-z</italic> projection of the 3D image at the cellular interface of the NK and target cells.</p>", "<title>Image Analysis</title>", "<p>Clustering or accumulation of a receptor at the NK cell immune synapse was defined as follows. Obvious conjugates between NK cells and target cells were first selected on DIC images and then fluorescent images were acquired. Clustering of NK cell receptors at the intercellular contact zone was defined as an obvious increase (usually 2 fold or greater) of the fluorescence intensity compared to the rest of the membrane. Fluorescence intensities were analyzed over the entire NK cell membrane on single confocal sections with the “Profile” function of the Zeiss LSM 510 software. This function allows the user to mark a line on a fluorescence microscope image and the software will report the intensity of every fluorophore at every position along the line. We used this function to draw a line around the entire membrane of the NK cell, starting at an arbitrary point roughly opposite the position of the NK:target cell contact. The software produces a graph of fluorescence intensity for each fluorophore at each point along the line vs. the relative position of each point along the line.</p>" ]

[ "<title>Results</title>", "<title>Detection of activating and inhibitory immune synapses between target cells and primary human NK cells</title>", "<p>We wished to study NK cell immune synapses in unmanipulated, polyclonal human NK cells in order to avoid complications or biases that may arise in the cloning of NK cells or the expression of exogenous proteins in NK cells. To do this, it was necessary to identify NK cells expressing the receptors of interest. All human NK cells express the β2 integrin LFA-1 and activation receptor 2B4, whereas a subset of NK cells express CD2. Expression of MHC class I-specific inhibitory receptors, including KIR, on NK cells is more complex. KIRs are clonally distributed among NK cells, with any given NK cell expressing its own KIR repertoire. Furthermore, monoclonal Abs to KIRs do not distinguish between inhibitory KIR2DL1 and the short–tailed, activating KIR2DS1, or between inhibitory KIR2DL2 and activating KIR2DS2. To identify inhibitory KIR2DL, a polyclonal antiserum against the conserved C-terminal amino acids of KIR2DL1 and KIR2DL2 was raised (cyt42/43 antiserum). The short cytoplasmic tails of KIR2DS1 and KIR2DS2 do not include the amino acid sequence reactive with cyt42/43 antiserum. Inhibitory synapses were identified by the clustering of KIR2DL1 towards target cells expressing its ligands HLA-Cw4 or HLA-Cw15, and by the clustering of KIR2DL2 towards target cells expressing its ligand HLA-Cw3. Several controls were performed to validate this approach.</p>", "<p>Primary NK cells were incubated with HLA class I-negative 721.221 cells, and 721.221 cells expressing HLA-Cw15. Cell conjugates were allowed to settle onto poly-L-lysine–coated coverslips, fixed, permeabilized, and stained for CD11a and inhibitory KIR2DL1/KIR2DL2. No clustering was detected in NK cells in contact with 721.221 cells, whereas 60% of cyt42/43-positive cells in contact with 221-Cw15 cells displayed clustered KIR (##FIG##0##Fig 1A##). Therefore, KIR2DL1 clustering occurs in at least 60% of KIR2DL1–positive NK cells. An NK cell population from a donor that expressed KIR2DL2 but not KIR2DL1 was also tested. In such NK cells, all of the cyt42/43-reactivity is directed at KIR2DL2. Sixty percent of KIR2DL2–expressing cells in contact with 221-Cw3 cells displayed inhibitory KIR clustering (##FIG##0##Fig 1B##). 17% of KIR2DL2⁺ NK cells formed clusters with 221-Cw15 cells (##FIG##0##Fig 1B##), which could be explained by the known crossreactivity of KIR2DL2 with HLA-Cw15 ##REF##18322206##[16]##, ##REF##9670929##[17]##.</p>", "<p>KIR clustering was also assessed with <italic>Drosophila</italic> S2 cells that express HLA-Cw4, a ligand of KIR2DL1. In this system, the target cells express well-defined combinations of ligands for human NK cell receptors. As shown previously, expression of peptide-loaded HLA-Cw4 on insect cells is sufficient to induce clustering of KIR on NK cells ##REF##12794140##[18]##. As insect cells cannot load peptides, HLA class I reaches the cell surface “empty” ##REF##1465448##[19]##. Clustering of inhibitory KIR was observed when HLA-Cw4 on S2 cells was loaded with a peptide that is permissive for KIR2DL1 binding ##REF##9126935##[20]## (##FIG##0##Fig 1C##). For comparison, HLA-Cw4 was also loaded with a peptide that is not permissive for KIR2DL1 binding ##REF##9126935##[20]##. In this case, KIR2DL1 clustering was much less frequent (##FIG##0##Fig 1C##). For the purpose of this study, immune synapses were considered inhibitory when conjugates between HLA-Cw4 or HLA-Cw15-expressing target cells and NK cells displayed clustering of cyt42/43-reactive KIR. NK cells in conjugates that lacked cyt42/43 reactivity altogether (i.e. negative for KIR2DL1 and KIR2DL2) were scored as activating immune synapses. By this approach, it was possible to distinguish activating and inhibitory immune synapses within the same population of primary NK cells that were in contact with target cells expressing well-defined ligands of NK cell activating and inhibitory receptors.</p>", "<title>Activation receptors CD2 and 2B4 accumulate at inhibitory immune synapses</title>", "<p>The distribution of most activation receptors at inhibitory NK cell immune synapses has not been examined. Accumulation of CD2 at activating immune synapses is dependent on the protein WASp and actin polymerization ##REF##14612578##[15]##. Rapid accumulation of 2B4 at activating immune synapses has been visualized in live cells ##REF##15356108##[21]##. The phosphorylation and recruitment of 2B4 to detergent-resistant membrane domains, which are also dependent on actin polymerization, are blocked by co-engagement of inhibitory KIR ##REF##12515815##[8]##. As inhibitory ITIM–containing receptors prevent actin cytoskeleton rearrangement ##REF##11160312##[10]##–##REF##15494512##[12]## and localization of GM-1–containing lipid rafts to NK cell immune synapses ##REF##11724921##[13]##, one would expect inhibition of the actin polymerization–dependent clustering of activation receptors CD2 and 2B4 by KIR. Here, we determined the localization of receptors CD2 and 2B4 in both activating and inhibitory NK cell immune synapses.</p>", "<p>The MHC class I–deficient cell line 721.221 expresses LFA-3 and CD48, which are ligands for CD2 and 2B4, respectively. 721.221 cells transfected with HLA-Cw15 (221-Cw15), which is a ligand for KIR2DL1, were used as targets to evaluate the distribution of CD2 and 2B4 in NK cell immune synapses. Contrary to expectations, CD2 accumulated at both activating (##FIG##1##Figure 2A, 2C##, cell #2) and inhibitory (##FIG##1##Figure 2A, 2C##, cell #1) NK cell immune synapses with 221-Cw15 target cells. Reconstruction of the zone of contact in inhibitory NK–target cell inhibitory synapses showed that the intensity of CD2 staining correlated well with the intensity of KIR2DL1 staining (##FIG##1##Figure 2D##, cell #1). Similar results were obtained with S2 insect cells expressing LFA-3 and HLA-Cw4 (##FIG##1##Figure 2B, 2C, 2D##, cell #3), indicating that engagement of other receptors is not required for CD2 accumulation at NK cell immune synapses. Surprisingly, the frequency of CD2 clustering in NK cell conjugates with 721.221 cells and transfected S2 cells was higher at inhibitory immune synapses than at activating synapses (##FIG##2##Figure 3A##). Accumulation of KIR at inhibitory synapses is very rapid ##REF##15494512##[12]##, ##REF##12794140##[18]##. To test whether KIR clustering may accelerate the accumulation of CD2, NK–target cell conjugates were allowed to form for only one minute. In contrast to activating immune synapses, in which CD2 accumulation was more limited at one minute, CD2 accumulation in inhibitory synapses at one minute was already as high as its accumulation at 10 minutes (##FIG##2##Figure 3B##). Therefore, KIR engagement with HLA class I on target cells promotes rapid accumulation of CD2 at inhibitory NK cell immune synapses.</p>", "<p>Accumulation of activation receptor 2B4 was also observed in NK cells that formed activating (##FIG##3##Figure 4A, 4B##, cell #2) and inhibitory (##FIG##3##Figure 4A, 4B##, cells #1 and #3) immune synapses with 221-Cw15 cells. Three-dimensional reconstruction of the contact zone at inhibitory synapses showed that the intensity of 2B4 staining correlated well with the intensity of KIR staining, implying that, similar to CD2, 2B4 colocalizes with KIR at inhibitory NK cell immune synapses (##FIG##3##Figure 4C##).</p>", "<title>Exclusion of LFA-1 at inhibitory immune synapses</title>", "<p>Although inhibitory KIR segregate from LFA-1 at inhibitory synapses, different LFA-1 and KIR distribution patterns have been reported, which could be due to different levels of HLA-C expression on target cells ##REF##15265884##[14]##, ##REF##10611338##[22]##–##REF##17082605##[24]##. <italic>Drosophila</italic> S2 cells expressing ICAM-1 and HLA-Cw4 were used to examine the distribution of LFA-1 and KIR2DL1 in the absence of the many other receptor–ligand interactions that occur between NK cells and mammalian target cells. LFA-1 in NK–S2 cell conjugates was detected with an anti-CD11a antibody. Confocal <italic>z</italic>-series of activating and inhibitory synapses were acquired. CD11a staining on the NK cells was often uneven, with patches of CD11a around the cell periphery. However, careful quantitation of CD11a fluorescence on NK cells at activating synapses (##FIG##4##Figure 5A##, cell #2) revealed a reproducible increase of CD11a at the site of cell–cell contact, when compared to the rest of the NK cell membrane (##FIG##4##Figure 5B##, cell #2). Analysis of a number of conjugates showed that CD11a accumulates at approximately 75% of activating NK cell synapses (##FIG##4##Figure 5D##).</p>", "<p>In contrast to the CD11a accumulation seen at activating synapses, CD11a at inhibitory synapses was reduced in the central zone of cell–cell contact (##FIG##4##Figure 5A, 5B##, cell #1). The overall distribution of LFA-1 on NK cells engaged in inhibitory synapses showed that KIR binding to HLA-C on target cells not only prevents accumulation of LFA-1, but actively excludes some of the LFA-1 from inhibitory synapses. Three–dimensional reconstruction of confocal z-stacks of inhibitory synapses revealed an obvious hole in the CD11a staining, which corresponded to the most intense KIR staining (##FIG##4##Figure 5C##). Quantitation of a number of conjugates showed that exclusion of CD11a from the zone of contact was observed in about half of the inhibitory synapse (##FIG##4##Figure 5D##). The other half of inhibitory synapses showed mostly unchanged CD11a fluorescence across the area of cell contact (##FIG##4##Fig 5D##). Therefore, the segregation of inhibitory KIR and LFA-1, which occurs at inhibitory synapses between NK cells and mammalian target cells, does also occur with insect cells expressing only ICAM-1 and HLA-C. In conclusion, the distribution of LFA-1 at NK cell immune synapses differs from that of activation receptors 2B4 and CD2. First, the extent of LFA-1 accumulation at activating immune synapses is more limited. Second, in contrast to activation receptors 2B4 and CD2, which accumulate at inhibitory synapses, LFA-1 is often excluded from the zone of KIR clustering.</p>" ]

[ "<title>Discussion</title>", "<p>The clustering of receptors that occurs upon ligand binding at cell–cell contacts is usually an energy-dependent process, the basis of which is still poorly understood ##REF##10963677##[25]##. Inhibition of actin polymerization blocks the accumulation of receptors CD2 and 2B4 at NK cell immune synapses, and the recruitment of 2B4 to detergent-resistant membrane domains ##REF##14612578##[15]##, ##REF##11034353##[26]##. In contrast, inhibitory KIRs have the very unusual property of clustering independently of actin polymerization and of ATP metabolism when binding to an HLA class I ligand on target cells ##REF##10611338##[22]##. Expression of a cognate HLA-C ligand on transfected <italic>Drosophila</italic> cells was sufficient to induce KIR clustering ##REF##12794140##[18]##. As ITIM-containing inhibitory receptors prevent actin dynamics ##REF##11160312##[10]##, ##REF##12351398##[11]##, it was predicted that KIR inhibitory signaling would prevent the energy- and actin-dependent clustering of activating receptors, thereby blocking activation of NK cells. Indeed, clustering of the activation receptor NKG2D is inhibited by KIR engagement ##REF##17442943##[9]##. However, we report here the unexpected accumulation of activation receptors CD2 and 2B4 at both activating and inhibitory NK cell immune synapses. CD2 clustered at inhibitory synapses even more frequently and more rapidly than at activating synapses. The sensitivity to cytochalasin D (an inhibitor of actin polymerization) and to azide (an inhibitor of cytochrome c oxidase) of CD2 and 2B4 clustering was in fact lifted by KIR co-engagement (data not shown). We conclude that KIR does not inhibit, but rather promotes accumulation of CD2 and 2B4 at inhibitory immune synapses.</p>", "<p>We have examined the clustering of receptors at activating and inhibitory synapses of primary, unmanipulated NK cells with target cells. To work with primary NK cells, one has to overcome the complication due to the heterogeneous expression of the family of inhibitory receptors, which are essentially randomly distributed on NK cells. To do so, inhibitory KIR2DL1 and KIR2DL2 were visualized in polyclonal NK cell populations with a specific antiserum raised against their conserved cytoplasmic tail. By this approach, inhibitory NK cell immune synapses were identified not only on the basis of the NK and target cell KIR and HLA phenotypes, but also by visible clustering of inhibitory KIR. In addition to the use of human target cells, which express ligands for many different NK cell receptors, experiments were also performed with transfected <italic>Drosophila</italic> insect cells. By expression of only one or a few ligands for human NK cell receptors, this insect cell system is ideally suited to dissect the contribution of individual receptors to NK cell activation ##REF##15356110##[27]##, ##REF##16203869##[28]##. Furthermore, transfected <italic>Drosophila</italic> cells provide a very stringent test for MHC class I-dependent function. As MHC class I folding at the surface of insect cells requires addition of specific exogenous peptide ##REF##1465448##[19]##, insect cells expressing MHC class I in the absence of peptide provide an ideal negative control. Using this system, we have shown that clustering of KIR correlated with reduced LFA-1 at the synapse, but did not prevent accumulation of CD2 or 2B4.</p>", "<p>The reason for the unexpected coclustering of CD2 and 2B4 with inhibitory KIR may be to facilitate inhibition by maintaining proximity of activating receptors with the tyrosine phosphatases SHP-1 and SHP-2 recruited by KIR, thus providing the opportunity for rapid deactivation of signaling initiated by CD2 or 2B4. Otherwise, 2B4 and CD2 accumulation in a region peripheral to the clusters of KIR could be dangerous, as signaling could proceed unimpeded by KIR-dependent inhibition. The predicted sizes of KIR2DL1/HLA-Cw4, 2B4/CD48, and CD2/LFA-3 complexes are very similar, and much smaller than LFA-1/ICAM-1 complexes. Co-clustering of 2B4, CD2, and KIR2DL1, as well as the exclusion of LFA-1 from these clusters, could therefore result from partitioning based on size, as has been proposed for the T cell synapse ##REF##10723794##[29]##.</p>", "<p>A recent report visualized inhibitory signaling by KIR at NK cell immune synapses by detecting phosphorylation of the KIR ITIMs through fluorescence resonance energy transfer (FRET) imaging ##REF##16801390##[30]##. Surprisingly, KIR phosphorylation does not occur uniformly across the inhibitory synapse but in small clusters, suggesting that inhibition may be transient and local. In such a case, colocalization of inhibitory KIR with clusters of activation receptors would greatly improve KIR-dependent inhibition. Alternatively, clusters of phosphorylated KIR may represent KIR molecules that have been phosphorylated due to their proximity to signaling clusters of activation receptors. However, the fact that KIR prevents tyrosine phosphorylation of 2B4 and movement of 2B4 into detergent-resistant membrane domains, suggests that KIR phosphorylation is not downstream of 2B4 signaling ##REF##18759921##[7]##, ##REF##11034353##[26]##. KIR prevents the recruitment of 2B4 and of NKG2D to detergent-resistant membrane domains ##REF##12515815##[8]##, ##REF##17442943##[9]##, and inhibits the accumulation of ganglioside GM1 at the synapse ##REF##17442943##[9]##, ##REF##11724921##[13]##, ##REF##15265884##[14]##. These observations suggest a model in which signaling by activation receptors is inhibited due to the failure of lipid raft-associated signaling molecules to coalesce at inhibitory synapses. The precise mechanism by which KIR inhibits NK cell activation will only be understood through the study of the unique biophysical properties of KIR, which lead to its unusual, energy- and actin-independent clustering.</p>" ]

[]

[ "<p>Conceived and designed the experiments: NS EOL. Performed the experiments: NS MEM. Analyzed the data: NS MEM EOL. Contributed reagents/materials/analysis tools: NS MEM. Wrote the paper: MEM EOL.</p>", "<p>Current address: Service de Médecine Interne, CHU Conception AP-HM, Marseille, France</p>", "<p>Natural killer (NK) cell activation receptors accumulate by an actin-dependent process at cytotoxic immune synapses where they provide synergistic signals that trigger NK cell effector functions. In contrast, NK cell inhibitory receptors, including members of the MHC class I-specific killer cell Ig-like receptor (KIR) family, accumulate at inhibitory immune synapses, block actin dynamics, and prevent actin-dependent phosphorylation of activation receptors. Therefore, one would predict inhibition of actin-dependent accumulation of activation receptors when inhibitory receptors are engaged. By confocal imaging of primary human NK cells in contact with target cells expressing physiological ligands of NK cell receptors, we show here that this prediction is incorrect. Target cells included a human cell line and transfected <italic>Drosophila</italic> insect cells that expressed ligands of NK cell activation receptors in combination with an MHC class I ligand of inhibitory KIR. The two NK cell activation receptors CD2 and 2B4 accumulated and co-localized with KIR at inhibitory immune synapses. In fact, KIR promoted CD2 and 2B4 clustering, as CD2 and 2B4 accumulated more efficiently at inhibitory synapses. In contrast, accumulation of KIR and of activation receptors at inhibitory synapses correlated with reduced density of the integrin LFA-1. These results imply that inhibitory KIR does not prevent CD2 and 2B4 signaling by blocking their accumulation at NK cell immune synapses, but by blocking their ability to signal within inhibitory synapses.</p>" ]

[]

[]

[ "<fig id=\"pone-0003278-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003278.g001</object-id><label>Figure 1</label><caption><title>Detection of inhibitory synapses using the cyt42/43 antiserum.</title><p>IL-2 activated polyclonal human NK cells were mixed with target cells for 10 minutes at 37°C, fixed, permeabilized and stained with the cyt42/43 rabbit polyclonal antiserum. NK cells and NK:target cell conjugates stained with the cyt42/43 antiserum were scored for KIR clustering. The number of cyt42/43-positive cell conjugates analyzed is given in parentheses over each bar. (A) Mixing with .221 and .221-Cw15 target cells, as indicated. (B) NK cells expressing KIR2DL2 and not KIR2DL1 mixed with .221-Cw3 and .221-Cw15 cells, as indicated. (C) Mixing with S2–Cw4 cells loaded with a peptide that is permissive for KIR2DL1 binding (peptide #1) or a peptide that is nonpermissive for KIR2DL1 binding (K8E), as indicated.</p></caption></fig>", "<fig id=\"pone-0003278-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003278.g002</object-id><label>Figure 2</label><caption><title>CD2 accumulates at both activating and inhibitory synapses.</title><p>IL-2 activated polyclonal human NK cells were mixed with target cells at 37°C for 10 minutes, fixed, permeabilized, and stained with the cyt42/43 antiserum and a mAb to CD2 followed by the relevant secondary antibodies. Confocal microscope <italic>z</italic>-series were obtained. (A) Mixed with .221-Cw15 as indicated. A single confocal section is shown. The cell labeled #1, which displays KIR expression and clustering, represents an inhibitory synapse while cell #2, which lacks KIR2DL1 expression, displays an activating synapse. (B) Mixed with S2–LFA-3/Cw4 target cells, as indicated. A single confocal section is shown. (C) The fluorescence intensity was scanned around the perimeter of conjugated NK cells. Profiles labeled 1, 2, and 3 are from the corresponding cells in ##FIG##1##Figures 2A and 2B##. The green and red lines represent the cyt42/43 and anti–CD2 fluorescence, respectively. Vertical red and blue lines mark the boundaries of cell contact as determined in DIC images. (D) Confocal <italic>z</italic>-stacks were used to create an <italic>en face</italic> view of the zone of cell contact in 2 inhibitory synapses.</p></caption></fig>", "<fig id=\"pone-0003278-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003278.g003</object-id><label>Figure 3</label><caption><title>CD2 accumulates more frequently at inhibitory synapses than at activating synapses.</title><p>(A) Conjugates between IL-2 activated polyclonal human NK cells and 721.221-Cw15 cells or S2–LFA-3/Cw4 cells were formed and stained as in ##FIG##1##Figure 2##. Activating (Act.) and inhibitory (Inhib.) synapses were scored for clustering of CD2 at the zone of contact. The number of conjugates scored in each condition is indicated in parentheses. (B) Conjugates between activated NK cells and 721.221-Cw15 cells were allowed to form for 1 minute or 10 minutes as indicated, stained with the cyt42/43 antiserum and an anti-CD2 antibody as in ##FIG##1##Figure 2##, and scored for CD2 clustering.</p></caption></fig>", "<fig id=\"pone-0003278-g004\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003278.g004</object-id><label>Figure 4</label><caption><title>2B4 accumulates at both activating and inhibitory synapses.</title><p>Activated NK cells were mixed with target cells at 37°C for 10 minutes, fixed, permeabilized, and stained with the cyt42/43 antiserum and a mAb to 2B4 followed by the appropriate secondary antibodies. (A) Mixed with .221-Cw15 target cells. Confocal microscope <italic>z</italic>-series were obtained, and single sections are shown. The cells labeled #1 and #3, which display KIR expression and clustering, represent inhibitory synapses while cell #2, which lacks KIR2DL1 expression, is an activating synapse. The NK cells without a number in the top image were not analyzed, because they did not appear to form tight conjugates with target cells. (B) The fluorescence intensity was scanned around the perimeter of conjugated NK cells. Profiles labeled 1, 2, and 3 are from the corresponding cells in ##FIG##3##Figure 4A##. The green and blue lines represent the cyt42/43 and anti–2B4 fluorescence, respectively. Vertical red and blue lines mark the boundaries of cell contact as determined in DIC images. (C) Confocal <italic>z</italic>-stacks were used to create an <italic>en face</italic> view of the zone of cell contact in 2 inhibitory synapses.</p></caption></fig>", "<fig id=\"pone-0003278-g005\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003278.g005</object-id><label>Figure 5</label><caption><title>Surface level of CD11a is reduced at inhibitory synapses.</title><p>IL-2 activated polyclonal human NK cells and S2 cells expressing ICAM-1 and peptide-loaded HLA-Cw4 were mixed at 37°C for 10 minutes, fixed, permeabilized, and stained with cyt42/43 and a mAb to CD11a followed by the appropriate secondary antibodies. (A) Confocal microscope <italic>z</italic>-series were obtained, and single sections are shown. The cell labeled #1, which shows KIR expression and clustering, represents an inhibitory synapse while cell #2, which lacks KIR2DL1 expression, displays an activating synapse. (B) The fluorescence intensity was scanned around the perimeter of conjugated NK cells. Profiles labeled 1 and 2 are from the corresponding cells in ##FIG##4##Figure 5A##. The third profile is another representative inhibitory synapse. The green and red lines represent the cyt42/43 and anti–CD11a fluorescence, respectively. Vertical red and blue lines mark the boundaries of cell contact as determined in DIC images. (C) Confocal <italic>z</italic>-stacks were used to create an <italic>en face</italic> view of the zone of cell contact in 2 inhibitory synapses. (D) The frequency of synapses displaying increased CD11a, reduced CD11a, or no change in CD11a intensity was determined for both activating and inhibitory synapses.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[ "<fn-group><fn fn-type=\"COI-statement\"><p><bold>Competing Interests: </bold>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p><bold>Funding: </bold>This research was supported by the Intramural Research Program of the NIAID, NIH. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"pone.0003278.g001\"/>", "<graphic xlink:href=\"pone.0003278.g002\"/>", "<graphic xlink:href=\"pone.0003278.g003\"/>", "<graphic xlink:href=\"pone.0003278.g004\"/>", "<graphic xlink:href=\"pone.0003278.g005\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Gene expression is regulated through a complex series of events that coordinate the synthesis, processing and export, and in some instances, degradation of the mRNA. As the polymerase reaches the end of the gene, these events must coincide to produce a functional mRNA. The transcription of sequences that specify cleavage/polyadenylation signals directs the termination phase of transcription ##REF##2658217##[1]##, ##REF##10859161##[2]##. In yeast, the cleavage/polyadenylation machinery, which is composed of the Cleavage Factor I (CFI) (CSTF in humans) and Cleavage/Polyadenylation Factor (CPF) (CPSF in humans) complexes, is responsible for the recognition of polyA/termination signals and their subsequent processing ##REF##10859161##[2]##–##REF##16651651##[6]##. Within CFI, Rna15, the homolog to mammalian CSTF64, specifically recognizes and binds sequences that direct the polyA/termination machinery to the 3′ end ##REF##11689695##[7]##. Lacking is a detailed understanding of how the recognition of polyA signals transitions towards transcription termination, and how this is linked to the preceding elongation phase. The allosteric or anti-termination model takes into account some of these events. In this model, association of a positive elongation factor or an anti-terminator with RNA Polymerase II (RNAP II) is essential. Dissociation of this factor from the polymerase at the end of a gene is predicted to have two effects: (1) it destabilizes transcriptionally active RNAP II, and (2) it promotes the recruitment of the polyA/termination machinery.</p>", "<p>The SR-like protein Npl3 functions in transcription, 3′ end processing, hnRNP formation, and mRNA export ##REF##11233987##[8]##–##UREF##0##[12]##. Previously, we proposed that the competition between Npl3 and Rna15 for binding to RNA targets is central to the function of Npl3 in termination and 3′ end processing ##REF##15902270##[11]##, ##UREF##0##[12]##. Increased binding of Npl3 to a target sequence can shift the competition away from processing, strongly inhibiting cleavage/polyadenylation <italic>in vitro</italic>\n##UREF##0##[12]##. <italic>In vivo,</italic> mutations or deletion of <italic>npl3</italic> can result in the utilization of weak or otherwise poorly recognized polyA sites, while strong polyA signals are recognized and stably bound by cleavage/polyadenylation factors for processing ##REF##15902270##[11]##, ##UREF##1##[13]##. As we demonstrated previously, shifting the competition in favor of protection instead of processing by overexpressing Npl3 strongly inhibits cleavage/polyadenylation ##UREF##0##[12]##. These results are consistent with a model of competition driven by the relative binding affinities of Rna15 and Npl3, yet an additional layer of regulation could not be excluded.</p>", "<p>Phosphorylation plays an important role in the association/dissociation of a number of activators and factors that regulate transcription and mRNA processing. Most notable is the co-transcriptional phosphorylation of the carboxy terminal domain (CTD) of RNAP II. In <italic>S. cerevisiae,</italic> phosphorylation of the CTD repeat at serine 5 recruits capping enzyme, while phosphorylation at serine 2 functions to coordinate transcription and mRNA processing ##REF##11018013##[14]##–##REF##12942140##[17]##. Another prominent transcription/RNA processing kinase is Casein Kinase 2 (CK2). In yeast, CK2 associates with elongation factors Spt16/Pob3 and Chd1 ##REF##12242279##[18]##, ##REF##14759368##[19]##. In higher eukaryotes, one third of CK2 phosphorylation targets are involved in gene expression, half of these being transcription factors ##REF##12631575##[20]##. These include the activator PC4 (Sub1 in yeast), which is required for the regulation of mammalian promoter-dependent elements (DPE), and was shown to have an anti-termination effect ##REF##15893730##[21]##–##REF##16858867##[23]##. CK2 also phosphorylates factors within the polyA/termination machinery that are important for 3′ end processing ##REF##16137619##[24]##, ##REF##17585202##[25]##.</p>", "<p>Here we show that Npl3 directly interacts with phosphorylated serine 2 of the CTD and stimulates the elongation rate of RNAP II <italic>in vitro.</italic> Association with polymerase is inhibited by phosphorylation of Npl3 on S411. Stimulation of elongation is also reduced by a mutation in the Npl3 RNA binding domain, indicating that RNA binding is important. CK2 is shown to be required for Npl3's phosphorylation, and alter the competition between Npl3 and Rna15 for binding to the RNA. These results demonstrate that Npl3 functions as both a positive transcription elongation factor and an anti-terminator and that these activities are regulated by phosphorylation.</p>" ]

[ "<title>Materials and Methods</title>", "<title>Plasmids and Strains</title>", "<p>Plasmids pSBEThis7-<italic>NPL3</italic> and pSBEThis7-<italic>npl3-120</italic> were described previously ##REF##15902270##[11]##. For plasmids pSBEThis7-<italic>npl3-S411A,</italic> -<italic>npl3-S411D</italic>, and -<italic>npl3-S411E,</italic> the ORF of <italic>NPL3</italic> was amplified by PCR using the oligo: <named-content content-type=\"gene\">5′-CATGCCATGGCTGAAGCTCAAGAAACTCACG-3′</named-content> and <named-content content-type=\"gene\">5′- CGCGGATCCGCTTACCTGGTTGGTG<bold>C</bold>TCTTTCACG-3′</named-content> for the S411A; or <named-content content-type=\"gene\">5′- CGCGGATCCGCTTACCTGGTTGGT<bold>TC</bold>TCTTTCACG-3′</named-content> for the S411D; and <named-content content-type=\"gene\">5′-CGCGGATCCGCTTACCTGGTTGG<bold>GTC</bold>TCTTTCACG-3′</named-content> for the S411E substitution, and cloned into <italic>BamH</italic>I/<italic>Nco</italic>I site of pSBEThis7. Plasmid pET21b-<italic>CKA1</italic>-His6 was described previously ##REF##16137619##[24]##. DNA template pAdGR220 used for transcription assays was described previously ##REF##16006523##[28]##. Yeast strains used are shown in ##SUPPL##0##Table S1##. The <italic>npl3-S411A</italic> strain was kindly provided by M. Lund (UCSF).</p>", "<title>Protein Purifications</title>", "<p>\n<italic>E. coli</italic> strain Rosetta (DE3) or BL21(DE3) were transformed with pSBEThis7-<italic>npl3-S411A, -S411D</italic>, <italic>-S411E</italic>, pET21b-<italic>RNA15</italic> or -<italic>CKA1-His6</italic> and purified as described previously ##REF##15902270##[11]##. Wild-type RNAP II was purified from yeast as described previously ##REF##14522989##[27]##.</p>", "<title>Pull-down and immunoprecipitation assays</title>", "<p>Npl3 peptides were conjugated to resin using the Carboxylink Immobilization Kit (Pierce). Peptides containing the amino acid sequences: GGYGGYSRGGYGGY (RGG/RS1), RGGYDSPRGGY (RGG/RS2), YRTRDAPRERSPTR (RGG/RS3), or YRTRDAPRERpSPTR (RGG/RS3p) were synthesized (BCMP/HMS Biopolymers Lab). Peptides were normalized by OD 280nm. 100 ng of purified RNAP II (A. Ponticelli) was incubated with each of the conjugated peptides overnight at 4°C in IPP-150 buffer (10 mM Tris [pH 7.9], 150 mM NaCl, 1 mM MgOAc, 2 mM CaCl₂, 0.1% NP-40, 1 mM DTT with protease inhibitors: 1 µg/ml of aprotinin, leupeptin, antipain and pepstatin-A), and washed using buffer IPP-150 plus 0.01% SDS (sodium dodecyl sulfate) ##REF##16137619##[24]##. Amino-terminal biotinylated CTD peptides were previously described ##REF##15565157##[57]##. For immunoprecipitations, IgM agarose (Sigma) was conjugated to a His antibody. 100 ng of purified RNAP II was incubated with recombinant His-Npl3, -S411A, -S411D or -S411E for 2 hours at 4°C in IPP-150 buffer and washed as described above. Alternatively, 5 µl of streptavidin-coated magnetic beads (Dynal) were used to bind 1 µg of a CTD peptide repeat (YSPTSPS) containing unphosphorylated, Ser 2, Ser 5 or Ser 2/Ser 5 phosphorylated peptides. 100 ng of recombinant Npl3 was incubated with each CTD peptide and 500 ng Bovine Serum Albumin (BSA), as described previously ##REF##15565157##[57]##. For immunoprecipitation of endogenous Npl3 for mass spectrometry analysis, Protein A (Sigma) was conjugated to polyclonal rabbit anti-Npl3 (kindly provided by Lithgow, La Trobe University, Australia) and incubated with 5 mg of whole cell extracts prepared as described previously ##REF##10523662##[58]##. All protein was eluted by boiling in the presence of SDS-PAGE loading buffer and resolved using 8% SDS-PAGE. Immunoblotting was done using standard methods using polyclonal mouse anti-CTD (8WG16).</p>", "<title>Oligo(dC)-tailed template transcription assays</title>", "<p>Transcription reactions were performed as described previously ##REF##16006523##[28]##. Briefly, reactions were carried out at room temperature in the presence of 20 mM HEPES-KOH, pH 7.9, 20 mM Tris-HCl, pH 7.9, 8 mM MgOAc, 100 mM KOAc, 1 µM DTT, 0.5 mg/ml BSA, 3% (vol/vol) glycerol, 8 U of RNasin, 100 ng template DNA and 100 ng RNAP II. Pulse was carried out by the addition of nucleotides (50 µM ATP, 50 µM GTP, 2 µM CTP, and 10 µCi [α-³²P] CTP (3000Ci/mmol), followed by incubation for 30 minutes. Chase was performed by the addition of 1 µM UTP and 50 µM unlabeled CTP ##REF##16006523##[28]##. Reactions were stopped by the addition of 10 mM Tris-HCl, pH 7.2, 0.5 mM EDTA and 0.3 M NaCl, 0.2% SDS and 20 µg proteinase K. The reaction mixture was extracted once with phenol-chloroform-isoamyl alcohol. Transcripts were resolved on an 8% polyacrylamide–7 M urea gel and then visualized using a PhosphorImager.</p>", "<title>Cka1 Phosphorylation and Phosphatase <italic>in vitro</italic> Assays</title>", "<p>500 ng of recombinant Npl3, Npl3-S411A, -S411D, -S411E and Rna15 were phosphorylated in kinase buffer (20 mM HEPES-KOH, pH 7.0, 7.5 mM MgOAc, 100 mM KOAc, 2 mM DTT, 20% glycerol) at 30°C for 60 min with [<italic>γ</italic>-³²P] –ATP (or cold ATP for MS analysis) and 500 ng recombinant Cka1-His6. For protein phosphatase assays, reactions were carried out in 1× Lambda Phosphatase (<italic>λ</italic>-PPase) reaction buffer ##UREF##1##[13]## with 500 ng of His-Npl3 and titrated (200–600 U) Lambda Protein Phosphatase ##UREF##1##[13]##. These were then incubated at 30°C for 60 min and stopped by the addition of SDS sample buffer. Proteins were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and detected using a PhosphorImager.</p>", "<title>Mass spectrometry analysis</title>", "<p>For identification of endogenous phosphorylated peptides in Npl3, the proteinwas immunoprecipitated from whole cell extracts. This <italic>in vivo</italic> phosphorylated Npl3 and the <italic>in vitro</italic> phosphorylated protein were excised from Coomassie-stained gels, digested with trypsin, and analyzed by LC-MS/MS in a linear ion trap (LTQ) mass spectrometer. MS/MS spectra were searched against the <italic>S. cerevisiae</italic> protein database.</p>", "<title>Protein analysis</title>", "<p>For the preparation of yeast whole-cell extracts, 10 ml of cells were grown overnight, inoculated to an optical density (600 nm) of 1.5, allowed to grow at the permissive (25°C) or non-permissive (37°C) temperature for 2 hours, and pellet. Glass beads were used to disrupt cells in lysis buffer (20 mM HEPES [pH 7.6], 200 mM potassium acetate [KOAc], 10% glycerol, 1 mM EDTA) supplemented with protease inhibitors (1 mM phenylmethylsulfonyl fluoride and 1 µg/ml of aprotinin, leupeptin, antipain and pepstatin-A). Equal amounts of protein were then resolved using SDS-PAGE. Immunoblotting was performed using standard methods with polyclonal rabbit anti-phosphorylated Npl3 (kindly provided by C. Guthrie, UCSF) and anti-Npl3. Polyclonal rabbit anti-Rpb3 was from Neoclone, Horseradish peroxidase-conjugated polyclonal goat anti-rabbit was from SouthernBiotech, and goat anti-mouse HRP antibodies were from Jackson ImmunoResearch Laboratories.</p>", "<title>UV cross-linking assays</title>", "<p>UV cross-linking experiments were performed as described previously ##UREF##0##[12]##. RNA oligos containing the sequences: <named-content content-type=\"gene\">5′-UAAUAAUGACUAUAUAUG-3′</named-content> (N4) or <named-content content-type=\"gene\">5′-UUGCCUGGUUGCCUGGUU-3′</named-content> (N2) were synthesized (Invitrogen), and [<italic>α</italic>-³²P]ATP 5′ end-labeled using T4 Polynucleotide Kinase (Invitrogen) as described previously ##REF##18022637##[33]##. RNA (∼100,000 cpm) was mixed with recombinant Npl3, Rna15, and Cka1 in kinase buffer (200 mM HEPES-KOH, pH 7.0, 75 mM MgOAc, 1 M KOAc, 20 mM DTT, 20% glycerol) with 200 µM ATP. All reactions were incubated for 20 minutes at room temperature and then UV irradiated in an Ultra-lum/UVC-515 ultraviolet multilinker set at 1800 µJ (×100). Loading buffer was added and reactions were resolved on a 10% SDS-polyacrylamide gel (30∶0.8 acrylamide∶bis), dried and exposed to a PhosphorImager screen.</p>", "<title>Expression analysis with tiled arrays</title>", "<p>RNA for <italic>NPL3, npl3-120, npl3-S411A</italic> and <italic>rna15-2</italic> was extracted as described previously ##REF##17157255##[59]##. Sense RNA was synthesized using whole transcript sense target labeling assay (Affymetrix, HMS Biopolymers Facility), and hybridized to <italic>S. cerevisiae</italic> Tiling 1.0R Arrays, and intensities were analyzed using the Tiling Analysis Software (TAS) (<ext-link ext-link-type=\"uri\" xlink:href=\"https://www.affymetrix.com/support/learning/tiling_analysis/tiling_analysis_sw_tutorial.affx\">https://www.affymetrix.com/support/learning/tiling_analysis/tiling_analysis_sw_tutorial.affx</ext-link>). Representative graphs were generated using the R software ##UREF##4##[60]##. The microarray datasets in ##FIG##4##Figure 5## have been deposited in the GEO database (<ext-link ext-link-type=\"uri\" xlink:href=\"http://www.ncbi.nlm.nih.gov/projects/geo/\">http://www.ncbi.nlm.nih.gov/projects/geo/</ext-link>) under accession number GSE12677.</p>" ]

[ "<title>Results</title>", "<title>Npl3 affects the rate of transcription elongation</title>", "<p>Mutations in or near the second RNA Recognition Motif (RRM) of <italic>NPL3</italic> have been shown to enhance termination ##REF##15902270##[11]##. These same <italic>npl3</italic> alleles confer sensitivity to MPA (mycophenolic acid) and 6AU (6-azauracil), phenotypes suggestive of elongation defects. To test for an effect of Npl3 on the rate of RNAP II elongation, an <italic>in vitro</italic> transcription assay containing purified RNAP II and a promoterless dC-tailed template was used ##REF##7068686##[26]##–##REF##16006523##[28]##. In this assay, the polymerase binds to the single-stranded oligo(dC)-tail and initiates transcription in the presence of ATP/GTP/[<italic>α</italic>-³²P]CTP ##REF##7068686##[26]##. In the absence of UTP, the polymerase is forced to pause at the first non-template T stretch (135 nt = T₃) <bold>(</bold>\n##FIG##0##\nFigure 1, lanes 1 and 2\n##\n<bold>)</bold>. After a 30-minute labeling incubation, the reaction is split and incubated for an additional 5-minutes with buffer or Npl3. The labeled transcripts are then extended with limiting UTP and excess unlabeled CTP for 2 to 60 minutes in the absence <bold>(</bold>\n##FIG##0##\nFigure 1, odd lanes\n##\n<bold>)</bold>, or presence of Npl3 <bold>(</bold>\n##FIG##0##\nFigure 1, even lanes\n##\n<bold>)</bold>. Limiting the chase lowers the rate of nucleotide addition through dTTP regions on the template, causing the accumulation of transcripts that are ∼250 nt in length. These products are eventually chased into longer products. A notable increase in the rate of elongation by RNAP II was observed in the presence of Npl3 <bold>(</bold>\n##FIG##0##\nFigure 1\n##\n<bold>)</bold>. These results suggest that Npl3 functions as a positive elongation factor in the absence of termination factors.</p>", "<title>Unphosphorylated Npl3 interacts with RNAP II and mediates the elongation effect</title>", "<p>The C-terminal RS domain of Npl3 contains three serine/proline ##REF##9786943##[29]## and six serine/arginine (SR) repeats that are interspersed within a region rich in arginines and glycines (RGG/RS domain). RS domains in other proteins can mediate protein-protein interactions ##REF##9030686##[30]##. Phosphorylation of RS domains also functions in the recruitment of these proteins to sites of transcription (<italic>e.g.,</italic> SF2/ASF) ##REF##9786943##[29]##. In addition, RS domain phosphorylation can affect RNA binding affinity ##REF##11233987##[8]##, ##REF##9030686##[30]##, ##REF##15525510##[31]##.</p>", "<p>Npl3 and RNAP II co-immunoprecipitate from yeast extracts, although it is unclear if this interaction is direct or indirect ##REF##11459827##[9]##. In order to test the role of the RGG/RS domain of Npl3 plus its phosphorylation in this interaction, peptides corresponding to three repeats within the RGG/RS domain were conjugated to a resin and used in pull-down experiments with purified RNAP II <bold>(</bold>\n##FIG##1##\nFigure 2A\n##\n<bold>)</bold>. Peptide RGG/RS1 corresponds to the SR repeat (GGYGGYSRGGYGGY), peptide RGG/RS2 corresponds to the SP repeat (RGGYDSPRGGY), and peptide RGG/RS3 corresponds to the SP repeat containing serine 411 (YRTRDAPRERSPTR). Since S411 is phosphorylated <italic>in vivo,</italic> and it has been extensively characterized ##REF##14759366##[10]##, ##UREF##2##[32]##, we also included a phosphorylated form of this peptide (YRTRDAPRERpSPTR). No RNAP II was detected as interacting with the RGG/RS1 or RGG/RS2 peptides <bold>(</bold>\n##FIG##1##\nFigure 2A\n##\n<bold>)</bold>. However, significant binding was observed with the unphosphorylated RGG/RS3 peptide. Binding to the phosphorylated RGG/RS3 was much weaker. These results suggest that Npl3 interacts directly with RNAP II primarily through RGG/RS3 in its unphosphorylated form.</p>", "<p>The contribution of S411 in RGG/RS3 to the interaction between Npl3 and RNAP II was further tested by immunoprecipitation with full-length Npl3 and three different mutants of S411 (S411A, S411E and S411D), which resemble unphosphorylated or phosphorylated forms of the protein. To rule out possible conformational changes due to the mutations, the point mutants were tested for RNA binding and found to retain their ability to bind RNA (data not shown). Purified RNAP II was incubated with recombinant histidine-tagged Npl3, Npl3-S411A, -S411D or -S411E, and reactions were immunoprecipitated using anti-His antibody and immunoblotted for RNAP II. Wild-type Npl3 bound to RNAP II <bold>(</bold>\n##FIG##1##\nFigure 2B\n##\n<bold>)</bold>, while mutants Npl3-S411A, -S411D and -S411E showed further decrease. This result suggests that changes to S411 affect the interaction, particularly those mutations that resemble a phosphorylated state.</p>", "<p>The interaction between unphosphorylated RGG/RS3-S411 and RNAP II suggested that Npl3's effect on elongation might be regulated through phosphorylation of S411. The three mutants of S411 (S411A, S411E and S411D) were used in the elongation assay to test for this possibility. In addition, a mutant that was previously shown to have a defect in RNA binding (Npl3-120) and a truncated form of Npl3 containing only the two RRMs (aa 121–280) were also tested ##REF##15902270##[11]##, ##UREF##0##[12]##, ##REF##18022637##[33]##. Elongation stimulation by Npl3-S411D and Npl3-S411E, which may resemble the phosphorylated form of Npl3, was reduced as compared to the wild-type <bold>(</bold>\n##FIG##1##\nFigure 2C\n##\n<bold>)</bold>. In addition, the Npl3-S411A mutant was slightly reduced. This result further highlights the importance of maintaining the integrity of the S411 residue since the S411A substitution differs only in the hydroxyl group present in the serine residue. These results are consistent with the model that S411 phosphorylation can modulate the effect on elongation. Notably, the truncated form of the protein did not stimulate elongation, which reinforces the importance of the RGG/RS domain and S411 in the interaction with RNAP II. Finally, the RRM mutant Npl3-120 also has reduced elongation stimulation, suggesting that RNA binding contributes to this activity. We conclude from these experiments that the positive effect of Npl3 on elongation requires both binding to the nascent RNA and its physical interaction with RNAP II, which may be inhibited by phosphorylation of S411.</p>", "<title>Npl3 interacts with phosphorylated serine 2 of the CTD</title>", "<p>The phosphorylated form of serine 2 of the CTD repeat YSPTSPS is associated with elongation ##REF##12942140##[17]##. Previous crosslinking experiments demonstrated the presence of Npl3 throughout elongation ##REF##11459827##[9]##, suggesting a possible link to the CTD and possibly phosphorylated serine 2. An interaction with the CTD was first tested using the RGG/RS peptides. As with purified RNAP II, the unphosphorylated RGG/RS3 was observed to interact with GST-CTD (data not shown). In addition, unphosphorylated or phosphorylated Ser 2, Ser 5 or Ser 2/Ser 5 CTD peptides were used in pull-down experiments with unphosphorylated full-length Npl3 <bold>(</bold>\n##FIG##1##\nFigure 2D\n##\n<bold>)</bold>. The CTD repeat peptide containing the Ser 2 phosphorylation was observed to interact with Npl3, and this result was reproducible. We concluded from these experiments that the phosphorylated form of Ser 2 CTD mediates the physical interaction between unphosphorylated Npl3 and RNAP II.</p>", "<title>Cka1 phosphorylates Npl3 <italic>in vitro</italic>\n</title>", "<p>In yeast, the cytosolic kinase Sky1 phosphorylates the SP motif closest to the C-terminus of Npl3 (S411) ##REF##14759366##[10]##. Phosphorylation of Npl3 at S411 in a <italic>sky1</italic> deletion strain is reduced, but not abolished ##REF##11233987##[8]## (data not shown), raising the possibility that another Npl3-specific kinase remained to be identified, and that other phosphorylation sites in Npl3 might exist. To ascertain the later possibility, endogenous Npl3 was immunoprecipitated from whole cell extracts and phosphorylation sites were analyzed by mass spectrometry. Multiple phosphorylation sites were identified in the endogenous Npl3 <bold>(</bold>\n##TAB##0##\nTable 1\n##\n<bold>)</bold>. Our results confirmed phosphorylation of serine 224, 349 and 356, which where identified previously in two large-scale phosphorylation analyses ##REF##17330950##[34]##, ##REF##11875433##[35]##. One new additional site was shown to be phosphorylated, serine 212. This data is summarized in ##TAB##0##\nTable 1\n##.</p>", "<p>We speculated that a putative kinase might be able to modulate the activity of Npl3 in the nucleus. As noted earlier, CK2 kinase had been shown to phosphorylate several targets within the cleavage/polyadenylation machinery (Cft1, Fip1, Brr5 and Pta1) ##REF##16137619##[24]##, ##REF##16496213##[36]##. Therefore, this kinase was a good candidate and was tested for its ability to phosphorylate Npl3 <italic>in vitro</italic>. Recombinant His-Cka1 and His-Npl3 were purified from <italic>E. coli</italic> and incubated in the presence of [<italic>γ</italic>-³²P] labeled ATP. Cka1 phosphorylates Npl3 and this modification is reversed by addition of Lambda-Phosphatase <bold>(</bold>\n##FIG##2##\nFigure 3A\n##\n<bold>)</bold>. A combination of mass spectrometry and point mutants was used to further analyze phosphorylated peptides in Npl3. <italic>In vitro</italic>, CK2 phosphorylation resulted in modification of multiples sites in Npl3, including Ser 212, Ser 349 and Ser 356 sites identified <italic>in vivo</italic>\n<bold>(</bold>\n##TAB##0##\nTable 1\n##\n<bold>)</bold>. Phosphorylation of S411 was tested using the S411 substitutions (S411A, S411D and S411E) in the kinase assay <bold>(</bold>\n##FIG##2##\nFigure 3B\n##\n<bold>)</bold>. A significant decrease was observed in the phosphorylation of Npl3-S411A, -S411D and -S411E as compared to wild-type. The residual phosphorylation signal observed for the mutant Npl3s is probably due to phosphorylation of the other sites reported in ##TAB##0##\nTable 1\n##. We conclude from these results that Npl3 is hyperphosphorylated <italic>in vivo</italic> and that S411 is likely a major target of CK2 phosphorylation.</p>", "<p>Given that the phosphorylation of sites in Npl3 did not strictly fit the known consensus sequence for mammalian CK2 ([ST]XX[EDpTpS]) ##REF##12631575##[20]##, and that the RGG/RS1-3 peptides were not efficiently phosphorylated when tested in the <italic>in vitro</italic> CK2 kinase assay (data not shown), the specificity of the yeast Cka1 was assayed by mass spectrometry using peptide libraries <bold>(</bold>\n##SUPPL##2##\nFigure S1\n##\n<bold>)</bold>. Quantification by mass spectrometry of the phosphorylation efficiency of Cka1 confirmed the specificity of yeast Cka1 used in our study for peptides with serine or threonine residues within acidic motifs, as previously reported for the mammalian CK2 counterpart ##REF##12631575##[20]##, ##REF##9305961##[37]##. Conversely, peptides containing SP motifs were phosphorylated very poorly <bold>(</bold>\n##SUPPL##2##\nFigure S1\n##\n<bold>)</bold>. In the context of the entire Npl3 protein, the SP or SR motifs of Npl3 that were detected as phosphorylated <italic>in vitro</italic> might reside in an acidic environment that favors phosphorylation by Cka1. This possibility is supported by evidence from the structure of the central RRM domain of Npl3, which revealed that all acidic residues lie on the surface of the protein ##REF##18022637##[33]##. Npl3's reported self-association may also be a factor, especially given recent evidence that dimerization is in some instances required for the phosphorylation of targets with poor similarity to recognition motifs ##REF##15998636##[38]##, ##REF##18239682##[39]##.</p>", "<title>CK2 phosphorylates Npl3 <italic>in vivo</italic>\n</title>", "<p>The ability of CK2 to phosphorylate Npl3 was next tested in the context of the cellular environment using strains that have the genes for both of the catalytic subunits of CK2 deleted and carry complementing plasmids for either a wild-type (<italic>CKA1)</italic> or a mutant allele of <italic>CKA1</italic> (<italic>cka1-12</italic>(<italic>ts</italic>) or <italic>cka1-45</italic>(<italic>ts</italic>)) ##REF##15143168##[40]##. These strains were grown and assayed for the steady state phosphorylation of Npl3 by immunoblotting with general anti-Npl3 and anti-phospho-S411, which was demonstrated previously to be specific for the S411 phosphorylation <italic>in vivo</italic>\n##REF##14759366##[10]##. The wild-type <italic>CKA1</italic> strain showed robust phosphorylation of Npl3 at both permissive (25°C) and non-permissive temperatures (37°C), whereas <italic>cka1-12</italic> showed reduced phosphorylation at both temperatures, and <italic>cka1-45</italic> showed a decrease upon the shift to the non-permissive temperature <bold>(</bold>\n##FIG##2##\nFigure 3C\n##\n<bold>)</bold>. At steady-state-levels, Npl3 localizes to the nucleus ##REF##7969175##[41]##. Therefore, the phosphorylation decrease observed in the <italic>cka1</italic> mutant strains was assumed to be due to CK2 mutations. The residual phosphorylation of S411 is presumed to be due to other kinases, including Sky1. While indirect phosphorylation of Npl3 cannot be absolutely excluded, we conclude from these results that Cka1 is required for the phosphorylation of Npl3-S411 <italic>in vivo</italic>.</p>", "<title>Phosphorylation of Npl3 affects its competition for RNA binding</title>", "<p>As mentioned earlier, phosphorylation of Npl3 by Sky1 reduces RNA binding affinity ##REF##11233987##[8]##. Therefore, Npl3 phosphorylation might be expected to reduce its ability to compete with Rna15. Using a modified UV cross-linking assay with a labeled RNA oligo (N4) thought to be a preferred binding site for Rna15 at <italic>GAL7</italic>\n##REF##11689695##[7]##, recombinant Npl3 and Rna15 were tested for binding. Npl3 was observed to compete strongly for binding to the polyA signal <bold>(</bold>\n##FIG##3##\nFigure 4A\n##\n<bold>)</bold>. Titration of Cka1 resulted in an ATP-dependent reduction of Npl3 binding to the RNA, while binding to an oligo corresponding to an Npl3-preferred site was unaffected (N2) (##FIG##3##\nFigure 4B\n##\n<bold>)</bold>. No cross-linking of Rna15 was detected with the N2 oligo. Rna15 was not phosphorylated by Cka1, nor was binding to the polyA signal affected in reactions with Cka1 lacking Npl3 (data not shown). Thus, we concluded that phosphorylation of Npl3 by Cka1 had a dramatic effect on the RNA binding competition at Rna15-preferred sequences.</p>", "<title>Competition for RNA binding decreases stimulation of elongation</title>", "<p>Previous characterization of alleles of <italic>NPL3</italic> demonstrated that the function of Npl3 in transcription termination relies on its ability to bind RNA ##REF##15902270##[11]##, ##REF##18022637##[33]##. This defect is due to mutations in RRM2, which result in reduced the binding specificity of Npl3 ##REF##18022637##[33]##, ##REF##9199325##[42]##, and reduce the ability of the protein to compete effectively for binding to polyA/termination sequences ##UREF##0##[12]##. In the elongation assay, the RRM mutant Npl3-120 had slightly reduced elongation stimulation, which suggested that RNA binding was also important for elongation <bold>(</bold>\n##FIG##1##\nFigure 2C\n##\n<bold>)</bold>. We exploited the unspecific RNA binding of Rna15 and its competition with Npl3 to test the RNA binding requirement of Npl3, presuming that binding of Rna15 to the nascent RNA might prevent a stable Npl3/RNA interaction. Stimulation of elongation was therefore measured in the presence of the competing polyA factor Rna15. A two-fold molar excess of Rna15 (relative to Npl3) was added together with Npl3 during the 5-min incubation between the pulse and chase. ##FIG##3##\nFigure 4C\n## shows the results for transcription reactions for 0 and 30-minutes after UTP addition. Rna15 alone did not affect the rate of elongation (##FIG##3##\nFigure 4C, lane 8\n##). Npl3 still stimulates elongation in the presence of Rna15, but a decrease in the intensity of the signal for the longer transcripts was seen. Therefore, competition for binding to the RNA transcript between Rna15 and Npl3 may decrease the positive effect on elongation, and strengthens the suggestion that RNA binding plays an important role in the stimulatory effect of Npl3.</p>", "<title>Termination defects associated with Npl3 RNA binding and phosphorylation</title>", "<p>Our previous chromatin immunoprecipitation experiments demonstrated that mutations in the RNA binding domain of Npl3, such as the <italic>npl3-120</italic> allele, lead to more efficient termination and increased cross-linking of polyA/termination factors to the 3′ end of various genes ##REF##15902270##[11]##. Phosphorylation of Npl3 reduces its association with RNA <italic>in vivo</italic>\n##REF##14759366##[10]##, and its association with an Rna15 preferred binding site <italic>in vitro</italic>\n<bold>(</bold>\n##FIG##3##\nFigure 4A\n##\n<bold> and</bold>\n##UREF##0##[12]##. Our results further suggest it is also important for regulating its interaction with RNAP II. Therefore, the expectation was that a mutation that abolishes S411 phosphorylation might result in a termination defect. The S411A mutation was selected for analysis since it more closely resembles an unphosphorylated state, although slight decreases in the elongation and binding assays were detected upon loss of the serine hydroxyl group. The predicted outcome <italic>in vivo</italic> was that a sustained interaction between unphosphorylated Npl3, the polymerase and RNA, would result in failure of RNAP II to release and terminate, and a more effective competition with Rna15 for recognition of termination/polyA signals, based on the assumption that both mechanisms are coupled <italic>in vivo</italic>. This prediction was tested by RNA expression analysis using whole genome tiled arrays consisting of 25-mer reverse probes with 5 base pair spacing. RNA was extracted from <italic>NPL3, npl3-120</italic> and <italic>npl3-S411A</italic> strains, followed by synthesis of labeled sense-strand cDNA and hybridization to the tiled arrays ##UREF##3##[43]##. The intensities of two independent experiments were used to plot the ratio of mutant to wild-type.</p>", "<p>Shown in ##FIG##4##\nFigure 5A–B\n## are plots for three representative genes (<italic>FBA1</italic>, <italic>MPE1</italic> and <italic>TDH3</italic>). The ratio of mutant to wild-type expression levels showed a clear difference at the 3′ ends. For the <italic>npl3-120</italic> strain, a decrease in transcript levels near the 3′end of the genes was observed <bold>(</bold>\n##FIG##4##\nFigure 5\n##\n<bold>and </bold>\n##SUPPL##3##\nFigures S2\n##\n<bold>, </bold>\n##SUPPL##4##\nS3\n##\n<bold> and </bold>\n##SUPPL##5##\nS4\n##\n<bold>)</bold>, consistent with previous experiments showing that this mutation leads to more efficient termination ##REF##15902270##[11]##. In contrast, the <italic>npl3-S411A</italic> mutant had the opposite effect, showing increased hybridization at the 3′ end, indicating the presence of readthrough transcription beyond the polyA site. A list of genes with a similar transcription readthrough defect for the <italic>npl3-S411A</italic> strain and additional sample plots are included in ##SUPPL##1##\nTable S2\n##\n<bold> and </bold>\n##SUPPL##3##\nFigures S2\n##\n<bold>, </bold>\n##SUPPL##4##\nS3\n##\n<bold> and </bold>\n##SUPPL##5##\nS4\n##. For reference, the plot corresponding to the ratio of the <italic>rna15-2 vs.</italic> wild-type is included. The <italic>rna15-2</italic> strain has a well-characterized termination defect that results in readthrough transcription ##REF##9535662##[44]##–##REF##12086625##[46]##. At the <italic>FBA1</italic> gene, changes were observed in the <italic>npl3-S411A</italic> and <italic>rna15-2</italic> mutants at both ends of the gene denoting a terminating effect in the flanking <italic>MPE1</italic> gene <bold>(</bold>\n##FIG##4##\nFigure 5A\n##\n<bold>)</bold>. For two of the examples shown (<italic>FBA1</italic> and <italic>MPE1</italic>), the ratio of <italic>rna15-2</italic> strain was similar to that of the <italic>npl3-S411A</italic>\n<bold>(</bold>\n##FIG##4##\nFigure 5A\n##\n<bold>)</bold>. The <italic>PDX1</italic> gene lying upstream of <italic>TDH3,</italic> had a clear termination defect in the <italic>rna15-2</italic> strain, but was unaffected by the <italic>S411A</italic> or <italic>npl3-120</italic> mutations <bold>(</bold>\n##FIG##4##\nFigure 5B\n##\n<bold>)</bold>. This suggests that the competition between Npl3 and Rna15 might be dependent on <italic>cis-</italic>elements; <italic>i.e.,</italic> strong polyA signals may override the competition ##UREF##0##[12]##. The <italic>npl3-S411A</italic> mutation resulted clearly in readthrough transcription of &gt;800 genes <bold>(</bold>\n##SUPPL##1##\nTable S2\n##\n<bold>)</bold>. A scatter-plot is provided for these genes using the mean value for the log₂ ratio of <italic>npl3-S411A vs.</italic> wild-type for a 300 bp region within the ORF and at the 3′UTR of each gene <bold>(</bold>\n##FIG##4##\nFigure 5C\n##\n<bold>)</bold>. An increase in mean value of the probes was observed at the 3′UTR. A scatter-plot in ##FIG##4##\nFigure 5D\n## was also generated for the total number of genes in the Crick strand. The majority of these genes showed an increase mean value at the 3′UTR relative to the region within the ORF for the <italic>npl3-S411A</italic> mutant <bold>(</bold>\n##FIG##4##\nFigure 5D\n##\n<bold>)</bold>. Our results are consistent with the involvement of Npl3 in a widely distributed mechanism of transcription termination.</p>" ]

[ "<title>Discussion</title>", "<p>Termination of transcription is a critical mechanistic step in regulating gene expression, particularly for genes with multiple polyadenylation sites. Here we report that the SR/hnRNP protein Npl3 directly enhances the elongation rate of RNAP II. Although co-transcriptional recruitment of eukaryotic RNA binding proteins is now well established, to our knowledge a direct stimulation of elongation by SR or hnRNP proteins has not been previously reported. No such stimulation is seen with the RNA binding protein Rna15 or the truncated Npl3 RRM domain, arguing that the Npl3 effect is not due to non-specific interaction with RNA.</p>", "<p>The data presented is consistent with a model where Npl3 functions in two ways: it directly interacts with phosphorylated serine 2 of the polymerase CTD to promote elongation and at the same time binds RNA to antagonize binding of the polyA/termination factor Rna15 <bold>(</bold>\n##FIG##5##\nFigure 6\n##\n<bold>)</bold>\n##REF##11459827##[9]##, ##REF##15902270##[11]##. This combination of activities resembles the l-related phage 82 anti-terminator Q⁸² that, together with the NusA protein, decreases pausing of the bacterial RNA polymerase and protects the RNA transcript from termination factors ##REF##17889665##[47]##.</p>", "<p>Phosphorylation of Npl3 reduces its interaction with RNAP II and RNA. It is interesting to speculate that the targeted phosphorylation of Npl3 at 3′ ends of genes might help trigger termination. This model is supported by chromatin immunoprecipitation experiments where crosslinking of Npl3 was observed to decrease downstream of the polyA site ##REF##15902270##[11]##, by the expression analysis presented here where mutations at S411 or in the second RRM of Npl3 were shown to specifically affect transcript signals at 3′ ends of genes, and by UV-crosslinking where the Cka1 phosphorylation decreased Npl3's competition for binding to polyA signals. In addition, a recent report by Lund <italic>et al</italic> suggests that phosphorylation is involved the autoregulation of the <italic>NPL3</italic> transcript ##UREF##2##[32]##. The cytosolic kinase Sky1 phosphorylates Npl3 ##REF##11233987##[8]##. Thus far, no other kinase had been linked to the phosphorylation of Npl3. CK2 is the only kinase reported to phosphorylate proteins involved in mRNA 3′ end formation. Phosphorylation of 3′ end processing factors Pta1 and Fip1 by CK2 affects their ability to cleave and polyadenylate the mRNA ##REF##16137619##[24]##. Therefore, the phosphorylation of Npl3 by CK2 shown here would support a targeted phosphorylation at the 3′ end. The existence of nuclear and cytosolic kinases that phosphorylate an mRNA shuttling protein at an identical residue suggests that the same phosphorylation switch, which functions by modulating protein-protein and -RNA interactions can be activated in different compartments of the cell. It remains to be determined what other kinases operate at the 3′ end or whether additional transcription-associated kinases can control the phosphorylation state of Npl3 to regulate its activity at sites of transcription.</p>", "<p>The model postulates several events that would affect assembly of the polyA/termination machinery at the 3′ end <bold>(</bold>\n##FIG##5##\nFigure 6\n##\n<bold>)</bold>. First, upon dissociation of Npl3, the elongation rate of RNAP II would decrease to enhance the window of opportunity for polyA factors to bind to signals in the nascent RNA. It is known that the rate of transcription elongation can affect alternative splicing decisions ##REF##17028590##[48]##, ##REF##15383674##[49]##, so it is reasonable to assume that a similar relationship can exist between elongation rate and polyadenylation. Interestingly, the elongation promoting activity of Npl3 is weaker in an RNA binding mutant and is inhibited by Rna15, which suggests that binding to the RNA transcript by Npl3 is important for stimulating RNAP II. Binding of Npl3 to the nascent transcript might stabilize interactions between the CTD, RNAP II and the RNA and prevent premature 3′ end formation. Accordingly, interactions between the CTD of the mammalian RNAP II with the RNA were shown to be critical to suppress premature termination ##REF##16209948##[50]##. Additionally, binding of Npl3 to the RNA may prevent the formation of arrest-inducing secondary structures, or suppress pausing at cryptic polyA sites.</p>", "<p>A second event upon dissociation of Npl3 would be a shift in the balance of the competition between Npl3 and Rna14/15 for RNA-binding. Several lines of evidence have shown that this competition is important for proper polyadenylation. Transcription readthrough of weakened polyA signals can be suppressed by mutation/deletion of Npl3 or Cbp20/80, or by overexpression of the polyadenylation factors Rna14 or Hrp1 ##REF##15902270##[11]##–##UREF##1##[13]##. This competition between Npl3 and Rna15 can be reconstituted <italic>in vitro</italic>\n##UREF##0##[12]##, and here we show that this competition can be shifted by the CK2 phosphorylation of Npl3.</p>", "<p>There are many examples of SR proteins being regulated by phosphorylation. In the case of Srp40, phosphorylation of the RS domain is required for sequence-specific RNA binding ##REF##9037021##[51]##. For another RS protein, ASF/SF2, phosphorylation enhances contacts essential for splicing ##REF##11233987##[8]##, ##REF##9030686##[30]##, ##REF##16766678##[52]##–##REF##17188683##[55]##. The identification by mass spectrometry of multiple phosphorylated serines <italic>in vivo</italic> and <italic>in vitro</italic> suggests that Npl3 is hyperphosphorylated at the RRM and RGG/RS domains. In the case of Npl3-S411, its phosphorylation decreases crosslinking of the protein to RNA ##REF##11233987##[8]##(this study), and also functions in autoregulation of the <italic>NPL3</italic> transcript ##UREF##2##[32]##. Our study suggests that CK2 is required for the phosphorylation of S411, and we also show that this residue appears to affect Npl3's interaction with RNAP II. Thus, S411 participates in the binding for both RNA and RNAP II.</p>", "<p>While this manuscript was in preparation Lund <italic>et al</italic> reported of an effect of Npl3 in autoregulation of its own transcript ##UREF##2##[32]##. In this study, phosphorylation of Npl3-S411 was observed to increase termination efficiency, thus modulating its own protein levels. Our expression analysis suggests that the termination of <italic>NPL3</italic> represents a unique case likely shared by a limited number of genes. Previous work by Steinmetz and co-workers suggested that the choice of termination sites at the <italic>NPL3</italic> locus involved a Sen1 anti-termination pathway ##REF##17157256##[56]##. In addition, the phosphorylation decrease observed for S411 in the <italic>cka1</italic> alleles reported in this study did not significantly affect the steady state levels of Npl3, as observed by Lund and co-workers. Therefore, we believe that the autoregulation of Npl3 is likely the result of an activity that is distinct from that mediated by CK2. At this time, the mechanism by which Npl3 functions in its autoregulation by allowing the selection of an alternative termination site remains unresolved.</p>", "<p>Our study suggests that Npl3 affects elongation and termination in a high percentage of genes (∼30%) making this a widely used mechanism in yeast that is likely to function in higher eukaryotes. We predict that metazoan SR-proteins with known functions in hnRNP formation, splicing and mRNA 3′end processing might also be able to affect elongation in a manner similar to Npl3.</p>" ]

[]

[ "<p>Conceived and designed the experiments: MEB. Performed the experiments: JLD JV BO MEB. Analyzed the data: JLD JD JV MEB. Contributed reagents/materials/analysis tools: JD JV SPG PJP ASP CM SB MEB. Wrote the paper: JLD MEB. ##FIG##1##Figures 2## &amp; ##FIG##3##4##: JLD. Oligo(dC)-tail template elongation assay: JLD. ##FIG##4##Figure 5##, Expression analysis: JD PJP. ##SUPPL##3##Figure S2##, ##SUPPL##4##S3## and ##SUPPL##5##S4##, Expression analysis: JD PJP. ##SUPPL##2##Figure S1##: JV SPG. Mass spectrometry analysis of Cka1 phosphorylation specificity using peptide libraries: JV. Analysis of Npl3 phosphorylation sites: JV. Editing of paper: JV. ##FIG##2##Figure 3##: BO. Cka1 phosphorylation in vitro analysis: BO. Cka1 phosphorylation in vivo analysis: BO. Mass spectrometry analysis of Cka1 phosphorylation specificity using peptide libraries: SPG. Analysis of Npl3 phosphorylation sites: SPG. ##FIG##1##Figure 2##: ASP. RNAP II purification: ASP. Editing of paper and guidance: CM SB.</p>", "<p>The production of a functional mRNA is regulated at every step of transcription. An area not well-understood is the transition of RNA polymerase II from elongation to termination. The <italic>S. cerevisiae</italic> SR-like protein Npl3 functions to negatively regulate transcription termination by antagonizing the binding of polyA/termination proteins to the mRNA. In this study, Npl3 is shown to interact with the CTD and have a direct stimulatory effect on the elongation activity of the polymerase. The interaction is inhibited by phosphorylation of Npl3. In addition, Casein Kinase 2 was found to be required for the phosphorylation of Npl3 and affect its ability to compete against Rna15 (Cleavage Factor I) for binding to polyA signals. Our results suggest that phosphorylation of Npl3 promotes its dissociation from the mRNA/RNAP II, and contributes to the association of the polyA/termination factor Rna15. This work defines a novel role for Npl3 in elongation and its regulation by phosphorylation.</p>" ]

[ "<title>Supporting Information</title>" ]

[ "<p>We are grateful to D. Brow J. Manley, and B. Schwartz for comments. We thank M. Lund/C. Guthrie lab for yeast strain; L. Vasilieva, Bret Redwine, the Conaway lab, and S. Annis (HMS Biopolymers Facility) for technical advice; and K. Struhl and E. Harlow for their support.</p>" ]

[ "<fig id=\"pone-0003273-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003273.g001</object-id><label>Figure 1</label><caption><title>Npl3 stimulates RNAP II elongation.</title><p>Oligo(dC)-tail templates were used in transcription reactions with or without 78 nM Npl3, as indicated. Reactions were pulsed by the addition of ATP, GTP, and <italic>α</italic>\n³²P-labeled CTP for 30 minutes as described in <xref ref-type=\"sec\" rid=\"s4\">Materials &amp; Methods</xref>. Transcripts were chased by the addition of excess CTP and limiting UTP for the indicated times. A schematic of the experimental scheme is shown on top. The oligo(dC)-template is represented next to the gel and the positions of the pause sites at stretches of Ts are indicated.</p></caption></fig>", "<fig id=\"pone-0003273-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003273.g002</object-id><label>Figure 2</label><caption><title>Npl3 interaction with RNAP II is affected by S411 phosphorylation.</title><p>(A) Unphosphorylated peptide RGG/RS1 (lane 1), RGG/RS2 (lane 2), RGG/RS3 (lane 3) and phosphorylated RGG/RS3 (lane 4) were used in pull-down assays with purified RNAP II, as indicated, and one fifth of the input is shown in lane 5. The precipitated sample (P) and supernatant (S) were analyzed by immunoblotting. (B) Full-length Npl3 or S411 mutants were used in immunoprecipitation assays with purified RNAP II (100 ng), as indicated. A control lane with RNAP II (10 ng) is shown in lane 13. (C) Oligo(dC)-tail transcription reactions were performed as described in ##FIG##0##Figure 1## for 30 minutes with RNAP II only (lane 1) or with equivalent concentrations of Npl3 (lane 2); Npl3-S411A (lane 3); -S411D (lane 4); -S411E (lane 5); Npl3-truncated (RRMs only, aa 121–280) (lane 6); or Npl3-120 (lane 7). Quantification of the transcription reactions was performed by calculating the ratio of the 450 nt to the 250 nt bands and normalized using the RNAP II only lane, as shown in the graph. The experiment was done twice, and a representative gel was chosen. (D) Unphosphorylated (lane 1 and 6), or phosphorylated Ser 5 (lane 2 and 7), Ser 2 (lane 3 and 8), and Ser 2/Ser 5 (lane 4 and 9) CTD peptides were used in pull-down assays with full-length Npl3, as indicated, and input is shown in lane 11. The IP and S shown were analyzed as described for (C) using antibodies specific for Npl3. Control Npl3 with no peptide is shown in lanes 5 and 10.</p></caption></fig>", "<fig id=\"pone-0003273-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003273.g003</object-id><label>Figure 3</label><caption><title>Cka1, the alpha catalytic subunit of CK2, phosphorylates Npl3.</title><p>(A) Recombinant His-Cka1 was incubated with His-Npl3 in the presence of radiolabeled ATP. Phosphorylation of Npl3 is reversed by the addition of increasing concentrations of <italic>λ</italic>-Phosphatase. (B) Using the <italic>in vitro</italic> kinase assay with mutants of Npl3-S411, this residue was uncovered as an additional phosphorylation site. Wild-type His-Npl3 or S411 point mutants, His-Npl3-S411A, -S411D or -S411E were incubated with His-Cka1, as described for (A) in the presence of radiolabeled-ATP. Coomasie stain representing the concentration of recombinant Npl3 proteins used is shown below. (C) <italic>CKA1</italic> mutations show reduced phosphorylation of Npl3. Whole-cell extracts were prepared for cells grown for one hour at the permissive (25°C) or non-permissive (37°C) temperature for wild-type <italic>CKA1, cka1-12</italic> or <italic>cka1-45</italic>, and immunoblot analysis was performed using antibodies specific for phosphorylated or non-phosphorylated Npl3, as indicated. Detection of Rpb3 with specific antibodies is shown as a loading control.</p></caption></fig>", "<fig id=\"pone-0003273-g004\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003273.g004</object-id><label>Figure 4</label><caption><title>Cka1 disrupts the ability of Npl3 to effectively compete for binding to an Rna15-preferred sequence.</title><p>(A) Recombinant Npl3, Rna15 and Cka1 were incubated with a radiolabeled RNA oligo (N4), UV cross-linked, and resolved in denaturing 10% SDS-PAGE gels. This RNA oligo consists of an A-rich repeat (described in <xref ref-type=\"sec\" rid=\"s4\">Materials and Methods</xref>), which is the preferred binding site for Rna15 and is commonly found upstream of polyA sites ##REF##11689695##[7]##. Representative UV-cross-linking experiments are shown where increasing Cka1 is added to reactions containing Npl3 with Rna15. The graph below each gel shows quantification for the average of three experiments (Npl3, white bars; Rna15, black bars). Binding levels were calculated as fractions relative to a reaction containing the highest concentration of the individual RNA-bound protein (lane 1 for Npl3 and lane 2 for Rna15). Control UV-crosslinking experiment where Cka1 has been added with or without ATP is also shown. Values represent total PhosphorImager units (PIU). (B) Recombinant Npl3, Rna15 and Cka1 were incubated with a radiolabeled U/G/C-rich RNA oligo (N2), UV-cross-linked, and resolved in denaturing 10% SDS-PAGE gels and quantified as described for (A). (C) Similar reactions were performed as described for ##FIG##0##Figure 1## for 0 and 30 minutes without (lane 1 and 5) or with 78 nM Npl3 (lanes 2–3 for 0 min time-point; and 6–7 for the 30 min time-point). 200 nM Rna15 was added to the transcription reactions in the presence or absence of Npl3 (lanes 3–4 for 0 min time-point; and 7–8 for the 30 min time-point). The oligo(dC)-template is represented next to the gel and the positions of the pause sites at stretches of Ts are indicated. Quantification of the transcription reactions was performed as shown for ##FIG##1##Figure 2C##.</p></caption></fig>", "<fig id=\"pone-0003273-g005\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003273.g005</object-id><label>Figure 5</label><caption><title>Opposing effects in termination demonstrated using RNA-binding or phosphorylation defective <italic>npl3</italic> alleles.</title><p>RNA was extracted from <italic>NPL3, npl3-120</italic> and <italic>npl3-S411A</italic> and whole transcript sense strand cDNA was synthesized and used for hybridization to tiled arrays. The hybridization signals for (A) <italic>FBA1</italic>, <italic>MPE1</italic> and (B) <italic>TDH3</italic> for two independent experiments were used to calculate the ratio of mutant <italic>vs.</italic> wild-type. The top panel shows the ratio of <italic>rna15-2/RNA15;</italic> bottom panel shows the ratio of each <italic>npl3</italic> mutant<italic>/NPL3</italic>; a solid line for <italic>npl3-120/NPL3</italic>, or dashed line for <italic>npl3-S411A/NPL3.</italic> The corresponding position and orientation (Watson (W+), dashed bars, or Crick (C-) strand, black solid bars) of the genes is also shown. (C–D) A 300 bp region at the beginning of each ORF (+50 to +350 relative to the start codon), was selected for comparison to the same size region at the 3′UTR (−100 to +200 relative to the stop codon) of the genes that showed increase readthrough for the <italic>npl3-S411A</italic> strain. The mean value of the log₂ ratio of <italic>npl3-S411A/NPL3</italic> was calculated and graphed in a scatter plot. (D) The total number of genes corresponding to the Crick strand were used to calculate the mean value of the same region described for (C).</p></caption></fig>", "<fig id=\"pone-0003273-g006\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003273.g006</object-id><label>Figure 6</label><caption><title>Two activities of Npl3 promote elongation.</title><p>Npl3 directly stimulates the elongation rate of RNAP II by physically interacting with phospho-Ser 2 of the CTD and the RNA. Phosphorylation regulates Npl3's interaction with RNAP II and RNA, and promotes binding of polyadenylation/termination factors. Binding of Npl3 to the nascent RNA may stabilize interactions between CTD-RNAP II and the RNA. Npl3, white circle; Rna15, small black diamond; CFI, large black diamond; RNAPII, grey shaded oval.</p></caption></fig>" ]

[ "<table-wrap id=\"pone-0003273-t001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003273.t001</object-id><label>Table 1</label><caption><title>Npl3 phosphorylation sites.</title></caption><alternatives><table frame=\"hsides\" rules=\"groups\"><colgroup span=\"1\"><col align=\"left\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/><col align=\"center\" span=\"1\"/></colgroup><thead><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">Npl3 (414 aa)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Phosphorylated Serines</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Treatment</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Analysis</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Reference</td></tr></thead><tbody><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>Endogenous</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>Ser212,</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">WCE (whole cell extract)</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">MS</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">This study</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Ser224</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">WCE</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">MS-IMAC</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">This study, ##REF##17330950##[34]##\n</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>Ser349, Ser356</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">WCE</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">MS-IMAC</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">(Ficarro <italic>et al</italic>, 2002)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Ser411</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n³²P-<italic>in vivo</italic> labeling</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">2D gel</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">(Gilbert <italic>et al</italic>, 2001)</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>CK2 in vitro phosphorylated</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Ser77, Ser79,</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">Kinase assay</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">MS</td><td align=\"left\" rowspan=\"1\" colspan=\"1\">This study</td></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>Ser212</bold>, Ser230,</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">“</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Ser328, Ser336,</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">“</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">\n<bold>Ser349, Ser356</bold>\n</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">“</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/></tr><tr><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Ser411</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/><td align=\"left\" rowspan=\"1\" colspan=\"1\">Point mutants</td><td align=\"left\" rowspan=\"1\" colspan=\"1\"/></tr></tbody></table></alternatives></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"pone.0003273.s001\"><label>Table S1</label><caption><p>Yeast strains used in this study.</p><p>(0.04 MB DOC)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003273.s002\"><label>Table S2</label><caption><p>Genes with increased Log2 Ratio at their 3′UTR.</p><p>(1.47 MB DOC)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003273.s003\"><label>Figure S1</label><caption><p>Distribution of the log2 ratios of the peptide libraries phosphorylated in vitro with Cka1 and measured by MS. Kinase assays utilizing peptide libraries representing acidic (L[LE]D[KDN]D[DA][LE][ST]D[EL]E[LEN][EL]K) and proline-directed ([KP]L[VKE]L[AP][NE][ST]P[KI][LKP]VV[KL]) motifs were performed. 20 µg of the non-phosphorylated libraries were used, and 0.2 µg of the heavy phosphorylated standard library was added upon reaction quenching with 0.15 % TFA. Reactions were desalted in a tC18 SepPak and enriched for phosphopeptides with PhosSelect IMAC resin (Sigma). Samples were desalted again prior to injection into a capillary (125 µm×18 cm) C18 column and analyzed by LC-MS/MS in a LTQ-Orbitrap mass spectrometer using a 60 minute-gradient and data dependent TOP10 method. MS/MS spectra were searched against a database containing the sequences for all the peptides in the libraries and all the sequences from E. coli protein sequence database (used for distraction purposes) in the forward and reverse directions. Results were filtered to &lt;1% false positives and peptides were further quantified using the VistaQUANT algorithm.</p><p>(0.07 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003273.s004\"><label>Figure S2</label><caption><p>Opposing effects in termination demonstrated using RNA-binding or phosphorylation defective npl3 alleles. RNA was extracted from NPL3, npl3-120 and npl3-S411A and whole transcript sense strand cDNA was synthesized and used for hybridization to tiled arrays. Shown are the hybridization signals for twelve genes. The three top panels represent the individual expression intensities for each strain, with the black and gray dotted lines showing two independent experiments. The bottom panel shows the ratio of each mutant vs. wild-type; a solid line for npl3-120 vs. NPL3, or dashed line for npl3-S411A vs. NPL3, and the corresponding position and orientation (Watson (W+), shown for reference only, or Crick (C-) strand) of the genes (black solid bars).</p><p>(0.17 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003273.s005\"><label>Figure S3</label><caption><p>Opposing effects in termination demonstrated using RNA-binding or phosphorylation defective npl3 alleles. RNA was extracted from NPL3, npl3-120 and npl3-S411A and whole transcript sense strand cDNA was synthesized and used for hybridization to tiled arrays. Shown are the hybridization signals for twelve genes. The three top panels represent the individual expression intensities for each strain, with the black and gray dotted lines showing two independent experiments. The bottom panel shows the ratio of each mutant vs. wild-type; a solid line for npl3-120 vs. NPL3, or dashed line for npl3-S411A vs. NPL3, and the corresponding position and orientation (Watson (W+), shown for reference only, or Crick (C-) strand) of the genes (black solid bars).</p><p>(0.18 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003273.s006\"><label>Figure S4</label><caption><p>Opposing effects in termination demonstrated using RNA-binding or phosphorylation defective npl3 alleles. RNA was extracted from NPL3, npl3-120 and npl3-S411A and whole transcript sense strand cDNA was synthesized and used for hybridization to tiled arrays. Shown are the hybridization signals for twelve genes. The three top panels represent the individual expression intensities for each strain, with the black and gray dotted lines showing two independent experiments. The bottom panel shows the ratio of each mutant vs. wild-type; a solid line for npl3-120 vs. NPL3, or dashed line for npl3-S411A vs. NPL3, and the corresponding position and orientation (Watson (W+), shown for reference only, or Crick (C-) strand) of the genes (black solid bars).</p><p>(0.17 MB TIF)</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><fn id=\"nt101\"><p>Phosphorylation sites identified in Npl3.</p></fn><fn id=\"nt102\"><label>*</label><p>Highlighted residues where phosphorylation sites detected in both the endogenous and <italic>in vitro</italic> CK2 phosphorylated Npl3.</p></fn></table-wrap-foot>", "<fn-group><fn fn-type=\"COI-statement\"><p><bold>Competing Interests: </bold>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p><bold>Funding: </bold>This work was supported by NIH grants 1K01CA115515-01A1 to M.B., GM51124 to A.S.P., GM46498 and GM56663 to S.B., GM68887 to C.M., and fellowship 5F31GM075383-03 to J.D.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"pone.0003273.g001\"/>", "<graphic xlink:href=\"pone.0003273.g002\"/>", "<graphic id=\"pone-0003273-t001-1\" xlink:href=\"pone.0003273.t001\"/>", "<graphic xlink:href=\"pone.0003273.g003\"/>", "<graphic xlink:href=\"pone.0003273.g004\"/>", "<graphic xlink:href=\"pone.0003273.g005\"/>", "<graphic xlink:href=\"pone.0003273.g006\"/>" ]

[ "<media xlink:href=\"pone.0003273.s001.doc\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003273.s002.doc\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003273.s003.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003273.s004.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003273.s005.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003273.s006.tif\"><caption><p>Click here for additional data file.</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>As professional antigen-presenting cells of the immune-system, DCs are instrumental in the initiation of adaptive immunity by presenting antigen to T cells ##REF##15771591##[1]##. Immature DCs endocytose soluble molecules and particulate material from the surrounding medium via macropinocytosis and adsorptive endocytosis, and then migrate from peripheral tissues to draining lymph nodes ##REF##10426272##[2]##. Antigenic proteins are processed into peptides, and loaded to major histocompatibility (MHC) molecules class I and class II. In immature DCs, MHC-II is present in late endosomes or lysosomes ##REF##10426272##[2]##–##REF##9286358##[5]##. However, with time, MHC-II molecules are trafficked via class-II vesicles towards the plasma membrane, where the pMHC-II is presented to T cells ##REF##9285592##[4]##. DC maturation, characterized by reduced endocytosis and by upregulation of the cell-surface activation markers CD80 and CD86, is promoted by “danger” signals or ligands for toll-like receptors ##REF##15728447##[6]##. Although the chemistry and subcellular distribution of compartments related to the antigen-presentation pathway in DCs have been studied extensively, the real-time dynamics of early endocytic vesicles in DCs have not yet been characterized.</p>", "<p>Previous investigations have established two-photon microscopy as an elegant method for imaging deep inside a living tissue or organ for long periods (hours) ##UREF##0##[7]##. The spatial and temporal resolution achieved by two-photon microscopy is unparalleled by other methods of non-optical deep-tissue imaging, including PET and FMRI ##REF##12415310##[8]##. Previous imaging studies in the lymph node utilizing fluorescent ‘cell-tracker’ dyes have visualized skin-draining DCs carrying cognate antigen, and the associated dynamic changes in T cell motility induced by DC contact ##REF##15466619##[9]##–##REF##15357946##[12]##. Initially, T cells make transient contacts (stochastic scanning) with antigen-presenting DCs, and their behavior then progresses from serial interactions to formation of dynamic T cell clusters, T cell dissociation and swarming, leading ultimately to several rounds of T cell proliferation.</p>", "<p>Quantum dots (QDs) are semiconductor nanocrystals that exhibit very bright and photostable fluorescence. QDs have broad absorption spectra, and narrow Gaussian emission spectra with peak emission wavelengths that depend upon their size ##REF##17237532##[13]##, ##UREF##1##[14]##. They are available encapsulated with biopolymeric coatings that allow conjugation with biomolecular moieties, and have been used for imaging in several <italic>in vitro</italic> and <italic>in vivo</italic> biological systems ##REF##17237532##[13]##, ##REF##17388641##[15]##–##REF##16512782##[17]##. When observed by two-photon microscopy QD fluorescence is 100–1,000 times brighter than conventional fluorophores, and is maintained for long periods of time without photobleaching. Proteins can be stably conjugated to the surface of encapsulated QDs by charge-charge interaction, or by using linkers such as streptavidin and maltose-binding protein. In addition to fluorescence imaging, individual QDs can be observed using electron microscopy, and correlated with fluorescence microscopy ##REF##16004566##[18]##–##REF##14733586##[20]##. Although QDs have been used for high-resolution imaging in biological systems, they have not yet been used for simultaneous modulation of biological function.</p>", "<p>We have harnessed the fluorescence and bioconjugation potential of QDs for real-time imaging of dendritic cells (DCs) <italic>in vitro</italic>, for deep tissue imaging of DCs <italic>in vivo</italic>, and for concomitant modulation of DCs to elicit an immune response to a specific antigen both <italic>in vitro</italic> and <italic>in vivo</italic>. The broad excitation spectra and narrow emission spectra of QDs make them ideally suited for two-photon fluorescence excitation and color separation from other probes. In addition, the high two-photon cross-section of QDs greatly increases their signal to noise ratio as compared to fluorescent proteins. These two properties facilitate detection of QD labeled cells by two-photon microscopy during real-time tissue imaging. Moreover, the ability to conjugate QDs with bioactive proteins makes them potentially useful for modulating biological responses and for cell tracking. Here, we first illustrate the use of QDs for <italic>in vitro</italic> imaging of intracellular compartments inside DCs following endocytosis. We then demonstrate that QDs facilitate real-time tissue imaging of DCs inside the lymph node. Finally, we show that quantum dots can be used effectively as a particle-based antigen delivery system, employing specific antigenic proteins conjugated to QDs to trigger T cell activation with high efficiency in two transgenic TCR models.</p>" ]

[ "<title>Methods</title>", "<title>Mice</title>", "<p>BALB/c (I-A^d), C57BL/6J (I-A^b), and congenic CD45.1-expressing BL6/SJL (C57BL/6) mice were purchased from Jackson Laboratory. Ova peptide^323–339/I-A^d-specific TCR-transgenic DO11.10 mice (KJ1.26⁺) were progeny of homozygotic DO11.10 parents purchased from Jackson Laboratory. Ova peptide^323–339/I-A^b-specific TCR-transgenic OT-II mice were progeny of homozygotic OT-II parents purchased from Jackson Laboratory. Mice expressing enhanced yellow fluorescent protein (EYFP) on CD11c promoter were a kind gift from M. Nussensweig ##REF##15543150##[34]##. These mice were back-crossed for 10 generations to a C57BL/6 background. Mice were housed in a pathogen-free animal facility and all procedures were performed in accordance with protocols approved by the institutional animal care and use committee of UCI.</p>", "<title>Conjugation of ovalbumin to quantum dots</title>", "<p>Ovalbumin was biotinylated using EZ-Link™ Sulfo-NHS-Biotin (Pierce™). Excess biotin and salts were separated by gel filtration using D-Salt™ dextran desalting columns (Pierce™). Biotinylation of ovalbumin was verified by Western blotting. Quantum dot (QD) 655-streptavidin conjugates (100 nmol; Quantum Dot Corp./Invitrogen™) were mixed with biotinylated ovalbumin at a molar ratio of 1∶10 in 2 ml PBS at 6°C for 2 hr. Excess ovalbumin was removed by at least 4 rounds of ultrafiltration using Amicon Ultra™ 4 100,000 MWCO ultrafiltration units (Millipore Corp.). Each QD-streptavidin conjugate contains an average of 10 molecules of surface-bound streptavidin (Quantum Dot Corp.). Thus, a 1 nM solution of QD-streptavidin conjugates bound to biotinylated ovalbumin, as used for priming DCs <italic>in vitro</italic>, contains 10 nM ovalbumin, which is equivalent to 0.5 µg/ml of ovalbumin.</p>", "<title>\n<italic>In vitro</italic> dendritic cell culture</title>", "<p>Dendritic cells were cultured from tibial and femoral bone marrow extracts of 8–12 week old mice as described ##REF##1460426##[39]##. Briefly, the bone marrow extracts were cultured in non TC-treated polystyrene culture dishes (Corning™) using IMDM (Lonza™) substituted with 10% fetal calf serum (FCS; Hyclone Inc.), ∼1,000 units/ml (20 ng/ml) recombinant mouse granulocyte/macrophage colony stimulating factor (GM-CSF; Pharmingen™), 100 units/ml penicillin, and 100 µg/ml streptomycin. DCs were harvested between six and ten days of culture and used in experiments. For flow cytometry of QD⁺ cells, harvested DCs were incubated in QD-containing medium (2 nM QD in RPMI substituted with 10% FCS), for 30 min at 37°C, and subsequently labeled with FITC-conjugated anti-mouse CD11c (Pharmingen™), PE-conjugated anti-mouse CD11b (Ebioscience™)To induce maturation, DCs were incubated with 1 µg/ml lipopolysachharide (LPS) (Sigma-Aldrich Inc.) for 12–16 hr, and maturation was verified by upregulation of MHC class-II (PE-conjugated anti-mouse MHC-II; Pharmingen™) and CD86 (PE-conjugated anti-mouse CD86; Pharmingen™).</p>", "<title>\n<italic>In vitro</italic> imaging of DCs, quantification of QD uptake, and vesicular dynamics</title>", "<p>For <italic>in vitro</italic> imaging, bone marrow-derived DCs were plated on cover glass chambers, and incubated with 2 nM QD-containing medium. Unless otherwise mentioned, temperature was maintained at 37°C by perfusing temperature-controlled PBS below the cover slip. For assessment of QD-toxicity in DCs propidium iodide (Calbiochem™) staining was used. Fluorescence and DIC images were acquired with a Zeiss Axiovert 35 microscope using a 40× 1.30 n.a. oil-immersion objective, equipped with a temperature controller. Higher resolution images were acquired with a custom-built video-rate two-photon microscope based on a Olympus BX50 confocal laser-scanning system, equipped with a titanium-sapphire femtosecond laser (Tsunami, Spectra-Physics) tuned to 780 nm, photomultiplier tubes for detection, and a 60× 1.10 NA water-immersion objective as previously described ##REF##11728133##[40]##. Every image acquired was an average of 15 video-rate frames (acquired at a rate of 30 frames/s using Metamorph™ software (Universal Imaging / Molecular Devices) and Video Savant (IO Industries™). Metamorph™ was used to create time-lapse videos from sequential two-photon images, and to observe and analyze QD uptake and the trafficking of QD-containing endosomes inside DCs. Images were acquired at a spatial resolution of 4.5 pixels/µm using 110 mW mean laser power for quantification of QD uptake, and at a spatial resolution of 12 pixels/µm for measurement of kinetics of QD-containing vesicles. For evaluation of antigen uptake in immature and mature DCs, QD-fluorescence intensity was measured over time. After background subtraction, intensities were normalized to the intensity measured 400 s after incubation of DCs in QD-containing medium. The linear slopes of intensities were measured to quantify rate of uptake of QDs by DCs. Velocities and displacements of individual vesicles were measured from unprocessed sequential two-photon images taken at intervals of 700 ms. Normalized displacements of individual vesicles were plotted to depict trajectories. For experiments involving temperature-variation of vesicular dynamics, temperatures were varied between 23°C and 37°C. Q₁₀ was calculated as Q₁₀ = 10^ΔlogV, where ΔlogV is the change in the value of log₁₀velocity for a 10°C rise in temperature. To observe the effects of cytoskeletal inhibitors on uptake of QDs, bone marrow-derived DCs were labeled with 10 µM CFSE (Invitrogen™) for 15 min at 37°C, and subsequently treated with calyculin A (Sigma-Aldrich™; 200 nM for 20 min), cytochalasin D (Sigma-Aldrich™; 1 µM for 1 hr), or nocodazole (Sigma-Aldrich™; 1 µM for 1 hr). These DCs were then incubated with 2 nM QD-containing medium at 37°C in the maintained presence of each of the reagents at the same concentration as used for pre-treatment. Using Metamorph™ QD-fluorescence and CFSE-fluorescence were measured within regions containing &gt;5 cells after 10 min of incubation. Fluorescence intensity was normalized to intensity of CFSE and has been presented as a ratio of QD-fluorescence intensity of untreated control DCs. To observe the effects of cytoskeletal inhibitors on vesicular motion, bone marrow-derived DCs were first allowed to take up QD for 30 min at 37°C, and subsequently treated with calyculin A (200 nM), cytochalasin D (1 µM), or nocodazole (1 µM) during imaging. For lysosomal co-localization, harvested DCs were cultured overnight in glass-bottom culture dishes (Mattek™), incubated with 2 nM QD-containing medium for varying times at 37°C. Subsequently, DCs were fixed with 2% paraformaldehyde in PBS for 30 min at 4°C, and then stained with 20 µg/ml FITC-conjugated anti-mouse LAMP-2 (Pharmingen) for 2 hr at 4°C to stain lysosomes. DCs were washed with PBS, dehydrated in 100% ethanol, mounted using Vectashield™ hardest mounting medium (Vector Labs), and left overnight at 4°C before imaging. To observe the effects of cytoskeletal inhibitors on F-actin organization and QD-uptake, bone marrow-derived DCs were treated with cytochalasin D, nocodazole, or calyculin A, and then incubated with 2 nM QD-containing medium for 15 min at 37°C in the maintained presence of these reagents, as described earlier. Subsequently these DCs were plated on glass-bottom culture dishes, fixed with 4% paraformaldehyde and stained with 50 µg/ml FITC-phalloidin for 1 hr at room temperature. These samples were washed with PBS, dehydrated in 100% ethanol, mounted using Vectashield™ hardest mounting medium (Vector Labs), and left overnight at 4°C before imaging. Fixed DCs were imaged using a Zeiss LSM-510 confocal microscope equipped with a Zeiss™ PlanApochromat™ 63× 1.40 N.A. oil-immersion objective.</p>", "<title>\n<italic>In vivo</italic> imaging of DC and determination of DC phenotype</title>", "<p>BALB/c mice were s.c. injected in lower flank with 100 nmol CFSE (green) and 20 pmol QD (red) included in 50 µl of complete Freund's adjuvant (CFA, Sigma™). Draining inguinal and brachial nodes from injected mice were harvested 24 hr later and observed using two-photon microscopy as described ##REF##14722354##[10]##. Briefly, the nodes were placed on plastic cover slips (Fisher™) using Vetbond™ (3M™), perfused with oxygenated RPMI (Hyclone Inc.) maintained at 37°C, and observed using two-photon microscopy. Images of several z sections (each an average of 15 frames acquired at 30 frames/s) at varying distances of separation were combined to create individual 3D time-points using Metamorph™. For locating DCs with respect to B cell follicles, mice were adoptively transferred with MACS™ magnetic separation chromatography-enriched (Miltenyi Biotec.) B cells from BALB/c mice labeled with 25 µM 7-amino-4-chloromethylcoumarin (CMAC, blue; Molecular Probes™ / Invitrogen™) for 45 min at 37°C; 24 hr later these mice were s.c. injected with 100 nmol CFSE (green) and 20 pmol QD (red) included in 50 µl CFA. 24 hr after injection, draining lymph nodes were harvested for two-photon imaging. DCs were recovered from lymph nodes for flow cytometry, as described ##REF##16461338##[41]##. Briefly, lymph nodes from injected mice were harvested 24 hr after s.c. injection of dye-CFA mixture, digested with 1 mg/ml Collagenase, Type IV (Worthington Biochemical) for 30 min at 37°C in DMEM substituted with 2% FBS. Cells were subsequently labeled with PE-conjugated anti-mouse CD11c (Ebioscience™), PE-conjugated anti-mouse GR-1 (Ebioscience™), or PE-conjugated anti-mouse CD8α (Ebioscience™), fixed with 2% paraformaldehyde in PBS, and analyzed by flow cytometry using Facscalibur™ flow cytometer (BD Biosciences).</p>", "<title>Electron microscopy</title>", "<p>Harvested DCs were cultured overnight in glass-bottom culture dishes (Mattek™), incubated with 2 nM QD-containing medium for varying times at 37°C, and processed as described ##REF##14597416##[23]##. Briefly, DCs were fixed with 2% glutaraldehyde (Electron Microscopy Sciences) in 0.1 M sodium cacodylate (Ted Pella™) buffer (pH 7.3) for 20 min, washed with 0.1% cacodylate buffer, and postfixed with 1% osmium tetroxide (Electron Microscopy Sciences) solution for 30 min. Subsequently, the DCs were counterstained with 4% uranyl acetate (Electron Microscopy Sciences) for 30 min, washed with distilled water, dehydrated in 100% ethanol, and embedded in Durcupan™ ACM resin (Fluka™).</p>", "<title>\n<italic>In vitro</italic> T cell proliferation assay</title>", "<p>DCs (∼4×10⁵ cells) were incubated with varying concentrations of ovalbumin or QD^ova in U-bottom polystyrene tubes (Fisher Scientific™). Antigen-pulsed DCs were incubated 6 hr later with 1 µg/ml LPS for 12–16 hr. CD4⁺ T cells were purified from spleen and lymph nodes of DO11.10 mice by depleting CD8⁺ T cells, B cells, NK cells, DCs, macrophages, granulocytes, and RBCs using MACS™ magnetic cell separation kit (Miltenyi Biotec.). Subsequently, the T cells were labeled with 4 µM carboxyfluorescein diacetate succinimidyl ester (CFSE) (Molecular Probes™/Invitrogen™) at 37°C for 10 min and ∼2×10⁶ cells (5 T cells per DC) were put into each U-bottom tube containing antigen-pulsed or control DCs in T cell medium (RPMI, 10% FCS, 1 mM sodium pyruvate, 1% non-essential amino acids, 1% l-glutamine, 50 µM β-mercaptoethanol, 100 units/ml penicillin, and 100 µg/ml streptomycin). After 72–96 hr at 37°C T cells were harvested and stained using PE-conjugated anti-mouse DO11.10 clonotypic T cell receptor (KJ1-26) antibody (Pharmingen™). CFSE counts were gated on KJ1-26⁺ cells and CFSE dilution was analyzed using flow cytometry to measure proliferation due to T cell activation. T cell activation was measured in terms of an activation index (<italic>AI</italic>) using Equation 1, derived as described in ##SUPPL##0##Methods S1##. <italic>M₁</italic> through <italic>M₇</italic> are the cumulative counts of T cells within the gates M₁ through M₇, shown in ##FIG##5##Figure 6A–6C##. For comparison and evaluation of varying antigen doses, the activation indices were normalized to the activation index of 50 µg/ml ovalbumin.\n</p>", "<title>\n<italic>In vivo</italic> T cell proliferation assay</title>", "<p>T cells from DO11.10 (ova peptide^323–339/I-A^d) or OT-II mice (ova peptide^323–339/I-A^b) were enriched using MACS™ as described above and labeled with 4 µM CFSE at 37°C for 15 min. ∼4×10⁶ CFSE-labeled cells were adoptively transferred into BALB/c (for DO11.10 T cells) or C57BL/6 (for OT-II T cells) recipients, and allowed to equilibrate for 24 hr. Subsequently the recipient mice were s.c. injected with varying dose of ovalbumin or QD^ova, or QDs included in 50 µl CFA, or CFA alone in the lower flank. Lymphocytes were recovered from lymph nodes and spleens harvested 2–4 days later. Harvested lymphocytes from BALB/c recipients were stained with PE-conjugated anti-mouse KJ1-26 antibody (Pharmingen) and analyzed for proliferation using flow cytometry by gating on KJ1-26. Harvested lymphocytes from C57BL/6 recipients were stained with PE-conjugated Vβ5.1/5.2 and APC-conjugated Vα2, and analyzed for proliferation using flow cytometry by gating on Vβ5.1/5.2 and Vα2. T cell activation was measured in terms of activation index (<italic>AI</italic>) using Equation 1 with five terms.</p>", "<title>Imaging ova-specific T cell-DC interactions</title>", "<p>T cells from DO11.10 mice were labeled with blue CMAC (20 µM at 37°C for 45 min) or green CFSE (10 µM at 37°C for 15 min) and ∼4×10⁶ cells were adoptively transferred into BALB/c recipients. 24 hr later mice were s.c. injected on one flank with QD^ova (containing 50 µg ovalbumin), and on the opposite flank with QD included in 50 µl complete Freund's adjuvant. Lymph nodes were harvested 18 hr later and imaged using two-photon microscopy, as described earlier. In separate experiments, mice were immunized with QD^ova or QD, and 12 hr later adoptive transferred with blue CMAC-labeled T cells. Lymph nodes were harvested 8 hr after adoptive transfer of T cells and imaged using two-photon microscopy.</p>", "<title>Assessment of IFN-γ production in T cells in immunized mice</title>", "<p>T cells from OT-II mice (expressing CD45.2) were enriched as described above, and labeled with 4 µM CFSE at 37°C for 15 min. ∼2×10⁶ CFSE-labeled cells were adoptively transferred into congenic CD45.1-expressing BL6 recipients, and allowed to equilibrate for 24 hr. Subsequently, the recipient mice were s.c. injected with 50 µg ovalbumin or QD^ova, in both cases in 50 µl CFA. Lymphocytes were harvested from draining inguinal lymph nodes or control mesenteric lymph nodes 4 days after immunization. Harvested lymphocytes were re-stimulated with PMA and ionomycin <italic>in vitro</italic> for 6 hr. For the last 2 hr BD™ Golgistop™ was added to the medium. Subsequently, lymphocytes were stained with PE-conjugated anti-mouse CD45.2 (Ebioscience), fixed using BD™ cytofix-cytoperm™ cytokine staining kit, and labeled with APC-conjugated anti-mouse IFN-γ (Ebioscience) or isotype control. Labeled cells were assessed for IFN-γ production by gating on CD45.2⁺ cells.</p>" ]

[ "<title>Results</title>", "<title>Quantum dot uptake by dendritic cells</title>", "<p>Murine DCs derived from bone marrow avidly endocytosed streptavidin-conjugated or unconjugated QDs during imaging <italic>in vitro</italic> at 37°C. QDs differing in size from 15–20 nm were taken up at approximately the same rate (not shown), and were retained for at least 48 hr without toxicity (##SUPPL##1##Figure S1##). In these mixed cultures, cells that became brightly labeled by taking up QDs also exhibited dynamic dendritic processes (##FIG##0##Figure 1A##; ##SUPPL##7##Video S1##), whereas the smaller, round cells (presumptive monocytes) did not take up QDs. Flow cytometric analysis showed that QD⁺ cells also expressed the characteristic DC marker CD11c (##SUPPL##2##Figure S2##). Furthermore, in experiments on bone marrow-derived cells cultured from EYFP-CD11c reporter mice, EYFP⁺ DCs became brightly labeled with endocytosed QDs (##FIG##0##Figure 1B##). Together, these results show that QDs are avidly and preferentially taken up by CD11c⁺ DCs.</p>", "<p>We monitored the kinetics of QD uptake following addition of QDs to the bathing medium (##FIG##1##Figure 2A–2F##; ##SUPPL##8##Video S2##). QD-fluorescence appeared within 2 min as a band adjacent to the plasma membrane of DCs (##FIG##1##Figure 2A and 2D##). Subsequently, distinct QD-containing vesicles were apparent inside the cells, initially close to the plasma membrane (##FIG##1##Figure 2B and 2E##), and later throughout the cytoplasm (##FIG##1##Figure 2C and 2F##). At longer times, the number of QD-containing vesicles per cell increased, as did their size and fluorescence intensity. Close inspection of single cells showed individual vesicles of varying size (up to ∼1 µm in diameter) ‘jiggling’ in the cytoplasm. Occasionally, vesicles fused with one another inside the cell (##FIG##1##Figure 2G–2I##; ##SUPPL##9##Video S3##), but vesicle fusion with the plasma membrane or release of their contents was never observed. Individual vesicles moved in a “stop and go” manner, exhibiting occasional bursts of high velocity followed by periods of jiggling (##FIG##1##Figure 2J and 2K##; ##SUPPL##10##Video S4##). The mean vesicular velocity was 9±1.5 µm/min at 37°C, whereas peak velocities during bursts were 5–10 times faster. Vesicular velocities were independent of the apparent variation in vesicular sizes. Vesicles did not exhibit any directional preference in their motion, as shown by their trajectories (##FIG##1##Figure 2L##). Increased temperature reversibly increased vesicular velocities (##FIG##1##Figure 2M##; ##SUPPL##10##Video S4##) with a Q₁₀ of ∼2. The rate of endocytosis is known to slow following maturation of DCs ##REF##10426272##[2]##; and, consistent with this, we observed a dramatically (10-fold) reduced rate of quantum dot uptake in DCs matured by treatment with LPS (##FIG##1##Figure 2N##). QD uptake was observed in a greater proportion of cells in the immature-DC population compared to matured DCs (Data not shown). The number of QD-containing vesicles was reduced in LPS-treated DCs. However, velocities and subcellular kinetics of vesicles was not altered.</p>", "<p>To investigate the involvement of cytoskeletal elements on QD uptake and vesicle dynamics, we observed effects of pre-treatment and acute addition of cytochalasin D, calyculin A, or nocodazole in DCs (##FIG##2##Figure 3##). Untreated control DCs displayed QD-containing vesicles, as well as intact actin filaments (##FIG##2##Figures 3A–3C## and ##SUPPL##3##S3A–S3C##). Actin filaments were absent inside cytochalasin D-treated DCs, whereas patches of condensed actin were observed near the plasma membrane along with vesicles that contained QD (##SUPPL##3##Figure S3D–S3F##), and total uptake of QD was partially inhibited (##FIG##2##Figure 3D–3F, 3M##). Inside nocodazole-treated DCs actin filaments were intact and QD-containing vesicles were apparent, as in untreated DCs (##FIG##2##Figures 3G–3I## and ##SUPPL##3##S3G–S3I##). Nocodazole-treated DCs rounded up and total QD uptake in these DCs was again partially reduced compared to untreated DCs (##FIG##2##Figure 3M##). Calyculin A-treated DCs rounded up, and cortical actin was observed condensed near the periphery (##SUPPL##3##Figure S3J##). QD-uptake was abolished in most calyculin A-treated DCs (##FIG##2##Figure 3J–3M##, ##SUPPL##3##S3J–S3L##). When present, QD fluorescence was restricted to the cell perimeter. However, deeper intracellular QD-containing vesicles were not formed. Together, these results indicate that the presence of free cortical actin is necessary for QD-uptake by DCs.</p>", "<p>We next studied the effects of these cytoskeletal inhibitors on the dynamics of QD-containing vesicles within DCs. Cytochalasin D did not affect vesicular dynamics (##FIG##2##Figure 3O##, ##SUPPL##11##Video S5##). However, nocodazole produced vesicle arrest (##FIG##2##Figure 3P##, ##SUPPL##12##Video S6##), implying that microtubular assembly is necessary for motion of individual vesicles inside DCs. Moreover, in untreated DCs, vesicles did not freely diffuse inside the cytoplasm, but exhibited constrained “jiggling” movement (##SUPPL##10##Video S4##), consistent with an attachment to microtubules. Calyculin A did not inhibit vesicular motion. (##FIG##2##Figure 3Q##, ##SUPPL##13##Video S7##). In calyculin A-treated DCs, vesicles were occasionally observed to traverse longer paths compared to untreated DCs, indicating the absence of interactions with actin for relatively longer periods of time.</p>", "<title>Compartmentalization of endocytosed QDs</title>", "<p>Co-labeling of QD-containing DCs for the lysosomal protein LAMP-2 showed that QD-containing vesicles (within 0–5 min of incubation) did not colocalize with lysosomes initially (##FIG##3##Figure 4A–4I##). However, at later time points (&gt;∼45 min) an increasing number of QD-containing vesicles colocalized with lysosomes (##FIG##3##Figure 4J–4O##), indicating that eventually QDs are retained inside lysosomes. We also visualized the distribution of single QDs inside DCs by electron microscopy (##FIG##3##Figure 4P–4U##). QD-containing vesicles were localized near the plasma membrane after 5 min of incubation (##FIG##3##Figure 4P and 4Q##); at this time vesicles typically contained ∼5–6 QDs and had sizes &lt;500 nm. After 45 min of incubation the numbers of QD-containing vesicles increased, each containing ∼20–30 QDs, with sizes up to 1000 nm (##FIG##3##Figure 4R and 4S##). Vesicles of varying size were distributed throughout the cytoplasm. As expected, QDs were excluded from the nuclei and mitochondria. Together, light microscopy and em visualization indicate that DCs continually endocytose QDs into small vesicles that subsequently fuse to form larger vesicles that at later times mature into lysosomes with higher electron density containing ∼50–100 QDs (##FIG##3##Figure 4T and 4U##).</p>", "<title>\n<italic>In vivo</italic> labeling of DCs and trafficking to peripheral lymph nodes</title>", "<p>Based on our <italic>in vitro</italic> results and previous <italic>in vivo</italic> labeling of DCs with cell tracker dyes at an adjuvant depot ##REF##14722354##[10]##, we anticipated that skin-resident DCs would take up subcutaneously injected QDs and subsequently traffic into draining lymph nodes. We injected mice subcutaneously with QD in complete Freund's adjuvant, either in eYFP-CD11c reporter mice or in BALB/c mice together with CFSE, and evaluated subsequent QD together with YFP or CFSE fluorescence in draining lymph nodes. At 4 hr, a time too early for skin-draining DCs to arrive in draining lymph nodes, QDs were observed in the lymph-node subcapsular sinus, but not inside the cortex (##SUPPL##4##Figure S4##). This is expected, as streptavidin-conjugated QDs have a size ∼2000 kD, larger than the approximate cutoff of ∼70 kD for free diffusion into the cortex from the subcapsular sinus ##REF##11085745##[21]##. QDs were also observed inside eYFP-expressing resident DCs lining the subcapsular sinus. One day after injection, DCs that had taken up either CFSE or QD, or both, were distributed within the T cell cortical region, and were excluded from the follicle (##FIG##4##Figure 5A–5C##). Flow cytometric analysis showed that QDs, as well as CFSE, preferentially labeled CD11c⁺ DCs (##SUPPL##5##Figure S5A and S5B##). Lymph node DCs labeled with QDs represent a mixture of migratory and resident DCs (##SUPPL##5##Figure S5C##), as indicated by the distribution of CD8α. Thus, consistent with our <italic>in vitro</italic> findings, skin-resident DCs near the site of injection take up QDs; and in addition, QDs that had migrated via lymphatics to the draining lymph node, are taken up by lymph node-resident DCs. The injected fluorophores also labeled a small percentage of GR-1⁺ granulocytes (##SUPPL##5##Figure S5D and S5E##). CFSE labeling of lymph node DCs was uniform through the cell body and processes, whereas QDs appeared compartmentalized in the cell body. DCs moved with an average velocity of ∼4 µm/min within the node, with individual DCs showing mean velocities distributed between 2–8 µm/min (##FIG##4##Figure 5D##). DCs frequently changed direction as they moved and did not exhibit any particular directional preference (##FIG##4##Figure 5E##). Movements were generally led by dynamic dendritic processes, followed by the cell body which often extended cytoplasmic “tails” (##SUPPL##14##Video S8## and ##SUPPL##15##S9##). Both CSFE and QD fluorescence could be resolved at depths down to ∼200 µm beneath the capsule, but only QD fluorescence remained visible below this, at depths up to 400 µm (data not shown). All QD⁺ cells were also CFSE⁺ (##FIG##4##Figure 5F##); thus QDs enabled deeper imaging by being brighter than CFSE under two-photon excitation.</p>", "<title>Antigen-conjugated QDs potently activate T cells <italic>in vitro</italic> and <italic>in vivo</italic>\n</title>", "<p>To explore the potential use of QDs for particle-based antigen presentation, we measured the <italic>in vitro</italic> proliferation of TCR-transgenic ovalbumin-specific T cells in the presence of bone marrow-derived DCs pulsed with ovalbumin or ovalbumin-conjugated QDs (QD^ova) (##FIG##5##Figure 6A and 6B##, respectively). Both treatments induced robust proliferation, whereas DCs pulsed with QDs alone failed to activate T cells (##FIG##5##Figure 6C##). Quantification of T cell proliferation using an activation index (<italic>AI</italic>; see <xref ref-type=\"sec\" rid=\"s4\">Methods</xref>), revealed that QD^ova was ∼20 fold more effective in activating ova-specific T cells compared to equivalent amounts of free ovalbumin (##FIG##5##Figure 6D##). T cell activation plateaued at ovalbumin concentrations &gt;50 µg/ml, and at QD^ova concentrations equivalent to &gt;1.25 µg/ml ovalbumin. Moreover, the maximal proliferation was higher with QD^ova than with ovalbumin alone.</p>", "<p>We examined dynamic T cell / DC interactions in mice bearing adoptively transferred ova-specific T cells with and without immunization with QD^ova. In mice injected with QDs without ova, T cells migrated freely, interacting transiently with DCs and covering a broad territory, as indicated by time overlays of superimposed frames (##FIG##6##Figure 7A##; ##SUPPL##16##Video S10##). In contrast, clusters of T cells were observed surrounding QD^ova-bearing DCs within draining lymph nodes from mice immunized with QD^ova (##FIG##6##Figure 7B##; ##SUPPL##17##Video S11##). Time overlays showed that blue T cell tracks corresponded with red tracks of QD^ova-bearing DCs. These T cell-DC clusters were observed at depths up to 150 µm. These results show that antigen-conjugated QD-labeled DCs can functionally engage in prolonged interactions with antigen-specific T cells to initiate an immune response.</p>", "<p>To assess the ability of QD^ova to induce T cell proliferation <italic>in vivo</italic>, CFSE-labeled ova-specific T cells were adoptively into mice that were then immunized with either ovalbumin or QD^ova. Upon immunization of these mice with ovalbumin and complete Freund's adjuvant (CFA), ova-specific T cells inside draining lymph nodes divided up to five times within 2 days (##FIG##6##Figure 7C–7E##; ##SUPPL##6##Figure S6A##). Beginning ∼3 days after immunization, T cells that had divided were released into the circulation and were recovered from the spleen and non-draining lymph nodes (##SUPPL##6##Figure S6##). Ova-specific T cells also proliferated robustly upon immunization of mice with QD^ova (7F–7H). Assessment of relative <italic>in vivo</italic> T cell proliferation prior to T cell egress showed that QD^ova was &gt;5 fold more effective than free ovalbumin in inducing ova-specific T cell proliferation (##FIG##6##Figure 7I##). T cells did not proliferate when injected with QDs (without antigen) or CFA alone (data not shown).</p>", "<p>To evaluate the generality of T cell priming by QD^ova, we examined a second transgenic TCR mouse also having sensitivity to ova but on an entirely different MHC background. To examine different stages of T cell activation, we evaluated CD69 expression and production of IFN-γ, in addition to cell proliferation (##FIG##7##Figure 8##). CD69 expression was up-regulated to a similar extent in mice immunized with either free ovalbumin+CFA or QD^ova+CFA (##FIG##7##Figure 8 A–C##). QD^ova stimulated OT-II T cells to proliferate in C57BL/6 recipient mice (I-A^b) more efficiently compared to free ovalbumin (##FIG##7##Figure 8D–8I##). Furthermore, production of IFN-γ was strongly enhanced in T cells inside draining lymph nodes of mice immunized with QD^ova, compared to mice immunized with free ovalbumin (##FIG##7##Figure 8K and 8J##). Collectively, these results demonstrate that <italic>in vivo</italic> T cell priming is more effective using antigen-conjugated QDs compared to a standard immunization regimen with free antigen.</p>" ]

[ "<title>Discussion</title>", "<p>Our results demonstrate the utility of QDs for functional immuno-imaging of DCs and concomitant priming of T cell responses <italic>in vitro</italic> and <italic>in vivo</italic>. DCs were shown preferentially to acquire QDs by endocytosis; single-vesicle dynamics were tracked inside the cell as small endocytic vesicles jiggled and fused to form larger vesicles up to 1 µm in diameter. Single QDs were detected by electron microscopy inside endocytic vesicles that become increasingly electron dense during maturation. Rates of quantum dot uptake were reduced during toll-like receptor stimulation with LPS, consistent with a decrease in the rate of endocytosis during DC maturation. When included in a subcutaneous adjuvant depot, QDs were taken up by dendritic cells that migrated into the draining lymph node, where the bright and stable fluorescence of QDs facilitated two-photon imaging of DCs with detection limits at approximately twice the depth as other labels, although not revealing the dynamics of dendritic processes. Moreover, QDs bearing conjugated antigen were taken up by DCs and efficiently evoked T cell proliferative responses <italic>in vitro</italic> and <italic>in vivo</italic>.</p>", "<p>DCs continually endocytose substances from the surrounding medium ##REF##10426272##[2]##. Previously, QDs have been used for real-time measurement of receptor-mediated endocytosis in transfected CHO cells ##REF##14704683##[22]##. Human DCs were recently reported to take up QDs conjugated to a ligand for the specific DC-receptor DC-SIGN by endocytosis of QDs ##REF##17388641##[15]##. In our study, we show that DCs avidly take up unconjugated or antigen-conjugated QDs, but without a requirement for conjugated endocytic DC cell-surface receptor-specific ligands. At a subcellular level, QD-uptake by DCs was not dependent on microfilaments, since disruption of F-actin by cytochalasin D did not inhibit uptake, consistent with previous reports showing that cytochalasin D does not inhibit endocytosis in mammalian macrophages and leukocytes ##REF##14597416##[23]##, ##REF##12724139##[24]##. Endocytosis of small particles (&lt;100 nm) by mammalian macrophages have been shown to be insensitive to cytochalasin D ##REF##15528351##[25]##, ##REF##16263511##[26]##, and the hydrodynamic diameter of QDs (&lt;20 nm) falls well below this cutoff. Moreover, cytochalasin is known to inhibit phagocytosis but not pinocytosis ##REF##2119341##[27]##. QD uptake by DCs was not completely inhibited by nocodazole, indicating the existence of a microtubule-independent mechanism of QD uptake. Previous evidence shows that nocodazole does not inhibit endocytosis in bone marrow-derived phagocytes ##REF##9495017##[28]##. In contrast, polymerization of cortical actin by calyculin A completely inhibited QD uptake by DCs. Similar inhibition of endocytosis by calyculin A has been previously observed in T cells ##REF##14597416##[23]##. Together, these results indicate that QDs are endocytosed by DCs mainly via pinocytosis, and the presence of free actin is necessary for QD uptake. Endocytosed QDs are compartmentalized by DCs into vesicles. These QD-containing endosomes alternated between periods of high velocity and stationary ‘jiggling’. Endosomes are known to traffic along microtubules and can interact with actin via motor proteins ##REF##7755990##[29]##, ##REF##11549946##30##. Thus, as expected, vesicular motion was arrested by microtubule disruption using nocodazole. The periods of jiggling likely represent interactions of QD-containing endosomes with actin while trafficking along microtubules. Moreover, the absence of free actin in calyculin A-treated DCs probably caused vesicles to move freely along microtubles for long distances (##SUPPL##13##Video S7##). Eventually, endocytosed QDs are retained by DCs inside lysosomes for at least 48 hr.</p>", "<p>Previous studies to visualize DCs <italic>in vivo</italic> have employed conventional fluorophores and fluorescent proteins, including <italic>in vivo</italic> labeling with cell tracker dyes included in alum adjuvant mixture ##REF##14722354##[10]##, <italic>in vitro</italic> labeling with subsequent cell transfer ##REF##14712275##[11]##, ##REF##12730692##[31]##, ##REF##12052961##[32]##, microinjection of fluorophores into lymph node ##REF##15516925##[33]##, and expression of YFP driven by a CD11c promoter element in transgenic mice ##REF##15543150##[34]##. Here, we show that QDs are selectively taken up by dendritic cells <italic>in vivo</italic>, and can be used as a fluorescent marker for antigen-bearing DCs that traffic to draining lymph nodes from the site of immunization. Moreover, QD-bearing DCs can be imaged to greater depth inside draining lymph nodes without phototoxicity to the tissue. Thus, although not revealing the dynamics of DC dendrites, QDs are superior to commonly used fluorophores in terms of imaging depth in intact tissue.</p>", "<p>Particle-based antigen delivery systems are being investigated for the development of efficient DC-based vaccines ##UREF##2##[35]##, ##UREF##3##[36]##. In addition to using QDs for imaging of DCs, we further demonstrate that antigen-conjugated QDs can induce robust DC-mediated CD4⁺ T cell proliferation both <italic>in vitro</italic> and <italic>in vivo</italic> in two different strains of mice. QD delivery enhances T cell priming, leading to a 5–20 fold shift in the dose required to stimulate proliferation and a strong enhancement of IFN-γ production, compared to a standard immunization using equivalent amounts of free antigen (ovalbumin). Since QDs are avidly endocytosed by DCs, conjugation to QDs likely increases the rate of antigen uptake by DCs as compared to free antigen. The ovalbumin bound to QD is subsequently processed by the DCs and presented to T cells, whereas the QDs themselves are not processed by lysosomal enzymes or trafficked further into the MHC class-II pathway, but are instead retained inside DCs. Concentrations (5–10 nM) of QDs sufficient to induce robust T cell proliferation did not produce any evident toxic effects in DCs observed over a period of 48 hr, and QD⁺ cells migrated to the lymph node 12–16 hr after subcutaneous injection, consistent with previous studies showing that QDs can be used for several days <italic>in vivo</italic> without deleterious effects ##REF##12459582##[37]##, ##REF##12459736##[38]##.</p>", "<p>In conclusion, QDs can serve as multi-functional probes with unique advantages for both imaging studies employing correlated fluorescence and electron microscopy, and as a potent particle-based system for antigen delivery. QDs are versatile fluorophores that can be used to fine-tune important functional aspects of the immune system, in conjunction with correlated fluorescence and electron microscopy, and may have potential for use in immunotherapy and vaccine development.</p>" ]

[]

[ "<p>Conceived and designed the experiments: DS MDC. Performed the experiments: DS. Analyzed the data: DS. Wrote the paper: DS MDC. Electron microscopy: TJD. Supervised electron microscopy: MHE. Supervised two-photon imaging: IP.</p>", "<p>Dendritic cells (DCs) play a key role in initiating adaptive immune response by presenting antigen to T cells in lymphoid organs. Here, we investigate the potential of quantum dots (QDs) as fluorescent nanoparticles for <italic>in vitro</italic> and <italic>in vivo</italic> imaging of DCs, and as a particle-based antigen-delivery system to enhance DC-mediated immune responses. We used confocal, two-photon, and electron microscopies to visualize QD uptake into DCs and compared CD69 expression, T cell proliferation, and IFN-γ production by DO11.10 and OT-II T cells <italic>in vivo</italic> in response to free antigen or antigen-conjugated to QDs. CD11c⁺ DCs avidly and preferentially endocytosed QDs, initially into small vesicles near the plasma membrane by an actin-dependent mechanism. Within 10 min DCs contained vesicles of varying size, motion, and brightness distributed throughout the cytoplasm. At later times, endocytosed QDs were compartmentalized inside lysosomes. LPS-induced maturation of DCs reduced the rate of endocytosis and the proportion of cells taking up QDs. Following subcutaneous injection of QDs in an adjuvant depot, DCs that had endocytosed QDs were visualized up to 400 µm deep within draining lymph nodes. When antigen-conjugated QDs were used, T cells formed stable clusters in contact with DCs. Antigen-conjugated QDs induced CD69 expression, T cell proliferation, and IFN-γ production <italic>in vivo</italic> with greater efficiency than equivalent amounts of free antigen. These results establish QDs as a versatile platform for immunoimaging of dendritic cells and as an efficient nanoparticle-based antigen delivery system for priming an immune response.</p>" ]

[ "<title>Supporting Information</title>" ]

[ "<p>We thank Luette Forest for care and breeding of the DO11.10 mouse strain and for assistance in cell culture, Kym Garrod for valuable discussion and assistance in flow cytometry, Sindy Wei and Melanie Matheu for assistance in two-photon imaging. We also wish to thank Dr. Ed Nelson and Bali Pulendran for helpful comments on the manuscript.</p>" ]

[ "<fig id=\"pone-0003290-g001\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003290.g001</object-id><label>Figure 1</label><caption><title>DCs avidly endocytose QDs.</title><p>(A) Overlay of brightfield and fluorescent images showing endocytosed QD-655 (red) within DCs with dendritic processes (indicated by arrows), after 30 min of incubation of DCs in QD-containing medium. Image is a single frame from ##SUPPL##7##Video S1##. Scale bar = 10 µm. (B) Two-photon image showing endocytosed QDs (red) inside CD11c⁺ DCs (green) cultured from bone marrow of transgenic EYFP-CD11c reporter mice. Scale bar = 5 µm.</p></caption></fig>", "<fig id=\"pone-0003290-g002\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003290.g002</object-id><label>Figure 2</label><caption><title>Dynamics of QD-uptake and QD-containing vesicles.</title><p>(A–F) Two-photon images of QD uptake by DCs at ∼2 min, 9 min, and 18 min after beginning incubation with 2 nM QD-containing medium at 37°C. These frames have been taken from from ##SUPPL##8##Video S2##. Indicated areas of A–C are magnified in D–F respectively to show individual vesicles. Scale bars = 10 µm in A–C, and = 1 µm in D–F. (G–I) Fusion of QD-containing vesicles inside DCs. Three vesicles, indicated by arrows, come closer, and eventually fuse into a single larger vesicle. These frames have been taken from ##SUPPL##9##Video S3##. Scale bars = 2 µm. (J) DCs with QD-containing. The image is a single frame taken from ##SUPPL##10##Video S4##. Scale bar = 4 µm. Arrowheads with numbers represent three individual vesicles, whose velocities are shown in K. (K) Velocity profiles of the individual vesicles indicated by numbers in J. Velocity bursts (with peak velocities &gt;5 times the average velocity) are marked in gray. (L) Trajectories of 10 individual vesicles that could be tracked for &gt;140 sec, normalized to their starting coordinates. (M) Temperature-dependence of vesicular velocities in DCs. Data from 3 independent experiments, 10–12 vesicles per experiment. (N) Downregulation of rate of QD-uptake by DC maturation. Plot of normalized mean QD-fluorescence intensity over time for LPS-treated or mature (diamonds) and untreated or immature (squares) DCs. Lines are linear regressions, with slopes of 32.8×10⁻⁴ a.u./s for immature cells and 4.4×10⁻⁴ a.u./s for mature cells. Data from 4 independent experiments for each condition, 6–10 cells per experiment.</p></caption></fig>", "<fig id=\"pone-0003290-g003\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003290.g003</object-id><label>Figure 3</label><caption><title>Effects of cytoskeletal inhibitors on QD-endocytosis by DCs.</title><p>QD (red) uptake in untreated control DCs (A–C), or DCs treated with cytochalasin D (D–F), nocodazole (G–I), or calyculin A (J–L). Scale bar = 10 µm. (M) Histogram showing mean normalized QD-fluorescence intensity of cells treated under different conditions as indicated. QD-fluorescence intensity has been expressed as a ratio of the mean QD-fluorescence intensity of untreated DCs. (N–Q) Trajectories of QD-containing vesicles inside untreated control DCs (N), or acutely treated with 1 µM cytochalasin D (O), 1 µM nocodazole (P), or 200 nM calyculin A (Q), respectively. The effects of these reagents on the dynamics of QD-containing vesicles inside DCs is further illustrated in ##SUPPL##11##Video S5##, ##SUPPL##12##S6## and ##SUPPL##13##S7##.</p></caption></fig>", "<fig id=\"pone-0003290-g004\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003290.g004</object-id><label>Figure 4</label><caption><title>Compartmentation of single QDs inside endocytic vesicles and lysosomes.</title><p>(A–O) Confocal images showing colocalization of QD-containing vesicles with the lysosomal protein LAMP-2 inside bone marrow-derived DCs after different times of incubation in QD-containing medium as indicated on the left. Columns show LAMP-2 staining (green), QD staining (red), and overlays of LAMP-2 and QD images, respectively. Scale bar in A–O = 10 µm. (P–U) Electron mircographs showing individual QDs within DCs, and associated structures and location after different times of incubation in QD-containing media as indicated on the right.. Indicated regions in P, R, and T, respectively, shown at higher magnification. Note the correspondence between electron micrographs P, R, and T with confocal images I, L and O, respectively. Scale bars in P, R, and T = 1 µm and in Q, S, and U = 200 nm.</p></caption></fig>", "<fig id=\"pone-0003290-g005\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003290.g005</object-id><label>Figure 5</label><caption><title>QD-labeled DCs visualized <italic>in situ</italic> in murine lymph nodes.</title><p>(A–C) Sequential two-photon images of <italic>in situ</italic>-labeled DCs inside draining lymph nodes 24 hr after subcutaneous injection of CFSE (green) and QD (red) included in CFA. The time elapsed (min) after beginning imaging is indicated. The images were taken from ##SUPPL##14##Video S8##. Scale bar = 20 µm. (D) Velocity distribution of DCs in lymph nodes. Arrow indicates average velocity (4.3±1.2 µm/min). (E) Trajectories of DCs in lymph nodes, normalized to their starting coordinates. (F) Profiles of CFSE⁺ and QD⁺ cells in draining lymph. Percentage of cells (except double negatives) have been indicated on the figure.</p></caption></fig>", "<fig id=\"pone-0003290-g006\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003290.g006</object-id><label>Figure 6</label><caption><title>Activation of DO11.10 T cells <italic>in vitro</italic> by DCs pulsed with ovalbumin or QD^ova.</title><p>(A, B, and C) CFSE peaks representing consecutive cycles of division of CFSE-labeled T cells that were activated by DCs pulsed with 25 µg/ml ovalbumin (A), 5 nM QD^ova (2.5 µg/ml ovalbumin) (B), or QD alone (Streptavidin-conjugated QDs without ovalbumin) (C). Lines delineate sequential divisions M1–M7 used in Equation 1 to derive the activation index (see <xref ref-type=\"sec\" rid=\"s4\">Methods</xref>). The amount of free ovalbumin present in the QD^ova sample was limited by serial dilution to &lt;0.1 µg/ml during purification, and was insufficient to induce T cell activation and proliferation (not shown). (D) Normalized activation indices of T cell proliferation as functions of concentration of ovalbumin (diamonds) and QD^ova (squares). Data from at least 3 independent experiments for each of 6 different QD and QD^ova concentrations are presented.</p></caption></fig>", "<fig id=\"pone-0003290-g007\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003290.g007</object-id><label>Figure 7</label><caption><title>Activation of DO11.10 T cells <italic>in vivo</italic> by DCs pulsed with QD^ova.</title><p>Overlays of successive two-photon images of draining lymph nodes 8 hr after adoptive transfer of CMAC-labeled ova-specific T cells into mice immunized with QD (A) or QD^ova (B) included in 50 µl CFA. These images are single frames from ##SUPPL##16##Video S10## and ##SUPPL##17##S11##, respectively. Scale bars = 20 µm. (C–H) Flow cytometry profiles showing activation of naïve T cells in draining lymph nodes of mice, into which ∼4×10⁶ CFSE-labeled ova-specific T cells were adoptively transferred, and which were subsequently s.c. injected in the lower flank with varying doses of free ovalbumin (C–E, light gray), or ovalbumin conjugated to QDs (dark gray) included in 50 µl CFA. (I) Normalized activation index as a function of total amount of ovalbumin delivered either unconjugated (diamonds) or conjugated with QDs (squares). Data are representative of at least 3 independent experiments.</p></caption></fig>", "<fig id=\"pone-0003290-g008\" position=\"float\"><object-id pub-id-type=\"doi\">10.1371/journal.pone.0003290.g008</object-id><label>Figure 8</label><caption><title>Activation of OT-II T cells <italic>in vivo</italic> by DCs pulsed with QD^ova.</title><p>(A, B, and C) CD69 upregulation in OT-II T cells following immunization with CFA containing either 50 µg free ovalbumin (A), QD^ova (B), or CFA alone (C). CD69 expression, evaluated 18 hr after immunization, was similar with free ova (∼60±7%, n = 3 experiments) and with QD^ova (∼65±6%, n = 3). (D–I) Proliferation of naïve OT-II T cells in draining lymph nodes of C57BL/6 mice s.c. injected with varying doses of either free ovalbumin (D–F, light gray), or ovalbumin conjugated to QDs (G–I, dark gray) included in 50 µl of CFA. Total amount of ovalbumin delivered in free form or conjugated to QDs is indicated. (J, K, and L) IFN-γ expression in mice adoptively transferred with OT-II T cells and immunized with CFA containing 50 µg free ovalbumin (J) or conjugated with QDs (K), or CFA alone (L). IFN-γ expression was evaluated 4 days after immunization. In three separate experiments, immunization with QD^ova+CFA showed 2.4±0.6 greater upregulation of IFN-γ, compared to immunization with ovalbumin+CFA (p&lt;0.05).</p></caption></fig>" ]

[]

[ "<disp-formula><label>(1)</label></disp-formula>" ]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"pone.0003290.s001\"><label>Methods S1</label><caption><p>Equipment and Settings, Calculation of T cell activation index (AI)</p><p>(0.07 MB DOC)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003290.s002\"><label>Figure S1</label><caption><p>Quantum dots are not toxic to DCs. DCs were incubated with 10 nM QD 525-Streptavidin conjugate for 48 hours, stained with 20 µg/ml of propidium iodide, and assessed for toxicity using flow cytometry. Histograms showing propidium iodide staining in untreated cells (green), QD-treated cells (red), and cells fixed with 2% glutaraldehyde (black). Isotype control has been shown in gray. While fixed DCs were ∼100% PI+, ∼15–20% of QD-treated DCs were PI+, comparable to untreated DCs. Thus QDs did not show toxicity at a concentration of 10 nM, higher than any of the concentrations used in our experiments</p><p>(0.37 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003290.s003\"><label>Figure S2</label><caption><p>Flow cytometry profile of cells that endocytosed QDs. Bone marrow-derived cells were incubated for 30 min with 2 nM QD at 37 µC, then washed and stained with (A) FITC-conjugated anti-CD11c or (B) FITC-conjugated anti-CD11b, and analyzed by flow cytometry by gating on the corresponding markers as shown in figure. ∼70% QD+ cells were CD11c+, and ∼75% QD+ cells were CD11b+.</p><p>(0.39 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003290.s004\"><label>Figure S3</label><caption><p>Confocal images showing effects of cytoskeletal inhibitors on F-actin organization and QD-uptake by DCs. DCs were either (A–C) untreated, or treated with (D–F) cytochalasin D (1 µM for 1 hr at 37°C), (G–I) nocodazole (1 µM for 1 hr at 37°C), or (J–L) calyculin A (200 nM for 20 min at 37°C), and incubated with media containing 2 nM QD (red) for 20 min in the maintained presence of these reagents. Subsequently, these DCs were fixed and stained with FITC-phalloidin to label F-actin (green). Confocal images are consistent with results obtained using real-time two-photon imaging. Scale bar = 10 µm.</p><p>(3.51 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003290.s005\"><label>Figure S4</label><caption><p>QDs in subcapsular sinus of draining lymph nodes. DCs were labeled in situ by subcutaneously injecting EYFP-CD11c mice with QDs included in 50 µl CFA. Draining lymph nodes were harvested 4 hrs later for imaging. Subcapsular fibers appear blue due to second harmonics. (A) QDs (red) are trapped inside vessels in the capsule and are presumably taken up by subcapsular DCs (green) and macrophages. (B, C) Z sectional views of the capsule show that QDs are not yet present inside the node below the capsule. Scale bar = 20 µm.</p><p>(5.15 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003290.s006\"><label>Figure S5</label><caption><p>Phenotypic profile of DCs observed by flow cytometry. Phenotypic markers are indicated on the Y-axis of each plot. Distribution of QD+ and CFSE+ cells (percent) for each phenotypic marker is shown on the graph.Note the similarity in distribution of QD+ and CFSE+ cells for each phenotypic marker.</p><p>(0.86 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003290.s007\"><label>Figure S6</label><caption><p>In vivo T cell response at different times following immunization. BALB/c mice were adoptively transferred with ∼4×10⁶ CFSE-labeled DO11.10 T cells, and subsequently immunized with 50 µg of ovalbumin included in 50 µl CFA. Control and draining lymph nodes, and spleens were harvested (A–C) 2 days, (D–F) 3 days, or (G–I) 4 days after injection, and analyzed for T cell activation using flow cytometry. Day 2 (just prior to egress of activated T cells) was chosen as the time point for analysis of T cell activation.</p><p>(0.32 MB TIF)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003290.s008\"><label>Video S1</label><caption><p>QD-labeled DCs <italic>in vitro</italic>\n</p><p>(5.34 MB AVI)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003290.s009\"><label>Video S2</label><caption><p>QD-uptake by DCs <italic>in vitro</italic>\n</p><p>(7.53 MB AVI)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003290.s010\"><label>Video S3</label><caption><p>Fusion of QD-containing vesicles inside DCs</p><p>(2.56 MB AVI)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003290.s011\"><label>Video S4</label><caption><p>Vesicles jiggling inside a single DC</p><p>(8.19 MB AVI)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003290.s012\"><label>Video S5</label><caption><p>Effect of cytochalasin D on vesicular motion</p><p>(7.58 MB AVI)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003290.s013\"><label>Video S6</label><caption><p>Effect of nocodazole on vesicular motion</p><p>(6.22 MB AVI)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003290.s014\"><label>Video S7</label><caption><p>Effect of calyculin A on vesicular motion</p><p>(10.30 MB AVI)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003290.s015\"><label>Video S8</label><caption><p>In situ QD and CFSE-labeled DCs inside draining lymph nodes</p><p>(4.61 MB AVI)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003290.s016\"><label>Video S9</label><caption><p>Localization of in situ-labeled DCs inside draining lymph nodes relative to B cell follicles</p><p>(4.15 MB AVI)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003290.s017\"><label>Video S10</label><caption><p>Ova-specific T cells alongside QD-labeled DCs inside lymph nodes</p><p>(7.69 MB AVI)</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"pone.0003290.s018\"><label>Video S11</label><caption><p>Ova-specific T cells cluster around QD^ova-labeled DCs inside lymph nodes</p><p>(4.94 MB AVI)</p></caption></supplementary-material>" ]

[ "<fn-group><fn fn-type=\"COI-statement\"><p><bold>Competing Interests: </bold>The authors have declared that no competing interests exist.</p></fn><fn fn-type=\"financial-disclosure\"><p><bold>Funding: </bold>This research was supported by the Graduate Research and Education in Adaptive bioTechnology (GREAT) Training Program of the UC Systemwide Biotechnology Research and Education Program, (grant 2004-11), and by National Institutes of Health grants GM-41514 (to M.D.C.), GM-48071 (to I.P.), biodefense subcontract N01-AI-50019 (to M.D.C), and the National Center for Microscopy and Imaging Research.</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"pone.0003290.g001\"/>", "<graphic xlink:href=\"pone.0003290.g002\"/>", "<graphic xlink:href=\"pone.0003290.g003\"/>", "<graphic xlink:href=\"pone.0003290.g004\"/>", "<graphic xlink:href=\"pone.0003290.g005\"/>", "<graphic xlink:href=\"pone.0003290.g006\"/>", "<graphic xlink:href=\"pone.0003290.g007\"/>", "<graphic xlink:href=\"pone.0003290.g008\"/>", "<graphic xlink:href=\"pone.0003290.e001.jpg\" mimetype=\"image\" position=\"float\"/>" ]

[ "<media xlink:href=\"pone.0003290.s001.doc\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003290.s002.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003290.s003.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003290.s004.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003290.s005.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003290.s006.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003290.s007.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003290.s008.avi\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003290.s009.avi\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003290.s010.avi\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003290.s011.avi\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003290.s012.avi\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003290.s013.avi\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003290.s014.avi\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003290.s015.avi\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003290.s016.avi\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003290.s017.avi\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"pone.0003290.s018.avi\"><caption><p>Click here for additional data file.</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[]

[ "<title>Patients and methods</title>", "<title>Eligibility criteria</title>", "<p>Patients aged &gt;18 years with histologically proven, locally advanced or metastatic carcinoma of the biliary tract (gallbladder, intrahepatic bile ducts, extrahepatic bile ducts and ampulla of Vater) were enroled in the study. Other eligibility criteria included: Eastern Cooperative Oncology Group performance status ⩽2; unidimensionally measurable disease (##REF##10655437##Therasse et al, 2000##); no prior chemotherapy for advanced disease; and adequate haematological (absolute neutrophil count &gt;1.5 × 10⁹ l⁻¹, platelets &gt;100 × 10⁹ l⁻¹), renal (creatinine &lt;1.5 × the upper limit of normal; ULN), and hepatic function (alanine aminotransferase &lt;5 × ULN; bilirubin &lt;2.5 × ULN). Patients with jaundice or evidence of bile duct obstruction and in whom the biliary tree could be decompressed by endoscopic percutaneous endoprosthesis, with a subsequent reduction in bilirubin to &lt;2.5 × ULN, were also eligible.</p>", "<p>Patients with prior malignancy or prior chemotherapy for advanced disease, central nervous system metastases or peripheral neuropathy grade ⩾2 were excluded from the study. Prior radiation therapy within 4 weeks of the first gemcitabine administration was not permitted. Women of childbearing potential were required to be neither pregnant nor breastfeeding and to be under active contraception.</p>", "<p>The study was conducted in compliance with Good Clinical Practice. All patients provided written informed consent.</p>", "<title>Treatment plan</title>", "<p>Treatment consisted of gemcitabine 1000 mg m⁻² as a 100-min i.v. infusion on day 1 followed by oxaliplatin 100 mg m⁻² as a 2-h i.v. infusion on day 2. Cycles were repeated every 2 weeks. Treatment was continued until disease progression, unacceptable toxicity, patient withdrawal of consent, or treatment delay of more than 3 weeks.</p>", "<p>Oxaliplatin was stopped altogether for grade 3 neurological symptoms but could be reintroduced upon recovery (to grade ⩽2). For paraesthesia without pain that persisted between cycles (lasting greater than 14 days), oxaliplatin was stopped until recovery, and then restarted at 75 mg m⁻². For paraesthesia with pain or functional impairment that persisted for greater than 7 but fewer than 14 days, the dose of oxaliplatin was reduced to 75 mg m⁻²; if the pain or functional impairment persisted between cycles, oxaliplatin was stopped. In the event of pharyngolaryngeal dysaesthesia during an infusion, the duration of oxaliplatin infusion was extended to 6 h for that and for subsequent infusions.</p>", "<p>For patients with a platelet count &lt;75 × 10⁹ l⁻¹, treatment was delayed for up to 3 weeks to allow recovery, and the oxaliplatin dose was reduced by 15% in subsequent cycles. For grade 3/4 thrombocytopenia, neutropenia, mucositis, diarrhoea or asthenia, the dose of gemcitabine was reduced to 800 mg m⁻² over 80 min and the dose of oxaliplatin was reduced to 85 mg m⁻² over 2 h. If grade 3/4 toxicity developed after dose reduction, the patient could be withdrawn from the study. In the event of oxaliplatin discontinuation for any toxicity, gemcitabine could be continued at the same dose and same schedule (every 2 weeks).</p>", "<title>Assessments and follow-up</title>", "<p>Tumour response was evaluated after four treatment cycles and every four cycles thereafter using the Response Evaluation Criteria in Solid Tumours (RECIST) unidimensional criteria (##REF##10655437##Therasse et al, 2000##). Objective responses were to be confirmed after at least 4 weeks. Tumour burden was assessed in both target and non-target lesions. Target lesions were defined as lesions that were measurable in at least one dimension, with the longest diameter ⩾20 mm using conventional techniques or the longest diameter ⩾10 mm measured using spiral computed tomographic (CT) scan. Computed tomographic and magnetic resonance imaging scans were the only accepted imaging techniques for the determination of response. Non-target lesions were defined as all other lesions and non-measurable lesions. The integrated response was derived from the investigator's assessment of target and non-target lesions, taking into account the appearance of new lesions in accordance with RECIST. For target lesions, SD was categorised as neither sufficient shrinkage to qualify for PR nor sufficient increase to qualify for progressive disease (PD), taking the the smallest sum of the longest diameter as reference since the treatment started; for non-target lesions, SD was categorised as neither CR or PD. Assignment to the SD category could only be made at least 6 weeks (two cycles) after the start of treatment. The best overall response was determined following a sequential review of all integrated responses recorded from the start of treatment until disease progression, with the last evaluation taken as the 30-day post-treatment follow-up. Sequential evaluations were compared, and the best response from each pair was compared with the subsequent pair to derive the best overall response.</p>", "<p>For patients who discontinued treatment for reasons other than disease progression, tumour evaluations were performed every 2 months. Whenever possible, the follow-up was continued until death.</p>", "<p>Safety evaluations were performed in the exposed population before administration of each cycle and toxicity was graded using National Cancer Institute Common Toxicity Criteria version 2.</p>", "<title>End points and statistics</title>", "<p>The primary objective of this study was to evaluate the efficacy (based on RR) of the GEMOX regimen as first-line therapy in patients with advanced BTCs and rejecting the treatment for further study, if it was found to be insufficiently active. The study employed a one-stage design (##REF##7082756##Fleming, 1982##) incorporating the following assumptions: H0: RR ⩽20% and H1: RR ⩾35%. On the assumption that of 56 evaluable patients, using a significance level of 0.05 and a power of 80%, GEMOX would be declared insufficiently active if ⩽12 responses were observed, and declared active if &gt;12 responses were observed. To be evaluable for response, patients were to receive a minimum of two cycles of GEMOX (ie, 6 weeks on study) and were to have had at least one post-baseline tumour assessment unless early progression occurred first, in which case, patients were considered evaluable for response. Assuming that 25% of patients would have non-measurable disease, a sample of 70 patients was required, so that a total of 56 patients attained the requirements of the evaluable patient population.</p>", "<p>An exploratory analysis was conducted to evaluate RRs by tumour type (GBC and CC). Kaplan–Meier survival curves were computed for time-to-event variables. Progression-free survival was calculated from the date of first treatment administration to the earliest date of disease progression, death or data cutoff. Overall survival was calculated from the date of first treatment administration to the date of death. Safety was reported for all subjects who received at least one dose of study drug.</p>" ]

[ "<title>Results</title>", "<title>Patient characteristics</title>", "<p>Between April 2003 and April 2005, 70 patients were enroled in the study, which was conducted in seven centres in France (2), Germany (2), Austria (1), Chile (1) and the United Kingdom (1). Patient and tumour characteristics at baseline are shown in ##TAB##0##Table 1##.</p>", "<title>Treatment received</title>", "<p>Three patients did not receive study treatment: two died before starting treatment (one with GBC, one with CC) and one patient with CC had hyperbilirubinaemia. The exposed population, therefore, comprised 67 patients. In total, 479 cycles of oxaliplatin and 521 cycles of gemcitabine were administered. The median number of cycles was five for both oxaliplatin (range: 1–18 cycles) and gemcitabine (range: 1–30 cycles). Mean relative dose intensities were 94.1% for oxaliplatin and 98.4% for gemcitabine in the exposed population.</p>", "<p>The primary reason for treatment discontinuation was disease progression (48 patients; 68.6%). Ten patients (14.3%) discontinued treatment as a result of adverse events (AEs), which were considered to be oxaliplatin-related hypersensitivity reactions in three patients (4.3%). Five patients continued treatment with gemcitabine alone after discontinuing oxaliplatin.</p>", "<title>Efficacy</title>", "<p>There were 10 PRs (14.9%;. 95% CI, 7.4–25.7%) in the exposed population (##TAB##1##Table 2##). A further five unconfirmed PRs were observed in the exposed population (three GBCs and three CCs). The majority of responses were observed in patients with CC: PRs were observed for 9/44 patients (20.5%) with CC and 1/23 patients (4.3%) with GBC.</p>", "<p>Median PFS was 3.4 months (95% CI, 2.5–4.6 months) for both the ITT and exposed populations (##FIG##0##Figure 1A##). Median PFS was 2.5 months for patients with GBC (95% CI, 1.6–4.3 months) and 3.8 months for patients with CC (95% CI, 2.7–5.6 months; ##FIG##0##Figure 1B##).</p>", "<p>Overall, 59 deaths were reported during the study period in the ITT population, 57 of which were in the exposed population. One patient died 1 month after the study cutoff date, but was considered alive for the purpose of the survival analysis. Median OS was 8.8 months (95% CI, 6.9–11.1 months) in the ITT population and 9.3 months (95% CI, 6.9–11.4 months) in the exposed population (##FIG##1##Figure 2A##). For both populations, median OS was 11.0 months for patients with non-GBC and 6.1 months for patients with GBC (##FIG##1##Figure 2B##).</p>", "<title>Safety</title>", "<p>Sixty-seven patients received at least one administration of study treatment and were included in the safety analysis. There were six deaths during the treatment period (between cycle 1 and cycle 4), none of which was considered by the investigators to be treatment-related. However, after reviewing the patient data, the sponsor could not rule out a relationship to study treatment for three deaths; one patient developed septicaemia and had global status deterioration, one patient had diarrhoea and vomiting leading to the rapid health deterioration and respiratory arrest, and one patient died from general health deterioration.</p>", "<p>Overall, nausea (82.1%) and vomiting (56.7%) of all grades were frequent side effects, despite systemic prophylactic measures (##TAB##2##Table 3##). Grade 3 nausea and vomiting occurred in 4.5% and 10.4% of patients, respectively, although there were no grade 4 events of this type. Overall, grade 3/4 AEs occurred in 47 patients (70.1%).</p>", "<p>Peripheral sensory neuropathies were observed in 67.2% of patients, with grade 3 neuropathy in 6.0%, including one case of laryngospasm. No grade 4 neuropathy was reported. In general, neuropathies were mild (grade 1) and intermittent during the first treatment cycles but tended to worsen and become more persistent as the number of treatment cycles increased. Neurotoxicity grade ⩾2 occurred in 40% of patients who received ⩾9 treatment cycles. Other frequently reported AEs included anaemia (77.6%), fatigue (73.1%), thrombocytopenia (68.7%), liver enzyme increase (62.7%), and weight loss (61.2%), although the majority of these events were grade 1/2 in severity.</p>" ]

[ "<title>Discussion</title>", "<p>This study is one among the largest of the phase II studies conducted to date to evaluate the efficacy and tolerability of chemotherapy in patients with advanced BTCs. The majority of patients enroled in this study had metastatic disease. The GEMOX administered every 2 weeks was well tolerated and conferred an ORR of 14.9%, with 50.7 % of patients achieving a PR or SD.</p>", "<p>The ORR is somewhat lower than previously reported for GEMOX (##REF##15319238##André et al, 2004##) or for gemcitabine in combination with other chemotherapy agents (##REF##16475213##Kim et al, 2006##; ##REF##17325704##Eckel and Schmid, 2007##; ##REF##17628484##Riechelmann et al, 2007##; ##REF##17364190##Lee et al, 2008##). Our goal of achieving a 20% RR in this phase II study was not reached, when responses were assessed using RECIST. However, the percentage of combined confirmed and unconfirmed PR (23.9%) was more in line with other published phase II studies in advanced BTCs, the majority of which report RRs that are often unconfirmed, as opposed to using RECIST confirmed responses. In our study, most unconfirmed PRs were those that did not persist from one assessment to the next (tumour response evaluation every 2 months).</p>", "<p>There is currently no standard chemotherapy regimen for advanced BTCs. Gemcitabine is one of the most widely used agents in this setting, with RRs of 12–35% for gemcitabine-based combination regimens, although these rates are unconfirmed in the majority of studies. A pooled analysis of studies (112 studies, including one phase III study; 2810 patients) performed between 1985 and 2006 demonstrated that the combination of gemcitabine with platinum compounds increased RRs and tumour-control rates compared with gemcitabine alone (##REF##17325704##Eckel and Schmid, 2007##). Response rates of 22–50% have been observed in single-institution studies, with OS of 7.6–15.4 months (##REF##15319238##André et al, 2004##; ##REF##16969352##Harder et al, 2006##; ##REF##16760298##Verderame et al, 2006##). However, single-institution studies often report higher RRs than multinational studies.</p>", "<p>For this study, we conducted an experimental subgroup analysis to assess response to treatment in patients with GBC and CC. Our study population comprised approximately one-third of the patients with GBC, who are generally considered to have a worse prognosis than patients with other BTCs (##REF##16234545##Gallardo et al, 2005##). In BTC, prognostic factors could be more important than therapy itself in determining treatment outcome. This study confirms the differential outcome between the gallbladder compared with other cancers of the biliary tree, as observed by ##REF##16234545##Gallardo et al (2005)##.</p>", "<p>In the pooled analysis by ##REF##17325704##Eckel and Schmid (2007)##, the median RR for trials in patients with GBC was higher than that in patients with CC (35.5 <italic>vs</italic> 17.7%; <italic>P</italic>=0.008). However, the response duration is short for GBC, and OS was significantly longer in trials of patients with CC compared with GBC (median 9.3 <italic>vs</italic> 7.2 months; <italic>P</italic>=0.048) (##REF##17325704##Eckel and Schmid, 2007##). In our study, RRs and RR+SD values in the GBC and CC subgroups were 4.0 <italic>vs</italic> 20.0% and 40 <italic>vs</italic> 53%, respectively. Concerning the survival, GBC was associated with a shorter PFS duration than CC (2.5 <italic>vs</italic> 3.8 months, respectively) and shorter median OS (6.1 <italic>vs</italic> 11.0 months, respectively). The reported efficacy,especially for GBC, differed markedly from a previous study of GEMOX in BTC, where the RR (unconfirmed) and OS were higher for GBC (<italic>n</italic>=11) <italic>vs</italic> CC (<italic>n</italic>=16; 54.4 <italic>vs</italic> 21.4%; 16.0 <italic>vs</italic> 14.5 months; ##REF##15319238##André et al, 2004##). The differences in the results of the two studies are surprising and could be related to the small number of patients, a centre effect, or the inclusion of unconfirmed responses.</p>", "<p>The combination of gemcitabine and oxaliplatin was generally well tolerated. Grade 3/4 AEs were reported in 70.1% of patients, although all patients experienced at least one AE of any grade. As previously observed with the GEMOX combination (##REF##15319238##André et al, 2004##), the most frequent grade 3/4 AEs were neutropenia, thrombocytopenia, pain and vomiting, all of which occurred in &lt;15% of patients. Neutropenia and thrombocytopenia appear to occur less frequently with the GEMOX regimen than with the combination of gemcitabine and cisplatin (##REF##16724985##Park et al, 2006##; ##REF##17364190##Lee et al, 2008##; ##REF##17597402##Meyerhardt et al, 2008##). Patients with advanced BTCs are very susceptible to infection and disease progression. Therefore, caution will be necessary with GEMOX chemotherapy like all chemotherapies in this disease. Grade 3 sensory neuropathy was uncommon (6.0%).</p>", "<p>In conclusion, this multinational study provides further evidence for the activity of GEMOX as a treatment for non-GBC, but also demonstrates the poor activity of this agent in GBCs. The combination of gemcitabine and oxaliplatin is well tolerated and provides a treatment option for patients with advanced BTCs, and in particular non-GBCs. A phase III study comparing GEMOX to gemcitabine is necessary to further establish the role of GEMOX in advanced BTCs. The design of such a study should include stratification for the location of the carcinoma (non-GBCs <italic>vs</italic> GBCs).</p>" ]

[]

[ "<p>Advanced biliary tract carcinomas (BTCs) are often diagnosed at an advanced/metastatic stage and have a poor prognosis. The combination of gemcitabine and oxaliplatin (GEMOX) has shown promising activity in this setting. This international phase II study evaluated the efficacy and safety of GEMOX as first-line therapy in patients with advanced BTCs. Eligible patients with previously untreated locally advanced or metastatic BTC received gemcitabine 1000 mg m⁻² (day 1) and oxaliplatin 100 mg m⁻² (day 2), every 2 weeks. Seventy patients were enroled; 72.9% had metastatic disease. Sixty-seven patients were treated. There were 10 confirmed partial responses (14.9%; 95% confidence interval (CI), 7.4–25.7%) in the treated population (RECIST). Twenty-four patients (35.8 %) had stable disease. The objective response rate was 20.5% in patients with non-gallbladder cancers (9/44 patients) and 4.3% in patients with gallbladder cancers (1/23). Median overall survival for the intent-to-treat population was 8.8 months (95% CI, 6.9–11.1%) and progression-free survival was 3.4 months (95% CI, 2.5–4.6%). Grade 3/4 toxicities included thrombocytopenia (14.9% of patients), alanine aminotransferase elevation (13.4%), anaemia (10.4%), neutropenia (11.9%) and pain (11.9%). In this study, GEMOX demonstrated activity in non-gallbladder carcinoma, but poor activity in gallbladder carcinoma. GEMOX is well tolerated in advanced BTCs.</p>" ]

[ "<p>Biliary tract carcinomas (BTCs), comprising gallbladder carcinoma (GBC) and intra- and extrahepatic cholangiocarcinoma (CC), are relatively rare in the United States and Europe (##REF##10536130##de Groen et al, 1999##). For example, approximately 5000 cases of GBC and 2000–3000 cases of CC are diagnosed annually in the United States. There are marked geographical variations in the incidence of GBC, although it is consistently more common in women than in men (##REF##11760569##Lazcano-Ponce et al, 2001##). The symptoms of BTC are non-specific and tumours have often reached an advanced stage at diagnosis. As such, the prognosis for patients with BTC is extremely poor: median survival is generally lesser than 6 months and estimated 1- and 2-year survival rates are 25 and 13%, respectively (##REF##11991810##Patel, 2002##). Chemotherapy is a palliative treatment option for patients with advanced disease. Owing to the lack of randomised phase III studies, there is no standard chemotherapy for advanced BTC. One clinical trial has demonstrated the improved survival for chemotherapy (5-fluorouracil plus leucovorin with or without etoposide) <italic>vs</italic> best supportive care (##REF##8879373##Glimelius et al, 1996##), although the ability of chemotherapy to prolong survival remains to be confirmed.</p>", "<p>Tolerability is of major importance when selecting palliative treatment regimens. Gemcitabine has palliative benefits and is generally well tolerated as therapy for advanced pancreatic carcinoma (##UREF##0##Heinemann, 2002##). Gemcitabine is also widely used as palliative therapy for advanced BTCs because of histogenetic similarities between the pancreas and biliary tract (##UREF##1##Scheithauer, 2002##).</p>", "<p>Gemcitabine has shown promising activity against advanced BTCs, with response rates (RRs) in the range of 12–35% when used in combination with agents such as 5-fluorouracil, cisplatin, mitomycin C, or capecitabine (##REF##14998852##Kornek et al, 2004##; ##REF##15558814##Alberts et al, 2005##; ##REF##16475213##Kim et al, 2006##; ##REF##17628484##Riechelmann et al, 2007##; ##REF##17364190##Lee et al, 2008##). A recent randomised phase II study suggested that combination chemotherapy with gemcitabine and cisplatin may be more effective than gemcitabine alone (##UREF##2##Valle et al, 2006##). The overall RR (ORR) was 24.3% for combination therapy and 15.2% for gemcitabine alone (complete response (CR)+partial response (PR)+stable disease (SD) was 75.7 <italic>vs</italic> 57.6%, respectively). Time to progression was also longer in the combination group (8.0 months) than in the monotherapy group (5.5 months). A follow-up study has been initiated with adequate power to assess the potential survival benefit of adding cisplatin to gemcitabine.</p>", "<p>Oxaliplatin was used as monotherapy in one phase II study as first-line treatment for patients with BTC. An objective RR of 20.6% was observed with an overall survival (OS) of 7 months (##REF##17047399##Androulakis et al, 2006##). Preclinical studies have demonstrated antitumour activity for the combination of gemcitabine and oxaliplatin (GEMOX) in human leukaemia and colorectal cancer cell lines and provide the rationale for using this combination in clinical studies. An optimal sequence-dependent synergy is apparent, with exposure to gemcitabine first and oxaliplatin later (##REF##10412945##Faivre et al, 1999##).</p>", "<p>A French phase II study (conducted in two centres) showed that the GEMOX combination was active and well tolerated as first-line chemotherapy in 36 patients with advanced BTCs (##REF##15319238##André et al, 2004##); ORR (without confirmed response for all patients) was 35.5% (95% confidence interval (CI), 18.7–52.3%), progression-free survival (PFS) was 5.7 months and OS was 15.4 months. We undertook the present international phase II study to evaluate the efficacy and tolerability of GEMOX as first-line chemotherapy in a larger group of patients with advanced BTCs.</p>" ]

[ "<p>The study was sponsored by sanofi-aventis.</p>" ]

[ "<fig id=\"fig1\"><label>Figure 1</label><caption><p>Progression-free survival (<bold>A</bold>) for the intent-to-treat population (<italic>n</italic>=70) and the (<bold>B</bold>) subgroup analysis of patients with gallbladder (<italic>n</italic>=25) and non-gallbladder (<italic>n</italic>=45) tumours.</p></caption></fig>", "<fig id=\"fig2\"><label>Figure 2</label><caption><p>Overall survival for the (<bold>A</bold>) Intent-to-treat population (<italic>n</italic>=70) and the (<bold>B</bold>) subgroup analysis of patients with gallbladder (<italic>n</italic>=25) and non-gallbladder (<italic>n</italic>=45) tumours.</p></caption></fig>" ]

[ "<table-wrap id=\"tbl1\"><label>Table 1</label><caption><title>Patient and tumour characteristics at baseline (intent-to-treat population)</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th colspan=\"3\" align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Tumour type</bold>\n<hr/></th></tr><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Characteristics</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Non-gallbladder (<italic>n</italic>=45)</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Gallbladder (<italic>n</italic>=25)</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>All (<italic>n</italic>=70)</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\"><italic>Gender,</italic> no. <italic>(%)</italic></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Male</td><td align=\"center\" valign=\"top\" charoff=\"50\">22 (48.9)</td><td align=\"center\" valign=\"top\" charoff=\"50\">6 (24.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">28 (40.0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Female</td><td align=\"center\" valign=\"top\" charoff=\"50\">23 (51.1)</td><td align=\"center\" valign=\"top\" charoff=\"50\">19 (76.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">42 (60.0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Median age, years (range)</td><td align=\"center\" valign=\"top\" charoff=\"50\">62 (30–83)</td><td align=\"center\" valign=\"top\" charoff=\"50\">55 (30–72)</td><td align=\"center\" valign=\"top\" charoff=\"50\">62 (30–83)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\"><italic>ECOG PS,</italic>\n<italic>no.</italic>\n<italic>(%)</italic></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 0</td><td align=\"center\" valign=\"top\" charoff=\"50\">23 (51.1)</td><td align=\"center\" valign=\"top\" charoff=\"50\">12 (48.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">35 (50.0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 1</td><td align=\"center\" valign=\"top\" charoff=\"50\">21 (46.7)</td><td align=\"center\" valign=\"top\" charoff=\"50\">10 (40.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">31 (44.3)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 2</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (2.2)</td><td align=\"center\" valign=\"top\" charoff=\"50\">3 (12.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">4 (5.7)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\"><italic>Primary tumour location,</italic>\n<italic>no.</italic>\n<italic>(%)</italic></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Gallbladder</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">25 (100.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">25 (35.7)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Intrahepatic bile ducts</td><td align=\"center\" valign=\"top\" charoff=\"50\">30 (66.7)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">30 (42.9)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Extrahepatic bile ducts</td><td align=\"center\" valign=\"top\" charoff=\"50\">13 (28.9)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">13 (18.6)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Intra/extrahepatic bile ducts</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (2.2)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (1.4)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Ampulla of Vater</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (2.2)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (1.4)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\"><italic>Prior treatment for BTC,</italic>\n<italic>no.</italic>\n<italic>(%)</italic></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Surgery</td><td align=\"center\" valign=\"top\" charoff=\"50\">18 (40.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">20 (80.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">38 (54.3)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Radiotherapy</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (4.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (1.4)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\"><italic>Disease status,</italic>\n<italic>no.</italic>\n<italic>(%)</italic></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Metastatic</td><td align=\"center\" valign=\"top\" charoff=\"50\">32 (71.1)</td><td align=\"center\" valign=\"top\" charoff=\"50\">19 (76.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">51 (72.9)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Locally advanced</td><td align=\"center\" valign=\"top\" charoff=\"50\">13 (28.9)</td><td align=\"center\" valign=\"top\" charoff=\"50\">6 (24.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">19 (27.1)</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl2\"><label>Table 2</label><caption><title>Best overall response to treatment by RECIST (exposed population)</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Response, <italic>n</italic> (%)</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Non-gallbladder (<italic>n</italic>=44)</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Gallbladder (<italic>n</italic>=23)</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Total (<italic>n</italic>=67)</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">CR</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0.0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Confirmed PR</td><td align=\"center\" valign=\"top\" charoff=\"50\">9 (20.5)</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (4.3)</td><td align=\"center\" valign=\"top\" charoff=\"50\">10 (14.9)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">SD</td><td align=\"center\" valign=\"top\" charoff=\"50\">15 (34.1)</td><td align=\"center\" valign=\"top\" charoff=\"50\">9 (39.1)</td><td align=\"center\" valign=\"top\" charoff=\"50\">24 (35.8)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">PD</td><td align=\"center\" valign=\"top\" charoff=\"50\">17 (38.6)</td><td align=\"center\" valign=\"top\" charoff=\"50\">10 (43.5)</td><td align=\"center\" valign=\"top\" charoff=\"50\">27 (40.3)<xref ref-type=\"fn\" rid=\"t2-fn2\">a</xref></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Not assessable</td><td align=\"center\" valign=\"top\" charoff=\"50\">3 (6.8)</td><td align=\"center\" valign=\"top\" charoff=\"50\">3 (13.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">6 (9.0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Overall tumour control rate (CR+PR+SD)</td><td align=\"center\" valign=\"top\" charoff=\"50\">24 (54.5)</td><td align=\"center\" valign=\"top\" charoff=\"50\">10 (43.5)</td><td align=\"center\" valign=\"top\" charoff=\"50\">34 (50.7)</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl3\"><label>Table 3</label><caption><title>Main National Cancer Institute Common Toxicity Criteria adverse events (exposed population)</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th colspan=\"3\" align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Number of patients (%) (<italic>n</italic>=67)</bold>\n<hr/></th></tr><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Adverse events</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>All grades 1–4</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Grade 3</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Grade 4</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Haematological</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Anaemia</td><td align=\"center\" valign=\"top\" charoff=\"50\">52 (77.6)</td><td align=\"center\" valign=\"top\" charoff=\"50\">5 (7.5)</td><td align=\"center\" valign=\"top\" charoff=\"50\">2 (3.0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Thrombocytopenia</td><td align=\"center\" valign=\"top\" charoff=\"50\">46 (68.7)</td><td align=\"center\" valign=\"top\" charoff=\"50\">9 (13.4)</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (1.5)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Neutropenia</td><td align=\"center\" valign=\"top\" charoff=\"50\">26 (38.8)</td><td align=\"center\" valign=\"top\" charoff=\"50\">5 (7.5)</td><td align=\"center\" valign=\"top\" charoff=\"50\">3 (4.5)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Neutropenic infection</td><td align=\"center\" valign=\"top\" charoff=\"50\">2 (3.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">2 (3.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0.0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Febrile neutropenia</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (1.5)</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (1.5)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0.0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Non-haematological</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Alanine aminotransferase increase</td><td align=\"center\" valign=\"top\" charoff=\"50\">42 (62.7)</td><td align=\"center\" valign=\"top\" charoff=\"50\">8 (11.9)</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (1.5)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Aspartate aminotransferase increase</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (1.5)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0.0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Hyperbilirubinaemia</td><td align=\"center\" valign=\"top\" charoff=\"50\">3 (4.5)</td><td align=\"center\" valign=\"top\" charoff=\"50\">3 (4.5)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0.0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Nausea</td><td align=\"center\" valign=\"top\" charoff=\"50\">55 (82.1)</td><td align=\"center\" valign=\"top\" charoff=\"50\">3 (4.5)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0.0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Vomiting</td><td align=\"center\" valign=\"top\" charoff=\"50\">38 (56.7)</td><td align=\"center\" valign=\"top\" charoff=\"50\">7 (10.4)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0.0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Weight loss</td><td align=\"center\" valign=\"top\" charoff=\"50\">41 (61.2)</td><td align=\"center\" valign=\"top\" charoff=\"50\">3 (4.5)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0.0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Fatigue</td><td align=\"center\" valign=\"top\" charoff=\"50\">49 (73.1)</td><td align=\"center\" valign=\"top\" charoff=\"50\">4 (6.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">3 (4.5)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Peripheral sensory neuropathy</td><td align=\"center\" valign=\"top\" charoff=\"50\">45 (67.2)</td><td align=\"center\" valign=\"top\" charoff=\"50\">4 (6.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0.0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Pain</td><td align=\"center\" valign=\"top\" charoff=\"50\">40 (59.7)</td><td align=\"center\" valign=\"top\" charoff=\"50\">8 (11.9)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0.0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Infection</td><td align=\"center\" valign=\"top\" charoff=\"50\">14 (20.9)</td><td align=\"center\" valign=\"top\" charoff=\"50\">5 (7.5)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0.0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Thrombosis</td><td align=\"center\" valign=\"top\" charoff=\"50\">3 (4.5)</td><td align=\"center\" valign=\"top\" charoff=\"50\">2 (3.0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (1.5)</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><fn id=\"t1-fn1\"><p>BTC=biliary tract carcinoma; ECOG PS=Eastern Cooperative Oncology Group performance status.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"t2-fn1\"><p>CR=complete response; PD=progressive disease; PR=partial response; RECIST=Response Evaluation Criteria in Solid Tumors; SD=stable disease.</p></fn><fn id=\"t2-fn2\"><label>a</label><p>For the intent-to-treat analysis, three patients who were not exposed to treatment were considered to have PD: two died before starting treatment (gallbladder carcinoma, one patient; cholangiocarcinoma, one patient) and one patient with cholangiocarcinoma had hyperbilirubinaemia.</p></fn></table-wrap-foot>" ]

[ "<graphic xlink:href=\"6604628f1\"/>", "<graphic xlink:href=\"6604628f2\"/>" ]

[]

{ "acronym": [], "definition": [] }

[]

[ "<title>Materials and Methods</title>", "<title>Patients, healthy blood donors, and breast cancer cell lines</title>", "<p>A total number of 136 consecutive, consenting breast cancer patients were included in the study. Operable patients (<italic>n</italic>=34) were admitted to the Mammazentrum at the Jerusalem Hospital in Hamburg (Germany), and tumour samples as well as normal adjacent tissues were collected during surgery. Non-malignant tonsils were collected from consenting patients during routine tonsillectomy performed at the Department of Otorhinolaryngology at the University Medical Center Hamburg-Eppendorf (Germany). Tissues were stored in RNAlater (Ambion, Austin, TX, USA) at −80°C until further use. Serum samples were obtained from breast cancer patients receiving adjuvant treatment or chemotherapy for metastasised disease at the Department of Oncology/Hematology at the University Medical Center Hamburg-Eppendorf. Serum samples from 44 blood donors were obtained as controls. The study protocol had received approval by the local ethics committee.</p>", "<title>Breast cancer cell lines and short-term-activating culture</title>", "<p>Human breast cancer cell lines, MDA-MB-231, MDA-MB-453, MDA-MB-468, MCF-7, ZR-75, BT-20, and CAMA-1, were kindly provided by the New York branch of the Ludwig Institute for Cancer Research. Cell lines were maintained in DMEM supplemented with 10% foetal bovine serum, 100 U ml⁻¹ penicillin and streptomycin (Invitrogen, Carlsbad, CA, USA). For cytokine-induced activation, cell lines were cultured for 72 h in complete medium with or without activating cytokines TNF-<italic>α</italic> (25 ng ml⁻¹; BD Bioscience, San Jose, CA, USA), INF-<italic>γ</italic> (100 U ml⁻¹; R&amp;D Systems, Minneapolis, MN), or both. Cells were harvested and were analysed for the expressions of CXCL13 and CXCR5 at baseline as well as after 24, 48, and 72 h.</p>", "<title>Reverse-transcription PCR</title>", "<p>Total RNA was extracted from cell lines applying the RNeasy Mini Kit (Qiagen, Hilden, Germany) and from tissue samples using guanidium isothiocyanate for denaturation, followed by caesium chloride gradient ultracentrifugation overnight (36 000 r.p.m., 4°C). Reverse transcription was performed using AMV reverse transcriptase (Promega, Madison, WI, USA) and was run at 42°C for 45 min with heat inactivation of the enzyme at 95°C for 5 min. Polymerase chain reaction primers and conditions used for the analysis are as follows: CXCL13 forward primer (F): 5′-GAGGCAGATGGAACTTGAGC-3′; CXCL13 reverse primer (R): 5′-CTGGGGATCTTCGAATGCTA-3′; CXCR5-F: 5′-CTTCGCCAAAGTCAGCCAAG-3′; CXCR5-R: 5′-TGGTAGAGGAATCGGGAGGT-3′; GADPH-F: 5′-TGATGACATCAAGAAGGTGG-3′; and GADPH-R: 5′-TTTCTTACTCCTTGGAGGCC-3′. For PCR analysis of single genes, 4 <italic>μ</italic>l first-strand cDNA (equivalent to 0.1 <italic>μ</italic>g RNA) was amplified after the preparation of 25 <italic>μ</italic>l of PCR reaction mixtures containing transcript-specific oligonucleotides (10 p<sc>M</sc>), 2 U of AmpliTaq Gold (Perkin Elmer, Weiterstadt, Germany), 10 n<sc>M</sc> of each dNTP (dATP, dTTP, dCTP, and dGTP), and 1.67 m<sc>M</sc> MgCl₂. After 35 PCR cycles, products were separated on 1.5% agarose gels, stained with ethidium bromide, visualised with ultraviolet light, recorded using a CCD camera, and assessed for expected size. Quality of cDNA was tested by reverse-transcription PCR (RT–PCR) using primers for housekeeping gene GAPDH. All RT–PCR experiments were performed at least twice. Negative controls without cDNA were integrated into all PCR reactions. To assess primer specificity, PCR products were analysed repeatedly by sequence analysis.</p>", "<title>Real-time PCR</title>", "<p>Primer sequences for target genes used in real-time are the same as the ones used in conventional PCR. A master mix of the following components was prepared at the indicated final concentrations: 4.0 m<sc>M</sc> MgCl₂, 400 n<sc>M</sc> forward and reverse primers, 200 n<sc>M</sc> dNTP (Invitrogen, Karlsruhe, Germany), 1% DMF, BSA at 250 <italic>μ</italic>g ml⁻¹, SYBR Green I (Sigma, St Louis, MO, USA) diluted at 1 : 20 000, and 1 Unit FastStart Taq DNA Polymerase (Roche Diagnostics, Branchburg, NJ, USA) in a total volume of 20 <italic>μ</italic>l. Next, samples were analysed using a LightCycler (Roche Diagnostics). After an initial denaturation at 95°C for 10 min, PCR reactions were cycled 40 times as follows: 10 s at 95°C, 5 s at adequate annealing temperature (CXCL13 and CXCR5 at 61°C; GADPH at 61°C) and 15 s at 72°C (elongation). Fluorescence intensity was measured at the end of each elongation phase. A melting curve analysis was carried out immediately after amplification. A standard curve prepared of the PCR product cloned into a pcDNA2.1 vector using the TA cloning kit (Invitrogen, Karlsruhe, Germany) was prepared to determine the concentration of target transcripts in cDNA samples. Results are given as copy numbers of the target gene/1 × 10⁶ copies of housekeeping gene GAPDH.</p>", "<title>Real-time PCR array</title>", "<p>A quantitative mRNA expression analysis of 84 chemokines/cytokines and their receptors was performed on 10 tumour tissue samples and 10 healthy breast tissue samples applying the chemokines/chemokine receptors RT2 profiler™ PCR Array (SuperArray, Frederick, MD). The following chemokine and related genes were analysed: chemokines: CCL1, CCL2, CCL3, CCL4, CCL5, CCL7, CCL8, CCL11, CCL13, CCL15, CCL16, CCL17, CCL18, CCL19, CXCL1, CXCL2, CXCL3, CXCL5, CXCL6, CXCL8, CXCL9, CXCL10, CXCL11, CXCL12, and CXCL13; chemokine receptors: CCR1, CCR2, CCR3, CCR4, CCR5, CCR6, CCR7, CCR8, CCR10, CCRL1, CCRL2, BLR1, CXCR3, CXCR4, CXCR6, and CYFIP2; other chemokines and related genes: AGTRL1, BDNF, C5, C5R1 (GPR77), CCBP2, CKLF, CKLFSF1, CKLFSF2, CKLFSF3, CKLFSF4, CMKLR1, CSF3, CX3CL1, CX3CR1, ECGF1, GDF5, GPR31, GPR77, GPR81, HIF1A, IL-13, IL-16, IL-18, IL-1A, IL-4, IL-8RA, LTB4R, MMP2, MMP7, MYD88, NFKB1, SCYE1, SDF2, SLIT2, TCP10, TLR2, TLR4, TNF, TNFRSF1A, TNFSF14, TREM1, VHL, XCL1, and XCR1. Extraction of RNA and reverse transcription were preformed as described above. PCR array analysis was performed according to the manufacturer's instructions using an iCycler System (Bio-Rad, Waltham, MA, USA). RNA expression of target genes was normalised for the expression of housekeeping gene 18S rRNA, and results were compared between malignant and normal samples according to the 2^{−ΔΔ<italic>C</italic>t} method. Results were considered significant if tumour-related relative mRNA expression was at least threefold higher or lower than that of autologous healthy tissue.</p>", "<title>Western blotting</title>", "<p>Protein lysates were prepared from tumour samples, healthy breast tissues, cell lines, and non-malignant tonsils using standard lysis buffer containing a protease inhibitor cocktail (Sigma-Aldrich, St Louis, MO, USA) and were subsequently denaturated for 10 min at 70°C. Samples of lysates containing 30 <italic>μ</italic>g of total protein were resolved on 4–12% Bis-Tris SDS–PAGE gels (Invitrogen, Carlsbad, CA, USA) under reducing conditions. Proteins were blotted on Hybond-ECL nitrocellulose membranes (Amersham Biosciences, Piscataway, NJ, USA), blocked overnight at 4°C with Top-Block (Fluka, Buchs, Switzerland), and incubated with 1 <italic>μ</italic>g of primary antibody directed against CXCL13 (clone 53610; R&amp;D Systems), CXCR5 (clone 51505; R&amp;D Systems), and <italic>β</italic>-actin (clone SC-47778; Santa Cruz Biotechnology, Santa Cruz, CA, USA) for 4 h at room temperature. Next, secondary HRP-labeled anti-mouse monoclonal antibody (R&amp;D Systems) was applied for 1 h at room temperature. Specific binding was visualised by chemiluminescence (ECL Western Blotting Analysis System, Amersham Biosciences).</p>", "<title>Flow cytometry</title>", "<p>For the analysis of CXCR5 protein expression, cell lines were fixed using FACS Lysing Solution (BD Bioscience) and were permeabilised using Permeabilising Solution (BD Bioscience). For analysis of cell surface expression, cells were stained and then fixed; for cytoplasmatic staining, cells were fixed and permeabilised before staining with a PE-conjugated anti-CXCR5 antibody (Clone 51505, R&amp;D Systems) or an appropriate isotype control. Samples were analysed using an FACSCalibur cytometer and CELLQuest software (BD Biosciences).</p>", "<title>Immunohistochemistry</title>", "<p>Immunohistochemistry for CXCL13 and CXCR5 was performed on formalin-fixed, paraffin-embedded tissue sections, which had been obtained for routine diagnostics. Briefly, slides were deparaffinised and pretreated with 10 mmol l⁻¹ citrate, pH 6.0 (Zymed, South San Francisco, CA, USA) in a steam pressure cooker (Decloaking Chamber; BioCare Medical, Walnut Creek, CA, USA) followed by washing in distiled water. All further steps were performed at room temperature in a hydrated chamber. The slides were pretreated with peroxidase block (Dako, Glostrup, Denmark), followed by blocking with goat serum diluted at 1 : 5 in 50 mmol l⁻¹ Tris-HCl, pH 7.4 for 20 min. Primary murine anti-human CXCL13 (clone 53610) and CXCR5 (clone 51505; both R&amp;D Systems) antibodies were applied at a 1 : 10 dilution in 50 mmol l⁻¹ Tris-HCl, pH 7.4 with 3% goat serum for 1 h. The slides were washed in 50 mmol l⁻¹ Tris-HCl, and goat anti-mouse horseradish peroxidase-conjugated antibody (Envision detection kit; Dako) was applied for 30 min. After further washing, immunoperoxidase staining was performed using a Diaminobenzidine Chromogen Kit (Dako), as per the manufacturer's instructions, and the slides were counterstained with haematoxylin.</p>", "<title>Enzyme-linked immunosorbent assay</title>", "<p>Serum concentrations of soluble CXCL13 protein were determined using a commercially available Quantikine kit (R&amp;D Systems) according to the manufacturer's instructions. After development of the enzyme-linked immunosorbent assay (ELISA) plates, absorbance was read at 450 nm using a spectrophotometer (SLT Labinstruments, Salzburg, Austria). The concentration of CXCL13 in the sera was interpolated from a standard curve, which was generated using the respective recombinant protein.</p>", "<title>Statistical analysis</title>", "<p>The Wilcoxon test was applied to results from tumour samples <italic>vs</italic> autologous healthy tissues, whereas the Mann–Whitney <italic>U</italic>-test was used to examine differences between samples of tumour patients and controls. Spearson's rank test was used to analyse correlations between gene expression and clinicopathological characteristics of the patients. Statistical analysis was performed using SPSS software (SPSS Inc., Chicago, IL, USA). Results were considered significant with <italic>P</italic>&lt;0.05.</p>" ]

[ "<title>Results</title>", "<title>Microarray analysis demonstrates strong CXCL13 overexpression in breast cancer tissue as compared with normal breast tissue</title>", "<p>Given the increasingly important role of chemokine/chemokine receptor interactions in the development and spreading of various malignancies, we aimed at applying a screening approach to look for chemokines that might have significant functions in the biology, and possibly in the pathophysiology, of breast cancer. We therefore performed a real-time PCR-based and pathway-focused microarray analysis on malignant and autologous healthy breast tissue samples from 10 patients whose tumours had been surgically removed. All patients included showed the same status of disease (T₂N₀M₀; stage II). Simultaneously examining mRNA expression levels of 84 chemokines/cytokines and their receptors in malignant and non-malignant tissue, we found 16 target genes that were differently expressed in breast cancer tissues as compared with their normal counterparts (##FIG##0##Figure 1A and B##). Six genes were significantly higher expressed in tumour tissue than in healthy tissue. Of these, chemokine CXCL13 was by far the most strongly and consistently overexpressed gene with a mean expression level in tumour samples that was 18 times higher than the one observed in autologous benign breast samples (##FIG##0##Figure 1B##).</p>", "<title>CXCL13 is consistently overexpressed at the RNA and protein levels in tumour samples of breast cancer patients</title>", "<p>To confirm our findings obtained by the RT–PCR array, we applied a CXCL13-specific real-time PCR to a larger set of samples from 34 breast cancer patients comprising all stages of disease (##TAB##0##Table 1##). Performing a comparative analysis of malignant and autologous healthy breast tissues, we observed readily detectable levels of CXCL13 in malignant as well as in non-malignant samples (##FIG##1##Figure 2A##). Comparing CXCL13 mRNA expression levels of malignant and normal breast tissue samples, however, we found that CXCL13 mRNA levels were significantly elevated in the malignant samples, confirming that increased expression of this chemokine is indeed a feature distinguishing tumour-infiltrated breast tissues from their healthy counterparts. Examining the presence of CXCR5, which is the only known receptor for chemokine CXCL13, within breast cancer tissues, we detected only a comparably weak expression (##FIG##1##Figure 2A##). We also did not observe any significant correlations between tumour RNA expression of CXCL13 or CXCR5 and clinical characteristics of individual patients (data not shown). However, we detected a strong and significant (<italic>P</italic>=0.0004) correlation between copy numbers of both genes analysed within breast cancer tissue (##FIG##1##Figure 2B##), indicating a biologically relevant role of CXCR5 as a receptor for CXCL13 in this solid tumour.</p>", "<p>Performing Western blot analyses of CXCL13 and CXCR5 expressions on lysates of six randomly selected breast cancer samples and four healthy breast tissue samples, we observed a weak but constantly detectable expression of the chemokine receptor CXCR5 in tumour samples, whereas protein expression of this chemokine receptor was not detectable in the healthy samples analysed (##FIG##1##Figure 2C##). In accordance with RNA expression levels, CXCL13 protein was commonly found in tumour samples, whereas it was absent from all healthy breast tissues analysed (##FIG##1##Figure 2C##). Marked overexpression of CXCL13 protein in breast cancer tissue was further confirmed performing a quantitative ELISA analysis with the same lysates of tumours and healthy controls. Remarkably, in this quantitative assay, protein concentrations of CXCL13 in breast cancer samples almost approached the levels found in non-malignant inflamed tonsils, which served as a positive control (##FIG##1##Figure 2D##).</p>", "<title>CXCL13 and CXCR5 are expressed intracellularly in breast cancer cell lines but are not detectable on the cell surface</title>", "<p>Given the significant overexpression of CXCL13 at the mRNA and protein levels in breast cancer tumour samples, we sought to determine the expression patterns of the chemokine ligand and its receptor in human breast cancer cell lines. We analysed six different breast cancer cell lines and found RNA and protein expressions of CXCL13 in all but one and CXCR5 expression in four of six cell lines (##FIG##2##Figure 3A##). Analysing the breast cancer lines by flow cytometry, we failed to detect a significant surface expression of the chemokine receptor, however, by performing intracellular staining, we observed strong expression of CXCR5 in the cytoplasma of cell lines ZR-75, BT-20, and MCF-7 (##FIG##2##Figure 3B##).</p>", "<p>Technical difficulties, mainly due to unspecific intracellular staining, did not allow for accurate analysis of CXCL13 expression by flow cytometry (data not shown). Therefore, we performed extensive analyses of CXCL13 protein concentrations in the supernatant of breast cancer cell lines that had been cultured for 72 h under different conditions to find shed CXCL13. However, culturing of cell lines with or without the addition of activating cytokine TNF-<italic>α</italic> or INF-<italic>γ</italic> did not result in the detection of CXCL13 in the supernatant (##FIG##2##Figure 3C##), whereas intracellular CXCL13 mRNA was consistently detectable using real-time PCR, independently from the addition of cytokines. Accordingly, CXCR5 mRNA was consistently detectable using real-time PCR, but the tumour cells remained negative for surface expression of the protein even after the addition of activating cytokines (##FIG##2##Figure 3C##).</p>", "<title>Epithelial tumour cells of patients with breast cancer show immunohistochemically detectable overexpression of CXCL13</title>", "<p>To further validate our findings of CXCL13 protein overexpression in human breast cancer and to localise the expression within a given tissue sample, we performed immunohistochemistry on breast cancer samples and autologous breast tissue samples of the same patients using lymph follicles as positive controls. Importantly, we did not observe significant numbers of tumour-infiltrating leukocytes expressing CXCL13 or CXCR5. However, we found a strong cytoplasmatic expression of CXCL13 in epithelial tumour cells, which was comparable with levels found in lymph follicles of draining lymph nodes from the same patients determined to be free of tumour (##FIG##3##Figure 4##). Benign epithelial cells also expressed CXCL13, however, to a much lesser extent. CXCR5 staining of the same specimens revealed a weak cytoplasmatic staining of epithelial cells with comparable levels between malignant and benign samples.</p>", "<title>Levels of CXCL13 protein are increased in the serum of breast cancer patients</title>", "<p>Breast cancer cell lines might not reflect the biological behaviour of original tumour tells and certain features, such as chemokines secretion or release might diminish or wear off during cell culture and repeated passages. Therefore, we next asked the question whether soluble CXCL13 protein could be detected in the <italic>in vivo</italic> setting, given the high levels of protein expression within breast cancer tissue samples as shown above. To this end, we analysed serum concentrations of CXCL13 protein in 44 healthy blood donors, 48 breast cancer patients without evidence of disease after surgical resection of their tumour, and 54 patients with metastatic breast cancer. Although there was no difference in CXCL13 serum concentrations between the healthy control group and patients without evidence of disease (##FIG##4##Figure 5##), we found significantly elevated serum concentrations of CXCL13 in patients with metastatic disease as compared with normal controls and with breast cancer patients without evidence of disease after surgical removal of their original tumour.</p>" ]

[ "<title>Discussion</title>", "<p>Besides our current report, there are very limited data on CXCL13 or CXCR5 expression in non-migratory cells, let alone in solid tumours. In one report, immunohistochemistry revealed CXCR5 expression in a significant proportion of human colon carcinoma specimens (##REF##15828050##Gunther et al, 2005##), whereas another group demonstrated cytoplasmatic expression in pancreatic cancer cells and cell lines (##REF##17018614##Meijer et al, 2006##). We show here for the first time that the ligand for CXCR5, CXCL13, is the most significantly overexpressed chemokine in breast cancer, supporting the idea of a role of CXCL13/CXCR5 interactions in promoting initiation and/or progression of this tumour type and, possibly, other human cancers. Furthermore, given the elevated CXCL13 expression in the serum of patients with metastatic disease as compared with patients without evidence of tumour burden, this chemokine may also serve as a tumour marker.</p>", "<p>In our analysis of breast cancer cell lines and tumour samples, we found the expression of chemokine receptor CXCR5 to be restricted to the cytoplasma, a finding that is in line with observations of ##REF##16331601##Muller et al (2006)## who detected CXCR5 intracellulary but did not detect surface expression of CXCR5 in cell lines and tumour samples from patients with metastatic head and neck cancer. Although these findings, at first glance, do not support the concept of CXCR5 as the main target for CXCL13 overexpression in breast cancer, we believe that they do not reflect the potential role this chemokine receptor might play <italic>in vivo</italic>. It may very well be that tumour cells express surface CXCR5 only during certain stages of tumour organisation and internalise the receptor after arrival within an area of maximum concentration of ligand CXCL13. This idea is supported by the findings of others who observed that, in the case of chronic lymphatic leukaemia, CXCR5 and its ligand were overexpressed in malignant B cells but that CXCR5 was downregulated after stimulation with soluble CXCL13 (##REF##17652619##Burkle et al, 2007##). Importantly, a concentration gradient- and time-dependent downregulation would also explain our observation of a strong correlation between CXCL13 expression and the presence of its receptor within breast cancer tissues despite the lack of CXCR5 overexpression in the same tissues.</p>", "<p>In addition, the hypothesis that an externalisation of receptor CXCR5 is restricted to certain phases of cancer development <italic>in vivo</italic> is also supported by the findings of ##REF##17018614##Meijer et al (2006)## who did not detect CXCR5 expression on a variety of tumour cell lines cultured <italic>in vitro</italic> but who found this chemokine receptor to be expressed on the surface of the same cells several days after injection into mice.</p>", "<p>What might the biological role of CXCR5/CXCL13 interactions be in the case of human cancers? Our finding of increased levels of soluble CXCL13 protein in the peripheral blood of breast cancer patients with advanced disease clearly suggests that this chemokine might be involved in the process of metastasisation. One might imagine at least three possible ways for CXCL13 to mediate a promoting effect on tumour development, the first two of them involving cellular immunity. First, it could be that breast cancer cells produce and release chemokines to allure cells, such as lymphocytes, monocytes, or dendritic cells, that are capable of secretion of cytokines directly promoting tumour cell growth and survival. Such an effect has been demonstrated for Hodgkin's lymphoma (##REF##10194423##Teruya-Feldstein et al, 1999##; ##REF##11238088##Buri et al, 2001##). Another possibility might be that CXCL13 mediates some form of protection against immune-mediated anti-tumour immunity. Such a phenomenon has been reported for leukaemias, where CXCL13-expressing malignant B cells showed an increased resistance against TNF-<italic>α</italic>-mediated apoptosis (##REF##15580304##Qiuping et al, 2005##; ##REF##17082584##Chunsong et al, 2006##).</p>", "<p>Our observation of a very low frequency of CXCL13-expressing leukocytes present within breast cancer tissue, however, would argue against the latter two hypotheses. Therefore, we favour the third possible explanation, suggesting that breast cancers cells might supply themselves with a growth advantage by the expression of CXCR5 and the release of CXCL13 <italic>in vivo</italic>. This view is also supported by a recent study demonstrating a growth advantage of CXCR5-positive cancer cells within the liver of mice presumably through CXCL13 produced by liver cells in their microenvironment (##REF##17018614##Meijer et al, 2006##).</p>", "<p>In our view, the most perspicuous explanation for the increased presence of CXCL13 in breast cancer is that CXCR5/CXCL13 interactions might function in a similar way as they do in lymphatic tissue. Breast cancer cells might interact through CXCR5/CXCL13 interactions to organise cellular cluster formation and compartmentalisation as has been shown for B cells in renal allografts (##REF##15780119##Steinmetz et al, 2005##) and in salivary glands of patients with Sjogren's syndrome (##REF##15934082##Barone et al, 2005##). This way, CXCL13 and its ligand might contribute significantly to tumour formation, and therapeutic interventions aiming at interrupting CXCL13/CXCR5 interactions might be of benefit for the clinical course of patients with breast cancer.</p>", "<p>It has recently been shown (##REF##17652619##Burkle et al, 2007##) that the stimulation of leukaemic B cells with CXCL13 results in a prolonged activation of p44/42 mitogen-activated protein kinases. Interestingly, p44/42 mitogen-activated protein kinase activation has also been shown to be involved in oncogenic transformation of epithelial breast cancer cells and their protection from oxidative stress (##REF##15782123##Zhu et al, 2005##; ##REF##17213808##Mohankumar et al, 2007##). Therefore, we suggest that this kinase pathway represents one possible route through which CXCL13/CXCR5 interactions might exert their biological function on a molecular level. Unfortunately, the lack of CXCR5 surface expression on breast cancer cell lines did not allow for consequent functional analyses of the role of CXCL13 in breast cancer <italic>in vitro</italic>. Further studies, however, are underway in our laboratory examining the exact consequences of CXCL13/CXCR5 interactions for the development and/or progression of breast cancer <italic>in vivo</italic>.</p>" ]

[]

[ "<p>The abilities of chemokines in orchestrating cellular migration are utilised by different (patho-)biological networks including malignancies. However, except for CXCR4/CXCL12, little is known about the relation between tumour-related chemokine expression and the development and progression of solid tumours like breast cancer. In this study, microarray analyses revealed the overexpression of chemokine CXCL13 in breast cancer specimens. This finding was confirmed by real-time polymerase chain reaction in a larger set of samples (<italic>n</italic>=34) and cell lines, and was validated on the protein level performing Western blot, ELISA, and immunohistochemistry. Levels of CXCR5, the receptor for CXCL13, were low in malignant and healthy breast tissues, and surface expression was not detected <italic>in vitro</italic>. However, we observed a strong (<italic>P</italic>=0.0004) correlation between the expressions of CXCL13 and CXCR5 in breast cancer tissues, indicating a biologically relevant role of CXCR5 <italic>in vivo</italic>. Finally, we detected significantly elevated serum concentrations of CXCL13 in patients with metastatic disease (<italic>n</italic>=54) as compared with controls (<italic>n</italic>=44) and disease-free patients (<italic>n</italic>=48). In conclusion, CXCL13 is overexpressed within breast cancer tissues, and increased serum levels of this cytokine can be found in breast cancer patients with metastatic disease pointing to a role of CXCL13 in the progression of breast cancer, suggesting that CXCL13 might serve as a useful therapeutic target and/or diagnostic marker in this malignancy.</p>" ]

[ "<p>Originally, chemokines (‘chemotactic cytokines’) and their receptors gained substantial scientific interest because of their central role in orchestrating immune responses by directing lymphocyte movement to the thymus, lymphoid tissues, and sites of inflammation (##REF##9459648##Luster, 1998##). Recent evidence, however, has shown that the unique abilities of chemokines in regulating cellular migration are utilised by a much wider variety of (patho-)biological networks, including developing haematological malignancies and solid tumours (##REF##15229479##Balkwill, 2004##).</p>", "<p>The more than 40 human chemokines described are divided into four groups as follows (##REF##12433287##Bacon et al, 2002##): the ‘CXC’ chemokines, defined by the presence of a single amino acid residue surrounded by two cysteines; the ‘CC’ chemokines, defined by two adjoined cysteine molecules; the ‘XC’ chemokines, defined by the loss of one cysteine residue; and the ‘C3XC’ chemokines (three interposed amino acids). Chemokines and their receptors are described by their family (e.g., CXC) combined with the letter R or L (‘receptor’ and ‘ligand’) and a member-specific number.</p>", "<p>Although it was self-evident that proteins guiding lymphocyte migration and formation should be involved in the development of haematological malignancies – as has been demonstrated for T- and B-cell lymphomas (##REF##12452848##Fujii et al, 2002##; ##REF##14707115##Basso et al, 2004##; ##REF##16287062##Ek et al, 2006##; ##REF##17071491##Harasawa et al, 2006##), multiple myeloma (##REF##17119115##Alsayed et al, 2007##), and leukaemias (##REF##11960346##Voermans et al, 2002##; ##REF##17082584##Chunsong et al, 2006##) – reports about the role of chemokines and their receptors in solid cancers have emerged only recently. For example, the expression of CXCR4 on melanoma cells was associated with increased rates of metastasis and patient mortality (##REF##15981210##Longo-Imedio et al, 2005##), whereas another study suggests that interleukin-8 (IL-8) and its two chemokine receptors CXCR1 and CXCR2 might promote prostate cancer progression through autocrine signalling of prostate cancer cells (##REF##15930347##Murphy et al, 2005##). Other authors found that CXCL12 might stimulate cell migration, cell growth, and invasion of ovarian cancer cells (##REF##12384559##Scotton et al, 2002##), and that CXCR4 was found to be expressed on the majority of glioma cell lines studied and on patients' tissue samples. After adding the CXCR4 agonist CXCL12, glioma cell lines were prevented from apoptosis and showed increased chemotaxis (##REF##12388552##Zhou et al, 2002##). A role of the chemokine/receptor pair CXCL12/CXCR4, normally controlling cell trafficking within the marrow (##REF##11197205##Nagasawa, 2000##), in the development of solid tumour bone marrow metastases has been shown for small-cell lung cancer (##REF##14603250##Burger et al, 2003##), breast cancer (##REF##11242036##Muller et al, 2001##), and renal cell cancer (##REF##16024619##Zagzag et al, 2005##). Accordingly, an entanglement of chemokine signalling pathways with tumour cell properties, such as proliferation, survival, adhesion, and chemotaxis, has been suggested (##REF##16409290##Laurence, 2006##).</p>", "<p>To gain insight into mechanisms by which chemokines might affect breast cancer development through local and microenviromental migration signalling, we analysed malignant and normal tissue samples from patients with breast cancer regarding their chemokine expression profiles.</p>", "<p>CXCL13 is a chemokine ligand originally termed B-cell-attracting chemokine 1 (bca-1), which is known to be expressed by stromal cells within B-cell follicles in secondary lymphoid tissues (##REF##9486651##Gunn et al, 1998##). It has a crucial function in germinal centre formation (##REF##15300245##Allen et al, 2004##) through interaction with its receptor CXCR5 expressed on follicular B cells (##REF##10770798##Bowman et al, 2000##). Under normal conditions, CXCL13 is furthermore expressed by follicular dendritic cells (##REF##11465112##Vissers et al, 2001##), macrophages (##REF##15284119##Carlsen et al, 2004##), and germinal centre T cells.</p>", "<p>We found that CXCL13 is overexpressed within breast cancer tissues at the mRNA and protein levels and that increased serum levels of this cytokine can be found in breast cancer patients with metastatic disease pointing to a possible role of CXCL13 within the development and progression of breast cancer.</p>" ]

[ "<p>This work was supported by grants from the Erich und Gertrud Roggenbuck-Stiftung, Eppendorfer Krebs- und Leukämiehilfe eV, Deutsche Krebshilfe, and from the Cancer Research Institute (to DA).</p>" ]

[ "<fig id=\"fig1\"><label>Figure 1</label><caption><p>CXCL13 is the most strongly overexpressed chemokine in breast cancer tissue as compared with normal breast tissue. Using a real-time PCR-based and pathway-focused microarray analysis on tissue samples from 10 patients whose tumours had been surgically removed, we simultaneously analysed mRNA expression levels of 84 chemokines/cytokines and their receptors in malignant and autologous healthy breast tissue. All patients included were stratified according to their status of disease (T₂N₀M₀, stage II). (<bold>A</bold>) Dots represent mean expression of single genes in malignant and healthy tissues after normalisation for housekeeping gene 18S rRNA. Dotted lines indicate the arbitrary cutoff value of threefold over- or underexpression in malignant <italic>vs</italic> normal tissues. (<bold>B</bold>) The microarray analysis revealed 16 target genes that showed at least three times higher or lower mRNA expression levels in malignant than in autologous healthy breast tissue. Black columns represent genes upregulated in tumours and white columns indicate genes that were downregulated as compared with normal breast tissue.</p></caption></fig>", "<fig id=\"fig2\"><label>Figure 2</label><caption><p>CXCL13 is consistently overexpressed in tumours of breast cancer patients. (<bold>A</bold>) Tumours and non-malignant breast tissues of 34 patients with breast cancer were analysed regarding the expression levels of CXCL13 and its receptor CXCR5 applying real-time PCR and results were normalised to the expression levels of housekeeping gene GAPDH. Black dots represent copy numbers of the target gene for each benign or malignant sample, respectively, and bars represent medians calculated. Benign and malignant samples were compared using Wilcoxon's test. (<bold>B</bold>) Performing a correlative analysis of CXCL13 and CXCR5 RNA expressions in breast cancer samples, a significant association between expression levels of both genes was observed. (<bold>C</bold>) Lysates of six tumour samples, four healthy breast tissues, and one non-malignant tonsil were analysed by Western blot for the protein expressions of CXCL13 and CXCR5. (<bold>D</bold>) To confirm findings obtained in the Western blot analysis, an ELISA was performed quantifying the absolute concentration of CXCL13 protein in lysates consisting of whole breast cancer protein.</p></caption></fig>", "<fig id=\"fig3\"><label>Figure 3</label><caption><p>CXCL13 and its receptor CXCR5 are expressed intracellularly in breast cancer cell lines. (<bold>A</bold>) Expressions of chemokine CXCL13 and its receptor CXCR5 were examined in six breast cancer cell lines applying conventional RT–PCR (upper rows) and Western blot (lower rows). Housekeeping genes GAPDH and <italic>β</italic>-actin served as internal controls. (<bold>B</bold>) Breast cancer cell lines ZR-75, BT-10, and MCF-7 were examined regarding the protein expression of CXCR5 using flow cytometry. Histograms indicate staining intensity applying anti-CXCR5 antibody (black) or an appropriate isotype control (grey). (<bold>C</bold>) Seven breast cancer cell lines were cultured for 72 h in complete medium with or without activating cytokine TNF-<italic>α</italic> or INF-<italic>γ</italic>. mRNA levels of CXCL13 and CXCR5 were evaluated at baseline as well as after 24, 48, and 72 h applying real-time PCR. At the same time points, the concentration of CXCL13 protein in the culture supernatant was quantified using an ELISA. Cell surface expression of CXCR5 protein was evaluated by flow cytometry.</p></caption></fig>", "<fig id=\"fig4\"><label>Figure 4</label><caption><p>Immunohistochemistry localises the overexpression of CXCL13 to epithelial tumour cells in tissue samples of breast cancer patients. Immunohistochemical staining using appropriate anti-CXCL13 and anti-CXCR5 antibodies was performed on 10 paraffin-embedded malignant or non-malignant tissue samples, respectively. Non-malignant lymph follicles removed from breast tissue served as a positive control for CXCL13 and CXCR5 stainings and showed homogenous staining throughout the node.</p></caption></fig>", "<fig id=\"fig5\"><label>Figure 5</label><caption><p>Serum concentrations of CXCL13 are elevated in breast cancer patients with metastatic disease. Serum concentrations of CXCL13 were analysed by ELISA in 44 healthy blood donors (white bar), 48 breast cancer patients without evidence of disease after surgical resection of their tumour and 54 patients with metastatic breast cancer (black bars). Bars show mean concentrations of CXCL13 plus standard deviations. Results were compared between groups applying the Mann–Whitney <italic>U</italic>-test.</p></caption></fig>" ]

[ "<table-wrap id=\"tbl1\"><label>Table 1</label><caption><title>Clinicopathological characteristics of breast cancer patients analysed for CXCL13/CXCR5 expression by real-time PCR</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"char\" char=\".\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Characteristics</bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>Number of patients per group</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Total</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">34</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Age (median and range) years</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">63 (43–85)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>pT status</italic>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> pT₁</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">13</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> pT₂</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">19</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> pT₃</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> pT₄</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>pN status</italic>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> pN₀</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">20</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> pN₁</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">9</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> pN₂</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> pN₃</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>M status</italic>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> M₀</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">30</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> M₁</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Stage</italic>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> I</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">10</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> II</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">15</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> III</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">4</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> IV</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Grading</italic>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Well differentiated</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">4</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Moderately differentiated</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">21</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Poorly differentiated</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">8</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><fn id=\"t1-fn1\"><p>A total of 34 patients with breast cancer were classified according to clinical features and pathological characteristics of their tumour samples. Assessment of T status and grading was only available for 33 patients, M status and classification into stages of the disease for 32 and 31 patients, respectively.</p></fn></table-wrap-foot>" ]

[ "<graphic xlink:href=\"6604621f1\"/>", "<graphic xlink:href=\"6604621f2\"/>", "<graphic xlink:href=\"6604621f3\"/>", "<graphic xlink:href=\"6604621f4\"/>", "<graphic xlink:href=\"6604621f5\"/>" ]

[]

{ "acronym": [], "definition": [] }

[]

[ "<title>Materials and Methods</title>", "<title>Patients</title>", "<p>A total of 2088 <italic>BRCA1</italic> mutation carriers, 841 <italic>BRCA2</italic> mutation carriers and 3 carriers of mutations in both genes ascertained from eight centres participating in CIMBA were included in this study (##TAB##0##Table 1##). The inclusion criteria for subjects is described elsewhere (##REF##17466083##Chenevix-Trench et al, 2007##).</p>", "<title>Genotyping</title>", "<p>Genotypes for the two polymorphisms – Ins16bp and Arg72Pro – were determined for each sample using previously described methodology (##REF##16419081##Osorio et al, 2006##). In some cases, the Ins16bp SNP was genotyped by DHPLC on the WAVE HT system (Transgenomic, Omaha, NE, USA) using an acetonitrile gradient and profiles analysed with the Navigator™ software (Transgenomic), and the Arg72Pro SNP was genotyped by Taqman (Applied Biosystems, Foster City, CA, USA). Hardy–Weinberg equilibrium (HWE) for each polymorphism was tested using the likelihood ratio test among unrelated individuals. The German Consortium of Hereditary Breast and Ovarian Cancer (GCHBOC) study gave HWE <italic>P</italic>-values of 0.005 and 0.043 for Ins16bp and Arg72Pro, respectively; therefore, all genotypes from that patient subset were confirmed by an alternative technique (DHPLC). In addition, individuals homozygous for Ins16bp were directly sequenced. The concordance rate was 100% in both instances; accordingly, the GCHBOC mutation carriers were included in all subsequent analyses. Call rates ranged between 92 and 100% across studies.</p>", "<title>Statistical analysis</title>", "<p>Haplotypes were imputed using the R-package ‘hapassoc’ (##UREF##0##Burkett and McNeney, 2006##). Associations of individual haplotypes with time to breast cancer or ovarian cancer diagnosis were evaluated using weighted Cox proportional hazards models, using age as the time variable (##REF##15880399##Antoniou et al, 2005##). Carriers were censored at the first occurrence of breast or ovarian cancer or bilateral prophylactic mastectomy. To allow for correlations between members of the same family, Huber and White robust estimators of variance were used, considering women clustered within families (##UREF##1##Huber, 1967##). The most frequent haplotype was taken as the reference and all other haplotypes were included in the multivariate model considering the number of copies of that particular variant. Models were adjusted for ethnicity, birth cohort and centre of recruitment. The analysis considered <italic>BRCA1</italic> and <italic>BRCA2</italic> mutation carriers separately (##TAB##1##Table 2##) and all carriers combined (data not shown).</p>" ]

[ "<title>Results and Discussion</title>", "<p>Genotype distributions and frequencies for the Ins16bp and Arg72Pro polymorphisms are shown in ##TAB##0##Table 1##. Allele frequencies were similar to those previously published (##REF##16419081##Osorio et al, 2006##), and genotype frequencies were consistent with HWE, except for the carriers from GCHBOC (see Materials and Methods). Haplotypes were inferred, and haplotype- and genotype-specific hazard ratios were estimated separately for each of breast (##TAB##1##Table 2##) and ovarian cancer (data not shown), among <italic>BRCA1</italic> and <italic>BRCA2</italic> mutation carriers. No evidence of association was found for any of the genotypes or haplotypes analysed with either breast or ovarian cancer risk, including the No Ins-72Pro haplotype, previously reported to be associated with an increased risk to develop a first primary tumour before 35 years of age in <italic>BRCA2</italic> mutation carriers (##REF##16419081##Osorio et al, 2006##).</p>", "<p>To confirm that this negative result was not due to the different analytic approach performed in this study, we carried out a logistic regression analysis, as was done in the original study (##REF##16419081##Osorio et al, 2006##), considering those with age at diagnosis younger than 35 as cases, and did not find a positive association between early diagnosis and this haplotype. In the original study, the result was corroborated by a functional assay (##REF##16419081##Osorio et al, 2006##), in which a decrease in apoptotic rate was found to be associated with the No Ins-72Pro haplotype. However, although concordant, both the genetic and the functional studies were limited by the small sample size (265 and 24 individuals, respectively), as reflected in the marginal statistically significant results described in that report.</p>", "<p>In summary, the previously reported association of the No Ins-72Pro haplotype in <italic>p53</italic> with an increased cancer risk in <italic>BRCA2</italic> mutation carriers (##REF##16419081##Osorio et al, 2006##) has not been validated in a larger series proceeding from the Consortium of Investigators of Modifiers of BRCA1 and BRCA2 (CIMBA). In this series of 2932 BRCA1/2 mutation carriers, no evidence of modification of breast or ovarian cancer risk by any of the two polymorphisms, Ins16bp and Arg72Pro, or their haplotype combinations has been detected. The lack of confirmation of a previously reported association found in a much smaller series highlights the necessity of international collaborative efforts aimed at achieving the statistical power required to reach reliable definitive conclusions in genetic association studies.</p>" ]

[ "<title>Results and Discussion</title>", "<p>Genotype distributions and frequencies for the Ins16bp and Arg72Pro polymorphisms are shown in ##TAB##0##Table 1##. Allele frequencies were similar to those previously published (##REF##16419081##Osorio et al, 2006##), and genotype frequencies were consistent with HWE, except for the carriers from GCHBOC (see Materials and Methods). Haplotypes were inferred, and haplotype- and genotype-specific hazard ratios were estimated separately for each of breast (##TAB##1##Table 2##) and ovarian cancer (data not shown), among <italic>BRCA1</italic> and <italic>BRCA2</italic> mutation carriers. No evidence of association was found for any of the genotypes or haplotypes analysed with either breast or ovarian cancer risk, including the No Ins-72Pro haplotype, previously reported to be associated with an increased risk to develop a first primary tumour before 35 years of age in <italic>BRCA2</italic> mutation carriers (##REF##16419081##Osorio et al, 2006##).</p>", "<p>To confirm that this negative result was not due to the different analytic approach performed in this study, we carried out a logistic regression analysis, as was done in the original study (##REF##16419081##Osorio et al, 2006##), considering those with age at diagnosis younger than 35 as cases, and did not find a positive association between early diagnosis and this haplotype. In the original study, the result was corroborated by a functional assay (##REF##16419081##Osorio et al, 2006##), in which a decrease in apoptotic rate was found to be associated with the No Ins-72Pro haplotype. However, although concordant, both the genetic and the functional studies were limited by the small sample size (265 and 24 individuals, respectively), as reflected in the marginal statistically significant results described in that report.</p>", "<p>In summary, the previously reported association of the No Ins-72Pro haplotype in <italic>p53</italic> with an increased cancer risk in <italic>BRCA2</italic> mutation carriers (##REF##16419081##Osorio et al, 2006##) has not been validated in a larger series proceeding from the Consortium of Investigators of Modifiers of BRCA1 and BRCA2 (CIMBA). In this series of 2932 BRCA1/2 mutation carriers, no evidence of modification of breast or ovarian cancer risk by any of the two polymorphisms, Ins16bp and Arg72Pro, or their haplotype combinations has been detected. The lack of confirmation of a previously reported association found in a much smaller series highlights the necessity of international collaborative efforts aimed at achieving the statistical power required to reach reliable definitive conclusions in genetic association studies.</p>" ]

[]

[ "<p>The close functional relationship between p53 and the breast cancer susceptibility genes <italic>BRCA1</italic> and <italic>BRCA2</italic> has promoted the investigation of various polymorphisms in the <italic>p53</italic> gene as possible risk modifiers in BRCA1/2 mutation carriers. Specifically, two polymorphisms in <italic>p53</italic>, c.97-147ins16bp and p.Arg72Pro have been analysed as putative breast cancer susceptibility variants, and it has been recently reported that a <italic>p53</italic> haplotype combining the absence of the 16-bp insertion and the presence of proline at codon 72 (No Ins-72Pro) was associated with an earlier age at the onset of the first primary tumour in <italic>BRCA2</italic> mutation carriers in the Spanish population. In this study, we have evaluated this association in a series of 2932 BRCA1/2 mutation carriers from the Consortium of Investigators of Modifiers of <italic>BRCA1</italic> and <italic>BRCA2</italic>.</p>" ]

[ "<p>Given the involvement of <italic>p53</italic> in cell cycle control, DNA repair and apoptosis, the role of this gene in cancer susceptibility has been extensively studied. Specifically, two polymorphisms in <italic>p53</italic>, c.97-147ins16bp and p.Arg72Pro have been analysed as putative breast cancer susceptibility variants, although not all studies have yielded consistent results (##REF##9037561##Weston et al, 1997##; ##REF##9677474##Wang-Gohrke et al, 1998##; ##REF##12471629##Suspitsin et al, 2003##; ##REF##17113725##Damin et al, 2006##; ##REF##17428325##Baynes et al, 2007##; ##REF##18230179##Costa et al, 2008##). The Arg72Pro single-nucleotide polymorphism (SNP) has gained special attention, as there is consistent evidence of functional differences in apoptotic rates between the Arg and Pro variants (##REF##11844595##Biros et al, 2002##; ##REF##11983757##Wu et al, 2002##; ##REF##12567188##Dumont et al, 2003##). In addition, the close functional relationship between <italic>p53</italic> and the breast cancer susceptibility genes <italic>BRCA1</italic> and <italic>BRCA</italic>2 (##REF##11694875##Jonkers et al, 2001##; ##REF##12802282##Ongusaha et al, 2003##; ##REF##17626182##Liu et al, 2007##) has promoted the investigation of the Arg72Pro SNP as a possible risk modifier in <italic>BRCA1/2</italic> mutation carriers (##REF##12676907##Martin et al, 2003##). Indeed, it was recently reported that a <italic>p53</italic> haplotype combining the absence of the 16-bp insertion and the presence of proline at codon 72 (No Ins-72Pro) was associated with an earlier age at onset of the first primary tumour in <italic>BRCA2</italic> mutation carriers in the Spanish population (##REF##16419081##Osorio et al, 2006##). In this study, we have evaluated this association in a series of 2932 <italic>BRCA1/2</italic> mutation carriers from the Consortium of Investigators of Modifiers of <italic>BRCA1</italic> and <italic>BRCA2</italic> (CIMBA) (##REF##17466083##Chenevix-Trench et al, 2007##).</p>" ]

[ "<p>CNIO study: we thank RM Alonso and RL Milne for their assistance. This study was partially supported by Mutua Madrileña, Genome Spain and Marato TV Foundations. The Milan study is supported by the Fondazione Italiana per la Ricerca sul Cancro. DKFZ study: we thank D Torres and MU Rashid for providing DNA samples and supplying data. AC Antoniou, L McGuffog and the CIMBA data management are funded by Cancer Research UK.</p>" ]

[]

[ "<table-wrap id=\"tbl1\"><label>Table 1</label><caption><title>Genotype distribution of the two <italic>p53</italic> polymorphisms in the <italic>BRCA1</italic> and <italic>BRCA2</italic> mutation carriers by participating study</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"left\"/><col align=\"left\"/><col align=\"char\" char=\"(\"/><col align=\"char\" char=\"(\"/><col align=\"char\" char=\"(\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\"(\"/><col align=\"char\" char=\"(\"/><col align=\"char\" char=\"(\"/><col align=\"char\" char=\".\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th colspan=\"4\" align=\"center\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>Ins16bp <italic>N</italic> (%)</bold>\n<hr/></th><th colspan=\"4\" align=\"center\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>Arg72Pro <italic>N</italic> (%)</bold>\n<hr/></th></tr><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Study</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Country of residence</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Ascertainment basis</bold>\n</th><th align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>No Ins</bold>\n</th><th align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>No Ins/16bp Ins</bold>\n</th><th align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>16bpIns</bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>Total</bold>\n</th><th align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>Arg72Arg</bold>\n</th><th align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>Arg72Pro</bold>\n</th><th align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>Pro72Pro</bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>Total</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">CNIO</td><td align=\"left\" valign=\"top\" charoff=\"50\">Spain and Greece<xref ref-type=\"fn\" rid=\"t1-fn2\">a</xref></td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinic</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">335 (74.12%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">105 (23.23%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">12 (2.65%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">452</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">281 (56.31%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">176 (35.27%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">42 (8.42%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">499</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">MBCSG</td><td align=\"left\" valign=\"top\" charoff=\"50\">Italy</td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinic</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">190 (65.07%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">91 (31.16%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">11 (3.77%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">292</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">156 (50.81%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">135 (43.97%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">16 (5.21%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">307</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">DKFZ</td><td align=\"left\" valign=\"top\" charoff=\"50\">Germany, Pakistan, Colombia</td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinic</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">128 (74.42%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">41 (23.84%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">3 (1.74%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">172</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">87 (51.18%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">67 (39.41%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">16 (9.41%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">170</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">GCHBOC<xref ref-type=\"fn\" rid=\"t1-fn3\">b</xref></td><td align=\"left\" valign=\"top\" charoff=\"50\">Germany</td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinic</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">593 (75.16%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">171 (21.67%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">25 (3.17%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">789</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">474 (56.97%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">294 (35.34%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">64 (7.69%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">832</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">HEBCS</td><td align=\"left\" valign=\"top\" charoff=\"50\">Finland</td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinic</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">148 (78.72%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">39 (20.74%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1 (0.53%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">188</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">96 (51.06%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">79 (42.02%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">13 (6.91%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">188</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">NCI</td><td align=\"left\" valign=\"top\" charoff=\"50\">United States</td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinic</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">160 (73.06%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">56 (25.57%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">3 (1.37%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">219</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">96 (50.26%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">81 (42.41%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">14 (7.33%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">191</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">IHCC</td><td align=\"left\" valign=\"top\" charoff=\"50\">Poland</td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinic</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">458 (67.25%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">202 (29.66%</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">21 (3.08%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">681</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">328 (48.16%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">289 (42.44%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">64 (9.40%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">681</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Total</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2012 (72.04%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">705 (25.24%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">76 (2.72%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2793<xref ref-type=\"fn\" rid=\"t1-fn4\">c</xref></td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1518 (52.93%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1121 (39.09%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">229 (7.98%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2869</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl2\"><label>Table 2</label><caption><title>Haplotype frequencies<xref ref-type=\"fn\" rid=\"t2-fn3\">a</xref> by mutation and disease status and HR estimates for breast cancer</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/><col align=\"center\"/><col align=\"char\" char=\".\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>Unaffected (%)</bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>Affected (%)</bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>HR</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>95% CI</bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold><italic>P</italic>-value</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td colspan=\"6\" align=\"left\" valign=\"top\" charoff=\"50\">BRCA1 <italic>mutation carriers</italic></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> <italic>p53</italic> haplotype</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> No Ins-Arg72/No Ins Arg72<xref ref-type=\"fn\" rid=\"t2-fn4\">b</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">49.60</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">50.50</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1.00</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> <bold>No Ins-72Pro</bold></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>One</bold><xref ref-type=\"fn\" rid=\"t2-fn5\">c</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>23.30</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>23.30</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>1.05</bold>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>0.84–1.32</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>0.64</bold>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>Two</bold><xref ref-type=\"fn\" rid=\"t2-fn6\">d</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>2.80</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>2</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>0.80</bold>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>0.47–1.38</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>0.42</bold>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Ins16bp-72Pro</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  One</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">23.80</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">23.10</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1.03</td><td align=\"center\" valign=\"top\" charoff=\"50\">0.83–1.28</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.79</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Two</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2.20</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2.10</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1.16</td><td align=\"center\" valign=\"top\" charoff=\"50\">0.54–2.50</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.70</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Ins16bp-Arg72</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  One</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">4</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3.30</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1.42</td><td align=\"center\" valign=\"top\" charoff=\"50\">0.96–2.10</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.08</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Two</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">—</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">—</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">—</td><td align=\"center\" valign=\"top\" charoff=\"50\">—</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td colspan=\"6\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>BRCA2 mutation carriers</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> <italic>p53</italic> haplotype</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> No Ins-Arg72/No Ins Arg72</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">47.50</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">55.90</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1.00</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> <bold>No Ins-72Pro</bold></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>One</bold></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>26.50</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>23.60</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>0.82</bold>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>0.53–1.26</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>0.35</bold>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>Two</bold></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>0.90</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>0.90</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>1.41</bold>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>0.56–3.55</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>0.46</bold>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Ins16bp-72Pro</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  One</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">26.50</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">19.40</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.81</td><td align=\"center\" valign=\"top\" charoff=\"50\">0.52–1.27</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.36</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Two</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2.10</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2.20</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.72</td><td align=\"center\" valign=\"top\" charoff=\"50\">0.14–3.86</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.70</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Ins16bp-Arg72</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  One</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2.70</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1.11</td><td align=\"center\" valign=\"top\" charoff=\"50\">0.42–2.97</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.83</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Two</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">—</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">—</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">—</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><fn id=\"t1-fn1\"><p>Abbreviations: GCHBOC=German Consortium of Hereditary Breast and Ovarian Cancer; HWE=Hardy–Weinberg equilibrium.</p></fn><fn id=\"t1-fn2\"><label>a</label><p>The CNIO series consisted of samples from the Spanish Consortium for the Study of Genetic Modifiers of <italic>BRCA1</italic> and <italic>BRCA2</italic> and the NCSR Demokritos, Athens (Greece). Cases from the original study were included in the analysis (##REF##16419081##Osorio et al, 2006##).</p></fn><fn id=\"t1-fn3\"><label>b</label><p>Deviation from HWE with <italic>P</italic>-values of 0.005 and 0.043 was observed for Ins16bp and Arg72Pro, respectively.</p></fn><fn id=\"t1-fn4\"><label>c</label><p>Missing genotypes are not included in the totals. Owing to technical difficulties, more failed genotypes were observed for the Ins16bp polymorphism.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"t2-fn1\"><p>Abbreviations: CI=confidence interval; HR=hazard ratio.</p></fn><fn id=\"t2-fn2\"><p>HRs corresponding to the haplotype associated with increased cancer risk in the original study are in bold.</p></fn><fn id=\"t2-fn3\"><label>a</label><p>Haplotypes were established or inferred only in those cases who had data for both polymorphisms.</p></fn><fn id=\"t2-fn4\"><label>b</label><p>Those individuals who were homozygous for the haplotype containing the common allele for both polymorphisms were considered as the reference group.</p></fn><fn id=\"t2-fn5\"><label>c</label><p>Individuals harbouring at least one given haplotype (heterozygous or homozygous)</p></fn><fn id=\"t2-fn6\"><label>d</label><p>Individuals homozygous for a given haplotype.</p></fn></table-wrap-foot>" ]

[]

[]

{ "acronym": [], "definition": [] }

[]

[ "<title>Materials and methods</title>", "<title>Cell cultures</title>", "<p>The following 10 CRC cell lines were used: DLD-1, HCT-8, HCT-116, HT-29, LoVo, Ls174T, SK-CO-1, SW48, SW480 and SW620 (ATCC, Manassas, VA, USA). Cells were cultivated in α-MEM supplemented with 10% foetal bovine serum, 100 IU ml⁻¹ penicillin and 100 <italic>μ</italic>g ml⁻¹ streptomycin (GIBCO-invitrogen, CA, USA). Two cell cultures were established for each cell line for RNA preparation. The first culture was harvested at semi-confluence (50–60%) when cells were actively cycling. Cells in the second culture were harvested 48 h after reaching full confluence, when cells were slow-growing. A replicate experiment was performed to obtain RNA for dye-reversed hybridisations.</p>", "<title>Patients</title>", "<p>Tumours from two cohorts of colon cancer patients were analysed. Patient follow-up was a minimum of 5 years. Disease-free survival (DFS) was calculated from surgery to the date of distant disease recurrence. Patients with rectal cancer, patients for whom the date of recurrence was unknown or received pre-operation adjuvant therapy were not included in this study.</p>", "<p>Cohort A consisted of 108 New Zealand patients, who underwent surgery at Dunedin and Auckland hospitals between 1995 and 2000, and included all disease stages. Thirty patients received 5-FU-based post-operative adjuvant chemotherapy. Ethical approval was obtained from the Ethics Committee, Otago University.</p>", "<p>Cohort B consisted of a group of 37 stage II German patients, who underwent surgery at the Technical University of Munich between 1995 and 2001. None of the patients received adjuvant therapy. Clinico-pathologic variables of both cohorts are summarised in ##TAB##0##Table 1##.</p>", "<title>Array preparation and gene expression analysis</title>", "<title>Cohort A tumours and cell lines</title>", "<p>Gene expression profiling of cohort A tumours and cell lines was performed using arrays spotted with the MWG 30 K Oligo Set (MWG Biotech, NC, USA). RNA was extracted from cell lines and fresh-frozen tissues, using Tri-Reagent (Progenz, Auckland, NZ) and purified using RNeasy mini columns (Qiagen, Victoria, Australia). Ten micrograms of total RNA were oligo-dT primed and cDNA synthesis was carried out in the presence of aa-dUTP and Superscript II RNase H-Reverse Transcriptase (Invitrogen, CA, USA). Cy dyes were incorporated into cDNA using the indirect amino-allyl cDNA labelling method. A pooled RNA sample derived from a mixture of 12 cell lines was used as a reference for all hybridisations.</p>", "<p>The Cy5-dUTP-tagged cDNA from an individual colorectal cell line or tissue sample was combined with Cy3-dUTP-tagged cDNA from the reference sample. The mixture was then purified using a QiaQuick PCR purification kit and co-hybridised to a microarray. Duplicate hybridisations for cell lines were performed following dye reversal.</p>", "<p>After scanning with a GenePix 4000B microarray scanner (Axon, CA, USA), the foreground intensities from each channel were log₂-transformed and normalised using the SNOMAD software (##REF##12424128##Colantuoni et al, 2002##). Normalised values were collated and filtered using BRB-Array Tools version 3.6.0-β_3 (by Dr Richard Simon and Amy Peng Lam, National Cancer Institute). Genes of low signal intensity, or for which more than 20% of measurements across tissue samples or cell lines were missing, were excluded from further analysis.</p>", "<title>Cohort B tumours</title>", "<p>Total RNA was prepared from each tumour using the RNeasy mini kit (Qiagen, Hilden, Germany). Ten micrograms of total RNA were used to synthesise double-stranded cDNA with SuperScript II reverse transcriptase (Gibco-Invitrogen, NY, USA) and an oligo-dT-T7 primer (Eurogentec, Koeln, Germany). The biotinylated cRNA was synthesised from the double-stranded cDNA using the Promega RiboMax T7-kit (Promega, Madison, WI, USA) and Biotin-NTP labelling mix (Loxo, Dossenheim, Germany). The biotinylated cRNA was then purified and fragmented. The fragmented cRNA was hybridised to Affymetrix HGU133A GeneChips (Affymetrix, Santa Clara, CA, USA) and stained with streptavidin–phycoerythrin. The arrays were scanned with an HP-argon laser confocal microscope and the digitised image data were processed using the Affymetrix® microarray suite 5.0 software. Background correction and normalisation were performed in the R computing environment (##UREF##0##Ihaka and Gentleman, 1996##) using the robust multi-array average algorithm.</p>", "<title>Quantitative real-time PCR</title>", "<p>The expression of seven randomly selected genes from the GPS (<italic>MAD2L1</italic>, <italic>POLE2</italic>, <italic>CDC2</italic>, <italic>MCM6</italic>, <italic>MCM3</italic>, <italic>TOPK</italic> and <italic>GMNN</italic>) was validated by quantitative real-time PCR (qRT–PCR) on an ABI Prism 7900HT Sequence Detection System using cell line cDNA and Taqman Gene Expression assays (Applied Biosystems, CA, USA). Relative fold changes were calculated using the 2^−ΔΔCT method (##REF##11846609##Livak and Schmittgen, 2001##) with Topoisomerase 3A as the internal control. Reference RNA was used as the calibrator to enable comparison between different experiments.</p>", "<title>Immunohistochemical analysis</title>", "<p>Ki-67 immunohistochemistry was performed on 40 cohort A tumours for which paraffin blocks were available, and an additional set of 33 rectal/rectosigmoid tumours was included to increase statistical power. Antigens were retrieved on 4-<italic>μ</italic>m sections in boiling citrate buffer (pH 6). Primary antibody (MIB-1, DakoCytomation, Denmark; dilution 1 : 50) was detected using the EnVision system (Dako EnVision) and the DAB substrate kit (Vector Laboratories, CA, USA). Five high-power fields were counted by two observers in a blinded manner. The Ki-67 proliferation index (PI) was presented for each tumour as the percentage of positively stained nuclei.</p>", "<title>Statistical analysis</title>", "<p>The K-means clustering method was applied to stratify clinical samples on the basis of the GPS expression level using the TIGR MeV 4.0 software (##REF##12613259##Saeed et al, 2003##). Using Pearson uncentered correlation, tumours from each cohort were assigned to two clusters (i.e., K-means clustering with <italic>K</italic>=2, 1000 iterations of clustering) with the threshold of occurrence in the same cluster set to 80%. The consensus clusters each contained tumours with similar GPS expression, resulting in two patient groups differing in their GPS levels. Either Fisher's exact test or the Kruskal–Wallis test was then used to evaluate the associations between the dichotomous GPS variable and clinico-pathologic parameters. Statistical analyses were performed using SPSS 15.0.0 (SPSS Inc., Chicago, IL, USA). For Ki-67 analysis, tumours were stratified into two clusters with the mean Ki-67 value as a cutoff point.</p>", "<p>The gene set comparison function from the BRB-Array tools software was used to analyse the GPS for differential expression among subgroups defined by the clinico-pathologic variables. The GPS was considered to have a higher-than-expected number of genes differentially expressed (DE) between the classes being compared if the Kolmogorov–Smirnov (KS) resampling <italic>P</italic>-value was less than 0.005 (default value). The distribution of KS statistics was obtained by 100 000 iterations of the random resampling process.</p>", "<p>Disease-free survival was plotted using the method of Kaplan and Meier, and a log-rank test was used to test for differences in survival time between defined clusters of patients according to the GPS or Ki-67 PI. A multivariate Cox proportional hazards model was developed using forward stepwise regression with predictive variables that were significant in the univariate analysis. Cox multivariate regression was not relevant for measuring the performance of the GPS in cohort B, as this population of tumours included only stage II colon cancers.</p>" ]

[ "<title>Results</title>", "<title>Derivation of a GPS</title>", "<p>To identify a set of genes whose expression was associated with tumour cell proliferative activity, two proliferation-based systems were analysed and integrated as described below and illustrated in ##FIG##0##Figure 1##.</p>", "<p>\n<italic>A. Gene expression analysis of a CRC in vitro model</italic>\n</p>", "<p>The <italic>in vitro</italic> system involved identification of genes that were to reflect the proliferative activity of CRC cell lines. Genes DE between exponentially growing (nonconfluent) and growth-inhibited (confluent) cell lines were identified (##FIG##0##Figure 1A–C##). Firstly, DE genes between Cy5-labelled nonconfluent and confluent samples were identified by statistical analysis of microarray (two-class paired, FDR&lt;1%; ##REF##11309499##Tusher et al, 2001##). Secondly, DE genes between Cy3-labelled nonconfluent and confluent samples (biological replicates) were identified using the same approach. To minimise gene-specific dye bias and other sources of variation, only genes that were present in both SAM-generated gene sets were selected. The gene set was further reduced to genes whose expression was consistently altered in the same direction in at least 8 out of 10 cell lines, yielding a total 881 DE genes with known annotation (##FIG##0##Figure 1D##).</p>", "<p>Gene ontology (GO) analysis showed that categories related to cell cycle and DNA metabolism were the most over-represented biological themes among the DE genes, mainly because of genes that were overexpressed in exponentially growing cells (<italic>P</italic>&lt;10⁻⁵).</p>", "<p>\n<italic>B. Gene expression profile of the proliferative compartment of colon crypts</italic>\n</p>", "<p>The second gene set used for the design of a GPS was based on the physiological expression profile of human colon crypts. ##REF##17881565##Kosinski et al (2007)## compared the proliferative bottom part of crypts with the differentiated crypt top, and identified 299 DE genes highly expressed in the proliferative bottom (##FIG##0##Figure 1E##). The GO terms that were over-represented within this gene list were related to cell proliferation and renewal, consistent with the physiological function of the bottom crypt compartment.</p>", "<p>\n<italic>C. Definition of the GPS</italic>\n</p>", "<p>To define a final GPS enriched in key-proliferative genes, the <italic>in vitro</italic> and <italic>in vivo</italic>-derived DE gene lists were screened for common genes (##FIG##0##Figure 1F##). Thirty-six genes were found to be overexpressed in both exponentially growing CRC cell lines and the proliferative crypt compartment. These genes were defined as GPS that included 15 cell cycle-related genes (<xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Table 2</xref>).</p>", "<p>The expression of 7 genes randomly selected from the 15 cell cycle-related genes was validated by qRT–PCR on the cell line cDNAs. A close correlation between qRT–PCR and array data was observed (<xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Figure 1</xref>).</p>", "<title>Classification of colon tumours according to the GPS expression level</title>", "<p>To investigate the association between the GPS and patient parameters, expression values of the GPS genes were first obtained from the array-generated expression profiles. Expression data were available for all 36 genes in cohort A, and 35 genes (except CDCA5) in cohort B tumours. For each cohort, tumours were split into two consensus clusters on the basis of their GPS expression using K-means clustering. Tumours with relatively high or low GPS expression were defined as having high or low proliferative activity, respectively (##FIG##0##Figure 1G##).</p>", "<title>Reduced expression of the GPS is associated with unfavourable clinico-pathologic variables</title>", "<p>Intriguingly, we observed an association between <italic>reduced</italic> GPS expression and an increased risk of recurrence in both cohorts (##TAB##0##Table 1##). Groups with reduced GPS expression were significantly enriched for recurrent tumours (<italic>P</italic>=0.021 and 0.005 for cohort A and B, respectively). In cohort A, reduced GPS expression was also associated with higher disease stage (<italic>P</italic>=0.015). Further, reduced GPS expression was significantly more frequent in cohort A tumours with lymphatic invasion compared with those without lymphatic invasion (<italic>P</italic>=0.018). Gene set comparison analysis confirmed that GPS contained a higher-than-random proportion of DE genes among subgroups with clinico-pathologic parameters for which a significant association with the GPS was found in nonparametric tests (<xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Table 3</xref>).</p>", "<title><italic>Reduced</italic> GPS expression is associated with <italic>shorter</italic> DFS in colon cancer</title>", "<p>To examine whether a difference in cell proliferation determined by the GPS may be associated with time to recurrence, DFS was plotted for low GPS and high GPS tumours (##FIG##1##Figure 2##). DFS was significantly shorter in patients with reduced GPS expression (<italic>P</italic>=0.033 and 0.011 for cohort A and B patients, respectively; ##FIG##1##Figure 2A and C##). This association remained significant in cohort A, when patients with adjuvant therapy were excluded (<italic>P</italic>=0.029; ##FIG##1##Figure 2B##).</p>", "<p>If prolifetative activity is a crucial factor influencing DFS, then a GPS consisting of only genes directly implicated in cell cycle process should be sufficient to separate patients into groups with different survival rates. Therefore, the GPS was reduced to the 15 cell cycle-regulated genes and survival analysis was repeated. The modified GPS stratified patients into groups that differed in their DFS more stringently (<italic>P</italic>=0.022 and 0.003 for cohort A and B patients, respectively; ##FIG##1##Figure 2D–F##) than the original GPS including all 36 genes. Therefore, colon cancer patients with tumours of lower proliferative activity are more likely to recur.</p>", "<p>When the parameters predicting patient outcome in univariate analysis were investigated in a multivariate model, disease stage was the only independant predictor of 5-year DFS in cohort A (##TAB##1##Table 2##).</p>", "<title><italic>Increased</italic> GPS expression is associated with <italic>shorter</italic> DFS in breast cancer</title>", "<p>As the association between reduced GPS expression and poor colon cancer prognosis was an unexpected finding, we tested the validity of our GPS on public array data from two independant breast cancer cohorts. Using these data, an association between increased proliferation and bad outcome has been established earlier (##REF##12490681##van de Vijver et al, 2002##; ##REF##16280042##Pawitan et al, 2005##).</p>", "<p>For each breast cancer data set, patients were stratified into two groups with either low or high GPS expression and their DFS was plotted. In both data sets, patients with increased GPS expression had significantly shorter DFS compared with patients with reduced GPS expression (<italic>P</italic>&lt;0.0001; ##FIG##2##Figure 3##), confirming the previously established association. Therefore, the association between low proliferation and bad colon cancer outcome is not the result of biased methodology.</p>", "<title>Ki-67 PI is not associated with outcome</title>", "<p>Ki-67 PI ranged from 25 to 96%, with a mean value of 76.3±17.5 and a median value of 81.8%. When these 73 patients were stratified into two groups differing in their GPS expression, a significant difference in DFS was apparent (<italic>P</italic>=0.01; <xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Figure 2C</xref>). However, when patients were stratified into two groups according to the mean Ki-67 PI, the DFS of the group with a low PI was similar to that with a high PI (<italic>P</italic>=0.55; <xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Figure 2D</xref>). Furthermore, when analysis was limited to patients with the highest and the lowest Ki-67 values, no statistical difference in DFS was observed (data not shown). No correlation was found between the Ki-67 PI and mean GPS expression of tumours (Spearman <italic>R</italic>=0.06; <italic>P</italic>=0.63).</p>" ]

[ "<title>Discussion</title>", "<p>Cancer is regarded as a proliferative disorder, where a selective growth advantage is believed to crucially contribute to its development and progression. Supporting this concept are many studies that have shown an association between poor clinical outcome and high expression of proliferation-associated genes (##UREF##1##Krasnoselsky et al, 2004##; ##REF##15899795##Dai et al, 2005##; ##REF##16899601##Kirschner-Schwabe et al, 2006##).</p>", "<p>Our study is the first to report an inverse relation between cancer malignancy and the expression of a multi-gene proliferation signature. In two independant cohorts of colon cancer patients, we observed an association between an increased recurrence risk and the reduced expression of a GPS. This finding challenges the long-held belief that rapidly dividing cancer cells are a harbinger of poor prognosis, and suggests that low proliferative activity is a biological feature of the majority of aggressive colon cancers.</p>", "<p>The GPS we designed is highly likely to directly reflect the proliferative state of a tumour, as it was derived from both human colon crypts and an <italic>in vitro</italic> CRC model. All genes included in the GPS were overexpressed in actively proliferating cells of the both systems. With respect to the <italic>in vitro</italic> system, the comparison of exponentially growing cancer cells with contact-inhibited cancer cells has limitations; however, well established evidence indicates that many tumour cell lines maintain a variable degree of density-dependant growth suppression that is characteristic of the stationary phase (##REF##1506857##Couldwell et al, 1992##; ##REF##12789269##Steinman et al, 2003##; ##REF##15492842##Kuhn et al, 2004##; ##REF##15718252##Motti et al, 2005##). Ten CRC cell lines were included to ensure an overall growth suppression, and only genes were considered that were altered in at least 80% of the cell lines. By the inclusion of only DE genes that overlapped with the human colon proliferation signature, the final GPS consisted of genes that strongly correlate with both colon cancer cell growth and physiological colon proliferation. Consistent with this, ontology analysis indicated that proliferation-associated genes represent the main biological theme in both the <italic>in vitro</italic> and the <italic>in vivo</italic> system.</p>", "<p>Further evidence supporting the association of the GPS with cell proliferation stems from a considerable overlap in genes or gene families identified between our GPS and other proliferation signatures defined for tumours of the breast (##REF##10430922##Perou et al, 1999##, ##REF##10963602##2000##), ovary (##REF##11158614##Welsh et al, 2001##), liver (##REF##12058060##Chen et al, 2002##), acute lymphoblastic leukaemia (##REF##16899601##Kirschner-Schwabe et al, 2006##), neuroblastoma (##UREF##1##Krasnoselsky et al, 2004##), lung squamous cell carcinoma (##REF##16007138##Inamura et al, 2005##), head and neck (##REF##15144956##Chung et al, 2004##), prostate (##REF##12154061##LaTulippe et al, 2002##) and stomach (##REF##11782383##Hippo et al, 2002##). Comparing these published data, ##REF##16491069##Whitfield et al (2006)## identified a core set of genes common to various proliferation signatures. As expected, these genes (MYBL2, PLK1, CDC2 and MCM genes) are also contained within our GPS (see <xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Table 2</xref>), reflecting the universal mechanisms that govern human cell division. Indeed, by reanalysis of public breast cancer data, our GPS was shown to perform properly in other cancer types as well. Therefore, our GPS appears to be a reliable tool for the assessment of proliferation in colonic tumours.</p>", "<p>Application of our GPS to colon cancer patient data revealed a robust association between low proliferative activity and increased likelihood of recurrence. Firstly, the low GPS group had reduced DFS in two independant cohorts derived from different populations. Secondly, expression data from the two cohorts were obtained using two different array platforms, indicating that the observed association was not subject to methodological bias. Thirdly, reduced GPS expression in cohort A also correlated with clinico-pathological variables related to poor outcome (stage, lymphatic invasion). A possible confounding factor in our study was the chemotherapy treatment as given in 28% of cohort A patients. Exclusion of these patients from analysis had no effect on the association strength, suggesting that proliferation affects patient outcome independant of adjuvant chemotherapy.</p>", "<p>Notably, the observed association was not independant of tumour stage. In other words, higher disease stages were enriched for slowly proliferating tumours, but tumours with high GPS expression were also present. It remains possible, however, that these fast proliferating tumours had progressed slowly before they were diagnosed at an advanced stage. It is likely that tumour stage is a better prognostic factor because the presence of lymph node and distant organ involvement is a direct manifestation of metastasis. It is noteworthy that the goal of this study was not to develop a prognostic tool based on the GPS, but to determine the nature of the relationship between proliferation and the degree of malignancy in colon cancer.</p>", "<p>Together, the analyses we performed are all consistent with a marked effect of colon cancer proliferation on the rate of its recurrence. This is an important finding because studies using conventional proliferation markers have not been able to establish a clear-cut association between colon cancer proliferation and outcome. On account of these reported inconsistencies, we also performed DFS analysis on the basis of proliferation estimated by conventional Ki-67 labelling. Unlike the GPS, Ki-67 PI failed to separate a set of 73 CRCs into groups associated with distinct survival (<xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Figure 2</xref>). Although the advantage of assessing multiple genes as opposed to one proliferation marker appears obvious, additional parameters are likely to contribute to the inferior performance of the Ki-67 PI. Ki-67 positivity reflects the number of cells in cycle, rather than their cycling speed (##REF##10837136##Endl and Gerdes, 2000##). Furthermore, the visual scoring of Ki-67 positive nuclei is inherently subjective, confounding correlations particularly in the case of small sample sizes (##REF##10845252##Barzanti et al, 2000##). It is of interest, however, that a low Ki-67 PI has been associated with poor outcome in the three studies with the highest statistical power (i.e., largest sample sizes) (##REF##12525515##Allegra et al, 2003##; ##REF##15117979##Garrity et al, 2004##; ##REF##16400510##Hilska et al, 2005##).</p>", "<p>It is difficult to assess from our data whether slow proliferation directly enhances the ability of a colon cancer to metastasise or fast proliferation may be disadvantageous for colon cancers to acquire metastatic capacity. Although it appears intuitively logical that accelerated cell division allows for more genetic events required for progression to accumulate, fast proliferation might also have a negative impact on a tumour ability to survive and spread. For example, rapidly growing tumours may elicit a stronger and more effective immune response. Further, highly proliferative tumours may contain a lower proportion of cancer stem cells thought to undergo relatively slow divisions. Alternatively, a high level of genetic instability may increase the invasive potential of cancer cells or their resistance towards apoptosis, but may interfere with smooth replication. Also, hypoxic conditions may slow down the growth rate of tumours but could promote the onset of epithelial–mesenchymal transitions leading to invasion. Possible underlying mechanisms are currently under investigation in our laboratory.</p>", "<p>The intriguing finding of an inverse relationship between tumour proliferative activity and disease aggressiveness suggests that fundamental biological differences exist for the mechanisms that drive disease progression in colon cancer compared with other epithelial malignancies. The delineation of the underlying processes will be important not only for the understanding of colon cancer biology, but also for the design of new therapeutic strategies. In this context, the development of alternative therapies might be a pertinent issue, given that current chemotherapeutic agents are usually designed to target colon cancer progression by killing rapidly proliferating cells.</p>" ]

[]

[ "<p>The association between cell proliferation and the malignant potential of colon cancer is not well understood. Here, we evaluated this association using a colon-specific gene proliferation signature (GPS). The GPS was derived by combining gene expression data obtained from the analysis of a cancer cell line model and a published colon crypt profile. The GPS was overexpressed in both actively cycling cells <italic>in vitro</italic> and the proliferate compartment of colon crypts. K-means clustering was used to independantly stratify two cohorts of colon tumours into two groups with high and low GPS expression. Notably, we observed a significant association between reduced GPS expression and an increased likelihood of recurrence (<italic>P</italic>&lt;0.05), leading to shorter disease-free survival in both cohorts. This finding was not a result of methodological bias as we verified the well-established association between breast cancer malignancy and increased proliferation, by applying our GPS to public breast cancer data. In this study, we show that reduced proliferation is a biological feature characterizing the majority of aggressive colon cancers. This contrasts with many other carcinomas such as breast cancer. Investigating the reasons underlying this unusual observation may provide important insight into the biology of colon cancer progression and putative novel therapy options.</p>" ]

[ "<p>Defective regulation of cell proliferation is a fundamental feature of cancer. By providing a genome-wide insight, microarray technology has revealed that conserved tumour expression patterns include many proliferation-associated genes termed as ‘proliferation signatures’. In most cancers, an increased expression of proliferation signatures has been associated with enhanced malignancy (##REF##12620412##Rosenwald et al, 2003##; ##UREF##1##Krasnoselsky et al, 2004##; ##REF##15899795##Dai et al, 2005##) suggesting cell proliferation as a driving force for cancer progression. Breast cancer is a typical example where the hyper-proliferative nature of more aggressive subtypes has been confirmed by array-based proliferation assays (##REF##11553815##Sorlie et al, 2001##; ##REF##15899795##Dai et al, 2005##). These studies have also proved the superiority of cell proliferation analysis by microarrays over conventional methods regarding its objectivity and because of the assessment of multiple genes (##REF##15591335##Paik et al, 2004##; ##REF##15931389##Glinsky et al, 2005##; ##REF##16491069##Whitfield et al, 2006##).</p>", "<p>In colorectal cancer (CRC), the impact of tumour proliferation rate on malignancy is unclear because the conventional proliferation markers have produced conflicting results (<xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Table 1</xref>). To date, no proliferation signature has been established to address such an association.</p>", "<p>To investigate the association between proliferation and aggressiveness of colon cancer, we designed a colon-specific gene proliferation signature (GPS) that was derived from common expression patterns of an <italic>in vitro</italic> CRC model, and of human colon crypt compartments as published by ##REF##17881565##Kosinski et al (2007)##. The GPS was applied to evaluate two independant cohorts of colonic tumours and revealed an inverse relationship between tumour cell proliferation and unfavourable clinico-pathological parameters.</p>" ]

[ "<p><italic>Grant Support:</italic> Health Research Council of New Zealand, the Lottery Grants Board of New Zealand (Y-HL, AER) and the Kommission für Klinische Forschung des Klinikums rechts der Isar (JF, RR, JRS, BH, JM).</p>", "<title>Supplementary Material</title>" ]

[ "<fig id=\"fig1\"><label>Figure 1</label><caption><p>Overview methodology. A gene proliferation signature (GPS) was derived by combining the gene expression data from the analysis of a cell line model and colon crypts. (<bold>A</bold>) Ten colorectal cancer cell lines were cultured and harvested at semi-confluence and full confluence. Each cell culture was performed in duplicate. (<bold>B</bold>) Using a reference design, a mixture of sample and reference dye-labelled cDNAs was hybridised to a 30 K Oligo array. Dye orientation was reversed for biological replicates. (<bold>C</bold>) Statistical analysis of microarrays (SAM) was performed to identify differentially expressed (DE) genes between two stages of growth in cultures. Two gene sets were generated through the analysis of samples with identical dye labelling. (<bold>D</bold>) Only 881 genes that were presented in both SAM-generated gene sets and DE gene sets were selected. (<bold>E</bold>) Human colon crypt profiling resulted in identification of 299 DE genes with overexpression in the proliferation zone compared with the differentiation zone. (<bold>F</bold>) The GPS was generated by taking the overlapping genes between <bold>D</bold> and <bold>E</bold> gene sets. (<bold>G</bold>) Two cohorts of colon cancer patients were stratified into low and high groups according to the GPS expression using K-means clustering method. (<bold>H</bold>) Disease-free survival difference was calculated between the two defined groups.</p></caption></fig>", "<fig id=\"fig2\"><label>Figure 2</label><caption><p>Disease-free survival analysis of colon cancer patients stratified into high and low groups according to the GPS expression or 15 cell cycle-regulated genes included in the GPS. In both cohorts, the low GPS groups had significantly shorter DFS compared with the high GPS groups (<bold>A</bold> and <bold>C</bold>). This difference was more significant when only cell cycle-regulated genes were used to stratify patients (<bold>D</bold> and <bold>F</bold>). The same association was found when the analysis was limited to those cohort A patients who received no adjuvant therapy (<bold>B</bold> and <bold>E</bold>).</p></caption></fig>", "<fig id=\"fig3\"><label>Figure 3</label><caption><p>Stratification effect of the GPS on two cohorts of breast cancer patients. The heat maps represent the normalised gene expression values of the GPS genes across samples. Each row represents one gene and each column represents one sample. Colour bars on the top of heat maps represent proliferation groups (high proliferation is indicated in red, and low proliferation is indicated in green as defined by K-means clustering) and recurrence status (black, recurrence; grey, nonrecurrence). There is a close correlation between high GPS expression and recurrence in both cohorts. Disease-free survival is significantly shorter in the high GPS groups compared with the low GPS groups.</p></caption></fig>" ]

[ "<table-wrap id=\"tbl1\"><label>Table 1</label><caption><title>Clinico-pathologic characteristics of two cohorts of colon cancer patients and associations with the GPS</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th colspan=\"2\" align=\"center\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>Numbers</bold>\n<hr/></th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>Cohort A</bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>Cohort B</bold>\n</th></tr><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Parameters</bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>Cohort A</bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>Cohort B</bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>(<italic>P</italic>-value)<xref ref-type=\"fn\" rid=\"t1-fn2\">a</xref></bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>(<italic>P</italic>-value)<xref ref-type=\"fn\" rid=\"t1-fn2\">a</xref></bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Age</italic>\n<xref ref-type=\"fn\" rid=\"t1-fn3\">b</xref>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> &lt;Mean</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">49</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">20</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.33</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.79</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> &gt;Mean</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">59</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">17</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Sex</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Male</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">62</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">21</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.56</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.74</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Female</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">46</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">16</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Differentiation</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Well/moderate</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">83</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">22</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.17</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.20</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Poor</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">25</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">15</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Disease stage</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> I</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">12</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>0.015</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">NA</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> II</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">64</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">37</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> III</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">29</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> IV</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Vascular invasion</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Yes</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">9</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.065</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">NA</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> No</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">99</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">36</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Lymphatic invasion</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Yes</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">23</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>0.018</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> No</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">85</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">34</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Lymphocyte infiltration</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Nil/mild</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">53</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">9</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.67</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.48</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Moderate</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">42</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">17</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Prominent</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">13</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">11</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Margin</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Infiltrative</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">56</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">NA</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.84</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">NA</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Expansive</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">52</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Chemotherapy</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Yes</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">30</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.08</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">NA</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> No</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">78</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">37</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Recurrence at 5 years</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Yes</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">24</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">16</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>0.021</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>0.005</bold>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> No</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">84</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">21</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Total</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">108</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">37</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl2\"><label>Table 2</label><caption><title>Cox regression analysis of determinants of DFS in cohort A cancer patients</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"char\" char=\".\"/><col align=\"center\"/><col align=\"char\" char=\".\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>Univariate analysis</bold>\n</th><th colspan=\"2\" align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Multivariate analysis<xref ref-type=\"fn\" rid=\"t2-fn2\">a</xref></bold>\n<hr/></th></tr><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Parameters</bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold><italic>P</italic>-value<xref ref-type=\"fn\" rid=\"t2-fn3\">b</xref></bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>HR (CI)<xref ref-type=\"fn\" rid=\"t2-fn4\">c</xref></bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold><italic>P</italic>-value</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Disease stage \n(I+II vs III+IV)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"center\" valign=\"top\" charoff=\"50\">4.5 (1.8–10.8)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.001</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Lymphatic invasion \n(− vs +)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"center\" valign=\"top\" charoff=\"50\">—</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Vascular invasion \n(− vs +)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.012</td><td align=\"center\" valign=\"top\" charoff=\"50\">—</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Margin \n(expansive vs infiltrative)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.015</td><td align=\"center\" valign=\"top\" charoff=\"50\">—</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">GPS expression \n(high vs low)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.022</td><td align=\"center\" valign=\"top\" charoff=\"50\">—</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">—</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"xob1\"><label>Supplementary Figure 1</label></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"xob2\"><label>Supplementary Figure 2</label></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"xob3\"><label>Supplementary Table 1</label></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"xob4\"><label>Supplementary Table 2</label></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"xob5\"><label>Supplementary Table 3</label></supplementary-material>" ]

[ "<fn-group><fn><p><xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Information</xref> accompanies the paper on British Journal of Cancer website (<ext-link xmlns:xlink=\"http://www.w3.org/1999/xlink\" ext-link-type=\"uri\" xlink:href=\"http://www.nature.com/bjc\">http://www.nature.com/bjc</ext-link>)</p></fn><fn><p>\n<bold>Financial Disclosures</bold>\n</p><p>Yu-Hsin Lin, Michael A Black are consultants and Anthony E Reeve is a director of Pacific Edge Biotechnology Ltd.</p></fn></fn-group>", "<table-wrap-foot><fn id=\"t1-fn1\"><p>Abbreviations: GPS=gene proliferation signature; NA=not applicable.</p></fn><fn id=\"t1-fn2\"><label>a</label><p>A Fisher's Exact test or Kruskal–Wallis test were used for testing association between clinico-pathologic parameters and the dichotomous GPS variable.</p></fn><fn id=\"t1-fn3\"><label>b</label><p>Average age 68 and 63 years for cohort A and B patients, respectively. Bold numbers represent significant <italic>P</italic>-values.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"t2-fn1\"><p>Abbreviation: DFS=disease-free survival</p></fn><fn id=\"t2-fn2\"><label>a</label><p>Final results of Cox regression analysis using a forward stepwise method (enter limit=0.05, remove limit=0.10).</p></fn><fn id=\"t2-fn3\"><label>b</label><p>Log-rank test <italic>P</italic>-value.</p></fn><fn id=\"t2-fn4\"><label>c</label><p>Hazard ratio (HR) determined by Cox regression model; confidence interval (CI)=95%.</p></fn></table-wrap-foot>" ]

[ "<graphic xlink:href=\"6604560f1\"/>", "<graphic xlink:href=\"6604560f2\"/>", "<graphic xlink:href=\"6604560f3\"/>" ]

[ "<media xlink:href=\"6604560x1.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"6604560x2.tif\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"6604560x3.doc\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"6604560x4.xls\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"6604560x5.doc\"><caption><p>Click here for additional data file.</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[]

[ "<title>Materials and methods</title>", "<title>Cell culture and transfection</title>", "<p>AGS, MKN45 and KATOIII cell lines were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). AGS cells were grown in Ham's F/12 supplemented with 10% fetal bovine serum, MKN45 cells were grown in RPMI-1640 supplemented with 10% fetal bovine serum and KATOIII cells were grown in RPMI-1640 supplemented with 20% fetal bovine serum. In all the media 100 g ml⁻¹ penicillin and streptomycin were added. Cell lines were maintained in a 37°C incubator with 5% CO₂ and in a humidified atmosphere. For the overexpression of MUC4 an engineered MUC4 construct (MUC4minigene) was made, which contains 10% of the tandem repeat sequence. The MUC4minigene construct was stably transfected in AGS cell line for constitutive expression of MUC4. For control, cells were transfected with empty vector (pSecTagC). Stable clones were then selected in a medium containing Zeocin (400 mg ml⁻¹). The Zeocin-resistant colonies were isolated by the ring cloning method and maintained in the medium supplemented with Zeocin. The medium was replaced with a complete medium without antibiotic supplement for at least 5 days before any analysis.</p>", "<title>Immunoblot assay</title>", "<p>The AGS, MKN45 and KATOIII cell lines were processed for extraction of whole cell protein and using standard protocol western blotting was carried out. The cells were washed two times with phosphate-buffered saline (PBS) and scraped in radioimmunoprecipitation assay buffer (50 m<sc>M</sc> Tris, 5 m<sc>M</sc> ethylenediaminetetraacetic acid (EDTA), 150m<sc>M</sc> NaCl, 0.25% sodium deoxycholate, 1% NP40 pH 7.5), supplemented with protease inhibitor mixture (Roche Diagnostics, Mannheim, Germany) and kept at 4°C for at least 30 min. Cell lysates were passed through the 28G tuberculin needle or alternatively subjected to one freeze thaw cycle to facilitate the disruption of the cell membranes. Cell lysates were centrifuged at 14 000 r.p.m. for 20 min at 4°C and supernatants were collected. Using a BIO-RAD DC protein estimation kit the samples were quantified. Owing to the large size of MUC4, the protein samples were resolved by electrophoresis on a 2% SDS-agarose gel under reducing condition. Resolved proteins were transferred onto the polyvinylidene difluorided membrane. For MUC4 detection anti-MUC4 mouse monoclonal antibody (8G7) was used. For the detection of total ErbB2/HER2 and phosphorylated ErbB2/HER2, anti-HER2 rabbit polyclonal (Santa Cruz Biotechnology, Santa Cruz, CA, USA) and pY1248-HER2 (Upstate Biotech, Lake Placid, NY, USA) antibodies were used respectively. Secondary antibodies consisted of horseradish peroxidase-conjugated anti-mouse/anti-rabbit, which were used for the immunodetection. The blots were processed with ECL chemiluminescence kit (Amersham Biosciences, Piscataway, NJ, USA), and the signal was detected by exposing the processed blots to X-ray films (Biomax Films, Kodak, NY, USA).</p>", "<title>Confocal immunofluorescence microscopy</title>", "<p>Cells were grown at low density on sterile coverslips for 20 h. After washing with 0.1 <sc>M</sc> HEPES containing Hanks buffer, the cells were fixed in ice-cold methanol at −20°C for 2 min. Nonspecific blocking was done by using 10% goat serum containing 0.05% Tween-20 for 30 min, followed by incubation with anti-MUC4 monoclonal antibody (8G7) in PBS for 90 min at room temperature. Cells were washed 3–4 times with PBS containing 0.05% Tween-20 (PBS-T) and then incubated with FITC-conjugated goat anti-mouse secondary antibodies for 60 min. The cells were counterstained with propidium iodide. Finally slides were washed two times with PBS and mounted on glass slides in antifade Vectashield mounting medium (Vector Laboratories, Burlingame, CA, USA). The slides were observed under a ZEISS confocal laser-scanning microscope, and photographs were captured digitally using 510 software.</p>", "<title>Tissue microarrays and immunohistochemistry</title>", "<p>To have more number of sample sizes, tissue microarray slides were ordered from two different companies. The AccuMax TM (cat. no. A209) has total of 50 different cancer cases and four non-neoplastic tissues (one corresponding, three non-corresponding). The TMA slide had two spots for each cancer tissue and one spot for each non-neoplastic tissue. Tissue microarray slide from US Biomax Inc. (ST801) had a total of 40 individual cases and each normal adjacent tissue is placed next to its matched cancer tissue. As per recommendations of the manufacturer, before conducting immunohistochemistry, the slides were baked at 60°C for 2 h. Then slides were deparaffinized by using EzDewax (Bio Genex, CA, USA) for 30 min. Sections were hydrated through graded alcohol and endogenous peroxidase activity was quenched by incubating the sections in 0.3% H₂O₂ in methanol for 30 min. After washing the slides in PBS (5 min × 2), antigen retrieval was done by heating the slides in citrate buffer (0.01 <sc>M</sc>, pH 6.0) at 80°C for 20 min. After heating the samples were allowed to cool for 15–20 min in room temperature. This was followed by washing with PBS (5 min × 2). Nonspecific binding was blocked by incubating the sections with 2.5% horse serum for 30 min (Impress reagent Kit,Vector).This was followed by washing with PBS (5 min × 2). Sections were then incubated with anti-MUC4 monoclonal antibody (1 : 2500) at 4°C overnight. Then slides were washed and incubated with secondary antibody (peroxidase labeled Universal anti-mouse/anti-rabbit IgG (Vector, CA)) for 30 min. Then sections were washed with PBS followed by treatment with DAB reagents (0.2 mg ml⁻¹) and incubated for 10 min. After washing with distilled water, counter staining was done by using haematoxylin (Vector, CA, USA). After washing in tap water, sections were dehydrated in graded alcohol and after air drying the slides were mounted in permount permanent mounting media (Fisher Scientific, Fair Lawn, NJ, USA). All slides were observed under Nikon E400 light microscope and representative photographs were taken.</p>", "<title>Motility assay</title>", "<p>For motility assay, 1 × 10⁶ cells were plated on the top chamber of a non-coated polyethylene teraphthalate membrane (six-well inset, pore size 8 <italic>μ</italic>m; Becton Dickinson, Franklin Lakes, NJ, USA). The bottom chamber contained 1.0 ml DMEM supplemented with 10% fetal bovine serum. The cells were incubated for 24 h, and the cells that did not migrate through the pores in the membrane were removed by scraping the membrane with a cotton swab. Cells that transversed the membrane were stained with Diff-Quick cell staining kit (Dade Behring Inc., Newark, DE, USA). Cells in 10 random fields of view at × 100 magnifications were counted and expressed as the average number of cells/field of view. Three independent experiments were done in each case. These data were represented as the average of the three independent experiments with the s.d. of the average indicated.</p>", "<title>Aggregation assay</title>", "<p>Cells were tested for their ability to aggregate in hanging drop suspension cultures. Cells were trypsinized in the presence of EDTA, washed two times in PBS, and resuspended at 2.5 × 10⁵ cells ml⁻¹ in the appropriate medium containing 10% fetal bovine serum. Drops (20 <italic>μ</italic>l each) of medium, containing 5000 cells/drop, were pipetted onto the inner surface of the lid of a Petri dish. The lid was then placed on the Petri dish so that the drops were hanging from the lid with the cells suspended within them. To eliminate evaporation, 8 ml of serum-free culture medium were placed in the bottom of the Petri dish. After overnight incubation at 37°C, the lid of the Petri dish was inverted and photographed using a Nikon TS100 inverted tissue culture microscope at × 40 magnifications.</p>", "<title><italic>In vivo</italic> tumorigenicity assay</title>", "<p>To test the tumorigenic capacity, the MUC4-transfected AGS cells along with the control cells were harvested from subconfluent cultures by a brief exposure to 0.25% trypsin and 0.02% EDTA. After neutralising the effect of trypsin with 10% fetal bovine serum, the cells were washed once in PBS. Cell viability and number were determined by trypan blue staining using a hemocytometer. Cells were resuspended in a normal saline solution at a concentration of 25 × 10⁶ cells ml⁻¹. Single-cell suspensions of &gt;90% viability was used for the injections. Immunodeficient mice were purchased from the Animal Production Area of the National Cancer Institute-Frederick Cancer Research and Development Center (Frederick, MD, USA). The mice were housed in specific pathogen-free conditions and fed sterile water and food <italic>ad libitum</italic>. The mice were treated in accordance with the Institutional Animal Care and Use Committee guidelines. 5 × 10⁶ viable MUC4-transfected AGS-MUC4 cells, resuspended in a normal saline solution, were injected subcutaneously in six immunodeficient mice. Empty vector transfected cells (AGS-pSecTagC) were used as a control (<italic>n</italic>=6). The animals were monitored two times weekly for tumour formation up to 4 months after inoculation. A palpable mass was observed at the inoculation site at around 80 days of post-injection, followed by rapid growth of the tumour. All mice were killed on day 120 after implantation, and the incidence of tumour was determined.</p>", "<title>Statistical analysis</title>", "<p>Subjects with normal or adenocarcinoma or signet ring cell carcinoma were included in the analysis. The distribution of type, grade and stage was compared between positive and negative intensity groups using a <italic>χ</italic>² test or Fisher's Exact test where appropriate. Tumour incidence was compared between groups using Fisher's Exact test.</p>" ]

[ "<title>Results</title>", "<title>Immunohistochemical analysis of MUC4 expression in gastric cancer tissues</title>", "<p>As a first step toward studying the role of MUC4 in gastric cancer, we did immunohistochemical analysis of MUC4 expression on TMA samples of gastric adenocarcinoma and normal adjacent area. MUC4 staining showed a diffuse staining pattern (membrane/cytoplasmic) in most of the tissue sections. Representative pictures of the stained gastric tumour tissue sections are presented in ##FIG##0##Figure 1##. Statistical analysis was conducted to determine the association of MUC4 expression pattern with cancer type, differentiation and stage of the tumour (##TAB##0##Tables 1## and ##TAB##1##2##). Out of a total of 128 tissue spots, we found that the proportion of MUC4-positive staining is lower for normal adjacent spots (<italic>n</italic>=45, 9%) compared with patients with adenocarcinoma (<italic>n</italic>=58, 43%) and signet ring cell carcinoma (<italic>n</italic>=25, 32%) (<italic>P</italic>&lt;0.001). However, no significant difference was found in MUC4 expression between adenocarcinomas and signet ring cell carcinomas (<italic>P</italic>=0.34). Furthermore, no significant correlation was found between tumour differentiation and MUC4 expression for all cancer patients combined (<italic>P</italic>=0.34) (##TAB##0##Table 1##). MUC4 expression pattern did not significantly correlate with the stage of tumour, both individually (<italic>P</italic>=0.16) or in groups, that is, I/II (early stage) and III/IV (late stage) (<italic>P</italic>=0.21) (##TAB##1##Table 2##).</p>", "<title>Expression of MUC4 in different gastric cancer cell lines</title>", "<p>Using immunoblot assay, MUC4 expression was checked in three different types of gastric adenocarcinoma cell lines (AGS, MKN45 and KATOIII). MUC4 expression was detected in KATOIII (SRCC) but was undetectable in AGS and MKN45 cell lines (##FIG##1##Figure 2##). To further analyse the role of MUC4 in the aggressiveness of non-signet ring cell type poorly differentiated gastric adenocarcinomas, MUC4 was ectopically expressed in AGS gastric cancer cell line (using an engineered MUC4 cDNA construct, MUC4minigene (##REF##17595659##Moniaux et al, 2007##)). The MUC4minigene's deduced protein (320 kDa) is analogous to that of wild-type MUC4 protein (930 kDa) (##REF##17595659##Moniaux et al, 2007##). Western blot analysis (##FIG##2##Figure 3##) and confocal study showed the overexpression of MUC4 in selected clones transfected with MUC4 construct in comparison to empty vector transfected clones (##FIG##2##Figure 3##). Furthermore, confocal analysis showed the membrane localisation of MUC4 in AGS-MUC4 clones.</p>", "<title>MUC4 overexpression increases cell motility in AGS gastric cancer cells</title>", "<p>The aggressiveness of a malignant cell depends on its potential to invade the ECM and its ability to metastasize to distant sites. Different studies have shown that the invasive and metastasis potential of cancer cells are strongly related to a variety of phenotypic characteristics. Among these characteristics, motility of cells highly influences the metastatic property of cells (##REF##16098726##Yamaguchi et al, 2005##). As the MUC4 overexpression is associated with an increased motility of pancreatic cancer cells, we examined whether the MUC4 overexpression in gastric cancer is associated with an increase in cell motility or not. Cell motility was determined on top of the uncoated porous membrane. The number of MUC4 overexpressing AGS gastric cancer cells (AGS-MUC4) migrated to the lower surface of the porous membrane was significantly high (<italic>P</italic>&lt;0.005) than that of the vector control (AGS-vector) cells (##FIG##3##Figure 4A##).</p>", "<title>Overexpression of MUC4 decreases aggregation property of AGS gastric cancer cells</title>", "<p>Like cell motility, aggregation property of cells is also a critical factor, which affects the metastasis property of tumour cells. The aggregation property is usually deregulated in the tumour cells because of alteration in the expression of different cell surface molecules (##REF##10887495##Sommers, 1996##; ##REF##9407114##Komatsu et al, 1997##; ##REF##12640674##Truant et al, 2003##). Therefore, to test the effect of MUC4 overexpression on adhesiveness of AGS cells, we used the aggregation assay described in Materials and methods. MUC4 overexpression showed a decrease in aggregation of AGS-MUC4-transfected cells in comparison to vector-transfected control cells (##FIG##3##Figure 4B##).</p>", "<title>Overexpression of MUC4 enhances tumorigenicity of AGS gastric cancer cells in nude mice</title>", "<p>To investigate the role of MUC4 on tumorigenic property of gastric cancer cells, AGS-MUC4 and AGS-vector cells, were injected subcutaneously into immune deficient nude mice. 5 × 10⁶ cells were injected and a palpable mass was first noticed at the eightieth day of post-inoculation and continued to grow up to 120 days. Among the six mice inoculated with AGS-MUC4 cells, incidence of tumour was observed in five mice (83%), whereas only one mouse had a tumour among the six mice inoculated with AGS-vector cells (17%). Further, statistical analysis showed that this difference in the incidence of the tumour between the two groups is marginally significant (<italic>P</italic>=0.08) (##TAB##2##Table 3##).</p>", "<title>Overexpression of MUC4 activates ErbB2 oncoprotein in AGS gastric cancer cells</title>", "<p>shRNA-mediated knock-down of MUC4 in pancreatic cancer cells showed a decrease in the total levels and phosphorylated form of ErbB2 protein (at Tyr¹²⁴⁸) and was shown that overexpression of MUC4 plays a crucial role in stabilising ErbB2 (##REF##17406026##Chaturvedi et al, 2007##, ##REF##18381409##2008##). Other studies have shown that in SRCC type gastric cancer cells, MUC4 is required for the activation of ErbB2 (##REF##17292332##Yokoyama et al, 2007##). On the basis of these earlier observations, we wanted to determine whether MUC4 exerts its function through regulating ErbB2 expression and activation in AGS-MUC4 gastric cancer cells. The expression of total and active form of ErbB2 was measured by western blot analysis. Cell lysate of AGS-MUC4 showed increased level of total and phosphorylated (at Tyr¹²⁴⁸) ErbB2 protein compared with the control AGS-vector cell lysate (##FIG##4##Figure 5##).</p>" ]

[ "<title>Discussion</title>", "<p>MUC4 is one of the most widely studied membrane-bound mucins having a significant function in the pathogenesis of several cancers (##REF##14752841##Shibahara et al, 2004##; ##REF##14744777##Singh et al, 2004##). The aberrant expression of MUC4 has been reported in different types of carcinomas (##REF##11751498##Andrianifahanana et al, 2001##; ##REF##14657954##Llinares et al, 2004##; ##UREF##1##Weed et al, 2004##). It has been shown that MUC4 expression correlates with cancer progression (##REF##12090430##Swartz et al, 2002##; ##REF##12657964##Park et al, 2003##). In this study, we showed that MUC4 is overexpressed in gastric cancer tissues as compared with normal adjacent tissues. Overexpression of MUC4 was associated with an aggressive phenotype of gastric cancer cells. MUC4 overexpression also increased the activation of ErbB2 oncoprotein. Hence, our study provides for the first time, the importance of MUC4 in gastric cancer and explains the possible mechanism through which MUC4 can promote aggressive property of poorly differentiated gastric non-SRCC cells.</p>", "<p>In this study, we examined and compared the expression of MUC4 in gastric cancer tissues. MUC4 showed a significant overexpression in gastric cancer tissues compared with the normal adjacent tissues. Similar findings showing the overexpression of MUC4 in gastric adenocarcinoma has been reported earlier (##REF##10940270##Lopez-Ferrer et al, 2000##). MUC4 is known to be expressed in the embryonic gastric tissues around 8 weeks of gestation (##REF##11101634##Buisine et al, 2000##). It has also been shown that many embryogenesis phenomena like cell proliferation, lineage allocation, cell migration and differentiation of cells are also seen during cancer progression. Therefore, overexpression of MUC4 in adult gastric carcinoma supports the concept of ‘fetal antigen’ expression during malignant condition and indicates its possible role in gastric cancer progression. Further, our statistical analysis showed that there is no significant association between MUC4 expression with type, grade of differentiation and stage of gastric cancer. This suggests that MUC4 itself may not be a potential marker for early diagnosis of gastric cancer. Previous studies performed regarding the role of mucins as diagnostic and prognosis markers in gastric carcinoma tissue showed contradictory results (##REF##8137316##Correa and Shiao, 1994##; ##REF##9247464##Byrd et al, 1997##). Therefore, for higher accuracy, many studies have been done to evaluate the combined expression pattern of mucins with other molecules such as E-cadherin with MUC1 expression (##REF##12520583##Tanaka et al, 2003##). Similarly, comparing the expression pattern of MUC4 with other proteins like E-cadherin or other mucins will strengthen the study and may potentiate the possible use of MUC4 as a diagnostic and prognostic marker for gastric adenocarcinomas.</p>", "<p>Different studies have shown the functional role of MUC4 in tumorigenicity and metastasis property of cancer cells (##REF##14744777##Singh et al, 2004##; ##REF##17406026##Chaturvedi et al, 2007##). Recently, we have also shown that in pancreatic cancer cells MUC4 interacts with ErbB2 and stabilizes its localization on the cell membrane (##REF##18381409##Chaturvedi et al, 2008##). In another very recent study, it has been shown that MUC4 interacts with ErbB2 in human gallbladder carcinoma and helps in the activation of erbB2 (##REF##18397823##Miyahara et al, 2008##).Till date, information regarding the role of MUC4 in gastric cancer is very less, recently, it has been shown that specifically in poorly differentiated type gastric signet ring cell carcinoma cells, MUC4 is required for the activation of ErbB2. Because MUC4 is expressed even in poorly differentiated gastric non-SRCC cells, we therefore reasoned that MUC4 might have a possible role in those kinds of cells. Here, we report that out of the three poorly differentiated cells (KATOIII, MKN45 and AGS), only KATOIII, which is a signet ring cell carcinoma cell line, expresses MUC4.</p>", "<p>Further, to check the actual role of MUC4 in non-SRCC type poorly differentiated gastric cancer cells, MUC4 was ectopically overexpressed in a gastric adenocarcinoma cell line (AGS), which has an undetectable expression level of MUC4. In different <italic>in vitro</italic> studies, we found that, MUC4 causes an increase in the motility of AGS-MUC4 cells, and a decrease in the adhesive property of AGS-MUC4 cells. This finding supports other studies where MUC4 has a similar function in other cancers, such as pancreatic cancer (##REF##14744777##Singh et al, 2004##; ##REF##17406026##Chaturvedi et al, 2007##). The increase in cell motility (<italic>P</italic>&lt;0.05), which has a major role during the dynamic process of tumour invasion and metastasis, may be a direct result of MUC4-mediated changes in the actin organisation, or indirectly through an ErbB2-mediated pathway (##REF##17234748##Singh et al, 2007##). During cancer cell metastasis, cells remain loosely attached to the ECM or to the other cells. This property is essential to make the cells more migratory and to increase the invasiveness of cancer cells. In our aggregation assay, we found that MUC4 overexpression in AGS cell line, decreases its aggregation property, or in other words enhances its metastatic property. This decreased adhesive property among cells may be because of charge–charge repulsion on account of the presence of negatively charged O-glycosidic chains present in the central repetitive domain of MUC4 or owing to disruption of integrin-mediated cell adhesion. Furthermore, the high incidence of tumours in animals injected with AGS-MUC4 cells than animals inoculated with AGS-vector cells, indicate the role of MUC4 in tumorigenicity of gastric cancer cells.</p>", "<p>Decrease in cell death and increase in cell proliferation are two major regulatory elements in enhancing the tumorigenicity of cancer cells. Overexpression of ErbB2 oncoprotein has been shown to correlate with tumour aggressiveness in various tumours (##REF##15864276##Hynes and Lane, 2005##). In gastric cancer, a correlation between ErbB2 gene amplification and prognosis of patients has been reported (##REF##10223227##Nakajima et al, 1999##; ##REF##11114748##Lin et al, 2000##). In this study, similar to our previous report, we also observed an increase in total ErbB2 and phosphorylated ErbB2 expression in MUC4 expressing AGS cells. AGS, is a well studied poorly differentiated cell line (##REF##6831414##Barranco et al, 1983##). The tumorigenicity of this cell line in animals has been already reported (##REF##6831414##Barranco et al, 1983##). Here, the increase in tumorigenicity of AGS cells mediated by MUC4 may be on account of the interaction of MUC4 with ErbB2 and further stabilisation and activation of ErbB2-mediated oncogenic signaling.</p>", "<p>In conclusion, our <italic>in vitro</italic> and <italic>in vivo</italic> studies showed a significant role of MUC4 in promoting the aggressiveness and tumorigenicity of poorly differentiated gastric non-SRCC cells. In addition, we showed a possible mechanism through which MUC4 can increase the tumorigenicity property of poorly differentiated gastric non-SRCC cells. Validation for MUC4 expression as a prognostic marker requires further studies. A study on the combined expression pattern of MUC4 and other molecules like E-cadherin or other mucins may provide more accuracy and specificity to use MUC4 as a diagnostic or prognostic marker for gastric cancer. Our study also indicates that MUC4 can be targeted for treatment of gastric cancer.</p>" ]

[]

[ "<p>MUC4 is a large, heavily glycosylated transmembrane mucin, that is implicated in the pathogenesis of various types of cancers. To date, no extensive study has been done to check the expression and functional significance of MUC4 in different types of gastric adenocarcinomas. Here, we report the expression profile of MUC4 in gastric adenocarcinomas and its function in poorly differentiated gastric non-signet ring cell carcinoma (non-SRCC) type cells. Immunohistochemical analysis using tissue microarray (TMA) showed a significant difference in MUC4 expression between normal adjacent (<italic>n</italic>=45) and gastric adenocarcinoma (<italic>n</italic>=83; <italic>P</italic>&lt;0.001). MUC4 expression was not associated with tumour type, stage or with the degree of differentiation. To gain further insight into the significance of MUC4 expression in gastric non-SRCC cells, MUC4 was ectopically expressed in AGS, a poorly differentiated gastric non-signet ring cell line. The MUC4 overexpressing cells (AGS-MUC4) showed a significant increase (<italic>P</italic>&lt;0.005) in cell motility and a decrease in cellular aggregation as compared with the vector-transfected cells. Furthermore, <italic>in vivo</italic> tumorigenicity analysis revealed that animals transplanted with the MUC4 overexpressing cells (AGS-MUC4) had a greater incidence of tumours (83%) in comparison to empty vector control (17%). In addition, the expression of MUC4 resulted in enhanced expression of total cellular ErbB2 and phosphorylated ErbB2. In conclusion, our results showed that MUC4 is overexpressed in gastric adenocarcinoma tissues, and that it has a role in promoting aggressive properties in poorly differentiated gastric non-SRCC cells through the activation of the ErbB2 oncoprotein.</p>" ]

[ "<p>Gastric cancer is the fourth most commonly found cancer worldwide and more than 90% of gastric cancers are adenocarcinomas (##REF##15052434##Correa et al, 2004##). According to recent statistical information, in the United States, 21 500 new gastric cancer cases are estimated for the year 2008 (##REF##18287387##Jemal et al, 2008##). Despite advances in diagnostic techniques such as imaging, esophagogastroduodenoscopy, magnetic resonance imaging and dual-phase spiral computer tomography, early diagnosis of gastric adenocarcinoma is still a diagnostic problem for clinicians. An early detection with accurate diagnosis and effective surgical or endoscopic treatment can result in a better prognosis.</p>", "<p>Gastric adenocarcinoma is classified into intestinal and diffuse type of adenocarcinomas (##REF##14320675##Lauren, 1965##). Morphologically, the intestinal-type gastric adenocarcinoma has well-defined glands with epithelial lining and diffuse-type gastric adenocarcinoma, which mainly consists of scattered individual cells or clusters of cells. It has been shown that the prevalence of poorly differentiated gastric carcinomas is higher than well differentiated gastric cancers (##REF##10440758##Nakamura et al, 1999##).</p>", "<p>Deregulation of mucins has been shown to be critical in a number of gastro-intestinal malignancies including gastric cancer. Normal gastric epithelial cells express a variety of mucins and have different functions, such as protection against mechanical and infectious insults, lubrication and acid resistance (##UREF##0##Tasman-Jones, 1985##; ##REF##14681689##Hollingsworth and Swanson, 2004##). Various reports have shown that altered mucin carbohydrate and peptide residues of mucins may be used as molecular markers of an increased risk of malignant transformation (##REF##2471698##Girling et al, 1989##; ##REF##2662714##Hakomori, 1989##; ##REF##2582438##Merlo et al, 1989##; ##REF##7678777##Ho et al, 1993##; ##REF##7750105##Springer et al, 1995##; ##REF##14657954##Llinares et al, 2004##). Recent studies have provided strong evidences, which potentiate the role of mucins in the pathogenesis of various malignancies (##REF##12955090##Li et al, 2003##; ##REF##12826677##Yin et al, 2003##; ##REF##14744777##Singh et al, 2004##). In gastric adenocarcinoma, mucin expression pattern is heterogeneous. Mucins in gastric carcinoma include normal mucins of stomach like MUC1, MUC5AC, MUC6 and <italic>de novo</italic> expression of the intestinal mucins MUC2 (##REF##8020658##Carrato et al, 1994##; ##REF##7780985##Ho et al, 1995##; ##REF##9308730##Sakamoto et al, 1997##; ##REF##9583726##Baldus et al, 1998##; ##REF##9699534##Reis et al, 1998##, ##REF##10681391##2000##; ##REF##9829723##Utsunomiya et al, 1998##).</p>", "<p>MUC4, is a membrane-bound mucin and has a significant role in different cancers including pancreatic and breast cancers (##REF##14744777##Singh et al, 2004##). MUC4 expression is also associated with the poor prognosis for pancreatic, lung and bile duct cancer patients. Overexpression of MUC4 in pancreatic cancer potentiates pancreatic tumour cell proliferation, survival and invasive properties and also interferes with its interaction to extracellular matrix proteins (##REF##17406026##Chaturvedi et al, 2007##). MUC4 also interacts with ErbB2, a growth factor receptor, stabilizes it at the cell surface and hence has an important function in modulating ErbB2-mediated oncogenic signaling in pancreatic cancer cells (##REF##18381409##Chaturvedi et al, 2008##). The importance of MUC4 for activation of ErbB2 in poorly differentiated gastric SRCC has been reported (##REF##17292332##Yokoyama et al, 2007##). However, to date there is a lack of knowledge about the functional significance of MUC4 in poorly differentiated gastric cancers other than the SRCC subtypes.</p>", "<p>Realizing the importance of MUC4 in different malignancies, we prompted to investigate the expression pattern of MUC4 in different gastric adenocarcinomas. In this study, we have shown that MUC4 is overexpressed in the gastric cancer and its expression pattern does not correlate with type, differentiation or stage of cancer. In consideration of MUC4 expression in poorly differentiated gastric non-SRCC cells and its role in the activation of ErbB2 in gastric SRCC cells, we did <italic>in vitro</italic> and <italic>in vivo</italic> studies to check the significance of MUC4 in non-SRCC cells. Here, we have shown that MUC4 overexpression in poorly differentiated AGS, gastric cancer cells, increases its aggressive cancer property in both <italic>in vitro</italic> and <italic>in vivo</italic> experiments. In addition, overexpression of MUC4 in AGS, gastric cancer cells, increases both total and phosphorylated form of ErbB2.</p>" ]

[ "<p>This work was supported by grants from the National Institutes of Health (CA78590, CA111294). We thank Ms Kristi L Berger for editing the paper. The invaluable technical support of Mr Erik Moore was greatly appreciated.</p>" ]

[ "<fig id=\"fig1\"><label>Figure 1</label><caption><p>Immunohistochemical analysis of gastric tissues for MUC4 expression by using gastric cancer tissue microarray (TMA) slides. Tissue sections were stained for MUC4 using anti-MUC4 monoclonal antibody followed by biotinylated secondary antibody incubation and streptavidin peroxidase 3,3′-diaminobenzidine-chromogen detection. All the sections were examined under microscope and the immunoreactivity was judged by dark brown staining. (<bold>A</bold>) Representative picture of stained gastric normal adjacent tissues showing no visible MUC4 staining. (<bold>B</bold>) Representative picture of gastric adenocarcinoma tissues showing diffused MUC4 staining. All sections were counter stained with haematoxylin. In all the top panels, original magnification is × 10 and in bottom panels, original magnification is × 40.</p></caption></fig>", "<fig id=\"fig2\"><label>Figure 2</label><caption><p>Western blot analysis of MUC4 expression in gastric cancer cell lines. Total protein lysates from AGS, KATOIII and MKN45 gastric cancer cells were prepared. Protein lysate from FG (pancreatic cancer cell line) cells was taken as a positive control. Protein lysates were electrophoretically resolved on 2% agarose gel. Resolved proteins were transferred onto PVDF membrane and probed with MUC4 MAb (8G7) and detected using Amersham HRP-conjugated secondary antibody and ECL kit. Immunoblot of <italic>β</italic>-actin, obtained from 10% SDS–PAGE/Western, was used as an internal control to correct for the loading variation.</p></caption></fig>", "<fig id=\"fig3\"><label>Figure 3</label><caption><p>Expression of MUC4 in AGS and its derived sublines: AGS-vector (empty vector transfected) and AGS-MUC4 (MUC4-transfected) cells. (<bold>A</bold>) Western blot analysis: total protein lysates were prepared from the subconfluent cells. A total of 20 <italic>μ</italic>g protein from cell extracts was electrophoretically resolved on 2% Agarose gel. Resolved proteins were transferred onto PVDF membrane and probed with MUC4 MAb (8G7). Protein from FG (pancreatic cancer cell line) cells was taken as a positive control. (<bold>B</bold>) Expression analysis of MUC4 using confocal microscopy: Cells were grown at a low density on sterilised cover slips; after methanol fixation, slides were incubated with MUC4 MAb (8G7), followed by FITC-conjugated secondary antibody, and were observed under a ZEISS confocal laser-scanning microscope (magnification, × 630).</p></caption></fig>", "<fig id=\"fig4\"><label>Figure 4</label><caption><p>Phenotypic changes of AGS-MUC4 cells compared with AGS-vector cells. (<bold>A</bold>) Cell motility assay: MUC4 expression correlates with the cell motility. Cells (1 × 10⁶) were plated in the top chamber of noncoated polyethylene teraphthalate membranes and incubated for 20 h. Cells that transversed the membranes were stained with a Diff-Quick cell staining kit. The number of cells transversing the membrane was determined by averaging 10 random fields of view at × 100 and expressed as the average number of cells/field of view and is the average of two independent experiments. Mean±s.e.; <italic>n</italic>=20; ^*<italic>P</italic>&lt;0.005. Cell motility was significantly (<italic>P</italic>&lt;0.005) increased in MUC4-transfected AGS cells. (<bold>B</bold>) Aggregation assay: drops of medium (20 <italic>μ</italic>l each) containing 500 cells/drop were pipetted onto the inner surface of the lid of a Petri dish. After overnight incubation at 37°C, the lid of the Petri dish was inverted and photographed using a Nikon TS100 inverted tissue culture microscope at × 40 magnification. An increased cellular aggregation observed in AGS-MUC4 cells.</p></caption></fig>", "<fig id=\"fig5\"><label>Figure 5</label><caption><p>Effect of MUC4 expression on ErbB2 expression and phosphorylation. (<bold>A</bold>) Western blot analysis: a total of 20 <italic>μ</italic>g of protein from AGS-derived cell lines, were resolved by SDS-PAGE and transferred to PVDF membrane and probed with antibodies against, ErbB2, Phospho-tyr 1248 ErbB2, and <italic>β</italic>-actin. MUC4-transfected AGS cells showed an increased level of total and phosphorylated ErbB2 in comparison to vector-transfected cells.</p></caption></fig>" ]

[ "<table-wrap id=\"tbl1\"><label>Table 1</label><caption><title>Association of MUC4 expression pattern with types and grades of gastric cancer</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Tissues</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>MUC4-negative <italic>N</italic> (%)</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>MUC4-positive <italic>N</italic> (%)</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold><italic>P</italic>-value</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Types</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Normal</td><td align=\"center\" valign=\"top\" charoff=\"50\">41 (91%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">4 (9%)</td><td align=\"center\" valign=\"top\" charoff=\"50\"><italic>P</italic>&lt;0.001</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Adenocarcinoma</td><td align=\"center\" valign=\"top\" charoff=\"50\">33 (57%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">25 (43%)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> SRCC</td><td align=\"center\" valign=\"top\" charoff=\"50\">17 (68%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">8 (32%)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Differentiation</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Moderate</td><td align=\"center\" valign=\"top\" charoff=\"50\">9 (53%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">8 (47%)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Poor</td><td align=\"center\" valign=\"top\" charoff=\"50\">30 (58%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">22 (42%)</td><td align=\"center\" valign=\"top\" charoff=\"50\"><italic>P</italic>=0.34</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Not</td><td align=\"center\" valign=\"top\" charoff=\"50\">4 (100%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0 (0%)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Well</td><td align=\"center\" valign=\"top\" charoff=\"50\">7 (70%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">3 (30%)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl2\"><label>Table 2</label><caption><title>Association of MUC4 expression pattern with stages of gastric cancer</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"char\" char=\"(\"/><col align=\"char\" char=\"(\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Cancer stage</bold>\n</th><th align=\"center\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>MUC4-negative <italic>N</italic>=28</bold>\n</th><th align=\"center\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>MUC4-positive <italic>N</italic>=19</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold><italic>P</italic>-value</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">I</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">11 (58%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">8 (42%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0.16^***</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">II</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">6 (100%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0 (0%)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">III</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">9 (50%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">9 (50%)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">IV</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2 (50%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2 (50%)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl3\"><label>Table 3</label><caption><title>Incidence of tumours in animals injected with AGS-MUC4 and AGS-pSecTagC cells</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\"(\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Name of the group</bold>\n</th><th align=\"center\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>Number of animals injected</bold>\n</th><th align=\"center\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>Number of animals having tumour</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold><italic>P</italic>-value</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">AGS-MUC4</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5 (83%)</td><td align=\"center\" valign=\"top\" charoff=\"50\"><italic>P</italic>=0.08^***</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">AGS-Vector</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1 (17%)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><fn id=\"t2-fn1\"><p>^***Fisher's Exact test.</p></fn><fn id=\"t2-fn2\"><p>Information regarding the tumour stage was not available for all the spots.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"t3-fn1\"><p>^*** Fisher's Exact test.</p></fn></table-wrap-foot>" ]

[ "<graphic xlink:href=\"6604632f1\"/>", "<graphic xlink:href=\"6604632f2\"/>", "<graphic xlink:href=\"6604632f3\"/>", "<graphic xlink:href=\"6604632f4\"/>", "<graphic xlink:href=\"6604632f5\"/>" ]

[]

{ "acronym": [], "definition": [] }

[]

[ "<title>Materials and methods</title>", "<title>Patient selection</title>", "<p>Eligible patients were required to have histological verified non-resectable gastric or gastro-oesophageal junction adenocarcinoma and at least one measurable lesion according to RECIST criteria.</p>", "<p>Eligibility criterias also included WHO performance status 0–2; age between 18 and 75 years; estimated life expectancy of at least 3 months; adequate hepatic function (serum bilirubin &lt;1.5 × upper normal limit (UNL); transaminase &lt;3 × UNL, however in cases of livermetastases, there was no upper limit for transaminases); adequate renal function (calculated creatinine clearance &gt;30 ml min⁻¹ by the Cockroft and Gault formula); adequate haematological function (neutrophil count &gt;1.5 × 10⁹ l⁻¹; platelets &gt;100 × 10⁹ l⁻¹ and no prior chemotherapy other than adjuvant chemotherapy, completed at least six months before inclusion.</p>", "<p>Other inclusion criterias were ability to tolerate and comply with oral medication, no sign of peripheral neuropathy, no co-existent severe medical illness, no sign of brain metastases and no concomitant treatment with other anticancer therapy. Females were not included if they were pregnant or lactating.</p>", "<p>The study was approved by the local ethics committee and Danish health authority and all participants gave written informed consent before entering the study.</p>", "<title>Study design and treatment</title>", "<p>Patients were planned to receive a combination of epirubicin 50 mg m⁻² as a 20 min. i.v. infusion day 1, oxaliplatin 130 mg m⁻² as a 30 min. i.v. infusion day 1 and capecitabine 500 mg m⁻² × 2 daily continuously each 3 weeks (##UREF##2##Dupont et al, 2006##).</p>", "<p>Therapy was repeated every 3 weeks up to a maximum of eight cycles of treatment unless stopped before because of disease progression, unacceptable toxicity or patient refusal.</p>", "<p>Reduction of 25% in all drug doses was recommended in the event of occurrence of febrile neutropenia, grade 4 thrombocytopenia (platelet count &lt;25 × 10⁹ l⁻¹) or grade 4 neutropenia (absolute neutrophil count &lt;0.5 × 10⁹ l⁻¹), grade 3–4 mucositis, diarrhoea, or nausea/vomiting in spite of optimal antiemetic treatment. Toxicity was graded according to NCIC-CTC version 2.0. Peripheral sensitive neuropathy to oxaliplatin was graded according to the following oxaliplatin-specific scale. Grade 1, parestesias/hypoesthesias of short duration with complete recovery before the next cycle; grade 2, parestesias/hypoesthesias persisting between two cycles without functional impairment; and grade 3 permanent parestesias/hypoesthesias resulting in functional impairment (##REF##2348469##Caussanel et al, 1990##). In case of grade 2 neuropathy oxaliplatin was reduced 25%.</p>", "<title>Patient evaluation</title>", "<p>Baseline evaluation included physical examination, assessment of medical history, evaluation of performance status and blood counts. During treatment patients were evaluated before each cycle of therapy with the above parameters. All patients had an abdominal and thoracic CT-scan performed at baseline and every third cycle to assess tumour response. Tumour response was classified according to RECIST guidelines and confirmed as lasting longer than 4 weeks.</p>", "<p>After the end of the therapy patients were followed every third month until progression or death.</p>", "<title>Statisistical methods</title>", "<p>The primary end point of this phase II study was response rate to EXE and secondary end points were progression-free survival (PFS), and overall survival (OS). Progression-free survival was defined as the time from inclusion to progressive disease occurred (according to the RECIST criteria) or death of any cause. Overall survival was defined as the time from inclusion to death of any cause. Progression-free survival and OS were updated until 1 September 2007.</p>", "<p>Non-parametric statistics was applied. All median values are followed by range in brackets. After cessation of treatment patients without documented progression were followed every 3 months with clinical and radiological evaluation. Progression-free survival and OS were generated according to the Kaplan–Meier method. Data were recorded and analysed in a Medlog® database. All analyses were done on an intention-to-treat population.</p>" ]

[ "<title>Results</title>", "<title>Patient charasteristics</title>", "<p>Between June 2004 and August 2005, 54 patients with gastric and gastro-oesophageal adenocarcinoma were treated at three Danish oncology centres. Baseline patient characteristics are shown in ##TAB##0##Table 1##. The median age was 59 years (range, 31–74 years). Performance status was 0 in 25 patients, one in 26 patients and two in three patients. No patients had previously received adjuvant chemotherapy or radiotherapy. Forty-three patients had adenocarcinoma located in cardia and 11 patients had adenocarcinoma located in the corpus or antrum of the stomach. Nine patients (17%) had locally advanced disease and 45 patients (83%) had metastatic disease. Twenty-four patients had more than one organ involved (range, 2–4).</p>", "<title>Toxicity</title>", "<p>Fifty-three patients were evaluable for toxicity as one died of gastrointestinal bleeding 5 days after the first cycle of chemotherapy. Worst toxicity for all patients and all cycles are listed in ##TAB##1##Table 2##. Neutropenia was the principal haematologic toxicity, 13% of the patients experienced grade 2 neutropenia, 15% grade 3 and 2% grade 4 neutropenia, but no patient had febrile neutropenia. Infection grade 2–3 without neutropenia was seen in 12%. In general grade 4 toxicity was rare, one patient experienced grade 4 neutropenia and one had vomiting in grade 4.</p>", "<p>Non-haematologic toxicities were primary grade 2 toxicity, diarrhoea (13%), nausea (32%), vomiting (22%), hand food syndrome (PPE) (10%) and peripheral neuropathy (36%). No patients had peripheral neuropathy or PPE grade 3 or 4, and only 6, 4, and 6% had diarrhoea, vomiting or nausea grade 3, respectively. Except for the patient who died of gastrointestinal bleeding there was no treatment-related death in this study.</p>", "<title>Efficacy</title>", "<p>The median number of EXE was 6 (range, 1–8), 19 patients completed eight cycles of treatment. The reasons for discontinuation of EXE were progressive disease or deterioration of health (<italic>n</italic>=21), toxicity (<italic>n</italic>=7), patient refusal (<italic>n</italic>=5) and other (<italic>n</italic>=2). One patient had complete response and 23 obtained partial response, giving an overall response rate of 45% in the ITT population.</p>", "<p>Progression-free survival was 6.8 months (range, 5.2–7.9) and median survival was 10.1 months (range, 7.9–11.1) (##FIG##0##Figure 1##). Five patients are still alive 24–37 months after inclusion, one without active disease and four patients who have either had surgery or second- and third-line chemotherapy after progression.</p>" ]

[ "<title>Discussion</title>", "<p>Combination chemotherapy prolongs survival and improves quality of life in patients with advanced GC but still they have a poor prognosis and no standard regimen has yet been established.</p>", "<p>In many areas of Europe, ECF has been regarded as a reference regimen in patients with advanced gastric or gastrooesophageal cancer as randomized studies have confirmed a high response rate around 45% and long overall survival around 9–10 months (##REF##8996151##Webb et al, 1997##; ##REF##11956258##Ross et al, 2002##).</p>", "<p>The introduction of the new anticancer drugs oxaliplatin and capecitabine made it possible to overcome problems with hydration and central venous catheter. A large randomized phase lll study, REAL-2, used ECF as control and substituted cisplatin with oxaliplatin and CVI 5-FU with capecitabine in a 2 × 2 factorial design. An impressive 1002 patients were enroled in REAL-2. Response rates were highest in the EOX group (47.9%) but did not differ significantly between the four groups. There was a nonsignificant difference in PFS among the groups, but there was a trend towards prolonged OS in patients receiving oxaliplatin and capecitabine EOX (##REF##18172173##Cunningham et al, 2008##). In another phase III study, 220 patients were randomised to receive 24-h infusion 5-FU, leukovorin and cisplatin or 24-h infusion 5-FU, leukovorin and oxaliplatin (FLO). The authors also observed a trend towards a longer TTP in the FLO arm (median 5, 7 <italic>vs</italic> 3.8 months) (##UREF##0##Al-Batran et al, 2006##). Recently, a phase II study with 36 patients also evaluated a therapy with epirubicin, oxaliplatin and 5-FU. The response rate was 46%, PFS was 8.2 months and OS 12.2 months (##REF##17353926##Neri et al, 2007##) Toxicity profile was very similar to the data in this study.</p>", "<p>In recent years, ongoing or completed studies are evaluating new drugs or combinations of drugs in locally advanced or metastatic gastric cancer. Docetaxel, one of the ‘new’ drugs, was evaluated as first-line therapy in a phase III study (V325) with 457 patients receiving docetaxal, cisplatin and 5-FU (DCF) or cisplatin and 5-FU (CF). TTP (5.6 <italic>vs</italic> 3.7 months) and OS (9.2 <italic>vs</italic> 8.6 months) were significantly longer for patients receiving DCF (##REF##17075117##Van Custem et al, 2006##). Docetaxel has also been used in combination with capecitabine, irinotecan and epirubicin in different phase II studies. An ongoing phase I and II study in five Danish oncology centres evaluates a combination of docetaxel, epirubicin and capecitabine. The primary end point for phase I is dose-limited toxicity. Primary end point for phase II study is response rate and secondary end points are PFS and OS.</p>", "<p>The aims of the present phase II study, was to confirm the efficacy, convenience and tolerability of short-time EXE. We found a promising response rate of 46%, PFS of 6.8 months and OS of 10.1 months and all these efficacy data are comparable to the data observed in similar studies.</p>", "<p>Short-time EXE is very well tolerated. Only seven patients (13%) stopped chemotherapy because of toxicity. Worst haematologic toxicity was neutropenia (grade 3–4 toxicity in 17 patients) but no patient had febrile neutropenia. Non-haematologic toxicity was primarily diarrhoea, nausea and vomiting but only two patients experienced nausea/vomiting grade 3–4. Despite the fact that oxaliplatin was infused in 30 min no patient developed peripheral neuropathy grade 3 and no patient experienced laryngopharyngeal dysaesthesia.</p>", "<p>We therefore conclude, that a combination of epirubicin, capecitabine and oxaliplatin every third week is a convenient regimen that easily can be administrated in the outpatient setting but the regimen needs further evaluation in a phase III study.</p>" ]

[]

[ "<p>Epirubicin, cisplatin and continuous infusion of 5-FU is a widely used palliative regimen in patients with gastric cancer. If cisplatin is substituted by oxaliplatin and 5-FU by capecitabine this regimen can be administered in the outpatient setting. Dose-limiting toxicity of oxaliplatin is peripheral sensory neuropathy and it is recommended to give oxaliplatin as a 120 min infusion. However, in patients with colorectal cancer a 30 min infusion of oxaliplatin can safely be administered without increasing neurotoxicity, standard infusion time is 30 min at our departments. In our phase I study the recommended doses of EXE was established (Dupont <italic>et al</italic>, 2006). Patients with non-resectable gastric adenocarcinoma were eligible. Patients received EXE (epirubicin 50 mg m⁻² day 1; capecitabine 1000 mg m⁻² day⁻¹ continuously and oxaliplatin 130 mg m⁻² day 1) as outpatient therapy every third week for a maximum of 8 cycles. From June 2004 to September 2005, we enroled 54 patients. Median age was 60 years (31–74 years) Median number of courses was 6 (1–8). Response rate was 45%. Median PFS was 6.8 (5.2–7.9) months and median survival was 10.1 (7.9–11.1) months. Most important grade 3 toxicities were as follows: nausea, vomiting, and diarrhoea (6%). Neurotoxicity grade 2 was seen in 36.5%. We therefore conclude, that EXE every third week is a convenient regimen that easily can be administrated in the outpatient setting but the regimen needs further evaluation in a phase III study.</p>" ]

[ "<p>Gastric cancer (GC) remains a significant global health problem and is one of the most frequent causes of cancer-related deaths. Although the incidence has been declining in Europe during the last century it is still the fifth most common cancer and an estimated 171 000 new cases of GC will be diagnosed each year (##REF##15718248##Boyle and Ferlay, 2005##). The prognosis is very poor, with a 5-year overall survival of approximately 10% in Denmark (##UREF##1##Coleman et al, 2003##). The only potentially curative treatment for GC is surgery, but unfortunately most cases are diagnosed at an advanced stage because of very few organ-specific symptoms and therefore only about 20% of the patients are suitable for curative surgery (##UREF##4##Janunger et al, 2001##).</p>", "<p>Therefore the need for palliative treatment is huge for these patients. The efficiency of palliative chemotherapy is widely accepted. Data from randomised studies have shown that combination chemotherapy results in a significant survival advantage and improves quality of life in patients with advanced GC (##REF##8508427##Murad et al, 1993##; ##REF##7533517##Pyrhonen et al, 1995##; ##UREF##3##Glimelius et al, 1997##; ##REF##16782930##Wagner et al, 2006##).</p>", "<p>In the past decade a number of effective chemotherapy regimens have been established worldwide. However, these studies have not resulted in a standard regimen for gastric cancer, but the backbone of these regimens is most often a combination of 5-fluorouracil (5-FU) and platinum. In Europe ECF (epirubicin 50 mg m⁻² day 1, cisplatin 60 mg m⁻² day 1 and continuous infusion (CVI) of 5-FU 200 mg m⁻² daily) is widely used as it was shown that ECF improves response rate and prolongs survival when compared with FAMtx which was considered to be a standard therapy a decade ago (##REF##8996151##Webb et al, 1997##; ##REF##10390007##Waters et al, 1999##).</p>", "<p>We sought to find a regimen that was comparable with ECF in efficacy, easy to administer in an outpatient setting and with low toxicity.</p>", "<p>Therapy with 5-FU requires a central venous catheter which is associated with risk of venous thromboses and infection (##REF##7993836##Findlay et al, 1994##; ##REF##8630289##Evans et al, 1996##). Capecitabine is an oral fluoropyrimidine, which is absorbed from the gastrointestinal tract as an intact molecule and afterwards converted into 5-FU in a cascade of three enzymatic steps (##REF##9849491##Miwa et al, 1998##). Capecitabine is a well-established alternative which simplifies the administration of 5-FU and overcomes the problem with the central venous catheter.</p>", "<p>Another cornerstone in ECF is cisplatin, which is highly emetic and in addition intravenous hydration is required to prevent renal toxicity.</p>", "<p>Oxaliplatin is a third-generation platinum that easily can be administered in outpatients without hydration. When compared with cisplatin, oxaliplatin has a better safety profile with regard to renal toxicity and emesis. Recent studies have shown synergism between oxaliplatin and 5-FU and a combination of these two drugs has proven effective as first- or second-line treatment for advanced colorectal cancer (##REF##10944126##De Gramont et al, 2000##; ##UREF##6##Raymond et al, 2001##; ##REF##15169795##Cassidy et al, 2004##). The dose-limiting toxicity of oxaliplatin is peripheral sensory neuropathy, which is reversible but cumulative. Usually it is recommended to give oxaliplatin as a 120 min infusion to minimize peripheral neuropathy. However, we have recently in patients with colorectal cancer demonstrated that a 30 min infusion apparently does not result in more neurotoxicity (##REF##14968944##Pfeiffer et al, 2003##, ##UREF##5##2006##).</p>", "<p>We therefore used 30 min infusion of oxaliplatin in both our phase I and phase II trial (##UREF##2##Dupont et al, 2006##). The primary aims of our phase II study were to evaluate response rate and toxicity of EXE in patients with advanced gastric cancer. Secondary end points were progression-free survival and overall survival.</p>" ]

[]

[ "<fig id=\"fig1\"><label>Figure 1</label><caption><p>Time-to-progression and overall survival. Kaplan–Meier curves of time-to-progression (median 6.8 months; 95% CI 5.2–7.9 months) and overall survival (median 10.1 months; 95% CI 7.9–11.1 months).</p></caption></fig>" ]

[ "<table-wrap id=\"tbl1\"><label>Table 1</label><caption><title>Patient characteristics at baseline</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Characteristic</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>\n<italic>N</italic>\n</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td colspan=\"2\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Age</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Median</td><td align=\"center\" valign=\"top\" charoff=\"50\">59</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Range</td><td align=\"center\" valign=\"top\" charoff=\"50\">31–74</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"2\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>WHO performance status</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 0</td><td align=\"center\" valign=\"top\" charoff=\"50\">25</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 1</td><td align=\"center\" valign=\"top\" charoff=\"50\">26</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 2</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"2\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Location</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Cardia</td><td align=\"center\" valign=\"top\" charoff=\"50\">43</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Corpus</td><td align=\"center\" valign=\"top\" charoff=\"50\">11</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Prior adjuvant chemotherapy</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"2\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>No. of organs involved</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 1</td><td align=\"center\" valign=\"top\" charoff=\"50\">30</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 2</td><td align=\"center\" valign=\"top\" charoff=\"50\">17</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> &gt;3</td><td align=\"center\" valign=\"top\" charoff=\"50\">7</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"2\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Stage</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Locally advanced disease</td><td align=\"center\" valign=\"top\" charoff=\"50\">9</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Metastatic disease</td><td align=\"center\" valign=\"top\" charoff=\"50\">45</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"2\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Status of primary tumour</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> R0 resection</td><td align=\"center\" valign=\"top\" charoff=\"50\">9</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> R1 resection</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> R2 resection</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> No surgery</td><td align=\"center\" valign=\"top\" charoff=\"50\">39</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"><italic>Increased alkaline phosphatase</italic> (&gt;300 U l⁻¹)</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"><italic>Increased ALAT</italic> (0 U l⁻¹)</td><td align=\"center\" valign=\"top\" charoff=\"50\">8</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl2\"><label>Table 2</label><caption><title>Worst toxicity after median 6 cycles of EXE</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Grade 2 <italic>n</italic> (%)</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Grade 3 <italic>n</italic> (%)</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Grade 4/5 <italic>n</italic> (%)</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Haematologic toxicity</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Neutropenia</td><td align=\"center\" valign=\"top\" charoff=\"50\">7 (13)</td><td align=\"center\" valign=\"top\" charoff=\"50\">8 (15)</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (2)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Thrombocytopenia</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (2)</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (2)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Infection (no neutropenia)</td><td align=\"center\" valign=\"top\" charoff=\"50\">3 (6)</td><td align=\"center\" valign=\"top\" charoff=\"50\">3 (6)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Febrile neutropenia</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Nonhaematologic toxicity</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Diarrhoea</td><td align=\"center\" valign=\"top\" charoff=\"50\">7 (13)</td><td align=\"center\" valign=\"top\" charoff=\"50\">3 (6)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> PPE</td><td align=\"center\" valign=\"top\" charoff=\"50\">5 (10)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Nausea</td><td align=\"center\" valign=\"top\" charoff=\"50\">17 (32)</td><td align=\"center\" valign=\"top\" charoff=\"50\">3 (6)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Vomiting</td><td align=\"center\" valign=\"top\" charoff=\"50\">12 (22)</td><td align=\"center\" valign=\"top\" charoff=\"50\">2 (4)</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (2)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Neuropathy</td><td align=\"center\" valign=\"top\" charoff=\"50\">19 (36)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Bleeding</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (2)</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"6604569f1\"/>" ]

[]

{ "acronym": [], "definition": [] }

[]

[ "<title>MATERIALS AND METHODS</title>", "<title>Data mining of publicly available prostate cancer mRNA expression data</title>", "<p>We interrogated the common gene expression databases, Oncomine and Arrayexpress, for differential expression of <italic>GOLPH2</italic> mRNA in human prostate cancer and normal tissue (##REF##15068665##Rhodes et al, 2004##; ##REF##17132828##Parkinson et al, 2007##). We identified nine studies within Oncomine (##REF##11518967##Dhanasekaran et al, 2001##, ##REF##15548588##2005##; ##REF##11406537##Luo et al, 2001##, ##REF##11807955##2002##; ##REF##12873976##Vanaja et al, 2003##; ##REF##14711987##Lapointe et al, 2004##; ##REF##16286247##Varambally et al, 2005##; ##REF##16513839##Nanni et al, 2006##; ##REF##17173048##Tomlins et al, 2007##) and one study within Arrayexpress (##REF##16618720##Liu et al, 2006##). Altogether, these studies interrogated 305 samples of prostate cancer in combination with 148 of benign prostate tissues. From the Oncomine database, the normalised expression values for the nine studies were extracted and analysed using SPSS.</p>", "<title>Prostate cancer patients</title>", "<p>Six hundred and fourteen prostate cancer patients who underwent radical prostatectomy between 1999 and 2005 were enclosed in this study. Patient age ranged between 43 and 74 years (median 62 years). Preoperative PSA levels ranged from 0.8 to 39 ng ml⁻¹ (median 7.2). Forty-four patients (7.2%) had received gonadotropin-releasing hormone analogues at the discretion of the referring urologist before surgery (median 4 weeks, range 2–16 weeks). Clinical follow-up data were assessed annually. Prostate-specific antigen relapse-free survival time was available for 479 patients. The median follow-up time of all cases was 17 months (range 1–68 months). The median follow-up time of patients without a PSA relapse was 18 months (range 4–68 months). Forty-three patients (9%) experienced a PSA relapse after a median time of 5 months (range 1–52). The Gleason scores (GS) in the cohort were distributed as follows: GS 2–6: 217 (35.3%) GS 7: 291 (47.4%), GS 8–10: 106 (17.3%). Four hundred –and twenty cases had organ-confined carcinomas (pT2); 191 cases showed extracapsular tumour extension (pT3). The surgical margins were clear (R0) in 444 cases; 167 cases had positive margins (R1) and 3 cases were Rx. Use of this tissue has been approved by the Charite’ University Ethics Committee under the title ‘Retrospektive Untersuchung von Gewebeproben mittels immunhistochemischer Färbung und molekularbiologischer Methoden’ (‘Retrospective analysis of tissue samples by immunohistochemistry and molecular biological techniques’) (EA1/06/2004) on 20 September 2004.</p>", "<title>Screening tissue microarray construction</title>", "<p>Formalin-fixed and paraffin-embedded material of a representative variety (185 spots) of normal and malignant human tissues and tumour cell lines was compiled and assembled on a single block, as described (##REF##16675566##Varga et al, 2006##).</p>", "<title>Prostate tissue microarray construction</title>", "<p>Formalin-fixed, paraffin-embedded tissue blocks of radical prostatectomy specimens were selected according to tissue availability for construction of a TMA. Each case was represented by five tissue cores. In all cases, benign prostatic hyperplasia (BPH) of the transitional zone, normal tissue from the peripheral zone, prostatic intraepithelial neoplasia (PIN), if present (otherwise another core from the peripheral zone), and two cores of invasive carcinoma, ideally of primary and secondary GS, were selected for TMA construction. The core diameter was 1.0 mm. All cases were arranged in 40 TMA recipient paraffin blocks.</p>", "<title>Immunohistochemistry</title>", "<p>The TMA blocks were freshly cut (3 <italic>μ</italic>m) and mounted on superfrost slides (Menzel Gläser, Braunschweig, Germany). Immunohistochemistry was conducted with the Ventana Benchmark automated staining system (Ventana Medical Systems, Tucson, AZ, USA) using Ventana reagents for the entire procedure. To detect GOLPH2, two commercially available antibodies (mouse monoclonal, clone 5B10; Abnova Corp., Taipei, Taiwan, catalogue no. H00051280-M06, dilution 1 : 1000 and rabbit polyclonal; Abcam, Cambridge, UK, catalogue no. Ab22209, dilution 1 : 100) were diluted in a Ventana diluent. To detect racemase and p63, we created a cocktail of racemase (rabbit polyclonal; Biologo, Kronshagen, Germany, dilution 1 : 30) and p63 (clone mix 4A4/Y4A3; Neomarkers, Fremont, CA, USA, dilution 1 : 200) in a Ventana diluent. Primary antibodies were detected using the UltraVIEW DAB detection kit using the benchmarks CC1m- heat-induced epitope retrieval. For the racemase/p63 cocktail, the signal was further enhanced with the amplification kit. Slides were counterstained with haematoxylin, dehydrated and mounted.</p>", "<title>Evaluation of the immunohistochemical stainings</title>", "<p>Chromogenic immunohistochemistry using both GOLPH2 antibodies was primarily conducted on a multitissue array constructed for antibody testing comprising 185 human tissue spots and cell lines. The immunostainings were evaluated by two genitourinary pathologists (GK, FFR) and one histopathology resident (CJ) simultaneously on a multiheaded microscope.</p>", "<p>For both GOLPH2 and racemase, we evaluated staining intensity with a four-tiered system: 0 (negative), 1+ (weak), 2++ (moderate), 3+++ (strong) in benign tissue, PIN and invasive carcinoma. To also detect very subtle staining intensity differences, we further created a dichotomous (‘tumour&gt;normal’) ratio to better indicate upregulation in tumour in comparison with adjacent normal tissue. Equal or less GOLPH2 staining intensity in carcinomatous tissue was reported as ratio 0, higher staining intensities than in normal glands were regarded as ratio 1.</p>", "<p>Heterogeneity of marker expression in invasive carcinoma was also recorded and diagnosed if more than 25% of the tumour showed a variation of staining intensity exceeding one scoring category. P63 immunoreactivity of the racemase/p63 cocktail was sometimes used to clearly distinguish benign and malignant glands.</p>", "<title>Monoclonal and polyclonal GOLPH2 double staining by immunofluorescence</title>", "<p>GOLPH2 is a Golgi protein. To better assess the specificity of the polyclonal and the monoclonal antibody, a double staining by immunofluorescence was conducted. Primary antibodies (mouse-anti GOLPH2, Abnova Corp., 1 : 4000; rabbit-anti GOLPH2, Abcam Ltd., 1 : 200) were coincubated on a de-paraffinised prostate tissue slide after heat-induced antigen retrieval (5 min, citrate buffer, pH 6.0, 110°C) at room temperature for 30 min. Binding was detected by fluorescence-labelled secondary antibodies (goat anti-rabbit-Alexa546 and goat anti-mouse-Alexa488, both from Molecular probes, catalog nos. A11010 and A11029) under a fluorescence microscope.</p>", "<title>Antibody preincubation with immunogenic peptide</title>", "<p>To further assess antibody specificity, the monoclonal antibody was incubated with an excess of the immunogenic peptide provided by the antibody supplier (partial recombinant protein (NP_057632, 302 aa–402 aa) with GST, Abnova Corp.) at 4°C overnight before application to the control tissue (Figure 2F).</p>", "<title>Statistical analysis</title>", "<p>Statistical analysis was performed using SPSS version 15.0. <italic>P</italic>-values &lt;0.05 were considered significant.</p>" ]

[ "<title>RESULTS</title>", "<title>GOLPH2 mRNA expression in prostate cancer</title>", "<p>We have reported earlier that <italic>GOLPH2</italic> mRNA is overexpressed in microdissected prostate cancer epithelium compared with the adjacent normal prostate epithelium from the same patient by a fold change of 2.2 (##REF##15532095##Kristiansen et al, 2005##). ##REF##16618720##Liu et al (2006)## described <italic>GOLPH2</italic> mRNA as overexpressed by a fold change of 3.14 in their samples (13 normal; 45 cancer), which did not correlate to tumour differentiation according to GS. A comprehensive analysis of the studies from Oncomine combining 260 samples from CaP and 135 from benign prostate normal revealed an overexpression of <italic>GOLPH2</italic> by a factor of 2.7 in prostate cancer (<italic>P</italic>&lt;0.001, ##FIG##0##Figure 1A##).</p>", "<title>GOLPH2 protein expression in normal and neoplastic human tissues</title>", "<p>Both GOLPH2 antibodies showed identical stainings on a multitissue array comprising 185 tissue spots and cell lines (##FIG##1##Figure 2A and B##). Highest rates of GOLPH2 expression were seen not only in adenocarcinomas of the prostate, colon and breast, but also in renal cell cancer and hepatocellular carcinoma (##TAB##0##Table 1##). Prostate cancer showed the strongest staining.</p>", "<p>GOLPH2 expression can be observed in mesenchymal cells and epithelia, but with strongly differing intensities. As can be expected from a Golgi-associated protein, it shows a distinct semigranular dot-like staining pattern and is localised perinuclearly towards the cell apex in epithelia, whereas the rest of the cytoplasm is remarkably clear (##FIG##1##Figure 2A and B##, ##FIG##2##Figures 3##, ##FIG##3##4## and ##FIG##4##5##). To further cross-validate antibody specificity, a double immunofluorescent staining using both antibodies was conducted and demonstrated a clear colocalisation of the antigens in the Golgi apparatus (##FIG##1##Figure 2C–E##). Again, the monoclonal and the polyclonal antibody showed identical staining localisations. The monoclonal antibody was preferred for further immunostaining our prostate cancer cohort as it yielded slightly less background and a more intense signal at lower concentrations. In addition, the preincubation of the monoclonal antibody with an excess of recombinant GOLPH2 protein completely abolished immunoreactivity (##FIG##1##Figure 2F##).</p>", "<title>GOLPH2 immunostaining in prostate tissues</title>", "<p>Perinuclear GOLPH2 expression is present in normal and neoplastic prostate glands with unequivocal upregulation in most hyperplastic and neoplastic glands in comparison with normal glands. Some cases of PIN and carcinoma, however, display an attenuation of the Golgi staining and an additional diffuse strong cytoplasmic immunoreactivity, which was seen in 68 cases (11.1%) (##FIG##2##Figure 3C and F##).</p>", "<p>Normal prostatic glands show finely granular GOLPH2 staining, in some instances with an almost linear pattern (##FIG##2##Figure 3A##). The staining intensity is relatively weak yielding a more golden than brownish DAB precipitate. The median GOLPH2 intensity in normal tissue was 1+ (##FIG##0##Figure 1B##). Hyperplastic glands of BPH show a moderate-to-strong staining intensity (##FIG##2##Figure 3B##). Glands with slightly atypical epithelia also show a stronger immunoreactivity than the normal glands. This becomes even more pronounced in high-grade PIN (median intensity 2+) and invasive carcinoma (median intensity 3+) wherein the granules are much coarser and stain deeply brown, which yields a well-discernable contrast to adjacent benign glands (##FIG##2##Figures 3C–G##, ##FIG##4##5C and F##). The statistical differences between GOLPH2 expression in normal, PIN and carcinoma were highly significant (##FIG##0##Figure 1B##; Wilcoxon's signed rank test, <italic>P</italic>&lt;0.001).</p>", "<title>GOLPH2 histopathology and survival</title>", "<p>GOLPH2 protein expression in prostate cancer was not associated with pT stage, differentiation grade (GS) and preoperative PSA levels. There was no association with disease-free survival (Cox regression, relative risk 0.969, <italic>P</italic>=0.910).</p>", "<title>GOLPH2 as a potential tool for prostate cancer diagnosis</title>", "<p>The most impressive finding of this expression analysis was a striking difference in GOLPH2 expression in normal and neoplastic prostate glands. In 237 cases (38.6%), GOLPH2 intensity was two scoring points higher in tumour epithelia than in normal glands; in another 324 cases (52.8%), tumoral GOLPH2 expression excelled by one scoring point; in 51 cases (8.3%), no differences between normal and tumour were noted; and in only two cases (0.3%), normal tissue showed a stronger GOLPH2 staining than adjacent tumour. In summary, 91.4% of cases showed an upregulation of GOLPH2 in the tumour by at least one scoring point. This rate is even higher in the separately scored ‘tumour&gt;normal’ ratio. This is able to measure even subtle differences: 567 of 614 cases (92.3%) had a ratio of 1 and only 47 cases (7.7%) had a ratio of 0.</p>", "<p>To characterise GOLPH2 as a new diagnostic tissue marker of prostate cancer, we conducted a careful comparison with the well-established AMACR immunohistochemistry. As expected, AMACR was found overexpressed in high-grade PIN (median score 1+) and invasive prostate cancer (median score 2+), whereas normal tissues were found to be negative (median score 0) (##FIG##0##Figure 1B##). AMACR overexpression in the tumour in direct comparison with adjacent normal tissue (‘tumour&gt;normal’ ratio) was seen in 95% of cases. AMACR expression was significantly but not highly correlated to GOLPH2 expression (##TAB##1##Table 2##, Spearman rank correlation coefficient 0.113, <italic>P</italic>=0.005). However, both markers also showed remarkable differences, particularly, when the tumour/normal ratio of GOLPH2 and AMACR was considered. Here, 26 of 31 AMACR-negative cases (84%) were identified by GOLPH2. On the other hand, 42 of 47 cases (89%) without GOLPH2 upregulation were AMACR-positive. Five cases were concordantly negative and 541 cases were positive for both markers (##TAB##2##Table 3##). Four of the five cases, negative for both markers, were of higher GS. Examples of the comparison of AMACR and GOLPH2 expression in prostate tissues are shown in ##FIG##3##Figures 4## and ##FIG##4##5##.</p>", "<p>The histologically evident intratumoral heterogeneity of prostate cancer is also reflected in biomarker expression. In this study, intratumoral AMACR and GOLPH2 heterogeneity of expression was also evaluated. AMACR has a considerably higher degree of heterogeneous expression (45% of cases) than GOLPH2 (25%). This heterogeneity (##FIG##3##Figure 4E##) can be troublesome in small tumour foci. In 43 cases, one of two TMA tumour cores was completely AMACR-negative, whereas the other core of the same case showed some immunoreactivity. Of these AMACR-negative cores, GOLPH2 was upregulated in 36 cases (84%).</p>", "<p>The combination of GOLPH2 and AMACR showed expression of either marker in 99.2% of cancer cases, which advocates a combined use of AMACR and GOLPH2 as positive confirmative markers of prostate cancer.</p>" ]

[ "<title>DISCUSSION</title>", "<p>This is the first report on GOLPH2 (Golgi protein 73, GP73) protein expression in prostate tissues validated on a large cohort of clinically detected prostate cancer specimens following radical prostatectomy. We have recently shown that <italic>GOLPH2</italic> mRNA is among the top upregulated transcripts in prostate cancer (##REF##15532095##Kristiansen et al, 2005##), which is in line with other profiling studies (##REF##11518967##Dhanasekaran et al, 2001##, ##REF##15548588##2005##; ##REF##11406537##Luo et al, 2001##, ##REF##11807955##2002##; ##REF##12873976##Vanaja et al, 2003##; ##REF##14711987##Lapointe et al, 2004##; ##REF##16286247##Varambally et al, 2005##; ##REF##16618720##Liu et al, 2006##; ##REF##16513839##Nanni et al, 2006##; ##REF##17173048##Tomlins et al, 2007##). In our meta-analysis of publicly available expression data encompassing 260 prostate cancer cases, a mean fold change of 2.7 for <italic>GOLPH2</italic> upregulation in cancerous tissues was found. However, a detailed tissue-based <italic>in situ</italic> analysis of GOLPH2 protein in prostate tissues was lacking so far. Very recently, this widely acknowledged upregulation of <italic>GOLPH2</italic> was put into practise: ##REF##18245462##Laxman et al (2008)## included <italic>GOLPH2</italic> in a multiplex RT–PCR panel of markers composed of transcripts known to be overexpressed in prostate cancer, which, as a urine-based screening test, allows detecting prostate cancer with a higher sensitivity than a classical PSA blood test.</p>", "<p>GOLPH2 is a 73-kDa Golgi apparatus-associated protein coded by the gene <italic>GOLM1</italic> located on chromosome 9q21.33 and was originally cloned from a library derived from liver tissue of a patient with adult giant-cell hepatitis (##REF##10831838##Kladney et al, 2000##). The initial report also described GOLPH2 expression in a variety of other human tissues at RNA and protein level and demonstrated colocalisation of GOLPH2 with giantin, a type II Golgi membrane protein located at the <italic>cis</italic> and medial Golgi compartment. Structurally, GOLPH2 protein consists of a short cytoplasmic N terminus, a membrane-spanning region, some coiled-coil domains and a longer luminal C terminus with several potential glycosylation sites. The functions and the mechanisms of GOLPH2 regulation in normal and neoplastic tissues are still unclear. It can be generally assumed that it is either involved in post-translational protein modification, transport of secretory proteins, cell signalling regulation or simply maintenance of Golgi apparatus function. Functional assays are necessary to clarify whether GOLPH2 overexpression confers pro-tumorigenic properties to tumour cells and how it is regulated. First colocalisation experiments with GPP130, another Golgi marker, hinted at a differential colocalisation with GOLPH2 in normal and malignant prostate tissues, which deserves further study. GOLPH2 has several potential glycosylation sites and up to 75% of GOLPH2 secreted from hepatocytes is fucosylated, but so far the glycosylation patterns of GOLPH2 in malignant and normal prostatic epithelia have not been analysed (##UREF##0##Norton et al, 2007##).</p>", "<p>In the liver cancer cell line HepG2, GOLPH2 was found strongly upregulated after adenoviral infection, which suggested GOLPH2 as a marker of viral infection in liver tissue and which was confirmed in following studies incorporating clinical samples (##REF##12029628##Kladney et al, 2002a##, ##REF##12359426##2002b##). More recently, GOLPH2 was found upregulated in the serum of patients with hepatocellular carcinoma (HCC) compared with healthy individuals and has been proposed as a new serum marker of HCC, which is more sensitive than <italic>α</italic>-fetoprotein (##REF##15642945##Block et al, 2005##; ##REF##16137783##Marrero et al, 2005##). Apparently, GOLPH2-overexpressing hepatocytes secrete this normally membrane-bound Golgi protein after cleavage into the serum, which can be diagnostically utilised (##REF##17662025##Bachert et al, 2007##). We can confirm the GOLPH2 expression in HCC; however, the finding that adenocarcinomas of the colorectum, the breast and the prostate showed equally strong or even stronger immunostainings argues against GOLPH2/GP73 as a HCC-specific tissue marker. This finding also implies that further serum analysis of non-HCC cancer patients, especially prostate cancer patients, is clearly necessary, before the role of GOLPH2/GP73 as a serum marker specific for HCC can be further established.</p>", "<p>Histological diagnosis of prostate cancer mainly rests on the conventional parameters of morphological architecture and cytology. Prostate-specific antigen serum screening has led to an increase of prostate needle biopsies in the last two decades, which in turn increased the rate of difficult diagnostic situations (small carcinoma infiltrates <italic>vs</italic> benign mimickers of carcinoma) where immunohistochemical tests are necessary. Loss of basal cells is a hallmark of prostate cancer; hence, high molecular weight cytokeratins and p63 have become widely used basal cell tissue markers. However, even with a loss of basal cells, cancer diagnosis can be problematic in some cases. Additional markers of prostate cancer are desirable. So far only AMACR/racemase has gained wider acceptance as a positive marker of prostate cancer, although is has two well-known limitations: intratumoral heterogeneity, which was confirmed in 45% of our cases, and AMACR-negative carcinomas (##REF##16491480##Wang et al, 2006##; ##REF##17222253##Murphy et al, 2007##). In our series, 31 completely AMACR-negative carcinomas (5%) and another 43 cases (7%), in which one of both tumour cores on the TMA was negative, were seen. In these 12% of cases, which might have been considered negative on a needle biopsy, an additional GOLPH2 immunostaining would have allowed a cancer diagnosis in 84% of cases. This is partially because of the considerably lower rate of intratumoral heterogeneity of GOLPH2, which was 25% in our series. These findings clearly advocate the use of GOLPH2 as an additional ancillary positive marker for the histological detection of prostate cancer. Comparable with the introduction of AMACR, we would expect that the number of unclear cases can be further lowered by GOLPH2, which would help to avoid costly and unnecessary rebiopsies (##REF##14750247##Jiang et al, 2004##). Although GOLPH2 immunostaining is not as easy to read as an AMACR staining at first sight, mainly because of the physiological basal GOLPH2 expression in normal tissues, we think that the internal positive control of immunoreactivity in normal tissues can also be seen as an advantage. In addition, the characteristic Golgi pattern is another indicator of specific immunoreactivity, whereas a general overstaining of a slide is often more diffusely cytoplasmic.</p>", "<p>In spite of our comprehensive description of GOLPH2 as a positive marker of malignancy, we would hesitate to recommend using GOLPH2 as the primary second-line antibody after basal cell markers for determining malignancy. First, its sensitivity is slightly lower (92.3%) than AMACR (95.0%), which is, of course, compensated for by its higher homogeneity. Secondly, and more importantly, definition of a positive test result requires adjacent normal glands for direct comparison. As high-grade PIN and hyperplastic benign glands can also show GOLPH2 upregulation, it can be difficult, or even impossible, to diagnose an atypical focus that lacks adjacent normal glands by GOLPH2 immunohistochemistry alone. The comparison with normal tissue is mandatory to obtain a valid result. Another caveat stems from the construction of our TMA, which has been compiled after central review of 640 fully embedded prostatectomy specimens to allow an immunohistochemical evaluation of representative normal and tumour tissue. Benign cancer mimickers, which can be particularly problematic to diagnose on needle biopsies, were not intentionally sampled. Further validation of the diagnostic value of GOLPH2 in rare cancer variants and benign cancer mimickers is necessary.</p>", "<p>In summary, this study is the first to comprehensively confirm at protein level the GOLPH2 upregulation in prostate cancer, which has been suggested in preceding mRNA profiling studies. The high rate of GOLPH2 protein overexpression, which is also seen in AMACR-negative prostate cancer cases, suggests its use as an additional ancillary positive tissue marker of prostate cancer.</p>" ]

[]

[ "<p><italic>GOLPH2</italic> is coding the 73-kDa type II Golgi membrane antigen GOLPH2/GP73. Upregulation of <italic>GOLPH2</italic> mRNA has been recently reported in expression array analyses of prostate cancer. As GOLPH2 protein expression in prostate tissues is currently unknown, this study aimed at a comprehensive analysis of GOLPH2 protein in benign and malignant prostate lesions. Immunohistochemically detected GOLPH2 protein expression was compared with the basal cell marker p63 and the prostate cancer marker <italic>α</italic>-methylacyl-CoA racemase (AMACR) in 614 radical prostatectomy specimens. GOLPH2 exhibited a perinuclear Golgi-type staining pattern and was preferentially seen in prostatic gland epithelia. Using a semiquantitative staining intensity score, GOLPH2 expression was significantly higher in prostate cancer glands compared with normal glands (<italic>P</italic>&lt;0.001). GOLPH2 protein was upregulated in 567 of 614 tumours (92.3%) and AMACR in 583 of 614 tumours (95%) (correlation coefficient 0.113, <italic>P</italic>=0.005). Importantly, GOLPH2 immunohistochemistry exhibited a lower level of intratumoral heterogeneity (25 <italic>vs</italic> 45%). Further, GOLPH2 upregulation was detected in 26 of 31 (84%) AMACR-negative prostate cancer cases. These data clearly suggest GOLPH2 as an additional ancillary positive marker for tissue-based diagnosis of prostate cancer.</p>" ]

[ "<p>The identification of sensitive and specific biomarkers in tissue and serum is of utmost importance to reduce the mortality of prostate cancer (##REF##17936845##Parekh et al, 2007##). Expression arrays, SNP analyses and mass spectrometry are new tools for biomarker identification (##REF##17925536##Zheng et al, 2007##). Such high-throughput analyses have recently identified new prostate cancer biomarkers, including, for example, HEPSIN, EZH2 and <italic>α</italic>-methyl-Co-racemase (AMACR) (##REF##11518967##Dhanasekaran et al, 2001##; ##REF##11684956##Jiang et al, 2001##; ##REF##11406537##Luo et al, 2001##; ##REF##11479199##Magee et al, 2001##; ##REF##11696729##Stamey et al, 2001##; ##REF##12374981##Varambally et al, 2002##; ##REF##12734317##Rhodes et al, 2003##; ##REF##17936845##Parekh et al, 2007##). AMACR has first been found upregulated in prostate cancer by ##REF##10749139##Xu et al (2000)## using suppressive subtractive hybridisation, and AMACR antibodies have become available quickly thereafter (##REF##11684956##Jiang et al, 2001##; ##REF##11926890##Rubin et al, 2002##). Thus far, it is the only new tissue biomarker of prostate cancer that has gained clinical acceptance. AMACR is frequently used in combination with the basal cell markers p63, CK5/6 and 34-<italic>β</italic>E12. In diagnostic histopathology, the absence of these basal cells, which usually line the periphery of normal prostate glands, is (with very rare exceptions) a defining criterion of invasive tumour growth (##REF##2410099##Brawer et al, 1985##; ##REF##9620026##Kaleem et al, 1998##; ##REF##11106548##Signoretti et al, 2000##). However, it can be difficult to ascertain a cancer diagnosis in prostate needle biopsies. Use of an additional positive prostate cancer marker is desirable. AMACR immunohistochemistry can show dramatic pictures of strongly positive cancer glands infiltrating perfectly negative benign prostatic parenchyma and in these cases its use may turn a diagnosis of atypical glands into a straightforward diagnosis of cancer (##REF##12766580##Zhou et al, 2003##, ##REF##15043314##2004##; ##REF##14739905##Epstein, 2004##; ##REF##16469560##Epstein and Herawi, 2006##). However, it has been recognised that AMACR may be false-negative in up to 18% of prostate cancer foci on biopsies and even higher in some carcinoma subtypes (##REF##14739905##Epstein, 2004##; ##REF##15043314##Zhou et al, 2004##).</p>", "<p>Recently, <italic>GOLPH2</italic> mRNA expression has been reported to be upregulated in prostate cancer tissues (##REF##11807955##Luo et al, 2002##; ##REF##14711987##Lapointe et al, 2004##; ##REF##15532095##Kristiansen et al, 2005##). GOLPH2 is a Golgi phosphoprotein of yet unknown function that has until very recently only been described in liver disease as a potential serum marker of hepatocellular carcinoma (##REF##10831838##Kladney et al, 2000##, ##REF##12029628##2002a##; ##REF##15180730##Iftikhar et al, 2004##; ##REF##16137783##Marrero et al, 2005##; ##REF##17662025##Bachert et al, 2007##). <italic>GOLPH2</italic> mRNA has recently been described as an integral part of a multiplex marker to detect prostate cancer from urine samples that even outperformed a prostate-specific antigen (PSA) blood test (##REF##18245462##Laxman et al, 2008##).</p>", "<p>In this study, we performed a comprehensive GOLPH2 protein expression analysis in a broad spectrum of normal and malignant tissues. Further, GOLPH2 expression patterns were studied in detail in different prostatic lesions. We demonstrate that GOLPH2 protein is upregulated in most prostate cancer cases. In addition to AMACR and p63, GOLPH2 antibodies will be helpful in the correct histological diagnosis of prostate cancer.</p>" ]

[ "<p>We are grateful to Britta Beyer and Silvia Behnke for excellent technical assistance. The photographical support of Norbert Wey and André Wethmar is greatly acknowledged. We also thank Christian Zuber, PhD, and Jürgen Roth, MD, (Division of Cell and Molecular Pathology, Department of Pathology, USZ) for helpful discussions on GOLPH2 localisation.</p>" ]

[ "<fig id=\"fig1\"><label>Figure 1</label><caption><p>GOLPH2 expression in prostate tissues at mRNA and protein level. (<bold>A</bold>) Boxplot of the combined normalised expression values of the nine studies from Oncomine interrogating normal and cancerous prostate tissues. The fold changes and the respective <italic>P</italic>-values are indicated above the brackets. CaP=prostate cancer tissue; FC=fold change; N=normal prostate. The open circles indicate outliers. (<bold>B</bold>) Illustration of the progression of GOLPH2 (on the left) and AMACR expression (on the right) from normal tissue through PIN to invasive carcinoma (immunohistochemical data).</p></caption></fig>", "<fig id=\"fig2\"><label>Figure 2</label><caption><p>Characterisation of GOLPH2 antibodies. (<bold>A</bold> and <bold>B</bold>) Chromogenic immunocytochemistry of the paraffin-embedded melanoma cell line PF2000. Both antibodies (<bold>A</bold> – mouse monoclonal, Abnova; <bold>B</bold> – rabbit polyclonal, Abcam) show a strong semigranular perinuclear staining, which is suggestive of a Golgi pattern. (<bold>C</bold> and <bold>D</bold>) Immunofluorescent double staining of a prostate cancer gland using both GOLPH2 antibodies (<bold>C</bold> – mouse monoclonal, <bold>D</bold> – rabbit polyclonal). The signal of both antibodies is clearly located to the golgi apparatus, which can now be appreciated by the higher resolution of immunofluorescence. (<bold>E</bold>) The colocalisation of the immunoreactivity of both antibodies (plus DAPI staining), which shows that the polyclonal antibody (red signal) has a less favourable signal to background ratio. (<bold>F1</bold>) A GOLPH2 immunohistochemistry (monoclonal antibody) of prostate cancer tissue (lower part – malignant glands, upper part – normal glands) and (<bold>F2</bold>) a consecutive section of the same case was immunostained after preincubation of the antibody with an excess of the immunogenic GOLPH2 peptide, which abolishes immunoreactivity.</p></caption></fig>", "<fig id=\"fig3\"><label>Figure 3</label><caption><p>GOLPH2 expression in prostate tissues. (<bold>A</bold>) Normal secretory epithelium of normal prostate glands (immunoreactivity score 1+). (<bold>B</bold>) Hyperplastic gland with stronger GOLPH2 expression (score 2+). (<bold>C</bold>) Transition of normal epithelium (arrowheads) to high-grade PIN. Note prominent nucleoli (arrows). This PIN has a strong GOLPH2 immunoreactivity (3+) and shows an additional diffuse cytoplasmic staining. (<bold>D</bold>) Gleason 3+3=6 adenocarcinoma (central) infiltrating in between normal glands (marked ‘N’). Note the upregulation of GOLPH2 (3+) in comparison with normal glands. (<bold>E</bold>) Same case at a higher magnification. Note the characteristic Golgi pattern. (<bold>F</bold>) Gleason 3+3=6 adenocarcinoma, with a more diffuse cytoplasmic GOLPH2 staining (3+). Note neural invasion (lower left). (<bold>G</bold>) High-grade adenocarcinoma (Gleason score 3+4=7) with a strong and coarse GOLPH2 staining (3+).</p></caption></fig>", "<fig id=\"fig4\"><label>Figure 4</label><caption><p>Comparison between AMACR/p63 and GOLPH2 immunohistochemistry. (<bold>A</bold>) AMACR expression in invasive cancer glands. Epithelium of normal glands, with a p63-positive basal cell layer, is AMACR-negative. (<bold>B</bold>) Sequential section showing GOLPH2 upregulation in matching cancer glands (score 2+); adjacent normal glands are weakly GOLPH2-positive (score 1+). (<bold>C</bold>) Shows an AMACR-negative example of invasive prostate cancer, whereas the same tumour has a significant upregulation of GOLPH2 (<bold>D</bold>) in comparison with normal glands (upper left corner, lower right corner). The case depicted in (<bold>E</bold>) and (<bold>F</bold>) has no included normal glands, but nonetheless a very strong GOLPH2 expression (3+) that is rarely seen in normal glands.</p></caption></fig>", "<fig id=\"fig5\"><label>Figure 5</label><caption><p>Two examples (<bold>A</bold>–<bold>C</bold>, <bold>D</bold>–<bold>F</bold>) of prostate needle biopsies (H&amp;E, AMACR/p63, GOLPH2). (<bold>A</bold>) Prostate needle biopsy with a small focus of a Gleason 3+3 adenocarcinoma (arrow). Sequential sections of this focus show a lack of p63-positive basal cells and a moderate AMACR immunoreactivity (<bold>B</bold>). GOLPH2 is moderately strongly expressed in these glands, compared with adjacent normal glands (arrowheads), which have a weaker GOLPH2 staining (<bold>C</bold>). (<bold>D</bold>) Another example of a prostate needle biopsy with atypical glands, some are macroacinar (arrow), some (lower right) are smaller (^*). (<bold>E</bold>) The AMACR/p63 cocktail demonstrates a continuous basal cell layer in larger normal gland on top (marked ‘N’), the macroacinar glands directly adjacent to it and the microacinar proliferates in the lower right corner have no basal cells. In between is a larger gland with a disrupted basal cell layer, probably diagnostic of a high-grade PIN. All these glands are strongly positive for AMACR and for GOLPH2 (<bold>F</bold>). It is of importance to note that in this case, both markers (AMACR and GOLPH2) do not differentiate between the high-grade PIN and the invasive carcinoma.</p></caption></fig>" ]

[ "<table-wrap id=\"tbl1\"><label>Table 1</label><caption><title>GOLPH2 expression in normal and neoplastic human tissues and cell lines</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Tissue/cell line (<italic>n</italic>)</bold>\n</th><th align=\"center\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>GOLPH2−</bold>\n</th><th align=\"center\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>GOLPH2+</bold>\n</th><th align=\"center\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>GOLPH2++</bold>\n</th><th align=\"center\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>GOLPH2+++</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Normal testis (2)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Seminoma (2)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Teratoma (2)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Placenta (2)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Invasive lobular breast carcinoma (4)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Invasive ductal breast cancer (4)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Cholangiocarcinoma (2)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Hepatocellular carcinoma (HCC) (2)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Lung adenocarcinoma (1)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Lung squamous cell carcinoma (1)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Lung small cell carcinoma (1)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Serous ovarian carcinoma (2)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Ovarian endometrioid carcinoma (1)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Ovarian mucinous carcinoma (1)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Endometrium endometrioid carcinoma (2)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Endometrium serous carcinoma (2)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Colon adenocarcinoma (4)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">GIST (1)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Skin squamous cell carcinoma (2)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Merkel cell carcinoma (1)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Anaplastic oligodendroglioma (1)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Anaplastic astrocytoma (1)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Glioblastoma multiforme (1)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Thyroid papillary carcinoma (2)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Thyroid follicular carcinoma (1)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Thyoid anaplastic carcinoma (1)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Normal kidney (2)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Clear cell renal cell carcinoma (4)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Papillary renal cell carcinoma (2)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Urothelial carcinoma, bladder (4)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Adenocarcinoma, prostate (4)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Benign prostatic hyperplasia (2)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Normal liver (2)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Tonsils (3)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Non-Hodgkin's lymphoma (4)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Hodgkin's lymphoma (1)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Melanoma (1)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">HA98 (2) (melanoma)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">HN2004 (2) (melanoma)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">PF2000 (2) (melanoma)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">MET5A (2) (mesothelioma)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">SW480 (2) (colon cancer)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">786-O (2) (renal cell cancer)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">H69 (2) (lung cancer)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">MCF-7 (2) (breast cancer)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">SK BR 7 (2) (breast cancer)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">HELA (2) (cervical cancer)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">PC3 (2) (prostate cancer)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">293-T (2) (human embryonal kidney)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl2\"><label>Table 2</label><caption><title>GOLPH2 protein expression in prostate cancer</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"char\" char=\"(\"/><col align=\"char\" char=\"(\"/><col align=\"char\" char=\"(\"/><col align=\"char\" char=\".\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th colspan=\"4\" align=\"center\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>GOLPH2 expression</bold>\n<hr/></th></tr><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th align=\"center\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>1+</bold>\n</th><th align=\"center\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>2+</bold>\n</th><th align=\"center\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>3+</bold>\n</th><th align=\"center\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold><italic>P</italic>-value</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">All cases</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">10 (1.6%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">275 (44.8%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">329 (53.6%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Age</italic>\n</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.321</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> ⩽62</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">6 (1.9%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">143 (46.3%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">160 (51.8%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> &gt;62</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">4 (1.3%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">132 (43.3%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">169 (55.4%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Pre-OP PSA</italic>\n<xref ref-type=\"fn\" rid=\"t2-fna\">a</xref>\n</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.475</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> ⩽10 ng ml⁻¹</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5 (1.1%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">197 (44.6%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">240 (54.3%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> &gt;10 ng ml⁻¹</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5 (3.0%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">73 (44.2%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">87 (52.7%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>pT status</italic>\n</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.267</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> pT2</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">8 (1.9%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">194 (45.9%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">221 (52.2%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> pT3/4<xref ref-type=\"fn\" rid=\"t2-fnb\">b</xref></td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2 (1.0%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">81 (42.4%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">108 (56.5%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Gleason score</italic>\n</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.264</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 3–6</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1 (0.5%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">92 (42.4%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">124 (57.1%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 7</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">7 (2.4%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">136 (46.7%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">148 (50.9%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 8–10</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2 (1.9%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">47 (44.3%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">57 (53.8%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Residual tumour</italic>\n<xref ref-type=\"fn\" rid=\"t2-fnc\">c</xref>\n</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.457</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> R0</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">6 (1.4%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">206 (46.4%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">232 (52.3%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> R1</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">4 (2.4%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">68 (40.7%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">95 (56.9%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>AMACR expression</italic>\n</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.005</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 0</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2 (10.5%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">7 (36.8%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">10 (52.6%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 1+</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">3 (3%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">53 (52.5%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">45 (44.6%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 2+</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2 (0.6%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">152 (47.2%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">168 (52.2%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 3+</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">3 (1.7%</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">63 (36.6%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">106 (61.6%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl3\"><label>Table 3</label><caption><title>Tumour/normal ratio of AMACR and GOLPH2 expression in prostate cancer</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th colspan=\"2\" align=\"center\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>GOLPH2 tumour&gt;normal</bold>\n<hr/></th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </th></tr><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th align=\"center\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>\n<italic>No</italic>\n</bold>\n</th><th align=\"center\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>\n<italic>Yes</italic>\n</bold>\n</th><th align=\"center\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>Total AMACR</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>AMACR tumour&gt;normal</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> <bold>No</bold></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">26</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">31 (5%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> <bold>Yes</bold></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">42</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">541</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">583 (95%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Total GOLPH2</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">47 (7.7%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">567 (92.3%)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<fn-group><fn><p>\n<bold>Conflict of interest</bold>\n</p><p>The authors declare no conflict of interest.</p></fn></fn-group>", "<table-wrap-foot><fn id=\"t2-fn1\"><p>Abbreviations: AMACR=<italic>α</italic>-methylacyl-CoA racemase; Pre-OP PSA=preoperative PSA; pT-status=tumour stage.</p></fn><fn id=\"t2-fna\"><label>a</label><p>Preoperative PSA was not available for seven cases.</p></fn><fn id=\"t2-fnb\"><label>b</label><p>One case was pT4.</p></fn><fn id=\"t2-fnc\"><label>c</label><p>Three cases were Rx.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"t3-fn1\"><p>Abbreviation: AMACR=α-methylacyl-CoA racemase.</p></fn></table-wrap-foot>" ]

[ "<graphic xlink:href=\"6604614f1\"/>", "<graphic xlink:href=\"6604614f2\"/>", "<graphic xlink:href=\"6604614f3\"/>", "<graphic xlink:href=\"6604614f4\"/>", "<graphic xlink:href=\"6604614f5\"/>" ]

[]

{ "acronym": [], "definition": [] }

[]

[ "<title>Patients and methods</title>", "<title>Questionnaire survey</title>", "<p>A survey was conducted by means of a questionnaire circulated to all 21 CCLG paediatric oncology centres in the United Kingdom. The questionnaire was sent to centres in May 2006 and all replies were returned by October 2006. The questionnaire was completed by a paediatric oncology consultant or paediatric oncology research nurse in conjunction with the local medical physicist at each centre. The survey requested detailed information regarding the methodology currently practiced in estimating GFR in paediatric oncology patients. The questionnaire also collected details on the existing practice of estimating carboplatin dose based on patient renal function in children with cancer. A follow-up survey was carried out in February 2008 to establish whether any changes had been made to centre practices since the completion of the original questionnaire.</p>", "<title>Impact of BNMS recommendations on carboplatin dosing</title>", "<p>Retrospective data from 337 GFR tests carried out on a total of 178 children in Newcastle upon Tyne were used to study the impact of the new BNMS recommendations on carboplatin dosing. These patients had a median weight of 36.6 kg (range: 5.0–120.6 kg) and a median age of 12.4 years (range: 0.2–27.4 years). The raw data used for this analysis were from tests conducted in children with blood samples taken at 1, 2, 3 and 4 h after injection of ⁵¹Cr-EDTA and covered a wide range of GFR values. A median GFR value of 64.7 ml min⁻¹ (range: 9.7–202.1 ml min⁻¹), or 95.9 ml min⁻¹ 1.73 m⁻² (range: 14.7–171.2 ml min⁻¹ 1.73 m⁻²), was determined in this patient population, based on the data obtained from blood samples taken at 2, 3 and 4 h and application of the BM correction. This involved fitting the data from the 2, 3 and 4 h samples to a single exponential equation by taking the natural logarithm of the plasma concentrations and use of linear regression analysis against time to determine the slope and intercept at time zero for the exponential. GFR_(SI), the clearance obtained from these data ((−1 × dose × slope)/intercept), was then modified, according to the BM equation developed for use in children (##UREF##0##Brochner-Mortensen et al, 1974##), to account for the more rapid early phase of the clearance. To obtain the BM modified estimate of clearance GFR_(BM), the following three steps were followed:</p>", "<p>(1) the BSA corrected, slope–intercept estimate of GFR is given by: </p>", "<p>(2) the BSA corrected and BM-modified estimate of GFR is given by: </p>", "<p>(3) the BM modified estimate of GFR is given by: </p>", "<p>This data set from 337 GFR tests was used to estimate the difference in carboplatin dose that would be seen after implementation of the new regulations in individual centres. Separate analyses were performed to assess the impact of omitting the 1-h blood sample and the implementation of the BM correction in centres where this was not current practice, i.e., in centres either applying the Chantler correction or using no correction factor. The Newell formula based on patient weight and ⁵¹Cr-EDTA <italic>t</italic>_1/2, equation ##FORMU##3##(1)##, was used to estimate the carboplatin dose to assess the impact of changing sampling times. This involved the determination of ⁵¹Cr-EDTA <italic>t</italic>_1/2 using the equation: <italic>t</italic>_1/2=0.693/elimination rate constant, i.e., using the slope determined from linear regression of the exponential as described above. The alternative Newell formula based on uncorrected GFR and patient weight (modified Calvert formula), equation ##FORMU##4##(2)##, was used to estimate the carboplatin dose when centres implemented a change in the correction factor used (##REF##8246021##Newell et al, 1993##). Both of these formulae were developed for use in a paediatric patient population and the comparisons selected, i.e., the effect of blood sampling on use of the Newell formula and effect of correction factor on use of the modified Calvert formula, were based on the most practical combinations with likely clinical relevance. </p>", "<p>\n\n</p>", "<p>For the purpose of this analysis, a change in carboplatin dose of more than 10% was considered to be of potential clinical relevance.</p>" ]

[ "<title>Results</title>", "<title>Measurement of GFR</title>", "<p>A summary of the findings from the questionnaire relating to current practices in determining renal function in children with cancer at the UK centres is provided in ##TAB##0##Table 1##. The findings indicated that at the time the survey was carried out all centres followed the BNMS guidelines with respect to the use of a radioisotopic method to measure GFR, with 19 of 21 centres using ⁵¹Cr-EDTA and the remaining two centres using ^99mTc-DTPA. Either the central or peripheral route was used to inject the radioisotope tracer in two-thirds of centres, depending on whether the patient had a single or a double-lumen central venous line, whereas the remaining third used the peripheral route only. The BNMS recommendations indicate the use of one lumen of a double-lumen central venous line to inject the radioisotope and the second for blood sampling wherever possible. In situations where only a single-lumen central venous line is present, the peripheral route should be used to inject the radioisotope tracer and the blood samples taken from the central line.</p>", "<p>The number and timing of blood samples taken after injection of radioisotopic tracer varied markedly between centres, with six centres taking two blood samples, 10 centres taking three samples and five centres collecting four blood samples. An early blood sample was taken within 2 h of radioisotope administration in 28% (6/21) of centres, potentially falling within the initial rapid exponential phase of disappearance of the radioisotope. Inappropriate inclusion of this early sample in renal function estimation could potentially lead to the inaccurate determination of a reduced ⁵¹Cr-EDTA <italic>t</italic>_1/2, an increased GFR and resultant overdosing of carboplatin.</p>", "<p>To obtain GFR estimates from the ⁵¹Cr-EDTA raw data, all but one centre (20/21) applied the slope–intercept method, whereas the remaining centre used an undefined ‘in-house’ method to determine renal function. Among those centres who applied the slope–intercept method, 40% (8/20) used the BM correction, whereas 30% (6/20) applied the Chantler correction. The remaining six centres (30%) applied no correction factor. In the majority of centres (16/21), the Medical Physics department was responsible for estimating GFR from the ⁵¹Cr-EDTA data obtained, whereas this task was carried out by the Biochemistry department in the remaining five centres.</p>", "<title>Renal function-based carboplatin dosing</title>", "<p>All CCLG centres involved in the treatment of children with cancer in the United Kingdom routinely use renal function-based carboplatin dosing as part of chemotherapy regimens for many different tumour types. Estimation of carboplatin dose was carried out solely using ⁵¹Cr-EDTA <italic>t</italic>_1/2 values in 52% (11/21) of centres, whereas an additional six centres (29%) used either ⁵¹Cr-EDTA <italic>t</italic>_1/2 or uncorrected GFR obtained from the EDTA data. The remaining 10% (2/21) of centres exclusively used uncorrected GFR as the renal parameter, in conjunction with patient body weight, to estimate the carboplatin dose to be administered.</p>", "<title>Glomerular filtration rate measurement and carboplatin dosing</title>", "<p>Implementation of the BNMS guidelines, i.e., using data obtained from blood samples taken at 2, 3 and 4 h and application of the BM correction, resulted in a median EDTA <italic>t</italic>_1/2 value of 93.5 min (range: 47.4–481.8 min) and a median carboplatin dose (calculated to achieve an AUC of 5.0 mg ml⁻¹ min⁻¹) of 416 mg (range: 72–1207 mg) in this patient population. On the basis of the above findings from the questionnaire survey, implementation of these guidelines would result in the omission of data obtained from early blood samples taken within 2 h of radioisotope administration in six centres. In addition, 12 centres would shift from either applying no correction, or using the Chantler correction, to implement the BM correction. The consequences of these changes on renal function-based carboplatin dosing are summarised in ##TAB##1##Tables 2## and ##TAB##2##3##.</p>", "<p>These data show the percentage of children who would have a &gt;10% change in carboplatin dose following the implementation of the BNMS recommendations, including both increases and decreases to the calculated carboplatin dose. Following omission of the early blood sample in those centres where this was taken, the ⁵¹Cr-EDTA <italic>t</italic>_1/2 increased in the majority of patients, with a resultant decrease in dose of carboplatin calculated on the basis of ⁵¹Cr-EDTA <italic>t</italic>_1/2 and weight as expected. A change in carboplatin dose of &gt;10% would be seen in 23–52% of patients treated at these centres, depending on the existing sampling times in place. The vast majority of patients would have a decrease in the dose of carboplatin administered following the omission of the early blood sample, as illustrated in ##FIG##0##Figure 1A##. In a small group of patients (3%), the dose difference could be as high as 25%.</p>", "<p>With regards to the slope–intercept correction method used, a shift from the Chantler correction to the BM correction would result in a &gt;10% change in carboplatin dose in at least 15% of children (##TAB##2##Table 3##). In a small group of patients (2%), the change in dose could be as high as 25%. In centres applying no correction factor, introduction of the BM correction would impact on a greater proportion of patients, with a &gt;10% change in carboplatin dose observed in 72–85% of children; the change in dose could be as high as 25% in 6–14% of patients. Although the proportion of patients who would experience a significant carboplatin dose change is substantial, the actual mean percentage dose change would be relatively small, ranging from 14–17% across different centres.</p>", "<p>Introduction of the BM correction (instead of the Chantler or no correction) would be expected to result in a higher estimate of AUC and the clearance of ⁵¹Cr-EDTA would decrease. Consequently, this would lead to a decrease in the dose of carboplatin calculated using the Newell formula on the basis of uncorrected GFR and weight. An example of the distribution of carboplatin dose adjustment when the Chantler correction is substituted by the BM correction is illustrated in ##FIG##0##Figure 1B##.</p>" ]

[ "<title>Discussion</title>", "<p>There are numerous techniques described in the literature for estimating renal function or GFR in children and adults (##UREF##1##Price and Finney, 2000##; ##REF##11207037##Wright et al, 2001##; ##REF##12432431##Leger et al, 2002##; ##REF##14710207##Cole et al, 2004##; ##REF##17013606##Brandt et al, 2006##). Although creatinine clearance is a reliable marker in day-to-day practice in adults, its use in paediatrics has been questioned and there are situations where a more accurate measure of GFR is preferred (##REF##10630919##Finney et al, 2000##; ##REF##16012866##Holweger et al, 2005##). This information may be particularly important in clinical situations where potentially nephrotoxic drugs are administered, in addition to allowing for accurate renal function-based dosing of anticancer drugs such as carboplatin (##REF##2681557##Calvert et al, 1989##; ##REF##16012866##Holweger et al, 2005##). It has been well established that the use of radioisotope tracers, such as ⁵¹Cr-EDTA and ^99mTc-DTPA, are reliable methods for the accurate estimation of GFR (##UREF##2##Rehling et al, 1984##; ##REF##3899658##Fawdry et al, 1985##). Techniques of limited sampling restricted to the slow exponential phase of radioisotope clearance provide a reasonable compromise to avoid the complexity of measuring GFR using the biexponential method where multiple blood samples need to be taken at earlier time points. This is particularly important in paediatric practice where blood sampling can be a traumatic experience to a child and where blood volumes need to be minimised. However, this approach necessitates the use of an appropriate correction factor to offset the overestimation of GFR by the slope–intercept technique.</p>", "<p>An audit carried out by the BNMS previously highlighted the problems of widespread variation in hospital practices for measuring GFR and guidelines have been published in an attempt to standardise the methodologies used. In view of the BNMS recommendations, it was prudent to investigate the potential effects of changes in methodologies for determining GFR on carboplatin dosing in children with cancer. Therefore, this study aimed to investigate current practices for determining GFR in paediatric oncology centres across the United Kingdom and to use the information obtained to investigate the potential impact of the BNMS recommendations on renal function-based dosing of carboplatin in patients.</p>", "<p>Information obtained from all 21 CCLG centres responsible for the treatment of children with cancer in the United Kingdom showed that, to comply with the BNMS recommendations, six centres would be required to stop taking early blood samples (less than 2 h after radioisotope injection) and 12 centres would need to implement the use of the BM slope–intercept correction method. These factors are important when we consider that use of a sample taken 1 h after radioisotope injection, potentially falling within the rapid early phase of elimination, in combination with the slope–intercept method, could lead to inaccurate GFR determinations and potential overdosing of carboplatin. Using raw ⁵¹Cr-EDTA data from 337 GFR tests carried out in children with cancer, in conjunction with the findings of the questionnaire survey, it was possible to study the potential impact of the new recommendations on carboplatin dosing.</p>", "<p>On performing the analyses of various possible combinations of sampling times, it was found that changing from the Chantler to the BM correction factor altered the dose of carboplatin estimated by more than 10% in at least 15% of the children studied. In a small group of patients (2%), the dose difference was as high as 25%. However, the impact was even more marked when changing from no correction to the BM correction, with a &gt;10% change in carboplatin dose observed in 72–85% of children. Omitting the 1-h blood sample after injection of the radioisotope increased the estimate of ⁵¹Cr-EDTA half-life and consequently decreased the dose of carboplatin estimated using ⁵¹Cr-EDTA half-life and weight. The alteration in the estimated dose was &gt;10% in 23–52% of the children studied and in 3% of patients, the dose would decrease by &gt;25%.</p>", "<p>Whether or not the magnitude of these changes in carboplatin dosage would impact on the ultimate clinical outcome has not been prospectively evaluated and the heterogeneity of the cases in our study population (from whom GFR study data were obtained) would make such an analysis difficult. Also, the 10% change of carboplatin dose used for our analysis represents an arbitrary cutoff point. Although subjective, in view of the narrow therapeutic windows for chemotherapy agents, most clinicians would consider a &gt;10% dose change for any chemotherapeutic drug as being significant. As the predominant shift is towards a decrease in carboplatin dose, implementation of the BNMS recommendations should not raise any concerns regarding potential increases in drug toxicity but an impact on efficacy cannot be ruled out.</p>", "<p>These findings highlight the potential problem that patients with similar physical characteristics and renal function could receive different carboplatin doses at different centres on the basis of the methodology used to estimate GFR or ⁵¹Cr-EDTA half-life. Indeed, it is not unlikely for such an occurrence to be observed in patients being treated on the same clinical protocol. As many paediatric oncology patients in the United Kingdom are recruited to national and international trials, it is logical that the methodology of estimating GFR and carboplatin dose estimation should be standardized, so that the outcome data for these trials are not confounded by this factor. This task could arguably be made more straightforward if the Newell formula was universally used for the determination of renal function-based carboplatin dosing in children, as the ⁵¹Cr-EDTA half-life on which it is based is only dependent on the radioisotope sampling times and does not depend on the choice of correction factor. Implementation of a change in the type of slope–intercept correction method used for determining GFR, in addition to the changes in sampling times, would impact on the majority of CCLG centres (16 of 21) and there may be some reluctance to implement such changes. Despite the publication of the BNMS recommendations in 2004, there remains substantial variation in practice in paediatric oncology centres across the United Kindom. Indeed, in a follow-up survey carried out as part of this study in February 2008, only three centres reported a change to the previously used method of GFR determination. While one centre had adopted the BNMS recommendations during this time interval, changes in the additional two centres related to the inclusion of a third ⁵¹Cr-EDTA sampling time point and a shift from the use of ⁵¹Cr-EDTA to iohexol as the radioisotopic tracer, respectively. There has also been confusion between the use of ‘uncorrected GFR’, the parameter used in the modified Calvert formula for estimating carboplatin dose, and body surface area or weight ‘standardized’ or ‘normalized’ GFR with reports in the literature where serious dosing errors have occurred when these values have been misused (##REF##14528109##Liem et al, 2003##). The standardised use of ⁵¹Cr-EDTA half-life as the renal parameter to estimate carboplatin dose would prevent such problems occurring.</p>", "<p>In summary, our findings suggest that steps should be taken to implement the new BNMS guidelines and standardise methodologies of GFR estimation and renal function-based carboplatin dose calculation in all centres. Ideally, we recommend the determination of carboplatin dose in children with cancer on the basis of ⁵¹Cr-EDTA half-life and weight wherever possible.</p>" ]

[]

[ "<p>Renal function-based carboplatin dosing is used routinely in paediatric oncology clinical practice. It is important that accurate assessments of renal function are carried out consistently across clinical centres, a view supported by recently published British Nuclear Medicine Society (BNMS) guidelines for measuring glomerular filtration rate (GFR). These guidelines recommend the use of a radioisotope method for GFR determination, with between two and five blood samples taken starting 2 h after radioisotope injection and application of the Brochner-Mortensen (BM) correction factor. To study the likely impact of these guidelines, we have investigated current practices of measuring GFR in all 21 Children's Cancer and Leukaemia Group (CCLG) paediatric oncology centres in the United Kingdom. This information was used to evaluate the potential impact on renal function-based carboplatin dosing using raw ⁵¹Cr-EDTA clearance data from 337 GFR tests carried out in children with cancer. A questionnaire survey revealed that between two and four samples were taken after isotope administration, with BM and Chantler corrections used in 38% (8/21) and 28% (6/21) of centres, respectively. A change from Chantler to BM correction, based on the BNMS guidelines, would result in a &gt;10% decrease in carboplatin dose in at least 15% of patients and a &gt;25% decrease in 2% of patients. A greater proportion of patients would have an alteration in carboplatin dose when centres not using any correction factor implement the BM correction. The increase in estimated ⁵¹Cr-EDTA half-life observed by omitting the 1 h sample decreases carboplatin dose by &gt;10% in 23–52% of patients and by &gt;25% in 3% of patients. This study highlights current variations in renal function measurement between clinical centres and the potential impact on carboplatin dosing. A standard methodology for estimating GFR should be followed to achieve uniform dosing in children with cancer.</p>" ]

[ "<p>Carboplatin dosing based on renal function has been shown to result in more reproducible and reliable drug exposures than dosing based on surface area in children (##REF##11054434##Thomas et al, 2000##). This approach not only minimises the risk of underdosing and hence inadequate treatment, but also reduces overdosing, which may be associated with the risk of increased toxicity. As renal function-based carboplatin dosing is now used routinely in clinical practice in many multicentre protocols, it is important that accurate assessments of renal function in children are carried out consistently across paediatric oncology clinical centres.</p>", "<p>Glomerular filtration rate (GFR) estimation is widely used as a standard measure of renal function in both children and adults. There are many techniques described in the literature to estimate GFR, using clearance of either exogenous or endogenous markers (##REF##951142##Schwartz et al, 1976##; ##UREF##1##Price and Finney, 2000##; ##REF##11207037##Wright et al, 2001##; ##REF##12432431##Leger et al, 2002##; ##REF##14710207##Cole et al, 2004##; ##REF##17013606##Brandt et al, 2006##). Use of exogenous markers for estimating GFR involves administration of the marker as either a bolus injection or an infusion and measurement of concentrations in serial blood and/or urine samples over a period of time. Radioisotopically labelled inert substances have been in widespread use and the most common tracers that have been used are iothalmate (¹³¹I) (##UREF##3##Sigman et al, 1966##), iodothalmate, chromium ethylenediamine tetracetic acid (⁵¹Cr-EDTA) and technetium diethylenetriamine pentacetic acid (⁹⁹Tc-DTPA) (##REF##4625784##Chantler and Barratt, 1972##; ##REF##822911##Hilson et al, 1976##; ##UREF##1##Price and Finney, 2000##). The radioisotope tracers ⁵¹Cr-EDTA and ⁹⁹Tc-DTPA have found widespread use in paediatric oncology and they are now accepted as a simple, safe and accurate way of measuring GFR. In the United Kingdom, it is standard practice in the majority of Children's Cancer and Leukaemia Group (CCLG) centres to estimate GFR using ⁵¹Cr-EDTA, whereas in the United States, ⁹⁹Tc-DTPA is the radioisotope tracer of choice (##REF##15266169##Fleming et al, 2004##).</p>", "<p>Estimation of GFR from the clearance of radiolabelled tracer is most accurate if multiple blood samples are taken after administration as the decrease in plasma concentration of the tracer is often biexponential, with a rapid early phase and a later slower phase (##REF##1547805##Piciotto et al, 1992##). However, for routine clinical practice, sampling is usually restricted to the second phase of the clearance, with the slope–intercept method commonly used for the determination of GFR. As the more rapid early phase of the clearance is not accounted for when using this method, it is necessary to correct for systematic errors introduced into the derived GFR values. The most popular methods used to correct for these errors are known as the Chantler and Brochner-Mortensen (BM) corrections (##REF##4629674##Brochner-Mortensen, 1972##; ##REF##4625784##Chantler and Barratt, 1972##). Whereas the former uses a constant multiplicative correction factor to adjust the GFR values obtained, the latter uses a quadratic correction and is dependent on the individual's body surface area.</p>", "<p>In 2004, the British Nuclear Medicine Society (BNMS) published guidelines to assist nuclear medicine specialists in performing GFR studies and interpreting and reporting the results obtained (##REF##15266169##Fleming et al, 2004##). These guidelines recommended that all centres should take between 2 and 5 samples, starting 2 h after injection of radioisotope, and that the GFR should be calculated using the slope–intercept method with the BM correction factor applied.</p>", "<p>As carboplatin dosing represents the main practical outcome of estimating GFR in paediatric oncology, it would seem sensible to consider the potential effect that a change in methodology for estimating GFR may have on carboplatin dosing in children with cancer. Therefore, to study the likely impact of the BNMS recommendations, a questionnaire survey concerning the current practices of estimating GFR in children was sent to all CCLG paediatric oncology centres in the United Kingdom. This information was then used to evaluate the potential impact of these changes on renal function-based carboplatin dosing, using ⁵¹Cr-EDTA clearance data from over 300 GFR tests carried out in a paediatric patient population.</p>" ]

[ "<p>We thank the research nurses, clinicians and all other staff who participated in the study at the following CCLG centres: Royal Victoria Infirmary, Newcastle upon Tyne; Great Ormond Street Hospital, London; Manchester Children's Hospital; St James's Hospital, Leeds; Royal Liverpool Children's Hospital; Royal Marsden Hospital, Surrey; Addenbrooke's Hospital, Cambridge; Bristol Royal Hospital for Children; Southampton General Hospital; Birmingham Children's Hospital; Children's Hospital for Wales, Cardiff; Royal Aberdeen Children's Hospital; Royal Hospital for Sick Children, Edinburgh; Leicester Royal Infirmary; Queen's Medical Centre, Nottingham; Royal Hospital for Sick Children, Glasgow; University College London Hospital; Our Lady's Hospital for Sick Children, Dublin; Sheffield Children's Hospital; John Radcliffe Hospital, Oxford; Royal Hospital for Sick Children, Belfast. This study was supported by a grant from Cancer Research UK.</p>" ]

[ "<fig id=\"fig1\"><label>Figure 1</label><caption><p>Distribution of percentage change in carboplatin dose following a change in sampling times from 1, 2, 3 and 4 h to 2, 3 and 4 h (<bold>A</bold>), and following a shift from the Chantler correction to the Brochner-Mortensen correction with sampling times at 2, 3 and 4 h (<bold>B</bold>). A decrease in dose is seen for the majority of patients in both cases.</p></caption></fig>" ]

[ "<table-wrap id=\"tbl1\"><label>Table 1</label><caption><title>Results of the questionnaire survey showing the methodology of estimation of GFR in the UK CCLG centres</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"left\"/><col align=\"left\"/><col align=\"center\"/><col align=\"left\"/><col align=\"left\"/><col align=\"left\"/><col align=\"left\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Centre</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Radioisotope</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Route of injection</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>No of samples</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Early sampling (&lt;2 h)</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>GFR estimation</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Methodology</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Correction factor</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 1</td><td align=\"left\" valign=\"top\" charoff=\"50\">⁵¹Cr-EDTA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Peripheral</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"left\" valign=\"top\" charoff=\"50\">Yes</td><td align=\"left\" valign=\"top\" charoff=\"50\">Med Physics</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 2</td><td align=\"left\" valign=\"top\" charoff=\"50\">⁵¹Cr-EDTA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Central</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">Biochemistry</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">Chantler</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 3</td><td align=\"left\" valign=\"top\" charoff=\"50\">⁵¹Cr-EDTA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Central</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">Biochemistry</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">Chantler</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 4</td><td align=\"left\" valign=\"top\" charoff=\"50\">⁵¹Cr-EDTA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Central</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">Med Physics</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">Chantler</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 5</td><td align=\"left\" valign=\"top\" charoff=\"50\">⁵¹Cr-EDTA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Peripheral</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">Med Physics</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 6</td><td align=\"left\" valign=\"top\" charoff=\"50\">⁵¹Cr-EDTA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Both</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td><td align=\"left\" valign=\"top\" charoff=\"50\">Yes</td><td align=\"left\" valign=\"top\" charoff=\"50\">Med Physics</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 7</td><td align=\"left\" valign=\"top\" charoff=\"50\">⁵¹Cr-EDTA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Both</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">Med Physics</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 8</td><td align=\"left\" valign=\"top\" charoff=\"50\">⁵¹Cr-EDTA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Both</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">Med Physics</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">Modification</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 9</td><td align=\"left\" valign=\"top\" charoff=\"50\">^99mTc-DTPA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Both</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">Med Physics</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">10</td><td align=\"left\" valign=\"top\" charoff=\"50\">⁵¹Cr-EDTA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Both</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">Med Physics</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">Chantler</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">11</td><td align=\"left\" valign=\"top\" charoff=\"50\">⁵¹Cr-EDTA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Peripheral</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">Med Physics</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">12</td><td align=\"left\" valign=\"top\" charoff=\"50\">⁵¹Cr-EDTA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Peripheral</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">Biochemistry</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">13</td><td align=\"left\" valign=\"top\" charoff=\"50\">⁵¹Cr-EDTA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Both</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">Biochemistry</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">Chantler</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">14</td><td align=\"left\" valign=\"top\" charoff=\"50\">⁵¹Cr-EDTA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Both</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"left\" valign=\"top\" charoff=\"50\">Yes</td><td align=\"left\" valign=\"top\" charoff=\"50\">Med Physics</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">15</td><td align=\"left\" valign=\"top\" charoff=\"50\">⁵¹Cr-EDTA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Peripheral</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">Med Physics</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">16</td><td align=\"left\" valign=\"top\" charoff=\"50\">⁵¹Cr-EDTA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Both</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">Med Physics</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">17</td><td align=\"left\" valign=\"top\" charoff=\"50\">^99mTc-DTPA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Both</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">Med Physics</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">18</td><td align=\"left\" valign=\"top\" charoff=\"50\">⁵¹Cr-EDTA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Both</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"left\" valign=\"top\" charoff=\"50\">Yes</td><td align=\"left\" valign=\"top\" charoff=\"50\">Med Physics</td><td align=\"left\" valign=\"top\" charoff=\"50\">In-house method</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">19</td><td align=\"left\" valign=\"top\" charoff=\"50\">⁵¹Cr-EDTA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Peripheral</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"left\" valign=\"top\" charoff=\"50\">Yes</td><td align=\"left\" valign=\"top\" charoff=\"50\">Biochemistry</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">20</td><td align=\"left\" valign=\"top\" charoff=\"50\">⁵¹Cr-EDTA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Central</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td><td align=\"left\" valign=\"top\" charoff=\"50\">Yes</td><td align=\"left\" valign=\"top\" charoff=\"50\">Med Physics</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">Chantler</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">21</td><td align=\"left\" valign=\"top\" charoff=\"50\">⁵¹Cr-EDTA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Peripheral</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">Med Physics</td><td align=\"left\" valign=\"top\" charoff=\"50\">Slope–intercept</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl2\"><label>Table 2</label><caption><title>Predicted impact on renal function-based carboplatin dosing of a change from existing sampling times to the new BNMS guidelines</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/><col align=\"left\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Existing sampling times (h)</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>New sampling times (h)</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Number of centres</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Percentage of children who have a &gt;10% change in carboplatin dose</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">1, 2, 3 and 4</td><td align=\"center\" valign=\"top\" charoff=\"50\">2,3 and 4</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"left\" valign=\"top\" charoff=\"50\">23 (21% increase/2% decrease)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">1, 2 and 3</td><td align=\"center\" valign=\"top\" charoff=\"50\">2,3 and 4</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"left\" valign=\"top\" charoff=\"50\">52 (47% increase/5% decrease)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">2 and 4</td><td align=\"center\" valign=\"top\" charoff=\"50\">2,3 and 4</td><td align=\"center\" valign=\"top\" charoff=\"50\">5</td><td align=\"left\" valign=\"top\" charoff=\"50\">No change</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">2, 3 and 4</td><td align=\"center\" valign=\"top\" charoff=\"50\">2,3 and 4</td><td align=\"center\" valign=\"top\" charoff=\"50\">8</td><td align=\"left\" valign=\"top\" charoff=\"50\">No change</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">1, 3 and 4</td><td align=\"center\" valign=\"top\" charoff=\"50\">2,3 and 4</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"left\" valign=\"top\" charoff=\"50\">27 (25% increase/2% decrease)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">2 and 3</td><td align=\"center\" valign=\"top\" charoff=\"50\">2,3 and 4</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"left\" valign=\"top\" charoff=\"50\">33 (23% increase/10% decrease)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Others</td><td align=\"center\" valign=\"top\" charoff=\"50\">—</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl3\"><label>Table 3</label><caption><title>Predicted impact on renal function-based carboplatin dosing of a change from current practices to the new sampling times and correction factor recommended by BNMS</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"left\"/><col align=\"center\"/><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Existing sampling times (h)</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Existing correction factor</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>New sampling times (h)</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>New correction factor</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Number of centres</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Percentage of children with &gt;10% carboplatin dose change</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">1, 2, 3 and 4</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td><td align=\"center\" valign=\"top\" charoff=\"50\">2,3 and 4</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">72</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">1, 2 and 3</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td><td align=\"center\" valign=\"top\" charoff=\"50\">2,3 and 4</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\">80</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">2 and 4</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td><td align=\"center\" valign=\"top\" charoff=\"50\">2,3 and 4</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\">85</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">2, 3 and 4</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td><td align=\"center\" valign=\"top\" charoff=\"50\">2,3 and 4</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">85</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">2, 3 and 4</td><td align=\"left\" valign=\"top\" charoff=\"50\">Chantler</td><td align=\"center\" valign=\"top\" charoff=\"50\">2,3 and 4</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\">15</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">2 and 4</td><td align=\"left\" valign=\"top\" charoff=\"50\">Chantler</td><td align=\"center\" valign=\"top\" charoff=\"50\">2,3 and 4</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\">15</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">1, 3 and 4</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td><td align=\"center\" valign=\"top\" charoff=\"50\">2,3 and 4</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">2 and 3</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td><td align=\"center\" valign=\"top\" charoff=\"50\">2,3 and 4</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">2, 3 and 4</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td><td align=\"center\" valign=\"top\" charoff=\"50\">2,3 and 4</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td><td align=\"center\" valign=\"top\" charoff=\"50\">5</td><td align=\"center\" valign=\"top\" charoff=\"50\">No change</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Others</td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td><td align=\"center\" valign=\"top\" charoff=\"50\">2,3 and 4</td><td align=\"left\" valign=\"top\" charoff=\"50\">BM</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\">Not applicable</td></tr></tbody></table></table-wrap>" ]

[ "<disp-formula id=\"equ1\"></disp-formula>", "<disp-formula id=\"equ2\"></disp-formula>", "<disp-formula id=\"equ3\"></disp-formula>", "<disp-formula id=\"equ4\"></disp-formula>", "<disp-formula id=\"equ5\"></disp-formula>" ]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><fn id=\"t1-fn1\"><p>BM=Brochner-Mortensen; CCLG=Children's Cancer and Leukaemia Group; GFR=glomerular filtration rate.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"t2-fn1\"><p>BNMS=British Nuclear Medicine Society.</p></fn><fn id=\"t2-fn2\"><p>Dosing on the basis of ⁵¹Cr-EDTA <italic>t</italic>_1/2 and weight.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"t3-fn1\"><p>BM=Brochner-Mortensen; BNMS=British Nuclear Medicine Society.</p></fn><fn id=\"t3-fn2\"><p>Dosing based on uncorrected GFR and weight.</p></fn></table-wrap-foot>" ]

[ "<graphic xlink:href=\"6604612e1.jpg\"/>", "<graphic xlink:href=\"6604612e2.jpg\"/>", "<graphic xlink:href=\"6604612e3.jpg\"/>", "<graphic xlink:href=\"6604612e4.jpg\"/>", "<graphic xlink:href=\"6604612e5.jpg\"/>", "<graphic xlink:href=\"6604612f1\"/>" ]

[]

{ "acronym": [], "definition": [] }

[]

[ "<title>PATIENTS AND METHODS</title>", "<title>Data</title>", "<p>Two international phase III British Biotech studies (BB128, ##REF##11481349##Bramhall et al, 2001##; BB193, ##REF##12107836##Bramhall et al, 2002##) randomised 414 and 239 patients with advanced pancreatic cancer, respectively: BB128 randomised patients between marimistat and gemcitabine; BB193 randomised patients between marimistat with gemcitabine and gemcitabine alone. The studies had similar eligibility criteria: histologically or cytologically unresectable pancreatic cancer, within 8 weeks of diagnosis or disease recurrence and Karnofsky performance status of ⩾50% (BB128) or ⩾60% (BB193). Previous therapy for metastatic or locally advanced disease was an exclusion criterion. The primary outcome measure in both studies was survival time calculated from the date of randomisation to the date of death from any cause. Randomisation was stratified by cancer stage (stage I/II, III or IV), Karnofsky performance status (50–70%, 80–100%), sex and study centre. The first stage of data reduction was considering only factors that were clinically relevant and available within an NHS outpatient clinic. Eighteen baseline clinical, histological, biochemical and demographic variables (including trial and randomised treatment group) were considered appropriate for analysis as possible prognostic factors (##TAB##1##Table 2##).</p>", "<title>Statistical analysis</title>", "<p>We followed a strategy aimed at maximising model performance and avoiding poorly fitted and overfitted regression models in the development (##REF##8668867##Harrell et al, 1996##) and reporting (##REF##16106245##McShane et al, 2005##) of multivariable prognostic models. Initial analysis was based on standard methodology comparing Kaplan–Meier survival estimates using the log-rank test and estimating univariate hazard ratios for levels of each factor. The hazard of death was assessed in the multivariable setting using Cox proportional hazards regression modelling with variable reduction by backward elimination. The proportional hazards assumption was investigated for each covariate using log cumulative hazard (##UREF##0##Collett, 1994##) and martingale residual plots, and incorporating a time-dependent covariate (<italic>X</italic>=factor(LN (survival)−LN (mean survival))) and did not indicate any significant violation.</p>", "<p>Ten of the 18 possible prognostic factors were collected as continuous measurements. Continuous data were investigated by assessing three different assumptions of the underlying relationship between survival and predictor, they are: (a) a linear relationship between the predictor and log hazard, (b) a step functional relationship using dichotomised covariates (laboratory measures based on central laboratory reference ranges) and (c) a nonlinear relationships based on either a simple log or more complex nonlinear FP transformation (##UREF##4##Royston and Altman, 1994##). For the third model, the functional form of each variable was assessed univariately comparing the Akaike's Information Criterion (AIC) (##UREF##0##Collett, 1994##) of a model based on the simple log transformation with the AIC of a model based on the best fitting FP transformation. First- and second-degree FP transformations (##UREF##2##Meier-Hirmer et al, 2003##) were considered using a selection level of 0.05 for input of variables based on power values of the polynomial ranging (−2, −1, −0.5, 0 (log), 0.5, 1, 2, 3). The best FP for each predictor was selected if it resulted in a significantly better fit (significantly smaller AIC) than the log transformation. The most appropriate (log or FP) transformation, if any, was applied to each variable and all variables were considered multivariately using Cox proportional hazards regression based on a backward selection method using a nominal significance level of 0.05 for elimination and including trial, sex, cancer stage (stratification factors at randomisation) and randomised treatment in each model.</p>", "<p>The majority of variables had ⩽5% missing values (##TAB##1##Table 2##). Tumour stage, CA19-9 and WBC had 5–10% missing values, and lymph node status was missing for 24% of patients. Metastases or lymph node status was considered in the analysis as dummy variables using ‘negative’ as a reference level. Primary analysis was based on complete cases and a secondary analysis used multiple imputation to investigate the possible influence of variables with larger amounts of missing data (##UREF##5##Rubin, 1987##; ##UREF##6##Schafer, 1997##) and provided valid inferential alternative results.</p>", "<p>Model fit was assessed comparing AIC statistics, deviance residuals and Kaplan–Meier survival statistics for four predictive groups. The four predictive groups were based on quartiles of linear predictor scores, assessed comparing median survival estimates and hazard ratios. The bootstrap resampling approach described by ##REF##8668867##Harrell et al (1996)## was applied to assess the extent of overfitting in the final model, using 200 bootstrap resamples. This approach repeats the model selection methods used in the original model development in a series of bootstrapped resamples, freezing the derived model and applying to the original sample. Model optimism (overfitting) is described by the difference in the rank correlation coefficient relating predicted and observed survival times between the model derived in the bootstrap resample and that from the frozen model applied to the original sample averaged over 200 resamples. This provides an honest estimate of internal validity penalised for overfitting (##REF##8668867##Harrell et al, 1996##).</p>", "<p>Analyses were carried out using SAS and R using a two-sided significance level of 0.05 throughout.</p>" ]

[ "<title>RESULTS</title>", "<title>Patient characteristics</title>", "<p>A total of 653 patients were randomised. The eighteen clinically appropriate factors for analysis are presented in ##TAB##1##Table 2## and appear balanced across the two studies. On average, patients in the two trials were randomised 20 and 15 days after diagnosis and started treatment the day following randomisation. The average age of patients was 63 years (range 29–89), 368 (56%) were male, 439 (68%) had cancer stage IV disease, 436 (67%) presenting with metastases and 251 (39%) had lymph node involvement.</p>", "<title>Survival</title>", "<p>The majority of patients (612, 94%) had died by the time of analysis with a median follow-up time of 21 months for the 41 patients still alive (##TAB##1##Table 2##, ##FIG##0##Figure 1##). The median survival estimate for the group is 4.7 months (95% CI: 4.2, 5.1) with 12-month survival estimate of 17% (##FIG##0##Figure 1##). Hazard functions estimated for 1-monthly time intervals to 18 months from trial entry were similar for both trials and reasonably constant over time. No significant survival benefit for marimastat was identified in the BB128 trial (<italic>P</italic>=0.19) when compared with gemcitabine (##REF##11481349##Bramhall et al, 2001##). Similarly, no significant survival benefit was seen for a combination of gemcitabine and marimistat when compared with gemcitabine alone in the BB193 trial (<italic>P</italic>=0.95) (##REF##12107836##Bramhall et al, 2002##).</p>", "<title>Univariate analyses</title>", "<p>Log-rank analyses (##TAB##2##Table 3##) indicated that potentially important factors were age (split at median, <italic>P</italic>=0.036), nodal status (<italic>P</italic>=0.035), cancer stage (I/II <italic>vs</italic> III/IV), metastases (both <italic>P</italic>&lt;0.001) and laboratory measures (split as normal/abnormal according to laboratory reference ranges as per current clinical practise) as AST, alkaline phosphatase, albumin, LDH, WBC (all <italic>P</italic>&lt;0.001), bilirubin (<italic>P</italic>=0.002), CA19-9 (<italic>P</italic>=0.005) and haemoglobin (<italic>P</italic>=0.009). Trial, treatment (gemcitabine <italic>vs</italic> marimistat), race (white <italic>vs</italic> not-white), sex, tumour stage (T0, 1, 2 <italic>vs</italic> T3, 4) and BUN were not significantly related to survival.</p>", "<title>Multivariate analyses</title>", "<p>Three Cox proportional hazards regression models were developed (##TAB##3##Table 4##) using 556 patients (520 deaths) with complete data (excluding patients with missing data) based on the assumption of (a) a linear relationship between continuous covariates and log hazard, (b) a step functional dichotomisation of continuous covariates and (c) a nonlinear transformation of continuous covariates. All three models included trial, sex, cancer stage (stratification factors at randomisation) and randomised treatment group.</p>", "<p>The ‘linear’ model (##TAB##3##Table 4##, Model 1) identified five highly significant prognostic factors, namely albumin, alkaline phosphatase, LDH, WBC and metastases. The ‘categorical’ model (##TAB##3##Table 4##, Model 2) identified six highly significant prognostic factors, namely LDH, albumin, metastases, WBC, CA19-9 and bilirubin. Univariate analysis of the 10 continuous variables identified that nonlinear transformations were appropriate for 3 variables in their relationship with survival: bilirubin and LDH both as log transformations and CA19-9 as a second-degree FP transformation (CA19-9^0.5+(CA19-9^0.5 × log(CA19-9))). The seven remaining continuous covariates were analysed assuming a linear relationship with log hazard, as in Model 1. The ‘transformed’ model (##TAB##3##Table 4##, Model 3) identified eight prognostic factors. Five factors were highly significant with <italic>P</italic>&lt;0.01, namely albumin, CA19-9, LDH, alkaline phosphatase and WBC with AST, BUN and metastases being more borderline in the model (<italic>P</italic>=0.023, 0.026 and 0.047, respectively).</p>", "<p>Nonlinear transformations were appropriate for two variables, LDH and CA19-9, and the estimated log hazard ratio functions are shown graphically in ##FIG##1##Figures 2## and ##FIG##2##3##. The second-degree FP function for CA19-9 estimates increasing risk up to an approximate CA19-9 value of 14 000 and then decreases with increasing CA19-9. The log function for LDH estimates increasing risk for increasing values of LDH.</p>", "<title>Model comparison</title>", "<p>In all three models, albumin, LDH and WBC were highly statistically significant and influential prognostic factors. Metastases were also an important variable but its parameter estimate and overall significance were reduced in the ‘transformed’ model when continuous covariates were included in a more appropriate format. In both the ‘linear’ and ‘transformed’ models, alkaline phosphatase was also a highly significant and influential prognostic factor. CA19-9 was also a highly significant and influential prognostic factor in both the ‘categorical’ and ‘transformed’ models. This variation is largely explained by the nonlinear relation of CA19-9 to survival (##FIG##1##Figure 2##), which could explain why it was considered important when dichotomised but not when included as linear. When considered as a transformed second-degree FP, its significance was much greater. Bilirubin was selected as a highly significant factor in the ‘categorical’ model but was not included in either the ‘linear’ or ‘transformed’ models. AST and BUN were only selected as prognostic in the ‘transformed’ model.</p>", "<title>Model performance</title>", "<p>The AIC statistic was smallest for the ‘transformed’ model (##TAB##3##Table 4##, Model 3), indicating a better fit to the data. Deviance residuals for this model were plotted against the linear predictor and were randomly scattered centred around a residual value of zero ranging between −3.86 and 3.33, which suggests the data have not been mis-modelled.</p>", "<p>Patients were split into four groups based on quartiles of the distribution of linear predictor scores from the ‘transformed’ model. Kaplan–Meier survival estimates (##FIG##3##Figure 4##) show four distinct predictive groups with descending median survival estimates of 9.1 (95% CI: 7.4, 10.9), 7.0 (95% CI: 5.9, 8.3), 4.0 (95% CI: 3.4, 4.9) and 2.0 (95% CI: 1.6, 2.4) months. The hazard ratios for groups 2, 3 and 4 using predictive group 1 as the baseline were 1.35 (95% CI: 1.09, 1.66), 2.08 (95% CI: 1.64, 2.64) and 4.21 (95% CI: 3.11, 5.68), respectively.</p>", "<p>When assessing model validity, the <italic>R</italic>² measure of model fit was estimated as 0.30. The bootstrap resampled estimate of <italic>R</italic>² of 0.26 described model optimism (overfitting) under 5% and gives an improved estimate of model accuracy.</p>", "<p>Multiple imputation allowed all 653 patients to be included in the modelling process and confirmed all the variables included in the ‘transformed’ model with increased significance for metastases (<italic>P</italic>=0.001), and the model also included nodal status (<italic>P</italic>=0.016) that had been excluded from all models prior to imputation, suggesting a strong link with other variables already in the model.</p>" ]

[ "<title>DISCUSSION</title>", "<p>Large, prospective, phase III randomised controlled trials aim to provide robust statistical evidence for new treatment combinations. Stratification is important to control for known important variability in the data. Generally, patients with pancreatic cancer are not clinically separated into prognostic groups, with the exception of surgical status, before consideration for treatment. This study investigated potentially important baseline prognostic factors for survival as possible stratification variables for randomisation and analysis. Data from 653 patients included in two international randomised controlled trials in advanced pancreatic cancer (##REF##11481349##Bramhall et al, 2001##, ##REF##12107836##2002##) were analysed investigating multiple clinical, histological, biochemical and demographic variables in the form of both binary and continuous measurements. Valid statistical analyses are necessary to make best use of the data and optimise clinical results. As such, a multivariable approach was used to account for the functional form of the relationship between continuous prognostic variable factors and survival. Misspecification of functional form may lead to inappropriate conclusions but has not been previously investigated in pancreatic cancer studies. Continuous variables are often simplified by assuming a linear relationship between predictor and log hazard, that is the log risk increases or decreases linearly as the value of the factor increases, which may not be appropriate. Dichotomisation of continuous data is common but is problematic and unnecessary. As the variability in outcome within groups is ignored by categorisation, the variability between groups may be significantly underestimated as patients close to the cut point are analysed as being very different rather than being very similar, resulting in a serious reduction of statistical power to detect relationships between predictors and outcome, residual confounding and serious bias (##REF##16675816##Altman and Royston, 2006##; ##REF##16217841##Royston et al, 2006a##). Regression using FPs of continuous covariates has been used in data from breast cancer (##REF##10206288##Sauerbrei et al, 1999##) and metastatic renal carcinoma (##REF##16736003##Royston et al, 2006b##) trials. Our study supported these in showing that this approach allowed important additional prognostic information to be extracted with less sophisticated approaches missed. FPs provide a flexible, parametric approach for modelling nonlinear relationships, making full use of the information available within each variable and as such can provide a clearer insight into the nature of the underlying relationship (##UREF##4##Royston and Altman, 1994##; ##REF##10066088##Altman and Lyman, 1998##).</p>", "<p>Pancreatic ductal adenocarcinoma is the fifth most common cause of death from cancer in the Western world (##REF##7881926##Bramhall et al, 1995##; ##REF##11602373##Parkin et al, 2001##; ##REF##12568441##Jemal et al, 2003##). It is particularly difficult to treat because of its remote location, late presentation and resistance to conventional chemotherapy. Long-term survival remains poor with a 5-year survival rate between 0.4 and 4% (##REF##7881926##Bramhall et al, 1995##; ##REF##12568441##Jemal et al, 2003##). Resection is associated with improved survival but this is only possible in approximately 10% of patients (##REF##10401733##Sener et al, 1999##; ##REF##15499648##Alexakis et al, 2004##). Although significant improvements in surgical outcome have been obtained with increasing specialisation (##REF##9361591##Neoptolemos et al, 1997##; ##REF##10455881##Birkmeyer et al, 1999##), further benefits are anticipated by identifying high-risk groups. A validated prognostic index would identify subgroups of patients for specific treatments and predict survival, but there is no tool in routine use. Also, many possible prognostic factors have been identified in advanced pancreatic cancer (##TAB##0##Table 1##), most derived from retrospective studies based on small numbers of patients resulting in analyses that may be underpowered. A total of 34 possible factors were identified from 36 studies (4 randomised controlled trials; ##REF##11350436##Johnson et al, 2001##; ##REF##12149301##Berlin et al, 2002##; ##REF##12181240##Ducreux et al, 2002##; ##REF##12118027##Maisey et al, 2002##) and 32 consecutive series (##REF##8449644##Friedman and van den Eeden, 1993##; ##REF##7846017##Yasue et al, 1994##; ##REF##7535184##Falconer et al, 1995##; ##REF##8572621##Lundin et al, 1995##; ##REF##8830333##Ishii et al, 1996##; ##REF##8805925##Rothenberg et al, 1996##; ##REF##12118565##Shibamoto et al, 1996##; ##REF##10235206##Cubiella et al, 1999##; ##REF##10189130##Storniolo et al, 1999##; ##REF##10737382##Halm et al, 2000##; ##REF##11034467##Terwee et al, 2000##; ##REF##11051691##Trigui et al, 2000##; ##UREF##7##Ueno et al, 2000##; ##REF##11169931##Ikeda et al, 2001##; ##UREF##3##Ridwelski et al, 2001##; ##REF##11801751##Tas et al, 2001##; ##REF##11745808##Tsuruta et al, 2001##; ##REF##12630768##Saad et al, 2002##; ##REF##12555292##Bachmann et al, 2003##; ##REF##12711291##Engelken et al, 2003##; ##UREF##1##Fujino et al, 2003##; ##REF##12706865##Karayiannakis et al, 2003##; ##REF##12909220##Micke et al, 2003##; ##REF##14527913##Ohigashi et al, 2003##; ##REF##12791502##Paillaud et al, 2003##; ##REF##14605463##Stemmler et al, 2003##; ##REF##14628094##Talar-Wojnarowska et al, 2003##; ##REF##14562009##Ziske et al, 2003##; ##REF##15613258##Gupta et al, 2004##; ##REF##14962722##Kuhlmann et al, 2004##; ##REF##15028948##Watanabe et al, 2004##; ##REF##15698733##Ni et al, 2005##) of which 15 had &lt;100 patients (##REF##7846017##Yasue et al, 1994##; ##REF##8830333##Ishii et al, 1996##; ##REF##8805925##Rothenberg et al, 1996##; ##REF##10737382##Halm et al, 2000##; ##REF##11169931##Ikeda et al, 2001##; ##REF##11745808##Tsuruta et al, 2001##; ##REF##12630768##Saad et al, 2002##; ##REF##12706865##Karayiannakis et al, 2003##; ##REF##12909220##Micke et al, 2003##; ##REF##14527913##Ohigashi et al, 2003##; ##REF##12791502##Paillaud et al, 2003##; ##REF##14605463##Stemmler et al, 2003##; ##REF##14628094##Talar-Wojnarowska et al, 2003##; ##REF##14562009##Ziske et al, 2003##; ##REF##15613258##Gupta et al, 2004##), including demographic, clinical (including, performance status, weight loss and treatment), surgical (including, palliative procedures, site and stage of disease) and laboratory (including, CA19-9, LDH, alkaline phosphatase and albumin). Further concerns include the inadequate use of statistical methods, model comparison when different factors are being investigated and the differing format of factors across studies.</p>", "<p>We developed three prognostic models, two based on standard assumptions of log linear or step functional relationships with survival and a novel approach based on nonlinear relationships using more complex FP transformations. The model based on transformed covariates (##TAB##3##Table 4##, Model 3) was the best-fitting model, better utilising the information within nonlinear covariates. This model confirmed five previously reported prognostic factors, namely albumin, CA19-9, alkaline phosphatase, LDH and metastases; and also identified three additional possible prognostic factors not previously reported: WBC, AST and BUN. Nonlinear transformations were appropriate for two variables indicating strong nonlinear effects on survival: CA19-9 as a second-degree FP and LDH under a log transformation. Importantly, the effect of CA19-9 was not apparent in the ‘linear’ model, the effect of alkaline phosphatase was not apparent in the ‘categorical’ model and the effects of AST and BUN were not apparent in either the ‘linear’ or ‘categorical’ models, indicating how the significant effect of these variables may go unrecognised due to simplistic assumptions made in statistical modelling.</p>", "<p>Shrinkage represents the degree to which a plot of predicted and observed values is flattened from the 45° line attributable to overfitting. Overfitting leads to inflated estimates of model fit and is a potentially important source of bias in prognostic models. Overfitting may be minimised through sensible model selection, which for survival models implies avoiding attempting to fit models with more than 1 candidate variable (degree of freedom) for each 10 events of interest (e.g., death) included in the analysis. The degree to which overfitting is present in the fitted model may be estimated either directly through validation in an external data set or through a bootstrap process. In practise, it is rare for an external data set to be available, and if data are scarce, it becomes attractive to use all available data to derive the prognostic model. Thus, bootstrap resampling approaches may become the model validation methods of choice. In our model, bootstrap resampling methods (##REF##8668867##Harrell et al, 1996##) suggested minimal optimism. As shrinking estimators will not increase the real discrimination of the model, and the degree of overfitting estimated for the model is minimal, rescaling the model estimates appears neither helpful nor necessary.</p>", "<p>A model based on multiple imputation methods (##UREF##5##Rubin, 1987##; ##UREF##6##Schafer, 1997##) to control for missing covariate data selected an additional variable nodal status as prognostic (<italic>P</italic>=0.016), which had been excluded from all models prior to imputation. The true importance of this variable requires further investigation, suggesting a strong link with other variables already in the model. All prognostic models ideally require external validation to determine the generality across different data sets, and our results may be seen as provisional until replicated on independent data. Performance status and tumour size at randomisation are well-documented factors (##TAB##0##Table 1##) but unfortunately were not available in this data set and should be included in any external validation.</p>", "<p>This research was based on data from two large, phase III randomised controlled trials representative of patients with advanced pancreatic cancer with a high event rate, long follow-up and an overall 1 year survival rate of 17% (##REF##11481349##Bramhall et al, 2001##, ##REF##12107836##2002##). Analyses were based on a multivariable approach and utilised the information contained within continuous variables appropriately. The functional form of the relationship between continuous covariates and survival should always be assessed when investigating potential prognostic value. Models were based on information readily available in clinic and once validated should have the ability to aid decision-making by identifying patients with borderline disease for surgery and patients for inclusion into clinical trials or off-study treatment, especially since a greater number of palliative and more toxic treatments are becoming available and being trialed in this disease.</p>" ]

[]

[ "<p>Pancreatic cancer is the fifth most common cause of cancer death. Identification of defined patient groups based on a prognostic index may improve the prediction of survival and selection of therapy. Many prognostic factors have been identified often based on retrospective, underpowered studies with unclear analyses. Data from 653 patients were analysed. Continuous variables are often simplified assuming a linear relationship with log hazard or introducing a step function (dichotomising). Misspecification may lead to inappropriate conclusions but has not been previously investigated in pancreatic cancer studies. Models based on standard assumptions were compared with a novel approach using nonlinear fractional polynomial (FP) transformations. The model based on FP-transformed covariates was most appropriate and confirmed five previously reported prognostic factors: albumin, CA19-9, alkaline phosphatase, LDH and metastases, and identified three additional factors not previously reported: WBC, AST and BUN. The effects of CA19-9, alkaline phosphatase, AST and BUN may go unrecognised due to simplistic assumptions made in statistical modelling. We advocate a multivariable approach that uses information contained within continuous variables appropriately. The functional form of the relationship between continuous covariates and survival should always be assessed. Our model should aid individual patient risk stratification and the design and analysis of future trials in pancreatic cancer.</p>" ]

[ "<p>Pancreatic ductal adenocarcinoma is a common cause of cancer death and is difficult to treat because clinical presentation is often late, and the disease is resistant to conventional chemotherapy. Long-term survival remains poor with a 5-year survival rate of 0.4–4% (##REF##7881926##Bramhall et al, 1995##; ##REF##12568441##Jemal et al, 2003##). Multivariable prognostic models are important for grouping patients into risk sets for predicting survival and treating appropriately. There is currently no prognostic tool in routine use to identify subgroups of pancreatic cancer patients for selection and stratification of treatment and prediction of survival.</p>", "<p>Because of its poor prognosis, few prognostic factors may be expected for patients with advanced pancreatic cancer; however, many possible factors have been identified. The majority of prognostic factor studies are questionable in terms of sample size and statistical methods, most based on small retrospective analyses. Literature searches (ISI Web of Science and Ovid Technologies databases) identified 36 prognostic factor studies reporting a total of 34 possible prognostic factors for advanced pancreatic cancer patients (##TAB##0##Table 1##) grouped as surgical, clinical, laboratory or demographic. Four studies (##REF##11350436##Johnson et al, 2001##; ##REF##12149301##Berlin et al, 2002##; ##REF##12181240##Ducreux et al, 2002##; ##REF##12118027##Maisey et al, 2002##) were randomised controlled trials reporting five prognostic factors from multivariate analyses, namely metastases, tumour site, performance status, alkaline phosphatase and treatment. The remaining 32 studies (##REF##8449644##Friedman and van den Eeden, 1993##; ##REF##7846017##Yasue et al, 1994##; ##REF##7535184##Falconer et al, 1995##; ##REF##8572621##Lundin et al, 1995##; ##REF##8830333##Ishii et al, 1996##; ##REF##8805925##Rothenberg et al, 1996##; ##REF##12118565##Shibamoto et al, 1996##; ##REF##10235206##Cubiella et al, 1999##; ##REF##10189130##Storniolo et al, 1999##; ##REF##10737382##Halm et al, 2000##; ##REF##11034467##Terwee et al, 2000##; ##REF##11051691##Trigui et al, 2000##; ##UREF##7##Ueno et al, 2000##; ##REF##11169931##Ikeda et al, 2001##; ##UREF##3##Ridwelski et al, 2001##; ##REF##11801751##Tas et al, 2001##; ##REF##11745808##Tsuruta et al, 2001##; ##REF##12630768##Saad et al, 2002##; ##REF##12555292##Bachmann et al, 2003##; ##REF##12711291##Engelken et al, 2003##; ##UREF##1##Fujino et al, 2003##; ##REF##12706865##Karayiannakis et al, 2003##; ##REF##12909220##Micke et al, 2003##; ##REF##14527913##Ohigashi et al, 2003##; ##REF##12791502##Paillaud et al, 2003##; ##REF##14605463##Stemmler et al, 2003##; ##REF##14628094##Talar-Wojnarowska et al, 2003##; ##REF##14562009##Ziske et al, 2003##; ##REF##15613258##Gupta et al, 2004##; ##REF##14962722##Kuhlmann et al, 2004##; ##REF##15028948##Watanabe et al, 2004##; ##REF##15698733##Ni et al, 2005##) were based on consecutive series of patients, often retrospective, often single-centre, of which 15 studies were based on fewer than 100 patients (##REF##7846017##Yasue et al, 1994##; ##REF##8830333##Ishii et al, 1996##; ##REF##8805925##Rothenberg et al, 1996##; ##REF##10737382##Halm et al, 2000##; ##REF##11169931##Ikeda et al, 2001##; ##REF##11745808##Tsuruta et al, 2001##; ##REF##12630768##Saad et al, 2002##; ##REF##12706865##Karayiannakis et al, 2003##; ##REF##12909220##Micke et al, 2003##; ##REF##14527913##Ohigashi et al, 2003##; ##REF##12791502##Paillaud et al, 2003##; ##REF##14605463##Stemmler et al, 2003##; ##REF##14628094##Talar-Wojnarowska et al, 2003##; ##REF##14562009##Ziske et al, 2003##; ##REF##15613258##Gupta et al, 2004##). One was based on five observational studies with varied inclusion criteria, inconsistent results and no prospective verification (##REF##11034467##Terwee et al, 2000##). The largest series (2380 patients) identified factors based on univariate analyses and data containing a large proportion (57%) of censored patients (##REF##10189130##Storniolo et al, 1999##).</p>", "<p>An important issue in prognostic factor studies is the nature of the relationship between the factor and survival (functional form). Continuous variables are often simplified at analysis by assuming a linear relationship with log-hazard or by introducing a step function through categorisation (frequently dichotomisation). If the linearity assumption is not correct, the final prognostic model could be misspecified. Misspecification of the functional form may lead to inappropriate conclusions but has not been previously investigated in pancreatic cancer studies. Many researchers avoid this problem by dichotomising, with a consequent loss of power. There is also the risk of important bias when the choice of cutoff is data-driven and the use of different cutoff points across multiple studies hinders direct comparisons.</p>", "<p>The aim of this study was to evaluate potentially important baseline prognostic factors for survival in advanced pancreatic cancer using prospective data from two randomised controlled trials and a total of 653 patients (##REF##11481349##Bramhall et al, 2001##, ##REF##12107836##2002##). The study investigated clinical, histological, biochemical and demographic variables. A multivariable approach was used accounting for the functional form of the relationship between continuous factors and survival. Models were developed either on the basis of standard assumptions of log linear or step functional relationships with survival, or a novel approach based on nonlinear relationships, using more complex fractional polynomial (FP) transformations: a flexible, parametric method for modelling nonlinear relationships (##UREF##4##Royston and Altman, 1994##; ##REF##10066088##Altman and Lyman, 1998##).</p>" ]

[ "<p>We are grateful to British Biotech for providing the individual patient data and to all patients who took part in these trials. We are grateful to Cancer Research UK who supported this research under a Population and Behavioural Sciences project grant (reference: C19491/A6150).</p>" ]

[ "<fig id=\"fig1\"><label>Figure 1</label><caption><p>Survival function by trial.</p></caption></fig>", "<fig id=\"fig2\"><label>Figure 2</label><caption><p>Estimated functional form for CA19-9. Dots indicate actual data values.</p></caption></fig>", "<fig id=\"fig3\"><label>Figure 3</label><caption><p>Estimated functional form for LDH. Dots indicate actual data values.</p></caption></fig>", "<fig id=\"fig4\"><label>Figure 4</label><caption><p>Survival by predictive group.</p></caption></fig>" ]

[ "<table-wrap id=\"tbl1\"><label>Table 1</label><caption><title>Literature review</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"left\"/><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Type (number) of studies</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Number of patients per study</bold>\n</th><th colspan=\"3\" align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Prognostic factors reported (frequency of reporting)</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>U or MV analysis</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Randomised controlled trial (<italic>n</italic>=4)^3–6</td><td align=\"center\" valign=\"top\" charoff=\"50\">207–322</td><td align=\"left\" valign=\"top\" charoff=\"50\">Surgical</td><td align=\"left\" valign=\"top\" charoff=\"50\">Metastases</td><td align=\"center\" valign=\"top\" charoff=\"50\">(3)</td><td align=\"center\" valign=\"top\" charoff=\"50\">4 MV</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Tumour location</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinical</td><td align=\"left\" valign=\"top\" charoff=\"50\">Performance status</td><td align=\"center\" valign=\"top\" charoff=\"50\">(3)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Treatment</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Laboratory</td><td align=\"left\" valign=\"top\" charoff=\"50\">Alkaline phosphatase</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Consecutive series &gt;500 patients (<italic>n</italic>=3)^7–9</td><td align=\"center\" valign=\"top\" charoff=\"50\">782–2380</td><td align=\"left\" valign=\"top\" charoff=\"50\">Surgical</td><td align=\"left\" valign=\"top\" charoff=\"50\">Metastases</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\">2 MV</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Stage of disease</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 U</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Operation</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinical</td><td align=\"left\" valign=\"top\" charoff=\"50\">Performance status</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Diabetes</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Pain</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Appetite/weight</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Jaundice</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Treatment</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Laboratory</td><td align=\"left\" valign=\"top\" charoff=\"50\">Albumin</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Demographic</td><td align=\"left\" valign=\"top\" charoff=\"50\">Age</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Specialist centre</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Consecutive series 100–500 patients (<italic>n</italic>=14)^10–23</td><td align=\"center\" valign=\"top\" charoff=\"50\">102–450</td><td align=\"left\" valign=\"top\" charoff=\"50\">Surgical</td><td align=\"left\" valign=\"top\" charoff=\"50\">Metastases</td><td align=\"center\" valign=\"top\" charoff=\"50\">(4)</td><td align=\"center\" valign=\"top\" charoff=\"50\">13 MV</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Stage of disease</td><td align=\"center\" valign=\"top\" charoff=\"50\">(2)</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 U</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Tumour location</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Operation</td><td align=\"center\" valign=\"top\" charoff=\"50\">(2)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Tumour size</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Duodenal invasion</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Peridissemination</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Ascites</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinical</td><td align=\"left\" valign=\"top\" charoff=\"50\">Performance status</td><td align=\"center\" valign=\"top\" charoff=\"50\">(2)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Diabetes</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Pain</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Appetite/weight</td><td align=\"center\" valign=\"top\" charoff=\"50\">(3)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Symptom onset</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Treatment</td><td align=\"center\" valign=\"top\" charoff=\"50\">(2)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Laboratory</td><td align=\"left\" valign=\"top\" charoff=\"50\">CA242</td><td align=\"center\" valign=\"top\" charoff=\"50\">(2)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">CA19-9</td><td align=\"center\" valign=\"top\" charoff=\"50\">(2)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Leukocytes</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Gamma GT</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Albumin</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">LDH</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">CRP</td><td align=\"center\" valign=\"top\" charoff=\"50\">(3)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Iron</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Demographic</td><td align=\"left\" valign=\"top\" charoff=\"50\">Age</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Consecutive series</td><td align=\"center\" valign=\"top\" charoff=\"50\">28–95</td><td align=\"left\" valign=\"top\" charoff=\"50\">Surgical</td><td align=\"left\" valign=\"top\" charoff=\"50\">Metastases</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\">8 MV</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">&lt;100 patients (<italic>n</italic>=15)^24–38</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Stage of disease</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\">7 U</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Grade of disease</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Nodal status</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Operation</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Tumour size</td><td align=\"center\" valign=\"top\" charoff=\"50\">(2)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Fibrosis</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinical</td><td align=\"left\" valign=\"top\" charoff=\"50\">Performance status Inflammation</td><td align=\"center\" valign=\"top\" charoff=\"50\">(4)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Appetite/weight</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Treatment</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Laboratory</td><td align=\"left\" valign=\"top\" charoff=\"50\">CA19-9</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">VEGF</td><td align=\"center\" valign=\"top\" charoff=\"50\">(7)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">CEA</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Phase angle BIA</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">SCA</td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">(1)</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl2\"><label>Table 2</label><caption><title>Patient characteristics by trial</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Variable</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>BB128³⁹\n<italic>N</italic>=414 (63%)</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>BB193⁴⁰\n<italic>N</italic>=239 (37%)</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Total <italic>N</italic>=653 (100%)</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Demographics</bold>\n<hr/></td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Age at entry (years)</italic>\n<xref ref-type=\"fn\" rid=\"t2-fn1\">a</xref>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Median</td><td align=\"center\" valign=\"top\" charoff=\"50\">63</td><td align=\"center\" valign=\"top\" charoff=\"50\">62</td><td align=\"center\" valign=\"top\" charoff=\"50\">63</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Range</td><td align=\"center\" valign=\"top\" charoff=\"50\">29–89</td><td align=\"center\" valign=\"top\" charoff=\"50\">32–85</td><td align=\"center\" valign=\"top\" charoff=\"50\">29–89</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Ethnic race</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> White</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">364 (88%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">226 (95%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">590 (90%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Black</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">27 (6%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">8 (3%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">35 (6%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Oriental</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">7 (2%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">7 (1%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Other</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">15 (4%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5 (2%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">20 (3%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Missing</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Sex</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Male</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">228 (55%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">140 (59%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">368 (56%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Female</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">186 (45%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">99 (41%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">285 (44%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Treatment</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Gemcitabine</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">103 (25%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">119 (50%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">222 (34%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Marimistat</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">311 (75%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">120 (50%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">431 (66%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Tumour information</bold>\n<hr/></td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Cancer stage</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> I</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">19 (4%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">13 (5%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">32 (5%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> II</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">45 (11%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">27 (11%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">72 (11%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> III</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">76 (19%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">28 (12%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">104 (16%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> IV</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">268 (66%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">171 (72%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">439 (68%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Missing</td><td align=\"center\" valign=\"top\" charoff=\"50\">6</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">6</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Distant metastases</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> M0</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">129 (31%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">65 (27%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">194 (30%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> M1</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">265 (64%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">171 (72%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">436 (67%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Missing</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">20 (5%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">3 (1%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">23 (3%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Regional lymph nodes</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> N0</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">153 (37%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">90 (38%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">243 (37%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> N1</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">164 (40%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">87 (36%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">251 (39%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Missing</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">97 (23%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">62 (26%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">159 (24%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Primary tumour T stage</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> T0</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5 (1%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">3 (1.5%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">8 (1.5%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> T1</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">114 (30%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">44 (20%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">158 (26%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> T2</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">90 (24%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">54 (25%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">144 (24%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> T3</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">167 (44%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">113 (53%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">280 (47%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> T4</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">6 (1%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1 (0.5%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">7 (1.5%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Missing</td><td align=\"center\" valign=\"top\" charoff=\"50\">32</td><td align=\"center\" valign=\"top\" charoff=\"50\">24</td><td align=\"center\" valign=\"top\" charoff=\"50\">56</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Serum chemistry and haematology</bold>\n<hr/></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Laboratory variables</bold>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Median (range), missing</bold>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Median (range), missing</bold>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Median (range), missing</bold>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">AST (SGOT)<xref ref-type=\"fn\" rid=\"t2-fn1\">a</xref></td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">24 (6–365), 17</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">26 (9–538), 12</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">25 (6–538), 29</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Total bilirubin<xref ref-type=\"fn\" rid=\"t2-fn1\">a</xref></td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">13.7 (3.4–277.0), 16</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">13.7 (3.0–135.1), 8</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">13.7 (3.0–277.0), 24</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Alkaline phosphatase<xref ref-type=\"fn\" rid=\"t2-fn1\">a</xref></td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">136 (36–1660), 16</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">157 (35–2064), 8</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">140 (35–2064), 24</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Albumin<xref ref-type=\"fn\" rid=\"t2-fn1\">a</xref></td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">38 (22–47), 17</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">38 (24–47), 8</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">38 (22–47), 25</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">LDH<xref ref-type=\"fn\" rid=\"t2-fn1\">a</xref></td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">163 (77–1074), 21</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">169 (29–1495), 11</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">164 (29–1495), 32</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">BUN<xref ref-type=\"fn\" rid=\"t2-fn1\">a</xref></td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">9.2 (2.9–34.3), 17</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">9.3 (4.3–27.9), 16</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">9.3 (2.9–34.3), 33</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">CA19/9<xref ref-type=\"fn\" rid=\"t2-fn1\">a</xref></td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">686 (5–1 000 000), 17</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">800 (8–1 000 000), 30</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">710 (5–1 000 000), 47</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Haemoglobin<xref ref-type=\"fn\" rid=\"t2-fn1\">a</xref></td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">12.5 (5.5–16.1), 28</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">12.4 (8.3–19.1), 13</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">12.4 (5.5–19.1), 41</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">WBC<xref ref-type=\"fn\" rid=\"t2-fn1\">a</xref></td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">7.6 (2.3–31.6), 28</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">8.3 (2.4–23.7), 13</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">7.9 (2.3–31.6), 41</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Outcome</bold>\n<hr/></td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Event indicator</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Alive</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">22 (5%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">19 (8%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">41 (6%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Dead</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">392 (95%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">220 (92%)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">612 (94%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Follow-up of alive (months)</italic>\n<xref ref-type=\"fn\" rid=\"t2-fn1\">a</xref>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Median</td><td align=\"center\" valign=\"top\" charoff=\"50\">20.1</td><td align=\"center\" valign=\"top\" charoff=\"50\">19.4</td><td align=\"center\" valign=\"top\" charoff=\"50\">20.7</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Range</td><td align=\"center\" valign=\"top\" charoff=\"50\">0.9–24.6</td><td align=\"center\" valign=\"top\" charoff=\"50\">1.9–23.3</td><td align=\"center\" valign=\"top\" charoff=\"50\">0.9–24.6</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl3\"><label>Table 3</label><caption><title>Univariate log-rank analyses</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/><col align=\"center\"/><col align=\"char\" char=\"(\"/><col align=\"left\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>Patients</bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>Deaths</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>12-month survival (%)</bold>\n</th><th align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>Median survival (95% CI)</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold><italic>χ</italic>²_LR, p (<italic>χ</italic>²_W, p)</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>HR (95% CI)</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Overall survival</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">653</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">612</td><td align=\"center\" valign=\"top\" charoff=\"50\">17</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">4.7 (4.2, 5.1)</td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td><td align=\"center\" valign=\"top\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Trial</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> BB128</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">414</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">392</td><td align=\"center\" valign=\"top\" charoff=\"50\">17</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">4.2 (3.6, 4.8)</td><td align=\"left\" valign=\"top\" charoff=\"50\">2.28, <italic>P</italic>=0.13</td><td align=\"center\" valign=\"top\" charoff=\"50\">1.0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> BB193</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">239</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">220</td><td align=\"center\" valign=\"top\" charoff=\"50\">18</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5.4 (4.8, 6.0)</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.88 (0.75, 1.04)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Demographics</bold>\n<hr/></td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Age group (years)</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> &lt;=63</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">343</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">320</td><td align=\"center\" valign=\"top\" charoff=\"50\">21</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5.1 (4.3, 5.8)</td><td align=\"left\" valign=\"top\" charoff=\"50\">4.42, <italic>P</italic>=0.036</td><td align=\"center\" valign=\"top\" charoff=\"50\">1.0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> &gt;63</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">310</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">292</td><td align=\"center\" valign=\"top\" charoff=\"50\">13</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">4.3 (3.5, 4.9)</td><td align=\"left\" valign=\"top\" charoff=\"50\">(7.55, <italic>P</italic>=0.006)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.18 (1.01, 1.39)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Ethnic group</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> White</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">590</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">554</td><td align=\"center\" valign=\"top\" charoff=\"50\">18</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">4.6 (4.1, 5.1)</td><td align=\"left\" valign=\"top\" charoff=\"50\">0.20, <italic>P</italic>=0.65</td><td align=\"center\" valign=\"top\" charoff=\"50\">1.0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Other</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">62</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">57</td><td align=\"center\" valign=\"top\" charoff=\"50\">12</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5.2 (3.5, 6.0)</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.06 (0.80, 1.41)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Sex</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Female</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">285</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">267</td><td align=\"center\" valign=\"top\" charoff=\"50\">18</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">4.9 (4.2, 5.8)</td><td align=\"left\" valign=\"top\" charoff=\"50\">0.73, <italic>P</italic>=0.39</td><td align=\"center\" valign=\"top\" charoff=\"50\">1.0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Male</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">368</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">345</td><td align=\"center\" valign=\"top\" charoff=\"50\">17</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">4.5 (3.9, 5.1)</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.07 (0.91, 1.26)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Treatment</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Gemcitabine</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">222</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">204</td><td align=\"center\" valign=\"top\" charoff=\"50\">18</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5.5 (4.7, 5.9)</td><td align=\"left\" valign=\"top\" charoff=\"50\">2.79, <italic>P</italic>=0.095</td><td align=\"center\" valign=\"top\" charoff=\"50\">1.0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Marimistat</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">431</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">408</td><td align=\"center\" valign=\"top\" charoff=\"50\">17</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">4.2 (3.5, 4.9)</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.15 (0.98, 1.36)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Tumour information</bold>\n<hr/></td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Cancer stage</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Early (I/II)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">104</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">92</td><td align=\"center\" valign=\"top\" charoff=\"50\">26</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">6.8 (5.7, 8.2)</td><td align=\"left\" valign=\"top\" charoff=\"50\">14.72, <italic>P</italic>&lt;0.001</td><td align=\"center\" valign=\"top\" charoff=\"50\">1.0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Late (III/IV)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">543</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">514</td><td align=\"center\" valign=\"top\" charoff=\"50\">16</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">4.1 (3.5, 4.7)</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.53 (1.26, 1.86)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Metastases</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> M0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">194</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">176</td><td align=\"center\" valign=\"top\" charoff=\"50\">30</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">6.8 (5.9, 8.4)</td><td align=\"left\" valign=\"top\" charoff=\"50\">35.47, <italic>P</italic>&lt;0.001</td><td align=\"center\" valign=\"top\" charoff=\"50\">1.0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> M1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">436</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">414</td><td align=\"center\" valign=\"top\" charoff=\"50\">12</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">3.5 (3.2, 4.0)</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.69 (1.43, 1.99)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Missing</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">23</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">22</td><td align=\"center\" valign=\"top\" charoff=\"50\">17</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5.5 (4.9, 7.5)</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.35 (0.88, 2.09)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Lymph nodes</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> N0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">243</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">226</td><td align=\"center\" valign=\"top\" charoff=\"50\">20</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5.5 (4.8, 6.0)</td><td align=\"left\" valign=\"top\" charoff=\"50\">6.73, <italic>P</italic>=0.035</td><td align=\"center\" valign=\"top\" charoff=\"50\">1.0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> N1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">251</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">240</td><td align=\"center\" valign=\"top\" charoff=\"50\">18</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">4.5 (3.5, 5.4)</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.19 (1.00, 1.43)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Missing</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">159</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">146</td><td align=\"center\" valign=\"top\" charoff=\"50\">11</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">3.8 (3.2, 4.9)</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.29 (1.04, 1.59)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Tumour stage</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Early (0/1/2)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">310</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">291</td><td align=\"center\" valign=\"top\" charoff=\"50\">16</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">4.3 (3.7, 4.9)</td><td align=\"left\" valign=\"top\" charoff=\"50\">1.63, <italic>P</italic>=0.44</td><td align=\"center\" valign=\"top\" charoff=\"50\">1.0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Late (3/4)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">287</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">268</td><td align=\"center\" valign=\"top\" charoff=\"50\">18</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">4.9 (4.2, 5.8)</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.91 (0.77, 1.08)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Missing</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">56</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">53</td><td align=\"center\" valign=\"top\" charoff=\"50\">18</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5.8 (3.5, 7.9)</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.87 (0.66, 1.16)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Serum chemistry and haematology</bold>\n<hr/></td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>AST (SGOT)</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Normal</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">538</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">499</td><td align=\"center\" valign=\"top\" charoff=\"50\">19</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5.1 (4.6, 5.7)</td><td align=\"left\" valign=\"top\" charoff=\"50\">14.17, <italic>P</italic>&lt;0.001</td><td align=\"center\" valign=\"top\" charoff=\"50\">1.0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Abnormal</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">86</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">84</td><td align=\"center\" valign=\"top\" charoff=\"50\">12</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2.8 (2.2, 3.9)</td><td align=\"left\" valign=\"top\" charoff=\"50\">(5.99, <italic>P</italic>=0.014)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.55 (1.18, 2.04)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Total bilirubin</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Normal</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">464</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">429</td><td align=\"center\" valign=\"top\" charoff=\"50\">20</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5.1 (4.7, 5.7)</td><td align=\"left\" valign=\"top\" charoff=\"50\">9.32, <italic>P</italic>=0.002</td><td align=\"center\" valign=\"top\" charoff=\"50\">1.0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Abnormal</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">165</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">159</td><td align=\"center\" valign=\"top\" charoff=\"50\">11</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">3.8 (3.3, 4.4)</td><td align=\"left\" valign=\"top\" charoff=\"50\">(6.27, <italic>P</italic>=0.012)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.32 (1.09, 1.61)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Alkaline phosphatase</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Normal</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">442</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">411</td><td align=\"center\" valign=\"top\" charoff=\"50\">20</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5.5 (5.0, 6.1)</td><td align=\"left\" valign=\"top\" charoff=\"50\">20.20, <italic>P</italic>&lt;0.001</td><td align=\"center\" valign=\"top\" charoff=\"50\">1.0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Abnormal</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">187</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">177</td><td align=\"center\" valign=\"top\" charoff=\"50\">13</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">3.1 (2.6, 3.5)</td><td align=\"left\" valign=\"top\" charoff=\"50\">(56.05, <italic>P</italic>&lt;0.001)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.49 (1.23, 1.81)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Albumin</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Normal</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">583</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">544</td><td align=\"center\" valign=\"top\" charoff=\"50\">19</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5.1 (4.6, 5.6)</td><td align=\"left\" valign=\"top\" charoff=\"50\">31.37, <italic>P</italic>&lt;0.001</td><td align=\"center\" valign=\"top\" charoff=\"50\">1.0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Abnormal</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">45</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">43</td><td align=\"center\" valign=\"top\" charoff=\"50\">7</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.5 (1.0, 2.7)</td><td align=\"left\" valign=\"top\" charoff=\"50\">(74.34, <italic>P</italic>&lt;0.001)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2.36 (1.49, 3.72)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>LDH</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Normal</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">543</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">505</td><td align=\"center\" valign=\"top\" charoff=\"50\">20</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5.2 (4.8, 5.8)</td><td align=\"left\" valign=\"top\" charoff=\"50\">37.05, <italic>P</italic>&lt;0.001</td><td align=\"center\" valign=\"top\" charoff=\"50\">1.0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Abnormal</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">78</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">75</td><td align=\"center\" valign=\"top\" charoff=\"50\">5</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2.1 (1.5, 2.8)</td><td align=\"left\" valign=\"top\" charoff=\"50\">(36.16, <italic>P</italic>&lt;0.001)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2.08 (1.50, 2.88)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>BUN</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Normal</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">407</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">382</td><td align=\"center\" valign=\"top\" charoff=\"50\">20</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5.1 (4.3, 5.7)</td><td align=\"left\" valign=\"top\" charoff=\"50\">3.43, <italic>P</italic>=0.064</td><td align=\"center\" valign=\"top\" charoff=\"50\">1.0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Abnormal</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">213</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">199</td><td align=\"center\" valign=\"top\" charoff=\"50\">13</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">4.4 (3.5, 5.1)</td><td align=\"left\" valign=\"top\" charoff=\"50\">(5.28, <italic>P</italic>=0.022)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.17 (0.98, 1.40)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>CA19/9</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Normal</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">98</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">86</td><td align=\"center\" valign=\"top\" charoff=\"50\">28</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">6.3 (4.8, 8.0)</td><td align=\"left\" valign=\"top\" charoff=\"50\">7.74, <italic>P</italic>=0.005</td><td align=\"center\" valign=\"top\" charoff=\"50\">1.0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Abnormal</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">508</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">481</td><td align=\"center\" valign=\"top\" charoff=\"50\">16</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">4.6 (4.0, 5.1)</td><td align=\"left\" valign=\"top\" charoff=\"50\">(4.84, <italic>P</italic>=0.028)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.38 (1.12, 1.70)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Haemoglobin</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Normal</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">79</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">77</td><td align=\"center\" valign=\"top\" charoff=\"50\">8</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">3.7 (3.3, 5.1)</td><td align=\"left\" valign=\"top\" charoff=\"50\">6.88, <italic>P</italic>=0.009</td><td align=\"center\" valign=\"top\" charoff=\"50\">1.0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Abnormal</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">533</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">495</td><td align=\"center\" valign=\"top\" charoff=\"50\">20</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">4.9 (4.4, 5.6)</td><td align=\"left\" valign=\"top\" charoff=\"50\">(10.64, <italic>P</italic>=0.001)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.73 (0.55, 0.95)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"7\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>WBC</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Normal</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">483</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">446</td><td align=\"center\" valign=\"top\" charoff=\"50\">21</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5.5 (4.9, 5.9)</td><td align=\"left\" valign=\"top\" charoff=\"50\">34.36, <italic>P</italic>&lt;0.001</td><td align=\"center\" valign=\"top\" charoff=\"50\">1.0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Abnormal</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">129</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">126</td><td align=\"center\" valign=\"top\" charoff=\"50\">8</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2.9 (2.4, 4.0)</td><td align=\"left\" valign=\"top\" charoff=\"50\">(46.52, <italic>P</italic>&lt;0.001)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.78 (1.40, 2.26)</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl4\"><label>Table 4</label><caption><title>Cox proportional hazards regression models, <italic>n</italic>=556 patients, 520 deaths</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"left\"/><col align=\"left\"/><col align=\"left\"/><col align=\"left\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Variable</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>\n<italic>χ</italic>\n²\n</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold><italic>P</italic>-value</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>HR (95% CI)</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>\n<bold>(a) Model 1 – ‘Linear’ covariates</bold>\n</italic>\n</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Full model</bold>\n<hr/></td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\"><italic>AIC</italic>=<italic>5558.4</italic></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Stratification factors</td><td align=\"left\" valign=\"top\" charoff=\"50\">Trial</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">7.3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.007</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.76 (0.63, 0.93)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Cancer stage<xref ref-type=\"fn\" rid=\"t4-fn2\">a</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.0001</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.99</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.00 (0.69, 1.44)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Sex</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3.8</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.050</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.82 (0.68, 1.00)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">TRT</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2.6</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.11</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.17 (0.97, 1.42)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Independent factors</td><td align=\"left\" valign=\"top\" charoff=\"50\">Age</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3.0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.081</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.01 (1.00, 1.02)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Albumin</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">22.9</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.94 (0.92, 0.96)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Alkaline phosphatase</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">24.3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.00 (1.00, 1.00)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">AST</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5.3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.022</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.00 (0.99, 1.00)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Bilirubin</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.7</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.40</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.00 (1.00, 1.01)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">BUN</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2.3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.13</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.02 (0.99, 1.05)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">CA19/9</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.9</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.34</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.00 (1.00, 1.00)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Ethnic</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.0009</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.98</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.01 (0.74, 1.36)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Haemoglobin</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.01</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.91</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.00 (0.94, 1.08)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">LDH</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">19.3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.00 (1.00, 1.00)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">METS<xref ref-type=\"fn\" rid=\"t4-fn3\">b</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">8.5</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.004</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.50 (1.14, 1.96)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Nodes<xref ref-type=\"fn\" rid=\"t4-fn3\">b</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.4</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.51</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.08 (0.86, 1.37)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Tumour stage<xref ref-type=\"fn\" rid=\"t4-fn3\">b</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.03</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.86</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.02 (0.84, 1.23)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">WBC</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">10.2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.04 (1.02, 1.07)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Final model</bold>\n<hr/></td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\"><italic>AIC</italic>=<italic>5557.1</italic></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Stratification factors</td><td align=\"left\" valign=\"top\" charoff=\"50\">Trial</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">8.1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.005</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.76 (0.63, 0.92)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Cancer stage<xref ref-type=\"fn\" rid=\"t4-fn2\">a</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.008</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.93</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.01 (0.74, 1.38)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Sex</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3.8</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.051</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.84 (0.70, 1.00)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">TRT</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3.2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.073</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.19 (0.98, 1.44)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Independent factors</td><td align=\"left\" valign=\"top\" charoff=\"50\">Albumin</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">41.0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.94 (0.92, 0.96)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">LDH</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">13.3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.00 (1.00, 1.00)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">METS<xref ref-type=\"fn\" rid=\"t4-fn3\">b</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">10.5</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.50 (1.17, 1.92)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">WBC</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">11.7</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.04 (1.02, 1.07)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Alkaline phosphatase</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>21.1</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>&lt;0.001</bold>\n</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>1.00 (1.00, 1.00)</bold>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>\n<bold>(b) Model 2 – ‘Categorical’ covariates</bold>\n</italic>\n</td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Full model</bold>\n<hr/></td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\"><italic>AIC</italic>=<italic>5582.3</italic></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Stratification factors</td><td align=\"left\" valign=\"top\" charoff=\"50\">Trial</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5.8</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.016</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.79 (0.65, 0.96)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Cancer stage<xref ref-type=\"fn\" rid=\"t4-fn2\">a</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.0001</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.99</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.00 (0.70, 1.44)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Sex</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2.1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.14</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.87 (0.72, 1.05)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">TRT</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3.3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.07</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.19 (0.99, 1.44)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Independent factors</td><td align=\"left\" valign=\"top\" charoff=\"50\">Age</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2.2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.13</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.15 (0.96, 1.39)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Albumin</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">17.6</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2.08 (1.48, 2.93)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Alkaline phosphatase</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.57</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.07 (0.85, 1.34)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">AST</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1.5</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.22</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.20 (0.90, 1.59)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Bilirubin</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5.0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.025</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.28 (1.03, 1.58)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">BUN</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1.9</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.17</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.14 (0.94, 1.39)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">CA19/9</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">7.8</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.005</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.43 (1.11, 1.85)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Ethnic</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.5</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.48</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.89 (0.66, 1.22)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Haemoglobin</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2.2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.14</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.82 (0.62, 1.07)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">LDH</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">22.8</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2.07 (1.54, 2.79)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">METS<xref ref-type=\"fn\" rid=\"t4-fn3\">b</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">13.3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.64 (1.26, 2.14)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Nodes<xref ref-type=\"fn\" rid=\"t4-fn3\">b</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2.3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.13</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.20 (0.95, 1.51)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">T stage<xref ref-type=\"fn\" rid=\"t4-fn3\">b</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.06</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.80</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.98 (0.81, 1.18)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">WBC</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">9.2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.002</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.42 (1.13, 1.78)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Final model</bold>\n<hr/></td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\"><italic>AIC</italic>=<italic>5583.2</italic></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Stratification factors</td><td align=\"left\" valign=\"top\" charoff=\"50\">Trial</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">7.0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.008</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.77 (0.64, 0.94)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Cancer stage<xref ref-type=\"fn\" rid=\"t4-fn2\">a</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.9</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.36</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.16 (0.85, 1.58)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Sex</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1.8</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.18</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.89 (0.74, 1.06)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">TRT</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3.4</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.065</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.19 (0.99, 1.44)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Independent factors</td><td align=\"left\" valign=\"top\" charoff=\"50\">Albumin</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">24.2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2.30 (1.65, 3.21)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">LDH</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">25.1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2.05 (1.55, 2.72)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">METS<xref ref-type=\"fn\" rid=\"t4-fn3\">b</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">11.9</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.54 (1.21, 1.97)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">WBC</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">10.3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.44 (1.15, 1.79)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Bilirubin</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>8</bold>\n<bold>.4</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>0</bold>\n<bold>.004</bold>\n</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"><bold>1.34</bold>\n<bold>(1.10, 1.64)</bold></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>CA199</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>9</bold>\n<bold>.4</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>0</bold>\n<bold>.002</bold>\n</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"><bold>1.48</bold>\n<bold>(1.15, 1.89)</bold></td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>\n<bold>(c) Model 3–‘Transformed’ covariates</bold>\n</italic>\n</td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Full model</bold>\n<hr/></td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\"><italic>AIC</italic>=<italic>5509.6</italic></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Stratification factors</td><td align=\"left\" valign=\"top\" charoff=\"50\">Trial</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">12.5</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.70 (0.57, 0.85)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Cancer stage<xref ref-type=\"fn\" rid=\"t4-fn2\">a</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.68</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.93 (0.64, 1.33)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Sex</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">4.3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.038</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.82 (0.67, 0.99)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">TRT</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3.0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.082</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.19 (0.98, 1.44)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\"> <italic>Independent factors</italic></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Linear</td><td align=\"left\" valign=\"top\" charoff=\"50\">Age</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2.3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.13</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.01 (1.00, 1.02)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Linear</td><td align=\"left\" valign=\"top\" charoff=\"50\">Albumin</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">20.8</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.94 (0.92, 0.97)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Linear</td><td align=\"left\" valign=\"top\" charoff=\"50\">Alkaline phosphatase</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">11.3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.00 (1.00, 1.00)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Linear</td><td align=\"left\" valign=\"top\" charoff=\"50\">AST</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">7.5</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.0062</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.00 (0.99, 1.00)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  1st degree FP</td><td align=\"left\" valign=\"top\" charoff=\"50\">Log (bilirubin)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3.0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.083</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.16 (0.98, 1.37)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Linear</td><td align=\"left\" valign=\"top\" charoff=\"50\">BUN</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3.2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.075</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.02 (1.00, 1.05)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  2nd degree FP</td><td align=\"left\" valign=\"top\" charoff=\"50\">CA199^0.5</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">36.3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.03 (1.02, 1.03)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  2nd degree FP</td><td align=\"left\" valign=\"top\" charoff=\"50\">CA199^0.5 × log(CA199)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">30.9</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.00 (1.00, 1.00)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Categorical</td><td align=\"left\" valign=\"top\" charoff=\"50\">Ethnic</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.74</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.95 (0.70, 1.29)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Linear</td><td align=\"left\" valign=\"top\" charoff=\"50\">Haemoglobin</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.61</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.98 (0.92, 1.05)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  1st degree FP</td><td align=\"left\" valign=\"top\" charoff=\"50\">Log (LDH)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">12.3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.72 (1.27, 2.32)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Categorical</td><td align=\"left\" valign=\"top\" charoff=\"50\">METS<xref ref-type=\"fn\" rid=\"t4-fn3\">b</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6.1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.014</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.41 (1.07, 1.85)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Categorical</td><td align=\"left\" valign=\"top\" charoff=\"50\">Nodes<xref ref-type=\"fn\" rid=\"t4-fn3\">b</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1.4</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.24</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.15 (0.91, 1.46)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Categorical</td><td align=\"left\" valign=\"top\" charoff=\"50\">Tumour stage<xref ref-type=\"fn\" rid=\"t4-fn3\">b</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.60</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.05 (0.87, 1.28)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Linear</td><td align=\"left\" valign=\"top\" charoff=\"50\">WBC</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">10.5</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.0012</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.05 (1.02, 1.07)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"5\" align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Final model</bold>\n<hr/></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"><italic>AIC</italic>=<italic>5510.2</italic></td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Stratification factors</td><td align=\"left\" valign=\"top\" charoff=\"50\">Trial</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">15.1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.68 (0.56, 0.83)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Cancer stage<xref ref-type=\"fn\" rid=\"t4-fn2\">a</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.001</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.97</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.0 (0.73, 1.36)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Sex</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3.5</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.061</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.84 (0.70, 1.01)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">TRT</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3.16</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.075</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.19 (0.98, 1.44)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> <italic>Independent factors</italic></td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Linear</td><td align=\"left\" valign=\"top\" charoff=\"50\">Albumin</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">41.4</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.94 (0.92, 0.95)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  1st degree FP</td><td align=\"left\" valign=\"top\" charoff=\"50\">Log (LDH)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">12.8</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">&lt;0.001</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.70 (1.27, 2.27)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Categorical</td><td align=\"left\" valign=\"top\" charoff=\"50\">METS<xref ref-type=\"fn\" rid=\"t4-fn3\">b</xref></td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">4.0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.047</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.29 (1.00, 1.66)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Linear</td><td align=\"left\" valign=\"top\" charoff=\"50\">WBC</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">10.0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0.002</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.04 (1.02, 1.07)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>Linear</bold></td><td align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Alkaline phosphatase</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>14</bold>\n<bold>.6</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>&lt;0</bold>\n<bold>.001</bold>\n</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"><bold>1.00</bold>\n<bold>(1.00, 1.00)</bold></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>Linear</bold></td><td align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>AST</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>5</bold>\n<bold>.2</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>0</bold>\n<bold>.023</bold>\n</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"><bold>1.00</bold>\n<bold>(0.99, 1.00)</bold></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>Linear</bold></td><td align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>BUN</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>5</bold>\n<bold>.0</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>0</bold>\n<bold>.026</bold>\n</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"><bold>1.03</bold>\n<bold>(1.00, 1.06)</bold></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>2nd degree FP</bold></td><td align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>CA199</bold>\n^{\n<bold>0.5</bold>\n}\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>33</bold>\n<bold>.3</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>&lt;0</bold>\n<bold>.001</bold>\n</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\"><bold>1.02</bold>\n<bold>(1.02, 1.03)</bold></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>2nd degree FP</bold></td><td align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>CA199^0.5 × log(CA199)</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>28.4</bold>\n</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>&lt;0.001</bold>\n</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>1.00 (1.00, 1.00)</bold>\n</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<fn-group><fn><p>\n<bold>Conflict of interest</bold>\n</p><p>The authors state no conflict of interest.</p></fn></fn-group>", "<table-wrap-foot><fn id=\"t1-fn1\"><p>MV=multivariate; U=univariate.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"t2-fn1\"><label>a</label><p>Continuous measurements.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"t3-fn1\"><p>HR=hazard ratio; LR=log-rank statistic; W=Wald <italic>χ</italic>² statistic.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"t4-fn1\"><p>FP=fractional polynomial; HR=hazard ratio.</p></fn><fn id=\"t4-fn2\"><label>a</label><p>Stage (I/II <italic>vs</italic> III/IV).</p></fn><fn id=\"t4-fn3\"><label>b</label><p>Metastases (negative <italic>vs</italic> positive), Nodes (negative <italic>vs</italic> positive), Tumour stage (I/II <italic>vs</italic> III/IV): missing data included in analysis as a separate ‘dummy’ variable using lower level as the reference level.</p></fn></table-wrap-foot>" ]

[ "<graphic xlink:href=\"6604568f1\"/>", "<graphic xlink:href=\"6604568f2\"/>", "<graphic xlink:href=\"6604568f3\"/>", "<graphic xlink:href=\"6604568f4\"/>" ]

[]

{ "acronym": [], "definition": [] }

[]

[]

[]

[]

[]

[]

[ "<p>\n<bold>Sir,</bold>\n</p>", "<p>It is reassuring that Dr Ingleby reaches the same conclusion as we did, that black patients have earlier onset on average than white patients, even after taking into account the different age distributions of the populations. Unlike Dr Ingleby, we are reluctant to attempt to deduce absolute incidence from our figures, as hospital admissions do not exactly correspond to geographical areas. Therefore, we cannot be certain that the differences are entirely because of the ‘black other’ subgroup, as suggested by Dr Ingleby. Although the ‘black other’ category has the youngest age at onset in our data, the black African, Caribbean and British patients were also significantly younger than white patients (<italic>P</italic>&lt;0.001). The poorer survival in tumours 2 cm in size or less was also observed in all the black ethnic subgroups.</p>", "<p>Dr Ingleby queries the effect of socioeconomic status. As measured by the Index of Multiple Deprivation, this did not differ significantly between ethnic groups, nor did it have a significant effect on survival.</p>", "<p>We have taken care to avoid knee-jerk reactions, such as to recommend an early detection response to these observations. The fact that the poorer survival of black patients was confined to small tumours suggests that early detection approaches are not the solution, or at least not the whole solution. Our next task is the further study of aetiology and tumour biology to find out why there is earlier onset among black patients and why they have poorer survival in the case of small tumours. When these questions have been answered, we will be in a better position to prescribe appropriate control measures.</p>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

{ "acronym": [], "definition": [] }

[]

[ "<title>Materials and methods</title>", "<title>Cell culture, cell lysates preparation, immunoprecipitation, and immunoblotting</title>", "<p>Lung cancer cell lines were obtained from American Type Culture Collection and grown in RPMI 1640 (Hyclone, Logan, UT, USA), 10% (v/v) foetal bovine serum (FBS) as instructed under standard cell culture conditions. For growth factor stimulation studies, human HGF (50 ng ml⁻¹) (R&amp;D Systems, Minneapolis, MN, USA) and human EGF (100 ng ml⁻¹) (Calbiochem, Cambridge, MA, USA) were used as indicated. Cellular proteins were extracted from whole cells as previously described (##REF##16407879##Choong et al, 2006##). Immunoprecipitation (IP) studies and immunoblotting (WB) were performed as previously described (##REF##11839685##Maulik et al, 2002##; ##REF##16407879##Choong et al, 2006##) using the following primary antibodies as indicated: p-MET[Y1234/1235] (i.e., pY1252/1253 as in the full-length MET version) (Cell Signaling, Danvers, MA, USA), MET (C-12, Santa Cruz Biotechnology, Santa Cruz, CA, USA), p-EGFR[Y1068] (Cell Signaling), EGFR (Santa Cruz Biotechnology), p-ERBB3[Y1289] (Cell Signaling), p-AKT[S473] (Cell Signaling), AKT (Biosource-Invitrogen, Carlsbad, CA, USA), p-extracellular signal-regulated kinases 1 and 2 (ERK1/2)[T202/Y204] (Cell Signaling), ERK1/2 (Biosource-Invitrogen), p-STAT3[Y705] (Cell Signaling), STAT3 (Zymed-Invitrogen, Carlsbad, CA, USA), phosphotyrosine (p-Tyr; Upstate-Millipore, Billerica, MA, USA), and Actin (Santa Cruz Biotechnology).</p>", "<title>Chemicals</title>", "<p><italic>SU11274</italic>: [(3Z)-<italic>N</italic>-(3-chlorophenyl)-3-({3,5-dimethyl-4-[(4-methylpiperazin-1-yl)carbonyl]-1<italic>H</italic>-pyrrol-2-yl}methylene)-<italic>N</italic>-methyl-2-oxo-2,3-dihydro-1<italic>H</italic>-indole-5-sulphonamide] (##REF##15735036##Ma et al, 2005a##) and CL-387,785 were purchased from EMD-Calbiochem (Cambridge, MA, USA), dissolved in DMSO, and used at the indicated concentrations. Erlotinib was prepared as previously described (##REF##16407879##Choong et al, 2006##).</p>", "<p><italic>Genomic studies of</italic> MET <italic>gene DNA extraction and DNA sequencing</italic>: Genomic DNA of H1975 cells were extracted using Qiagen DNAeasy Kit (Qiagen, Valencia, CA, USA) according to the manufacturer's instructions. Direct DNA sequencing of the complete <italic>MET</italic> gene was performed as previously described (##REF##15735036##Ma et al, 2005a##).</p>", "<p><italic>Quantitative real-time polymerase chain reaction (QPCR)</italic>: Genomic copy number variation of the <italic>MET</italic> gene was determined in triplicate using QPCR with the RNaseP as the reference gene. Quantitative real-time polymerase chain reactions were performed in ABI PRISM 7900-HT System and the reaction conditions are available upon request. The QPCR primers for <italic>MET</italic> copy number variation determination were purchased from ABI (ABI assay no.: Hs01565582_g1).</p>", "<p><italic>Cellular apoptosis and viability assay</italic>: For cellular apoptosis assays, cells were plated separately in triplicate in six-well plates in 10% FBS-containing media. Drug inhibitor treatment using erlotinib, SU11274, or DMSO control as indicated was added the next day, with the cells incubated for 72 h further. Cellular apoptosis was determined by fluorescence-activated cell sorting (FACS) analysis using the Annexin-V-Fluos Staining Kit (Roche Diagnostics, Mannheim, Germany) according to the manufacturer's instructions. Cellular apoptosis assays were performed in triplicate. Cellular viability assays were performed using the Trypan-Blue Dye Exclusion assay with duplicate counting using standard haemacytometer-light microscopy, with each experiment further repeated in duplicate.</p>", "<title>Time-lapsed video microscopy and image analysis of cytoskeletal functions</title>", "<p>H1975 cells were plated on cell culture dishes and placed into a temperature-controlled chamber at 37°C in an atmosphere of 5% CO₂. The cells were examined by and recorded under video microscopy using an Leica 6000 B inverted microscope, Pecon incubation chamber, and Retiga EXI 12 bit camera (Q imaging, Vancouver, BC, Canada) with MetaMorph image analysis software (Universal Imaging, Downington, PA, USA) (details see also <xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Materials and Methods</xref>).</p>", "<p><italic>Lentivirus transduction of luciferase-expressing vector</italic> (a) Lentivirus production: <italic>Plasmids</italic>. The packaging plasmid pCMVÄR8.91, the vesicular stomatitis virus glycoprotein G (VSV.G), and encoding plasmid pCSO-rre-cppt-MCU3-LUC were kind gifts from Dr Donald B Kohn (University of Southern California). <italic>Virus production.</italic> Transfection with transfer vector, packaging plasmid and envelope plasmid were performed by calcium phosphate precipitate 12 h after planting package 293T cells into 10 cm cell culture dishes. (b) Lentiviral transduction of EGFR-TKI-resistant lung tumour cells: Medium from the package cell culture was then collected and centrifuged at 3000 r.p.m. for 5 min at room temperature, followed by filtering through 0.45 <italic>μ</italic>m filter. The filtered medium containing virus particles was then added to the target transduction cells (H1975), which were plated the day before transduction.</p>", "<title><italic>In vivo</italic> murine xenograft model</title>", "<p>Six-week-old female Ncr-nu (Nude) mice were purchased from Charles River Laboratories (Wilmington, MA, USA) and hosted in the pathogen-free animal facility at the Case Western Reserve University. <italic>In vivo</italic> animal studies were performed according to institution-approved protocols and guidelines. Xenografts of the luciferase-expressing H1975 lung cancer cells were established by intradermally injecting 3 × 10⁶ viable cells in RPMI 1640 media into the flank/leg region of nude mice to produce subcutaneous tumours. Indicated treatments with targeted TKIs were given at the time when tumour xenografts were beginning to be visible (corresponding to 7 days post-implantation of H1975 cells). <italic>In vivo</italic> daily inhibitor drug treatments were performed as indicated. SU11274 was administered as intratumoral injections, whereas erlotinib was administered using oral gavage. Body weight was recorded for each animal twice weekly to monitor potential toxicities. Tumour xenografts were subsequently dissected and harvested at the end of the experiments, formalin-fixed, and stained with haematoxylin and eosin (H&amp;E) using standard techniques.</p>", "<p><italic>Small animal</italic> in vivo <italic>imaging</italic> (a) Bioluminescence imaging (BLI): Xenograft tumour growth of H1975-luc cells were monitored by non-invasive luciferase-bioluminescence molecular imaging. Mice were imaged by BLI using a Xenogen IVIS® 200 bioluminescence scanner (Xenogen, Hopkinton, MA, USA) at indicated times on the pretreatment day as baseline, and then on various post-TKI treatment days as specified (details see also <xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Materials and Methods</xref>). (b) MicroPET/magnetic resonance imaging (MRI) imaging: For microPET/MRI imaging study, H1975 tumour xenografts were allowed to grow to a readily visible size in a total of 7 days post-implantation to ensure adequate baseline micro-PET uptake. H1975 tumour xenografts were treated with (a) diluent control and (b) SU11274 (100 <italic>μ</italic>g per xenograft). The nude mice were subjected to MRI (Bruker Biospec 7T MRI scanner, Bruker BioSpin MRI, Billerica, MA, USA) and microPET scanning (R4 micro-PET system, Siemens Medical Solutions, Knoxville, TN, USA) at the indicated time of treatment (details see also <xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Materials and Methods</xref>).</p>", "<title>siRNA inhibition</title>", "<p>Specific siRNAs targeting human <italic>MET</italic> or <italic>EGFR</italic> mRNA, ON-TARGET plus SMARTpool, were purchased from Dharmacon Inc. (Chicago, IL, USA). The siRNA duplexes were transiently transfected using DharmaFECT 1 Transfection reagent (Dharmacon Inc.) according to the manufacturer's instructions. Control transfection using scrambled siRNA was performed in parallel using ON-TARGETplus siCONTROL siRNA (Dharmacon Inc.).</p>", "<title>Statistical analysis</title>", "<p>Statistical significance was tested by two-tailed Student's <italic>t</italic>-test, with <italic>P</italic>-value of less than 0.05 considered statistically significant.</p>" ]

[ "<title>Results</title>", "<title>Co-expression of MET and EGFR in lung cancer</title>", "<p>We first examined the expression pattern of MET and EGFR in lung cancer cell lines. Nine of 11 NSCLC cell lines (82%) (except H520 and H661) (see <xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Table 1</xref> for baseline characteristics of the cell lines) co-expressed both MET and EGFR, including the H1975 cell line (##FIG##0##Figure 1##). Signal transducer and activator of transcription 3 (STAT3) is a common downstream signalling target for both MET and EGFR, and has been shown to be crucial in mediating the oncogenic potential of mutant EGFR signalling (##REF##12833138##Song et al, 2003##). STAT3 was ubiquitously expressed in all the cell lines examined.</p>", "<title>SU11274 induces apoptosis and inhibition of cytoskeletal functions in erlotinib-resistant H1975 lung cancer cells expressing L858R/T790M-EGFR</title>", "<p>As we found that many NSCLC cell lines co-express EGFR and MET, including H1975 cells, we first investigated if MET inhibition using the small-molecule inhibitor SU11274 can be effective in the erlotinib-resistant H1975 cells. H1975 cell line was chosen because it expresses the ‘gatekeeper’-resistant T790M-<italic>EGFR</italic> mutation (in-<italic>cis</italic> with L858R) in the receptor kinase domain hydrophobic pocket, representing a major mechanism of resistance to reversible EGFR-TKI (erlotinib/gefitinib) (##REF##15728811##Kobayashi et al, 2005a##; ##REF##15737014##Pao et al, 2005##). SU11274 was previously characterised as a reversible inhibitor of MET, inhibiting specific tyrosine phosphorylation of the juxtamembrane CBL-binding phosphosite (pY1003), the major kinase autophosphorylation sites (pY1234/1235), as well as downstream signalling (##REF##14500382##Sattler et al, 2003##; ##REF##15735036##Ma et al, 2005a##). It exhibits &gt;60-fold selectivity for MET over FLK and &gt;400-fold selectivity over RON, FGFR-1, SRC, CDK2, PDGFR-<italic>β</italic>, EGFR, and Tie-2 (##REF##15735036##Ma et al, 2005a##). Here, we tested the pro-apoptotic effect of SU11274 treatment, in comparison to erlotinib, in the EGFR-TKI-resistant cell line H1975 (L858R/T790M-<italic>EGFR</italic>, wild-type <italic>KRAS</italic>). SU11274 at 2 <italic>μ</italic><sc>M</sc> induced apoptosis (Annexin V- and propidium iodide-stained positive cells) in 14.8±2.4% of T790M-EGFR expressing H1975 cells, which is 5.5-fold (<italic>P</italic>&lt;0.001) higher than diluent control and 3.9-fold (<italic>P</italic>=0.0015) higher than erlotinib. To further demonstrate that the pro-apoptotic effect of SU11274 seen above was not a result of off-target effects, we also tested the EGFR-negative and MET-negative H520 cell line as negative control. SU11274 at 5 <italic>μ</italic><sc>M</sc> did not result in any significant apoptosis in H520 cells (0.44±0.30%, <italic>P</italic>=0.22). Similarly, H520 cells were also insensitive to erlotinib without any significant apoptosis induced by the drug (0.3±0.1%, <italic>P</italic>=0.35) (##FIG##1##Figure 2##).</p>", "<title>SU11274 induces cytoreduction of erlotinib-resistant H1975 tumour xenograft <italic>in vivo</italic></title>", "<p>To further test the role of MET inhibition in EGFR-TKI-resistant lung cancer <italic>in vivo</italic>, we developed stable luciferase-expressing H1975 lung cancer cells using lentivirus transduction. These cells were used in an <italic>in vivo</italic> xenograft model coupled with multimodal molecular imaging for non-invasive monitoring of xenograft growth and tumour response to TKI. Daily treatment with the MET inhibitor SU11274 caused statistically significant interval retardation of the xenograft tumour growth of H1975 cells with a ninefold reduction (<italic>P</italic>=0.0251) in the xenograft growth, when compared with the diluent control, during the treatment period (##FIG##2##Figure 3A-a,b##). At the end of treatment period, SU11274-treated H1975 xenograft tumour BLI flux remained essentially unchanged at 104% (<italic>P</italic>=0.0251), when compared to 905% as seen in the diluent control (##FIG##2##Figure 3A-a,b##). Histological analysis of the tumour xenografts harvested at the end of the experiment confirmed the presence of intense tumour necrosis in the SU11274-treated animals, but not in the diluent control (##FIG##2##Figure 3A-c##). Both EGFR and MET signal pathways are functional and ligand-sensitive in the erlotinib-resistant H1975 cells. HGF stimulated downstream signal path activation in AKT (survival) and the mitogen-activated protein kinase ERK1/2 (proliferation–differentiation), as surrogate markers for MET inhibition were both abrogated in the H1975 cells by SU11274 treatment <italic>in vitro</italic> (##FIG##2##Figure 3A-d##). Furthermore, we also showed that SU11274 inhibition effectively induced erlotinib-resistant A549 xenograft cytoreduction <italic>in vivo</italic> (<xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Figure 1</xref>).</p>", "<title>SU11274 inhibition induces early tumour response of the H1975 <italic>in vivo</italic> xenograft evident in microPET/MRI studies</title>", "<p>We focused further on the H1975 cell line to investigate whether MET inhibition with SU11274 induced H1975 xenograft tumour response in terms of glucose metabolism as monitored by <italic>in vivo</italic> FDG-PET (glucose analogue [¹⁸F]fluoro-2-deoxy-D-glucose-positron emission tomography) studies with MRI co-registration. <italic>In vivo</italic> SU11274 inhibition induced a metabolic tumour response in H1975 xenografts within 24 h of the first treatment dose (##FIG##2##Figure 3B and C##). Although the changes of the calculated xenograft tumour volumes between the two treatment groups did not differ significantly (<italic>P</italic>&gt;0.05) (##FIG##2##Figure 3B##), the SU11274-treated xenografts had statistically significant lower glucose metabolism by 45% (<italic>P</italic>=0.0226), when compared to diluent control (##FIG##2##Figure 3B and C##) (also see <xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Figure 2</xref>).</p>", "<title>MET–EGFR signalling cross-activation in lung cancer</title>", "<p>As MET and EGFR often co-express in lung cancer cells (##FIG##0##Figure 1##), we asked if there is signalling cross-activation between MET and EGFR pathways. Both A549 (wild-type <italic>EGFR</italic>) and H1975 (L858R/T790M-<italic>EGFR</italic>) cell lines were used as models for the signalling studies (##FIG##3##Figure 4##). Here, MET and EGFR signal transduction pathways were both shown to be functional and ligand-sensitive, although H1975 cells have higher serum-independent constitutively activated MET and EGFR. Enhanced and more durable downstream signalling activation was observed in phospho-AKT (survival), phospho-ERK1/2 (proliferation-differentiation), and phospho-STAT3 (transcriptional activation) when A549 and H1975 cells were co-stimulated with dual-ligand (HGF and EGF) (##FIG##3##Figure 4A##; see lanes 4, 7, 11, and 14). Immunoprecipitation studies of the MET and EGFR under single- or dual-ligand stimulation confirmed the presence of receptor cross-activation between MET and EGFR in these lung cancer cell lines (##FIG##3##Figure 4B##). In A549 cells, HGF was capable of activating EGFR in the presence of EGF, whereas in H1975, EGF activated MET with and without co-stimulation with HGF.</p>", "<title><italic>MET</italic> is activated with no genomic amplification or mutations in H1975 lung adenocarcinoma cells</title>", "<p>Using standard QPCR technique, we determined the <italic>MET</italic> genomic copy number in several NSCLC cell lines, namely A549, H1975, H441, H520, H596, and H661. None of the cell lines examined exhibited <italic>MET</italic> genomic amplification. <italic>MET</italic> genomic copy number in H1975 was found to be 1.1, whereas that of A549 being 1.0 for comparison (##FIG##3##Figure 4B##). Direct DNA sequencing of the <italic>MET</italic> gene in H1975 did not reveal any non-synonymous mutations.</p>", "<title>Dual inhibition with SU11274 plus erlotinib/CL-387,785 potentiates the MET-targeted inhibitory efficacy in erlotinib-resistant lung cancer cells</title>", "<p>As both MET and EGFR are functional in erlotinib-resistant cell lines such as A549 and H1975 cells, and there is signalling cross-activation between the two receptors (##FIG##3##Figure 4##), we next investigated if MET inhibition could be enhanced in combination with an EGFR inhibitor. The EGFR-TKI-resistant H1975 cells were tested against SU11274 inhibition, either alone or in combination with erlotinib. We determined the functional effect of combined MET–EGFR inhibition on the cytoskeletal functions of H1975 cells. Cell motility and migration are crucial cellular regulatory functions in tumour cell invasion and metastasis. Video microscopy studies showed that SU11274 inhibition substantially abrogated the constitutively activated cytoskeletal changes of the serum-starved H1975 cells (with activated p-MET and p-EGFR), as reflected in the cellular migrational trajectories and velocity (<xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Figure 3</xref>). Moreover, SU11274 alone had some inhibitory effect on cytoskeletal functions in H1975 cells under serum-stimulated (10% FBS) conditions, whereas concurrent dual TKI combinatorial inhibition using SU11274 plus erlotinib completely abrogated the cytoskeletal functions (##FIG##4##Figure 5A##).</p>", "<p>Using dual-ligand concurrent stimulation (HGF and EGF) to activate both MET and EGFR in A549 and H1975 cells, single TKI alone was relatively ineffective in inhibiting the downstream signalling completely (##FIG##4##Figure 5B##). On the other hand, concurrent dual inhibition with MET–EGFR TKIs (SU11274 plus erlotinib) effectively induced cooperative and enhanced inhibition of the key downstream proliferative/survival and anti-apoptotic signal paths (phospho-AKT, phospho-ERK1/2, and phospho-STAT3) (##FIG##4##Figure 5B##). Most interestingly, despite the fact that p-EGFR[Y1068] was not significantly inhibited under the dual SU11274/erlotinib combinatorial treatment in H1975 cells, p-ERBB3[Y1289] activation was effectively abrogated only under this dual inhibitory strategy (##FIG##4##Figure 5B##).</p>", "<p>The signalling experiment using <italic>MET</italic>-specific siRNA instead of SU11274 in A549 cells (##FIG##4##Figure 5C##) showed similar inhibition synergism using dual RTK inhibition with siRNA-<italic>MET</italic> and erlotinib. Using the A549 cell line as a model, we further demonstrated that alternative ligand-stimulated RTK signalling (MET–HGF and EGFR–EGF) indeed could rescue the downstream signalling activation from single targeted inhibitor (##FIG##4##Figure 5D##), supporting our hypothesis of the optimal efficacy of dual MET–EGFR inhibition especially in the <italic>in vivo</italic> setting where both RTK signal paths are functional and activated. Irreversible EGFR-TKIs (##REF##16103058##Kobayashi et al, 2005b##), such as CL-387,785 or HKI-272 (##REF##18056464##Ruhe et al, 2007##), have been shown to exhibit inhibitory efficacy against erlotinib-resistant T790M mutation. Here, in H1975 cells, results similar to erlotinib were obtained using the irreversible EGFR-TKI, CL-387,785, in combination with SU11274 (##FIG##4##Figure 5E##).</p>", "<p>As erlotinib is the only FDA-approved clinical EGFR-targeted inhibitor for the treatment of advanced lung cancer in the United States, we further investigated its potential use in combination with MET inhibition against erlotinib-resistant H1975 cells using an <italic>in vitro</italic> cellular viability assay and BLI tumour xenograft growth assay <italic>in vivo</italic>. Under serum-stimulated conditions, when both EGFR and MET were basally activated, SU11274 plus erlotinib (3 <italic>μ</italic><sc>M</sc> of each TKI) induced a significantly enhanced cell viability inhibition (3 <italic>μ</italic><sc>M</sc>: 58.0±6.8%, <italic>P</italic>&lt;0.05), when compared with either erlotinib or SU11274 treatment alone (##FIG##5##Figure 6A##). To further validate this combination dual TKI strategy, we also subjected the H1975 cells to pathway-specific siRNA knockdown of the MET and EGFR kinase signal paths, either alone or in combination, under serum-stimulated conditions (##FIG##5##Figure 6B##). Optimal downstream signalling inhibition (phospho-AKT and phospho-STAT3) and global phosphotyrosine (p-Tyr) cellular signalling inhibition were achieved through dual inhibition with siRNA knockdown of both <italic>MET</italic> and <italic>EGFR</italic> targets, when compared with single target knockdown. Taken together, these data provide support that cooperative enhanced inhibition using dual TKIs against MET and EGFR pathways may be an effective treatment strategy to inhibit lung cancer with intrinsic or acquired T790M-EGFR-mediated TKI resistance.</p>", "<p>Finally, we also tested concurrent dual SU11274 plus erlotinib combinatorial treatment in our <italic>in vivo</italic> H1975-luc BLI xenograft growth assay (##FIG##5##Figure 6C##). Here, a suboptimal daily dose of SU11274 (50 <italic>μ</italic>g per xenograft per day) was found to be partially effective in retarding the xenograft growth, when compared with either diluent control or erlotinib (100 mg kg⁻¹ day⁻¹) alone. Moreover, we found that combining SU11274 with erlotinib induced a complete tumour xenograft regression (0.12-fold BLI from baseline), evident within 2 weeks of dual inhibitor therapy. The difference seen with dual MET–EGFR-TKIs treatment is statistically significant when compared with either erlotinib alone (35.8-fold BLI increase, <italic>P</italic>=0.0006) or with SU11274 alone (12.2-fold, <italic>P</italic>=0.0003). These molecular imaging data were confirmed by xenograft H&amp;E-stained histology examination (##FIG##5##Figure 6D##). In particular, the combined SU11274 plus erlotinib treatment resulted in substantial tumour necrosis.</p>" ]

[ "<title>Discussion</title>", "<p>At least half of the acquired resistance to EGFR-TKI in advanced NSCLC patients is thought to be mediated by the ‘gatekeeper’ mutation T790M in exon 20 of <italic>EGFR</italic> (##REF##15728811##Kobayashi et al, 2005a##; ##REF##15737014##Pao et al, 2005##). Many tumours with intrinsic resistance to erlotinib/gefitinib were found to have wild-type <italic>EGFR</italic> and/or mutant <italic>KRAS</italic>. At present, no FDA-approved inhibitor drugs have been shown to be successful in overcoming T790M-mediated resistance clinically. Recent study suggested that T790M-EGFR-mediated resistance could even emerge from the irreversible EGFR/ERBB2 inhibitor HKI-272 treatment at maximally tolerated dosing, as it mediates resistance to low concentrations of the irreversible inhibitor (##REF##18413800##Godin-Heymann et al, 2008##). Alternative novel therapies to target lung cancer patients with intrinsic or acquired resistance to erlotinib are needed. MET has recently been affirmed to be an attractive anti-neoplastic therapeutic target (##REF##15949770##Corso et al, 2005##; ##REF##16778093##Peruzzi and Bottaro, 2006##; ##REF##17143251##Salgia, 2006##), including lung cancer (##REF##14612533##Christensen et al, 2003##, ##REF##15922853##2005##; ##REF##14559814##Ma et al, 2003a##, ##REF##12884908##2003b##, ##REF##15735036##2005a##, ##REF##15788682##2005b##; ##REF##14500382##Sattler et al, 2003##). MET was found overexpressed in up to 67% of lung adenocarcinomas in our previous study (##REF##15735036##Ma et al, 2005a##). Various targeted inhibitory strategies are being undertaken in drug development to antagonise MET/HGF signalling in human cancers, including small-molecule kinase inhibitors, antibodies to the ligand HGF, and receptor MET itself (##REF##15922853##Christensen et al, 2005##; ##REF##17062691##Martens et al, 2006##; ##REF##16778093##Peruzzi and Bottaro, 2006##). In this study, we identified that there is frequent co-expression of MET and EGFR in NSCLC cell lines. In the erlotinib-resistant H1975 cell line (L858R/T7980M-<italic>EGFR</italic>), <italic>MET</italic> is neither genomically amplified nor mutated. Yet, MET is activated in the cells, possessing both constitutive (serum/ligand-independent) and basal (serum-stimulated) receptor activation. Furthermore, MET also remains HGF ligand-sensitive. Owing to the unique intrinsic properties of MET regulating cellular ‘invasive signalling’, MET has been proposed as not merely playing a role in ‘oncogene addiction’ in a small subset of human cancers but can also play an essential role in ‘oncogene expedience’ by inducing an enhanced transformed tumour malignant ‘fitness’ in a much larger range of cancers leading to promotion of tumour progression (##REF##18511928##Comoglio et al, 2008##). And in the latter case, activated MET can intercept with various other oncogenic signals, including mutant-EGFR, in maintaining and enhancing the tumour invasive–progressive phenotype, thereby also allowing the opportunity for MET to be a therapeutic target even in late advanced metastatic disease. The MET inhibitor SU11274 was shown to promote apoptosis in H1975 cells, but was ineffective in the MET-negative/EGFR-negative H520 cells. SU11274 was previously characterised to be a selective, reversible ATP-competitive inhibitor of MET kinase (##REF##14500382##Sattler et al, 2003##; ##REF##15735036##Ma et al, 2005a##). Here, we show that SU11274 exhibited inhibitory efficacy in the EGFR-TKI-resistant H1975 cells both <italic>in vitro</italic> and <italic>in vivo</italic>. In particular, it inhibited MET signalling, induced cellular apoptosis, and abrogated cytoskeletal functions (key controlling step in tumour invasion and metastasis) <italic>in vitro</italic> (##REF##14685170##Birchmeier et al, 2003##), and was effective <italic>in vivo</italic>, leading to cytoreduction of murine tumour xenografts of the T790M-EGFR expressing erlotinib-resistant H1975 cells. Taken together, our study supports the hypothesis that MET may be targeted to circumvent T790M-EGFR-mediated intrinsic or acquired resistance to EGFR-TKI (erlotinib) in lung cancer.</p>", "<p>H1975 cells were tested further to provide a better understanding of the mechanism of dual MET–EGFR inhibition. A549 cell line was included as model in signalling studies for comparison with H1975 cells. A549 is an extensively studied lung cancer cell line and is known to have <italic>KRAS</italic> mutation but not any <italic>EGFR</italic> kinase domain mutations. In our <italic>in vitro</italic> signalling studies, we identified that there was signalling cross-activation between MET and EGFR in both A549 and H1975 cells. Nonetheless, the pattern of cross-activation appeared to be different between the two cell lines. In A549 cells, HGF brought about cooperative induction of p-EGFR[Y1068] in the presence of EGF, whereas in H1975 cells, EGF induced cross-activation of p-MET[Y1234/1235] by itself and further MET activation when combined with HGF in dual ligand stimulation. It is tempting to postulate that the different pattern of cross-activation observed in A549 and H1975 cells might be a result of the different <italic>EGFR</italic> kinase mutational status in the two cell lines, that is non-mutated in A549 but L858R/T790M in H1975. The mutant EGFR in H1975 evidently is capable of cross-activating MET in an EGF –ligand-dependent manner, indicating that MET could be ‘downstream’ of the mutant EGFR in H1975. Of interest, it has recently been shown that the MET receptor activating phosphorylation site was highly responsive to EGFRvIII levels in glioblastoma cells <italic>in vitro</italic>, suggesting downstream cross-activation of MET by mutant EGFRvIII (##REF##17646646##Huang et al, 2007##).</p>", "<p>L858R/T790M-<italic>EGFR</italic> mutations exist in H1975 cells in-<italic>cis</italic> (##REF##15737014##Pao et al, 2005##). The double mutations not only confer resistance to gefitinib/erlotinib but also result in markedly enhanced catalytic kinase and oncogenic activity (##REF##17332364##Mulloy et al, 2007##). Emerging evidence suggests that the T790M ‘gatekeeper’ mutation may exist in lung tumours before EGFR-TKI therapeutic selection (##REF##16258541##Bell et al, 2005##; ##REF##17671201##Godin-Heymann et al, 2007##), partly due to its enhanced oncogenicity, and accounts for the adverse clinical course and outcome in gefitinib/erlotinib-resistance lung cancers after a course of rapid TKI selection. Interestingly, we found that H1975 cells co-express EGFR and MET at high level, although without any <italic>MET</italic> genomic amplification, and were capable of serum-independent constitutive MET activation. Moreover, MET and EGFR engage in collaborative signalling cross-activation to transduce stronger and more durable downstream signals. A recent report utilised a mutant <italic>EGFR</italic> (deletion 19)-expressing NSCLC cell line HCC827, which is highly sensitive to gefitinib, an <italic>in vitro</italic> gefitinib-resistance long-term inhibition culture system to select for gefitinib-resistant cell subclones (##REF##17463250##Engelman et al, 2007##). Acquired <italic>MET</italic> amplification was identified to be a potential alternative mechanism to enable the mutant <italic>EGFR</italic>-expressing HCC827 to become secondarily resistant (HCC827-GR) without a T790M mutation. In the course of our study, ##REF##18093943##Bean et al (2007)## reported the presence of <italic>MET</italic> amplification that occurred independently with and without T790M-<italic>EGFR</italic> mutation in lung tumours. <italic>MET</italic> amplification in gastric cancer cell lines has recently been correlated with high sensitivity towards MET inhibitor (##REF##16461907##Smolen et al, 2006##). Interestingly, another recent report adopted a global phosphoproteomic approach using cell lines sensitive to gefitinib (HCC827) and sensitive to SU11274 (MKN45), and showed that besides p-EGFR inhibition, gefitinib also inhibited p-MET (that is indeed constitutively activated) in HCC827, but not vice versa (##REF##18180459##Guo et al, 2008##). In the case of SU11274, the MET inhibitor inhibited p-EGFR in MKN45 besides p-MET, but not vice versa (##REF##18180459##Guo et al, 2008##). Further studies to examine the signalling cross-talk between the EGFR and MET receptor pathways in the context of mutations and downstream signalling networking would be helpful in optimising combinational therapeutic strategies. Moreover, it would be useful to further investigate and catalogue the activating mechanisms of <italic>MET</italic> (such as activating mutations, transcriptional and protein overexpression), and their role in oncogenic signalling in the context of multiple receptor co-activation in lung cancer.</p>", "<p>Our report here demonstrates the efficacy of dual RTK targeted inhibition against MET (SU11274) and EGFR (erlotinib or CL-387,785) as a strategy to achieve optimised inhibition of cytoskeletal functions, cell viability, cellular signalling, and <italic>in vivo</italic> xenograft complete regression in T790M-EGFR-mediated erlotinib resistance. The concentrations of SU11274 used in our current study are consistent with previous results to be within the range of selectivity towards MET (##REF##14500382##Sattler et al, 2003##; ##REF##15735036##Ma et al, 2005a##). Finally, our dual siRNA knockdown experiment against <italic>MET</italic> and <italic>EGFR</italic> (##FIG##5##Figure 6B##) in H1975 cells provides further validation of this novel therapeutic approach. Dual inhibition may be of benefit over single target inhibition, especially in the context of serum and/or alternative ligand stimulation. This can be of clinical relevance considering that tumour cells often exist and adapt <italic>in vivo</italic> under a multitude of host stromal conditions during various stages of tumour progression, including serum starvation (e.g. in tumour core), serum-stimulation, and also potentially microenvironment-specific ligand(s)-stimulation.</p>", "<p>Dual TKI combinatorial approach may allow more effective target inhibition with a lower MET inhibitor concentration requirement, when compared with monotherapy alone, as suggested in our <italic>in vivo</italic> H1975-luc xenograft study. The ability to use lower drug concentrations than that in monotherapy would be beneficial and clinically relevant to minimise additive toxicity profiles of two inhibitor drugs of similar class used in combination. It is intriguing that we consistently observed a modest, but readily detectable degree of potentiated inhibition of MET phosphorylation by erlotinib, with and without SU11274, both in A549 and in H1975 cells. In addition, our siRNA-<italic>EGFR</italic> knockdown study in H1975 cells resulted in appreciable downregulation of p-MET[Y1234/1235], suggesting that the mutant EGFR in H1975 cells might signal into MET as a ‘downstream’ cross-talk collaborative signal partner (##FIG##5##Figure 6B##). It may account for some of the enhanced inhibitory effects seen in the dual TKI treatment. In our <italic>in vivo</italic> bioluminescence xenograft model, combined SU11274–erlotinib inhibition remarkably induced complete H1975-luc tumour xenograft regression associated with histological features of massive tumour necrosis–apoptosis. We were initially intrigued that despite the lack of effective inhibition of EGFR phosphorylation (at pY1068) by erlotinib alone, the drug enhanced inhibitory efficacy both <italic>in vitro</italic> and <italic>in vivo</italic> in H1975 cells when combined with SU11274. ##REF##17463250##Engelman et al (2007)## recently reported that <italic>MET</italic>, when amplified genomically as in the setting of acquired EGFR-TKI resistance, can capture the ERBB3 signal control from EGFR. Our data here suggest that MET inhibition (SU11274) in H1975 cells had a modest but detectible negative effect on ERBB3 activation, and that dual SU11274/erlotinib inhibition cooperatively abrogated p-ERBB3 signal activation completely (##FIG##4##Figure 5B##). This might partially explain the observed role of erlotinib in the dual inhibitory strategy against the T790M-<italic>EGFR</italic> mutant cells, even though it is ineffective against p-EGFR itself. Further studies to dissect the interplay between MET and ERBB3 signal paths in lung cancer are warranted. Besides the possible effect of erlotinib upon MET activation <italic>per se</italic>, one might not rule out other potential off-target effects of erlotinib as contributing factors. ##REF##15998836##Stegmaier et al (2005)## first reported a previously unrecognised EGFR-independent mechanism of gefitinib in inducing the differentiation and inhibiting proliferation of EGFR-negative acute myeloid leukaemia cells at clinically achievable doses (##REF##15998836##Stegmaier et al, 2005##). More recently, erlotinib was also found to exhibit off-target anti-neoplastic effects in acute myeloid leukaemia and myelodysplastic syndrome, supporting the potential clinical therapeutic utility of these EGFR-TKIs in haematologic malignancies (##REF##17925489##Boehrer et al, 2007##). As erlotinib is currently the only clinically approved EGFR-TKI for lung cancer in the United States, discovery of its utility in combinatorial inhibitory approaches would be of potential clinical benefit. Combination strategy to target EGFR and MET has recently been reported to show promise in overcoming mutant-EGFRvIII-driven glioblastoma, although using a higher concentration range of MET inhibitor (##REF##17646646##Huang et al, 2007##). Some success to overcome T790M-mutant EGFR resistance has been reported using irreversible EGFR/ERBB family inhibitors, such as CL-387,785 and HKI-272 (##REF##16103058##Kobayashi et al, 2005b##; ##REF##17671147##Wong, 2007##). Nonetheless, recent work in a L858R/T790M-<italic>EGFR</italic> transgenic mouse model suggests that the double mutant-<italic>EGFR</italic> responds only partially to HKI-272 alone (##REF##17613438##Li et al, 2007##), and enhanced inhibition was seen in combination with mammalian target of rapamycin inhibitor.</p>", "<p>In conclusion, the present study identified that despite having no genomic <italic>MET</italic> alterations in H1975 cells, the MET–HGF signal path is functional and activated in this EGFR-TKI-resistant cell line that already expresses the oncogenic mutant EGFR (L858R/T790M) signal axis. There are also receptor cross-activation and signalling circuitry cross-talk between MET and EGFR, and MET inhibition has efficacy <italic>in vitro</italic> and <italic>in vivo</italic> in the erlotinib-resistant H1975 cells. Our results also implicate that combination treatment using a MET inhibitor plus a reversible or irreversible EGFR kinase inhibitor to achieve dual MET–EGFR inhibition may represent an alternative strategy to circumvent T790M-<italic>EGFR</italic>-mediated resistance in lung cancer. Importantly, erlotinib may still have clinical utility in this context of combined inhibition with MET inhibitor in EGFR-TKI-resistant lung cancer. As T790M-EGFR may play a role in both intrinsic and acquired EGFR-TKI resistance in lung cancer, it would be useful to test concurrent combinatorial MET–EGFR inhibitors in clinical trials on lung cancer patients refractory to erlotinib/gefitinib. Other emerging pan-ERBB class inhibitors, such as PF00299804, or EGFR-targeting dual/multitargeted inhibitors, such as lapatinib (EGFR/ERBB2 inhibitor) or ZD6474 (EGFR/VEGFR2 inhibitor), might also be candidates for combination with MET inhibitors. Whether a combined MET–EGFR inhibitory strategy as upfront treatment is superior to MET inhibition used only after EGFR-TKI monotherapy failure should be the subject of further investigation.</p>" ]

[]

[ "<p>These authors contributed equally to the work.</p>", "<p>Despite clinical approval of erlotinib, most advanced lung cancer patients are primary non-responders. Initial responders invariably develop secondary resistance, which can be accounted for by T790M-<italic>EGFR</italic> mutation in half of the relapses. We show that MET is highly expressed in lung cancer, often concomitantly with epidermal growth factor receptor (EGFR), including H1975 cell line. The erlotinib-resistant lung cancer cell line H1975, which expresses L858R/T790M-EGFR in-<italic>cis</italic>, was used to test for the effect of MET inhibition using the small molecule inhibitor SU11274. H1975 cells express wild-type <italic>MET</italic>, without genomic amplification (CNV=1.1). At 2 <italic>μ</italic><sc>M</sc>, SU11274 had significant <italic>in vitro</italic> pro-apoptotic effect in H1975 cells, 3.9-fold (<italic>P</italic>=0.0015) higher than erlotinib, but had no effect on the MET and EGFR-negative H520 cells. <italic>In vivo</italic>, SU11274 also induced significant tumour cytoreduction in H1975 murine xenografts in our bioluminescence molecular imaging assay. Using small-animal microPET/MRI, SU11274 treatment was found to induce an early tumour metabolic response in H1975 tumour xenografts. MET and EGFR pathways were found to exhibit collaborative signalling with receptor cross-activation, which had different patterns between wild type (A549) and L858R/T790M-EGFR (H1975). SU11274 plus erlotinib/CL-387,785 potentiated MET inhibition of downstream cell proliferative survival signalling. Knockdown studies in H1975 cells using siRNA against <italic>MET</italic> alone, EGFR alone, or both, confirmed the enhanced downstream inhibition with dual MET–EGFR signal path inhibition. Finally, in our time-lapse video-microscopy and <italic>in vivo</italic> multimodal molecular imaging studies, dual SU11274-erlotinib concurrent treatment effectively inhibited H1975 cells with enhanced abrogation of cytoskeletal functions and complete regression of the xenograft growth. Together, our results suggest that MET-based targeted inhibition using small-molecule MET inhibitor can be a potential treatment strategy for T790M-EGFR-mediated erlotinib-resistant non-small-cell lung cancer. Furthermore, optimised inhibition may be further achieved with MET inhibition in combination with erlotinib or an irreversible EGFR-TKI.</p>" ]

[ "<p>Receptor tyrosine kinases (RTKs) play a key role in lung cancer tumorigenesis and progression (##REF##15948672##Choong et al, 2005##). Progress has been made in the treatment of advanced non-small-cell lung cancer (NSCLC) using small-molecule tyrosine kinase inhibitors (TKIs) gefitinib and erlotinib, targeting epidermal growth factor receptor (EGFR) (##UREF##0##Lynch et al, 2006##). <italic>EGFR</italic> kinase domain mutations (frequently L858R) and exon 19 deletions have been identified to be predictive of response to gefitinib/erlotinib (##REF##16231326##Shigematsu and Gazdar, 2006##; ##REF##17318210##Sharma et al, 2007##). Although erlotinib was shown to prolong survival in a large phase III randomised trial (NCIC-BR.21) (##REF##16014882##Shepherd et al, 2005##), the majority of unselected lung cancer patients are still primary non-responders. Patients whose cancer has wild-type <italic>EGFR</italic> genotype are generally non-responders but may at best derive stable disease from the TKIs. Initial responders with mutant <italic>EGFR</italic> invariably develop secondary resistance and soon succumb to the disease. At least half of the acquired resistance is mediated by the ‘gatekeeper’ mutation T790M-<italic>EGFR</italic> (##REF##15728811##Kobayashi et al, 2005a##; ##REF##15737014##Pao et al, 2005##). Moreover, T790M was found in the H1975 cell line, in combination with L858R, which was previously established without prior exposure to EGFR TKIs. Hence, T790M may also have a role in primary EGFR-TKI resistance. Currently, there are still no Federal Drug Administration (FDA)-approved clinical inhibitors that can overcome T790M-mediated EGFR-TKI resistance yet.</p>", "<p>The MET receptor has been shown to be an important molecule in a variety of malignancies (##REF##7505952##Tsarfaty et al, 1994##; ##REF##9140397##Schmidt et al, 1997##, ##REF##9563489##1998##, ##REF##10327054##1999##; ##REF##14685170##Birchmeier et al, 2003##; ##REF##12884908##Ma et al, 2003b##; ##REF##17607709##Benvenuti and Comoglio, 2007##), and has recently been validated as an attractive therapeutic target in cancer therapy, including lung cancer (##REF##14559814##Ma et al, 2003a##, ##REF##15735036##2005a##, ##REF##15788682##2005b##; ##REF##15922853##Christensen et al, 2005##; ##REF##17143251##Salgia, 2006##). Overexpression of MET or its ligand HGF have been found to confer a poor prognosis (##REF##12884908##Ma et al, 2003b##; ##REF##16914575##Miyata et al, 2006##; ##REF##17242702##Garcia et al, 2007##; ##REF##17308108##Sawada et al, 2007##). Dysregulation of the MET–HGF signalling axis upregulates diverse tumour cell functions, including cell proliferation, survival, cell scattering and motility, epithelial-mesenchymal transition, angiogenesis, invasion, and metastasis (##REF##15949770##Corso et al, 2005##; ##REF##16050800##Dietrich et al, 2005##; ##REF##15735036##Ma et al, 2005a##; ##REF##16778093##Peruzzi and Bottaro, 2006##). Reversible small-molecule inhibitors such as SU11274 targeting MET have been developed for therapeutic inhibition (##REF##14612533##Christensen et al, 2003##; ##REF##14500382##Sattler et al, 2003##; ##REF##15735036##Ma et al, 2005a##, ##REF##15788682##2005b##). We hypothesised that MET signalling plays a key role in lung cancer oncogenic signalling and optimised therapy targeting MET would be effective as a treatment strategy in the face of EGFR-TKI resistance. In this study, we sought to define the role of MET signalling in EGFR-TKI-resistant lung cancer. Furthermore, using a combination of small-molecule kinase inhibitors and short-interfering RNA (siRNA), we examined the role of MET inhibition, either alone or combined with EGFR inhibition, using both <italic>in vitro</italic> and <italic>in vivo</italic> assays against the EGFR-TKI-resistant lung cancer cell line H1975 (L858R/T790M-mutant EGFR). Our data support the potential role of dual TKI combinatorial inhibition using EGFR inhibitors to enhance MET inhibition in T790M-EGFR-mediated therapy resistance.</p>" ]

[ "<p>This work was supported by NIH/National Cancer Institute-K08 Mentored Career Development Award (1K08CA102545-01A2), Ohio Cancer Research Associates ‘Give New Ideas A Chance’ Grant Award (both to PC Ma), NIH HL075422 and VA Merit Award (both to JA Kern), the Case Comprehensive Cancer Center (Genotype and Gene Expression Core, Confocal Microscopy Core, Xenograft and Athymic Animal Core, and Animal Imaging Core Facilities) of CWRU/University Hospitals Case Medical Center (P30 CA43703-12), and Northern-East Ohio Small Animal Imaging Resource Program (R24 CA110943). We also thank Drs Sanford Markowitz and David Danielpour (Case Western Reserve University) for helpful discussions.</p>", "<title>Supplementary Material</title>" ]

[ "<fig id=\"fig1\"><label>Figure 1</label><caption><p>Co-expression pattern of MET and EGFR in lung cancer. The expression pattern of MET (1st panel) and EGFR (2nd panel) was examined using standard immunoblotting of the whole cell lysates (WCLs) from the following lung cancer cell lines cultured under serum-containing conditions (10% FBS): A549, H1975, H226, H441, H520, H596, H661, H1437, H1838, H2122, and SW900. The downstream signalling effector STAT3 (3rd panel) was also included in the immunoblot analysis. <italic>β</italic>-Actin was included as loading control (bottom panel).</p></caption></fig>", "<fig id=\"fig2\"><label>Figure 2</label><caption><p>MET inhibition with SU11274 in EGFR-TKI-resistant H1975 lung cancer cells: induction of apoptosis <italic>in vitro</italic> and inhibition of cytoskeletal functions. The MET kinase inhibitor SU11274 was used to treat H1975 cells (L858R/T790M-<italic>EGFR</italic>, wild-type <italic>KRAS</italic>) and H520 cells (negative expression for both <italic>EGFR</italic> and <italic>MET</italic>) as control. The effect of SU11274 was examined using Annexin-V/propidium iodide (PI)-FITC cellular apoptosis assay. Untreated diluent control (U) and erlotinib were included in the experiment as treatment controls for comparison. Erlotinib (EGFR inhibition) was ineffective in promoting apoptosis in any of these above cell lines at 72 h. On the other hand, MET inhibition by SU11274 at 2 <italic>μ</italic><sc>M</sc> induced significant cellular apoptosis in the EGFR-TKI-resistant H1975 cells (14.8±2.4%, <italic>P</italic>=0.0015), when compared with erlotinib (3.8±0.7%). For the EGFR-negative and MET-negative H520 cells, neither SU11274 (0.44±0.30%, <italic>P</italic>=0.22) nor erlotinib (0.28±0.13%, <italic>P</italic>=0.35) at 5 <italic>μ</italic><sc>M</sc> induced any significant cellular apoptosis when compared with diluent control (0.18±0.10%). Mean values of percent cells in early apoptosis (Annexin-V plus PI staining) from three independent experiments for each of the treatment conditions were plotted in the graphs shown here. Error bar, s.e.m. (<italic>N</italic>=3).</p></caption></fig>", "<fig id=\"fig3\"><label>Figure 3</label><caption><p>MET inhibition with SU11274 successfully induced <italic>in vivo</italic> tumour response in EGFR-TKI-resistant H1975 cells in murine xenograft model assessed by multimodal molecular imaging. (<bold>A</bold>) <italic>In vivo</italic> tumour xenografts for H1975-luc cells were established as described in the Materials and Methods section in 6-week old nude mice. Daily SU11274 (100 <italic>μ</italic>g per xenograft) treatment was administered to the H1975-luc lung cancer tumour xenografts in nude mice as described. DMSO diluent control was included for comparison. Imaging was performed using a Xenogen IVIS 200 System cooled CCD camera at indicated times. (a) Representative BLI digital pictures of nude mouse from each of the treatment conditions are illustrated. SU11274 significantly inhibited L858R/T790M-EGFR expressing H1975-luc <italic>in vivo</italic> tumorigenesis within the treatment durations (6 days). (b) Mean values of relative BLI flux of each group are plotted here (H1975-luc). <italic>N</italic>=4 per treatment group. Error bar, s.e.m. (^*), <italic>P</italic>=0.025 for H1975-luc. Representative tumour xenograft micrographs from H1975-luc (c) cell line under haematoxylin and eosin (H&amp;E) staining are also shown here for the control and SU11274 treatment animals. Magnification × 100 (inset, × 200). (d) SU11274 inhibited HGF-driven signalling activation in H1975 cells. H1975 cells were stimulated with HGF (50 ng ml⁻¹, 15 min) and inhibited by MET inhibitor SU11274 (1 <italic>μ</italic><sc>M</sc>, 4 h) <italic>in vitro</italic>, and analysed with 7.5% SDS–PAGE and immunoblotting with the indicated antibodies as described in Materials and Methods. (<bold>B</bold>, <bold>C</bold>) Magnetic resonance imaging (MRI) and microPET molecular imaging studies of MET inhibition of H1975 <italic>in vivo</italic> xenograft. H1975 <italic>in vivo</italic> xenografts were established as above for treatment with either diluent control (<italic>N</italic>=2) or SU11274 (<italic>N</italic>=2). The nude mice with H1975 xenografts were subjected to MRI and microPET imaging as described in the Materials and Methods section at 0, and 24 h with the MET inhibitor SU11274 treatment or diluent control. (<bold>B</bold>) Examples of the transverse sections of high-resolution MRI images of the tumour xenografts at baseline between the two treatment groups were shown here for illustration (left). The MRI tumour volumes were analysed digitally with the calculated tumour volume changes at the indicated time intervals (0 and 24 h) plotted. Comparing with baseline, the control xenograft tumour volume increased by 105.1±8.3% at 24 h, whereas the SU11274-treated xenografts increased by 120.2±16.2% at 24 h. The MRI tumour volumes changes at 24 h post-treatment between the two groups were not statistically significant (^*<italic>P</italic>=0.360). Error bar, s.e.m. Quantitative microPET radiotracer uptake of the H1975 tumour xenografts at 60 min of radiotracer tail-vein infusion in the animals' pretreatment baseline (0 h) and post-MET-TKI treatment at 24 h is shown graphically (right). <italic>N</italic>=2 in each treatment group: Control and SU11274. Error bar, s.e.m. Representative co-registered pictures of the microPET/MRI (low-resolution) images of each xenograft from the two treatment groups are shown in (<bold>C</bold>). SU11274 induced early tumour metabolic response, as early as 24-h post-TKI treatment, with statistically significant inhibition of glucose metabolism as evident in the decrease in microPET uptake signal intensity by 45% (<italic>P</italic>=0.0226) in SU11274-treated xenografts, when compared with diluent control. The degree of increase in the glucose uptake in the H1975 tumour xenograft in diluent control is also consistent with the average rate of the xenograft growth (increase of 50.8% BLI flux per day) as reflected in the bioluminescence imaging.</p></caption></fig>", "<fig id=\"fig4\"><label>Figure 4</label><caption><p>Signalling cross-activation between MET and EGFR signalling pathways. (<bold>A</bold>) Cross-activation between MET and EGFR signalling in lung cancer cells, A549 and H1975. A549 or H1975 cells were cultured in serum-starved conditions with exogenous stimulation with RTK ligands: EGF alone, HGF alone, or both EGF and HGF. Cells without any ligand stimulation were included as control. Both MET and EGFR signalling pathways are functional and ligand-sensitive in A549 and H1975 cells. There was augmented downstream signalling with combined EGF-HGF co-stimulation, with also more durable signalling induction. In A549 cells, although HGF alone did not activate EGFR phosphorylation appreciably, under co-stimulation conditions with EGF together, HGF further enhanced the EGFR phosphorylation in A549 cells to a level higher than that with EGF alone. On the other hand, EGF stimulation of H1975 cells co-activated MET receptor to enhance the level of MET phosphorylation. (<bold>B</bold>) MET–EGFR cross-activation in lung cancer. Left panel (A549), HGF cross-activated p-EGFR in A549 cells in the presence of co-stimulation with EGF. A549 cells were cultured in serum-starved conditions overnight, then stimulated with EGF alone (100 ng ml⁻¹, 15 min), HGF alone (50 ng/ml, 15 min), or both. Whole cell lysates were collected for immunoprecipitation with EGFR antibody, followed by immunoblotting (WB) with antibodies against p-EGFR[Y1068] (upper panel) and total EGFR (lower panel). Right panel (H1975), EGF cross-activates phospho-MET in H1975 cells. H1975 cells were cultured in starved media overnight, then stimulated with EGF alone (100 ng ml⁻¹, 15 min), HGF alone (50 ng ml⁻¹, 15 min), or both. Whole cell lysates were collected for immunoprecipitation with MET antibody (C-12), followed by immunoblotting with antibodies against p-MET[Y1234/1235] (upper panel) and total MET (lower panel). The <italic>MET</italic> and <italic>EGFR</italic> genotypes of the A549 and H1975 cells, as well as their <italic>MET</italic> genomic copy numbers, are shown in the bottom.</p></caption></fig>", "<fig id=\"fig5\"><label>Figure 5</label><caption><p>SU11274 inhibition of MET in combination with EGFR inhibitor in erlotinib-resistant NSCLC cell signalling. (<bold>A</bold>) Potentiated inhibition of cellular cytoskeletal functions by combined MET–EGFR inhibition (SU11274 plus erlotinib) in H1975 cells under video microscopy. H1975 cells had constitutively activated cytoskeletal functions with enhanced cell motility and migration under the serum-starved culture conditions. Comparing with the untreated control (left panel), drug treatment using SU11274 (right panel) substantially inhibited the constitutively activated cytoskeletal functions of H1975 cells (<xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Figure 3</xref>). H1975 cells were cultured in serum-stimulated conditions (10% FBS) and treated with the following for video microscopy digital video recording as described in the Materials and Methods: (a) DMSO diluent control, (b) Erlotinib alone (2 <italic>μ</italic><sc>M</sc>), (c) SU11275 alone (5 <italic>μ</italic><sc>M</sc>), and (d) combined concurrent SU11274 (5 <italic>μ</italic><sc>M</sc>)+erlotinib (2 <italic>μ</italic><sc>M</sc>). Complete abrogation of cytoskeletal functions with inhibition of cell motility and migration was only evident in the dual SU11274/erlotinib TKI-treated cells (d). (<bold>B</bold>) MET inhibition using SU11274, in combination with EGFR inhibition (erlotinib), induced cooperative downstream signalling inhibition in A549 (left panel) and H1975 (right panel) cells <italic>in vitro</italic>. EGFR-TKI-resistant A549 and H1975 cells were cultured in serum-starved conditions with EGF and HGF dual ligands stimulation as described in the Materials and Methods section. The cells were treated with SU11274 alone, erlotinib alone, or combination SU11274 plus erlotinib, then analysed in immunoblotting as indicated. (<bold>C</bold>) MET inhibition using specific siRNA-<italic>MET</italic>, in combination with EGFR inhibition (erlotinib) induced cooperative downstream signalling inhibition in A549 cells <italic>in vitro</italic>. Cells were transfected with control siRNA or siRNA-<italic>MET</italic> as described in Methods. Forty-eight hours after transfection, cells were cultured in starved media overnight, then treated with or without erlotinib and alone or in combination with siRNA-<italic>MET</italic> as indicated. After 4 h of inhibitor treatment, cells were then stimulated with both EGF (100 ng ml⁻¹) and HGF (50 ng ml⁻¹) ligands as indicated for 15 min. Whole cell lysates were then collected for immunoblotting analysis as in panel B above. (<bold>D</bold>) Rescue from alternative RTK ligand-stimulated signalling (MET–HGF <italic>vs</italic> EGFR–EGF) against TKI in A549 cells. A549 cells were cultured under serum-starved conditions, and then treated with either HGF or EGF, and in the presence or absence of the corresponding targeted inhibitor SU11274 or erlotinib as indicated (lanes 2, 4). Dual ligand stimulation (HGF and EGF) with single or dual TKIs treatment was included as indicated (lanes 1, 3, 5, 6). Although receptor-specific TKI was able to inhibit the downstream signalling driven by the corresponding ligand stimulation, alternative ligand stimulation in the form of dual ligand stimulation rescued the inhibited downstream signals. Dual TKI SU11274 plus erlotinib inhibition was required to fully knockdown the dual ligand-stimulated downstream signal activation of AKT, ERK1/2, and STAT3 (lane 6). (<bold>E</bold>) MET inhibition with SU11274, in combination with EGFR inhibition using CL-387,785 (irreversible EGFR-TKI), induced cooperative downstream signalling inhibition in H1975 cells <italic>in vitro</italic>. Cells were cultured in starved media overnight, then treated with or without CL-387,785 and alone or in combination with MET inhibitor SU11274 as indicated. After 4 h of inhibitor treatment, cells were then stimulated with both EGF (100 ng ml⁻¹) and HGF (50 ng ml⁻¹) ligands as indicated for 15 min. Whole cell lysates were then collected for immunoblotting analysis. Similar to erlotinib, CL-387,785 further sensitised H1975 cells to SU11274 inhibition with enhanced cooperative inhibition of signalling pathways downstream of the two RTKs.</p></caption></fig>", "<fig id=\"fig6\"><label>Figure 6</label><caption><p>Dual SU11274–erlotinib inhibition induced cooperative inhibition in H1975 cell viability <italic>in vitro</italic> and murine xenograft tumour growth <italic>in vivo</italic>. (<bold>A</bold>) <italic>In vitro</italic> inhibition using MET inhibitor SU11274 combined with erlotinib was more effective in H1975 cell viability inhibition under serum-stimulated conditions. Enhanced inhibition of cell viability was evident with dual SU11274-erlotinib treatment in combination (at 3 <italic>μ</italic><sc>M</sc> of each TKI as indicated). ^*<italic>P</italic>&lt;0.05 (SU11274/erlotinib <italic>vs</italic> SU11274); ^**<italic>P</italic>&lt;0.02 (SU11274/erlotinib <italic>vs</italic> erlotinib); and ^***<italic>P</italic>=0.281 (SU11274 <italic>vs</italic> erlotinib). Error bar, s.d. (<bold>B</bold>) Combined knockdown of <italic>MET</italic> and <italic>EGFR</italic> signalling using short-interfering RNA (siRNA) in H1975 cells resulted in enhanced downstream signal transduction inhibition. H1975 cells cultured under serum-stimulated conditions were treated with siRNAs specifically targeted against mRNA of <italic>MET</italic> alone, <italic>EGFR</italic> alone, or both <italic>MET</italic> and <italic>EGFR</italic> as described in the Materials and Methods section. Cells with siRNA knockdown as indicated were harvested for immunoblotting using antibodies against phosphotyrosine (left panel) to survey the effects on global cellular phosphotyrosine phosphoproteomic profiles. Cells were also immunoblotted with antibodies against the MET and EGFR signal paths including the downstream pro-survival AKT and STAT3 pathways (right panel). Concurrent dual knockdown of <italic>MET</italic> and <italic>EGFR</italic> signalling by siRNA in H1975 cells led to optimally enhanced downregulation of global phosphorylated cellular proteome (left panel) including the pro-survival downstream p-AKT and p-STAT3 signal activation (right panel). (<bold>C</bold>) <italic>In vivo</italic> treatment using SU11274 combined with erlotinib induced cooperative complete regression of EGFR-TKI-resistant H1975 tumour xenograft growth. EGFR-TKI-resistant H1975-luc cells were used to establish nude mouse xenograft <italic>in vivo</italic> as described in the Materials and Methods section. The nude mice with H1975-luc xenografts were then treated with diluent control, EGFR inhibitor (erlotinib, 100 mg/kg/day) alone, MET inhibitor (SU11274, 50 <italic>μ</italic>g per xenograft per day) alone, or both inhibitors concurrently (SU11274 plus erlotinib). Tumour xenograft growth was monitored by BLI at pretreatment baseline (day 0), and on post-treatment days 6 and 13. SU11274, in combination with erlotinib, induced complete tumour xenograft regression of H1975 cells <italic>in vivo</italic>. The mean relative BLI flux from each treatment group was plotted graphically (<italic>N</italic>=4 per treatment group). Error bar, s.e.m. (^*), (SU11274/erlotinib <italic>vs</italic> erlotinib) <italic>P</italic>=0.0006. (^**), (SU11274/erlotinib <italic>vs</italic> SU11274) <italic>P</italic>=0.0003. (^***), (SU11274 <italic>vs</italic> erlotinib) <italic>P</italic>=0.0070. (<bold>D</bold>) H1975 tumour xenograft micrographs under H&amp;E staining at × 100 magnification (and × 200, inset) showed substantial viable tumour cells in panel (a) DMSO control, and (b) erlotinib-treated animals, whereas there were necrotic and apoptotic tumour cells seen in panel (c) SU11274 (suboptimal dose: 50 <italic>μ</italic>g per xenograft) and massively so in panel (d) in combined SU11274 plus erlotinib-treated animals.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"xob1\"><label>Supplementary Figures</label></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"xob2\"><label>Supplementary Table</label></supplementary-material>" ]

[ "<fn-group><fn><p><xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Information</xref> accompanies the paper on British Journal of Cancer website (<ext-link xmlns:xlink=\"http://www.w3.org/1999/xlink\" ext-link-type=\"uri\" xlink:href=\"http://www.nature.com/bjc\">http://www.nature.com/bjc</ext-link>)</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"6604559f1\"/>", "<graphic xlink:href=\"6604559f2\"/>", "<graphic xlink:href=\"6604559f3a\"/>", "<graphic xlink:href=\"6604559f3b\"/>", "<graphic xlink:href=\"6604559f4\"/>", "<graphic xlink:href=\"6604559f5\"/>", "<graphic xlink:href=\"6604559f6\"/>" ]

[ "<media xlink:href=\"6604559x1.ppt\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"6604559x2.ppt\"><caption><p>Click here for additional data file.</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[]

[]

[]

[]

[ "<title>Conclusions</title>", "<p>In conclusion, the discovery by Tomlins <italic>et al</italic> in 2005 of a frequent genetic rearrangement in prostate cancers has changed our conception about the role of chromosomal rearrangements in the aetiologies of common solid tumours. In the short time since this discovery, several authors have confirmed the importance of this genetic fusion, and have expanded the class of fusion genes greatly. It now appears that these are among the most frequent recurrent rearrangements in cancer. The consequence of the various fusion transcripts is the overexpression of a member of the <italic>ETS</italic> family of oncogenes, initially under the control of androgen and the androgen receptor, but androgen dependence may be lost in advanced disease. It now appears that activation of this pathway may be central to prostate carcinogenesis, but the clinical implication of the various fusion products has not been worked out. It is hoped that this discovery will quickly lead to treatments tailored to patients in different risk classes, and possibly to a screening test, and ultimately it is hoped that the <italic>ETS</italic> family oncogenes will be molecular targets for novel therapies.</p>" ]

[ "<p>It has recently been shown that the majority of prostate cancers harbour a chromosomal rearrangement that fuses the gene for an androgen-regulated prostate-specific serine protease, <italic>TMPRSS2</italic>, with a member of the ETS family of transcription factors, most commonly <italic>ERG</italic>. These are among the most common genetic alterations in any human solid tumour. This knowledge may provide us with clues to prostate carcinogenesis, and may lead to the development of important molecular-based biomarkers for patients with localised prostate cancer. The most common variant is fusion between the 5′-untranslated region of <italic>TMPRSS2</italic> and the 3′ region of <italic>ERG</italic>. However, over 20 other fusion variants have now been described (involving over 10 different genes) and the number of variants continues to grow. Fusion products can be identified by several techniques, including FISH, RT–PCR, and expression profiling using exon arrays. The protein products associated with the fusion transcripts have not been characterised, and the phenotypic expression of the various products of gene fusion on prostate cancer histology, or on the clinical course of cancer, are not yet understood. Several early cohort studies suggest that the presence of the <italic>TMPRSS2:ERG</italic> fusion product is associated with relatively poor cancer-specific survival. Studies that examine how individual variants and their associated phenotypes affect prostate cancer presentation and progression are required.</p>" ]

[ "<title>Fusion of Tmprss2 and Ets Transcription Factor Genes</title>", "<p>In 2005, Tomlins <italic>et al</italic> identified and described a small number of fusion transcripts, specific to prostate cancer, that were the consequence of a chromosomal rearrangement involving two genes. One gene, <italic>TMPRSS2</italic> (androgen-regulated trans-membrane protease, serine 2), encodes for a serine protease that is secreted by prostate epithelial cells in response to androgen exposure (##REF##11245484##Afar et al, 2001##). <italic>TMPRSS2</italic> was fused to either <italic>ERG</italic> or <italic>ETV1</italic>, two members of the ETS family of oncogenes (##REF##16254181##Tomlins et al, 2005##). It had earlier been reported that the <italic>ERG</italic> gene was the most commonly overexpressed proto-oncogene in prostate cancer (present in about 72% of cases of prostate cancer) (##REF##15750627##Petrovics et al, 2005##), and Tomlins <italic>et al</italic> now proposed a mechanism to explain the overexpression. In this landmark paper, they found that both intra-chromosomal and inter-chromosomal genetic rearrangements led to the creation of a fusion transcript called <italic>TMPRSS2–ETS</italic>. ETS is a family of transcriptional activators and inhibitors. Their activity is regulated by phosphorylation and protein–protein interactions (##REF##16213704##Seth and Watson, 2005##). <italic>ERG</italic>, <italic>ETV1</italic>, <italic>ETV4</italic>, and <italic>ETV5</italic> are members of the <italic>ETS</italic> family. The <italic>ERG</italic> gene is located on chromosome 21q, <italic>ETV1</italic> is located on chromosome 7p, ETV4 on chromosome 17q, and <italic>ETV5</italic> on chromosome 3q. Using a novel and powerful bioinformatic technique, they first determined that either <italic>ETV1</italic> or <italic>ERG</italic> (but not both) was commonly overexpressed in prostate cancer cells. Furthermore, overexpression did not usually include all <italic>ERG</italic> exons – exons 4–7 were overexpressed much more commonly than were exons 1 and 2. This intriguing observation suggested to them that the gene had somehow been broken – and by sequencing the cDNA products, they confirmed that the 5′ part of the <italic>ERG</italic> gene had been replaced by the sequence derived from the <italic>TMPRSS2</italic> gene. They were able to verify the rearrangement at the DNA level with FISH techniques. Using expression arrays derived from prostate cancer specimens, they confirmed that one of the two <italic>ETS</italic> oncogenes was overexpressed in 57% of 167 prostate cancers. They concluded by showing that, in cells which carried either fusion product, expression of the <italic>ERG</italic> oncogene sequences was under the control of androgen. For example, VCaP cells, which harbour the fusion product <italic>TMPRSS2:ERG</italic>, expressed <italic>ERG</italic> transcript at a level 2000-fold greater than LNCAP cells, which do not harbour a fusion product. Because <italic>TMPRSS2</italic> is regulated in the prostate by androgens, it was proposed that this gene rearrangement could be a mechanism whereby the <italic>ETV1</italic> or <italic>ERG</italic> oncogenes were overexpressed, leading to prostate cancer.</p>", "<p>The <italic>TMPRSS2</italic> and <italic>ERG</italic> genes are about 3 megabases (mB) apart on chromosome 21. In about two-thirds of cases, fusion is the result of the deletion of the intervening DNA sequence, but fusion may also occur by a more complex rearrangement, such as a translocation (##REF##16820092##Yoshimoto et al, 2006##; ##REF##17632455##Tu et al, 2007##). The most common fusion is between the 5′-untranslated region of <italic>TMPRSS2</italic> and <italic>ERG</italic> (##REF##16951139##Perner et al, 2006##; ##REF##16820092##Yoshimoto et al, 2006##; ##REF##17632455##Tu et al, 2007##; ##REF##17334343##Mehra et al, 2007b##). The specific points of DNA breakage, and the exons retained in the fusion product, differ between patients. Over 20 <italic>TMPRSS2:ERG</italic> variants have now been described (##REF##16254181##Tomlins et al, 2005##, ##REF##16585160##2006##; ##REF##17043636##Clark et al, 2007##; ##REF##17654723##Liu et al, 2007##). A nomenclature has been proposed to describe the variant transcripts, on the basis which exons of the genes are involved (##REF##17043636##Clark et al, 2007##). Most variants are the result of a recombination between either exon 1 or exon 2 of <italic>TMPRSSR2</italic>, and exon 4 of <italic>ERG</italic> (although exons 2–5 may be involved). The most commonly reported fusion transcript is between exon 1 of <italic>TMPRSS2</italic> and exon 4 of <italic>ERG</italic> (##REF##17043636##Clark et al, 2007##). This particular rearrangement is designated as T1/E4 according to the nomenclature proposed by ##REF##17043636##Clark et al (2007)##. This transcript was originally described by ##REF##16254181##Tomlins et al (2005)##, and in some studies, accounts for up to 86% of all reported fusions (##REF##16951141##Wang et al, 2006##). Most patients who have been included in clinical and pathologic studies so far carry a variant involving these two genes (##REF##16951141##Wang et al, 2006##; ##REF##17237811##Demichelis et al, 2007##; ##REF##17334351##Lapointe et al, 2007##; ##REF##17971772##Nam et al, 2007b##; ##REF##17637754##Attard et al, 2008##).</p>", "<p>The number of genes and the variants involved in fusion transcripts continues to grow. Recently, it was found that other members of the <italic>ETS</italic> gene family (<italic>ETV4</italic> and <italic>ETV5</italic>) are involved in a small proportion of prostate cancer cases (##REF##16585160##Tomlins et al, 2006##; ##REF##18172298##Helgeson et al, 2008##). New partners have also been identified on the 5′ side of the translocation. ##REF##17334351##Lapointe et al (2007)## identified a fusion product derived from a variant isoform of <italic>TMPRSS2</italic>, which mapped 4 kb upstream of the more common start site (##REF##17334351##Lapointe et al, 2007##). One in 10 of the 63 prostate cancer cases in their series expressed this unique isoform. ##REF##17671502##Tomlins et al (2007)## and ##REF##18172298##Helgeson et al (2008)## implicated other 5′ fusion partners for ETV1, including <italic>SLC5A3</italic>, <italic>HERV-K22q11.23</italic>, <italic>C15orf21</italic>, and <italic>HNRPA2B1</italic>. <italic>SLC5A3</italic> appears to be capable of fusion to <italic>ETV5</italic>, as well as to <italic>ETV1</italic>, but not to <italic>ERG</italic> (##REF##17671502##Tomlins et al, 2007##; ##REF##18172298##Helgeson et al, 2008##). ##REF##18451133##Hermans et al (2008)## identified two additional fusion partners of <italic>ETV4</italic> (kallikrein 2 (KLK2) and calcium-activated nucleotidase 1 (<italic>CANT1</italic>)) (##REF##18451133##Hermans et al, 2008##).</p>", "<p>Considering the new reports, it appears that a member of the ETS family is overexpressed in the majority of prostate cancers and that there may be mechanisms for overexpression other than through gene fusion. In the original paper by ##REF##15750627##Petrovics et al (2005)## a high proportion of cancers overexpressed <italic>ERG</italic>, but the underlying genetic mechanism was not determined. ##REF##17505060##Cai et al (2007)## reported that <italic>ETV1</italic> was overexpressed in the majority of prostate cancers, but that only the minority of these had evidence of a translocation (##REF##17505060##Cai et al, 2007##). ##REF##17108102##Hermans et al (2006)## studied 11 prostate cancer xenografts, some of which exhibited androgen-dependent growth. In the androgen-dependent cases, the <italic>TMPRSS2:ERG</italic> fusion transcript was present, and overexpression of the <italic>ERG</italic> gene was associated with the expression of androgen receptor and PSA. Some androgen-independent cancers were also found to contain the <italic>TMPRSS2:ERG</italic> fusion transcript, but lacked androgen receptor – it is believed that these tumours have passed through an androgen-dependent phase. They also found two cases of advanced androgen-independent prostate cancer in which other members of the <italic>ETS</italic> family (specifically <italic>FlI-1</italic> and <italic>ETV4</italic>) were overexpressed due to a mechanism other than gene fusion. Of the 5′ fusion partners identified by ##REF##17671502##Tomlins et al (2007)##, three (<italic>TMPRSS2</italic>, <italic>SLC5A3</italic>, and <italic>HERV-K22q11.23</italic>) appear to be androgen-responsive and two (<italic>C15orf21</italic> and <italic>HNRPA2B1</italic>) appear to result in constitutive overexpression of <italic>ETV1</italic> in the absence of androgen stimulation. In the future, clinical studies should distinguish the course of the disease in cases of cancer with different fusion proteins and the response to androgen ablation treatment.</p>", "<title>Characterisation of Tmprss2erg Protein</title>", "<p>The consequence of the most common gene fusion is to generate a hybrid transcript that attaches the prostate-specific promoter sequence of the <italic>TMPRSS2</italic> gene to the ERG oncogene open reading frame (ORF). The proteins sequences have been predicted from the sequence of the fusion ORFs. ##REF##17043636##Clark et al (2007)## studied cDNAs prepared from <italic>ERG</italic> mRNAs isolated from 26 prostate cancers. They reported that of the 14 different fusion transcripts identified from the cDNA sequence, 5 would be predicted to generate premature stop codons and would be unlikely to encode for a functional ERG protein. In most cases, no amino acid sequence derived from TMPRSS2 is integrated in the hybrid ORF, and therefore a fusion protein is not created. In the case of the T1/E2 <italic>TMPRSS2:ERG</italic> variant, the full-length ERG protein is translated; the other fusion transcripts encode for a truncated protein (##REF##17043636##Clark et al, 2007##).</p>", "<title>Prevalence of Fusion Product Among Unselected Prostate Cancer Cases</title>", "<p>The presence of a gene fusion product can be determined by measuring the level of RNA expression using RT–PCR, by using FISH (which probes for inappropriate juxtaposition of non-adjacent sequences or the ‘break apart’ of a single gene to different chromosome locations), or by observing imbalance in the expression of individual exons using array technology. The prevalence rate of fusions among cancer patients depends on the assay used, the volume of the cancer, the number of cancer foci studied, and the number of fusion variants included in the screening panel. The question of prevalence is also complicated by the observation that a single cancer may have different foci that harbour rearrangements involving different genes, or no rearrangment at all (##REF##17043636##Clark et al, 2007##). The field is evolving, but on the whole, these data suggest that 70% or more of all prostate cancers harbour a fusion product (##REF##17108102##Hermans et al, 2006##; ##REF##16951139##Perner et al, 2006##; ##REF##16575875##Soller et al, 2006##; ##REF##17259299##Rajput et al, 2007##; ##REF##17632455##Tu et al, 2007##; ##UREF##0##Nam et al, 2007a##); individual estimates vary from 15% (##REF##17237811##Demichelis et al, 2007##) to 78% (##REF##16575875##Soller et al, 2006##). As the number of variant species continue to increase and as techniques for measuring them become increasingly diverse and more sensitive, we are likely to find that an even greater proportion of prostate cancer specimens contain one or more fusion variants.</p>", "<p>Prostate cancer is often a multifocal disease. ##REF##17804708##Mehra et al (2007a)## examined gene fusion status within different tumour foci from 43 patients with multifocal cancers, embedded in a tissue microarray (##REF##17804708##Mehra et al, 2007a##). Of these patients, 70% had cancers that exhibited a fusion product, but in many cases, different foci from the same tumour carried different fusion products. Surprisingly, 70% of the cases with fusion were discordant for the specific fusion products. When the largest focus of cancer from each patient was identified, 83% of samples had a rearrangement. A similar result was obtained by ##REF##17991527##Barry et al (2007)##, who studied 32 cases of multifocal prostate cancer. Of these cases, 41% exhibited heterogeneity (##REF##17991527##Barry et al, 2007##). Recently, ##REF##17043636##Clark et al (2007)## also found that different cancer foci from a single patient harboured different fusion protiens. In some cases, both <italic>ERG</italic> and <italic>ETV1</italic> were involved in different foci of cancer from the same patient. Thus, heterogeneity of <italic>TMPRSS2:ERG</italic> gene fusion suggest that the different foci of cancer that arise within a multifocal prostate cancer probably have different origins and represent different malignant clones – this observation also implies that the clinical interpretation of this biomarker is more complex than was originally thought.</p>", "<title>Pathological Characteristics</title>", "<p>Despite the lack of characterisation of the individual proteins, the phenotypic features of the <italic>TMPRSS2:ERG</italic>-associated cancers, as a class, have been described. ##REF##17385188##Mosquera et al (2007)## studied 120 cases of prostate cancer with the fusion gene, identified from a total of 253 cases (##REF##17385188##Mosquera et al, 2007##). Five histologic features were associated with the presence of gene fusion: the presence of blue-tinged mucin, a cribriform growth pattern, macronucleoli, intraductal tumour spread, and signet-ring cell features – these features are also associated with an aggressive clinical course of prostate cancer – but neither Gleason grade nor stage was significantly associated with the presence of the fusion gene. Several other studies have compared the characteristic of prostate cancers with and without gene fusion in terms of grade and stage, and PSA level (##REF##16951139##Perner et al, 2006##; ##REF##16951141##Wang et al, 2006##; ##REF##17334351##Lapointe et al, 2007##; ##REF##17259299##Rajput et al, 2007##; ##REF##17632455##Tu et al, 2007##). Results to date have been inconsistent, but most studies suggest that the presence of the fusion protein is not correlated with other markers of risk. ##REF##17259299##Rajput et al (2007)## found that fusions were less common in low-grade (Gleason 2) prostate cancers (1 of 17; 7%) than in moderate grade (Gleason 3–5) cancers (35 of 86; 41%; <italic>P</italic>-value for difference=0.02). ##REF##16951141##Wang et al (2006)## suggested that the T2/E4 variant might be associated with more aggressive disease than other fusion transcripts. This might be the case if the abundance of the oncogene transcript varied between cells with different fusion species, but this hypothesis is not confirmed. <italic>TMPRSS2:ERG</italic>-associated cancers have also been evaluated from the point of view of gene co-expression. Iljin <italic>et al</italic> have reviewed expression data derived from 410 different prostate tissue samples. They found that the most common gene that was co-expressed with ERG was histone deacetylase 1 (<italic>HDAC 1</italic>) and that, in fact, all the ERG-positive prostate cancers in that series were also strongly positive for <italic>HDAC</italic> (##REF##17079440##Iljin et al, 2006##). This observation suggested to them that anti-HDAC therapies might have a potential therapeutic application for this class of prostate tumours (##REF##15657340##Li et al, 2005##).</p>", "<title>Clinical Significance of Tmprss2:erg Gene Fusion</title>", "<p>To date, there are three clearly established prognostic factors for men with localised prostate cancer: histologic grade (measured by the Gleason scoring system, tumour stage, and PSA level at diagnosis. Men with tumours of higher grade (Gleason 8–10), stage (T3–T4), or PSA level (e.g., &gt;20 ng ml⁻¹) experience relatively high rates of progression to metastasis, when compared with men with tumours of lower grade, local stage, or low PSA level. A biomarker will be clinically useful if it allows one to select treatments for individuals; that is if it helps identify subgroups of patients who will, or who will not, benefit from a specific treatment. A patient may not benefit from a treatment because survival is excellent in the absence of treatment, or because the rate of progression is high despite treatment.</p>", "<p>In an early study, ##REF##15750627##Petrovics et al (2005)## found that <italic>ERG</italic> sequences were overexpressed in prostate cancer cells relative to adjacent benign prostate cells (at a level of twofold or greater) in approximately 80% of prostate cancers. In this study, patients with the highest level of expression in their cancer cells (relative to benign tissues) had the best prognosis, and the difference in survival between the groups was highly significant.</p>", "<p>The clinical significance of the presence of the <italic>TMPRSS2:ERG</italic> gene fusion product on prostate cancer presentation and progression is not fully understood, but studies to date suggest that this may be a biomarker of risk. The results of the various studies are inconsistent, possibly because different study designs, different biomarkers, and different end points are used. In general, the case series (##REF##17334351##Lapointe et al, 2007##; ##REF##17259299##Rajput et al, 2007##; ##REF##17632455##Tu et al, 2007##; ##REF##17637754##Attard et al, 2008##) and case–control studies (##REF##16951139##Perner et al, 2006##; ##REF##16951141##Wang et al, 2006##) did not identify a significant prognostic effect. Using a case–control approach, ##REF##16951141##Wang et al (2006)## examined 119 patients for fusion status. They found a significant correlation with tumour stage, but not with early recurrence. ##REF##17334351##Lapointe et al (2007)## found modest correlations between stage and grade and the presence of the fusion product, but the sample size was small (<italic>n</italic>=63 cases) and neither association was statistically significant.</p>", "<p>In contrast to the early study of <italic>ERG</italic> gene expression by ##REF##15750627##Petrovics et al (2005)##, three recent cohort studies that evaluated the presence of the fusion protein <italic>per se</italic> on patient outcome have found the translocation to be an adverse prognostic factor. Two of these studies included patients undergoing watchful waiting. These authors sought to determine whether patients who do not harbour a fusion product might be candidates for watchful waiting. Currently, it is estimated that up to 30% prostate cancer patients lack the fusion protein, and therefore the potential exists to avoid the morbidity associated with anti-androgen therapy or with surgery for a significant proportion of patients. However, it should also be remembered that more than half of prostate cancer patients have cancers that harbour a fusion product, and we expect that many patients with clinically localised disease will be cured of cancer by prostatectomy, even if the fusion product is present.</p>", "<p>##REF##17237811##Demichelis et al (2007)## followed 111 patients undergoing watchful waiting for localised prostate cancer (##REF##17237811##Demichelis et al, 2007##). Patients with the gene fusion had a 2.7-fold increase in prostate cancer-specific mortality, when compared with patients without fusion. However, 23% of 94 patients without the fusion protein experienced metastatic prostate cancer after 12 years of observation, when compared with 53% of 17 patients with fusion. Although the difference was statistically significant (<italic>P</italic>&lt;0.01), a recurrence risk of 23% is not sufficiently low to endorse watchful waiting as an alternative to surgery. In a similar study, ##REF##17637754##Attard et al (2008)## studied 445 patients under watchful waiting. Fusion analysis was conducted on prostate specimens embedded in a tissue microarray using FISH. In this study, patients without the fusion transcript had a good survival experience (90% survival at 10 years). ##REF##17637754##Attard et al (2008)## refined their study by sub-dividing patients with fusion transcripts into three categories: (1) those with retained 5′ and 3′ <italic>ERG</italic> DNA sequences; (2) those with one retained copy of 3′ <italic>ERG</italic> sequence but no retained 5′ <italic>ERG</italic> sequence; and (3) those with two copies of the retained 3′ <italic>ERG</italic> sequence (i.e., homozygous or duplicated) but no retained 5′ <italic>ERG</italic> sequence. The third group was noteworthy for its poor prognosis; after 8 years of follow-up, patients in this group were six times more likely to die from prostate cancer than those with no <italic>TMPRSS2:ERG</italic> gene fusion (hazard ratio=6.1, 95% CI: 3.3–11.1, <italic>P</italic>&lt;0.0001). Only 25% of patients in this class were alive at 8 years. The frequency of gene rearrangements in the ##REF##17637754##Attard et al (2008)## study was 30%, when compared with only 15% in the study by ##REF##17237811##Demichelis et al (2007)##. It may be that the Demichelis group used a less sensitive assay to screen for the presence of the fusion gene, or that a number of variant transcripts went undetected, and that this resulted in misclassification and attenuation of the true effect (however, in a later study from the same group, the low prevalence of rearrangements was confirmed). The patients in the Demichelis study were from Sweden, and it is possible that the prevalence of the fusion protein varies with ethnic group. It remains to be proven that patients who lack fusion genes may be candidates for watchful waiting. Future studies might benefit from including additional markers of risk in the prognostic model, including grade, stage, PSA level, ethnic group, and the presence of one or more markers of genetic susceptibility, such as the recently defined cluster of markers on chromosome 8q24 (##REF##16682969##Amundadottir et al, 2006##; ##REF##17401363##Yeager et al, 2007##; ##REF##17925536##Zheng et al, 2007##).</p>", "<p>In a prospective cohort study, ##REF##17971772##Nam et al (2007b)## examined the effect of the most common fusion variant (T1/E4) on disease recurrence (defined by a rising PSA level after surgery) among 165 patients who underwent surgery for localised prostate cancer (##REF##17971772##Nam et al, 2007b##). This particular gene fusion was present in 49% of patients, and 26% of the patients developed biochemical disease recurrence. Patients with fusion had a much higher rate of recurrence (54% at 5 years) than those without fusion (8%). Fusion status did not correlate with PSA, grade, or stage. After adjusting for PSA, grade, and stage, the hazard ratio for recurrence was 8.5 (95% CI: 3.6–20.6, <italic>P</italic>&lt;0.0001). This study implies that this biomarker may be an independent prognostic factor of disease recurrence. However, although all patients who die of prostate cancer experience biochemical recurrence before death, only a minority of patients with biochemical recurrence succumb to prostate cancer. It is important that large studies of prostatectomy patients be conducted with longer follow-up and that additional end points include distal recurrence and prostate-specific mortality. The T1/E4 variant is the most common of <italic>TMPRSS2:ERG</italic> gene fusions and is the best studied; it is not yet known if other variants are associated with prostate cancer prognosis to the same extent. Furthermore, it is not clear why the results of the three cohort studies (which examined the presence of the fusion protein) had results that differed dramatically from those of ##REF##15750627##Petrovics et al (2005)## who examined the overexpression of ERG. Petrovics <italic>et al</italic> measured overexpression in cancer tissues relative to benign prostate, and it is possible that expression levels in the surrounding stroma are clinically relevant as well. In an ideal study, one would perform both assays on a single group of patients.</p>", "<p>The <italic>TMPRSS2:ERG</italic> gene fusion is specific for prostate cancer, and the ability to identify this DNA rearrangement could be used as a screening test for prostate cancer in serum, prostatic fluid, or in urine. One study has been conducted on DNA specimens isolated from urine from men known to have prostate cancer with a gene rearrangement (##REF##17785564##Hessels et al, 2007##). The sensitivity of the urine test was only 37% and the specificity was 93%. It is possible that future assays will have comparatively better sensitivity or that the presence of the fusion gene in urine could supplement a panel of markers in a screening setting.</p>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

{ "acronym": [], "definition": [] }

[]

[ "<title>Patients and methods</title>", "<title>Study design</title>", "<p>This phase I open label study was divided into two parts. The first two cycles were designated as phase A, to determine the MTD, pharmacokinetic and pharmacodynamic parameters. In phase B, matuzumab plus ECX was continued on the DL selected in phase A until disease progression, unacceptable toxicity or for a further six cycles. Patients who did not complete phase A for any reason, except unacceptable toxicities or progressive disease, were replaced.</p>", "<title>Patients</title>", "<p>Eligibility requirements included histologically confirmed adenocarcinoma of the stomach or lower third of the oesophagus, locally advanced, metastatic or recurrent disease, measurable disease by computed tomography (CT), EGFR expression in tumour tissue, normal cardiac function defined by left ventricular ejection fraction, Karnofsky performance status (KPS) ⩾60%, life expectancy &gt;12 weeks, no prior chemotherapy at all, no radiotherapy or major surgery within 4 weeks before the first study treatment, adequate baseline bone marrow and liver function, a glomerular filtration rate &gt;60 ml min⁻¹, no severe uncontrolled comorbidities and signed informed consent.</p>", "<p>The study protocol was approved by the local ethics committee and was carried out according to the Declaration of Helsinki and good clinical practice guidelines. The subjects' informed consent was obtained before any study-related activities.</p>", "<title>EGFR expression</title>", "<p>Tumour material was obtained from the initial tumour resection or diagnostic biopsy. Epidermal growth factor receptor expression was determined by a central pathologist in representative paraffin-embedded tumour blocks using EGFR pharmDx test kit (from DakoCytomation KGaA, Darmstadt, Germany) as previously reported. Tumours were considered positive if any membrane staining was observed. Only patients with EGFR-positive tumours were enrolled. All assessments were performed and reviewed centrally.</p>", "<title>Pre-treatment and evaluation</title>", "<p>Pre-treatment evaluation consisted of medical history, physical examination, full blood count (FBC), serum biochemistry, serum tumour marker, urine analysis, CT scans of the chest abdomen pelvis, multiple gated acquisition scan and chest X-ray. During treatment, monitoring included clinical toxicities assessment, FBC, serum biochemistry and physical examination weekly. Computed tomography scans were performed at weeks 6 and 12 and at the end of the treatment. Flectrocardiogram (ECG), KPS assessment and FBC and biochemistry were repeated at the end of the treatment.</p>", "<title>Administration and dose escalation</title>", "<p>Matuzumab was supplied by Merck (Germany) as a lyophilisate of 200 mg per vial. Matuzumab was administered as a 1-h intravenous infusion without premedication in 250 ml of 0.9% normal saline solution. ECX comprised of epirubicin 50 mg m⁻² given as a 15-min infusion, cisplatin 60 mg m⁻² given as a 4-h infusion on day 1 and capecitabine 500 mg m⁻² twice daily given continuously, each cycle duration being 3 weeks. Pre-medication and hydration were administered as described previously (##REF##15928658##Sumpter et al, 2005##).</p>", "<p>Initially, the study was planned with two DLs of matuzumab 400 and 800 mg weekly combined with ECX. However, an amendment was made to the protocol after pharmacodynamic data from a phase I matuzumab monotherapy study revealed that 1200 mg three weekly was the target effective dose. This provided a strong rationale for extending the dose regimen from weekly to a three-weekly schedule (##UREF##6##Tabernero et al, 2003##). Thus two additional DLs of matuzumab 1200 and 1600 mg, administered every 3 weeks with ECX, were included (##TAB##0##Table 1##).</p>", "<p>No intrasubject dose escalation was performed. At each DL, six patients were initially enrolled. If ⩽1 of the patients experienced a DLT during the first two cycles, the next cohort of patients was treated at the subsequent DL. If ⩾2 of 6 patients at one DL experienced any DLT, additional patients were enrolled at the same DL. The MTD and RD were defined as the DL at which not more than one of six patients experienced a DLT.</p>", "<title>Evaluation of toxicities and response</title>", "<p>Toxicities were evaluated weekly and graded according to the National Cancer Institute Common Toxicity Criteria (NCI-CTC; version 2.0). The MTD was based on DLTs observed during the first two cycles. Dose-limiting toxicity was defined as follows: an adverse event related to matuzumab including any grade 3/4 non-haematological toxicities (excluding alopaecia, nausea, vomiting and skin reactions), grade 4 nausea, vomiting and skin reactions and toxicity-related discontinuation of treatment for more than 1 week.</p>", "<p>Tumour response was measured by CT scans according to RECIST criteria using unconfirmed responses (##REF##10655437##Therasse et al, 2000##). Progression-free survival (PFS) was defined as the interval between the date of administration of the first infusion and the confirmation of progressive disease or death, depending on which occurred first.</p>", "<title>Pharmacokinetics</title>", "<p>For pharmacokinetic analysis, blood samples were taken before and 1, 2, 5, 48, 96, 168 and 336 h after the start of the matuzumab infusion in weeks 1 and 4. Serum samples were obtained and handled as previously described (##REF##14701780##Vanhoefer et al, 2004##). Concentrations of matuzumab were measured in serum using a validated ELISA. On the basis of the resulting concentrations, PK parameters were calculated by compartmental and non-compartmental standard methods using the software package KINETICA™, version 4.1.1.</p>", "<title>Pharmacodynamics</title>", "<p>Normal skin tissue biopsies from the upper arm at the posterior surface were taken before the first cycle, after the second cycle and fourth cycle. Tumour biopsies were taken by endoscopy as part of the routine staging before the first cycle, at the end of the second and fourth cycles. The percentage of cells staining positive for proteins on skin and tumour biopsies were determined as biological markers of the treatment. These pharmacodynamic markers comprised phosphorylated EGFR (p-EGFR), phosphorylated p42/44 MAP kinase (p-MAPK), EGFR, Ki67, p27, phosphorylated STAT3 (p-STAT3) and cytokeratin 1. In addition, phosphorylated protein kinase B was measured in tumour biopsies only. The samples were prepared and investigated as previously described (##REF##11773160##Albanell et al, 2002##).</p>" ]

[ "<title>Results</title>", "<p>Between 2002 and 2005, 45 patients underwent EGFR testing for the study at the Royal Marsden Hospital, UK and 60% exhibited positive EGFR expression by immunohistochemical analysis. In total, 21 patients with EGFR-positive tumours received study treatment (##TAB##0##Table 1##). Baseline patient characteristics are shown in ##TAB##1##Table 2##.</p>", "<p>No DLT was observed in the initial cohort. At the 800 mg matuzumab DL, one patient experienced a DLT of grade 3 hypotension. One patient was replaced at each DL as per protocol. At the next DL of matuzumab 1200 mg every 3 weeks, 1 DLT of grade 3 pancreatitis and 1 DLT of grade 3 abdominal pain were reported; however, the DLT of grade 3 lethargy occurred in three of six patients, indicating that the MTD was exceeded. Thus the main DLT was grade 3 lethargy, and inconsistent with our previous experience with ECX chemotherapy, therefore, one further patient was entered at this DL. However, this patient experienced the same DLT, thus four of seven patients experienced the main DLT of grade 3 lethargy. Hence, the DL of 800 mg matuzumab weekly combined with ECX was defined as the MTD and the RD.</p>", "<p>During phase B, the median number of cycles of treatment was five for the first two cohorts and three for the 1200 mg matuzumab DL. Among the most frequent toxicities observed were diarrhoea, nausea, vomiting and stomatitis, which can be associated with ECX chemotherapy (##TAB##2##Table 3##). The most significant matuzumab-related side effects (grades 1–4) across all DLs were lethargy and rash affecting 11 and 13 patients, respectively.</p>", "<p>One patient with a recurrent anastomotic OGJ tumour treated at the first DL developed an oesophagobronchial fistula and died subsequent to aspiration pneumonia. The patient had received one dose of matuzumab (400 mg per week) and ECX; this event was deemed unrelated to study treatment.</p>", "<title>Efficacy</title>", "<p>Although efficacy was not a primary objective, 20 patients were assessable for tumour response. Thirteen of 20 patients achieved a partial response resulting in an overall objective response rate (ORR) of 65% (95% confidence interval (CI): 43–82). Five patients (25%) (95% CI: 11–47) demonstrated disease stabilisation and two (10%) developed progressive disease. The ORR according to DL is shown in ##TAB##3##Table 4##.</p>", "<p>One patient with a T3N1 gastric tumour was downstaged to T2N0 on endoscopic ultrasound after five cycles of ECX plus matuzumab (400 mg per week) and underwent surgical resection followed by post-operative ECX plus matuzumab.</p>", "<p>The overall median time to disease progression was 5.2 months (95% CI: 3.0–16.0).</p>", "<p>Pharmacokinetic analysis parameters are shown in ##TAB##4##Table 5##. Maximum serum concentrations <italic>C</italic>_max were achieved on average 1–2 h after the end of the infusion. The mean values for <italic>C</italic>_max for all three DLs ranged between 154 and 442 <italic>μ</italic>g ml⁻¹ in week 1 and were dose proportional. The AUC results in the first week AUC (0_-168) were also approximately dose proportional.</p>", "<p>The terminal elimination phase was best characterised in the 1200 mg dose group where the concentration–time profile could be assessed over a 3-week period after each matuzumab infusion. Mean terminal elimination half-lives determined after weeks 1 and 4 were in the range of 8–9 days. At lower weekly dosing, only apparent <italic>t</italic>_1/2 were determined with values of 5–7 days. The mean values for the volume of distribution were consistently small (∼4 l) and dose independent. The mean trough values increased over time towards the steady state (##FIG##0##Figure 1##). There was evidence of accumulation at the 800 mg weekly DL; there was no correlation between the incidence of DLTs and <italic>C</italic>_max and AUC.</p>", "<title>Pharmacodynamics</title>", "<p>Pharmacodynamic results were available from 18 subjects in total. Skin biopsy data were consistent between patients at each DL (##FIG##1##Figure 2##). Following the administration of matuzumab, the total EGFR expression remained in the range of 80–100%. In contrast, EGFR phosphorylation was inhibited and there was a similar reduction in pMAPK for all investigated DLs. Increased levels of p27 and p-STAT3 were detected. Baseline Ki67 decreased following matuzumab treatment in all but one patient. Cytokeratin 1 levels in skin biopsies generally increased during treatment although there were decreased levels in two patients. The changes in these marker proteins described were not dose dependent.</p>", "<p>Tumour biopsy data were available for 15 pre-treatment biopsies but limited for the treatment samples due to poor fixation of tumour tissue. This reflects the difficulties in sampling tumour tissue after the administration of chemotherapy. Thus it was not possible to evaluate the changes observed during treatment in tumour biopsies.</p>", "<p>Further investigations were performed to evaluate any correlation between the development of rash and PD changes particularly in the skin biopsies. It was not possible to identify a clear correlation between the change of any PD marker protein in skin or tumour biopsies during matuzumab therapy and the presence of skin rash or response outcome.</p>" ]

[ "<title>Discussion</title>", "<p>This study has demonstrated that ECX combined with matuzumab weekly at doses up to 800 mg per week were generally well-tolerated. At 1200 mg of matuzumab three weekly, the MTD was exceeded and the main DLT experienced by four of seven patients was grade 3 lethargy. This occurred after the first infusion and generally took several weeks to resolve completely. All four patients were of KPS 90 before commencing treatment.</p>", "<p>The interim analysis of the REAL-2 study reported 9% grade 3/4 lethargy for patients treated with ECX (##REF##15928658##Sumpter et al, 2005##). Asthenia has previously been reported with cetuximab (a chimeric anti-EGFR antibody) in colorectal cancer (CRC). ##UREF##0##Abubakr et al (2006)## observed 25% asthenia (6% grade 3/4) in the phase III EPIC study of irinotecan plus or minus cetuximab in 783 patients with refractory CRC. ##REF##16106027##Schrag et al (2005)## described a magnesium-wasting syndrome associated with severe fatigue-affecting patients with CRC treated with cetuximab and irinotecan. They recommended that serum magnesium levels be monitored for any patient with severe asthenia following administration of cetuximab. In this study (initiated before this publication), the lethargy observed did not appear to be associated with serum hypomagnesaemia, although magnesium was not routinely measured. Thus the mechanism of the fatigue remains unknown but may be due to an interaction between matuzumab at 1200 mg three weekly and ECX chemotherapy.</p>", "<p>No allergic reactions, other severe or unexpected adverse events were observed. NCI-CTC grade 1/2 rash was observed in 61.9% of patients in total, which is similar to that previously observed in a phase I study of matuzumab monotherapy (##REF##14701780##Vanhoefer et al, 2004##). However, the incidence of rash appears lower than reported with other anti-EGFR monoclonal antibodies including cetuximab or panitumumab, which range between 80 and 90% for all grades (##REF##15269313##Cunningham et al, 2004##; ##UREF##5##Malik et al, 2005##). The skin toxicity associated with anti-EGFR antibodies is commonly described as an acneiform rash but pathologically resembles an infectious folliculitis (##UREF##4##Lenz, 2006##) and its pathophysiological mechanism remains unclear.</p>", "<p>The pharmacokinetic analyses demonstrated a dose-proportional increase of AUC and <italic>C</italic>_max for matuzumab with accumulation suggesting linear pharmacokinetics within the dose range tested. There was no correlation between the incidence of the DLT of lethargy and <italic>C</italic>_max and AUC in the 1200 mg cohort. In addition, when comparing the <italic>C</italic>_max and AUC_0–168 of matuzumab plus ECX to a previous matuzumab monotherapy study (##REF##14701780##Vanhoefer et al, 2004##), it was found that the matuzumab exposure was similar. This suggests that the coadministration of ECX chemotherapy did not influence the pharmacokinetics of matuzumab.</p>", "<p>The PD results obtained from the skin samples were as expected and total EGFR expression was not altered; there was a decrease in p-EGFR, p-MAPK and Ki67, whereas p27 and p-STAT3 increased following the administration of matuzumab. Thus, overall, there was abrogation of EGFR downstream signalling and there was no dose–response relationship. Therefore, there is inhibition of the EGFR network at doses of matuzumab below the MTD. Similar findings have been reported utilising varying schedules of matuzumab (##UREF##6##Tabernero et al, 2003##; ##REF##14701780##Vanhoefer et al, 2004##). It is not possible in this study to comment on the use of skin as a surrogate for tumour, given the limitations of the tumour biopsies.</p>", "<p>There has been much controversy surrounding EGFR expression by immunohistochemistry and the use of anti-EGFR therapy. In this study, all patients were EGFR-positive according to immunohistochemistry; however, there was no apparent correlation between the degree of EGFR inhibition and objective response.</p>", "<p>Although efficacy was not a primary objective, only patients with measurable disease were included in the study. The unconfirmed ORR with the combination of ECX plus matuzumab was 65 with 25% disease stabilisation, that is a disease control rate (CR, PR+SD) of 90%, and the median PFS was 5.2 months. The unconfirmed ORR for ECX in the REAL-2 study was 46.4% (##REF##18172173##Cunningham et al, 2008##); hence there may be a synergistic or additive effect between matuzumab and ECX although no conclusions can be drawn from these data and further investigation is required in a randomised study. The median time to progression in this study was 5.2 months and thus shorter than that reported for all treatment arms in REAL-2 (##REF##18172173##Cunningham et al, 2008##). This may partly be accounted for by the 1200 mg matuzumab cohort who received only five median cycles of treatment and the dose delays incurred before restarting chemotherapy by those patients who experienced the main DLT of grade 3 lethargy.</p>", "<p>Several phase I and II studies of EGFR tyrosine kinase inhibitors as immunotherapy in previously treated OG cancers have been reported with response rates ranging from 2.85 to 12% (##UREF##3##Ferry et al, 2004##; ##UREF##8##Van Groeningen and Giaccone, 2004##; ##REF##16575012##Janmaat et al, 2006##). A phase II study of cetuximab combined with FOLFIRI in untreated gastric and OGJ cancers achieved an ORR of 44.1% (95% CI 27.5–60.9%) (##REF##17164226##Pinto et al, 2007##). ##REF##16575012##Janmaat et al (2006)## identified female gender, squamous histology and high EGFR expression to be associated with improved outcome following the administration of gefitinib in patients with advanced OG cancer. A study of erlotinib demonstrated activity in OGJ tumours but no objective responses in gastric cancers (##REF##17050876##Dragovich et al, 2006##). In this trial, the inclusion criteria stipulated adenocarcinoma and the responses to ECX and matuzumab were observed in oesophageal, OGJ and gastric tumours.</p>", "<p>In conclusion, this trial has demonstrated that the MTD of matuzumab in combination with ECX was 800 mg weekly and at this DL it was generally well tolerated. Grade 3 lethargy was the main DLT at 1200 mg three weekly and the mechanism for this remains unclear. The combination regimen was associated with clinically meaningful tumour response and stabilisation and the PD markers in skin reflected inhibition of the EGFR signalling at all DLs. Thus a randomised national multicentre phase II trial of ECX with or without the addition of matuzumab at 800 mg weekly in advanced untreated OG cancer has been conducted.</p>" ]

[]

[ "<p>To evaluate the safety, tolerability, efficacy, pharmacokinetics and pharmacodynamics of the humanised antiepidermal growth factor receptor monoclonal antibody matuzumab combined with epirubicin, cisplatin and capecitabine (ECX) in patients as first-line treatment for advanced oesophagogastric cancer that express epidermal growth factor receptor (EGFR). This was a phase I dose escalation study of matuzumab at 400 and 800 mg weekly and 1200 mg every 3 weeks combined with ECX (epirubicin 50 mg m⁻², cisplatin 60 mg m⁻² on day 1 and capecitabine 1000 mg m⁻² daily). Patients were treated until disease progression, unacceptable toxicity or for a maximum of eight cycles. Twenty-one patients were treated with matuzumab at three different dose levels (DLs) combined with ECX. The main dose-limiting toxicity (DLT) was grade 3 lethargy at 1200 mg matuzumab every 3 weeks and thus 800 mg matuzumab weekly was the maximum-tolerated dose (MTD). Other common toxicities included rash, nausea, stomatitis and diarrhoea. Pharmacokinetic evaluation demonstrated that the coadministration of ECX did not alter the exposure of matuzumab. Pharmacodynamic studies on skin biopsies demonstrated inhibition of the EGFR pathway. Objective response rates of 65% (95% confidence interval (CI): 43–82), disease stabilisation of 25% (95% CI: 11–47) and a disease control rate (CR+PR+SD) of 90% were achieved overall. The MTD of matuzumab in combination with ECX was 800 mg weekly, and at this DL it was well-tolerated and showed encouraging antitumour activity. At the doses evaluated in serial skin biopsies, matuzumab decreased phosphorylation of EGFR and MAPK, and increased phosphorylation of STAT-3.</p>" ]

[ "<p>Oesophagogastric (OG) cancer represents a major health burden worldwide (##REF##11905707##Parkin, 2001##). For patients with advanced disease combination, chemotherapy has shown a survival benefit compared to best supportive care (##REF##8508427##Murad et al, 1993##; ##REF##7533517##Pyrhonen et al, 1995##).</p>", "<p>ECF (epirubicin, cisplatin, infused 5FU) is the reference regimen in the United Kingdom and other parts of Europe for advanced OG cancer based on superior response rates, survival and global QOL in several phase III studies (##REF##8996151##Webb et al, 1997##; ##REF##10390007##Waters et al, 1999##; ##REF##11956258##Ross et al, 2002##). Furthermore, a recent meta-analysis concluded that the best survival results are achieved with regimens containing anthracyclines, cisplatin and 5FU and among these ECF was the most well-tolerated (##UREF##9##Wagner et al, 2006##).</p>", "<p>Recently, the V325 study demonstrated a survival benefit for TCF (docetaxel, cisplatin and 5FU) <italic>vs</italic> CF (cisplatin and 5FU) although TCF was associated with &gt;80% grade 3 and 4 neutropaenia (##REF##17075117##Van Cutsem et al, 2006##). The randomised phase III trial REAL-2 evaluated four treatment arms ECF, EOF, ECX and EOX (E, epirubicin; X, capecitabine; C, cisplatin; O, oxaliplatin; F, 5FU). Non-inferiority was demonstrated for capecitabine <italic>vs</italic> 5FU and oxaliplatin <italic>vs</italic> cisplatin with acceptable toxicity for all treatment arms (##REF##18172173##Cunningham et al, 2008##).</p>", "<p>Despite recent advances, the median overall survival with combination chemotherapy is approximately 10–11 months, thus newer treatment strategies are required. The epidermal growth factor receptor (EGFR) has previously been identified as a novel target for anticancer treatment. Epidermal growth factor receptor activation leads to a cascade of signal transduction pathways involved in cell proliferation, angiogenesis, metastasis and invasion (##REF##10579913##Kim and Muller, 1999##; ##REF##10358079##Olayioye et al, 1999##; ##REF##10707088##Sako et al, 2000##; ##REF##11057895##Schlessinger, 2000##). In oesophageal cancer, expression of EGFR has been reported to be 80–90% (##REF##8039107##Itakura et al, 1994##) and is associated with poorer survival.</p>", "<p>Matuzumab is a humanised antibody that competitively inhibits natural ligand binding to the EGF receptor with abrogation of EGFR downstream signalling. Matuzumab has also shown antibody-dependent cellular cytotoxicity in these models (##REF##9625540##Bier et al, 1998##). Antitumour activity of matuzumab has been observed in non-clinical xenograft models (##UREF##1##Amendt et al, 2003##; ##UREF##2##Burger et al, 2003##).</p>", "<p>In a phase I study of matuzumab monotherapy in solid tumours, grade 3 headache was identified as the main dose-limiting toxicity (DLT) at 2000 mg weekly, the maximum-tolerated dose (MTD) was 1600 mg weekly and antitumour activity was seen in one heavily pre-treated oesophageal cancer patient (##REF##14701780##Vanhoefer et al, 2004##). In recent phase I studies of chemotherapy plus matuzumab in lung and pancreatic cancer (at doses ranging from 100 to 800 mg weekly), the MTD was not reached although one DLT of grade 4 neutropaenia was observed at matuzumab 800 mg combined with paclitaxel. Antitumour activity was reported and pharmacodynamic data revealed blockade of the EGFR pathway (##REF##16622465##Graeven et al, 2006##; ##REF##16533873##Kollmannsberger et al, 2006##). Preliminary data of the phase I study of PFL (cisplatin, leucovorin and 5FU) and matuzumab (at doses of 400 or 800 mg weekly) in advanced OG cancer indicate good tolerability at the 400 mg dose level (DL) (##UREF##7##Trarbach et al, 2005##).</p>", "<p>The primary objective of this phase I study was to determine the MTD, recommended dose (RD), safety, tolerability, pharmacokinetic and pharmacodynamic profile of matuzumab combined with ECX in advanced OG tumours expressing EGFR.</p>" ]

[ "<p>Thus study was supported by a research grant from Merck KGaA.</p>" ]

[ "<fig id=\"fig1\"><label>Figure 1</label><caption><p>Concentration–time courses of matuzumab in the dose groups 400 mg weekly (dashed line and circles), 800 mg weekly (continuous line and triangles) and 1200 mg every 3 weeks (bold line and squares). The mean concentrations (symbols) are fitted with a two-compartment model per dose group.</p></caption></fig>", "<fig id=\"fig2\"><label>Figure 2</label><caption><p>Pharmacodynamics of matuzumab on EGFR downstream signalling events.</p></caption></fig>" ]

[ "<table-wrap id=\"tbl1\"><label>Table 1</label><caption><title>Study design and dose escalation</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"left\"/><col align=\"char\" char=\".\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Dose level</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Matuzumab absolute dose<xref ref-type=\"fn\" rid=\"t1-fn2\">a</xref></bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>Planned no. of patients</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">1</td><td align=\"left\" valign=\"top\" charoff=\"50\">400 mg per week</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">2</td><td align=\"left\" valign=\"top\" charoff=\"50\">800 mg per week</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">3</td><td align=\"left\" valign=\"top\" charoff=\"50\">1200 mg per 3 weeks</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">4</td><td align=\"left\" valign=\"top\" charoff=\"50\">1600 mg per 3 weeks</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl2\"><label>Table 2</label><caption><title>Patient demographics</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Characteristic</bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>Total (<italic>N</italic>=21)</bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>%</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td colspan=\"3\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Gender</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Male</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">16</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">76.2</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Female</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">23.8</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td colspan=\"3\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Age (years)</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Median</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">59</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td colspan=\"3\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Karnofsky performance status</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 60</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">4.8</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 70</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">9.5</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 80</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">23.8</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 90</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">13</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">61.9</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td colspan=\"3\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Primary tumour location</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Lower one-third of oesophagus</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">23.8</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> OGJ</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">7</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">33.3</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Gastric</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">9</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">42.9</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Adenocarcinoma</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">21</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">100</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td colspan=\"3\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Stage (AJCC)</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> IIIb</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> IV</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">20</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">95</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl3\"><label>Table 3</label><caption><title>NCI-CTC all grade toxicities</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/><col align=\"left\"/><col align=\"left\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th colspan=\"4\" align=\"center\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>ECX+matuzumab 400 mg weekly <italic>N</italic>=7</bold>\n<hr/></th><th colspan=\"4\" align=\"center\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>ECX+matuzumab 800 mg weekly <italic>N</italic>=7</bold>\n<hr/></th><th colspan=\"4\" align=\"center\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>ECX+matuzumab 1200 mg every 3 weeks <italic>N</italic>=7</bold>\n<hr/></th><th colspan=\"2\" align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Adverse events</bold>\n<hr/></th></tr><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th colspan=\"4\" align=\"center\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>NCI-CTC grade</bold>\n</th><th colspan=\"4\" align=\"center\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>NCI-CTC grade</bold>\n</th><th colspan=\"4\" align=\"center\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>NCI-CTC grade</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Total</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>(%)</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Toxicity</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">4</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">4</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">4</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Rash</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">4</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">13</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">61.9</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Headache</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">4.8</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Diarrhoea</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">14</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">66.7</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Abdo pain</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">8</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">38</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">PPE</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">28.6</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Stomatitis</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">16</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">76.2</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Nausea</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">16</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">76</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Vomiting</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">4</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">14</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">66.7</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Lethargy</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6Ψ</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">16</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">76</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Neutropaenia</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">2</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">4</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">12</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">57</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Febrile neutropaenia</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">4.8</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Thrombocytopaenia</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">0</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">1</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">4.8</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl4\"><label>Table 4</label><caption><title>Objective response to ECX plus matuzumab at all dose levels</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"char\" char=\"(\"/><col align=\"char\" char=\"(\"/><col align=\"char\" char=\"(\"/><col align=\"char\" char=\"(\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>400 mg matuzumab</bold>\n</th><th align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>800 mg matuzumab</bold>\n</th><th align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>1200 mg matuzumab</bold>\n</th><th align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>Total</bold>\n</th></tr><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Dose level</bold>\n</th><th align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>weekly <italic>N</italic>=7<xref ref-type=\"fn\" rid=\"t4-fn2\">a</xref></bold>\n</th><th align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>weekly <italic>N</italic>=7</bold>\n</th><th align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>every 3 weeks <italic>N</italic>=7</bold>\n</th><th align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold><italic>N</italic>=21<xref ref-type=\"fn\" rid=\"t4-fn2\">a</xref></bold>\n</th></tr><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Response</bold>\n</th><th align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold><italic>n</italic> (%)</bold>\n</th><th align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold><italic>n</italic> (%)</bold>\n</th><th align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold><italic>n</italic> (%)</bold>\n</th><th align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold><italic>n</italic> (%)</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Partial response (PR)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">4 (66.7 [30–90])</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">3 (42.8 [16–75])</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">6 (85.7 [49–97])</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">13 (65 [43–82])</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Stable disease (SD)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2 (33.3 [10–70])</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2 (28.6 [8–64])</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1 (14.3 [3–51])</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5 (25 [11–47])</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Tumour growth control (PR+SD)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">6 (100 [61–100])</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">5 (71.4 [36–92])</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">7 (100 [65–100])</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">18 (90 [70–97])</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Progressive disease (PD)</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0 (0 [0–39])</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2 (28.6 [8–64])</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0 (0 [0–35])</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2 (10 [3–30])</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl5\"><label>Table 5</label><caption><title>Pharmacokinetic parameters of matuzumab derived by non-compartmental analysis</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"char\" char=\"(\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\"(\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\"(\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\"(\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\"(\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\"(\"/><col align=\"char\" char=\".\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th colspan=\"6\" align=\"center\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>Week 1</bold>\n<hr/></th><th colspan=\"6\" align=\"center\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>Week 4</bold>\n<hr/></th></tr><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th colspan=\"2\" align=\"center\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>400 mg per week</bold>\n</th><th colspan=\"2\" align=\"center\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>800 mg per week</bold>\n</th><th colspan=\"2\" align=\"center\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>1200 mg per 3 weeks</bold>\n</th><th colspan=\"2\" align=\"center\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>400 mg per week</bold>\n</th><th colspan=\"2\" align=\"center\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>800 mg per week</bold>\n</th><th colspan=\"2\" align=\"center\" valign=\"top\" char=\"(\" charoff=\"50\">\n<bold>1200 mg per 3 wk</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\"><italic>C</italic>_max, <italic>μ</italic>g ml⁻¹</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">154 (44)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">7</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">294 (89)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">7</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">442 (108)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">7</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">224 (54)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">495 (166)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">534 (125)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"><italic>t</italic>_max, h</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">2.04 (1.05)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">7</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.88 (1.45)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">7</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.61 (0.55)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">7</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">3.91 (1.81)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">3.88 (1.88)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">1.68 (1.63)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">AUC_{<italic>τ</italic>}, <italic>μ</italic>g ml⁻¹ h⁻¹</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">10717 (1553)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">23347 (8748)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">79189 (19217)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">19674 (4720)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">52797 (20448)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">94868 (22029)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">AUC_0–∞ <xref ref-type=\"fn\" rid=\"t5-fn1\">a</xref>, <italic>μ</italic>g ml⁻¹ h⁻¹</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">12721 (3519)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">35406 (10593)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">7</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">88066 (24277)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">7</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">NA</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">NA</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">NA</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"><italic>t</italic>_1/2, h</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">80.5 (15)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">110.8 (36.2)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">7</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">189.5 (23.2)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">7</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">131.4 (31.1)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">165 (35)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">221.3 (70.8)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">CL, h⁻¹</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.034 (0.0115)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.0243 (0.0069)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">7</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.0145 (0.0038)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">7</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.0214 (0.0054)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.0170 (0.0061)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">0.0131 (0.0025)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"><italic>V</italic><xref ref-type=\"fn\" rid=\"t5-fn2\">b</xref>, l</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">3.64 (0.53)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">6</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">3.76 (1.51)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">7</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">3.83 (1.14)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">7</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">3.92 (0.44)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">3.43 (0.96)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5</td><td align=\"char\" valign=\"top\" char=\"(\" charoff=\"50\">4.64 (1.22)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><fn id=\"t1-fn1\"><p>ECX=epirubicin, cisplatin and capecitabine.</p></fn><fn id=\"t1-fn2\"><label>a</label><p>Administered in combination with fixed-dose ECX.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"t2-fn1\"><p>ECX=epirubicin, cisplatin and capecitabine; OGJ=oesophagogastric junction.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"t3-fn1\"><p>ECX=epirubicin, cisplatin and capecitabine; NCI-CTC=National Cancer Institute Common Toxicity Criteria; Ψ=Main DLT.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"t4-fn1\"><p>ECX=epirubicin, cisplatin and capecitabine.</p></fn><fn id=\"t4-fn2\"><label>a</label><p>Response data missing from one patient.</p></fn><fn id=\"t4-fn3\"><p>The values inside square brackets denote 95% confidence interval.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"t5-fn1\"><label>a</label><p>AUC_{0–<italic>∞</italic>} is not applicable (NA) for week 4.</p></fn><fn id=\"t5-fn2\"><label>b</label><p>The values reported are <italic>V</italic>_ss for week 1 and <italic>V</italic>_Z for week 4.</p></fn><fn id=\"t5-fn3\"><p>The symbols not explained in the text are <italic>t</italic>_max, time of <italic>C</italic>_max; AUC_{<italic>τ</italic>}/AUC_{0−<italic>∞</italic>}, AUC within one dosing interval/from time 0 to infinity after single administration; CL, clearance; <italic>V</italic> (<italic>V</italic>_ss/<italic>V</italic>_Z), volume of distribution (at steady state/in the terminal phase).</p></fn><fn id=\"t5-fn4\"><p>Mean (±s.d.) is given, together with the number of underlying values in the adjacent column.</p></fn></table-wrap-foot>" ]

[ "<graphic xlink:href=\"6604622f1\"/>", "<graphic xlink:href=\"6604622f2\"/>" ]

[]

{ "acronym": [], "definition": [] }

[]

[ "<title>Patients and methods</title>", "<title>Patient eligibility criteria</title>", "<p>Patients with untreated, histologically or cytologically documented NSCLC with clinical stage IB (c-T2N0M0), IIA (c-T1N1M0) or IIB (c-T2N1M0 or T3N0M0) were eligible for study entry. Each patient was required to meet the following criteria: 20–74 years of age, Eastern Cooperative Oncology Group (ECOG) performance status (PS) of 0 or 1; measurable disease; and adequate organ function (leukocyte count ⩾4000/<italic>μ</italic>l and ⩽12 000/<italic>μ</italic>l, neutrophil count ⩾2000/<italic>μ</italic>l, platelet count ⩾10⁵/<italic>μ</italic>l, haemoglobin ⩾10.0 g dl⁻¹, serum creatinine ⩽the upper limit of the institutional normal range (ULN), creatinine clearance calculated by the Cockcroft–Gault formula ⩾60 ml min⁻¹, serum bilirubin ⩽ULN, serum ALT and AST ⩽2 × ULN and PaO₂ ⩾70 mm Hg). Women who were pregnant or lactating were excluded from the study. Other exclusion criteria included patients with active infection, unstable angina or a history of myocardial infarction within 6 months, interstitial pneumonia or active lung fibrosis, uncontrolled diabetes or hypertension, systemic use of corticosteroid or active concomitant malignancy. Patients with tumour invading the first rib or more superior chest wall (Pancoast type) were also excluded. All mediastinal nodes measuring 1.0 cm or more in size on computed tomographic (CT) scans were required to be biopsied to be histologically benign before patient entry.</p>", "<p>Patient eligibility was confirmed by the Japan Clinical Oncology Group Data Centre before registration. The study protocol was approved by the institutional review boards at each participating centre, and all patients provided written informed consent.</p>", "<title>Treatment plan</title>", "<p>This was an open-label, randomised trial. Patients were randomly assigned to one of two treatment arms. Dosages of the chemotherapy were based on the regulatory notes and clinical data in Japan (##REF##14722033##Kubota et al, 2004##). In the DP combination arm, patients received D at 60 mg m⁻² as a 1-h intravenous infusion followed by P at 80 mg m⁻² as a 2-h infusion on day 1. Two cycles of the chemotherapy were repeated at an interval of 4 weeks. The interval was permitted to be shortened to 3 weeks, if the patient was judged to have adequately recovered enough from the first cycle. Surgery (lobectomy or pneumonectomy with systematic lymph node dissection) was performed 4–5 weeks after completion or early termination of the chemotherapy. Patients in the D monotherapy arm received D at 70 mg m⁻² as a 1-h intravenous infusion on day 1. Three cycles of the chemotherapy were repeated at 3 weeks intervals. Surgery in the D arm was performed 3–4 weeks after completion or early termination of chemotherapy. The preoperative periods were thus set at 8–10 weeks in each arm, which was designed to be easier to accept for the patients and the surgeons.</p>", "<p>In each arm, when chemotherapy was judged to be ineffective with ⩾10% unidirectional tumour growth, or when the patient experienced unacceptable toxicity (such as, grade 3 neurotoxicity, grade 2 pulmonary toxicity, grade 3 cardiac toxicity or other grade 4 non-haematological toxicities), chemotherapy was discontinued and the patient was taken up for surgery as clinically indicated. With minor toxicities, such as uncomplicated grade 4 haematologic or grade 3 non-critical, non-haematological toxicities, dosages of subsequent chemotherapy courses were reduced (P by 20 mg m⁻² and D by 10 mg m⁻²).</p>", "<p>No protocol therapy was predetermined for those with unresectable tumours, either during chemotherapy or at operation, and those with microscopically or macroscopically incompletely resected tumours. Those who underwent curative resection were observed until recurrence without additional therapy.</p>", "<p>Chemotherapy was supported with routine premedication for hypersensitivity and antiemetics. For the DP arm, ample hydration was ensured. Recombinant human granulocyte colony-stimulating factor was administered when grade 4 neutropaenia or neutropaenic fever occurred.</p>", "<title>Patient evaluation and follow-up</title>", "<p>Before study enrolment, a complete medical history and physical examination, blood cell count determinations, biochemistry testing, chest X-ray, ECG, CT scan of the chest and CT scan or ultrasound of upper abdomen were conducted for each patient. Whole-brain CT or magnetic resonance imaging (MRI) or isotope bone scanning was performed if clinically indicated. Positron emission tomography (PET) was not widely available in Japan at the time of the protocol activation and was not routinely used for staging. Blood cell counts, differential WBC counts and biochemistry testing were performed weekly during each course of chemotherapy.</p>", "<p>Toxicity of the chemotherapy was evaluated with the National Cancer Institute Common Toxicity Criteria Tumour (NCI-CTC; version 2.0). Tumour responses were assessed radiographically according to the RECIST guideline (##REF##10655437##Therasse et al, 2000##). Response confirmation at 4 weeks or longer intervals was not necessitated. Response was assessed by the attending physicians in each participating institution, and no central confirmation was performed. Chest X-ray was taken at each course, and when suggested for even minor tumour growth (⩾10%), confirmatory chest CT was performed to decide on the continuation of chemotherapy.</p>", "<p>After curative resection, the patients were followed up with periodic reevaluation. This included chest CT every 6 months for the first 2 years and annually thereafter, until 5 years or tumour recurrence.</p>", "<title>Statistical considerations</title>", "<p>This trial was designed as a randomised phase II selection design. Therefore, formal statistical hypothesis testing of the differences between the arms, including the calculation of <italic>P</italic>-values, was not to be performed. The aim was to select the ‘preferable’ preoperative chemotherapy arm for a future definitive phase III trial, with the DFS rate at 1 year as primary end point. The DFS was calculated from the date of enrolment by the Kaplan–Meier method, as was the overall survival (OS).The ‘events’ for the determination of the DFS included tumour relapse after curative surgery, death from any cause and non-curative operation. Those with non-curative operation include patients without surgery and those with incomplete resection, either microscopically or macroscopically. Non-curative operation was to be counted as an event on the date of registration, not on that of surgery. The sample size was determined to provide sufficient probability to choose the ‘preferable’ arm (##REF##4075313##Simon et al, 1985##). Assuming DFS rates at 1 year of 70 and 80%, 40 patients per arm were required to correctly select the arm that is not inferior with the probability of 84.9%. The ‘minimal’ DFS rate of 70% was assumed with the prior report from North America, in which the 1- and 2-year survival rates were reported to be 85 and 56%, respectively (##REF##10694600##Pisters et al, 2000##). The assumption was rough and might well be inaccurate, for no DFS data were available from the literature. The randomisation was carried out by the JCOG data centre using a minimisation method with c-stage (IB <italic>vs</italic> II) and institutions as balancing factors.</p>", "<p>The secondary end points included the objective tumour response to chemotherapy, complete resection rate, intra- and post-surgical morbidity/mortality and the OS rate. Tumour responses in both arms were compared using Fisher's exact test.</p>", "<p>During the accrual period, an interim analysis was planned after 40 patients were randomised and followed up for at least 4 months. All analyses were performed with the SAS software version 9.1 (SAS Institute, Cary, NC, USA).</p>" ]

[ "<title>Results</title>", "<title>Patient characteristics</title>", "<p>From October 2002 to October 2003, 80 patients from 18 institutions were enroled and randomised. After 40 patients were randomised, an interim analysis was carried out. Following the JCOG Data and Safety Monitoring Committee's review, the study was continued. One patient in the D arm was found to be ineligible because of the wrong histology (sarcoma). All 80 patients were analysed for characteristics and chemotherapy toxicity, and the 79 eligible patients were analysed for the clinical and pathological response to chemotherapy, surgical results, DFS and OS.</p>", "<p>##TAB##0##Table 1## lists the characteristics of the patients, which were well balanced between the arms.</p>", "<title>Chemotherapy delivery and toxicity</title>", "<p>##TAB##1##Table 2## summarises the chemotherapy delivery, and ##TAB##2##Table 3## summarises toxicity in the subject group. Only 60% in the D arm completed the planned chemotherapy courses, mainly arising from the clinical ineffectiveness of the therapy. On the other hand, compliance was very good in the DP arm, and the toxicity was not greater. Hyponatraemia, probably due to hydration with P administration, was an unexpected toxicity in the DP arm, but it was clinically silent and transient in all the cases. All patients recovered without any particular management, with no clinically relevant sequelae. Other toxicities were mainly haematologic, and both chemotherapy arms were generally well tolerated by the patients.</p>", "<title>Clinical response and pathological results</title>", "<p>##TAB##3##Table 4## shows the clinical responses to the chemotherapy. The overall response rates, 45% in the DP arm and 15% in the D arm, were compatible with earlier reports for each of the chemotherapy regimen in patients with NSCLC.</p>", "<p>Thoracotomy was performed in 39 of the 40 patients in the DP arm, and in 37 of the 39 patients in the D arm. The tumour was surgically resected in 39 (98%) patients in the DP arm, including pneumonectomy in 3 cases, bi-lobectomy in 2 cases and lobectomy in 34 cases. Tumour resection was performed in 35 (90%) patients of the D arm, including pneumonectomy in 1 case, bi-lobectomy in 4 cases and lobectomy in 30 cases. Five patients, including four in the DP arm and one in the D arm, suffered from massive (⩾1 l) intraoperative bleeding: due to severe adhesion in three cases (two in DP and one in D arm), to incomplete suture of the autostapler resulting in injury of pulmonary artery in one case (DP arm) and accidental injury to the aorta in one case (DP arm). None was judged to be related to preoperative therapy. The postoperative complications included one patient with empyema and another with pulmonary oedema, both in the DP arm. There were two surgical deaths, both in the DP arm; one died on postoperative day 59 because of empyema, and another on postoperative day 2 because of massive intraoperative bleeding resulting from surgical injury to the aorta.</p>", "<p>Pathological complete resection (R0), without residual tumour found either macroscopically or microscopically, was achieved in 38 (95%) cases in the DP arm, and 34 (87%) cases in the D arm. On pathological examination, 23% of the 75 patients who underwent surgery were found to have N2 or N3 status. Pathologic CR was achieved in two patients, both in the DP arm. Clinical N-stage was poorly correlated to pathological nodal status (##TAB##4##Table 5##).</p>", "<title>DFS and OS</title>", "<p>The DFS and OS were updated in November 2007. The DFS rates at 1, 2 and 4 years were 78, 65 and 57% in the DP arm, and were 62, 44 and 36% in the D arm, respectively (##FIG##0##Figure 1##). ##TAB##5##Table 6## summarises the outcome at 1 year, the primary end point of the study. The DFS rate at 1 year was 78% (31 out of 40) in the DP arm, which was consistent with the study assumption that it would be 80% in the ‘better’ arm, whereas it was a disappointing 62% (24 out of 39) in the D arm. The 16% difference was more than presumed in the protocol.</p>", "<p>The OS rates at 1, 2 and 4 years were 88, 83 and 75% in the DP arm, and were 87, 72 and 57% in the D arm, respectively (##FIG##1##Figure 2##). Both the DFS and the OS rates were better in the DP arm. The OS was better in the DP arm in both adenocarcinoma and non-adenocarcinoma histological subtypes.</p>" ]

[ "<title>Discussion</title>", "<p>As compared with post-surgical adjuvant therapy, preoperative chemotherapy has several practical as well as theoretical advantages (##REF##10694600##Pisters et al, 2000##; ##REF##12781072##Pisters, 2003##). The practical advantages include better patient tolerance and clinical visualisation of chemotherapy effect.</p>", "<p>There are very few reports as to the optimal preoperative therapy strategy. The majority of trials have used ‘standard’ platinum-based doublets (##REF##12781072##Pisters, 2003##). Although they are the ‘standard’ for advanced, stage IV NSCLC, a trade-off between the cytotoxic effect and toxicity of the chemotherapy, not only toxicity itself but also its influence on surgery and post-surgical morbidity and mortality (##REF##11773176##Depierre et al, 2002##; ##UREF##0##Pisters et al, 2007##), must be considered for preoperative therapy.</p>", "<p>In this randomised phase II study, we evaluated DP combination chemotherapy and D monotherapy as preoperative treatment for early stage NSCLC. Although the DFS assumptions of the protocol, 70 <italic>vs</italic> 80% at 1 year, were rough and arbitrary due to lack of historical data, subsequent S9900 trial (##UREF##0##Pisters et al, 2007##) showed DFS rate of 68% in the surgery alone group and 69% in those with preoperative carboplatin–paclitaxel therapy, consistent with our assumption.</p>", "<p>Our results showed that single-agent D was inadequate in this setting; an unexpectedly high progression rate led to an early chemotherapy termination rate of as high as 40%. The reason for the high PD rate is unknown. In addition, we tried to minimise the disadvantage of continuation of ineffective chemotherapy by defining the ineffectiveness as ⩾10% tumour size increase instead of ⩾20% in the RECIST guideline (##REF##10655437##Therasse et al, 2000##). This subtle decision rule might require centralised confirmation. The DFS rate in the D arm was disappointing and was, in fact, very similar to that in the surgery-alone arm in the S9900 study in the United States (##UREF##0##Pisters et al, 2007##).</p>", "<p>On the other hand, both the DFS and OS rates of the DP arm were promising. Disease-free survival at 1 year of 78% was fully consistent with the estimation in the study protocol. Although our data do not refute other platinum-based chemotherapy as candidates of preoperative treatment, it would be justified to conclude that DP was active and promising, regardless of disappointing data of D monotherapy. One might argue that DFS at 1 year was too premature as an end point. Because the DFS and OS curves of the DP arm seem to have reached to plateau at 2 years, DFS at 2 years might be a more appropriate end point.</p>", "<p>The number of chemotherapy courses of the DP combination was two, whereas many previous studies used three courses. In the North American trials with carboplatin and paclitaxel, three preoperative courses appeared to have no advantage when compared with two courses (##REF##10694600##Pisters et al, 2000##; ##REF##12781072##Pisters, 2003##). Although patients with ‘two preoperative courses’ were to have two additional courses after the operation, compliance to the post-surgical courses was very poor anyway (##REF##10694600##Pisters et al, 2000##). But, as the majority of the patients appeared fit enough after two courses of DP and a major operation, we could consider the addition of a couple of postoperative chemotherapy cycles at least for responders.</p>", "<p>One of the major disadvantages of preoperative therapy is the inaccuracy of the clinical staging, as reported by ##REF##11773176##Depierre et al (2002)##. In our trial, 23% of the 74 patients who underwent thoracotomy were found to have p-N2/N3 disease. In Japan, mediastinoscopy for patients with mediastinal nodes measuring 1 cm or less in size on CT is not performed as a routine clinical practise, and nor was it in our study. Although the introduction of PET may improve the accuracy of the clinical staging, it would still be unlikely to be comparable to surgical staging (##REF##12815135##Lardinois et al, 2003##; ##REF##15337041##Cerfolio et al, 2004##; ##REF##16014441##Shim et al, 2005##). This would inevitably lead to heterogeneity of the patient population, necessitating a sophisticated study design and large sample size for any future trial on preoperative therapy.</p>", "<p>We conclude that the DP combination regimen is active and well tolerated as preoperative chemotherapy, with highly promising survival data. Future clinical trials are warranted based on our results.</p>" ]

[]

[ "<p>Preoperative chemotherapy is a promising strategy in patients with early-stage resectable non-small-cell lung cancer (NSCLC); optimal chemotherapy remains unclear. Clinical (c-) stage IB/II NSCLC patients were randomised to receive either two cycles of docetaxel (D)–cisplatin (P) combination chemotherapy (D 60 mg m⁻² and P 80 mg m⁻² on day 1) every 3–4 weeks or three cycles of D monotherapy (70 mg m⁻²) every 3weeks. Thoracotomy was performed 4–5 weeks (DP) or 3–4 weeks (D) after chemotherapy. The primary end point was 1-year disease-free survival (DFS). From October 2002 to November 2003, 80 patients were randomised. Chemotherapy toxicities were mainly haematologic and well tolerated. There were two early postoperative deaths with DP (one intraoperative bleeding and one empyema). Pathologic complete response was observed in two DP patients. Docetaxel–cisplatin was superior to D in terms of response rate (45 <italic>vs</italic> 15%) and complete resection rate (95 <italic>vs</italic> 87%). Both DFS and overall survival were better in DP. Disease-free survival at 1, 2 and 4 years were 78, 65 and 57% with DP, and were 62, 44 and 36% with D, respectively. Preoperative DP was associated with encouraging resection rate and DFS data, and phase III trials for c-stage IB/II NSCLC are warranted.</p>" ]

[ "<p>Surgery is the standard of care for clinical (c-) stage IB/II non-small-cell lung cancer (NSCLC), but the treatment outcome remains poor, with 5-year survival rates of 50% or less (##REF##9187198##Mountain, 1997##; ##REF##16061304##Goya et al, 2005##). The majority of post-surgical relapse occurs as distant metastases (##REF##15886314##Pisters and Le Chevalier, 2005##); therefore, effective systemic therapy is necessary. Recently, a series of postoperative adjuvant chemotherapy trials reported modest but significant improvement in survival, mainly in patients with pathological stage II or IIIA NSCLC (##REF##14736927##Arriagada et al, 2004##; ##REF##16000605##Scagliotti, 2005##; ##REF##15972865##Winton et al, 2005##; ##REF##16945766##Douillard et al, 2006##). Compliance to the chemotherapy remains a problem (##REF##14736927##Arriagada et al, 2004##; ##REF##16000605##Scagliotti, 2005##; ##REF##15972865##Winton et al, 2005##; ##REF##16945766##Douillard et al, 2006##).</p>", "<p>On the other hand, previous small phase III trials had reported that preoperative chemotherapy was better than surgery alone in stage III NSCLC (##REF##8043059##Rosell et al, 1994##; ##REF##8158698##Roth et al, 1994##). Recent trials of preoperative platinum-based chemotherapy have reported promising results in c-stage IB/II NSCLC (##REF##10694600##Pisters et al, 2000##; ##REF##11773176##Depierre et al, 2002##; ##REF##12445742##Rosell et al, 2002##). One advantage of the preoperative chemotherapy is better tolerability and compliance.</p>", "<p>No data are available, however, as to the optimal preoperative therapy strategy for early-stage NSCLC. Although platinum-based ‘standard’ combination chemotherapy regimens have widely been used and reported to be generally safe, results of randomised trials reported nonsignificant but modest excess of post-surgical morbidity and mortality (##REF##11773176##Depierre et al, 2002##; ##UREF##0##Pisters et al, 2007##). Monotherapy with an active agent is associated with lower response rate, but less toxicity (##REF##15280345##Delbaldo et al, 2004##); it might well be favourable for preoperative therapy in early stage, when surgery must not be compromised by adjuvant therapy.</p>", "<p>Docetaxel (D) is a semisynthetic taxoid derived from the European yew <italic>Taxus baccata</italic>. It is active against NSCLC either in monotherapy (D)(##REF##7911160##Fossella et al, 1994##; ##REF##7911159##Francis et al, 1994##; ##REF##8622084##Kunitoh et al, 1996##) or in combination with cisplatin (DP) (##REF##9586914##Zalcberg et al, 1998##; ##REF##12837811##Fossella et al, 2003##). In advanced NSCLC, DP was reported to be better than P–vinca combination (##REF##12837811##Fossella et al, 2003##; ##REF##14722033##Kubota et al, 2004##), one of the ‘standard’ adjuvant therapies. The DP combination was also reported to be active and promising as preoperative chemotherapy in c-stage III NSCLC (##REF##16622435##Betticher et al, 2006##).</p>", "<p>Docetaxel monotherapy, on the other hand, was reported to be not inferior to DP, with better tolerability in advanced NSCLC (##REF##15226327##Georgoulias et al, 2004##). For stage III NSCLC, ##REF##12488303##Mattson et al (2003)## reported the results of D as preoperative chemotherapy; it was active, and did not compromise surgery.</p>", "<p>On the basis of this rationale, we undertook a randomised phase II trial of DP <italic>vs</italic> D in resectable, c-stage IB/II NSCLC. The objectives of the study were to evaluate the safety and efficacy of the preoperative chemotherapy and to select promising one for future phase III trials. The primary end point was the disease-free survival (DFS) rate at 1 year.</p>" ]

[ "<p>We thank Ms Mieko Imai for the data management, and Mr Takashi Asakawa and Dr Naoki Ishizuka for the statistical analyses. This study was supported by the Grant-in-Aid for Cancer Research (114S-2, 14S-4, 17S-2, 17S-5) and Health Sciences Research Grant from the Ministry of Health, Labour and Welfare of Japan. Presented in part at the 40th Annual Meeting of the American Society of Clinical Oncology, 5–8 June 2004, New Orleans, LA, and at the 11th World Conference on Lung Cancer, 3–6 July 2005, Barcelona, Spain. Registered in <ext-link xmlns:xlink=\"http://www.w3.org/1999/xlink\" ext-link-type=\"uri\" xlink:href=\"http://www.clinicaltrials.gov\">www.clinicaltrials.gov</ext-link>. ClinicalTrials.gov number, NCT00132639.</p>", "<title>Appendix</title>", "<p>The following institutions and investigators participated in the trial:</p>", "<p>Tohoku University Hospital (Takashi Kondo and Akira Sakurada), Tochigi Cancer Center (Haruhisa Matsuguma), Saitama Cancer Center (Hirohiko Akiyama), National Cancer Center Hospital East (Kanji Nagai, Junji Yoshida and Nagahiro Saijo), National Cancer Center Hospital (Hisao Asamura, Kenji Suzuki and Hideo Kunitoh), Kyorin University School of Medicine (Tomoyuki Goya and Yoshihiko Koshiishi), Tokyo Medical University (Harubumi Kato and Masahiro Tsuboi), Cancer Institute Hospital (Ken Nakagawa and Yukitoshi Satoh), Yokohama Munipical Citizen's Hospital (Koshiro Watanabe and Jun-ichi Nitadori), Niigata Cancer Center Hospital (Teruaki Koike and Yasushi Yamato), Aichi Cancer Center Hospital (Tetsuya Mitsudomi and Shoichi Mori), Osaka Prefectural Hospital Organization Osaka Medical Center for Cancer and Cardiovascular Diseases (Ken Kodama and Masahiko Higashiyama), Osaka Prefectural Hospital Organization Osaka Prefectural Medical Center for Respiratory and Allergic Disease (Mitsunori Ota), Osaka City General Hospital (Hirohito Tada and Ryoji Yamamoto), Hyogo Cancer Center (Morihito Okada, Masahiro Yoshimura and Koichiro Iwanaga), National Hospital Organization Shikoku Cancer Center (Motohiro Yamashita), National Kyushu Cancer Center (Yukito Ichinose and Koji Yamazaki), Nagasaki University School of Medicine (Takeshi Nagayasu and Tsutomu Tagawa).</p>" ]

[ "<fig id=\"fig1\"><label>Figure 1</label><caption><p>Disease-free survival.</p></caption></fig>", "<fig id=\"fig2\"><label>Figure 2</label><caption><p>Overall survival.</p></caption></fig>" ]

[ "<table-wrap id=\"tbl1\"><label>Table 1</label><caption><title>Patient characteristics</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Arm</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Cisplatin–docetaxel</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Docetaxel alone</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>N</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\">40</td><td align=\"center\" valign=\"top\" charoff=\"50\">40</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Clinical stage</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> IB</td><td align=\"center\" valign=\"top\" charoff=\"50\">22</td><td align=\"center\" valign=\"top\" charoff=\"50\">23</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> II</td><td align=\"center\" valign=\"top\" charoff=\"50\">18</td><td align=\"center\" valign=\"top\" charoff=\"50\">17</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Clinical T stage</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> T1</td><td align=\"center\" valign=\"top\" charoff=\"50\">5</td><td align=\"center\" valign=\"top\" charoff=\"50\">5</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> T2</td><td align=\"center\" valign=\"top\" charoff=\"50\">31</td><td align=\"center\" valign=\"top\" charoff=\"50\">29</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> T3</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td><td align=\"center\" valign=\"top\" charoff=\"50\">6</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Clinical N stage</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> N0</td><td align=\"center\" valign=\"top\" charoff=\"50\">26</td><td align=\"center\" valign=\"top\" charoff=\"50\">28</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> N1</td><td align=\"center\" valign=\"top\" charoff=\"50\">14</td><td align=\"center\" valign=\"top\" charoff=\"50\">12</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>ECOG performance status</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> PS0</td><td align=\"center\" valign=\"top\" charoff=\"50\">35</td><td align=\"center\" valign=\"top\" charoff=\"50\">31</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> PS1</td><td align=\"center\" valign=\"top\" charoff=\"50\">5</td><td align=\"center\" valign=\"top\" charoff=\"50\">9</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Histology</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Adenocarcinoma</td><td align=\"center\" valign=\"top\" charoff=\"50\">30</td><td align=\"center\" valign=\"top\" charoff=\"50\">24</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Squamous cell carcinoma</td><td align=\"center\" valign=\"top\" charoff=\"50\">10</td><td align=\"center\" valign=\"top\" charoff=\"50\">11</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Others</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">5</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Body weight loss within 6 months</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> None</td><td align=\"center\" valign=\"top\" charoff=\"50\">24</td><td align=\"center\" valign=\"top\" charoff=\"50\">22</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> ⩽5 kg</td><td align=\"center\" valign=\"top\" charoff=\"50\">13</td><td align=\"center\" valign=\"top\" charoff=\"50\">14</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> &gt;5 kg</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Missing</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Smoking</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Median smoking</td><td align=\"center\" valign=\"top\" charoff=\"50\">40 pack-years</td><td align=\"center\" valign=\"top\" charoff=\"50\">40 pack-years</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Never-smoker</td><td align=\"center\" valign=\"top\" charoff=\"50\">6</td><td align=\"center\" valign=\"top\" charoff=\"50\">8</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl2\"><label>Table 2</label><caption><title>Delivery of chemotherapy</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Arm</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Cisplatin–docetaxel</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Docetaxel alone</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>N</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\">40</td><td align=\"center\" valign=\"top\" charoff=\"50\">40</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Completed</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\">38 (95%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">24 (60%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Not completed</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\">16</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Ineffective<xref ref-type=\"fn\" rid=\"t2-fn1\">a</xref></td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">11</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Adverse event</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Patient refusal</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Found ineligible</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl3\"><label>Table 3</label><caption><title>Toxicity of chemotherapy</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Arm</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Cisplatin–docetaxel</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Docetaxel alone</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>N</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\">40</td><td align=\"center\" valign=\"top\" charoff=\"50\">40</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Grade</td><td align=\"center\" valign=\"top\" charoff=\"50\">2/3/4 (% grade 3+4)</td><td align=\"center\" valign=\"top\" charoff=\"50\">2/3/4 (% grade 3+4)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Haematological</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Leukopaenia</td><td align=\"center\" valign=\"top\" charoff=\"50\">18/14/1 (38)</td><td align=\"center\" valign=\"top\" charoff=\"50\">12/15/2 (43)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Neutropaenia</td><td align=\"center\" valign=\"top\" charoff=\"50\">5/16/17 (83)</td><td align=\"center\" valign=\"top\" charoff=\"50\">5/10/21 (78)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Anaemia</td><td align=\"center\" valign=\"top\" charoff=\"50\">4/0/0 (0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">7/0/0 (0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Thrombocytopenia</td><td align=\"center\" valign=\"top\" charoff=\"50\">1/0/0 (0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0/0/0 (0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Nonhaematological</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Total bilirubin</td><td align=\"center\" valign=\"top\" charoff=\"50\">4/0/0 (0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0/0/0 (0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Serum AST</td><td align=\"center\" valign=\"top\" charoff=\"50\">0/0/0 (0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">3/1/0 (3)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Serum ALT</td><td align=\"center\" valign=\"top\" charoff=\"50\">5/0/0 (0</td><td align=\"center\" valign=\"top\" charoff=\"50\">5/1/0 (3)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Serum creatinine</td><td align=\"center\" valign=\"top\" charoff=\"50\">3/0/0 (0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0/0/0 (0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Hypoxia</td><td align=\"center\" valign=\"top\" charoff=\"50\">0/0/0 (0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">3/0/0 (0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Hypercalcaemia</td><td align=\"center\" valign=\"top\" charoff=\"50\">0/0/0 (0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0/1/0 (3)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Hyponatraemia</td><td align=\"center\" valign=\"top\" charoff=\"50\">−/6/0 (15)</td><td align=\"center\" valign=\"top\" charoff=\"50\">−/1/0 (3)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Hypersensitivity</td><td align=\"center\" valign=\"top\" charoff=\"50\">0/0/0 (0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0/1/0 (3)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Fatigue</td><td align=\"center\" valign=\"top\" charoff=\"50\">3/1/0 (3)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0/0/0 (0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Constipation</td><td align=\"center\" valign=\"top\" charoff=\"50\">4/1/0 (3)</td><td align=\"center\" valign=\"top\" charoff=\"50\">5/0/0 (0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Diarrhea</td><td align=\"center\" valign=\"top\" charoff=\"50\">3/3/0 (8)</td><td align=\"center\" valign=\"top\" charoff=\"50\">2/0/0 (0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Nausea</td><td align=\"center\" valign=\"top\" charoff=\"50\">9/7/− (18)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0/0/− (0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Vomiting</td><td align=\"center\" valign=\"top\" charoff=\"50\">5/1/0 (3)</td><td align=\"center\" valign=\"top\" charoff=\"50\">0/0/0 (0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Febrile neutropaenia</td><td align=\"center\" valign=\"top\" charoff=\"50\">−/1/0 (3)</td><td align=\"center\" valign=\"top\" charoff=\"50\">−/0/0 (0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Infection with neutropaenia</td><td align=\"center\" valign=\"top\" charoff=\"50\">−/2/0 (5)</td><td align=\"center\" valign=\"top\" charoff=\"50\">−/3/0 (8)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Infection without neutropaenia</td><td align=\"center\" valign=\"top\" charoff=\"50\">1/0/0 (0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">4/2/0 (5)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> Neuropathy</td><td align=\"center\" valign=\"top\" charoff=\"50\">0/0/0 (0)</td><td align=\"center\" valign=\"top\" charoff=\"50\">1/0/0 (0)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Any grade 3/4 toxicity</td><td align=\"center\" valign=\"top\" charoff=\"50\">35 (88%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">32 (80%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Any grade 3/4</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Non-haematological toxicity</td><td align=\"center\" valign=\"top\" charoff=\"50\">15 (38%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">9 (23%)</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl4\"><label>Table 4</label><caption><title>Clinical response to chemotherapy based on RECIST</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Arm</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Cisplatin–docetaxel</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Docetaxel alone</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>N</italic>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\">40</td><td align=\"center\" valign=\"top\" charoff=\"50\">39</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Completed chemotherapy</td><td align=\"center\" valign=\"top\" charoff=\"50\">38 (95%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">24 (62%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">CR</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">PR</td><td align=\"center\" valign=\"top\" charoff=\"50\">17</td><td align=\"center\" valign=\"top\" charoff=\"50\">6</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">CR+PR</td><td align=\"center\" valign=\"top\" charoff=\"50\">18</td><td align=\"center\" valign=\"top\" charoff=\"50\">6</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">SD</td><td align=\"center\" valign=\"top\" charoff=\"50\">18</td><td align=\"center\" valign=\"top\" charoff=\"50\">23</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">PD</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td><td align=\"center\" valign=\"top\" charoff=\"50\">10</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">NE</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">ORR</td><td align=\"center\" valign=\"top\" charoff=\"50\">45%</td><td align=\"center\" valign=\"top\" charoff=\"50\">15%</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">(95% confidence interval)</td><td align=\"center\" valign=\"top\" charoff=\"50\">(29–62%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">(6–31%)</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl5\"><label>Table 5</label><caption><title>Pathological results</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Arm</bold>\n</th><th colspan=\"3\" align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Cisplatin–docetaxel</bold>\n<hr/></th><th colspan=\"3\" align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Docetaxel alone</bold>\n<hr/></th></tr><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>c-N stage</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>N0</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>N1</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Total</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>N0</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>N1</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Total</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Number of cases</td><td align=\"center\" valign=\"top\" charoff=\"50\">26</td><td align=\"center\" valign=\"top\" charoff=\"50\">14</td><td align=\"center\" valign=\"top\" charoff=\"50\">40</td><td align=\"center\" valign=\"top\" charoff=\"50\">27</td><td align=\"center\" valign=\"top\" charoff=\"50\">12</td><td align=\"center\" valign=\"top\" charoff=\"50\">39</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">p-N0</td><td align=\"center\" valign=\"top\" charoff=\"50\">17</td><td align=\"center\" valign=\"top\" charoff=\"50\">6</td><td align=\"center\" valign=\"top\" charoff=\"50\">23</td><td align=\"center\" valign=\"top\" charoff=\"50\">18</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td><td align=\"center\" valign=\"top\" charoff=\"50\">22</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">p-N1</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\">5</td><td align=\"center\" valign=\"top\" charoff=\"50\">8</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">p-N2</td><td align=\"center\" valign=\"top\" charoff=\"50\">5</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\">7</td><td align=\"center\" valign=\"top\" charoff=\"50\">5</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td><td align=\"center\" valign=\"top\" charoff=\"50\">9</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">p-N3</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Not assessable</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl6\"><label>Table 6</label><caption><title>Outcome at 1 year</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Arm</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Cisplatin–docetaxel</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Docetaxel alone</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Total</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Number of cases</td><td align=\"center\" valign=\"top\" charoff=\"50\">40</td><td align=\"center\" valign=\"top\" charoff=\"50\">39</td><td align=\"center\" valign=\"top\" charoff=\"50\">79</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Alive, disease-free</td><td align=\"center\" valign=\"top\" charoff=\"50\">31</td><td align=\"center\" valign=\"top\" charoff=\"50\">24</td><td align=\"center\" valign=\"top\" charoff=\"50\">55</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Alive with disease</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td><td align=\"center\" valign=\"top\" charoff=\"50\">11</td><td align=\"center\" valign=\"top\" charoff=\"50\">15</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Dead, due to cancer</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\">5</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Dead, treatment-related</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Dead, other causes</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<fn-group><fn><p>\n<bold>Conflict of interest</bold>\n</p><p>Hideo Kunitoh, Masahiro Tsuboi, Yukito Ichinose and Nagahiro Saijo have received honoraria from Sanofi-Aventis.</p></fn></fn-group>", "<table-wrap-foot><fn id=\"t2-fn1\"><label>a</label><p>Ineffectiveness was judged upon ⩾10% unidirectional increase in tumour size, and did not necessarily mean progressive disease by RECIST.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"t4-fn1\"><p>CR=complete response; NE=not evaluable; ORR=overall response rate; PD=progressive disease; PR=partial response; RECIST=Response Evaluation Criteria in Solid Tumor; SD=stable disease.</p></fn></table-wrap-foot>" ]

[ "<graphic xlink:href=\"6604613f1\"/>", "<graphic xlink:href=\"6604613f2\"/>" ]

[]

{ "acronym": [], "definition": [] }

[]

[]

[]

[]

[]

[ "<p>Within the era of molecularly targeted anticancer agents, it has become increasingly important to provide proof of mechanism as early on as possible in the drug development cycle, especially in the clinic. Selective activation of apoptosis is often cited as one of the major goals of cancer chemotherapy. Thus, the present minireview focuses on a discussion of the pros and cons of a variety of methodological approaches to detect different components of the apoptotic cascade as potential biomarkers of programmed cell death. The bulk of the discussion centres on serological assays utilising the technique of ELISA, since here there is an obvious advantage of sampling multiple time points. Potential biomarkers of apoptosis including circulating tumour cells, cytokeratins and DNA nucleosomes are discussed at length. However, accepting that a single biomarker may not have the power to predict proof of concept and patient outcome, it is clear that in the future more emphasis will be placed on technologies that can analyse panels of biomarkers in small volumes of samples. To this end the increased throughput afforded by multiplex ELISA technologies is discussed.</p>" ]

[ "<p>A biomarker is a characteristic objectively measured and evaluated to indicate normal or pathogenic biological processes or pharmacologic response. Its potential to enhance translational science progress and accelerate drug development is becoming recognised. Nowhere is this more pertinent than in the complex arena of anticancer drug development, where the rate of compound attrition is high and success rates in the clinic are low (##REF##15286737##Kola and Landis, 2004##). Biomarkers may facilitate rational decision-making during drug discovery and in pre-clinical drug evaluation (##REF##17108987##Collins and Workman, 2006##). In addition, pharmacodynamic biomarkers allow real-time monitoring of drug efficacy and identify early signs of toxicity during clinical drug evaluation, while stratification biomarkers should facilitate selection of patients most likely to respond (##REF##16239904##Ludwig and Weinstein, 2005##).</p>", "<p>Suppression of apoptosis is a hallmark of human cancer (##REF##12098689##Weinstein, 2002##) and a desired end point of many targeted therapies is induction of tumour cell death. Mechanism-based therapies under clinical evaluation in oncology may directly induce apoptosis by targeting molecular components of apoptosis regulatory pathways (##UREF##0##Taylor et al, 2006##), or do so indirectly, following drug target modulation that is then coupled to apoptosis. Either way, application of informative, validated biomarkers of apoptosis in clinical trials of anti-cancer therapies is urgently required.</p>", "<title>Apoptosis</title>", "<p>Cell death can occur by mechanisms including necrosis, mitotic catastrophe and autophagy (##REF##18000678##Taatjes et al, 2008##). However, apoptotic cell death regulators are currently considered to have significant potential as targets for cancer therapeutics. Morphological changes during apoptosis include plasma membrane blebbing, cell shrinkage, chromatin condensation and formation of apoptotic bodies (##REF##11684438##Makin and Dive, 2001##). The biochemistry of apoptosis is summarised in three stages, the activation of initiator caspases, mitochondrial release of ‘apoptogens’ and finally the activation of effector caspases, which cleave recognised substrates to dismantle the dying cell.</p>", "<p>Molecularly, apoptosis is activated via either the death-receptor-mediated extrinsic pathway or the mitochondria-directed intrinsic pathway. The extrinsic pathway, triggered by ligands binding plasma membrane death receptors, leads to activation of initiator caspase 8 (##REF##16394652##Fas et al, 2006##). In certain cell types, this directly activates effector caspases, such as caspase 3, whereas in others (and in most cancer cells) caspase 8 can amplify death signalling by engaging the intrinsic pathway. The latter is controlled by pro- and anti-apoptotic Bcl-2 family proteins where, upon an apoptotic stimulus, changes in intrafamily protein interactions at the mitochondrial surface determine the release of cytochrome <italic>c</italic>. Cytosolic cytochrome <italic>c</italic> activates the apoptosome complex, initiator caspase 9 and the effector caspases. Caspase cleavage of cytokeratins (CKs), poly(ADP-ribose) polymerase and activation of endonucleases (to generate nucleosomal DNA (nDNA)) form a cascade of irreversible events and lead to the formation of apoptotic bodies. They also promote exposure of phosphatidylserine on the external surface of the plasma membrane, which allows phagocyte recognition of the dying cell.</p>", "<p>Many of the above-mentioned molecular events are potential biomarkers of apoptosis (##FIG##0##Figure 1##), (##REF##15930332##Singhal et al, 2005##) (##TAB##0##Table 1##). However, detection of apoptosis <italic>in vivo</italic> is challenging; it is generally asynchronous and the half-life of apoptotic cells in tissues is short. Thus, time of analysis is critical regarding the levels of apoptosis detected in patient samples. Apoptosis kinetics are dependent upon the drug's mechanism of action, its pharmacokinetics and critically, the apoptotic threshold of the cells in question.</p>", "<title>Application of biomarkers in clinical trials</title>", "<p>Biomarker qualification is the preferred terminology for the evidentiary procedure of causally linking a biomarker to a biological process, pharmacodynamic (PD) effect or clinical end point (##REF##12364809##Wagner, 2002##). This is a lengthy process requiring retrospective and prospective clinical trials and large population screening. This duration may not be necessary for biomarkers intended for early-phase drug discovery (##REF##15846456##Lee et al, 2005##). Therefore, qualification and method validation requirements depend upon the inherent assay quantitation and the position of biomarkers in the spectrum toward the clinical end point (##REF##17876307##Cummings et al, 2008##). Nonetheless, emphasising on mechanistic studies at the pre-clinical stage with robust PD biomarker assays will increase prospects of successful outcomes in the clinic (##REF##17161682##Sarker and Workman, 2007##).</p>", "<title>Ideal characteristics of a biomarker</title>", "<p>Use of PD biomarkers in early clinical trials extends hypothesis testing and confirms whether a drug hits a tumour target (proof of mechanism, POM) and thereafter, the anticipated tumour outcome is reached (proof of concept, POC). The PD biomarkers compared before and after drug treatment may reflect changes in drug target (e.g., protein phosphorylation, direct DNA damage and enzyme activity), or those distal to target hitting (e.g., downstream signalling events, changes in gene expression). Cell fate after target modulation should then be apparent (e.g., changes in proliferation, apoptosis or angiogenesis) (##REF##15136593##Hidalgo, 2004##). Biomarkers measured in tumour and/or surrogate body fluids encompass a broad range of molecules (proteins, nucleic acids, lipids and sugars) as well as circulating intact cells. An ideal biomarker should provide a minimally invasive/non-invasive indirect continuous readout of disease/drug activity.</p>", "<p>Ideal biomarkers should therefore aspire to the following criteria:\n<list list-type=\"bullet\"><list-item><p>Specificity for the biological process/target</p></list-item><list-item><p>Accurately quantifiable in clinical samples with sufficient dynamic range to detect change upon drug treatment</p></list-item><list-item><p>Provide a rapid, reliable and robust measurement</p></list-item><list-item><p>Validateable and validated to internationally recognised standards</p></list-item><list-item><p>Exhibit little overlap in levels between untreated patients and treated patients</p></list-item><list-item><p>Have baseline levels not subject to wide variations between patients</p></list-item><list-item><p>Have levels that correlate with the total burden of disease and unaffected by unrelated conditions</p></list-item><list-item><p>Have levels that correlate closely with the proximal or distal effects of therapy, thereby aiding POM or POC</p></list-item><list-item><p>Measurable in a readily obtainable clinical sample</p></list-item></list></p>", "<p>The best approach to chart the progress of apoptosis in a patient's tumour directly before and after a therapeutic intervention would be to interrogate serial tumour biopsies. However, this is usually impractical and unacceptable to the patient. Biomarker analysis procedures that are less invasive must be adopted, including biomedical imaging and detection in more readily obtainable samples such as biological fluids. As the scope of imaging technologies expands and relationships between drug-induced changes in blood-borne biomarkers and tumour images (that detect changes in volume, metabolic activity, and so on) are revealed, a more comprehensive understanding of drug effects in tumours will emerge. There is no ideal biomarker that meets all the above criteria under all circumstances. This is precisely because fulfilling such criteria is so demanding that assays used to measure biomarkers should be validated to exacting standards. Panels of biomarkers must be defined that will in themselves collectively meet these exacting requirements. Such panels may represent a series of single measurements or, more likely, as modern technological platforms emerge, by multiplex measurements. The use of such multiplex measurements will itself generate significant problems for method validation, such as crossreactivity, interference, sensitivity and stability. Such issues will in turn require more complex, although no less stringent, validation strategies. The gulf between technology and the successful deployment of biomarkers in clinical trials is narrowing rapidly, although clinical pharmacology laboratories face significant challenges in implementing biomarker studies in early clinical trials (##REF##15585173##Cummings et al, 2004##).</p>", "<title>What we can currently measure</title>", "<p>Serial tumour sampling is rarely obtained in early-phase clinical trials, although fixed tissue from diagnosis is often available. As a consequence, the broad utility of tissue biomarkers is limited to prognosis in the first instance and resistance to therapy in the second. Resistance to therapy is often the result of multiple mechanisms such as overexpression of anti-apoptotic proteins, such as Bcl-2 family members (Bcl-2, Bcl-xL, Bcl-w, Mcl-1), or the inhibitors of apoptosis proteins (IAPs), such as survivin or XIAP. In addition, downregulation or mutation of pro-apoptotic proteins, such as Bax, caspase 8, death receptors, p53/p73/p21<italic>waf</italic>I, as well as alterations in NF-<italic>κ</italic>B expression/activity can all contribute to chemoresistance (##REF##12001989##Igney and Krammer, 2002##). Most, if not all, of these proteins have been detected in tissues using immunohistochemistry. However, there are now also many ELISA and cytometric bead-based assays available. The above listed markers remain potential predictors of therapy response rather than truly PD biomarkers, which change in response to therapy. Many of the established caspase substrates are potentially good candidate POC biomarkers of apoptosis that can be measured by immunohistochemistry and ELISA-based assays (e.g., cleaved poly(ADP-ribose) polymerase and cleaved caspase 3). However, if the products of the apoptotic cascade are released into the circulation of cancer patients following therapeutic intervention, they can be more readily measured serially and therefore dynamically. Cytokeratins form approximately 5% of intracellular proteins; therefore, by measuring these, even small numbers of apoptotic cells should be detectable (##REF##12766486##Biven et al, 2003##). Specific ELISAs have been developed to quantify CK18 and/or CK19 (e.g., tissue polypeptide antigen, tissue polypeptide-specific antigen and CYFRA21-1) (##REF##17108218##Holdenrieder et al, 2006##). These assays are not specific to apoptosis, as necrosis also leads to the release of intact soluble CKs. The M30 apoptosense ELISA uses an antibody to a caspase-cleaved neo-epitope on CK18, whereas the M65 ELISA detects both intact and cleaved soluble CK18. The combined use of M30 and M65 offers potential to dissect mechanisms of cell death in cancer patients (##REF##14996736##Kramer et al, 2004##).</p>", "<p>Apoptotic endonucleases preferentially cleave DNA between nucleosomes, and the resultant oligonucleosomes are detectable in serum where histones partially protect DNA from further nuclease degradation. In healthy subjects, nDNA has a short half-life; however, their levels are elevated in cancer patients suggesting high levels of production, altered catabolism or both (##REF##17108218##Holdenrieder et al, 2006##), and ELISA assays can detect nucleosomes in cancer patient sera. Since CKs do not provide any information on cell death from non-epithelial cells, their combined use with nDNA provides a biomarker panel to assess caspase-dependent and -independent cell deaths of all nucleated cells.</p>", "<p>Such assays are being integrated into trials of pro-apoptotic therapies as POC biomarkers (##REF##16804528##Cummings et al, 2006##). Although these biomarkers are elevated in cancer patients, they are not sufficiently specific for diagnosis. However, high levels appear to be associated with poor prognosis in certain tumour types, probably reflecting tumour burden. Indeed tissue polypeptide antigen and tissue polypeptide-specific antigen have been used as tumour markers (##REF##17108218##Holdenrieder et al, 2006##; ##REF##17316892##Ulukaya et al, 2007##), and increases in their levels after chemotherapy may be associated with therapeutic response, although this has not been consistently reproduced (##REF##14996736##Kramer et al, 2004##; ##REF##17545523##Olofsson et al, 2007##; ##REF##17316892##Ulukaya et al, 2007##). In one of the largest studies, nDNA and CYFRA21-1 were measured in 311 patients with NSCLC receiving chemotherapy (##REF##17108218##Holdenrieder et al, 2006##). Changes in nDNA and CYFRA21-1 predicted response independently from stage and performance status. These assays in combination demonstrated 100% specificity for response with a sensitivity of 29%, suggesting that they add clinically meaningful information to patient management. Clearly, panels of multiple validated biomarkers specifically tailored to particular treatment regimes or disease groups are the way forward. Currently, we are using existing tumour markers to follow dynamic fold changes in biomarker levels in response to therapy, a hitherto little explored approach. However, new biomarkers as predictors of response as well as patient survival are urgently needed. Such novel biomarkers will need to be tested in large trials with full clinical data available and follow-up, and the biomarker data subjected to rigorous statistical evaluation before implementation into routine clinical practice. The goal of stratifying patients based on biomarker expression is yet to be fulfilled and is an exciting goal for moving ahead.</p>", "<title>Biomarkers: the clinical challenges</title>", "<p>Cancer is complex and it is increasingly recognised that tumour cells are rarely addicted to a single pathway, and therefore targeting a single pathway is unlikely to be effective in producing durable remissions due to plasticity, redundancy and feedback mechanisms within molecular signalling pathways. Likewise, considering ‘biomarkers’ as a generic term is an oversimplification. Biomarkers that identify at-risk individuals, detect disease earlier, determine prognosis, detect recurrence/metastases and predict or monitor response/toxicity to treatment are needed (##REF##17433551##Voorzanger-Rousselot and Garnero, 2007##). It is essential that targeted therapies and their associated biomarkers co-evolve.</p>", "<p>One key barrier is the lack of high-quality reference material to define biomarker normality, as before you can detect an abnormality, it is essential to know the normal range of a biomarker. Moreover, to get an understanding of the biomarker dynamic range in tumours (given tumour heterogeneity) requires comprehensive international databases of healthy individual and cancer patient samples, collected by standardised methods at multiple time points, analysed retrospectively and prospectively, using validated protocols with quality controls combined with long-term clinical data. Currently, due to interinstitutional variability, this is barely achievable within a nation. Only with this international co-operation, may a consensus of ‘normal’ be obtained and biomarkers discovered, validated to identify false-positives and -negatives, and qualified rapidly.</p>", "<p>In current practice, it is extremely rare that biomarker changes accurately represent all of the effects of a therapy on the clinical outcome and, thus, it is essential that biomarker qualification does not distract from robust clinical end points. Finally, as stated above, although many biomarkers correlate statistically with disease end points, this does not automatically prove clinical usefulness (##REF##12734304##Kattan, 2003##). Before integration into the busy clinic, ‘novel’ biomarkers must demonstrate added value beyond that which is already available.</p>", "<title>New technology platforms: the promise of the future</title>", "<p>Protein microarray technology is a rapidly evolving field, driven by the need for high-throughput methodology to measure multiple biomarkers in clinical samples. Nearly all the fully tested and characterised protein microarrays are based on antibody technologies. These multiplex platforms have become widely used in the exploratory research arena. To date, three main formats exist: substrate-anchored sandwich ELISA, liquid-based bead assays and protein arrays. Although much effort has been invested in optimisation and instrumentation, these techniques are still only at an early stage of development for use in clinical trials.</p>", "<p>ELISAs use an immobilised antibody to capture a soluble ligand, with subsequent detection of captured ligand by a second antibody linked to a reporter molecule. Multiplex plate-based sandwich ELISAs, such as the MSD Mesoscale® (Gaithersburgh, MD, USA) and SearchLight® (Fishers Scientific, Pittsburgh, PA, USA) chemiluminescent arrays are capable of quantification of up to 16 different proteins in multi-well plates. This approach has the advantage that a greater number of analytes can be measured in the same or smaller volume of blood than a conventional single-plex ELISA. This economic use of samples represents a significant saving in both cost and blood/tumour lysate volumes (##REF##15261575##Nielsen and Geierstanger, 2004##). In bead-based assays, such as Luminex and Bead array, capture antibodies are conjugated to polystyrene beads that are uniquely tagged with a combination of two fluorescent dyes. Such dye combinations represent a unique signature for each bead. A second detection antibody tagged with a common fluorochrome is used to quantify the amount of analyte bound to each bead. Detection and quantification are achieved using conventional flow cytometry or dedicated bead-based bioanalysers. Multiplex assays can be created by mixing bead sets with different conjugated capture antibodies to simultaneously test for many (up to 50 or more) analytes in a single clinical sample (##REF##16481199##Elshal and McCoy, 2006##). There are several substantial differences between multiplex platforms in current use and little published work exists regarding validating the relative performances of each platform. Most of the comparisons that have been published to date compare the ‘gold standard’ single-plex ELISA assay to that of a new multiplex system. Comparisons between multiplex and single-plex platforms tend to show good correlations (0.6–0.96). Moreover, both intra- and interassay variations are generally less than 16% (##REF##16481199##Elshal and McCoy, 2006##). The main stumbling block with such comparisons is the source of the antibodies used to capture the analyte, the nature of the capture surface, crossreactivity of the antibodies and heterophilic reactions within the physiological matrix. If comparisons are made between platforms that use identical antibody pairs and detection reagents, correlations tend to be tight. However, even in this scenario, the nature of the substrate is important. Bead-based assays tend to have a lower available surface area to react with analytes than do microspot-based assays. As they flow through the analysis system, beads often have only a half or less of their surface area presented to the signal detector at any time. The uniform, high-density signal from microspots leads to lower levels of detection but higher levels of sensitivity, when compared with bead-based platforms. Despite these issues, multiplex assays are still capable of detection in the analyte nanogram range and a number of multiplex assay platforms have been approved by the FDA (##REF##17187487##Ling et al, 2007##).</p>", "<p>Protein arrays consist of a large number of regularly arranged discrete microspots of capture molecules, which are transferred onto a solid support using spotting robots. Spotted capture molecules may also be conjugated to fluorescent beads, thereby enabling existing cytometric bead-based technologies to capture nucleic acids, proteins and soluble receptors/ligands. Purified recombinant proteins, antibodies, antibody fragments, aptamers, peptides, nucleic acids or complex protein extracts have all been used as capture molecules. These arrays measure relative protein abundance and are analogous to the DNA arrays commonly used in expression profiling. Comparison of different biological samples in this way is increasingly important in the discovery of potential biomarkers and new targets for therapies.</p>", "<p>The development of stable, intensely fluorescent reporter molecules, such as quantum dots, will enhance the multiplexing capacities of protein microarrays (##REF##16097884##Kersten et al, 2005##).</p>", "<p>Currently, analysis of a single analyte with a single assay is the predominant method applied to most clinical trials. However, once validation strategies have proved multiplex assays to be as robust, reliable and reproducible as single-plex assays, they are destined to comprise a significant part of clinical trial activities. Potentially, these technologies could rapidly trawl through a subset of trial samples to identify the most informative biomarkers to be implemented in the context of that trial. However, one must always bear in mind that regardless of the method chosen to measure biomarkers (singly or multiply), they still exist in a complex biological matrix where they have the potential to interact. More importantly, the rate at which individual biomarkers degrade within such a matrix may vary significantly between biomarkers. Such possibilities mean that analysis of clinical samples should take place as soon as possible following collection. When the path for regulatory compliance for these assays is defined, the development of multiplex systems should accelerate and these approaches should be widely taken up for clinical trials.</p>", "<title>Serial collection of circulating tumour cells: a cell-based apoptotic biomarker?</title>", "<p>Circulating tumour cells (CTCs) have been detected in the peripheral blood of patients with solid carcinomas (##REF##15317891##Cristofanilli et al, 2004##). Although the development of apoptotic biomarkers has been predominantly focused on the molecular level, decreased CTC numbers may represent an apoptosis-associated biomarker. With the advent of automated, standardised technologies, further morphological and molecular characterisation of CTCs can be carried out in great detail. Accumulating evidence shows that CTCs represent a heterogeneous cell population, among which there exist both apoptotic cells and viable cells with metastatic potential. At the cellular level, the change of CTC number pre- and post-treatment correlates well with patient response to treatment in several cancer types (##REF##15317891##Cristofanilli et al, 2004##). Persistence of CTCs 3–4 weeks following therapy strongly suggested that the patients are relapsing with drug-resistant disease and additional chemotherapy would be futile (##REF##15317891##Cristofanilli et al, 2004##). At the level of morphology, incorporating Wright–Giemsa staining into the protocol allows fibre-optic array scanning technology to be applied. This approach has shown that CTCs detected from widely metastatic breast cancer patients exhibited early and late apoptotic changes (##REF##17188328##Marrinucci et al, 2007##). Further studies have shown that circulating breast cancer cells are frequently apoptotic based on CK staining pattern, nuclear condensation and DNA strand-breaks (##REF##11438448##Mehes et al, 2001##), and caspase-cleaved CK18 was detected in circulating prostate tumour cells (##REF##15472900##Larson et al, 2004##). In contrast to CT scans, which are expensive, and biopsies, which are difficult to serially collect, the assessment of CTCs provides a readily accessed and cheaper biomarker. Such biomarkers inform on early dynamic changes in the tumour population, which in turn help to evaluate therapeutic response and provide POM for novel pro-apoptotic drugs. Isolation of viable, intact CTCs in sufficient purity and quality may allow informative genomic profiling. Although CTCs may provide a practical useful source of biomarkers, it is not yet known precisely how they relate to the primary tumour, or which CTC markers might predict which cells will metastasise. Technologies for isolating CTCs are advancing rapidly and CTCs have a great potential for biomarker research.</p>", "<title>Summary</title>", "<p>In the era of molecular targeted agents, cell death pathways have become key players in drug discovery portfolios. Proteomics can be seen to be playing an ever-increasing role in both the discovery and measurement of biomarkers pertinent to cell death pathways. Apoptosis, therefore, represents not only a vital target in oncology but also a unique biomarker opportunity hitherto unexploited. The challenge ahead lies in discovering new biomarkers and understanding the biology of cellular release of existing apoptotic biomarkers into the circulation. More complex still is the relationship between clinical efficacy, biomarker measurement and overall survival of patients treated with novel molecular targeted agents. As with genomics, key technology tools are now available to fully exploit these opportunities and take apoptosis biomarkers to the forefront of drug discovery and future clinical trials.</p>" ]

[]

[ "<fig id=\"fig1\"><label>Figure 1</label><caption><p>Schematic diagram of the consequential accumulation of proteins following induction of apoptosis. These biomarker molecules are eventually released and can be detected in the circulation in patients undergoing therapy.</p></caption></fig>" ]

[ "<table-wrap id=\"tbl1\"><label>Table 1</label><caption><title>Most commonly described biomarkers of apoptosis can be measured in tissues and blood using a variety of technology platforms</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"left\"/><col align=\"left\"/><col align=\"left\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Biomarker</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Matrix</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Analysis platform</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Comment</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Activated caspases 2, 3, 7, 8 and 9</td><td align=\"left\" valign=\"top\" charoff=\"50\">Tissue</td><td align=\"left\" valign=\"top\" charoff=\"50\">IHC, Elisa, flow cytometry, cytometric bead arrays</td><td align=\"left\" valign=\"top\" charoff=\"50\">Detection by immunoreaction or substrate/active site interactions</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Cytochrome <italic>c</italic></td><td align=\"left\" valign=\"top\" charoff=\"50\">Tissue, serum</td><td align=\"left\" valign=\"top\" charoff=\"50\">ELISA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Useful biomarker measured serially in blood samples</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Externalised phosphatidylserine</td><td align=\"left\" valign=\"top\" charoff=\"50\">Cells</td><td align=\"left\" valign=\"top\" charoff=\"50\">ELISA, flow cytometry</td><td align=\"left\" valign=\"top\" charoff=\"50\">Measures Annexin binding to externalised ligand. Early apoptosis event</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Cytokeratins</td><td align=\"left\" valign=\"top\" charoff=\"50\">Tissue, serum plasma</td><td align=\"left\" valign=\"top\" charoff=\"50\">ELISA, IHC, Flow cytometry</td><td align=\"left\" valign=\"top\" charoff=\"50\">Useful biomarker measured serially in blood samples</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Nucleosomal DNA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Tissue, serum</td><td align=\"left\" valign=\"top\" charoff=\"50\">ELISA, DNA array, PCR</td><td align=\"left\" valign=\"top\" charoff=\"50\">Nucleosomal DNA can be measured serially in serum samples. Extracted DNA can be analysed using PCR</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Apo-1/Fas, Fas ligand (sFasL)</td><td align=\"left\" valign=\"top\" charoff=\"50\">Serum, follicular fluid, cells</td><td align=\"left\" valign=\"top\" charoff=\"50\">ELISA, flow cytometry, IHC</td><td align=\"left\" valign=\"top\" charoff=\"50\">Expressed on B and T cells as well as in normal and tumour tissue</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Bcl-2/Bcl-xl/Mcl-1</td><td align=\"left\" valign=\"top\" charoff=\"50\">Cells, tissue</td><td align=\"left\" valign=\"top\" charoff=\"50\">IHC, ELISA, flow cytometry</td><td align=\"left\" valign=\"top\" charoff=\"50\">Overexpression contributes to chemo-resistance.</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">p53, phospo-p53, p21^wafi, pH2AX</td><td align=\"left\" valign=\"top\" charoff=\"50\">Cells, tissues</td><td align=\"left\" valign=\"top\" charoff=\"50\">IHC, flow cytometry, ELISA</td><td align=\"left\" valign=\"top\" charoff=\"50\">Activation and stabilisation of these proteins informs on DNA damage and repair</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><fn id=\"t1-fn1\"><p>IHC=immunohistochemistry; PCR=polymerase chain reaction.</p></fn><fn id=\"t1-fn2\"><p>Currently, ELISA platforms and flow cytometry offer the highest throughput for clinical trial use.</p></fn></table-wrap-foot>" ]

[ "<graphic xlink:href=\"6604519f1\"/>" ]

[]

{ "acronym": [], "definition": [] }

[]

[ "<title>Materials and methods</title>", "<title>Cell lines, reagents and antibodies</title>", "<p>FET cells, vector control clone and Smad7 clones in FET cells that overexpressed Smad7 were described previously (##REF##15922743##Halder et al, 2005##). The TGF-<italic>β</italic> was purchased from R&amp;D Systems (Minneapolis, MN, USA), antibodies for Smad2, Smad3, Claudin-1, Claudin-4 and Claudin-7 were purchased from Zymed Laboratories Inc. (San Francisco, CA, USA). Anti-T<italic>β</italic>RII, anti-Smad4, anti-p21^Cip1, anti-p27^Kip1, anti-p53, anti-Cdk2, anti-Cdk4, anti-Cyclin D1, ERK and anti-Rb antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA) and anti-phospho-Smad2, anti-phospho-ERK, anti-phospho-Rb, anti-phospho-AKT and AKT antibodies were purchased from Cell Signaling (Danvers, MA, USA). Anti-<italic>β</italic>-actin antibody was purchased from Sigma Biochemicals (St Louis, MO, USA).</p>", "<title>Analysis of liver metastasis in mice</title>", "<p>Parental FET cells, vector control clone and three stable Smad7 clones (#2, #6 and #15) were assayed for metastasis in 5- to 6-week-old athymic nude mice using splenic injection model as described previously (##REF##16432186##Buchanan et al, 2006##). Cells (5 × 10⁶) from each cell line was injected with a 27.5-gauge needle into the spleen after pulling out through the incision made in the abdomen in a sterile condition. The spleen was removed by electrocautery and homoeostasis was assured. The area was thoroughly irrigated with warm sterile water and the abdominal cavity was closed in appropriate layers by using a 6–0 prolene suture. Thirty-three days after surgery, mice were sacrificed and the livers were obtained from mice. Liver weight was determined, and a portion of tumour area or a corresponding normal liver area was collected in a cassette and fixed overnight in 4% paraformaldehyde, stored in 70% ethanol and then processed for immunohistochemical (IHC) studies. The other portion of liver was stored at −80°C after immediate freezing in liquid nitrogen and used for western blot and PCR analyses.</p>", "<title>Immunohistochemistry</title>", "<p>Paraffin-embedded blocks were prepared from liver metastases and normal liver of mice injected with parental or vector control cells. A 5 <italic>μ</italic>m thick serial section for each slide was fixed with 20% xylene for 20 min, washed with 100 and 95% ethanol for 10 min and finally washed with water. A standard IHC method was applied to stain the slides for hematoxylin and eosin (H&amp;E), Ki67, Tunnel, phospho-Smad2, E-cadherin, Claudin-1 and Claudin-4.</p>", "<title>Western blot analysis</title>", "<p>Lysates from normal liver and Smad7-mediated liver metastases were prepared according to the method as described previously (##REF##15922743##Halder et al, 2005##). Briefly, liver tissue samples were cut into small pieces, homogenized by sonication for 20 s in lysis buffer containing protease inhibitor cocktail and then centrifuged at 14 000 r.p.m. for 20 min at 4°C. Protein concentrations of clear lysates were measured using the Bio-Rad Protein Assay kit (BioRad, Hercules, CA, USA). Equal amount of each lysate was analysed for western blotting with antibodies against Smad7, T<italic>β</italic>RII, phospho-Smad2, Smad2, Smad3, Smad4, phospho-AKT, AKT, phospho-ERK, ERK, E-cadherin, <italic>β</italic>-catenin, ZO-1, Claudin-1, Claudin-4, Claudin-7, nm23, p53, uPA, p21^Cip1, p27^Kip1, Cdk2, Cdk4, Cyclin D1, phospho-Rb and Rb. <italic>β</italic>-Actin antibody was used as loading control.</p>", "<title>PCR analysis</title>", "<p>Genomic DNA was extracted from cell lines and liver metastases using standard genomic DNA isolation protocol. Primers for hygromycin-specific gene were used for PCR analyses (forward: 5′-ATGGACGACACCGTCAGTG-3′ and reverse: 5′-GTCAACCAAGCTCTGATAGAG-3′). Polymerase chain reaction was performed using 2.5 <italic>μ</italic>l (∼100 ng) of DNA, 0.5 U Taq DNA polymerase, 200 m<sc>M</sc> dNTPs and 10 pmol of each primer. Polymerase chain reaction was carried out by initial denaturation at 94°C for 3 min followed by 30 cycles; each cycle has 94°C for 45 s, 52°C for 45 s and 72°C for 1 min.</p>" ]

[ "<title>Results</title>", "<title>Stable expression of Smad7 induces liver metastasis in nude mice</title>", "<p>The role of Smad pathway in TGF-<italic>β</italic>-mediated metastasis is controversial. We demonstrated previously that Smad7 induces tumorigenicity in colon tumour-derived FET cells by blocking TGF-<italic>β</italic>-induced growth inhibition and apoptosis (##REF##15922743##Halder et al, 2005##). In colorectal cancer, patients with impaired Smad signalling show poor prognosis and increased metastasis, whereas in mouse model of breast cancer (##REF##16172383##Kang et al, 2005##) and melanoma metastasis to bone (##REF##17332363##Javelaud et al, 2007##), Smad pathway plays a prometastatic role. These apparently contradictory results regarding the role of Smad pathway in metastasis prompted us to test whether inhibition of Smad pathway by Smad7 induces invasion and metastasis of colon cancer cells. For this purpose, we have used splenic injection model of liver metastasis with colon adenocarcinoma-derived FET cells that stably express functional Smad7 (##REF##15922743##Halder et al, 2005##). The FET cells are non-tumorigenic and highly responsive to TGF-<italic>β</italic>. We have injected parental FET cells, one vector control clone and three Smad7-expressing clones into the spleen of athymic nude mice, and the liver mets were obtained by dissecting the mice 33 days after injection (##FIG##0##Figure 1A##). We observed that 17 out of 18 control mice did not show any metastasis in the liver (##FIG##0##Figure 1B##). Interestingly, mice injected with all three Smad7 clones developed liver metastases. We compared the size and morphology of livers between mice injected with control cells and stable Smad7 clones. Interestingly, Smad7 clones developed aggressive liver metastases with increased liver size as compared with vector control liver (##FIG##0##Figure 1C##). To further test whether the liver metastasis is due to the splenic injection of Smad7 stable cells that are migrated to the liver, we performed PCR with genomic DNA using primers that amplify hygromycin gene in the backbone vector of Smad7 expression construct. We observed 345 bp PCR product in Smad7-induced liver metastases, but not in the livers of mice injected with parental FET and vector control cells (##FIG##0##Figure 1D##, right pannel). In a parallel experiment, PCR amplification was performed using genomic DNA from FET cells, vector control clone and three stable Smad7 clones. Polymerase chain reaction product was observed in the genomic DNA from control vector and Smad7 stable cells, but not in parental FET cells (##FIG##0##Figure 1D##, top left panel). We further verified the expression of exogenous Smad7 in Smad7-induced metastases by western blot analyses, using lysates from livers of mice injected with parental FET cells and a vector control clone as well as liver metastases from mice injected with stable Smad7 clones (##FIG##0##Figure 1D##, bottom panel). These results suggest that only Smad7 overexpressing FET cells go to the liver, survive and produce metastasis. Taken together, these results suggest that blockade of TGF-<italic>β</italic>/Smad pathway in these cells by overexpressing Smad7 induces liver metastasis.</p>", "<title>Regulation of Smad and non-Smad pathways in Smad7-induced liver metastases</title>", "<p>The TGF-<italic>β</italic> signalling is known to be activated in advanced stage of tumour progression including invasion and metastasis. To verify whether TGF-<italic>β</italic> signal mediators are affected in Smad7-induced liver metastasis, we performed western blot analyses using lysates from livers of mice injected with vector control clones as well as liver metastases from mice injected with stable Smad7 clones. We observed higher levels of T<italic>β</italic>RII in liver metastases obtained from mice injected with Smad7-expressing clones compared with that from control mice (##FIG##1##Figure 2A##). We further verified whether increased levels of T<italic>β</italic>RII in liver metastases affect the downstream Smad signalling. Interestingly, we observed increased phosphorylation of Smad2 in Smad7-induced liver metastases when compared with that in the livers from vector control clones-injected mice (##FIG##1##Figure 2A##). However, the levels of total Smad2, phospho-Smad3 (data not shown), total Smad3 and Smad4 remained unchanged irrespective of metastasis (##FIG##1##Figure 2A##). In an attempt to test whether any of the non-Smad pathways are activated in the liver metastases, we observed increased phosphorylation of AKT and ERK in liver metastases from mice inoculated with Smad7-expressing cells when compared with the mice inoculated with vector control clones (##FIG##1##Figure 2B##). However, the phosphorylation of p38 MAPK and c-Jun was unchanged (data not shown). These results suggest that Smad7 induces liver metastasis through the activation of TGF-<italic>β</italic> signalling.</p>", "<title>Expression profile of proteins in liver metastases including junctional and cell cycle regulatory proteins</title>", "<p>During tumour development and metastasis the expression of adherens junction and tight junction proteins are aberrantly expressed or delocalised, and cells become motile and metastasise into the target organs. Although contradictory, the level of expression of E-cadherin and catenins has been shown to be enhanced in metastatic lesions (##REF##15619205##Imai et al, 2004##; ##REF##17032390##Hung et al, 2006##). Some of the tight junction proteins including Claudin-1 (##REF##15965503##Dhawan et al, 2005##; ##REF##16990707##Soini and Talvensaari-Mattila, 2006##) and Claudin-4 (##REF##16253248##de Oliveira et al, 2005##; ##REF##16266975##Morin, 2005##; ##REF##16990707##Soini and Talvensaari-Mattila, 2006##) are upregulated in several cancers. To examine the expression of these junctional proteins in Smad7-mediated liver metastases, we observed that while E-cadherin and <italic>β</italic>-catenin were marginally upregulated, the levels of expression of ZO-1, Claudin-1, Claudin-4, Claudin-7, nm23, p53 and uPA were strongly elevated in Smad7-mediated liver metastases (##FIG##1##Figure 2C##). The downregulation of ZO-1 leads to increased motility. ZO-1 is upregulated in melanoma cells and that upregulation of ZO1 contributes to the oncogenic behaviour of this tumour (##REF##15855653##Smalley et al, 2005##). Another membranous protein, nm23, is increased in highly metastatic malignancies (##REF##15599350##do Nascimento Souza et al, 2005##). The primary function of p53 is to control checkpoints in cell cycle and to stimulate apoptosis. Mutations in this gene produce inactive p53 proteins that accumulate in the tumour cells, and the level of mutated protein is increased. The FET cells have a p53 mutation (C176F) that has been shown to be elevated during the progression of human cancers (##REF##11526487##Gayet et al, 2001##). We further tested the expression of these proteins in increasing levels of aggressiveness (determined by the number of tumour nodules and volume of the liver occupied by tumour) of liver metastases. The expression of p53, E-cadherin, nm23, Claudin-4 and Claudin-7 were correlated with the aggressiveness of liver metastasis (##FIG##2##Figure 3A and B##), which includes number of tumour nodules, volume of the liver occupied by tumours and liver weight. ##FIG##2##Figure 3C## shows the percentage of total number of mice, with or without liver metastases, inoculated with vector control cells or Smad7-expressing cells. Only one mouse (out of 18, 5.5%) from vector control cells produce liver metastasis that covered around 20% of the total liver. In contrast, 76% of mice from Smad7-expressing clones produced liver metastases of different aggressiveness and 39% of these mice produced metastases that covered &gt;60% of the liver.</p>", "<p>We next examined the expression of cell cycle regulatory proteins that are differentially modulated in Smad7-mediated liver metastases. We observed a marginal decrease in the expression of p21^Cip1 and a marginal increase in the expression of p27^Kip1 in liver metastases from Smad7 clones, when compared with that from vector control clones (##FIG##1##Figure 2D##). In addition, the levels of cyclin-dependent kinases, Cdk2 and Cdk4 and Cyclin D1 were found to be increased in liver metastases induced by Smad7, and this could result in the phosphorylation of pRb (##FIG##1##Figure 2D##). Regulation of these proteins may contribute to cell proliferation in the liver metastases.</p>", "<title>Smad7 changes the expression pattern and localisation of junctional proteins in liver metastases</title>", "<p>Aberrant expression of junctional proteins such as E-cadherin and Claudins in metastatic cancers has been reported. To test the expression and localisation of junctional proteins in liver metastases, IHC analyses were performed. The H&amp;E staining revealed poorly differentiated spindle cells in Smad7-mediated liver metastases as shown in ##FIG##3##Figure 4A##. Diffuse positive Ki67 (a cell proliferation marker) staining was detected in Smad7-mediated liver metastases, as compared to livers from parental FET and vector control cells (##FIG##3##Figure 4A##), suggesting the involvement of increased cell proliferation in metastatic tumour growth. However, we did not observe any significant change in apoptotic tunnel positive cells between hepatocytes from parental or vector control mice and Smad7-mediated liver metastases (##FIG##3##Figure 4A##). We observed that phosphorylated Smad2 was predominantly localised in the nucleus in Smad7-induced liver metastases, suggesting the activation of TGF-<italic>β</italic>/Smad signalling, as a result of metastasis (##FIG##3##Figure 4B##). Interestingly, we observed higher membranous staining of E-cadherin in liver metastases induced by Smad7 (##FIG##3##Figure 4B##), whereas the localisation of <italic>β</italic>-catenin remained unchanged (data not shown). Similarly, we observed higher immunoreactivity of Claudin-1 and Claudin-4 in Smad7-induced liver metastases when compared with livers of mice injected with parental or vector control cells (##FIG##3##Figure 4B##), whereas Claudin-7 was not changed (data not shown). These results suggest that restoration of expression of junctional proteins may play a role in establishing the metastatic growth at distant sites.</p>" ]

[ "<title>Discussion</title>", "<p>There is substantial evidence to demonstrate that Smad signalling is important for TGF-<italic>β</italic> tumour suppression function, and neoplastic transformation is often associated with the loss of this tumour suppressor function. During cancer progression, high levels of TGF-<italic>β</italic> can promote tumour growth in an autocrine and/or paracrine manner through the changes that favor invasion and metastasis. Smad4 mutation and its reduced level in colorectal cancer are directly correlated to poor prognosis and increased metastasis (##REF##10340381##Miyaki et al, 1999##), whereas Smad7 expression is associated with poor outcome in gastric carcinomas (##REF##15033661##Kim et al, 2004##). In contrast, the tumour suppressor Smad pathway has been shown to mediate the prometastatic function of TGF-<italic>β</italic> in the development of breast cancer bone metastasis (##REF##16172383##Kang et al, 2005##), and dominant negative Smad3 inhibits lung metastasis of breast cancer cells in animal models (##REF##14678987##Tian et al, 2003##). In addition, blockade of Smad pathway by overexpression of the inhibitory Smad, Smad7 impairs bone and lung metastases (##REF##16333029##Azuma et al, 2005##; ##REF##17332363##Javelaud et al, 2007##). Therefore, the role of Smad pathway in TGF-<italic>β</italic>-mediated metastasis remains poorly understood and controversial. We have previously reported that stable expression of Smad7 in human colon adenocarcinoma FET cells induces tumorigenicity (##REF##15922743##Halder et al, 2005##), whereas Smad7 inhibits tumorigenicity in melanoma cells (##REF##16007121##Javelaud et al, 2005##). In this report, we show that ectopic expression of Smad7 enhances colorectal cancer metastasis to liver in a splenic injection model. In addition, IHC analyses have suggested poorly differentiated spindle cell morphology and higher cell proliferation in Smad7-induced liver metastases.</p>", "<p>Although Smad signalling has been shown to play an active role in enhancing breast cancer and melanoma metastases in animal models, functional inactivation of Smad signalling in human colorectal cancer is associated with poor prognosis and distant metastasis. This is in agreement with our present study that blockade of Smad signalling by inhibitory Smad7 increases tumorigenicity and liver metastasis. Human colon adenocarcinoma-derived FET cells are growth inhibited by TGF-<italic>β</italic> and are non-tumorigenic. Increase in the metastatic potential of this cell line by Smad7 may involve (1) blockade of TGF-<italic>β</italic>-mediated tumour suppressor function by Smad7, (2) cooperation between mutated K-Ras (in FET cells) and activated Smad2 generated as a result of metastasis to the liver and (3) cooperation between activated K-Ras and higher levels of Smad7. Previously, we have shown that overexpression of Smad7 blocks TGF-<italic>β</italic>-induced growth inhibition and apoptosis of these cells (##REF##15922743##Halder et al, 2005##). Elevated levels of T<italic>β</italic>RII expression in the liver metastases from Smad7-expressing clones (##FIG##1##Figure 2A##) coupled with higher levels of TGF-<italic>β</italic> secretion (data not shown) may result in phosphorylation and nuclear accumulation of Smad2. In mouse keratinocytes during multistage tumorigenesis, elevated levels of activated Ras are important for the activation and nuclear accumulation of Smad2, which are essential for inducing EMT and metastasis (##REF##12105419##Oft et al, 2002##). The FET cells have an activating mutation in K-Ras, and activation of MAPK/ERK and AKT pathways (##FIG##1##Figure 2B##) in the liver metastases coupled with the suppression of growth inhibitory effects of TGF-<italic>β</italic> by Smad7 may exert a selective pressure for tumour outgrowth. Although, our earlier studies showed that the stable expression of Smad7 in FET cells inhibits TGF-<italic>β</italic>-induced phoshphorylation of Smad2 (##REF##15922743##Halder et al, 2005##), liver metastases resulting from these clones show higher levels of phosphorylation and nuclear accumulation of Smad2 (##FIG##1##Figure 2A and B##). This could be due to the increased expression of T<italic>β</italic>RII (##FIG##1##Figure 2A##) and higher levels of TGF-<italic>β</italic> in the liver metastases. At that point, ectopic expression of Smad7 in the FET cells has no influence on Smad2 phosphorylation. This is consistent with the previous study that suggested active Smad signalling with nuclear accumulation of phospho-Smad2 in breast cancer bone metastatic lesions (##REF##16172383##Kang et al, 2005##). It is possible that activated K-Ras in FET cells induces phosphorylation and nuclear accumulation of Smad2. Subsequent upregulation of Smad2-mediated gene expression can induce epithelial-to-mesenchymal transition. Therefore, higher level of activated Smad2 may cooperate with activated Ras to induce spindle cell morphology and invasiveness of tumour cells. Smad7 cooperates with oncogenic Ras to cause malignant conversion in a mouse model for squamous cell carcinoma, which is associated with the blockade of normal differentiation (##REF##14633701##Liu et al, 2003##). In agreement with this, it is also possible that elevated level of Smad7 cooperates with oncogenic Ras in FET cells to induce poorly differentiated spindle cell morphology (##FIG##3##Figure 4A##, top panel).</p>", "<p>There is compelling evidence indicating that downregulation or delocalisation of E-cadherin is critical for malignant progression of epithelial tumours (##REF##10505543##Behrens, 1999##). In contrast, E-cadherin expression has also been shown to be increased in the metastatic lesions when compared with corresponding primary site (##REF##10505543##Behrens, 1999##; ##REF##14580257##Kowalski et al, 2003##). Cancer cells re-express E-cadherin once they reach to distant metastasis sites and the re-expression of E-cadherin enables cancer cells to form a metastatic deposit by facilitating intracellular adhesion and colonisation (##REF##14580257##Kowalski et al, 2003##). E-cadherin has also been shown to be upregulated in ovarian cancer (##REF##9191009##Peralta Soler et al, 1997##), renal cell carcinoma (##REF##8550243##Tani et al, 1995##) and breast cancer (##REF##14580257##Kowalski et al, 2003##). In consistent with these results, we have observed upregulation and higher immunoreactivity of membranous E-cadherin in Smad7-induced liver metastases (##FIG##1##Figures 2C## and ##FIG##3##4B##). In addition, the expression of E-cadherin is enhanced with the increase in aggressiveness of liver metastases (##FIG##2##Figure 3A and B##). Increase in E-cadherin immunoreactivity is highly associated with a worse prognosis of oral squamous cell carcinoma (OSCC), and the re-expression of E-cadherin in lymph node confers advantages to OSCC in metastasis environment (##REF##15619205##Imai et al, 2004##; ##REF##17032390##Hung et al, 2006##). Thus, re-expression of E-cadherin in Smad7-induced liver metastasis may play a role in the establishment and successful growth of the metastatic cells in the liver by facilitating cell–cell adhesion.</p>", "<p>In most cancers, ZO-1 is typically downregulated that leads to increased cell motility. In contrast, ZO-1 has been shown to be upregulated in melanoma cells and is located at adherence junction, and its upregulation contributes to the oncogenic behaviour of this tumour (##REF##15855653##Smalley et al, 2005##). Therefore, E-cadherin and ZO-1 can function in either pro- or antioncogenic manner when expressed in different cellular contexts. It is possible that once in the circulation, cancer cells establish the expression and localisation of E-cadherin and ZO-1, facilitating intercellular adhesion and enabling the formation of cohesive tumour mass at distant sites. Claudins are large family of integral membrane proteins crucial for tight junction formation and cell polarity. Disruption of tight junction barrier function and changes in permeability properties has been shown to be associated with a number of pathologic conditions including cancers. Previous studies have shown that Claudin-1, Claudin-3, Claudin-4 and Claudin-7 proteins are highly expressed in ovarian carcinoma. These higher expressions of Claudin-1 and Claudin-7 correlated with shorter survival (##REF##18439941##Kleinberg et al, 2008##). We have observed the upregulation of Claudin-1, Claudin-4 and Claudin-7 in Smad7-induced liver metastases (##FIG##1##Figures 2C## and ##FIG##3##4B##), and the increased expression of Claudin-4 and Claudin-7 is directly correlated with the aggressiveness of metastasis as shown in ##FIG##2##Figure 3A and B##. The expression of Claudin-4 is upregulated in ovarian cancer (##REF##16714763##Honda et al, 2006##), gastric adenocarcinoma (##REF##16540726##Montgomery et al, 2006##) and colorectal cancer (##REF##16253248##de Oliveira et al, 2005##). In addition, it has been shown previously that overexpression of Claudin-4 promotes tumorigenecity and metastasis in ovarian cancer through the increased invasion and survival of tumour cells (##REF##16103090##Agarwal et al, 2005##). Claudin-1 has been shown to be upregulated in primary and metastatic colorectal cancer that may be important in enhancing the tumorigenicity (##REF##15965503##Dhawan et al, 2005##). These findings correlate with our present observations. p53 is mutated in FET cells (##REF##11526487##Gayet et al, 2001##) and the upregulation of mutated p53 may contribute to the metastatic phenotype of these cells. In addition, nm23 is upregulated in highly metastatic malignancies (##REF##15599350##do Nascimento Souza et al, 2005##) and it is known to be directly or inversely correlated with metastasis depending on the cancer type. Upregulation of nm23 in Smad7-induced liver metastasis (##FIG##1##Figures 2C## and ##FIG##2##3A, B##) may be involved in the metastatic growth of Smad7-expressing FET cells. Interestingly, none of these proteins including E-cadherin, ZO-1, Claudins and nm23 are upregulated in Smad7-expressing FET cells, whereas the expression of these proteins is strongly enhanced <italic>in vivo</italic> in the liver metastases.</p>", "<p>Recent studies have demonstrated that Smad7 expression is induced by TGF-<italic>β</italic>, EGF and the inflammatory cytokines, such as TNF-<italic>α</italic> and IFN-<italic>γ</italic> (##REF##9712726##Afrakhte et al, 1998##; ##REF##15885643##Takahara et al, 2005##). Upregulation of Smad7 may induce cell proliferation through the inhibition of p21^Cip1 (##REF##7626805##Harper et al, 1995##; ##REF##9924195##Robson et al, 1999##) and suppress TGF-<italic>β</italic>-mediated growth inhibition through the induction of c-Myc (##REF##11689553##Yagi et al, 2002##). The higher levels of Smad7 can induce cell survival (##REF##15684397##Edlund et al, 2005##) and inhibit apoptosis (##REF##15150118##Arnold et al, 2004##). Smad7 has been shown to induce tumorigenicity in cooperation with activated Ras (##REF##14633701##Liu et al, 2003##). Thus, increased cell proliferation, survival and tumorigenicity, as well as the inhibition of apoptosis, may contribute to Smad7-induced metastatic growth in the lever (##FIG##4##Figure 5##). In conclusion, we have established a mouse model for splenic injection of colon adenocarcinoma FET cells that develop liver metastasis when Smad7 is overexpressed. Smad7 not only blocks TGF-<italic>β</italic>-mediated antitumour function, but also promotes tumour progression and metastasis of colorectal cancer probably through the cooperation with oncogenic Ras. Although Smad pathway has been shown to mediate the prometastatic function of TGF-<italic>β</italic> in the development of metastases of breast cancer (##REF##16172383##Kang et al, 2005##) and melanoma (##REF##17332363##Javelaud et al, 2007##) in mouse model, our results provide the first evidence that blockade of Smad pathway by Smad7 in colon cancer cells increases liver metastasis. Thus, Smad7 could be a potential target for therapeutic intervention of colorectal cancers.</p>" ]

[]

[ "<p>These authors contributed equally to this study.</p>", "<p>Although Smad signalling is known to play a tumour suppressor role, it has been shown to play a prometastatic function also in breast cancer and melanoma metastasis to bone. In contrast, mutation or reduced level of Smad4 in colorectal cancer is directly correlated to poor survival and increased metastasis. However, the functional role of Smad signalling in metastasis of colorectal cancer has not been elucidated. We previously reported that overexpression of Smad7 in colon adenocarcinoma (FET) cells induces tumorigenicity by blocking TGF-<italic>β</italic>-induced growth inhibition and apoptosis. Here, we have observed that abrogation of Smad signalling by Smad7 induces liver metastasis in a splenic injection model. Polymerase chain reaction with genomic DNA from liver metastases indicates that cells expressing Smad7 migrated to the liver. Increased expression of TGF-<italic>β</italic> type II receptor in liver metastases is associated with phosphorylation and nuclear accumulation of Smad2. Immunohistochemical analyses have suggested poorly differentiated spindle cell morphology and higher cell proliferation in Smad7-induced liver metastases. Interestingly, we have observed increased expression and junctional staining of Claudin-1, Claudin-4 and E-cadherin in liver metastases. Therefore, this report demonstrates, for the first time, that blockade of TGF-<italic>β</italic>/Smad pathway in colon cancer cells induces metastasis, thus supporting an important role of Smad signalling in inhibiting colon cancer metastasis.</p>" ]

[ "<p>Metastasis is the major cause of cancer morbidity and mortality, and accounts for 90% of cancer deaths. Despite the fact that metastasis ultimately kills the host, the mechanisms leading to tumour invasion and metastasis have been less characterized than those resulting in tumour initiation. Cancer development and metastasis is a multistep process that involves local tumour growth and invasion followed by dissemination to, and re-establishment at, the distant sites (##REF##1703045##Liotta et al, 1991##; ##REF##12105419##Oft et al, 2002##). There is compelling evidence indicating that TGF-<italic>β</italic> has complex roles in tumour suppression and progression that are context- and stage-dependent. Therefore, elucidating the molecular pathways essential for tumour metastasis is a higher priority in the pathobiology of cancer to design small molecule drugs.</p>", "<p>Family members of TGF-<italic>β</italic> initiate signalling from the cell surface by binding to a heteromeric complex of two distinct but related serine/threonine kinase receptors. Binding of the ligand to the type II receptor (T<italic>β</italic>RII) results in recruitment and phosphorylation of the type I receptor (T<italic>β</italic>RI). After being activated, T<italic>β</italic>RI propagates the signal to a family of intracellular signal mediators known as Smads. Smad proteins are classified according to their structure and function in signalling by TGF-<italic>β</italic> family members. Receptor-regulated Smads, Smad2 and Smad3, are phosphorylated and activated by T<italic>β</italic>RI. Then they form complexes with common Smad (Smad4) and translocate to the nucleus for regulating the expression of target genes. Given the involvement of TGF-<italic>β</italic> in regulation of cellular homoeostasis, it is expected that there are also a number of feedback mechanisms regulating this process. The activity of the inhibitory Smad, Smad7, is regulated at many levels, suggesting that they serve as key regulators for fine-tuning the responses to TGF-<italic>β</italic> signalling. Smad7 normally resides in the nucleus and moves to the cytoplasm in response to TGF-<italic>β</italic>. TGF-<italic>β</italic> not only enhances the expression of Smad7 transcription but also mobilises a pre-existing nuclear pool of Smad7 to inhibit TGF-<italic>β</italic> receptors (##REF##9786930##Itoh et al, 1998##). Smad7 also interfere with TGF-<italic>β</italic>/Smad signalling through the recruitment of E3 ubiquitin ligases leading to the degradation of receptors and through the interaction with GADD34 that inactivate T<italic>β</italic>RI (##REF##14718519##Shi et al, 2004##). Smad7 expression can also be induced by other signalling inputs, which interfere with TGF-<italic>β</italic> signalling.</p>", "<p>Genes involved in oncogenic pathways are generally classified as either tumour suppressors or promoters, depending on their function in regulating cell growth, differentiation and death. TGF-<italic>β</italic> and its downstream signal transducers are well documented for such paradoxical characters. Genetic manipulation of the TGF-<italic>β</italic> pathway in tumour cells and experimental animal models validated the metastasis-promoting function of TGF-<italic>β</italic> in the late stage of cancer progression (##REF##15935406##Letterio, 2005##). Carcinogen-induced tumours that develop in TGF-<italic>β</italic> haploid mice often maintain the wild-type TGF-<italic>β</italic> allele and these tumours in fact produce higher level of TGF-<italic>β</italic> than tumours from the wild-type mice (##REF##9662371##Tang et al, 1998##). However, after development, the transgenic tumours rapidly acquire a spindle cell phenotype, overexpress TGF-<italic>β</italic>, and metastasise (##REF##8752208##Cui et al, 1996##). These observations have led to the speculations that during cancer progression, TGF-<italic>β</italic> may reverse its role from an inhibitor of tumour growth to a tumour promoter (##REF##9285120##Reiss and Barcellos-Hoff, 1997##; ##REF##11057902##Massague et al, 2000##; ##REF##11586292##Derynck et al, 2001##). Although complete or partial loss of TGF-<italic>β</italic> signals is permissive for early stages of tumour development, active TGF-<italic>β</italic> signalling with selective loss of growth inhibitory response of TGF-<italic>β</italic> may be advantageous for the progression and metastasis of cancer.</p>", "<p>The tumour suppressor Smad pathway has been shown to mediate the prometastatic function of TGF-<italic>β</italic> in the development of breast cancer bone metastasis (##REF##16172383##Kang et al, 2005##). In addition, blockade of Smad pathway by overexpression of the inhibitory Smad, Smad7 impairs bone and lung metastases (##REF##16333029##Azuma et al, 2005##; ##REF##17332363##Javelaud et al, 2007##). In contrast, Smad4 mutation and its reduced level in colorectal cancer are directly correlated to poor prognosis and increased metastasis (##REF##10340381##Miyaki et al, 1999##). Furthermore, upregulation of Smad7 in colorectal cancer has been correlated with poor survival (##REF##10389752##Korchynskyi et al, 1999##). However, nothing is known about the functional role of Smad signalling in colorectal cancer metastasis to the liver. The purpose of this study is to understand the stage-specific duality of TGF-<italic>β</italic> and Smad function, and the molecular mechanism underlying the role of Smad7 in the metastasis of colon cancer. In an experimental model of colon cancer liver metastasis, we have shown that the stable expression of Smad7 enhances liver metastasis. We have observed a diffuse positive Ki67 staining and poorly differentiated spindle cell morphology in the liver metastases. The expression of junctional proteins is increased in the liver metastases, the expression of some of which has been shown to be elevated in the metastases of human cancers. Our data provide the first evidence that Smad signalling plays a protective role in TGF-<italic>β</italic>-induced colorectal cancer metastasis.</p>" ]

[ "<p>We are grateful to Dr M Key Washington (Department of Pathology, Vanderbilt University Medical Center) for analysing our IHC data. This study was supported by R01 CA95195 and CA113519, NCI SPORE grant in lung cancer (5P50CA90949) and a Clinical Innovator Award from Flight Attendant Medical Research Institute (to PKD).</p>" ]

[ "<fig id=\"fig1\"><label>Figure 1</label><caption><p>Stable expression of Smad7 induces liver metastasis of colon adenocarcinoma (FET) cells after splenic injection in athymic nude mice. (<bold>A</bold>) Cells from each pool of parental FET, vector control and Smad7 overexpressing cells were injected into spleen of athymic nude mice. Mice were sacrificed 33 days after injection, liver tissues were separated and pictures were taken. (<bold>B</bold>) Represents total number of mice in each group used for splenic injection and total number of mice showing liver metastases. The average body weight and the average weight of liver tissues from each group with respective standard deviation were shown. (<bold>C</bold>) The graph shows the average of the ratio of liver weight and total body weight of each mouse in each group after metastases to the liver. (<bold>D</bold>) The PCR analyses were performed for hygromycin gene using genomic DNA from parental FET cells, vector control clone and three stable Smad7 clones as well as from the normal liver and liver metastases generated using these cell lines (top panel). Expression of Smad7 in liver metastasis was verified by western blotting using lysates prepared from normal livers of mice injected with FET cells and vector control clone as well as liver metastases generated by stable Smad7 clones (bottom panel).</p></caption></fig>", "<fig id=\"fig2\"><label>Figure 2</label><caption><p>Changes in protein expression in liver metastases. (<bold>A</bold>) Lysates were prepared from normal livers of mice injected with FET cells and vector control clone as well as liver metastases generated by stable Smad7 clones. Western blot analyses were performed using antibodies against T<italic>β</italic>RII, phospho-Smad2, Smad2, Smad3 and Smad4. <italic>β</italic>-actin was used as loading control. (<bold>B</bold>) Non-Smad pathway proteins were analysed by western blotting using antibodies against phospho-AKT, AKT, phospho-ERK and ERK. <italic>β</italic>-actin was used as loading control. (<bold>C</bold>) Junctional proteins such as E-cadherin, <italic>β</italic>-catenin, ZO-1, Claudin-1, Claudin-4, Claudin-7 and nm23, as well as uPA and p53 in normal and metastatic livers, were analysed by western blotting. <italic>β</italic>-actin was used as loading control. (<bold>D</bold>) The expression of cell cycle regulatory proteins such as p21^Cip1, p27^Kip1, Cdk2, Cdk4, Cyclin D1, phospho-Rb and Rb were analysed by western blotting. <italic>β</italic>-actin was used as loading control.</p></caption></fig>", "<fig id=\"fig3\"><label>Figure 3</label><caption><p>Correlation between higher levels of protein expression and aggressiveness of liver metastasis. (<bold>A</bold>) Lysates from livers with different degree of metastasis and unaffected normal liver tissues were tested for western blot analysis using antibodies against p53, E-cadherin, nm23, Claudin-4 and Claudin-7. <italic>β</italic>-actin was used as loading control. Aggressiveness in liver metastasis was assessed by the number of tumour nodules, volume of the liver occupied by tumour and total weight of the liver 33 days after splenic injection. (<bold>B</bold>) Intensity of protein bands for each tumour from mice in each group (<italic>n</italic>=5) with different degree of metastasis were calculated using TotalLab TL 100 software. The relative average band intensity for each protein from tumours with different degree of aggressiveness in liver metastasis was plotted. Data were presented as the mean±s.e. of five different tumours per group. The <italic>P</italic>-values were calculated by two-way ANOVA method using GraphPad Prism 4 software. ^*<italic>P</italic>&lt;0.0001; ^**<italic>P</italic>&lt;0.005; ^***<italic>P</italic>&lt;0.001 by comparison with corresponding control values were shown. (<bold>C</bold>) The graph shows the percentage of total number of mice, with or without liver metastases, inoculated with vector control cells or Smad7-expressing cells. Percentage in the parenthesis indicates the aggressiveness in liver metastasis assessed as above.</p></caption></fig>", "<fig id=\"fig4\"><label>Figure 4</label><caption><p>Localisation of junctional proteins and phospho-Smad2 in Smad7-induced liver metastases. Tissues from normal livers of parental and vector control cells injected mice and liver metastases from Smad7-expressing clones injected mice were collected 33 days after splenic injection. (<bold>A</bold>) Shows staining with H&amp;E, proliferation marker Ki67 and Tunnel. (<bold>B</bold>) Shows immunohistochemical staining for phospho-Smad2, E-cadherin, Claudin-1 and Claudin-4. Pictures were taken at original magnification of 630 × .</p></caption></fig>", "<fig id=\"fig5\"><label>Figure 5</label><caption><p>A model showing the role of Smad7 signalling in colon cancer metastasis. The TGF-<italic>β</italic> activates T<italic>β</italic>R heterotetrameric complex, leading to activation of the classical Smad2, Smad3 and Smad4 signalling cascade that induces cyclin-dependent kinase inhibitor p21^Cip1 and suppresses pro-oncogenic c-Myc expression. The p21^Cip1 expression leads to inhibition of cyclin-dependent kinases (CDKs) and cyclins that results in hypophosphorylation of Rb, repression of E2F transcriptional activity and inhibition of cell cycle progression. Smad7 negatively regulates TGF-<italic>β</italic>/Smad signalling pathways to induce cell proliferation by suppressing p21^Cip1 and by induction of c-Myc. Smad7 induces cell survival through the activation of AKT and inhibits apoptosis through the induction of TRX (thioredoxin-1) and ASK1 (apoptosis signal-regulating kinase-1). Smad7 also cooperates with activated Ras and induces tumorigenicity. All of these deregulations of cell behaviour may finally contribute to metastasis. Smad7 expression is induced by TGF-<italic>β</italic>, EGF, TNF-<italic>α</italic> and IFN-<italic>γ</italic>.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"6604562f1\"/>", "<graphic xlink:href=\"6604562f2\"/>", "<graphic xlink:href=\"6604562f3\"/>", "<graphic xlink:href=\"6604562f4\"/>", "<graphic xlink:href=\"6604562f5\"/>" ]

[]

{ "acronym": [], "definition": [] }

[]

[ "<title>Materials and methods</title>", "<title>Samples</title>", "<p>Foetal tissues were obtained from the MRC Tissue Bank and from the adult tissues from patients undergoing clinically indicated gastroscopy at University College London Hospitals NHS Foundation Trust (Joint UCL/UCLH Committees on the Ethics of Human Research approvals 04/0011 and 01/0237). Genomic DNA and RNA from nine cancer cell lines were also tested: lung cancer cell lines A427, A549, NCI-H522, NCI-H358, NCI-H441, NCI-H23, NCI-H460, laryngeal carcinoma and HEP-2, and were kindly supplied by Dr Jeremy Hull, University Department of Paediatrics, John Radcliffe Hospital, Oxford.</p>", "<title>RNA preparation</title>", "<p>Tissues were stored at −70°C. Frozen tissue (25–50 mg) was crushed into a fine powder with a pestle and mortar, followed by RNA extraction using RNA-Bee™ Biogenesis Ltd, Poole, UK, according to the manufacturer's protocol. The total RNA was re-suspended in RNase-free water, quantified using a spectrophotometer and the quality checked on 1% agarose gels.</p>", "<p>Southern blot analyses were performed as previously described (##REF##12942364##Fowler et al, 2003##).</p>", "<title>DNA preparation</title>", "<p>For PCR-based genotyping, genomic DNA was retrieved from the residue and organic phase of the RNA extraction mixture, after removal of the aqueous phase, by adding a back extraction buffer (4 <sc>M</sc> guanidine thiocyanate, 50 m<sc>M</sc> Na citrate, 1 <sc>M</sc> Tris, pH 10.5), followed by DNA precipitation using isopropanol, as described in the manufacturer's instructions. Higher quality tissue DNA, used for Southern blotting, was prepared using the Puregene® DNA Purification Kit (Gentra Systems, Minneapolis, MN, USA): 10–20 mg tissue was crushed in liquid nitrogen with a pestle and mortar and the DNA extracted according to the manufacturer's instructions.</p>", "<title>cDNA synthesis</title>", "<p>Complementary DNA was synthesised with M-MLV Reverse Transcriptase (Invitrogen, Paisley, Scotland) using 200 or 400 ng of total RNA, 0.2 m<sc>M</sc> of each dNTP, 0.25 <italic>μ</italic>g random hexamers and 20 U of RNAase inhibitor (RNaseOUT™, Invitrogen) in 20 <italic>μ</italic>l. For PCRs, a 1/10 dilution of the cDNA was used. In each case, control transcriptions were run with no reverse transcriptase; no-template blanks were also included in the PCRs as negative controls. Primers from a non-polymorphic region of <italic>MUC1</italic> and from the ribosomal S14 gene (<italic>RPS14</italic>) were used as controls for template quality.</p>", "<p>Oligonucleotide primers were purchased from Sigma-Genosys, Haverhill, UK. Thermocycling was conducted on an MJ Research PTC-200 Peltier Thermal Cycler (Genetic Research Instrumentation, Braintree, UK). Primer sequences and cycling conditions are shown in <xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Table 1</xref>.</p>", "<p>For rs4072037 typing, the primers 1946 AGS and Cy5-labelled Exon2AS (see <xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Table 1</xref>) were used to generate a 260 bp product, followed by restriction enzyme digestion using <italic>AlwN1</italic>, for which the sequence cuts when there is an A allele and fails to cut if there is a G allele, as was done earlier (##REF##12942364##Fowler et al, 2003##). Samples with no uncut product were interpreted as AA and no cut product, as GG, assuming no silent alleles.</p>", "<p>Electrophoresis was conducted on 4.8% acrylamide (19 : 1) UltraPure™ Sequagel gels using an ALF express™ sequencer (Amersham Pharmacia, Little Chalfont, UK). PCR products were run with molecular size markers mixed with the PCR product and loading buffer. The same technique was used for transcript analyses. The relative peak heights were also recorded to semi-quantify the relative amount of <bold>a</bold> and <bold>b</bold> transcripts.</p>", "<p>PHASE (version 2.1), which uses a Bayesian statistical method for haplotype construction (##REF##11254454##Stephens et al, 2001##), was downloaded from <ext-link xmlns:xlink=\"http://www.w3.org/1999/xlink\" ext-link-type=\"uri\" xlink:href=\"http://stephenslab.uchicago.edu/software.html\">http://stephenslab.uchicago.edu/software.html</ext-link>.</p>" ]

[ "<title>Results</title>", "<title>Relationship of splicing with genotype: variable exon 2 spliceoforms</title>", "<p>All the samples were first typed for rs4072037. Corresponding cDNA from 68 foetuses (total of 120 tissue samples), 36 adults (21 gastric and 15 duodenal samples of which 13 and 9, respectively, had normal histology and the rest had various degrees of inflammation) as well as the nine cancer cell lines were amplified using primers to detect the variable splicing at the 5′ end of exon 2 (##FIG##0##Figure 1A##).</p>", "<p>Up to four possible transcripts, <bold>a</bold>, <bold>b</bold>, <bold>c</bold> and <bold>d</bold>, could be detected in a single sample. ##FIG##0##Figure 1B## shows PCR products from three representative individuals of AA, AG and GG genotypes. The results are summarised in ##TAB##0##Table 1##.</p>", "<p>There was a clear correlation between the transcripts observed and the rs4072037 genotype. In all cases where the rs4072037 G allele was present, <bold>a</bold> transcripts were present in the RNA, and when the rs4072037 A allele was present, there were <bold>b</bold> transcripts. In the AA homozygotes, there was no <bold>a</bold> transcript, and in most GG homozygotes, there was no <bold>b</bold> transcript, although in two GG foetal lungs and one GG foetal stomach, one GG adult stomach and one GG adult duodenum, a trace amount of the <bold>b</bold> variant was detected. In the AG heterozygotes, both <bold>b</bold> and <bold>a</bold> transcripts were observed, though <bold>b</bold> was usually more prominent and the proportions of <bold>a</bold> and <bold>b</bold> were variable (##FIG##0##Figure 1C##).</p>", "<p>Expression of the two minor transcripts, <bold>c</bold> and <bold>d</bold>, was also associated with the rs4072037 allele. When detected, the smallest transcript <bold>d</bold> was always associated with G allele, which also produces the <bold>a</bold> transcript. The <bold>c</bold> transcript on the other hand was found with the A allele, which also generates the <bold>b</bold> transcript.</p>", "<title>Spliceoforms lacking the TR domain</title>", "<p>The expression of the ‘non-TR’ transcripts was examined using the foetal tissues and with the same protocol with the same exon 1 primer and a reverse primer located at the 3′ end of exon 2 (##FIG##0##Figure 1##), except that two extra rounds of PCR amplification and double the cDNA concentration were usually required to detect these components. Up to seven different transcripts were detected (##FIG##1##Figure 2##). A summary of the results is shown in ##TAB##1##Table 2##. On the basis of their sizes, the non-TR transcripts were identified as the previously reported <italic>MUC1</italic>/X, <italic>MUC1</italic>/Y, <italic>MUC1</italic>/Z, as well as <italic>MUC1</italic>/Y alt (##REF##11606095##Obermair et al, 2001##), the non-TR transcript analogous to <bold>a</bold>. Two additional components were 27 bases larger than <italic>MUC1</italic>/X and <italic>MUC1</italic>/Z, respectively, and thus are expected to be alt or <bold>a</bold>-like variants of transcripts X and Z, respectively, and were thus named <italic>MUC1</italic>/X alt and <italic>MUC1</italic>/Z alt. <italic>MUC1Yc</italic> was so named because it is 9 bases shorter than the <italic>MUC1</italic>/Y variant and thus likely to correspond to a non-TR version of <bold>c</bold>.</p>", "<p>A clear genotype-associated pattern of expression for these variants was observed (##TAB##1##Table 2##). In rs40723037 AA homozygotes, only the non-alt variants were present, whereas in rs40723037 GG homozygotes, only the alt variants were observed. In the AG heterozygotes, both the alt and non-alt forms were detected.</p>", "<title>Variability of proportions of the major transcripts <bold>a</bold> and <bold>b</bold></title>", "<p>The fact that in the rs4072037 heterozygotes, the peak heights of the <bold>a</bold> transcript were almost always lower than the <bold>b</bold> transcript (##FIG##0##Figure 1B and C##) is probably partly a reflection of the relative signal detection by the ALF Express™ machine and may also reflect the PCR of the different sized fragments (<bold>a</bold> being 27 nucleotides longer than <bold>b</bold>). However, the relative levels of the two transcripts (##FIG##0##Figure 1C##), though variable, were reproducible for a particular sample. Although measurements using the ALF Express could at best be considered semi-quantitative, peak heights were recorded to evaluate this variability in relation to tissue source and individual.</p>", "<title>Tissue differences</title>", "<p>Consideration of all heterozygous samples of a particular tissue showed that the <bold>a</bold>/<bold>b</bold> ratios were not normally distributed, particularly in the case of the liver and trachea. The median values obtained for each of the tissues were 0.47 (range 0.06–0.82) for lung, 0.49 (range 0.12–0.64) for stomach, 0.7 (range 0.08–1.33) for liver and 0.26 (range 0.04–0.35) for trachea, and the data sets were compared using the non-parametric Mann–Whitney test. There was a significant difference in transcript ratio between the liver and trachea (<italic>P</italic>=0.0128) and between each of these and the lung (<italic>P</italic>=0.011) (<italic>P</italic>=0.0047) or stomach (<italic>P</italic>=0.0053) (<italic>P</italic>=0.0016). There was no significant correlation of transcript proportions with gestational age, although only a relatively small range of gestational ages was represented within the sample set (11–20.4 weeks).</p>", "<title>Inter-individual differences: allelic origin of the transcripts</title>", "<p>The <bold>a</bold>/<bold>b</bold> transcript ratio in three of the heterozygotes was particularly imbalanced (##FIG##0##Figure 1C##, individual 2). In one of these three cases, two tissues were available from the same individual, and both had a very low expression of <bold>a</bold>. To interpret the source of this asymmetry, we determined the allelic origin of the <bold>a</bold> and <bold>b</bold> transcripts for this individual, by digesting the cDNA with <italic>AlwNI,</italic> the enzyme used for rs4072037 genotyping in genomic DNA. The enzyme cuts the rs4072037 A allele, which is expected to be present in the <bold>b</bold> transcript, whereas the <bold>a</bold> transcript is expected to have the rs4072037 G allele and would not be cut. Samples from both the trachea and lung were tested. A ‘typical’ rs4072037 AG heterozygote and an rs4072037 AA and rs4072037 GG homozygote were used as controls. The results (<xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Figure 1</xref>) show that the <bold>G</bold> allele is completely cut by the enzyme and thus does not contribute to the <bold>b</bold> transcript in either heterozygote, and implies that the G allele is less stable or poorly transcribed in the individual with very low amounts of <bold>a</bold> transcript.</p>", "<title>Possible genetic origin of inter-individual differences</title>", "<p>The detection of dramatically lower amounts of <bold>a</bold> transcript in two tissues from the same individual and the non-normal distributions of the <bold>a</bold>/<bold>b</bold> transcript ratios suggested the possibility of a genetic origin for some of the differences in transcript ratio. In support of this, ratios obtained from the lung and stomach from the same individuals showed a significant correlation (<italic>P</italic>=0.05, data not shown).</p>", "<p>The possibility that the rate of transcript synthesis or transcript stability is affected by the variable lengths of the TR domains of <italic>MUC1</italic> was therefore examined by testing the TR polymorphism in the same set of foetal samples and comparing allele length ratios with transcript ratios for the lung and stomach, the two tissues for which the most samples were tested. ##FIG##2##Figure 3A## shows an autoradiograph of a typical Southern blot.</p>", "<p>The haplotypes comprised of rs4072037 (G or A) and the TR length, binned as long (L) and short (S) intronic, were inferred using PHASE. Approximately 90% of the chromosomes have non-recombinant haplotypes, namely, A-S (58%) or G-L (33%) with only 5% AL and 4% GS, consistent with what was found previously in Europeans (##REF##12942364##Fowler et al, 2003##; ##UREF##0##Teixeira, 2005##). Thus, one can infer that in the majority of rs4072037 AG heterozygotes, the <bold>b</bold> transcript encoded by the A allele carries a short TR array, whilereas the <bold>a</bold> transcript encoded by the G allele carries a long array. Less than 0.2% double heterozygotes (5% × 4%) are expected to have the converse arrangement.</p>", "<p>##FIG##2##Figure 3B## shows the TR length ratios (length of the long TR domain, typically rs4072037 G allele carrying haplotype over the short TR domain, typically the rs4072037 A allele carrying haplotype, see above) plotted against the <bold>a</bold>/<bold>b</bold> transcript ratios present in the lung. There was no statistically significant correlation for either the lung or the stomach. There was also no significant correlation when differences in absolute length were plotted against the transcript ratio (data not shown).</p>", "<title>Data mining</title>", "<p>All human exon 1 and exon 2 boundary containing <italic>MUC1</italic> transcripts, annotated on the UCSC Browser as of June 2007, were retrieved from the GenEMBL database. Where available, all original literature references were examined to confirm that the full sequence was taken from the clones reported and checked to determine whether any composite sequences were included in the list. For 15 of 37 annotated <italic>MUC1</italic> sequences, literature references were available. In 14 of the 15 referenced sequences, appropriate sequencing appeared to have been done. In one case, the transcript submitted was declared as a composite sequence (##REF##2318825##Ligtenberg et al, 1990##).</p>", "<p>Sequences upstream of the TR domain in exon 2 were aligned with ClustalW. Four classes of transcripts were observed, which were named for convenience as <bold>a</bold>, <bold>b</bold>, <bold>c</bold> and <bold>d</bold>. As most were either short transcripts or incomplete sequences, it was not formally possible to determine how many came from transcripts that contain a full TR domain. The <bold>a</bold> variant is represented by eight transcripts and the <bold>b</bold> transcripts by 26 sequences (see <xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary information</xref> for list), 13 of which were from HeLa cells submitted from one laboratory (Zhang and Lu, 2003, database submission only).</p>", "<p>All 26 <bold>b</bold> transcripts have the rs4072037 A allele. Although five of eight annotated <bold>a</bold> transcripts have the rs4072037 G allele, the other three transcripts carried the A allele. All three (M32738, AY327587 and S81781) were from cancer-derived material, but it is noteworthy that one was from the composite sequence (##REF##2318825##Ligtenberg et al, 1990##), and one has no associated literature reference. Two <bold>c</bold> transcripts (AF125525 and Z17325) and one <bold>d</bold> transcript (Z17325) were also identified but the rs4072037 A/G SNP is not included within the shorter exonic sequence of <bold>c</bold> and <bold>d</bold> so that genotype information cannot be obtained.</p>", "<title>Splice-prediction programs</title>", "<p>A number of bioinformatic prediction tools (SplicePredictor: <ext-link xmlns:xlink=\"http://www.w3.org/1999/xlink\" ext-link-type=\"uri\" xlink:href=\"http://deepc2.psi.iastate.edu/cgi-bin/sp.cgi\">http://deepc2.psi.iastate.edu/cgi-bin/sp.cgi</ext-link>, GeneSplicer: <ext-link xmlns:xlink=\"http://www.w3.org/1999/xlink\" ext-link-type=\"uri\" xlink:href=\"http://www.tigr.org/tdb/GeneSplicer/gene_spl.html\">http://www.tigr.org/tdb/GeneSplicer/gene_spl.html</ext-link>, Berkeley Drosophila Genome Project: <ext-link xmlns:xlink=\"http://www.w3.org/1999/xlink\" ext-link-type=\"uri\" xlink:href=\"http://www.fruitfly.org/\">http://www.fruitfly.org/</ext-link> and NetGene2: <ext-link xmlns:xlink=\"http://www.w3.org/1999/xlink\" ext-link-type=\"uri\" xlink:href=\"http://www.cbs.dtu.dk/services/NetGene2/\">http://www.cbs.dtu.dk/services/NetGene2/</ext-link>) were also used to examine which if any of the alternative 3′ splice sites was predicted by the programs. The 3′ splice site for the <bold>b</bold> transcript was not predicted by any of these four programs. SplicePredictor, however, predicted the 3′ splice site that is used for the <bold>a</bold> transcript (regardless of nucleotide at rs4072037). This is consistent with the calculated strength of the splice sites based on the algorithm of ##REF##3658675##Shapiro and Senapathy (1987)##, which suggested that the strength of the 3′ splice site used by the <bold>a</bold> transcript is significantly stronger than the site used for the <bold>b</bold> transcript.</p>", "<p>The results of the analysis using the program Splice Sequences Finder version 2.2 (<ext-link xmlns:xlink=\"http://www.w3.org/1999/xlink\" ext-link-type=\"uri\" xlink:href=\"http://www.umd.be/SSF\">http://www.umd.be/SSF</ext-link>), which predicts exonic motifs, showed that three exonic splicing enhancer motifs and one exonic splicing silencer motif overlap the rs4072037 SNP in the presence of either the A or the G alleles. However, the two allelic variants were predicted to bind different proteins.</p>" ]

[ "<title>Discussion</title>", "<p>This is the first large scale study that shows a clear association of alternative splicing of the 5′ exon 2 region of <italic>MUC1</italic> with the rs4072037 A/G SNP. Both experimental and database trawling show that there is a strong association between transcript type and exon 2 allele status, implying that SNP rs4072037 controls splice acceptor usage as originally predicted from data obtained using cancer cell lines (##REF##2014168##Ligtenberg et al, 1991##). In addition to the major spliceoforms, the polymorphism affects all the minor ones, including ones with additional deletions in exon 2 (<bold>c</bold> and <bold>d</bold>) and those without a TR domain, the latter association being described here for the first time. Our study confirms that these components are present as a minor fraction of the <italic>MUC1</italic> transcripts in ‘normal’ as well as cancer tissues. The alternative splicing at the start of exon 2 is independent of alternative splicing of the TR region, which is not controlled by rs4072037. While this work was in progress, an association of rs4072037 with the <bold>a</bold>, <bold>b</bold>, <bold>c</bold> and <bold>d</bold> transcripts was also reported in adult corneal tissue (##REF##16631167##Imbert et al, 2006##), confirming the effect in another tissue. These authors did not, however, test genomic DNA but rather deduced genotype from cDNA.</p>", "<p>Analysis of haplotypes 310 000 bp downstream and 330 000 bp upstream of <italic>MUC1TR</italic> using CEPH trio data shows a number of SNPs in very strong linkage disequilibrium with rs4072037 (data not shown). However, none were completely associated with rs4072037. In addition, extensive re-sequencing (NIEHS (<ext-link xmlns:xlink=\"http://www.w3.org/1999/xlink\" ext-link-type=\"uri\" xlink:href=\"http://egp.gs.washington.edu/\">http://egp.gs.washington.edu/</ext-link>) and A Teixeira and DM Swallow, unpublished) failed to identify any other intragenic SNPs, notably there being none in intron 1 that might have been responsible for this splicing event. Indeed no other common SNPs have been identified within the gene, showing that rs4072037 must be directly responsible for the splicing polymorphism.</p>", "<p>However, this splicing polymorphism was not predicted by any of the exon prediction programs currently available. Indeed for both alleles, the longer transcript was predicted. On the other hand, the A-to-G substitution is predicted to alter protein binding in Splice Sequences Finder version 2.2. The secondary structure of the pre-mRNA is also predicted to be different, where only the G allele forms a physiologically stable stem loop structure (as predicted by the mfold program; <ext-link xmlns:xlink=\"http://www.w3.org/1999/xlink\" ext-link-type=\"uri\" xlink:href=\"http://frontend.bioinfo.rpi.edu/applications/mfold/cgi-bin/rna-form1.cgi\">http://frontend.bioinfo.rpi.edu/applications/mfold/cgi-bin/rna-form1.cgi</ext-link>) (##REF##10329189##Mathews et al, 1999##; ##REF##12824337##Zuker, 2003##). Indeed the possible importance of this difference was noted previously (##REF##2014168##Ligtenberg et al, 1991##). Whatever the mechanism, there appears to be some leakiness in the control: small amounts of the <bold>b</bold> transcript can be found in a few of the normal tissues from GG homozygotes, and in the sequences submitted to databases, a few <bold>a</bold> transcripts from cancer tissues were found that carry the A allele. The proportions of the transcripts are variable in heterozygotes, and this is not attributable to this leakiness as shown by the restriction enzyme digestion experiments. It seems to reflect cell-to-cell (and thus tissue-to-tissue) as well as person-to-person differences in allelic expression (as opposed to splicing), which may be epigenetic or genetic in origin. Comparison of the relative differences in allelic expression with relative length of the TR domain suggests that the genetically determined differences in transcript length do not significantly affect transcript quantity.</p>", "<p>Putting together all these observations, it can be concluded that none of these transcripts are tumour-specific. Their over-representation in tumour material cannot really be evaluated without information on the SNP genotype as well. It is possible that inflammation and cancer affect the relative expression of the alleles and/or splicing. However, it is also possible that the genotypes were unevenly represented in the cohorts under study. In most studies, the <bold>a</bold> transcript (normally encoded by the G allele) was more abundant. Studies on gastric cancer and gastritis (##REF##9076520##Carvalho et al, 1997##; ##REF##12105832##Vinall et al, 2002##) have shown that <italic>MUC1TR S</italic> alleles are over-represented, and the A allele (not the G allele) is normally associated with short TR alleles (##REF##8835095##Pratt et al, 1996##; ##REF##12105832##Vinall et al, 2002##). However, in our own work, we have shown that it is in fact the recombinant GS alleles that are over-represented in gastric cancer and gastritis (A Teixeira <italic>et al</italic>, unpublished). Unfortunately, in most of the studies that suggest an association of tumorigenesis with the <bold>a</bold> transcript, data for neither the SNP nor the TR length were collected. Thus, it is possible that the associated rs4072037 G allele is often present as a recombinant haplotype with the short TR domain. These recombinant haplotypes are present at a reasonable frequency in the population (7–10%).</p>", "<p>Interestingly, in one paper (##REF##12462382##Schmid et al, 2002##) where genotyping was done, substantial quantities of <bold>a</bold> transcript were detected in three cell lines, which were homozygous AA for rs4072037. This is suggestive of secondary ‘leakiness’ rather than allelic differences in expression and could be a genuine cancer-related change in splicing, which is consistent with the finding of some tumour-derived A allele <bold>a</bold> transcripts on the databases. These observations, together with the study described here, also indicate that transcript phenotype cannot be used as a reliable surrogate for genotype. It had been our intention to examine these splicing events in relation to <italic>H. pylori</italic> gastritis. It is clear, however, from the work reported here that there are too many variables to address this question without having a very large cohort of samples and better quantitative methods.</p>", "<p>It should be emphasised that the allelic variation of rs4072037 polymorphism may affect the function of MUC1 protein. The alternative splicing event occurs within the signal peptide of MUC1 in the vicinity of the known proteolytic cleavage sites found for the b variant (##REF##11341784##Parry et al, 2001##). Although there is no experimental evidence for the <bold>a</bold> variant, SignalP 3.0 site (<ext-link xmlns:xlink=\"http://www.w3.org/1999/xlink\" ext-link-type=\"uri\" xlink:href=\"http://www.cbs.dtu.dk/services/SignalP/\">http://www.cbs.dtu.dk/services/SignalP/</ext-link>), which accurately predicts cleavage of the <bold>b</bold> variant, suggests that the signal peptide for the <bold>a</bold> variant will end at residue 23 from the start of translation, and an above threshold signal peptide cleavage site was predicted to lie between the T and A residues at positions 22 and 23, respectively. This is within the region that contains the inserted amino-acid sequence. The predictions for all the 5′ transcript variants are shown in ##FIG##3##Figure 4##.</p>", "<p>The normal MUC1 biosynthesis pathway and glycosylation include targeting of the mature protein to the apical surfaces of epithelial cells, and we have previously shown that this is altered in <italic>H. pylori</italic> gastritis (##REF##12105832##Vinall et al, 2002##). The alternative splicing events within the signal peptide sequence domain could possibly lead to differences in cellular trafficking, as observed in interleukin-15 (##REF##10869346##Kurys et al, 2000##), altering the localisation of the mature protein. If, however, variation in the signal peptide does affect apical targeting, in either the normal or diseased mucosa, the effect is not ‘all or nothing’ because apical staining is seen in the normal mucosa of individuals of all three genotypes and no staining was seen in any of the gastritis specimens regardless of genotype (LE Vinall <italic>et al</italic>, unpublished).</p>" ]

[]

[ "<p>Current addresses: School of Biological Sciences, Nanyang Technological University, 60 Nanyang Drive, Singapore 637551, Republic of Singapore</p>", "<p>Current addresses: Molecular and Population Genetics Laboratory, Cancer Research UK – London Research Institute, 44 Lincoln's Inn Fields, London WC2A 3PX, UK</p>", "<p>The membrane mucin MUC1 is aberrantly expressed in a variety of cancers, and in stomach, it is a ligand for <italic>Helicobacter pylori</italic> where it plays a role in gastric carcinogenesis. Splicing variation, leading to a 9-amino acid insertion in the signal peptide region, was proposed to be because of a single-nucleotide polymorphism (rs4072037) at the 5′ end of exon 2, but is also reported to be cancer-associated. However, the effect of rs4072037 on this splicing event in healthy non-cancer tissues and on the additional spliceoforms of <italic>MUC1</italic>, including those lacking the polymorphic tandem repeat (TR) domain, has never been investigated. Here we show that in both foetal and adult tissues of known genotype, there is clear evidence for the role of rs4072037 in controlling alternative splicing of the 5′ exon 2 region of both full-length transcripts and those lacking the TR domain. Although there is some evidence for additional genetic and epigenetic influences, there is no indication of an effect of the TR domain on the proportions of the spliceoforms. In conclusion, over-representation of certain transcripts in tumour material cannot be evaluated without information on the SNP genotype as well.</p>" ]

[ "<p>MUC1 is a highly polymorphic membrane-associated mucin that is often aberrantly expressed in cancer (##REF##12463741##Taylor-Papadimitriou et al, 2002##). It has a centrally located tandem repeat (TR) domain (##REF##3965141##Lan et al, 1985##; ##REF##2888110##Gendler et al, 1987##, ##REF##3417635##1988##, ##REF##2307533##1990##; ##REF##3600778##Swallow et al, 1987##; ##REF##2112460##Hareuveni et al, 1990##) comprised of 20–120 or more repeat units of 60 nucleotides, which encode 20 amino acids. The repeating units include several serine and threonine residues, which carry most of the glycosylation, and this glycosylation, as well as the general pattern of expression, is altered in cancerous cells (##REF##12463741##Taylor-Papadimitriou et al, 2002##). Susceptibility to <italic>Helicobacter pylori</italic> gastritis and to gastric cancer appears to be associated with <italic>MUC1</italic> allele length (##REF##9076520##Carvalho et al, 1997##; ##REF##11464247##Silva et al, 2001##, ##REF##12734543##2003##; ##REF##12105832##Vinall et al, 2002##). MUC1 binds to <italic>H. pylori</italic> (##REF##15260802##Linden et al, 2004##; ##REF##17919495##McGuckin et al, 2007##), <italic>Muc1</italic> knockout mice are susceptible to <italic>H. pylori</italic> gastritis (##REF##17919495##McGuckin et al, 2007##) and there is an altered pattern of expression of MUC1 in <italic>H. pylori</italic> gastritis (##REF##12105832##Vinall et al, 2002##). The observations that MUC1 plays a role in the progression to gastric cancer highlight the importance of understanding all the aspects of the normal variation of this gene.</p>", "<p>During the course of various cancer-related studies, several variant <italic>MUC1</italic> transcripts were reported. Early cloning experiments revealed the presence of transcripts (<bold>a</bold> and <bold>b</bold>) with an alternative 27 bp intron retention event at the start of exon 2 (##REF##2318825##Ligtenberg et al, 1990##). Although a genetic basis for this variable splicing event was first suggested by ##REF##2014168##Ligtenberg et al (1991)##, who used cancer cell lines, the longer <bold>a</bold> transcript was also reported to be more abundant in cancer tissues (##REF##8608966##Weiss et al, 1996##; ##REF##11606095##Obermair et al, 2001##, ##REF##12115565##2002##; ##REF##12462382##Schmid et al, 2002##, ##REF##14534730##2003##) and to have potential prognostic value. Therefore, the function of the genetic polymorphism implicated (rs4072037 G/A, located at nucleotide position 8 of exon 2 of the <bold>b</bold> variant) was left in some doubt. An additional alternative splicing event is known to occur at the 5′ end of exon 2, which generates two minor transcripts denoted as <bold>c</bold> and <bold>d</bold>, and was first reported by ##REF##11606095##Obermair et al (2001)##. Variants <bold>c</bold> and <bold>d,</bold> respectively, have 9 and 27 fewer nucleotides, at the start of exon 2, than the <bold>b</bold> transcript. The splice variants observed in breast (##REF##12462382##Schmid et al, 2002##, ##REF##14534730##2003##), cervical (##REF##11606095##Obermair et al, 2001##) and ovarian tumours (##REF##12115565##Obermair et al, 2002##) were suggested to be associated with transcripts <bold>b</bold> and <bold>a</bold>, respectively, and thus are possibly also under the influence of rs4072037. These alternative splicing events lie within the signal peptide domain and are very close to two signal peptide cleavage sites that have been observed experimentally in <bold>b</bold> variant (##REF##11341784##Parry et al, 2001##).</p>", "<p><italic>MUC1</italic> is also known to have alternative spliceoforms that lack the TR domain (##REF##7988707##Zrihan-Licht et al, 1994a##, ##REF##7925397##1994b##; ##REF##9180140##Baruch et al, 1997##, ##REF##10197628##1999##; ##REF##11606095##Obermair et al, 2001##, ##REF##12115565##2002##). Each of these transcripts is generated through the use of one additional 5′ splice site and one of three different 3′ splice sites in the sequence, which flank the TR domain in exon 2. <italic>MUC1</italic>/Z is the longest transcript, followed by <italic>MUC1</italic>/Y, which is 54 bases shorter, followed by <italic>MUC1</italic>/X, which is a further 19 bases shorter, using the nomenclature adopted by ##REF##12115565##Obermair et al (2002)##. The three transcripts and an additional form called <italic>MUC1</italic>/Y alt, some of which are likely to be functional (##REF##15987679##Levitin et al, 2005##), have been detected in breast tumours (##REF##9180140##Baruch et al, 1997##, ##REF##10197628##1999##; ##REF##10389761##Hartman et al, 1999##; ##REF##11606095##Obermair et al, 2001##), cervical tumours (##REF##11606095##Obermair et al, 2001##) and ovarian tumours (##REF##12115565##Obermair et al, 2002##).</p>", "<p>The aim of this study was to investigate the role of rs4072037 in relation to the alternative splicing events of <italic>MUC1</italic>, in both TR-containing and TR-negative transcripts, for a range of normal non-cancer tissues, to help evaluate published results and as a platform for future cancer studies. Foetal samples were used to obtain several tissues from a single individual as well as the same tissue from many individuals. In addition, a number of adult gastric and duodenal samples and cancer cell lines were examined. As the TR domain makes up a significant proportion of the transcript for <italic>MUC1</italic> and is known to show considerable length variation in different individuals, this might affect the rate of transcript synthesis or transcript stability. The TR lengths of <italic>MUC1</italic> were therefore examined in the same set of foetal samples.</p>" ]

[ "<p>We acknowledge an MRC PhD Studentship and grants from the Annals of Human Genetics and the Oxford University Department of Paediatrics for W Ng, UCL ORS scholarship for AL, GABBA PhD program studentship and Fundação para a Ciência e a Tecnologia, Portugal, for AT, The Wellcome Trust grant ref 070395/Z/03/Z, for supporting the preliminary work, and Adil Elamin for help with sample collection and Jeremy Hull who provided the tumour cell lines. The work was undertaken at UCLH/UCL, which received a proportion of funding from the Department of Health's National Institute for Health Research (NIHR) Biomedical Research Centres funding scheme. We thank Sue Povey for helpful discussions.</p>", "<title>Supplementary Material</title>" ]

[ "<fig id=\"fig1\"><label>Figure 1</label><caption><p>(<bold>A</bold>) Diagram of the start of <italic>MUC1</italic> showing the positions of the PCR primers in relation to rs4072037 (asterisked). The variable extension of exon 2 is shown in white. Tandem repeat domain is shown in dark grey. The positions of the single-sense primer and two antisense primers are shown with arrows. The antisense primer downstream of the TR domain is used for amplifying the components lacking the TR domain. (<bold>B</bold>) ALF Express trace of the RT–PCR from three individuals with different rs4072037 genotypes, using the primer in exon 1 in combination with the primer upstream of the TR domain. The positions of the four transcripts, <bold>a</bold>, <bold>b</bold>, <bold>c</bold> and <bold>d</bold>, as well as the 100 and 300 bp markers are indicated. (<bold>C</bold>) ALF Express traces from two AG heterozygous individuals, one of whom shows very little <bold>a</bold> transcript. Samples from foetal lung; complete traces are shown in <xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Figure 1</xref>.</p></caption></fig>", "<fig id=\"fig2\"><label>Figure 2</label><caption><p>Variant transcripts lacking the TR domain, detected using primers Cy5 M1ProF4 and MUC1X2R. The positions of the transcripts X, Yc, X alt, Y alt, Z, Z alt, and the 150 and 300 bp markers are indicated. Samples from AA, AG and GG individuals.</p></caption></fig>", "<fig id=\"fig3\"><label>Figure 3</label><caption><p>(<bold>A</bold>) <italic>MUC1</italic> VNTR polymorphism detected on a Southern blot. The band sizes for the Raoul ladder size markers in the first and last tracks are shown on the right. Grouping of long (L) and short alleles (S) used for haplotype analysis is shown. (<bold>B</bold>) <italic>MUC1</italic>a/b transcript ratio (lung samples) plotted against <italic>MUC1TR</italic> length ratio. Similar results were obtained for stomach (data not shown).</p></caption></fig>", "<fig id=\"fig4\"><label>Figure 4</label><caption><p>Amino-acid sequences of the N-terminal ends of the <bold>a</bold>, <bold>b</bold>, <bold>c</bold>, and <bold>d</bold> encoded peptides and positions of the observed and/or predicted peptide cleavage. The sequence that undergoes alternative splicing is shown in grey on the <bold>a</bold> transcript. The upward arrows show the positions of predicted, or in the case of the <bold>b</bold> transcript the experimentally observed, peptide cleavage sites.</p></caption></fig>" ]

[ "<table-wrap id=\"tbl1\"><label>Table 1</label><caption><title>Summary of splice variants <bold>a</bold>, <bold>b</bold>, <bold>c</bold> and <bold>d</bold> observed in foetal and adult tissues and cell line samples from individuals of different genotypes for rs4072037</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Genotypes</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>AA</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>AG</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>GG</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>Splice variants present</italic>\n</td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\"> <italic>Foetal lung</italic></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Total no.</td><td align=\"center\" valign=\"top\" charoff=\"50\">22</td><td align=\"center\" valign=\"top\" charoff=\"50\">23</td><td align=\"center\" valign=\"top\" charoff=\"50\">5</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/<bold>d</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>b</bold>/<bold>c</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\">21</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/<bold>b</bold>/<bold>c</bold>/<bold>d</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">23</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/trace-<bold>b</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">2<xref ref-type=\"fn\" rid=\"t1-fn1\">a</xref></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\"> <italic>Foetal stomach</italic></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Total no.</td><td align=\"center\" valign=\"top\" charoff=\"50\">19</td><td align=\"center\" valign=\"top\" charoff=\"50\">18</td><td align=\"center\" valign=\"top\" charoff=\"50\">5</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/<bold>d</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>b</bold>/<bold>c</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\">19</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/<bold>b</bold>/<bold>c</bold>/<bold>d</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">18</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/trace-<bold>b</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1<xref ref-type=\"fn\" rid=\"t1-fn1\">a</xref></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\"> <italic>Foetal liver</italic></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Total no.</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\">6</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>b</bold>/<bold>c</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/<bold>b</bold>/<bold>c</bold>/<bold>d</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">5</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/<bold>b</bold>/<bold>c</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/<bold>b</bold>/<bold>d</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\"> <italic>Foetal trachea</italic></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Total no.</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\">7</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/<bold>d</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>b</bold>/<bold>c</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/<bold>b</bold>/<bold>c</bold>/<bold>d</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/<bold>b</bold>/<bold>c</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\"> <italic>Adult stomach</italic></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Total no.</td><td align=\"center\" valign=\"top\" charoff=\"50\">6</td><td align=\"center\" valign=\"top\" charoff=\"50\">11</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/<bold>d</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>b</bold>/<bold>c</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>b</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/<bold>b</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/trace-<bold>b</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1<xref ref-type=\"fn\" rid=\"t1-fn1\">a</xref></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/<bold>b</bold>/<bold>c</bold>/<bold>d</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">10</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\"> <italic>Adult duodenum</italic></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Total no.</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">8</td><td align=\"center\" valign=\"top\" charoff=\"50\">6</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/<bold>d</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>b</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/<bold>b</bold>/<bold>c</bold>/<bold>d</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">6</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/<bold>b</bold>/<bold>c</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/trace-<bold>b</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1<xref ref-type=\"fn\" rid=\"t1-fn1\">a</xref></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\"> <italic>Cell lines</italic></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  Total no.</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/<bold>d</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>b</bold>/<bold>c</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">  <bold>a</bold>/<bold>b</bold>/<bold>c</bold>/<bold>d</bold></td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl2\"><label>Table 2</label><caption><title><italic>MUC1</italic> ‘non-TR’ transcripts observed in the foetal sample series, subdivided according to genotype, and tissue</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th colspan=\"7\" align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Genotype rs4072037</bold>\n</th><th colspan=\"3\" align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>AA</bold>\n</th><th colspan=\"3\" align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>AG</bold>\n</th><th colspan=\"2\" align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>GG</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td colspan=\"7\" align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Total individuals</bold>\n<hr/></td><td colspan=\"3\" align=\"center\" valign=\"top\" charoff=\"50\">25<hr/></td><td colspan=\"3\" align=\"center\" valign=\"top\" charoff=\"50\">31<hr/></td><td colspan=\"2\" align=\"center\" valign=\"top\" charoff=\"50\">7<hr/></td></tr><tr><td colspan=\"7\" align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Total tissues</bold>\n<hr/></td><td colspan=\"3\" align=\"center\" valign=\"top\" charoff=\"50\">42<hr/></td><td colspan=\"3\" align=\"center\" valign=\"top\" charoff=\"50\">48<hr/></td><td colspan=\"2\" align=\"center\" valign=\"top\" charoff=\"50\">10<hr/></td></tr><tr><td colspan=\"7\" align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Spliceoforms</bold>\n</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr></tbody></table><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>X</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Yc</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Y</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Z</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>X alt</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Y alt</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Z alt</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Lung</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Stomach</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Liver</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Lung</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Stomach</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Liver</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Lung</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Stomach</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">10</td><td align=\"center\" valign=\"top\" charoff=\"50\">8</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">3</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">√</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Total</td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\"> </td><td align=\"center\" valign=\"top\" charoff=\"50\">26</td><td align=\"center\" valign=\"top\" charoff=\"50\">15</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"center\" valign=\"top\" charoff=\"50\">26</td><td align=\"center\" valign=\"top\" charoff=\"50\">16</td><td align=\"center\" valign=\"top\" charoff=\"50\">6</td><td align=\"center\" valign=\"top\" charoff=\"50\">6</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"xob1\"><label>Supplementary Figure 1</label></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"xob2\"><label>Supplementary Table 1, Supplementary Information and Supplementary Figure 1 Legend</label></supplementary-material>" ]

[ "<fn-group><fn><p><xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Information</xref> accompanies the paper on British Journal of Cancer website (<ext-link xmlns:xlink=\"http://www.w3.org/1999/xlink\" ext-link-type=\"uri\" xlink:href=\"http://www.nature.com/bjc\">http://www.nature.com/bjc</ext-link>)</p></fn></fn-group>", "<table-wrap-foot><fn id=\"t1-fn1\"><label>a</label><p>Samples that contain low amounts of transcripts not usually associated with their genotype.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"t2-fn1\"><p>Presence of the ‘alt’ splice variants is correlated with the presence of the rs4072037 G, and that of X, Y and Z transcripts is correlated with the A allele.</p></fn></table-wrap-foot>" ]

[ "<graphic xlink:href=\"6604617f1\"/>", "<graphic xlink:href=\"6604617f2\"/>", "<graphic xlink:href=\"6604617f3\"/>", "<graphic xlink:href=\"6604617f4\"/>" ]

[ "<media xlink:href=\"6604617x1.doc\"><caption><p>Click here for additional data file.</p></caption></media>", "<media xlink:href=\"6604617x2.doc\"><caption><p>Click here for additional data file.</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[]

[ "<title>Materials and methods</title>", "<title>Synthesis and characterisation of the copolymers</title>", "<p>The tyrosinamide (TyrNH₂)-, gadolinium (Gd)- and doxorubicin-containing <italic>N</italic>-(2-hydroxypropyl)methacrylamide (HPMA) copolymers were synthesised as described previously (##REF##16274831##Lammers et al, 2005##; ##REF##16417250##Kiessling et al, 2006##). Details on the synthesis of these copolymers, as well as on the preparation of the two newly generated gemcitabine-containing copolymers are provided as <xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Information</xref> online. The weight- and number-average molecular weights (<italic>M</italic>_w and <italic>M</italic>_n) and the polydispersity (<italic>M</italic>_w/<italic>M</italic>_n) of the copolymers were determined by size exclusion chromatography, and the amounts of doxorubicin and gemcitabine incorporated by means of spectrophotometry. The (weight-) average molecular weights of the two tyrosinamide-containing copolymers (i.e., poly(HPMA-co-MA-TyrNH₂)) were 30.5 and 64.5 kDa, their polydispersities were 1.3 and 1.2, and the amounts of tyrosinamide, included to allow for radiolabeling, were 0.8 and 1.1 mol%. The average molecular weight of the gadolinium-labelled copolymer (i.e., poly(HPMA-co-MA-AH-Asp-[(Asp-(OH)₂]₂)-gadolinium) was 24.8 kDa, its polydispersity was 1.9 and the amount of gadolinium was 5.2 wt% (as determined by inductively coupled plasma mass spectrometry). The average molecular weights of poly(HPMA)-GFLG-Dox (i.e., PK1) and human immunoglobulin G-modified poly(HPMA)-GFLG-Dox (i.e., IgG-PK1) were 27.9 and 900 kDa, their polydispersities were 1.5 and 4.0, and relative amounts of doxorubicin were 6.5 and 4.8 wt%, respectively. The average molecular weights of A-Gem (i.e., poly(HPMA)-AH-Gem; uncleavable) and of B-Gem (i.e., poly(HPMA)-GFLG-Gem; cleavable) were 20.2 and 24.0 kDa, their polydispersities were 1.5 and 1.6, and their drug contents were 6.8 and 10.9 wt%, respectively.</p>", "<title>Drug release and <italic>in vitro</italic> efficacy of the Gem-containing copolymers</title>", "<p>The release of Gem from A-Gem and from B-Gem was investigated at pH=7.4, at pH=6.0, and at pH=6.0 in the presence of the lysosomal cysteine protease cathepsin B. The temperature was set to resemble physiological conditions (i.e., 37°C). The concentrations of the two polymeric prodrugs were 2.3 × 10⁻³ mol l⁻¹ Gem equivalent. The concentration of cathepsin B was 1.9 × 10⁻⁷ mol l⁻¹, and its activity was standardised using the substrate <italic>α</italic>-<italic>N</italic>-benzoyl-DL-arginine <italic>β</italic>-naphthylamide. Drug release was quantified spectrophotometrically at 310 nm. The cytotoxicity of free and HPMA copolymer-bound gemcitabine was determined by seeding Dunning AT1 rat prostate carcinoma cells and A2780 human ovarian carcinoma cells into six-well plates, and by incubating them with increasing concentrations of the free drug, the two polymeric agents, and a drug-free control copolymer. Eight to ten days later, the cells were fixed and stained with crystal violet, and the number of surviving colonies was counted.</p>", "<title>Biodistributional analyses</title>", "<p>The tumour and organ accumulation of the copolymers were evaluated by means of magnetic resonance imaging (MRI) and <italic>γ</italic>-scintigraphy. The former was performed using a clinical 1.5 T whole-body MRI system (Siemens Symphony) and a custom-made radio frequency (RF) coil (##REF##16417250##Kiessling et al, 2006##). To optimise the signal-to-noise ratio, manual tuning and matching of the RF coil's circuitry was performed before each individual measurement. Additional details on the MRI-based biodistributional analyses are presented as <xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Information</xref>. For the scintigraphic analyses, two differently sized poly(HPMA-co-MA-TyrNH₂) conjugates were radiolabeled with iodine-131 (by means of the Iodogen method). Labelling efficacies were always &gt;95% (##REF##17032495##Lammers et al, 2006##). Five hundred microliters of a saline solution containing 0.1 m<sc>M</sc> of the labelled copolymers (∼300 <italic>μ</italic>Ci) were subsequently injected i.v. into male Copenhagen rats. At several different time points p.i., blood samples were collected for kinetic analyses, and the biodistribution of the copolymers was visualised by means of a Searle-Siemens scintillation camera. At 24 and 168 h, animals were killed, and tumours and organs were harvested for quantification. The residual amounts of radioactivity were determined using a gamma counter, they were corrected for radioactive decay and they were expressed as percent of the injected dose per gram tissue.</p>", "<title>Therapeutic analyses</title>", "<p>All experiments involving animals were approved by an external committee for animal welfare and were performed according to the guidelines for laboratory animals established by the German government. Experiments were performed on 6–12 month old male Copenhagen rats, using the syngeneic and radio- and chemoresistant Dunning AT1 prostate carcinoma model. Fresh pieces (∼10 mm³) of an AT1 donor tumour were transplanted subcutaneously into both hind limbs of the animals. Prior to treatment, tumours were grown for 6–8 days, until they reached an average diameter of 6 mm. Free doxorubicin was administered at its maximum tolerated dose (MTD), that is, at three doses of 2.5 mg kg⁻¹ (days 1, 8 and 15). PK1 was applied at three (doxorubicin-equivalent) doses of 5 mg kg⁻¹, and IgG-PK1 at both the 2.5 and 5 mg kg⁻¹ regimen. Local external beam radiotherapy was delivered by means of the Siemens Gammatron S (cobalt-60 <italic>γ</italic>-irradiation; dose rate ∼0.5 Gy min⁻¹) and it was applied at a regimen comparable to that routinely used in clinics: 2 Gy was given on every weekday for four consecutive weeks, i.e., for a total of 20 doses of 2 Gy (days 1–5, 8–12, 15–19, and 22–26). gemcitabine and the gemcitabine-containing copolymers were administered four times at a dose of 3 mg kg⁻¹ (days 1, 8, 15, and 22). In this case, radiotherapy was delivered three times weekly for 4 weeks at a dose of 3 Gy, that is, for a total of 12 doses of 3 Gy (days −1, 1, 3, 6, 8, 10, 13, 15, 17, 20, 22, and 24). Tumour volumes were calculated using the formula <italic>V</italic>=(<italic>a</italic> × (<italic>b</italic> × <italic>b</italic>))/2, and they were expressed relative to the volume determined on the first day of therapy. The toxicity of the combination regimens was assessed by determining the body weight loss of the animals and by analysing the number of white blood cells, red blood cells, and platelets (Bayer Advia 120 hematology analyzer).</p>", "<title>Statistical analysis</title>", "<p>Values are expressed as average±s.d. or as average±s.e.m. In the biodistributional analyses, the two-tailed <italic>t</italic>-test (for standard comparisons) or the paired <italic>t</italic>-test (for the left <italic>vs</italic> right comparisons) was used. In the therapeutic analyses, the Mann–Whitney <italic>U</italic>-test (i.e., Wilcoxon rank-sum) was used. Bonferroni–Holm <italic>post hoc</italic> analysis was used to correct for multiple comparisons. In all cases, <italic>P</italic>&lt;0.05 was considered to represent statistical significance.</p>" ]

[ "<title>Results</title>", "<title>Biodistributional analysis of HPMA copolymers</title>", "<p>To assess the validity and the therapeutic potential of carrier-based radiochemotherapy, HPMA copolymers were selected as a model drug targeting system. Copolymers of <italic>N</italic>-(2-hydroxypropyl)methacrylamide (<xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Figure 1A</xref>) are prototypic and well-characterised polymeric drug carriers that have been broadly implemented in the delivery of anticancer therapeutics (##REF##10840193##Kopecek et al, 2000##; ##REF##14529421##Rihova and Kubackova, 2003a##; ##REF##16900224##Duncan, 2006##; ##REF##18648371##Lammers et al, 2008##). <xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Figure 1B</xref> schematically depicts the different copolymers used in this study, functionalised, for example, with Gd and iodine-131 for imaging purposes, and with doxorubicin and gemcitabine for therapeutic purposes.</p>", "<p>The biodistribution of the copolymers was evaluated by means of MRI, <italic>γ</italic>-scintigraphy and HPLC. The MR angiography scans in ##FIG##1##Figure 2A## show that at 0.5 h post i.v. injection (p.i.), a 25 kDa gadolinium-labelled HPMA copolymer was localised predominantly to the vascular compartment. The color-coded maximum intensity projection (MIP; right panel in ##FIG##1##Figure 2A##) confirms the long-circulating properties of the copolymer, showing that also in tumours, the targeting system still resided predominantly within the vasculature at this time point. Up to 10 h p.i., the tumour, kidney, and liver concentrations (i.e., whole organ levels) of the conjugate were then compared to those of Gd-DTPA-BMA (i.e., gadolinium-diethylenetriaminepentaacetic-acid-bis-methylamide; Omniscan®; 0.5 kDa), which is a prototypic low molecular weight MR contrast agent that does not bind to plasma proteins and that is rapidly eliminated from blood by means of renal filtration. As shown in ##FIG##1##Figures 2B and C##, it was found that the implementation of the targeting system attenuated the renal clearance of the gadolinium label (reducing the initial peak in kidney accumulation from 60.1±7.5 to 28.5±5.0 <italic>μ</italic><sc>M;</sc>\n<italic>P</italic>&lt;0.05), and that it consequently – i.e., as a result of the EPR effect (22) – enhanced its accumulation in tumours over time. Significant localisation to liver was also noted (##FIG##1##Figure 2C##), but one needs to take into account that (I) the weight (and volume) of a rat liver tends to be ∼10 times higher than that of an average (10 × 10 mm) AT1 tumour, and that (II) at such ‘early’ time points, the signal quantified does not necessarily reflect contrast agent accumulation, as significant amounts of the copolymer are still present in circulation up to 24 h (see below), and as the liver is a highly vascularised organ.</p>", "<p>To more extensively evaluate the tumour and organ accumulation of the polymeric drug delivery system, and to do so at later time points, we next radiolabeled two differently sized tyrosinamide-containing HPMA copolymers with iodine-131, and we monitored their biodistribution scintigraphically. In line with the MR angiography data ##FIG##1##Figure 2A##, the images in ##FIG##2##Figure 3A## on the one hand again demonstrate that the polymeric drug carriers circulate for prolonged periods of time, with especially for the 65 kDa copolymer, substantial amounts still present in blood at 0.5 and 24 h p.i. (as exemplified by the high levels localised to heart). Quantification of the concentrations of the two copolymers in systemic circulation confirmed this observation, with at 24 h p.i., for instance, 11.2±0.7 and 23.7±1.2% of the injected dose still present in blood for 31 and 65 kDa pHPMA, respectively (##FIG##2##Figure 3B##). The scintigrams in ##FIG##2##Figure 3A## on the other hand also quite convincingly demonstrate that the polymeric drug delivery system presents with an acceptable biodistribution, with besides localisation to tumours, only indications for an accumulation in organs of the reticuloendothelial system (RES; i.e., liver, spleen, and lung), which is known to be involved in the clearance of long-circulating nanomedicines (##REF##10840193##Kopecek et al, 2000##; ##REF##10699287##Maeda et al, 2000##; ##REF##15688077##Torchilin, 2005##; ##REF##16900224##Duncan, 2006##). In line with this, when quantifying the tumour and organ concentrations of the smaller copolymer at 24 and 168 h p.i., actually only in spleen, levels were always significantly higher than in tumours (##FIG##2##Figure 3C##). In lung, comparable levels were found, and in all other organs, the concentrations of the copolymer were significantly lower than in tumours. For 65 kDa pHPMA, an identical pattern was observed, the only difference being that the targeting efficacy appeared to be lower at 24 h p.i. and higher at 168 h p.i. (##FIG##2##Figure 3C##). This can be explained by taking the basic principles of EPR and the prolonged circulation time of the larger copolymer into account (##FIG##2##Figure 3B##), and is exemplified by the fact that levels in tumour and spleen substantially increased and levels in healthy tissues substantially decreased over time (##FIG##2##Figure 3C##). The tumour-to-organ ratios in ##FIG##2##Figure 3D## confirm this observation, showing both higher overall values and larger increases over time for the 65 kDa copolymer, and they furthermore illustrate that HPMA copolymers localise to tumours relatively selectively, with throughout follow-up, always higher levels in tumours than in seven out of nine healthy tissues. Using HPLC, it was finally demonstrated that by means of their beneficial biodistributional properties, HPMA copolymers are able to improve the tumour-directed delivery of doxorubicin, increasing its target site accumulation at 24 h p.i. by more than a three-fold (22.1 <italic>vs</italic> 6.2 <italic>μ</italic>g doxorubicin per gram tumour; <xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Figure 2</xref>). Together, these findings show that HPMA copolymers are able to improve the temporal and spatial parameters of low molecular weight agents, and they thereby exemplify that HPMA copolymers are suitable targeting systems for assessing the validity and the therapeutic potential of carrier-based radiochemotherapy.</p>", "<title>Radiotherapy improves drug targeting</title>", "<p>To address the former aspect of carrier-based radiochemotherapy, that is, to evaluate whether radiotherapy is able to improve the tumour accumulation of drug targeting systems (##FIG##0##Figure 1B##), we next visualised and quantified the concentrations of the iodine- and gadolinium-labelled copolymers in tumours that were exposed to 20 Gy of radiotherapy 24 h before i.v. injection. The scintigrams in ##FIG##3##Figure 4A## show that as hypothesised, ionising radiation indeed significantly improved the tumour accumulation of the carrier systems. Quantifications at 24 and 168 h p.i. confirmed this notion, showing that for the radiolabeled 31 kDa copolymer, increases of 24 and 57% were found (##FIG##3##Figure 4B##), and for the 65 kDa copolymer, increases of 46 and 48%, respectively (##FIG##3##Figure 4C##). Using the abovementioned 25 kDa gadolinium-containing copolymer and a 1.5 T clinical MR scanner, we subsequently also quantified the tumour accumulation of the targeting system at several earlier time points, that is, between 0.5 and 24 h after i.v. administration. In this case, upon quantifying the T1 signal enhancements in irradiated <italic>vs</italic> control tumours, increases ranging from 31 to 44% were observed (##FIG##3##Figures 4C and D##). Even though such increases may intuitively seem to be modest, in line with previous findings, using different doses, tumour models, and carrier systems (##REF##10914731##Li et al, 2000##; ##UREF##0##Davies et al, 2004##; ##REF##17215057##Lammers et al, 2007##), they do convincingly demonstrate that the tumour accumulation of drug targeting systems can be improved by combining them with radiotherapy. This in contrast, for instance, to the majority of targeting ligands that have been evaluated for this purpose over the years, but that generally only tend to improve the internalisation of the systems (##UREF##4##Park et al, 2004##; ##REF##16818648##Kirpotin et al, 2006##).</p>", "<title>Drug targeting improves doxorubicin-based radiochemotherapy</title>", "<p>To address the latter (and clinically more relevant) aspect of carrier-based radiochemotherapy, that is, to investigate whether drug targeting systems are able to improve the interaction between radiotherapy and chemotherapy (##FIG##0##Figure 1D##), two different versions of HPMA copolymer-bound doxorubicin were synthesised, that is, poly(HPMA)-GFLG-Dox (PK1; 28 kDa) and human immunoglobulin G-modified poly(HPMA)-GFLG-Dox (IgG-PK1; 900 kDa). Both polymeric prodrugs have been tested in clinical trials, and both have been shown to be at least equally effective as free doxorubicin (##REF##10840193##Kopecek et al, 2000##; ##REF##12870445##Bilim, 2003##; ##REF##14529421##Rihova and Kubackova, 2003a##; ##REF##12932633##Rihova et al, 2003b##; ##REF##16900224##Duncan, 2006##). As the MTD of PK1 is known to be 4–5 times higher than that of free doxorubicin (both in humans and in rodents (##REF##1760851##Yeung et al, 1991##; ##REF##9918206##Vasey et al, 1999##)), in an attempt to simultaneously increase the efficacy and reduce the toxicity of the intervention, we chose to apply PK1 at three times 5 mg kg⁻¹ (days 1, 8, and 15), that is, at twice the dose used for free doxorubicin, which has a cumulative MTD of 7.5 mg kg⁻¹ in rats. IgG-PK1 was applied both at the 2.5 and at the 5 mg⁻¹kg regimen. As shown in ##FIG##4##Figure 5A##, in the aggressively growing and radio- and chemoresistant Dunning AT1 model, all three doxorubicin formulations were found to be significantly more effective than control, but the improvements were rather modest, and neither PK1 nor IgG-PK1 turned out to be better than the free drug. When combined with 20 daily doses of 2 Gy of local tumour radiotherapy, however, in line with our rationale (##FIG##0##Figure 1##), the two polymeric prodrugs did present with a significantly higher therapeutic index than did the free drug: PK1 applied at three 5 mg kg⁻¹ doses and IgG-PK1 applied at three 2.5 mg kg⁻¹ doses were both found to be substantially more effective than three 2.5 mg kg⁻¹ doses of free doxorubicin (##FIG##4##Figures 5B and C##), and both were also significantly less toxic (##FIG##4##Figure 5D##). For the former two regimens, a dose-enhancement factor (DEF; i.e., a routinely used parameter for assessing the radiosensitising potential of chemotherapeutic agents) of 1.50 was found, as compared to a DEF of only 1.28 for the free drug (<xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Table 1</xref>). With a DEF of 1.82, IgG-PK1 applied at three 5 mg kg⁻¹ doses was found to be by far the most effective regimen for improving the efficacy of fractionated radiotherapy (##FIG##4##Figures 5B and C##). It was, however, also the most toxic treatment (##FIG##4##Figure 5D##), which underlines the importance of determining an optimal dosing regimen for each individual agent. As it was found that both for PK1 (3 × 5 mg kg⁻¹) and for IgG-PK1 (3 × 2.5 mg kg⁻¹), the efficacy of the intervention was increased while its toxicity was reduced, these findings were considered to be an initial indication for the fact that drug targeting systems are able to improve the therapeutic index of radiochemotherapy.</p>", "<title>Preparation of gemcitabine-containing HPMA copolymers</title>", "<p>To provide additional evidence for the validity of carrier-based radiochemotherapy, we subsequently synthesised pHPMA-Gemcitabine. Gemcitabine is a well-known radiosensitiser (##UREF##2##Lawrence et al, 1997##), and it is used in the first-line treatment of various advanced solid malignancies. Two different gemcitabine-containing copolymers were prepared, termed A-Gem and B-Gem (##FIG##5##Figures 6A and B##). In A-Gem, the drug is conjugated to the copolymer by means of the uncleavable aminohexanoic acid spacer. In B-Gem, gemcitabine is attached to the polymeric backbone by means of the -GFLG- spacer, which is also used in PK1 and which is known to be cleaved by the lysosomal cysteine protease cathepsin B. ##FIG##5##Figure 6C## shows that A-Gem does not release the drug <italic>in vitro</italic>: independent of the conditions used, less than 1% of the agent was liberated upon 8 h of incubation. B-Gem, on the other hand, was quite effective in releasing gemcitabine: at pH=6, ∼10% was released after 8 h, at pH=7.4, ∼30% was released, and upon incubation with physiologically relevant concentrations of cathepsin B (at pH=6; resembling endo- and lysosomal conditions), the total amount of drug conjugated to the copolymer was released within less than 6 h (##FIG##5##Figure 6D##). To demonstrate that the extent of drug release correlates with the cytotoxicity of the conjugates, clonogenic survival assays were performed. As shown in ##FIG##5##Figures 6E and F##, as expected, both in Dunning AT1 rat prostate carcinoma cells and in A2780 human ovarian carcinoma cells, B-Gem was significantly more effective than A-Gem. In addition, also in line with our expectations, the two carrier-based agents were found to be less effective <italic>in vitro</italic> than the free agent. For a drug-free control copolymer, no cytotoxicity was observed.</p>", "<title>Drug targeting improves gemcitabine-based radiochemotherapy</title>", "<p>In the final set of experiments, we set out to investigate the <italic>in vivo</italic> potential of HPMA copolymer-bound gemcitabine. As shown in ##FIG##6##Figure 7A##, it was again found that without radiotherapy, neither the free drug nor its polymeric prodrugs were able to induce substantial growth inhibition in the therapy-resistant Dunning AT1 model: B-Gem applied at four 3 mg kg⁻¹ doses appeared to be the only regimen that was significantly more effective than control, but it was not more effective than free gemcitabine (<xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Table 2</xref>). In line with our rationale (##FIG##0##Figure 1##), however, upon again combining the agents with a clinically relevant regimen of fractionated radiotherapy (12 × 3 Gy), it could again be observed that the targeted formulation was significantly more effective than the free drug (##FIG##6##Figures 7B and C##): for B-Gem, a DEF of 2.79 was found, as compared to a DEF of ‘only’ 2.14 for free gemcitabine (<xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Table 2</xref>). ##FIG##6##Figure 7D## and <xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Figure 3</xref> finally show that the combination of B-Gem with fractionated radiotherapy was equally well tolerated as the combination of free gemcitabine with fractionated RT, with both agents inducing an identical degree of weight loss and of bone marrow suppression. In line with the results obtained for doxorubicin (##FIG##4##Figure 5##), these notions exemplify that drug targeting systems are able to increase the efficacy of radiochemotherapy without increasing its toxicity.</p>" ]

[ "<title>Discussion</title>", "<p>Over the years, a variety of different drug targeting systems have been developed, ranging in nature from simple polymers (##REF##16900224##Duncan, 2006##) and liposomes (##REF##15688077##Torchilin, 2005##), to stimuli-sensitive polymeric micelles (##REF##17623798##Rapoport et al, 2007##), bacterially derived Minicells (##REF##17482133##MacDiarmid et al, 2007##), and temporally targeted Nanocells (##REF##16049491##Sengupta et al, 2005##). Thus far, however, not very many have managed to reach the final stages of clinical evaluation, and only about a handful have been approved by the responsible regulatory authorities. Consequently, hardly any information is available on the combination of tumour-targeted nanomedicines with other well-established treatment modalities. The potential of combining them with a clinically relevant regimen of fractionated radiotherapy, for instance, has not yet been properly evaluated, even though there is an obvious rationale for doing so (##FIG##0##Figure 1##).</p>", "<p>We have here used the prototypic polymeric drug carrier pHPMA to validate the potential of this targeted combination regimen. In line with the literature (##REF##10840193##Kopecek et al, 2000##), HPMA copolymers were hereto first shown to be versatile and multifunctional drug carriers, that can be easily tracked <italic>in vivo</italic>, that circulate for prolonged periods of time, that localise to tumours both effectively and selectively, and that improve the tumour-targeted delivery of low molecular weight agents (##FIG##1##Figures 2## and ##FIG##2##3##). Subsequently, they were then shown to interact synergistically with radiotherapy, with radiotherapy increasing the tumour accumulation of the copolymers (##FIG##3##Figure 4##), and with the copolymers increasing the therapeutic index of radiochemotherapy (##FIG##4##Figures 5## and ##FIG##6##7##). Improvements were observed in a rapidly growing and therapy-resistant tumour model, and both for doxorubicin and for gemcitabine, together indicating that ‘carrier-based radiochemotherapy’ is indeed a promising approach for improving the efficacy of combined modality anticancer therapy.</p>", "<p>This notion is in line with the results of a recently published phase I trial, in which 12 patients with localised oesophageal and gastric cancer were treated with the combination of poly(<sc>L</sc>-glutamic acid) (PGA)-bound paclitaxel (Xyotax; 6 doses; weekly) and fractionated radiotherapy (28 cycles; 1.8 Gy; daily), and in which four complete responses and an additional seven partial responses (with reductions in tumour size of more than 50%) were achieved (##REF##16891865##Dipetrillo et al, 2006##). Prior to this trial, preclinical studies had already identified Xyotax as a highly potent radiosensitiser. When a single i.v. injection of Xyotax was combined with a single dose of radiotherapy, for instance, the dose required to produce 50% tumour cure (TCD₅₀) could be reduced substantially, from 53.9 to 7.5 Gy (##REF##12573758##Milas et al, 2003##). When radiotherapy was delivered as five daily fractions, the effect of Xyotax was even more pronounced reducing the TCD₅₀ from 66.6 to 7.9 Gy. In a follow-up study in the same tumour model (i.e., C3Hf/KamLaw mice bearing ∼7 mm syngeneic OCa-1 ovarian adenocarcinomas), similar results were reported for Abraxane, that is, for albumin-based paclitaxel, which also beneficially combined both with single dose and with fractionated radiotherapy, and which did not increase normal tissue radiotoxicity (##REF##17363543##Wiedenmann et al, 2007##).</p>", "<p>In comparable analyses, ##REF##11156255##Harrington et al (2000##) have demonstrated that also liposomes hold significant potential for combination with radiotherapy. They combined both PEGylated liposomal doxorubicin and PEGylated liposomal cisplatin with both single dose (4.5 and 9 Gy) and fractionated (3 × 3 Gy) radiotherapy, and they showed that animals treated with carrier-based radiochemotherapy survived for significantly longer periods of time than did animals treated with standard radiochemotherapy. ##UREF##0##Davies et al (2004)## recently confirmed and extended these findings, showing that PEGylated liposomal doxorubicin (Caelyx; Doxil) significantly improves the efficacy of both single-dose (8 Gy) and fractionated (3 × 3.6 Gy) radiotherapy, and that it does so, at least in part, by improving the penetration and the intratumoral distribution of the agent. In their studies on polymeric radiosensitisers, ##REF##10914731##Li et al (2000##) also observed that radiotherapy increases the tumour accumulation of passively targeted nanomedicines, attributing at least part of the supra-additively improved efficacy of PGA-paclitaxel (Xyotax) and radiotherapy to a radiotherapy-induced increase in tumour localisation. It is interesting to note in this regard that the overall improvement in the tumour concentration of PGA-paclitaxel was virtually identical to that observed here for HPMA copolymers, with as compared to sham-irradiated controls, increases ranging from 25 to 50%. Also in line with our findings, Li <italic>et al</italic> demonstrated that this radiotherapy-induced increase in tumour accumulation could already be observed almost immediately upon i.v. injection (i.e., at 1 h p.i. <italic>vs</italic> at 30 min p.i. here), and remains relatively constant over time. Together with Davies <italic>et al</italic>'s findings on the enhanced penetration and the improved intratumoral distribution of PEGylated liposomal doxorubicin in response to radiotherapy, these observations suggest that radiotherapy beneficially affects both arms of the EPR effect: on the one hand, by e.g. enhancing the expression of VEGF (##REF##10914731##Li et al, 2000##; ##UREF##3##Park et al, 2001##), it likely increases the permeability of the tumour blood vessels towards long-circulating nanomedicines, and on the other hand, by e.g. reducing the tumour cell density (##REF##10212807##Peschke et al, 1999##) and the interstitial fluid pressure (##REF##8640786##Znati et al, 1996##), it likely enhances their retention and their intratumoral distribution. Although clinically clearly less important than the improved therapeutic indices resulting from carrier-based radiochemotherapy, these improvements in EPR-mediated drug targeting to tumours provide an additional rationale for combining tumour-targeted nanomedicines with radiotherapy.</p>", "<p>In summary, using prototypic polymeric drug carriers, two different imaging techniques and two different chemotherapeutic agents, we here demonstrate that drug targeting and radiotherapy interact synergistically, with radiotherapy increasing the tumour accumulation of drug targeting systems, and with drug targeting systems increasing the therapeutic index of radiochemotherapy. We extend previous efforts by attempting to generalise the concept of ‘carrier-based radiochemotherapy’, by implementing clinically relevant regimens of radio- and chemotherapy, and by directly comparing the radiosensitising potential of polymeric prodrugs to that of free chemotherapeutic agents. The results presented and the insights obtained strongly suggest that carrier-based radiochemotherapy holds significant potential for improving the treatment of advanced solid malignancies.</p>" ]

[]

[ "<p>These authors contributed equally to this study.</p>", "<p>Drug targeting systems are nanometer-sized carrier materials designed for improving the biodistribution of systemically applied (chemo-) therapeutics. Reasoning that (I) the temporal and spatial interaction between systemically applied chemotherapy and clinically relevant fractionated radiotherapy is suboptimal, and that (II) drug targeting systems are able to improve the temporal and spatial parameters of this interaction, we have here set out to evaluate the potential of ‘carrier-based radiochemotherapy’. <italic>N</italic>-(2-hydroxypropyl)methacrylamide (HPMA) copolymers were used as a model drug targeting system, doxorubicin and gemcitabine as model drugs, and the syngeneic and radio- and chemoresistant Dunning AT1 rat prostate carcinoma as a model tumour model. Using magnetic resonance imaging and <italic>γ</italic>-scintigraphy, the polymeric drug carriers were first shown to circulate for prolonged periods of time, to localise to tumours both effectively and selectively, and to improve the tumour-directed delivery of low molecular weight agents. Subsequently, they were then shown to interact synergistically with radiotherapy, with radiotherapy increasing the tumour accumulation of the copolymers, and with the copolymers increasing the therapeutic index of radiochemotherapy (both for doxorubicin and for gemcitabine). Based on these findings, and on the fact that its principles are likely broadly applicable, we propose carrier-based radiochemotherapy as a novel concept for treating advanced solid malignancies.</p>" ]

[ "<p>The combination of radiotherapy and chemotherapy has been evaluated extensively in the past few decades (##REF##2185342##Vokes and Weichselbaum, 1990##; ##REF##17259930##Seiwert et al, 2007##). Besides significantly increasing the efficacy of radiotherapy, however, the simultaneous application of chemotherapy also substantially increases its toxicity (##REF##14585258##Maduro et al, 2003##; ##UREF##5##Rowell and O’rourke, 2004##). As external beam radiotherapy can nowadays be delivered with extremely high levels of spatial specificity (##REF##1869275##Fuks et al, 1991##; ##REF##15343280##Bernier et al, 2004##), this is likely mostly due to the low degree of spatial specificity that chemotherapeutic agents generally present upon intravenous (i.v.) administration. Based on this notion, and on the fact that drug targeting systems are known to be able to improve both the temporal (circulation time, tumour residence time) and the spatial (tumour accumulation, tumour-to-organ ratio) parameters of drug therapy (##REF##14667500##Moses et al, 2003##; ##REF##15688077##Torchilin, 2005##; ##REF##16900224##Duncan, 2006##; ##REF##18648371##Lammers et al, 2008##), we reasoned that the implementation of a drug targeting system might be able to increase the therapeutic index of radiochemotherapy. The rationale for this novel combination regimen, which we have termed ‘carrier-based radiochemotherapy’, relies on the notion that on the one hand, radiotherapy improves drug targeting (i.e. the tumour accumulation of drug targeting systems), and that on the other hand, drug targeting improves radiochemotherapy (i.e., the temporal and spatial interaction between daily radiotherapy and weekly chemotherapy; see ##FIG##0##Figure 1##).</p>", "<p>Concerning the former aspect of carrier-based radiochemotherapy, several different mechanisms can be envisioned by which radiotherapy increases the tumour accumulation of drug targeting systems (##FIG##0##Figure 1B##). Besides reflecting, for instance, on the integrity and function of the tumour vasculature (V), and on the expression of certain cellular receptors (R), it is also known to affect several cell membrane-related (C), nuclear (N), mitochondrial (M) and signalling (S) processes. By eliciting such effects, radiotherapy has been shown to induce (I) an increase in the production of vascular endothelial growth factor (VEGF) (##UREF##3##Park et al, 2001##) and fibroblast growth factor (FGF) (##REF##7499402##Lee et al, 1995##), (II) an increase in apoptosis and endothelial cell apoptosis (##REF##12750523##Garcia-Barros et al, 2003##), (III) a decrease in tumour cell density (##REF##10212807##Peschke et al, 1999##), and (IV) a reduction in interstitial fluid pressure (##REF##8640786##Znati et al, 1996##). By means of the former phenomenon, radiotherapy is considered to be able to increase the permeability of the vasculature towards long-circulating nanomedicines (##REF##9626048##Samaniego et al, 1998##; ##REF##10100187##Feng et al, 1999##), and by means of the latter three, it likely improves their penetration and their intratumoral distribution (##UREF##1##Jain, 1994##; ##REF##7585615##Netti et al, 1995##; ##REF##11489481##Au et al, 2001##).</p>", "<p>The rationale for the latter aspect of carrier-based radiochemotherapy, that is, for the assumption that drug targeting systems are able to improve the interaction between radiotherapy and chemotherapy, is depicted schematically in ##FIG##0##Figure 1C##. Drug targeting systems are generally designed to be stable in circulation, and to release the conjugated or entrapped active agent only at the target site (##REF##15688077##Torchilin, 2005##; ##REF##16900224##Duncan, 2006##; ##REF##18648371##Lammers et al, 2008##). As a result, as compared to an i.v. applied free drug, the peak plasma concentration of a tumour-targeted agent tends to be reduced, and the degree of systemic toxicity can often be attenuated (##FIG##0##Figure 1C##; upper two panels). For doxorubicin, for instance, both polymeric (##REF##1760851##Yeung et al, 1991##; ##REF##16900224##Duncan, 2006##) and liposomal (##REF##15717742##Ewer et al, 2004##; ##REF##15688077##Torchilin, 2005##) drug targeting systems have been shown to be able to reduce the incidence of cardiomyopathy. At the same time, by increasing the concentration of the active agent at the target site, drug targeting systems also tend to be able to improve the efficacy of the drug (##REF##14667500##Moses et al, 2003##; ##REF##15688077##Torchilin, 2005##; ##REF##16900224##Duncan, 2006##; ##REF##18648371##Lammers et al, 2008##). As a matter of fact, as depicted schematically in the lower two panels in ##FIG##0##Figure 1C##, the implementation of a drug targeting system generally not only increases the concentration of the active agent at the target site, but it also improves its availability over time (i.e., its area under the curve, as a result of the enhanced permeability and retention (EPR) effect (##REF##10699287##Maeda et al, 2000##)). When combining a clinically relevant regimen of (weekly) chemotherapy with a clinically relevant regimen of (daily) radiotherapy, it can therefore be expected that a drug delivered to the tumour by means of a drug targeting system interacts more effectively with radiotherapy than does a free, untargeted drug (##FIG##0##Figure 1D##). As will be outlined below, this was indeed found to be the case, and both for doxorubicin and for gemcitabine, it could be demonstrated that drug targeting systems increase the efficacy of radiochemotherapy without increasing its toxicity.</p>" ]

[ "<p>This study was supported by the German-Israeli Cooperation Program in Cancer Research (CA-105; TL, PP, JD, PEH), the Wieland-Stiftung (TL), the Program of Research Centers of the Ministry of Education, Youth and Sports of the Czech Republic (IM 4635608802 (VS, KU), and the European Commission (FP6-MediTrans; TL, WEH, GS). Blanka Rihova, Tomas Etrych, and Helena Misorcova are kindly acknowledged for the HPLC analyses, Werner Rittgen and Lutz Edler for the statistical analyses, Jochen Schumacher for radiolabeling the copolymers, and Ditmar Greulich for the <italic>in vitro</italic> analyses. We declare no competing financial interests.</p>", "<title>Supplementary Material</title>" ]

[ "<fig id=\"fig1\"><label>Figure 1</label><caption><p>Rationale for carrier-based radiochemotherapy. (<bold>A</bold>) Carrier-based radiochemotherapy is based on the notion that drug targeting and radiotherapy interact synergistically, with on the one hand, radiotherapy improving the tumour accumulation of drug targeting systems, and with on the other hand, drug targeting systems improving the therapeutic index of radiochemotherapy. (<bold>B</bold>) Potential physiological mechanisms by which radiotherapy increases the tumor accumulation of drug targeting systems. See text for details. (<bold>C</bold>) Schematic representation of the blood and tumour concentrations of an intravenously (i.v.) applied free drug (upper and lower left panels), and of a drug delivered to the tumour by means of an i.v. applied drug targeting system (upper and lower right panels). The arrows indicate the administration of fractionated radiotherapy, which is routinely applied on every weekday for several consecutive weeks. TT: toxicity threshold, AT: activity threshold. See text for details. (<bold>D</bold>) Schematic representation of the <italic>in vivo</italic> interaction between radiotherapy and chemotherapy upon standard and upon carrier-based radiochemotherapy, exemplifying that in case of the latter, the temporal and spatial interaction between the two treatment modalities is improved.</p></caption></fig>", "<fig id=\"fig2\"><label>Figure 2</label><caption><p>Magnetic resonance imaging (MRI)-based biodistributional analysis of pHPMA-gadolinium (Gd). (<bold>A</bold>) MR angiography scans of the chest and head region of a rat, of a tumour-bearing paw, and of an AT1 tumour, obtained at 0.5 h after the intravenous (i.v.) injection of 25 kDa pHPMA-Gd. A color-coded maximal intensity projection (MIP) of the polymer-visualised perfusion of the tumour is depicted in the right panel. (<bold>B</bold>) Dynamic, color-coded MRI T1 determination obtained for AT1 tumour, for kidney and for liver before contrast agent administration and at various time points after the i.v. injection of a low (0.5 kDa Gd-DTPA-BMA) and high (25 kDa Gd-pHPMA) molecular weight MR contrast agent. Note that contrast agent accumulation corresponds to a decrease in the T1 signal. (<bold>C</bold>) Quantification of the concentrations of gadolinium in AT1 tumour, kidney, and liver upon the i.v. injection of Gd-DTPA-BMA and pHPMA-Gd. Values represent average±s.d. (<italic>n</italic>=3). ^* Indicates <italic>P</italic>&lt;0.05 (paired <italic>t</italic>-test).</p></caption></fig>", "<fig id=\"fig3\"><label>Figure 3</label><caption><p>HPMA copolymers localise to tumours both effectively and selectively. (<bold>A</bold>) Scintigraphic analysis of the biodistribution of two differently sized iodine-131-labelled HPMA copolymers in Copenhagen rats bearing subcutaneously transplanted Dunning AT1 tumours, demonstrating prolonged circulation and effective tumour accumulation (H: heart (blood), B: bladder, S: spleen, L: liver, T: tumour). (<bold>B</bold>) Analysis of the blood concentrations of the two radiolabeled copolymers. Values represent average±s.d. (<italic>n</italic>=6). (<bold>C</bold>) Quantification of the tumour and organ concentrations of the two radiolabeled copolymers at 24 and 168 h post intravenous injection. Values represent average±s.d. (<italic>n</italic>=6). Except for lung and spleen, concentrations in tumours were always significantly higher than those in healthy organs (<italic>P</italic>&lt;0.05; two-tailed <italic>t</italic>-test). (<bold>D</bold>) Quantification of the tumour-to-organ ratios of the copolymers analysed in (<bold>C</bold>), pointing out (in green) that they accumulate more selectively in tumours than in seven out of nine healthy tissues.</p></caption></fig>", "<fig id=\"fig4\"><label>Figure 4</label><caption><p>Radiotherapy (RT) improves drug targeting. (<bold>A</bold>) Scintigraphic analysis of the effect of 20 Gy of local tumour RT on the tumour accumulation of an iodine-131-labelled 31 kDa HPMA copolymer, demonstrating that RT beneficially affects tumour targeting. (<bold>B</bold>) Quantification of the effect of RT on the tumour concentrations of the 31 kDa (left panel) and 65 kDa (right panel) copolymer at 24 and 168 h post intravenous injection. Values represent average±s.d. (<italic>n</italic>=3–6). ^* Indicates <italic>P</italic>&lt;0.05 (two-tailed <italic>t</italic>-test). (<bold>C</bold>) Magnetic resonance imaging analysis of the effect of 20 Gy of RT on the tumour accumulation of the 25 kDa gadolinium-labelled HPMA copolymer. The T1 images correspond to contrast agent accumulation, the T2 images were used for positioning and for morphological analysis. (<bold>D</bold>) Quantification of the effect of RT on the tumour accumulation of the 25 kDa gadolinium-labelled copolymer. Values represent average±s.d. (<italic>n</italic>=3). ^*Indicates <italic>P</italic>&lt;0.05 (paired <italic>t</italic>-test).</p></caption></fig>", "<fig id=\"fig5\"><label>Figure 5</label><caption><p>Drug targeting improves doxorubicin (Dox)-based radiochemotherapy. (<bold>A</bold>) Growth inhibition of Dunning AT1 tumours induced by three intravenous (i.v.) injections (days 1, 8, and 15; see vertical arrows) of saline, of free doxorubicin and HPMA copolymer-bound doxorubicin. PK1: pHPMA-GFLG-Dox (28 kDa). IgG-PK1: human IgG-modified pHPMA-GFLG-Dox (900 kDa). Values represent average ± s.e.m (<italic>n</italic>=6–12). ^*Indicates <italic>P</italic>&lt;0.05 <italic>vs</italic> control (Mann–Whitney <italic>U</italic>-test; Bonferroni–Holm <italic>post hoc</italic> analysis). (<bold>B</bold>) Tumour growth inhibition induced by three i.v. injections of the abovementioned chemotherapeutic agents in combination with a clinically relevant regimen of fractionated radiotherapy (20 × 2 Gy; see vertical lines). Values represent average±s.e.m. (<italic>n</italic>=8–10). ^* Indicates <italic>P</italic>&lt;0.05 <italic>vs</italic> control, ^# indicates <italic>P</italic>&lt;0.05 <italic>vs</italic> free Dox, and ^† indicates <italic>P</italic>&lt;0.005 <italic>vs</italic> free Dox (Mann–Whitney <italic>U</italic>-test; Bonferroni–Holm <italic>post hoc</italic> analysis). (<bold>C</bold>) Representative images (day 50) of tumours treated with the indicated combination regimens. (<bold>D</bold>) Weight loss induced by doxorubicin-based combined modality therapy. Values represent average±s.e.m. (<italic>n</italic>=4–5).</p></caption></fig>", "<fig id=\"fig6\"><label>Figure 6</label><caption><p>Characterisation of the gemcitabine (Gem)-containing HPMA copolymers. (<bold>A</bold>) and (<bold>B</bold>) Chemical structure of A-Gem (poly(HPMA)-AH-Gem) and B-Gem (poly(HPMA)-GFLG-Gem). (<bold>C</bold> and <bold>D</bold>) Release of gemcitabine from A-Gem and B-Gem at pH=7.4, at pH=6.0, and at pH=6.0 in the presence of the lysosomal cysteine protease cathepsin B (CB). Values are expressed relative to the total amount of drug conjugated to the copolymers, and they represent average±s.d. of three independent experiments. (<bold>E</bold> and <bold>F</bold>) Cytotoxicity of free gemcitabine, A-Gem, B-Gem and a drug-free control copolymer. Colony formation assays were performed using Dunning AT1 rat prostate carcinoma cells (<bold>E</bold>) and A2780 human ovarian carcinoma cells (<bold>F</bold>). Values represent average±s.d. (<italic>n</italic>=3).</p></caption></fig>", "<fig id=\"fig7\"><label>Figure 7</label><caption><p>Drug targeting improves gemcitabine (Gem)-based radiochemotherapy. (<bold>A</bold>) Growth inhibition of Dunning AT1 tumours induced by four intravenous (i.v.) injections (days 1, 8, 15, and 22; see vertical arrows) of saline, free gemcitabine and HPMA copolymer-bound gemcitabine. A-Gem: pHPMA-AH-Gem (20 kDa). B-Gem: pHPMA-GFLG-Gem (24 kDa). ^* Indicates <italic>P</italic>&lt;0.05 <italic>vs</italic> control (Mann–Whitney <italic>U</italic>-test; Bonferroni–Holm <italic>post hoc</italic> analysis). (<bold>B</bold>) Tumour growth inhibition induced by four i.v. injections of the abovementioned chemotherapeutic agents in combination with a clinically relevant regimen of fractionated radiotherapy (12 × 3 Gy; see vertical lines). Values represent average±s.e.m. (<italic>n</italic>=10–12). ^* Indicates <italic>P</italic>&lt;0.005 <italic>vs</italic> control, # indicates <italic>P</italic>&lt;0.0005 <italic>vs</italic> control, and ^† indicates <italic>P</italic>&lt;0.05 <italic>vs</italic> free Gem (Mann–Whitney <italic>U</italic>-test; Bonferroni–Holm <italic>post hoc</italic> analysis). (<bold>C</bold>) Representative images (day 45) of tumours treated with the indicated combination regimens. (<bold>D</bold>) Weight loss induced by gemcitabine-based combined modality therapy. Values represent average±s.e.m. (<italic>n</italic>=5–6).</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"xob1\"><label>Supplementary Information</label></supplementary-material>" ]

[ "<fn-group><fn><p><xref ref-type=\"supplementary-material\" rid=\"sup1\">Supplementary Information</xref> accompanies the paper on British Journal of Cancer website (<ext-link xmlns:xlink=\"http://www.w3.org/1999/xlink\" ext-link-type=\"uri\" xlink:href=\"http://www.nature.com/bjc\">http://www.nature.com/bjc</ext-link>)</p></fn></fn-group>" ]

[ "<graphic xlink:href=\"6604561f1\"/>", "<graphic xlink:href=\"6604561f2\"/>", "<graphic xlink:href=\"6604561f3\"/>", "<graphic xlink:href=\"6604561f4\"/>", "<graphic xlink:href=\"6604561f5\"/>", "<graphic xlink:href=\"6604561f6\"/>", "<graphic xlink:href=\"6604561f7\"/>" ]

[ "<media xlink:href=\"6604561x1.doc\"><caption><p>Click here for additional data file.</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[]

[ "<title>Materials and methods</title>", "<title>Study population</title>", "<p>Participants were disease-free rectal cancer patients who had participated in a study assessing rectal cancer treatment preferences (##REF##17848910##Pieterse et al, 2007##). For that study, a stratified random sample was selected from patients who had participated in a multicenter trial assessing the benefit of adding preoperative radiotherapy (PRT) to total mesorectal excision surgery between January 1996 and December 1999 (##REF##11547717##Kapiteijn et al, 2001##). Stratification was carried out to include equal numbers of patients from both treatment groups as well as those who were reported to suffer from side effects at follow-up in both groups. A total of 94 eligible patients were approached. Of these, four patients could not be reached and nine had other types of cancer or recurrent disease and were excluded. Of the remaining 81 patients, 70 (86%) agreed to participate. Reasons for refusal were the psychological burden (<italic>N</italic>=8), physical burden (<italic>N</italic>=1), time investment (<italic>N</italic>=1), or unknown (<italic>N</italic>=1).</p>", "<p>For the treatment preference study, we aimed further to include 60 oncologists specialised in gastroenterology. We predominantly aimed for surgical and radiation oncologists, as these specialties are most involved in primary rectal cancer treatment. Seventy eligible oncologists were randomly selected from surgeons, radiotherapists, and medical oncologists involved in the Simply Capecitabine in Rectal cancer after Irradiation Plus TME (SCRIPT) trial and were approached. Three clinicians could not be reached. Sixty (86%) oncologists agreed to participate (25 surgical, 25 radiation, and 10 medical). Reasons for refusal were time constraints (<italic>N</italic>=4), considering participation not meaningful (<italic>N</italic>=1 medical oncologist), being retired (<italic>N</italic>=1), or unknown (<italic>N</italic>=1).</p>", "<title>Procedure</title>", "<p>Eligible participants were informed about the study by letter and then asked by phone whether they agreed to participate. One of two trained interviewers (AHP and MCMB-T) conducted individual face-to-face interviews following a strict protocol. Interviews were held at home (patients) or at the institution (oncologists). All patients gave written informed consent at the start of the interview. Interviews with clinicians were audiotaped with their permission. Prior to the interview, sociodemographic, disease- (patients) and work- (oncologists) related data were collected using a self-administered paper-and-pencil questionnaire. The medical ethical board of the Leiden University Medical Centre approved the study. Participants were included between February and August 2006.</p>", "<title>Interviews</title>", "<p>We started the interview by assessing role preferences in treatment decision-making using the Control Preferences Scale (CPS) (##REF##1432023##Degner and Sloan, 1992##). The CPS was devised to assess the degree of control an individual wants to assume when decisions are being made about a specific medical treatment (##UREF##0##Degner et al, 1997b##). It has been widely used in previous studies of cancer patients (##REF##9301440##Rothenbacher et al, 1997##; ##REF##11041108##Wallberg et al, 2000##; ##REF##16041829##Mallinger et al, 2006##). We started out with this measure to help participants think about participation in decision making. Patients especially may never have thought about such issues. We asked participants to focus on the hypothetical decision whether or not to undergo PRT, had they presently been diagnosed with rectal cancer (patients), or were they seeing a patient diagnosed with rectal cancer (oncologists). The decision was not familiar to patients, as they had been randomised to receive either PRT or not at the time they were actually treated. Notably, we had not informed patients about benefits and side effects of PRT.</p>", "<p>The CPS consists of five cards, each of which portrays a different role in treatment decision making using a statement. The roles ranged from (A) the patient deciding on PRT, through (C) the individual making the decision jointly with the physician (or patient), to (E) the physician making the decision about PRT (##FIG##0##Figure 1##). Participants were presented with a series of paired comparisons of roles until the preferred role was established, following the same procedure as ##REF##14990371##Davey et al (2004)##. The statements thus indicate a preference for an active (A and B), collaborative (C), or passive (D and E) patient role in treatment decision making.</p>", "<p>Participants were then asked a set of closed questions relating to the desirability of patient participation and to the weighing of pros and cons of treatments (##TAB##0##Table 1##). Clinicians were asked whether they would offer the option of an adjuvant treatment that cures an additional 1–5% patients, but with a clinically relevant risk of side effects in patients treated with that adjuvant treatment (the absolute numbers and the nature of the treatment and side effects were not further specified). The example was chosen as a general case for current adjuvant treatments in cancer care. Clinicians were further questioned about their preference regarding the role of patients in the formulation of treatment guidelines. Participants were asked to explain their answer following each question. The interviewers noted participants' answers on paper.</p>", "<title>Coding</title>", "<p>A random selection of five taped clinician interviews was used to compare the recorded explanations with the paper notes taken during the interview. This showed that main themes had not been missed on paper.</p>", "<p>Answers to the open-ended questions were categorised. Each of the interviewers read a different random selection of five patient interviews and separately developed an initial list of codes intended to reflect the various views in participants' responses. The interviewers compared their list of codes and decided on a definitive set of codes and subheadings. In coding oncologists' answers, additional codes were decided on as needed. Both interviewers then coded all participants' explanations, reviewed their coding, and resolved disparities. Participants' answers could be categorised into one or more categories (see <xref ref-type=\"app\" rid=\"app1\">Appendix A</xref>), depending on the number of reasons they nominated. Answers were not categorised if they reiterated a previous ‘yes’ or ‘no’ answer to the closed-ended question, if it was open to interpretation, or if it did not refer to the question but went beyond.</p>", "<title>Analysis</title>", "<p>Descriptive statistics (frequencies and percentages) were used to describe sociodemographic, disease- (patients) and work- (oncologists) related data, role preferences, and the answers to the closed- and open-ended questions. Using Fisher's exact or <italic>χ</italic>² tests, as appropriate, proportions of patients and clinicians were compared on their answers to the closed-ended questions and on how often their explanation fell into one of the coding categories. Bivariate associations between participants' characteristics and role preferences (active, collaborative, and passive patient role) and responses to question 1 (##TAB##0##Table 1##) were assessed using <italic>t</italic>-tests, ANOVA, Kendall's <italic>τ</italic>-correlation, <italic>χ</italic>² tests, and Fisher's exact tests. Significance testing was done two-sided at <italic>α</italic>=0.05.</p>" ]

[ "<title>Results</title>", "<title>Participants</title>", "<p>Forty-eight male and 22 female patients participated. They were aged 64 years on average (s.d.=9.4; range: 41–84) at the time of the interview and had been treated for rectal cancer 6–10 (M=8; s.d.=1.0) years ago. Among those responding to demographic questions, 31 (45%) patients had completed 9 years or less of education, 23 (33%) 10–12 years, and 15 (22%) 13 years or more. All patients had undergone surgery and 38 (54%) had also been treated with PRT.</p>", "<p>All surgeons as well as 14 out of 25 radiation and 9 out of 10 medical oncologists were male. The clinicians were between 35 and 62 years old (M=48, s.d.=7.3). Time since specialisation was 13 years on average (s.d.=8.1, range, 1–31) and was not significantly different according to specialty.</p>", "<title>Control preferences scale</title>", "<p>##FIG##0##Figure 1## depicts participants' preferences regarding their role in the decision about PRT in the treatment of rectal cancer. Except for eight (11%) patients who preferred to leave the decision to their clinician, all participants would prefer both the patient and the clinician to share in that decision.</p>", "<p>There was a trend (<italic>P</italic>=0.05) for patients' and clinicians' preferences over the active, collaborative, and passive patient roles to differ. The clinicians preferred in majority the collaborative role (73%), whereas patients' role preferences were more equally spread out over the three levels of involvement.</p>", "<p>There was a significant association (<italic>P</italic>=0.03) between a lower educational level in patients and their preference to relinquish decisional control to their clinician. Also, preferences of male compared with female patients were significantly (<italic>P</italic>=0.04) different, with preferences of male patients being more equally distributed over the decisional roles. Clinicians' specialty was significantly (<italic>P</italic>=0.01) related to their role preferences. Medical, radiation, and surgical oncologists preferred the clinician to decide mainly in increasing proportions (10, 20, and 24%, respectively). No significant associations were found between decisional role preferences and participants' age, patients' past treatment with PRT (yes/no), or clinicians' time since specialisation.</p>", "<title>Desirability of patient participation in treatment-related decision making</title>", "<p>##TAB##0##Table 1## shows that overall almost all patients (<italic>N</italic>=66, 96%) and clinicians (<italic>N</italic>=57, 95%) thought that cancer patients should be involved in decision making regarding treatment (question 1). Responses did not significantly differ according to patients' gender, educational attainment, or past treatment, nor to clinicians' specialty.</p>", "<p>The patients and clinicians did not significantly differ in the frequency with which they explained their answer by referring to patient autonomy (patients: 44%, clinicians: 46%), the need to inform patients (patients: 39%, clinicians: 35%), and to openness in patient–clinician communication (patients: 14%, clinician: 5%). A significantly larger proportion of clinicians than patients (23 <italic>vs</italic> 5%; <italic>P</italic>&lt;0.01) referred to the outcome of the decision process, in that patients and clinicians should reach an agreement about treatment and/or that a participating patient is more motivated and therefore endures the treatment better. There was a trend for clinicians to refer more often to their responsibility and expertise (25 <italic>vs</italic> 12%; <italic>P</italic>=0.10). Eight (14%) clinicians compared with none of the patients referred to the necessity to comply with the informed consent procedure, and for six of the eight this was their sole explanation. Of the 66 patients, 15 (23%) motivated their response only as the need for patients to be informed, compared with none of the clinicians.</p>", "<p>Most participants (patients: <italic>N</italic>=51, 85%; clinicians: <italic>N</italic>=46, 81%) thought that not all patients are able to participate in such decision making (##TAB##0##Table 1##, question 2). There was a trend (<italic>P</italic>=0.07) for patients' and clinicians' responses to differ if we took the patient (male, 71 years old, with 13 years or more education, and irradiated prior to surgery) into account who did not have an opinion on the issue. If we discard this participant, patients' and clinicians' responses did not significantly differ.</p>", "<p>There was a trend (<italic>P</italic>=0.10) for patients (22%) to refer more often to patients' psychological inability (not further specified), when compared with clinicians (9%). Patients (43%) and clinicians (33%) referred in similar proportions to emotional incapacities, including patients' anxiety or lack of confidence. A significantly larger proportion of clinicians compared with patients (59 <italic>vs</italic> 22%; <italic>P</italic>&lt;0.001) referred to cognitive deficiencies, in that patients may not grasp the information provided and especially have difficulties with understanding the risks involved. Clinicians tended also to refer more often to co-morbidity, including dementia and psychiatric disorders, when compared with patients (28 <italic>vs</italic> 12%; <italic>P</italic>=0.07). Other explanations related to sociodemographic factors, including age, social support, and religion (patients: 12%, clinicians: 20%) and the novelty and complexity of the decision situation for patients (patients: 6%, clinicians: 4%).</p>", "<p>A large majority (patients: <italic>N</italic>=55, 79%; clinicians: <italic>N</italic>=41, 69%) of participants indicated that clinicians should try to involve patients in decisions about their treatment, even if the patient is reluctant to be involved (##TAB##0##Table 1##, question 3). Patients' and clinicians' responses significantly differed (<italic>P</italic>=0.04) if we take into account the two male respondents (70 and 75 years old, 9 years or less education, and irradiated preoperatively) who did not have an opinion on the issue. If we discard their response, patients and clinicians did not significantly differ in their opinion on the involvement in treatment decision making of reluctant patients.</p>", "<p>Participants justified their conviction by referring to patient responsibility (patients: 18%, clinicians: 7%) and the temporary nature of patient's reluctance (patients: 7%, clinicians: 2%). Others (patients: 13%, clinicians: 15%) viewed it as part of the clinician's task. Moreover, participants (patients: 46%, clinicians: 56%) often referred to the manner in which physicians may proceed, such as by providing information, reassuring the patient, or involving significant others of the patient. Several participants (patients: 16%, clinicians: 17%) stated that the clinician should try to involve a reluctant patient only within reason. In contrast, four patients compared with none of the clinicians considered that a clinician should try to involve a patient to the utmost. Participants' reasons for not trying to involve a reluctant patient referred to respect for the patient's wish not to participate. Participants indicated, moreover, that in such circumstances a clinician should adopt the role of providing information (<italic>N</italic>=5) or involving significant others of the patient (<italic>N</italic>=1).</p>", "<title>Weighing benefits and side effects of treatments</title>", "<p>A large majority of patients (<italic>N</italic>=52, 74%) and clinicians (<italic>N</italic>=49, 82%) thought that clinicians are not always able to weigh the pros and cons of treatment for their patients (##TAB##0##Table 1##, question 4).</p>", "<p>Participants referred to individual differences in patients' experience and the acceptance of treatments (patients: 39%, clinicians: 35%), the lack of skills and/or the subjective stance of the clinician (patients: 21%, clinicians: 29%), and insufficiencies in the dialogue between patients and clinicians (patients: 27%, clinicians: 29%). Participants thinking that a clinician always has this ability (patients: <italic>N</italic>=16, 23%; clinicians: <italic>N</italic>=11, 18%), in contrast, referred mostly to clinicians' skills and responsibility (patients: 81%, clinicians: 55%). Others referred to the clinician knowing patients sufficiently (patients: 13%, clinicians: 27%). One patient (7%) explained it by the patient's trust in the clinician.</p>", "<p>Overall, even higher proportions of participants (patients: <italic>N</italic>=61, 88%; clinicians: <italic>N</italic>=52, 87%) did not think that clinicians can always weigh the value of quality compared to length of life accurately for individual patients, even though this may be viewed as a specific case of weighing the pros and cons of treatment mentioned above (##TAB##0##Table 1##, question 5).</p>", "<p>Participants related clinicians' lack of insight in patient values often to individual patient differences. Patients did so significantly more often than clinicians (patients: 66 <italic>vs</italic> clinicians: 42%; <italic>P</italic>=0.02). A significantly larger proportion of clinicians compared to patients underlined the importance of dialogue, referring to the need for patients to bring their values forward in the interaction with the clinician (patients: 10 <italic>vs</italic> clinicians: 35%; <italic>P</italic>&lt;0.01). Other participants (patients: 7%, clinician: 10%) referred to limitations resulting from physicians' emotions, experience, and subjective stance.</p>", "<p>When asked explicitly about an adjuvant treatment that cures an additional 1–5% patients at the expense of a clinically relevant risk of side effects, a significant minority (17%) of clinicians would not even propose the treatment to their patient. None of the remaining 83% clinicians would prescribe that treatment without discussing it first with their patient. Of them, half (54%) would tell patients the absolute probabilities of benefits and side effects and other half (46%) would not. Moreover, a majority (63%) of clinicians indicated that in their opinion, patients should be involved in the formulation of treatment guidelines. A significant minority (37%) was opposed to patient involvement at that stage.</p>" ]

[ "<title>Discussion</title>", "<p>The results from this study suggest that both in case of a specific adjuvant treatment decision situation and more generally, a large majority of treated cancer patients and clinicians prefer the decision to be the outcome of a shared decision-making process. Our results support previous research in newly diagnosed cancer patients (##REF##9145723##Degner et al, 1997a##), in palliative cancer care patients (##REF##9301440##Rothenbacher et al, 1997##), and in a healthy population (##REF##17904327##O'Donnell and Hunskaar, 2007##) that a higher educational attainment is associated with a preference for a more active role in decision making. Our finding that gender may affect decisional role preferences adds to the evidence that is, as of yet, inconclusive (reviewed by ##UREF##1##Hubbard et al, 2008##). Further assessments are needed before firm conclusions can be drawn. Our results on surgical oncologists preferring to a greater extent that the clinician mainly decides on treatment compared with radiation and medical oncologists may seem to diverge from ##REF##12610196##Charles et al's (2003)## results. They found breast cancer surgeons to agree more often than radiation and medical oncologists to the statement that patients and physicians agreeing together to the treatment to be given, to be important to a shared decision-making process. However, surgeons' understanding of a shared process may differ from their preferences regarding actual decision-making in the clinic. Other determinants of decisional role preferences could not be traced.</p>", "<p>Up to one-fifth of the patients and one-third of the clinicians felt that reluctant patients need not be involved in the decision making, referring to respect for the patient's attitude. Bearing in mind that clinicians are not accurate in judging patient values for the outcome of care (##REF##11569703##Cotler et al, 2001##; ##REF##11533437##Montgomery and Fahey, 2001##; ##REF##15234224##Brothers et al, 2004##; ##REF##17634489##Stalmeier et al, 2007##), it seems in these cases critical to ascertain that the patient persists in preferring non-involvement.</p>", "<p>Importantly, these results do not as such shed light on whether patients and clinicians actually agree on how patient involvement should take place. The participants provided a variety of explanations for their opinions on participation. Clinicians more often than patients defined patient participation in terms of clinician–patient communication, namely reaching an agreement about treatment. Some clinicians compared to none of the patients explained the need for patient participation in terms of the need to comply with the informed consent procedure. For a significant minority (10%) of clinicians, this was their only motivation. It is unclear how these clinicians view the legal definition of patient participation and whether it goes beyond informing patients and asking them for consent to treatment plans that clinicians have proposed.</p>", "<p>There is no agreement on the definition of shared decision making in the literature (##REF##16051459##Makoul and Clayman, 2006##; ##REF##17873252##Moumjid et al, 2007##). However, essential elements incorporated in all prominent cited models of shared decision making include not only the presentation of options but also the discussion of patient values (##REF##16051459##Makoul and Clayman, 2006##). Accordingly, the focus of common definitions of shared decision making is both on information exchange between physician and patient and the involvement of both parties in the decision made (##REF##17873252##Moumjid et al, 2007##). As a result, patient participation presupposes that both clinicians and patients have responsibilities, which stands in sharp contrast to patients only being informed. Strikingly, a significant minority of patients explained participation in decision making exclusively in terms of being informed. Our finding is in accord with evidence from a qualitative study including 41 patients diagnosed with colorectal cancer. Patients in that study reported to perceive that there was a ‘right’ decision to be made and that doctors would make the right decision for them (##REF##15860051##Beaver et al, 2005##). Several other studies have also shown that patients more often prefer to receive information than to actually participate in decision making (##REF##3206248##Blanchard et al, 1988##; ##REF##2644407##Ende et al, 1989##; ##REF##8583261##Nease and Brooks, 1995##; ##REF##9695899##Benbassat et al, 1998##). Clearly, not all patients want to be involved (##REF##8539181##Davison et al, 1995##; ##REF##11281903##Beaver et al, 1999##; ##REF##16859860##de Haes, 2006##). ##REF##15045756##Schwartz (2004)## has even questioned the superiority of patient autonomy. Making choices may be a burden rather than a good option. By choosing one option, the patient dismisses the advantages of the options that are not selected, and this implies loss. Similarly, anticipated and post-decision regret may be at stake (##REF##16859860##de Haes, 2006##). None of the clinicians, in contrast, appeared to agree that informing a patient is sufficient to consider the patient involved in the process. This finding is in line with results of ##REF##12610196##Charles et al (2003)##, showing that only few physicians would define shared decision making in terms of information exchange alone.</p>", "<p>Where clinicians and patients wish to share in the decision-making process, patients should voice their values during their interaction with their clinician. However, patients may feel intimidated by their doctor because of the power differential (##REF##16480987##Bryant et al, 2006##), and therefore refrain from participating (##REF##9681902##Guadagnoli and Ward, 1998##; ##REF##17614177##Roter et al, 2007##). Physician use of supportive communication was found to be essential to facilitate patient involvement (##REF##16166865##Street et al, 2005##) and physicians' explicit encouragement of patient participation may foster patient participation in medical decision making (##REF##17443368##Fraenkel and McGraw, 2007##).</p>", "<p>More than 80% of patients and clinicians doubted as to all patients' ability to participate in treatment decision making, even though almost all participants considered it necessary. Clearly, participants' doubts as to patients' capacities could point to the exceptional cases of emotional disturbance or cognitive deficiencies. But the limitations that participants stated could apply to any ordinary patient, not the isolated case. Patients and clinicians often nominated emotional barriers. Indeed, distress may hamper patients' capacity to process information (##REF##7841058##McHugh et al, 1995##; ##REF##12790250##Erblich et al, 2003##). Clinicians more often than patients further viewed cognitive limitations in patients as a barrier. Limited educational attainment and literacy skills have been shown to relate to difficulties in understanding and recalling complex medical information (##REF##12038721##Williams et al, 2002##). As already suggested by some of the patients and clinicians in this study, clinicians may find ways to involve the more common patients, such as by addressing their fears, simplifying the information, or repeating it to facilitate its processing. Limited skills in patients to understand medical information should be an important motivation for clinicians to explain the information in even more simple language, take time to repeat the relevant details and to check understanding, and supply reference materials. Importantly, where inability is related to feelings of intimidation or lack of encouragement, clinicians could help patients to overcome patients' hesitations.</p>", "<p>Interestingly, none of the participants in our study suggested time constraints as a barrier to patient involvement. In ##REF##17443368##Fraenkel and McGraw's (2007)## study, time during medical encounters was nominated as an essential element to enable patients to become informed and to process information. Results in patients with heterogeneous cancers suggest though that clinical encounters need not be lengthened if the clinician proactively addresses patient questions (##REF##11720460##Brown et al, 2001##).</p>", "<p>Patients and clinicians seemed to agree that clinicians need patient input to assess accurately how the individual patient weighs pros and cons of treatment alternatives. Up to one-fifth of patients and almost one-third of clinicians made reference to clinicians' subjective stance. They recognised that the expertise of clinicians does not exist in a vacuum but is embedded in their interpretation of the situation and their perception of the patient. Interestingly, if pros and cons are specified as quality <italic>vs</italic> length of life, participants agreed even more strongly to clinicians' limitations in this regard. In their explanations, patients emphasised individual patient differences. Clinicians underlined patients' role in terms of the need for clinicians to receive patient input in order to assess accurately how an individual patient weighs risks and benefits of treatments. In practise, this standpoint implies that clinicians are willing to inform patients about risks and benefits of treatments, and maybe to a larger extent than they routinely do. In particular, this requires the use of precise vocabulary so as to facilitate patient recall of the discussion of relevant treatment options (##REF##12825857##Keating et al, 2003##) and guarantee the necessary specificity to help patients estimate the impact of the treatment on their lives (##REF##17265785##Davidson et al, 2007##). This perspective also requires clinicians to help patients to bring their values forward. Evidently, clinicians should then be prepared to acknowledge the legitimacy of patients' values in treatment decisions. Also, patients should then accept to share the responsibility for the treatment decision (##REF##10452420##Charles et al, 1999a##).</p>", "<p>A significant minority of clinicians indicated that they would decide against offering an existing treatment if the small probability of extra benefit would go hand in hand with a large probability of side effects. This is what may be termed a physician's silent decision (##UREF##2##Whitney and McCullough, 2007##) and may be justified by considerations of (lack of) clinical utility. Yet, clinicians' opinion about the clinical utility in this case varied. Also, almost half of the clinicians who would offer the treatment, would not state the absolute probabilities. These results are in line with those from ##REF##9469335##Ravdin et al (1998)##. Their study showed that women with breast cancer often are not given quantitative estimates of the magnitude of probable benefit and toxicity of adjuvant therapy. Not stating the probably impact of adjuvant therapy or using probability words instead of numbers may result in patients understanding poorly the trade-offs in benefits and side effects. Indeed, the breast cancer patients in Ravdin <italic>et al</italic>'s study overestimated their risk of early relapse and the effectiveness of their adjuvant therapy, even though they were younger and better educated than the average woman with breast cancer. For patients to be active, informed participants in the treatment-related decision-making process, probability information on benefits and risks of treatment alternatives is of critical value. It is questionable to what extent a patient can actually think about the pros and cons of treatment if the clinician is not willing to share relevant evidence.</p>", "<p>Our results suggest that at least some clinicians support the involvement of patients in the formulation of guidelines. Alternatively, guidelines may explicitly prescribe the need to elicit patient views at specific decision points. Dutch guidelines on cancer treatment (<ext-link xmlns:xlink=\"http://www.w3.org/1999/xlink\" ext-link-type=\"uri\" xlink:href=\"http://www.oncoline.nl\">www.oncoline.nl</ext-link>) including for example the recently updated guideline for rectal cancer and the one for breast cancer, include the recommendation that clinicians should extensively inform their patients about benefits and side effects of treatment alternatives. This recommendation underlines the need for clear information but does not go as far as stating that patient values for outcomes of care should explicitly be discussed when a treatment decision needs to be made, nor that treatment advice could vary, according to patient values. Yet, in order to obtain the best achievable care for individual patients, treatment choice should vary according to clinical circumstances and to patient values (##REF##7707631##Hlatky, 1995##). In the process of involving patients further than only informing them, patients may increasingly prefer to be treated in other ways than standard treatment. Guidelines may be explicit on this point and include the remark that given particular patient values, deviating from consensus treatment may be acceptable.</p>", "<p>In conclusion, our results run counter to guideline-based treatment of patients, in cases where clinicians' treatment advice is not based upon individual patient values. A majority of clinicians and patients in this study were in favour of clinicians and patients reaching an agreement on treatment. They held the conviction that clinicians cannot accurately predict how pros and cons of treatment weigh for their patients. Clinicians should inform patients extensively about treatment options and should acknowledge the legitimacy of patients' values in deciding about treatment. Ideally, treatment guidelines should include the recommendation that patient values about treatment benefits and side effects should explicitly be elicited at the time a treatment decision needs to be made.</p>" ]

[]

[ "<p>Patient participation in treatment decision-making is being increasingly advocated, although cancer treatments are often guideline-driven. Trade-offs between benefits and side effects underlying guidelines are made by clinicians. Evidence suggests that clinicians are inaccurate at predicting patient values. The aim was to assess what role oncologists and cancer patients prefer in deciding about treatment, and how they view patient participation in treatment decision-making. Seventy disease-free cancer patients and 60 oncologists (surgical, radiation, and medical) were interviewed about their role preferences using the Control Preferences Scale (CPS) and about their views on patient participation using closed- and open-ended questions. Almost all participants preferred treatment decisions to be the outcome of a shared process. Clinicians viewed participation more often as reaching an agreement, whereas 23% of patients defined participation exclusively as being informed. Of the participants, ⩾81% thought not all patients are able to participate and ⩾74% thought clinicians are not always able to weigh the pros and cons of treatment for patients, especially not quality as compared with length of life. Clinicians seemed reluctant to share probability information on the likely impact of adjuvant treatment. Clinicians should acknowledge the legitimacy of patients' values in treatment decisions. Guidelines should recommend elicitation of patient values at specific decision points.</p>" ]

[ "<p>Both clinicians and researchers have commented on the need to involve patients in treatment decision making, especially when a patient presents with a serious illness, different treatment options exist, the gains of treatment should be weighed against possible adverse effects, or outcomes are uncertain (##REF##9681902##Guadagnoli and Ward, 1998##; ##REF##10488014##Charles et al, 1999b##; ##REF##11533436##Robinson and Thomson, 2001##). Involving patients in treatment-related decision making is in line with the increasingly acknowledged patients' right to autonomy and self-determination. The need to involve patients is supported by evidence that physicians do not have the ability to adequately judge patients' values for outcomes of care (##REF##11569703##Cotler et al, 2001##; ##REF##11533437##Montgomery and Fahey, 2001##; ##REF##15234224##Brothers et al, 2004##; ##REF##17634489##Stalmeier et al, 2007##).</p>", "<p>Treatment in oncology is often guideline-based, at least in the Netherlands. Clinicians may access up-to-date information on nationwide guidelines in oncological and palliative care through <ext-link xmlns:xlink=\"http://www.w3.org/1999/xlink\" ext-link-type=\"uri\" xlink:href=\"http://www.oncoline.nl\">http://www.oncoline.nl</ext-link>, a publication of the Dutch Association of Comprehensive cancer centres (<ext-link xmlns:xlink=\"http://www.w3.org/1999/xlink\" ext-link-type=\"uri\" xlink:href=\"http://www.ikcnet.nl\">http://www.ikcnet.nl</ext-link>). In the guideline development process, trade-offs between expected benefits and side effects are almost exclusively made by clinicians, not by patients. Also, these trade-offs are usually not made explicit to patients. If patient values are to be incorporated in treatment decision making, which seems especially relevant in decisions on adjuvant treatment, this requires patients to participate in the consultation and voice their values. There is considerable uncertainty about what patients and clinicians understand by patient participation (##REF##9681902##Guadagnoli and Ward, 1998##). This study was set up to assess what role oncologists and cancer patients prefer in deciding about cancer treatment, and how they view patient participation in treatment-related decision making. Earlier studies have assessed decisional role preferences, but these are of limited interest to situations where treatment choices are mostly guideline-driven.</p>" ]

[ "<p>We thank the participating patients and oncologists for their efforts. This study was supported by a grant from the Dutch Cancer Society (UL 2005-3213).</p>", "<title>Appendix A Coding scheme</title>", "<p>\n##TAB##1##Table A1##\n</p>" ]

[ "<fig id=\"fig1\"><label>Figure 1</label><caption><p>Patients' (<italic>N</italic>=70) and clinicians' (<italic>N</italic>=60) control preferences in deciding about preoperative radiotherapy (PRT). Note: Phrasing of control preferences roles in <italic>patient</italic> interviews: (A) I prefer to make the decision about my treatment; (B) I prefer to make the decision about my treatment, after seriously considering my doctors' opinion; (C) I prefer that my doctor and I make the decision about my treatment jointly; (D) I prefer that my doctor makes the decision about my treatment, after seriously considering my opinion; (E) I prefer to leave the decision about my treatment to my doctor. Phrasing of control preferences roles in <italic>clinician</italic> interviews: (A) I prefer to leave the decision about treatment to my patient; (B) I prefer that my patient makes the decision about treatment, after seriously considering my opinion; (C) I prefer that my patient and I make the decision about treatment jointly; (D) I prefer to make the decision about treatment, after seriously considering my patient's opinion; (E) I prefer to make the decision about treatment.</p></caption></fig>" ]

[ "<table-wrap id=\"tbl1\"><label>Table 1</label><caption><title>Participantś answers to the closed-ended questions</title></caption></table-wrap>", "<table-wrap id=\"tbla1\"><label>Table A1</label><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"left\"/><col align=\"left\"/><col align=\"left\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Question</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Heading</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Subheading</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Examples of utterances coded</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">1. <italic>Do you think that cancer patients should be involved in decisions regarding their treatment?</italic><xref ref-type=\"fn\" rid=\"ta1-fn1\">a</xref></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Conditions for involvement</td><td align=\"left\" valign=\"top\" charoff=\"50\">Information</td><td align=\"left\" valign=\"top\" charoff=\"50\">Being able to understand information about alternatives; right to receive information</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Patient-clinician relationship</td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinician's honesty; patient's trust in clinician; clinician's clarity of information</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Decision</td><td align=\"left\" valign=\"top\" charoff=\"50\">There is a choice to be made</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Roles in decision-making process</td><td align=\"left\" valign=\"top\" charoff=\"50\">Patient</td><td align=\"left\" valign=\"top\" charoff=\"50\">Patient autonomy; patient chooses what he/she considers as quality; patient carries responsibility; patient thinks along</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinician</td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinician has the expertise; clinician advises; clinician is responsible</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Outcome of decision-making process</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Patient and clinician reach agreement; motivated patient endures the treatment better</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Legal obligation</td><td align=\"left\" valign=\"top\" charoff=\"50\">Informed consent procedure</td><td align=\"left\" valign=\"top\" charoff=\"50\">Need to comply to the informed consent procedure</td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">2. <italic>If so, do you think all patients are able to be involved in deciding about their treatment?</italic><xref ref-type=\"fn\" rid=\"ta1-fn2\">b</xref></td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Psychological inability<xref ref-type=\"fn\" rid=\"ta1-fn3\">c</xref></td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Emotional barriers</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Patient is too anxious; patient is emotionally unstable; patient lacks confidence</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Cognitive barriers</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Patient has limited intelligence; patient has difficulties with appraising risks; patient does not understand the information</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Socio-demographic factors</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Patient stems from an older generation; patient lacks social support; patient holds particular religious beliefs</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Co-morbidity</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Dementia; intellectual disability; psychiatric disorder</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Complex situation</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Patient is unfamiliar with decision situation; information is complex</td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>3. If a patient does not want to be involved in deciding about treatment, do you think that physicians should nevertheless try to involve their patient in deciding about treatment?</italic>\n<xref ref-type=\"fn\" rid=\"ta1-fn4\">d</xref>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Reason why</td><td align=\"left\" valign=\"top\" charoff=\"50\">Patient: Temporary evasive behaviour</td><td align=\"left\" valign=\"top\" charoff=\"50\">Patient feels panicky; it is important to help prevent regret in patient</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Patient: Responsibility</td><td align=\"left\" valign=\"top\" charoff=\"50\">Patient is (also) responsible; patient has to live with the consequences; patient is autonomous</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinician</td><td align=\"left\" valign=\"top\" charoff=\"50\">It is clinician's task; clinician needs patient agreement</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Patient-clinician relationship</td><td align=\"left\" valign=\"top\" charoff=\"50\">Creates mutual trust</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinician's devotion to involve patient</td><td align=\"left\" valign=\"top\" charoff=\"50\">To the utmost</td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinician should try to involve patient to the utmost</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Within reason</td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinician should try to involve patient within reason</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Respect patient's wish</td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinician should respect patient's wish not to participate</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Manner in which</td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinician gives information; clinician reassures the patient; clinician takes time; clinician involves significant others of patient; clinician gains patient's trust</td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>4. In deciding about treatment, one looks at advantages and disadvantages of various options, among other things. Do you think that physicians are always able to determine how these advantages and disadvantages weigh for a patient?</italic>\n<xref ref-type=\"fn\" rid=\"ta1-fn4\">e</xref>\n</td></tr><tr><td colspan=\"4\" align=\"left\" valign=\"top\" charoff=\"50\">\n<italic>5. If the decision is about quality vs length of life, do you think that physicians can always decide for patients how these should be weighed?</italic>\n</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Yes, the starting point is the…</td><td align=\"left\" valign=\"top\" charoff=\"50\">Patient</td><td align=\"left\" valign=\"top\" charoff=\"50\">Patient trusts the clinician; patient does not understand the situation well enough</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinician</td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinician has the expertise; clinician is responsible; clinician has the ability to estimate patient values</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Patient-clinician relationship</td><td align=\"left\" valign=\"top\" charoff=\"50\">Good communication; clinician knows the patient well enough</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">No, the starting point is the …</td><td align=\"left\" valign=\"top\" charoff=\"50\">Patient</td><td align=\"left\" valign=\"top\" charoff=\"50\">Individual patients differ too much from one another; patient makes own decision; patient knows best; patientś experience with health or health care differ; patients differ in their acceptance of treatments</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinician</td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinician has own subjective stance; clinician cannot infer importance for patient; clinician brings in own emotions</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\"> </td><td align=\"left\" valign=\"top\" charoff=\"50\">Patient-clinician relationship</td><td align=\"left\" valign=\"top\" charoff=\"50\">Clinician does not know patient well enough; clinician and patient need to consult together; patient needs to share own values with clinician</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><fn id=\"ta1-fn1\"><label>a</label><p>The participants' explanations to this question were only coded in those agreeing to patient involvement.</p></fn><fn id=\"ta1-fn2\"><label>b</label><p>The participants' explanations to this question were only coded in those agreeing to patient involvement (question 1) and disagreeing to question 2.</p></fn><fn id=\"ta1-fn3\"><label>c</label><p>This category includes references to psychological inabilities in patients that participants did not further specify.</p></fn><fn id=\"ta1-fn4\"><label>d</label><p>The participants' explanations to this question were coded both in those answering ‘yes’ and ‘no’ to the question.</p></fn><fn id=\"ta1-fn5\"><label>e</label><p>Coding categories for questions 4 and 5 were identical.</p></fn></table-wrap-foot>" ]

[ "<graphic xlink:href=\"6604611f1\"/>", "<graphic xlink:href=\"6604611t1\"/>" ]

[]

{ "acronym": [], "definition": [] }

[]

[]

[]

[]

[]

[]

[ "<p>\n<bold>Sir,</bold>\n</p>", "<p>The article by ##REF##18182985##Bowen et al (2008a)##, reporting that Black women in Hackney presented for breast cancer at a median age of 21 years younger than that of White women and had more serious pathology, has given rise to widespread concern and a petition to 10 Downing Street calling for a programme of early breast cancer screening for Black women. If the conclusions of the article are correct, the refusal of such a screening programme would amount to no less than a racial injustice. If they are not correct, its introduction would lead to unnecessary anxiety, discomfort and expense. All the more reason, then, to require from such a study the highest standards of scientific rigour.</p>", "<p>Two letters to the editor (##REF##18414474##Dindyal et al, 2008## and ##REF##18542067##Cichowska et al, 2008##) have raised concerns about this study. The authors' replies to these letters (##UREF##0##Bowen et al, 2008b##, ##UREF##1##2008c##) did not seem to me to deal with these points adequately, and I therefore subjected their published data to further analysis. The results indicate that it may be premature to conclude that Black women in general have a higher risk of contracting early breast cancer. The risk would seem to be confined to one ethnic subcategory, ‘Black Other’. In this category, however, the risk seems to be very high.</p>", "<p>##REF##18414474##Dindyal et al (2008)## and ##REF##18542067##Cichowska et al (2008)## emphasised the importance of distinguishing among Black women originating from different geographical regions. In their reply to ##REF##18414474##Dindyal et al (2008)##, the authors present a breakdown of the sample of Black patients according to their ethnic subcategory. However, this does not answer the question of whether there are significant differences among these groups. To find out, we would need to compare the numbers of patients in each subcategory with the size of the underlying population. This information is readily available from the excellent website maintained by the Office for National Statistics (<ext-link xmlns:xlink=\"http://www.w3.org/1999/xlink\" ext-link-type=\"uri\" xlink:href=\"http://www.nomisweb.co.uk\">www.nomisweb.co.uk</ext-link>).</p>", "<p>A slight problem arises in performing these calculations because the ethnic subcategories used by ##REF##18182985##Bowen et al (2008a)## are not all labelled in the same way as in the census. They classify their patients as ‘Black African, African Caribbean, Black British or Black Other’, whereas the census data refer to ‘Black or Black British’, subdivided into ‘Black African’, ‘Black Caribbean’ and ‘Other’. The four patients in the authors' category ‘Black British’ can be allocated to the census category ‘Other’, as 83% of this group are British-born. I have assumed that the authors' category ‘African Caribbean’ corresponds to the census category ‘Black Caribbean’, as it is hard to see what else it could mean.</p>", "<p>The results of this analysis were surprising. First, although the total Black population of Hackney reported by the authors was approximately the same as that reported in the census data, their figure for the White population of interest was 34% higher than in the census. This seemed to be because they had included the category ‘White Other’ in their population estimate, whereas they had <italic>excluded</italic> this category from their patient sample.</p>", "<p>In fact, ##REF##18182985##Bowen et al (2008a)## give contradictory information about the inclusion criteria for their White patients. In their article, they state that ‘data were obtained from … 191 White British women’, but in their reply to ##REF##18414474##Dindyal et al (2008)##, they say ‘all White women included reported themselves to be White English, Welsh or Scottish <italic>or White Irish</italic>’ (my italics). I have assumed that the second statement is more accurate and that White Irish patients were included.</p>", "<p>Excluding persons under 20 from the population totals, as the risk of contracting breast cancer at such an early age is negligible, ##TAB##0##Table 1## shows the overall incidence rates for each group over a 10-year period.</p>", "<p>Comparing the White and Black Other groups, the odds ratio (OR) is 3.43 (<italic>χ</italic>²=36.1; <italic>P</italic>&lt;0.0001). However, no significant difference is found between the White group and any of the remaining Black groups. The data also need to be adjusted for differences in socioeconomic status, which only the authors are in a position to do. (It would be interesting to know what influence this variable has: unfortunately no information is given on this point.) But these figures suggest strongly that only the ‘Black Other’ group has an increased overall risk of breast cancer. In this small group (2.5% of the total population), the increase is indeed dramatic and needs to be investigated urgently.</p>", "<p>Of course, ##REF##18182985##Bowen et al (2008a)## are not concerned with overall rates but with the age of onset and clinical features of breast cancer. There was a striking difference of 21 years between Black and White women in the median age of presentation. However, as ##REF##18542067##Cichowska et al (2008)## point out, this figure takes no account of possible differences in the age structure of the Black and White populations. According to the authors, no such differences were significant; furthermore, they say that they were unable to calculate detailed age-specific rates, because information about the age structure of the population was only available in three broad age groups (0–15, 16–59 and 60 years or more).</p>", "<p>Both statements are puzzling. The difference in age structure among the groups as reported by the authors in their ##TAB##0##Table 1## is, in fact, highly significant (<italic>χ</italic>²=374.6; <italic>P</italic>&lt;0.0001). Second, detailed information about the age structure of ethnic groups is readily available in the census data for Hackney. By combining Table 4 of the authors' original article with this data, it was possible to calculate age-specific incidence rates, although I could not perform this analysis for the ethnic subgroups separately or adjust for socioeconomic status. ##FIG##0##Figure 1## shows the results.</p>", "<p>Even after allowing in this way for the different age structure of the two groups, incidence rates below the age of 50 years were significantly higher in the Black group (age 20–39 years, OR=4.41, <italic>χ</italic>²=14.8, <italic>P</italic>&lt;0.0002; age 40–49 years, OR=1.95, <italic>χ</italic>²=8.0, <italic>P</italic>&lt;0.005; other differences not significant. When carrying out five tests simultaneously, the Bonferroni correction requires an <italic>α</italic> of 0.01 instead of 0.05). These results support the authors' conclusions, although adjustment for differences in socioeconomic status might change the picture to some extent. Nevertheless, the overall figures quoted above suggest that these effects are mainly due to the category ‘Black Other’.</p>", "<p>With the help of the census data, it is also possible to calculate how many patients would be expected in the Black group if it had the same age structure as the White group. The difference in median age would then fall to approximately 12 years: in other words, almost half of the difference reported by ##REF##18182985##Bowen et al (2008a)## is simply a consequence of the fact that there are relatively few Black women over the age of 60 in Hackney. It is regrettable that the figure of 21 years has received so much publicity in the media.</p>", "<p>These results show clearly that it does not make sense, either in research or screening policy, to treat Black women as a single group. Moreover, looking at the variations in incidence rates within different Black groups suggests that <italic>country of birth</italic> may be an important variable.</p>", "<p>As ##UREF##2##McCormack et al (2008)## point out, ‘breast cancer incidence rates vary six-fold between industrialised and less-developed countries, and migrants from low-risk countries to high-risk countries have an intermediate risk’. On this basis, we would expect a lower incidence in the Black subgroups containing a higher percentage of foreign-born migrants. As ##TAB##1##Table 2## shows, this relationship holds within this study, suggesting that future studies should pay attention not only to self-reported ethnicity, but also to place of birth.</p>", "<p>It should be borne in mind that data obtained from studies such as this one can never be entirely reliable, for the simple reason that the boundaries of local authorities are permeable. For various reasons, people may seek treatment at a hospital outside their own area. In this study, it is impossible to know how many women from Hackney obtained treatment at other hospitals, and the authors do not tell us how many of the patients in this study came from other areas. Moreover, census data are collected only once in 10 years, whereas populations can change rapidly.</p>", "<p>The comparison of pathological and tumour features yielded hardly any indications of a Black/White difference: of the nine variables studied, only two showed significant group differences. It is not reported whether a Bonferroni or similar correction was used – but if this was not the case, none of the reported differences are significant, as with nine tests the Bonferroni correction requires an <italic>α</italic> of 0.005 instead of 0.05.</p>", "<p>To sum up: in view of the concern their study has given rise to, it would seem highly advisable for the authors to reanalyse their data, taking care to match their categories as accurately as possible with those used in the census data and paying attention to the points raised by ##REF##18414474##Dindyal et al (2008)## and ##REF##18542067##Cichowska et al (2008)##.</p>" ]

[]

[ "<fig id=\"fig1\"><label>Figure 1</label><caption><p>Age-specific incidence of breast cancer in the Black and White groups (patients per 1000 of the population).</p></caption></fig>" ]

[ "<table-wrap id=\"tbl1\"><label>Table 1</label><caption><title>Overall incidence rates by ethnic subgroup (patients per 1000 of the population)</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"char\" char=\".\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Group</bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>Incidence</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">White British and Irish</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">4.9</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Black Caribbean</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5.5</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Black African</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3.6</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Black Other</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">16.9</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Black (total)</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5.5</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl2\"><label>Table 2</label><caption><title>Ethnic subgroup, % UK-born and overall incidence rates</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"char\" char=\".\"/><col align=\"char\" char=\".\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Subgroup</bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>% Born in United Kingdom</bold>\n</th><th align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">\n<bold>Overall incidence</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Black African</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">37</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">3.6</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Black Caribbean</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">50</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">5.5</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">Black Other</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">83</td><td align=\"char\" valign=\"top\" char=\".\" charoff=\"50\">16.9</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"6604625f1\"/>" ]

[]

{ "acronym": [], "definition": [] }

[]

[ "<title>Patients and methods</title>", "<title>Patients</title>", "<p>Patients, aged &gt;18 years old, with histologically confirmed NSCLC who underwent biopsy or surgical excision of the primary tumour and the corresponding metastases were included in this retrospective analysis. Histological type was determined according to the World Health Organization criteria, and the stage of the disease corresponds to that of the time of primary diagnosis. Smoking history was obtained during the patient's first evaluation. All patients gave their informed consent for using their tumour sample for molecular and pathologic analysis. The study has been approved by the Ethics and Scientific Committees of our institution.</p>", "<title>DNA extraction and mutation analysis</title>", "<p>All tumour samples were formalin-fixed paraffin-embedded tissues. Sections of a paraffin block corresponding to one representative area of the tumour were stained with haematoxylin/eosin, and the presence of tumour tissue was verified by an experienced pathologist. Subsequently, tissue samples from at least three serial sections were microdissected (piezo power Eppendorf Microdissector; Eppendorf, Germany) to ensure that specimens contained at least 80% tumour cells; sections of 5 <italic>μ</italic>m thickness were also collected from adjacent normal tissue when available, extracted with xylene and ethanol to remove paraffin and placed in 1% SDS/proteinase K (10 mg ml⁻¹) at 56°C overnight. DNA was extracted using the MasterPure Complete DNA/RNA Purification kit (Epicentre Biotechnologies, Madison, WI, USA) according to the manufacturer's instructions. Exons 18, 19, 20 and 21 of the EGFR and exon 1 of K-RAS were sequentially amplified by two rounds of polymerase chain reaction (PCR) and subjected to direct sequencing. The PCR primers for <italic>EGFR</italic> amplification were as follows: 155273L23: 5′-TCCCAAACACTCAGTGAAACAAA-3′; 155348L22: 5′-TGGTCTCACAGGACCACTGATT-3′; 154838U22: 5′-TCAGAGCCTGTGTTTCTACCAA-3′; 154899U20: 5′-TCCAAATGAGCTGGCAAGTG-3′; 55634U24: 5′-AAATAATCAGTGTGATTCGTGGAG-3′; 156027L20: 5′-TGTGGAGATGAGCAGGGTCT-3′; 156107L22: 5′-GAGGCCAGTGCTGTCTCTAAGG-3′; 155750U20: 5′-GTGCATCGCTGGTAACATCC-3′; 173160L22: 5′-CAGCTCTGGCTCACACTACCAG-3′; 173076L19: 5′-CATCCTCCCCTGCATGTGT-3′; 172656U22: 5′-GCAGCGGGTTACATCTTCTTTC-3′; and 172747U19: 5′-GCT CAGAGCCTGGCATGAA-3′. The PCR primers for <italic>K-RAS</italic> amplification were as follows: RASU1: 5′-AGGCCTGCTGAAAATGACTGAATA-3′; RASL1: 5′-CTGTATCAAAGAATGGTCCTGCAC-3′; RASU2: 5′-AAAATGACTGAATATAAACTTGTGG-3′; RASL2: 5′-CTCTATTGTTGGATCATATTCGTC-3′. The first PCR was carried out in a total volume of 10 <italic>μ</italic>l contained 1/10 of the extracted genomic DNA using 1 U of Platinum Taq DNA polymerase (Invitrogen Corporation, Carlsbad, CA, USA). The initial denaturing step was at 94°C for 15 min, followed by 35 cycles of denaturing step at 94°C for 20 s, annealing step at 60°C for 30 s and extension step at 72°C for 1 min, ending with a final extension step at 72°C for 7 min. Nested PCR was carried out in a total volume of 20 <italic>μ</italic>l and the conditions were identical to the first PCR. The PCR products are directly sequenced by dye terminator sequencing (ABI BigDye Terminator kit, v3.1, Applied Biosystems, Foster City, CA, USA), purified by ethanol precipitation and separated by capillary electrophoresis on an ABI 3100 Avant genetic analyzer (Applied Biosystems). Sequence analysis was carried out by Seqscape software (Applied Biosystems) and manually by two reviewers (AK and AV). All sequence variations were confirmed by sequencing in both directions and by an independent PCR amplification when sufficient material was available.</p>", "<p>The sensitivity of our methodology was evaluated by determining the minimum frequency of <italic>EGFR</italic> and <italic>K-RAS</italic> mutations required for detection in our system. This was accomplished by performing mixing experiments using cell lines with and without <italic>EGFR</italic> (H2073-wt-<italic>EGFR</italic> and HCC827-Del19-<italic>EGFR</italic>) or <italic>K-RAS</italic> (H2073-wt-<italic>K-RAS</italic> and A549-G12D-<italic>K-RAS</italic>) mutations. These experiments demonstrated that the Del19 and <italic>K-RAS</italic> mutations could be detected when present in 10 and 20% of the cells in the sample, respectively (data not shown).</p>", "<title>Immunohistochemistry for EGFR</title>", "<p>The paraffin-embedded tissues were cut at 4 <italic>μ</italic>m thickness and were deposited on SuperFrost/Plus Slides (O.Kindler GmbH, Freiburg, Germany). After deparaffinisation, the slides for EGFR were treated with proteinase K (Code S3020, DakoCytomation, Glostrup, Denmark) for 5 min at room temperature, and those for p-EGFR were treated with EDTA at pH 8 in a microwave oven three times for 5 min at 500 W for antigen retrieval. The primary anti-EGFR antibody (mouse monoclonal, clone H11, code M3563, DakoCytomation) was used at a dilution of 1 : 50 (v/v) and incubated for 1 h at room temperature. The primary anti-phospho-EGFR antibody (rabbit monoclonal, pY1173-EGFR, code no. 4407, Cell Signalling, Danvers, MA, USA) was used at a dilution of 1 : 200 (v/v) and incubated overnight at 4°C. For the detection of antigen-antibody reaction, the UltraVision detection system AP Polymer kit (Cat no. TL-125-AL, Lab Vision, Fremont, CA, USA) was used according to the manufacturer's instructions. Fast red was used as chromogen for 20 min; the sections were counterstained with Mayer's haematoxylin for 3 min, subsequently rinsed in ammonium and finally mounted with glycergel. A positive (an NSCLC tumour specimen with known positivity) and a negative (omission of primary antibody) control were used. The stained sections were independently evaluated by two pathologists (AK and ES). Immunoreactions of EGFR were graded as 3+ when strong complete membrane staining was observed in more than 10% of the tumour cells, as 2+ when more than 10% of the tumour cells showed weak-to-moderate complete membrane staining, as 1+ when partially, faint membrane stain was detected in more than 10% of the tumour cells and as 0 when no staining at all or membrane staining in less than 10% of the tumour cells was observed. In this study, we arbitrarily classified EGFR expression status in two subsets; the 2+ or 3+ signals were considered as EGFR overexpression and the 0 or 1+ signals as non-expression (##REF##17442448##Koutsopoulos et al, 2007##).</p>", "<p>A paraffin block from the HCC827 lung epithelial adenocarcinoma cell line and commercially available positive controls (SignalSlide Phospho-EGF Receptor (Tyr1173) IHC Controls, code no. 8102, Cell Signalling) was used to validate the antibody for p-EGFR. pY1173-EGFR immunostaining was mainly membranous and was graded as 0 (&lt;5% positive cells), 1+ (5–19% positive cells), 2+ (20–50% positive cells) and 3+ (&gt;50% positive cells) (##REF##17161498##Cortas et al, 2007##).</p>", "<title>Statistical analysis</title>", "<p>McNemar test was used to compare the <italic>EGFR</italic> and <italic>K-RAS</italic> status between primary tumours and related metastatic sites. Differences were considered statistically significant when the <italic>P</italic>-value was &lt;0.05. All statistical tests were two-sided.</p>" ]

[ "<title>Results</title>", "<title>Patient characteristics</title>", "<p>Twenty-two (88%) patients were men and 22 (88%) were active or former smokers and their median age was 55 years (range, 41–70). Eighteen (72%) patients had adenocarcinomas (ADC) and 21 (84%) patients had stage III or IV disease. Among the 50 samples analysed (25 primary tumours and 25 metastases), 26 (52%) samples were surgical and 24 (48%) biopsies. The primary tumour tissue was the lung (<italic>n</italic>=25 patients); and the origin of the metastatic sample was lung (<italic>n</italic>=9 patients), thoracic wall (<italic>n</italic>=5 patients), adrenal gland (<italic>n</italic>=4 patients), brain (<italic>n</italic>=3 patients), bone (<italic>n</italic>=2 patients), liver (<italic>n</italic>=1 patient) and skin (<italic>n</italic>=1 patient). Metastases were metachronous in all cases; the median time elapsed between resection of the primary tumour and the corresponding metastatic site was 30 months (range, 4–143). The patients' clinicopathologic characteristics are presented in ##TAB##0##Table 1##. Nine (36%) patients were treated with gefitinib in the context of an Expanded Access Program (##TAB##1##Table 2##).</p>", "<title><italic>EGFR</italic> mutation status of the primary tumours and the corresponding metastasis</title>", "<p>The <italic>EGFR</italic> mutation status of the primary tumours and the corresponding metastases is presented in ##TAB##1##Table 2##. Epidermal growth factor receptor mutations were detected in the primary tumours of five (20%) patients; of these, three of them were the well-characterised ‘hotspot’ mutations in exon 19 (Del746-750 and E746V; case nos. 20, 23 and 18) and the remaining two were novel point mutations in exons 18 and 21 (L692P and G857E; case nos. 17 and 19), respectively. The corresponding metastases of the Del746-750 (case no. 23), E746V (no. 18), L692P (no. 17) and G857E (no. 19) mutant primary tumours were wild type with respect to <italic>EGFR</italic> mutation status. Epidermal growth factor receptor mutations were detected in the metastatic tumours of three (12%) patients (##TAB##1##Table 2##). The metastasis of one of these patients showed the same <italic>EGFR</italic> mutation (Del19) as the primary tumour as well as an additional one, the T790M in exon 20 (no. 20); conversely, the other two patients carried two novel mutations in exon 18 (L692P and V717A; no. 12) and the T847A (no. 13) in exon 21, which could not be detected in the patients' primary tumour samples. We have confirmed that the non-classical mutations detected in our series are not single nucleotide polymorphisms by mutation analysis of matched normal tissue or blood (data not shown).</p>", "<p>Consequently, the <italic>EGFR</italic> gene status could be classified as: (i) <italic>EGFR</italic> wild type in both primary tumour and metastasis (<italic>n</italic>=18 patients; 72%), and (ii) <italic>EGFR</italic> mutations detected only in the primary tumour (<italic>n</italic>=4 patients; 16%) or the metastases (<italic>n</italic>=2 patients; 8%) or both (<italic>n</italic>=1 patient; 4%). Therefore, <italic>EGFR</italic> mutation status showed a discordance of 28% (7 of 25 patients) (McNemar test, <italic>P</italic>=0.688) between the primary tumour and corresponding metastasis (##TAB##2##Table 3##).</p>", "<title>EGFR and p-EGFR expression by IHC in the primary tumour and the corresponding metastases</title>", "<p>Én 19 patients sufficient tumour tissue was available for immunohistochemical analysis of EGFR (##TAB##1##Tables 2## and ##TAB##3##4##). The incidence of EGFR overexpression (grade 2+, 3+) was 26.5% in both primary and metastatic tumours. Concordance between the primary tumour and the corresponding metastases was observed in 17 (89.5) patients (Cohen's Kappa=0.729, <italic>P</italic>=0.001) and among them EGFR was overexpressed in four (21.1%). Discordance was observed in two (10.5%) patients (nos. 14 and 15) (McNemar test, <italic>P</italic>=1). Evaluable paired tissue specimens for p-EGFR analysis by IHC were available in 16 patients. Three out of 16 (18.8%) primary and seven out of 16 (43.8%) metastatic tumours expressed phosphorylated EGFR (pY1173-EGFR-positive). Discordance between the primary and metastatic tumours was observed in eight (50%) patients (McNemar test, <italic>P</italic>=1) (##TAB##1##Tables 2## and ##TAB##3##4##). There was no correlation between the expression of EGFR and pY1173-EGFR. In order to confirm our findings, we repeated the immunohistochemistry from serial sections in all tumour specimens that exhibited EGFR-negative and p-EGFR-positive staining and the obtained results were identical. Epidermal growth factor receptor gene copy number was also investigated by fluorescence <italic>in situ</italic> hybridisation in eight patients (nos. 2, 5, 7, 9, 10, 13, 19 and 24) for whom paired tissue specimens were available; amplification of <italic>EGFR</italic> gene was not detected in any of these tumour specimens.</p>", "<title><italic>K-RAS</italic> mutation status of the primary tumours and the corresponding metastases</title>", "<p>Primary and metastatic tumours were also assessed for <italic>K-RAS</italic> mutations (##TAB##1##Table 2##). <italic>K-RAS</italic> mutations were detected in the primary tumours of five (20%) patients (nos. 8, 10, 22, 23 and 25) and in the metastatic tumours of five (20%) patients (nos. 9, 10, 11, 16 and 23), respectively. Two patients (nos. 10 and 23) carried the same <italic>K-RAS</italic> mutations in both primary and metastatic tumours (##TAB##1##Table 2##). One of them (case no. 23) carried the Del746-750 EGFR mutation in the primary tumour but not in metastasis. This was confirmed by three independent PCRs from three genomic DNAs extracted from serial sections of the paraffin blocks. Discordance in <italic>K-RAS</italic> mutation status between the primary tumours and the corresponding metastases was observed in six (24%) patients (McNemar test, <italic>P</italic>=1) (##TAB##2##Table 3##).</p>", "<title>Response to gefitinib according to <italic>EGFR</italic> and <italic>K-RAS</italic> mutation status</title>", "<p>Nine patients received gefitinib as first (nos. 23 and 18), third (nos. 15, 16, 19 20 and 22) or fourth (nos. 14 and 17) line treatment. Three patients received gefitinib before the biopsy of metachronous metastases (nos. 17, 18 and 20) and six received gefitinib after the biopsy of metastases (nos. 14, 15, 16, 19, 22 and 23). Five patients (56%) achieved stable disease and four (44%) progressive disease (##TAB##1##Table 2##). All patients who experienced progressive disease on gefitinib were wild type regarding the <italic>EGFR</italic> mutation status in both primary tumours and metastases (nos. 14, 15, 16, and 22), of these, two patients carried <italic>K-RAS</italic> mutations in the primary tumour (no. 22) or metastasis (no. 16). All the five patients with stable disease on gefitinib carried EGFR mutations in their primary tumours (nos. 17, 18, 19, 20 and 23). Of these, three patients carried the well-characterised activating mutations in exon 19 (nos. 18, 20 and 23) and two carried <italic>EGFR</italic> mutations of unknown function (nos. 17 and 19). Patient nos. 18 and 20 who received gefitinib before metachronous metastasis developed metastatic tumours, which were either wild type in respect to <italic>EGFR</italic> mutation status (no. 18) or had acquired resistance because of the T790M <italic>EGFR</italic> mutation (no. 20). Patient no. 23, who received gefitinib after metastasis, carried both <italic>EGFR</italic> and <italic>K-RAS</italic> mutations in the primary tumour and the same <italic>K-RAS</italic> mutation in metastasis.</p>" ]

[ "<title>Discussion</title>", "<p>Several studies have shown that activating <italic>EGFR</italic> mutations in exons 18, 19 and 21 have been associated with a 75–95% objective response rate with EGFR TKIs, whereas the <italic>K-RAS</italic> mutations were associated with a lack of sensitivity to these agents (##REF##15696205##Pao et al, 2005##; ##REF##17318210##Sharma et al, 2007##). Epidermal growth factor receptor gene amplification and protein expression have also been considered as predictors of clinical benefit with gefitinib (##REF##17075123##Hirsch et al, 2006##). However, in clinical studies, a significant percentage of patients presented clinical benefit when treated with TKIs irrespective of the expression and mutational status of <italic>EGFR</italic> (##REF##17318210##Sharma et al, 2007##). As, in most studies, <italic>EGFR</italic> expression and mutations were determined on the primary tumour, the observed clinical benefit of patients with wild-type <italic>EGFR</italic> or the absence of response to TKIs of patients with <italic>EGFR</italic> mutations could be due to discordance in the <italic>EGFR</italic> mutation status or expression between the primary tumour and the corresponding metastasis.</p>", "<p>This study demonstrated the existence of a significant discordance of <italic>EGFR</italic> and <italic>K-RAS</italic> mutations occurring in primary tumours and their corresponding metastases in patients with NSCLC. The discordance in <italic>EGFR</italic> mutation status was 28% and that in <italic>K-RAS</italic> was 24%. Similarly, two other studies in paired NSCLC tumours showed a discordance of 32 and 27% regarding the <italic>EGFR</italic> gene copy number (##REF##16524970##Italiano et al, 2006##; ##REF##18166836##Bozzetti et al, 2008##), whereas another study including six NSCLC patients of Asian ethnicity reported a 100% concordance in regard to <italic>EGFR</italic> mutation status (##REF##16642476##Matsumoto et al, 2006##). This discrepancy could be related to the different sites of the metastatic tumours analysed in this study (five different distant metastases), whereas in the Matsumoto's <italic>et al</italic> study, only tumour samples from brain metastases were included. Diverse sites of metastases in NSCLC probably represent different clonal outgrowths. Alternatively, we cannot exclude that the different ethnicity of patients and/or the different types of <italic>EGFR</italic> mutations could be the reason for this discrepancy (##REF##17409862##Tsao et al, 2006##; ##REF##18000506##Pallis et al, 2007##). In this study including only Caucasian patients, three classical activating (Del19 and E746V) and four non-classical mutations (T847A, L692P-V717A, L692P and G857E) were detected, whereas in Matsumoto's <italic>et al</italic> study, only classical activating Del19 and L858R mutations were reported. Concerning the non-classical mutations, it is unlikely to represent PCR artifacts, as L692F, T847I and G857E <italic>EGFR</italic> mutations have been previously reported (##REF##16357156##Fujimoto et al, 2005##; ##REF##16014883##Tsao et al, 2005##; ##REF##16870303##Hsieh et al, 2006##). Furthermore, the expression of phosphorylated EGFR (pY1173-EGFR-positive) on tumour cells from metastatic lesions carrying the T847A and L692P-V717A mutations strongly suggests that the former might be activating EGFR mutation. The time elapsed between diagnoses of the primary tumour and corresponding metastasis in patients (nos. 13 and 12) carrying the abovementioned mutations was 74.5 and 10 months, respectively. Therefore, acquisition of new mutations may be developed during the evolution of the metastatic process.</p>", "<p>The administration of TKIs could be an additional explanation of the observed discordance of <italic>EGFR</italic> mutations. Three out of 25 patients received gefitinib before the development of a metachronous metastasis, whereas none of the patients reported in the Matsumoto's <italic>et al</italic> study had exposed to TKIs. It is known that NSCLC patients with EGFR-dependent primary tumours when treated with TKIs can develop metastases, in which either the EGFR signalling is negated or resistance is acquired due to secondary <italic>EGFR</italic> mutations like T790M or MET amplification (##REF##12855618##Daneshmand et al, 2003##; ##REF##16288556##Gazdar and Minna, 2005##; ##REF##17463250##Engelman et al, 2007##; ##REF##17332337##Lutterbach et al, 2007##). It is interesting to note that in patients who had not been exposed to TKIs before biopsy of metastatic lesions, a 18% (4 out of 22 patients) discordance was observed between primary tumours and related metastases. However, a final explanation of the observed discordance, which cannot be excluded, concerns the low-frequency intratumoral heterogeneity for the occurrence of <italic>EGFR</italic> mutations (##REF##18449007##Sakurada et al, 2008##).</p>", "<p>Our findings concerning the <italic>K-RAS</italic> mutation status are in agreement with a previous study demonstrating that the <italic>K-RAS</italic> mutational status of the primary tumour does not always predict the status of bone metastasis in NSCLC (##REF##17607370##Badalian et al, 2007##). A similar phenomenon was also reported in patients with colorectal cancer (##REF##11387355##Tortola et al, 2001##). Although <italic>K-RAS</italic> mutations seem to be associated with the early development of NSCLC, it cannot be excluded that <italic>K-RAS</italic> mutations are lost later during tumour progression (##REF##2648401##Burmer and Loeb, 1989##; ##REF##8081702##Li et al, 1994##). This may, in part, explain the discordance in the <italic>K-RAS</italic> mutation status between primary tumours and metachronous metastases.</p>", "<p>Another possibility for the observed discordance in the <italic>EGFR</italic> and <italic>K-RAS</italic> mutation status could be related to the administered chemotherapy. However, as shown in ##TAB##0##Table 1##, among the 10 patients who had not received any treatment before the mutation analysis of the metastatic lesions, 5 developed metastases with different mutation status from that of the corresponding primary tumours. Thus, although tumour clone selection through the various treatments could be an explanation for the different molecular pattern in the primary tumour and metastatic site, our findings suggest that the metastasis genotype could be different from that of the corresponding primary tumour irrespectively of administered chemotherapy.</p>", "<p>Previous studies have shown that <italic>EGFR</italic> and <italic>K-RAS</italic> mutations are mutually exclusive, suggesting the presence of different pathways of lung carcinogenesis. However, as previously reported, our data show that <italic>K-RAS</italic> mutation may coexist with <italic>EGFR</italic> mutation (##REF##16357156##Fujimoto et al, 2005##; ##REF##16638863##Han et al, 2006##). Among the five patients with <italic>EGFR</italic> mutations in primary tumours, one patient concomitantly had <italic>K-RAS</italic> mutation (G12C with deletion in exon 19). In the metachronous metastasis, only the <italic>K-RAS</italic> mutation was retained. This patient received gefitinib after the biopsy of metastatic lesion and had stable disease lasting for 3.5 months. The limited duration of response is compatible with the knowledge that the presence of <italic>K-RAS</italic> mutations is associated with resistance to TKIs.</p>", "<p>In this study, the expression of pY1173-EGFR was different between primary tumours and corresponding metastases in eight (50%) patients, whereas EGFR expression was discordant in two (10%) patients. The apparent lack of correlation between EGFR and p-EGFR expression has been previously reported and could be due to the different scoring systems and different sensitivities of antibodies used to evaluate the EGFR and p-EGFR expressions (##REF##15386420##Han et al, 2005##). It is well known that overexpression of <italic>EGFR</italic> is associated with Tyr phosphorylation of the receptor proteins and that mutations in the kinase domain may cause constitutive phosphorylation of EGFR (##REF##16205628##Chen et al, 2006##). However, biochemical studies have shown that variable phosphorylation rates were associated with different tyrosine phosphorylation sites in the receptor; G719S mutant receptor had less EGF-induced phosphorylation at Y845, Y992, Y1068 and Y1173 than did wild-type EGFR, whereas L858R mutant receptor preferentially phosphorylated at the Y1068 but not at Y1173 (##REF##15284455##Sordella et al, 2004##; ##REF##16205628##Chen et al, 2006##; ##REF##17273938##Liu et al, 2007##).</p>", "<p>Clinical investigations suggested a correlation between a high <italic>EGFR</italic> gene copy number and <italic>EGFR</italic> mutations (##REF##15870435##Cappuzzo et al, 2005##; ##REF##15998907##Takano et al, 2005##). Our study failed to demonstrate such a correlation probably due to limited number of cases analysed. Evaluable paired tissue specimens for fluorescence <italic>in situ</italic> hybridisation analysis were available for eight patients. Among them, two (nos 13 and 19) patients presented different <italic>EGFR</italic> mutation status in the primary tumours and corresponding metastases; however, these two patients harbour non-classical <italic>EGFR</italic> mutations (T847A and G857E) and did not show different <italic>EGFR</italic> amplification patterns between paired samples. On the basis of these findings, at the present time, we consider that only the genetic differences unequivocally distinguish EGFR-dependent tumours, which are likely to be sensitive to TKIs from the tumours that could be resistant to these agents (##REF##16900147##Papadopoulos et al, 2006##).</p>", "<p>In conclusion, our findings indicate a substantial discordance of <italic>EGFR</italic> and <italic>K-RAS</italic> mutations between the primary tumours and the corresponding metastases in NSCLC and underline the need to consider the genotype of both primary and metastatic tumours for selecting patients who will respond to therapy with TKIs.</p>" ]

[]

[ "<p>In non-small-cell lung cancer (NSCLC), epidermal growth factor receptor (<italic>EGFR</italic>) and <italic>K-RAS</italic> mutations of the primary tumour are associated with responsiveness and resistance to tyrosine kinase inhibitors (TKIs), respectively. However, the <italic>EGFR</italic> and <italic>K-RAS</italic> mutation status in metastases is not well studied. We compared the mutation status of these genes between the primary tumours and the corresponding metastases of 25 patients. Epidermal growth factor receptor and <italic>K-RAS</italic> mutation status was different between primary tumours and corresponding metastases in 7 (28%) and 6 (24%) of the 25 patients, respectively. Among the 25 primary tumours, three ‘hotspot’ and two non-classical <italic>EGFR</italic> mutations were found; none of the corresponding metastases had the same mutation pattern. Among the five (20%) <italic>K-RAS</italic> mutations detected in the primary tumours, two were maintained in the corresponding metastasis. Epidermal growth factor receptor and <italic>K-RAS</italic> mutations were detected in the metastatic tumours of three (12%) and five (20%) patients, respectively. The expressions of EGFR and phosphorylated EGFR showed 10 and 50% discordance, in that order. We conclude that there is substantial discordance in <italic>EGFR</italic> and <italic>K-RAS</italic> mutational status between the primary tumours and corresponding metastases in patients with NSCLC and this might have therapeutic implications when treatment with TKIs is considered.</p>" ]

[ "<p>Lung cancer is the most frequent solid tumour and represents the leading cause of cancer death throughout the developed world. Almost 70% of patients with non-small-cell lung carcinoma (NSCLC) present with locally advanced or metastatic disease at the time of diagnosis. Non-small-cell lung carcinoma is characterised by the accumulation of multiple genetic alterations (##REF##15574780##Marsit et al, 2004##; ##REF##15016317##Yokota and Kohno, 2004##; ##REF##16187286##Garnis et al, 2006##). Mutations within the tyrosine kinase domain of epidermal growth factor receptor (<italic>EGFR</italic>) account for increased sensitivity to tyrosine kinase inhibitors (TKIs; gefitinib and erlotinib) and they are associated with prolonged overall survival (##REF##15118073##Lynch et al, 2004##; ##REF##15118125##Paez et al, 2004##; ##REF##15310767##Perez-Soler et al, 2004##; ##REF##15897572##Chou et al, 2005##; ##REF##16115929##Taron et al, 2005##; ##REF##17409888##Hirsch, 2006##). However, the point mutation T790M and an insertion mutation in exon 20 were associated with resistance to TKIs (##REF##16288556##Gazdar and Minna, 2005##). Furthermore, recent studies have shown that the expression of EGFR as assessed by gene copy number, mRNA and protein levels could be used to predict responsiveness to therapy with TKIs (##REF##15998906##Hirsch et al, 2005##; ##REF##16115929##Taron et al, 2005##; ##REF##16690485##Dacic et al, 2006##; ##REF##16707605##Dziadziuszko et al, 2006##; ##REF##16865253##Endo et al, 2006##). In addition, several clinicopathological characteristics, such as adenocarcinoma histology, non-smoking history, female gender and Asian origin, are also associated with a higher probability of response to TKIs, whereas the presence of <italic>K-RAS</italic> mutations seems to be correlated with primary resistance to these agents (##REF##15709185##Tokumo et al, 2005##; ##REF##16014883##Tsao et al, 2005##; ##REF##17060486##van Zandwijk et al, 2007##). Thus, an emerging issue concerning EGFR-targeted therapy is to identify the best method for selecting patients who are more likely to benefit from EGFR inhibition.</p>", "<p>Advanced NSCLC metastasises systemically to diverse sites, such as the brain, bone, adrenal glands and liver. The classical model for the metastatic process suggests that most cells of a given primary tumour have low metastatic potential and only a few cells acquire enough somatic mutations to become metastatic (##REF##12192390##Bernards and Weinberg, 2002##). An alternative model proposes that the metastatic potential is encoded in the mass of a given primary tumour that has progressed to a pre-metastatic state, after which metastases may randomly occur without any further gene expression changes (##REF##11823860##van ‘t Veer et al, 2002##; ##REF##12837240##Hynes, 2003##; ##UREF##0##Van't Veer and Weigelt, 2003##). Taking into account these two models, a critical issue for the treatment of metastatic NSCLC is the question of genetic variability and differences between the primary tumour and the corresponding metastases. In the majority of studies, <italic>EGFR</italic> and <italic>K-RAS</italic> status was determined on the primary tumours and there are very few data concerning those of corresponding metastases (##REF##16524970##Italiano et al, 2006##; ##REF##16642476##Matsumoto et al, 2006##). Therefore, it is unclear whether the same <italic>EGFR</italic> and <italic>K-RAS</italic> mutations are also present in the metastatic lesions or whether clones with different mutations are responsible for the generation of metastases.</p>", "<p>In this study, the mutation status of <italic>EGFR</italic> and <italic>K-RAS</italic> as well as the EGFR and p-EGFR expressions on the primary tumours and the corresponding metastatic lesions were evaluated in 25 patients with advanced NSCLC. The objective of this study was to investigate the prevalence of <italic>EGFR</italic> and <italic>K-RAS</italic> mutations in metastases and to examine whether these mutations and the EGFR expression patterns are discordant between the primary tumours and the corresponding metastases. Secondary objectives were to explore whether the EGFR expression pattern correlated to <italic>EGFR</italic> and/or <italic>K-RAS</italic> mutations in both the primary tumours and corresponding metastases.</p>" ]

[ "<p>This study was supported by a research grant from the Cretan Association for Biomedical Research (CABR).</p>" ]

[]

[ "<table-wrap id=\"tbl1\"><label>Table 1</label><caption><title>Patient's clinicopathological characteristics</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/><col align=\"left\"/><col align=\"center\"/><col align=\"left\"/><col align=\"left\"/><col align=\"left\"/><col align=\"center\"/><col align=\"left\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Case</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Age</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Sex</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Histology</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Stage<xref ref-type=\"fn\" rid=\"t1-fn2\">a</xref></bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Smoking status</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Tissue sample<xref ref-type=\"fn\" rid=\"t1-fn3\">b</xref> P<xref ref-type=\"fn\" rid=\"t1-fn4\">c</xref>/M<xref ref-type=\"fn\" rid=\"t1-fn5\">d</xref></bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Metastatic site</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Time<xref ref-type=\"fn\" rid=\"t1-fn6\">e</xref> elapsed between P and M</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Treatment administered between P and M</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 1</td><td align=\"center\" valign=\"top\" charoff=\"50\">60</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC</td><td align=\"center\" valign=\"top\" charoff=\"50\">IV</td><td align=\"left\" valign=\"top\" charoff=\"50\">Active</td><td align=\"left\" valign=\"top\" charoff=\"50\">B/B</td><td align=\"left\" valign=\"top\" charoff=\"50\">Skin</td><td align=\"center\" valign=\"top\" charoff=\"50\">10</td><td align=\"left\" valign=\"top\" charoff=\"50\">Taxane–platinum</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 2</td><td align=\"center\" valign=\"top\" charoff=\"50\">54</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">SCC</td><td align=\"center\" valign=\"top\" charoff=\"50\">II</td><td align=\"left\" valign=\"top\" charoff=\"50\">Active</td><td align=\"left\" valign=\"top\" charoff=\"50\">S/S</td><td align=\"left\" valign=\"top\" charoff=\"50\">Lung</td><td align=\"center\" valign=\"top\" charoff=\"50\">20</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 3</td><td align=\"center\" valign=\"top\" charoff=\"50\">70</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC</td><td align=\"center\" valign=\"top\" charoff=\"50\">III</td><td align=\"left\" valign=\"top\" charoff=\"50\">Former</td><td align=\"left\" valign=\"top\" charoff=\"50\">B/B</td><td align=\"left\" valign=\"top\" charoff=\"50\">Lung</td><td align=\"center\" valign=\"top\" charoff=\"50\">55</td><td align=\"left\" valign=\"top\" charoff=\"50\">Taxane–gemcitabine</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 4</td><td align=\"center\" valign=\"top\" charoff=\"50\">44</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC</td><td align=\"center\" valign=\"top\" charoff=\"50\">III</td><td align=\"left\" valign=\"top\" charoff=\"50\">Active</td><td align=\"left\" valign=\"top\" charoff=\"50\">S/B</td><td align=\"left\" valign=\"top\" charoff=\"50\">Lung</td><td align=\"center\" valign=\"top\" charoff=\"50\">65</td><td align=\"left\" valign=\"top\" charoff=\"50\">Taxane–platinum–gemcitabine</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 5</td><td align=\"center\" valign=\"top\" charoff=\"50\">55</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC</td><td align=\"center\" valign=\"top\" charoff=\"50\">III</td><td align=\"left\" valign=\"top\" charoff=\"50\">Active</td><td align=\"left\" valign=\"top\" charoff=\"50\">S/S</td><td align=\"left\" valign=\"top\" charoff=\"50\">Lung</td><td align=\"center\" valign=\"top\" charoff=\"50\">23</td><td align=\"left\" valign=\"top\" charoff=\"50\">Taxane–platinum</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 6</td><td align=\"center\" valign=\"top\" charoff=\"50\">63</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC</td><td align=\"center\" valign=\"top\" charoff=\"50\">IV</td><td align=\"left\" valign=\"top\" charoff=\"50\">Active</td><td align=\"left\" valign=\"top\" charoff=\"50\">B/B</td><td align=\"left\" valign=\"top\" charoff=\"50\">Lung</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 7</td><td align=\"center\" valign=\"top\" charoff=\"50\">66</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC/BAC</td><td align=\"center\" valign=\"top\" charoff=\"50\">III</td><td align=\"left\" valign=\"top\" charoff=\"50\">Never</td><td align=\"left\" valign=\"top\" charoff=\"50\">S/B</td><td align=\"left\" valign=\"top\" charoff=\"50\">Thoracic wall</td><td align=\"center\" valign=\"top\" charoff=\"50\">12</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 8</td><td align=\"center\" valign=\"top\" charoff=\"50\">57</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">LCC</td><td align=\"center\" valign=\"top\" charoff=\"50\">III</td><td align=\"left\" valign=\"top\" charoff=\"50\">Active</td><td align=\"left\" valign=\"top\" charoff=\"50\">S/B</td><td align=\"left\" valign=\"top\" charoff=\"50\">Thoracic wall</td><td align=\"center\" valign=\"top\" charoff=\"50\">4</td><td align=\"left\" valign=\"top\" charoff=\"50\">Taxane–platinum</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 9</td><td align=\"center\" valign=\"top\" charoff=\"50\">55</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC</td><td align=\"center\" valign=\"top\" charoff=\"50\">III</td><td align=\"left\" valign=\"top\" charoff=\"50\">Former</td><td align=\"left\" valign=\"top\" charoff=\"50\">S/S</td><td align=\"left\" valign=\"top\" charoff=\"50\">Thoracic wall</td><td align=\"center\" valign=\"top\" charoff=\"50\">15</td><td align=\"left\" valign=\"top\" charoff=\"50\">Taxane–platinum</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">10</td><td align=\"center\" valign=\"top\" charoff=\"50\">49</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC</td><td align=\"center\" valign=\"top\" charoff=\"50\">II</td><td align=\"left\" valign=\"top\" charoff=\"50\">Active</td><td align=\"left\" valign=\"top\" charoff=\"50\">S/S</td><td align=\"left\" valign=\"top\" charoff=\"50\">Adrenal gland</td><td align=\"center\" valign=\"top\" charoff=\"50\">28</td><td align=\"left\" valign=\"top\" charoff=\"50\">Taxane–platinum</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">11</td><td align=\"center\" valign=\"top\" charoff=\"50\">50</td><td align=\"center\" valign=\"top\" charoff=\"50\">F</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC</td><td align=\"center\" valign=\"top\" charoff=\"50\">IV</td><td align=\"left\" valign=\"top\" charoff=\"50\">Active</td><td align=\"left\" valign=\"top\" charoff=\"50\">B/B</td><td align=\"left\" valign=\"top\" charoff=\"50\">Brain</td><td align=\"center\" valign=\"top\" charoff=\"50\">36</td><td align=\"left\" valign=\"top\" charoff=\"50\">Taxane–platinum</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">12</td><td align=\"center\" valign=\"top\" charoff=\"50\">68</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC</td><td align=\"center\" valign=\"top\" charoff=\"50\">III</td><td align=\"left\" valign=\"top\" charoff=\"50\">Active</td><td align=\"left\" valign=\"top\" charoff=\"50\">S/S</td><td align=\"left\" valign=\"top\" charoff=\"50\">Brain</td><td align=\"center\" valign=\"top\" charoff=\"50\">10</td><td align=\"left\" valign=\"top\" charoff=\"50\">Taxane–platinum</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">13</td><td align=\"center\" valign=\"top\" charoff=\"50\">44</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">GCC</td><td align=\"center\" valign=\"top\" charoff=\"50\">IV</td><td align=\"left\" valign=\"top\" charoff=\"50\">Active</td><td align=\"left\" valign=\"top\" charoff=\"50\">B/B</td><td align=\"left\" valign=\"top\" charoff=\"50\">Lung</td><td align=\"center\" valign=\"top\" charoff=\"50\">74</td><td align=\"left\" valign=\"top\" charoff=\"50\">Taxane–platinum</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">14</td><td align=\"center\" valign=\"top\" charoff=\"50\">56</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC</td><td align=\"center\" valign=\"top\" charoff=\"50\">IV</td><td align=\"left\" valign=\"top\" charoff=\"50\">Active</td><td align=\"left\" valign=\"top\" charoff=\"50\">B/S</td><td align=\"left\" valign=\"top\" charoff=\"50\">Adrenal gland</td><td align=\"center\" valign=\"top\" charoff=\"50\">17</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">15</td><td align=\"center\" valign=\"top\" charoff=\"50\">53</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC</td><td align=\"center\" valign=\"top\" charoff=\"50\">III</td><td align=\"left\" valign=\"top\" charoff=\"50\">Active</td><td align=\"left\" valign=\"top\" charoff=\"50\">S/B</td><td align=\"left\" valign=\"top\" charoff=\"50\">Thoracic wall</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"left\" valign=\"top\" charoff=\"50\">Taxane–platinum</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">16</td><td align=\"center\" valign=\"top\" charoff=\"50\">41</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC</td><td align=\"center\" valign=\"top\" charoff=\"50\">IV</td><td align=\"left\" valign=\"top\" charoff=\"50\">Active</td><td align=\"left\" valign=\"top\" charoff=\"50\">S/B</td><td align=\"left\" valign=\"top\" charoff=\"50\">Lung</td><td align=\"center\" valign=\"top\" charoff=\"50\">143</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">17</td><td align=\"center\" valign=\"top\" charoff=\"50\">56</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC</td><td align=\"center\" valign=\"top\" charoff=\"50\">IV</td><td align=\"left\" valign=\"top\" charoff=\"50\">Former</td><td align=\"left\" valign=\"top\" charoff=\"50\">B/S</td><td align=\"left\" valign=\"top\" charoff=\"50\">Adrenal gland</td><td align=\"center\" valign=\"top\" charoff=\"50\">36</td><td align=\"left\" valign=\"top\" charoff=\"50\">Taxane–platinum–gemcitabine, gefitinib</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">18</td><td align=\"center\" valign=\"top\" charoff=\"50\">42</td><td align=\"center\" valign=\"top\" charoff=\"50\">F</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC</td><td align=\"center\" valign=\"top\" charoff=\"50\">II</td><td align=\"left\" valign=\"top\" charoff=\"50\">Never</td><td align=\"left\" valign=\"top\" charoff=\"50\">S/S</td><td align=\"left\" valign=\"top\" charoff=\"50\">Liver</td><td align=\"center\" valign=\"top\" charoff=\"50\">30</td><td align=\"left\" valign=\"top\" charoff=\"50\">Taxane–platinum, gefitinib</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">19</td><td align=\"center\" valign=\"top\" charoff=\"50\">55</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC</td><td align=\"center\" valign=\"top\" charoff=\"50\">III</td><td align=\"left\" valign=\"top\" charoff=\"50\">Former</td><td align=\"left\" valign=\"top\" charoff=\"50\">S/B</td><td align=\"left\" valign=\"top\" charoff=\"50\">Bone</td><td align=\"center\" valign=\"top\" charoff=\"50\">2</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">20</td><td align=\"center\" valign=\"top\" charoff=\"50\">46</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">SCC</td><td align=\"center\" valign=\"top\" charoff=\"50\">III</td><td align=\"left\" valign=\"top\" charoff=\"50\">Active</td><td align=\"left\" valign=\"top\" charoff=\"50\">S/B</td><td align=\"left\" valign=\"top\" charoff=\"50\">Lung</td><td align=\"center\" valign=\"top\" charoff=\"50\">45</td><td align=\"left\" valign=\"top\" charoff=\"50\">Taxane–platinum–gefitinib</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">21</td><td align=\"center\" valign=\"top\" charoff=\"50\">62</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">LCC</td><td align=\"center\" valign=\"top\" charoff=\"50\">III</td><td align=\"left\" valign=\"top\" charoff=\"50\">Never</td><td align=\"left\" valign=\"top\" charoff=\"50\">S/B</td><td align=\"left\" valign=\"top\" charoff=\"50\">Bone</td><td align=\"center\" valign=\"top\" charoff=\"50\">48</td><td align=\"left\" valign=\"top\" charoff=\"50\">Platinum–gemcitabine</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">22</td><td align=\"center\" valign=\"top\" charoff=\"50\">67</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC</td><td align=\"center\" valign=\"top\" charoff=\"50\">IV</td><td align=\"left\" valign=\"top\" charoff=\"50\">Active</td><td align=\"left\" valign=\"top\" charoff=\"50\">B/S</td><td align=\"left\" valign=\"top\" charoff=\"50\">Adrenal gland</td><td align=\"center\" valign=\"top\" charoff=\"50\">14</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">23</td><td align=\"center\" valign=\"top\" charoff=\"50\">53</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC</td><td align=\"center\" valign=\"top\" charoff=\"50\">IV</td><td align=\"left\" valign=\"top\" charoff=\"50\">Active</td><td align=\"left\" valign=\"top\" charoff=\"50\">S/B</td><td align=\"left\" valign=\"top\" charoff=\"50\">Brain</td><td align=\"center\" valign=\"top\" charoff=\"50\">21</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">24</td><td align=\"center\" valign=\"top\" charoff=\"50\">52</td><td align=\"center\" valign=\"top\" charoff=\"50\">F</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC/BAC</td><td align=\"center\" valign=\"top\" charoff=\"50\">II</td><td align=\"left\" valign=\"top\" charoff=\"50\">Active</td><td align=\"left\" valign=\"top\" charoff=\"50\">S/S</td><td align=\"left\" valign=\"top\" charoff=\"50\">Lung</td><td align=\"center\" valign=\"top\" charoff=\"50\">51</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">25</td><td align=\"center\" valign=\"top\" charoff=\"50\">63</td><td align=\"center\" valign=\"top\" charoff=\"50\">M</td><td align=\"left\" valign=\"top\" charoff=\"50\">ADC</td><td align=\"center\" valign=\"top\" charoff=\"50\">IV</td><td align=\"left\" valign=\"top\" charoff=\"50\">Active</td><td align=\"left\" valign=\"top\" charoff=\"50\">B/B</td><td align=\"left\" valign=\"top\" charoff=\"50\">Thoracic wall</td><td align=\"center\" valign=\"top\" charoff=\"50\">1</td><td align=\"left\" valign=\"top\" charoff=\"50\">None</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl2\"><label>Table 2</label><caption><title><italic>EGFR</italic> and <italic>K-RAS</italic> status in paired primary and metastatic tumours</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"left\"/><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"left\"/><col align=\"left\"/><col align=\"left\"/><col align=\"left\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th colspan=\"2\" align=\"center\" valign=\"top\" charoff=\"50\">\n<bold><italic>EGFR</italic> mutation status</bold>\n<hr/></th><th colspan=\"2\" align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>EGFR expression</bold>\n<hr/></th><th colspan=\"2\" align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>p-EGFR expression</bold>\n<hr/></th><th colspan=\"2\" align=\"center\" valign=\"top\" charoff=\"50\">\n<bold><italic>K-RAS</italic> mutation status</bold>\n<hr/></th><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th align=\"left\" valign=\"top\" charoff=\"50\"> </th></tr><tr><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Case</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Primary</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Metastasis</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Primary</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Metastasis</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Primary</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Metastasis</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Primary</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Metastasis</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Gefitinib</bold>\n</th><th align=\"left\" valign=\"top\" charoff=\"50\">\n<bold>Response</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 1</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">1+</td><td align=\"center\" valign=\"top\" charoff=\"50\">1+</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">2+</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 2</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">3+</td><td align=\"center\" valign=\"top\" charoff=\"50\">3+</td><td align=\"center\" valign=\"top\" charoff=\"50\">1+</td><td align=\"center\" valign=\"top\" charoff=\"50\">1+</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 3</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">1+</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 4</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">3+</td><td align=\"center\" valign=\"top\" charoff=\"50\">2+</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">1+</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 5</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 6</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">1+</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 7</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">1+</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 8</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">1+</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"left\" valign=\"top\" charoff=\"50\">G12S</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\"> 9</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">1+</td><td align=\"center\" valign=\"top\" charoff=\"50\">1+</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">1+</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">G13S</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">10</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">2+</td><td align=\"center\" valign=\"top\" charoff=\"50\">3+</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"left\" valign=\"top\" charoff=\"50\">G12V</td><td align=\"left\" valign=\"top\" charoff=\"50\">G12V</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">11</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">2+</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">2+</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">G12S</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">12</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">L692P V717A</td><td align=\"center\" valign=\"top\" charoff=\"50\">1+</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">2+</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">13</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">T847A</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">1+</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">14</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">3+</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">1+</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">Yes</td><td align=\"left\" valign=\"top\" charoff=\"50\">PD</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">15</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">2+</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">2+</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">Yes</td><td align=\"left\" valign=\"top\" charoff=\"50\">PD</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">16</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">2+</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">G12A</td><td align=\"left\" valign=\"top\" charoff=\"50\">Yes</td><td align=\"left\" valign=\"top\" charoff=\"50\">PD</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">17</td><td align=\"left\" valign=\"top\" charoff=\"50\">L692P</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">1+</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">Yes</td><td align=\"left\" valign=\"top\" charoff=\"50\">SD</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">18</td><td align=\"left\" valign=\"top\" charoff=\"50\">E746V</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">Yes</td><td align=\"left\" valign=\"top\" charoff=\"50\">SD</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">19</td><td align=\"left\" valign=\"top\" charoff=\"50\">G857E</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"center\" valign=\"top\" charoff=\"50\">1+</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">Yes</td><td align=\"left\" valign=\"top\" charoff=\"50\">SD</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">20</td><td align=\"left\" valign=\"top\" charoff=\"50\">E746-A750 del</td><td align=\"left\" valign=\"top\" charoff=\"50\">E746-A750 del T790M</td><td align=\"center\" valign=\"top\" charoff=\"50\">2+</td><td align=\"center\" valign=\"top\" charoff=\"50\">3+</td><td align=\"center\" valign=\"top\" charoff=\"50\">2+</td><td align=\"center\" valign=\"top\" charoff=\"50\">0</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">Yes</td><td align=\"left\" valign=\"top\" charoff=\"50\">SD</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">21</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">22</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"left\" valign=\"top\" charoff=\"50\">G12D</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">Yes</td><td align=\"left\" valign=\"top\" charoff=\"50\">PD</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">23</td><td align=\"left\" valign=\"top\" charoff=\"50\">E746-A750 del</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"left\" valign=\"top\" charoff=\"50\">G12C</td><td align=\"left\" valign=\"top\" charoff=\"50\">G12C</td><td align=\"left\" valign=\"top\" charoff=\"50\">Yes</td><td align=\"left\" valign=\"top\" charoff=\"50\">SD</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">24</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">25</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"center\" valign=\"top\" charoff=\"50\">ND</td><td align=\"left\" valign=\"top\" charoff=\"50\">G12C</td><td align=\"left\" valign=\"top\" charoff=\"50\">wt</td><td align=\"left\" valign=\"top\" charoff=\"50\">No</td><td align=\"left\" valign=\"top\" charoff=\"50\">—</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl3\"><label>Table 3</label><caption><title>Combined analysis of <italic>EGFR</italic> and <italic>K-RAS</italic> mutation status</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th colspan=\"4\" align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Primary/metastatic tumour</bold>\n<hr/></th><th align=\"center\" valign=\"top\" charoff=\"50\"> </th></tr><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>wt/wt</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>wt/mut</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>mut/wt</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>mut/mut</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Discordance</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">EGFR</td><td align=\"center\" valign=\"top\" charoff=\"50\">18 (72%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">2 (8%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">4 (16%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (4%)<xref ref-type=\"fn\" rid=\"t3-fn1\">a</xref></td><td align=\"center\" valign=\"top\" charoff=\"50\">7 cases (28%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">K-RAS</td><td align=\"center\" valign=\"top\" charoff=\"50\">17 (68%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">3 (12%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">3 (12%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">2 (8%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">6 cases (24%)</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"tbl4\"><label>Table 4</label><caption><title>EGFR expression as assessed by IHC</title></caption><table frame=\"hsides\" rules=\"groups\" border=\"1\"><colgroup><col align=\"left\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/><col align=\"center\"/></colgroup><thead valign=\"bottom\"><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th colspan=\"4\" align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Primary tumour status/metastasis status</bold>\n<hr/></th><th align=\"center\" valign=\"top\" charoff=\"50\"> </th></tr><tr><th align=\"left\" valign=\"top\" charoff=\"50\"> </th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>+ve/+ve</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>−ve/−ve</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>+ve/−ve</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>−ve/+ve</bold>\n</th><th align=\"center\" valign=\"top\" charoff=\"50\">\n<bold>Discordance</bold>\n</th></tr></thead><tbody valign=\"top\"><tr><td align=\"left\" valign=\"top\" charoff=\"50\">EGFR</td><td align=\"center\" valign=\"top\" charoff=\"50\">4 (21.1%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">13 (68.4%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (5.3%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (5.3%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">2/19 (10.6%)</td></tr><tr><td align=\"left\" valign=\"top\" charoff=\"50\">p-EGFR (Y1173)</td><td align=\"center\" valign=\"top\" charoff=\"50\">1 (6.2%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">7 (43.8%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">2 (12.5%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">6 (37.5%)</td><td align=\"center\" valign=\"top\" charoff=\"50\">8/16 (50%)</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><fn id=\"t1-fn1\"><p>ADC=adenocarcinoma; ADC/BAC=adenocarcinoma with bronchoalveolar features; GCC=giant cell carcinoma; LCC=large cell carcinoma; F=female; M=male; SCC=squamous carcinoma.</p></fn><fn id=\"t1-fn2\"><label>a</label><p>Stage, corresponds to that of the time of primary diagnosis.</p></fn><fn id=\"t1-fn3\"><label>b</label><p>Tissue sample: B, biopsy; S, surgery.</p></fn><fn id=\"t1-fn4\"><label>c</label><p>P, primary tumour.</p></fn><fn id=\"t1-fn5\"><label>d</label><p>M, metastasis.</p></fn><fn id=\"t1-fn6\"><label>e</label><p>Time, months.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"t3-fn1\"><label>a</label><p>Del19/Del19 and T790M.</p></fn></table-wrap-foot>", "<table-wrap-foot><fn id=\"t4-fn1\"><p>IHC=immunohistochemistry.</p></fn></table-wrap-foot>" ]

[]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>C-reactive protein (CRP) is the prototypical acute phase protein in humans [##REF##17452737##1##,##REF##16214080##2##]. It is widely used as a marker of infection, and also to aid the differentiation between a bacterial and a viral origin. Many geriatricians have the clinical impression that the CRP response to serious invasive bacterial infections may be delayed in frail old patients. This view is not supported in the literature [##REF##17452737##1##], but studies in this field are rather scanty. Some relatively small studies exist [##REF##9177673##3##, ####REF##4003173##4##, ##REF##3776747##5####3776747##5##], but were not designed to compare the diagnostic sensitivity for bacterial infections in different ages, and do not give information on the proportion of false-negative tests.</p>", "<p>CRP is mainly produced by the liver [##REF##10852144##6##]. The synthetic function of the liver may be reduced due to physiological aging, possibly due to both decreases in liver volume, liver blood flow and perfusion, and to cellular changes [##UREF##0##7##]. The main stimulator of CRP synthesis in the liver is interleukin 6 (IL-6), which is associated with diseases of old age such as multiple myelomas, chronic lymphatic leukaemia and renal cancer [##REF##8218595##8##], and is higher in older than in younger subjects [##REF##1453878##9##]. In mice the regulation of IL-6 synthesis is dependent of dehydroepiandrosterone, a hormone that declines with age, and the dysregulation is reversed by replacement of the hormone [##REF##8515056##10##]. Studies have found a delayed IL-6 response in infection in the old [##REF##10395881##11##], as well as impaired production of proinflammatory cytokines [##REF##10540184##12##]. Thus, a diminished or delayed CRP response to bacterial infections in the old may be biologically plausible. The issue is clinically important, as the atypical clinical symptoms and signs of infection in geriatric patients make a correct diagnosis more dependent on a rational use of supplementary tests.</p>", "<p>Hogarth et al suggest a cut-off value of CRP at 40 mg/L to be sufficient in elderly patients with infection [##REF##9177673##13##], whereas one study has recommended that the cut-off value indicating sepsis, irrespective of age, should be 80 mg/L [##REF##15378239##14##]. Even higher cut-off levels are also commonly used in clinical decisions.</p>", "<p><italic>E. coli </italic>is the most frequent Gram-negative species in bloodstream infections [##REF##15551210##15##], whereas <italic>S. pneumoniae </italic>is the leading cause of community-acquired pneumonia and a significant cause of bacteraemia and meningitis [##REF##2739613##16##]. Bacteraemia with either of these species will in almost every instance represent a clinically important invasive infection in need of antibacterial treatment. A \"normal\" CRP-concentration in these conditions can therefore be considered a false negative test result if CRP should be regarded as an indicator of bacterial infection.</p>", "<p>The main aim of this study was to assess the diagnostic sensitivity of a raised CRP-concentration for bacteraemia with either <italic>E. coli </italic>or <italic>S. pneumoniae </italic>in different age groups. We also studied the impact of different cut-off points for the CRP-concentration as well as differences in the time course of the CRP-response in relation to age.</p>" ]

[ "<title>Methods</title>", "<title>Setting and selection of patients</title>", "<p>Aker University Hospital is a 350-bed hospital in Oslo, Norway, serving a population of 500 000 for urology and abdominal vascular surgery, and 180 000 for internal medicine, general surgery and psychiatry. Until the late 1990's, the hospital served the latter population in gynaecology, obstetrics and paediatrics as well.</p>", "<p>All adult (≥ 16 years) patients with culture-verified bacteraemia of <italic>E. coli </italic>or <italic>S. pneumoniae </italic>during the period 1994–2004 were retrieved from the database of the hospital's bacteriological laboratory. Patients having more than one episode of bacteraemia were registered only once. For analytical purposes, the patients were divided into three age groups, &lt; 65, 65–84, and ≥ 85, respectively. To be able to evaluate the need for age differentiated cut points for CRP, we selected patients admitted to the Medical department with ICD 10 diagnoses corresponding to either acute myocardial or acute cerebral infarction during 2003 to serve as controls. Patients with any ICD10 diagnoses indicating severe infection or a positive blood culture with a significant bacterium were, however, not used as controls.</p>", "<title>Microbiological methods</title>", "<p>Blood was cultured by the Bactec 9240 system (Becton &amp; Dickinson, Sparks, MD, USA). Further identification of the positive blood cultures followed standard procedures of the laboratory. Growth of possible <italic>S. pneumoniae </italic>was confirmed by a typical phenotype, sensitivity to optochin and by an agglutination test (Slidex pneumo-Kit, BioMériux, Lyon, France). <italic>E. coli </italic>was identified by a three tube method for identification of Gram-negative rods [##REF##1106114##17##], or by the api20E system (BioMériux).</p>", "<title>CRP analysis</title>", "<p>From the databases of the clinical chemistry laboratory, we retrieved the CRP concentration in blood drawn on the same date as the blood culture, if applicable (CRP1). When available, we also retrieved one CRP value measured on the second or third day of the hospital stay (CRP2), and one retrieved between day four and day seven (CRP3). All the CRP-measurements were performed on automated clinical chemistry analyzers (Roche Diagnostics, Mannheim, Germany): first on Hitachi 717, then Hitachi 917 and then on Modular. The assay vendor in the actual period was Roche diagnostics. The assay is based on an immunoturbidometric principle, where CRP is agglutinated with anti-CRP. In 2004 the assay was changed to a latex-particle-bound-anti-CRP reagent, which gave an enhanced agglutination signal with CRP. No significant shift in results could be detected at the time of change, comparing the old and new methods in selected clinical specimens. The analytic stability over the time period was documented by stable values of daily long-time quality control specimens. The reference range was unchanged during the period.</p>", "<p>In order to test for any shift in the analyses over time in the present material, we estimated the Pearson correlation coefficient between time of testing and CRP concentration in a subsample of 681 patients. The correlation coefficient was non-significant at 0.060 (p = 0.12). When sub grouping the material according to time of change of analyzers and type of reagent kits, no statistically significant differences were found neither in mean ranks (p = 0.065) nor in medians (p = 0.398).</p>", "<title>Statistical methods</title>", "<p>Since CRP-data and age were not normally distributed, data are presented as medians and interquartile ranges (IQR), and Mann-Whitney U tests were used to compare the ranks of age in the bacteraemic patient group versus the control group, and to compare the ranks of CRP1, CRP2 and CRP3 in the three different bacteraemic patient age groups. The Spearman rank correlation was used to examine a possible linear association between age and CRP1. The medians and IQRs of CRP1, CRP2 and CRP3 were depicted graphically. Contingency tables were used to calculate the sensitivities of CRP1 according to age group, bacteria, gender, and cut-off value (40, 80 and 120 mg/l, respectively). The confidence intervals of the sensitivities were manually calculated, and Chi-squared tests were used to compare the observed proportions. A 10% difference in sensitivity was considered as clinically relevant. A power analysis indicated that 294 patients in each age group would be required to detect such a difference with a power of 80% and a 5% significance level. Receiver operating characteristics (ROC) curves for the three age groups were constructed, and area under the curves (AUC) was calculated. Z-tests were used to assess whether the AUCs were statistically significantly different.</p>", "<p>All tests were computed using SPSS -15.0 (SPSS Inc. Chicago, IL, USA), or by hand. P-values less than 0.05 were regarded as a sign of statistical significance. No attempts were done to adjust for multiple comparisons.</p>" ]

[ "<title>Results</title>", "<p>1150 patients had a positive blood culture with either <italic>E. coli </italic>or <italic>S. pneumoniae </italic>during the study period. Of these, 260 had no registered CRP measurement on the same day as the blood culture, reducing the number of eligible patients to 890. Of these, 604 (68%) had had their blood culture and CRP drawn on the day of hospital admittance. 634 (71%) of the patients had <italic>E. coli </italic>whereas the remaining 256 had <italic>S. pneumoniae</italic>. Median age of bacteraemic patients was 75 years (IQR 58–82). 300 (34%) patients were &lt; 65 years, 443 (49%) were 65–84 years, and 147 (17%) were ≥ 85 years. The percentages of <italic>E. coli </italic>in the three age groups were 61%, 77% and 76%, respectively (p =&lt; 0.01), and the percentages of females were 58%, 50% and 71% (p =&lt; 0.01), respectively. 421 control patients were eligible. Their median age was 77 years (IQR 62–83), which was significantly higher than in the bacteraemic group (p = 0.005). 120 (28%) of the controls were &lt; 65 years, 223 (52%) were 65–84 years, and 84 (20%) were ≥ 85 years.</p>", "<p>The median concentration of CRP on the same day as the positive blood culture (CRP1) was 188 mg/L (IQR 97–288 mg/L). For the control group taken together the median CRP concentration at admittance was 6 mg/L (IQR 2–19 mg/L, range 1–303 mg/L), 5 mg/L (IQR 2–12 mg/L) for those with cerebral infarction and 7 mg/L (IQR 2–24 mg/L) for those with myocardial infarction. The relation between CRP1 and age with trendlines is shown in figure ##FIG##0##1##. The Spearman correlation coefficient was – 0.072 (p = 0.032) and 0.146 (p = 0.003) among the cases and the controls, respectively. For bacteraemic patients the median CRP1 was 209.5 (IQR 110–320), 175 (IQR 83–271) and 176 mg/L (IQR 103–278), in the youngest, median and oldest age group, respectively. Corresponding results among the controls were 4 (IQR 2 – 11), 6 (IQR 2 – 20) and 11 mg/L (IQR 3 – 43).</p>", "<p>For bacteraemic patients, the medians and IQRs of CRP1, CRP2 and CRP3 in the three age groups are shown in Figure ##FIG##1##2##. The youngest group had a significantly higher CRP1 level than the intermediate group (p = 0.003). For CRP2 both the difference between the youngest and the intermediate group (p = 0.04) and the difference between the youngest and the oldest group (p = 0.026), were statistically significant. There were no significant differences for CRP3.</p>", "<p>The sensitivity of CRP in detection of bacteraemia according to age group, bacterial species and gender at different cut-off values is shown in table ##TAB##0##1##. At all cut-off points, the sensitivity is significantly better for <italic>S. pneumoniae </italic>than for <italic>E. coli</italic>, but we were not able to identify any differences in diagnostic sensitivity regarding age groups.</p>", "<p>The ROC-curves for the different age groups are shown in figure ##FIG##2##3##. The AUCs are 0.968 (95% confidence interval (CI) 0.952–0.985), 0.940 (95% CI 0.922–0.958) and 0.923 (95% CI 0.888–0.958) for the age groups 16–64 years, 65–84 years, and ≥ 85 years, respectively. The AUC for the youngest age group was significantly greater than that for the intermediate as well as the oldest age group (z-test, p &lt; 0.05 for both comparisons).</p>" ]

[ "<title>Discussion</title>", "<p>We were not able to confirm the hypothesis that the CRP-response in bacteraemia is age-dependent to any clinically important degree. There was a slight, though statistically significant negative correlation between age and initial CRP value as well as a higher area under the ROC curve for the youngest patients, but this did not translate into a clinically meaningful difference in sensitivity for any relevant cut-off value for CRP. The study may have been underpowered for a comparison of sensitivity, but when the age groups were collapsed into broader classes (results not shown), thus improving the power, the negative results regarding sensitivity persisted.</p>", "<p>Accordingly, this study supports the use of CRP as an indicator of possible infection in the elderly with the same cut-off values as in younger adults. This result should, however, be interpreted with several reservations.</p>", "<p>First, we cannot completely rule out selection bias, since we do not know if the blood cultures in some instances were drawn after the results of the CRP-tests were known, thus increasing the possibility that some cases of bacteraemia remained undiagnosed in cases with low CRP. Moreover, fever is a very common indication for having a blood culture drawn, but is less frequent in elderly patients with infection [##REF##3905926##18##]. Both factors could bias the results towards higher CRP levels in the elderly patients, and this trend may be strengthened by the fact that elderly patients more often have non-infectious comorbid conditions characterized by high CRP concentrations.</p>", "<p>Second, we lack information on the duration of symptoms prior to CRP testing. A substantial proportion of the patients had their CRP1 drawn at admission. If there exists a systematically increased pre-hospital delay among the oldest, a slower CRP response in this group could be masked.</p>", "<p>Third, we have insufficient clinical information as to include the potential confounding factors of comorbidity, disability or frailty upon the CRP response, an association that has been suggested by others [##REF##16899118##19##,##REF##16611705##20##]. Neither do we have sufficient information on the clinical severity of the infection.</p>", "<p>It should be noted that our results are valid only for bacteraemia with <italic>E. coli </italic>or <italic>S. pneumoniae</italic>. Our study was not designed to assess the sensitivity of the CRP-response for severe bacterial infections that does not result in a positive blood culture for one of these bacteria. Thus, one should be careful in generalising our results to different clinical settings. Furthermore, the choice of control group can be criticized as both cerebral and coronary infarction may lead to elevated CRP. However, the choice was based on clinical grounds as these two conditions represent the main alternative diagnoses in the emergency room, especially when dealing with elderly patients.</p>", "<p>The small relative difference between the age strata persisted at day 2–3, but disappeared at day 4–7 (Figure ##FIG##1##2##). This weighs against the hypothesis that the CRP response among the oldest is delayed, as in this instance we would have expected the progress from the first to the second assessment to differ according to age. It also weighs against the hypothesis of a prolonged inflammatory response in human old age [##REF##10395881##11##], as is seen in aged mice [##REF##17849263##21##]. There may, however, have been a selection bias in the ordering of a second and a third measurement. We have no information on the reasons why these repeated measurements were ordered.</p>", "<p>Thus, taking these important reservations into account, there is still a real possibility that there may exist a clinically relevant age related difference in CRP response in systemic infection, at least in invasive bacterial infections not causing a positive blood culture. What factors are in favour of this hypothesis?</p>", "<p>In ageing research it is widely accepted that ageing is characterized by a systemic, low-grade inflammatory state [##REF##17173699##22##]. It is also acknowledged that this subclinical proinflammatory status, interacting with genetic background (evolutionary selection of proinflammatory genes beneficial in early age), may trigger the onset of frailty and age-associated diseases [##REF##17173699##22##, ####REF##17118425##23##, ##REF##17683521##24####17683521##24##]. Innate immunity is viewed as being mainly well preserved, but immune changes seen in old age also affect innate immunity [##REF##17683521##24##]. Studies on possible age-related changes in proinflammatory cytokines (and thus the CRP response) in endotoxiemia have yielded diverging results [##REF##11238217##25##, ####REF##8568256##26##, ##REF##8641780##27##, ##REF##14659593##28####14659593##28##]. A recent study on gene expression in endotoxin-stimulated macrophages from young and old mice showed reduced expression of proinflammatory cytokines as well as Toll like receptor signalling pathways with increasing age [##REF##16603589##29##]. Another study showed that the gene expression of proinflammatory cytokines was significantly lower in aged mice than in young at 24 hours after stimulation with endotoxin, but higher at 72 hours [##REF##17849263##21##]. Thus, although still controversial, experimental studies do open up for a weaker proinflammatory response and thus a limited production of CRP.</p>", "<p>There seems to be a convincing and clinically important difference in the sensitivity of CRP between patients infected with <italic>E. coli </italic>and <italic>S. pneumoniae</italic>. This finding accords well with the results of other studies. The CRP concentration is associated with mortality and organ failure [##REF##12796187##30##], and it is well known that <italic>S. pneumonia </italic>consistently causes serious infections, while the range of severity is much broader for infections with <italic>E. coli</italic>. The finding of a gender difference in sensitivity in one particular age stratum, on the other hand, seems not biologically plausible, and we suspect this association to be spurious.</p>" ]

[ "<title>Conclusion</title>", "<p>We conclude that there exists a statistically significant association between age and CRP in invasive bacterial infections. However, this association may not be of a magnitude that makes it clinically important for CRP as a diagnostic tool in bacterial infections associated with a positive blood culture. Further studies are needed to decide whether markers of the aging process such as comorbidity and frailty influence the diagnostic value of CRP as a marker of infection, as well as whether the diagnostic performance of CRP is affected by severity of infection.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>C-reactive protein (CRP) is an indicator of inflammation, and is often used in the diagnosis of bacterial infections. It is poorly known whether CRP in bacterial infection is age-dependent.</p>", "<title>Methods</title>", "<p>Adult patients with a positive blood culture with <italic>E. coli </italic>or <italic>S. pneumoniae </italic>during 1994–2004 were included. CRP measured on the same date as the blood cultures were drawn (CRP1), 2–3 days (CRP2) and 4–7 days later (CRP3), were retrieved. The patients were divided into three age groups, &lt; 65, 65–84, and ≥ 85, respectively. We studied three cut-off values for CRP and produced age-specific receiver operating characteristics (ROC) curves, using patients with acute coronary or cerebral infarction as controls.</p>", "<title>Results</title>", "<p>890 patients and 421 controls were available. There was a statistically significant negative correlation between age and CRP1 – 0.072 (p = 0.032). The median CRP1 and CRP2 were significantly higher in the youngest age group. The area under the ROC-curve for the youngest age group was significantly greater than that of the two other age groups, but we found no statistically significant differences in sensitivity related to age. The diagnostic sensitivity of CRP was better for <italic>S. pneumoniae </italic>than for <italic>E. coli</italic>, 92.6% vs. 88.0% (p = 0.046) for a cut-off value of 40 mg/L, and 82.4% vs. 61.5% (p =&lt; 0.01) for a cut-off value of 120 mg/L.</p>", "<title>Conclusion</title>", "<p>CRP is better in identifying infection with <italic>S. pneumoniae </italic>than with <italic>E. coli</italic>. We found a weakening of the CRP-response with age, but this is hardly of clinical significance.</p>" ]

[ "<title>Ethical considerations</title>", "<p>The study was approved by the Regional Committee for Ethics in Medical Research and by the Norwegian Data Inspectorate, which gave permission to carry out the study without the patients' consent.</p>", "<title>Competing interests</title>", "<p>This study was financed by Aker university hospital and by Ullevaal university hospital. The authors have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>ALW participated in all parts of the study, both in the design of the study, data collection, statistical analysis and interpretation, and in the writing of the manuscript. KGB participated in performing the statistical analysis, the interpretation of the data and in the writing of the manuscript. TBW participated in the design of the study, the interpretation of the statistical analysis and in the writing of the manuscript, including revising it critically for important intellectual content.</p>", "<p>All authors have read and approved the final manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>We thank Professor Signe Holta Ringertz for the initial idea of design of the study and Dr. Kari Lohne at the clinical chemistry laboratory for the possibility to obtain data. We also want to thank Dr. Solve Tjora for clarifying comments on quality control of CRP analysis and Professor Kjetil K. Melby for the opportunity to accomplish the study.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Scatter plot of the relation between age and CRP concentration on the day of a blood culture positive for <italic>S. pneumoniae </italic>or <italic>E. coli </italic>(cases), and for the controls.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>Median CRP1 (on the day of the positive blood culture), CRP2 (2–3 days later), and CRP3 (4–7 days later) in different age groups. The vertical bars indicate the interquartile range.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Receiver operating characteristics (ROC) curves for the relationship between initial CRP value and bacteraemia in different age groups</bold>. The sensitivity and the corresponding 1-specificity for the cut-points 40, 80 and 120 mg/L are indicated by circles, squares and diamonds, respectively.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Sensitivity of a \"high\" concentration of C-reactive protein (CRP) as indicator of bacteraemia by gender, species, and age</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td/><td align=\"center\" colspan=\"2\"><bold>Sex</bold></td><td align=\"center\" colspan=\"2\"><bold>Species</bold></td><td align=\"center\" colspan=\"3\"><bold>Age group</bold></td></tr><tr><td/><td/><td align=\"center\">Female</td><td align=\"center\">Male</td><td align=\"center\"><italic>S. pneumoniae</italic></td><td align=\"center\"><italic>E. coli</italic></td><td align=\"center\">16–64</td><td align=\"center\">65–84</td><td align=\"center\">≥ 85</td></tr></thead><tbody><tr><td align=\"left\"><bold>Cut-off (CRP) 40 mg/L</bold></td><td align=\"left\">Sensitivity (%)</td><td align=\"center\">90.2</td><td align=\"center\">88.2</td><td align=\"center\">92.6</td><td align=\"center\">88.0</td><td align=\"center\">90.4</td><td align=\"center\">88.5</td><td align=\"center\">90.5</td></tr><tr><td/><td align=\"left\">95% CI</td><td align=\"center\">87.6 – 92.8</td><td align=\"center\">85.0 – 91.4</td><td align=\"center\">89.4 – 95.8</td><td align=\"center\">85.5 – 90.5</td><td align=\"center\">86.6 – 93.4</td><td align=\"center\">85.5 – 91.5</td><td align=\"center\">85.8 – 95.2</td></tr><tr><td/><td align=\"left\">p-value</td><td align=\"center\" colspan=\"2\">0.339</td><td align=\"center\" colspan=\"2\"><bold>0.046</bold></td><td align=\"center\" colspan=\"3\">0.714</td></tr><tr><td/><td/><td/><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"><bold>Cut-off (CRP) 80 mg/L</bold></td><td align=\"left\">Sensitivity (%)</td><td align=\"center\">81.4</td><td align=\"center\">75.6</td><td align=\"center\">89.1</td><td align=\"center\">74.8</td><td align=\"center\">82.0</td><td align=\"center\">75.6</td><td align=\"center\">82.3</td></tr><tr><td/><td align=\"left\">95% CI</td><td align=\"center\">78.0 – 84.8</td><td align=\"center\">71.3 – 79.9</td><td align=\"center\">84.8 – 93.4</td><td align=\"center\">71.4 – 78.2</td><td align=\"center\">77.7 – 86.3</td><td align=\"center\">71.6 – 79.6</td><td align=\"center\">76.1 – 88.5</td></tr><tr><td/><td align=\"left\">p-value</td><td align=\"center\" colspan=\"2\"><bold>0.037</bold></td><td align=\"center\" colspan=\"2\"><bold>&lt; 0.01</bold></td><td align=\"center\" colspan=\"3\">0.060</td></tr><tr><td/><td/><td/><td/><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"><bold>Cut-off (CRP) 120 mg/L</bold></td><td align=\"left\">Sensitivity (%)</td><td align=\"center\">69.7</td><td align=\"center\">64.6</td><td align=\"center\">82.4</td><td align=\"center\">61.5</td><td align=\"center\">71.7</td><td align=\"center\">64.8</td><td align=\"center\">67.3</td></tr><tr><td/><td align=\"left\">95% CI</td><td align=\"center\">65.7 – 73.7</td><td align=\"center\">59.9 – 69.3</td><td align=\"center\">77.7 – 87.1</td><td align=\"center\">57.7 – 65.3</td><td align=\"center\">66.7 – 76.8</td><td align=\"center\">60.4 – 69.2</td><td align=\"center\">59.7 – 74.9</td></tr><tr><td/><td align=\"left\">p-value</td><td align=\"center\" colspan=\"2\">0.106</td><td align=\"center\" colspan=\"2\"><bold>&lt; 0.01</bold></td><td align=\"center\" colspan=\"3\">0.145</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>CI = confidence interval, p-values for between-group comparison of sensitivity (chi-square test)</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1742-4933-5-8-1\"/>", "<graphic xlink:href=\"1742-4933-5-8-2\"/>", "<graphic xlink:href=\"1742-4933-5-8-3\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Accommodation of the eye refers to changes in the refraction of the ocular lens in order to provide a sharp retinal image at any viewing distance of the visual target. Accommodation – like the pupil size – is controlled by the autonomous nervous system, predominantly mediated by the parasympathetic branch. However, there is anatomical, pharmacological and physiological evidence for an additional sympathetic input – via adrenoceptors [##UREF##0##1##,##REF##12358303##2##]. From a human factors perspective, measurements of accommodation can be relevant for two reasons: first, the design of complex visual displays (for example, virtual image displays) may include conditions, where the accommodative response is not appropriate or mislead, i.e. blurred vision may result [##UREF##1##3##,##REF##17915601##4##]. This question is complicated by the fact that accommodation is affected by factors like contrast, blur or perceived distance [##REF##2371062##5##]. Second, cognitive demand can influence the accommodative response via an activation change in the autonomous nervous system [##UREF##0##1##,##REF##9394631##6##,##REF##2216473##7##]. One should know how stable and large such \"cognitive-induced shifts\" in accommodation might be, to evaluate expected blurred vision under high cognitive load. However, the main purpose of the present paper is to evaluate the possibility that shifts in accommodation might be an indicator of the amount of cognitive load imposed on a task operator. In addition to the well-known pupillary response to cognitive demand [##REF##5021049##8##], ocular accommodation as a second ocular indicator may improve the identification and clarification of autonomic activation. This combined approach might be advantageous, because pupillary responses alone have some disadvantages: for example, the pupil is directly dependant on the illumination level, which leads to ceiling effects in dim surroundings; further, the pupil is thought to be an unspecific indicator of autonomic changes – it reflects general states of mood, motivation, emotions and so on. The accommodation, in contrast, might be strictly sensitive to cognitive demand changes and is not directly influenced by surrounding light or involved in homeostatic regulation of the body (which is typically true for other classical autonomic indicators like cardiovascular measures). Modern autorefractors allow for measuring both accommodation and pupil size, even dynamically with frequencies up to 25 Hz; these video techniques measure the eyes from a remote position. The measurement of both indicators – pupil size and accommodation – with a simple video recording system is a tempting possibility, that additionally motivated this paper.</p>", "<p>Cognitive demand can affect open-loop accommodation, i.e. when no appropriate stimulus is presented [##REF##2216473##7##,##REF##1771070##9##,##UREF##2##10##]. Unfortunately, for closed-loop accommodation at near viewing distances – a situation more relevant for human factor applications – the results for cognitive effects are conflicting: Wolffsohn, Gilmartin, Thomas &amp; Mallen (2003), for example, reported no change of accommodation while subjects checked summation-tasks for correctness. Otherwise, Winn, Gilmartin, Mortimer &amp; Edwards (1991) described a mean increase (i.e. a near shift) in accommodation by 0.17 D when subjects had to respond to a target letter rather than reading the letters to themselves. Further, Kruger (1980) showed that the average accommodation increased by 0.28 D when the subjects changed from reading to adding two-digit numbers. On the contrary, Malmstrom, Randle, Bendix &amp; Weber (1980) found a decrease in accommodation when subjects fixated a target and additionally counted backwards, in contrast to pure fixation. Bullimore &amp; Gilmartin (1988) described an accommodative response while numbers were presented in rows and columns: for the 5 D viewing distance, the accommodative responses decreased by 0.04 D with the alteration from reading to adding [##REF##1425129##11##]. The recent study of Davies, Wolffsohn &amp; Gilmartin (2005) used an independent physiological indicator (heart period): a reduction in accommodation with increasing cognitive demand coincided with a reduction in heart period (the correlation of both effects amounted to r = 0.98). Cognitive demand was altered by varying the speed of a two-alternative forced choice task. The change in accommodation and heart period was interpreted as an increase in sympathetic activation in autonomic control of the body. Taken together – no general answer appeared to the question about the effects of cognitive demand on accommodation.</p>", "<p>The aim of the present studies was therefore to ensure – for a possible human factor application – that closed-loop accommodation is an indicator of cognitive induced changes in autonomic balance. We tried to clarify the confusing findings of previous research by considering all previously reported information about stimulus conditions [##UREF##0##1##], instructions [##REF##2371062##5##] or subject's refractive status [##REF##15851584##12##]. Additionally, we considered the effect of two modulating factors: gaze shifts and performance measures, which endanger the correct interpretation of accommodative changes. Cognitive operations could induce different eye movement patterns [##REF##5032358##13##,##REF##3211676##14##]; consequently, possible changes in gaze direction may alter the measured accommodation without any change in curvature of the lens [##REF##15829857##15##, ####REF##12367740##16##, ##REF##15005679##17####15005679##17##]. None of the previous research included measurements of gaze direction. Further, in order to confirm that accommodative effects are really induced by cognitive demand, the intended demand has to be validated by means of behavioral changes, i.e. performance measures. Increasing cognitive demand should result in increased performance times or higher error rates. In most of the studies described above, the accommodative response was pooled across correct or false results and no further control of errors was implemented, whereas – usually – the acceptable level of performance should not exceed 25% error rates, when all trials are considered in the calculation of means [##REF##9347483##18##,##REF##12814572##19##]. One should consider that errors might occur due to intermittent blurred vision of the targets: such short-term accommodative far-drifts or stares are somewhat likely in extreme near viewing conditions [##REF##12929791##20##,##REF##4094868##21##]. Thus, in such conditions it remains unclear whether errors are the result of erroneous task processing or not-perceived targets. To avoid such uncertainties, the \"cognitive-induced\" shift in accommodation should be based on accommodative data collected during correct task performance.</p>", "<p>For correct interpretation of possible changes in the autonomic nervous system, we included the following reliable indicators of autonomic balance: heart period or heart rate and pulse transit time as cardiovascular parameters [##UREF##0##1##,##REF##8598068##22##, ####REF##9401419##23##, ##REF##6828608##24##, ##REF##7375621##25##, ##REF##7291449##26##, ##REF##1467394##27####1467394##27##] and the well-documented pupillary response [##REF##5021049##8##,##REF##5924930##28##, ####UREF##3##29##, ##REF##17833905##30##, ##UREF##4##31##, ##REF##17806274##32##, ##REF##17455055##33####17455055##33##]. We compared these classical indicators of autonomic activation with accommodative effects, looking for evidence of conformity. We collected data during 4 experiments, including prior reported tasks like reading, adding and multiplying numbers and a variation of the \"n-back\"-task, which is known to demand processes of short-term memory [##REF##9038284##34##,##REF##9121583##35##].</p>" ]

[ "<title>Methods</title>", "<title>Targets and subjects</title>", "<p>The targets were composed of numbers or letters and were presented on a TFT screen as black on white numbers with a mean background luminance of 30 cd/m². Each number/letter subtended 0.29 deg × 0.37 deg (width × height) at a viewing distance of 5 D (20 cm; as suggested by Gilmartin (1988)). We presented the targets at this very close viewing distances (accommodation demand relative to the individual resting state &gt; 4 D) in order to elicit a strong initial parasympathetic response and to ensure accommodative changes per se. The surrounding room lighting was adjusted individually in order to set the initial pupil size at an individual intermediate size; resulting room lighting varied between 2 and 15 lux.</p>", "<p>In general, we used tasks most prevalent in the literature: reading, adding or multiplying of numbers and a \"n-back\" task. Each task period lasted 160 s; when different tasks were compared with each other, these task periods were presented without any time gap to avoid artefacts due to normal physiological variations of autonomic parameters during homeostasis [##REF##13521047##41##,##UREF##6##46##]. Additionally, to avoid short time effects of orientation or attentional shifts at the moments of task switching, the first 32 s of each task period were excluded from data analysis and all parameters are described as means across the remaining 128 s. Analyses of variance with repeated measures (with Greenhouse-Geisser adjusted error probabilities) were then calculated.</p>", "<p>For each task period, the percentage of correct responses was taken as general performance measure; subjects were told to exert equally effort on focusing the numbers/letters and performing as correctly as possible.</p>", "<p>Participants were selected from a pool of 40 male (mean age ± SD: 23 ± 3.6 years), almost emmetropic subjects (best spheres and astigmatic component did not exceed 0.5 D) – with a visual acuity &gt; 0.8 (in decimal units) in the right eye. The near point of accommodation was typical for this age, i.e. close to 10 D on average, and the mean (± SD) of resting accommodation (dark focus) was 0.79 D (± 0.27). Each subject gave informed consent before experiments; the research followed the tenets of the Declaration of Helsinki.</p>", "<title>Measurement of accommodation, gaze direction and pupil size</title>", "<p>Dynamic photorefractor measurements: In experiments 1, 2, and 3, accommodation, pupil size and gaze direction were measured objectively and dynamically (25 Hz) using a remote, automatic eccentric infrared photorefractor, the PowerRefractor (MultiChannel) in its standard version [##REF##12637836##47##, ####REF##14660462##48##, ##REF##10615454##49####10615454##49##] and its successor, the PowerRef II (PlusoptiX) [##REF##15491480##50##]. No difference between readings of the two photorefractors for different subject subgroups was found, so that the data were pooled. In the current study, all measurements of accommodation reflect sphere data determined in the vertical meridian of the right eye under conditions of monocular viewing. The horizontal gaze direction was determined from the position of the corneal Purkinje image of the refractor with respect to the pupil centre. The PowerRefractor and the PowerRef II are specified to have an accuracy of 0.25 D (for pupil size within 3 – 11 mm) [##UREF##7##51##]. In order to keep the required fixed 1 m distance from the eyes to the video camera and to present the targets at 5 D, we placed the camera above the subject's eye level and used two dichroic mirrors, which pass visible light and reflect infrared light [##REF##15491480##52##]. The camera was placed in line with the right eye and a chin and forehead rest was used.</p>", "<p>During data screening, blink artefacts were removed from the accommodation, pupil size and gaze direction records by eliminating all data points within an interval of 100 ms before and after each eye blink.</p>", "<p>Autorefractor measurements: In experiment 4, accommodation was measured with the open-view autorefractor Shin-Nippon SRW 5000 (Canon Inc., Tokyo, Japan) in its standard version [##REF##15005679##17##]. On pressing a button at the joystick of the SRW 5000, the instrument can take static measurements of refractive error in the range of ± 22 D in steps of 0.125 D. When the button is kept pressed, about 45 static readings can be collected within 1 min (data analysis is performed in 0.15 s). A built-in display provides an image of the pupil to allow alignment of the instrument with respect to the subject's visual axis; refractive data were transmitted to a PC [##REF##11261343##53##]. The manufacturer's designation of a minimal pupil size is 2.9 mm. Accommodative data were corrected for outliers (beyond a ± 4 SD range).</p>", "<title>Measurement of electrocardiogram (ECG) and pulse transit time (PTT)</title>", "<p>All cardiovascular data passed an eight pole Bessel-lowpass-filter with a corner frequency of 220 Hz. The sampling rate for each channel was set to 500 Hz. To determine the heart period (HP) we used a bipolar record (Einthoven II). The ECG-signal was amplified (Siemens Mingograf 710) and filtered with a time constant of 3.2 s. To identify the inter-beat-intervals (IBI) out of the ECG signal, the QRS complex was determined by a template-matching-algorithm which consisted of a QRS complex template: a mean of 3 to 4 selected intervals of 27 ms after and before the R-wave was calculated and a cross-correlation function identified the remaining QRS complexes. Artefacts were selected automatically [##REF##9401419##23##] and visually by checking the distribution of the IBI for outliers. The resulting mean IBI represented the mean heart period. The pulse transit time (PTT) was defined as time between the R-wave of the QRS-complex of the ECG signal and the initial upstroke of the pulse curve at a peripheral site. The arrival of the pulse in the left pointing finger was measured with a special photoplethysmograph. The signal of the PTT was low-pass filtered with 24 Hz. The upstroke of the pulse curve was detected by an algorithm selecting the minimum of the peripheral pulse curve in an interval of 300 ms after the appearance of the R-wave, which was already determined by the QRS-algorithm. Outliers were defined as values being greater than 260 ms or smaller than 140 ms.</p>", "<p>According to Weiss, Del Bo, Reichek and Engelman (1980), we focused on calculating a quotient for the change in cardiovascular parameters, indicating selective branch activity due to a shift in autonomic balance. This quotient (ΔPulseTransitTime/ΔHeartRate; we transferred heart period in heart rate) reflects relative changes of both parameters and is based on a general increase in heart rate accompanied by a decrease in pulse transit time; when this quotient is more negative than -0.9, increased beta-sympathetic activity is deduced, while for a quotient more positive than -0.9 a decreased parasympathetic activity is supposed.</p>" ]

[ "<title>Results of experiments 1 to 4</title>", "<title>Experiment 1: Reading and adding within a number-matrix</title>", "<p>40 subjects were asked to read or add one-digit numbers arranged in rows and columns of a 5 × 5 number matrix (2.4 deg width × 2.8 deg height; see Figure ##FIG##0##1##); our task was closely related to the one of Bullimore and Gilmartin (1988). We had periods of reading and adding that lasted 160 s. Reading-adding and adding-reading task sequences were presented and the order was counterbalanced within subjects. After each task sequence of 320 s (160 s reading + 160 s adding, and vice versa) a break of 10 min reduced possible carry-over effects. The instructions \"read\" or \"add\" were given at the beginning of each block and indicated which row/column was to add/read. No timing protocol was enforced, so that reading and adding was completely self-paced. For the adding period, a possible result (randomly correct or incorrect by ± 1 in half of the blocks) had to be indicated as correct or not; in half of the subjects, the right (left) button was assigned as \"correct\" (\"incorrect\") and vice versa for the other half of the sample. In order to have similar procedures in the adding and reading task, the following response was used after the reading task: randomly, either a \"11\" or \"22\" appeared at the end of the task; half the participants had to press the right button for the response \"11\" and the left button for \"22\", while the other half had the reversed assignment. Accommodation and gaze direction were measured with the PowerRefractor and the PowerRef II (see General Methods).</p>", "<title>Results of experiment 1</title>", "<p>The sequence of tasks had no effect, thus the two repetitions were averaged. When changing from reading to adding, the performance data showed a mean decrease in correct responses of 4% (F(1,39) = 7.70; p = 0.01) and the pupil dilated by 0.3 mm (F(1,39) = 36.76; p &lt; 0.01) relative to a diameter of 4.91 mm during reading; these results indicated an increase in task difficulty [##REF##5785627##36##]. Both cardiovascular parameters decreased: the heart period by 13 ms (F(1,39) = 9.09; p &lt; 0.01) and the pulse transit time by 1.27 ms (F(1,39) = 7.31; p = 0.01). According to Weiss, Del Bo, Reichek and Engelman (1980), we calculated a quotient of -1.35 for the change in cardiovascular parameters, indicating an increase in sympathetic activity. Additionally, changing the task from reading to adding induced an apparent decrease in accommodation of 0.07 D (F(1,39) = 4.17; p = 0.04; CI 95%: (-0.15, +0.01)) and a change in gaze direction of 0.14 deg to the right (F(1,39) = 4.42; p = 0.04) (see Figure ##FIG##1##2##), although the target position was exactly the same for both instructions.</p>", "<p>Because of this significant change in gaze direction, we examined the extent to which the accommodative measure depends on gaze direction (see Appendix) and decided, theoretically and data motivated, to consider the gaze direction as covariate for accommodation in the analysis of variance: the resulting accommodative effect by changing the task became non-significant (F(1,38) = 1.52; p = 0.23). We considered the best-fit linear equation between gaze direction and accommodation (see Appendix) to obtain a corrected accommodation level and calculated a 95%-confidence interval (-0.04; +0.01) for the remaining mean change of -0.02 D from reading to adding. Further, only 2% of variance could be explained by individual differences, i.e. by susceptible individuals, in this presumed \"cognitive-induced effect\" in accommodation, reflected by the interaction \"subject × task\" [##REF##14660221##37##]. Additionally, the individual accommodative effects were neither correlated with those in pupil size (r = 0.14; p = 0.36), in heart period (r = -0.03; p = 0.84) nor in pulse transit time (r = -0.11; p = 0.49).</p>", "<p>In sum, experiment 1 showed no evidence for the \"cognitive-induced shift in accommodation\". However, the change in correct response was only 4%; thus, our task might have been too easy to provoke an adequate sympathetic reaction in the ciliary muscle. Therefore, in experiment 2 we adopted another, more difficult task from the literature.</p>", "<title>Experiment 2: Reading and adding two-digit numbers</title>", "<p>40 subjects viewed two-digit numbers (see Kruger (1980)); each number was presented for 1.5 s and they were separated by pauses of 1.5 s (see Figure ##FIG##2##3##). This paced 160 s task period comprised 5 blocks of 30 s and each block contained the presentation of 10 two-digit numbers; these numbers were exactly the same for the reading and adding task and 10 numbers contained six \"easy\", like 02 or 09, and four \"difficult\" numbers, like 17 or 14; the latter were restricted to vary between 11 and 19. We had reading-adding and adding-reading task sequences in counterbalanced order. The instructions \"read\" or \"add\" were given at the beginning of each 30s-block and after 30 s a possible result was presented. For the reading period the subjects had to quit randomly presented numbers (\"11\" or \"22\"), while for the adding period a possible result (incorrect by ± 1 in half of the blocks) had to be indicated as correct or not. Accommodation and gaze direction were measured using the PowerRef II (see General Methods).</p>", "<title>Results of experiment 2</title>", "<p>The amount of correct responses decreased by 16% when switching the task from reading to adding (F(1,39) = 50.72; p &lt; 0.01), refecting a higher cognitive demand then in experiment 1. Accordingly [##REF##5785627##36##], the pupil dilated even more, i.e. by 0.5 mm, (F(1,39) = 38.92; p &lt; 0.01; reading pupil size: 4.83 mm). Heart period and pulse transit time decreased by 14 ms and 2.8 ms, respectively (F(1,39) = 6.51; p = 0.01 and F(1,39) = 8.39; p &lt; 0.01) and the quotient of these changes (q = -2.63) indicated an increase in sympathetic activity as in experiment 1 [##REF##7375621##25##]. Generally, the sequence of tasks had no effect, except for a statistically significant interaction of task sequence × task for correct responses (F(1,39) = 9.42; p &lt; 0.01: subjects made 11% more errors when they added first). When the task changed from reading to adding, gaze direction changed by 0.31 deg to the right (F(1,39) = 13.76; p &lt; 0.01) and accommodative raw data decreased by 0.10 D (CI 95%: (-0.19; +0.01)). The latter accommodative effect diminished to 0.03 D (F(1,38) = 3.67; CI 95%: (-0.05; +0.01)), when gaze direction was used as covariate. Further, only about 5% of variance [##REF##14660221##37##] was explained by the variation in \"cognitive-induced\" effects between the subjects. Again, accommodative effects were neither correlated with changes in pupil size (r = 0.16; p = 0.32), in heart period (r = -0.06; p = 0.71) nor pulse transit time (r = 0.12; p = 0.48).</p>", "<p>In sum, the cognitive demand was higher in experiment 2, but, nevertheless, the accommodative change did not reach statistical significance after the change in gaze direction was taken into account.</p>", "<p>Unfortunately, in experiments 1 &amp; 2, separate post hoc analyses of correct and incorrect trials – as claimed for in the introduction – were not possible since a task contained a block of number presentations and summarized performance measures. Therefore, we changed the task design in the following experiment 3.</p>", "<title>Experiment 3: Reading, adding and multiplying numbers</title>", "<p>Twenty subjects had to read, add or multiply a two-digit and a one-digit number; number combinations were identical for all three tasks and selected in order to avoid trivial combinations like \"20*1\". The arrangement of the numbers is shown in Figure ##FIG##3##4##; for reading the numbers, a \"L\" or \"R\" was placed between them and subjects had to react with the left or right mouse button, respectively (see Figure ##FIG##3##4##). During adding and multiplying periods, the presented result could be incorrect by ± 1 or ± 10; subjects had to quit the result as correct or false. Each number combination was presented for 2 s and the complete reading, adding or multiplying period lasted again 160 s.</p>", "<p>We had two reading periods: one before (or after) adding and one before (or after) multiplying periods; task sequences was counterbalanced across subjects. For statistical analysis, we distinguished between an experimental phase – reading versus calculating – and a calculation content – adding versus multiplying.</p>", "<p>Accommodation and gaze direction were measured using the PowerRef II (see General Methods). In addition to the averages of 128 s task periods, the accommodation and gaze data were clustered into 8 periods of 20 s to trace accommodative changes throughout the task period.</p>", "<title>Results of experiment 3</title>", "<p>The reading and adding task did not produce significant differences in the number of correct responses, cardiovascular parameters, pupil size or accommodation. Maybe the task of adding within 2 s was as easy as reading the numbers. However, the multiplication task induced significantly more errors (by 30%) than the reading task (F(1,19) = 56.84; p &lt; 0.01) and than the adding task (F(1,19) = 53.79; p &lt; 0.01), as shown by simple-effect analysis of variance [##UREF##5##38##]. The ratio of heart period to pulse transit time showed a relative change from reading/adding to multiplying (quotient: -1.40 and -1.62, respectively) indicating an increase in sympathetic activation [##REF##7375621##25##]; the pupil size increased by 0.7 mm for the same comparison (reading-multiplying: F(1,19) = 19.91; p &lt; 0.01;adding-multiplying: F(1,19) = 14.08; p &lt; 0.01).</p>", "<p>In a first analysis, the accommodative data were analyzed considering gaze direction as covariate and irrespective whether responses were correct or incorrect. The analysis of the 8 periods of 20 s showed that accommodation increased slightly by 0.10 D over time during the whole 160 s task period – regardless of the experimental phase or calculation content (F(7,132) = 3.44; p &lt; 0.01). The multiplying task produced a significant decrease of accommodation by about 0.25 D – both relative to the reading task (F(1,18) = 5.94; p = 0.02; CI 95%:(-0.46; -0.08)) and relative to the adding task (F(1,18) = 8.63; p &lt; 0.01; CI 95%: (-0.54; -0.10); see Figure ##FIG##4##5a##).</p>", "<p>For a further analysis, we included only accommodative measures during correct task processing and confined the sample to 16 subjects with an individual mean error rate smaller than 40%. Again, a temporal increase of accommodation during the 160 s periods was observed (F(7,104) = 2.64; p = 0.02). The mean decrease in accommodation during multiplying relative to reading/adding shrank to 0.04 D (CI 95%: (-0.09; +0.01)); the interaction between experimental phase (reading or calculating) and calculation content (adding or multiplying) remained non-significant (F(1,14) = 2.04; p = 0.13; see Figure ##FIG##4##5b##), indicating no statistical difference between the three tasks for the accommodative response.</p>", "<p>In sum, eventhough the cognitive demand varied between reading/adding and multiplying number, experiment 3 showed also no evidence for a \"cognitive-induced shift\" in accommodation. In the following and last experiment we used a task, which comprised only a single target character – to avoid shifts of gaze direction directly- and induced different levels of cognitive demand.</p>", "<title>Experiment 4: the \" n-back\" task</title>", "<p>Presenting a central target and varying cognitive demand is easily done within an adaptation of the \"n-back\"-task: a series of characters is presented in random order for 1000 ms (at 600 ms intervals) and the subjects have to indicate whether the letter in the present step n was the same (or not) as the one before in step n-1 (or n-2) [##REF##9038284##34##,##REF##9121583##35##,##REF##7691540##39##] (see Figure ##FIG##5##6##). By increasing the number of steps backwards, the demand on processes of the short-term memory is increased, indicated by an increase in reaction time and errors. Typically, the reaction time increases mostly by changing the task from n-1 to n-2, whereas the errors continuously increase with n steps backwards [##REF##9038284##34##,##REF##9121583##35##,##REF##7691540##39##]. To our knowledge, the accommodative response to this \"n-back\" task is not reported elsewhere yet. We started the letter presentation with a short instruction line and three green letters (A or H); then a series of As and Hs was presented for a 160 s period. After each letter a response was given with a button and the reaction time was measured. Mainly, we varied cognitive demand using n-1 and n-2 tasks (N = 20), but had an additional control run with a n-4 task.</p>", "<p>To ensure that the measured physiological effects were due to cognitive demand, we compared the lower cognitive demand (baseline: n-1; 160 s) directly with the same or higher demand (task: n-1, n-2 or n-4; 160 s) [##REF##13373240##40##,##REF##13521047##41##], resulting again in 320 s task processing. The task sequences were counterbalanced across subjects. For statistical analysis, we distinguished between an experimental phase (baseline versus task) and a task content (n-1 versus n-2).</p>", "<p>We changed the measurement technique for the fourth experiment in order to implement a system which is reported to have an highest standard accuracy of accommodation measurement: accommodation was measured with the open-view autorefractor Shin-Nippon SRW 5000 (Canon Inc., Tokyo, Japan; see General methods). In a separate control experiment, we tested for possible gaze shifts during the n-back task: for 10 subjects we measured the gaze direction with the PowerRef II (described above) while they performed the n-1 and n-2 task. No change in gaze direction between the two tasks was observed (t (9) = 0.56; p = 0.59) and, therefore, for this experiment gaze induced effects on accommodation were unlikely.</p>", "<title>Results of experiment 4</title>", "<p>An interaction of experimental phase (baseline vs. task) and task content (n-1 vs. n-2) was significant for errors, reaction time and cardiovascular parameters throughout analyses of variance. We therefore calculated simple-effect analysis of variance [##UREF##5##38##] to clarify the source of differences: obviously, the n-1-baseline and the n-1-task did not differ significantly. As expected from literature [##REF##9038284##34##,##REF##9121583##35##,##REF##7691540##39##], when the n-2-task was compared with the n-1-baseline the error rate increased by 8% (F(1,19) = 29.63; p &lt; 0.01) and the reaction time increased by 360 ms (F(1,19) = 77.65; p &lt; 0.01).</p>", "<p>Additionally, the cardiovascular quotient indicated a decrease in parasympathetic activity between n-1-baseline versus n-2-task (q = -0.24) [##REF##7375621##25##].</p>", "<p>For the \"n-back\" task, average refraction as indicator of accommodative changes varied non-systematically between 3.83 D and 3.91 D (mean SD: 0.28 D) regardless of experimental phases and task contents.</p>", "<p>In an additional experiment, we applied a larger task demand, i.e. n-4, in a sample of 19 subjects. The results are shown in Figure ##FIG##6##7##: as expected from previous research [##REF##9038284##34##,##REF##9121583##35##,##REF##7691540##39##], the mean amount of correct responses decreased (monotonously) by 25% (t(18) = 9.79; p &lt; 0.01) when changing the task from n-1 to n-4. Reaction time increased by 340 ms (t(18) = -11.65; p &lt; 0.01) indicating the same increase as for the n-1 to n-2 variation. However, refraction (including accommodation) remained on a constant level of around 3.98 D regardless of the task demand (t(18) = 0.76; p = 0.45).</p>", "<p>In sum, in experiment 4 we varied successfully the demand on the short term memory, but accommodation – measured during correct task processing – was still not systematically affected by these demand changes.</p>" ]

[ "<title>General Discussion</title>", "<p>Accommodation depends on parasympathetic and sympathetic innervation of the ciliary muscle [##UREF##0##1##,##REF##15851584##12##,##REF##3203603##42##,##REF##7208254##43##] and one might expect that accommodation will be influenced by non-optical stimuli, e.g. cognitive demand, just like pupil size or cardiovascular parameters. Since accommodation can recently be measured with commercially available video-based refractors (installed remote from the eyes), we had started this research with the expectation that an easy access to both, accommodation and pupil size, would provide a more complete assessment of the actual activation of the autonomous nervous system – for both, human factor applications and experimental research. Specifically, accommodation was thought to refine the interpretations based on classic autonomic indicators, as pupil size, heart rate, etc.. However, the prerequisite for such an approach is the explanation of previous conflicting results: decreases in accommodation with increasing task difficulty have been interpreted as evidence for a so called \"cognitive-induced shift\" in accommodation [##UREF##0##1##,##REF##5021049##8##,##REF##15851584##12##,##REF##3203603##42##,##REF##7208254##43##]. However, increases of accommodation due to cognitive tasks were reported as well [##REF##1771070##9##,##REF##7406012##44##]. In this situation of conflicting literature, we tried to replicate and confirm previous research including classical cardiovascular measures and pupil size [##REF##9401419##23##,##REF##5785627##36##]. As reference, our performance data confirmed task demand variations as well.</p>", "<p>In sum, pupil size and performance measures reflected increased task demand throughout all 4 experiments. The same was true for cardiovascular measures – besides the fact, that in experiment 4 a decrease of parasympathetic activity instead of an increase in sympathetic activity with increasing cognitive demand was observed. It remains unclear, why increased demands on short-term memory elicited other autonomic activation pattern than arithmetic tasks.</p>", "<p>However, for accommodation, the initially observed decrease was marginally, particularly, after the confounding effect of a cognitive-induced shift in gaze direction was included into analysis [##REF##3211676##14##]: changing cognitive demand resulted in a reliable change in gaze direction which in turn led, for methodological reasons, to systematic errors in accommodation measures (see Appendix). We were able to do an ex-post statistical control of gaze direction which could be measured with our apparatus (PowerRefractor and PowerRef II) for three of our experiments. Moreover, in experiment 3, which contained a more pronounced variation of cognitive demand, a remaining small shift in accommodation after gaze correction disappeared as well, when erroneous trials were excluded from data analysis. In near viewing conditions, it could be that short moments of inattention induce far shifts of accommodation or moments of blurred vision impairs proper target viewing and task performance [##REF##12929791##20##,##REF##4094868##21##]. In any case, it is important that the possibility of unintended influences on the accommodative data (like stares due to, for example, inattention) is reduced when data are collected only during correct task performance. In experiment 4, the performance data showed that the demand on short-term memory was increased as intended by our n-back task variations [##REF##9038284##34##,##REF##9121583##35##,##REF##7691540##39##], but we still did not find a corresponding change in accommodation – no \"cognitive-induced shift\" occurred.</p>", "<p>For all data sets, the partly observed minor (non-significant) accommodative changes were neither correlated with cardiovascular changes nor pupil size variations; this observation confirms the disbelief that our subtle accommodative shifts were mediated by autonomic functions.</p>", "<p>Last but no least, reducing our initial sample size of 40 subjects in experiment 1 &amp; 2 to remaining 20 subjects in experiment 3 &amp; 4 (19 subjects for the control run), may have questioned the statistical power. Therefore, we calculated post-hoc power estimates for all our analysis of variances [##REF##17695343##45##]; we found all power values to be larger than 0.95 with one exception of 0.60 for the t-test in experiment 4, where we compared the results of accommodation for n-1 task with the n-4 task. Nevertheless, we conclude, that sample size was adequate to reveal accommodative changes, if they were of physiologically relevant size and inherent in our data.</p>" ]

[ "<title>Conclusion</title>", "<p>Our data showed that the variation of gaze direction and the correctness of task responses might have contributed – probably – to the inhomogeneity of previous results besides other aspects as instructions [##REF##2371062##5##] or initial pupil sizes (corresponding to different depth-of-focus conditions). Finally, our data leave us to doubt changes in closed-loop accommodation due to cognitive demand – at least in near viewing conditions with visually presented targets; at longer viewing distances, effects are even less likely. For practical application, we can draw the positive conclusion that operators using visual displays are unlikely to experience blurred vision due to cognitive demand since we did not find evidence for changes in accommodation that reached practically relevant levels. Although the expectation of accommodation as possible autonomic indicator of cognitive demand was not confirmed, the present results are informative for the field of applied psychophysiology noting that it seems not to be worthwhile to include closed-loop accommodation in future studies.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>The aim of the present study was to assess accommodation as a possible indicator of changes in the autonomic balance caused by altered cognitive demand. Accounting for accommodative responses from a human factors perspective may be motivated by the interest of designing virtual image displays or by establishing an autonomic indicator that allows for remote measurement at the human eye. Heart period, pulse transit time, and the pupillary response were considered as reference for possible closed-loop accommodative effects. Cognitive demand was varied by presenting monocularly numbers at a viewing distance of 5 D (20 cm) which had to be read, added or multiplied; further, letters were presented in a \"n-back\" task.</p>", "<title>Results</title>", "<p>Cardiovascular parameters and pupil size indicated a change in autonomic balance, while error rates and reaction time confirmed the increased cognitive demand during task processing. An observed decrease in accommodation could not be attributed to the cognitive demand itself for two reasons: (1) the cognitive demand induced a shift in gaze direction which, for methodological reasons, accounted for a substantial part of the observed accommodative changes. (2) Remaining effects disappeared when the correctness of task processing was taken into account.</p>", "<title>Conclusion</title>", "<p>Although the expectation of accommodation as possible autonomic indicator of cognitive demand was not confirmed, the present results are informative for the field of applied psychophysiology noting that it seems not to be worthwhile to include closed-loop accommodation in future studies. From a human factors perspective, expected changes of accommodation due to cognitive demand are of minor importance for design specifications – of, for example, complex visual displays.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that there is only one competing, academic interest: we are aware of the fact that we replicated accommodative changes, which other researchers published as \"cognitive-induced\". However, these effects disappeared, after possible alternative effect sources (gaze changes and the correctness of task processing) were taken into account. Trying for publication of these results on the edge of two research disciplines, i.e. human factor research and optometry/vision research, unfortunately, sometimes led to comments as \"not of interest for our community\" with this particular negative result. A more general journal (on negative results) may be appropriate so that other researchers can take advantage of our experience.</p>", "<title>Authors' contributions</title>", "<p>SJ conceived the design, made the final statistical analysis of the data and drafted the manuscript. JH carried out the data aquision and statistical analysis of the cardiovascular data. WJ participated in the design of the study and has been involved in drafting the manuscript. All authors read and approved the final manuscript.</p>", "<title>Appendix: Measures of accommodation at different horizontal gaze directions</title>", "<p>The direction of gaze may affect the measured refraction value without any change in curvature of the lens [##REF##15829857##15##, ####REF##12367740##16##, ##REF##15005679##17####15005679##17##]. In order to quantify this possible measurement artefact, we presented small targets (0.16 deg × 0.16 deg) at 9 horizontal gaze positions to the right eye at eye level. The total range of presentation was ± 2 deg with gaps between each gaze position of 0.5 deg. Refraction and gaze direction were measured monocularly for 10 subjects with the PowerRefractor and for 15 subjects with the PowerRef II (subject pool described above). Each gaze position was fixated for 2 s and refraction and gaze direction were sampled with 25 Hz; for analysis we extracted the medial second of each stable fixation by cutting off the first and last 500 ms. The measured gaze positions reflected the presented gaze positions (see Figure ##FIG##7##8##); the deviation of both regression lines along the y-axes reflected a mean difference in the position of the two cameras (PowerRefractor/PowerRef II); both cameras were fixed in 1 m distance – relative to the right eye. Additionally, the stimulus range of ± 2 deg was not completely reflected by the measured data.</p>", "<p>More important, the measured refraction decreased by 0.16 D (PowerRefractor) and by 0.09 D (PowerRef II) when the measured gaze direction changed by 1 deg to the right (see Figure ##FIG##8##9##). This indicated an artefact: off axis refractive errors where directly influenced by gaze direction; therefore, gaze direction should be considered in the statistical analysis of refraction data (as indicator of accommodation).</p>", "<p>Most studies on \"cognitive-induced shifts\" in accommodation (reported in the Introduction) used a different technique for measuring accommodation, an autorefractor like the Shin-Nippon SRW 5000 (Canon Inc., Tokyo, Japan; [##REF##15005679##17##]); therefore, we investigated whether the standard version (6.15) of this device also produces the gaze direction artefact. In the same experimental setting, 10 subjects fixated each gaze position until 20 readings of the SRW 5000 were collected. For calculating an average refractive value, the first five and last five readings were excluded. As shown in Figure ##FIG##9##10##, for gaze changes from central to the right and to the left, the refraction measures decreased with a slope of 0.03 D/deg.</p>", "<p>Even though these changes are very small, one should bear in mind that the SRW 5000 provides no opportunity to control for gaze changes. Minor changes in refraction as indicator of possible accommodative changes may reflect gaze changes even when the optic of the refractor is permanently adjusted to reach alignment of the instrument with respect to the subject's visual axis, as it was in our experiments.</p>" ]

[ "<title>Acknowledgements</title>", "<p>We wish to thank Alexandra Konrad, Carola Reiffen and Ewald Alshuth for collecting the data. This research was supported by the Deutsche Forschungsgemeinschaft (DFG Ja 747/3-1).</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Time scheme of experiment 1</bold>. In a. the reading and in b. the adding period is shown in time-dependant details.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Box-and-Whisker Plots of the results of experiment 1</bold>. In a. the accommodation response (D) and in b. the gaze direction (deg) as function of task (reading &amp; adding) are shown; both were statistically significant but – as shown in the Appendix – not independent of each other: the measured change in accommodation was mainly induced by the observed change in gaze direction.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Time scheme of experiment 2</bold>. In a. the reading and in b. the adding period is shown in time-dependant details.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p><bold>Task presentations for experiment 3</bold>. The task layout for reading (a), adding (b) and multiplying periods (c), for comparison.</p></caption></fig>", "<fig position=\"float\" id=\"F5\"><label>Figure 5</label><caption><p><bold>Accommodation results of experiment 3</bold>. Accommodation (D) as function of experimental phase (reading or calculating) and calculation content (adding or multiplying). a. shows accommodation data for all 20 subjects regardless of the correctness of task processing while b. shows the accommodation data of 16 subjects collected only during correct task processing. (Note that in b the remaining difference in accommodation between multiplying and reading/adding is mostly due to gaze drifts, which were accounted for statistically BUT not graphically.)</p></caption></fig>", "<fig position=\"float\" id=\"F6\"><label>Figure 6</label><caption><p><bold>Time scheme of experiment 4</bold>. The n-back task contained a letter sequence and subjects had to indicate if the letter in the present step \"n\" was the same as the one before in step \"n-1\" or \"n-2\".</p></caption></fig>", "<fig position=\"float\" id=\"F7\"><label>Figure 7</label><caption><p><bold>Results of Experiment 4</bold>. Average results for the additional control block of experiment 4 as function of task content (n-1 or n-4): in a. the amount of correct responses (%), in b. the reaction time (ms) and in c. the accommodation response (D) are shown; SDs are indicated as error bars. Only the amount of correct responses and the reaction time changed significantly with the task content.</p></caption></fig>", "<fig position=\"float\" id=\"F8\"><label>Figure 8</label><caption><p><bold>Gaze direction measurements</bold>. Measured horizontal gaze direction (deg; mean ± SD) as a function of presented horizontal position (deg) for the PowerRefractor and PowerRef II. Regression equations are shown separately. As indicated by the slope of the regression equations, the distance between the presented horizontal positions was underestimated by the measured gaze direction, even though the measurement distance of 1 m was correctly maintained; maybe the Hirschberg ratio set by the default routine was to low for our sample.</p></caption></fig>", "<fig position=\"float\" id=\"F9\"><label>Figure 9</label><caption><p><bold>Accommodation measures (PowerRefractor/PowerRef II)</bold>. Measured refraction in the vertical pupil meridian (D; mean ± SD) as a function of measured horizontal gaze direction (deg) for the PowerRefractor and PowerRef II. Regression equations are shown separately.</p></caption></fig>", "<fig position=\"float\" id=\"F10\"><label>Figure 10</label><caption><p><bold>Accommodation measures (SRW 5000)</bold>. Measured refraction in the vertical pupil meridian (D; mean ± SD) as a function of presented horizontal position (deg) for the SRW 5000. Regression equations are shown separately.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1477-5751-7-6-1\"/>", "<graphic xlink:href=\"1477-5751-7-6-2\"/>", "<graphic xlink:href=\"1477-5751-7-6-3\"/>", "<graphic xlink:href=\"1477-5751-7-6-4\"/>", "<graphic xlink:href=\"1477-5751-7-6-5\"/>", "<graphic xlink:href=\"1477-5751-7-6-6\"/>", "<graphic xlink:href=\"1477-5751-7-6-7\"/>", "<graphic xlink:href=\"1477-5751-7-6-8\"/>", "<graphic xlink:href=\"1477-5751-7-6-9\"/>", "<graphic xlink:href=\"1477-5751-7-6-10\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Three-dimensional (3D) visualization is thought to ameliorate anatomical understanding among surgeons as well as medical students [##REF##11074877##1##, ####REF##12013693##2##, ##REF##15455893##3####15455893##3##] Better perception of anatomy improves the surgeon's ability to accurately plan and perform surgical procedures [##REF##11074877##1##,##REF##16640519##4##], an effect which can result in lower morbidity and mortality rates after surgical interventions.</p>", "<p>Our group has been working on 3D operation planning in liver surgery for many years [##REF##16686764##5##, ####REF##15664288##6##, ##REF##10719482##7##, ##REF##12488744##8####12488744##8##]. Three-dimensional visualization based on CT scans is an efficient and fast tool for analyzing liver anatomy, possible resection proposals and volumetric consequences of the planned resections [##REF##16686764##5##,##REF##11894216##9##,##REF##17080812##10##]. Particularly, the assessment of post-resectional functioning liver parenchyma is an important issue in planning of liver resections. Is has been shown that computer-assisted operation planning has a potential use of for assessment of functional respectability [##REF##16334748##11##].</p>", "<p>The existence of intrahepatic venous anastomoses (IVA) between the main stems of the hepatic veins (i.e. middle, right, and left) has been known for many years [##REF##13594065##12##,##UREF##0##13##]. Couinaud [##REF##13594065##12##] found anastomoses between intrahepatic veins in 25 out of 30 casts and such anastomoses must be considered a reality from the anatomical point of view [##UREF##0##13##]. The question whether these IVA exist only in pathological livers or also reflect physiological vascular patterns has been discussed heatedly from the beginning [##UREF##0##13##,##UREF##1##14##]. However, there are a few case reports describing the importance of such anastomoses under clinical conditions [##REF##11965052##15##, ####REF##11014654##16##, ##REF##12239918##17##, ##REF##11100313##18####11100313##18##].</p>", "<p>To our knowledge, this is the first published report of 3D visualization of intravenous anastomoses in the human liver. Because 3D imaging leads to a faster and easier understanding of the individual anatomical situation, it may be an appropriate tool to potentially increase patient safety not only in visceral surgery but also in other surgical areas [##REF##17483684##19##, ####REF##17468928##20##, ##REF##17377327##21##, ##REF##16105517##22####16105517##22##].</p>" ]

[]

[]

[]

[ "<title>Conclusion</title>", "<p>Even though 3D visualization is now available for a couple of organ systems, the technique must still be described as rather new [##REF##14600451##25##, ####REF##16123867##26##, ##REF##11153028##27####11153028##27##]. It has been shown in the field of liver surgery that 3D visualization leads to a better understanding of the liver anatomy and in turn improves surgical operation planning [##REF##11074877##1##,##REF##12013693##2##]. In order to optimize the conception of the anatomical situation and thereby potentially improve patient safety, we routinely perform 3D analyses in preparation for liver surgery in patients with complicated liver anatomy and/or demanding surgical procedures including living-related liver transplantation [##REF##16640519##4##].</p>", "<p>The 3D images of the here presented data showed a missing middle hepatic vein (Figure ##FIG##2##3A##). Interestingly, the distal branches of the middle hepatic vein were still open and served as the origin of intervenous anastomoses to both the left and right hepatic veins (Figure ##FIG##2##3B##). Overall, six IVA were detectable, two to the left hepatic vein and four to the right hepatic vein. Because the proximal part of the middle hepatic vein was not detectable intraoperatively, one could speculate that the anatomical situation represents an anatomical variation rather than an after-effect of an old thrombosis.</p>", "<p>Even though the CT images were sufficient to detect IVA, the 3D visualization improved the ability to detect IVA (Figures ##FIG##2##3A## and ##FIG##2##3B##), especially by using the features of the relocalization tool (description in Materials and Method section) which enables parallel searching for IVA in the CT scan and the 3D visualization. This parallel approach enabled us to detect intrahepatic anastomoses which could have been missed if only two-dimensional CT scans had been examined (figures ##FIG##2##3A## and ##FIG##2##3B##). Further, within the freely movable 3D images, the IVA could be detected very easily and fast. In contrast, the determination of IVA is time consuming with two-dimensional CT scans because one has to go back and forth within the image slices to follow the course of the IVA in order to find their drainage vessels.</p>", "<p>The here presented data suggests that the information offered by 3D visualization is not redundant but rather serves as a true source of additional information especially in patients with complicated liver anatomy.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Three-dimensional (3D) visualization is thought to improve the anatomical understanding of clinicians, thus improving patient safety.</p>", "<title>Case presentation</title>", "<p>A 58-year-old male was admitted to our hospital in April 2007 with a suspected metastasis of a sigmoid cancer in the Couinaud segment (CS) 7. The anatomical situation of this patient was analyzed using both a CT scan and 3D images. The initial CT scan revealed that the proximal part of the middle hepatic vein was completely missing and the metastasis in the CS 7 was closely attached to the right hepatic vein. After analyzing additional 3D images, it became clear that due to the close proximity of metastasis and right hepatic vein, the resection of the right hepatic vein was inevitable. Based on this 3D analysis, it was decided to perform a right-sided hemihepatectomy. In this case report, 3D visualization resulted in a faster and clearer understanding of the unique anatomical situation in a patient with complicated liver anatomy than transversal CT images.</p>", "<title>Conclusion</title>", "<p>The here presented data shows for the first time 3D visualization of intravenous anastomoses in the human liver. The information offered by 3D visualization is not redundant but rather serves as a true source of additional information, indicating the potential benefit of 3D visualization in surgical operation planning.</p>" ]

[ "<title>Case presentation</title>", "<title>Description of the patient</title>", "<p>A 58-year-old male was admitted to our hospital in April 2007 with a suspected metastasis of a sigmoid cancer in the Couinaud segment (CS) 7 found during a follow-up examination using ultrasound. The patient had undergone a resection of the colon sigmoideum due to carcinoma in 2003. In addition to the liver metastasis, the patient suffered from mild hypertension and non-insulin-dependent diabetes mellitus. The only preoperative imaging was an abdominal ultrasound; thus, an additional CT scan of the abdomen was performed in preparation for the surgical intervention. The metastasis in the CS 7 had a size of 2.2 × 2.4 cm and was closely attached to the right hepatic vein (Figure ##FIG##0##1##). After careful analysis of the CT scan, it became clear that the proximal part of the middle hepatic vein was completely missing, whereas its distal branches seemed to be open. In order to get more detailed information about the anatomical situation, an additional 3D imaging of this case was performed. It was evident that the middle hepatic vein was missing. Most likely this anatomical pattern was not tumour related because there was a close proximity of metastasis and right hepatic vein (Figure ##FIG##1##2##) making the resection of the right hepatic vein inevitable. As the non-existence of a middle hepatic vein may have led to insufficient drainage of remnant hepatic parenchyma following a bisegmental resection of CS 6 and 7, it was decided to perform a right-sided hemihepatectomy. The pathologic examination confirmed the diagnosis of a colorectal liver metastasis. The surrounding liver parenchyma did not show any distinctive features, and in particular, no evidence of liver cirrhosis. The intra- and postoperative courses of the patient were uneventful. The patient was discharged on the eighth postoperative day.</p>", "<title>CT scans</title>", "<p>The according CT scan was made available by the Radiology Department of the German Cancer Research Center, Heidelberg. The image data was acquired preoperatively during the routinely performed CT scan.</p>", "<title>Image data</title>", "<p>Image data were acquired with a Toshiba Aquilion 16 slice multidetector CT scanner (Toshiba, Japan). A standard bi- or triphasic liver scan with an optimized portal venous phase was performed. SureStart bolus tracking technique (130 ml Imeron 300, Altana, Germany, flow rate 4–5 ml/s) was used in all protocols to optimize vascular contrast [##REF##16123867##26##].</p>", "<title>Data transfer and digital postprocessing</title>", "<p>Imaging data sets were transferred to the Department of Medical and Biological Informatics of the German Cancer Research Center using the CHILI [##REF##11963253##23##] teleradiology system (CHILI GmbH, Heidelberg, Germany). Segmentation of the liver was performed using interactive region growing techniques as previously described by our group [##REF##15664288##6##]. The vessels were segmented using a grey-value-based volume-growing technique. The segmentation was then transformed automatically into a symbolic representation of the vascular anatomy containing the vessel paths, and locations of bifurcations as well as the vessels diameters [##UREF##2##24##]. The origin of the left, middle, and right hepatic veins were identified, clicked upon and all depending vessels were marked automatically as branches of that particular vein. Portal veins and arteries were defined in the same manner. All results were validated by a team consisting of surgeons, radiologists and medical computing specialists.</p>", "<p>Since the whole process from CT data to 3D viszualization preserves all positional information, 3D visualization and original CT data can be presented in a consistent manner. Several tools are available for the interactive exploration of the individual patient anatomy. One of these tools, the \"re-localization tool\", allowed us to define any volume element (voxel) in the CT scan and find this particular voxel within the 3D visualization. This is of special importance because the localization of intrahepatic venous shunts is more easily achieved with the help of 3D images.</p>", "<title>Consent</title>", "<p>Written informed consent was obtained from the patient for publication of this case report and any accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal.</p>", "<title>Competing interests</title>", "<p>The authors hereby confirm that there exists no competing interest regarding their personal or financial relationship with other people or organizations. The authors disclose any financial competing interests but also any non-financial competing interests that may cause them embarrassment were they to become public after the publication of the manuscript.</p>", "<title>Authors' contributions</title>", "<p>LF and BMS both conceived of the case report and drafted the manuscript. MS and H-PM post processed the image data to 3D images. SM and JON collected the patient's data and patients informed consent. MWB was involved in the planning of the study and the critical revision of the manuscript for important intellectual content. All authors read and approved the final manuscript.</p>" ]

[]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Transversal CT image of the patient's liver showing the right (arrowhead) in proximity to the tumor (arrow).</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>3D visualization of the venous anatomy. </bold>The observer has a view from above the liver. Here, the proximal part of the middle hepatic vein is missing. The tumor is colored in red.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>A: 3D visualization of an intrahepatic shunt between the distal part of the middle hepatic vein and the right and left hepatic veins.</bold> Since both CT images and 3D images are based on the same data set, every volume element (voxel) is uniquely defined in all three dimensions (x-, y- and z-axes). Our operations planning system allows us to identify the exact same voxel in both representations, i.e. in CT scans and 3D visualization. Here, the arrow directs to the yellow dot that is set as the original landmark in the 3D image. <bold>B:</bold> The crosshairs point to the exact same position as the yellow dot shown in figure 3A, indicating the position of the identical intrahepatic anastomosis in the CT image. One can speculate whether this vessel structure would have been correctly identified as an intravenous anastomosis if only the CT rather than the 3D images had been available.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1754-9493-2-19-1\"/>", "<graphic xlink:href=\"1754-9493-2-19-2\"/>", "<graphic xlink:href=\"1754-9493-2-19-3\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>The main clinical features in patients with Roberts sydrome (pseudothalidomide)are severe shortening of the limbs, with radial defects and oligodactyly or syndactyly, and a characteristic face with hypertelorism, severe cleft lip, a prominent premaxilla, a mid-face capillary haemangioma, cloudy corneae or cataracts and dysplastic or small ears. Other defects may be seen such as large genitalia, congenital heart defects and cystic kidneys [##UREF##0##1##, ####UREF##1##2##, ##REF##7536395##3##, ##REF##872834##4####872834##4##]. Opitz and Lowry [##REF##3812580##5##] stated that early on they had the impression that the Roberts syndrome and the SC pseudothalidomide syndrome may occur in different members of the same sibships and that the identification of identical cytological markers complements the conclusion of nosologic identity as well as effective prenatal diagnosis. Mutations in both Roberts and SC phocomelia were reported by Schule et al. [##REF##16380922##6##]. They found no obvious phenotype-genotype correlation.</p>" ]

[]

[]

[ "<title>Discussion</title>", "<p>Typically the symptoms of phocomelia syndromes are undeveloped limbs and absent pelvic bones; however, various abnormalities can occur to the limbs and bones. Usually the upper limbs are not fully formed and sections of the \"hands and arms may be missing.\" Short arm bones, fused fingers, and missing thumbs will often occur. Legs and feet are also affected similarly to that of the arms in hands. Individuals with phocomelia will often experience missing thigh bones, and the hands or feet may be of an unordinary petite size or appear as stumps due to their close \"attachment to the body. There were several and different clinical entities of children born with phocomelia syndrome [##UREF##0##1##,##UREF##1##2##,##REF##16380922##6##, ####REF##1612213##7##, ##UREF##2##8####2##8##].</p>", "<p>The Roberts (pseudothalidomide syndrome) is a rare autosomal recessive disorder. Roberts initially described it in 1919 [##UREF##0##1##]. It is characterised by pre and postnatal growth deficiency, symmetric limb reductions of variable severity and craniofacial anomalies including hypertelorism, hypoplastic nasal alae, cleft lip and palate. About half of the reported cases presented chromosomal abnormalities. Roberts syndrome is probably the same as SC-phocomelia in view of the fact that in both conditions chromosome analysis shows premature centromere separation. However, not all cases show this feature. Hermann et al., [##UREF##1##2##] published a family with surname begening with S and another with surname begening with C. He descibed a very similar entity called pseudothalidomide or SC syndrome in 1969.</p>", "<p>Allingham-Hawkins and Tomkins [##REF##7536395##3##] review heterogeneity in the condition. Patients with increased premature centromere separation also have cellular hypersensitivity to mitomycin C and there is evidence for different complementation groups. SC are the initial of the two families that were originally described [##UREF##1##2##]. Thalidomide primarily prescribed as a sedative or hypnotic. In Germany, between 5000 and 7000 infants were born with the qualities of Phocomelia. Thalidomide became effectively linked to death or severe disabilities among babies [##UREF##2##8##].</p>", "<p>Previous reports described variable association of malformation complex in connection with Roberts syndrome. Satar et al., [##REF##8004795##9##] reported a case with an accessory spleen and rudimentary gallbladder. Fryns et al., [##REF##7188929##10##] described 2 sibs with tetraphocomelia typical of Roberts syndrome: there was almost complete reduction of the midparts of the upper and lower limbs, and characteristic oligodactyly with absent nails. Neither cleft lip/cleft palate nor eye anomalies were present. Furthermore, premature centromere separation was not observed. The facies was unusual, consisting of a beaked nose, short philtrum, and triangular mouth. Occasional cases can have craniosynostosis, causing confusion with Baller-Gerold syndrome [##REF##2359099##11##]. Many affected infants die in the newborn period, and survivors may have mental retardation (although intelligence can also be normal, [##REF##1612213##12##]. Femoral-tibial synostosis was not a feature. Eylon et al., [##REF##18209617##13##] described femoral-tibial ankylosis in a 9.5-year-old girl who presented with Roberts syndrome.</p>", "<p>Defective development of the limbs and femoral-tibial synostosis is a feature in connection with thrombocytopenia-absent radii (TAR) syndrome, and Schinzel phocomelia syndrome [##REF##9788553##14##,##REF##8123061##15##].</p>" ]

[ "<title>Conclusion</title>", "<p>This case report emphasizes the importance of recognizing infants born with phocomelia syndrome. Differentiating infants with Roberts-SC phocomelia from other multiple malformation syndromes that feature intercalary limb defects, including thalidomide embryopathy, TAR, and Schinzel phocomelia is fundamental.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Roberts syndrome (Pseudothalidomide) is a rare birth defect that causes severe bone malformation complex. The bones of the arms, and in some cases other appendages, may be extremely shortened and even absent. The fingers of the hands may be fused. An extreme case results in the absence of the upper bones of both the arms and legs so that the hands and feet appear attached directly to the body. This is called tetraphocomelia.</p>", "<title>Case presentation</title>", "<p>We report on a two-year-old boy of Austrian origin who manifests a constellation of malformation complex include prenatal and postnatal growth retardation, craniofacial anomalies and defective development of all four extremities. The overall clinico-radiographic features were compatible with Roberts syndrome (Pseudothalidomide). Significant unilateral femoral-tibial synostosis was additional malformation.</p>", "<title>Conclusion</title>", "<p>Associated malformations and symptoms may be the key factor in the differential diagnosis of neonatal malformation complex. Roberts's syndrome may be genetically transmitted within families as an autosomal recessive trait or may be the result of spontaneous/sporadic changes in the gene. Because the signs of the disorder so closely mimic those caused by the ingestion of thalidomide, the term \"pseudo-thalidomide\" is frequently used.</p>", "<p>In this report we describe total femorotibial fusion in a child manifesting the phenotypic features consistent with Roberts syndrome from a healthy parents but first cousins in Austria. Aggressive medical intervention is of prime importance, as is forthright parental counselling when discussing the possible outcome for these patients.</p>" ]

[ "<title>Case presentation</title>", "<p>A two-year-old child was referred to the department of orthopaedics to asssess his multiple malformation complex. The mother was a 27-year old primigarvida woman married to a31-year-old related man (first cousins). Parents did not report any history of congenital abnormalities in the family. Pregnancy history was negative for exposure to drugs or teratogens. At birth the growth parameters were below the 4 th pecentile. A characteristic dysmorphic facies associated with severe defective development of all four extremities were the major malformation complex. Cleft lip and palate was repaired when the child was 10 weeks old.</p>", "<p>Dysmorphology examination showed a child with prenatal and postnatal growth retardation. Craniofacial abnormalities such as sparse silvery blond hair, prominent frontal bones, hypertelorism, shallow orbits, prominent eyes, a median cleft lip and palate, micrognathia, hypoplastic alae nasi, small and low set ears associated with severe bone malformation defects (figure ##FIG##0##1##).</p>", "<p>Severe and fixed flexion deformities of the knees were evident with maximal deformity being encountered over the left knee in connection with femoral-tibial synostosis. Cytogenetic studies were normal and no premature centromere separation has been encountered. A normal 46, XY karyotype was demonstrated, and autosomal recessive inheritance was presumed on the basis of parental consanguinity. Echocardiodoppler and abdominal ultrasound were normal.</p>", "<p>Radiographic examination showed, hypoplastic humeri, agenesis of the radii and ulnae, defective ossification of the carpals and hypoplasia of the distal phalanges (figure ##FIG##1##2##).</p>", "<p>The pelvis and the lower limbs showed hypoplastic iliac bones, a hypoplastic left femur with subsequent development of femoral-tibial synostosis. Fixed flexion deformities of the knees was apparent, but with maximal intensity over the left knee. Bilateral fibular aplasia, the ankles showed defective ossification associated with hypoplasia of the distal phalanges (figure ##FIG##2##3##). His subsequent course of developmental milestones has been of severe retardation, but his mental growth was nearly normal. His primary obstacle to improve his motor development was the severe fixed flexion deformity of the left knee. The planned surgery will involve splitting of the quadriceps, exposure and then excision of the synostosis.</p>", "<title>Abbreviations</title>", "<p>SC-Syndrome: SC is the initial of the two families that were originally described; TAR: Thrombocytopenia-absent radii syndrome.</p>", "<title>Consent</title>", "<p>Written informed consent was obtained from the parents for the purpose of publication of the manuscript and figures of their child. A copy of the written consent is available for review by the editor-in-Chief of this journal.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>All of the authors were involved in the clinico-radiographic assessment and finalising the paper. All authors have red and approved the final version of the paper.</p>" ]

[ "<title>Acknowledgements</title>", "<p>We thank the parents for their remarkable cooperation.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Craniofacial abnormalities such as sparse silvery blond hair, prominent frontal bones, hypertelorism, shallow orbits, prominent eyes, a median cleft lip and palate, micrognathia, hypoplastic alae nasi, small and low set ears associated with severe bone defects.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>Radiographic examination showed, hypoplastic humeri, agenesis of the radii and ulnae, defective ossification of the carpals and hypoplasia of the distal phalanges.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Radiographic examination of the pelvis and the lower limbs showed hypoplastic iliac bones, a hypoplastic left femur with subsequent development of femoral-tibial synostosis.</bold> Fixed flexion deformities of the knees was apparent, but with maximal intensity over the left knee. Bilateral fibular aplasia, the ankles showed defective ossification associated with hypoplasia of the distal phalange.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1757-1626-1-109-1\"/>", "<graphic xlink:href=\"1757-1626-1-109-2\"/>", "<graphic xlink:href=\"1757-1626-1-109-3\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>An aneurysm of the subclavian artery is uncommon. In the younger age group the cause is usually due to thoracic outlet compression often due to an extraneous cervical rib. One such exceptional case of arterial thoracic outlet syndrome presenting in a young man is described below.</p>" ]

[]

[]

[ "<title>Discussion</title>", "<p>True subclavian artery aneurysms are relatively rare lesions, accounting for approximately 0.13% of all aneurysms [##REF##9544142##1##,##REF##15920658##2##]. Atherosclerotic aneurysms are more commonly found in males over the age of 60 years [##UREF##0##3##,##UREF##1##4##]. Thoracic outlet compression is responsible for 75% of aneurysms which arise from the distal part of the subclavian artery, also known as extrathoracic subclavian artery or subclavian-axillary artery aneurysms [##UREF##0##3##]. They are formed as a result of compression by, for example a cervical rib, malalignment post-clavicular fracture or congenital bands on the artery as it passes through the interscalene triangle. This causes stenosis and angulation of the artery, which with time will lead to post-stenotic dilatations [##UREF##1##4##,##REF##2500246##5##]. However, these aneurysms are only found in 1.1% of patients with thoracic outlet syndrome (TOS) [##REF##10743737##6##]. Patients may present with symptoms of thromboembolism, such as digital ischaemia, Raynaud's phenomena, pulsatile masses in the supraclavicular fossa or axilla, brachial plexopathy or Horner's syndrome [##UREF##1##4##].</p>", "<p>The remainder of the aneurysms arise from the proximal part of the subclavian artery, also known as intrathoracic or proximal subclavian artery aneurysms [##UREF##0##3##]. They are mainly the result of atherosclerosis and are usually asymptomatic [##UREF##0##3##]. However, symptoms of compression, including discomfort in the neck, dysphagia and dyspnoea have been reported. Other causes of subclavian artery aneurysms include trauma, iatrogenic; and rarely arteritis, cystic medial necrosis, congenital, syphilis and tuberculosis [##UREF##1##4##,##REF##17669880##7##].</p>", "<p>Plain films and ultrasound scanning are used to identify bony abnormalities and confirm the presence of an aneurysm. However CT or magnetic resonance imaging angiography remain the investigations of choice [##UREF##0##3##,##UREF##1##4##].</p>", "<p>Open surgical repair is the standard management, even for asymptomatic subclavian artery aneurysms due to their tendency to increase in size or rupture [##REF##17669880##7##]. This involves resection of the aneurysm and reconstruction of the artery with or without graft insertion. Decompression of the TOS with resection of the cervical and/or 1^strib, anterior scalene muscle or constricting fibrous bands; or median sternotomy and lateral thoracotomy may be necessary for intrathoracic aneurysms [##UREF##1##4##,##REF##17669880##7##]. This depends on the cause and site of the aneurysm. Distal arterial occlusion secondary to emboli will warrant angioplasty or thromboembolectomy to restore patency of the vessels [##UREF##0##3##]. Overall, surgical management has very favourable outcomes in these patients [##UREF##1##4##].</p>" ]

[ "<title>Conclusion</title>", "<p>Cervical rib leading to subclavian artery aneurysms should always be included in the differential diagnosis of an acutely ischaemic hand. It is a rare event especially in the young adult. This report illustrates the importance and difficulty of correctly diagnosing a rare pathology that has a common presentation. It is important to recognise subclavian artery aneurysms secondary to cervical rib in patients with TOS as they can be safely managed preventing rupture and further thrombosis.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Introduction</title>", "<p>True subclavian artery aneurysms are relatively rare lesions. Thoracic outlet compression is responsible for 75% of aneurysms. They are formed as a result of compression by, for example a cervical rib.</p>", "<title>Case Presentation</title>", "<p>We present a case of subclavian artery aneurysm secondary to a cervical rib in a 25-year-old young Asian adult, who presented with an acutely ischaemic upper limb. A Computed Tomography angiogram revealed a right sided cervical rib and sacular aneurysm in the mid-portion of the subclavian artery.</p>", "<title>Conclusion</title>", "<p>Although it is a rare condition, it is important to be aware especially in the young age group. Surgical management has very favourable outcomes in these patients.</p>" ]

[ "<title>Case presentation</title>", "<p>A, healthy 25 year old Asian male presented with a 3 day history of sudden onset worsening pain of the right forearm and hand. He was a non-smoker with no history of trauma and no past medical history of note. On examination, he had classical signs of acute limb ischaemia consisting of pallor, coldness with prolonged capillary refill of about 10 seconds and sensory and motor impairment. Pulsatile mass in the root of the right side of the neck with a bruit and flow murmur was also elicited. On hand held Doppler ultrasound of the right arm, a diminished ulnar artery signal and a monophasic radial artery signal were detected, along with a normal brachial artery signal. A clinical diagnosis of aneurysm of the subclavian artery was made with embolic occlusion of the brachial artery.</p>", "<p>At this point the patient was commenced on intravenous heparin and the symptoms partially improved. A brachial angiogram (Figure ##FIG##0##1##) and Computed Tomography angiogram revealed a right sided cervical rib and saccular aneurysm in the mid-portion of the subclavian artery, proximal to the clavicle. The aneurysm measured 1.1 × 2 cm. Brachial artery occlusion due to a thrombus extending from the mid-upper arm to the distal part of the elbow was also demonstrated. Intra-arterial thrombolysis was attempted but failed due to extravasation following direct puncture of the axillary artery. 6 days after initial presentation, the patient underwent a subclavian artery aneurysm repair with a cervical rib excision and brachial embolectomy. He was treated with warfarin for 3 months post-op. Post operative Doppler scan revealed good signals in the right ulnar and radial arteries. Follow-up of the patient in clinic 3 months post-discharge, revealed no complications and complete resolution of the symptoms.</p>", "<title>Consent</title>", "<p>Written informed consent was obtained from the patient for publication of this case report and accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>KT and KS undertook writing, literature review, manuscript preparation and literature search. KS submitted the article. RP was responsible for diagnosis, patient management and review. KT and KS managed the patient on the ward. AJ reviewed the final manuscript. All authors read and approved the final manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>Vascular surgical secretary of RP for obtaining patient notes post-discharge.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Right Brachial arteriogram (A) Occlusion of the brachial artery with an abrupt cut-off point. (B) Saccular aneurysm arising from the mid right subclavian artery; The first image (A) is of the Right brachial artery, its is an arteriogram and was taken a day after admission. Image (B) shows the right subclavian artery and its branches, it is also an arteriogram and was taken the day after admission.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1757-1626-1-140-1\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Nocardia organism is filamentous gram positive rods. They are aerobic. Human pathogens infect by inhalation of airborne bacilli or the traumatic inoculation of organism into skin. Overall 80% present as invasive pulmonary infection, disseminated disease abscess; 20% present as cellulitis. Debilitated patients have a 45% mortality rate even with appropriate therapy.</p>" ]

[]

[]

[ "<title>Discussion</title>", "<p>In this report, we present a rare case of a non-immunocompromised patient with old Tuberculosis but she has been on low dose of steroids. Nocardia is a weakly aerobic gram positive acid fast, filamentous bacteria. They are an important opportunistic infection in elderly and immunocompromised affecting lung, brain and skin. Nocardiasis is sporadic and have higher incidence in the immunocompromised population. There is no age, race predilection [##REF##16825752##1##]. Pre-existing lung condition, in this of case of tuberculosis increase the risk of contracting the infection. Bronchiectasis is an important risk factor of Nocardia colonisation. The presentations are cough, fever and difficulties in breathing. The spread is sporadic, and are usually found in dust and soil. The use of Ziehl Neelson technique with a weaker acid concentration can result in identification of a variety of acid fast organisms [##REF##9797732##2##]. Generalised infection involves cutaneous joint and pulmonary cause. Further dissemination involves acute and chronic symptoms. There are no person to person transmission. Core diagnostic test include sputum culture and bronchoscopy.</p>", "<p>Standard therapy is trimethoprim-sulphamethoxazole for 6 months to 1 year. Refractory case needs imipenem and amikacin. Nocardiais should be suspected in immunosuppressed patients.</p>" ]

[ "<title>Conclusion</title>", "<p>Pulmonary bronchiectasis is difficult to diagnose, which delays its diagnosis and a high level of suspicion is required in patients with underlying chronic condition or chronic steroid use [##REF##9230244##3##]. Nocardia when involves the central nervous system leads to a poor prognosis, implies an early diagnosis and prompt treatment. Death occurs from sepsis, overwhelming pneumonia or brain abscess. Mortality is increased in disseminated disease involving 2 or more organs [##REF##8924542##4##]. Mortality is more in patients on corticosteroids or chemotherapy. Nocardiosis should be an important differential of any chronic pneumonia not responding to the antibiotic treatment.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<p>In this rare case a non-immunocompromised patient with old Tuberculosis on low dose of steroids presents with opportunistic infection of a weakly aerobic gram positive acid fast, filamentous bacteria called Nocardia.</p>", "<p>An 80 year old non-smoking white female presented with cough, shortness of breath and purulent sputum.</p>", "<p>Initial antibiotics given were not helpful. Later microbial diagnosis was Nocardia in sputum sample which was uncommon in a non-immunocompromised. She responded to co-trimoxazole therapy.</p>" ]

[ "<title>Case presentation</title>", "<p>An 80 year old non-smoker white female was admitted with shortness of breath and cough with purulent sputum. Relevant past history of polymyalgia rheumatica, for which she was on regular low dose steroids. She also informed past history of Tuberculosis (in 1951) treated in sanatorium. She was hypoxic with oxygen saturation 79.9% on air. Her inflammatory markers were high. Chest X-ray [Fig ##FIG##0##1##] and Computed tomographic scan [Fig ##FIG##1##2##] revealed old calcification of right apical lobe and consolidation and bronchiectatic changes of right lower lobe. She was treated with bronchodilators and antibiotics. She was still hypoxic with high inflammatory markers. The microbial diagnosis was established after isolating Nocardia in sputum sample. Subsequently she was subsequently treated with trimethoprim and sulphamethoxazole for 6 months and offered further follow-up. Further follow-up clinic we found out that she responded to the treatment.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>KS carried out literature search. KS participated in the sequence alignment and draft of manuscript. All authors read and approved the final manuscript</p>" ]

[ "<title>Acknowledgements</title>", "<p>Written consent has been obtained from the patient for submission of this manuscript and figures for publication. Funding was neither sought nor obtained.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Chest X-ray showing consolidation and bronchiectatic changes.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>C<bold>T scan showing bronchiectatic cavity of old tuberculosis which grew Nocardia.</bold></p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1757-1626-1-122-1\"/>", "<graphic xlink:href=\"1757-1626-1-122-2\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>The microembolic injury is one of the determining factors to cognitive dysfunction after cardiac surgery [##REF##8023354##1##] as hypoperfusion, iper-rewarming and inflammation due to cardiopulmonary bypass (CPB) [##REF##15920636##2##].</p>", "<p>In literature the transcranial doppler (TCD) is proposed to determine the occurrence and the frequency of cerebral microembolic signals during different kinds of cardiac surgery and may alert the surgical team when microemboli enter into the cerebral circulation during surgery, thus allowing preventive measures to be taken [##REF##9341708##3##].</p>", "<p>Advances in doppler technology have made possible to detect not only gaseous microemboli but also the solid ones [##REF##12154248##4##], derived from pericardial blood suction [##REF##11081888##5##] and platelets aggregation on gas microbubbles [##REF##16236969##6##].</p>", "<p>Microbubbles are normally seen with transoesophageal echocardiography (TEE) in the heart after declamping the aorta especially in patients submitted to valve surgery. Normally the TEE evaluation stops when CPB starts and the surgeon clamps the aorta: at this time it is possible to explore the descending thoracic aorta long and short axis with TEE. We suggest to utilize these echo windows to detect microbubbles coming from the extracorporeal circulation.</p>", "<p>We have documented this case because it describes an interesting example of TCD and TEE interaction.</p>" ]

[]

[ "<title>Results</title>", "<p>No significant microembolic activity was recorded until the extracorporeal circulation started. After the aorta was clamped and cardioplegic perfusion made, the surgeon opened and mechanically fixed the left atrial wall to visualize the mitral valve. During this time the TCD revealed bilateral microembolic signals reaching the brain. During six minutes of monitoring the TCD software recorded 213 MES on the left middle cerebral artery (178 gaseous and 35 solid) and 234 MES on the right one (220 gaseous and 49 solid) (Tab ##TAB##0##1##). The perfusionist noted gas bubbles like foam on the venous return line (Fig ##FIG##0##1##). We appreciated MES on TCD (Fig ##FIG##1##2##) and gas microbubbles on TEE (Fig ##FIG##2##3##) and simultaneously these audio/video informations were recorded in the neurophysiological monitoring system [see Additional file ##SUPPL##0##1##].</p>", "<p>This microembolic activity disappeared after eight minutes; during this time the vacuum assist venous drainage was set below 40 mmHg.</p>" ]

[ "<title>Discussion</title>", "<p>Several studies have demonstrated the impact of microemboli during CPB on postoperative neurological dysfunction [##REF##8023354##1##,##REF##9124978##7##].</p>", "<p>TCD is capable of detecting microembolic material, both gaseous and solid, within the intracranial cerebral arteries and his utility in determining the occurrence and the frequency of high intensity transient signals (HITS) during different cardiac surgery procedures has been documented [##REF##9341708##3##].</p>", "<p>TCD is operator dependent and requires training and experience neurologist to perform and to interpret results [##REF##15136667##8##]. In the daily clinical practice most of the cardiac anaesthesiologists is able to perform a routine intraoperative TEE according to the recommendations of the American Society of Echocardiography published in the 1999 [##REF##10512257##9##]. In the clinical practice the identification of gas microbubbles by TEE is limited to the dearing time before declamping the aorta and the outpatient test of patent forame ovale since the TCD and TEE showed an almost perfect concordance in detection and quantification of right-left shunt [##REF##16629735##10##]. In both two different clinical scenarios the echo window utilized is the four chamber mid esophageal. So far nobody has proposed to identify HITS during CPB with TEE on descending thoracic aorta views.</p>", "<p>The TEE could asses the systemic embolic load during the extracorporeal circulation. Moreover TEE could, like TCD, monitors the surgical, perfusionist and anesthesiologist procedures in respect to the air contamination of CPB. Like the TCD the Pulse Wave Doppler of the descending thoracic aorta blood flow can visualize the microbubbles as HITS because of a different intensity of the Doppler signal due to microemboli (Fig ##FIG##3##4##). The current TCD software with an automatic emboli detection and count is not implemented in the echocardiography devices.</p>", "<p>In our patient a big amount of MES is recorded during the CPB when an higher negative pressure over 60 mmHg is necessary to achieve the venous drainage trough a 25 Fr trans-femoral cannula. In literature an high level of vacuum is reported to be a risk factor for gas embolism during the extracorporeal circulation [##REF##10543494##11##]. A reduced emboli count during CPB is observed when the venous drainage is obtained with a double (femoral and jugular) venous cannulation and a reduced negative pressure is applied to the venous return [##REF##17670507##12##].</p>" ]

[ "<title>Conclusion</title>", "<p>The TEE monitoring of the descending thoracic aorta during the CPB seems to be an alternative method in respect to TCD to assess the microembolic activity and could be a new approach to monitor the efficiency of the surgical team and of the bypass circuit regarding the systemic gas microembolization.</p>", "<p>Our case seems to sustain higher microembolic load in patient in which minimally invasive mitral valve surgery is performed with one percutaneous venous cannula and high level of vacuum in the venous return.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Introduction</title>", "<p>Microembolic signals are usually detected with transcranial doppler during cardiac surgery.</p>", "<p>This report focuses on suggesting the transesophageal echocardiography as a different diagnostic approach to detect microemboli during cardiopulmonary bypass.</p>", "<title>Case presentation</title>", "<p>A 58 year old male patient, caucasian race, was operated on video assisted minimally invasive mitral valve repair using right minithoracotomy approach. His past medical history included an uncontrolled hypertension, dyslipidemia, insulin dependent diabetes mellitus, carotid arteries stenosis. The extracorporeal circulation was performed with femoral-femoral artery and venous approach. Negative pressure for vacuum assist venous drainage was applied in order to facilitate venous blood return. The patient had a brain monitoring with bilateral transcranial doppler of middle cerebral arteries and a double channels electroencephalogram. A three dimensional transesophageal echocardiography to evaluate the mitral valve repair was performed.</p>", "<p>During the cardiopulmonary bypass a significant microembolic activity was detected in the middle cerebral arteries spectrum velocities due to gas embolism from venous return. Simultaneous recording of microbubbles was also observed on the descending thoracic aorta transesophageal echo views.</p>", "<title>Conclusion</title>", "<p>During the aortic cross-clamping time the transesophageal echocardiography can be useful as an alternative method to assess the amount of gas embolism coming from cardiopulmonary bypass. These informations can promote immediate interaction between perfusionist, surgeon and anesthesiologist to perform adequate manoeuvres in order to reduce the microembolism during extracorporeal circulation.</p>" ]

[ "<title>Case presentation</title>", "<p>A 58 year old male patient with severe mitral valve insufficiency was scheduled for a minimally invasive mitral valve surgery (MIMVS) repair which consists of a video assist right minithoracotomy.</p>", "<p>His past medical history included uncontrolled hypertension, dyslipidemia, insulin dependent diabetes mellitus, epiaortic vessel stenosis (50% stenosis in the left and in the right internal carotid arteries) and a recent acute heart failure.</p>", "<p>The preoperative echocardiography revealed a mitral valve insufficiency because of a posterior leaflet prolapse with a moderate reduction of the ejection fraction.</p>", "<p>The three dimensional TEE helped to assess the function of mitral valve before and after surgical repair.</p>", "<p>Bilateral middle cerebral arteries velocity, emboli count and differentiation were recorded by TCD (Doppler Box – DWL). A longitudinal bipolar electroencephalogram montage (2 channels: F3-C3 and F4-C4)) was used, based on the International 10–20 System. The neurophysiological monitoring system (Eclipse – Axon System) simultaneously recorded the raw, the spectral EEG and two videos from TCD and from TEE.</p>", "<p>A moderate hypothermic (34°C) phosporilcoline coated circuit (Dideco Avant) with venous (25 Fr Cardiovation) and arterial (22 Fr DLP) femoral cannulations was performed. A vacuum assist venous drainage of at least 60 mmHg was applied to the external reservoir system to facilitate the venous return according to the surgeon demand. The circuit had a bubble trap of 40 μm on the arterial line and a roller pump (Stokert SV). The aorta was clamped by the surgeon and the Custodiol cardioplegic solution was perfused on the ascending aorta as a single shot. CO₂was continuously delivered at 3 to 5 l/min into operative field until the left atrium was de-aired and closed.</p>", "<title>Abbreviations</title>", "<p>TCD: Transcranial Doppler; EEG: Electroencephalogram; TEE: Transoesophageal Echocardiography; MES: Microembolic Signals; MIMVS: Minimally Invasive Mitral Valve Surgery; CPB: Cardiopulmonary Bypass.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>PZ conceived the work, carried out the study, collected and analyzed the data and wrote the article. EB, VS and LS analyzed the data and helped to write the article. CV and CS analysed the data. All authors read and approved the final manuscript.</p>", "<title>Consent</title>", "<p>Written informed consent was obtained from the patient for publication of this case report and accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal.</p>", "<title>Supplementary Material</title>" ]

[]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Gas bubbles like foam in the venous return line.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>MES from TCD obtained with double sampling boxes placed over the two middle cerebral arteries.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p>Microbubbles recorded from descending thoracic aorta three dimensional TEE.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p>Two dimensional TEE Pulse Wave Doppler recording of HITS in the descending thoracic aorta long axis view during CPB.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Gaseous and solid MES differentiation on the left and right meddle cerebral arteries.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\">Left</td><td align=\"center\">Right</td><td align=\"center\">Total</td></tr></thead><tbody><tr><td align=\"left\">Gasseous</td><td align=\"center\">178</td><td align=\"center\">220</td><td align=\"center\">398</td></tr><tr><td align=\"left\">Solid</td><td align=\"center\">35</td><td align=\"center\">14</td><td align=\"center\">49</td></tr><tr><td align=\"left\">Total</td><td align=\"center\">213</td><td align=\"center\">234</td><td align=\"center\">447</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional file 1</title><p>Movie 1. Simultaneously recording of EEG, microbubbles with 3d TEE and MES with TCD during CPB. The audio is from the TCD: is it possible to distinguish the sound from MES and the one from blood flow because of the roller pump.</p></caption></supplementary-material>" ]

[]

[ "<graphic xlink:href=\"1757-1626-1-141-1\"/>", "<graphic xlink:href=\"1757-1626-1-141-2\"/>", "<graphic xlink:href=\"1757-1626-1-141-3\"/>", "<graphic xlink:href=\"1757-1626-1-141-4\"/>" ]

[ "<media xlink:href=\"1757-1626-1-141-S1.AVI\" mimetype=\"text\" mime-subtype=\"plain\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Dietary supplements, herbs, and vitamins are used widely among HIV-infected patients [##REF##11304425##1##]. However, the safety or efficacy of many of these therapies has not been formally evaluated.</p>", "<p>These agents are often used among HIV-infected patients to prevent or treat the adverse effects of antiretroviral therapy [##REF##12737639##2##]. Among these adverse effects are disorders of glucose metabolism, including insulin resistance[##REF##15911733##3##], glucose intolerance [##REF##11118392##4##], and diabetes mellitus[##REF##15911733##3##], which were described soon after the introduction of highly active antiretroviral therapy (HAART)[##REF##9244318##5##]. Although the etiology of these problems is multifactorial [##REF##17064644##6##], exposure to protease inhibitors (PIs) likely contributes directly. Indinavir (IDV), for example, can worsen insulin sensitivity even after a single dose in healthy volunteers[##REF##11964551##7##].</p>", "<p>Ginseng is one of the most commonly used herbs in the United States [##REF##16092887##8##] and has long been used for the treatment of hyperglycemia in Traditional Chinese Medicine [##UREF##0##9##]. Experimental evidence for its efficacy comes from several studies in both animals and humans [##REF##16047557##10##]. In two small clinical trials, <italic>Panax ginseng </italic>[##REF##8721940##11##] and AG [##REF##11603646##12##] given daily over 8 weeks significantly reduced fasting blood glucose and hemoglobin A1c in patients with type 2 diabetes. While the mechanisms by which ginseng affects glucose metabolism have not been fully elucidated, most recent evidence from animal models suggests that improvement in insulin sensitivity may underlie its hypoglycemic effects [##REF##12031973##13##]. It is not known whether ginseng can reverse insulin resistance induced by PIs.</p>", "<p>Herbals treatments, like ginseng, are perceived to be safe by patients, but pharmacologic interactions with concomitant conventional drugs are a major concern. Ginseng was recently found to reduce the anticoagulant effect of warfarin [##REF##15238367##14##]. However, its potential interaction with antiretrovirals has not been investigated. Other herbs, such as St. Johns' wort [##REF##10683007##15##] and garlic [##REF##11740713##16##] dramatically reduce the concentrations of PIs in healthy volunteers, in the case of St. John's wort, through the induction of their common metabolic pathway, cytochrome P450 3A4 (CYP450 3A4). This could lead to a decrease in effectiveness and potentially result in treatment failure.</p>", "<p>The goals of this study were to evaluate potential PK interactions between IDV and AG, and assess whether AG improves IDV-induced insulin resistance.</p>" ]

[ "<title>Methods</title>", "<title>Study Participants</title>", "<p>Healthy volunteers, male and female, 18–64 years of age, were recruited through newspaper advertisement, flyers and word of mouth. Exclusion criteria included a history of nephrolithiasis, use of any prescription medications, over the counter medications, or dietary supplement within 14 days of enrollment, body mass index (BMI, weight (kg)/height (m²)) ≥ 30 kg/m², or blood donation within 30 days of study enrollment. Prior to enrollment, subjects were required to have a negative HIV-1 antibody test, normal values for serum creatinine, aspartate aminotransferase (AST), alanine aminotransferase (ALT), and hemoglobin, and a normal physical examination. The study was approved by the Institutional Review Board at Johns Hopkins University School of Medicine and all participants signed informed consent.</p>", "<title>Study Design</title>", "<p>After initial screening, participants were admitted to the Johns Hopkins inpatient General Clinical Research Center (GCRC) on three separate occasions over a 21 day period for PK and metabolic studies. During the first inpatient visit, baseline insulin sensitivity was assessed. Subjects were then discharged with instructions to take IDV 800 mg by mouth (Crixivan, Merck and Company, Rahway, New Jersey, USA) every 8 hours (0600, 1400, 2200) on an empty stomach beginning three days before the subsequent inpatient visit. During the second inpatient visit, insulin sensitivity was reassessed and an 8-hour PK IDV sampling was obtained. Subjects were then instructed to take IDV, as taken previously, and to co-administer encapsulated AG ground root, 1 gram by mouth every 8 hours. Outpatient safety visits with laboratory assessments occurred 7 days before and 7 days after the final inpatient visit.</p>", "<p>Dried whole-root AG was obtained from a single batch through the Wisconsin Ginseng Board (Wausau, Wisconsin). The identity of the dried whole root of AG was verified by the Wisconsin ginseng Board prior to encapsulation. Following verification the dried whole-root AG was ground and encapsulated in 500 mg capsules by the American Pharmaceutical Nutraceutical Laboratories (Wausau, Wisconsin) and dispensed by the Johns Hopkins Hospital Investigational Drug service. AG was administered under Investigational New Drug Application # 69,866 granted by the Food and Drug Administration. The dose of AG was selected based on previous studies investigating the effect of AG on glucose tolerance in humans[##REF##11603646##12##,##REF##11273850##17##]. The dose of IDV was selected according to its package insert. Following 14 days of IDV and AG, subjects returned to the inpatient GCRC for reassessment of insulin sensitivity and IDV PK.</p>", "<title>Ginseng Analyses</title>", "<p>The concentrations of six common ginsenosides (Rg1, Re, Rb1, Rc, Rb2, and Rd) were measured, following procedures described by Chuang [##REF##17238100##18##], in the Analytical Laboratory of the Johns Hopkins University Division of Clinical Pharmacology. Briefly, five different capsules were randomly selected from five separate bottles. The content of each capsule was extracted with 3 mL of ethanol:water (50:50, vol:vol), sonicated, and centrifuged. The pellet was extracted two more times and the resulting liquid was filtered. Samples and ginsenoside standards (Sigma-Aldrich, St. Louis, Missouri) were analyzed by High Performance Liquid Chromatography (HPLC) using Beckman Ultrasphere 5 μm, 250 × 4.6 mm column. The amount of each ginsenoside assayed in the capsules was as follows (expressed as percent per gram of ginseng): Rg1 0.11, Re 1.06, Rb1 0.91, Rc 0.08, Rb2 0.19, Rd 0.08, Total ginsenosides 2.43. The intra- and interassay coefficients of variation (CV) were &lt; 10.5% and &lt; 8.5%, respectively with &gt; 85% accuracy.</p>", "<title>Bioassay for Hypoglycemic Activity</title>", "<p>The hypoglycemic activity of the batch of AG used in the study was assessed in diabetic mice prior to initiation of the clinical studies, as previously described [##REF##12031973##13##]. Briefly, male <italic>ob/ob </italic>mice (n = 6) received AG extract dissolved in distilled water by daily intraperitoneal injection at a dose of 250 mg/kg. Fasting serum glucose was measured from blood obtained from tail samples at baseline, day 5 and day 12. The hypoglycemic response in AG-treated mice was compared to that observed in <italic>ob/ob </italic>mice who received vehicle following the same protocol. Animals were fed rodent chow and housed in environmentally stable conditions in metabolic cages. Food consumption and weight were monitored daily. All animal experiments were carried out at the Tang Center for Herbal Medicine Research, University of Chicago in the laboratory of Chun-Su Yuan, MD, PhD.</p>", "<title>PK Sampling and Analysis</title>", "<p>During the second and third inpatient GCRC visits, nine blood samples were collected immediately before administration of the morning dose of IDV and at 0.5, 1,2,3,4,5,6, and 8 hours post-dose after an overnight fast. The previous observed dose was given at 2200. PK data were analyzed using non-compartmental methods using WinNonlin^®(Pharsight, Cary, NC). IDV PK parameter estimates included area under the plasma-concentration-time curve from time 0 to 8 hours after the dose, (AUC_{0–8 hrs}), maximum plasma concentration (C_max), minimum plasma concentration (C_min), and time to maximum plasma concentration (T_max).</p>", "<title>Metabolic Assessments</title>", "<p>Body composition, including fat-free mass (FFM), was assessed at baseline using dual x-ray absorptiometry (Hologic-QDR 4500W, Hologic Co, Bedford, MA, intra-subject CV &lt; 2%). BMI was assessed during each inpatient admission and was calculated as the weight in kilograms/height in meters squared.</p>", "<p>Peripheral sensitivity to glucose was assessed by the hyperinsulinemic euglycemic clamp technique [##REF##382871##19##]. Subjects were admitted to the GCRC in the afternoon prior to the clamp procedure and maintained on a eucaloric diet (50% carbohydrate, 30% fat, and 20% protein). After an overnight fast, an intravenous catheter was placed in the antecubital vein for infusion of glucose and insulin. A second catheter for blood sampling was inserted in a dorsal vein of the hand in a retrograde fashion. The hand was placed into a 50°C warming box to \"arterialize\" the samples [##REF##985402##20##]. After baseline blood samples were obtained, a primed, continuous infusion of regular insulin (40 mU/m²/min) was started, followed by a constant infusion of 20% dextrose at a variable rate, based on 5-minute measurements of blood glucose, in order to maintain blood glucose concentrations at each subject's baseline level ± 5%. Infusion continued for a total of 180 minutes.</p>", "<p>Glucose utilization (M) was calculated as the average glucose infusion rate (milligrams of glucose/kilograms of FFM/minute) during the last 60 minutes of the infusion (i.e. steady state portion). FFM was derived from the baseline evaluation by dual energy X-ray absorptiometry (DXA).</p>", "<p>During the final 60 minutes of the insulin clamp procedure, samples were obtained at 10 minutes intervals to determine insulin concentrations. M divided by the average insulin concentration in the final 60 minutes (I), which represents the amount of glucose metabolized per unit of insulin, was used as the primary measure of insulin sensitivity. During the second and third inpatient visits, the morning dose of IDV (and AG for the third inpatient visit) was given immediately prior to the beginning of the insulin clamp procedure. Fasting serum glucose and homeostasis model assessment of insulin resistance (HOMA-IR), a widely used marker of insulin sensitivity [##REF##3899825##21##], were assessed prior to the insulin clamp procedure at each of the inpatient study visits.</p>", "<p>The serum glucose concentration during the final 60 minutes of the clamp was 89.8 ± 1.0 mg/dL which was 97.8 ± 0.3% of the targeted glucose concentration with a coefficient of variation of 4.2 ± 0.3% (n = 32 clamp studies). The average insulin concentration during the final 60 minutes of the infusion was 69.2 ± 1.7 μU/mL.</p>", "<title>Measurements</title>", "<p>Serum glucose was measured using the glucose oxidase method (Beckman Instruments, Fullerton, CA). Serum insulin concentrations were determined by enzyme immunoassay, using a TOSOH 1800 (Tosoh Corporation, Tokyo, Japan). Intra-assay and inter-assay CVs range from 1.4–2.3% and 5–6%, respectively. IDV concentrations were measured by HPLC-mass spectrometry. The intra- and inter-assay CVs were &lt; 9% and &lt; 8%, respectively with &gt; 94% accuracy.</p>", "<title>Statistical Calculations</title>", "<p>Paired comparisons between outcome measures obtained during the first and second inpatients visits (i.e. Baseline vs. IDV Condition) and during the second and third inpatient visits (IDV vs IDV + AG) were made using a paired t-test. Univariate relationships between continuous variables were assessed with linear regression. All data are presented as mean ± standard error of the mean (SEM). Differences were regarded as statistically significant if p &lt; 0.05. All data analysis was performed using Stata (Version 8.1, Stata Corporation, College Station, TX). Pharmacokinetic parameter estimates were summarized using geometric means with 95% confidence intervals; values obtained with and without AG were compared using a geometric mean ratio and 90% confidence interval.</p>" ]

[ "<title>Results</title>", "<title>Preclinical Studies</title>", "<p>In <italic>ob/ob </italic>mice receiving AG extract, fasting glucose concentrations decreased (Baseline vs Day 12; 235 ± 11 mg/dL vs 193 ± 9 mg/dL, between group t-test, p = 0.003), but remained constant in the vehicle-treated group (243 ± 10 mg/dL vs 251 ± 12 mg/dL) (Figure ##FIG##0##1##). Weight did not change over the study interval in the AG-treated mice (data not shown).</p>", "<title>Clinical Studies</title>", "<p>Fourteen healthy volunteers were evaluated (Table ##TAB##0##1##). All participants were male, 12 were African-American, and two were white. Age ranged from 26–53 years with a mean of 42.9 ± 1.9. The average BMI was 26.1 ± 0.8 kg/m²(range 21.9–29.9).</p>", "<title>PK Analysis</title>", "<p>Thirteen participants were included in the analysis of the comparison of IDV PK with and without AG. Subject 9 was noted to develop transaminitis after the second inpatient study visit and did not complete the remainder of the study. Table ##TAB##1##2## shows the average IDV PK parameters after 3 days of IDV following the morning dose, and after co-administration of 14 days of IDV and AG. All PK parameters were similar in the two periods.</p>", "<title>Metabolic Studies</title>", "<title>Insulin Sensitivity: Effect of IDV</title>", "<p>Eleven subjects were included in the analysis to assess the effect of IDV on insulin sensitivity. The calculated M and M/I values for subject 4–14 is presented in Table ##TAB##2##3##. The data from subjects 1–3 were excluded since the protocol was changed after these subjects completed their participation in protocol. The protocol was changed to better match peak IDV concentrations with the steady state portion of hyperinsulinemia during the insulin clamp. Specifically, for subjects 4 – 14, the morning dose of IDV was administered immediately prior to the beginning of the insulin clamp procedure, rather than two hours before the start of the insulin clamp as originally planned and executed for subjects 1 – 3.</p>", "<p>Compared to baseline measurements, IDV was associated with an average decrease in insulin sensitivity (M/I) of 14.8 ± 5.9%, from 0.113 ± 0.012 to 0.096 ± 0.014 mg/kg FFM/min per μU/ml of insulin, p = 0.03) (Figure ##FIG##1##2##). We also detected a decrease in average glucose utilization (M) with IDV administration when compared to baseline (from 7.73 ± 0.75 to 6.32 ± 0.78 mg/kg FFM/min, p = 0.005). Fasting glucose and HOMA-IR did not change when comparing baseline to the second admission (92.4 ± 1.8 mg/dL vs 95.0 ± 2.4 mg/dL, p = 0.27; 1.34 ± 0.32 vs 1.34 ± 0.23, p = 0.97, respectively).</p>", "<title>Insulin Sensitivity: Effect of AG</title>", "<p>Ten subjects were included to assess the effect of AG on IDV-induced insulin resistance (Table ##TAB##2##3##). One subject (subject 9) did not have complete data for this analysis because he was discontinued from the study after the second insulin clamp procedure, as previously noted. There was no difference in insulin sensitivity between the second (IDV alone) and third (IDV + AG) insulin clamp procedures, 0.102 ± 0.013 vs 0.099 ± 0.016 mg/kg FFM/min per μU/ml of insulin, respectively (p = 0.72) (Figure ##FIG##1##2##). Glucose utilization (M) with IDV administration also did not change when AG was co-administered, 6.68 ± 0.76 vs 6.47 ± 0.84 mg/kg FFM/min, respectively (p = 0.62).</p>", "<p>Fasting glucose and HOMA-IR did not change after 2 weeks of AG (93.0 ± 1.4 mg/dL vs 92.0 ± 2.6 mg/dL, p = 0.65; 1.4± 0.25 vs 1.4 ± 0.28, p = 0.95). Body weight did not change during the trial (baseline vs third inpatient study visit, 83.1 ± 3.1 kg vs 83.2 ± 3.1 kg, p = 0.88).</p>", "<title>Insulin Sensitivity: Effect of AG after Normalization for IDV Concentrations</title>", "<p>Because of the variability of IDV concentrations with and without co-administration of AG, we evaluated the potential effect of AG on insulin sensitivity after adjustment for IDV concentrations. In a <italic>post hoc </italic>analysis, we standardized the measurement of insulin sensitivity for IDV concentrations by dividing M/I for the second and third insulin clamp procedures by the corresponding IDV AUC between 2–3 hours (AUC_2–3). We then multiplied this value by 10⁶for ease of interpretation.</p>", "<p>We chose the 2–3 hour IDV AUC since it is the same time period during the clamp procedure in which insulin sensitivity is assessed and there was a trend toward correlation between insulin sensitivity (M/I) and IDV AUC_2–3during the third insulin clamp procedure (r = 0.58, p = 0.078). The IDV AUC_2–3was not correlated with M/I (r = 0.04, p = 0.90) during the second clamp procedure (IDV alone). No other IDV PK parameters or other fractional AUCs were correlated with M/I for either the second or third clamp procedure (data not shown).</p>", "<p>For the 10 subjects with measurable M/I estimates normalized for IDV AUC_2–3, the average difference in IDV AUC_2–3was similar when comparing the period of IDV alone to the period with concomitant IDV and AG (2353 ± 389 vs 1923 ± 483 ng·hr/mL; difference: 430 ± 330 ng·hr/mL, p = 0.22). When each subject's M/I was divided by the IDV AUC_2–3, which can be interpreted as insulin sensitivity per unit of plasma IDV, a significant increase was seen after AG co-administration (60.9 ± 15.4 vs 92.2 ± 20.9 mg Glucose·10⁶/kg FFM/min/μU/ml of insulin/ng·hr/mL IDV, p = 0.03). A similar increase was seen when comparing M (multiplied times 10⁶) divided by the IDV AUC_2–3between the IDV alone period and the concomitant IDV plus AG period (4240.6 ± 1180.0 vs 6100.5 ± 1526.8 mg Glucose·10⁶/kg FFM/min/ng·hr/mL IDV, p = 0.05).</p>", "<title>Safety Assessment</title>", "<p>Both IDV and AG were well tolerated. There were no serious adverse events. Three subjects developed transaminitis. In one subject (as previously mentioned), an AST elevation 3 times the upper limit of normal (Grade 2) after 3 days of IDV required study discontinuation. Three subjects were noted to have mild elevations in bilirubin. One subject had a mild increase in serum creatinine. One patient reported an episode of vomiting and one subject reported dyspepsia after dose administration. All laboratory abnormalities and symptoms normalized after discontinuation of the study drugs. With the exception of the Grade 2 transaminites, all adverse events occurred during the co-administration of IDV and AG.</p>", "<title>Adherence Assessment</title>", "<p>By pill count assessed on three occasions during the study (i.e., the second and third inpatient visit and the intervening outpatient safety visit), IDV overall mean adherence was 97.7 ± 1.5%. AG adherence, as assessed on two occasions (interim outpatient visit and visit three), was 96.6 ± 2.5%.</p>" ]

[ "<title>Discussion</title>", "<p>In this prospective study of healthy volunteers, we found that 14 days of co-administration of AG did not significantly impact IDV PK. We also found that, although IDV acutely reduced insulin sensitivity by an average of 15%, AG did not change insulin sensitivity in IDV-treated healthy volunteers, providing evidence against its use in the treatment of PI-induced disorders of glucose metabolism.</p>", "<p>Potential drug-herb interactions are an important safety concern for many clinicians and HIV-infected patients. Although herbal remedies are perceived as safe by many patients, some can inhibit and/or induce the CYP3A4 enzyme, the main metabolic pathway for most PIs and non-nucleoside reverse transcriptase inhibitors (NNRTIs), potentially leading to increased toxicity or therapeutic failure [##REF##10683007##15##,##REF##11740713##16##]. The metabolic pathways of ginseng components are not well known. Ginsenosides, which are steroid-like molecules of the saponin class, are thought to be the active compounds of all ginseng species [##REF##10571242##22##]. There are more than 20 different types of ginsenosides in ginseng root and their relative concentrations varies depending on the species, the batch, and the part of the plant assayed (eg. root vs berry)[##REF##12571655##23##,##REF##13678250##24##]. The ginsenosides also differ in their effects on metabolism. In <italic>in vitro </italic>models investigating the catalytic activity of c-DNA expressed CYP450 isoforms, the ginsenoside, Rd, has been shown to be a weak inhibitor of CYP3A4, while Rf increased CYP3A4 activity by 54% [##UREF##1##25##]. In addition, metabolites of ginsenosides as well as non-ginsenoside components of ginseng have also been shown to inhibit CYP3A4 in experimental models [##REF##16547074##26##,##REF##15266218##27##]. The clinical significance of these <italic>in vitro </italic>observations is not clear.</p>", "<p>In previous human studies, Siberian ginseng (<italic>Eleutherococcus senticosus</italic>) has been shown to increase concentrations of nifedipine [##UREF##2##28##], a CYP3A4 substrate, but does not affect concentrations of midazolam [##REF##12695337##29##], another CYP3A4 probe drug. It should be noted that Siberian ginseng is not a species of the genus <italic>Panax</italic>, contains no ginsenosides, and therefore the effects of Siberian ginseng are not generalizable to members of the <italic>Panax </italic>genus, such as AG. However, human studies in healthy volunteers using <italic>Panax ginseng </italic>(C.A. Meyer) have shown no interaction with midazolam [##REF##15974642##30##] and no change in urinary 6-beta-OH-cortisol/cortisol ratio [##REF##12817527##31##], both measures of CYP3A4 activity.</p>", "<p>We also found that 3 days of IDV administration decreased insulin sensitivity by 15% in healthy volunteers. Our findings are similar to another healthy volunteer study, showing a 17% decrease in insulin sensitivity with IDV administration (800 mg q 8 hours)[##REF##11399973##32##], but smaller in magnitude when compared to another study using a higher dose (34% decrease with a single dose of 1200 mg) [##UREF##3##33##]. In <italic>in vitro </italic>studies and animal models, PIs, such as IDV, impede glucose movement through the major glucose transporter in skeletal muscle, GLUT4, thereby inducing insulin resistance[##REF##17186064##34##]. The observations in humans, including ours, are consistent with this mechanism, although the extent to which it is clinically relevant in the pathogenesis of hyperglycemia among HIV-infected, HAART-treated patients remains unclear.</p>", "<p>Although we observed that insulin sensitivity decreased with IDV administration, we did not find any change in insulin sensitivity with co-administration of AG. In previous studies, Vuksan, <italic>et al</italic>. showed that AG administration immediately prior to an oral glucose tolerance test was associated with 20% decrease in glucose AUC in both patients with type 2 diabetes mellitus and healthy volunteers [##REF##10977009##35##], but this effect appears to be dependent on the batch of ginseng used [##REF##12571655##23##]. To rule out an inert batch of AG, we conducted a bioassay of our batch using an <italic>ob/ob </italic>mouse model and a hypoglycemic effect was observed, suggesting that our batch possessed some active components.</p>", "<p>The mechanism underlying the previously observed effects of AG on glucose metabolism remains unknown and may be multifactorial. The modulation of digestion and enhancement of insulin secretion have been proposed based on findings in some animals models [##REF##6759629##36##,##REF##11440080##37##], but ginseng administration has also been shown to improve insulin sensitivity in <italic>ob/ob </italic>mice by more than two-fold [##REF##12031973##13##]. It has been postulated that ginsenosides may intercalate into the cellular plasma membrane, thus modulating the cell signaling, electrolyte transport, and receptor binding [##REF##10571242##22##]. It is not known whether this effect could also modulate the effect of IDV on the glucose flux through the GLUT4 transporter.</p>", "<p>Another factor that may have contributed to the lack of effect is the variability of IDV plasma concentrations between the two clamps, which has been previously observed [##REF##9562584##38##]. The effect of AG on IDV-induced insulin resistance would be best determined if IDV plasma concentrations were the same when it was given alone and when it was co-administered with AG. For this reason, we normalized the measure of insulin sensitivity for drug concentration in a <italic>post hoc </italic>analysis, by dividing M/I by the IDV concentration during the steady state portion of the clamp and found that this ratio was significantly higher in the IDV + AG condition.</p>", "<p>Although the interpretation of ratios can be challenging [##REF##8574275##39##] and should be done with caution, one explanation of this finding is that insulin sensitivity per unit of IDV concentration modestly improved with the administration of AG. Given the known effect of IDV on GLUT4 blockade, AG components may work directly at the plasma membrane to allow glucose to enter cells, either by improving glucose movement through GLUT4, increasing the concentration of GLUT4 in the plasma membrane, or facilitating glucose entry through alternate pathways. Another possible mechanism of AGs effect may be through the enhancement of local blood flow. Increased capillary recruitment mediated though nitric oxide is an important mechanism of increasing glucose and insulin delivery to skeletal muscle [##REF##12791603##40##,##REF##16682488##41##]. IDV has been shown to cause endothelial dysfunction in healthy volunteers, likely by reducing nitric oxide production [##REF##16290967##42##]. Ginseng species have been shown in experimental models to increase nitric oxide production [##REF##12940876##43##,##REF##9296344##44##] and therefore, may effect insulin sensitivity by this mechanism. This hypothesis requires further investigation.</p>", "<p>Our study had additional limitations which may have implications for the generalizability of its findings. Although both men and women were eligible for participation, only men enrolled. In previous studies, inducibility of CYP3A4 by herbal compounds has shown an interaction by sex, whereby women showed a 74% increased effect of St. John's wort on CYP3A4 activity using a midazolam probe compared to men [##REF##12235448##45##]. In the same study, however, <italic>Panax ginseng </italic>showed no effect on the midazolam metabolism in either men or women. Further PK interaction studies in women may be necessary. In addition, our population was mostly African-American. Although there is evidence for genotypic differences in CYP3A4 between Caucasians and African-Americans, phenotypic differences in CYP3A4 activity have not been found [##REF##14515058##46##]. Finally, although we quantified the amount of common ginsenosides, there is substantial variability of composition even within species. As a result, as is the case in all botanical research using non-standardized complex compounds whose active component is unknown, the extent to which our findings are generalizable to other AG products is not clear.</p>", "<p>In conclusion, this study in healthy volunteers found no evidence of a significant PK interaction between AG and IDV. Because CYP3A4 is the principal metabolic pathway of most HIV PIs and NNRTIs, significant drug-herb interactions with these antiretrovirals are unlikely. The metabolic effects of AG are more difficult to interpret. Although IDV significantly reduced insulin sensitivity, there was no change in insulin sensitivity with AG administration. However, after normalization for IDV concentrations, insulin sensitivity improved in the AG condition, leaving open the possibility of a modest effect on glucose metabolism. Further studies to clarify this question should attempt to reduce variability in IDV plasma concentrations, either by giving multiple doses of IDV at shorter intervals to maintain plasma IDV concentrations constant while insulin sensitivity is assessed, or using an IDV regimen boosted with ritonavir. Until these issues are resolved, there is no clear scientific basis to recommend AG for the treatment of glucose abnormalities in HIV-infected patients.</p>" ]

[]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Complementary and alternative medicine (CAM) use is prevalent among HIV-infected patients to reduce the toxicity of antiretroviral therapy. Ginseng has been used for treatment of hyperglycemia and insulin resistance, a common side effect of some HIV-1 protease inhibitors (PI). However, it is unknown whether American ginseng (AG) can reverse insulin resistance induced by the PI indinavir (IDV), and whether these two agents interact pharmacologically. We evaluated potential pharmacokinetic interactions between IDV and AG, and assessed whether AG improves IDV-induced insulin resistance.</p>", "<title>Methods</title>", "<p>After baseline assessment of insulin sensitivity using the insulin clamp technique, healthy volunteers received IDV 800 mg q8 h for 3 days and then IDV and AG 1g q8h for 14 days. IDV pharmacokinetics and insulin sensitivity were assessed before and after AG co-administration.</p>", "<title>Results</title>", "<p>There was no difference in the area-under the plasma-concentration-time curve after the co-administration of AG, compared to IDV alone (n = 13). Although insulin-stimulated glucose disposal per unit of insulin (M/I) decreased by an average of 14.8 ± 5.9% after 3 days of IDV (from 0.113 ± 0.012 to 0.096 ± 0.014 mg/kgFFM/min per μU/ml of insulin, p = 0.03, n = 11), M/I remained unchanged after co-administration of IDV and AG.</p>", "<title>Conclusion</title>", "<p>IDV decreases insulin sensitivity, which is unaltered by AG co-administration. AG does not significantly affect IDV pharmacokinetics.</p>" ]

[ "<title>Competing interests</title>", "<p>Drs. Dobs, Parsons, Caballero, and Yuan declared that they have no competing interests. Dr. Hendrix received research support from Merck and Company. Dr. Andrade served as a consultant for Abbott Laboratories. Dr. Brown served as a consultant or has received research support from Merck and Company, Abbott Laboratories, Reliant Pharmaceuticals, Glaxo Smith Kline, EMD-Serono, and Theratechnologies. Dr. Flexner served on a Scientific Advisory Board for Merck and Company.</p>", "<title>Authors' contributions</title>", "<p>TTB performed the insulin clamps. C–SY carried out the animal studies. TLP carried out the measurements of ginsenoside and IDV plasma concentrations. TTB and ASAA drafted the manuscript. ASAA, TTB, CWF, CH C–SY, ASD and BC participated in the design of the study. TTB, ASAA, and CH performed the statistical analysis. ASAA, TTB, ASD, CH and BC participated in the data interpretation. TTB and ASAA conceived of the study and participated in its coordination. All authors read and approved the final manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1472-6882/8/50/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>We would like to thank Chryssanthi Stylianopoulos, PhD and Margia Arguello for their technical expertise in measuring glucose during the insulin clamp procedure and Alice Ryan, PhD for her guidance with the insulin clamp procedure. We are thankful to Andrea Weiss for dispensing study drugs and preparation of solutions used in the insulin clamp procedure. Finally, we also would like to acknowledge Dr Angela Kashuba for the analysis of the indinavir assays. Her work is supported by #9P30 AI 50410-04, University of North Carolina at Chapel Hill Center for AIDS Research (CFAR). This work was supported by a National Center for Complementary and Alternative Medicine supplement to the National Cancer Institute Grant # P30CA06973, the National Center for Complementary and Alternative Medicine K23 AT002862-01 (TTB), and the Johns Hopkins University School of Medicine General Clinical Research Grant M01-RR00052 from the National Center for Research Resources/National Institutes of Health.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Fasting Glucose Concentrations in Leptin-Deficient (<italic>ob/ob</italic>) Mice Treated with AG-Extract (250 mg/kg) and Vehicle Over 12 Days (n = 6 in both groups).</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>Insulin Sensitivity (Glucose Infusion Rate Per Kilogram of Free-Fat Mass Per Unit of Insulin) at Baseline, After 3 Days of Indinavir, and After 14 Days of Co-Administration of AG in Healthy Volunteers (p-values represent the average mean difference by paired t-test).</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Demographic Characteristics of Study Participants</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\"><bold>Subject #</bold></td><td align=\"center\"><bold>Race</bold></td><td align=\"center\"><bold>Age (years)</bold></td><td align=\"center\"><bold>BMI (kg/m²)</bold></td><td align=\"center\"><bold>Fat (%)</bold></td></tr></thead><tbody><tr><td align=\"center\">1</td><td align=\"center\">AA</td><td align=\"center\">41</td><td align=\"center\">29.9</td><td align=\"center\">29.5</td></tr><tr><td align=\"center\">2</td><td align=\"center\">AA</td><td align=\"center\">43</td><td align=\"center\">22.4</td><td align=\"center\">12.2</td></tr><tr><td align=\"center\">3</td><td align=\"center\">AA</td><td align=\"center\">41</td><td align=\"center\">22.3</td><td align=\"center\">12.6</td></tr><tr><td align=\"center\">4</td><td align=\"center\">AA</td><td align=\"center\">53</td><td align=\"center\">26.6</td><td align=\"center\">23.0</td></tr><tr><td align=\"center\">5</td><td align=\"center\">W</td><td align=\"center\">26</td><td align=\"center\">29.0</td><td align=\"center\">25.2</td></tr><tr><td align=\"center\">6</td><td align=\"center\">AA</td><td align=\"center\">49</td><td align=\"center\">29.9</td><td align=\"center\">32.1</td></tr><tr><td align=\"center\">7</td><td align=\"center\">AA</td><td align=\"center\">43</td><td align=\"center\">24.7</td><td align=\"center\">22.3</td></tr><tr><td align=\"center\">8</td><td align=\"center\">AA</td><td align=\"center\">44</td><td align=\"center\">21.9</td><td align=\"center\">21.1</td></tr><tr><td align=\"center\">9</td><td align=\"center\">AA</td><td align=\"center\">36</td><td align=\"center\">29.1</td><td align=\"center\">29.9</td></tr><tr><td align=\"center\">10</td><td align=\"center\">AA</td><td align=\"center\">37</td><td align=\"center\">24.0</td><td align=\"center\">24.7</td></tr><tr><td align=\"center\">11</td><td align=\"center\">AA</td><td align=\"center\">52</td><td align=\"center\">23.4</td><td align=\"center\">15.8</td></tr><tr><td align=\"center\">12</td><td align=\"center\">AA</td><td align=\"center\">42</td><td align=\"center\">29.3</td><td align=\"center\">19.7</td></tr><tr><td align=\"center\">13</td><td align=\"center\">AA</td><td align=\"center\">46</td><td align=\"center\">26.7</td><td align=\"center\">26.7</td></tr><tr><td align=\"center\">14</td><td align=\"center\">W</td><td align=\"center\">49</td><td align=\"center\">26.5</td><td align=\"center\">26.1</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>PK parameter estimates of IDV before and after co-administration with AG (1 gram every 8 hours), n = 13</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><bold>IDV alone (Day 7) </bold><break/><bold>GM (95%CI)</bold></td><td align=\"center\"><bold>IDV + AG (Day 22) </bold><break/><bold>GM (95%CI)</bold></td><td align=\"center\"><bold>Day 7/Day 22 </bold><break/><bold>Geometric Mean Ratio </bold><break/><bold>(90% CI)</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>C_max(ng/mL)</bold></td><td align=\"center\">5623 (4027–7853)</td><td align=\"center\">5633 (4908–6464)</td><td align=\"center\">1.00 (0.77–1.29)</td></tr><tr><td align=\"left\"><bold>C_min(ng/mL)</bold></td><td align=\"center\">61.97 (42.72–89.88)</td><td align=\"center\">62.78 (39.72–99.24)</td><td align=\"center\">1.00 (0.77–1.29)</td></tr><tr><td align=\"left\"><bold>t_max(hours)</bold></td><td align=\"center\">0.98 (0.7189–1.3403)</td><td align=\"center\">0.74 (0.5527–0.9870)</td><td align=\"center\">0.75 (0.53–1.06)</td></tr><tr><td align=\"left\"><bold>AUC_0–8(ng·h/dL)</bold></td><td align=\"center\">12171 (8628–17168)</td><td align=\"center\">10700 (8515–13649)</td><td align=\"center\">0.89 (0.72–1.08)</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Glucose Utilization (M) and Insulin Sensitivity (M/I) Measured by the Hyperinsulinemic Euglycemic Clamp in Healthy Volunteers at Baseline, After 3 Days of IDV, and After Co-Administration of IDV and AG (1 gram every 8 hours) for 14 Days</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Subject</bold></td><td align=\"center\" colspan=\"3\"><bold>M </bold><break/>(mg/kg FFM/min)</td><td align=\"center\" colspan=\"3\"><bold>M/I </bold><break/>(mg/kg FFM/min per μU/ml of insulin)</td></tr></thead><tbody><tr><td/><td align=\"center\"><bold>Baseline</bold></td><td align=\"center\"><bold>IDV Alone</bold></td><td align=\"center\"><bold>IDV + AG</bold></td><td align=\"center\"><bold>Baseline</bold></td><td align=\"center\"><bold>IDV Alone</bold></td><td align=\"center\"><bold>IDV + AG</bold></td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"left\"><bold>4</bold></td><td align=\"center\">8.25</td><td align=\"center\">5.69</td><td align=\"center\">5.79</td><td align=\"center\">0.121</td><td align=\"center\">0.072</td><td align=\"center\">0.072</td></tr><tr><td align=\"left\"><bold>5</bold></td><td align=\"center\">8.46</td><td align=\"center\">8.20</td><td align=\"center\">10.35</td><td align=\"center\">0.135</td><td align=\"center\">0.138</td><td align=\"center\">0.193</td></tr><tr><td align=\"left\"><bold>6</bold></td><td align=\"center\">4.58</td><td align=\"center\">3.93</td><td align=\"center\">4.01</td><td align=\"center\">0.055</td><td align=\"center\">0.049</td><td align=\"center\">0.059</td></tr><tr><td align=\"left\"><bold>7</bold></td><td align=\"center\">11.52</td><td align=\"center\">7.14</td><td align=\"center\">7.82</td><td align=\"center\">0.173</td><td align=\"center\">0.141</td><td align=\"center\">0.134</td></tr><tr><td align=\"left\"><bold>8</bold></td><td align=\"center\">8.88</td><td align=\"center\">9.02</td><td align=\"center\">7.29</td><td align=\"center\">0.132</td><td align=\"center\">0.155</td><td align=\"center\">0.118</td></tr><tr><td align=\"left\"><bold>9</bold></td><td align=\"center\">4.10</td><td align=\"center\">2.74</td><td align=\"center\">-</td><td align=\"center\">0.062</td><td align=\"center\">0.037</td><td align=\"center\">-</td></tr><tr><td align=\"left\"><bold>10</bold></td><td align=\"center\">10.14</td><td align=\"center\">9.54</td><td align=\"center\">9.94</td><td align=\"center\">0.119</td><td align=\"center\">0.126</td><td align=\"center\">0.123</td></tr><tr><td align=\"left\"><bold>11</bold></td><td align=\"center\">9.60</td><td align=\"center\">8.62</td><td align=\"center\">5.93</td><td align=\"center\">0.157</td><td align=\"center\">0.115</td><td align=\"center\">0.107</td></tr><tr><td align=\"left\"><bold>12</bold></td><td align=\"center\">6.19</td><td align=\"center\">4.24</td><td align=\"center\">3.51</td><td align=\"center\">0.078</td><td align=\"center\">0.062</td><td align=\"center\">0.042</td></tr><tr><td align=\"left\"><bold>13</bold></td><td align=\"center\">8.62</td><td align=\"center\">7.82</td><td align=\"center\">7.66</td><td align=\"center\">0.137</td><td align=\"center\">0.123</td><td align=\"center\">0.113</td></tr><tr><td align=\"left\"><bold>14</bold></td><td align=\"center\">4.60</td><td align=\"center\">2.62</td><td align=\"center\">2.37</td><td align=\"center\">0.068</td><td align=\"center\">0.038</td><td align=\"center\">0.030</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>AA – African-American; W-White</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1472-6882-8-50-1\"/>", "<graphic xlink:href=\"1472-6882-8-50-2\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Lactational mastitis is a painful, debilitating condition that can adversely affect mothers in their efforts to breastfeed their babies [##REF##9699467##1##,##REF##12019051##2##]. Despite being a relatively common complication of lactation, surprisingly few studies documenting the incidence of, and risk factors for, the condition have been reported. To our knowledge no large longitudinal study of mastitis has been conducted in the UK in recent times. Observational studies conducted in the USA [##REF##11790672##3##], Finland [##REF##8092782##4##] New Zealand [##REF##10655826##5##] and Australia [##REF##9785526##6##,##REF##9775842##7##] suggest however, that up to 20–25% of breastfeeding women will develop mastitis during the course of lactation and approximately 20–35% of women who develop mastitis will experience recurrent episodes [##REF##10655826##5##, ####REF##9785526##6##, ##REF##9775842##7##, ##UREF##0##8####0##8##].</p>", "<p>In some studies, mastitis has been associated with the premature cessation of breastfeeding [##REF##9699467##1##,##REF##12019051##2##]. Sufferers of mastitis may stop breastfeeding because of the pain associated with the condition or because they have been inappropriately advised by a health professional to do so. Health professionals therefore, have a responsibility to provide appropriate advice and support to assist women to manage the condition successfully and to continue breastfeeding.</p>", "<p>The purpose of this paper is to report the incidence of mastitis in the first six months postpartum in a Scottish population, its impact on breastfeeding duration and to describe the type and appropriateness of the support and management received by affected women from health professionals.</p>" ]

[ "<title>Methods</title>", "<p>Study participants were recruited from women who had given birth at the Princess Royal Maternity Hospital in Glasgow between April 2004 and January 2005. Women who were breastfeeding at the time of recruitment and had delivered a full-term, healthy singleton were eligible to participate. Those who were fully formula feeding at the time of recruitment, lived outside the Greater Glasgow Conurbation, had a multiple delivery, a low birth weight infant (&lt; 2500 g), a premature delivery (&lt; 37 weeks gestation) or who could not read and speak English were excluded from the study.</p>", "<p>Women meeting the inclusion criteria were visited on the maternity ward by the research midwife (SM) within 24–48 hours of delivery. They were given a written and verbal description of the study and were informed that they could decline to participate in, or withdraw from, the study without prejudice to their care. The study was approved by the Local Research Ethics Committee of the North Glasgow University Hospitals NHS Trust and written informed consent was obtained from participants.</p>", "<p>Participants completed a baseline questionnaire prior to or shortly after discharge from hospital, providing information on demographic characteristics and previous births, breastfeeding experience and episodes of mastitis. Women were followed up by telephone interview at 3, 8, 18 and 26 weeks or up to the time that they discontinued breastfeeding. Follow-up interviews included questions on breastfeeding practices, breast and feeding related problems, breast care, use of breast pumps and maternal and infant health.</p>", "<title>Case definition</title>", "<p>Mastitis was defined as a red, tender, hot, swollen area of the breast, accompanied by one or more of the following [##REF##9785526##6##]:</p>", "<p>i) an elevated temperature (either estimated or measured as being ≥ 38°C ) or</p>", "<p>ii) one or more of the constitutional symptoms of fever (body aches, headaches and chills) or</p>", "<p>iii) diagnosis of mastitis from a medical practitioner.</p>", "<p>Symptoms had to have been present for a minimum duration of 24 hours [##REF##9775842##7##].</p>", "<p>Women who experienced an episode of mastitis post-discharge were instructed to telephone the research team to inform them of the event if their symptoms persisted for 24 hours or more. Additional cases of mastitis were identified retrospectively at the time of the telephone follow-up interviews. Those identified by either method as having mastitis were requested to complete and return a 'Mastitis Case' questionnaire, a copy of which had been given to them as part of an information pack at the time of recruitment (see Additional file). This questionnaire elicited information about events leading up to the episode of mastitis, management advice received and followed, and outcome.</p>", "<title>Statistical analysis</title>", "<p>Data were analysed using the Statistical Package for the Social Sciences (SPSS for Windows Version 15) [##UREF##1##9##]. The estimated incidence of mastitis was based on women who had symptoms meeting our criteria who either directly contacted the research team or who were identified at one of the follow-up interviews. The target sample size of 500 had the power to detect an incidence of mastitis of 20% with an accuracy of 3.5%. The achieved analysis population of 420 had the power to detect an incidence of 20% with an accuracy of 3.8%.</p>", "<p>Descriptive statistics include means and standard deviations or percentages. Survival analysis and the log-rank test were performed to determine if there was a difference in the duration of breastfeeding between women who developed mastitis and women who did not.</p>" ]

[ "<title>Results</title>", "<p>During the study period 1141 breastfeeding women were identified. Of these, 183 were missed or discharged prior to completion of recruitment and 372 did not meet the eligibility criteria. The remaining 586 women were invited to participate and of these 40 women refused to participate, a further 46 agreed to participate but provided incorrect or insufficient follow-up contact details and a further 80 discontinued breastfeeding within two weeks of delivery. In most cases women discontinuing breastfeeding before 2 weeks failed to establish breastfeeding successfully due to feeding difficulties. As an objective of the study was to identify determinants of mastitis (to be reported separately), any woman discontinuing breastfeeding before 2 weeks for reasons other than mastitis was excluded from the analysis population as their inclusion might reduce or mask the association of feeding difficulties and mastitis.</p>", "<title>Incidence of mastitis</title>", "<p>In total, 420 women were included in the analysis population (response rate 72%) and 74 women (18%, 95%CI: 14%, 21%) were identified as having experienced at least one episode of mastitis in the first 26 weeks postpartum. Mastitis was not significantly more common in women with a prior history of mastitis compared with those with no prior history (Table ##TAB##0##1##). Fifty-seven women (77% of cases) completed the Mastitis Case questionnaire, 39 of whom (68%) reported only one episode of mastitis but 13 women (23%) reported two episodes and five women (9%) three or more episodes of mastitis. In total 76 episodes of mastitis were reported by the 57 women completing the Mastitis Case Questionnaire. Thirty initial episodes (53%) of mastitis, and 33 (43%) of all episodes, occurred within the first 4 weeks postpartum (Figure ##FIG##0##1##). The mean time to first episode of mastitis was 6.3 (SD 6.6) weeks.</p>", "<title>Mastitis and breastfeeding duration</title>", "<p>Of the 74 women who developed mastitis 28 (38%) stopped breastfeeding before the end of the study (26 weeks). The duration from time of mastitis to stopping breastfeeding was 6.1 (SD 6.6) weeks. Of the other 46 women who developed mastitis, 41 were continuing to breastfeed at 26 weeks while the remaining 5 were lost to follow-up during the study but were breastfeeding at the time of last study contact. The occurrence of mastitis was not negatively associated with breastfeeding duration (Figure ##FIG##1##2##). In fact, those women who suffered an episode of mastitis were significantly more likely to be breastfeeding at 26 weeks than those who did not suffer mastitis (Log-rank test χ²= 8.81, df = 1, p = 0.003).</p>", "<title>Advice and treatment</title>", "<p>Information on advice and treatment was available for the 57 women who completed the Mastitis Case questionnaire. Of these women, 21 (37%) were able to manage their first episode of mastitis without consulting a health professional. Thirty six women sought advice from a health professional: 21 (37%) from their General Practitioner (GP), 18 (32%) from a Community Midwife and 12 (21%) from a Health Visitor. Twelve women (21%) consulted more than one health professional, possibly indicating that more serious cases were referred on to a woman's GP by her Community Midwife or Health Visitor.</p>", "<p>Thirty (53%) of those women who completed a questionnaire and 78% (28/36) of women who reported consulting a health professional were prescribed an antibiotic. Of these, nine women (30%) were prescribed flucloxacillin, five (17%) amoxicillin, two (7%) erythromycin, two (7%) co-amoxiclav (amoxicillin and clavulanic acid), one (3%) ceftriaxone (a third generation cephalosporin) and one (3%) co-fluampicil (ampicillin and flucloxacillin). Ten women (33%) could not recall what antibiotic they were prescribed.</p>", "<p>The type of advice given to women by their health professional is listed in Table ##TAB##1##2##. Six women (10%) were told to stop feeding either from the affected breast or altogether.</p>" ]

[ "<title>Discussion</title>", "<p>The incidence of mastitis reported in this study is comparable to the incidence rates of 17.3% and 20% reported in recent Australian studies [##REF##9699467##1##,##REF##9785526##6##,##REF##1957190##10##]. The only other study of Scottish women identified was conducted in the 1940s and reported an incidence of breast abscess of 8.6% [##REF##1174212##11##]. However, as non-suppurative forms of mastitis were not identified, the real incidence of mastitis was likely to be higher. This and other earlier studies [##UREF##0##8##,##UREF##2##12##] probably underestimated the true incidence due to limitations in case ascertainment and the short time period that women were followed postpartum. For instance, earlier studies identified only those women with mastitis who sought medical treatment from the hospital where they were delivered and used hospital medical records as their source of data [##UREF##0##8##,##UREF##2##12##]. A more recent US study [##REF##11790672##3##] reported a 9.5% incidence of health care provider-diagnosed lactational mastitis. In our study only one woman reported visiting a hospital casualty department with approximately one third of women visiting their GP. The majority of women either self-managed their mastitis or consulted only their community midwife and/or health visitor for management advice. In general, incidence rates for mastitis are below 10% when medical records and women seeking medical advice are used as a source of data, whereas incidence rates of around 20% are seen in studies where diagnosis is based on self-reported symptoms [##REF##9699467##1##].</p>", "<p>The recommended management of mastitis is usually conservative, the key recommendation being that mothers continue to breastfeed and to feed more frequently or express milk from the affected breast(s), in an effort to clear blocked ducts and engorgement [##UREF##3##13##, ####UREF##4##14##, ##UREF##5##15####5##15##]. Most women received advice consistent with current recommendations, however one in ten affected women were inappropriately advised by a health professional either to stop breastfeeding from the affected breast or to stop breastfeeding altogether. While in relative terms this may seem small, in absolute terms this represents approximately 680 women across Scotland receiving inappropriate advice annually. This estimation is based on the assumption that 70% of Scottish women delivering in 2004 [##UREF##6##16##] (n = 53957) initiated breastfeeding [##REF##9699468##17##] and that 18% of these women experienced mastitis, of whom 10% received inappropriate advice.</p>", "<p>There appear to be differences between countries in the extent to which antibiotics are prescribed to treat mastitis. In our study just over half (53%) of women who reported mastitis were prescribed antibiotics. This is higher than the 38% reported in a Finnish study [##REF##8092782##4##] but lower than the 75% or more in Australian studies [##REF##9785526##6##,##REF##18394188##18##] and 86% in a recent US study [##REF##11790672##3##]. In a recent Swedish study just under 15% of women with mastitis were prescribed antibiotics [##REF##7755694##19##]. Of these, 3.3% of cases were prescribed antibiotics on the basis of their symptoms and the remaining cases (11.4%) were prescribed antibiotics on the basis of culture results.</p>", "<p>The bacteriological analysis of breast milk is not routinely practiced in the UK with women usually being prescribed antibiotics on the basis of the severity and duration of their symptoms. Potentially pathogenic bacteria are found in the breast milk of healthy breastfeeding women and because the results from bacterial cultures may be difficult to interpret, it has been suggested that the bacteriological examination of breast milk is not particularly informative in the decision to treat mastitis with antibiotics [##REF##7755694##19##]. However, in light of the fact that community acquired methicillin resistant <italic>Staphylococcus aureus </italic>(MRSA) is becoming more common, breast milk culture and sensitivity testing is recommended if the condition does not respond to antibiotic therapy within two days or if the mastitis recurs [##UREF##4##14##,##UREF##5##15##].</p>", "<p>The difference in prescribing rates may be related to the number of women who self-manage their condition or seek advice from a heath professional other than their GP. Scottish women tended to consult their Community Midwife or Health Visitor, some of whom may have organised a prescription for antibiotics, with only just over a third consulting their GP. Whereas in an Australian study the majority of women (73%) had sought treatment and advice from their GP [##REF##9785526##6##] and all of the women in a US study [##REF##11790672##3##] were diagnosed following a medical consultation, thus increasing the likelihood of antibiotics being prescribed.</p>", "<p><italic>Staphylococcus aureus </italic>is the most common organism responsible for mastitis [##UREF##7##20##] and recent Clinical Practice Guidelines [##UREF##5##15##,##REF##17456243##21##] recommend penicillinase-resistant penicillins such as flucloxacillin and dicloxacillin as the drug of first choice, or cephalexin and clindamycin in women who are allergic to penicillin. While the WHO publication on mastitis also recommends amoxicillin and erythromycin [##UREF##4##14##] more recent guidelines advise against the use of these drugs on the basis that a significant proportion of isolates of <italic>Staphylococcus aureus </italic>are resistant to these antibiotics [##UREF##5##15##,##REF##17456243##21##]. Of the women who could recall the antibiotic they were prescribed (20/30) almost half (9/20) were prescribed an antibiotic that was not consistent with current practice guidelines. Kvist et al. recommend that the \"imprudent use of antibiotics be avoided because of the spread of MRSA and other multi-resistant pathogens\" [##REF##7755694##19##]. Both their and our results suggest that a relatively large proportion of women can conservatively manage their mastitis without resorting to taking antibiotics.</p>", "<p>Mastitis has been associated with the premature cessation of breastfeeding [##REF##9699467##1##,##REF##12019051##2##]. However, this was not the case in a recent study of Australian women where no association between mastitis and breastfeeding duration was found [##REF##1957190##10##]. In our study, women who experienced mastitis were significantly more likely to be breastfeeding at 26 weeks than those who did not experience mastitis, which is similar to the finding of a study of New Zealand mothers [##REF##10655826##5##]. Vogel et al. concluded that mastitis is more likely to occur in mothers with ample milk supply, who may be more at risk of milk stasis if they delay or miss a feed [##REF##10655826##5##].</p>", "<p>The strengths of this study are the relatively high response (72%) and the high follow-up rate (95%). In addition, we had frequent and regular contact with women allowing us to pinpoint the timing of onset of mastitis. There are also a number of limitations to this study. Firstly, women identified as cases through the follow-up interviews were only identified if they answered yes to having had mastitis specifically. They were not asked if they had experienced any symptoms suggestive of mastitis. However, the results of our study are strikingly similar to those of Amir et al. who, in order to reduce bias, avoided asking about mastitis directly but collected information about mastitis symptoms [##REF##1957190##10##]. A further limitation of this study is that almost half of participants were continuing to breastfeed at 6 months, compared with the national average of 25% [##REF##9699468##17##], suggesting that our sample was not necessarily representative of all breastfeeding women in Scotland. Despite these limitations, the mastitis incidence rate in this study is reasonably consistent with the incidence rates from studies of women in other Western countries.</p>" ]

[ "<title>Conclusion</title>", "<p>The findings of this study suggest that one in six women may develop lactational mastitis. While most women receive appropriate management advice from health professionals, a clinically significant number of women are advised to stop breastfeeding from the affected breast or to stop breastfeeding altogether. If followed, this advice could lead to women unnecessarily depriving their infants prematurely of the known nutritional and immunological benefits of breast milk.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Lactational mastitis is a painful, debilitating condition that if inappropriately managed, may lead women to discontinue breastfeeding prematurely. The aim of this paper is to report the incidence of mastitis in the first six months postpartum in a Scottish population, its impact on breastfeeding duration and to describe the type and appropriateness of the support and management received by affected women from health professionals.</p>", "<title>Methods</title>", "<p>A longitudinal study of 420 breastfeeding women was undertaken in Glasgow in 2004/05. Participants were recruited and completed a baseline questionnaire before discharge from hospital. Cases of mastitis were reported either directly to the researchers or were detected during regular follow-up telephone interviews at weeks 3, 8, 18 and 26. Women experiencing mastitis provided further information of their symptoms and the management and advice they received from health professionals.</p>", "<title>Results</title>", "<p>In total, 74 women (18%) experienced at least one episode of mastitis. More than one half of initial episodes (53%) occurred within the first four weeks postpartum. One in ten women (6/57) were inappropriately advised to either stop breastfeeding from the affected breast or to discontinue breastfeeding altogether.</p>", "<title>Conclusion</title>", "<p>Approximately one in six women is likely to experience one or more episodes of mastitis whilst breastfeeding. A small but clinically important proportion of women continue to receive inappropriate management advice from health professionals which, if followed, could lead them to unnecessarily deprive their infants prematurely of the known nutritional and immunological benefits of breast milk.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>JAS designed the study, analysed data and drafted the manuscript. MR analysed data. JF and CK contributed to the design of the study. SM collected the data. All authors revised the manuscript and approved the final version.</p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>The authors acknowledge the willing assistance given by the mothers participating in this study. They would like to thank Catherine Fetherston for assistance with the development of the Mastitis Case Questionnaire and the Journal Editor and Reviewers for their helpful comments. This study was supported by a Wellcome Trust Health Services Project grant #070744/03/Z.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Proportion (%) of episodes of mastitis occurring in each time period (n = 57).</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>Duration of breastfeeding, by mastitis (n = 420).</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Characteristics of women (n = 420)</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Characteristics</bold></td><td align=\"center\" colspan=\"2\"><bold>No mastitis (n = 346)</bold></td><td align=\"center\" colspan=\"2\"><bold>Mastitis (n = 70)</bold></td><td align=\"right\"><bold>Chi-square test p value</bold></td></tr><tr><td/><td align=\"right\"><bold>n</bold></td><td align=\"right\"><bold>%</bold>¹</td><td align=\"right\"><bold>N</bold></td><td align=\"right\"><bold>%</bold>¹</td><td/></tr></thead><tbody><tr><td align=\"left\">Age (years)</td><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"> &lt;25</td><td align=\"right\">34</td><td align=\"right\">10</td><td align=\"right\">11</td><td align=\"right\">16</td><td align=\"right\">0.106</td></tr><tr><td align=\"left\"> 25–29</td><td align=\"right\">84</td><td align=\"right\">26</td><td align=\"right\">11</td><td align=\"right\">16</td><td/></tr><tr><td align=\"left\"> 30–34</td><td align=\"right\">118</td><td align=\"right\">36</td><td align=\"right\">22</td><td align=\"right\">31</td><td/></tr><tr><td align=\"left\"> ≥ 35</td><td align=\"right\">89</td><td align=\"right\">27</td><td align=\"right\">26</td><td align=\"right\">37</td><td/></tr><tr><td align=\"left\">Marital status</td><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"> Married or living with partner</td><td align=\"right\">306</td><td align=\"right\">94</td><td align=\"right\">66</td><td align=\"right\">96</td><td align=\"right\">0.623</td></tr><tr><td align=\"left\"> Other</td><td align=\"right\">19</td><td align=\"right\">6</td><td align=\"right\">3</td><td align=\"right\">4</td><td/></tr><tr><td align=\"left\">Deprivation category²</td><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"> 1–2</td><td align=\"right\">69</td><td align=\"right\">20</td><td align=\"right\">24</td><td align=\"right\">32</td><td align=\"right\">0.044</td></tr><tr><td align=\"left\"> 3–5</td><td align=\"right\">164</td><td align=\"right\">47</td><td align=\"right\">33</td><td align=\"right\">45</td><td/></tr><tr><td align=\"left\"> 6–7</td><td align=\"right\">113</td><td align=\"right\">33</td><td align=\"right\">17</td><td align=\"right\">23</td><td/></tr><tr><td align=\"left\">Education</td><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"> Did not complete high school</td><td align=\"right\">49</td><td align=\"right\">15</td><td align=\"right\">12</td><td align=\"right\">17</td><td align=\"right\">0.622</td></tr><tr><td align=\"left\"> Completed high school</td><td align=\"right\">14</td><td align=\"right\">4</td><td align=\"right\">4</td><td align=\"right\">6</td><td/></tr><tr><td align=\"left\"> Professional or technical Diploma/certificate</td><td align=\"right\">145</td><td align=\"right\">45</td><td align=\"right\">34</td><td align=\"right\">49</td><td/></tr><tr><td align=\"left\"> University degree</td><td align=\"right\">115</td><td align=\"right\">36</td><td align=\"right\">19</td><td align=\"right\">27</td><td/></tr><tr><td align=\"left\">Number of children</td><td/><td/><td/><td/><td align=\"right\">0.405</td></tr><tr><td align=\"left\"> 1</td><td align=\"right\">184</td><td align=\"right\">57</td><td align=\"right\">33</td><td align=\"right\">48</td><td/></tr><tr><td align=\"left\"> 2</td><td align=\"right\">96</td><td align=\"right\">29</td><td align=\"right\">24</td><td align=\"right\">35</td><td/></tr><tr><td align=\"left\"> ≥ 3</td><td align=\"right\">45</td><td align=\"right\">45</td><td align=\"right\">12</td><td align=\"right\">17</td><td/></tr><tr><td align=\"left\">Previous history of mastitis</td><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"> First time mother</td><td align=\"right\">184</td><td align=\"right\">57</td><td align=\"right\">33</td><td align=\"right\">48</td><td align=\"right\">0.447</td></tr><tr><td align=\"left\"> Parity &gt; 1, never breastfed before</td><td align=\"right\">13</td><td align=\"right\">4</td><td align=\"right\">3</td><td align=\"right\">4</td><td/></tr><tr><td align=\"left\"> Breastfed before, no mastitis</td><td align=\"right\">95</td><td align=\"right\">29</td><td align=\"right\">22</td><td align=\"right\">32</td><td/></tr><tr><td align=\"left\"> Mastitis in past</td><td align=\"right\">33</td><td align=\"right\">10</td><td align=\"right\">11</td><td align=\"right\">16</td><td/></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Advice received by women for management of mastitis from a health professional (n = 57)</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Advice</bold></td><td align=\"center\"><bold>n</bold></td><td align=\"center\"><bold>%</bold></td></tr></thead><tbody><tr><td align=\"left\">Feed frequently from the affected breast</td><td align=\"center\">31</td><td align=\"center\">54</td></tr><tr><td align=\"left\">Feed frequently from the affected breast first/empty the affected breast</td><td align=\"center\">30</td><td align=\"center\">53</td></tr><tr><td align=\"left\">Express milk between feeds or when infant not interested in feeding</td><td align=\"center\">4</td><td align=\"center\">7</td></tr><tr><td align=\"left\">Do NOT stop breastfeeding</td><td align=\"center\">1</td><td align=\"center\">2</td></tr><tr><td align=\"left\">Stop breastfeeding or discontinue feeding from the affected breast</td><td align=\"center\">6</td><td align=\"center\">10</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional file 1</title><p>Mastitis Case Questionnaire.</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><p>¹Percentages may not total 100 due to rounding</p><p>²Where 1 = least deprived group and 7 = most deprived</p></table-wrap-foot>", "<table-wrap-foot><p>Percentages add to more than 100 as some women may have received more than one form of advice</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1746-4358-3-21-1\"/>", "<graphic xlink:href=\"1746-4358-3-21-2\"/>" ]

[ "<media xlink:href=\"1746-4358-3-21-S1.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>As a \"barometer\" for cardiovascular health status, brachial artery flow-mediated dilation (FMD) provides a bioassay for <italic>in vivo </italic>endothelial function [##UREF##0##1##]. The evidence supporting the dependency of FMD on the endothelium is based on the observation that, after removal of the endothelial lining, arteries lose their ability to dilate in response to an increase in flow [##REF##3080370##2##]. Furthermore, FMD is partially to totally abolished after intra-arterial administration of a NO synthase blocker (i.e. L-NMMA) [##REF##11724650##3##,##REF##11157665##4##]. Nitric oxide is an anti-atherosclerotic molecule, and decreased NO bioavailability is a hallmark of pro-atherogenic states [##REF##12675638##5##]. The biological significance of FMD, the relative simplicity and the non-invasive nature of the technique has motivated the dramatic growth in FMD research over the past 15 years. The widespread adoption of this ultrasound technique has been supported in part by published data suggesting that FMD correlates with invasive measurements of endothelial function in the coronary arteries [##UREF##1##6##,##REF##9874063##7##] and may predict future cardiovascular events [##REF##14530195##8##,##UREF##2##9##]. However, data are also available indicating that reduced FMD has weak or no association with other cardiovascular risk factors [##REF##15451155##10##,##REF##15963396##11##], is a questionable marker of the presence and severity of coronary artery disease [##REF##12323158##12##], fails to show a prognostic value [##REF##17045673##13##], and has poor diagnostic accuracy for identifying older adults with subclinical cardiovascular disease [##REF##17714717##14##]. These contradictory results may be due to substantial discrepancies in FMD protocols across labs, but also due to the inter-subject variability in hyperemic shear stress (the stimulus for FMD). Large variability in reactive hyperemia has been documented among individuals and populations [##REF##16051630##15##]. The unadjusted FMD outcome may reflect conduit artery endothelial function as well as the magnitude of the hyperemic stimulus. Current work is moving to adjust the measured vasodilation response for the applied stimulus; in other words, dividing the peak FMD by the magnitude of stimulus achieved with reactive hyperemia. In 2005, Pyke and Tschakovsky [##REF##16051630##15##], the initiators of this approach, suggested the utilization of shear rate (or shear stress (SS)) area under the curve (SS_AUC) (individual area until peak FMD) as a method to quantify the entire and \"relevant\" hyperemic stimulus; and thus to be used for FMD normalization purposes. Two years later [##REF##17170205##16##], the same authors provided experimental evidence of the physiological appropriateness of using this method. They concluded that the SS_AUC, but not the peak shear, was the critical determinant of the peak FMD response. Following publication of this paper, a radical change in clinical practice within the FMD community was expected; however, to our surprise, since the time of Pyke and Tschakovsky's publication (April 2007), only two [##REF##18086954##17##,##REF##17556495##18##] out of 92 FMD studies (PubMed search) have incorporated normalization of FMD to SS_AUC. Reasons for this slow adoption of the recently proposed method to express FMD data may include 1) insufficient lead time after publication of the article for dissemination and implementation; 2) inability to simultaneously capture arterial diameter and velocity; 3) lack of clinical evidence supporting the utility of this approach; and 4) complicated comparison with the large body of evidence collected using the traditional FMD approach.</p>", "<p>A \"proof of concept\" study was conducted herein to further evaluate the efficacy of FMD normalization. In a group of healthy individuals, we elicited five different magnitudes of reactive hyperemia-induced shear stress to <italic>mimic </italic>real-life inter-population differences. Considering this analogy, we created the following experimental scenario: comparison of five hypothetical \"populations\" with different magnitudes of reactive hyperemia-induced shear stress stimuli. Because the five \"populations\" do in fact have identical endothelial function, the correct outcome should be no detectable differences. We hypothesized that peak FMD:SS_AUCratio (normalization approach), but not peak FMD (traditional approach), would be the same among the five \"populations\" with identical endothelial function. This observation would corroborate that normalization of FMD to SS_AUCeliminates the influences of variable shear stress found among populations, and solidify the utility of FMD:SS_AUCratio as an index of endothelial function.</p>" ]

[ "<title>Methods</title>", "<title>Subjects</title>", "<p>Twenty healthy, physically active young adults (10 men, 10 women) volunteered for this study. All subjects were free of recognized cardiovascular, pulmonary and metabolic diseases, non-hypertensive (resting blood pressure &lt; 140/80 mm Hg), non-obese (body mass index &lt; 30 kg/m²), non-smokers, and had no family history of heart diseases. No subjects were taking medications with vaso-active effects, including contraceptives. All procedures were approved by Indiana University Committee for the Protection of Human Subjects. Written informed consent was obtained from each subject prior to participation in the study.</p>", "<title>Study procedures</title>", "<p>Procedures of the study consisted of a screening session and a vascular testing session. The screening session included completion of a medical history/health habits questionnaire, measurements of height, weight, resting blood pressure, and a fasting venous blood draw to obtain total cholesterol, low density lipoprotein cholesterol, high density lipoprotein cholesterol, triglycerides, blood glucose, and high sensitivity C-reactive protein. In addition, all subjects were familiarized with the equipment/testing room to be used during the vascular session.</p>", "<p>On the vascular session day, subjects were instructed to report to the laboratory between 6:00 and 10:00 am having 1) fasted for 12 h, 2) abstained from caffeine, tobacco products, and vitamin supplements for 12 h, and 3) abstained from exercise for 12 h [##UREF##0##1##]. Women were studied during days 1–7 of their menstrual cycle to minimize the influence of cyclical changes in female hormones. Subjects were instructed to lie supine in a dark, climate-controlled quiet room (22–24°C), with their right arm extended out laterally. A venous blood draw was performed from the antecubital vein. Samples were collected into 6-mL heparin vacutainer tubes for determination of whole blood viscosity and hematocrit. Each subject underwent an acclimation phase (20 min) to obtain a hemodynamic steady state. Heart rate was continuously monitored using a three-lead ECG. Blood pressures were taken in the left arm with a mercury sphygmomanometer to confirm a steady state. A 5 × 84 cm automatic cuff (E-20 rapid cuff inflator, D.E. Hokanson, Bellevue, WA) was placed around the forearm immediately distal to the olecranon process following established guidelines for assessing FMD [##UREF##0##1##]. To quantify the magnitude of occlusion-induced ischemia (volume and peak), a dual-wavelength near-infrared spectrometry (NIRS) (ISS, Champaign, IL) probe was positioned over the right extensor digitorum muscle (medial aspect, distally from the cuff). The probe was held in place with an elastic bandage. The forearm oxygen tissue saturation (%StO₂) was recorded throughout the study. The ultrasound image of the brachial artery was obtained longitudinally 2–10 cm above the antecubital fossa by 2D high resolution (Terason T3000, Teratech Corporation, Burlington, MA) ultrasound system, using a 5 to 12-MHz multifrequency linear-array transducer. Once a satisfactory image was obtained, the right arm was secured using a custom-designed arm immobilizer and the transducer was stabilized using a clamp (Figure ##FIG##0##1##). Minor corrections of transducer placement were made to maintain optimal imaging. Doppler velocity was also measured via ultrasound. Doppler flow signals were corrected at an insonation angle of 60° and measurements were performed with the sample volume placed in mid-artery. Ultrasound parameters were not changed during the study. Simultaneous Doppler measurements for blood velocity and 2D ultrasound imaging for diameter were continuously recorded for 30 sec at baseline. The automatic forearm cuff was then inflated to 250 mm Hg and maintained for 1, 2, 3, 4 or 5 min; in a randomized order. Diameter and velocity recordings resumed before cuff deflation and continued for 2 min thereafter. Ultrasound images were recorded at 5 frames/second using Camtasia (TechSmith, Okemos, MI) and converted into an AVI file. R-wave gated frames were not captured exclusively because internal data in our laboratory shows that continuous assessment of diameter at 5 frames/second yields the same FMD results (n = 10; FMD = 7.58 ± 0.9 vs. 7.62 ± 0.9%, p = 0.655; Intraclass correlation coefficient = 0.998, p &lt; 0.0001). The occlusion conditions were applied 10 min apart from each other and baseline measurements were re-established prior to each condition.</p>", "<title>Data analysis</title>", "<title>Brachial Artery Diameter and Blood Velocity</title>", "<p>Off-line analyses of diameters and velocities were performed using an automated edge-detection Brachial Analyzer software (Medical Imaging Applications, LLC, Coralville, IA); briefly, the software allows the user to identify a region of interest (ROI) on the portion of the image where the vessel walls are most clear. The arterial wall borders were detected by an optimal graph search-based segmentation that uses a combination of pixel density and image gradient as an objective function. Each sequence of images was reviewed by the technician and interactively edited when needed to ensure that diameter measurements were always calculated from the intima-lumen interface at the distal and proximal vessel wall. Similarly, for determination of blood velocity, the ROI was selected around the Doppler waveform and the trace of the velocity-time integral was used to calculate mean velocity for each cardiac cycle (Figure ##FIG##1##2##). All measurements were performed by a single technician who was blinded to the trial condition for each image sequence. Reproducibility of our measurements has been reported previously [##REF##16669406##19##]. The time course of diameters and velocities were determined using a 3-sec moving average. The peak dilation postocclusion was determined as the highest 3-sec average and was presented as a percent change from baseline diameter (peak FMD; %). Brachial artery blood flow was calculated using the following formula: Vm • <italic>π </italic>• (D²/4) 60, where Vm is mean blood velocity (cm • s^-1), <italic>π </italic>is 3.14, and D is mean arterial diameter (cm) [##REF##16118583##20##].</p>", "<title>Whole blood viscosity</title>", "<p>With the use of a pipette, 1 mL of blood sample was removed from the vacutainer tube, the mass of the sample was determined, and the density calculated. An additional blood sample was removed and transferred to a glass capillary viscometer [##UREF##3##21##] (Cannon-Manning Semi-Micro Viscometer, Cannon Instruments, Philadelphia, PA) and placed in a constant-temperature water bath (37°C). The viscosity of the sample was determined by measuring the time required for the sample to pass between two fixed points on the viscometer. Kinematic viscosity (mm²• s^-1) was obtained by multiplying a viscometer constant (0.007927 mm²• s^-2) by the efflux time in seconds. To obtain viscosity in mPa • s, kinematic viscosity was multiplied by the density in grams per milliliter. Measures were performed in duplicate.</p>", "<title>Brachial artery shear stress</title>", "<p>Brachial artery shear stress (dynes • cm^-2) was calculated using the following formula: (4ηVm) • D^-1, where η is blood viscosity (mPa • s), Vm is mean blood velocity (cm • s^-1), and D is mean arterial diameter (cm) [##REF##16046216##22##]. To describe the magnitude of reactive hyperemia-induced shear stress elicited with increased duration of occlusion, shear stress AUC was calculated for each occlusion condition. Briefly, the AUC was calculated by summing the areas of successive postocclusion trapezoids (each with a base of 3-sec) for 60 sec (SS60sec_AUC; a.u.). To quantify the \"relevant\" hyperemic stimulus responsible for the peak FMD response, the shear stress AUC above baseline was individually calculated for the duration of time-to-peak dilation (SS_AUC; a.u.). Normalization of FMD to shear stress was expressed as the peak FMD:SS_AUCratio (a.u).</p>", "<title>Hematocrit</title>", "<p>A 50–75 μL blood sample was removed from the vacutainer tube, transferred into a capillary tube and centrifuged for 5 min using a micro-hematocrit centrifuge (Clay-Adams, New York, NY). Hematocrit (%) was read using a micro-capillary reader (International Equipment Company, Needham Heights, MA). Measures were performed in duplicate.</p>", "<title>Magnitude of ischemia</title>", "<p>Forearm oxygen tissue saturation (StO₂; %) was monitored at baseline (30 sec), throughout each occlusion period, and during the reperfusion phase. The time course of StO₂was determined using a 3-sec moving average. To quantify the volume of ischemia, the StO₂AUC (area below baseline) was calculated by summing the areas of successive trapezoids (each with a base of 3-sec) for the total duration of the occlusion period (StO_2AUC; a.u.). Peak ischemia (StO_2peak; %) was considered as the StO₂change from baseline to immediately before cuff release.</p>", "<title>Statistical analysis</title>", "<p>Descriptive statistics were used to summarize the subject demographic data. One-way repeated measures ANOVA (5 levels) were performed to determine the effect of forearm occlusion duration on volume of ischemia (StO_2AUC; a.u.), peak ischemia (StO_2peak; a.u.), magnitude of reactive hyperemia-induced shear stress (SS60sec_AUC; a.u.), peak FMD response (%), and FMD normalized to shear stress (FMD:SS_AUCratio; a.u.). Tukey's HSD procedure was used when a significant F-ratio was found. All data are presented as mean ± standard error of the mean (SEM). For all statistical tests, the alpha level was set at 0.05. Statistical analyses were performed with SPSS v.15.0. (SPSS, Inc. Chicago, IL, USA).</p>" ]

[ "<title>Results</title>", "<p>Demographic information of the subjects is summarized in Table ##TAB##0##1##. No effect of sex was found in any of the main outcome variables (SS60sec_AUC, peak FMD response, FMD:SS_AUCratio), thus data were pooled across sex. Table ##TAB##1##2## displays mean baseline values for brachial artery diameter, blood velocity, shear stress, heart rate and forearm oxygen tissue saturation. No differences in baseline values were found (p &gt; 0.05) for any of the variables. NIRS data on one subject was lost, thus sample size was 19 for all NIRS-related analysis. Figure ##FIG##2##3## illustrates the time course of forearm oxygen tissue saturation in conjunction with brachial artery blood flow across the different forearm occlusion conditions. As indicated, the volume of ischemia and peak ischemia increased incrementally with duration of forearm occlusion (F_(4,72)= 102.6; p &lt; 0.0001 and F_(4,72)= 69.85; p &lt; 0.0001, respectively). Figure ##FIG##3##4## illustrates the effects of forearm occlusion duration on the magnitude of reactive hyperemia-induced shear stress (panel A), peak FMD response (panel B), and FMD normalized to shear stress (panel C). As shown, varying forearm occlusion duration effectively elicited five different magnitudes of reactive hyperemia-induced shear stress (F_(4,76)= 97.6; p &lt; 0.0001). As a result, differences in peak FMD response were detected among conditions (F_(4,76)= 40.5; p &lt; 0.0001); however, these discrepancies were abolished when normalizing FMD to the shear stress (peak FMD:SS_AUCratio) (F_(4,76)= 0.43; p = 0.785).</p>" ]

[ "<title>Discussion</title>", "<p>Given that reactive hyperemia varies among individuals and populations [##REF##16051630##15##], the FMD outcome is reflective of both conduit artery endothelial function and magnitude of the hyperemic stimulus. Normalization of FMD to SS_AUChas recently been proposed [##REF##17170205##16##] to control for the presence of the large inter-subject variability in reactive hyperemia-induced shear stress and solely reflect conduit artery endothelial function. The present study was designed to further examine the efficacy of FMD normalization. In a group of healthy individuals, via manipulation of duration of cuff occlusion, we evoked five different magnitudes of reactive hyperemia-induced shear stress (Figure ##FIG##3##4A##) to create an idealized experimental scenario: comparison of five hypothetical \"populations\" with known identical endothelial function but different magnitude of reactive hyperemia-induced shear stress stimuli. Our findings demonstrate that, when presenting the data using the traditional approach (peak FMD), differences in FMD are detected across the varying shear stimuli (Figure ##FIG##3##4B##), which would suggest the erroneous conclusion that discrepancies in endothelial function exist among these populations. However, when presenting the data using the normalization approach (FMD:SS_AUCratio), these differences are abolished (Figure ##FIG##3##4C##); thus confirming that normalization of FMD to SS_AUCeliminates the influences of variable shear stress on measured conduit artery endothelial function.</p>", "<p>Pyke and Tshakovsky [##REF##17170205##16##] were the first to examine the effect of normalizing FMD to SS_AUC. The purpose of their study was to investigate the independent contributions of the peak and continued reactive hyperemia-induced shear stress on FMD. This was elegantly conducted by maintaining the peak shear stimulus constant while manipulating the shear stimulus duration (re-inflation of the cuff) or by maintaining the shear stimulus constant while manipulating the magnitude of the peak shear stimulus (application of arterial pressure). From these experiments, the authors concluded that the SS_AUC, but not the peak shear, was the critical determinant of the peak FMD response; and thus to be used for normalization purposes. Because their findings were convincing and physiologically sound, we did not consider further exploration of the relative contribution of peak vs continued shear. Instead, we opted to manipulate the duration of cuff occlusion because this is an effective method to globally modify the hyperemic stimulus (both peak and duration); thus closely mimicking real-life between-subject differences in shear.</p>", "<p>The only two ultrasound studies manipulating the duration of cuff occlusion as strategy to create a range in hyperemic stimuli and assess the subsequent vasodilation response at the conduit artery (radial and brachial) were published in 1997 [##REF##9247757##23##,##REF##9290397##24##]. Both groups concluded that the duration of hyperemic stimulus played an important role on the vasodilatory response, observations that prompted Pyke and Tshakovsky's study [##REF##17170205##16##] 10 years later. Our data are in agreement with Leeson et al. [##REF##9290397##24##] and Joannides et al. [##REF##9247757##23##] in that longer duration of cuff occlusion was associated with greater hyperemic and vasodilation responses; however, these groups did not test whether normalization of FMD to SS_AUCremoved differences among trials; this is not surprising as the concept of FMD normalization was not introduced at that time. A limitation to the studies of Leeson et al. and Joannides et al. is the utilization of blood flow, instead of shear stress, as representation of the vasodilatory stimulus. Unfortunately, blood flow (velocity • <italic>π </italic>• (diameter²/4) • 60) and shear stress (viscosity • 4 • velocity/diameter) are parallel only in conditions where arterial diameters are constant, a situation that does not occur during the hyperemic phase where arterial diameters are significantly and dynamically altered. Furthermore, in contrast to Leeson et al. and Joannides et al., our study incorporated the NIRS technology to quantify the magnitude (volume and peak) of forearm occlusion-induced ischemia. Given that the forearm ischemia is the driving force for reactive hyperemia, characterization of the principal stimulus for FMD should not be neglected.</p>", "<p>There are a few limitations to this study. First, because all trials were performed on the same day, it is possible that repeated episodes of ischemia-induced hyperemia influenced subsequent measures; however evidence suggests that serial FMD measurements do not affect subsequent FMD outcomes [##REF##16669406##19##]. We chose to include all trials on one day in order to avoid day-to-day endothelial function variability [##REF##10483941##25##] and to allow for consistent placement of the ultrasound and NIRS probes across trials, which minimizes these recognized sources of variability in these measurements. Sufficient time was allowed between trials for diameters, velocities, and tissue oxygen saturation to reach baseline values. Most importantly, the order of trials was randomized to minimize any confounding influence from this potential carry-over effect. Second, although the dilation following the conventional 5-min forearm occlusion is largely attributed to NO[##REF##11724650##3##,##REF##11157665##4##], it remains unknown if dilation following shorter periods of occlusion (1, 2, 3, and 4 min) is NO dependent as well. In addition, the effects of reduced shear and/or metabolites released during the ischemic period on the FMD response are not well understood, thus manipulation of occlusion duration may be considered a limitation to the study. Third, our measurements of whole blood viscosity were performed using a glass capillary viscometer, thus we were unable to assess viscosity at varying shear rates. Although we believed that measurements of viscosity using this method would provide more accurate shear stress information than using a constant (assumed) viscosity of 4 mPa • s for all subjects [##UREF##4##26##], there were no differences between both approaches (Table ##TAB##2##3##). This observation suggests that the addition of viscosity measurements does not have a significant impact on shear stress calculations and it does not alter the interpretation of the FMD results. The validity of our viscosity measures may be demonstrated by the small coefficient of variation (4.47 ± 0.65%) and high correlation between hematocrit and viscosity (r = 0.878, p &lt; 0.001). Fourth, because our sample was composed of 20 healthy asymptomatic men and women, extrapolation of these findings across clinical populations should be made with caution.</p>" ]

[ "<title>Conclusion</title>", "<p>The significance of the present findings is notable. Our results support the view of Pyke and Tschakovsky [##REF##17170205##16##], the original advocates of FMD normalization to SS_AUCapproach. The obvious lack of compliance to the recent guidelines in the clinical literature is intriguing and perhaps problematic. From a theoretical and physiological standpoint, there is robust evidence that FMD should be corrected by its dilator stimulus; however, the following question remains to be resolved: Does FMD:SS_AUCratio predict cardiovascular risk and events more accurately than peak FMD (the traditional FMD approach)? Further research is warranted to confirm that the apparent physiological appropriateness of normalization is in fact clinically pertinent. Given that arterial diameter and velocity measurements are typically performed continuously and simultaneously during the hyperemic phase, we encourage investigators to, at the minimum, report FMD data using both the traditional (peak FMD response) and normalization (peak FMD:SS_AUCratio) approaches. In supporting this new approach, while we do not intend to invalidate the large body of evidence collected using the traditional FMD approach, we advise the reader to interpret the existing literature with caution. Special attention should be devoted to studies that did not report any form of hyperemic stimulus.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Normalization of brachial artery flow-mediated dilation (FMD) to individual shear stress area under the curve (peak FMD:SS_AUCratio) has recently been proposed as an approach to control for the large inter-subject variability in reactive hyperemia-induced shear stress; however, the adoption of this approach among researchers has been slow. The present study was designed to further examine the efficacy of FMD normalization to shear stress in reducing measurement variability.</p>", "<title>Methods</title>", "<p>Five different magnitudes of reactive hyperemia-induced shear stress were applied to 20 healthy, physically active young adults (25.3 ± 0. 6 yrs; 10 men, 10 women) by manipulating forearm cuff occlusion duration: 1, 2, 3, 4, and 5 min, in a randomized order. A venous blood draw was performed for determination of baseline whole blood viscosity and hematocrit. The magnitude of occlusion-induced forearm ischemia was quantified by dual-wavelength near-infrared spectrometry (NIRS). Brachial artery diameters and velocities were obtained via high-resolution ultrasound. The SS_AUCwas individually calculated for the duration of time-to-peak dilation.</p>", "<title>Results</title>", "<p>One-way repeated measures ANOVA demonstrated distinct magnitudes of occlusion-induced ischemia (volume and peak), hyperemic shear stress, and peak FMD responses (all p &lt; 0.0001) across forearm occlusion durations. Differences in peak FMD were abolished when normalizing FMD to SS_AUC(p = 0.785).</p>", "<title>Conclusion</title>", "<p>Our data confirm that normalization of FMD to SS_AUCeliminates the influences of variable shear stress and solidifies the utility of FMD:SS_AUCratio as an index of endothelial function.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>JP designed the study, recruited the subjects, collected the data, analyzed the data, and wrote the original manuscript draft. BDJ and DPW assisted in data collection and edited the manuscript draft. SCN, TDM, ADF and KJM assisted in study design and edited the manuscript draft. JPW assisted in study design, supervised the study, and edited the manuscript draft. All authors read and approved the manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>The authors thank all the subjects for their time, effort, and willingness to participate in the study. This research was supported in part by the IU Graduate Student Research Grant-in-Aid, the IU HPER Research Grant-in-Aid, and the IU AAU/Bell-Updyke-Willett Kinesiology Research Fund. JP is sponsored by a fellowship from the Ministerio de Educación y Cultura de España.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Experimental set-up.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>Analysis of brachial diameter and velocity (using the edge-detection Brachial Analyzer software) of a representative subject immediately following cuff deflation.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Time course of forearm oxygen tissue saturation (NIRS) in conjunction with brachial artery blood flow (ultrasound)</bold>. Values are means. *Volume of ischemia (StO_2AUC) significantly different from all other trials; ^#Volume of ischemia (StO_2AUC) significantly different from 3-min, 4-min, and 5-min; ^&amp;Peak ischemia (StO_2peak) significantly different from all other trials; ^{^}Peak ischemia (StO_2peak) significantly different from 1-min, 2-min, and 3-min. All p &lt; 0.0001. Ultrasound data were not collected during the occlusion periods.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p><bold>Reactive hyperemia-induced shear stress (panel A), peak FMD response (panel B), and FMD normalized to shear stress (panel C) for the five occlusion conditions</bold>. Values are means ± SEM. <italic>Panel A: </italic>*Magnitude of reactive hyperemia-induced shear stress (SS60sec_AUC) significantly different among all trials. <italic>Panel B: </italic>*Peak FMD significantly different from all other trials; ^#Peak FMD significantly different from 1-min, 2-min, and 5-min; ^&amp;Peak FMD significantly different from 1-min and 2-min; ^{^}Peak FMD significantly different from 1-min, 2-min, and 3-min. All p &lt; 0.0001.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Demographic characteristics of the subjects.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Variable</td><td align=\"center\">Value</td></tr></thead><tbody><tr><td align=\"left\">N</td><td align=\"center\">20</td></tr><tr><td align=\"left\">Sex, M/W</td><td align=\"center\">10/10</td></tr><tr><td align=\"left\">Age, years</td><td align=\"center\">25.3 ± 0. 6</td></tr><tr><td align=\"left\">Height, cm</td><td align=\"center\">175.0 ± 2.1</td></tr><tr><td align=\"left\">Weight, kg</td><td align=\"center\">68.6 ± 2.5</td></tr><tr><td align=\"left\">Body mass index, Kg/m²</td><td align=\"center\">22.3 ± 0.4</td></tr><tr><td align=\"left\">Resting systolic blood pressure, mmHg</td><td align=\"center\">112.0 ± 2.6</td></tr><tr><td align=\"left\">Resting diastolic blood pressure, mmHg</td><td align=\"center\">71.4 ± 2.2</td></tr><tr><td align=\"left\">Total cholesterol, mg/dL</td><td align=\"center\">159.3 ± 7.2</td></tr><tr><td align=\"left\">High density lipoprotein cholesterol, mg/dL</td><td align=\"center\">57.2 ± 2.4</td></tr><tr><td align=\"left\">Low density lipoprotein cholesterol, mg/dL</td><td align=\"center\">89.1 ± 6.2</td></tr><tr><td align=\"left\">Triglycerides, mg/dL</td><td align=\"center\">64.9 ± 5.4</td></tr><tr><td align=\"left\">Fasting glucose, mg/dL</td><td align=\"center\">90.3 ± 1.5</td></tr><tr><td align=\"left\">High sensitivity C-reactive protein, mg/L</td><td align=\"center\">0.50 ± 0.1</td></tr><tr><td align=\"left\">Whole blood viscosity, mPa·s</td><td align=\"center\">3.74 ± 0.1</td></tr><tr><td align=\"left\">Hematocrit, %</td><td align=\"center\">42.9 ± 0.8</td></tr><tr><td align=\"left\">Reported physical activity¹, days/week</td><td align=\"center\">4.3 ± 0.2</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Baseline hemodynamic information among forearm occlusion durations.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Variable</td><td align=\"center\">1-min</td><td align=\"center\">2-min</td><td align=\"center\">3-min</td><td align=\"center\">4-min</td><td align=\"center\">5-min</td><td align=\"center\"><italic>p </italic>value</td></tr></thead><tbody><tr><td align=\"left\">Arterial diameter, cm</td><td align=\"center\">0.352 ± 0.01</td><td align=\"center\">0.349 ± 0.01</td><td align=\"center\">0.349 ± 0.01</td><td align=\"center\">0.350 ± 0.01</td><td align=\"center\">0.350 ± 0.01</td><td align=\"center\">0.308</td></tr><tr><td align=\"left\">Blood velocity, cm • s^-1</td><td align=\"center\">8.48 ± 1.9</td><td align=\"center\">8.99 ± 2.0</td><td align=\"center\">8.18 ± 1.8</td><td align=\"center\">8.81 ± 2.0</td><td align=\"center\">9.00 ± 2.0</td><td align=\"center\">0.67</td></tr><tr><td align=\"left\">Shear stress, dynes • cm^-2</td><td align=\"center\">367.0 ± 32</td><td align=\"center\">392.4 ± 42</td><td align=\"center\">359.9 ± 32</td><td align=\"center\">385.1 ± 42</td><td align=\"center\">393.7 ± 40</td><td align=\"center\">0.648</td></tr><tr><td align=\"left\">Heart rate, bpm</td><td align=\"center\">51.5 ± 2</td><td align=\"center\">51.1 ± 2</td><td align=\"center\">52.5 ± 2</td><td align=\"center\">52.6 ± 2</td><td align=\"center\">52.4 ± 2</td><td align=\"center\">0.412</td></tr><tr><td align=\"left\">Forearm oxygen tissue saturation, %</td><td align=\"center\">56.87 ± 2.4</td><td align=\"center\">58.16 ± 2.3</td><td align=\"center\">58.46 ± 2.5</td><td align=\"center\">57.98 ± 2.5</td><td align=\"center\">58.41 ± 2.5</td><td align=\"center\">0.974</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Shear stress area under the curve calculated using measured viscosity vs. assumed viscosity.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"2\">Shear stress area under the curve (a.u.)</td><td/><td/></tr><tr><td/><td colspan=\"2\"><hr/></td><td/><td/></tr><tr><td/><td align=\"center\">Measured viscosity used</td><td align=\"center\">Assumed viscosity (4 mPa·s) used</td><td align=\"center\">t-test p value</td><td align=\"center\">ICC coefficient (p value)</td></tr></thead><tbody><tr><td align=\"center\">1-min</td><td align=\"center\">11.7 ± 1.3</td><td align=\"center\">12.7 ± 1.5</td><td align=\"center\">0.610</td><td align=\"center\">0.985 (&lt; 0.0001)</td></tr><tr><td align=\"center\">2-min</td><td align=\"center\">20.0 ± 1.4</td><td align=\"center\">21.7 ± 1.7</td><td align=\"center\">0.457</td><td align=\"center\">0.950 (&lt; 0.0001)</td></tr><tr><td align=\"center\">3-min</td><td align=\"center\">26.3 ± 1.6</td><td align=\"center\">28.7 ± 2.2</td><td align=\"center\">0.386</td><td align=\"center\">0.941 (&lt; 0.0001)</td></tr><tr><td align=\"center\">4-min</td><td align=\"center\">35.4 ± 2.1</td><td align=\"center\">38.3 ± 2.5</td><td align=\"center\">0.377</td><td align=\"center\">0.930 (&lt; 0.0001)</td></tr><tr><td align=\"center\">5-min</td><td align=\"center\">41.5 ± 2.2</td><td align=\"center\">45.0 ± 2.8</td><td align=\"center\">0.340</td><td align=\"center\">0.919 (&lt; 0.0001)</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>Values are mean ± SEM.</p><p>¹Aerobic physical activity of moderate intensity for 30 min or more.</p></table-wrap-foot>", "<table-wrap-foot><p>Values are means ± SEM. P value indicates the one-way ANOVA comparing the five baseline periods.</p></table-wrap-foot>", "<table-wrap-foot><p>Values are means ± SEM.</p><p>ICC = Intraclass correlation</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1476-7120-6-44-1\"/>", "<graphic xlink:href=\"1476-7120-6-44-2\"/>", "<graphic xlink:href=\"1476-7120-6-44-3\"/>", "<graphic xlink:href=\"1476-7120-6-44-4\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Palliative care is expected to be holistic and multidisciplinary; it is provided to both the patient and their family [##UREF##0##1##]. Effective communication between the patient, the family and health care providers is integral to optimal palliative care. One method of facilitating communication is a family meeting, also referred to as a family conference [##REF##15662189##2##]. Family meetings between the patient, their family and health care professionals are undertaken for multiple purposes including the sharing of information and concerns, clarifying the goals of care, discussing diagnosis, treatment, prognosis and developing a plan of care for the patient and family carers [##REF##15662189##2##, ####REF##16128661##3##, ##REF##9654615##4####9654615##4##].</p>", "<p>However despite the promotion of family meetings as an essential tool for information sharing and goal clarification in specialist palliative care settings, it has been reported that sparse evidence exists to demonstrate the process for training staff to conduct or participate in them [##REF##16128661##3##]. It has also been claimed that there is a dearth of published literature describing when such meetings should be initiated, who should attend them, how they should be conducted and evaluated [##REF##15381505##5##].</p>", "<p>We set out to develop practice guidelines for conducting family meetings within the context of the specialist palliative care setting, based on the best available evidence and complemented by consensus based expert opinion. For the purposes of these guidelines we define a specialist palliative care setting as a health care environment or service that specifically focuses on the care of patients with an advanced incurable disease and their family. An evaluation of the effectiveness of these guidelines in clinical practice will be reported separately.</p>" ]

[ "<title>Methods</title>", "<p>Key questions guiding the development of the guidelines were: (a) How should family meetings be convened and structured? (b) What content is essential? and (c) Who should attend and lead them?</p>", "<p>Development of the guidelines was based on the following three methods: (1) A literature search using MEDLINE (1995 – 2007), CINAHL (1995 – 2007), and PsycINFO (1995–2007) databases. The principal search terms used either singly or in combination, were: family, carers, case, conference, meeting, hospice and palliative care. Key palliative care textbooks (eg [##UREF##1##6##]) and their reference sections were also hand searched for relevant information; (2) The research team considered relevant models and theories to support the conceptual framework to underpin the guidelines; (3) (a) The research team prepared a draft version of the guidelines which was then reviewed by an expert panel alongside the conceptual framework. Members of the panel were purposively chosen (based on their experience in conducting family meetings) and invited by mail to participate; (b) The guidelines were further refined and their clinical applicability was assessed via two focus groups. The first (focus group 1) comprised multidisciplinary specialists from three palliative care units and the other (focus group 2) comprised multidisciplinary palliative care specialists from three large Melbourne hospitals. The inclusion criteria required that participants had experience in participating in family meetings. Letters of invitation were sent to the managers of each of the services requesting that potential participants contact the research team. Ethical approval was provided by St Vincent's Hospital Melbourne, Australia and written consent was obtained from all participants. The focus group questions centred on asking participants to briefly describe: current practice in relation to family meetings in their setting; key elements of family meetings and then to provide comment on the draft version of the clinical guidelines.</p>", "<p>Data from the expert panel and focus groups were audiotaped and transcribed verbatim. These data were then content analysed in accordance with the structured questions set by the research team.</p>" ]

[ "<title>Results</title>", "<title>Literature Review</title>", "<p>The literature search revealed only three published articles in peer reviewed journals specifically related to family meetings in the specialist palliative care setting [##REF##15662189##2##,##REF##16128661##3##,##UREF##2##7##]. Due to the limited amount of palliative care specific literature; evidence from non-specialist palliative care settings, including intensive care and aged care was also reviewed and included for the purposes of the literature review.</p>", "<p>One palliative care related article acknowledged that most health professionals do not receive sufficient training to conduct family meetings [##REF##16128661##3##]. The authors reported on the evaluation of a training program focused on preparing medical and social work students for conducting family meetings in palliative care. However, the study was primarily about the outcomes of the educational initiative rather than specific details about how to convene and conduct family meetings or whether or not family meetings were beneficial [##REF##16128661##3##]. One article was prepared as an educative tool for health professionals and provided recommendations for conducting family meetings in palliative care, seemingly based on expert opinion in the most part [##UREF##2##7##]. The final article was an informational tool about family meetings designed for patients and families [##REF##15662189##2##], although the process for determining the content was not outlined. Recommendations for conducting family meetings were identified in some palliative care books, but these were also based on expert opinion (for example [##UREF##3##8##]).</p>", "<p>The opinion based family meeting guidelines for health professionals advocated suitable planning and an overt purpose in order to conduct an effective meeting. Other key recommendations included giving consideration to who should attend, a meeting place that ensures privacy and a method for disseminating outcomes of the meeting.</p>", "<p>Similar recommendations on how to conduct a family meeting exist in the aged care setting [##UREF##4##9##]. While these suggested approaches appear useful (as was the case for palliative care based suggestions), the basis for their content is not explicitly outlined and it is unclear as to whether their utility has been tested. One qualitative study was identified which aimed to explore the opinions of participants in family meetings in a geriatric rehabilitation hospital. Staff, patients and families participated in focus groups, completed surveys and individual semi-structured interviews [##REF##15381505##5##]. Although high levels of satisfaction with family meetings were reported, patient interviews revealed a lack of informed consent and lack of clarity regarding the purpose of meetings. An unclear agenda was identified by staff, patients and families as the underlying reason for unsatisfactory meetings.</p>", "<p>A practice model for nurses working with families has been developed [##UREF##5##10##]. It includes family interviews conducted over approximately an hour, as well as 15 minute interviews designed to be conducted within the course of normal nursing duties. The key elements of a family interview identified by these authors included: therapeutic conversations; manners; family genograms and ecomaps; therapeutic questions and commending family and individual strengths. However, there does not appear to be research evidence supporting the effectiveness of this model in practice.</p>", "<p>The conduct of family meetings has also been described in the context of training medical practitioners [##UREF##6##11##]. Two related qualitative studies investigated the attitudes of residents (junior doctors) and their educators towards family conferences. A survey of 65 residents revealed that family meetings are: valued as a communication tool to provide medical information and reach consensus on care; occur primarily in inpatient settings in the context of crisis; fail to occur due to significant barriers; and are best taught experientially with the involvement of physician role models [##UREF##6##11##]. The accompanying survey of 12 teaching staff revealed major themes including: family conferences occur on a continuum of formality; experiential, in vivo training for family conferences is the primary training modality; specific skills are needed to conduct family conferences; and the barriers to conducting family conferences are formidable [##UREF##6##11##].</p>", "<p>Our review identified several studies that explored family meetings within the context of the intensive care unit (ICU) [##REF##17267907##12##, ####REF##15827431##13##, ##REF##16625131##14##, ##UREF##7##15##, ##REF##15241092##16##, ##REF##17079999##17##, ##REF##17057600##18##, ##REF##17353493##19##, ##REF##11042236##20####11042236##20##]. Even though the ICU setting is considerably different from a typical palliative care unit, given the lack of research related to family meetings in the specialist palliative care setting we deemed it pertinent to highlight these studies. Furthermore, there are some similarities between the two care settings: a considerable proportion of the patients die in both ICU and palliative care units and patients are often too unwell to be involved in the family meeting.</p>", "<p>A study involving qualitative analysis of 51 family conferences across four ICUs in the USA and a survey of 169 family members regarding their satisfaction with communication found that increasing the frequency of three types of clinicians' statements during family conferences was associated with increased family satisfaction[##REF##16625131##14##]. These statements included (a) assurances that the patient will not be abandoned before death, (b) assurances that the patient will be comfortable and will not suffer and (c) support for family's decisions about end-of-life care, including support for family's decision to withdraw or not to withdraw life-support. A related study used the same data to explore missed opportunities for providing support and information to families [##UREF##7##15##]. The researchers determined that in 29% of family conferences opportunities were missed by the convenors of the meetings. These missed opportunities fell into three broad categories: listening and responding to family; acknowledging and addressing emotions; pursuing key principles of medical ethics and palliative care such as exploration of patient preferences, explanation of surrogate decision making and affirmation of non-abandonment. Another sub study, (drawing on similar data) explored shared decision making strategies in the ICU and concluded there was a small degree of empirical support for this approach for family meetings [##REF##17353493##19##].</p>", "<p>Lilly and colleagues examined the effectiveness of a proactive patient and family centred communication strategy within the ICU [##REF##11042236##20##]. The intervention involved focused health professional meetings with the patient and/or their family within 72 hours of admission and incorporated a review of: the medical facts, the patient's perspective on impact of the illness/situation, and development of a care plan. Using a before- and- after study design, the intervention demonstrated reduction in the time spent in ICU and earlier withdrawal of advanced supportive technology.</p>", "<p>A study conducted in France aimed to measure the amount of time physicians in ICU spent with patients' families by examining contact over a 24 hour period with 951 patients [##REF##17079999##17##]. The median time spent providing information to families was 16 minutes per patient, with 20% of the time spent explaining the diagnosis, 20% on explaining treatments, and 60% on explaining the prognosis. One third of the time was spent listening to family members. Subsequent analysis revealed that the presence of more than one bed in the room was associated with less time spent on providing information, while a range of patient and family factors were associated with more time being spent on providing information. In an apparently related study, a randomised controlled trial demonstrated that family members who participated in a structured ICU family meeting format and received a brochure on bereavement, had less anxiety and depression than those receiving standard care[##REF##17267907##12##].</p>", "<p>A cross sectional study involving audiotaping family meetings in the ICU concluded that family satisfaction with communication was increased when they were given ample opportunity to voice their concerns, rather than just listening to the medical staff [##REF##15241092##16##]. Lautrette and colleagues [##REF##17057600##18##] claim from their review of ICU based family meeting literature that family meetings improve communication between relevant parties and help to alleviate families' emotional burden.</p>", "<p>In summary, from our review of the literature, while family meetings are promoted as a common and valuable tool within the context of specialist palliative care there is limited evidence related to how they should be conducted and indeed whether they are valuable, most evidence being extrapolated from other settings. Furthermore, despite the apparent support for family meetings, there is limited evidence from the literature that explores ways of preparing or educating health professionals to conduct them [##REF##16128661##3##]. While the findings from other clinical settings such as ICU help inform family meeting guideline development it would be unwise to generalise these results to the specialist palliative care setting given the different population (even though there are some similarities) and dissimilar foci of the meetings.</p>", "<title>Conceptual framework</title>", "<p>The conceptual foundation for the family meeting guidelines for palliative care was derived initially by the research team from the available literature and then with input from the expert panel. Research team and panel members drew upon their knowledge of various conceptual and theoretical models which appeared to be applicable to underpinning the guidelines. The final conceptual foundation was informed by: (1) The transactional model of stress and coping, (2) Single session therapy, (3) Family consultation model and (4) Therapeutic communication principles. A brief overview of each is now provided.</p>", "<title>Transactional model of stress and coping</title>", "<p>The diversity of responses related to end-of-life issues from patient and family carers can be understood from a psychological perspective based on a transactional model of coping in which cognitive appraisals are made to determine the possible impact of a potentially stressful event [##UREF##8##21##, ####REF##16594225##22##, ##REF##9381234##23####9381234##23##]. The carer's perception of self-efficacy, for example, or the greater the number of resources at their disposal to manage an event, the more likely the individual will more favourably respond to the situation. Such resources include feelings of preparedness, competence, having sufficient social support, and adequate information. This model was utilised favourably in other interventions that have assisted family carers to feel more prepared and better informed about the role of supporting a dying relative [##REF##16256897##24##,##REF##18477722##25##]. Given that family meetings tend to involve information provision about the patient and family's current status, future care options and resources the transactional model provides a useful means of approaching families.</p>", "<title>Single Session Therapy</title>", "<p>The psychologist, Moshe Talmon is most often associated with the development of Single Session Therapy which he articulated in his texts in the early 1990s in the United States [##UREF##9##26##]. Talmon initially researched the number of sessions attended by individuals seeking treatment in a mental health outpatient setting; the most common number of sessions (or mode) attended was one. Although usually viewed as 'drop outs' or treatment failures, Talmon discovered on follow up of a group of these clients that three quarters of them reported that they were improved or much improved after one unplanned session. He and his colleagues also found that when this single session was planned, the proportion of those reporting benefit increased by approximately 10% [##UREF##9##26##]. As interest in Single Session Therapy increased, its application was extended to family therapy and other settings, particularly within Australia where the results were similarly encouraging. (See, for example [##UREF##10##27##, ####UREF##11##28##, ##UREF##12##29##, ##UREF##13##30####13##30##]).</p>", "<p>Over time the practice of Single Session Therapy has developed into an approach and set of techniques that emphasizes both the client and the therapist approaching their single session with an attitude of trying to make the most of each meeting (even if further sessions are needed). It also sees Single Session Therapy as part of a process that begins with initial contact and includes phone contact after the session with the client. An advantage of the single session method is that it has been used in combination with a range of counselling approaches with differing theoretical orientations.</p>", "<p>More recently, there have been attempts to integrate Single Session approaches, including the types of questions used, into existing work roles; for example, as part of intake or assessment processes in mental health settings. In other contexts it has been used to inform meeting practices with families, where ongoing treatment and contact occurs with an individual but where professional contact with their family may be occasional or episodic. Family meetings in palliative care typically occur only once (usually as a consequence of limited time). The tenets of single session therapy therefore seem relevant to the development of family meetings clinical guidelines. Although we contend that conducting a routine family meeting should not be considered as 'therapy', some principles of single session therapy appear applicable.</p>", "<title>Family Consultation</title>", "<p>Family Consultation is a model of family intervention that emerged in the mental health field in the United States in the late 1980s [##UREF##14##31##]. It arose in response to a lack of responsiveness by services to the needs of families whose relatives were experiencing mental health problems. There was also a sense that these families were often 'co-opted' into particular forms of involvement deemed helpful to the client by professionals without any consideration of family members' own needs or wishes [##UREF##15##32##]. Typically, family consultations happen on a 'one off' or 'as needed' basis usually defined by the family. The consultations may be provided by professionals within the service setting where their family member is being treated or in separate family support services. The person with the mental health problem may or may not participate in the consultation. In contrast to family meetings in many settings, there is a conscious shift by the practitioner from an exclusive focus on how families support treatment of their ill relative to recognising and responding to the impact of the condition on other family members.</p>", "<p>The model assumes a position of family competence and seeks to cultivate a respectful and reciprocal relationship between health professionals and families [##UREF##16##33##]. In practice the professional conducting a family consultation seeks to clarify how families want to be involved in various elements of the treatment process and to develop a plan to meet the expressed needs of family members. [##UREF##17##34##] Research in relation to family consultation is limited although in one study the use of this approach resulted in an increased sense of self-efficacy for family members in managing their relatives' mental illness [##REF##8685662##35##]. Hence, given palliative care's remit to support the family alongside the patient, the family consultation model offers a useful framework to support the development of guidelines for conducting family meetings.</p>", "<title>Therapeutic communication</title>", "<p>Guidelines for therapeutic communication and psychosocial support for adults with cancer [##UREF##18##36##] offer strategies for eliciting and responding to cues from patients and family carers, which are highly relevant for palliative care related family meetings. Moreover, the guidelines highlight evidence from systematic reviews of randomised controlled trials with people with cancer that show that: (a) expressing empathy and listening actively improves psychological adjustment; (b) provision of comprehensive information about what to expect in the future promotes psychological well-being; and (c) an opportunity to discuss feelings with a health professional reduces psychosocial distress. Effective communication between health professionals and families is crucial to constructive family meeting outcomes. Consequently, the principles of therapeutic communication [##UREF##18##36##] were a core component of the guideline development.</p>", "<title>Refinement of the guidelines (expert panel and focus groups)</title>", "<p>The research team developed the initial draft of the clinical guidelines based on their clinical experience, the aforementioned literature review and the conceptual framework. The guidelines were then refined by an expert panel comprising: family therapist, social worker, psychologist, clinical nurse consultant, pastoral care consultant, consumer representative, and a palliative care medical consultant.</p>", "<p>After several iterations a penultimate version of the guidelines was presented at two focus groups for feedback to inform the final version of the guidelines. The first focus group consisted of multi-disciplinary team members from three metropolitan palliative care in-patient units. There were seven participants in this focus group. The health professionals represented the following disciplines: medicine (2); nursing (2); social work (1); occupational therapy (1); and pastoral care (1). The second focus group consisted of consultative palliative care service representatives from three major metropolitan hospitals. Eight participants contributed to this focus group including palliative care medical consultants (3); clinical nurses consultants (4); and social worker (1). Commensurate with the focus group questions (developed by the research team) the results were categorised in the following content areas: (a) description of current practice related to family meetings; (b) recommended core elements of family meetings and (c) feedback on the draft version of the family meeting clinical practice guidelines.</p>", "<p>A summary of the results is now outlined. In relation to current practice regarding family meetings it was reported that:</p>", "<p>• Meetings were typically offered based on need and not routinely;</p>", "<p>• Meetings were typically convened and chaired by a social worker;</p>", "<p>• The treating team usually determined who attended the meeting;</p>", "<p>• Formal training in conducting family meetings does not occur as part of general training;</p>", "<p>• Some units use a family meeting guide that they had developed internally;</p>", "<p>• The typical purpose of meetings was to find out what a family already knows about patient's prognosis and whether there are any gaps in knowledge;</p>", "<p>• Relevant health professionals generally try to meet before the meeting to clarify the goal of the meeting;</p>", "<p>• The social worker (who is commonly the chairperson) usually documents the outcomes of the meeting in the patient' medical record;</p>", "<p>• Documentation of the meeting discussion and outcome is not usually provided to the family;</p>", "<p>• Meetings are not formally evaluated as part of a quality improvement strategies.</p>", "<p>In relation to recommendations regarding core elements of family meetings it was reported that:</p>", "<p>• Good communication skills were considered essential;</p>", "<p>• A friendly and comfortable environment should be promoted;</p>", "<p>• Common goals and objectives should be established for the meeting;</p>", "<p>• Only key health professionals should be present at the meeting;</p>", "<p>• A designated chairperson is required for the meeting;</p>", "<p>• An agenda, specified aims and objectives should be confirmed at the start of the meeting;</p>", "<p>• The meeting should also be used as a forum to provide and gather information;</p>", "<p>• Allowing time at the beginning of the family meeting for introductions was considered essential.</p>", "<p>In relation to feedback regarding the draft version of the clinical guidelines for family meetings there was general consensus that the guidelines were applicable to the clinical setting and contained the aforementioned key elements. However the following points were considered important to be included in the guidelines:</p>", "<p>• Clarification of some terms throughout the protocol to reduce ambiguity was recommended;</p>", "<p>• Limiting the number to 'one or two' family members and/or friends was highlighted as a potential challenge in practice;</p>", "<p>• Ensuring that each discipline represented at the meeting has an opportunity to contribute was considered important;</p>", "<p>• Where pertinent and if possible (resources permitting), offer a family meeting via teleconference.</p>", "<title>Clinical practice guidelines for conducting family meetings</title>", "<p>The guidelines for conducting family meetings are outlined in Table ##TAB##0##1##. The following principles for conducting family meetings were also developed alongside the guidelines preparation process.</p>", "<title>Guiding principles for conducting family meetings</title>", "<p>• Family meetings can be a useful way to assist patients and family members to clarify goals of care, consider site of care options, and to share information. Ideally they provide a safe environment where issues and questions can be raised and appropriate strategies agreed upon.</p>", "<p>• Strategies to support family carers are a core component of palliative care; hence service providers have a responsibility to <italic>offer </italic>family meetings based on need.</p>", "<p>• Service providers should view family meetings as mutually beneficial. They are not only potentially valuable for patients and family carers; they may also provide a resource effective way to explain what the service can and cannot offer. Such meetings provide an opportunity to triage priority issues and a way to make referrals to other health professionals or other institutions early in the care planning phase.</p>", "<p>• Family meetings should <italic>not </italic>be used as an opportunity for health care professionals to debate a patient's medical status; in this situation, a case conference should be convened prior to the family meeting.</p>", "<p>• Family meetings should not be saved for 'crisis' situations. Instead, a preventative approach is advocated where issues are anticipated before they become major dilemmas. Hence a proactive rather than reactive approach to care is fostered.</p>", "<p>• Ideally, family meetings are <italic>offered </italic>routinely on admission, and conducted at a pertinent time thereafter.</p>", "<p>• Facilitators of family meetings require appropriate skills in group work, therapeutic communication and palliative care. We contend that the decision about who (i.e. which discipline) should convene and facilitate a family meeting is best determined on pragmatic grounds (local, site specific reasons) and not based on hierarchical reasons (i.e. based on authority). Hence the multidisciplinary team should determine who conducts the family meeting and presumably this may change depending upon skills, knowledge of the family and resources.</p>", "<p>• Occasionally, family members may want to withhold details of the patient's prognosis from the patient; there may be incongruent wishes about the site of care; 'desire to die' statements may have been made by the patient; or there may be conflict within the family or difficulties regarding the transition from curative treatment to palliative care. In these circumstances we recommend the key resources and references to support therapeutic communication outlined in Table ##TAB##1##2##. Additionally, if it is known in advance that there is significant conflict (or other major issues) within the family, involving a family therapist or health psychologist may be appropriate.</p>", "<p>• Pre-planning for the actual meeting is imperative as outlined in Table ##TAB##2##3##; so is comprehensive follow up after the meeting, as outlined in Tables ##TAB##3##4## and ##TAB##4##5##.</p>", "<p>• Suitable resources should be available to patients and family members who attend the meeting in order to complement the verbal information (e.g brochures about services available, carer guidebooks, treatment and drug information, etc).</p>" ]

[ "<title>Discussion and Conclusion</title>", "<p>Family meetings provide an ideal avenue to inform, deliberate, clarify and set goals for future care, based on discussions between health professionals and the patient and family. While there is consensus in the literature that family meetings are necessary and valuable, our study identified an absence of empirical evidence, within the context of specialist palliative care settings, to guide when they are required, who should attend, how they should be conducted and whether or not they are beneficial. Evidence from the intensive care context appears to offer the best guidance to date. While these data are important, caution is needed when considering their applicability to specialist palliative care settings where the population is somewhat different.</p>", "<p>The feedback from clinicians involved in the focus groups in this study also complemented other claims that most palliative care health professionals receive little or no preparation to conduct family meetings [##REF##16128661##3##]. Focus group participants also highlighted variations in current practice with regard to how family meetings were facilitated and by whom.</p>", "<p>The clinical guidelines described herewith (based on a multidisciplinary research based approach) aim to assist health care professionals working in specialist palliative care to convene and conduct family meetings. Nonetheless, our study had several limitations and consequently more research is required. The clinical guidelines presented were in the most part based on expert opinion owing to the paucity of available research evidence. Accordingly, a formal evaluation of the benefits (or otherwise) of family meetings conducted using the guidelines is required. Furthermore, it would be advantageous to know whether the guidelines are applicable to other palliative care settings such as home care and aged care. Moreover, testing the utility of the guidelines in other countries is warranted. Our literature review included studies undertaken after 1995, hence there may have been evaluations of family meetings in palliative care conducted prior to this period. However, if so, the relevance of research findings almost 15 years on may be questionable. Evaluating strategies to prepare health care professionals to conduct family meetings would also be valuable [##REF##16128661##3##].</p>", "<p>The essence of palliative care provision is effective communication amongst health professionals, patients and their family carers. Family meetings are one potential method of interaction that may facilitate optimal care planning and support and seem to be commonly used in palliative care. To our surprise however, according to our review, there have been no research studies within specialist palliative care settings that have examined: when meetings should be convened, how they should be conducted; who should attend and whether or not they are effective. It is intended that the clinical guidelines presented here will aid health care professionals to plan, conduct and evaluate family meetings. However, we recommend that the utility of these guidelines undergo additional examination.</p>" ]

[ "<title>Discussion and Conclusion</title>", "<p>Family meetings provide an ideal avenue to inform, deliberate, clarify and set goals for future care, based on discussions between health professionals and the patient and family. While there is consensus in the literature that family meetings are necessary and valuable, our study identified an absence of empirical evidence, within the context of specialist palliative care settings, to guide when they are required, who should attend, how they should be conducted and whether or not they are beneficial. Evidence from the intensive care context appears to offer the best guidance to date. While these data are important, caution is needed when considering their applicability to specialist palliative care settings where the population is somewhat different.</p>", "<p>The feedback from clinicians involved in the focus groups in this study also complemented other claims that most palliative care health professionals receive little or no preparation to conduct family meetings [##REF##16128661##3##]. Focus group participants also highlighted variations in current practice with regard to how family meetings were facilitated and by whom.</p>", "<p>The clinical guidelines described herewith (based on a multidisciplinary research based approach) aim to assist health care professionals working in specialist palliative care to convene and conduct family meetings. Nonetheless, our study had several limitations and consequently more research is required. The clinical guidelines presented were in the most part based on expert opinion owing to the paucity of available research evidence. Accordingly, a formal evaluation of the benefits (or otherwise) of family meetings conducted using the guidelines is required. Furthermore, it would be advantageous to know whether the guidelines are applicable to other palliative care settings such as home care and aged care. Moreover, testing the utility of the guidelines in other countries is warranted. Our literature review included studies undertaken after 1995, hence there may have been evaluations of family meetings in palliative care conducted prior to this period. However, if so, the relevance of research findings almost 15 years on may be questionable. Evaluating strategies to prepare health care professionals to conduct family meetings would also be valuable [##REF##16128661##3##].</p>", "<p>The essence of palliative care provision is effective communication amongst health professionals, patients and their family carers. Family meetings are one potential method of interaction that may facilitate optimal care planning and support and seem to be commonly used in palliative care. To our surprise however, according to our review, there have been no research studies within specialist palliative care settings that have examined: when meetings should be convened, how they should be conducted; who should attend and whether or not they are effective. It is intended that the clinical guidelines presented here will aid health care professionals to plan, conduct and evaluate family meetings. However, we recommend that the utility of these guidelines undergo additional examination.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Support for family carers is a core function of palliative care. Family meetings are commonly recommended as a useful way for health care professionals to convey information, discuss goals of care and plan care strategies with patients and family carers. Yet it seems there is insufficient research to demonstrate the utlility of family meetings or the best way to conduct them. This study sought to develop multidisciplinary clinical practice guidelines for conducting family meetings in the specialist palliative care setting based on available evidence and consensus based expert opinion.</p>", "<title>Methods</title>", "<p>The guidelines were developed via the following methods: (1) A literature review; (2) Conceptual framework; (3) Refinement of the guidelines based on feedback from an expert panel and focus groups with multidisciplinary specialists from three palliative care units and three major teaching hospitals in Melbourne, Australia.</p>", "<title>Results</title>", "<p>The literature review revealed that no comprehensive exploration of the conduct and utility of family meetings in the specialist palliative care setting has occurred. Preliminary clinical guidelines were developed by the research team, based on relevant literature and a conceptual framework informed by: single session therapy, principles of therapeutic communication and models of coping and family consultation. A multidisciplinary expert panel refined the content of the guidelines and the applicability of the guidelines was then assessed via two focus groups of multidisciplinary palliative care specialists. The complete version of the guidelines is presented.</p>", "<title>Conclusion</title>", "<p>Family meetings provide an opportunity to enhance the quality of care provided to palliative care patients and their family carers. The clinical guidelines developed from this study offer a framework for preparing, conducting and evaluating family meetings. Future research and clinical implications are outlined.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>PH was the principal investigator for the study and was awarded a Postdoctoral Research Grant to undertake the project. He oversaw the grant application, management of the research project and was responsible for drafting the manuscript and preparing the final submission and responsible for coordinating feedback regarding reviewer and editorial queries. KQ was a chief investigator for the project and assisted with design, data collection, project oversight and provided input into the manuscript focusing on the literature review. BO'H was a consultant for the project focusing on the areas of single session therapy and assisted primarily with the conceptual framework and literature review sections of the manuscript. SA was a chief investigator for the project and helped conceptualise the study design and provided feedback on draft versions of the manuscript. All authors read and approved the final manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1472-684X/7/12/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>• Nurses Board of Victoria for the Mona Menzies Postdoctoral Research Grant awarded to Dr Peter Hudson</p>", "<p>• Project officers: Dr Tina Thomas and Dr Colleen Nordstrom</p>", "<p>• Expert panel members: Dr Maxine Braithwaite; Ms Helen Kean; Dr Tamsin Bryan; Dr Carrie Lethborg; Dr Michael Summers; A/Professor Rosalie Hudson; Dr Annabel Pollard</p>", "<p>• Focus group participants from: Monash Medical Centre; The Alfred Hospital; Caritas Christi Hospice; Royal Melbourne Hospital, Calvary Health Care Bethlehem; Peter MacCallum Cancer Centre, Australia</p>" ]

[]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Clinical Practice Guidelines for Conducting Family Meetings in Palliative Care</p></caption><table frame=\"hsides\" rules=\"groups\"><tbody><tr><td align=\"left\"><bold>1. Preparing for a family meeting</bold></td></tr><tr><td align=\"left\"> <bold>a) </bold>On admission to the palliative care unit the relevant health professional should introduce the purpose of a family meeting and offer a family meeting to all lucid patients. This discussion should incorporate the role that palliative care has in supporting families as well as the patient.</td></tr><tr><td align=\"left\"> <bold>b) </bold>Ask the patient to confirm one or two key family carers and/or friends who they approve to be involved in medical and care planning discussions. Note this in the medical record.</td></tr><tr><td align=\"left\"> <bold>c) </bold>Conduct a family genogram to determine key relationships within the patient's family. It could be introduced thus: \"Can I spend a few minutes just working out who is in your family?\"</td></tr><tr><td align=\"left\"> <bold>d) </bold>Seek the patient's permission to arrange a family meeting and ask if they have any particular issues/concerns or questions they would like discussed at the meeting. If the patient does not want to attend, seek their permission to conduct a meeting with key family and/or friends (as above). If the patient is unable to make an informed decision, offer the meeting to the next of kin or key family/friends who have been identified to receive information and care planning decisions related to the patient. Note: Where a patient has no family or appropriate proxy a legal guardian may need to be appointed.</td></tr><tr><td align=\"left\"> <bold>e) </bold>Identify the most appropriately skilled person from the multidisciplinary team to convene the family meeting. This person will take responsibility for scheduling, invitations and coordination. Ideally this person should also act as the primary contact point for the key family carer(s).</td></tr><tr><td align=\"left\"> <bold>f) </bold>Contact the primary family carer(s): provide an overview of purpose of the family meeting; offer to convene a meeting at a mutually acceptable time. Advise the carer that the meeting time will be confirmed in due course (i.e., once other attendees are arranged). Where pertinent, and if resources allow offer to conduct the meeting via teleconference. Establish the main questions and issues that the family carer would like discussed (refer Table 3). If the patient is participating in the meeting ask him/her to identify their key concerns.<break/><underline>Note:</underline> If significant family conflict (or other major issue) is identified consider referral to a practitioner who is trained to work with complex issues within families (e.g. family therapist or health psychologist).</td></tr><tr><td align=\"left\"> <bold>g) </bold>Determine which health care professionals should attend the family meeting. Invite key health care professionals based on the identified needs of the patient and family carer. The number of staff should be restricted, inviting only the relevant health professionals, so that the patient and family/friends do not feel overwhelmed. Note: Include a professional interpreter if required.</td></tr><tr><td align=\"left\"> <bold>h) </bold>Confirm the family meeting time and location. Inform attendees of the scheduled start and finish time for the meeting. A comfortable room free of interruptions (including pagers and phones), tissues made available and conducive seating arrangements is recommended.</td></tr><tr><td align=\"left\"><bold>2. Conducting a family meeting</bold></td></tr><tr><td align=\"left\"> <bold>a) </bold>Introduction</td></tr><tr><td align=\"left\">  Chairperson to:</td></tr><tr><td align=\"left\">  <bold>i) </bold>Thank everyone for attending and introduce him/herself and invite others to introduce themselves and state their role.</td></tr><tr><td align=\"left\">  <bold>ii) </bold>Establish ground rules in a non patronsing way e.g. <italic>\"We would like to hear from all of you, however if possible could one person please speak at a time, each person will have a chance to ask questions and express views.\" </italic>Request no interruptions such as phones etc.</td></tr><tr><td align=\"left\">  <bold>iii) </bold>Indicate the duration of meeting (recommended maximum time of 60 minutes).</td></tr><tr><td align=\"left\"> <bold>b) </bold>Determine the understanding of the purpose of the family meeting.</td></tr><tr><td align=\"left\">  Chairperson to:</td></tr><tr><td align=\"left\">  <bold>i) </bold>Briefly <bold>outline </bold>the broad purpose of the family meeting (based on previous steps), and then confirm with the family and patient that their interpretation of the purpose of the meeting concurs.</td></tr><tr><td align=\"left\">   For example:</td></tr><tr><td align=\"left\">   \"<italic>We arranged this meeting to consider discharge planning options. Is this your understanding of the purpose of the meeting?\" (If not reframe the meeting's purpose)</italic></td></tr><tr><td align=\"left\">   <italic>or</italic></td></tr><tr><td align=\"left\">   <italic>\"From the things you mentioned on the questionnaire what is the most important thing you would like to discuss?\"</italic></td></tr><tr><td align=\"left\">   <italic>or</italic></td></tr><tr><td align=\"left\">   <italic>\"How could we be most helpful to you today?\"</italic></td></tr><tr><td align=\"left\">  <bold>ii) </bold>Ask the patient/family if there are any additional key concerns, and if pertinent, prioritise these and confirm which ones will be attempted to be dealt with at this meeting (others can be discussed at a future meeting or can perhaps dealt with on a one on one basis).</td></tr><tr><td align=\"left\">  <bold>iii) </bold>Clarify if specific decisions need to be made (e.g. if the patient is to go home or not).</td></tr><tr><td align=\"left\"> <bold>c) </bold>Determine what the patient and family already know. Possible questions may include,</td></tr><tr><td align=\"left\">  \"<italic>What have you been told about palliative care</italic>\" as a way of clarifying, confirming etc.</td></tr><tr><td align=\"left\">  \"<italic>Tell me your understanding of the current medical condition or current situation</italic>?\"</td></tr><tr><td align=\"left\">  If pertinent provide information (in accordance with desire) on the patient's current status, prognosis and treatment options.</td></tr><tr><td align=\"left\">  Ask each family member in turn if they have any questions about current status, plan and prognosis. Helpful questions may include, \"<italic>Do you have questions or concerns about the treatment or care plan?\"</italic></td></tr><tr><td align=\"left\">  <underline>For family discussion with non-competent patient (i.e. cognitively impaired or imminently dying).</underline></td></tr><tr><td align=\"left\">  Ask each family member in turn:</td></tr><tr><td align=\"left\">  \"<italic>What do you believe your relative/friend would choose if they could speak for himself/herself?\"</italic></td></tr><tr><td align=\"left\">  <italic>\"In the light of that knowledge, what do you think should be done?\"</italic></td></tr><tr><td align=\"left\"> <bold>d) </bold>Address specific objectives of the meeting (as previously determined).</td></tr><tr><td align=\"left\"> <bold>e) </bold>'Check in' periodically throughout with the patient and family carer to see if the discussion seems to be valuable and is in keeping with their needs e.g. <italic>\"Are we on track?\"; \"Is this what you wanted from today's meeting?\"; \"What haven't we touched on that's important to you?\"</italic></td></tr><tr><td align=\"left\">  Also consider taking a short break during the meeting (to give participants time to digest information) and then allow some time to refocus.</td></tr><tr><td align=\"left\"> <bold>f) </bold>Offer relevant written or audiovisual resources. Examples include guidebooks, brochures, enduring power of attorney documents, advance care directive information and so forth.</td></tr><tr><td align=\"left\"> <bold>g) </bold>Identify other resources, including possible referral to other members of the multidisciplinary team. Suggest scheduling a follow-up meeting if pertinent.</td></tr><tr><td align=\"left\"> <bold>h) </bold>Concluding the discussion.</td></tr><tr><td align=\"left\">  Summarize any areas of consensus, disagreements, decisions and the ongoing plan (i.e. clarify next steps) and seek endorsement from attendees (e.g. <italic>\"Are we all clear on the next steps?\")</italic></td></tr><tr><td align=\"left\">  Emphasize positive outcomes arising from the meeting.</td></tr><tr><td align=\"left\">  Offer final opportunity for questions, concerns, or comments. E.g. <italic>\"What hasn't been covered today that you would have like to discuss?\" </italic>or <italic>\"Are there any questions you had that haven't been answered yet?\"</italic></td></tr><tr><td align=\"left\">  Remind patient and family carers to review the recommended written resources.</td></tr><tr><td align=\"left\">  Identify one family spokesperson for ongoing communication.</td></tr><tr><td align=\"left\">  Thank everyone for attending.</td></tr><tr><td align=\"left\"><bold>3. Documentation and follow-up</bold></td></tr><tr><td align=\"left\"> <bold>a) </bold>Document who was present, what decisions were made, what the follow-up plan is and share this with the care team (see Table 4).</td></tr><tr><td align=\"left\"> <bold>b) </bold>Offer the patient/family a copy of the main content of the meeting and file a copy of this document in the patient's medical record.</td></tr><tr><td align=\"left\"> <bold>c) </bold>Liaise with the primary family carer within a few days after the meeting to determine if the meeting was helpful (see Table 5)</td></tr><tr><td align=\"left\"> <bold>d) </bold>Maintain contact with the key family spokesperson, including attending scheduled follow-up meetings or telephone calls as needed.</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Recommended key references and resources for conducting family meetings</p></caption><table frame=\"hsides\" rules=\"groups\"><tbody><tr><td align=\"left\">• Clinical practice guidelines for the psychosocial care of adults with cancer [##UREF##18##36##]</td></tr><tr><td align=\"left\">• Key communication skills and how to acquire them [##REF##12351365##37##,##UREF##19##38##]</td></tr><tr><td align=\"left\">• Clinical practice guidelines for communicating prognosis and end-of-life issues with adults in the advanced stages of a life-limiting illness, and their carers [##REF##17727340##39##]</td></tr><tr><td align=\"left\">• Responding to desire to die statements from patients with advanced disease: recommendations for health professionals [##REF##17060269##40##]</td></tr><tr><td align=\"left\">• Supporting a person who requires palliative care: A guide for family and friends [##UREF##20##41##] (via: <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.pallcarevic.asn.au/\"/>)</td></tr><tr><td align=\"left\">• 'Would you like to talk about your future treatment options?' discussing the transition from curative cancer treatment to palliative care [##REF##16875109##42##]</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Pre-Family Meeting Primary Family Carer Questionnaire</p></caption><table frame=\"hsides\" rules=\"groups\"><tbody><tr><td align=\"left\" colspan=\"2\">Nb Conducted by phone [] or face to face [] by Family meeting convenor ............ [insert name]</td></tr><tr><td align=\"left\" colspan=\"2\">Now that I have explained about the family meeting and you have agreed to attend it would be useful for us if we had some more information in order to prepare for the family meeting.</td></tr><tr><td align=\"left\" colspan=\"2\">What are the main issues for you at the moment?</td></tr><tr><td align=\"left\" colspan=\"2\">(a) Greatest concern:</td></tr><tr><td align=\"left\" colspan=\"2\">................................................................................................................................................................</td></tr><tr><td align=\"left\" colspan=\"2\">(b) Second greatest concern:</td></tr><tr><td align=\"left\" colspan=\"2\">................................................................................................................................................................</td></tr><tr><td align=\"left\" colspan=\"2\">How upset/worried are you about these concerns? <italic>(Place a cross on the line)</italic></td></tr><tr><td align=\"left\" colspan=\"2\">--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------</td></tr><tr><td align=\"left\"><italic>(1) Not at all</italic></td><td align=\"right\"><italic>As worried as I could possibly be (10)</italic></td></tr><tr><td align=\"left\" colspan=\"2\">How often do these concerns arise? <italic>(Place a cross on the line)</italic></td></tr><tr><td align=\"left\" colspan=\"2\">--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------</td></tr><tr><td align=\"left\"><italic>(1) Not at all</italic></td><td align=\"right\"><italic>All the time (10)</italic></td></tr><tr><td align=\"left\" colspan=\"2\">Are there other difficulties you are coping with now? Please outline below:</td></tr><tr><td align=\"left\" colspan=\"2\">...............................................................................................................................................................</td></tr><tr><td align=\"left\" colspan=\"2\">...............................................................................................................................................................</td></tr><tr><td align=\"left\" colspan=\"2\">How much is the problem (or problems) interfering in your life? <italic>(Place a cross on the line)</italic></td></tr><tr><td align=\"left\" colspan=\"2\">--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------</td></tr><tr><td align=\"left\"><italic>(1) Not at all</italic></td><td align=\"right\"><italic>Dominating my life completely (10)</italic></td></tr><tr><td align=\"left\" colspan=\"2\">How confident do you feel in dealing with the problem(s)? <italic>(Place a cross on the line)</italic></td></tr><tr><td align=\"left\" colspan=\"2\">--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------</td></tr><tr><td align=\"left\"><italic>(1) Not at all</italic></td><td align=\"right\"><italic>Extremely (10)</italic></td></tr><tr><td align=\"left\" colspan=\"2\">What questions would you like to ask at the family meeting?</td></tr><tr><td align=\"left\" colspan=\"2\">...............................................................................................................................................................</td></tr><tr><td align=\"left\" colspan=\"2\">If you think of other questions between now and the family meeting, please write them down and bring them with you to the meeting.</td></tr><tr><td align=\"left\" colspan=\"2\"><italic>Adapted with permission from Single Session Therapy Resource Guide (The Bouverie Centre 2006)</italic></td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Outcome of the Family Meeting</p></caption><table frame=\"hsides\" rules=\"groups\"><tbody><tr><td align=\"left\" colspan=\"5\">Below are key points to be recorded at the completion of the family meeting by the Family Meeting's Facilitator.</td></tr><tr><td align=\"left\" colspan=\"5\">A copy should be provided to the patient and family carer and one copy kept in the medical record.</td></tr><tr><td align=\"left\" colspan=\"5\"><bold>Date of meeting: </bold>__________________________</td></tr><tr><td align=\"left\" colspan=\"5\"><bold>Name of family meeting facilitator: </bold>_________________________________</td></tr><tr><td align=\"left\" colspan=\"5\"><bold>Proposed purpose of the meeting: </bold>__________________________________</td></tr><tr><td align=\"left\" colspan=\"5\"><bold>FAMILY MEMBERS PRESENT</bold></td></tr><tr><td align=\"left\" colspan=\"2\">Name:</td><td align=\"left\" colspan=\"3\">Relationship:</td></tr><tr><td align=\"left\" colspan=\"2\">Name:</td><td align=\"left\" colspan=\"3\">Relationship:</td></tr><tr><td align=\"left\" colspan=\"2\">Name:</td><td align=\"left\" colspan=\"3\">Relationship:</td></tr><tr><td align=\"left\" colspan=\"5\"><bold>STAFF MEMBERS PRESENT</bold></td></tr><tr><td align=\"left\" colspan=\"2\">Name:</td><td align=\"left\" colspan=\"3\">Role/Discipline:</td></tr><tr><td align=\"left\" colspan=\"2\">Name:</td><td align=\"left\" colspan=\"3\">Role/Discipline:</td></tr><tr><td align=\"left\" colspan=\"2\">Name:</td><td align=\"left\" colspan=\"3\">Role/Discipline:</td></tr><tr><td align=\"left\" colspan=\"5\"><bold>KEY ISSUES RAISED AT THE MEETING</bold></td></tr><tr><td align=\"left\" colspan=\"5\">_________________________________________________________________________________</td></tr><tr><td align=\"left\" colspan=\"5\">_________________________________________________________________________________</td></tr><tr><td align=\"left\" colspan=\"5\"><bold>KEY ACTIONS FROM THE MEETING</bold></td></tr><tr><td align=\"left\">Current Situation</td><td align=\"left\">Goal</td><td align=\"left\">Action</td><td align=\"left\">Key Person to follow up</td><td align=\"left\">Review Date</td></tr><tr><td/><td/><td/><td/><td/></tr><tr><td/><td/><td/><td/><td/></tr><tr><td/><td/><td/><td/><td/></tr><tr><td align=\"left\" colspan=\"5\"><italic>Adapted (with permission) from Single Session Therapy Resource Guide (The Bouverie Centre 2006)</italic></td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T5\"><label>Table 5</label><caption><p>Post-Family Meeting Primary Family Carer Questionnaire</p></caption><table frame=\"hsides\" rules=\"groups\"><tbody><tr><td align=\"left\" colspan=\"4\">Nb Conducted by phone [] or face to face []. Completed by ............ [insert name]</td></tr><tr><td align=\"left\" colspan=\"4\">As a follow up to the recent family meeting we are interested in finding out how things are for you at the moment. Before the family meeting</td></tr><tr><td align=\"left\" colspan=\"4\">You nominated:</td></tr><tr><td align=\"left\" colspan=\"4\">.................................................................................................................................................................</td></tr><tr><td align=\"left\" colspan=\"4\">as the main problem to be discussed at the family meeting, and</td></tr><tr><td align=\"left\" colspan=\"4\">.................................................................................................................................................................</td></tr><tr><td align=\"left\" colspan=\"4\">as your second greatest problem.</td></tr><tr><td align=\"left\" colspan=\"4\">How upset/worried are you about this problem (or these problems) at the present time? <italic>(Place a cross on the line)</italic></td></tr><tr><td align=\"left\" colspan=\"4\">----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------</td></tr><tr><td align=\"left\" colspan=\"2\"><italic>(1) Not at all</italic></td><td align=\"right\" colspan=\"2\"><italic>As worried as I could possibly be (10)</italic></td></tr><tr><td align=\"left\" colspan=\"4\">How often do these problems happen? <italic>(Place a cross on the line)</italic></td></tr><tr><td align=\"left\" colspan=\"4\">----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------</td></tr><tr><td align=\"left\" colspan=\"2\"><italic>(1) Not at all</italic></td><td align=\"right\" colspan=\"2\"><italic>All the time (10)</italic></td></tr><tr><td align=\"left\" colspan=\"4\">How much is the problem (or problems) interfering in your life? <italic>(Place a cross on the line)</italic></td></tr><tr><td align=\"left\" colspan=\"4\">----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------</td></tr><tr><td align=\"left\" colspan=\"2\"><italic>(1) Not at all</italic></td><td align=\"right\" colspan=\"2\"><italic>Dominating my life completely (10)</italic></td></tr><tr><td align=\"left\" colspan=\"4\">In what ways?...........................................................................................................................................</td></tr><tr><td align=\"left\" colspan=\"4\">How confident do you feel in dealing with the problem(s)? <italic>(Place a cross on the line)</italic></td></tr><tr><td align=\"left\" colspan=\"4\">----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------</td></tr><tr><td align=\"left\" colspan=\"2\"><italic>(1) Not at all</italic></td><td align=\"right\" colspan=\"2\"><italic>Extremely (10)</italic></td></tr><tr><td align=\"left\" colspan=\"4\">You nominated the following questions as those you would like addressed in the family meeting:</td></tr><tr><td align=\"left\" colspan=\"4\">.................................................................................................................................................................</td></tr><tr><td align=\"left\" colspan=\"4\">To what extent do you feel these questions were addressed?</td></tr><tr><td align=\"left\" colspan=\"4\">.................................................................................................................................................................</td></tr><tr><td align=\"left\"><italic>Office use only:</italic></td><td/><td/><td/></tr><tr><td/><td align=\"left\">Pre-session</td><td align=\"left\">Post-session</td><td align=\"left\">Difference</td></tr><tr><td align=\"left\">How upset/worried:</td><td align=\"left\">........................</td><td align=\"left\">........................</td><td align=\"left\">........................</td></tr><tr><td align=\"left\">Problem frequency:</td><td align=\"left\">........................</td><td align=\"left\">........................</td><td align=\"left\">........................</td></tr><tr><td align=\"left\">Life interference:</td><td align=\"left\">........................</td><td align=\"left\">........................</td><td align=\"left\">........................</td></tr><tr><td align=\"left\">Confidence:</td><td align=\"left\">........................</td><td align=\"left\">........................</td><td align=\"left\">........................</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>Note: Publishers please aware that when inserting this box in the body of the text the ordering of the citations and reference list will need to be adjusted accordingly</p></table-wrap-foot>" ]

[]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Over the last 10 years the use of antiretroviral therapy (ART) in HIV-infected children has resulted in noticeable benefit as indicated by increased survival [##REF##14758118##1##,##REF##12553485##2##]. However, poor adherence to prescriptions and the high rates of virus replication that are characteristic of perinatal infection [##REF##10225229##3##] often lead to higher virological set points in children compared to adults and lower rates of attainment of undetectable viral loads. Therefore, the identification of immunological correlates of immune reconstitution and early predictors of antiretroviral failure in HIV-treated children are needed. Concomitantly with loss of CD4⁺cells, children with HIV infection display a profound impairment of the innate branch of immune system as indicated by the progressive decrease of circulating dendritic cells (DC). DC are a heterogeneous population of antigen-presenting cells that are required to draw the first line of host defense against viral infections [##REF##9521319##4##, ####REF##10545981##5##, ##REF##9345009##6####9345009##6##]. Two subsets of DC were originally identified in peripheral blood on the basis of the β2-integrin expression (CD11c marker): the CD11c⁺myeloid DC (mDC) and the CD11c- plasmacytoid DC (pDC). DC phenotype is defined on the basis of the two monoclonal antibodies (mAb) BDCA-1 and BDCA-2, which identify blood mDC and pDC, respectively [##REF##11748283##7##,##REF##15332126##8##]. But, BDCA-1 is also expressed on monocytes and is not sufficient by itself to define dendritic cells [##REF##15332126##8##].</p>", "<p>Although blood DC represent less than 1% of total peripheral blood mononuclear cells (PBMC), they exert a relevant protective role against invading pathogens by producing IL-12 and interpheron-alpha (IFN-α) and by inducing T-cell immunity via presentation of pathogen-specific antigens on their cell surface [##REF##15771572##9##, ####REF##15990333##10##, ##REF##15549123##11##, ##REF##15596797##12##, ##REF##10364556##13####10364556##13##]. IFN-α has been shown to decrease HIV replication by induction of IFN-stimulated genes. One such gene is Myxovirus resistance 1, which encodes for the Myxovirus resistance protein A (MxA), a protein capable of inhibiting the replication of several viruses, including HIV [##REF##2120071##14##,##REF##15585849##15##]. However, it is controversial whether the increase of IFN-α secretion, which is observed during disease progression, contributes to HIV pathogenesis. It has been described that high levels of IFN-α might exert deleterious effects on the immune cells by inducing depletion of uninfected CD4⁺lymphocytes [##REF##9520456##16##, ####REF##16632604##17##, ##REF##17112786##18####17112786##18##]. Like CD4⁺cells, DC express the receptor machinery necessary for HIV entry, and therefore are vulnerable to the detrimental effects of HIV infection and are functionally impaired in HIV-1-infected patients [##REF##11588058##19##, ####REF##11698285##20##, ##REF##11971031##21##, ##REF##12508143##22##, ##REF##11493432##23##, ##REF##11683579##24####11683579##24##]. Furthermore, pDC loss is believed to be a predictor of disease progression, of increased risk of opportunistic infections and of Kaposi sarcoma development [##REF##11588058##19##,##REF##11698285##20##,##REF##11493432##23##,##REF##11683579##24##].</p>", "<p>It is known that immune reconstitution after ART is mostly due to newly released lymphocytes from the thymus [##REF##12134227##25##]. However, at the same time, antiretroviral regimens influence DC pool in both adults and children by completely restoring mDC and only partially recovering the frequency and function of pDC [##REF##11971031##21##,##REF##16622034##26##].</p>", "<p>In this study we evaluated pDC and IFN-α involvement in a cohort of 33 perinatally HIV-infected patients undergoing ART at various stages of immune-reconstitution, as determined by analysis of their viral load and CD4⁺T-cell counts.</p>" ]

[ "<title>Patients and methods</title>", "<title>Patients and controls</title>", "<p>Informed consent was obtained, in accordance with institutional review board guidelines for the protection of human subjects, from parents or legal guardians of 33 subjects with perinatal HIV infection (age: from 2 up to 19 years), and from 15 age-matched healthy controls (mean age: 11.8 years, range: 2–19 years). Patients with acute infectious diseases or opportunistic infections at the time of the analysis were excluded from the study.</p>", "<title>Circulating pDC and serum IFN-α levels determination</title>", "<p>The identification of pDC was performed using Blood Dendritic Cell enumeration Kit (Miltenyi Biotec GmbH, Bergish Gladbach, Germany) as previously described [##REF##12893749##27##]. Blood samples were stained simultaneously with anti-BDCA-2-FITC mAb to identify pDC, anti-CD19-PE-Cy5 and anti-CD14-PE-Cy5 mAbs to exclude B cells and monocytes, and with a fluorescent cell-impermeant dye (Dead cells Discriminator), which binds covalently and irreversibly to nucleic acids of dead cells. After cells staining, erythrocytes were lysed with ammonium chloride buffer (Red Blood Cell Lysis), and cells were washed and fixed using formaldehyde (Fix Solution). Finally, cells were analyzed by 3-color flow cytometry using FACScan (Becton Dickinson, San Jose, CA, USA).</p>", "<p>The level of IFN-α in the serum was measured by Sandwich ELISA using an IFN-α kit (PBL Biomedical Laboratories, Piscataway, NJ, USA) and following the manufacturer's instructions.</p>", "<title>IFN-α, BDCA-2, MxA and TRECs quantification by real-time PCR</title>", "<p>Blood samples (2.5 ml) from patients and controls were drawn into PAXgene tubes (preAnalytiX GmbH, Hombrechtikon, CH), which contain an additive that stabilizes gene transcription profile by reducing <italic>in vitro </italic>RNA degradation and minimizing gene induction. RNA was prepared using PAXgene Blood RNA kit following manufacturer's instructions (preAnalytiX). To remove contaminating DNA, samples were treated with DNase (Qiagen Inc. Valencia, CA, USA). Five hundred ng of total RNA were reverse transcribed with random hexamers in a final volume of 20 μl using the TaqMan RT kit (Applied Biosystems, Foster City, CA, USA). Real-time PCR analysis was performed with the GeneAmp 5700 Sequence Detection System (Applied Biosystems). For IFN-α and BDCA-2 mRNA measure a pre-developed TaqMan assay, which included the appropriate primer/probe combination, was purchased from Applied Biosystems and the test was performed following manufacturer's instructions. For MxA mRNA quantification we used primers and probes described by Pachner et al. [##REF##14529316##28##]; the experiments were performed after standardization of the assay in terms of precision, accuracy and reproducibility [##REF##17619058##29##]. All PCR reactions were set up in 96-well optical reaction plates (ABgene, Epsom, UK) in a final volume of 25 μl with 12.5 μl of double concentrated TaqMan universal PCR Master mix (Applied Biosystems), 3 μl of cDNA, 1.25 μl of primer/probe mix for IFN-α and BDCA-2 and 600 nM of primers and 200 nM of probe for MxA. Probe and primer concentration for the housekeeping gene, Glyceraldehyde-3-phosphate dehydrogenase (GAPDH), were respectively 400 and 200 nM. PCR amplification started with a first step at 50°C for 2 min., followed by an initial heating at 95°C for 10 min.; samples were then subjected to 45 cycles of denaturation at 95°C for 15 sec and annealing at 60°C for 1 min. Each sample was run in duplicate.</p>", "<p>The relative quantification of BDCA-2, IFN-α and MxA mRNA expression was calculated using the Comparative cycle threshold (Ct) method according to the following formula: Normalization Ratio (NR) = 2^-ΔΔCtCt. First, for each sample, ΔCt value was calculated as the Ct of the target gene minus the Ct of GAPDH and then the ΔΔCt value was obtained as the difference between the ΔCt of the sample and the ΔCt of the calibrator. According to the formula, the normalization ratio of the calibrator in each run is 1. As calibrator in each sample run, we utilized the same RNA extracted from a single healthy control, and stored at -80°C.</p>", "<p>The quantification of the signal-joint T-cell receptor excision circles (TRECs) was performed on PBMC separated by Ficoll-Hypaque density gradient centrifugation using the standard curve method described by Pirovano et al. [##REF##15075079##30##].</p>", "<title>Immunohistochemistry</title>", "<p>Lymph node and tonsil specimens from 8 patients with HIV infection and from 2 age-matched controls were analyzed. Each specimen was fixed in buffered formalin and embedded in paraffin. Sections were stained overnight with a dilution 1:50 of mouse anti-CD123 mAb (clone 7G3, IgG2a, Becton Dickinson PharMingen Biosciences Europe, Erembodegem-AALST, Belgium), after inhibition of endogenous peroxidase activity with 0.3% H₂O₂in methanol for 20 min and heat-induced epitope retrieval in Tris-EDTA buffer (pH 9.0). Reaction was developed with the polymer peroxidase conjugated technique (ChemMate, Dako, Glostrup, DK) and diaminobenzydine as chromogen. Sections were counterstained with haematoxylin. Cell counting was performed using an Olympus BX60 microscope equipped with the Olympus DP-70 digital camera and the Olympus Soft Imaging System Cell-F 2.5; two medium-power fields (corresponding to 0.3 mm²), selected on the basis of highest content of positive cells, were evaluated and values were expressed as mean per one mm².</p>", "<title>Statistical analysis</title>", "<p>Since real-time PCR data did not follow a Gaussian distribution, results were expressed as median and range and analyzed using nonparametric statistical tests. Differences between values from patients and controls were examined by Mann-Whitney test, while comparisons between more than 2 groups of results were assessed by Kruskal-Wallis ANOVA. In case of significance, Dunn's multiple columns post-test was applied to compare the sub-groups of treatment. Moreover, Spearman's correlation test was used to find an association between variables. Two-tailed P values &lt; 0.05 were considered to be statistically significant.</p>" ]

[ "<title>Results</title>", "<title>Clinical and immune features of perinatally HIV- infected patients</title>", "<p>We evaluated pDC number and function in a population of 33 perinatally HIV-infected patients, 17 of them were female and 16 male. CD4⁺T lymphocytes ranged from 18.6% to 43.4% (median of 31.6%; Table ##TAB##0##1##). In particular, 26 of 33 subjects showed an adequate immune reconstitution (Category 1, CD4⁺% ≥ 25%), while the remaining patients had a low number of CD4⁺and a high viral load. They all were under ART at the time of the analysis according to PENTA guidelines for treatment of HIV infection in children [##REF##15239717##31##], but the adopted treatment regimen varied from subject to subject depending on the pattern of genotypic resistance and on the adherence to the therapy [##REF##16094240##32##]. Twenty-four patients were subjected to ART regimen containing at least one protease inhibitor, 7 patients were treated with a non-nucleoside reverse transcriptase inhibitor, and the remaining 2 patients with two nucleoside reverse transcriptase inhibitors because of their poor adherence to other regimens. Despite the adoption of ART, complete suppression of viral load (less than 50 copies/ml) was achieved only in 18 out of 33 patients (Table ##TAB##0##1##). This unsatisfactory result was probably the consequence of several contributing reasons, including the common observance of genotypic resistance in children failing the treatment, the long length of the disease in older patients, the slow reduction of viral load following the introduction of ART in children, and possibly, the difficulty in obtaining long-term adherence to therapy in adolescents [##REF##12553485##2##,##REF##10225229##3##].</p>", "<title>pDC in HIV-infected perinatally HIV-infected patients</title>", "<p>We evaluated both the proportion and the absolute number of pDC in HIV-infected patients by using an assay that selectively identified the BDCA-2⁺CD14^-CD19^-cell subset. As observed in healthy controls [##REF##11975756##33##], there is an inverse correlation (r = -0.503, p = 0.003) of pDC values with age (Figure ##FIG##0##1A##) and a direct correlation (r = 0.537; p = 0.002) with the absolute cell number of CD4⁺T lymphocytes (Figure ##FIG##0##1B##). The progressive decline of pDC (values ranging from 33,300 cells/μl to 3,402 cells/μl) can reflect the \"natural\" decline in pDC counts during ageing [##REF##11975756##33##] and is related to the reduction of CD4⁺cell number. This is in agreement with data from Azzoni et al. [##REF##15809903##34##], demonstrating a greater depletion of pDC in HIV⁺children with a clinical history of decreasing CD4⁺cell percentage. Median values of pDC, both in terms of percentage and absolute number, are not statistically different from those measured in age-matched controls, although pDC number of HIV⁺patients receiving ART is heterogeneous and varies from extremely low blood counts up to levels observed in controls (Figure ##FIG##0##1C##).</p>", "<p>Since BDCA-2 was shown to be modulated following pDC activation [##REF##11748283##7##], we have evaluated the amount of BDCA-2 mRNA by real-time PCR. Comparable values were found in patients and healthy controls (median value of: 1.13 in HIV⁺patients vs. 0.93 in healthy children, p = NS), thus confirming the data obtained by cytofluorimetry. We did not assess pDC counts before treatment in our patients, but based on previous observations we speculate that pDC counts, possibly lower before treatment, might be increased after starting the ART.</p>", "<title>IFN-α and MxA mRNA induction during HIV infection</title>", "<p>Since pDC are specialized cells that produce large amounts of IFN-α in response to viral infections, we sought to determine whether HIV infection induces the expression of this cytokine. We observed a significant increase of IFN-α mRNA in PBMC in HIV-infected patients as compared to healthy control subjects (Figure ##FIG##0##1D##), suggesting that the chronic infection induces IFN-α production. Next, we analyzed the expression of MxA mRNA, which is selectively induced after type-I IFN receptor binding and is a marker for cell responsiveness to IFN-α [##REF##2120071##14##,##REF##2481229##35##]. We found that MxA mRNA levels were significantly increased in HIV-infected patients with abnormal HIV viral load RNA. Indeed, we observed a direct correlation between MxA expression and HIV RNA copies in HIV infected patients (Figure ##FIG##1##2B##), suggesting that a high number of viral copies may directly influence the induction of MxA mRNA. Due to the fact that pDC are one of the principal producers of IFN-α [##REF##15771572##9##] and that MxA is a marker of type I IFN bioactivity [##REF##17619058##29##,##REF##2481229##35##,##REF##9562412##36##], one might predict that patients with the largest number of circulating pDC might express more MxA as compared to age-matched controls (Figure ##FIG##1##2A##). On the contrary, we have observed that higher levels of MxA mRNA are found in those patients with lower number of circulating pDC, as indicated by the strong negative correlation between these two parameters (Figure ##FIG##1##2C##). In particular, we observed that 15 patients displayed MxA mRNA levels above the cut-off, established as the 99^thpercentile of the distribution of MxA in healthy control subjects (NR = 2.7, shown as dotted line in the figure), while the remaining presented normal MxA mRNA expression. Surprisingly, patients with MxA mRNA levels above the cut-off are those with lower numbers of pDC (Figure ##FIG##2##3A##) and significantly higher serum levels of IFN-α (Figure ##FIG##2##3B##). It is of note that the median level of this cytokine is low in both groups, at levels observed in control children, but it has been demonstrated that circulating IFN-α is elevated in the sera just at the moment of HIV antigen appearance [##REF##2890005##37##]. Moreover, we have also observed that children with low MxA mRNA (&lt;2.7) below the cut-off show an increased number of CD4⁺lymphocyte (Figure ##FIG##2##3C##) and that children with increased MxA mRNA (&gt;2.7) are of older age (Figure ##FIG##2##3D##). Since MxA levels do not vary among ages [##REF##17619058##29##,##REF##9562412##36##], we speculate that increased levels of MxA mRNA are associated with the length of HIV infection. Finally, since pDC are present within the thymus [##REF##15585849##15##] and immune-reconstitution is related to thymic function [##REF##16989615##38##], we utilized real-time PCR to determine the levels of DNA episomes created in the thymus during T-cell receptor rearrangement process, known as TRECS. We observed that TREC values were comparable in HIV-infected patients and age-matched controls (63,345 TRECs/1 × 10⁶cells, range: 39,470–173,194 vs. 54,560 TRECs/1 × 10⁶, range: 7,279–299,500), without any detectable difference between patients with normal or increased MxA mRNA values (median: 65,925 TRECs/1 × 10⁶cells, range: 15,259–155,003 vs. 51,125 TRECs/1 × 10⁶cells, range: 17,586–128,900). Similarly, there was no significant correlation between TRECs and pDC cell number (data not shown).</p>", "<title>pDC in lymphoid tissues of HIV infected children</title>", "<p>The decline of circulating pDC, correlating with reduction of CD4⁺and increased IFN-α-dependent MxA gene expression, in children with advanced HIV infection suggests that pDC might accumulate in secondary lymphatic tissues [##REF##18300699##39##]. Thus we sought to determine if pDC (identified as CD123⁺cells) are present in lymph nodes and tonsils, previously obtained for diagnostic purposes from 8 HIV-infected children. The number of pDC identified in lymphoid tissues of HIV-infected children was extremely variable: in some specimens the pDC were abundant in the interfollicular areas of tonsils and lymph nodes of both age-matched control subjects (a section of tonsil tissue is shown as representative example in Figure ##FIG##3##4A##) and HIV⁺patients (a representative example is shown in Figure ##FIG##3##4B##), while in other HIV-infected children they were severely depleted (Figure ##FIG##3##4C##). It should be noted that pDC were regularly scant in cases showing general lymphoid depletion, but they are found to be rare also in cases with preserved lymphoid tissue, as indicated in Figure ##FIG##3##4C##. After plotting the number of pDC detected in lymphoid tissues of the HIV-infected children against the percentage of CD4⁺or against the HIV-RNA copy number, we observed a direct correlation between the number of lymphatic tissue-resident pDC and CD4⁺blood counts, while there was no correlation with HIV viral load (Figure ##FIG##3##4D##).</p>" ]

[ "<title>Discussion</title>", "<p>The relationship between pDC and CD4⁺cells is based on numerous concordant observations. Following HIV infection, both cell subsets decrease to levels associated with increased susceptibility to opportunistic infections [##REF##11683579##24##]. Due to the fact that pDC infiltrate the medulla and the corticomedullary junction of the thymus, where they inhibit the replication of the virus [##REF##15585849##15##], and due to the relationship between HIV infection and the release of newly generated T lymphocytes from the thymus into peripheral blood [##REF##16622034##26##], we determined whether there was a direct relationship between the number of pDC and the extent of thymic output. We did not observe a correlation between the number of newly produced T lymphocytes containing TRECs and the number of pDC, most likely due to the fact that lymphocyte homeostasis is also regulated in peripheral lymphoid organs, depending on cell replication and removal [##REF##16963128##40##].</p>", "<p>Although higher pDC frequency found in long-term non-progressors [##REF##15718836##41##] indicate that pDC also have a role in HIV infection control, a possible link also exists between disease outcome and reduced pDC number and function, including a decreased production of IFN-α by pDC during AIDS progression [##REF##11683579##24##].</p>", "<p>High plasma levels of IFN-α can be detected during acute HIV infection [##REF##1906289##42##], but it is usually barely detectable in chronic HIV infection due to its short half-life or the rapid binding to its receptor [##REF##1702693##43##, ####REF##9848789##44##, ##REF##3919398##45####3919398##45##]. Accordingly, we found low serum levels of IFN-α in HIV-infected children, but we detected significantly increased levels of IFN-α mRNA in blood cells, suggesting that pDC were in an active state and potentially capable to deliver the cytokine. The measurement of serum IFN-α and/or IFN-α mRNA might be used as markers of HIV disease progression and/or of therapy failure [##REF##1906289##42##]. However, the biochemical and biological properties of the cytokine, including its limited distribution to blood circulation [##REF##16632604##17##,##REF##9848789##44##], present a serious limitation for its use in a clinical setting. On the contrary, the evaluation of MxA expression by real-time PCR might constitute a sensitive and reliable assay for measuring the extent of IFN-α biological activity. Since MxA mRNA is induced in response to IFN-α, the expression of MxA is strictly related to the amount of the cytokine [##REF##9562412##36##]. Indeed, we have demonstrated that MxA mRNA expression increased above normal levels in ART-treated HIV-infected patients who had low numbers of both pDC and CD4⁺cells, suggesting that analysis of MxA induction could serve as a convenient marker for the evaluation of extent of pDC immune-restoration. In particular, MxA mRNA was distinctively increased in HIV-infected children with marked depletion of circulating pDC at levels that were proportional to viral load. This indicates that the failure to reach viral suppression in children with low CD4⁺and pDC leads to high increased levels of IFN-α. According to this hypothesis, we have observed that serum IFN-α concentration was significantly higher in HIV-infected children who have increased MxA mRNA values as compared to patients with MxA expression below the cut-off.</p>", "<p>We have also observed that perinatally HIV-infected patients with low pDC number displayed the highest levels of MxA mRNA production. We speculate that MxA induction might be influenced by the interaction of IFN receptor with its other ligand, IFN-β, by an augmented IFN-α synthetic activity of fewer circulating pDC, or by the increased production of IFN-α by sources other than circulating pDC, such as human monocytes [##REF##18342575##46##] or tissue-resident pDC. Indeed, since HIV infection up-regulates the pDC chemokine receptor CCR7 [##REF##15113904##47##], these cells might accumulate in T-cell rich lymphoid tissues, such as tonsils or lymph nodes. Supporting this Dillon <italic>et al</italic>. [##REF##18300699##39##] demonstrated an accumulation of partially activated pDC in lymphoid tissues. We did not observe a reduction of circulating pDC, but the immunohistochemistry results appear to argue against the hypothesis that enhanced tissue pDC migration is the main factor explaining the inverse correlation with MxA in perinatally HIV-infected patients. These issues will need further research, considering that Biancotto <italic>et al</italic>. [##REF##17289812##48##] have demonstrated a reduction of pDC in nodes of chronically infected adults. Further, the number of pDC in lymphatic tissues of HIV⁺individuals might depend on the stage of the disease and on the extent of pDC activation and susceptibility to HIV-dependent cell death [##REF##18300699##39##].</p>", "<p>High levels of MxA have been reported in peripheral blood cells of patients with acute viral infections, including influenza, herpes, cytomegalovirus, rotavirus, adenovirus and RSV [##REF##9562412##36##,##REF##7551417##49##,##REF##10950979##50##], but not in patients with HCV infection, therefore, this virus is considered a poor inducer of IFN-α [##REF##10950979##50##,##REF##8138257##51##]. Thus, MxA mRNA levels above the cut-off found in 3 out of 4 of our patients with HIV-HCV co-infection were probably correlated to the outcome of HIV infection.</p>" ]

[ "<title>Conclusion</title>", "<p>Although the interpretation of an elevated MxA level should be made on the basis of the complete clinical and laboratory data, our results indicate that the analysis of MxA may represent a valuable tool for the management of ART in perinatal HIV infection.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>To determine the role of interferon-alpha in controlling HIV infection we phenotypically and functionally analyzed circulating plasmacytoid dendritic cells (pDC), which are known to be the highest interferon-alpha producing cells, in 33 perinatally infected HIV⁺patients undergoing standard antiretroviral therapy.</p>", "<title>Methods</title>", "<p>Circulating pDC were identified by flow cytometry using anti-BDCA-2 monoclonal antibody and by measuring BDCA-2 mRNA by real-time PCR, while tissue-resident pDC were identified by immunohistochemistry. mRNA for interferon-alpha and MxA, a gene that is specifically induced by interferon-alpha, was quantified in peripheral blood cells by real-time PCR, while serum interferon-alpha protein was measured by ELISA.</p>", "<title>Results</title>", "<p>While median values of pDC, both in terms of percentage and absolute number, were not statistically different from age-matched controls, interferon-alpha mRNA was increased in HIV-infected patients. However, in a group of patients with long disease duration, having a low number of both pDC and CD4⁺lymphocytes and a significant increase of serum interferon-alpha, MxA mRNA was produced at high level and its expression directly correlated with HIV RNA copy numbers. Furthermore in patients displaying a low CD4⁺blood cell count, a severe depletion of pDC in the tonsils could be documented.</p>", "<title>Conclusion</title>", "<p>HIV replication unresponsive to antiretroviral treatment in perinatal-infected patients with advanced disease and pDC depletion may lead to interferon-alpha expression and subsequent induction of MxA mRNA. Thus, the latter measurement may represent a valuable marker to monitor the clinical response to therapy in HIV patients.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>" ]

[ "<title>Acknowledgements</title>", "<p><italic>Financial Support</italic>. This work was supported by grants from \"VI Programma Nazionale di Ricerca sull'AIDS 2006\" (Istituto Superiore di Sanità) to LI, PRIN 2007ACZMMZ_005, EU Grant FP7 HLH-cure (project n. 201461) and Fondazione Cariplo NOBEL Grant to RB, and PRIN 2006063183-2006 to FF. FS and CG are supported by \"Sovvenzione Globale INGENIO\" of Fondo Sociale Europeo, Ministero del Lavoro e della Previdenza Sociale and Regione Lombardia.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>(A) Correlation of pDC with age and (B) with CD4⁺cell number in ART-treated HIV⁺patients. (C)</bold> Percentage and number of pDC in ART-treated HIV⁺children and in age-matched controls (CTRL). <bold>(D)</bold> Values of IFN-α mRNA, obtained by real-time PCR and expressed as normalization ratio (NR), in HIV⁺children and CTRL.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>(A) MxA mRNA levels in ART-treated HIV⁺children and controls (CTRL).</bold> Correlation between MxA mRNA levels and HIV viral load <bold>(B)</bold> and pDC number <bold>(C)</bold>. NR: normalization ratio.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Immunological parameters of 2 groups of ART-treated HIV⁺patients divided according the production of MxA mRNA in MxA-non induced (cut-off value NR &lt; 2.7) and MxA-induced (NR &gt; 2.7) patients. (A)</bold> Number of pDC, <bold>(B)</bold> level of serum IFN-α, <bold>(C)</bold> number of CD4⁺lymphocytes, and <bold>(D)</bold> age of the patients. NR: normalization ratio.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p><bold>Analysis of tissue pDC.</bold> pDC counting was performed using an Olympus BX60 microscope equipped with the Olympus DP-70 digital camera and the Olympus Soft Imaging System Cell-F 2.5. Sections were stained with anti-CD123 by immunoperoxidase technique and counterstained with Mayer's hematoxylin. Magnifications: ×100 (left panels), ×500 (right panels). Two medium-power fields (corresponding to 0.3 mm²), selected on the basis of highest content of CD123+ positive cells, were evaluated and values were expressed as mean per mm². Panels A, B and C respectively show tonsil sections from one representative healthy child, an HIV-infected child with abundant pDC and a HIV-infected child showing pDC depletion. In panel D are the number of pDC (CD123⁺cells) of tonsil specimens plotted against the CD4⁺cell percentage or the HIV RNA copy number of 8 HIV⁺children.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Clinical and immunological features of HIV-infected children</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\"><bold>Patient ID</bold></td><td align=\"center\"><bold>Age (years)</bold></td><td align=\"center\"><bold>Sex</bold></td><td align=\"center\"><bold>CD4⁺(%)</bold></td><td align=\"center\"><bold>HIV RNA (copies/ml)</bold></td><td align=\"center\"><bold>CDC</bold></td></tr></thead><tbody><tr><td align=\"center\">1</td><td align=\"center\">16</td><td align=\"center\">F</td><td align=\"center\">18.6</td><td align=\"center\">18527</td><td align=\"center\">B 3</td></tr><tr><td align=\"center\">2</td><td align=\"center\">14</td><td align=\"center\">M</td><td align=\"center\">18.8</td><td align=\"center\">12461</td><td align=\"center\">B 2</td></tr><tr><td align=\"center\">3</td><td align=\"center\">19</td><td align=\"center\">F</td><td align=\"center\">20</td><td align=\"center\">50</td><td align=\"center\">B 3</td></tr><tr><td align=\"center\">4</td><td align=\"center\">19</td><td align=\"center\">F</td><td align=\"center\">20</td><td align=\"center\">4506</td><td align=\"center\">B 2</td></tr><tr><td align=\"center\">5</td><td align=\"center\">17</td><td align=\"center\">F</td><td align=\"center\">21.3</td><td align=\"center\">78</td><td align=\"center\">C2</td></tr><tr><td align=\"center\">6</td><td align=\"center\">8</td><td align=\"center\">M</td><td align=\"center\">21.9</td><td align=\"center\">4366</td><td align=\"center\">B 2</td></tr><tr><td align=\"center\">7</td><td align=\"center\">18</td><td align=\"center\">M</td><td align=\"center\">23.9</td><td align=\"center\">65</td><td align=\"center\">A 3</td></tr><tr><td align=\"center\">8</td><td align=\"center\">16</td><td align=\"center\">M</td><td align=\"center\">25.2</td><td align=\"center\">50</td><td align=\"center\">B 3</td></tr><tr><td align=\"center\">9</td><td align=\"center\">4</td><td align=\"center\">F</td><td align=\"center\">25.6</td><td align=\"center\">166</td><td align=\"center\">B 1</td></tr><tr><td align=\"center\">10</td><td align=\"center\">16</td><td align=\"center\">F</td><td align=\"center\">28.9</td><td align=\"center\">50</td><td align=\"center\">B 3</td></tr><tr><td align=\"center\">11</td><td align=\"center\">16</td><td align=\"center\">M</td><td align=\"center\">29.1</td><td align=\"center\">14627</td><td align=\"center\">A 2</td></tr><tr><td align=\"center\">12</td><td align=\"center\">14</td><td align=\"center\">F</td><td align=\"center\">30</td><td align=\"center\">50</td><td align=\"center\">B 3</td></tr><tr><td align=\"center\">13</td><td align=\"center\">11</td><td align=\"center\">F</td><td align=\"center\">30.2</td><td align=\"center\">50</td><td align=\"center\">C 2</td></tr><tr><td align=\"center\">14</td><td align=\"center\">8</td><td align=\"center\">M</td><td align=\"center\">30.5</td><td align=\"center\">50</td><td align=\"center\">A 2</td></tr><tr><td align=\"center\">15</td><td align=\"center\">8</td><td align=\"center\">M</td><td align=\"center\">30.7</td><td align=\"center\">50</td><td align=\"center\">A 2</td></tr><tr><td align=\"center\">16</td><td align=\"center\">4</td><td align=\"center\">M</td><td align=\"center\">30.8</td><td align=\"center\">2214</td><td align=\"center\">C 2</td></tr><tr><td align=\"center\">17</td><td align=\"center\">12</td><td align=\"center\">M</td><td align=\"center\">31.6</td><td align=\"center\">50</td><td align=\"center\">C 3</td></tr><tr><td align=\"center\">18</td><td align=\"center\">17</td><td align=\"center\">M</td><td align=\"center\">31.9</td><td align=\"center\">50</td><td align=\"center\">B 3</td></tr><tr><td align=\"center\">19^a</td><td align=\"center\">12</td><td align=\"center\">F</td><td align=\"center\">32</td><td align=\"center\">50</td><td align=\"center\">C 2</td></tr><tr><td align=\"center\">20</td><td align=\"center\">11</td><td align=\"center\">F</td><td align=\"center\">32.6</td><td align=\"center\">50</td><td align=\"center\">A 2</td></tr><tr><td align=\"center\">21</td><td align=\"center\">2</td><td align=\"center\">M</td><td align=\"center\">33</td><td align=\"center\">2154</td><td align=\"center\">A1</td></tr><tr><td align=\"center\">22</td><td align=\"center\">13</td><td align=\"center\">F</td><td align=\"center\">33.3</td><td align=\"center\">50</td><td align=\"center\">A 2</td></tr><tr><td align=\"center\">23</td><td align=\"center\">4</td><td align=\"center\">M</td><td align=\"center\">33.5</td><td align=\"center\">6984</td><td align=\"center\">N 1</td></tr><tr><td align=\"center\">24</td><td align=\"center\">6</td><td align=\"center\">M</td><td align=\"center\">33.9</td><td align=\"center\">50</td><td align=\"center\">B 2</td></tr><tr><td align=\"center\">25^a</td><td align=\"center\">16</td><td align=\"center\">F</td><td align=\"center\">34.5</td><td align=\"center\">50</td><td align=\"center\">A 2</td></tr><tr><td align=\"center\">26</td><td align=\"center\">10</td><td align=\"center\">M</td><td align=\"center\">34.6</td><td align=\"center\">77</td><td align=\"center\">A 2</td></tr><tr><td align=\"center\">27</td><td align=\"center\">11</td><td align=\"center\">F</td><td align=\"center\">38.2</td><td align=\"center\">516</td><td align=\"center\">B 1</td></tr><tr><td align=\"center\">28</td><td align=\"center\">11</td><td align=\"center\">M</td><td align=\"center\">39.3</td><td align=\"center\">50</td><td align=\"center\">C 1</td></tr><tr><td align=\"center\">29</td><td align=\"center\">9</td><td align=\"center\">F</td><td align=\"center\">40.9</td><td align=\"center\">50</td><td align=\"center\">B 1</td></tr><tr><td align=\"center\">30</td><td align=\"center\">9</td><td align=\"center\">F</td><td align=\"center\">41</td><td align=\"center\">5674</td><td align=\"center\">B 1</td></tr><tr><td align=\"center\">31</td><td align=\"center\">14</td><td align=\"center\">M</td><td align=\"center\">42.2</td><td align=\"center\">50</td><td align=\"center\">C 3</td></tr><tr><td align=\"center\">32^a</td><td align=\"center\">9</td><td align=\"center\">F</td><td align=\"center\">43.4</td><td align=\"center\">50</td><td align=\"center\">B 2</td></tr><tr><td align=\"center\">33^a</td><td align=\"center\">15</td><td align=\"center\">F</td><td align=\"center\">43.4</td><td align=\"center\">21422</td><td align=\"center\">B 2</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>^aChildren with HCV co-infection</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1479-5876-6-49-1\"/>", "<graphic xlink:href=\"1479-5876-6-49-2\"/>", "<graphic xlink:href=\"1479-5876-6-49-3\"/>", "<graphic xlink:href=\"1479-5876-6-49-4\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Overlapping genes were found in all cellular domains of life, as well as in viruses [##REF##15680581##1##, ####REF##17900620##2##, ##REF##17877818##3####17877818##3##]. Overlapping genes are thought to have unique evolutionary constraints [##REF##10937248##4##,##REF##692657##5##] and regulatory properties [##REF##9570947##6##,##REF##6198955##7##]. Genes can overlap on the same strand (→ →) or on the complementary strand (\"tail-to-tail\" → ←, or \"head-to-head\" ← →, Figure ##FIG##0##1##). Different nomenclatures have been used in the literature to denote \"same-strand\" (\"unidirectional,\" \"codirected,\" \"parallel,\" and \"tandem\"), \"tail-to-tail\" (\"convergent,\" \"anti-parallel,\" and \"end-on\"), and \"head-to-head\" (\"divergent\" and \"head-on\") overlapping genes [##REF##17479344##8##, ####REF##14659892##9##, ##REF##10101192##10##, ##REF##15520290##11####15520290##11##]. Here, we use the self-explanatory terms \"same-strand\" and \"opposite-strand\" overlapping genes.</p>", "<p>In bacteria, overlaps on the same strand are by far the most abundant [##REF##10101192##10##,##REF##15520290##11##], most likely because, on average, 70% of the genes in bacterial genomes, are located on one strand [##REF##14659892##9##]. Same-strand overlaps occur in frameshifts of one nucleotide (phase 1) or two nucleotides (phase 2). Overlaps in the same frame (phase 0) are rare [##REF##15520290##11##], and since the reading frame is unaffected, they may be thought of as genes with alternative initiation or termination sites rather than overlapping genes. Phase-0 overlaps are not dealt with here. Several studies have shown that there are significant differences between the frequencies of phase-1 and phase-2 overlapping genes [##REF##17877818##3##,##REF##17479344##8##,##REF##15520290##11##]. Overlapping-gene pairs, in which the overlap sequence is of length one to five bases (short overlaps), are abundant in phase 2, but rare in phase 1. This difference is dictated by the sequence of termination codons of the upstream gene [##REF##17479344##8##]. Since none of the stop codons (TGA, TAG, and TAA) ends in AT, GT, or TT (needed to create the initiation codons ATG, GTG or TTG in phase-1 two-nucleotide overlap) or start with G (needed to create an initiation codon in phase-1 five-nucleotide overlap), short phase-1 overlaps can only use alternative initiation codons. In contrast, as far as long overlaps (seven nucleotides or longer) are concerned, phase-1 overlapping gene pairs are more frequent than those of phase 2 [##REF##17479344##8##,##REF##15520290##11##]. Cock and Whitworth [##REF##17479344##8##] suggested that the phase bias in long overlaps is due to some unspecified selective advantage of phase-1 over phase-2 overlapping genes. They also hypothesized that since the bias was found to be universal and independent of gene function, it might be a property of gene location. Krakauer [##REF##10937248##4##] introduced a model in which the frequencies of overlapping genes in different phases are determined by their level of interdependence with respect to selective constraints. That model assumes an adaptive advantage for overlapping genes in evolvable phases [##REF##10937248##4##]. For example, in phase-1 opposite-strand overlaps, in which the second codon position of one gene corresponds to the third codon position of the second gene (and vice versa), the freedom of each gene to evolve independently is maximized [##REF##10937248##4##] (Figure ##FIG##0##1##). Indeed, Rogozin et al. [##REF##12047938##12##] found that among opposite-strand overlaps in bacteria, the least constrained overlap phase (phase 1) was the most abundant. Kingsford et al. [##REF##17642473##13##] explained this phase distribution in opposite-strand overlapping genes by the frequency of reverse-complementary stop codons in coding sequences. For same-strand overlaps, phase-1 and phase-2 overlaps have equal selective constraints and are predicted by this model, to occur in equal frequencies [##REF##10937248##4##].</p>", "<p>Previous studies [##REF##14659892##9##,##REF##15520290##11##] have found that the number of overlapping genes in bacterial genomes is positively correlated with the number of genes, implying that gene overlap may be mainly the result of accidental or random \"trespassing\" of one gene into another. There can be two scenarios for the creation of same-strand overlapping genes from pre-existing neighboring genes: (1) a mutation in the termination codon of the upstream gene, resulting in an extension of the gene downstream to the first in-frame termination codon and (2) a mutation in the initiation codon of the downstream gene, resulting in an extension of the gene upstream to the first in-frame functional initiation codon [##REF##14659892##9##]. As in point mutations, where the effect of nonsynonymous mutation is expected to be stronger than that of synonymous ones, the impact of mutations that cause extension is expected to vary according to the length of the extension. Since most mutations are deleterious, long extensions of genes are expected to be under stronger purifying selection than short ones [##REF##17642473##13##] and the frequency of initiation and termination codons in a certain phase is an upper-limit constraint to the possible number of overlapping genes in that phase.</p>", "<p>Here, we tested the influence of initiation- and termination-codon frequencies as well as genomic GC-content on the number of overlapping genes in the two phases.</p>" ]

[ "<title>Methods</title>", "<p>Data of overlapping genes from 167 bacterial genomes that employ the universal genetic code were acquired from the BPhyOG overlapping-genes database [##REF##17650344##14##]. Same-strand overlapping genes in each genome were classified according to phase and the length of the intersecting segment. We defined overlap frequency as the number of same-strand overlapping genes divided by the number of same-strand neighboring gene pairs (i.e., adjacent genes, which are located on the same strand and in between them there are no genes on the opposite strand, Figure ##FIG##1##2##) in the genome. In our analysis, we explicitly ignored recombination and therefore we used the number of same-strand neighboring gene pairs, rather than the number of genes, because a neighboring gene pair located on opposite strands cannot become overlapping on the same strand as a result of point mutation. Short overlaps (two and five bases in phase 1 and one and four bases in phase 2) were dealt separately from long overlaps of seven bases or longer.</p>", "<p>The coding sequences of the studied genomes were downloaded from NCBI. Codon and amino-acid frequencies, as well as initiation and termination codon frequencies in phase 1 and phase 2, were calculated from the coding sequences of each genome. We denote the frequency of a codon or a group of codons with a superscript for the codon's phase and a subscript for the codon. For example, denotes the frequency of ATG in phase 1 and denotes the frequencies of codons in phase 0 that end in AT, where N denotes any of the four nucleotides. The expected frequencies of each start and stop codons are calculated as the products of the frequencies of the codons that combine them, i.e., and for ATG in phase 1 and phase 2, respectively. If the codons frequencies in phase 1 and phase 2 are primarily determined by the frequencies of the codons in phase 0 that combine them, the expected frequencies would match the observed frequencies.</p>" ]

[ "<title>Results</title>", "<p>We identified 71,210 same-strand overlapping gene pairs (Table ##TAB##0##1##). Short overlaps (of length two or five bases) are rare in phase 1. In our sample, we found only 18 phase-1 short overlaps (0.08%, Table ##TAB##0##1##). In contrast, the majority of phase-2 overlaps are of length one or four bases (20% and 65%, respectively).</p>", "<p>The frequency of long phase-1 overlaps exceeds that of long phase-2 overlaps by a factor of almost 3 (Table ##TAB##0##1##, Figure ##FIG##2##3##, two-sample paired Student t-test, p &lt; 0.001). The frequency of long phase-1 overlaps is negatively correlated with genomic GC content (Figure ##FIG##2##3##, <italic>r </italic>= -0.39, <italic>p </italic>&lt; 0.001). In contrast, the correlation between the frequency of long phase-2 overlaps and GC content is not significant (<italic>p </italic>= 0.4). The frequencies of start and stop codons in phase 1 and phase 2 in the coding regions of the genomes are presented in Figure ##FIG##3##4##. Pooling together phase 1 and phase 2, the frequency of stop codons (average of 13.16%) is significantly higher than that of start codons (average of 9.36%, two-sample paired Student t-test, <italic>p </italic>&lt; 0.001). We found that the frequency of start codons in phase 1 is significantly higher than that in phase 2 by a factor of 5.2 on average (Figure ##FIG##3##4a##, two-sample paired Student t-test, <italic>p </italic>&lt; 0.001). There is no significant difference between the frequencies of stop codons in the two phases (Figure ##FIG##3##4b##, two-sample paired Student t-test, <italic>p </italic>= 0.13). These results suggest that the difference between the number of long overlaps in phase 1 and phase 2 is primarily influenced by the frequencies of start codons in the two reading frames.</p>", "<p>The difference in start codon frequencies between phase 1 and phase 2 can be explained by the codons in phase 0 that may potentially lend a dinucleotide to a start codon (ATG, GTG, and TTG) in each of the phases. In phase 2, all start codons consist of phase-0 TGN codons, which may lend TG to form a phase-2 start codon. One of these codons, TGA, is a stop codon that cannot be a part of long overlap. The remaining three codons (TGT, TGC, TGG) encode for two amino acids (cysteine and tryptophan), which are among the rarest in protein-coding genes, with a mean frequency of ~1% (Table ##TAB##1##2##). In contrast, in phase 1, the amino acids coded by NAT, NGT, and NTT codons that may lend a dinucleotide to one of the start codons (ATG, GTG, and TTG, respectively), are found in moderate to high frequencies in proteins (Table ##TAB##1##2##). Interestingly, the abundance of NAT-, NGT-, and NTT-encoded amino acids is inversely correlated with the frequency of start codons (Table ##TAB##1##2##). Moreover, amino acids encoded by NAT codons which can form the most common start codon, ATG, appear in lower frequencies than amino acids encoded by NGT- and NTT-encoded amino acids. For all bacteria and for all GC contents the frequencies of amino acids coded by TGN codons are lower than each of the amino acid groups encoded by NAT, NGT, and NTT (Figure ##FIG##4##5##, all pairwise two-sample paired Student t-tests, <italic>p </italic>&lt; 0.001).</p>", "<p>Thus, consideration of the number of amino acids and their frequencies alone will lead us to expect start codons to occur much more frequently in phase 1 than in phase 2. However, the difference in amino acids usage does not provide a very good fit to the observed frequencies. This can be achieved by a more detailed compositional argument, one that is based on codon frequencies. Such a model will accommodate differences in GC content and codon usage among the bacteria under study. We found that the frequencies of the codons that combine to form start and stop codons (e.g., and for ATG), are strongly correlated with the frequencies of start and stop codons in both phases, as well as with genomic GC content (Table ##TAB##2##3##).</p>", "<p>To control for potential annotation errors, we used a subset of overlapping genes that were not annotated as \"hypothetical,\" \"putative\" or \"pseudogene\" in the NCBI genome data. This subset of overlapping genes, which we assume to be more accurately annotated, contains 31,767 gene pairs (45% of the complete set). As in the complete set, the frequency of long phase-1 overlaps exceeds the frequency of long phase-2 overlaps by a factor of 3.1 and the frequency of long phase-1 overlaps is negatively correlated with genomic GC content (<italic>r </italic>= -0.28, <italic>p </italic>&lt; 0.001), whereas the frequency of long phase-2 overlaps is not (<italic>p </italic>= 0.6). Therefore, the influence of misannotation seems not to be significant.</p>" ]

[ "<title>Discussion</title>", "<p>Understanding the distribution of overlapping genes in different phases is a key step towards distinguishing between the effects of selection and mutation on the evolution of overlapping genes. Krakauer [##REF##10937248##4##] showed that overlapping genes in different orientations and phases differ in the freedom for each gene to evolve independently. Therefore, he suggested that the variation in selective constraints would be reflected in the frequency of the overlap phases. In the case of same-strand overlapping genes, his model predicted no difference between the frequency of phase-1 and phase-2 overlaps [##REF##10937248##4##]. However, in agreement with previous studies [##REF##17877818##3##,##REF##17479344##8##,##REF##15520290##11##], our results indicate a preponderance of long phase-1 overlaps over long phase-2 overlaps. Cock and Whitworth [##REF##17479344##8##] attributed the difference between the number of long overlaps in the two phases to either gene location or to an unspecified selective advantage. These hypotheses cannot be quantifiably tested.</p>", "<p>Considering the two scenarios for the creation of same-strand overlapping genes, we showed that the phase bias in long overlaps might be attributed to a great extant to overlaps created by 5'-end mutation of the downstream gene. Since there is purifying selection against long overlaps, the frequency of start codons in phase 2 constrains the number of overlap that can be created in that phase and leads to the phase bias. In addition, we showed that the difference in start codon frequencies between phase 1 and phase 2 is dictated by the frequencies of amino acids whose codons may combine to form start codons in the two phases. Finally, the dependency of frame-shift start and stop codons on species-specific codon usage result in a correlation between long phase-1 overlap frequency and genomic GC content.</p>", "<p>Although our model explains the phase bias in overlap frequency, we do not have a full explanation for the absence of correlation between GC content and long phase-2 overlaps as expected from the frequency of frame-shift start and stop codons. This correlation is expected to have lower statistical significance than that of phase-1 overlaps because of the smaller sample size, but it is also possible that other factors affect the potential for overlap as well. A more complex compositional model for overlapping genes frequency, might include the length distribution of overlaps, the frequencies of regulatory elements (e.g., Shine-Delgarno sequences) and the strand-specific composition bias, since bacterial genomes have an asymmetrical chirochoric base composition [##REF##10570985##15##, ####REF##8676740##16##, ##REF##8614839##17####8614839##17##].</p>", "<p>The wide abundance of overlapping genes and the straightforward definition of phase evolvability make the phase distribution of overlapping genes an interesting case study. If evolvability is selected for, the expectation is for a positive correlation to exist between the frequency of an overlap phase and its evolvability. Evolvability considerations predict phase-1 and phase-2 overlaps to occur at equal frequencies [##REF##10937248##4##]. Therefore, our data does not support a role for evolvability in the evolution of same-strand overlapping genes.</p>", "<p>Fukuda et al. [##REF##14659892##9##] examined homologous overlapping genes in related bacterial species and found that the rate of accumulation and degradation of overlapping pairs is higher for overlaps caused by mutation at the 3'-end of the upstream gene compared to overlaps caused by mutation at the 5'-end of the downstream gene. The difference in rates was suggested to be a result of an evolutionary constraint imposed on the 5'-end of genes [##REF##14659892##9##]. Our model predicts a difference in these rates simply because of the higher frequency of frame-shift stop codons compared to the frequency of frame-shift start codons. It would be interesting to test whether the rate difference of accumulation and degradation of overlapping gene pairs in the two scenarios holds even when accounting for the difference in frequency of frame-shift stop codons compared to frame-shift start codons.</p>", "<p>The high frequency of frame-shift stop codons was previously suggested to be under positive selection for minimization of frame-shift translation errors [##REF##17293451##18##,##REF##15585128##19##]. We found that the frequency of frame-shift stop codons is strongly correlated with genomic GC content leading to AT-rich genomes having five times more frame-shift stop codons than GC-rich genomes. Therefore, it seems that the mutation pattern is a major player in determining frame-shift stop-codon frequencies, while selection does not seem to play a major role.</p>", "<p>Viral genomes also exhibit high frequencies of overlapping genes. In a study of RNA viruses, Belshaw et al. [##REF##17785537##20##] distinguished between internal overlaps, in which one gene is embedded within the other, and terminal overlaps. For internal overlaps, it was found that, similar to bacteria, there is a predominance of phase-1 overlaps [##REF##17785537##20##]. In the case of terminal overlaps, Belshaw et al. [##REF##17785537##20##] reported no frequency difference between phase 1 and phase 2. However, Belshaw et al. [##REF##17785537##20##] did not distinguish between short overlaps, in which phase-1 overlaps are extremely rare, and long overlaps. We showed that at least as far as bacteria are concerned, pooling short and long overlaps together results in obscuring the pattern for long overlaps (Table ##TAB##0##1##). Therefore, the similar frequencies of overall overlaps in phase 1 and phase 2 in RNA viruses [##REF##17785537##20##], suggests that the phase bias in long overlaps was most likely unnoticed.</p>" ]

[ "<title>Conclusion</title>", "<p>1. The phase-distribution of same-strand overlapping genes in bacteria is determined by the frame-shift frequencies of start and stop codons in protein-coding genes.</p>", "<p>2. The predominance of long phase-1 overlaps results from a lower frequency of start codons in phase 2 that limits the potential overlaps created by an upstream extension of the downstream gene.</p>", "<p>3. The difference in the frequency of start codons is dictated by the abundance of those amino acids that are encoded by codons that combine to form start codons in phase 1 and phase 2. This difference is conserved among all the bacterial genomes in the study.</p>", "<p>4. The variability of codon usage across bacterial genomes leads to a correlation between long phase-1 overlaps and genomic GC content.</p>", "<p>5. Our model explains the phase bias in same-strand overlapping genes by compositional factors without invoking selection. Therefore, it can be used as a null model of neutral evolution for testing selection hypotheses affecting the evolution of overlapping genes.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Same-strand overlapping genes may occur in frameshifts of one (phase 1) or two nucleotides (phase 2). In previous studies of bacterial genomes, long phase-1 overlaps were found to be more numerous than long phase-2 overlaps. This bias was explained by either genomic location or an unspecified selection advantage. Models that focused on the ability of the two genes to evolve independently did not predict this phase bias. Here, we propose that a purely compositional model explains the phase bias in a more parsimonious manner. Same-strand overlapping genes may arise through either a mutation at the termination codon of the upstream gene or a mutation at the initiation codon of the downstream gene. We hypothesized that given these two scenarios, the frequencies of initiation and termination codons in the two phases may determine the number for overlapping genes.</p>", "<title>Results</title>", "<p>We examined the frequencies of initiation- and termination-codons in the two phases, and found that termination codons do not significantly differ between the two phases, whereas initiation codons are more abundant in phase 1. We found that the primary factors explaining the phase inequality are the frequencies of amino acids whose codons may combine to form start codons in the two phases. We show that the frequencies of start codons in each of the two phases, and, hence, the potential for the creation of overlapping genes, are determined by a universal amino-acid frequency and species-specific codon usage, leading to a correlation between long phase-1 overlaps and genomic GC content.</p>", "<title>Conclusion</title>", "<p>Our model explains the phase bias in same-strand overlapping genes by compositional factors without invoking selection. Therefore, it can be used as a null model of neutral evolution to test selection hypotheses concerning the evolution of overlapping genes.</p>", "<title>Reviewers</title>", "<p>This article was reviewed by Bill Martin, Itai Yanai, and Mikhail Gelfand.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>NS performed the analyses and wrote the draft manuscript. DG and GL contributed to the interpretation of the results and the final version.</p>", "<title>Reviewers' comments</title>", "<title>Reviewer's report 1</title>", "<p>Review by Bill Martin, University of Dusseldorf.</p>", "<p>This is an interesting and straightforward paper showing that the main patterns shown by overlapping genes can be simply explained with constraints posed by base compositional factors and the nature of the genetic code. I had not thought much about overlapping genes and the conundrum that they entail, and I suspect that many other readers have not either, so the present paper was a very worthwhile read for me and I suspect that others will see it similarly.</p>", "<title>Reviewer's report 2</title>", "<p>Review by Itai Yanai, Department of Biology, Technion – Israel Institute of Technology</p>", "<p>1) In this paper, Sabath et al. propose a convincingly simple explanation for a known genomic bias without recourse to positive selection. This is a significant achievement and a sobering one too given that it offers a minimal mechanism to a process where only complicated explanations were previously available. The coding regions of neighboring same-strand genes sometimes overlap, and for this overlap to consist of a different open reading frame a frame-shift of one (phase 1) or two (phase 2) base-pairs may be introduced. While it might be expected that both phases occur equally frequently, Sabath et al. confirm, using a large set of 167 genomes, the previously reported observation that long overlaps (=7 bp), phase 1's are favored 3 to 1 to phase 2's. This trend has been previously attributed to an unknown selective advantage or genomic location; however the authors here provide evidence for the preference of phase 1 codons from a simple base-pair compositional perspective.</p>", "<p>2) The results can be essentially seen here as two themes: 1. Sabath et al. show that when examining coding regions, the codons in phase 1 contain more start codons than in the phase 2; and 2. that this trend holds across 167 genomes, although an impressive dependency with GC content is also revealed. For the former, the authors make the argument that the formation of a start codon in phase 2 is less probably since it requires rare phase 0 codons. This is a simple and brilliant explanation that appears well supported by the data. It is an explanation which does not require special selective biases and I fully support the authors claim that this is a neutral model which ought to be considered the null-hypothesis for the formation of overlapping genes.</p>", "<p>3) As noted by the authors however, there seems to be another layer to this puzzle that remains unsolved. Throughout, Sabath et al. demonstrate the correlations across an axis of GC content, where genomes with a high GC content contains less fraction overlapping genes, of start codons in phase 1, and of stop codons in both phase 1, and 2. These strong correlation are the elephant in the room, especially contrasted with the lack of correlation of GC content with phase-2 long overlaps. It would be interesting to test whether frequent phase 0 codons lead to more popular codons in phase 1 than in phase 2. Since a gene with less frequent codons may also have low expression, purifying selection would tend to select against overlaps with unpopular codons. This analysis would generalize Sabath et al.'s analysis of the start/stop codons to the entire genetic code.</p>", "<title>Author's response</title>", "<p><italic>The lack of correlation between phase-2 long overlaps and genomic GC content is, indeed, unresolved. When trying to resolve this issue, one has to keep in mind that the observed negative correlation between start and stop codons and GC content is a result of these codons being AT rich in sequence. However, this overall negative correlation contains particular positive correlations between GC content and some phase-0 codons that combine to yield start or stop codons in phase 1 and phase 2. For example, there is a negative correlation between GC content and the start codons ATG, GTG, and TTG in phase 2, whereas, the correlation between GC content and phase-0 TGC and TGG codons that may combine to yield a start codon in phase 2 is positive (data not shown). Unfortunately, this issue cannot be simply resolved by focusing on overlaps with one start codon at a time, since the factors governing start codon usage are not well understood for either overlapping or non-overlapping genes. The suggestion that the frequencies of overlap phases are influenced by codon-bias in phase-0 codons is important and should be studied in the future. In fact, as noted in the discussion, it would be important to consider other compositional factors (such as the length distribution of overlaps, the frequencies of regulatory elements, and the strand-specific composition bias) as well</italic>.</p>", "<p>4) On a final note, I do not agree with the authors statements on the evolvability of overlap in the Discussion section. Sabath et al. write: \"If evolvability is selected for, the expectation is for a positive correlation to exist between the frequency of an overlap phase and its evolvability. Evolvability considerations predict phase-1 and phase-2 overlaps to occur at equal frequencies [##REF##10937248##4##]. Therefore, our data does not support a role for evolvability in the evolution of same-strand overlapping genes.\" It is not clear what exactly is meant by evolvability in this context, and why an equal frequency among the phases would support this. I would have expected the authors to conclude here that evolvability is an inappropriate issue when discussing overlapping genes since the evidence provided here point to a predominantly neutral process.</p>", "<title>Author's response</title>", "<p><italic>For better or worse, the topic of evolvability of biological entities has been a subject of great interest in the recent years [reviewed in </italic>[##REF##18059367##21##]]. <italic>However, the quantification of evolvability has been a difficult task. Overlapping genes are unique in that their evolvability can be quantified objectively. A biological system is evolvable if it can acquire novel functions through genetic change. In the case of overlapping genes, evolvability was defined as the degree in which each of the overlapping genes can evolve independently </italic>[##REF##10937248##4##], <italic>i.e., the proportion of changes that are nonsynonymous in one gene and synonymous in the overlapping gene. Given that the three codon-positions are different in the proportion of changes that are synonymous (~5%, 0%, and ~70% for first, second, and third codon positions, respectively), phase evolvability depends on the combination of the three codon-positions in the two overlapping genes. For example, opposite-strand phase 2 is the least evolvable phase with the third codon position in one gene corresponds to the third codon position of the second gene (and vice versa), leading to maximization of the proportion of changes that are nonsynonymous in both genes </italic>(Figure ##FIG##0##1##). <italic>In contrast, opposite-strand phase 1 is the most evolvable phase, with the second codon position of one gene corresponds to the third codon position of the second gene (and vice versa) which maximizes the proportion of changes that are nonsynonymous in one gene and synonymous in the overlapping genes. Krakauer </italic>[##REF##10937248##4##]<italic>suggested that overlapping genes in evolvable phases have an adaptive advantage over overlapping genes in less evolvable phases, since they allow for higher degrees of independent evolution. Therefore, he predicted a positive correlation between the frequency of an overlap phase and its evolvability </italic>[##REF##10937248##4##]. <italic>Indeed, Rogozin et al</italic>. [##REF##12047938##12##]<italic>found that among opposite-strand overlaps in bacteria, the least constrained overlap phase (phase 1) was the most abundant. This result was later questioned by Kingsford et al</italic>. [##REF##17642473##13##]<italic>who used a similar approach to ours. In the case of same-strand overlaps, phase-1 and phase-2 overlaps have equal degree of evolvability </italic>[##REF##10937248##4##]. <italic>The reason is that from the point of view of one gene, there is an equivalence of overlap phase. For example, if gene A overlaps gene B on the same strand in phase 1, than gene B overlaps gene A in phase 2. Therefore, phase-1 and phase-2 overlaps are predicted by this model to occur at equal frequencies. Since our data shows unequal frequencies of phase 1 and phase 2, evolvability does not seem to play an important role in the evolution of same-strand overlapping genes</italic>.</p>", "<title>Reviewer's report 3</title>", "<p>Review by Mikhail Gelfand, Department of Bioinformatics, Institute of Information Transfer Problems</p>", "<p>1) It is common knowledge that in many cases it is much more difficult to prove a negative result than a positive one. Thus, the authors have set themselves a hard problem: to show that the frequencies of gene pairs overlapping in different frames can be explained by simple consideration of amino acid frequencies and codon usage and do not require more complicated evolutionary explanation.</p>", "<title>Author's response</title>", "<p><italic>Any scientific explanation should make as few assumptions as possible. We provided an explanation for the phase bias in same-strand overlapping genes that is simpler than previous ones and does not invoke selection for phase of overlap. A more complicated model will only be required if it can explain significantly more of the variation in the observed data than our simple model. In this case, the more complex model (i.e., that overlap phase frequency is determined by selective constraints) fails to explain the data and can, therefore, be discarded</italic>.</p>", "<p>2) While the point is well taken and the approach clearly interesting, there still seem to be some technical issues that have not been addressed. Of course, the main problem plaguing all large-scale genome analysis projects is reliance on existing annotations: one may find himself studying idiosyncrasies of annotation software rather than biologically relevant features. For some analyses the authors exclude genes annotated as hypothetical, but this does not guarantee that gene starts have been predicted correctly.</p>", "<title>Author's response</title>", "<p><italic>Annotation errors are a major concern in any computational analysis. Our approach of using a subset of genes for which there is higher confidence in the annotation is common in the literature in general, as well as in studies that deal with overlapping genes (e.g</italic>., [##REF##15520290##11##].).</p>", "<p>3) At that, it is noteworthy that all non-trivial observations have been made for 5'-extensions, but not 3'-extension: it is fairly easy to mispredict the start codon (some annotation projects routinely consider the most distal codon to serve as the start), but not the stop codon.</p>", "<p>I do not see an easy way out of this difficulty. One possible control is to consider separately overlaps caused by 5'-most start codons for the downstream gene (open reading frame) and internal start codons.</p>", "<title>Author's response</title>", "<p><italic>Our observations on 5'-extensions are not based on the annotation of start codons, but on the observed frequency of start codons in phase 1 and phase 2 of coding sequences. This difference in the frequency of start codons explains the difference in the frequency of long same-strand overlapping genes</italic>.</p>", "<p>4) Another approach is much more time-consuming, but it might provide interesting biological insight per se. The authors state that overlaps are caused by mutations in either start or stop codons. For the stop codons this should be not very difficult to trace to these mutations to specific branches of the evolutionary tree. Then the entire analysis might be repeated for the overlaps where the causing mutation is known. It is likely that it would seriously decrease the sample size, but it would also make the sample much more reliable. In particular, one might consider separately established overlaps persisting for some time and very recent overlaps caused by species-specific mutations (or, for that matter, sequencing errors).</p>", "<p>With start codons it might be more difficult. Indeed, one has to consider separately two types of mutations. One is the loss of a pre-existing start codon, and this can be treated in a manner similar to the one when stops are considered. A useful addition would be considering separately cases where there are candidate start codons upstream (in the previous reading frame, either on the same strand, or the complementary strand) and when candidate start codons can be found within the gene whose original start codon is mutated. The second type of mutations is gain of function, that is, emergence of a new functional upstream start codon. However, in this case it would be very difficult to prove by purely computational means that the new start really functions.</p>", "<p>There are also other possibilities for a more detailed analysis. A common problem for all of them is that they require considerable effort to prove a rather simple point, and thus it is not clear whether they are worth pursuing.</p>", "<title>Author's response</title>", "<p><italic>We agree that a phylogenetic approach may be beneficial. Unfortunately, the phylogenetic topology of bacteria is unresolved, so that a phylogenetic approach may introduce a new source of error into the analyses</italic>.</p>", "<p>5) Another important problem is, however, necessary to be addressed, as it clearly lies in the framework of the suggested approach. The point is, for a new upstream codon to be functional, it needs to occur in the same open reading frame as the old one, that is, there should be no stop codons in the region between the new and old starts. Since the frequency of candidate stop codons is not the same in the two shifted reading phases of the upstream gene, this might influence the general conclusions made in the paper. It looks like the authors have something like that in mind when they write about \"stronger purifying selection\" in long extensions, but this point is never quantified, and the applied term looks somewhat misleading and inviting further criticisms: if there is stronger purifying selection, one should observe decrease in the substitution rate in the longer-overlap regions compared to shorter-overlap ones – is this the case?</p>", "<title>Author's response</title>", "<p><italic>Dr. Gelfand wrote: \"Since the frequency of candidate stop codons is not the same in the two shifted reading phases of the upstream gene, this might influence the general conclusions made in the paper.\" However, as shown in Figure </italic>##FIG##3##4##, <italic>there is no significant difference between the frequencies of stop codons in the two phases, while the frequency of start codons in phase 1 is significantly higher than that of phase 2. Regarding the stronger purifying selection in long extensions, we have clearly failed to convey the idea. All we meant was to convey the common-sense assumption that in molecular evolution \"big changes\" are selected against more frequently and more stringently than \"small changes.\" The strength of the negative selection is expected to be positively correlated with the length of the extension following the obliteration of a stop codon</italic>.</p>", "<p>6) Background, second paragraph: \"Overlaps in the same frame are rare\": that depends on how one quantifies it; gene fusions do not seem to be very rare in bacterial genomes, especially conserved with long overlaps.</p>", "<title>Author's response</title>", "<p><italic>In our dataset, there are 187 phase-0 same-strand overlaps (0.26%). One reason for the paucity may be that in phase-0 same-strand overlaps, stop codons should be unstable or subjected to readthrough. Another reason may be the one raised by Dr. Gelfand, i.e., the ease with which gene fusion occurs in bacteria</italic>.</p>", "<p>7) Results, first paragraph: It might be interesting to learn more about 18 non-standard start codons yielding short phase-1 overlaps. Are they functional? Are they conserved? Are they regulatory?</p>", "<title>Author's response</title>", "<p><italic>True. However, these might be also a result of annotation or sequencing errors</italic>.</p>", "<p>8) Discussion, fifth paragraph: One of the reasons for relative scarcity of 3'-extensions might be that many bacterial genes contain tandem stop codons. This has been ascribed to avoidance of translational readthrough, but an evolutionary consequence is that mutation in one stop codon from a tandem pair does not create overlapping genes.</p>", "<title>Author's response</title>", "<p><italic>There is no relative scarcity of 3'-extensions. In fact, the rate of accumulation and degradation of overlapping pairs is higher for overlaps caused by mutation at the 3'-end of the upstream gene compared to overlaps caused by mutation at the 5'-end of the downstream gene </italic>[##REF##14659892##9##].</p>", "<p>9) Discussion, sixth paragraph: Correlation between the GC-content and the frequency of stop codons in frames 1 and 2 does not prove the absence of selection for such stop codons: one needs to demonstrate that the number of observed stops coincides with the number of expected ones, while controlling for dependencies between adjacent codons.</p>", "<title>Author's response</title>", "<p><italic>Given that AT-rich genomes have, on average, five times more frame-shift stop codons than GC-rich genomes, we believe that the impact of selection on frame-shift stop codon frequency should be small compared to the impact of the mutation pattern that affects composition</italic>.</p>" ]

[ "<title>Acknowledgements</title>", "<p>This article is dedicated to Prof. Lev Fishelson on his 85th birthday. We thank Dr. Yingqin Luo for providing the data from the BPhyOG. This work was supported in part by grant DBI-0543342 from the National Science Foundation. We thank the three reviewers for their constructive comments.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Orientations and phases of gene overlap. Genes can overlap on the same strand and on the opposite strand. The reference gene in a pair of overlapping genes is called phase 0. Same-strand overlaps can be in two phases (1 and 2); opposite-strand overlaps can be in three phases (0, 1, and 2). First and second codon positions, in which ~5% and 0% of the changes are synonymous, are marked in red. Third codon positions, in which ~70% of the changes are synonymous, are marked in blue.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>Same-strand neighboring gene pairs (marked with the letter N) are defined as two adjacent genes that are located on the same strand and in between them there are no genes on the opposite strand.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p>Frequency of overlapping genes in 167 bacterial genomes plotted against genomic GC content. Long phase-1 overlaps are marked in blue. Long phase-2 overlaps are marked in red.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p>a. Start codon frequencies in phase-1 (blue) and phase-2 (red) reading frames plotted against genomic GC content. b. Stop codon frequencies in phase-1 (blue) and phase-2 (red) reading frames plotted against genomic GC content.</p></caption></fig>", "<fig position=\"float\" id=\"F5\"><label>Figure 5</label><caption><p>Mean frequencies of groups of amino acids in the 167 bacterial genomes plotted against genomic GC content. Mean frequency of amino acids, which are encoded by TGN, NAT, NGT, or NTT codons, are marked in red, blue, green, and black, respectively. NAT, NGT, and NTT codons may lend a dinucleotide to one of the start codons in phase 1. TGN codons may lend a dinucleotide to one of the start codons in phase 2.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Number of same-strand overlapping genes.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\">Short overlaps (1–5 bases)</td><td align=\"center\">Long overlaps (7 bases or more)</td><td align=\"center\">Total</td></tr></thead><tbody><tr><td align=\"left\">Phase 1</td><td align=\"center\">18</td><td align=\"center\">21,550</td><td align=\"center\">21,568</td></tr><tr><td align=\"left\">Phase 2</td><td align=\"center\">42,177</td><td align=\"center\">7,465</td><td align=\"center\">49,642</td></tr><tr><td align=\"left\">Total</td><td align=\"center\">42,195</td><td align=\"center\">29,015</td><td align=\"center\">71,210</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Codons in phase 0 that may lend a dinucleotide to form a start codon in phase 1 and phase 2. The usage of each start codon in (a) all genes; (b) the downstream gene of long phase-1 overlaps; and (c) the downstream gene of long phase-2 overlaps, is noted.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\">Start Codon (usage in: all genes, phase1, phase 2)</td><td align=\"center\">Phase</td><td align=\"center\">Codon Group</td><td align=\"center\">Amino Acids</td><td align=\"center\">Mean amino acid frequency</td></tr></thead><tbody><tr><td align=\"left\">ATG (^a77%, ^b73%, ^c64%)</td><td align=\"center\">1</td><td align=\"center\">NAT</td><td align=\"center\">Tyr, His, Asn, Asp</td><td align=\"center\">3.67%</td></tr><tr><td/><td align=\"center\">2</td><td align=\"center\">TGN</td><td align=\"center\">Cys, Trp</td><td align=\"center\">1.06%</td></tr><tr><td align=\"left\">GTG (^a14%, ^b15%, ^c23%)</td><td align=\"center\">1</td><td align=\"center\">NGT</td><td align=\"center\">Cys, Arg, Ser, Gly</td><td align=\"center\">4.87%</td></tr><tr><td/><td align=\"center\">2</td><td align=\"center\">TGN</td><td align=\"center\">Cys, Trp</td><td align=\"center\">1.06%</td></tr><tr><td align=\"left\">TTG (^a9%, ^b12%, ^c14%)</td><td align=\"center\">1</td><td align=\"center\">NTT</td><td align=\"center\">Phe, Leu, Ile, Val</td><td align=\"center\">7.12%</td></tr><tr><td/><td align=\"center\">2</td><td align=\"center\">TGN</td><td align=\"center\">Cys, Trp</td><td align=\"center\">1.06%</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>The correlation between the frequency of frame-shift start and stop codons and (a) their expected frequencies; and (b) the genomic GC content. All correlations are significant at the <italic>p </italic>&lt; 0.001 level (sample size is 167).</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\" colspan=\"2\">Frame-Shift Codon</td><td align=\"center\">Phase</td><td align=\"center\">Combining Codons</td><td align=\"center\">^aCorrelation Observed-Expected</td><td align=\"center\">^bCorrelation Observed-GC%</td></tr></thead><tbody><tr><td align=\"center\">Start</td><td align=\"center\">ATG</td><td align=\"center\">1</td><td align=\"center\">N<bold>AT,G</bold>NN</td><td align=\"center\">0.96</td><td align=\"center\">-0.84</td></tr><tr><td/><td/><td align=\"center\">2</td><td align=\"center\">NN<bold>A,TG</bold>N</td><td align=\"center\">0.89</td><td align=\"center\">-0.76</td></tr><tr><td/><td align=\"center\">GTG</td><td align=\"center\">1</td><td align=\"center\">N<bold>GT,G</bold>NN</td><td align=\"center\">0.94</td><td align=\"center\">-0.34</td></tr><tr><td/><td/><td align=\"center\">2</td><td align=\"center\">NN<bold>G,TG</bold>N</td><td align=\"center\">0.86</td><td align=\"center\">0.80</td></tr><tr><td/><td align=\"center\">TTG</td><td align=\"center\">1</td><td align=\"center\">N<bold>TT,G</bold>NN</td><td align=\"center\">0.96</td><td align=\"center\">-0.80</td></tr><tr><td/><td/><td align=\"center\">2</td><td align=\"center\">NN<bold>T,TG</bold>N</td><td align=\"center\">0.87</td><td align=\"center\">-0.70</td></tr><tr><td colspan=\"6\"><hr/></td></tr><tr><td align=\"center\">Stop</td><td align=\"center\">TAA</td><td align=\"center\">1</td><td align=\"center\">N<bold>TA,A</bold>NN</td><td align=\"center\">0.98</td><td align=\"center\">-0.87</td></tr><tr><td/><td/><td align=\"center\">2</td><td align=\"center\">NN<bold>T,AA</bold>N</td><td align=\"center\">0.97</td><td align=\"center\">-0.93</td></tr><tr><td/><td align=\"center\">TAG</td><td align=\"center\">1</td><td align=\"center\">N<bold>TA,G</bold>NN</td><td align=\"center\">0.96</td><td align=\"center\">-0.89</td></tr><tr><td/><td/><td align=\"center\">2</td><td align=\"center\">NN<bold>T,AG</bold>N</td><td align=\"center\">0.90</td><td align=\"center\">-0.84</td></tr><tr><td/><td align=\"center\">TGA</td><td align=\"center\">1</td><td align=\"center\">N<bold>TG,A</bold>NN</td><td align=\"center\">0.86</td><td align=\"center\">0.51</td></tr><tr><td/><td/><td align=\"center\">2</td><td align=\"center\">NN<bold>T,GA</bold>N</td><td align=\"center\">0.92</td><td align=\"center\">-0.84</td></tr></tbody></table></table-wrap>" ]

[ "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M1\" name=\"1745-6150-3-36-i1\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msubsup><mml:mi>f</mml:mi><mml:mrow><mml:mi>A</mml:mi><mml:mi>T</mml:mi><mml:mi>G</mml:mi></mml:mrow><mml:mn>1</mml:mn></mml:msubsup></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M2\" name=\"1745-6150-3-36-i2\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msubsup><mml:mi>f</mml:mi><mml:mrow><mml:mi>N</mml:mi><mml:mi>A</mml:mi><mml:mi>T</mml:mi></mml:mrow><mml:mn>0</mml:mn></mml:msubsup></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M3\" name=\"1745-6150-3-36-i3\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msubsup><mml:mi>f</mml:mi><mml:mrow><mml:mi>N</mml:mi><mml:mi>A</mml:mi><mml:mi>T</mml:mi></mml:mrow><mml:mn>0</mml:mn></mml:msubsup><mml:mo>×</mml:mo><mml:msubsup><mml:mi>f</mml:mi><mml:mrow><mml:mi>G</mml:mi><mml:mi>N</mml:mi><mml:mi>N</mml:mi></mml:mrow><mml:mn>0</mml:mn></mml:msubsup></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M4\" name=\"1745-6150-3-36-i4\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msubsup><mml:mi>f</mml:mi><mml:mrow><mml:mi>N</mml:mi><mml:mi>N</mml:mi><mml:mi>A</mml:mi></mml:mrow><mml:mn>0</mml:mn></mml:msubsup><mml:mo>×</mml:mo><mml:msubsup><mml:mi>f</mml:mi><mml:mrow><mml:mi>T</mml:mi><mml:mi>G</mml:mi><mml:mi>N</mml:mi></mml:mrow><mml:mn>0</mml:mn></mml:msubsup></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M5\" name=\"1745-6150-3-36-i3\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msubsup><mml:mi>f</mml:mi><mml:mrow><mml:mi>N</mml:mi><mml:mi>A</mml:mi><mml:mi>T</mml:mi></mml:mrow><mml:mn>0</mml:mn></mml:msubsup><mml:mo>×</mml:mo><mml:msubsup><mml:mi>f</mml:mi><mml:mrow><mml:mi>G</mml:mi><mml:mi>N</mml:mi><mml:mi>N</mml:mi></mml:mrow><mml:mn>0</mml:mn></mml:msubsup></mml:mrow></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M6\" name=\"1745-6150-3-36-i4\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msubsup><mml:mi>f</mml:mi><mml:mrow><mml:mi>N</mml:mi><mml:mi>N</mml:mi><mml:mi>A</mml:mi></mml:mrow><mml:mn>0</mml:mn></mml:msubsup><mml:mo>×</mml:mo><mml:msubsup><mml:mi>f</mml:mi><mml:mrow><mml:mi>T</mml:mi><mml:mi>G</mml:mi><mml:mi>N</mml:mi></mml:mrow><mml:mn>0</mml:mn></mml:msubsup></mml:mrow></mml:semantics></mml:math></inline-formula>" ]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>We used 167 bacterial genomes from Luo et al. [##REF##17650344##14##]. Nine genomes GenBank:<ext-link ext-link-type=\"gen\" xlink:href=\"NC_000908\">NC_000908</ext-link>, GenBank:<ext-link ext-link-type=\"gen\" xlink:href=\"NC_000912\">NC_000912</ext-link>, GenBank:<ext-link ext-link-type=\"gen\" xlink:href=\"NC_002162\">NC_002162</ext-link>, GenBank:<ext-link ext-link-type=\"gen\" xlink:href=\"NC_002771\">NC_002771</ext-link>, GenBank:<ext-link ext-link-type=\"gen\" xlink:href=\"NC_004432\">NC_004432</ext-link>, GenBank:<ext-link ext-link-type=\"gen\" xlink:href=\"NC_004829\">NC_004829</ext-link>, GenBank:<ext-link ext-link-type=\"gen\" xlink:href=\"NC_005364\">NC_005364</ext-link>, GenBank:<ext-link ext-link-type=\"gen\" xlink:href=\"NC_006055\">NC_006055</ext-link>, and GenBank:<ext-link ext-link-type=\"gen\" xlink:href=\"NC_006908\">NC_006908</ext-link> that do not employ the universal genetic code were excluded.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1745-6150-3-36-1\"/>", "<graphic xlink:href=\"1745-6150-3-36-2\"/>", "<graphic xlink:href=\"1745-6150-3-36-3\"/>", "<graphic xlink:href=\"1745-6150-3-36-4\"/>", "<graphic xlink:href=\"1745-6150-3-36-5\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Fatigue is a common symptom of multiple sclerosis [##REF##9007405##1##, ####REF##8180914##2##, ##REF##6703889##3##, ##REF##8929171##4##, ##REF##12051465##5####12051465##5##], reported by around three-quarters of affected patients [##REF##6703889##3##], and considered one of the most distressing symptoms of disease by over half [##REF##8180914##2##]. Many patients experience debilitating fatigue every day [##REF##8180914##2##]. In multiple sclerosis, fatigue has a major detrimental impact on quality of life [##REF##11724451##6##, ####REF##12409184##7##, ##REF##15180801##8####15180801##8##], is frequently associated with depression [##REF##12474995##9##,##REF##15794395##10##] and is a leading cause of absence from work or impaired work performance [##REF##11724451##6##,##UREF##0##11##,##REF##16193900##12##]. The pathophysiology of fatigue in multiple sclerosis is poorly understood, but is likely to be multifactorial [##REF##11355149##13##, ####REF##12665396##14##, ##REF##10101582##15##, ##REF##17881389##16####17881389##16##]</p>", "<p>Treatment of fatigue in multiple sclerosis is thus a major challenge, which cannot be adequately achieved at the present time. Both non-pharmacological and pharmacological interventions have been proposed for the management of fatigue in multiple sclerosis patients [##REF##10101582##15##,##REF##15200345##17##], although the benefits of drugs such as modafenil and amantadine have not been demonstrated unequivocally [##REF##17253480##18##, ####REF##11796766##19##, ##REF##15824337##20####15824337##20##].</p>", "<p>Immunomodulatory treatments for relapsing-remitting multiple sclerosis, namely glatiramer acetate and the β-interferons, provide a marked reduction in relapse rates and in MRI markers of disease activity [##REF##12521562##21##]. It is therefore of interest to explore whether such treatments might influence fatigue symptoms as well. A retrospective chart review of 218 Canadian patients receiving an immunomodulatory treatment during the late 1990s revealed that fatigue improved over the six months following treatment initiation [##REF##15201369##22##]. Of particular interest was the observation that a significantly higher proportion of glatiramer acetate treated patients than β-interferon-treated patients improved by at least one standard deviation of the Fatigue Impact Scale (FIS).</p>", "<p>In order to investigate further the potential impact of immunomodulatory treatment on fatigue in multiple sclerosis, we initiated a prospective, observational, non-interventional study to monitor fatigue in treatment-naive RRMS patients initiating therapy with glatiramer acetate under conditions of daily practice. The primary objective of study was to determine the impact of initiating treatment with glatiramer acetate on fatigue and absenteeism. Secondary objectives were to evaluate the effect of treatment on clinical and MRI outcomes and to determine the tolerability of treatment.</p>" ]

[ "<title>Methods</title>", "<p>This study was a prospective, observational, non-interventional study of patients with relapsing remitting multiple sclerosis treated with glatiramer acetate conducted in Germany. 130 ambulatory and hospital neurologists participated in the study. The study was performed between November 2002 and October 2004.</p>", "<title>Patients</title>", "<p>The study included patients with a diagnosis of relapsing-remitting multiple sclerosis by the McDonald criteria [##REF##11456302##23##] who had not previously been treated with an immunomodulatory treatment and in whom the investigator had decided to initiate therapy with glatiramer acetate. Patients were followed for twelve months following treatment initiation.</p>", "<title>Clinical assessment</title>", "<p>Patients were evaluated at inclusion and after 3, 6, 9 and 12 months of treatment. At each visit, patients underwent a full neurological assessment, any relapses occurring since the previous visit were ascertained and disability assessed with the Expanded Disability Status Scale (EDSS) [##REF##6685237##24##]. Fatigue was assessed by the patient using a visual analogue scale scored from 0 (no fatigue) to 10 (maximum possible fatigue) and with the Modified Fatigue Impact Scale (MFIS) [##UREF##1##25##] in its validated German translation. This is a 21-item questionnaire which yields a total score ranging from 0 (no impact of fatigue) to 84 points (maximum impact of fatigue), as well as three subscales representing the physical (score range 0 to 36), cognitive (score range 0 to 40) and psychosocial (score range 0 to 8) dimensions of fatigue.</p>", "<p>Patients were questioned about any time spent off work due to their multiple sclerosis. Due to the study protocol, the reasons for work absentism (relapse, fatigue) could not be differentiated. Any adverse events occurring since the previous visit were recorded.</p>", "<title>Statistical analysis</title>", "<p>Number of work days lost and fatigue scores over the course of the study were evaluated with the Wilcoxon rank test. All comparisons were two-tailed and a <italic>p </italic>value of &lt; 0.05 was taken as being statistically significant.</p>", "<title>Ethics</title>", "<p>This study was conducted according to the Declaration of Helsinki (Hong Kong Amendment) and pertinent national legal and regulatory requirements. Each patient provided written, informed consent and was free to withdraw from the study at any time for any reason without consequences on the care provided.</p>" ]

[ "<title>Results</title>", "<title>Study sample</title>", "<p>A total of 338 patients were included in the study. Of these, 53 were excluded from the analysis due to a protocol violation (24 patients treated previously with an immunomodulatory therapy and 29 patients for whom certain data were recorded retrospectively) and 47 failed to provide complete questionnaire data. The study population thus consisted of 291 subjects (86.1% of included patients).</p>", "<p>The baseline demographic and disease variables of the study subjects are presented in Table ##TAB##0##1##. At inclusion, their median age was 36.9 years and 74.9% were female. The median time since diagnosis was 4.31 years. In the year preceding inclusion, patients had experienced a mean of 1.71 relapses (retrospectively assessed) and their mean EDSS score at inclusion was 2.58. Forty patients (13.7%) discontinued treatment during the course of the study, principally due to the occurrence of an adverse event (sixteen patients).</p>", "<title>Clinical outcome</title>", "<p>Clinical outcome at the end of the study are presented in Table ##TAB##1##2##. Information on relapses was missing for 24 patients. Of the remaining 267 patients, 61 (22.8%) experienced a single relapse during the twelve-month study period and 23 patients (8.6%) more than one relapse. The mean annual relapse rate during the year of treatment with glatiramer acetate was 0.46. The mean EDSS score at the end of the study was 2.45, representing a mean decrease from baseline of 0.55 points. The change in EDSS score between baseline and twelve months was statistically significant (<italic>p </italic>&lt; 0.05; Wilcoxon rank test). A sustained reduction in EDSS score of &gt; 1 point was observed in fifteen patients (5.2%) and a sustained increase of &gt; 1 point in three patients (1%).</p>", "<title>Fatigue</title>", "<p>Overall, 220 patients provided exploitable data from the MFIS questionnaire at both inclusion and study end. Measures were compared between the three-month period before inclusion and the last three months of the treatment period. Significant decreases were observed in the total score as well as in all three dimension scores of the MFIS (Table ##TAB##2##3##). Similarly, the VAS rating of fatigue was reduced by around one quarter following initiation of treatment with glatiramer acetate (Table ##TAB##2##3##), between baseline and study end</p>", "<title>Work absenteeism</title>", "<p>The number of days missed from work due to multiple sclerosis was evaluated in the patients who were in employment (72.9% of the study population). In the three month period preceding inclusion, 138 patients (65.1%) had taken at least one day off work (Tables ##TAB##3##4## and ##TAB##4##5##). This number decreased to 64 patients (30.1%) in the year following initiation of treatment with glatiramer acetate. The number of days lost was significantly lower in the second year (<italic>p </italic>≤ 0.001; Wilcoxon rank test).</p>", "<title>Safety</title>", "<p>Safety was assessed in all 338 included patients. Overall, 51 patients (15.1%) experienced at least one adverse event during the treatment period. These were most frequently injection site reactions or symptoms of a systemic immediate post-injection reaction such as dyspnoea or tachycardia. No single event was reported in more than ten patients. The immediate post-injection reaction was classified as serious in one patient.</p>" ]

[ "<title>Discussion</title>", "<p>In this study, immunomodulatory treatment of relapsing-remitting multiple sclerosis with glatiramer acetate was associated with a reduction in subjective perceptions of fatigue and with the numbers of days taken off work due to illness. We observed a reduction of approximately one-quarter in both MFIS scores and in a VAS measure of fatigue. These findings are consistent with an earlier retrospective study, which also reported an improvement in fatigue measured with the FIS following initiation of glatiramer acetate treatment in 24.8% of patients [##REF##15201369##22##]. The two studies cannot, however, be directly compared due differences in methodology.</p>", "<p>The amelioration observed following treatment with glatiramer acetate may be a non-specific consequence of improved overall disease status in treated patients or alternatively result from a specific action of the medication on the pathophysiology of multiple sclerosis fatigue. For example, it has been suggested that fatigue may be aggravated by the production of high levels of pro-inflammatory cytokines [##REF##16361589##26##,##REF##15124762##27##]. The ability of glatiramer acetate to attenuate the secretion and activity of these cytokines within the central nervous system [##REF##12390966##28##,##REF##17349813##29##] may provide such a specific mechanism. Others have proposed, on the basis of magnetic resonance spectroscopy (MRS) findings, that fatigue may be associated with axonal injury in the cortex rather than inflammatory white matter lesions <italic>per se </italic>[##REF##14967766##30##]. In a recent trial, Tedeschi et al. could demonstrate that among MS patients with low disability those with high-fatigue show higher white and gray matter atrophy and higher lesion load. They suggest that in MS, independent of disability, white and gray matter atrophy is a risk factor to have fatigue [##REF##17673234##31##]. In additon, a recent trial of Rocca et al. using functional imaging in MS patients with fatigue and interferon beta-1a treatment pointed out that an abnormal recruitment of the fronto-thalamic circuitry is associated with interferon-induced fatigue in MS patients [##REF##16933299##32##]. In contrast to the interferon's, the specific action of glatiramer acetate to improve MRS markers of axonal injury in multiple sclerosis might contribute to a reduction in fatigue [##REF##17548572##33##,##UREF##2##34##].</p>", "<p>We also observed a dramatic reduction of over fifty percent in the number of patients who needed to take time off work due to their multiple sclerosis. This is consistent with findings from an American study, which also reported a marked decrease in days off work in patients treated with glatiramer acetate [##REF##16971761##35##], but less so with beta-interferons. This is an important functional effect of treatment since the ability to hold down a job satisfactorily is critical for self-esteem and because, in certain countries such as the USA, remaining in full-time employment is an important determinant of obtaining insurance for reimbursement of treatment costs.</p>", "<p>Again, the impact of glatiramer acetate on time off work may be an indirect consequence of reduced relapse frequency, although the data from the US study showing a differential effect on time off work between glatiramer acetate and β-interferons would argue against this. Alternatively, the observed effect may be secondary to a reduction in fatigue, which has been identified in other studies to be a major reason why patients with multiple sclerosis need to take time off work [##UREF##0##11##,##REF##16193900##12##]. Finally, it should be noted that the low incidence of debilitating side-effects reported with glatiramer acetate [##REF##11735654##36##] means that patients are unlikely to need to take time off work due to treatment side-effects.</p>", "<p>The strength of this study include the naturalistic design, which means that the findings can probably be generalised to standard care, at least in Europe, with confidence, the prospective nature of the data collection and the relatively large numbers of patients evaluated. Limitations include the absence of a comparator group against which the magnitude of the observed treatment effects could be assessed, and data collection during physician consultations rather than with patients' diaries, which may have introduced some degree of anamnestic error into the findings. The absence of a control group might overestimate the improvement in fatigue symptoms. As a placebo group is probably not ethical it will be further of interest to compare prospectively the benefit on fatigue in a group of naive MS patients treated with GA vs. a group treated with IFN-beta in a next study.</p>" ]

[ "<title>Conclusion</title>", "<p>In conclusion, this non-interventional prospective study demonstrated that treatment with glatiramer acetate was associated with a reduction in patient-reported fatigue ratings and in days missing from work, concomitant with an improvement in clinical manifestations of disease activity. These functional outcomes are of critical importance for overall patient well-being.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Objectives</title>", "<p>Treatment of multiple sclerosis patients with glatiramer acetate has been demonstrated a beneficial effect on disease activity. The objective of this prospective naturalistic study was to evaluate the impact of glatiramer acetate on fatigue and work absenteeism.</p>", "<title>Methods</title>", "<p>291 treatment-naïve patients with relapsing remitting multiple sclerosis were included and treated with glatiramer acetate for twelve months. Relapse rates, disability, fatigue symptoms, days of absence from work and adverse events were monitored. Fatigue was measured with the MFIS scale and with a visual analogue scale.</p>", "<title>Results</title>", "<p>Total MFIS scores decreased by 7.6 ± 16.4 from 34.6 to 27.0 (<italic>p </italic>≤ 0.001). Significant reductions were observed on all three subscales of the MFIS. Fatigue symptoms, assessed using a visual analogue scale, decreased by 1.04 ± 2.88 cm from 4.47 cm to 3.43 cm (<italic>p </italic>≤ 0.001). The proportion of patients absent from work at least once was reduced by a factor of two from 65.1% to 30.1% (<italic>p </italic>≤ 0.001). Tolerance to treatment was rated as very good or good in 78.3% of patients. Adverse effects, most frequently local injection site reactions, were reported in 15.1% of patients.</p>", "<title>Conclusion</title>", "<p>Treatment with glatiramer acetate was associated with a significant improvement in fatigue symptoms and a marked reduction in absence from work. Treatment was well-tolerated. Such benefits are of relevance to overall patient well-being.</p>" ]

[ "<title>Competing interests</title>", "<p>JH and RA are employed by TEVA Germany. TZ has received honoraria and financial compensation by Bayer Healthcare, Biogen Idec, Merck Serono, Pfizer, Sanofi-Aventis and Teva. SK has received honoraria and financial compensation by Bayer Healthcare, Biogen Idec, Sanofi-Aventis and Teva. Research Projects of TZ and SK were funded by the Roland-Ernst-Foundation, Robert-Pfleger-Foundation, Sanofi-Aventis/TEVA and Bayer Healthcare. In the MS center Dresden, clinical studies are performed for Bayer Healthcare, Biogen Idec, BioMS, Genzyme, Glaxo Smith Kline, Sanofi-Aventis and Teva.</p>", "<title>Authors' contributions</title>", "<p>RA, JH and TZ were responsible for the conception of the study. TZ drafted the article. All authors contributed to the interpretation of the results and revising the article for important intellectual content. All authors read and approved the final manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>This was an investigator-driven, only observational study supported by an unrestricted grant by TEVA Germany, purveyors of glatiramer acetate. The unrestricted grant was spent for the production of the study material, distribution, compensation of the subinvestigator, collecting the data by a clinical research associate and statistical analysis. TZ and SK received no financial compensation for their role in the study and manuscript preparation.</p>" ]

[]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Demographic and clinical characteristics of patients at inclusion</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\">Population (N = 291)</td></tr></thead><tbody><tr><td align=\"left\">Age (mean ± SD; years)</td><td align=\"center\">36.9 ± 9.3</td></tr><tr><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\">Gender</td><td/></tr><tr><td align=\"left\"> Women</td><td align=\"center\">218 (74.9%)</td></tr><tr><td align=\"left\"> Men</td><td align=\"center\">67 (23.0%)</td></tr><tr><td align=\"left\"> <italic>Missing data</italic></td><td align=\"center\"><italic>6 (2.1%)</italic></td></tr><tr><td/><td/></tr><tr><td align=\"left\">Time since diagnosis (mean ± SD; years)</td><td align=\"center\">4.31 ± 5.47</td></tr><tr><td/><td/></tr><tr><td align=\"left\">ARR since diagnosis (mean ± SD)</td><td align=\"center\">3.82 ± 3.54</td></tr><tr><td align=\"left\"> No relapses</td><td align=\"center\">14 (4.8%)</td></tr><tr><td align=\"left\"> Up to 2 relapses</td><td align=\"center\">111 (38.1%)</td></tr><tr><td align=\"left\"> 3–5 relapses</td><td align=\"center\">100 (34.4%)</td></tr><tr><td align=\"left\"> More than 5 relapses</td><td align=\"center\">57 (19.6%)</td></tr><tr><td/><td/></tr><tr><td align=\"left\">ARR within previous 12 months (mean ± SD)</td><td align=\"center\">1.71 ± 0.88</td></tr><tr><td/><td/></tr><tr><td align=\"left\">EDSS at treatment initiation (mean ± SD)</td><td align=\"center\">2.58 ± 1.44</td></tr><tr><td align=\"left\"> EDSS 0–2</td><td align=\"center\">127 (43.6%)</td></tr><tr><td align=\"left\"> EDSS 3–5</td><td align=\"center\">121 (41.6%)</td></tr><tr><td align=\"left\"> EDSS 6–7</td><td align=\"center\">16 (5.5%)</td></tr><tr><td align=\"left\"> <italic>Missing data</italic></td><td align=\"center\"><italic>27 (9.3%)</italic></td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Clinical outcome</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\">Population (N = 291)</td></tr></thead><tbody><tr><td align=\"left\">Relapses during study (<italic>n = 267</italic>)</td><td/></tr><tr><td align=\"left\"> No relapse</td><td align=\"center\">180 (67.4%)</td></tr><tr><td align=\"left\"> 1 relapse</td><td align=\"center\">61 (22.8%)</td></tr><tr><td align=\"left\"> 2 relapses</td><td align=\"center\">12 (4.5%)</td></tr><tr><td align=\"left\"> 3 relapses</td><td align=\"center\">8 (3.0%)</td></tr><tr><td align=\"left\"> 4–5 relapses</td><td align=\"center\">3 (1.1%)</td></tr><tr><td/><td/></tr><tr><td align=\"left\">Mean EDSS scores (<italic>n = 235</italic>)</td><td/></tr><tr><td align=\"left\"> Baseline</td><td align=\"center\">2.58 ± 1.45</td></tr><tr><td align=\"left\"> Study end</td><td align=\"center\">2.45 ± 1.52</td></tr><tr><td align=\"left\"> Change from baseline</td><td align=\"center\">-0.13 ± 0.73*</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Fatigue ratings.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><bold>Baseline</bold></td><td align=\"center\"><bold>On treatment</bold></td><td align=\"center\"><bold>Mean Change</bold></td><td align=\"center\"><italic>p</italic></td></tr></thead><tbody><tr><td align=\"left\">MFIS Total score (n = 220)</td><td align=\"center\">34.6 ± 18.7</td><td align=\"center\">27.0 ± 18.6</td><td align=\"center\">-7.6 ± 16.4</td><td align=\"center\"><italic>p </italic>≤ 0.001</td></tr><tr><td align=\"left\"> Physical dimension score</td><td align=\"center\">17.6 ± 9.1</td><td align=\"center\">13.5 ± 9.0</td><td align=\"center\">-4.1 ± 8.1</td><td align=\"center\"><italic>p </italic>≤ 0.001</td></tr><tr><td align=\"left\"> Cognitive dimension score</td><td align=\"center\">13.9 ± 9.2</td><td align=\"center\">11.2 ± 8.6</td><td align=\"center\">2.7 ± 8.0</td><td align=\"center\"><italic>p </italic>≤ 0.001</td></tr><tr><td align=\"left\"> Psycho-social dimension score</td><td align=\"center\">3.1 ± 2.1</td><td align=\"center\">2.4 ± 2.0</td><td align=\"center\">-0.7 ± 2.0</td><td align=\"center\"><italic>p </italic>≤ 0.001</td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td align=\"left\">VAS score (n = 198)</td><td align=\"center\">4.47 ± 2.53</td><td align=\"center\">3.43 ± 2.55</td><td align=\"center\">-1.04 ± 2.88</td><td align=\"center\"><italic>p </italic>≤ 0.001</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Number of days missing from work in the previous year at baseline and one year after start of treatment.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"2\">Baseline</td><td align=\"center\" colspan=\"2\">After 12 Months</td></tr></thead><tbody><tr><td/><td align=\"center\">N</td><td align=\"center\">%</td><td align=\"center\">N</td><td align=\"center\">%</td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td align=\"left\">No</td><td align=\"center\">76</td><td align=\"center\">26.1%</td><td align=\"center\">148</td><td align=\"center\">50.9%</td></tr><tr><td align=\"left\">≤ 5 days</td><td align=\"center\">26</td><td align=\"center\">8.9%</td><td align=\"center\">27</td><td align=\"center\">9.3%</td></tr><tr><td align=\"left\">6–10 days</td><td align=\"center\">39</td><td align=\"center\">13.4%</td><td align=\"center\">14</td><td align=\"center\">4.8%</td></tr><tr><td align=\"left\">11–20 days</td><td align=\"center\">32</td><td align=\"center\">11.0%</td><td align=\"center\">8</td><td align=\"center\">2.8%</td></tr><tr><td align=\"left\">&gt; 20 days</td><td align=\"center\">50</td><td align=\"center\">17.2%</td><td align=\"center\">18</td><td align=\"center\">6.2%</td></tr><tr><td align=\"left\">Not in employment</td><td align=\"center\">68</td><td align=\"center\">23.4%</td><td align=\"center\">62</td><td align=\"center\">21.3%</td></tr><tr><td align=\"left\">Missing information</td><td align=\"center\">0</td><td align=\"center\">0%</td><td align=\"center\">14</td><td align=\"center\">4.8%</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T5\"><label>Table 5</label><caption><p>Development of the different groups at baseline (No work absentism, less than 5 days,...) one year after start of treatment with glatiramer acetate using the same categories (No work absentism, less than 5 days,...).</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Baseline</td><td align=\"center\" colspan=\"2\">No work absentism</td><td align=\"center\" colspan=\"2\">≤ 5 days absent</td><td align=\"center\" colspan=\"2\">6–10 days absent</td><td align=\"center\" colspan=\"2\">11–20 days absent</td><td align=\"center\" colspan=\"2\">&gt; 20 days absent</td><td align=\"center\" colspan=\"2\">Not in employment</td></tr></thead><tbody><tr><td align=\"left\">After 12 Months</td><td align=\"center\">N</td><td align=\"center\">%</td><td align=\"center\">N</td><td align=\"center\">%</td><td align=\"center\">N</td><td align=\"center\">%</td><td align=\"center\">N</td><td align=\"center\">%</td><td align=\"center\">N</td><td align=\"center\">%</td><td align=\"center\">N</td><td align=\"center\">%</td></tr><tr><td colspan=\"13\"><hr/></td></tr><tr><td align=\"left\">No work absentism</td><td align=\"center\">59</td><td align=\"center\">21.3%</td><td align=\"center\">15</td><td align=\"center\">5.4%</td><td align=\"center\">20</td><td align=\"center\">7.2%</td><td align=\"center\">23</td><td align=\"center\">8.3%</td><td align=\"center\">22</td><td align=\"center\">7.9%</td><td align=\"center\">9</td><td align=\"center\">3.3%</td></tr><tr><td align=\"left\">≤ 5 days absent</td><td align=\"center\">6</td><td align=\"center\">2.2%</td><td align=\"center\">8</td><td align=\"center\">2.9%</td><td align=\"center\">8</td><td align=\"center\">2.9%</td><td align=\"center\">1</td><td align=\"center\">0.4%</td><td align=\"center\">4</td><td align=\"center\">1.4%</td><td align=\"center\">0</td><td align=\"center\">0%</td></tr><tr><td align=\"left\">6–10 days absent</td><td align=\"center\">4</td><td align=\"center\">1.4%</td><td align=\"center\">1</td><td align=\"center\">0.4%</td><td align=\"center\">3</td><td align=\"center\">1.1%</td><td align=\"center\">2</td><td align=\"center\">0.7%</td><td align=\"center\">4</td><td align=\"center\">1.4%</td><td align=\"center\">0</td><td align=\"center\">0%</td></tr><tr><td align=\"left\">11–20 days absent</td><td align=\"center\">2</td><td align=\"center\">0.7%</td><td align=\"center\">2</td><td align=\"center\">0.7%</td><td align=\"center\">1</td><td align=\"center\">0.4%</td><td align=\"center\">2</td><td align=\"center\">0.7%</td><td align=\"center\">1</td><td align=\"center\">0.4%</td><td align=\"center\">0</td><td align=\"center\">0%</td></tr><tr><td align=\"left\">&gt; 20 days absent</td><td align=\"center\">0</td><td align=\"center\">0%</td><td align=\"center\">0</td><td align=\"center\">0%</td><td align=\"center\">3</td><td align=\"center\">1.1%</td><td align=\"center\">1</td><td align=\"center\">0.4%</td><td align=\"center\">13</td><td align=\"center\">4.7%</td><td align=\"center\">1</td><td align=\"center\">0.4%</td></tr><tr><td align=\"left\">Not in employment</td><td align=\"center\">3</td><td align=\"center\">1.1%</td><td align=\"center\">0</td><td align=\"center\">0%</td><td align=\"center\">2</td><td align=\"center\">0.7%</td><td align=\"center\">1</td><td align=\"center\">0.4%</td><td align=\"center\">4</td><td align=\"center\">1.4%</td><td align=\"center\">52</td><td align=\"center\">18.8%</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>ARR: annualised relapse rate: EDSS: Expanded Disability Status Scale; SD: standard deviation.</p></table-wrap-foot>", "<table-wrap-foot><p>Data are presented as number of patients (%) for relapses and as mean ± SD for Expanded Disability Status Scale (EDSS) scores. The asterisk indicates a significant change from baseline (<italic>p </italic>&lt; 0.05; Wilcoxon signed rank test).</p></table-wrap-foot>", "<table-wrap-foot><p>Fatigue over three months was measured with the Modified Fatigue Impact Scale (MFIS) and with a visual analogue scale (VAS). Data are presented as mean ± SD for those patients providing exploitable data both at inclusion and at study end. Probabilities were calculated with the Wilcoxon rank test.</p></table-wrap-foot>" ]

[]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Various methods have been developed for predicting human pharmacokinetics, including Dedrick's approach, non-compartment analysis, and an in vitro-in vivo extrapolation (IVIVE) approach used for drug discovery. Dedrick's approach is an animal scaling-up method, which is used to extrapolate human pharmacokinetic parameters from at least 2 animal species [##REF##7120049##1##,##REF##15919852##2##]. In contrast, the IVIVE approach, which is also used to extrapolate clinical pharmacokinetic parameters, uses in vitro materials such as hepatocytes and microsomes to scale up to an actual target pharmacokinetic parameter such as organ clearance [##REF##16864504##3##,##REF##17968734##4##]. Among these options, two different models have been used for many years. The compartment model, which has a long history, is still the preferred choice because it is easy to apply. However, this approach consumes considerable resources when an animal scale-up approach is used, as many animal experiments are required for proper analysis; also, the range of application is limited [##REF##15733214##5##]. In contrast, whole body physiologically-based pharmacokinetic (WBPBPK) models for simulating human pharmacokinetics [##UREF##0##6##] enable the time-course of the tissue concentrations of various drugs to be simulated using data from only one species. A WBPBPK model can also be used for pharmacokinetic/pharmacodynamic (PK/PD) analysis at a target site. However, such models have not been commonly used because they are complex. Thus, it would be advantageous to develop a WBPBPK model based on a simple concept that is easy to implement.</p>", "<p>WBPBPK models have been much investigated. They exhibit comparatively satisfactory precision in predicting human pharmacokinetics [##REF##16640456##7##,##REF##17620347##8##]. They are generic, consisting of already well-known methods applicable to rational PK/PD simulation. However, they do not include solutions for correction when the data used as input parameters show considerable divergence (e.g. as a result of factors associated with in vitro and in vivo studies). Therefore, improvement in the precision of predictions cannot be expected from previous models. Recently, several WBPBPK models have also been analyzed using a single simplified method [##REF##14650375##9##]. Unfortunately, the more simplified versions do not account for the complexity of biological systems, as mixed models consist of individual organs as well as multiple organs considered together. Thus, it has remained difficult to apply PK/PD analysis at the level of a target organ, although this method can be useful since it is relatively simple.</p>", "<p>It remains desirable to develop a generic, simple, and more precise WBPBPK model that is useful at the preclinical stage. Although generic WBPBPK models satisfy the conditions mentioned above (i.e. they can apply to PK/PD analysis), the ones currently in use are difficult to apply to the analysis of various compounds owing to poor predictive precision and the lack of solutions for correction. However, if these faults could be rectified, the generic WBPBPK model would be a more useful method. To improve the precision of prediction, it is important to use the available experimental data more efficiently. For example, preclinical in vivo experiments on rats are essential for Investigating New Drug (IND) applications. Such data are useful for predicting human pharmacokinetics using the generic WBPBPK model, even when the findings are derived from in silico or in vivo experiments [##REF##17372687##10##]. They should ideally be used prior to the initiation of clinical trials by the pharmaceutical industry. However, it is possible that the aforementioned data are insufficient for satisfactory prediction, because a more convenient supplementary method for improving the precision of human pharmacokinetics prediction with only slight modifications is not currently available.</p>", "<p>The aim of the present study was to construct a WBPBPK model that will enable human pharmacokinetics to be predicted with high precision using only in vivo data from rat studies and in vitro data from liver microsomes or hepatocytes, and will be supplemented by straightforward mathematical methods devoid of highly complex concepts. We also used the method developed here indirectly to investigate the potential reasons why the predictions achieved to date with precursors of the method have been poor. To these ends, we used the following procedures. 1. We speculated about the possible causes of poor precision of prediction and changed part of a generic WBPBPK model accordingly. 2. We developed a novel method and deployed it to identify and ameliorate the causes of poor prediction. The utility of the new method was demonstrated by comparing the precision with which it predicted pharmacokinetic parameters to evaluate its validity. 3. We elucidated the causes of poor precision of prediction using the developed method. Because this method involves only physiology-related parameters, it can show whether any of these parameters contribute to the lack of precision in prediction. This is the first investigation aimed at improving the precision of prediction by WBPBPK models by attempting to elucidate the reasons for the lack of such precision.</p>" ]

[ "<title>Materials and methods</title>", "<title>Experimentation and data collection</title>", "<p>Fourteen drugs with various physicochemical properties were selected for this study. Tolbutamide [##REF##3938813##11##, ####REF##15771236##12##, ##REF##7175701##13####7175701##13##] and diclofenac [##REF##15771236##12##,##REF##23275##14##, ####UREF##1##15##, ##REF##12438516##16##, ##UREF##2##17##, ##UREF##3##18##, ##REF##12124306##19####12124306##19##] were used as acidic drugs. Midazolam [##REF##15771236##12##,##REF##12490579##20##, ####REF##7293228##21##, ##REF##11745776##22####11745776##22##] and diazepam [##REF##15771236##12##,##REF##4156305##23##,##REF##11560875##24##] were used as neutral drugs. Phenytoin [##REF##3938813##11##,##REF##15771236##12##,##REF##23275##14##,##REF##11977112##25##,##REF##3171873##26##], imipramine [##REF##15771236##12##,##REF##6475108##27##, ####REF##15858854##28##, ##REF##2894451##29##, ##REF##3965690##30####3965690##30##] and lidocaine [##REF##15771236##12##,##REF##12913267##31##, ####REF##5010683##32##, ##REF##3254976##33##, ##REF##6863582##34##, ##REF##2663301##35####2663301##35##] were used as basic drugs. Gatifloxacin [##UREF##4##36##], grepafloxacin [##REF##10993208##37##, ####UREF##5##38##, ##UREF##6##39####6##39##], gemifloxacin [##REF##11259328##40##,##UREF##7##41##], pazufloxacin [##UREF##5##38##,##UREF##8##42##, ####REF##7637202##43##, ##UREF##9##44##, ##UREF##10##45####10##45##], enoxacin [##UREF##5##38##,##UREF##11##46##, ####UREF##12##47##, ##UREF##13##48####13##48##], fleroxacin [##UREF##4##36##,##UREF##5##38##,##UREF##14##49##] and lomefloxacin [##UREF##15##50##,##UREF##16##51##] were used as zwitterionic drugs. Data collected from the published literature about these drugs are shown in Table ##TAB##0##1##. Kp values (steady state tissue-plasma partition coefficients) were also obtained from the literature and are described in the reference column of Table ##TAB##0##1##. Physicochemical parameters such as molecular weight (M.W.), calculated logP (clogP), topological polar surface area (tPSA) and calculated molecular reflectability (cMR) were determined using ChemOffice Ultra 9.0 (Cambridge Software, USA).</p>", "<p>All the observed human data in this study were obtained from the literature and were used as published or with the proper corrections. The total plasma clearance was corrected to the total blood clearance using the blood-plasma concentration ratio for calculations.</p>", "<title>Model development</title>", "<title>Generic WBPBPK model</title>", "<p>The simple WBPBPK model without membrane permeation was used (equations 1–7). This model incorporated veins (v), arteries (a), lung, pancreas (panc), heart, liver (h), kidney (r), small intestine (gi), brain, adipose tissue, muscle and bone, as well as a single adjusting compartment (Figure ##FIG##0##1##). The well-stirred model was used for modelling each organ and tissue type. The rat Kp values were used without correction. Organ clearance was used to describe system clearance. It was assumed that the excreting organs were the liver, kidney and small intestine. Physiological input parameters (e.g. the blood flow rate in each organ or tissue [Q_i] and the volume of the organ or tissue [V_i]) were obtained from the literature [##REF##9249929##52##].</p>", "<p>A system of three ordinary linear differential equations was proposed for liver, kidney and small intestine, which are organs with elimination processes such as metabolism and excretion of bile and urine. The following equations were used [##REF##16640456##7##]:</p>", "<p></p>", "<p></p>", "<p></p>", "<p>where C is the concentration, Q is the blood flow rate, V is the volume of tissue or organ, and Kp is the steady-state tissue-plasma partition coefficient.</p>", "<p>Another system of linear ordinary differential equations was proposed for the lung and other organs, including a single adjusting compartment, with no elimination process. The following equations were used:</p>", "<p></p>", "<p></p>", "<p>where i represents the other organ.</p>", "<p>Two linear ordinary differential equations were proposed for veins and arteries, and the following equations were used:</p>", "<p></p>", "<p></p>", "<p>Pancreas and bone were not incorporated in the 8-organ model, and the adipose tissue and muscle were omitted from the 6-organ model.</p>", "<p>The system of linear ordinary differential equations describing the WBPBPK model was solved numerically using the Runge-Kutta-Gill method [##REF##1663838##53##].</p>", "<p>A correction for intrinsic clearance in the liver was performed for acidic, neutral and basic compounds, using the in vitro intrinsic liver clearance of both rats and humans [##REF##15771236##12##]. This correction was necessary because of the large species differences in metabolism. The following equation was used for scaling up from the rat to the human model:</p>", "<p></p>", "<p>In this equation, sf represents a scaling factor, and the human:rat hepatic blood flow rate ratio was taken as 0.325.</p>", "<p>Renal and secretion clearance corrections for the blood flow were performed for scaling up from a rat model to a human model because it has been reported that blood flow rate is useful for correcting some pharmacokinetic parameters [##REF##15155551##54##, ####REF##15155552##55##, ##REF##16621936##56####16621936##56##]:</p>", "<p></p>", "<p>where CL_orgrepresents clearance in the kidney or small intestine, and Q_jrepresents the blood flow rate in these organs.</p>", "<title>Single adjusting compartment</title>", "<p>A single adjusting compartment (SAC) was incorporated into the present model as a potential function that can offset the lack of predictive precision. The SAC was incorporated as a newly-developed virtual organ possessing the same functions as other organs in place of the \"rest of the body\" (carcass) previously used in WBPBPK modelling. However, the physiological parameters of the SAC were set up so that they could be adjusted arbitrarily. It was assumed that the lack of precision in simulating human pharmacokinetics has typically been caused by certain physiological factors. Thus, to describe the SAC, its blood flow rate (Q_SAC), organ/tissue volume (V_SAC) and steady-state tissue-plasma partition coefficient (Kp_SAC) were selected as input parameters. The SAC was also described using the well-stirred model (equation 5). Simulated rat pharmacokinetics were fitted to the observed pharmacokinetics using Q_SAC, V_Sand a Kp_SAC, all of which could be changed arbitrarily. These SAC values used for fitting were fixed as data derived from rat studies.</p>", "<p>When the Q_SACof a rat was transformed to a human value, the following equation was used:</p>", "<p></p>", "<p>where Q_riis the blood flow rate in the isolated organ. P is a factor that depends on the individual model; P = 15 was used for this study. This value was fixed after optimising the 6- and 8-organ model simulations for correcting the Q_{SAC, rat}where the values were lager than the human Q_tot. This value is intrinsically different for each compound, but was assumed to be constant in order to give the model generality.</p>", "<p>The following equation was used to transform rat to human V_SAC:</p>", "<p></p>", "<p>Veins and arteries were not incorporated into the total volume for each organ or tissue in a SAC. In addition, Kp_SAC, which was used as a parameter to describe the tissue distribution of a SAC, was assumed to be the same as the value obtained from the rat. This method was used as an alternative compartment in place of the \"rest of the body\". The ability to be arbitrary is its main advantage. In contrast, the \"rest of the body\" has only a fixed parameter, which could be a major cause of poor prediction.</p>", "<title>Calculation of pharmacokinetic parameters</title>", "<p>In general, the half-life (T_1/2) and the total clearance (CL_tot) are used to compare the precision of prediction of human pharmacokinetics among models [##REF##16640456##7##, ####REF##17620347##8##, ##REF##14650375##9####14650375##9##]. Therefore, we used these parameters for this purpose. The T_1/2was calculated using equation 12, and k_el(the terminal phase rate constant) was obtained by linear regression analysis of the log-transformed concentration-time data. The total area under the blood concentration-time curve (AUC_inf) was obtained according to the following procedure. Blood AUC_0-tvalues (where t is the time of the last blood concentration collected) were estimated using Simpson's rule [##UREF##17##57##], a more reasonable method than the trapezoidal method for calculating the AUC precisely. AUC_t-infwas estimated by dividing the final blood concentration measured by the terminal-phase rate constant. AUC_infis the sum of AUC_0-tand AUC_t-inf. CL_totwas calculated according to equation 13.</p>", "<p></p>", "<p></p>", "<title>Statistical analysis</title>", "<p>The accuracy and precision of the calculated values were confirmed by considering the ratio of the observed to the predicted values. Average values were used to confirm accuracy, and the average-fold error (AFE) [##REF##11560875##24##] and the within-2-fold error were used to confirm precision. The AFE was calculated using the following equation:</p>", "<p></p>", "<p>where N represents the number of data inputs used for the calculation.</p>", "<p>In order to clarify the major cause of poor predictions by WBPBPK models, we confirmed the correlations between certain SAC input parameters and various physicochemical parameters, which were calculated on the basis of the structures of the selected compounds.</p>" ]

[ "<title>Results</title>", "<p>A generic WBPBPK model and the single adjusting compartment (SAC)-WBPBPK model were constructed with parameters that depended on each compound. The precision of predictions was confirmed for each model. The influence of the following two factors on the precision of simulation of human pharmacokinetics was investigated: the number of organs incorporated and the presence or absence of a SAC. The human blood concentration of each compound was simulated using the constructed model. The half-life (T_1/2) and total clearance (CL_tot) values were calculated from the simulated human blood concentration. Figure ##FIG##1##2a–c## shows the relationship of the observed and predicted CL_totand T_1/2values when a SAC was not incorporated and the number of organs changed. The predicted values differed widely from the observed values. No satisfactory improvement in divergence was observed in spite of the addition of organs. Figure ##FIG##2##3a–c## shows the relationship observed when a SAC was incorporated and the number of organs altered. The predicted values resembled the observed values more closely in the model incorporating a SAC than in the models lacking a SAC. The precision of the simulated values in each model was confirmed by comparing the average fold error (AFE) and the within-2-fold error. These results (Table ##TAB##1##2##) showed that the precision of predictions of human T_1/2values decreased when some organs were removed from the model, regardless of the incorporation of a SAC. In the case of CL_tot, the SAC-incorporated model yielded highly precise predictions in each of the three organ-number models, even the 6-organ model; the within 2-fold error was 92%. The AFE and within-2-fold error values were compared to those obtained from previous generic WBPBPK models and with those obtained by the conventional method for predicting human pharmacokinetics (Table ##TAB##2##3##). The predictions obtained with the SAC-WBPBPK model were more precise than those yielded by the other models.</p>", "<p>Significant correlations or non-significant trends were observed between Q_SAC, the blood flow rate of a SAC (Table ##TAB##3##4##), and four physicochemical parameters (tPSA, clogP, M.W. and cMR). The correlation coefficients between Q_SACand tPSA, clogP, M.W. and cMR were 0.78, 0.57, 0.73 and 0.52, respectively (Figure ##FIG##3##4a–d##).</p>" ]

[ "<title>Discussion</title>", "<title>Investigation of the lack of precision in simulations of human pharmacokinetics using the generic WBPBPK model</title>", "<p>This study was conducted to clarify the main cause of the poor predictions obtained with the generic WBPBPK model and to enable a model to be constructed that could address this problem easily. We initially attempted to elucidate the divergence in the precision of predictions with the number of organs investigated, i.e. in the 6-, 8- and 10-organ models. Poor precision and discrepancies may be related to one or more of the following: active versus passive transportation systems, species differences in metabolism, and physiological factors such as blood flow rate, tissue volume and the number of organs involved. Other factors could also be involved. The results of this series are shown in Figure ##FIG##1##2##: increasing the number of organs in the model improved the precision of prediction. These results indicate that failure to account for particular physiological factors may contribute to the poor predicted values from the generic WBPBPK model.</p>", "<p>On the basis of the present findings, we inferred that not only species differences in active transportation systems, metabolism, etc., but also failure to account for the physiological parameters of each individual and each species, were responsible for the poor predicted values by previous WBPBPK models. Therefore, the precision with which human pharmacokinetics were predicted was examined by adding a single adjusting compartment (SAC), a newly developed virtual organ that could be expected to improve the precision of predictions if added to the generic WBPBPK model. The results are shown in Figure ##FIG##2##3##. Fitting of the simulated to the observed rat pharmacokinetics before scaling up to the human was successful and the AFE values of T_1/2and CL_totwere lower than 1.1 for almost all compounds. These findings supported our initial assumptions, because the improvement in precision observed with the model incorporating the SAC implicated the previous failure to account for blood flow rate, tissue volume and tissue distribution.</p>", "<p>The parameters for elucidating the precision of prediction were calculated (Table ##TAB##1##2##): the AFEs of CL_totand T_1/2were greatly improved by incorporating a SAC into the 10-organ model. If the only major cause of poor predictive precision had been differences in the active transportation systems of different species, then it would not have been possible to correct for differences in predictive precision. However, inclusion of a SAC in the model corrected for the divergence resulting from active transportation systems and metabolism, provided that no species differences were involved. These findings did not contradict the assumptions made for the present series, because use of actual hepatic clearance values did not improve the precision of predictions. It is therefore reasonable to conclude that the poor predictive value of the previous methods is due to their failure to account for physiological factors.</p>", "<p>The predictions of CL_totwere less precise for tolbutamide, diclofenac, diazepam, grepafloxacin and lomefloxacin than for the other compounds tested, even when a SAC was incorporated into the 10-organ model. The divergence of prediction for the two acidic drugs is thought to have been caused by drug binding to plasma proteins, i.e. acidic drugs have a high affinity for plasma albumin, which leads to a lower contribution to tissue distribution. Consequently, most of the total pharmacokinetics of a drug can be described by a SAC and a clearance equation, together with a scaling-up equation to adjust for the results obtained from rats. However, a SAC acts only in a supporting role. The scaling-up equation also acts only in a supporting role. Therefore, the precision of prediction for the two acidic drugs tested here might have been worse than that for the other drugs. Specifically, in order to obtain precise predictions, the tissue distribution must have a large influence on the model.</p>", "<p>Diazepam, a drug for which predictions show considerable divergence in precision, is known to be a substrate of human MDR1 [##REF##11785684##58##]. Moreover, grepafloxacin is known to be a substrate of human MRP1 and rat Mrp2 [##REF##10991972##59##,##REF##9495864##60##]. However, there are no data regarding the contribution of rat Mdr1 to diazepam pharmacokinetics or of rat Mrp1 and human MRP2 in the case of grepafloxacin. In addition, the differences between observed and predicted values were smaller than those obtained when no SAC was incorporated. Previously reported findings, taken together with the results of the present study, indicate the involvement of both an active transportation system and species differences. However, these factors play only a minor role in the predictive precision of the generic WBPBPK model.</p>", "<p>Table ##TAB##2##3## compares the predictive precision of the SAC-WBPBPK model with previous methods. The best within-2-fold error for predicting human T_1/2values was achieved with the 10-organ model with a SAC, and the results were even better for CL_tot. Regardless of the AFE values associated with each of the previous methods (2 in both cases), the values for T_1/2and CL_totin the SAC-WBPBPK model showed more precise predictions; both were approximately 1.5.</p>", "<p>In summary, this series revealed that a major factor leading to the poor precision observed with the generic WBPBPK model was the failure to account for human physiological parameters. The precision of a generic WBPBPK model was improved by incorporating a SAC, which included such physiological parameters. The results also indicated that the SAC-WBPBPK model will be more useful than previous WBPBPK models for predicting human pharmacokinetics, particularly in cases when predictions are made with data obtained before the onset of clinical trials.</p>", "<title>Indirect investigation of the lack of precision of simulations of human pharmacokinetics using SAC-related parameters</title>", "<p>The input parameters for the SAC in this study were useful not only in terms of fitting the data to rat pharmacokinetics, but also for investigating factors that were missing from previous models. Initially, it was confirmed that Q_SAC, V_SACand Kp_SACeach correlated with various physicochemical parameters (Table ##TAB##3##4##). Significant correlations were confirmed between Q_SACand three physicochemical parameters (topological polar surface area (tPSA), molecular weight (M.W.), and calculated logP (clogP)) and a non-significant trend was observed between Q_SACand calculated molecular reflectability (cMR) (Figure ##FIG##3##4##). In particular, for the correlations between Q_SACand tPSA, a negative slope below the 0.1% significance criterion was observed. Generally, compounds with larger tPSA values are known to permeate the cell membrane with more difficulty. The finding of large Q_SACvalues indicated that the previous WBPBPK model does not take sufficient account of organs with high blood flow rates. On the other hand, small Q_SACvalues indicate that the previous model was unable to account for organs with low blood flow rates. The incorporation of a SAC in the model improved this issue. The negative-slope correlation between Q_SACand tPSA indicated the following: a compound with a low tPSA value (i.e. a compound that easily permeates the cell membrane and is therefore readily distributed among tissues) does not account for the factor of relative blood flow rate. Thus, high blood flow rates could affect the pharmacokinetics of such a compound because cell membrane permeation is not a major factor. Accordingly, it is reasonable to assume that the physiological factor of blood flow rate, such as blood flow-rate limitation, is related to the outcomes obtained from models. In contrast, for compounds associated with large tPSA values, membrane permeability contributes more than blood flow rate because permeability is low. The problem caused by a large Q_SAC(small tPSA) could be resolved by incorporating a membrane permeation process into the WBPBPK model. However, the problem caused by a small Q_SAC(large tPSA) cannot be resolved easily: it is difficult to choose an adequate blood flow rate for each model because of variation among individuals. This factor could be the cause of poor predictions for large Q_SACdrugs. Therefore, we should keep these points in mind when we perform a proper human pharmacokinetics simulation. In short, previous models did not sufficiently account for the relationship between physiological factors and the unique distribution that is caused by an individual compound's physicochemical properties. Moreover, adding considerations such as a permeation process and individual differences in blood flow rate for constructing a generic WBPBPK model could improve the precision of prediction.</p>", "<p>The significant correlations that we found between clogP and Q_SACare also considered reasonable, as was the case with tPSA, because when a drug is more lipophilic, its ability to permeate the cell membrane increases, resulting in a smooth distribution to certain tissues. Moreover, this factor is not related to the presence of an active transportation system. However, the simple incorporation of organs did not account for a precise system, because drug metabolism contributes more when lipophilicity increases. On the other hand, the present findings indicate that differences in active transportation systems and metabolism between species did not play a major role in the model's predictions; the improvement in predictive precision when correcting for physiological factors by incorporating a SAC played a larger role. These conclusions were supported by the correlation between Q_SACand M.W., and by the tendency of Q_SACand cMR to reflect molecular size. Q_SACand cMR showed no significant correlations. However, the bias of cMR values of selected compounds in this study could explain why no significant correlations were found. The correlation between Q_SACand cMR could be significant, provided the number of test compounds was increased. These results indicate that physiological limitations such as blood flow and membrane permeability were involved in improving the predictive precision of the WBPBPK model. Furthermore, such physiological limitations were not accounted for sufficiently in previous WBPBPK models.</p>", "<p>No significant correlations were observed between V_SACor Kp_SACand the physicochemical parameters. However, V_SACand Kp_SACtended to overestimate T_1/2as the values increased (data not shown). Moreover, the tendency toward overestimation was especially marked when the product of V_SACand Kp_SAC, which represented the degree of tissue distribution, was considered. These results indicate that the SAC was incorporated into this WBPBPK model as an organ with relatively slow drug transportation and slow drug elimination. Therefore, estimates of T_1/2tended to be longer when more of the drug is distributed to a SAC. With regard to the generic WBPBPK model without a SAC, the precision of prediction of T_1/2was relatively good. However, the prediction of CL_totshowed low precision. From these results, it is possible that the volume of distribution (Vd) value was not accurately predicted. This assumption indicates that the related factors V_SACand Kp_SACin the SAC-WBPBPK model were not present in the previous generic WBPBPK model because, fundamentally, Vd is predicted using organ volumes and the Kp value of each organ. In the present study, the Kp value was not corrected by the blood free fraction (f_B) in rat or human when the model was constructed. Therefore, the actual Kp values for humans were different from the experimental values for the rat, which were used in the present study. Moreover, inter-individual differences in organ volume are not considered in the generic WBPBPK model. Accordingly, organ volume as a physiological parameter should have been accounted for in more detail, including the inter-individual variability of the data set, as well as drug-specific parameters such as Kp values.</p>", "<p>The addition of a SAC, such as that developed for this study, to various generic WBPBPK models may enhance the precision of human pharmacokinetics simulations. This approach may also facilitate with the handling of certain species differences (e.g. intrinsic clearance) because the SAC can be used as the \"rest of the body (carcass)\", i.e. as a non-specific compartment. Furthermore, this approach did not require arbitrary alterations of the actual experimental data, which distinguishes it from methods in which the observed data must be altered to fit the animal (rat) findings. Thus, the present approach is a more rational methodology for prediction. In this regard, we will discuss the concept underlying the model presented here. Dedrick's animal scaling-up is an empirical approach. In contrast, a WBPBPK model entails a mechanistic approach. However, the generic WBPBPK model, which has been used at the preclinical stage, contains empirical factors such as Kp values, and a clearance prediction method for scaling up to the human. Moreover, if membrane permeation processes are incorporated into the model, we have to rely on empirical methods to scale up to human permeation rate constants. Nevertheless, the generic WBPBPK model is applicable for predicting human pharmacokinetics. That is because almost all parts of this system consist of actual human physiological parameters and are linked mechanistically. Therefore, the WBPBPK approach can elucidate kinetics in organs and is applicable for a variety of uses. The SAC approach is a hybrid of an empirical and a mechanistic approach. Using a SAC, we found that the primary cause of poor prediction was a failure to consider physiological systems. Therefore, a SAC approach is compatible with a mechanistic approach because it complements previous problems. On the other hand, a SAC is not just described as a physiological system. In this context, it is more empirical than the generic WBPBPK model used previously. However, despite including an empirical factor, the SAC-WBPBPK model is more rational than the previous generic WBPBPK models. Moreover, our model addresses the cause of poor prediction in previous generic models, and does not need to manipulate observed experimental values to adjust to rat pharmacokinetics.</p>", "<p>Some limitations are associated with the addition of a SAC. In this study, tolbutamide kinetics could only be simulated in a 10-organ model. If no upper or lower limits could be set as input parameters for a SAC, then the model would be unable to deal adequately with outliers. This problem has not yet been resolved, even when corrections were made using a scaling-up equation for human Q_SAC. This matter will require further study.</p>" ]

[ "<title>Conclusion</title>", "<p>Incorporation of a SAC into a generic WBPBPK model, as performed in this study, significantly improved the precision of predictions of human pharmacokinetics. For the first time, failure to account for certain physiological parameters was identified as a major problem in previous generic WBPBPK models, in addition to confounders such as species differences in terms of metabolism and the presence/absence of active transportation systems (i.e. transporters). Moreover, the SAC-WBPBPK model performed better than all previous methods in terms of precision of prediction. Moreover, this newly developed model entails a simpler and more straightforward methodology than older models. It is likely that the present model will be useful not only for predicting clinical pharmacokinetics, but also for analyzing PBPK/PD at the preclinical stage in simulations of drug efficacy.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>There are various methods for predicting human pharmacokinetics. Among these, a whole body physiologically-based pharmacokinetic (WBPBPK) model is useful because it gives a mechanistic description. However, WBPBPK models cannot predict human pharmacokinetics with enough precision. This study was conducted to elucidate the primary reason for poor predictions by WBPBPK models, and to enable better predictions to be made without reliance on complex concepts.</p>", "<title>Methods</title>", "<p>The primary reasons for poor predictions of human pharmacokinetics were investigated using a generic WBPBPK model that incorporated a single adjusting compartment (SAC), a virtual organ compartment with physiological parameters that can be adjusted arbitrarily. The blood flow rate, organ volume, and the steady state tissue-plasma partition coefficient of a SAC were calculated to fit simulated to observed pharmacokinetics in the rat. The adjusted SAC parameters were fixed and scaled up to the human using a newly developed equation. Using the scaled-up SAC parameters, human pharmacokinetics were simulated and each pharmacokinetic parameter was calculated. These simulated parameters were compared to the observed data. Simulations were performed to confirm the relationship between the precision of prediction and the number of tissue compartments, including a SAC.</p>", "<title>Results</title>", "<p>Increasing the number of tissue compartments led to an improvement of the average-fold error (AFE) of total body clearances (CL_tot) and half-lives (T_1/2) calculated from the simulated human blood concentrations of 14 drugs. The presence of a SAC also improved the AFE values of a ten-organ model from 6.74 to 1.56 in CL_tot, and from 4.74 to 1.48 in T_1/2. Moreover, the within-2-fold errors were improved in all models; incorporating a SAC gave results from 0 to 79% in CL_tot, and from 14 to 93% in T_1/2of the ten-organ model.</p>", "<title>Conclusion</title>", "<p>By using a SAC in this study, we were able to show that poor prediction resulted mainly from such physiological factors as organ blood flow rate and organ volume, which were not satisfactorily accounted for in previous WBPBPK models. The SAC also improved precision in the prediction of human pharmacokinetics. This finding showed that the methodology of our study may be useful for functionally reinforcing a WBPBPK model.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>HA conceived the idea, developed the model, performed the analysis, and drafted the manuscript. SI and WH assisted in the analysis and preparation of the manuscript. IN assisted in the analysis and preparation of the manuscript, and participated in helpful discussion. All authors read and approved the final manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>The authors wish to thank Kyorin Pharmaceutical Co., Ltd. for allowing us to perform this study. We also thank Mr. Yoshiaki Kitamura for helpful discussions. Dr. Paul S Agutter is thanked for improving and editing our manuscript.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Concept of the SAC-WBPBPK model</bold>. The compartment \"other organs\" contained brain, muscle, adipose tissue and bone. Pancreas and bone were not incorporated in the 8-organ model, and adipose tissue and muscle were omitted from the 6-organ model.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Correlation between the observed and simulated pharmacokinetic parameters predicted without a SAC</bold>. (a) Six-organ model without a SAC, (b) 8-organ model without a SAC, (c) 10-organ model without a SAC. The solid line represents unity, whereas the dashed lines represent the 2-fold prediction error.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Correlation between the observed and simulated pharmacokinetic parameters predicted with a SAC</bold>. (a) Six-organ model with a SAC, (b) 8-organ model with a SAC, (c) 10-organ model with a SAC. The solid line represents unity, whereas the dashed lines represent the 2-fold prediction error.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p><bold>Correlation of Q_SACwith physicochemical parameters</bold>. (a) tPSA, (b) clogP, (c) M.W., (d) cMR. The solid lines represent regression.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Pharmacokinetic parameters of various compounds used as inputs for each WBPBPK model simulation</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Compound</td><td align=\"center\" colspan=\"7\">Rat</td><td align=\"center\" colspan=\"4\">Human</td><td align=\"left\">References</td></tr><tr><td/><td colspan=\"7\"><hr/></td><td colspan=\"4\"><hr/></td><td/></tr><tr><td/><td align=\"right\">CL_tot<break/>(mL/h/kg)</td><td align=\"right\">CL_h<break/>(mL/h/kg)</td><td align=\"right\">CL_r<break/>(mL/h/kg)</td><td align=\"right\">CL_s<break/>(mL/h/kg)</td><td align=\"left\">T_1/2<break/>(h)</td><td align=\"left\">R_B^a</td><td align=\"left\">f_B</td><td align=\"left\">CL_tot<break/>(mL/h/kg)</td><td align=\"right\">T_1/2<break/>(h)</td><td align=\"left\">R_B^{<italic>a</italic>}</td><td align=\"left\">f_B</td><td/></tr></thead><tbody><tr><td align=\"left\">Tolbutamide</td><td align=\"right\">109</td><td align=\"right\">109</td><td align=\"right\">0</td><td align=\"right\">0</td><td align=\"left\">1.8</td><td align=\"left\">0.75</td><td align=\"left\">0.36</td><td align=\"center\">24.0</td><td align=\"right\">7.0</td><td align=\"left\">0.75</td><td align=\"left\">0.12</td><td align=\"left\">11–13</td></tr><tr><td align=\"left\">Diclofenac</td><td align=\"right\">1809</td><td align=\"right\">1176</td><td align=\"right\">633</td><td align=\"right\">0</td><td align=\"left\">0.14</td><td align=\"left\">0.55</td><td align=\"left\">0.009</td><td align=\"center\">447</td><td align=\"right\">1.2</td><td align=\"left\">0.55</td><td align=\"left\">0.009</td><td align=\"left\">12, 14–19</td></tr><tr><td align=\"left\">Midazolam</td><td align=\"right\">3024</td><td align=\"right\">1542</td><td align=\"right\">269</td><td align=\"right\">1213</td><td align=\"left\">0.53</td><td align=\"left\">1</td><td align=\"left\">0.066</td><td align=\"center\">473</td><td align=\"right\">2.8</td><td align=\"left\">0.80</td><td align=\"left\">0.043</td><td align=\"left\">12, 20–22</td></tr><tr><td align=\"left\">Diazepam</td><td align=\"right\">2492</td><td align=\"right\">2343</td><td align=\"right\">149</td><td align=\"right\">0</td><td align=\"left\">1.1</td><td align=\"left\">1.04</td><td align=\"left\">0.13</td><td align=\"center\">40.4</td><td align=\"right\">32.8</td><td align=\"left\">1.04</td><td align=\"left\">0.03</td><td align=\"left\">12, 23, 24</td></tr><tr><td align=\"left\">Phenytoin</td><td align=\"right\">1806</td><td align=\"right\">1246</td><td align=\"right\">181</td><td align=\"right\">379</td><td align=\"left\">0.37</td><td align=\"left\">0.99</td><td align=\"left\">0.23</td><td align=\"center\">187</td><td align=\"right\">13.2</td><td align=\"left\">0.61</td><td align=\"left\">0.20</td><td align=\"left\">11, 12, 14, 25, 26</td></tr><tr><td align=\"left\">Imipramine</td><td align=\"right\">2544</td><td align=\"right\">1649</td><td align=\"right\">895</td><td align=\"right\">0</td><td align=\"left\">3.5</td><td align=\"left\">1.67</td><td align=\"left\">0.01</td><td align=\"center\">424</td><td align=\"right\">16.5</td><td align=\"left\">1.67</td><td align=\"left\">0.14</td><td align=\"left\">12, 27–30</td></tr><tr><td align=\"left\">Lidocaine</td><td align=\"right\">4252</td><td align=\"right\">1276</td><td align=\"right\">2764</td><td align=\"right\">213</td><td align=\"left\">0.57</td><td align=\"left\">1.27</td><td align=\"left\">0.30</td><td align=\"center\">938</td><td align=\"right\">2.1</td><td align=\"left\">0.80</td><td align=\"left\">0.81</td><td align=\"left\">12, 31–35</td></tr><tr><td align=\"left\">Gatifloxacin</td><td align=\"right\">1101</td><td align=\"right\">341</td><td align=\"right\">574</td><td align=\"right\">186</td><td align=\"left\">1.8</td><td align=\"left\">1.07</td><td align=\"left\">0.68</td><td align=\"center\">252</td><td align=\"right\">6.5</td><td align=\"left\">1.07</td><td align=\"left\">0.75</td><td align=\"left\">36</td></tr><tr><td align=\"left\">Grepafloxacin</td><td align=\"right\">1079</td><td align=\"right\">917</td><td align=\"right\">151</td><td align=\"right\">11</td><td align=\"left\">3.4</td><td align=\"left\">1.34</td><td align=\"left\">0.44</td><td align=\"center\">245</td><td align=\"right\">11.6</td><td align=\"left\">1.1</td><td align=\"left\">0.45</td><td align=\"left\">37–39</td></tr><tr><td align=\"left\">Gemifloxacin</td><td align=\"right\">1300</td><td align=\"right\">163</td><td align=\"right\">599</td><td align=\"right\">432</td><td align=\"left\">1.6</td><td align=\"left\">1</td><td align=\"left\">0.57</td><td align=\"center\">500</td><td align=\"right\">7.0</td><td align=\"left\">1.2</td><td align=\"left\">0.30</td><td align=\"left\">40, 41</td></tr><tr><td align=\"left\">Pazufloxacin</td><td align=\"right\">970</td><td align=\"right\">90</td><td align=\"right\">721</td><td align=\"right\">159</td><td align=\"left\">0.88</td><td align=\"left\">1</td><td align=\"left\">0.74</td><td align=\"center\">384</td><td align=\"right\">1.8</td><td align=\"left\">1</td><td align=\"left\">0.77</td><td align=\"left\">38, 42–45</td></tr><tr><td align=\"left\">Enoxacin</td><td align=\"right\">1794</td><td align=\"right\">57</td><td align=\"right\">940</td><td align=\"right\">797</td><td align=\"left\">1.8</td><td align=\"left\">0.91</td><td align=\"left\">0.71</td><td align=\"center\">527</td><td align=\"right\">6.0</td><td align=\"left\">0.91</td><td align=\"left\">0.57</td><td align=\"left\">38, 46–48</td></tr><tr><td align=\"left\">Fleroxacin</td><td align=\"right\">285</td><td align=\"right\">57</td><td align=\"right\">195</td><td align=\"right\">34</td><td align=\"left\">2.6</td><td align=\"left\">1.29</td><td align=\"left\">0.40</td><td align=\"center\">120</td><td align=\"right\">9.5</td><td align=\"left\">1</td><td align=\"left\">0.77</td><td align=\"left\">36, 38, 49</td></tr><tr><td align=\"left\">Lomefloxacin</td><td align=\"right\">1243</td><td align=\"right\">973</td><td align=\"right\">270</td><td align=\"right\">0</td><td align=\"left\">4.0</td><td align=\"left\">1</td><td align=\"left\">0.69</td><td align=\"center\">252</td><td align=\"right\">7.1</td><td align=\"left\">1</td><td align=\"left\">0.79</td><td align=\"left\">50, 51</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Human pharmacokinetic prediction results for 14 compounds</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\">Parameter</td><td align=\"center\">Group</td><td align=\"center\" colspan=\"6\">T_1/2</td><td align=\"center\" colspan=\"6\">CL_tot</td></tr><tr><td/><td/><td colspan=\"6\"><hr/></td><td colspan=\"6\"><hr/></td></tr><tr><td/><td/><td align=\"center\" colspan=\"2\">6 organs</td><td align=\"center\" colspan=\"2\">8 organs</td><td align=\"center\" colspan=\"2\">10 organs</td><td align=\"center\" colspan=\"2\">6 organs</td><td align=\"center\" colspan=\"2\">8 organs</td><td align=\"center\" colspan=\"2\">10 organs</td></tr><tr><td/><td/><td colspan=\"2\"><hr/></td><td colspan=\"2\"><hr/></td><td colspan=\"2\"><hr/></td><td colspan=\"2\"><hr/></td><td colspan=\"2\"><hr/></td><td colspan=\"2\"><hr/></td></tr><tr><td/><td/><td align=\"left\">+</td><td align=\"left\">-</td><td align=\"left\">+</td><td align=\"left\">-</td><td align=\"left\">+</td><td align=\"left\">-</td><td align=\"left\">+</td><td align=\"left\">-</td><td align=\"left\">+</td><td align=\"left\">-</td><td align=\"left\">+</td><td align=\"left\">-</td></tr></thead><tbody><tr><td align=\"center\">AFE</td><td align=\"center\">acidic</td><td align=\"left\">5.45</td><td align=\"left\">38.7</td><td align=\"left\">2.26</td><td align=\"left\">16.6</td><td align=\"left\">1.36</td><td align=\"left\">13.5</td><td align=\"left\">1.39</td><td align=\"left\">305</td><td align=\"left\">1.07</td><td align=\"left\">26.5</td><td align=\"left\">2.56</td><td align=\"left\">14.7</td></tr><tr><td/><td align=\"center\">neutral</td><td align=\"left\">3.58</td><td align=\"left\">36.6</td><td align=\"left\">2.97</td><td align=\"left\">2.12</td><td align=\"left\">1.64</td><td align=\"left\">2.25</td><td align=\"left\">4.01</td><td align=\"left\">68.5</td><td align=\"left\">1.47</td><td align=\"left\">14.8</td><td align=\"left\">1.38</td><td align=\"left\">10.8</td></tr><tr><td/><td align=\"center\">basic</td><td align=\"left\">4.13</td><td align=\"left\">36.8</td><td align=\"left\">1.85</td><td align=\"left\">7.48</td><td align=\"left\">1.76</td><td align=\"left\">6.28</td><td align=\"left\">1.34</td><td align=\"left\">24.4</td><td align=\"left\">1.62</td><td align=\"left\">5.64</td><td align=\"left\">1.10</td><td align=\"left\">4.37</td></tr><tr><td/><td align=\"center\">zwitterionic</td><td align=\"left\">2.80</td><td align=\"left\">14.4</td><td align=\"left\">1.41</td><td align=\"left\">5.10</td><td align=\"left\">1.36</td><td align=\"left\">3.85</td><td align=\"left\">1.40</td><td align=\"left\">33.3</td><td align=\"left\">1.48</td><td align=\"left\">7.40</td><td align=\"left\">1.63</td><td align=\"left\">5.67</td></tr><tr><td colspan=\"14\"><hr/></td></tr><tr><td align=\"center\">AFE<break/></td><td align=\"center\">Overall</td><td align=\"left\">3.35</td><td align=\"left\">23.2</td><td align=\"left\">1.75</td><td align=\"left\">5.78</td><td align=\"left\">1.48</td><td align=\"left\">4.74</td><td align=\"left\">1.63</td><td align=\"left\">47.4</td><td align=\"left\">1.47</td><td align=\"left\">9.24</td><td align=\"left\">1.56</td><td align=\"left\">6.74</td></tr><tr><td align=\"center\">within-2-fold (%)</td><td/><td align=\"left\">0</td><td align=\"left\">0</td><td align=\"left\">62</td><td align=\"left\">14</td><td align=\"left\">93</td><td align=\"left\">14</td><td align=\"left\">92</td><td align=\"left\">0</td><td align=\"left\">92</td><td align=\"left\">0</td><td align=\"left\">79</td><td align=\"left\">0</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>AFE values and within-2-fold errors from the present study and previous studies</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\">Method</td><td align=\"center\">n</td><td align=\"center\" colspan=\"2\">AFE</td><td align=\"center\" colspan=\"2\">Within-2-fold error (%)</td></tr><tr><td/><td/><td/><td colspan=\"2\"><hr/></td><td colspan=\"2\"><hr/></td></tr><tr><td/><td/><td/><td align=\"center\">T_1/2</td><td align=\"center\">CL_tot</td><td align=\"center\">T_1/2</td><td align=\"center\">CL_tot</td></tr></thead><tbody><tr><td align=\"center\">Present work</td><td align=\"center\">SAC-WBPBPK</td><td align=\"center\">14</td><td align=\"center\">1.5</td><td align=\"center\">1.6</td><td align=\"center\">93</td><td align=\"center\">79</td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td/><td align=\"center\">generic WBPBPK</td><td align=\"center\">19⁷⁾</td><td align=\"center\">2.2</td><td align=\"center\">2.7</td><td align=\"center\">71</td><td align=\"center\">71</td></tr><tr><td/><td/><td align=\"center\">19 or 26⁸⁾</td><td align=\"center\">1.5</td><td align=\"center\">1.1</td><td align=\"center\">69</td><td align=\"center\">74</td></tr><tr><td align=\"center\">Previous work</td><td align=\"center\"><italic>in silico</italic>⁶¹⁾</td><td align=\"center\">18</td><td align=\"center\">N/A</td><td align=\"center\">2.8</td><td align=\"center\">N/A</td><td align=\"center\">50</td></tr><tr><td/><td align=\"center\">animal scale-up</td><td align=\"center\">19¹⁾</td><td align=\"center\">2.4</td><td align=\"center\">3.4</td><td align=\"center\">53</td><td align=\"center\">37</td></tr><tr><td/><td/><td align=\"center\">18⁶²⁾</td><td align=\"center\">N/A</td><td align=\"center\">2.5</td><td align=\"center\">N/A</td><td align=\"center\">50</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Values of Q_SAC, V_SAC, Kp_SAC, and various physicochemical parameters</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Compound</td><td align=\"center\" colspan=\"4\">SAC input parameter</td><td align=\"center\" colspan=\"4\">Physicochemical parameter</td><td align=\"center\">acidic/neutral/basic/zwitterionic</td></tr><tr><td/><td colspan=\"4\"><hr/></td><td colspan=\"4\"><hr/></td><td/></tr><tr><td/><td align=\"right\">Q</td><td align=\"right\">V</td><td align=\"left\">Kp</td><td align=\"right\">VKp^{<italic>a</italic>}</td><td align=\"left\">M.W.</td><td align=\"left\">clogP</td><td align=\"left\">cMR</td><td align=\"right\">tPSA</td><td/></tr></thead><tbody><tr><td align=\"left\">Tolbutamide</td><td align=\"right\">2850</td><td align=\"right\">330</td><td align=\"left\">0.5</td><td align=\"right\">165</td><td align=\"left\">270</td><td align=\"left\">2.5</td><td align=\"left\">7.1</td><td align=\"right\">93</td><td align=\"center\">acidic</td></tr><tr><td align=\"left\">Diclofenac</td><td align=\"right\">3120</td><td align=\"right\">300</td><td align=\"left\">0.3</td><td align=\"right\">90</td><td align=\"left\">296</td><td align=\"left\">4.7</td><td align=\"left\">7.7</td><td align=\"right\">59</td><td align=\"center\">acidic</td></tr><tr><td align=\"left\">Midazolam</td><td align=\"right\">3400</td><td align=\"right\">500</td><td align=\"left\">1</td><td align=\"right\">500</td><td align=\"left\">326</td><td align=\"left\">3.2</td><td align=\"left\">9.1</td><td align=\"right\">25</td><td align=\"center\">neutral</td></tr><tr><td align=\"left\">Diazepam</td><td align=\"right\">3130</td><td align=\"right\">380</td><td align=\"left\">2</td><td align=\"right\">760</td><td align=\"left\">285</td><td align=\"left\">3.2</td><td align=\"left\">8.1</td><td align=\"right\">36</td><td align=\"center\">neutral</td></tr><tr><td align=\"left\">Phenytoin</td><td align=\"right\">3120</td><td align=\"right\">5</td><td align=\"left\">0.3</td><td align=\"right\">1.5</td><td align=\"left\">252</td><td align=\"left\">2.1</td><td align=\"left\">7.2</td><td align=\"right\">70</td><td align=\"center\">basic</td></tr><tr><td align=\"left\">Imipramine</td><td align=\"right\">3120</td><td align=\"right\">790</td><td align=\"left\">8</td><td align=\"right\">6320</td><td align=\"left\">280</td><td align=\"left\">5.0</td><td align=\"left\">9.0</td><td align=\"right\">5</td><td align=\"center\">basic</td></tr><tr><td align=\"left\">Lidocaine</td><td align=\"right\">3670</td><td align=\"right\">500</td><td align=\"left\">2</td><td align=\"right\">1000</td><td align=\"left\">234</td><td align=\"left\">2.0</td><td align=\"left\">7.2</td><td align=\"right\">38</td><td align=\"center\">basic</td></tr><tr><td align=\"left\">Gatifloxacin</td><td align=\"right\">2830</td><td align=\"right\">800</td><td align=\"left\">1</td><td align=\"right\">800</td><td align=\"left\">375</td><td align=\"left\">-0.69</td><td align=\"left\">9.8</td><td align=\"right\">101</td><td align=\"center\">zwitterionic</td></tr><tr><td align=\"left\">Grepafloxacin</td><td align=\"right\">2810</td><td align=\"right\">300</td><td align=\"left\">1</td><td align=\"right\">300</td><td align=\"left\">359</td><td align=\"left\">-0.13</td><td align=\"left\">9.6</td><td align=\"right\">87</td><td align=\"center\">zwitterionic</td></tr><tr><td align=\"left\">Gemifloxacin</td><td align=\"right\">2470</td><td align=\"right\">100</td><td align=\"left\">1</td><td align=\"right\">100</td><td align=\"left\">389</td><td align=\"left\">-0.89</td><td align=\"left\">9.9</td><td align=\"right\">133</td><td align=\"center\">zwitterionic</td></tr><tr><td align=\"left\">Pazufloxacin</td><td align=\"right\">2920</td><td align=\"right\">430</td><td align=\"left\">1</td><td align=\"right\">430</td><td align=\"left\">318</td><td align=\"left\">-0.87</td><td align=\"left\">8.0</td><td align=\"right\">108</td><td align=\"center\">zwitterionic</td></tr><tr><td align=\"left\">Enoxacin</td><td align=\"right\">2970</td><td align=\"right\">455</td><td align=\"left\">5</td><td align=\"right\">2275</td><td align=\"left\">320</td><td align=\"left\">-1.8</td><td align=\"left\">8.2</td><td align=\"right\">98</td><td align=\"center\">zwitterionic</td></tr><tr><td align=\"left\">Fleroxacin</td><td align=\"right\">2840</td><td align=\"right\">100</td><td align=\"left\">1</td><td align=\"right\">100</td><td align=\"left\">369</td><td align=\"left\">-0.65</td><td align=\"left\">8.9</td><td align=\"right\">77</td><td align=\"center\">zwitterionic</td></tr><tr><td align=\"left\">Lomefloxacin</td><td align=\"right\">2870</td><td align=\"right\">650</td><td align=\"left\">6</td><td align=\"right\">3900</td><td align=\"left\">351</td><td align=\"left\">-0.30</td><td align=\"left\">8.9</td><td align=\"right\">87</td><td align=\"center\">zwitterionic</td></tr></tbody></table></table-wrap>" ]

[ "<disp-formula id=\"bmcM1\"><label>(1)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M1\" name=\"1742-4682-5-19-i1\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mfrac><mml:mrow><mml:mi>d</mml:mi><mml:msub><mml:mi>C</mml:mi><mml:mi>h</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:mi>d</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>C</mml:mi><mml:mi>a</mml:mi></mml:msub><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:msub><mml:mi>Q</mml:mi><mml:mi>h</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:msub><mml:mi>Q</mml:mi><mml:mrow><mml:mi>g</mml:mi><mml:mi>i</mml:mi></mml:mrow></mml:msub><mml:mo>−</mml:mo><mml:msub><mml:mi>Q</mml:mi><mml:mrow><mml:mi>p</mml:mi><mml:mi>a</mml:mi><mml:mi>n</mml:mi><mml:mi>c</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow><mml:mrow><mml:msub><mml:mi>V</mml:mi><mml:mi>h</mml:mi></mml:msub></mml:mrow></mml:mfrac><mml:mo>+</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>Q</mml:mi><mml:mrow><mml:mi>g</mml:mi><mml:mi>i</mml:mi></mml:mrow></mml:msub><mml:mo>⋅</mml:mo><mml:msub><mml:mi>C</mml:mi><mml:mrow><mml:mi>g</mml:mi><mml:mi>i</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mi>V</mml:mi><mml:mi>h</mml:mi></mml:msub><mml:mo>⋅</mml:mo><mml:mi>K</mml:mi><mml:msub><mml:mi>p</mml:mi><mml:mrow><mml:mi>g</mml:mi><mml:mi>i</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mfrac><mml:mo>+</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>Q</mml:mi><mml:mrow><mml:mi>p</mml:mi><mml:mi>a</mml:mi><mml:mi>n</mml:mi><mml:mi>c</mml:mi></mml:mrow></mml:msub><mml:mo>⋅</mml:mo><mml:msub><mml:mi>C</mml:mi><mml:mrow><mml:mi>p</mml:mi><mml:mi>a</mml:mi><mml:mi>n</mml:mi><mml:mi>c</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mi>V</mml:mi><mml:mi>h</mml:mi></mml:msub><mml:mo>⋅</mml:mo><mml:mi>K</mml:mi><mml:msub><mml:mi>p</mml:mi><mml:mrow><mml:mi>p</mml:mi><mml:mi>a</mml:mi><mml:mi>n</mml:mi><mml:mi>c</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mfrac><mml:mo>−</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>Q</mml:mi><mml:mi>h</mml:mi></mml:msub><mml:mo>⋅</mml:mo><mml:msub><mml:mi>C</mml:mi><mml:mi>h</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mi>V</mml:mi><mml:mi>h</mml:mi></mml:msub><mml:mo>⋅</mml:mo><mml:mi>K</mml:mi><mml:msub><mml:mi>p</mml:mi><mml:mi>h</mml:mi></mml:msub></mml:mrow></mml:mfrac><mml:mo>−</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>C</mml:mi><mml:mi>a</mml:mi></mml:msub><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:msub><mml:mi>Q</mml:mi><mml:mi>h</mml:mi></mml:msub><mml:mo>⋅</mml:mo><mml:msub><mml:mi>f</mml:mi><mml:mi>B</mml:mi></mml:msub><mml:mo>⋅</mml:mo><mml:mi>C</mml:mi><mml:msub><mml:mi>L</mml:mi><mml:mrow><mml:mi>int</mml:mi><mml:mo>⁡</mml:mo><mml:mo>,</mml:mo><mml:mi>h</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow><mml:mrow><mml:msub><mml:mi>V</mml:mi><mml:mi>h</mml:mi></mml:msub><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:msub><mml:mi>Q</mml:mi><mml:mi>h</mml:mi></mml:msub><mml:mo>+</mml:mo><mml:msub><mml:mi>f</mml:mi><mml:mi>B</mml:mi></mml:msub><mml:mo>⋅</mml:mo><mml:mi>C</mml:mi><mml:msub><mml:mi>L</mml:mi><mml:mrow><mml:mi>int</mml:mi><mml:mo>⁡</mml:mo><mml:mo>,</mml:mo><mml:mi>h</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM2\"><label>(2)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M2\" name=\"1742-4682-5-19-i2\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mfrac><mml:mrow><mml:mi>d</mml:mi><mml:msub><mml:mi>C</mml:mi><mml:mi>r</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:mi>d</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>Q</mml:mi><mml:mi>r</mml:mi></mml:msub><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:msub><mml:mi>C</mml:mi><mml:mi>a</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:msub><mml:mi>C</mml:mi><mml:mi>r</mml:mi></mml:msub></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow><mml:mrow><mml:msub><mml:mi>V</mml:mi><mml:mi>r</mml:mi></mml:msub><mml:mo>⋅</mml:mo><mml:mi>K</mml:mi><mml:msub><mml:mi>p</mml:mi><mml:mi>r</mml:mi></mml:msub></mml:mrow></mml:mfrac><mml:mo>−</mml:mo><mml:mfrac><mml:mrow><mml:mi>C</mml:mi><mml:msub><mml:mi>L</mml:mi><mml:mi>r</mml:mi></mml:msub><mml:mo>⋅</mml:mo><mml:msub><mml:mi>C</mml:mi><mml:mi>a</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mi>V</mml:mi><mml:mi>r</mml:mi></mml:msub></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM3\"><label>(3)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M3\" name=\"1742-4682-5-19-i3\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mfrac><mml:mrow><mml:mi>d</mml:mi><mml:msub><mml:mi>C</mml:mi><mml:mrow><mml:mi>g</mml:mi><mml:mi>i</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mrow><mml:mi>d</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>Q</mml:mi><mml:mrow><mml:mi>g</mml:mi><mml:mi>i</mml:mi></mml:mrow></mml:msub><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:msub><mml:mi>C</mml:mi><mml:mi>a</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:msub><mml:mi>C</mml:mi><mml:mrow><mml:mi>g</mml:mi><mml:mi>i</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow><mml:mrow><mml:msub><mml:mi>V</mml:mi><mml:mrow><mml:mi>g</mml:mi><mml:mi>i</mml:mi></mml:mrow></mml:msub><mml:mo>⋅</mml:mo><mml:mi>K</mml:mi><mml:msub><mml:mi>p</mml:mi><mml:mrow><mml:mi>g</mml:mi><mml:mi>i</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mfrac><mml:mo>−</mml:mo><mml:mfrac><mml:mrow><mml:mi>C</mml:mi><mml:msub><mml:mi>L</mml:mi><mml:mrow><mml:mi>g</mml:mi><mml:mi>i</mml:mi></mml:mrow></mml:msub><mml:mo>⋅</mml:mo><mml:msub><mml:mi>C</mml:mi><mml:mi>a</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mi>V</mml:mi><mml:mrow><mml:mi>g</mml:mi><mml:mi>i</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM4\"><label>(4)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M4\" name=\"1742-4682-5-19-i4\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mfrac><mml:mrow><mml:mi>d</mml:mi><mml:msub><mml:mi>C</mml:mi><mml:mrow><mml:mi>l</mml:mi><mml:mi>u</mml:mi><mml:mi>n</mml:mi><mml:mi>g</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mrow><mml:mi>d</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>Q</mml:mi><mml:mrow><mml:mi>t</mml:mi><mml:mi>o</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:msub><mml:mi>C</mml:mi><mml:mi>v</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:msub><mml:mi>C</mml:mi><mml:mrow><mml:mi>l</mml:mi><mml:mi>u</mml:mi><mml:mi>n</mml:mi><mml:mi>g</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow><mml:mrow><mml:msub><mml:mi>V</mml:mi><mml:mrow><mml:mi>l</mml:mi><mml:mi>u</mml:mi><mml:mi>n</mml:mi><mml:mi>g</mml:mi></mml:mrow></mml:msub><mml:mo>⋅</mml:mo><mml:mi>K</mml:mi><mml:msub><mml:mi>p</mml:mi><mml:mrow><mml:mi>l</mml:mi><mml:mi>u</mml:mi><mml:mi>n</mml:mi><mml:mi>g</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM5\"><label>(5)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M5\" name=\"1742-4682-5-19-i5\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mfrac><mml:mrow><mml:mi>d</mml:mi><mml:msub><mml:mi>C</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:mi>d</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>Q</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:msub><mml:mi>C</mml:mi><mml:mi>a</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:msub><mml:mi>C</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow><mml:mrow><mml:msub><mml:mi>V</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>⋅</mml:mo><mml:mi>K</mml:mi><mml:msub><mml:mi>p</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM6\"><label>(6)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M6\" name=\"1742-4682-5-19-i6\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mfrac><mml:mrow><mml:mi>d</mml:mi><mml:msub><mml:mi>C</mml:mi><mml:mi>v</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:mi>d</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:mstyle displaystyle=\"true\"><mml:mo>∑</mml:mo><mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mfrac><mml:mrow><mml:msub><mml:mi>Q</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>⋅</mml:mo><mml:msub><mml:mi>C</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mi>V</mml:mi><mml:mi>v</mml:mi></mml:msub><mml:mo>⋅</mml:mo><mml:mi>K</mml:mi><mml:msub><mml:mi>p</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow></mml:mfrac></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow></mml:mstyle><mml:mo>−</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>Q</mml:mi><mml:mrow><mml:mi>t</mml:mi><mml:mi>o</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub><mml:mo>⋅</mml:mo><mml:msub><mml:mi>C</mml:mi><mml:mi>v</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mi>V</mml:mi><mml:mi>v</mml:mi></mml:msub></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM7\"><label>(7)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M7\" name=\"1742-4682-5-19-i7\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mfrac><mml:mrow><mml:mi>d</mml:mi><mml:msub><mml:mi>C</mml:mi><mml:mi>a</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:mi>d</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mfrac><mml:mrow><mml:msub><mml:mi>C</mml:mi><mml:mrow><mml:mi>l</mml:mi><mml:mi>u</mml:mi><mml:mi>n</mml:mi><mml:mi>g</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mrow><mml:mi>K</mml:mi><mml:msub><mml:mi>p</mml:mi><mml:mrow><mml:mi>l</mml:mi><mml:mi>u</mml:mi><mml:mi>n</mml:mi><mml:mi>g</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mfrac><mml:mo>−</mml:mo><mml:msub><mml:mi>C</mml:mi><mml:mi>a</mml:mi></mml:msub></mml:mrow><mml:mo>)</mml:mo></mml:mrow><mml:mfrac><mml:mrow><mml:msub><mml:mi>Q</mml:mi><mml:mrow><mml:mi>t</mml:mi><mml:mi>o</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mi>V</mml:mi><mml:mi>a</mml:mi></mml:msub></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM8\"><label>(8)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M8\" name=\"1742-4682-5-19-i8\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>C</mml:mi><mml:msub><mml:mi>L</mml:mi><mml:mrow><mml:mi>int</mml:mi><mml:mo>⁡</mml:mo><mml:mo>,</mml:mo><mml:mi>h</mml:mi><mml:mi>u</mml:mi><mml:mi>m</mml:mi><mml:mi>a</mml:mi><mml:mi>n</mml:mi><mml:mo>,</mml:mo><mml:mi>i</mml:mi><mml:mi>n</mml:mi><mml:mi>v</mml:mi><mml:mi>i</mml:mi><mml:mi>v</mml:mi><mml:mi>o</mml:mi></mml:mrow></mml:msub><mml:mo>=</mml:mo><mml:mi>C</mml:mi><mml:msub><mml:mi>L</mml:mi><mml:mrow><mml:mi>int</mml:mi><mml:mo>⁡</mml:mo><mml:mo>,</mml:mo><mml:mi>r</mml:mi><mml:mi>a</mml:mi><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>i</mml:mi><mml:mi>n</mml:mi><mml:mi>v</mml:mi><mml:mi>i</mml:mi><mml:mi>v</mml:mi><mml:mi>o</mml:mi></mml:mrow></mml:msub><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mfrac><mml:mrow><mml:mi>C</mml:mi><mml:msub><mml:mi>L</mml:mi><mml:mrow><mml:mi>int</mml:mi><mml:mo>⁡</mml:mo><mml:mo>,</mml:mo><mml:mi>h</mml:mi><mml:mi>u</mml:mi><mml:mi>m</mml:mi><mml:mi>a</mml:mi><mml:mi>n</mml:mi><mml:mo>,</mml:mo><mml:mi>i</mml:mi><mml:mi>n</mml:mi><mml:mi>v</mml:mi><mml:mi>i</mml:mi><mml:mi>t</mml:mi><mml:mi>r</mml:mi><mml:mi>o</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mrow><mml:mi>C</mml:mi><mml:msub><mml:mi>L</mml:mi><mml:mrow><mml:mi>int</mml:mi><mml:mo>⁡</mml:mo><mml:mo>,</mml:mo><mml:mi>r</mml:mi><mml:mi>a</mml:mi><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>i</mml:mi><mml:mi>n</mml:mi><mml:mi>v</mml:mi><mml:mi>i</mml:mi><mml:mi>t</mml:mi><mml:mi>r</mml:mi><mml:mi>o</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mfrac></mml:mrow><mml:mo>)</mml:mo></mml:mrow><mml:mo>⋅</mml:mo><mml:mi>s</mml:mi><mml:mi>f</mml:mi></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM9\"><label>(9)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M9\" name=\"1742-4682-5-19-i9\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>C</mml:mi><mml:msub><mml:mi>L</mml:mi><mml:mrow><mml:mi>o</mml:mi><mml:mi>r</mml:mi><mml:mi>g</mml:mi><mml:mo>,</mml:mo><mml:mi>h</mml:mi><mml:mi>u</mml:mi><mml:mi>m</mml:mi><mml:mi>a</mml:mi><mml:mi>n</mml:mi></mml:mrow></mml:msub><mml:mo>=</mml:mo><mml:mi>C</mml:mi><mml:msub><mml:mi>L</mml:mi><mml:mrow><mml:mi>o</mml:mi><mml:mi>r</mml:mi><mml:mi>g</mml:mi><mml:mo>,</mml:mo><mml:mi>r</mml:mi><mml:mi>a</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mfrac><mml:mrow><mml:msub><mml:mi>Q</mml:mi><mml:mrow><mml:mi>j</mml:mi><mml:mo>,</mml:mo><mml:mi>h</mml:mi><mml:mi>u</mml:mi><mml:mi>m</mml:mi><mml:mi>a</mml:mi><mml:mi>n</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mi>Q</mml:mi><mml:mrow><mml:mi>j</mml:mi><mml:mo>,</mml:mo><mml:mi>r</mml:mi><mml:mi>a</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mfrac></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM10\"><label>(10)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M10\" name=\"1742-4682-5-19-i10\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msub><mml:mi>Q</mml:mi><mml:mrow><mml:mi>S</mml:mi><mml:mi>A</mml:mi><mml:mi>C</mml:mi><mml:mo>,</mml:mo><mml:mi>h</mml:mi><mml:mi>u</mml:mi><mml:mi>m</mml:mi><mml:mi>a</mml:mi><mml:mi>n</mml:mi></mml:mrow></mml:msub><mml:mo>=</mml:mo><mml:msub><mml:mi>Q</mml:mi><mml:mrow><mml:mi>S</mml:mi><mml:mi>A</mml:mi><mml:mi>C</mml:mi><mml:mo>,</mml:mo><mml:mi>r</mml:mi><mml:mi>a</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub><mml:mrow><mml:mo>[</mml:mo><mml:mrow><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mstyle displaystyle=\"true\"><mml:mo>∑</mml:mo><mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mfrac><mml:mi>P</mml:mi><mml:mrow><mml:msub><mml:mi>Q</mml:mi><mml:mrow><mml:mi>r</mml:mi><mml:mi>i</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mfrac></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow></mml:mstyle></mml:mrow><mml:mo>]</mml:mo></mml:mrow><mml:mo>⋅</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>Q</mml:mi><mml:mrow><mml:mi>t</mml:mi><mml:mi>o</mml:mi><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>h</mml:mi><mml:mi>u</mml:mi><mml:mi>m</mml:mi><mml:mi>a</mml:mi><mml:mi>n</mml:mi></mml:mrow></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mi>Q</mml:mi><mml:mrow><mml:mi>t</mml:mi><mml:mi>o</mml:mi><mml:mi>t</mml:mi><mml:mo>,</mml:mo><mml:mi>r</mml:mi><mml:mi>a</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM11\"><label>(11)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M11\" name=\"1742-4682-5-19-i11\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msub><mml:mi>V</mml:mi><mml:mrow><mml:mi>S</mml:mi><mml:mi>A</mml:mi><mml:mi>C</mml:mi><mml:mo>,</mml:mo><mml:mi>h</mml:mi><mml:mi>u</mml:mi><mml:mi>m</mml:mi><mml:mi>a</mml:mi><mml:mi>n</mml:mi></mml:mrow></mml:msub><mml:mo>=</mml:mo><mml:msub><mml:mi>V</mml:mi><mml:mrow><mml:mi>S</mml:mi><mml:mi>A</mml:mi><mml:mi>C</mml:mi><mml:mo>,</mml:mo><mml:mi>r</mml:mi><mml:mi>a</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub><mml:mfrac><mml:mrow><mml:mstyle displaystyle=\"true\"><mml:mo>∑</mml:mo><mml:mrow><mml:msub><mml:mi>V</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mo>,</mml:mo><mml:mi>h</mml:mi><mml:mi>u</mml:mi><mml:mi>m</mml:mi><mml:mi>a</mml:mi><mml:mi>n</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mstyle></mml:mrow><mml:mrow><mml:mstyle displaystyle=\"true\"><mml:mo>∑</mml:mo><mml:mrow><mml:msub><mml:mi>V</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mo>,</mml:mo><mml:mi>r</mml:mi><mml:mi>a</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mstyle></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM12\"><label>(12)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M12\" name=\"1742-4682-5-19-i12\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:msub><mml:mi>T</mml:mi><mml:mrow><mml:mn>1</mml:mn><mml:mo>/</mml:mo><mml:mn>2</mml:mn></mml:mrow></mml:msub><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mi>ln</mml:mi><mml:mo>⁡</mml:mo><mml:mn>2</mml:mn></mml:mrow><mml:mrow><mml:msub><mml:mi>k</mml:mi><mml:mrow><mml:mi>e</mml:mi><mml:mi>l</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM13\"><label>(13)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M13\" name=\"1742-4682-5-19-i13\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>C</mml:mi><mml:msub><mml:mi>L</mml:mi><mml:mrow><mml:mi>t</mml:mi><mml:mi>o</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:msub><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mi>D</mml:mi><mml:mi>o</mml:mi><mml:mi>s</mml:mi><mml:mi>e</mml:mi></mml:mrow><mml:mrow><mml:mi>A</mml:mi><mml:mi>U</mml:mi><mml:msub><mml:mi>C</mml:mi><mml:mrow><mml:mi>inf</mml:mi><mml:mo>⁡</mml:mo></mml:mrow></mml:msub></mml:mrow></mml:mfrac></mml:mrow></mml:semantics></mml:math></disp-formula>", "<disp-formula id=\"bmcM14\"><label>(14)</label><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M14\" name=\"1742-4682-5-19-i14\" overflow=\"scroll\"><mml:semantics><mml:mrow><mml:mi>A</mml:mi><mml:mi>F</mml:mi><mml:mi>E</mml:mi><mml:mo>=</mml:mo><mml:msup><mml:mrow><mml:mn>10</mml:mn></mml:mrow><mml:mrow><mml:mfrac><mml:mn>1</mml:mn><mml:mi>N</mml:mi></mml:mfrac><mml:mstyle displaystyle=\"true\"><mml:mo>∑</mml:mo><mml:mrow><mml:mrow><mml:mo>|</mml:mo><mml:mrow><mml:mi>log</mml:mi><mml:mo>⁡</mml:mo><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mfrac><mml:mrow><mml:mi>o</mml:mi><mml:mi>b</mml:mi><mml:mi>s</mml:mi><mml:mi>e</mml:mi><mml:mi>r</mml:mi><mml:mi>v</mml:mi><mml:mi>e</mml:mi><mml:mi>d</mml:mi></mml:mrow><mml:mrow><mml:mi>s</mml:mi><mml:mi>i</mml:mi><mml:mi>m</mml:mi><mml:mi>u</mml:mi><mml:mi>l</mml:mi><mml:mi>a</mml:mi><mml:mi>t</mml:mi><mml:mi>e</mml:mi><mml:mi>d</mml:mi></mml:mrow></mml:mfrac></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow><mml:mo>|</mml:mo></mml:mrow></mml:mrow></mml:mstyle></mml:mrow></mml:msup></mml:mrow></mml:semantics></mml:math></disp-formula>" ]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>^{<italic>a</italic>}R_B(blood-plasma concentration ratio) assumed to be 1 when there were no data in the literature.</p></table-wrap-foot>", "<table-wrap-foot><p>+: WBPBPK model with SAC, -: WBPBPK model without SAC.</p></table-wrap-foot>", "<table-wrap-foot><p>N/A: not available.</p></table-wrap-foot>", "<table-wrap-foot><p>^{<italic>a</italic>}VKp represents the product of V and Kp.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1742-4682-5-19-1\"/>", "<graphic xlink:href=\"1742-4682-5-19-2\"/>", "<graphic xlink:href=\"1742-4682-5-19-3\"/>", "<graphic xlink:href=\"1742-4682-5-19-4\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Increasingly, automated monitoring methods are being applied to routinely collected population health data to facilitate the recognition of significant changes in the health indicators under surveillance. The implementation of automated monitoring methods has been associated with improved awareness of trends in health-related data, improved data sharing and integration, and an improved ability to detect and respond to health events [##REF##15958924##1##].</p>", "<p>Statistical process control methods such as cumulative sums (cusums) are among the most commonly used methods to monitor population health surveillance data. Cusums are powerful, yet reasonably straightforward to design and implement, and are considered well-suited to the task of detecting changes in surveillance data early [##UREF##0##2##, ####REF##9242040##3##, ##REF##12033676##4####12033676##4##]. Using cusums to monitor disease data traditionally assumes that the parameters which adequately describe the observed outcome of the disease processes when it is at an ideal level can be specified [##UREF##1##5##]. In practice, specifying parameters which adequately describe the process under surveillance can be difficult, particularly when the process and systems associated with data generation do not appear to be stable, and disturbances can vary in size and area. Over time, changes in disease surveillance methods can also produce apparent changes in disease incidence when no real change has occurred.</p>", "<p>One alternative to using fixed parameters to describe the ideal behaviour of the process under surveillance is to use historical observations to estimate the parameters to be employed in monitoring [##UREF##1##5##]. This approach has been used in the cusum-based monitoring methods developed as part of the Early Aberration Reporting System (EARS) [##REF##15678442##6##]. The EARS cusum-based aberration detection algorithms use recently observed data to inform the expected level of the data under surveillance. As such, these algorithms signal change from the recent past rather than from theoretically derived parameters or ideal values.</p>", "<p>Determining an appropriate baseline period from which to derive parameters for automated monitoring is complex, as changes in surveillance methods and trends associated with changing disease epidemiology can limit the usefulness of long series of historical data for the identification of ideal process levels. For example, longer baselines can be useful as a mechanism to assist in identifying or down-weighting the influence of previous outbreaks on summary statistics; however, for some diseases, seasonal outbreaks may produce elevated baseline estimates over many months. There is little specific information available to guide the selection of appropriate baselines in individual public health surveillance applications.</p>", "<p>Routine national notifiable disease surveillance methods in Australia do not yet include the use of automated algorithms for outbreak detection; however, the EARS algorithms (C1, C2 and C3) have been implemented at a national level in New Zealand for the surveillance of notifiable disease data [##UREF##2##7##]. The EARS algorithms provide early detection of real and simulated disease outbreaks [##UREF##3##8##], and these cusum methods, which require little baseline data, have been found to perform as well as methods that require greater amounts of historical data for baseline estimation [##REF##15752454##9##]. Although the EARS C1, C2 and C3 algorithms have been found to detect outbreaks of public health interest, including the start of the influenza season [##REF##15752454##9##], little is known about how the signalling pattern of these algorithms corresponds to the identification of events of public health interest among epidemiologists.</p>", "<p>In contrast to the industrial applications for which cusums were originally designed [##UREF##0##2##], the use of cusums to monitor population health data provides additional challenges associated with monitoring the complex and variable process of disease transmission and detection. Cusums applied to public health surveillance data which have high variance have been associated with a greater than expected number of false alarms in comparison to data with low variance [##REF##16372902##10##]. Similarly, public health surveillance data with fluctuating variance has also been associated with variable specificity of surveillance algorithms [##REF##17567912##11##]. Although a negative binomial cusum may provide a means to moderate the reported high false alarm rates associated with the use of established cusum based on other statistical models, the performance of negative binomial cusums has not been widely investigated. This analysis aimed to investigate the correspondence between retrospective epidemiological evaluation of notifications of RRv disease in Western Australia, and the signals produced by two cusum-based automated monitoring methods; the widely used EARS C1, C2 and C3 cusums, and a negative binomial cusum.</p>" ]

[ "<title>Methods</title>", "<p>The performance of the EARS C1, C2 and C3 algorithms [##REF##15678442##6##], and a negative-binomial cusum were compared with the occurrence of events deemed by two experienced epidemiologists to be of potential public health importance using historical daily RRv disease notification data. RRv disease is among the most commonly notified diseases in Western Australia, with a total of 1099 cases notified in Western Australia (55.4 per 100,000 population) during 2004 [##REF##16639808##12##]. RRv infections most frequently occur among middle-aged adults, and produce a range of clinical symptoms including fatigue and polyarthritis that typically last from months to years [##REF##11729067##13##].</p>", "<title>Data</title>", "<p>Historical daily RRv disease case notification data for Western Australia are not publicly available, and de-identified data were obtained from the National Notifiable Diseases Surveillance System (NNDSS) under agreement from the Department of Health Western Australia and the Australian Government Department of Health and Ageing. Data were available for analysis from the 1^stof January 1991 to the 10^thof September 2004 (5002 days). Cases of RRv disease are notified based on laboratory evidence of infection, however, this does not always imply a definitive diagnosis [##UREF##4##14##]. Due to under-presentation and the underuse of laboratory testing in endemic areas, notified rates of RRv disease are considered to underestimate true disease rates [##REF##11729067##13##].</p>", "<p>The use of the date of report as the reference date in this investigation is based on the rationale outlined by Farrington and coworkers [##UREF##5##15##], including its availability for all notified cases. The use of the report date will influence timeliness and sensitivity of outbreak detection as it incorporates additional variability between the onset of illness and the date of report [##UREF##5##15##], however, they are the most feasible data to monitor in prospective surveillance.</p>", "<title>Definition of outbreaks</title>", "<p>As detailed historical records of prospectively-identified outbreaks of RRv disease in Western Australia were not available, retrospective expert epidemiological evaluation was used to identify the occurrence of events of epidemiological significance in the data. Daily time series data for the study period were reviewed by two independent epidemiologists. These highly experienced medical epidemiologists, although employed within the same university faculty for a year, both have a history of independent employment both within Australia and internationally. The time series graphs used to identify RRv disease outbreaks also displayed the approximate timing of any outbreaks that were documented in the Communicable Diseases Intelligence quarterly and annual surveillance reports [##UREF##6##16##], or were identified in a review of RRv in Australia [##REF##11729067##13##].</p>", "<p>Both epidemiologists independently identified events of epidemiological significance by labelling outbreak and non-outbreak periods based on a visual review of the data, information provided on documented outbreaks, and their knowledge of RRv disease epidemiology in Australia. Start and end dates of the retrospectively identified events were specified. A meeting between the researcher and the two epidemiologists was subsequently used to discuss any differences in the identification of events or their timing and produce an agreed standard.</p>", "<p>There were two areas of difference in opinion among the epidemiologists. The first area related to minor differences in the timing of specified start and end dates for each event identified, and was resolved through a joint review of the data. The second area of difference was the labelling of hyper-endemic periods during August to December 1992 and February to June 1994 by one expert. The first epidemiologist indicated that the observed level of disease activity in practice could be of some concern; however, chose to concur with the second expert who believed that although the level of disease was different, it was difficult to label as an event of significance based on the level of notifications alone. No information was provided to expert reviewers on epidemiological linkages between cases or geographical distribution. Following the joint discussion, a total of 15 events of public health significance were identified, and these were considered to define 'outbreak' periods for the purposes of algorithm comparison.</p>", "<title>EARS algorithms</title>", "<p>The EARS algorithms C1, C2 and C3 [##REF##15678442##6##] were selected for evaluation due to their widespread use. We evaluated the performance of the EARS algorithms over a range of alarm levels. If the conventional alarm level (C1 = 2) is used, the C1 algorithm simplifies to the current value being greater than the baseline mean plus three standard deviations, which is based on the previous 7 days of data. The C2 algorithm differs from C1 in the use of a guard band of two days duration between the baseline and the current day being evaluated. The C3 algorithm also uses a two-day guard band, but calculates a partial sum for the last three days of the positive deviation of the current value from the mean [##REF##15678442##6##]. The EARS algorithms are designed to signal when the cusum values exceed 2, which implies that the algorithm statistics have exceeded a level which is three standard deviations greater than the baseline mean.</p>", "<p>The algorithms were run using the R statistical software based on the implementation of the algorithms in the EARS-X Excel software version [##UREF##7##17##]. To check for coding errors in the R implementation of EARS developed for this study, the outputs of both versions were compared based on six semi-synthetic datasets of 150 days in length with randomly inserted outbreaks of different magnitudes. No differences in performance between the excel and R versions of the algorithms were detected. The R code is provided in Additional file ##SUPPL##0##1##.</p>", "<title>Negative binomial cusum algorithms</title>", "<p>A negative binomial cusum [##UREF##0##2##] was selected for testing due to the potential ability of this method to minimise false alarms associated with over-dispersed data. The negative binomial distribution can be described by two parameters, r and c, where over-dispersion is determined by the parameter c. The following two equations were used to determine the negative binomial parameter values based on the means (u) and variances (σ²) derived from selected baseline periods:</p>", "<p></p>", "<p></p>", "<p>If we consider r as given and monitor for changes in c from an in-control c₀to an out of control level c₁where c₁&gt; c₀, the decision interval cusum is given by [##UREF##0##2##]:</p>", "<p></p>", "<p></p>", "<p></p>", "<p>The out of control level c₁was determined by calculating a fixed interval of two baseline standard deviations greater than the baseline-derived in control level c₀, effectively setting the negative binomial cusum to detect a shift magnitude of two standard deviations above the baseline mean. To allow comparison with the EARS algorithms, the negative binomial cusum was evaluated with the baseline mean estimated using just 7 days of baseline data. Like the EARS algorithms C2 and C3, a guard band of 2 non-analysis days was used between the 7 days of data used to establish the baseline mean and variance, and the data for the current day. This guard band prevents the most recent data being included in the baseline estimates, which may be detrimental in the case of slowly increasing incidence being incorporated into the baseline. The negative binomial cusum was also tested using three additional baseline data period lengths: 14, 28 and 56 days. A zero-start method was evaluated, and cusums were not reset following alarms to allow exploration of the effects of different signal thresholds on performance. The R code implementation of the negative binomial cusum is provided in Additional File ##SUPPL##0##1##.</p>", "<title>Evaluation</title>", "<p>To compare algorithm performance, empirical methods were used to determine the cut-off values for each algorithm that would produce equivalent false alarm rates. We investigate the performance of both algorithm types using false alarm rates of approximately 0.005 and 0.001, which implies that the cusums are expected on average to produce a false positive alarm approximately once every six months and three years respectively.</p>", "<p>Although the cut-off value used to determine signalling of the cusums was varied to allow exploration of performance at different false alarm rates, the implementation of the C3 algorithm retains the threshold of 2 used to exclude large observed counts on the current or previous two days from the cusum total score, as implemented in the EARS-X software (see Additional File ##SUPPL##0##1##).</p>", "<p>Performance comparisons were based on two main indicators: sensitivity, which describes the ability of the algorithm to detect outbreaks in the data at any time during each outbreak period; and timeliness, which describes the number of days from the beginning of each outbreak until the first signal for each outbreak. As sensitivity was defined as the signalling of the algorithm at any point during the defined outbreak periods, timeliness was determined only for outbreaks that were detected, and defined as the number of days between the first nominated outbreak day, and the day of the first signal during the outbreak period. Any signals that occurred during outbreak periods were considered valid and the first of these during each outbreak period was used to calculate the time to detection. Signals that occurred on non-outbreak days (days that were not considered to be of epidemiological significance), were considered false alarms, and the average proportion of false alarms was calculated as the total number of false alarms divided by the total number of non-outbreak days. A one per cent false alarm rate (0.01) is equivalent to an alarm occurring on average on one out of every one hundred non-outbreak days.</p>", "<p>As the early detection of events of interest is important, timeliness was also summarised for each cusum as the proportion of outbreaks detected within the first 7 days. As performed for a previous evaluation of the EARS algorithms [##REF##16177703##18##], we calculated a conditional average run length indicator based on the detection of an event within the first 7 days. To enable the generation of complete timeliness data for algorithm comparison, an additional outcome variable (adjusted timeliness) was also derived by allocating the total duration of each outbreak as the timeliness result if an outbreak was undetected.</p>", "<p>The performance of each algorithm was also evaluated by comparing the area under the receiver operating characteristic (ROC) curve (AUC) [##REF##17089948##19##]. The AUC calculations were performed using the trapz function of the CaTools package for the R statistical software [##UREF##8##20##]. To allow the consideration of both sensitivity and timeliness in AUC comparisons, a weighted AUC indicator (WtAUC) was also generated [##REF##17089948##19##], which weights the contribution of the sensitivity component of the AUC indicator based on the proportion of time saved relative to a reference value. As no historical data were available to provide a reference value, a fixed value of 7 days was used. The weightings applied to the sensitivity data were calculated as (7-timeliness)/7, with a lower limit of zero. As such, outbreaks detected more than seven days after their commencement do not contribute to the weighted AUC indicator.</p>", "<p>Formal statistical comparisons of sensitivity, timeliness, adjusted timeliness, AUC and weighted AUC for the EARS and negative binomial cusums were performed using Friedman Rank Sum Tests as implemented in the Stats package for the R software version 2.6.0 [##UREF##9##21##]. Selected multiple comparisons were performed to investigate difference in performance between the three EARS algorithms and the 7-day and 28-day negative binomial cusums using paired Wilcoxon Signed Rank tests. As these six analyses were conducted for each multiple comparisons procedure, the two-tailed p-value used to determine significance for the Wilcoxon Signed Rank tests was adjusted to reflect the repeated testing and set at 0.05/6, or 0.0083. Nonparametric methods were used due to the non-normailty of the data and small sample size available for analysis. This research was approved by the Human Research Ethics Committee of Curtin University of Technology.</p>" ]

[ "<title>Results</title>", "<p>We observed a good level of agreement between epidemiologists based on their independent evaluations, and good agreement between retrospective epidemiological review and larger documented outbreaks. Divergence between documented outbreaks at a national level and expert opinion only occurred for the smallest events identified by the epidemiologists, which are less likely to be of interest, or documented, at a national level.</p>", "<title>Descriptive statistics</title>", "<p>Major outbreaks of RRv disease occurred approximately every four years between 1991 and 2004. Case notifications often remain elevated for 6 months following seasonal outbreaks, which generally occurred with a mid-point around the month of March. The multiple-year disease cycle was associated with large long term variation in baseline statistics due to the timing of the epidemic and inter-epidemic years. There was no clear trend towards increased case counts during the period analysed.</p>", "<p>The mean of the daily count of RRv disease case notifications (1.53) was considerably lower than the variance (13.76), indicating that the data are overdispersed. When epidemiologically-defined outbreak periods were excluded (n = 2636), the mean RRv disease case count (0.26) remained lower than the variance (0.31). The mean number of cases reported per day for each expert-identified outbreak period varied between less than 1 case per day for smaller outbreaks in 1993 and 1995, to more than 8 cases per day for a large outbreak in 1996. The number of cases reported on the first day of identified outbreaks ranged between 2 and 4. The historical data are displayed in Figure ##FIG##0##1## and Additional File ##SUPPL##1##2##, with the division of the data into 15 'outbreak' datasets based on equal bisection of each of the intervening non-outbreak periods.</p>", "<p>On average, the use of longer baseline periods for the negative binomial cusum was associated with decreased variation in the out of control mean, and a higher average out of control mean. The negative binomial cusum out of control mean for the 7-day baseline model (4.56) was lower than the mean for the 14-day (4.64), 28-day (4.74) and 56-day (4.97) baseline models. The maximum value of the negative binomial cusum out of control mean for the 7-day baseline model (60.1) was higher than the maximum for the 14-day (55.2), 28-day (46.1) and 56-day (40.1) baseline models.</p>", "<p>Large EARS and negative binomial cusum algorithm signals occurred on a number of days which were not classified as of epidemiological importance. Most commonly these false alarms occurred following isolated small-scale increases in notifications, or increases prior to a defined outbreak period. In July 2001 (Additional File ##SUPPL##1##2##: 11^thoutbreak dataset, day 279) a large signal was issued by all algorithms on a day where an isolated spike of one then four cases were reported.</p>", "<title>Sensitivity</title>", "<p>Based on epidemiological opinion, the sensitivity of the EARS and negative binomial cusums were more similar at higher false alarm rates, as illustrated in Figure ##FIG##1##2##. At false alarm rates of less than 0.01, the negative binomial cusum had higher sensitivity than the EARS algorithms, and the sensitivity of the negative binomial cusum was generally higher when longer baseline periods were used. At a false alarm level of 0.005 the 28-day negative binomial cusum also had the highest sensitivity during the first 7 days of the outbreak (60%), followed by the C3 algorithm and the 7-day negative binomial cusum (both 40%) (Table ##TAB##0##1##). At a false alarm rate of approx 1 every 1000 days (0.001), the 7-day negative binomial cusum had the highest 7-day sensitivity (27%) (Table ##TAB##1##2##).</p>", "<p>Comparison of the sensitivity of the seven cusum algorithms found significant differences at the 0.005 and 0.001 false alarm levels (Friedman χ²= 25.8, df = 6, p = 0.0002 and Friedman χ²= 39.2, df = 6, p &lt; 0.0001 respectively, Tables ##TAB##0##1## and ##TAB##1##2##). When the sensitivity of the EARS algorithms were individually compared with the sensitivity of the 7-day and 28-day negative binomial cusums at the 0.005 false alarm level, no comparisons were significant following adjustment for the number of comparisons performed, with p ≤ 0.02 for all comparisons with the 28-day negative binomial cusum, and p ≤ 0.07 for all comparisons with the 7-day negative binomial cusum. Multiple comparisons at the 0.001 false alarm level were significant for EARS C1 and the 7-day and 28-day negative binomial cusums, and C2 and the 28-day negative binomial cusum following adjustment for the number of comparisons performed (p = 0.006, p = 0.002 and p = 0.006 respectively), and p ≤ 0.04 for all remaining comparisons.</p>", "<title>Timeliness</title>", "<p>The timeliness of the EARS algorithms exceeded that of the negative binomial cusum given that the outbreak was detected within the first 7 days (Tables ##TAB##0##1## and ##TAB##1##2##). The negative binomial cusum was more likely than the EARS algorithms to signal between days 2 and 7. The negative binomial cusum also generally had a greater overall median time to detection at the 0.005 false alarm rate (Table ##TAB##0##1##), although the overall median time to detection for the negative binomial cusum was less than the EARS algorithms at the 0.001 false alarm level (Table ##TAB##1##2##). The 28-day negative binomial cusum was the most timely negative binomial cusum model tested based on all outbreaks detected at the 0.005 and 0.001 false alarm levels.</p>", "<p>Comparison of the timeliness of the seven cusum algorithms found no significant difference at the 0.005 and 0.001 false alarm levels (Friedman χ²= 3.8, df = 6, p = 0.71 and Friedman χ²= 2.1, df = 6, p = 0.91 respectively, Tables ##TAB##0##1## and ##TAB##1##2##). However, comparison of the adjusted timeliness of the seven cusum algorithms, which allocates the duration of the outbreak as the timeliness result when outbreaks are undetected, found a significant difference at the 0.005 and 0.001 levels (Friedman χ²= 17.3, p = 0.008 and Friedman χ²= 25.0, p = 0.0003 respectively, Tables ##TAB##0##1## and ##TAB##1##2##). Multiple comparisons at both the 0.005 and 0.001 false alarm levels were only significant for EARS C1 and the 28-day negative binomial cusum following adjustment for the number of comparisons performed (p = 0.001 and p = 0.007 respectively), with p ≤ 0.05 for all remaining comparisons.</p>", "<title>AUC</title>", "<p>Comparison of the AUC analyses for the seven cusum algorithms found a significant difference for the 0–1, 0–0.01, and the 0–0.005 false alarm ranges (Friedman χ²= 70.3, p &lt; 0.0001, Friedman χ²= 70.6, p &lt; 0.0001 and Friedman χ²= 61.9, p &lt; 0.0001 respectively, Table ##TAB##2##3##). With the exception of the comparison of EARS C3 and the 7-day negative binomial cusum (p = 0.01 and p = 0.04 respectively), all multiple comparisons were significant after adjustment for repeated testing for the false alarm ranges 0–1 and 0–0.01 (all p ≤ 0.002). All multiple comparisons were significant after adjustment for repeated testing for the false alarm range 0–0.005 (all p ≤ 0.005).</p>", "<p>Comparison of the weighted AUC analyses for the seven cusum algorithms found a significant difference for the 0–1 and 0–0.01 false alarm ranges (Friedman χ²= 12.8, p = 0.046 and Friedman χ²= 18.4, p = 0.006 respectively, Table ##TAB##2##3##). However, no multiple comparisons were significant after adjustment for repeated testing for the false alarm ranges 0–1 (all 0.17 ≤ p ≤ 0.85), and 0–0.01 (C1 all p ≤ 0.02, C2 all p ≤ 0.08, C3 all p ≤ 1.00). Comparison of the weighted AUC analyses for the seven cusum algorithms found no significant difference for the 0–0.005 false alarm range (Friedman χ²= 10.7, p = 0.10).</p>", "<title>Negative binomial cusum calibration</title>", "<p>To investigate the influence of the negative binomial cusum settings on outbreak detection performance, we also evaluated the performance of the negative binomial cusum when the out of control state was defined as 3 standard deviations greater than the mean. Overall the timeliness and 7-day sensitivity of both the 2 and 3 standard deviation negative binomial cusum algorithms were comparable, with descriptive performance characteristics summarised in Table ##TAB##3##4##. At a false alarm level of 0.001 both the longer baseline 28-day and 56-day negative binomial cusums had an overall sensitivity of 80% based on a 3 standard deviation control limit.</p>" ]

[ "<title>Discussion</title>", "<p>This study explored the agreement between retrospective epidemiological opinion and the performance of two types of cusum algorithms for the detection of RRv disease outbreaks. The negative binomial cusum algorithm showed greater congruence with epidemiological opinion in terms of sensitivity, with the EARS algorithms having significantly lower AUC scores than the negative binomial cusum. However, when timeliness was incorporated into the AUC analyses, multiple comparisons between the EARS algorithms and the negative binomial cusum were no longer significant. This finding is associated with the ability of the EARS algorithms to signal an outbreak earlier if the outbreak was detected within the first seven days.</p>", "<p>Our findings suggest that the use of a negative binomial distribution allows accommodation of the over-dispersion evident in disease notification data, and provides a lower rate of false alarms for a given sensitivity. However, improved sensitivity is associated with decreased early timeliness performance, particularly at higher false alarm rates. For the surveillance of RRv disease notification data, the improved sensitivity offered by the negative binomial cusum may outweigh the decrease in the timeliness of detection, as formal epidemiological review of the notification data does not routinely occur daily. Although the results of this study are based on a limited amount of historical data, they indicate the importance of examining the correspondence between the underlying model of the algorithm used and the data being monitored.</p>", "<p>Despite the tendency of the EARS algorithms to have comparably high false alarm rates [##UREF##3##8##,##REF##18240128##22##], previous studies have found cusums helpful for predicting and monitoring trends in influenza surveillance data [##REF##15958924##1##,##REF##16926216##23##]. The EARS system has been well-received as a method of automated monitoring as it is implemented using software likely to be familiar to epidemiologists, is relatively straightforward to set up, and has been favoured for its flexibility [##REF##15958924##1##]. Selection of the most appropriate algorithms to implement in a specific surveillance context will depend on the surveillance objectives as well as disease-specific and operational considerations. These factors will determine the relative importance of the speed of detection, sensitivity and acceptable false alarm levels. In the case of cusums being applied to routinely collected disease notification data, the cost of false alarms may not be high if they are linked with procedures which facilitate efficient epidemiological review of relevant data to determine if further investigation or heightened monitoring is warranted.</p>", "<p>Unlike a large-scale simulation study which found that algorithms, including the EARS C1, C2 and C3, did not reliably detect outbreaks of interest across a wide range of scenarios [##REF##17331250##24##], our findings suggest that for RRv disease, a negative binomial cusum can reliably identify events of interest, although this method does not generally provide the same capability for the early detection of outbreaks as the EARS algorithms. The large nature of many RRv disease outbreaks, and the generally low level of baseline disease activity in the absence of recognised outbreaks provides conditions that have been associated with the more consistent detection of outbreaks using automated analysis methods [##REF##17331250##24##]. Large-scale simulation studies are required to more fully define specific differences in the detection characteristics of the EARS and negative binomial cusum algorithms.</p>", "<p>A limitation of this analysis in the context of evaluating the performance of cusum models is that cusums are optimal for the early detection of small sustained increases in the data being monitored [##UREF##0##2##], and the epidemiologist-nominated outbreak start dates may occur after the time of cusum signalling. As such, the methods used are likely to have resulted in the interpretation of several early outbreak signals as false alarms. An examination of false alarms occurring within 7 days of the commencement of the outbreaks revealed no consistent signalling of any single algorithm prior to the outbreak start dates, indicating that the specific start dates selected are unlikely to have biased the evaluation in favour of any particular algorithm.</p>", "<p>The evaluation of outbreak detection algorithms in the absence of a well-accepted gold standard is challenging, and the small retrospective nature of this study is an important limitation of the current analysis. RRv disease notifications generally exhibit well defined outbreaks which are reasonably easily identified retrospectively; however, retrospective evaluation is unable to provide a definitive indicator of whether or when an outbreak has occurred, particularly for previously unrecognised or smaller-scale events. Furthermore, retrospective epidemiological judgements are not dependent on the case notification counts alone, but were made in the context of the epidemiologists' considerable history of experience in infectious disease epidemiology in Australia, and prior knowledge of the occurrence and epidemiological significance of previous outbreaks.</p>", "<p>Despite the limitations of retrospective analysis, investigation of the correspondence between algorithm signals and epidemiological opinion can provide important information about the potential usefulness of outbreak detection algorithms in practice. Algorithms that are potentially useful for disease control can be expected to signal in a way that is consistent with retrospective epidemiological opinion. This investigation indicates that cusum algorithms can produce signals that are consistent with epidemiological opinion through the identification of a high proportion of RRv disease outbreaks at relatively low false alarm levels. The automated analysis of disease notification data has advantages associated with the consistency of data review, particularly given the large number of diseases under routine surveillance and the lack of resources for regular review of routinely collected data in the absence of other indications for heightened monitoring.</p>", "<p>An advantage of the cusum implementations evaluated here is that signalling departures from the recent past removes the need to specify fixed parameters to describe the baseline level of disease activity, although there remains a need to define the amount of baseline data to be used for baseline estimation and the magnitude of shift of interest, which can both influence performance. Consistent with previous work [##REF##18240128##22##], our findings demonstrate that the amount of historical data used to estimate the baseline level can have an important effect on algorithm performance. A 28-day baseline period for the negative binomial cusum generally produced improved overall performance with respect to sensitivity and timeliness in the case of RRv disease when compared with a 7-day baseline.</p>", "<p>The challenge in designing algorithms to facilitate disease surveillance is to ensure that they quickly and consistently detect events of interest in the data. The evaluation of algorithms using historical data provides valuable information about performance in routine surveillance applications, contingent upon the outbreak definitions used. However, prospective studies are required to evaluate the value of automated surveillance systems in practice. Although this evaluation used a large proportion of the available historical data, performance evaluation was only based on the detection of 15 outbreaks. Further work is required to investigate the performance of negative binomial cusums more systematically using both large sample and prospective methods, and investigate the integration of this monitoring approach with other methods which may improve the sensitivity and timeliness of alerts and minimise the false alarms generated. Given the strong evidence of seasonal trends in RRv disease notifications, and previous studies indicating that time series approaches provide an effective method for automated surveillance [##REF##16926216##23##,##REF##10602151##25##], the investigation of these methods in future studies may also be worthwhile.</p>" ]

[ "<title>Conclusion</title>", "<p>We found reasonable agreement between negative binomial cusum performance and retrospective epidemiological opinion at low false alarm rates. The negative binomial cusum had a significantly greater ability to identify outbreaks of RRv disease that are considered epidemiologically significant than the EARS algorithms. However, when the timeliness of outbreak detection was considered in addition to sensitivity, there were no significant differences found between the performance of individual EARS and negative binomial cusum algorithms. Given that an outbreak was detected within the first seven days, the EARS algorithms were able to detect outbreaks more quickly when compared with the negative binomial cusum algorithms. Further work is required to explore the performance differences between the cusum models and determine if the application of these methods for automated surveillance are useful in practice.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>The automated monitoring of routinely collected disease surveillance data has the potential to ensure that important changes in disease incidence are promptly recognised. However, few studies have established whether the signals produced by automated monitoring methods correspond with events considered by epidemiologists to be of public health importance. This study investigates the correspondence between retrospective epidemiological evaluation of notifications of Ross River virus (RRv) disease in Western Australia, and the signals produced by two cumulative sum (cusum)-based automated monitoring methods.</p>", "<title>Methods</title>", "<p>RRv disease case notification data between 1991 and 2004 were assessed retrospectively by two experienced epidemiologists, and the timing of identified outbreaks was compared with signals generated from two different types of cusum-based automated monitoring algorithms; the three Early Aberration Reporting System (EARS) cusum algorithms (C1, C2 and C3), and a negative binomial cusum.</p>", "<title>Results</title>", "<p>We found the negative binomial cusum to have a significantly greater area under the receiver operator characteristic curve when compared with the EARS algorithms, suggesting that the negative binomial cusum has a greater level of agreement with epidemiological opinion than the EARS algorithms with respect to the existence of outbreaks of RRv disease, particularly at low false alarm rates. However, the performance of individual EARS and negative binomial cusum algorithms were not significantly different when timeliness was also incorporated into the area under the curve analyses.</p>", "<title>Conclusion</title>", "<p>Our retrospective analysis of historical data suggests that, compared with the EARS algorithms, the negative binomial cusum provides greater sensitivity for the detection of outbreaks of RRv disease at low false alarm levels, and decreased timeliness early in the outbreak period. Prospective studies are required to investigate the potential usefulness of these algorithms in practice.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>REW and AJP designed the study, REW conducted the study and drafted the manuscript, and SE, GW and BV were involved in revising the manuscript. All surviving authors read and approved the final manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1472-6947/8/37/prepub\"/></p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>This study was funded by a project grant from the Australian Biosecurity Cooperative Research Centre for Emerging Infectious Disease and a NHMRC Capacity Building Grant in Population Health #358424. We thank the Department of Health Western Australia for providing the data for analysis, and the two expert epidemiologists for their review of the data. We dedicate this work to our greatly missed colleague and much loved friend Aileen J Plant.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Ross River virus case notifications, expert-defined outbreak period (shaded) and cusum scores by day for the first outbreak dataset (days 1–235).</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>Sensitivity of Early Aberration Reporting System (EARS) and negative binomial cusum (NBC) algorithms according to false alarm rate.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Algorithm summary performance statistics for false alarm rates approximating 0.005^†</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Algorithm</td><td align=\"center\">false alarm rate</td><td align=\"center\">median (mean) sensitivity</td><td align=\"center\">Day 1 sensitivity</td><td align=\"center\">Day 2 sensitivity</td><td align=\"center\">Days 1–7 sensitivity</td><td align=\"center\">CARL</td><td align=\"center\">median (mean) timeliness^‡</td><td align=\"center\">median (mean) adjusted timeliness^‡</td></tr></thead><tbody><tr><td align=\"left\">EARS C1</td><td align=\"center\">0.0049</td><td align=\"center\">1 (0.53)</td><td align=\"center\">0.13</td><td align=\"center\">0</td><td align=\"center\">0.13</td><td align=\"center\">1.0</td><td align=\"center\">18.5 (54.6)</td><td align=\"center\">34.0 (78.8)</td></tr><tr><td align=\"left\">EARS C2</td><td align=\"center\">0.0042</td><td align=\"center\">1 (0.53)</td><td align=\"center\">0.33</td><td align=\"center\">0</td><td align=\"center\">0.33</td><td align=\"center\">1.0</td><td align=\"center\">0.0 (24.1)</td><td align=\"center\">32.0 (71.1)</td></tr><tr><td align=\"left\">EARS C3</td><td align=\"center\">0.0049</td><td align=\"center\">1 (0.60)</td><td align=\"center\">0.40</td><td align=\"center\">0</td><td align=\"center\">0.40</td><td align=\"center\">1.0</td><td align=\"center\">0.0 (17.9)</td><td align=\"center\">32.0 (68.1)</td></tr><tr><td align=\"left\">NBC 7-day</td><td align=\"center\">0.0049</td><td align=\"center\">1 (0.87)</td><td align=\"center\">0.2</td><td align=\"center\">0.07</td><td align=\"center\">0.40</td><td align=\"center\">2.0</td><td align=\"center\">12.0 (16.3)</td><td align=\"center\">14.0 (18.5)</td></tr><tr><td align=\"left\">NBC 14-day</td><td align=\"center\">0.0049</td><td align=\"center\">1 (0.93)</td><td align=\"center\">0.2</td><td align=\"center\">0.07</td><td align=\"center\">0.33</td><td align=\"center\">1.6</td><td align=\"center\">9.5 (13.9)</td><td align=\"center\">11.0 (15.2)</td></tr><tr><td align=\"left\">NBC 28-day</td><td align=\"center\">0.0049</td><td align=\"center\">1 (1.00)</td><td align=\"center\">0.27</td><td align=\"center\">0.07</td><td align=\"center\">0.60</td><td align=\"center\">2.9</td><td align=\"center\">5.0 (8.5)</td><td align=\"center\">5.0 (8.5)</td></tr><tr><td align=\"left\">NBC 56-day</td><td align=\"center\">0.0042</td><td align=\"center\">1 (0.93)</td><td align=\"center\">0.13</td><td align=\"center\">0</td><td align=\"center\">0.27</td><td align=\"center\">3.0</td><td align=\"center\">14.5 (19.1)</td><td align=\"center\">14.0 (18.8)</td></tr><tr><td align=\"left\">p-value^¥</td><td align=\"center\">-</td><td align=\"center\">0.0002</td><td align=\"center\">-</td><td align=\"center\">-</td><td align=\"center\">-</td><td align=\"center\">-</td><td align=\"center\">0.71</td><td align=\"center\">0.008</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Algorithm summary performance statistics for false alarm rates approximating 0.001^†</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Algorithm</td><td align=\"center\">false alarm rate</td><td align=\"center\">median (mean) sensitivity</td><td align=\"center\">Day 1 sensitivity</td><td align=\"center\">Day 2 sensitivity</td><td align=\"center\">Days 1–7 sensitivity</td><td align=\"center\">CARL</td><td align=\"center\">median (mean) timeliness^‡</td><td align=\"center\">median (mean) adjusted timeliness^‡</td></tr></thead><tbody><tr><td align=\"left\">EARS C1</td><td align=\"center\">0.0008</td><td align=\"center\">0 (0.27)</td><td align=\"center\">0.13</td><td align=\"center\">0</td><td align=\"center\">0.13</td><td align=\"center\">1.0</td><td align=\"center\">36.5 (49.0)</td><td align=\"center\">123 (113.1)</td></tr><tr><td align=\"left\">EARS C2</td><td align=\"center\">0.0008</td><td align=\"center\">0 (0.40)</td><td align=\"center\">0.13</td><td align=\"center\">0</td><td align=\"center\">0.13</td><td align=\"center\">1.0</td><td align=\"center\">44.5 (52.7)</td><td align=\"center\">90 (96.0)</td></tr><tr><td align=\"left\">EARS C3</td><td align=\"center\">0.0008</td><td align=\"center\">0 (0.47)</td><td align=\"center\">0.13</td><td align=\"center\">0</td><td align=\"center\">0.13</td><td align=\"center\">1.0</td><td align=\"center\">51.0 (52.4)</td><td align=\"center\">73 (93.4)</td></tr><tr><td align=\"left\">NBC 7-day</td><td align=\"center\">0.0008</td><td align=\"center\">1 (0.80)</td><td align=\"center\">0.13</td><td align=\"center\">0.07</td><td align=\"center\">0.27</td><td align=\"center\">1.8</td><td align=\"center\">19.5 (32.3)</td><td align=\"center\">21 (31.1)</td></tr><tr><td align=\"left\">NBC 14-day</td><td align=\"center\">0.0008</td><td align=\"center\">1 (0.80)</td><td align=\"center\">0.07</td><td align=\"center\">0</td><td align=\"center\">0.13</td><td align=\"center\">2.0</td><td align=\"center\">33.5 (39.0)</td><td align=\"center\">32 (36.5)</td></tr><tr><td align=\"left\">NBC 28-day</td><td align=\"center\">0.0008</td><td align=\"center\">1 (0.93)</td><td align=\"center\">0.07</td><td align=\"center\">0</td><td align=\"center\">0.2</td><td align=\"center\">3.0</td><td align=\"center\">16.5 (26.3)</td><td align=\"center\">18 (26.8)</td></tr><tr><td align=\"left\">NBC 56-day</td><td align=\"center\">0.0008</td><td align=\"center\">1 (0.80)</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">0.13</td><td align=\"center\">5.0</td><td align=\"center\">19.0 (31.0)</td><td align=\"center\">20 (30.1)</td></tr><tr><td align=\"left\">p-value^¥</td><td align=\"center\">-</td><td align=\"center\">&lt;0.0001</td><td align=\"center\">-</td><td align=\"center\">-</td><td align=\"center\">-</td><td align=\"center\">-</td><td align=\"center\">0.91</td><td align=\"center\">0.0003</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Descriptive statistics for summary performance characteristics for the EARS and NBC algorithms: receiver operating characteristic area under the curve (AUC) analyses</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Algorithm</td><td align=\"center\">median (mean) AUC_0–1</td><td align=\"center\">median (mean) AUC_0–0.01^†</td><td align=\"center\">median (mean) AUC_0–0.005^‡</td><td align=\"center\">median (mean) WtAUC_0–1</td><td align=\"center\">median (mean) WtAUC_0–0.01^‡</td></tr></thead><tbody><tr><td align=\"left\">EARS C1</td><td align=\"center\">0.995 (0.989)</td><td align=\"center\">0.0011 (0.0019)</td><td align=\"center\">0.0004 (0.0015)</td><td align=\"center\">0.959 (0.968)</td><td align=\"center\">&lt;0.0001 (0.0008)</td></tr><tr><td align=\"left\">EARS C2</td><td align=\"center\">0.996 (0.990)</td><td align=\"center\">0.0017 (0.0026)</td><td align=\"center\">0.0002 (0.0016)</td><td align=\"center\">0.994 (0.978)</td><td align=\"center\">0.0002 (0.0013)</td></tr><tr><td align=\"left\">EARS C3</td><td align=\"center\">0.996 (0.993)</td><td align=\"center\">0.0061 (0.0059)</td><td align=\"center\">0.0011 (0.0022)</td><td align=\"center\">0.991 (0.980)</td><td align=\"center\">0.0011 (0.0033)</td></tr><tr><td align=\"left\">NBC 7-day</td><td align=\"center\">1.0 (0.998)</td><td align=\"center\">0.0095 (0.0081)</td><td align=\"center\">0.0049 (0.0040)</td><td align=\"center\">0.990 (0.979)</td><td align=\"center\">&lt;0.0001 (0.0032)</td></tr><tr><td align=\"left\">NBC 14-day</td><td align=\"center\">1.0 (0.999)</td><td align=\"center\">0.0099 (0.0088)</td><td align=\"center\">0.0049 (0.0040)</td><td align=\"center\">0.988 (0.981)</td><td align=\"center\">&lt;0.0001 (0.0026)</td></tr><tr><td align=\"left\">NBC 28-day</td><td align=\"center\">1.0 (1.0)</td><td align=\"center\">0.0095 (0.0092)</td><td align=\"center\">0.0049 (0.0047)</td><td align=\"center\">0.990 (0.983)</td><td align=\"center\">0.0014 (0.0032)</td></tr><tr><td align=\"left\">NBC 56-day</td><td align=\"center\">1.0 (0.999)</td><td align=\"center\">0.0095 (0.0088)</td><td align=\"center\">0.0042 (0.0036)</td><td align=\"center\">0.977 (0.928)</td><td align=\"center\">&lt;0.0001 (0.0018)</td></tr><tr><td align=\"left\">p-value^¥</td><td align=\"center\">&lt;0.0001</td><td align=\"center\">&lt;0.0001</td><td align=\"center\">&lt;0.0001</td><td align=\"center\">0.046</td><td align=\"center\">0.006</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Summary performance statistics for false alarm rates approximating 0.005^†for the negative binomial cusum with an out of control state defined as 3 standard deviations greater than the mean</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Algorithm</td><td align=\"center\">false alarm rate</td><td align=\"center\">mean sensitivity</td><td align=\"center\">Day 1 sensitivity</td><td align=\"center\">Day 2 sensitivity</td><td align=\"center\">Days 1–7 sensitivity</td><td align=\"center\">CARL</td><td align=\"center\">median timeliness^‡</td></tr></thead><tbody><tr><td align=\"left\">NBC 7-day</td><td align=\"center\">0.0049</td><td align=\"center\">0.80</td><td align=\"center\">0.2</td><td align=\"center\">0.07</td><td align=\"center\">0.33</td><td align=\"center\">1.6</td><td align=\"center\">13.0</td></tr><tr><td align=\"left\">NBC 14-day</td><td align=\"center\">0.0049</td><td align=\"center\">0.97</td><td align=\"center\">0.2</td><td align=\"center\">0.07</td><td align=\"center\">0.33</td><td align=\"center\">1.6</td><td align=\"center\">12.0</td></tr><tr><td align=\"left\">NBC 28-day</td><td align=\"center\">0.0049</td><td align=\"center\">1.00</td><td align=\"center\">0.27</td><td align=\"center\">0.07</td><td align=\"center\">0.60</td><td align=\"center\">3.0</td><td align=\"center\">6.0</td></tr><tr><td align=\"left\">NBC 56-day</td><td align=\"center\">0.0038</td><td align=\"center\">1.00</td><td align=\"center\">0.27</td><td align=\"center\">0.07</td><td align=\"center\">0.40</td><td align=\"center\">1.5</td><td align=\"center\">12.0</td></tr></tbody></table></table-wrap>" ]

[ "<disp-formula>c₀= u/(σ²- u)</disp-formula>", "<disp-formula>r = u²/(σ²- u)</disp-formula>", "<disp-formula>C₀= 0</disp-formula>", "<disp-formula>C_n⁺= max(0, C_n-1⁺+ X_n- k⁺)</disp-formula>", "<disp-formula>where k⁺= r.ln [c₀(1+c₁)/c₁(1+c₀)]/ln [(1+c₀)/(1+c₁)]</disp-formula>" ]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional file 1</title><p>R code for Early Aberration Reporting System (EARS) and negative binomial cusum (NBC) algorithms.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S2\"><caption><title>Additional file 2</title><p>RRv case notifications, expert-defined outbreak period and cusum scores by day for outbreak datasets 2 to 15.</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><p>^†Test threshold that produced a false alarm rate ≤ 0.005</p><p>^‡Detection on the first outbreak day is equivalent to a timeliness of 0 days</p><p>^¥Friedman rank sum test</p><p>CARL: Conditional Average Run Length – conditional on the detection of the outbreak during the first 7 days</p><p>EARS: Early Aberration Reporting System</p><p>NBC: Negative binomial cusum with an out of control state defined as 2 standard deviations greater than the mean</p></table-wrap-foot>", "<table-wrap-foot><p>^†Test threshold that produced a false alarm rate ≤ 0.001</p><p>^‡Detection on the first outbreak day is equivalent to a timeliness of 0 days</p><p>^¥Friedman rank sum test</p><p>CARL: Conditional Average Run Length – conditional on the detection of the outbreak during the first 7 days</p><p>EARS: Early Aberration Reporting System</p><p>NBC: Negative binomial cusum with an out of control state defined as 2 standard deviations greater than the mean</p></table-wrap-foot>", "<table-wrap-foot><p>^†Evaluated at a false alarm rate ≤ 0.01</p><p>^‡Evaluated at a false alarm rate ≤ 0.005</p><p>^¥Friedman rank sum test</p><p>WtAUC: Weighted AUC using a reference value of 7 days for valid detection</p><p>EARS: Early Aberration Reporting System</p><p>NBC: Negative binomial cusum with an out of control state defined as 2 standard deviations greater than the mean</p></table-wrap-foot>", "<table-wrap-foot><p>^†Test threshold that produced a false alarm rate ≤ 0.00</p><p>^‡Detection on the first outbreak day is equivalent to a timeliness of 0 days</p><p>CARL: Conditional Average Run Length – conditional on the detection of the outbreak during the first 7 days</p><p>NBC: Negative binomial cusum</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1472-6947-8-37-1\"/>", "<graphic xlink:href=\"1472-6947-8-37-2\"/>" ]

[ "<media xlink:href=\"1472-6947-8-37-S1.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1472-6947-8-37-S2.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Henoch Schonlein purpura (HSP) is a well known systemic small vessel vasculitis characterized by major manifestations; arthritis, nonthrombocytopenic purpura, abdominal pain and renal disease. The latter is of major concern since it may result in lifelong problems. Although the pathogenesis of HSP remains largely unknown there is strong evidence that IgA has a pivotal role given the increased serum IgA concentrations, IgA with concomitant circulating immune complexes and IgA deposition in vessel walls and renal mesangium have been observed during the course of the disease. Furthermore, activation of the alternative pathway of complement and cytokine abnormalities has also been implicated as having a role in HSP pathogenesis [##UREF##0##1##].</p>", "<p>Endothelins (ETs) were first found by Yanagisawa et al. in 1988 [##REF##2451132##2##] and 3 isoforms of this peptide, ET-1, ET-2, ET-3, have been isolated [##UREF##1##3##]. Their biological activities cover a wide spectrum which includes regulation of hormones and neurotransmitter, cellular growth and proliferation, bronchoconstriction, natriuresis and water diuresis (4). Urine contains higher concentrations of ET compared to those of plasma which is mainly derived from the in situ production by the kidneys [##REF##9260260##4##].</p>", "<p>Endothelin-1 (ET-1) is a potent vasoconstrictor and its concentrations in plasma are increased markedly in a number of pathologies, such as ischemia induced damage and reperfusion, vasculities of various types, congestive heart failure, systemic inflammatory response seen in septic shock syndrome and similar pathology [##REF##8907584##5##]. In a study by Muslu A et al. it was shown that ET-1 plasma levels were significantly higher in HSP patients during the acute phase compared to levels in remission but also to levels in healthy controls [##REF##12432435##6##].</p>", "<p>The purpose of this study was to determine whether ET-1 levels in plasma and urine are related to the severity of the clinical presentation and the outcome of HSP.</p>" ]

[ "<title>Methods</title>", "<p>Thirty patients with HSP were recruited during a 2 year period from January 2005 to December 2006. The control group consisted of an equal number of healthy children matched for age and gender. The age range was 2–12.6 years with a mean of 6.3 ± 3 years. The respective ages for the controls were 2–12.7 years and 6.2 ± 2.6 years. Male to female ratio was 14/16 in both groups. Informed consent was obtained from the parents of all participants and the study was approved by the Local Ethics Committee. The diagnosis of HSP was based on the criteria established by the American College of Rheumatology [##REF##2202310##7##]. A punch skin biopsy was performed in all patients in order to verify the diagnosis. Biopsies were obtained from affected and non affected skin within 48 hours of the appearance of the lesions.</p>", "<p>A detailed history and a complete physical examination were obtained from all patients. A clinical scoring system adjusted from De Matia D et al. and Muslu A et al. which consisted of the sum of three distinct scores for arthritis, abdominal symptoms and renal involvement were used to assess disease activity and severity [##REF##12432435##6##,##REF##8580019##8##]. This score was modified by using more objective criteria (Table ##TAB##0##1##). Severity of the disease was determined as mild or severe, if the clinical score was ≤ 4 or &gt; 4, respectively</p>", "<p>Laboratory measurements included complete blood count, erythrocyte sedimentation rate (ESR), C reactive protein (CRP), total protein and albumin, blood and urine electrolytes, immunoglobulins (IgA, IgG, IgM), C3, C4, ASO, ANA, anti-DNA, Ra test, p-ANCA, blood urea nitrogen (BUN), creatinine, culture of pharyngeal swabbing and stool guiac test and routine urinalysis (which was performed every 15 days during the study period). In a random urine specimen taken at the same time with the creatinine blood sample, ET-1(expressed as creatinine ratio) and N-acetyl-b-D glucosaminidase (NAG- expressed as creatinine ratio), was measured as a sensitive marker of tubular damage. Creatinine, total protein and a1 and b2 microglobulin were also calculated in a 24 hours urine collection as tubular markers and microalbumin as a marker of glomerular damage. The children who presented with gross or mild haematuria and/or proteinuria and/or tubular or glomelural involvement were considered as having renal involvement.</p>", "<p>Venous blood samples for plasma ET-1 determination were collected in frozen plastic tubes containing EDTA (0.1 ml EDTA/ml of blood) after 30 min of supine rest. They were promptly centrifuged and stored in tubes containing aprotinine at -70°C until analyzed. Plasma ET-1 levels were determined by radioimmunoassay (RIA) (Endothelin 1–21 specific [¹²⁵I] columns (Amersham Biosciences Amprep™ 500 mg C2 columns). The sensitivity of the assay was 1.2 pg/ml. The intra and inter assay variability were 4.8% and 13.8% respectively. Urinary NAG levels were measured following a fluorometric method described by Leadback and Walker [##REF##13759894##9##] as well as by Woolen and Walker [##REF##5880047##10##] and subsequently modified by Gnanadurai et al [##REF##920915##11##]. The urinary creatinine levels were analyzed by Jaffe's method for calculating NAG/creatinine quotients [##REF##998546##12##].</p>", "<p>Patients were examined and underwent all the aforementioned laboratory tests at 3 distinct time points. The first at the time of diagnosis of HSP corresponding to the acute phase. The second, 1 month later or when renal involvement was diagnosed (which ever happened first) and the third 1 year later which roughly corresponded to the remission of the disease. Four patients continued to present with renal impairment even after 15, 24, 14 and 18 months respectively and for that reason their last examination did not correspond to their remission phase.</p>", "<title>Statistical methods</title>", "<p>Continuous variables are expressed as mean ± SD whereas time variables as median (p50), 25^thpercentile (p25) and 75^thpercentile (p75). In univariate analysis we compared means between different groups with t-test and one way Anova and we used Spearman's rho to search for any correlations between variables. In multivariate analysis, we used logistic regression to see if ET-1 was a predictive factor for renal damage and Cox Proportional Hazard models which were used to estimate relative risk [expressed as hazard Ratio (HR)] of remission associated with a unit change in a covariate, to explore for any correlations between ET-1 and the length of the disease. We used the backward procedure for selecting the best models. The statistical significance threshold was set at p &lt; 0.05.</p>" ]

[ "<title>Results</title>", "<p>Twenty-three (76.6%) of our patients had mild and 7 (23.4%) had severe disease. Rash was evident in all patients at presentation. In 12 (40%) of them it was preceded by: arthritis, in 6 (20%) by gastrointestinal complaints, <underline>in 5 (17%) hematuria in 1 (3%).</underline> Histological examination of the skin biopsies revealed typical leukocytoclastic vasculitis in all patients; whereas immunofluorescence studies showed perivascular IgA deposition in 28 (93.3%), C3 deposition in 13(43.3%) and IgM deposition in 5 (16.6%) patients.</p>", "<p>During the course of the disease, arthritis occurred in 25 (83%), gastrointestinal involvement in 24 (80%) and renal involvement in 18 (60%) patients. Blood pressure was within normal limits in all patients apart from 1 who developed nephritic syndrome.</p>", "<p>Only one patient who finally developed nephrotic syndrome, had low total protein and albumin. Four more patients had proteinuria, which was not at nephrotic levels and 10 microalbuminuria. Levels of NAG, a-1 and b-2 microglobulin were within normal limits in all patients. The clinical hallmark of HSP nephritis is haematuria. In our study 12/30 (40%) patients had microscopic and 5/30 (16.6%) developed gross haematuria. Four patients of the latter group finally developed proteinuria and two of them underwent renal biopsy which revealed glomerulonephritis characterized by diffuse hypercellularity and mesangial proliferation.</p>", "<p>Laboratory assays taken at presentation, were within normal limits with the exception of raised IgA in 14/30 (46%), elevated C₃values in 1/30 (3.3%) (Table ##TAB##1##2##), and high ASO titers (&gt; 200 IU/ml) in 17/30 (56%). The culture of pharyngeal swabbing was positive for group A beta hemolytic streptococcus in 11/30 (36%).</p>", "<p>Positive stool guiac test was detected in 16 (53.3%), haematemesis in 2 (6.6%), melena in 4 (13.3%) and intussusception in 1 patient. The duration of time variables expressed as median (P50) and 25^thand 75^thpercentile (p75, p25) in days were 8 (14,5) for the acute phase of the disease, 45(362,5) for microhaematuria and 34(92,5) for microalbuminuria.</p>", "<p>No differences were found in ET-1 levels in plasma and urine between patient and control groups at each of the three distinct time points (Table ##TAB##2##3##). Also there was no correlation between renal or overall clinical score and ET-1 values in plasma and urine on the 3 time points (Table ##TAB##3##4##, ##TAB##4##5##) between patients with and without renal involvement (Table ##TAB##5##6##).</p>", "<p>Of note, in order to investigate if any correlation existed between ET-1 in acute phase and renal involvement, we used two logistic regression models (one for ET-1 in plasma and one for ET-1 in urine) and in each one we adjusted for potential confounders namely, age, gender and clinical and laboratory parameters that were measured during the acute phase of the disease i.e. clinical score, ESR, CRP, IgA and C3. Our results show that renal involvement and ET-1 were not correlated (OR = 1.05, 95% CI 0.79–1.383, p = 0.74 and OR = 0.99 95% CI 0.95–1.03, p = 0.71 for ET-1 in plasma and urine respectively).</p>", "<p>Using Cox Proportional Hazards models and controlling for the same potential confounders as above, we tried to investigate whether ET-1 was correlated with the length of the acute phase of the disease. We found that there was no significant correlation with ET-1 in plasma but ET-1 in urine was a significant predictor for the length of the acute phase (HR = 1.01, p = 0.79, 95% CI 0.91–1.11 and HR = 0.98, p = 0.032, 95% CI 0.96–0.99 for ET-1 in plasma and urine respectively). The latter result implies that the higher the levels of ET-1 in urine, the longer the acute phase. A more precise expression of the latter result is that for one unit increase in ET-1 in urine, the mean rate of remission of acute phase (instantaneous remission rate) is reduced about 2%.</p>", "<p>We also investigated the relation between ET-1 levels during the acute phase and the length of microhaematuria and microalbuminuria adjusting for the same potential confounders as before. We found that ET-1 in plasma and urine did not correlate with renal involvement although ET-1 in plasma showed a trend toward a significant correlation with the duration of microhaematuria (H.R. = 1.51, p = 0.055, 95% CI 0.99–2.34).</p>" ]

[ "<title>Discussion</title>", "<p>HSP is the most common vasculitis in childhood. It is a multisystem disease most commonly affecting skin, joints, gastrointestinal tract and kidneys although other organs may also be affected [##REF##10086951##13##]. Its pathogenesis remains largely unknown. However, it is generally considered to be an immune complex-mediated disease characterized by the presence of polymeric IgA1 containing immune complexes which are mainly found in dermal, gastrointestinal and glomerular capillaries [##UREF##2##14##]. IgA aggregates or IgA complexes with complement deposits in target organs, result in elaboration of inflammatory mediators, including vascular prostaglandins such as prostacyclins which are thought to play a role in the pathogenesis of HSP vasculitis [##REF##2420289##15##].</p>", "<p>Alterations in the production of interleukins and growth factors may also have a role in the pathogenesis of HSP. Tumor necrosis factor (TNF), interleukin 1(IL-1) and interleukin-6 (IL-6) may mediate the inflammatory process present in HSP [##UREF##0##1##,##REF##9844059##16##,##REF##8883024##17##]. There is only one study which demonstrated that plasma ET-1 levels in the acute phase were significantly higher compared to the remission phase and to the controls. However plasma ET-1 levels did not correlate with the clinical and laboratory findings with the exception of a minority of patients with severe disease [##REF##12432435##6##]. In our study we did not find any differences of plasma and urine ET-1 levels between patients and controls.</p>", "<p>ET was initially described as a peptide released from large vessel endothelial cells [##REF##2451132##2##]. However other sites of synthesis have subsequently been discovered. ET is also expressed in brain, lung, heart and kidney. As far as the latter location is concerned endothelial, glomerular, tubular epithelial and mesangial cells of the kidney can all synthesize ET [##REF##1656130##18##].</p>", "<p>The circulating half life of ET is short (2–3 min) and it is almost completely metabolized in the pulmonary capillaries by the enzyme neutral endopeptidase. A small amount of circulating ET is cleared in the urine [##REF##1618556##19##]. Depending on the nature and the magnitude of the stimulus circulating ET may be elevated in different pathophysiological or experimental conditions. However it is not clear whether an elevated level of circulating ET is necessary for its actions [##REF##1644793##20##]. The effects of ET are detected long after the normalization of the plasma level. These data suggest that the concentration of plasma ET does not correlate with its activity in different cells and tissues and the circulating levels depend on the balance of the luminal endothelial synthesis and metabolism [##REF##9260260##4##].</p>", "<p>The kidney and particularly the renal medulla expresses the highest concentration of ET-1 receptors of any organ and also synthesizes ET-1 [##REF##15838285##21##]. Glomerular epithelial and mesangial cells, renal tubular cells and medullary collecting duct cells synthesize and release ET-1. ET-1 is a major mediator of renal vascular tone, tubular secretion of electrolytes and water and vascular smooth muscle and mesangial cell proliferation [##REF##15466627##22##]. ET-1 maintains normal kidney perfusion by increasing renal blood flow, effecting on natriuresis and diuresis [##UREF##3##23##].</p>", "<p>Despite the higher concentration of ET in urine compared to plasma, the urinary excretion of ET does not correlate with the glomerular filtration rate filtered load or plasma levels [##REF##2002644##24##]. It has been suggested that the filtered ET is subject to proteolytic degradation by neutral endopeptidase across the brush border of the proximal tubuli [##REF##1618556##19##]. ET can be a marker of renal injury in different pathological processes in children and in adults but its specificity is low. Therefore ET should be used with caution as a marker of distal tubular or collecting duct epithelial function [##REF##9260260##4##]. ET/creatinine ratio in random urine samples could be used as a reliable index of urinary excretion of ET [##REF##9650466##25##,##REF##14523636##26##]. In our study this urinary ET-1 was correlated with the duration of the acute phase of the disease.</p>", "<p>Several systemic rheumatic diseases in which vascular pathology is related to endothelial cell activation appear to be <underline>causally</underline> related to endothelin.</p>", "<p>They include systemic sclerosis, systemic lupus erythematosus, Takayasu and giant cell arteritis, Raynaud's phenomenon, mixed cryoglobulinaemia, aortoarteritis and Buerger's, Bechcet, Kawasaki, mixed connective tissue and Chaga's disease [##REF##7625853##27##, ####REF##8158316##28##, ##UREF##4##29##, ##REF##8793164##30##, ##REF##9132327##31##, ##REF##8357099##32##, ##REF##7586267##33##, ##UREF##5##34##, ##REF##1352096##35##, ##REF##11016811##36##, ##REF##11755863##37##, ##REF##11334935##38####11334935##38##]. All these studies measured only plasma ET-1 levels in contrast to our study which measured plasma and urine ET-1 levels (although no differences between patients and controls were found).</p>" ]

[ "<title>Conclusion</title>", "<p>In conclusion, plasma ET-1 levels cannot be utilized as a predictor of HSP duration or severity whereas urine ET-1 levels are correlated with the duration of the acute phase of the disease.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Henoch Schonlein purpura (HSP) is a common vasculitis of small vessels whereas endothelin-1 (ET-1) is usually reported elevated in vasculities and systematic inflammation. The aim of the present study was to investigate whether ET-1 levels are correlated with the clinical presentation and the outcome of HSP.</p>", "<title>Methods</title>", "<p>The study sample consisted of thirty consecutive patients with HSP. An equal number of healthy patients of similar age and the same gender were served as controls. The patients' age range was 2–12.6 years with a mean ± SD = 6.3 ± 3 years. All patients had a physical examination with a renal, and an overall clinical score. Blood and urinary biochemistry, immunology investigation, a skin biopsy and ET-1 measurements in blood and urine samples were made at presentation, 1 month later and 1 year after the appearance of HSP. The controls underwent the same investigation with the exception of skin biopsy.</p>", "<title>Results</title>", "<p>ET-1 levels in plasma and urine did not differ between patients and controls at three distinct time points. Furthermore the ET-1 were not correlated with the clinical score and renal involvement was independent from the ET-1 measurements. However, the urinary ET-1 levels were a significant predictor of the duration of the acute phase of HSP (HR = 0.98, p = 0.032, CI0.96–0.99). The ET-1 levels did not correlate with the duration of renal involvement.</p>", "<title>Conclusion</title>", "<p>Urinary ET-1 levels are a useful marker for the duration of the acute phase of HSP but not for the length of renal involvement.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>SF and PN had substantial contributions to conception and design and analysis and interpretation of data and drafting of the manuscript. DG and HG made all the necessary biochemical measurements MM and AF have been involved in drafting and revising the manuscript. KD performed the statistical analysis of the data</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2431/8/33/prepub\"/></p>" ]

[]

[]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>The clinical scoring system in patients with HSP (6, 8)</p></caption><table frame=\"hsides\" rules=\"groups\"><tbody><tr><td align=\"left\">Arthritis score</td></tr><tr><td align=\"left\">0 = No symptom</td></tr><tr><td align=\"left\">1 = arthralgia and/or slight swelling (normal walk)</td></tr><tr><td align=\"left\">2 = arthralgia and/or moderate swelling (limp walk)</td></tr><tr><td align=\"left\">3 = arthralgia and/or severe swelling (refuse to walk)</td></tr><tr><td/></tr><tr><td align=\"left\">Abdominal score</td></tr><tr><td align=\"left\">0 = No symptom</td></tr><tr><td align=\"left\">1 = mild abdominal pain(medically elicited) and/or occult blood in stool (+)</td></tr><tr><td align=\"left\">2 = moderate abdominal pain (transient complaints brought to medical attention) and/or occult blood in stool (++/+++)</td></tr><tr><td align=\"left\">3 = severe abdominal pain and/or melena and/or hematemesis and/or intussusception</td></tr><tr><td/></tr><tr><td align=\"left\">Renal score</td></tr><tr><td align=\"left\">0 = No proteinuria and/or 3–5 RBC/HPF</td></tr><tr><td align=\"left\">1 = proteinuria 30 mg/dl and/or microalbuminuria and/or 10–15 RBC/HPF)</td></tr><tr><td align=\"left\">2 = proteinuria 30–150 mg/dl and/or &gt; 50RBC/HPF</td></tr><tr><td align=\"left\">3 = proteinuria 150 mg/dl and/or macroscopic haematuria</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Clinical score and selected laboratory values in patients with HSP, at presentation (IgA, C₃) and at three distinct time points (ET-1).</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td/><td/><td/><td/><td align=\"left\" colspan=\"2\">1^sttime point</td><td align=\"left\" colspan=\"2\">2^ndtime point</td><td align=\"left\" colspan=\"2\">3^rdtime point</td></tr><tr><td align=\"left\">Patient No<break/><break/></td><td align=\"left\">Age (years) <break/>/Gender<break/>(M/F)</td><td align=\"left\">IgA <break/>(mg/dl)<break/></td><td align=\"left\">C3 <break/>(mg/dl)<break/></td><td align=\"left\">Clinical <break/>score<break/></td><td align=\"left\">ET-1 <break/> in plasma<break/>(pg/ml)</td><td align=\"left\">ET-1 <break/>in urine <break/>(pg/ml)</td><td align=\"left\">ET-1 <break/>in plasma <break/>(pg/ml)</td><td align=\"left\">ET-1 <break/>in urine <break/>(pg/ml)</td><td align=\"left\">ET-1 <break/>in plasma<break/>(pg/ml)</td><td align=\"left\">ET-1 <break/>in urine<break/>(pg/ml)</td></tr></thead><tbody><tr><td align=\"left\">1</td><td align=\"left\">5./F</td><td align=\"left\">261</td><td align=\"left\">204</td><td align=\"left\">2</td><td align=\"left\">15.388</td><td align=\"left\">9.147</td><td align=\"left\">16.2</td><td align=\"left\">40.112</td><td align=\"left\">16.57</td><td align=\"left\">39.742</td></tr><tr><td align=\"left\">2</td><td align=\"left\">9.5/F</td><td align=\"left\">162</td><td align=\"left\">141</td><td align=\"left\">1</td><td align=\"left\">17.568</td><td align=\"left\">22.843</td><td align=\"left\">16.75</td><td align=\"left\">60.149</td><td align=\"left\">16.78</td><td align=\"left\">56.631</td></tr><tr><td align=\"left\">3</td><td align=\"left\">10/F</td><td align=\"left\">802</td><td align=\"left\">139</td><td align=\"left\">2</td><td align=\"left\">9.96</td><td align=\"left\">75.957</td><td align=\"left\">21.55</td><td align=\"left\">24.874</td><td align=\"left\">15.64</td><td align=\"left\">39.401</td></tr><tr><td align=\"left\">4</td><td align=\"left\">2/M</td><td align=\"left\">101</td><td align=\"left\">116</td><td align=\"left\">3</td><td align=\"left\">13.519</td><td align=\"left\">55.413</td><td align=\"left\">16.64</td><td align=\"left\">25.85</td><td align=\"left\">15.89</td><td align=\"left\">38.46</td></tr><tr><td align=\"left\">5</td><td align=\"left\">2/M</td><td align=\"left\">107</td><td align=\"left\">118</td><td align=\"left\">1</td><td align=\"left\">9.158</td><td align=\"left\">53.148</td><td align=\"left\">14.56</td><td align=\"left\">23.69</td><td align=\"left\">15.6</td><td align=\"left\">37.75</td></tr><tr><td align=\"left\">6</td><td align=\"left\">11.5/M</td><td align=\"left\">278</td><td align=\"left\">128</td><td align=\"left\">3</td><td align=\"left\">15.637</td><td align=\"left\">23.54</td><td align=\"left\">16.63</td><td align=\"left\">30.71</td><td align=\"left\">16.97</td><td align=\"left\">25.589</td></tr><tr><td align=\"left\">7</td><td align=\"left\">10.5/F</td><td align=\"left\">254</td><td align=\"left\">141</td><td align=\"left\">3</td><td align=\"left\">8.099</td><td align=\"left\">21.07</td><td align=\"left\">14.26</td><td align=\"left\">85.07</td><td align=\"left\">15.97</td><td align=\"left\">61.322</td></tr><tr><td align=\"left\">8</td><td align=\"left\">4/F</td><td align=\"left\">171</td><td align=\"left\">108</td><td align=\"left\">4</td><td align=\"left\">14.329</td><td align=\"left\">34.932</td><td align=\"left\">17.5</td><td align=\"left\">27.258</td><td align=\"left\">15.72</td><td align=\"left\">45.46</td></tr><tr><td align=\"left\">9</td><td align=\"left\">5.5/M</td><td align=\"left\">323</td><td align=\"left\">121</td><td align=\"left\">6</td><td align=\"left\">18.316</td><td align=\"left\">43.31</td><td align=\"left\">16.69</td><td align=\"left\">52.272</td><td align=\"left\">16.47</td><td align=\"left\">48.37</td></tr><tr><td align=\"left\">10</td><td align=\"left\">5.5/M</td><td align=\"left\">261</td><td align=\"left\">116</td><td align=\"left\">7</td><td align=\"left\">15.45</td><td align=\"left\">40.02</td><td align=\"left\">11.4</td><td align=\"left\">30.616</td><td align=\"left\">11.08</td><td align=\"left\">11.505</td></tr><tr><td align=\"left\">11</td><td align=\"left\">12.5/M</td><td align=\"left\">223</td><td align=\"left\">172</td><td align=\"left\">5</td><td align=\"left\">8.659</td><td align=\"left\">12.346</td><td align=\"left\">10.65</td><td align=\"left\">20.706</td><td align=\"left\">19.17</td><td align=\"left\">113.235</td></tr><tr><td align=\"left\">12</td><td align=\"left\">2/F</td><td align=\"left\">81.7</td><td align=\"left\">147</td><td align=\"left\">2</td><td align=\"left\">14.515</td><td align=\"left\">35.614</td><td align=\"left\">15.26</td><td align=\"left\">42.23</td><td align=\"left\">15.97</td><td align=\"left\">38.46</td></tr><tr><td align=\"left\">13</td><td align=\"left\">11/M</td><td align=\"left\">218</td><td align=\"left\">148</td><td align=\"left\">1</td><td align=\"left\">16.69</td><td align=\"left\">31.3</td><td align=\"left\">14.39</td><td align=\"left\">22.357</td><td align=\"left\">15.9</td><td align=\"left\">34.093</td></tr><tr><td align=\"left\">14</td><td align=\"left\">3.5/F</td><td align=\"left\">112</td><td align=\"left\">127</td><td align=\"left\">1</td><td align=\"left\">16.63</td><td align=\"left\">34.646</td><td align=\"left\">19.62</td><td align=\"left\">42.578</td><td align=\"left\">23.5</td><td align=\"left\">41.68</td></tr><tr><td align=\"left\">15</td><td align=\"left\">3.5/F</td><td align=\"left\">210</td><td align=\"left\">104</td><td align=\"left\">1</td><td align=\"left\">12.46</td><td align=\"left\">16.329</td><td align=\"left\">17.75</td><td align=\"left\">73.666</td><td align=\"left\">16.26</td><td align=\"left\">45.03</td></tr><tr><td align=\"left\">16</td><td align=\"left\">8/F</td><td align=\"left\">288</td><td align=\"left\">137</td><td align=\"left\">2</td><td align=\"left\">14.57</td><td align=\"left\">48.38</td><td align=\"left\">17.38</td><td align=\"left\">45.133</td><td align=\"left\">16.34</td><td align=\"left\">44.627</td></tr><tr><td align=\"left\">17</td><td align=\"left\">8/M</td><td align=\"left\">204</td><td align=\"left\">151</td><td align=\"left\">2</td><td align=\"left\">15.64</td><td align=\"left\">23.78</td><td align=\"left\">17.88</td><td align=\"left\">28.657</td><td align=\"left\">15.98</td><td align=\"left\">36.95</td></tr><tr><td align=\"left\">18</td><td align=\"left\">8/F</td><td align=\"left\">402</td><td align=\"left\">116</td><td align=\"left\">3</td><td align=\"left\">26.53</td><td align=\"left\">32.51</td><td align=\"left\">18.25</td><td align=\"left\">86.823</td><td align=\"left\">17.62</td><td align=\"left\">89.72</td></tr><tr><td align=\"left\">19</td><td align=\"left\">5..5/F</td><td align=\"left\">133</td><td align=\"left\">152</td><td align=\"left\">3</td><td align=\"left\">17.13</td><td align=\"left\">25.918</td><td align=\"left\">14.82</td><td align=\"left\">30.406</td><td align=\"left\">14.98</td><td align=\"left\">42.78</td></tr><tr><td align=\"left\">20</td><td align=\"left\">2/M</td><td align=\"left\">129</td><td align=\"left\">119</td><td align=\"left\">5</td><td align=\"left\">13.7</td><td align=\"left\">22.774</td><td align=\"left\">17.88</td><td align=\"left\">52.463</td><td align=\"left\">23.45</td><td align=\"left\">30.821</td></tr><tr><td align=\"left\">21</td><td align=\"left\">7/F</td><td align=\"left\">239</td><td align=\"left\">164</td><td align=\"left\">4</td><td align=\"left\">16.19</td><td align=\"left\">43.74</td><td align=\"left\">17.38</td><td align=\"left\">38.13</td><td align=\"left\">7.78</td><td align=\"left\">97.409</td></tr><tr><td align=\"left\">22</td><td align=\"left\">8.5/M</td><td align=\"left\">143</td><td align=\"left\">143</td><td align=\"left\">2</td><td align=\"left\">16.57</td><td align=\"left\">20.225</td><td align=\"left\">16.01</td><td align=\"left\">37.31</td><td align=\"left\">15.79</td><td align=\"left\">44.63</td></tr><tr><td align=\"left\">23</td><td align=\"left\">7.5/M</td><td align=\"left\">156</td><td align=\"left\">106</td><td align=\"left\">8</td><td align=\"left\">11.21</td><td align=\"left\">23.737</td><td align=\"left\">18</td><td align=\"left\">63.22</td><td align=\"left\">23.31</td><td align=\"left\">95.465</td></tr><tr><td align=\"left\">24</td><td align=\"left\">4/F</td><td align=\"left\">193</td><td align=\"left\">141</td><td align=\"left\">3</td><td align=\"left\">18.87</td><td align=\"left\">16.731</td><td align=\"left\">8.72</td><td align=\"left\">23.641</td><td align=\"left\">15.34</td><td align=\"left\">38.538</td></tr><tr><td align=\"left\">25</td><td align=\"left\">4/M</td><td align=\"left\">377</td><td align=\"left\">117</td><td align=\"left\">4</td><td align=\"left\">18.87</td><td align=\"left\">33.195</td><td align=\"left\">22.75</td><td align=\"left\">58.51</td><td align=\"left\">19.41</td><td align=\"left\">44.75</td></tr><tr><td align=\"left\">26</td><td align=\"left\">5/M</td><td align=\"left\">179</td><td align=\"left\">147</td><td align=\"left\">3</td><td align=\"left\">18.06</td><td align=\"left\">48.372</td><td align=\"left\">13.39</td><td align=\"left\">28.513</td><td align=\"left\">8.16</td><td align=\"left\">28.113</td></tr><tr><td align=\"left\">27</td><td align=\"left\">7.5/F</td><td align=\"left\">174</td><td align=\"left\">139</td><td align=\"left\">3</td><td align=\"left\">26.8</td><td align=\"left\">20.227</td><td align=\"left\">21.5</td><td align=\"left\">59.855</td><td align=\"left\">15.5</td><td align=\"left\">19.893</td></tr><tr><td align=\"left\">28</td><td align=\"left\">6.5/M</td><td align=\"left\">284</td><td align=\"left\">147</td><td align=\"left\">5</td><td align=\"left\">19.49</td><td align=\"left\">31.143</td><td align=\"left\">11.21</td><td align=\"left\">51.412</td><td align=\"left\">10.77</td><td align=\"left\">31.48</td></tr><tr><td align=\"left\">29</td><td align=\"left\">4/F</td><td align=\"left\">114</td><td align=\"left\">96</td><td align=\"left\">1</td><td align=\"left\">12.27</td><td align=\"left\">181.192</td><td align=\"left\">16.82</td><td align=\"left\">25.594</td><td align=\"left\">17.55</td><td align=\"left\">37.985</td></tr><tr><td align=\"left\">30</td><td align=\"left\">3.5/F</td><td align=\"left\">197</td><td align=\"left\">130</td><td align=\"left\">7</td><td align=\"left\">17.56</td><td align=\"left\">41.909</td><td align=\"left\">11.58</td><td align=\"left\">28.575</td><td align=\"left\">14.84</td><td align=\"left\">48.859</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>ET-1 plasma and urine levels at 3 selected time points in HSP patients and controls</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td/><td align=\"left\">ET-1 in patients (mean ± SD))</td><td align=\"left\">ET-1 in controls (mean ± SD)</td><td align=\"left\">p</td></tr></thead><tbody><tr><td align=\"left\">1^stmeasurement</td><td align=\"left\">Plasma</td><td align=\"left\">15.66 ± 4.30</td><td align=\"left\">16.75 ± 5.96</td><td align=\"left\">0.41</td></tr><tr><td/><td align=\"left\">Urine</td><td align=\"left\">37.44 ± 30.78</td><td align=\"left\">45.31 ± 20.33</td><td align=\"left\">0.24</td></tr><tr><td align=\"left\">2^ndmeasurement</td><td align=\"left\">Plasma</td><td align=\"left\">16.11 ± 3.26</td><td align=\"left\">16.75 ± 5.96</td><td align=\"left\">0.60</td></tr><tr><td/><td align=\"left\">Urine</td><td align=\"left\">42.01 ± 18.64</td><td align=\"left\">45.31 ± 20.33</td><td align=\"left\">0.51</td></tr><tr><td align=\"left\">3^rdmeasurement</td><td align=\"left\">Plasma</td><td align=\"left\">16.14 ± 3.60</td><td align=\"left\">16.75 ± 5.96</td><td align=\"left\">0.63</td></tr><tr><td/><td align=\"left\">Urine</td><td align=\"left\">46.95 ± 23.15</td><td align=\"left\">45.31 ± 20.33</td><td align=\"left\">0.77</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>ET-1 values in plasma and urine at three distinct measurements in relation to renal score</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td/><td align=\"center\">ET-1 (mean ± SD, range)</td><td align=\"center\">Renal score (Mean ± SD, range)</td><td align=\"center\">rho</td><td align=\"center\">p</td></tr></thead><tbody><tr><td align=\"left\">1^stmeasurement</td><td align=\"center\">Plasma</td><td align=\"center\">16.20 ± 5.18, 2.18–26.80</td><td align=\"center\">0.70 ± 0.91, 0–3</td><td align=\"center\">-0.011</td><td align=\"center\">0.95</td></tr><tr><td/><td align=\"center\">Urine</td><td align=\"center\">41.38 ± 26.17, 9.14–181.19</td><td/><td align=\"center\">0.099</td><td align=\"center\">0.59</td></tr><tr><td align=\"left\">2^ndmeasurement</td><td align=\"center\">Plasma</td><td align=\"center\">16.11 ± 3.26, 8.72–22.75</td><td align=\"center\">0.73 ± 1.08, 0–3</td><td align=\"center\">-0.180</td><td align=\"center\">0.92</td></tr><tr><td/><td align=\"center\">Urine</td><td align=\"center\">42.01 ± 18.64, 20.70–86.82</td><td/><td align=\"center\">0.248</td><td align=\"center\">0.57</td></tr><tr><td align=\"left\">3^rdmeasurement</td><td align=\"center\">Plasma</td><td align=\"center\">16.14 ± 3.60, 7.78–23.5</td><td align=\"center\">0.1 ± 0.30, 0–1</td><td align=\"center\">-0.288</td><td align=\"center\">0.52</td></tr><tr><td/><td align=\"center\">Urine</td><td align=\"center\">46.95 ± 23.15, 11.50–113.23</td><td/><td align=\"center\">0.473</td><td align=\"center\">0.66</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T5\"><label>Table 5</label><caption><p>ET-1 values in plasma and urine at three distinct time point measurements in relation to clinical score.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td/><td align=\"left\">ET-1 (mean ± SD, range)</td><td align=\"left\">Clinical score (mean ± SD, range)</td><td align=\"left\">rho</td><td align=\"left\">p</td></tr></thead><tbody><tr><td align=\"left\">1^stmeasurement</td><td align=\"left\">Plasma</td><td align=\"left\">16.20 ± 5.18, 2.18–26.80</td><td align=\"left\">3.23 ± 1.94, 1–8</td><td align=\"left\">0.145</td><td align=\"left\">0.443</td></tr><tr><td/><td align=\"left\">Urine</td><td align=\"left\">41.38 ± 26.17, 9.14–181.19</td><td/><td align=\"left\">-0.032</td><td align=\"left\">0.864</td></tr><tr><td align=\"left\">2^ndmeasurement</td><td align=\"left\">Plasma</td><td align=\"left\">16.11 ± 3.26, 8.72–22.75</td><td align=\"left\">0.73 ± 1.08, 0–3</td><td align=\"left\">-0.018</td><td align=\"left\">0.923</td></tr><tr><td/><td align=\"left\">Urine</td><td align=\"left\">42.01 ± 18.64, 20.70–86.82</td><td/><td align=\"left\">-0.106</td><td align=\"left\">0.576</td></tr><tr><td align=\"left\">3^rdmeasurement</td><td align=\"left\">Plasma</td><td align=\"left\">16.14 ± 3.60, 7.78–23.5</td><td align=\"left\">0.13 ± 0.34, 0–1</td><td align=\"left\">0.011</td><td align=\"left\">0.952</td></tr><tr><td/><td align=\"left\">Urine</td><td align=\"left\">46.95 ± 23.15, 11.50–113.23</td><td/><td align=\"left\">0.237</td><td align=\"left\">0.205</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T6\"><label>Table 6</label><caption><p>ET-1 values in plasma and urine between patients with and without renal involvement (RI).</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td/><td align=\"left\">Patients with RI</td><td align=\"left\">Patients without RI</td><td align=\"left\">p</td></tr></thead><tbody><tr><td align=\"left\">1^stmeasurement</td><td align=\"left\">Plasma</td><td align=\"left\">15.77 ± 4.48, 8.6–26.8</td><td align=\"left\">15.56 ± 4.29, 8.09–26.53</td><td align=\"left\">0.851</td></tr><tr><td/><td align=\"left\">Urine</td><td align=\"left\">35.59 ± 16.80, 12.34–75.95</td><td align=\"left\">39.06 ± 39.77, 9.147–181.19</td><td align=\"left\">0.506</td></tr><tr><td align=\"left\">2^ndmeasurement</td><td align=\"left\">Plasma</td><td align=\"left\">16.11 ± 3.26, 8.72–22.75</td><td align=\"left\">16.20 ± 3.25, 8.72–22.75</td><td align=\"left\">0.891</td></tr><tr><td/><td align=\"left\">Urine</td><td align=\"left\">38.40 ± 14.64, 20.70–63.22</td><td align=\"left\">44.41 ± 20.94, 23.64–86.823</td><td align=\"left\">0.582</td></tr><tr><td align=\"left\">3^rdmeasurement</td><td align=\"left\">Plasma</td><td align=\"left\">16.86 ± 4.61, 11.08–23.31</td><td align=\"left\">16.00 ± 3.47, 7.78–23.5</td><td align=\"left\">0.977</td></tr><tr><td/><td align=\"left\">Urine</td><td align=\"left\">60.63 ± 42.53, 11.50–113.23</td><td align=\"left\">44.22 ± 17.30, 19.89–97.40</td><td align=\"left\">0.486</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>RBC: red blood cells</p></table-wrap-foot>", "<table-wrap-foot><p>Normal values of IgA (mg/dl)</p><p>4–6 months 4.4–84</p><p>7–12 months 11–106</p><p>1–5 years 14–159</p><p>6–10 years 33–236</p><p>Adults 70–312</p><p>Normal values of C₃(mg/dl)</p><p>1–3 months 53–31</p><p>3–12 months 62–180</p><p>1–10 years 77–195</p><p>Adult 83–177</p></table-wrap-foot>", "<table-wrap-foot><p>Results from univariate analysis (rho = spearman's correlation coefficient)</p></table-wrap-foot>", "<table-wrap-foot><p>Results from univariate analysis (rho = spearman's correlation coefficient).</p></table-wrap-foot>" ]

[]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p><italic>Learning is the process by which experience is transformed into recognition </italic>[##UREF##0##1##].</p>", "<p>Family physicians are challenged to maintain their professional competences in an ever changing environment characterized by a continuing evidence expansion in their field [##REF##11557716##2##,##REF##9518922##3##]. This encourages educational providers to generate learning methods that equally develop the family physicians' general competences and their capabilities to adapt these to real life contexts [##REF##11557716##2##,##REF##11588088##4##].</p>", "<p>Recent developments in adult learning theories, mainly inspired by complexity theory and experiential learning theories have taken up this challenge [##UREF##0##1##,##REF##11557716##2##,##UREF##1##5##]. According to these theories focus on professional capability is best enhanced through process-oriented learning methods by which learning is driven by needs, is situated in real-life contexts, and allows time to reflect on own performances [##REF##11557716##2##,##REF##11588088##4##,##UREF##2##6##]. This means that learning is seen as a continuous process in which every new experience builds on, and integrates with, previously accumulated experiences. Reviewing and reflecting on patient cases is in line with this learning approach [##REF##11588088##4##,##REF##11555216##7##]. There has, however, been remarkably little research into how patient cases might be applied in professional education [##REF##11588088##4##].</p>", "<p>The study presented is part of a larger project on delay in cancer diagnosis in family practice (the CAP-project) currently carried out at the Research Unit for General Practice, University of Aarhus. 467 (81%) family physicians in the County of Aarhus (640 000 inhabitants) completed detailed questionnaires on 2,212 (83%) newly diagnosed patients with cancer encountered in their practices during a one year period (2004–2005) [##REF##18307790##8##]. In order to complete the questionnaire the family physicians were required to perform a systematic case review of each patient: they had to consult their records to provide dates of symptom-presentation, investigations and treatments initiated, and reflect on previous encounters with the patients to give detailed information on his/hers knowledge of the patients' care seeking behaviour, mental health and risk factors.</p>", "<p>Because the participating family physicians had systematically performed case reviews of the majority of all patients with cancer identified in their practices in 2004–2005, the CAP-project posed a unique opportunity to study the learning potentials of reviewing patient cases. We therefore decided to explore family physicians' experiences from the CAP-project through indebt interviews (Table ##TAB##0##1##). The purpose of this article is to present family physicians' perceptions of the learning process initiated by reviewing patient cases, and their evaluations of using patient case reviews as a learning method in family practice.</p>" ]

[ "<title>Methods</title>", "<title>Setting</title>", "<p>Denmark has a publicly funded health care system providing free access to family practice and hospital care. The family physicians function as gatekeepers to the rest of the health care system, carrying out initial investigations and referring patients to hospital or outpatient clinic treatment if necessary. Within an overall legal and logistic framework the individual family physicians' practices and work procedures are subject to internal variations.</p>", "<title>Participant selection</title>", "<p>Our study is based on semi-structured, individual interviews. In order to obtain the broadest range of information and perspectives we selected family physicians according to gender, age and practice type. Initially we selected 18 family physicians from the CAP-database. Fourteen agreed to participate in the study; three participated in pilot interviews, one cancelled the interview due to illness, and ten were interviewed according to the final interview guide.</p>", "<title>Interviews</title>", "<p>The first author (RSA), a trained anthropologist, conducted semi-structured interviews of approximately one hour duration with participating family practitioners. RSA was supervised by the co-authors, JS and FB, both family physicians and experienced in qualitative research methods. Before the interview each family physician received a written invitation to participate and subsequently gave verbal consent. To ensure reliability, the three first interviews were conducted during the development of the interview guide that was continually modified as new issues and themes emerged from the data. The interviews took place at the individual family physician's practice and were all tape-recorded. We conducted the interviews according to ethnographic principles for open-ended questions, meaning that the physicians were encouraged to speak freely and to raise issues of importance to them, also if the issues went beyond the topic-guide themes [##REF##2037216##9##]. During the interviews, the family physicians were given paper copies of their CAP-questionnaires. Other studies have applied the same methods when exploring a similar setting [##REF##16549036##10##].</p>", "<title>Data analysis</title>", "<p>The interviews were transcribed verbatim by RSA, and all transcripts were read repeatedly to get an overall impression of the material before the initial coding. Initially, all meaningful text units in the transcripts were coded by theme [##UREF##3##11##]. Agreement was reached among the authors after thorough discussions of the initial coding. Subsequently, the themes were condensed and compared across groups, and main analytical categories were identified. During the last part of the analytic phase, theories of adult learning were used as an analytical basis for interpretation [##UREF##0##1##,##REF##11557716##2##,##UREF##1##5##,##UREF##2##6##,##REF##8373649##12##]. In order to clarify and link the different analytical categories with relevant theories, RSA wrote memos throughout the coding process [##UREF##4##13##]. We used the software package <italic>NVivo, ed. 7 </italic>to assist with coding, sorting and retrieval (Table ##TAB##1##2##).</p>", "<title>Ethical approval</title>", "<p>According to the Scientific Committee for the County of Aarhus, the Biomedical Research Ethics Committee System Act does not apply to this project. The study was approved by the Danish Data Protection Agency (J. nr. 2004-41-3705).</p>" ]

[ "<title>Results</title>", "<title>Learning processes initiated by patient case reviews</title>", "<p>The majority of the family physicians in our sample responded that reviewing patient cases gave rise to reflections on how their work was organized and carried out. They reported that the reflections resulted in more or less clear images of the structural and behavioural changes that were needed in their own practises, and it increased their capabilities to implement these changes, as it raised their awareness of how to navigate within the health care system. Two themes centred on the family physicians' case manager function and their communication strategies illustrate this.</p>", "<title>Case manager</title>", "<p>According to the family physicians, they are rarely able to predict their patients' clinical pathway when they are referred to hospital for further investigations, which impairs their case manager function.</p>", "<p>\"Looking back was good. We had lost contact with some of the patients. When we refer them [to the hospital] they often disappear from your system. It was good to see: well, what happened. It gives you something, it really does.\" (Informant 3, female, 25 years in family practice)</p>", "<p>Reviewing patient cases gave them better insight into the time span and distribution of responsibility in relation to clinical pathways, thereby potentially improving their ability to perform this task. E.g., some family physicians stated that it had increased their attention on the delay patients experienced during hospital investigations. For some these insights had resulted in a change of strategy for dealing with delay and attempts to accelerate patients' pathway through the health care system:</p>", "<p>\"The questionnaires once again demonstrated that the time from referral to initial patient appointment and until the result of the examinations is known is long (...). You must try to keep this process as short as possible. Sometimes I ask the patients to help, to tell me if they wait too long. I think this [reviewing patient cases] has been a catalyst for me in making me do this even more.\" (Informant 5, male, 12 years in family practice)</p>", "<p>To ask the patient to take on a more active role is just one in a series of strategies initiated to improve the family physician's case manager function. Many stated that they followed more pro-active referral procedures, such as phoning hospital departments to make appointments for their patients, or had become more selective in choosing which hospital departments or clinical investigations to refer to.</p>", "<p>\"I think we have to be very careful what we write in our referrals. If I suspect it is cancer, I have to write cancer. It has to b e very clear. Reviewing these patient cases has confirmed me in this.\" (Informant 9, male, 17 years in family practice)</p>", "<title>Communication</title>", "<p>The success of clinical pathways depends on the family physicians' communication skills and strategies at various levels: with patients, within the clinic and with hospital staff. According to the physicians reviewing patient cases may increase focus on all levels of communication.</p>", "<p>\"This [reviewing patient cases] makes you wiser professionally speaking, and it opens your eyes towards the weaknesses of our system, especially in relation to how we communicate with others. Sometimes they surprise you. 'Did this really take six months?\"' (Informant 5, male, 12 years in family practice)</p>", "<p>Individual family physicians described patient encounters where patients had failed to comply with their recommendations either by refusing further examinations or by failing to turn up at their appointments. In some of these cases lack of compliance was due to misunderstandings and to the fact that the family physician was not sure whether he or she had clearly indicated a suspicion of cancer.</p>", "<p>\"Well, she [the patient] didn't turn up for re-examinations as I had advised her to. And the fact that I had to go through her case again made me think if I had been too imprecise in my communications with her.\" (Informant 11, female, 23 years in family practice)</p>", "<p>Reflections on these experiences had led to considerations about how to make their wording more precise, or how to ensure better continuity, e.g. by making sure patients keep future appointments and by explaining to the patient to contact the family physician again if symptoms become worse.</p>", "<p>Reviewing patient cases also focused attention to communication within the clinic. In some partnership practices it resulted in spontaneous discussions about individual patients thus increasing the awareness of the beneficial use of group discussions:</p>", "<p>\"I believe that we consequently ... – well at least we do it more – discuss patients at our daily conference, if we think it went wrong, if things went too slow, or if we need to share experiences.\" (Informant 8, male, 18 years in family practice)</p>", "<p>\"Some of the questionnaires resulted in informal discussions during our breaks, and as such, they made us share experiences with each other in our practice.\" (Informant 1, female, 7 years in family practice)</p>", "<p>For some, focus was turned towards the more specific procedures for handling external communication such as discharge letters, x-ray results and other test results. A family physician from a large practice told how incoming answers from hospitals were sometimes mislaid, and how they had previously considered how to improve their communication system to avoid this. Reviewing patient cases had convinced them that these changes were necessary and consequently the process had been accelerated.</p>", "<p>To summarize. The process of reflection initiated by reviewing patient cases enabled family physicians to reconsider their clinical work procedures which potentially supported the transition from individual competence to personal capability. According to the physicians, they were not only able to identify needed changes, some reported that they were able to transform these ideas into action and do things more effectively. According to our data this transition takes place, because the learning processes initiated were based on real life experiences which equally initiated reflections on <italic>what </italic>to improve, as well as <italic>how </italic>to improve their work (Figure ##FIG##0##1##).</p>", "<title>Family physicians' evaluations of patient case reviews as a learning method</title>", "<p>Our data imply that the family physicians are in favour of using patient case reviews as a learning method. The fact that the patient cases are patient-centred and based on the family physicians' real-life experiences highly increase the level of motivation to take up learning.</p>", "<title>Patient-centred</title>", "<p>When asked about their perceptions of using patient case reviews as a learning method, the family physicians made a variety of emotional responses such as \"it feels very intrusive\" and \"it touches me as a human being, which makes it really efficient.\" Our data suggest that this is due to the fact that family physicians perceive of their clinical work as embedded in social and moral obligations. They do not have a conclusive view on their patients as 'just patients', and their clinical work does not take place in a vacuum, but is part of a wider social and professional framework, according to which they evaluate their work. According to the family physicians the fact that they were reviewing their own patients potentially improved the impact of learning.</p>", "<p>\"The examples always stand stronger when they are 'in flesh and blood', so to speak. I always listen to the patient stories of my colleagues, but of course, when I am faced with the stories of the patients I know personally it has a stronger effect on me. This man for example, I knew his family. That means something.\" (Informant 1, female, 7 years in family practice)</p>", "<p>\"It is more relevant, simple because it is my own patients. I like that.\" (Informant 2, female, 20 years in family practice)</p>", "<p>Some of the family physicians' responses suggest that the fact that the patient cases reviewed were cancer-related, and thus often long-term and severe, increased the learning impact of the patient case reviews.</p>", "<p>\"Well, these patients they have really made an impression on you, because you come to know them and their families. And then when they die....Of course that concerns me and my colleagues. But that was the powerful element in going through these cases again.\" (Informant 7, female, 9 years in family practice)</p>", "<title>Real life experiences</title>", "<p>The family physicians were also in favour of using patient case reviews as a learning method because they embrace the complexity they experience in their daily practice. Various factors such as age, medical history, physician-patient relation etc. influence their work, and no patients are alike.</p>", "<p>\"When we see a classic case, one that fits the descriptions in the medical textbooks, we tell each other because it is so rare.\" (Informant 5, male, 12 years in family practice)</p>", "<p>\"The good thing about focusing on your own patients is that focus is on <italic>how do you make things happen </italic>in real life, more than just information on how to do this and this. When I look back at my own work and my own experiences it becomes a kind of inspiration to me: what can I do in order to improve my work.\" (Informant 12, male, 7 years in family practice)</p>", "<p>Another aspect of this was that family physicians expressed that the process of 'looking back' and reviewing individual patient cases was well-known to many of them. When talking in abstract terms about their work, and when referring to their clinical experiences such as establishing a cancer diagnosis, many family physicians said that the management of patients is often based on clinical experiences of similar cases. However, they welcomed a means of systematising this part of their job:</p>", "<p>\"I often look back. What did I do? When did I do what? But this [reviewing patient cases] was good, more systematic. And when you act on your own initiative, perhaps you only see what you want to see. This made you go through the whole process.\" (Informant 1, female, 7 years in family practice)</p>" ]

[ "<title>Discussion</title>", "<p>Our study illustrates that reviewing patient cases has a learning potential because it initiates a reflective process that equally provide feedback about performance in real life situations and potentially enhance family physicians' capabilities implement changes. It also demonstrates that family physicians are in favour of this method, as it embraces the complexities they encounter in their daily practice (Table ##TAB##2##3##).</p>", "<p>Because of its qualitative perspective the study provides insights into the family physicians' subjective opinions, expectations and motivations and give in depth descriptions of potential learning processes initiated by patient case reviews. However, our results are based on family physicians' self-reported perceptions which call for further empirical studies, as our data do not allow us to evaluate whether changes were actually implemented. Finally our results may be biased toward a positive view on reviewing patient cases, because the participating physicians were selected among participants from the CAP-project, who had indirectly taken a positive interest in reviewing patient cases.</p>", "<p>A growing list of publications confirm that it is advantageous to apply learning methods that are structured around individual experiences because it increases the learners' potentials for acquiring new competences and improve their ability to apply these in daily practice [##REF##11557716##2##,##REF##11588088##4##,##REF##8373649##12##,##REF##9492684##14##, ####REF##12613060##15##, ##REF##1501333##16##, ##REF##16986148##17##, ##REF##11201063##18####11201063##18##]. The development of methods complying with this learning approach has been defined as a future challenge to continuing medical education [##REF##11588088##4##]. Our study shows that reviewing patient cases initiated a learning process that may meet the required demands, as they made family physicians reflect on work procedures, identify needed changes, and raised their awareness of how to implement changes.</p>", "<p>Another theme addressed in the paper was the family physicians' perceptions of using patient case reviews as a learning method. Our paper demonstrates that family physicians are in favour of applying patient case reviews as a learning method, because they embrace the complexities they encounter in their daily practice. Common claims against traditional learning methods have been that they are 'check-list driven' and fail to recognise the complex context into which new knowledge is integrated [##REF##12383461##19##,##REF##11701576##20##]. Our study indicate that reviewing patient cases evoke the complexity of the clinical situation and potentially assist in paying attention to how 'new knowledge' integrates with complex clinical reasoning [##REF##11555224##21##].</p>", "<p>However, learning based on patient cases represents real life dilemmas and it is inevitably based on the subjective experiences and decisions of the participating physicians. Therefore, the future challenge to educators and physicians is to develop methods that ensure reliable learning from the experiences of daily practice. Conducting patient case reviews in groups with a facilitator or in an audit-group may improve the likelihood of a fruitful outcome, and hence improve the level of quality assurance.</p>" ]

[ "<title>Conclusion</title>", "<p>Our results illustrate that patient case reviews initiate reflective processes providing feedback about performance in real life situations. It also demonstrates that family physicians are in favour of patient case reviews as a learning method, because it embraces the complexities they encounter in their daily practice and is based on personal experiences.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Recent theories on adult learning recommend that learning is situated in real-life contexts. Learning is considered a continuous process in which every new experience builds on, and integrates with, previously accumulated experiences. Reviewing and reflecting on patient cases is in line with this learning approach. There has, however, been remarkably little research into how patient cases might be applied in professional education. The purpose of this article is to present family physicians' perceptions of the learning process initiated by reviewing patient cases.</p>", "<title>Methods</title>", "<p>Thirteen family physicians, who had all participated in a large project on cancer diagnosis in family practice (the CAP-project), currently carried out at the Research Unit for General Practice, University of Aarhus were interviewed on their experiences of reviewing patient cases. In the CAP-project family physicians (n = 467, 81%) in the County of Aarhus (640 000 inhabitants) completed 2,212 (83%) detailed questionnaires on all newly diagnosed patients with cancer encountered in their practices during a one year period (2004–2005). In order to complete the questionnaire the family physicians were required to perform a systematic case review of each patient: they had to consult their records to provide dates of symptom-presentation, investigations and treatments initiated, and reflect on previous encounters with the patients to give detailed information on his/hers knowledge of the patients' care seeking behaviour, mental health and risk factors.</p>", "<p>The purpose of this article is to present indebt interview-data on family physicians' perceptions of the learning process initiated by reviewing patient cases, and their evaluations of using patient case reviews as a learning method in family practice.</p>", "<title>Results</title>", "<p>The process of reflection initiated by reviewing patient cases enabled family physicians to reconsider their clinical work procedures which potentially supported the transition from individual competence to personal capability. According to the physicians, they were not only able to identify needed changes, some reported that they were able to transform these ideas into action and do things more effectively. According to our data this transition takes place, because the learning processes initiated were based on real life experiences which equally initiated reflections on <italic>what </italic>to improve, as well as <italic>how </italic>to improve their work.</p>", "<title>Conclusion</title>", "<p>Patient case reviews initiate reflective processes providing feedback about performance in real life situations. Family physicians are in favour of patient case reviews as a learning method, because it embraces the complexities they encounter in their daily practice and is based on personal experiences.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>RSA has made contributions to the study design, has been in charge of the acquisition of data, contributed to data-analysis and written the final manuscript. JS has contributed with analysis and interpretation of data, critically revised the article for important intellectual content, and read and approved the final version of the manuscript. RPH has contributed with analysis and interpretation of data, critically revised the article, and read and approved the final version of the manuscript. FB Has made substantial contributions to the design of the study, interpretation of data, critically revised the article for important intellectual content and read and approved of the final version of the manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1472-6920/8/43/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>The study and all the authors of the manuscript were funded by The Research Unit for General Practice, Department of Public Health, University of Aarhus, Vennelyst Boulevard 6, 8000 Århus C.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Learning process initiated by reviewing patient cases.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Study design</p></caption><table frame=\"hsides\" rules=\"groups\"><tbody><tr><td align=\"left\"><bold>The CAP-study</bold></td></tr><tr><td align=\"left\">467 family physicians in the county of Aarhus completed 2,212 questionnaires in which they performed a systematic review of all newly diagnosed patients with cancer encountered in their practices during a one year period (2004–2005)</td></tr><tr><td/></tr><tr><td align=\"left\"><underline>The main themes of the CAP questionnaire:</underline></td></tr><tr><td align=\"left\">A description of the symptoms presented by the patients</td></tr><tr><td align=\"left\">Dates when patients first presented their symptoms to the family physician</td></tr><tr><td align=\"left\">Dates and descriptions of further investigations and treatments initiated</td></tr><tr><td align=\"left\">Responsibility for possible diagnostic delay in the health care system</td></tr><tr><td align=\"left\">A description of the family physicians' prior knowledge of the patient (care-seeking behaviour, mental health, risk factors)</td></tr><tr><td/></tr><tr><td align=\"left\"><bold>Study presented</bold></td></tr><tr><td align=\"left\">In order to study the learning potentials of reviewing patient cases thirteen family physicians from the CAP-study were interviewed on their perceptions of the learning processes initiated by reviewing patient cases, and their evaluations of using patient case reviews as a learning method in family practice</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Analysis</p></caption><table frame=\"hsides\" rules=\"groups\"><tbody><tr><td align=\"left\">We conducted a two-phased analysis:</td></tr><tr><td/></tr><tr><td align=\"left\"><bold>Phase one: Theme development.</bold></td></tr><tr><td align=\"left\"><underline>Themes developed:</underline></td></tr><tr><td align=\"left\">1. Reactions to receiving questionnaires</td></tr><tr><td align=\"left\">2. Potential outcome</td></tr><tr><td align=\"left\"> Diagnostic process</td></tr><tr><td align=\"left\"> Evaluation of competences</td></tr><tr><td align=\"left\"> Referral system</td></tr><tr><td align=\"left\"> Awareness of delay</td></tr><tr><td align=\"left\"> Organisational overview</td></tr><tr><td align=\"left\">3. Talk of change</td></tr><tr><td align=\"left\">4. Patient continuity</td></tr><tr><td align=\"left\">5. Family physician-patient relationship</td></tr><tr><td align=\"left\">6. Decision-making processes</td></tr><tr><td align=\"left\">7. Learning processes</td></tr><tr><td align=\"left\"> Based on patient records</td></tr><tr><td align=\"left\"> Based on daily experience</td></tr><tr><td align=\"left\">8. Learning by feedback</td></tr><tr><td align=\"left\"> Experiences from other learning interventions</td></tr><tr><td align=\"left\"> Positive and negative feedback</td></tr><tr><td align=\"left\"> Learning from patient records</td></tr><tr><td align=\"left\">9. Clinical failures</td></tr><tr><td align=\"left\">All data were coded according to these themes in Nvivo ed. 7.</td></tr><tr><td/></tr><tr><td align=\"left\"><bold>Phase two: Development of analytical categories.</bold></td></tr><tr><td align=\"left\">The interviews as well as the code books of each identified theme were re-read.</td></tr><tr><td align=\"left\">The analytical categories were developed in a dialectic process between identified themes and the applied theoretical framework.</td></tr><tr><td align=\"left\"><underline>Analytical categories:</underline></td></tr><tr><td align=\"left\">1. Learning processes described</td></tr><tr><td align=\"left\"> Reflections on work procedures</td></tr><tr><td align=\"left\"> Communication</td></tr><tr><td align=\"left\"> Case manager function</td></tr><tr><td align=\"left\">2. Evaluation of patient case reviews</td></tr><tr><td align=\"left\"> Patient-centeredness</td></tr><tr><td align=\"left\"> Real life experiences</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>What this study adds</p></caption><table frame=\"hsides\" rules=\"groups\"><tbody><tr><td align=\"left\"><bold>What is already known:</bold></td></tr><tr><td align=\"left\">Learning is best enhanced through process-oriented learning methods where learning is driven by needs, is situated in real-life contexts, and allows time to reflect on own performances.</td></tr><tr><td/></tr><tr><td align=\"left\"><bold>What this study adds:</bold></td></tr><tr><td align=\"left\">Reviewing patient cases allowed family physicians to reconsider their clinical work procedures, which enabled them to identify needed changes, as well as to transform these into action and do things more effectively.</td></tr><tr><td/></tr><tr><td align=\"left\"><bold>Suggestions for future research:</bold></td></tr><tr><td align=\"left\">Learning based on patient cases represents real life dilemmas and it is inevitably based on the subjective experiences of the participating physicians. Therefore, the future challenge is to develop methods that ensure reliable learning from the experiences of daily practice.</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1472-6920-8-43-1\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>The yeast <italic>Saccharomyces cerevisiae </italic>is widely used for production of many different commercial compounds such as food, feed, beverages and pharmaceuticals [##REF##17727659##1##]. It also serves as a model eukaryotic organism and has been the subject of more than 40,000 research publications [##REF##9297238##2##,##REF##17435240##3##]. After the complete genome sequence for yeast was released in 1996 [##REF##8849441##4##], about 4,600 ORFs were characterized [##REF##17435240##3##] and yeast contains many genes with human homologs [##REF##9297238##2##]. This has allowed for comparative functional genomics and comparative systems biology between yeast and human. Yeast, for example, has been used to understand the function of complex metabolic pathways that are related to the development of human diseases [##REF##8628392##5##, ####REF##9300813##6##, ##REF##12134146##7####12134146##7##].</p>", "<p>Several human diseases (e.g. cancer, atherosclerosis, Alzheimer's disease, and Parkinson's disease) are associated with disorders in lipid metabolism [##REF##17218835##8##, ####REF##16214218##9##, ##REF##15219388##10####15219388##10##]. The emergence of lipidomics has allowed analysis of lipid metabolism at the systems level [##REF##17218835##8##,##REF##17045192##11##]. Lipidomics promises to make a significant impact in our understanding of lipid related disease development [##REF##16920401##12##]. As with other high-throughput techniques, however, we hypothesize that one of the main challenges for utilization of lipidome data will be our ability to develop appropriate frameworks to integrate and map data for studying relations between lipid metabolism and other cellular networks.</p>", "<p>Previous work has shown that genome-scale metabolic models provide an excellent scaffold for integrating data into single, coherent models [##REF##15102469##13##]. The calculation of Reporter Metabolites using genome-scale metabolic models is an example of how metabolic models can be used to upgrade the information content of omics data [##REF##15710883##14##]. This approach allows mapping of key metabolites and reactions in large metabolic networks when combined with transcriptome [##REF##15710883##14##] or metabolome data [##REF##17016516##15##]. However, pathways, reactions, and genes that are not included in the metabolic network cannot be queried. Therefore, the Reporter Metabolite algorithm requires a reliable and global genome scale-model to achieve precise and accurate data interpretation.</p>", "<p>So far, three yeast genome-scale metabolic models, <italic>iFF708</italic>, <italic>iND750 </italic>and <italic>iLL672</italic>, have been published. All three models, however, lack a detailed description of the lipid metabolism. The first model, <italic>iFF708 </italic>[##REF##12566402##16##], consists of 1175 reactions linked to 708 ORFs. <italic>iFF708 </italic>shows good predictions of many different cellular functions [##REF##14578455##17##] and gene essentiality predictions [##REF##14506848##18##]. However, almost all intermediate reactions in lipid metabolism were either lumped or neglected. The second model published was <italic>iND750 </italic>[##REF##15197165##19##]. <italic>iND750 </italic>is fully compartmentalized, consisting of 1498 reactions linked to 750 ORFs. The model was validated by a large-scale gene deletion study and metabolic phenotypes [##REF##15355549##20##] and was expanded to include regulation for predicting gene expression and phenotypes of different transcription factor mutants [##REF##16606697##21##]. <italic>iND750 </italic>contains more reactions and metabolites in lipid metabolism than <italic>iFF708</italic>, but still lacks a comprehensive description of lipid metabolism. The third published model is <italic>iLL672</italic>, which is derived from <italic>iFF708 </italic>and comprises 1038 reactions. Several dead-end reactions of <italic>iFF708 </italic>were eliminated leading to an improved accuracy of the single gene deletion prediction [##REF##16204195##22##]. However, only minor improvements were made to reactions involved in lipid metabolism. The model was validated using ¹³C-labeling experiments to study the robustness of different yeast mutants [##REF##15960801##23##].</p>", "<p>Here our objective was to expand the genome-scale metabolic model of yeast to include a detailed description of lipid metabolism for use as a scaffold to integrate omics data. We used <italic>iFF708 </italic>as a template for building a model based on recent literature that contains new reactions in lipid metabolism and transport relative to all previous models. The new model named <italic>iIN800 </italic>includes 92 additional ORFs and provides a more detailed structure of lipid metabolism, tRNA synthesis and transport processes than previous models. The biomass composition, which is very important for flux balance analysis and predicting lethality, was also recalculated and improved. <italic>iIN800 </italic>was validated with large-scale gene deletion data and growth simulation predictions. Simulated intracellular fluxes were also supported by ¹³C-labeling flux experimental data. Finally, we show that the transcriptome data of yeast cultivated under various growth conditions can be integrated with <italic>iIN800 </italic>to identify lipid related Reporter Metabolites. We anticipate that <italic>iIN800 </italic>will be useful as a scaffold for integrating multilevel omic data and that this new model will have a significant impact in the emerging field of lipidomics.</p>" ]

[ "<title>Methods</title>", "<title>Model reconstruction and visualization</title>", "<p>Reconstruction of the <italic>S. cerevisiae </italic>genome-scale metabolic model was done by expanding <italic>iFF708 </italic>[##REF##12566402##16##]. The additional ORFs included in the expansion procedure were involved in lipid metabolism, tRNA synthesis and lipoamide dehydrogenase. These ORFs were added based on publications listed in Additional file ##SUPPL##0##1##. Online resources related to <italic>S. cerevisiae</italic>, such as SGD [##REF##16381907##41##], MIPS [##REF##15608217##42##] and YPD [##REF##12073323##43##], were also used to confirm the existence of the ORFs and their function. Pathway and reaction databases including KEGG [##REF##16268787##44##], ExPASy [##REF##12824418##45##], and Reactome [##REF##17367534##46##], were used together with research papers to identify relevant information of the additional reactions and metabolites, e.g. stoichiometry and co-factor usage. The expanded <italic>iFF708</italic>, called <italic>iIN800</italic>, was visualized by Adobe Illustrator software (Adobe Systems), and then converted to EPS format (Adobe Systems) format which is downloadable as Additional file ##SUPPL##5##6##. In this visualization file, it is possible to overlay information about transcription, fluxes etc. A detailed list of metabolic reactions in <italic>iIN800 </italic>is provided as Additional file ##SUPPL##6##7##.</p>", "<title>Metabolic modeling and simulations</title>", "<p>The reaction set in <italic>iIN800 </italic>was used for construction of a stoichiometric matrix <bold><italic>S </italic></bold>(m × n). In the stoichiometric matrix, m = 1013, which is the number of metabolites, and n = 1446, which is the number of metabolic reactions. With an assumption of steady state for all metabolite pools, a linear equation constraining the fluxes in the metabolic network is obtained [##REF##11175725##30##,##REF##10805808##47##]:</p>", "<p></p>", "<p>Here <bold><italic>v </italic></bold>is a vector that contains all the fluxes in the model. Equation (1) has a large number of degrees of freedom, i.e. it is an underdetermined problem, and linear programming was employed to solve the equation system by maximizing an objective function Z (equal to the growth rate), an approach generally referred to as flux balance analysis (FBA) [##REF##11175725##30##,##REF##10805808##47##]. The problem formulation is described below.</p>", "<p>Maximize:</p>", "<p></p>", "<p>Subject to:</p>", "<p></p>", "<p></p>", "<p>where <bold><italic>α </italic></bold>and <bold><italic>β </italic></bold>are lower and upper bounds of fluxes respectively, <bold><italic>ω </italic></bold>is a weight vector indicating an amount of desired metabolites for biomass synthesis. For irreversible fluxes semi-positive infinite boundary was applied as <bold><italic>0 </italic></bold>≤ <bold><italic>v </italic></bold>≤ ∞, and fully infinite boundaries was applied as -∞ ≤ <bold><italic>v </italic></bold>≤ ∞ for reversible fluxes. The problem was solved by using the commercial linear programming software package LINDO (Lindo systems Inc., Chicago, IL, USA). The calculated intracellular fluxes were overlaid on the visualized genome-scale map as described previously by the ReMapper software (The software has been developed for visualization of multilevel omics data onto a metabolic map.).</p>", "<title>Calculation of biomass composition and sensitivity analysis</title>", "<p>The biomass composition was re-calculated in order to improve the prediction of the model during growth at different nutrition-limitations, i.e. carbon- and nitrogen-limited growth condition. The contents of macro-molecules were extracted from the thesis of Schulze [##UREF##0##27##] who measured the biomass composition at a dilution rate of 0.1 h^-1. The calculations were performed as described previously [##REF##12566402##16##]. The calculation of protein precursors, i.e. amino acids, and carbohydrate precursors, i.e. trehalose, glycogen, manna and glucan, were adopted from Schulze's work [##UREF##0##27##]. Deoxyribonucleotide and ribonucleotide compositions were calculated from the study of Vaughan-Martini and co-workers [##REF##8259831##48##]. Lipid compositions were calculated from our own measurements of structural lipidomics, which contains phospholipids, triacylglycerol, sterols, sterol-esters, sphingolipids, free fatty acids and fatty acids composition of all measured lipid classes (unpublished data). The impact of the macromolecular composition on biomass yield was explored in aerobically glucose- and ammonium-limited conditions by fixing the specific growth rate and then minimizing the glucose and ammonium uptake rates at both glucose- and ammonium-limited growth conditions. Four parameters were evaluated, namely the protein, RNA, carbohydrate and lipid content of the biomass.</p>", "<title>Growth simulations</title>", "<p>The metabolic capabilities of <italic>iIN800 </italic>were evaluated by using FBA and linear programming to simulate the biomass flux representing the <italic>in silico </italic>growth rate, which were derived by maximizing the biomass production. Data from various carbon-limited and nitrogen-limited chemostat experiments performed at either aerobic or anaerobic growth condition were taken from the literature for comparisons (see references in Additional file ##SUPPL##2##3##). These data were used to validate the metabolic capabilities of the model by comparing <italic>in silico </italic>biomass yields with <italic>in vivo </italic>biomass yields. The <italic>in silico </italic>biomass yields were calculated by fixing measurable uptake rates of extracellular metabolites, such as glucose, ammonium and oxygen, as well as secretions rates of acetate, glycerol, ethanol, succinate, pyruvate and carbon dioxide. The biomass equation (or flux), which was the objective function, was changed depending on the growth conditions evaluated according to the data provide in Table ##TAB##3##4##.</p>", "<title>Large-scale gene essentiality simulations</title>", "<p>The impact of individual gene deletions on cell growth of <italic>iIN800 </italic>was evaluated by eliminating the reaction(s) corresponding to each gene in the model from the stoichiometric matrix <bold><italic>S </italic></bold>and then simulating growth of the mutant by FBA. The <italic>in silico </italic>gene essentialities were simulated for growth on rich- and minimal-medium. For minimal media, different carbon sources (glucose, galactose, glycerol and ethanol), ammonium, sulphate and phosphate were evaluated. For rich media, the uptake fluxes of amino acids, purines and pyrimidines were added as additional constraints as previously described [##REF##14506848##18##]. The <italic>in silico </italic>simulations were compared to experimental data available in the MIPS and SGD databases and from competitive growth assays [##REF##10436161##34##] as well as yeast mutant array experiments [##REF##16204195##22##]. The power of <italic>iIN800 </italic>to predict gene essentiality was evaluated based on the criteria defined as follows:</p>", "<p></p>", "<p></p>", "<p></p>", "<p></p>", "<p></p>", "<p></p>", "<p>where TP = true positive, TN = true negative, FP = false positive, FN = false negative. Positive and negative values referred to viable and lethal phenotype, respectively.</p>", "<title>Reporter Metabolite determination</title>", "<p>Published microarray data were retrieved from Gene Expression Omnibus (GEO) [##REF##16939800##49##]. The CEL files were normalized by the dChip software [##REF##11842437##50##] in order to minimize overall intensity variation among a set of chips. The statistical test of significance was done by ANOVA or student t-test for p-value calculation.</p>", "<p>Briefly, we describe the Reporter Metabolite calculations. The genome-scale model was converted to a bipartite undirected graph. In this graph each metabolite node has as neighbors the enzymes catalyzing the formation and consumption of the metabolite. The transcriptome data were mapped on the enzyme nodes using the significant values of gene expression. The normal commutative distribution was used to convert the p-values to a Z-score for further calculations. To identify an importance of metabolites in the metabolic network of the particular experimental conditions, the reporter algorithm was applied as described earlier [##REF##15710883##14##].</p>", "<title>Inferring regulatory modules from Reporter Metabolites</title>", "<p>The interactome network was initially constructed with data obtained from YPD [##REF##12073323##43##], ChIP-chip databases [##REF##15343339##51##] (protein-DNA interaction) and BioGRID [##REF##16381927##52##] (protein-protein interaction). The candidate genes of high scoring Reporter Metabolites were retrieved from the bipartite metabolite-gene encoding enzyme interaction graph. They were then used to identify subnetworks from the interactome network [##REF##10902179##53##]. Significantly changing p-values from microarray data were mapped on the subnetwork and then also genes having a p-value &lt; 0.01 directly connected with the Reporter Metabolites. The module was visualized by Cytoscape software [##REF##14597658##54##].</p>" ]

[ "<title>Results and discussion</title>", "<title>Model reconstruction and characteristics of <italic>iIN800</italic></title>", "<p>Due to the complexity of compartmentalization used in <italic>iND750 </italic>and the smaller scope of <italic>iLL672</italic>, the metabolic model <italic>iFF708 </italic>was selected as a template for the development of the model <italic>iIN800</italic>. Pathway and reaction databases (e.g. KEGG), online resources (e.g. SGD), and literature were used to expand <italic>iFF708</italic>, with particular focus on lipid metabolism. <italic>iIN800 </italic>contains 340 total reactions in lipid metabolism, more than at least 143 reactions greater than previous models (Table ##TAB##0##1##).</p>", "<p>To compare metabolic characteristics of the different <italic>in silico </italic>models, lipid metabolism was classified into unique sub-categories (e.g. mitochondrial fatty acid synthesis, ergosterol biosynthesis) (Table ##TAB##0##1##). Fatty acid synthesis and elongation accounted for three of these sub-categories. In contrast to previous models, <italic>iIN800 </italic>incorporates fatty acid biosynthesis in both mitochondria and the cytosol. Fatty acid synthesis, which involves iterative malonyl-CoA condensations that result in a growing chain of fatty acids, is catalyzed by four major enzymes: β-ketoacyl-ACP synthase (a condensing enzyme), β-ketoacyl-ACP reductase, β-dehydroxyacyl-ACP dehydratase and enoyl-ACP reductase. In the cytosol, these enzymes are encoded by the multifunctional <italic>FAS1 </italic>and <italic>FAS2</italic>. In the mitochondria, however, fatty acid synthesis is carried out by the products encoded by <italic>CEM1, OAR1, HTD2 </italic>and <italic>ETR1</italic>. These ORFs were missing from previous models, which prevented simulation of mitochondrial fatty acid synthesis. Fatty acid elongation, which leads to the production of long-chain fatty acids, was not included in <italic>iFF708</italic>, but was also updated in <italic>iIN800</italic>. Including fatty acid elongation resulted in the addition of four major biochemical reaction steps: condensing enzyme, 3-ketoacyl-CoA reductase, enoyl-CoA dehydratase and enoyl-CoA reductase [##REF##12087109##24##]. These reactions are carried out by the enzymes encoded by <italic>ELO1, ELO2, ELO3, IFA38 </italic>and <italic>TSC13</italic>. While the gene encoding enoyl-CoA dehydratase has not been identified in <italic>S. cerevisiae</italic>, the reaction was inferred due to the identification of long chain fatty acids in yeast.</p>", "<p>β-oxidation is the process where fatty acids, after becoming activated in the form of acyl-CoAs, are broken down to make acetyl-CoA, and ultimately energy. <italic>FAT1</italic>, encoding an enzyme for long-chain fatty acid activation was missing in <italic>iFF708 </italic>and <italic>iLL672</italic>. The genes <italic>SPS19, ECI1 </italic>and <italic>DCI1 </italic>are also now included in <italic>iIN800</italic>. As a result, <italic>iIN800 </italic>can simulate the oxidation of unsaturated fatty acids.</p>", "<p>Sphingolipid synthesis reactions were added to <italic>iIN800 </italic>according to a recently reported model [##REF##15674294##25##], resulting in more sphingolipid reactions than the template <italic>iFF708</italic>. Sphingolipid synthesis is the only sub-category in <italic>iIN800 </italic>with a significantly lower reaction tally than <italic>iND750</italic>. This is because <italic>iND750 </italic>incorporated both C24:0 and C26:0 as very long-chain fatty acids (the back bone of sphingolipids) to produce ceramides. Because the amount of very long chain fatty acids in <italic>S. cerevisiae </italic>is so low relative to other fatty acid species (&lt;2% of total fatty acid pool) [##REF##12087109##24##,##REF##4577752##26##], <italic>iIN800 </italic>treats very long chain fatty acids as a single metabolite. As a result, fewer reactions are present in sphingolipid synthesis.</p>", "<p>Relative to other models, only minor changes in the biosynthesis of phospholipids and triacylglycerides as well as ergosterol were introduced in <italic>iIN800</italic>. However, esterification of sterols and degradation of lipids, which were not included in all other previous models, are present in <italic>iIN800 </italic>(Table ##TAB##0##1##). Finally, 26 ORFs encoding for tRNA synthesis and one related enzyme, lipoamide dehydrogenase as well as 14 ORFs encoding transporters were also included in <italic>iIN800</italic>. The additionally included ORFs and their related references as well as detailed comparisons of reactions in lipid metabolism of all reported models are given in Additional files ##SUPPL##0##1## and ##SUPPL##1##2##, respectively.</p>", "<p>In summary, <italic>iIN800 </italic>was reconstructed from 17.2% of the characterized ORFs in yeast and contains 1446 metabolic reactions and 1013 metabolites in total. This model is relatively more comprehensive as compared with previously described models (Table ##TAB##1##2##). The network characteristics of <italic>iIN800 </italic>and the starting model <italic>iFF708 </italic>are shown in Table ##TAB##2##3##. Within lipid metabolism, we have incorporated many new reactions in mitochondrial fatty acid synthesis, cytosolic fatty acid synthesis, fatty acid elongation, fatty acid activation and β-oxidation, sphingolipid synthesis, ergosterol esterification, and lipid degradation (Table ##TAB##0##1##). 96 new reactions are derived from biochemical and physical considerations. These reactions mostly describe transportation of fatty acids and lipids across the mitochondria and the plasma membrane. To visualize the model <italic>iIN800</italic>, we constructed a comprehensive metabolic map using ReMapper software (Figure ##FIG##0##1##). This visualized map provides a method for globally plotting transcript and flux data onto <italic>iIN800</italic>. The source file is available for download (see Methods).</p>", "<title>Improved biomass equation</title>", "<p>The biomass equation is crucial for using genome-scale models to simulate growth using flux balance analysis (FBA). Therefore, an important consideration in the development of <italic>iIN800 </italic>was to address the concern that the biomass composition of <italic>S. cerevisiae </italic>changes under different growth conditions. For example, during growth on excess glucose the carbohydrate content increases and during growth on excess ammonium the protein content increases.</p>", "<p>To assess the sensitivity of flux simulations using <italic>iIN800 </italic>towards changes in the macro-molecular composition, we performed constraint-based simulations by varying the protein, RNA, carbohydrate and lipid content of the biomass in physiological relevant ranges based on previous experimental reports [##UREF##0##27##, ####REF##12111150##28##, ##REF##4297785##29####4297785##29##], from 35–65%, 3.5–12%, 15–50% and 2–15%, respectively. Specifically, glucose and ammonium uptake rates were minimized for both glucose- and ammonium-limited growth conditions, respectively, using different macromolecular compositions at fixed growth rates, (note: this is the same mathematical problem as fixing uptake rates and maximizing growth rate). In this way, we could compare the differences between glucose- and ammonium-limited growth conditions. The results are illustrated in Figure ##FIG##1##2##. An interesting finding was that the protein content strongly affects the uptake rates at both glucose- and ammonium-limited conditions, albeit to a greater extent in ammonium-limited conditions (Fig. ##FIG##1##2A##). The carbohydrate content on the other hand does not have an impact on the ammonium uptake rate, it strongly impacts the glucose uptake rate (Fig. ##FIG##1##2C##). The RNA content and the lipid content have only a minor impact on growth (Figures ##FIG##1##2B## and ##FIG##1##2D##).</p>", "<p>In summary, the sensitivity analysis shows that the biomass composition can significantly impact predictions made with genome-scale metabolic models to varying degrees based on different growth conditions. We therefore present new biomass equations to be used under C-limited and N-limited growth conditions, respectively. These compositions result from previous studies and our own measurements of lipids and fatty acids across multiple N-limited and C-limited growth conditions (data not shown). Using a separate biomass composition for N-limited cultures has not been proposed previously. The N-limited biomass equation is therefore new. Relative to previous C-limited biomass compositions, the most dramatic changes in our here proposed biomass equation is with respect to the lipids and fatty acids (Table ##TAB##3##4##). While our sensitivity analysis suggests that these components will most likely only lead to a small improvement in the accuracy of C-limited flux simulations, they may play an important role in lethality prediction by the model, as the addition of extra components in the biomass equation will give a higher resolution.</p>", "<title>Growth simulation capability</title>", "<p><italic>In silico </italic>genome-scale models are most generally used to predict various phenotypes. These include growth rates and extracellular secretion rates of metabolite products, as well as uptake rates of nutrients. In addition, models can be employed to explore active route(s) in metabolic pathways under certain growth conditions as illustrated for a genome-scale metabolic model of <italic>E. coli </italic>[##REF##11175725##30##, ####REF##17593909##31##, ##REF##12952533##32####12952533##32##] as well as for one of the <italic>S. cerevisiae </italic>genome-scale metabolic models [##REF##14578455##17##].</p>", "<p>To validate <italic>iIN800</italic>, we first investigated the model's ability to simulate aerobic and anaerobic growth in glucose- or ammonium-limited conditions. Several published chemostat datasets were used as experimental references. As shown in Figure ##FIG##2##3##, the results from the computational growth prediction agreed with experimental measurements. Less than 10% relative error was observed (Figure ##FIG##2##3##). The details of the simulations and the corresponding reference data are given in Additional file ##SUPPL##2##3##. Intracellular fluxes can be easily visualized using the ReMapper software and our model (Additional files ##SUPPL##3##4## and ##SUPPL##4##5##).</p>", "<p>Since the new biomass equations would be expected to impact the overall flux distributions, we used ¹³C-flux analysis data to further confirm the computed intracellular fluxes. Specifically, fluxes in the central carbon metabolism at two different growth conditions were both measured by ¹³C-labeling experiments and calculated by FBA using <italic>iIN800</italic>. The model validation is shown in Figure ##FIG##3##4##. There is a high degree of agreement between the predicted and experimental fluxes in the central metabolism, with the exception of fluxes through the pentose phosphate pathway (PPP). Using FBA, the flux through the PPP is largely determined by the requirement for NADPH, and it has earlier been shown difficult to balance NADPH production and consumption [##REF##11157958##33##]. This may explain why the FBA simulations under-predict the flux through this pathway.</p>", "<title>Evaluation of large-scale gene deletion</title>", "<p>To verify further <italic>iIN800</italic>, we investigated the ability of the model to predict for growth viability due to a single gene deletion. <italic>In silico </italic>deletion phenotype predictions were examined for the new model with cells grown in both minimal media with a sole carbon source (glucose, galactose, glycerol and ethanol) and with rich media (YPD). <italic>iIN800 </italic>was assessed for its ability to make correct predictions based on experimental data [##REF##16204195##22##,##REF##10436161##34##]. A summary of the <italic>in silico </italic>single gene deletion predictions are given in Table ##TAB##4##5##. The overall prediction rate of <italic>iIN800</italic>, derived from 3392 total predictions, was 89.36%, with 95.50% sensitivity and 38.69% selectivity. The evaluation of the mean of a confusion matrix as the geometric mean of <italic>iIN800 </italic>equals 60.79%. The performance of the <italic>iIN800 </italic>model has improved by ~2% and ~7% in terms of overall prediction rate compared with the models <italic>iFF708 </italic>and <italic>iND750</italic>, respectively. We believe that the improvement is mainly due to upgrades in the biomass equation, which is consistent with results from Kuepfer <italic>et al</italic>. demonstrating that more accurate biomass compositions lead to improved lethality predictions [##REF##16204195##22##]. The false predictions might be due to missing information in gene regulation, biomass compositions, dead-end reactions and medium composition, especially in the rich medium [##REF##14506848##18##,##REF##15197165##19##]</p>", "<title>Integration of transcriptome data with genome-scale metabolic models</title>", "<p>Genome-scale metabolic models have shown promise for identifying Reporter Metabolites, defined as metabolites whose neighboring genes in a bipartite metabolic graph are most significantly affected and respond as a group to genetic or environmental perturbations [##REF##15710883##14##]. Such an approach has previously been used to reveal important regulatory hot-spots in metabolism from genome-wide expression data and has demonstrated promise for integrating omic data using network topology. To highlight the importance and utility of having a more complete metabolic model in this integrated analysis, the genome-scale models <italic>iIN800 </italic>and <italic>iFF708 </italic>were used to calculate Reporter Metabolites. Multiple sets of transcriptome data were used for analysis. Lists of the top thirty most significant Reporter Metabolites for several perturbations are compared between <italic>iIN800 </italic>and <italic>iFF708 </italic>in Table ##TAB##5##6##, and Reporter Metabolites unique to <italic>iIN800 </italic>are marked in bold.</p>", "<p>First, transcriptome data from the yeast metabolic cycle [##REF##16254148##35##] were analyzed. Notably, the reporter algorithm identified unique Reporter Metabolites with <italic>iIN800 </italic>that would have been missed if <italic>iFF708 </italic>was used as the scaffold (Table ##TAB##5##6##). The most dramatic difference was observed for the reductive charging phase of the metabolic cell cycle. While both models revealed the importance of regulation controlling the cellular response at glycogen, trehalose, UDP-glucose, glucose-6-P and glucose nodes, only <italic>iIN800 </italic>was able to identify key intermediates in β-oxidation. For example, <italic>iIN800 </italic>identified <italic>trans</italic>-3-acyl-CoAs, <italic>trans</italic>-2-acyl-CoAs, 3-keto-acyl-CoAs and some fatty acids as Reporter Metabolites (Table ##TAB##5##6##). This result demonstrates the advantage of expanding the metabolic model to include a much more detailed description of lipid metabolism. Namely, we can now use the genome-scale metabolic model to identify the regulatory importance of lipid precursors and intermediates at different physiological conditions or at different phases of cellular growth. Searching for highly co-regulated subnetworks that implicate lipid genes is also now possible.</p>", "<p>Further demonstrations of the applicability of <italic>iIN800 </italic>as a scaffold to integrate omic data were performed by analyzing transcriptome data derived from nutrient-limited [##REF##15496405##36##], oxygen-limited [##REF##15496405##36##] and temperature stress conditions [##UREF##1##37##] Previously, mRNA and protein levels of genes and enzymes in fatty acid catabolism have been shown to be significantly different between carbon-limited and nitrogen limited growth [##REF##16738570##38##]. When comparing these conditions, only <italic>iIN800 </italic>was able to identify fatty acids as Reporter Metabolites (Table ##TAB##5##6##). In anaerobic yeast cultivation, oleic acid has to be added to the medium because unsaturated fatty acids synthesis is not possible; therefore, the expression of genes in this pathway is induced by the function of the ORE element [##REF##9813046##39##]. Consistent with this observed cellular response, only <italic>iIN800</italic>, with identified Reporter Metabolites involved in β-oxidation (Table ##TAB##5##6##). Similarly, <italic>iIN800 </italic>was able to highlight the importance of unsaturated fatty acids when comparing high and low temperature cultivations (Table ##TAB##5##6##), which is known to be important for maintaining proper membrane structure and fluidity [##REF##17071783##40##].</p>", "<p>Without the expanded model, the importance of cellular regulation stemming from lipid metabolism would be missed in analyses where metabolic topology is used for integrating data. As an illustration, we integrated results from our Reporter Metabolite analysis with known protein-protein and protein-DNA interaction networks to infer regulatory structure. First, genes associated to Reporter Metabolites in lipid metabolism unique to <italic>iIN800 </italic>and determined when comparing carbon- and nitrogen-limited growth (decanoyl-CoA, dodecanoyl-CoA, <italic>trans</italic>-2-C141-CoA, <italic>trans</italic>-2-C161-CoA, <italic>trans</italic>-2-C181-CoA) were identified. These genes were then used to search for highly regulated subnetworks within a protein-protein and protein-DNA interaction network. By applying a p-value threshold of 0.01 to filter for genes with significant gene expression, we inferred a regulatory network controlling the expression of lipid metabolism genes associated to the Reporter Metabolites (Figure ##FIG##4##5##). Strikingly, regulators at the top of this hierarchy are consistent with those previously known to be significantly changed between carbon- and nitrogen-limited growth. These include: <italic>SNF1, SNF4, MIG1 </italic>and <italic>ADR1 </italic>(glucose repression), <italic>OAF1 </italic>(β-oxidation), and <italic>INO1 </italic>and <italic>INO4 </italic>(phospholipid synthesis), among others. Previously reported genome-scale models are not capable of being used as scaffolds for implicating the conditional response of these lipid metabolism regulators because they lack a detailed description of lipid metabolism.</p>" ]

[ "<title>Results and discussion</title>", "<title>Model reconstruction and characteristics of <italic>iIN800</italic></title>", "<p>Due to the complexity of compartmentalization used in <italic>iND750 </italic>and the smaller scope of <italic>iLL672</italic>, the metabolic model <italic>iFF708 </italic>was selected as a template for the development of the model <italic>iIN800</italic>. Pathway and reaction databases (e.g. KEGG), online resources (e.g. SGD), and literature were used to expand <italic>iFF708</italic>, with particular focus on lipid metabolism. <italic>iIN800 </italic>contains 340 total reactions in lipid metabolism, more than at least 143 reactions greater than previous models (Table ##TAB##0##1##).</p>", "<p>To compare metabolic characteristics of the different <italic>in silico </italic>models, lipid metabolism was classified into unique sub-categories (e.g. mitochondrial fatty acid synthesis, ergosterol biosynthesis) (Table ##TAB##0##1##). Fatty acid synthesis and elongation accounted for three of these sub-categories. In contrast to previous models, <italic>iIN800 </italic>incorporates fatty acid biosynthesis in both mitochondria and the cytosol. Fatty acid synthesis, which involves iterative malonyl-CoA condensations that result in a growing chain of fatty acids, is catalyzed by four major enzymes: β-ketoacyl-ACP synthase (a condensing enzyme), β-ketoacyl-ACP reductase, β-dehydroxyacyl-ACP dehydratase and enoyl-ACP reductase. In the cytosol, these enzymes are encoded by the multifunctional <italic>FAS1 </italic>and <italic>FAS2</italic>. In the mitochondria, however, fatty acid synthesis is carried out by the products encoded by <italic>CEM1, OAR1, HTD2 </italic>and <italic>ETR1</italic>. These ORFs were missing from previous models, which prevented simulation of mitochondrial fatty acid synthesis. Fatty acid elongation, which leads to the production of long-chain fatty acids, was not included in <italic>iFF708</italic>, but was also updated in <italic>iIN800</italic>. Including fatty acid elongation resulted in the addition of four major biochemical reaction steps: condensing enzyme, 3-ketoacyl-CoA reductase, enoyl-CoA dehydratase and enoyl-CoA reductase [##REF##12087109##24##]. These reactions are carried out by the enzymes encoded by <italic>ELO1, ELO2, ELO3, IFA38 </italic>and <italic>TSC13</italic>. While the gene encoding enoyl-CoA dehydratase has not been identified in <italic>S. cerevisiae</italic>, the reaction was inferred due to the identification of long chain fatty acids in yeast.</p>", "<p>β-oxidation is the process where fatty acids, after becoming activated in the form of acyl-CoAs, are broken down to make acetyl-CoA, and ultimately energy. <italic>FAT1</italic>, encoding an enzyme for long-chain fatty acid activation was missing in <italic>iFF708 </italic>and <italic>iLL672</italic>. The genes <italic>SPS19, ECI1 </italic>and <italic>DCI1 </italic>are also now included in <italic>iIN800</italic>. As a result, <italic>iIN800 </italic>can simulate the oxidation of unsaturated fatty acids.</p>", "<p>Sphingolipid synthesis reactions were added to <italic>iIN800 </italic>according to a recently reported model [##REF##15674294##25##], resulting in more sphingolipid reactions than the template <italic>iFF708</italic>. Sphingolipid synthesis is the only sub-category in <italic>iIN800 </italic>with a significantly lower reaction tally than <italic>iND750</italic>. This is because <italic>iND750 </italic>incorporated both C24:0 and C26:0 as very long-chain fatty acids (the back bone of sphingolipids) to produce ceramides. Because the amount of very long chain fatty acids in <italic>S. cerevisiae </italic>is so low relative to other fatty acid species (&lt;2% of total fatty acid pool) [##REF##12087109##24##,##REF##4577752##26##], <italic>iIN800 </italic>treats very long chain fatty acids as a single metabolite. As a result, fewer reactions are present in sphingolipid synthesis.</p>", "<p>Relative to other models, only minor changes in the biosynthesis of phospholipids and triacylglycerides as well as ergosterol were introduced in <italic>iIN800</italic>. However, esterification of sterols and degradation of lipids, which were not included in all other previous models, are present in <italic>iIN800 </italic>(Table ##TAB##0##1##). Finally, 26 ORFs encoding for tRNA synthesis and one related enzyme, lipoamide dehydrogenase as well as 14 ORFs encoding transporters were also included in <italic>iIN800</italic>. The additionally included ORFs and their related references as well as detailed comparisons of reactions in lipid metabolism of all reported models are given in Additional files ##SUPPL##0##1## and ##SUPPL##1##2##, respectively.</p>", "<p>In summary, <italic>iIN800 </italic>was reconstructed from 17.2% of the characterized ORFs in yeast and contains 1446 metabolic reactions and 1013 metabolites in total. This model is relatively more comprehensive as compared with previously described models (Table ##TAB##1##2##). The network characteristics of <italic>iIN800 </italic>and the starting model <italic>iFF708 </italic>are shown in Table ##TAB##2##3##. Within lipid metabolism, we have incorporated many new reactions in mitochondrial fatty acid synthesis, cytosolic fatty acid synthesis, fatty acid elongation, fatty acid activation and β-oxidation, sphingolipid synthesis, ergosterol esterification, and lipid degradation (Table ##TAB##0##1##). 96 new reactions are derived from biochemical and physical considerations. These reactions mostly describe transportation of fatty acids and lipids across the mitochondria and the plasma membrane. To visualize the model <italic>iIN800</italic>, we constructed a comprehensive metabolic map using ReMapper software (Figure ##FIG##0##1##). This visualized map provides a method for globally plotting transcript and flux data onto <italic>iIN800</italic>. The source file is available for download (see Methods).</p>", "<title>Improved biomass equation</title>", "<p>The biomass equation is crucial for using genome-scale models to simulate growth using flux balance analysis (FBA). Therefore, an important consideration in the development of <italic>iIN800 </italic>was to address the concern that the biomass composition of <italic>S. cerevisiae </italic>changes under different growth conditions. For example, during growth on excess glucose the carbohydrate content increases and during growth on excess ammonium the protein content increases.</p>", "<p>To assess the sensitivity of flux simulations using <italic>iIN800 </italic>towards changes in the macro-molecular composition, we performed constraint-based simulations by varying the protein, RNA, carbohydrate and lipid content of the biomass in physiological relevant ranges based on previous experimental reports [##UREF##0##27##, ####REF##12111150##28##, ##REF##4297785##29####4297785##29##], from 35–65%, 3.5–12%, 15–50% and 2–15%, respectively. Specifically, glucose and ammonium uptake rates were minimized for both glucose- and ammonium-limited growth conditions, respectively, using different macromolecular compositions at fixed growth rates, (note: this is the same mathematical problem as fixing uptake rates and maximizing growth rate). In this way, we could compare the differences between glucose- and ammonium-limited growth conditions. The results are illustrated in Figure ##FIG##1##2##. An interesting finding was that the protein content strongly affects the uptake rates at both glucose- and ammonium-limited conditions, albeit to a greater extent in ammonium-limited conditions (Fig. ##FIG##1##2A##). The carbohydrate content on the other hand does not have an impact on the ammonium uptake rate, it strongly impacts the glucose uptake rate (Fig. ##FIG##1##2C##). The RNA content and the lipid content have only a minor impact on growth (Figures ##FIG##1##2B## and ##FIG##1##2D##).</p>", "<p>In summary, the sensitivity analysis shows that the biomass composition can significantly impact predictions made with genome-scale metabolic models to varying degrees based on different growth conditions. We therefore present new biomass equations to be used under C-limited and N-limited growth conditions, respectively. These compositions result from previous studies and our own measurements of lipids and fatty acids across multiple N-limited and C-limited growth conditions (data not shown). Using a separate biomass composition for N-limited cultures has not been proposed previously. The N-limited biomass equation is therefore new. Relative to previous C-limited biomass compositions, the most dramatic changes in our here proposed biomass equation is with respect to the lipids and fatty acids (Table ##TAB##3##4##). While our sensitivity analysis suggests that these components will most likely only lead to a small improvement in the accuracy of C-limited flux simulations, they may play an important role in lethality prediction by the model, as the addition of extra components in the biomass equation will give a higher resolution.</p>", "<title>Growth simulation capability</title>", "<p><italic>In silico </italic>genome-scale models are most generally used to predict various phenotypes. These include growth rates and extracellular secretion rates of metabolite products, as well as uptake rates of nutrients. In addition, models can be employed to explore active route(s) in metabolic pathways under certain growth conditions as illustrated for a genome-scale metabolic model of <italic>E. coli </italic>[##REF##11175725##30##, ####REF##17593909##31##, ##REF##12952533##32####12952533##32##] as well as for one of the <italic>S. cerevisiae </italic>genome-scale metabolic models [##REF##14578455##17##].</p>", "<p>To validate <italic>iIN800</italic>, we first investigated the model's ability to simulate aerobic and anaerobic growth in glucose- or ammonium-limited conditions. Several published chemostat datasets were used as experimental references. As shown in Figure ##FIG##2##3##, the results from the computational growth prediction agreed with experimental measurements. Less than 10% relative error was observed (Figure ##FIG##2##3##). The details of the simulations and the corresponding reference data are given in Additional file ##SUPPL##2##3##. Intracellular fluxes can be easily visualized using the ReMapper software and our model (Additional files ##SUPPL##3##4## and ##SUPPL##4##5##).</p>", "<p>Since the new biomass equations would be expected to impact the overall flux distributions, we used ¹³C-flux analysis data to further confirm the computed intracellular fluxes. Specifically, fluxes in the central carbon metabolism at two different growth conditions were both measured by ¹³C-labeling experiments and calculated by FBA using <italic>iIN800</italic>. The model validation is shown in Figure ##FIG##3##4##. There is a high degree of agreement between the predicted and experimental fluxes in the central metabolism, with the exception of fluxes through the pentose phosphate pathway (PPP). Using FBA, the flux through the PPP is largely determined by the requirement for NADPH, and it has earlier been shown difficult to balance NADPH production and consumption [##REF##11157958##33##]. This may explain why the FBA simulations under-predict the flux through this pathway.</p>", "<title>Evaluation of large-scale gene deletion</title>", "<p>To verify further <italic>iIN800</italic>, we investigated the ability of the model to predict for growth viability due to a single gene deletion. <italic>In silico </italic>deletion phenotype predictions were examined for the new model with cells grown in both minimal media with a sole carbon source (glucose, galactose, glycerol and ethanol) and with rich media (YPD). <italic>iIN800 </italic>was assessed for its ability to make correct predictions based on experimental data [##REF##16204195##22##,##REF##10436161##34##]. A summary of the <italic>in silico </italic>single gene deletion predictions are given in Table ##TAB##4##5##. The overall prediction rate of <italic>iIN800</italic>, derived from 3392 total predictions, was 89.36%, with 95.50% sensitivity and 38.69% selectivity. The evaluation of the mean of a confusion matrix as the geometric mean of <italic>iIN800 </italic>equals 60.79%. The performance of the <italic>iIN800 </italic>model has improved by ~2% and ~7% in terms of overall prediction rate compared with the models <italic>iFF708 </italic>and <italic>iND750</italic>, respectively. We believe that the improvement is mainly due to upgrades in the biomass equation, which is consistent with results from Kuepfer <italic>et al</italic>. demonstrating that more accurate biomass compositions lead to improved lethality predictions [##REF##16204195##22##]. The false predictions might be due to missing information in gene regulation, biomass compositions, dead-end reactions and medium composition, especially in the rich medium [##REF##14506848##18##,##REF##15197165##19##]</p>", "<title>Integration of transcriptome data with genome-scale metabolic models</title>", "<p>Genome-scale metabolic models have shown promise for identifying Reporter Metabolites, defined as metabolites whose neighboring genes in a bipartite metabolic graph are most significantly affected and respond as a group to genetic or environmental perturbations [##REF##15710883##14##]. Such an approach has previously been used to reveal important regulatory hot-spots in metabolism from genome-wide expression data and has demonstrated promise for integrating omic data using network topology. To highlight the importance and utility of having a more complete metabolic model in this integrated analysis, the genome-scale models <italic>iIN800 </italic>and <italic>iFF708 </italic>were used to calculate Reporter Metabolites. Multiple sets of transcriptome data were used for analysis. Lists of the top thirty most significant Reporter Metabolites for several perturbations are compared between <italic>iIN800 </italic>and <italic>iFF708 </italic>in Table ##TAB##5##6##, and Reporter Metabolites unique to <italic>iIN800 </italic>are marked in bold.</p>", "<p>First, transcriptome data from the yeast metabolic cycle [##REF##16254148##35##] were analyzed. Notably, the reporter algorithm identified unique Reporter Metabolites with <italic>iIN800 </italic>that would have been missed if <italic>iFF708 </italic>was used as the scaffold (Table ##TAB##5##6##). The most dramatic difference was observed for the reductive charging phase of the metabolic cell cycle. While both models revealed the importance of regulation controlling the cellular response at glycogen, trehalose, UDP-glucose, glucose-6-P and glucose nodes, only <italic>iIN800 </italic>was able to identify key intermediates in β-oxidation. For example, <italic>iIN800 </italic>identified <italic>trans</italic>-3-acyl-CoAs, <italic>trans</italic>-2-acyl-CoAs, 3-keto-acyl-CoAs and some fatty acids as Reporter Metabolites (Table ##TAB##5##6##). This result demonstrates the advantage of expanding the metabolic model to include a much more detailed description of lipid metabolism. Namely, we can now use the genome-scale metabolic model to identify the regulatory importance of lipid precursors and intermediates at different physiological conditions or at different phases of cellular growth. Searching for highly co-regulated subnetworks that implicate lipid genes is also now possible.</p>", "<p>Further demonstrations of the applicability of <italic>iIN800 </italic>as a scaffold to integrate omic data were performed by analyzing transcriptome data derived from nutrient-limited [##REF##15496405##36##], oxygen-limited [##REF##15496405##36##] and temperature stress conditions [##UREF##1##37##] Previously, mRNA and protein levels of genes and enzymes in fatty acid catabolism have been shown to be significantly different between carbon-limited and nitrogen limited growth [##REF##16738570##38##]. When comparing these conditions, only <italic>iIN800 </italic>was able to identify fatty acids as Reporter Metabolites (Table ##TAB##5##6##). In anaerobic yeast cultivation, oleic acid has to be added to the medium because unsaturated fatty acids synthesis is not possible; therefore, the expression of genes in this pathway is induced by the function of the ORE element [##REF##9813046##39##]. Consistent with this observed cellular response, only <italic>iIN800</italic>, with identified Reporter Metabolites involved in β-oxidation (Table ##TAB##5##6##). Similarly, <italic>iIN800 </italic>was able to highlight the importance of unsaturated fatty acids when comparing high and low temperature cultivations (Table ##TAB##5##6##), which is known to be important for maintaining proper membrane structure and fluidity [##REF##17071783##40##].</p>", "<p>Without the expanded model, the importance of cellular regulation stemming from lipid metabolism would be missed in analyses where metabolic topology is used for integrating data. As an illustration, we integrated results from our Reporter Metabolite analysis with known protein-protein and protein-DNA interaction networks to infer regulatory structure. First, genes associated to Reporter Metabolites in lipid metabolism unique to <italic>iIN800 </italic>and determined when comparing carbon- and nitrogen-limited growth (decanoyl-CoA, dodecanoyl-CoA, <italic>trans</italic>-2-C141-CoA, <italic>trans</italic>-2-C161-CoA, <italic>trans</italic>-2-C181-CoA) were identified. These genes were then used to search for highly regulated subnetworks within a protein-protein and protein-DNA interaction network. By applying a p-value threshold of 0.01 to filter for genes with significant gene expression, we inferred a regulatory network controlling the expression of lipid metabolism genes associated to the Reporter Metabolites (Figure ##FIG##4##5##). Strikingly, regulators at the top of this hierarchy are consistent with those previously known to be significantly changed between carbon- and nitrogen-limited growth. These include: <italic>SNF1, SNF4, MIG1 </italic>and <italic>ADR1 </italic>(glucose repression), <italic>OAF1 </italic>(β-oxidation), and <italic>INO1 </italic>and <italic>INO4 </italic>(phospholipid synthesis), among others. Previously reported genome-scale models are not capable of being used as scaffolds for implicating the conditional response of these lipid metabolism regulators because they lack a detailed description of lipid metabolism.</p>" ]

[ "<title>Conclusion</title>", "<p>Genome-scale metabolic models have emerged as a valuable tool in the post-genomic era for illustrating whole-cell functions based on the complete network of biochemical reactions. An iterative reconstruction process is required to achieve a comprehensive <italic>S. cerevisiae </italic>genome-scale metabolic model. In this work, we focused on improving the formulation of lipid metabolism relative to previously published <italic>S. cerevisiae </italic>genome-scale metabolic models. Validating the model and new biomass equations, the constraint-based simulation of <italic>iIN800 </italic>showed accurate predictions of cellular growth and is also consistent with ¹³C-labeling experiments. Furthermore, <italic>in silico </italic>gene essentialness predictions were found to be in high agreement with <italic>in vivo </italic>results. Finally, we show that <italic>iIN800</italic>, being more complete, is a better network scaffold for integration of multilevel omics data.</p>", "<p>In conclusion, by incorporating a more complete description of lipid metabolism, <italic>iIN800 </italic>is positioned to have a broader impact than previously described yeast models. Its capability of predictions were consistent with a number of experimental data both quantitatively (growth rate) and qualitatively (gene essentialness). Moreover, the new model is positioned to be used for studying the regulation and role of lipid metabolism during different growth conditions. With the high degree of homology in lipid metabolism between yeast and humans and emergence of lipidomics, this is expected to allow for new insights into the connection between lipid metabolism and overall cellular function for industrial and medical applications.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Up to now, there have been three published versions of a yeast genome-scale metabolic model: <italic>iFF708</italic>, <italic>iND750 </italic>and <italic>iLL672</italic>. All three models, however, lack a detailed description of lipid metabolism and thus are unable to be used as integrated scaffolds for gaining insights into lipid metabolism from multilevel omic measurement technologies (e.g. genome-wide mRNA levels). To overcome this limitation, we reconstructed a new version of the <italic>Saccharomyces cerevisiae </italic>genome-scale model, <italic>iIN800 </italic>that includes a more rigorous and detailed description of lipid metabolism.</p>", "<title>Results</title>", "<p>The reconstructed metabolic model comprises 1446 reactions and 1013 metabolites. Beyond incorporating new reactions involved in lipid metabolism, we also present new biomass equations that improve the predictive power of flux balance analysis simulations. Predictions of both growth capability and large scale <italic>in silico </italic>single gene deletions by <italic>iIN800 </italic>were consistent with experimental data. In addition, ¹³C-labeling experiments validated the new biomass equations and calculated intracellular fluxes. To demonstrate the applicability of <italic>iIN800</italic>, we show that the model can be used as a scaffold to reveal the regulatory importance of lipid metabolism precursors and intermediates that would have been missed in previous models from transcriptome datasets.</p>", "<title>Conclusion</title>", "<p>Performing integrated analyses using <italic>iIN800 </italic>as a network scaffold is shown to be a valuable tool for elucidating the behavior of complex metabolic networks, particularly for identifying regulatory targets in lipid metabolism that can be used for industrial applications or for understanding lipid disease states.</p>" ]

[ "<title>Authors' contributions</title>", "<p>IN designed the study, performed the metabolic reconstruction and validation, and contributed to manuscript writing. MCJ carried out the C¹³-labeling flux experiments, helped curate the model and contributed to manuscript writing. AM, CT, KL and SC contributed to the manuscript preparations, JN and SB participated in the concept and design of the study. All authors read and approved the final manuscript.</p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>The authors gratefully thank Mikael Rørdam Andersen and Kiran Raosaheb Patil for providing the ReMapper and the Reporter software, respectively. This work is supported by a grant from the National Center for Genetic Engineering and Biotechnology (BIOTEC) (grant number BT-B-06-NG-B5-4602). Intawat Nookaew gratefully acknowledges financial support by Thai Graduate Student Institute Science and Technology (TGIST). Michael C. Jewett is grateful to the NSF International Research Fellowship Program for supporting his work.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>The reconstructed <italic>S. cerevisiae </italic>genome-scale metabolic model <italic>iIN800</italic>.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Sensitivity analysis shows the influence of macromolecular composition on the simulated growth rate using <italic>iIN800</italic>.</bold> The simulations were performed for aerobic glucose- and ammonium-limited cultivations by varying (A) the protein content (35–65%), (B) the RNA content (3.5–12%), (C) the carbohydrate content (15–50%) and (D) the lipid content (2 – 15%).</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Comparison demonstrating <italic>in silico </italic>and <italic>in vivo </italic>growth rates at various cultivation conditions.</bold><italic>In silico </italic>predictions were performed using FBA with <italic>iIN800</italic>. Experimental measurements were taken from the literature (see text for references).</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p>Comparisons of the major intracellular fluxes in the central metabolism calculated using FBA with <italic>iIN800 </italic>and ¹³C-labeling metabolic flux analysis at a dilution rate of 0.05 h^-1of either aerobic or anaerobic glucose-limited conditions.</p></caption></fig>", "<fig position=\"float\" id=\"F5\"><label>Figure 5</label><caption><p><bold>Regulatory module implicated in the control of lipid metabolism genes associated to <italic>iIN800 </italic>Reporter Metabolites, which were determined by comparing N-limited and C-limited growth.</bold> Without the expanded model <italic>iIN800</italic>, the importance of cellular regulation stemming from lipid metabolism would be missed. High scoring Reporter Metabolites (diamonds), metabolic genes associated to Reporter Metabolites (circles), and genes encoding regulators (triangles). The blue, red, gray and green edges represent metabolite-gene interactions from the genome-scale metabolic model, protein-DNA interactions from ChIP-CHIP data, protein-DNA interactions from YPD and protein-protein interactions from BioGRID, respectively.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Comparison of the number of lipid metabolism reactions among yeast genome-scale metabolic models</p></caption><table frame=\"hsides\" rules=\"groups\"><tbody><tr><td align=\"left\"><bold>Model</bold></td><td align=\"center\"><bold><italic>iFF708</italic></bold></td><td align=\"center\"><bold><italic>iLL672</italic></bold></td><td align=\"center\"><bold><italic>iND750</italic></bold></td><td align=\"center\"><bold><italic>iIN800</italic></bold></td></tr><tr><td/><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\"><bold>Mitochondrial fatty acid synthesis</bold></td><td align=\"center\">14</td><td align=\"center\"><italic>0</italic></td><td align=\"center\">13</td><td align=\"center\">45</td></tr><tr><td align=\"left\"><bold>Cytosolic fatty acid synthesis</bold></td><td align=\"center\">17</td><td align=\"center\"><italic>18</italic></td><td align=\"center\">27</td><td align=\"center\">48</td></tr><tr><td align=\"left\"><bold>Fatty acid elongation</bold></td><td align=\"center\">0</td><td align=\"center\"><italic>4</italic></td><td align=\"center\">2</td><td align=\"center\">33</td></tr><tr><td align=\"left\"><bold>Fatty acid activation and beta-oxidation</bold></td><td align=\"center\">9</td><td align=\"center\"><italic>19</italic></td><td align=\"center\">53</td><td align=\"center\">65</td></tr><tr><td align=\"left\"><bold>Sphingolipid synthesis</bold></td><td align=\"center\">18</td><td align=\"center\"><italic>23</italic></td><td align=\"center\">37</td><td align=\"center\">27</td></tr><tr><td align=\"left\"><bold>Phospholipid and TAG synthesis</bold></td><td align=\"center\">37</td><td align=\"center\"><italic>37</italic></td><td align=\"center\">35</td><td align=\"center\">68</td></tr><tr><td align=\"left\"><bold>Ergosterol biosynthesis</bold></td><td align=\"center\">31</td><td align=\"center\"><italic>28</italic></td><td align=\"center\">30</td><td align=\"center\">29</td></tr><tr><td align=\"left\"><bold>Ergosterol esterification</bold></td><td align=\"center\">0</td><td align=\"center\"><italic>0</italic></td><td align=\"center\">0</td><td align=\"center\">2</td></tr><tr><td align=\"left\"><bold>Lipid degradation</bold></td><td align=\"center\">0</td><td align=\"center\"><italic>0</italic></td><td align=\"center\">0</td><td align=\"center\">23</td></tr><tr><td/><td colspan=\"4\"><hr/></td></tr><tr><td align=\"left\"><bold>Total</bold></td><td align=\"center\"><bold>126</bold></td><td align=\"center\"><bold><italic>129</italic></bold></td><td align=\"center\"><bold>197</bold></td><td align=\"center\"><bold>340</bold></td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Structure comparison of <italic>S. cerevisiae </italic>genome-scale metabolic models</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Model</bold></td><td align=\"center\"><bold>Genes</bold></td><td align=\"center\"><bold>Reactions</bold></td><td align=\"center\"><bold>Metabolites</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold><italic>iFF708</italic></bold></td><td align=\"center\">708(15.2%)*</td><td align=\"center\">1175</td><td align=\"center\">825</td></tr><tr><td align=\"left\"><bold><italic>iLL672</italic></bold></td><td align=\"center\">672(14.1%)*</td><td align=\"center\">1038</td><td align=\"center\">636</td></tr><tr><td align=\"left\"><bold><italic>iND750</italic></bold></td><td align=\"center\">750(16.1%)*</td><td align=\"center\">1489</td><td align=\"center\">972</td></tr><tr><td align=\"left\"><bold><italic>iIN800</italic></bold></td><td align=\"center\">800(17.2%)*</td><td align=\"center\">1446</td><td align=\"center\">1013</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Network characteristics of the reconstructed metabolic network of <italic>S. cerevisiae </italic>strain <italic>iFF708 </italic>and <italic>iIN800</italic></p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Model</bold></td><td align=\"center\"><bold><italic>iFF708</italic></bold></td><td align=\"center\"><bold><italic>iIN800</italic></bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Metabolites</bold></td><td align=\"center\"><bold>825</bold></td><td align=\"center\"><bold>1013</bold></td></tr><tr><td align=\"left\">Cytosolic metabolites</td><td align=\"center\">518</td><td align=\"center\">631</td></tr><tr><td align=\"left\">Mitochondrial metabolites</td><td align=\"center\">170</td><td align=\"center\">228</td></tr><tr><td align=\"left\">Extracellular metabolites</td><td align=\"center\">137</td><td align=\"center\">154</td></tr><tr><td colspan=\"3\"><hr/></td></tr><tr><td align=\"left\"><bold>Reactions</bold></td><td align=\"center\"><bold>1175</bold></td><td align=\"center\"><bold>1446</bold></td></tr><tr><td align=\"left\">Mitochondrial reactions</td><td align=\"center\">104</td><td align=\"center\">161</td></tr><tr><td align=\"left\">Cytosolic reactions</td><td align=\"center\">723</td><td align=\"center\">906</td></tr><tr><td colspan=\"3\"><hr/></td></tr><tr><td align=\"left\"><bold>Exchange fluxes</bold></td><td align=\"center\"><bold>348</bold></td><td align=\"center\"><bold>379</bold></td></tr><tr><td align=\"left\">Cytosolic exchange fluxes</td><td align=\"center\">286</td><td align=\"center\">304</td></tr><tr><td align=\"left\">Mitochondrial exchange fluxes</td><td align=\"center\">62</td><td align=\"center\">75</td></tr><tr><td colspan=\"3\"><hr/></td></tr><tr><td align=\"left\"><bold>Reactions with ORF assignments</bold></td><td align=\"center\">1075</td><td align=\"center\">1209</td></tr><tr><td align=\"left\"><bold>Biochemical and Physical consideration</bold></td><td align=\"center\">140</td><td align=\"center\">237</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Biomass composition</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Metabolites</bold></td><td align=\"center\" colspan=\"2\"><bold>Amount (mmol/gDW)</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold><underline>Amino acids</underline></bold></td><td align=\"center\"><bold>Carbon-limited</bold></td><td align=\"center\"><bold>Nitrogen-limited</bold></td></tr><tr><td align=\"left\">L-Alanine</td><td align=\"center\">0.357</td><td align=\"center\">0.252</td></tr><tr><td align=\"left\">L-Arginine</td><td align=\"center\">0.136</td><td align=\"center\">0.098</td></tr><tr><td align=\"left\">L-Asparagine</td><td align=\"center\">0.172</td><td align=\"center\">0.153</td></tr><tr><td align=\"left\">L-Aspartate</td><td align=\"center\">0.172</td><td align=\"center\">0.153</td></tr><tr><td align=\"left\">L-Cysteine</td><td align=\"center\">0.043</td><td align=\"center\">0.044</td></tr><tr><td align=\"left\">L-Glutamate</td><td align=\"center\">0.268</td><td align=\"center\">0.231</td></tr><tr><td align=\"left\">L-Glutamine</td><td align=\"center\">0.268</td><td align=\"center\">0.231</td></tr><tr><td align=\"left\">Glycine</td><td align=\"center\">0.325</td><td align=\"center\">0.278</td></tr><tr><td align=\"left\">L-Histidine</td><td align=\"center\">0.075</td><td align=\"center\">0.071</td></tr><tr><td align=\"left\">L-Isoleucine</td><td align=\"center\">0.172</td><td align=\"center\">0.142</td></tr><tr><td align=\"left\">L-Leucine</td><td align=\"center\">0.250</td><td align=\"center\">0.207</td></tr><tr><td align=\"left\">L-Lysine</td><td align=\"center\">0.239</td><td align=\"center\">0.204</td></tr><tr><td align=\"left\">L-Methionine</td><td align=\"center\">0.050</td><td align=\"center\">0.044</td></tr><tr><td align=\"left\">L-Phenylalanine</td><td align=\"center\">0.114</td><td align=\"center\">0.092</td></tr><tr><td align=\"left\">L-Proline</td><td align=\"center\">0.129</td><td align=\"center\">0.118</td></tr><tr><td align=\"left\">L-Serine</td><td align=\"center\">0.254</td><td align=\"center\">0.225</td></tr><tr><td align=\"left\">L-Threonine</td><td align=\"center\">0.197</td><td align=\"center\">0.160</td></tr><tr><td align=\"left\">L-Tryptophan</td><td align=\"center\">0.027</td><td align=\"center\">0.028</td></tr><tr><td align=\"left\">L-Tyrosine</td><td align=\"center\">0.096</td><td align=\"center\">0.068</td></tr><tr><td/><td/><td/></tr><tr><td align=\"left\"><bold><underline>Carbohydrates</underline></bold></td><td align=\"center\"><bold>Carbon-limited</bold></td><td align=\"center\"><bold>Nitrogen-limited</bold></td></tr><tr><td align=\"left\">Glycogen</td><td align=\"center\">0.519</td><td align=\"center\">0.667</td></tr><tr><td align=\"left\">alpha,alpha-Trehalose</td><td align=\"center\">0.023</td><td align=\"center\">0.085</td></tr><tr><td align=\"left\">Mannan</td><td align=\"center\">0.821</td><td align=\"center\">0.994</td></tr><tr><td align=\"left\">1,3-beta-D-Glucan</td><td align=\"center\">1.136</td><td align=\"center\">0.963</td></tr><tr><td/><td/><td/></tr><tr><td align=\"left\"><bold><underline>RNA</underline></bold></td><td align=\"center\"><bold>Carbon-limited</bold></td><td align=\"center\"><bold>Nitrogen-limited</bold></td></tr><tr><td align=\"left\">AMP</td><td align=\"center\">0.051</td><td align=\"center\">0.040</td></tr><tr><td align=\"left\">GMP</td><td align=\"center\">0.051</td><td align=\"center\">0.040</td></tr><tr><td align=\"left\">CMP</td><td align=\"center\">0.050</td><td align=\"center\">0.039</td></tr><tr><td align=\"left\">UMP</td><td align=\"center\">0.067</td><td align=\"center\">0.052</td></tr><tr><td/><td/><td/></tr><tr><td align=\"left\"><bold><underline>DNA</underline></bold></td><td align=\"center\"><bold>Carbon-limited</bold></td><td align=\"center\"><bold>Nitrogen-limited</bold></td></tr><tr><td align=\"left\">dAMP</td><td align=\"center\">0.004</td><td align=\"center\">0.004</td></tr><tr><td align=\"left\">dCMP</td><td align=\"center\">0.002</td><td align=\"center\">0.003</td></tr><tr><td align=\"left\">dTMP</td><td align=\"center\">0.004</td><td align=\"center\">0.004</td></tr><tr><td align=\"left\">dGMP</td><td align=\"center\">0.002</td><td align=\"center\">0.003</td></tr><tr><td/><td/><td/></tr><tr><td align=\"left\"><bold><underline>Lipids</underline></bold></td><td align=\"center\"><bold>Carbon-limited</bold></td><td align=\"center\"><bold>Nitrogen-limited</bold></td></tr><tr><td align=\"left\">Phosphatidylcholine</td><td align=\"center\">0.002884</td><td align=\"center\">0.001660</td></tr><tr><td align=\"left\">1-Phosphatidyl-D-myo-inositol</td><td align=\"center\">0.001531</td><td align=\"center\">0.001656</td></tr><tr><td align=\"left\">Phosphatidylserine</td><td align=\"center\">0.000373</td><td align=\"center\">0.000302</td></tr><tr><td align=\"left\">Phosphatidylethanolamine</td><td align=\"center\">0.000697</td><td align=\"center\">0.000083</td></tr><tr><td align=\"left\">Acyl_acids</td><td align=\"center\">0.000206</td><td align=\"center\">0.000723</td></tr><tr><td align=\"left\">Triacylglycerol</td><td align=\"center\">0.000781</td><td align=\"center\">0.003618</td></tr><tr><td align=\"left\">Ergosterol-ester</td><td align=\"center\">0.000812</td><td align=\"center\">0.004632</td></tr><tr><td align=\"left\">Ergosta-5,7,22,24(28)-tetraenol</td><td align=\"center\">0.000125</td><td align=\"center\">0.000167</td></tr><tr><td align=\"left\">Ergosterol</td><td align=\"center\">0.005603</td><td align=\"center\">0.005155</td></tr><tr><td align=\"left\">Zymosterol</td><td align=\"center\">0.000015</td><td align=\"center\">0.000051</td></tr><tr><td align=\"left\">Episterol</td><td align=\"center\">0.000096</td><td align=\"center\">0.000062</td></tr><tr><td align=\"left\">Fecosterol</td><td align=\"center\">0.000114</td><td align=\"center\">0.000068</td></tr><tr><td align=\"left\">Lanosterol</td><td align=\"center\">0.000032</td><td align=\"center\">0.000074</td></tr><tr><td align=\"left\">4,4-Dimethylzymosterol</td><td align=\"center\">0.000056</td><td align=\"center\">0.000046</td></tr><tr><td align=\"left\">Ceramide-I</td><td align=\"center\">0.000351</td><td align=\"center\">0.000075</td></tr><tr><td align=\"left\">Ceramide-II</td><td align=\"center\">0.000066</td><td align=\"center\">0.000009</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T5\"><label>Table 5</label><caption><p>Summary of large-scale single gene deletion evaluation of <italic>S. cerevisiae iIN800</italic>.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"5\">Minimal media on</td><td align=\"center\">Rich media</td><td/></tr><tr><td align=\"left\"><bold>Evaluation</bold></td><td align=\"center\"><bold>Glucose*</bold></td><td align=\"center\"><bold>Glucose**</bold></td><td align=\"center\"><bold>Galactose</bold></td><td align=\"center\"><bold>Glycerol</bold></td><td align=\"center\"><bold>Ethanol</bold></td><td align=\"center\"><bold>YPD</bold></td><td align=\"center\"><bold>Total</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>TP</bold></td><td align=\"center\">468</td><td align=\"center\">469</td><td align=\"center\">461</td><td align=\"center\">461</td><td align=\"center\">463</td><td align=\"center\">567</td><td align=\"center\"><bold>2889</bold></td></tr><tr><td align=\"left\"><bold>TN</bold></td><td align=\"center\">23</td><td align=\"center\">23</td><td align=\"center\">20</td><td align=\"center\">17</td><td align=\"center\">21</td><td align=\"center\">38</td><td align=\"center\"><bold>142</bold></td></tr><tr><td align=\"left\"><bold>FP</bold></td><td align=\"center\">37</td><td align=\"center\">37</td><td align=\"center\">42</td><td align=\"center\">45</td><td align=\"center\">43</td><td align=\"center\">21</td><td align=\"center\"><bold>225</bold></td></tr><tr><td align=\"left\"><bold>FN</bold></td><td align=\"center\">14</td><td align=\"center\">13</td><td align=\"center\">19</td><td align=\"center\">19</td><td align=\"center\">15</td><td align=\"center\">56</td><td align=\"center\"><bold>136</bold></td></tr><tr><td align=\"left\"><bold>Number of deletions</bold></td><td align=\"center\">542</td><td align=\"center\">542</td><td align=\"center\">542</td><td align=\"center\">542</td><td align=\"center\">542</td><td align=\"center\">682</td><td align=\"center\"><bold>3392</bold></td></tr><tr><td align=\"left\"><bold>Positive prediction rate</bold></td><td align=\"center\">92.67</td><td align=\"center\">92.69</td><td align=\"center\">91.65</td><td align=\"center\">91.11</td><td align=\"center\">91.50</td><td align=\"center\">96.43</td><td align=\"center\"><bold>92.77</bold></td></tr><tr><td align=\"left\"><bold>Negative prediction rate</bold></td><td align=\"center\">62.16</td><td align=\"center\">63.89</td><td align=\"center\">51.28</td><td align=\"center\">47.22</td><td align=\"center\">58.33</td><td align=\"center\">40.43</td><td align=\"center\"><bold>51.08</bold></td></tr><tr><td align=\"left\"><bold>Accuracy</bold></td><td align=\"center\">90.59</td><td align=\"center\">90.77</td><td align=\"center\">88.75</td><td align=\"center\">88.19</td><td align=\"center\">89.30</td><td align=\"center\">88.71</td><td align=\"center\"><bold>89.36</bold></td></tr><tr><td align=\"left\"><bold>Sensitivity</bold></td><td align=\"center\">97.10</td><td align=\"center\">97.30</td><td align=\"center\">96.04</td><td align=\"center\">96.04</td><td align=\"center\">96.86</td><td align=\"center\">91.01</td><td align=\"center\"><bold>95.50</bold></td></tr><tr><td align=\"left\"><bold>Selectivity</bold></td><td align=\"center\">38.33</td><td align=\"center\">38.33</td><td align=\"center\">32.26</td><td align=\"center\">27.42</td><td align=\"center\">32.81</td><td align=\"center\">64.41</td><td align=\"center\"><bold>38.69</bold></td></tr><tr><td align=\"left\"><bold>Geometric mean</bold></td><td align=\"center\">61.01</td><td align=\"center\">61.07</td><td align=\"center\">55.66</td><td align=\"center\">51.32</td><td align=\"center\">56.38</td><td align=\"center\">76.56</td><td align=\"center\"><bold>60.79</bold></td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T6\"><label>Table 6</label><caption><p>Top thirty Reporter Metabolites calculated from various perturbations. The Reporter Metabolite algorithm was performed with <italic>iIN800 </italic>and <italic>iFF708</italic>.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\" colspan=\"2\"><bold>Oxidative phase</bold>¹</td><td align=\"center\" colspan=\"2\"><bold>Reductive building phase</bold>¹</td><td align=\"center\" colspan=\"2\"><bold>Reductive charging Phase</bold>¹</td></tr></thead><tbody><tr><td align=\"left\"><bold><italic>iIN800</italic></bold></td><td align=\"left\"><bold><italic>iFF708</italic></bold></td><td align=\"left\"><bold><italic>iIN800</italic></bold></td><td align=\"left\"><bold><italic>iFF708</italic></bold></td><td align=\"left\"><bold><italic>iIN800</italic></bold></td><td align=\"left\"><bold><italic>iFF708</italic></bold></td></tr><tr><td colspan=\"6\"><hr/></td></tr><tr><td align=\"left\">AMP</td><td align=\"left\">IMP</td><td align=\"left\">AMPM</td><td align=\"left\">AMPM</td><td align=\"left\"><bold>Dodecanoyl-CoA*</bold></td><td align=\"left\">Acyl-CoA</td></tr><tr><td align=\"left\">IMP</td><td align=\"left\">Xanthosine 5'-phosphate</td><td align=\"left\">PyrophosphateM</td><td align=\"left\">tRNAM</td><td align=\"left\"><bold>Decanoyl-CoA*</bold></td><td align=\"left\">alpha,alpha-Trehalose</td></tr><tr><td align=\"left\">Pyrophosphate</td><td align=\"left\">L-Methionine</td><td align=\"left\">ATPM</td><td align=\"left\">PyrophosphateM</td><td align=\"left\"><bold>Trans-3-C16-CoA*</bold></td><td align=\"left\">Glycogen</td></tr><tr><td align=\"left\">L-Methionine</td><td align=\"left\">5-Phospho-alpha-D-ribose 1-PP</td><td align=\"left\">tRNAM</td><td align=\"left\">Porphobilinogen</td><td align=\"left\"><bold>Trans-3-C18-CoA*</bold></td><td align=\"left\">alpha,alpha'-Trehalose 6-phosphate</td></tr><tr><td align=\"left\">Xanthosine 5'-phosphate</td><td align=\"left\">L-Aspartate</td><td align=\"left\">H+M</td><td align=\"left\">L-TryptophanM</td><td align=\"left\"><bold>Trans-3-C14-CoA*</bold></td><td align=\"left\">alpha-D-Glucose</td></tr><tr><td align=\"left\">ATP</td><td align=\"left\">Sulfate</td><td align=\"left\">NADHM</td><td align=\"left\">L-Tryptophanyl-tRNA(Trp)M</td><td align=\"left\">alpha,alpha-Trehalose</td><td align=\"left\">Oxalosuccinate</td></tr><tr><td align=\"left\">5-Phospho-alpha-D-ribose 1-PP</td><td align=\"left\">Homocysteine</td><td align=\"left\">Porphobilinogen</td><td align=\"left\">Dolichyl beta-D-mannosyl-P</td><td align=\"left\">Glycogen</td><td align=\"left\">3-Oxoacyl-CoA</td></tr><tr><td align=\"left\">L-Serine</td><td align=\"left\">AMP</td><td align=\"left\">Dolichyl beta-D-mannosyl-P</td><td align=\"left\">Mannan</td><td align=\"left\">alpha,alpha'-Trehalose 6-phosphate</td><td align=\"left\">a Long-chain carboxylic acid</td></tr><tr><td align=\"left\">L-Aspartate</td><td align=\"left\">H+EXT</td><td align=\"left\">L-Tryptophanyl-tRNA(Trp)M</td><td align=\"left\">Xanthine</td><td align=\"left\">Oxalosuccinate</td><td align=\"left\">Carnitine</td></tr><tr><td align=\"left\">H+EXT</td><td align=\"left\">3-Phosphonooxypyruvate</td><td align=\"left\">Mannan</td><td align=\"left\">L-Asparaginyl-tRNA(Asn)M</td><td align=\"left\"><bold>Trans-2-C14-CoA*</bold></td><td align=\"left\">alpha-D-Glucose 6-phosphate</td></tr><tr><td align=\"left\">Homocysteine</td><td align=\"left\">N6-(L-1,3-Dicarboxypropyl)-L-lysine</td><td align=\"left\"><bold>tRNA(Ile)M*</bold></td><td align=\"left\">H+M</td><td align=\"left\"><bold>Trans-2-C16-CoA*</bold></td><td align=\"left\">UDPglucose</td></tr><tr><td align=\"left\">Sulfate</td><td align=\"left\">5,10-Methylenetetrahydrofolate</td><td align=\"left\"><bold>L-Isoleucyl-tRNA(Ile)M*</bold></td><td align=\"left\">Dolichyl phosphate</td><td align=\"left\"><bold>Trans-2-C18-CoA*</bold></td><td align=\"left\">Isocitrate</td></tr><tr><td align=\"left\">L-Glutamine</td><td align=\"left\">Aminoimidazole ribotide</td><td align=\"left\"><bold>tRNA(Thr)M*</bold></td><td align=\"left\">all-trans-Nonaprenyl-PP</td><td align=\"left\"><bold>3-keto-Dodecanoyl-CoA*</bold></td><td align=\"left\">D-Glucose 1-phosphate</td></tr><tr><td align=\"left\">L-Cysteine</td><td align=\"left\">L-Cystathionine</td><td align=\"left\"><bold>L-Threonyl-tRNA(Thr)M*</bold></td><td align=\"left\">NADHM</td><td align=\"left\"><bold>3-keto-Decanoyl-CoA*</bold></td><td align=\"left\">CoAM</td></tr><tr><td align=\"left\">L-Asparagine</td><td align=\"left\">L-Serine</td><td align=\"left\">Xanthine</td><td align=\"left\">ATPM</td><td align=\"left\"><bold>3-keto-Octanoyl-CoA*</bold></td><td align=\"left\">Acetyl-CoAM</td></tr><tr><td align=\"left\">S-Adenosyl-L-methionine</td><td align=\"left\">Uracil</td><td align=\"left\">Dolichyl phosphate</td><td align=\"left\">D-Mannose 6-phosphate</td><td align=\"left\"><bold>3-keto-Hexanoyl-CoA*</bold></td><td align=\"left\">CoA</td></tr><tr><td align=\"left\">Uracil</td><td align=\"left\">Sulfite</td><td align=\"left\">all-trans-Nonaprenyl-PP</td><td align=\"left\">UbiquinolM</td><td align=\"left\"><bold>3-keto-Butanoyl-CoA*</bold></td><td align=\"left\">O-Acetylcarnitine</td></tr><tr><td align=\"left\">5,10-Methylenetetrahydrofolate</td><td align=\"left\">5-amino-4-imidazolecarboxylate</td><td align=\"left\">L-Asparaginyl-tRNA(Asn)M</td><td align=\"left\">Ubiquinone-9M</td><td align=\"left\"><bold>Dodecanoic_acid*</bold></td><td align=\"left\">Succinate</td></tr><tr><td align=\"left\">3-Phosphonooxypyruvate</td><td align=\"left\">2-Hydroxybutane-1,2,4-tricarboxylate</td><td align=\"left\"><bold>tRNA(Phe)M*</bold></td><td align=\"left\">CO2M</td><td align=\"left\">Carnitine</td><td align=\"left\">(S)-3-Hydroxy-3-methylglutaryl-CoA</td></tr><tr><td align=\"left\">N6-(L-1,3-Dicarboxypropyl)-L-lysine</td><td align=\"left\">S-Adenosyl-L-methionine</td><td align=\"left\"><bold>L-Phenylalanyl-tRNA(Phe)M*</bold></td><td align=\"left\">Guanosine</td><td align=\"left\">alpha-D-Glucose</td><td align=\"left\">NAD+</td></tr><tr><td align=\"left\">L-Cystathionine</td><td align=\"left\">L-Asparagine</td><td align=\"left\">Intermediate_Methylzymosterol_II</td><td align=\"left\">IsocitrateM</td><td align=\"left\"><bold>Trans-2-4-diene-CoA*</bold></td><td align=\"left\">H2O2</td></tr><tr><td align=\"left\">NH3</td><td align=\"left\">5-Phosphoribosylamine</td><td align=\"left\">Intermediate_Zymosterol_II</td><td align=\"left\">GTPM</td><td align=\"left\">Isocitrate</td><td align=\"left\">Malate</td></tr><tr><td align=\"left\"><bold>tRNA(Phe)*</bold></td><td align=\"left\">GlycineM</td><td align=\"left\">UbiquinolM</td><td align=\"left\">GDPM</td><td align=\"left\">alpha-D-Glucose 6-phosphate</td><td align=\"left\">Maltose</td></tr><tr><td align=\"left\"><bold>L-Phenylalanyl-tRNA(Phe)*</bold></td><td align=\"left\">Guanine</td><td align=\"left\">D-Mannose 6-phosphate</td><td align=\"left\">ITPM</td><td align=\"left\">UDPglucose</td><td align=\"left\">(3S)-3-Hydroxyacyl-CoA</td></tr><tr><td align=\"left\">Tetrahydrofolate</td><td align=\"left\">L-Histidine</td><td align=\"left\">Ubiquinone-9M</td><td align=\"left\">IDPM</td><td align=\"left\">D-Glucose 1-phosphate</td><td align=\"left\">GLCxt</td></tr><tr><td align=\"left\">Guanine</td><td align=\"left\">N1-(5'-phosphoribosyl)acetamidine</td><td align=\"left\"><bold>tRNA(Asp)M*</bold></td><td align=\"left\">ITP</td><td align=\"left\">O-Acetylcarnitine</td><td align=\"left\">Glycerone phosphate</td></tr><tr><td align=\"left\">Sulfite</td><td align=\"left\">Tetrahydrofolate</td><td align=\"left\"><bold>L-Aspartyl-tRNA(Asp)M*</bold></td><td align=\"left\">IDP</td><td align=\"left\"><bold>Tetradecanoyl-CoA*</bold></td><td align=\"left\">O-AcetylcarnitineM</td></tr><tr><td align=\"left\">L-Histidine</td><td align=\"left\">alpha-D-Glutamyl phosphate</td><td align=\"left\"><bold>tRNA(Pro)M*</bold></td><td align=\"left\">Phosphatidate</td><td align=\"left\"><bold>Decanoic_acid*</bold></td><td align=\"left\">CarnitineM</td></tr><tr><td align=\"left\">5-amino-4-imidazolecarboxylate</td><td align=\"left\">HomoisocitrateM</td><td align=\"left\"><bold>L-Prolinyl-tRNA(Pro)M*</bold></td><td align=\"left\">C100ACPm</td><td align=\"left\">(S)-3-Hydroxy-3-methylglutaryl-CoA</td><td align=\"left\">D-Galactose</td></tr><tr><td align=\"left\">GlycineM</td><td align=\"left\">GMP</td><td align=\"left\">Pyrophosphate</td><td align=\"left\">Dodecanoyl-ACPM</td><td align=\"left\">H2O2</td><td align=\"left\">SuccinateM</td></tr><tr><td colspan=\"6\"><hr/></td></tr><tr><td align=\"left\" colspan=\"6\">* Metabolite is contained in <italic>iIN800 </italic>only<break/>1 = data from Tu, B. P., A. Kudlicki, et al. (2005)</td></tr><tr><td colspan=\"6\"><hr/></td></tr><tr><td align=\"center\" colspan=\"2\"><bold>Carbon- and Nitrogen-limited</bold>²</td><td align=\"center\" colspan=\"2\"><bold>Aerobic and Anaerobic</bold>²</td><td align=\"center\" colspan=\"2\"><bold>Temperature(30°C and 15°C)</bold>³</td></tr><tr><td colspan=\"6\"><hr/></td></tr><tr><td align=\"left\"><bold><italic>iIN800</italic></bold></td><td align=\"left\"><bold><italic>iFF708</italic></bold></td><td align=\"left\"><bold><italic>iIN800</italic></bold></td><td align=\"left\"><bold><italic>iFF708</italic></bold></td><td align=\"left\"><bold><italic>iIN800</italic></bold></td><td align=\"left\"><bold><italic>iFF708</italic></bold></td></tr><tr><td colspan=\"6\"><hr/></td></tr><tr><td align=\"left\">Glyoxylate</td><td align=\"left\">Glyoxylate</td><td align=\"left\">Oxygen</td><td align=\"left\">Ferricytochrome cM</td><td align=\"left\">IMP</td><td align=\"left\">IMP</td></tr><tr><td align=\"left\">GLUxt</td><td align=\"left\">L-Phenylalanine</td><td align=\"left\">Ferricytochrome cM</td><td align=\"left\">Ferrocytochrome cM</td><td align=\"left\">Tetrahydrofolate</td><td align=\"left\">Tetrahydrofolate</td></tr><tr><td align=\"left\">Isocitrate</td><td align=\"left\">GLUxt</td><td align=\"left\">Ferrocytochrome cM</td><td align=\"left\">Ubiquinone-9M</td><td align=\"left\">alpha,alpha-Trehalose</td><td align=\"left\">alpha,alpha-Trehalose</td></tr><tr><td align=\"left\">ALAxt</td><td align=\"left\">Isocitrate</td><td align=\"left\">Ubiquinone-9M</td><td align=\"left\">Oxygen</td><td align=\"left\"><bold>Hexadecanoyl-9-ene-CoA*</bold></td><td align=\"left\">D-Erythrose 4-phosphate</td></tr><tr><td align=\"left\">Malate</td><td align=\"left\">ALAxt</td><td align=\"left\">UbiquinolM</td><td align=\"left\">UbiquinolM</td><td align=\"left\"><bold>Octadecanoyl-9-ene-CoA*</bold></td><td align=\"left\">L-OrnithineM</td></tr><tr><td align=\"left\">Allantoate</td><td align=\"left\">Allantoate</td><td align=\"left\">ADPM</td><td align=\"left\">ADPM</td><td align=\"left\"><bold>Tetradecanoyl-9-ene-CoA*</bold></td><td align=\"left\">Xanthosine 5'-phosphate</td></tr><tr><td align=\"left\">SERxt</td><td align=\"left\">Malate</td><td align=\"left\">H+M</td><td align=\"left\">H+M</td><td align=\"left\">D-Erythrose 4-phosphate</td><td align=\"left\">N6-(L-1,3-Dicarboxypropyl)-L-lysine</td></tr><tr><td align=\"left\">L-Alanine</td><td align=\"left\">L-Alanine</td><td align=\"left\"><bold>Dodecanoyl-CoA*</bold></td><td align=\"left\">FADH2M</td><td align=\"left\">L-OrnithineM</td><td align=\"left\">NADH</td></tr><tr><td align=\"left\"><bold>Decanoyl-CoA*</bold></td><td align=\"left\">SERxt</td><td align=\"left\">FumarateM</td><td align=\"left\">FumarateM</td><td align=\"left\">Xanthosine 5'-phosphate</td><td align=\"left\">URIxt</td></tr><tr><td align=\"left\">ASNxt</td><td align=\"left\">ASNxt</td><td align=\"left\">OrthophosphateM</td><td align=\"left\">OrthophosphateM</td><td align=\"left\">N6-(L-1,3-Dicarboxypropyl)-L-lysine</td><td align=\"left\">1-Phosphatidyl-1D-myo-inositol 4-P</td></tr><tr><td align=\"left\">GLNxt</td><td align=\"left\">GLNxt</td><td align=\"left\">FADH2M</td><td align=\"left\">Sphinganine 1-phosphate</td><td align=\"left\">URIxt</td><td align=\"left\">Homocysteine</td></tr><tr><td align=\"left\">ILExt</td><td align=\"left\">ILExt</td><td align=\"left\"><bold>Hexadecanoyl-9-ene-CoA*</bold></td><td align=\"left\">ATPM</td><td align=\"left\">Homocysteine</td><td align=\"left\">1-Phosphatidyl-D-myo-inositol 4,5-PP</td></tr><tr><td align=\"left\">VALxt</td><td align=\"left\">VALxt</td><td align=\"left\"><bold>Octadecanoyl-9-ene-CoA*</bold></td><td align=\"left\">Fumarate</td><td align=\"left\">Octadecanoyl-CoA</td><td align=\"left\">N-Acetyl-L-glutamateM</td></tr><tr><td align=\"left\"><bold>Trans-2-C161-CoA*</bold></td><td align=\"left\">Ferricytochrome cM</td><td align=\"left\"><bold>Tetradecanoyl-9-ene-CoA*</bold></td><td align=\"left\">Glyoxylate</td><td align=\"left\">N-Acetyl-L-glutamateM</td><td align=\"left\">Dihydrofolate</td></tr><tr><td align=\"left\"><bold>Trans-2-C181-CoA*</bold></td><td align=\"left\">Ferrocytochrome cM</td><td align=\"left\">Sphinganine 1-phosphate</td><td align=\"left\">Isocitrate</td><td align=\"left\">Dihydrofolate</td><td align=\"left\">N2-Acetyl-L-ornithineM</td></tr><tr><td align=\"left\"><bold>Trans-2-C141-CoA*</bold></td><td align=\"left\">PHExt</td><td align=\"left\">Phytosphingosine 1-phosphate</td><td align=\"left\">ERGOSTxt</td><td align=\"left\">N2-Acetyl-L-ornithineM</td><td align=\"left\">Anthranilate</td></tr><tr><td align=\"left\">PHExt</td><td align=\"left\">L-Asparagine</td><td align=\"left\">Tetradecanoyl-Co</td><td align=\"left\">ZYMSTxt</td><td align=\"left\">Anthranilate</td><td align=\"left\">S-Adenosyl-L-homocysteine</td></tr><tr><td align=\"left\">Ferricytochrome cM</td><td align=\"left\">Allantoin</td><td align=\"left\">Fumarate</td><td align=\"left\">NAD+</td><td align=\"left\"><bold>Hexadecanoyl-9-ene_acid*</bold></td><td align=\"left\">UREAxt</td></tr><tr><td align=\"left\">Ferrocytochrome cM</td><td align=\"left\">LEUxt</td><td align=\"left\"><bold>Trans-2-C161-CoA*</bold></td><td align=\"left\">FADM</td><td align=\"left\"><bold>Octadecanoyl-9-ene_acid*</bold></td><td align=\"left\">L-Aspartate</td></tr><tr><td align=\"left\">FRUxt</td><td align=\"left\">FRUxt</td><td align=\"left\"><bold>Trans-2-C181-CoA*</bold></td><td align=\"left\">6-Phospho-D-gluconate</td><td align=\"left\">S-Adenosyl-L-homocysteine</td><td align=\"left\">N(pai)-Methyl-L-histidine</td></tr><tr><td align=\"left\">Allantoin</td><td align=\"left\">Succinate</td><td align=\"left\"><bold>Trans-2-C141-CoA*</bold></td><td align=\"left\">1,3-Diaminopropane</td><td align=\"left\">NADH</td><td align=\"left\">Adenosine 3',5'-bisphosphate</td></tr><tr><td align=\"left\">LEUxt</td><td align=\"left\">HISxt</td><td align=\"left\">Glyoxylate</td><td align=\"left\">sn-Glycerol 3-phosphate</td><td align=\"left\">UREAxt</td><td align=\"left\">4-imidazolecarboxylate</td></tr><tr><td align=\"left\">Succinate</td><td align=\"left\">TYRxt</td><td align=\"left\">sn-Glycerol 3-phosphate</td><td align=\"left\">O-Acetylcarnitine</td><td align=\"left\">L-Aspartate</td><td align=\"left\">3-Methyl-2-oxobutanoateM</td></tr><tr><td align=\"left\">HISxt</td><td align=\"left\">METxt</td><td align=\"left\">Isocitrate</td><td align=\"left\">Ethanol</td><td align=\"left\">1-Phosphatidyl-D-myo-inositol-3-P</td><td align=\"left\">Tetrahydrofolyl-[Glu](n)</td></tr><tr><td align=\"left\"><bold>Dodecanoyl-CoA*</bold></td><td align=\"left\">GLYxt</td><td align=\"left\">ERGOSTxt</td><td align=\"left\">DIPEPxt</td><td align=\"left\">N(pai)-Methyl-L-histidine</td><td align=\"left\">2-Phenylacetamide</td></tr><tr><td align=\"left\">PROxt</td><td align=\"left\">ASPxt</td><td align=\"left\">ZYMSTxt</td><td align=\"left\">Dipeptide</td><td align=\"left\">Adenosine 3',5'-bisphosphate</td><td align=\"left\">Phenylacetic acid</td></tr><tr><td align=\"left\">alpha-D-Mannose</td><td align=\"left\">GLCxt</td><td align=\"left\">1,3-Diaminopropane</td><td align=\"left\">OPEPxt</td><td align=\"left\">L-Asparagine</td><td align=\"left\">Indole-3-acetamide</td></tr><tr><td align=\"left\"><bold>Trans-3-C16-CoA*</bold></td><td align=\"left\">L-Tyrosine</td><td align=\"left\">6-Phospho-D-gluconate</td><td align=\"left\">Oligopeptide</td><td align=\"left\"><bold>C24-CoA*</bold></td><td align=\"left\">Indole-3-acetate</td></tr><tr><td align=\"left\"><bold>Trans-3-C18-CoA*</bold></td><td align=\"left\">PROxt</td><td align=\"left\">H2O2</td><td align=\"left\">PEPTxt</td><td align=\"left\">1-(5-Phospho-D-ribosyl)-5-amino</td><td align=\"left\">Urea-1-carboxylate</td></tr><tr><td align=\"left\"><bold>Trans-3-C14-CoA*</bold></td><td align=\"left\">alpha-D-Mannose</td><td align=\"left\"><bold>Trans-3-C16-CoA*</bold></td><td align=\"left\">Sphinganine</td><td align=\"left\">3-Methyl-2-oxobutanoateM</td><td align=\"left\">PhosphatidylserineM</td></tr></tbody></table></table-wrap>" ]

[ "<disp-formula id=\"bmcM1\"><label>(1)</label><bold><italic>S·v </italic></bold>= <bold>0</bold></disp-formula>", "<disp-formula><bold><italic>Z </italic></bold>= <bold><italic>ω·v</italic></bold></disp-formula>", "<disp-formula><bold><italic>S·v </italic></bold>= <bold>0</bold></disp-formula>", "<disp-formula><bold><italic>α </italic></bold>≤ <bold><italic>v </italic></bold>≤ <bold><italic>β</italic></bold></disp-formula>", "<disp-formula>Accuracy = (TP + TN)/(TP + TN + FP +FN)</disp-formula>", "<disp-formula>Sensitivity = TP/(TP + FN)</disp-formula>", "<disp-formula>Specificity = TN/(TN + FP)</disp-formula>", "<disp-formula>Positive predictive value = TN/(TP + FP)</disp-formula>", "<disp-formula>Negative predictive value = TN/(TN+FN)</disp-formula>", "<disp-formula>Geometric mean = (Sensitivity·Specificity)^1/2</disp-formula>" ]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional file 1</title><p><bold>Additional ORFs</bold>. List if additional ORFs and their references containing in <italic>iIN800</italic>.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S2\"><caption><title>Additional file 2</title><p><bold>Lipid metabolism reactions and comparison</bold>. Comparison of lipid metabolism reactions of all <italic>S. cerevisiae </italic>genome-scale models.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S3\"><caption><title>Additional file 3</title><p><bold>Growth simulation results</bold>. Growth simulations and comparison with experimental measurements.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S4\"><caption><title>Additional file 4</title><p><bold>Aerobic Flux distribution</bold>. Visualization of flux distribution of aerobic growth mapping on Figure ##FIG##0##1##.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S5\"><caption><title>Additional file 5</title><p><bold>Anaerobic Flux distribution</bold>. Visualization of flux distribution of anaerobic growth mapping on Figure ##FIG##0##1##.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S6\"><caption><title>Additional file 6</title><p><bold>High resolution file of Figure </bold>##FIG##0##1##. High resolution of <italic>S. cerevisiae </italic>metabolic map is provided as EPS format.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S7\"><caption><title>Additional file 7</title><p><bold><italic>iIN800 </italic>model</bold>. List of all participated reactions in <italic>iIN800 </italic>model.</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><p>* percentage of associated ORFs in the model relative to characterized ORFs in the yeast genome</p></table-wrap-foot>", "<table-wrap-foot><p>* = Aerobic growth, ** = Anaerobic growth</p><p>TP = true positive, TN = true negative, FP = false positive, FN = false negative.</p></table-wrap-foot>", "<table-wrap-foot><p>* Metabolite is contained in <italic>iIN800 </italic>only</p><p>2 = Tai, S. L., V. M. Boer, et al. (2005)</p><p>3 = Pizarro, F., M.C. Jewett, et al. (2008)</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1752-0509-2-71-1\"/>", "<graphic xlink:href=\"1752-0509-2-71-2\"/>", "<graphic xlink:href=\"1752-0509-2-71-3\"/>", "<graphic xlink:href=\"1752-0509-2-71-4\"/>", "<graphic xlink:href=\"1752-0509-2-71-5\"/>" ]

[ "<media xlink:href=\"1752-0509-2-71-S1.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1752-0509-2-71-S2.zip\" mimetype=\"application\" mime-subtype=\"x-zip-compressed\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1752-0509-2-71-S3.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1752-0509-2-71-S4.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1752-0509-2-71-S5.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1752-0509-2-71-S6.eps\" mimetype=\"application\" mime-subtype=\"postscript\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1752-0509-2-71-S7.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>DNA-based vaccines represent a new and rapidly progressing area in vaccinology. So far, plasmid DNA (pDNA) vaccines have been reported to induce protective immunity in numerous animal models of parasitic, viral and bacterial diseases [##REF##10050276##1##]. Moreover, pDNA vaccines appear to be well tolerated and exhibit a minimal risk of <italic>in vivo </italic>genome integration [##REF##16218776##2##, ####REF##10210143##3##, ##REF##11251381##4##, ##REF##14724672##5##, ##REF##16218775##6##, ##REF##16445866##7##, ##REF##16219399##8####16219399##8##]. In addition, persistent plasmid does not replicate inside the cells [##REF##16445866##7##] and there are no significant increases in anti-DNA antibodies leading to autoimmune reactions [##REF##10210142##9##]. Although preclinical studies on animal models document overall safety, some issues and potential risks related to food-producing animals need to be addressed directly on target species since these represent separate issues to clinical applications. Thus far, data on the rates of clearance, or conversely persistence, of pDNA post injection into animals is only limited, therefore potential risks must be extrapolated from model animal studies. Quantitative biodistribution studies have been performed in mice [##REF##10210143##3##, ####REF##11251381##4##, ##REF##14724672##5##, ##REF##16218775##6##, ##REF##16445866##7####16445866##7##,##REF##10210142##9##, ####REF##1301910##10##, ##REF##8546411##11##, ##REF##11251382##12##, ##REF##11438836##13##, ##REF##12877561##14##, ##REF##12885344##15##, ##REF##15208453##16##, ##REF##16125283##17####16125283##17##], rats [##REF##15603889##18##], rabbits [##REF##16218776##2##,##REF##16219399##8##,##REF##10210142##9##,##REF##11438836##13##,##REF##15005876##19##], sheep [##REF##11168628##20##], dog [##REF##12968706##21##] and macaques [##REF##15670887##22##], all post intramuscular (i.m.) administration of pDNA. Gratifyingly, all the studies have given evidence for overall safety as well.</p>", "<p>Quantitative real-time PCR (QRTPCR) is the most widely used method for specific quantitative assay of ultra low concentration of pDNA in biological materials. Such data are necessary for the assessment of the risk of residual plasmid presence in consumable parts of DNA vaccinated livestock, mainly in muscles. Nowadays, there are no definitive guidelines available to approve usage of DNA vaccines in food- producing animals. In this work, the QRTPCR method was used for the study of the persistence of pDNA at the injection sites in mice and beef cattle. For this reason we developed an isolation and detection QRTPCR based methodology for the accurate quantification of residual levels of vaccine pDNAX (pVAX-Hsp60 TM814) in the muscles after various approaches to vaccine application (naked pDNA, pDNA with electroporation, pDNA complexed with cationic liposomes). The primary motivation for this study was to obtain data for further negotiations with the State Veterinary Authority (Czech Republic) to get the approval for field trials with pDNAX against ringworm (<italic>Trichophyton mentagrophytes</italic>)[##REF##15504125##23##].</p>" ]

[ "<title>Materials and methods</title>", "<title>Plasmids</title>", "<p>The plasmid pDNAX (pVAX-Hsp60 TM814), encoding the heat shock protein 60 (Hsp60) from <italic>Trichophyton mentagrophytes </italic>[##REF##15601454##24##] and the plasmid pLacZ (pcDNA3.1/LacZ), expressing β-galactosidase, were used in this study. The plasmid DNA was produced in XL-1 Blue <italic>E. coli </italic>strain and purified with Qiagen Giga prep kit (Qiagen, Germany) to provide endotoxin free plasmid. Plasmid integrity was confirmed by electrophoresis on 0.8% agarose gel. The UV absorbance was used for quantification of DNA (A₂₆₀) and purity (A₂₆₀/₂₈₀) of plasmid preparation. The concentration of stock plasmid preparation was 2 mg/ml, the content of supercoil form was more than 90%, and the A₂₆₀/A₂₈₀was between 1.8–1.90.</p>", "<title>Preparation of liposomes and pDNAX-liposome complex</title>", "<p>Positively charged lipid <italic>N</italic>¹-cholesteryloxycarbonyl-3,7-diazanonan-1,9-diamine (CDAN) and neutral colipid dioleoyl L-α-phosphatidylethanolamine (DOPE) in 1:1 molar ratio were used for preparation of liposomes. Fluorescently labelled liposomes were prepared by addition of 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-lissamine rhodamine B (PE-rd)(1 mol % of total lipids). Lipids used in this study were purchased from Avanti Polar Lipids, Inc., USA. The lipid mixture was dissolved in freshly distilled chloroform and the solvent was evaporated under reduced pressure using rotary evaporator Laborota 4000 (Heidolph, Germany). Dry lipid film was hydrated in 4 mM HEPES buffer pH 7.2. Monodisperse liposomal preparation was obtained by extrusion through 100 nm Isopore filters (Millipore, Czech Republic). The size distribution and the zeta potential of resulting liposomes were measured using Zetasizer Nano ZS (Malvern, UK). Complexes of pDNAX with liposomes were prepared by incubation of the mixture of DNA with liposomes in 1:5 weight ratio at room temperature for 20 min [##REF##15491141##25##].</p>", "<title>pDNA application to mice</title>", "<p>The vaccination experiments were approved by the Ethical Committee of the Veterinary Research Institute, Brno, Czech Republic.</p>", "<title>Experiment I</title>", "<p>BALB/c mice (7–8 weeks of age) were divided into one control and three test groups. Various formulations of pDNAX (naked pDNAX, naked pDNAX followed by electroporation, liposomal complex pDNAX:CDAN/DOPE) were applied by i.m. injection route. On day 0, the tested animals received single injection into the right calf muscle. In each experimental group, pDNAX (10 μg comprising approximately 10¹²-10¹³copies) in a total volume of 50 μl was applied. An electroporator (developed in the laboratory of Prof. Yuhong Xu at Shanghai Jiao Tong University, Shanghai) was used in these experiments. Six electric pulses (duration 20 ms, field strength 150 V/cm, the interval between the pulses 1 s, the gap distance between electrodes 3 mm) were applied by two parallel needle electrodes (distance of the needles was 3 mm) immediately after i.m. injection. Injection point was in the middle between the electrodes. 50 μl of PBS were applied to mice of the control group. The animals were kept under standard conditions during the whole experimental period. Neither lost of weight nor pathological changes in the skin, somatomotoric activity or behaviour pattern were observed. At the end of each experimental period i.e.: 1, 7, 28, 90, 180 and 365 days after administration, 4 animals from each test group and 2 animals from the control group were sacrificed. Both quadriceps muscles from each mouse were collected for the evaluation of the persistence of pDNAX. The samples of muscles were homogenised, weighted, frozen in liquid nitrogen and stored at -70°C until further processing.</p>", "<title>Experiment II</title>", "<p>The influence of the age of the mice on the dynamics of plasmid clearance during 1 month period after administration was tested on BALB/c mice 5 weeks of age. Experimental design was the same as in Experiment I.</p>", "<title>Experiment III – fluorescent liposomes and analysis of gene expression</title>", "<p>Single dose of pLacZ (10 μg) was injected into calf muscle of BALB/c mice (5 weeks of age). Plasmid pLacZ was delivered in the following forms: naked DNA, naked DNA followed by electroporation, and pDNA complexed with fluorescent cationic liposomes (CDAN/DOPE/PE-rd). The samples of muscles were taken at the day 1, 7, 14 and 28 after the administration. Tissue sections of the thickness 7 μm were prepared by cryocat Leica CM1900 (Leica, Germany) and stained for β-galactosidase expression using the substrate X-gal (Sigma, Czech Republic). The distribution and persistence of fluorescently labelled pDNA:(CDAN/DOPE/PE-rd) complexes were evaluated using fluorescence microscope Eclipse TM200 with CCD camera (Nikon, Japan) and the images were recorded using Lucia software (Laboratory Imaging Ltd., Czech Republic).</p>", "<title>pDNA application to beef cattle</title>", "<p>The vaccination experiment was approved by the Ethical Committee of the Veterinary Research Institute, Brno and University of Palacky, Medicinal Faculty, Olomouc. Ten beef cattle bulls (3 months of age) were divided into three experimental groups. In each experimental group, pDNAX (500 μg per dose; this dose was found to be sufficient for induction of the immune response in calves [##REF##15504125##23##]) in various formulations (naked pDNAX, pDNAX in combination with liposomal adjuvant B30-norAbu-MDP (lipophilic derivative of muramyl dipeptide entrapped into liposomes; this compound was synthetised at the Institute of Organic Chemistry and Biochemistry, Prague), complex pDNAX:CDAN/DOPE) was administered by i.m. single needle injection into right coccygeus muscle. The animals were re-vaccinated after three weeks by the same dose, formulation, and procedure. The bulls were slaughtered 242–292 days after the second vaccination and whole right coccygeus muscle (injection site), whole left coccygeus muscle (opposite-to-injection site), random tissue samples from gluteus muscle (distant muscle tissue), and poplitheal lymph nodes were collected. The samples of muscles were cut into small pieces, homogenised by blender and stored at -70°C before further processing. Various numbers of samples from particular tissues were prepared and taken for analyses: injection site (n = 5), opposite-to-injection site (n = 4), distant muscle tissue (n = 3), each draining lymph node (n = 2).</p>", "<title>DNA extraction from tissue sample</title>", "<p>The isolation of genomic DNA (gDNA) from the samples of tissue taken from mice or beef cattle was performed by modification of guanidine thiocyanate (GuSCN) lysis method followed by binding of DNA to SiO₂[##REF##15954711##26##]. The average weights of mice muscle samples and the samples from beef cattle muscles were 100–150 mg and 200 mg, respectively. The samples were mixed in 2-ml tubes with 1 ml of lysis buffer (5 M GuSCN; 0.05 M Tris-HCL, pH 6.4; 0.02 M EDTA, pH 8.0; 1.3% Triton X-100) and about 10 pcs. of 2.5 mm glass beads. The mixture was homogenised twice in Magnalyser (Roche, Germany) for 30 s at 6000 rpm. Then the suspension was centrifuged (14000 g, 10 min.); 1 ml of the supernatant from mice tissue samples or 700 μl of the supernatant from beef cattle tissue samples was transferred in 1.5 ml tube, filled with lysis buffer to the total volume of 1.2 ml, and then 50 μl of silica suspension (freshly prepared on the preceding day by mixing 100 mg of Celite with 500 μl of water and 5 μl of 32% HCl) was added. The tubes were vortexed for 30 s. The mixture was incubated at room temperature for 10 min., centrifuged (14000 g, 1 min.), and the supernatant was discarded. The silica pellet was washed twice with 1 ml of washing buffer (5 M GuSCN; 0.05 M Tris-HCL, pH 6.4; 0.02 M EDTA, pH 8.0), twice with 1 ml of 70% ethanol, and once with 1 ml of acetone. Subsequently, silica pellet was dried in heated block at 56°C for 15 min, followed by extraction step performed twice: mixing with 80 μl of tempered (56°C) TE-buffer (10 mM Tris-HCl, 1 mM EDTA pH 8.0), incubation in heated block for 10 min., and centrifugation (14000 g, 1 min.). 80 μl of the recovered supernatant was transferred into clean tube, centrifuged again (14000 g, 1 min.), and used for QRTPCR analysis. 20-μl volumes were taken from each extracted DNA sample to measure DNA concentration (A₂₆₀), purity (A₂₆₀/A₂₈₀), and integrity (0.6% agarose gel electrophoresis).</p>", "<title>QRTPCR analysis</title>", "<p>The Genecompare software (Applied-Maths, Belgium) was used to design primers amplifying a sequence stretch that contains plasmid specific promoter sequence (CMV) as well as sequence from <italic>hsp60 </italic>gene, generating 161 bp specific product. 500 ng of genomic DNA (gDNA) template was amplified in duplicate in glass capillaries in a final volume of 20 μl using 2× Real time PCR Syber green master mix (Qiagen, Germany) with 0.5 μM primers: CMV-Hsp60-F: 5'-ACTATAGGGAGACCCAAGCT-3' CMV-Hsp60 R: 5'-GCCTGTAGGTACTCGACAAC-3' Optimal PCR cycling conditions were: 15 min. pre-incubation at 95°C, 45 amplification cycles consisting of denaturation at 95°C for 10 s, annealing at 61°C for 25 s, extension at 72°C for 10 s and data acquisition at 78°C for 1 s using a temperature transition rate of 20°C/s in the LightCycler 1.5 instrument (Roche, Germany). Second derivative maximum method was used for Ct calculation from amplification curves. The amount of pDNAX in the tested samples was calculated by the comparison of the sample's Ct value with Ct values of the titration curve of genomic samples artificially spiked with pDNAX. The results for each mouse group were recalculated and are expressed as mean plasmid copy number per μg of gDNA (PCN/μg gDNA).</p>", "<title>Precautions to prevent contamination</title>", "<p>All the manipulations with stock plasmid, tissue sampling, QRTPCR set up and template addition were done in separated working areas [##REF##10918521##27##]. To prevent cross-contamination, the non-treated animals were handled before the vaccinated animals. Samples from the vaccinated animals were processed in the following manner: distant muscle tissues (beef cattle), muscle tissue from opposite-to-injection site (mice: left calf muscle, beef cattle: left coccygeus muscle), injection site (mice: right calf muscle, beef cattle: right coccygeus muscle). Disposable materials were used whenever possible. The work surfaces and equipment were decontaminated by either 10% bleach or DNAoff (Fluka, Germany).</p>" ]

[ "<title>Results</title>", "<title>Validation of QRTPCR method</title>", "<p>Persistence of pDNAX was determined by a QRTPCR methodology designed to specifically recognize the stretch of promoter-insert from the pDNAX plasmid. The methodology was initially investigated for sensitivity, specificity and linearity, in the detection of pDNAX plasmid. Firstly, the detection method was studied as part of the protocol for isolation of genomic DNA (gDNA) from mouse and beef muscle tissue. This protocol for isolation was found to be scalable up to 200 mg of muscle tissue, and in repeated applications of the QRTPCR methodology no inhibition due to sample matrix or presence of inhibitors was observed. Thereafter, pDNAX was introduced to gDNA allowing the detection limit (DL) and linear quantification range (LQL) of the QRTPCR methodology to be determined. In this instance, the LQL was found to be within the range of 40-4 × 10⁹ag (10-1 × 10⁹PCN/500 ng gDNA and the DL was shown to be 10 ag (3 PCN/500 ng gDNA) (Fig. ##FIG##0##1##). Finally, mouse and beef muscle tissue samples were spiked with quantities of pDNAX in the range from 10-4 × 10⁹ag. Thereafter, complete pDNAX isolation procedures were performed demonstrating that pDNA recovery was in the range of 65–95%. The detection limit of pDNAX isolation from tissue samples was found to be 800 ag (100 PCN/500 ng gDNA). This parameter represents the lowest amount of pDNAX that could be detected in all replicates of spiked samples by QRTPCR.</p>", "<title>Biodistribution and persistence in mice</title>", "<title>Experiment I</title>", "<p>The pDNAX plasmid (10 μg) was injected i.m. to 8-week-old mice and then detectable levels of plasmid were assayed as a function of time by QRTPCR. As shown (Fig. ##FIG##1##2##), pDNAX introduced i.m. to 8-week-old mice persisted at detectable levels in the region of the injection site for up to one year after administration regardless of the plasmid formulation and method of application. However, rates of clearance of pDNAX varied with the mode of administration. One day post injection, pDNAX remaining in muscle samples from three different groups was in the following order: pDNAX:CDAN/DOPE: 374 ng/μg gDNA (4.60 × 10⁷PCN/500 ng gDNA) &gt; pDNAX electroporation: 2600 pg/μg gDNA (3.20 × 10⁵PCN/500 ng gDNA) &gt; naked pDNAX: 689 pg/μg gDNA (1.70 × 10⁵PCN/500 ng gDNA). In the first group, pDNAX was injected in complex with CDAN/DOPE cationic liposomes; in the second group, pDNAX was injected with electroporation; in the third group naked pDNAX was injected alone. Thereafter, in the case of the pDNAX:CDAN/DOPE group levels of pDNAX were found to undergo a 10-fold decline between the day 7 and the day 28, followed by a further 100-fold decline by the day 90, so that by the day 365 a detectable level of only 535 ag/μg gDNA (1.35 × 10²PCN/500 ng gDNA) was determined by QRTPCR (Fig. ##FIG##1##2##). By contrast, in the case of both pDNAX electroporation and naked pDNAX groups, clearance rates were more considerable. In the case of the naked DNAX group, final plasmid levels were found to be below the quantification limit of 40 ag/μg gDNA (10 PCN/500 ng gDNA (Fig. ##FIG##1##2##).</p>", "<title>Experiment II</title>", "<p>Identical experiment was performed with 5-week-old mice to evaluate a possible relationship between the animal age and the rate of clearance of pDNAX from the site of injection. In both cases, naked pDNAX and pDNAX electroporation groups, the rates of clearance of pDNAX were found to be slower for 5-week-old mice in comparison to the corresponding situation in 8-week-old mice (compare Fig. ##FIG##1##2## and Fig. ##FIG##2##3##). Nevertheless, the final differences in pDNAX levels between pDNAX:CDAN/DOPE and the pDNAX electroporation groups were still in the range of 100-fold, with an even greater gap of over 10⁴-fold between pDNAX:CDAN/DOPE and naked pDNAX groups. In this instance too, a difference of 1–2 orders of magnitude also existed between the measured plasmid levels in the pDNAX electroporation group and the naked pDNAX group at all time points analyzed (Fig. ##FIG##2##3##), in partial contrast to our observations with 8-week animals (Fig. ##FIG##1##2##).</p>", "<title>Experiment III- analysis of gene expression and distribution of fluorescent complex of pDNA/cationic liposomes</title>", "<p>Flourescently labelled pDNAX:CDAN/DOPE complexes were prepared and injected i.m. into 5-week old mice in order to make comparison with the QRTPCR data (Fig. ##FIG##2##3##). Post administration, complexes were clearly visible, localised at the site of application, and persisted for more than four weeks as shown in histological sections by fluorescent microscopy (Fig. ##FIG##3##4##). This is in a good correlation with the persistence of pDNAX as determined by QRTPCR (Fig. ##FIG##2##3##). Similar data were found in the group of 8-week-old mice (data not shown). Transfection experiments were then performed by the administration of naked pLacZ injected <italic>i.m</italic>. into 5-week and 8-week old mice. Histological analyses of muscle tissue sections revealed that β-galactosidase expression was undetectable after the injection to 8-week old mice with naked pLacZ (10 μg) (data not shown). However, when pLacZ (10 μg) was introduced together with electroporation pulse, transfection was detectable, but only a few myocytes were found to be positive for β-galactosidase expression. In contrast, β-galactosidase expression was much more evident with 5-week old mice. Myocyte bundles expressing β-galactosidase were clearly localised around the site of injection and there was little tissue damage associated with electroporation. Several β-galactosidase positive myocytes were found also four weeks after electroporation. Micrographs of the tissue sections documenting β-galactosidase expression are presented (Fig. ##FIG##3##4##).</p>", "<title>Biodistribution and persistence in beef cattle</title>", "<p>Residual pDNAX levels in various samples of tissues taken from beef cattle slaughtered 9 months after application of plasmid are summarized (Table ##TAB##0##1##). QRTPCR examinations of muscle tissue taken from the injection site revealed very low residual or nearly zero pDNAX levels in all animals tested. Plasmid levels detected in animals injected with naked pDNAX group were predominantly below quantification 40 ag/μg gDNA (10 PCN/500 ng gDNA) or detection 13 ag/μg gDNA (3 PCN/500 ng gDNA) limit. Slightly higher residual plasmid levels, but mostly close to quantification limit, were also detected in the cases where pDNAX was injected with a liposomal formulation of adjuvant B30-norAbu-MDP. The highest levels of retention (288 PCN/500 ng gDNA) were recorded at the injection site in the muscle samples from beef cattle injected with pDNAX:CDAN/DOPE. However, plasmid levels from all slaughtered animals showed progressive decreases in pDNAX levels below the quantification limit after longer time periods. Gratifyingly, essentially no plasmid was found at either distant muscle tissue or in draining lymph node samples. Muscle samples from opposite-to-injection site (internal negative control) were also negative for the presence of pDNAX.</p>", "<title>General safety</title>", "<p>After the injection of pDNAX (or pLacZ), both mice and cows from all the tested groups survived throughout the duration of the experiments and neither any apparent pathological changes at the site of injection nor loss of body weight were observed indicating that pDNAX (pVAX-Hsp60 TM814) vaccine and its formulations as a complex with cationic liposomes or liposomal adjuvant B30-norAbu-MDP were well tolerated in both species. Application of electroporation with or without previous local or general anesthesia did not lead to any changes of somatomotoric activity or even paraplegia in mice.</p>" ]

[ "<title>Discussion</title>", "<p>Limited data on the examination of the effect of pDNA vaccines on food-producing animals have been reported so far and we can only extrapolate the results obtained in the model animals. Different regulation acts on genetically modified organisms and their interpretation by national authorities represent serious obstacles for the field of DNA vaccination experiments on large animals. DNA vaccines have not yet been licensed in many countries, therefore national authorities are not experienced with this kind of product and do not differentiate between gene medication and gene modification. Within the EU, two opposite points of view are maintained as regards DNA vaccinated animals. The first one, held by The British Agriculture and Environment Biotechnology Committee, does not consider DNA vaccinated animals as genetically modified ones due to the low risk of insertion of pDNA into genome. The second one, held by The Norwegian Directorate for Nature Management, states that DNA vaccinated animals should be considered as genetically modified for as long as the added DNA is present. In other words, gene medication is the subset of gene modification [##REF##14595352##28##]. The safety concerns raised by the use of plasmid DNA for immunization of food producing animals, livestock and poultry are obviously distinct from those in humans. The addition of foreign products e.g. pDNA into the food chain should be carefully considered to ensure that neither livestock animals nor consumers develop unpredicted or undesirable side-effects. While the safety of DNA vaccines was documented in animal and human trials, the problem of residual plasmid in consumable parts of livestock and poultry has not yet been solved on the level of the State Veterinary Authority and regulatory veterinarians. In contrast to experiments performed on small rodents, vaccination field trials on large animals, e.g. cows, are more expensive and are subjected to more strict regulations.</p>", "<p>The condemnation of whole animals and the processing of their cadavers in rendering plants pose not only an economic problem but also an ethic one. The presented study has shown that pVAX-Hsp60 TM814 vaccine and its formulations as a complex with cationic liposomes or liposomal adjuvant B30-norAbu-MDP were well tolerated by both species. From the practical point of view, the regulatory authorities will demand a reliable, sensitive and cost effective method for the determination of the amount of residual plasmid and its localization in the body at the time of the slaughter. The detection method based on QRTPCR was proved to be suitable for the exact quantification of residual plasmid levels in muscle tissues after i.m. application of pDNA vaccine. By the use of the artificially spiked muscle tissue samples we documented, that pDNA was efficiently recovered (65–95% of the initial amount) within the wide range of plasmid concentrations that might occur in real tested samples (Fig. ##FIG##0##1##). The quality of the isolated DNA was sufficient for the development of QRTPCR assay providing parameters ensuring high specificity, sensitivity and reproducibility for the precise pDNA quantification. The sensitivity of our assay was comparable to that published by Tuomela for the pDNA GTU^®-MultiHIV [##REF##15603889##18##].</p>", "<title>Biodistribution and persistence of pDNA in mice</title>", "<p>Model studies on rodents covering overall biodistribution and safety features are required before DNA vaccines enter human clinical trials [##UREF##0##29##]. We used mouse model to provide information about plasmid clearance kinetics, which is useful for further extrapolation for beef cattle. Biodistribution studies, primarily those performed with naked pDNA applied i.m., show that pDNA is completely cleared from the injection site within 28 days or even sooner. However, long-term persistence was reported as well – by qualitative PCR: 18 wks [##REF##8546411##11##], 180 days [##REF##16445866##7##], 19 months [##REF##1301910##10##], and 2 years [##REF##15208453##16##] after application.</p>", "<p>Our results confirm the previous observations that plasmid DNA is rapidly cleared from the injection site [##REF##12885344##15##,##REF##16125283##17##,##UREF##1##30##]. Depending on the type of application, the amount of pDNA found in mice after 24 hours in electroporated and naked group was less than 0.1% and less than 0.01%, respectively. Naked pDNA is immediately subjected to degradation, therefore only limited fraction of the applied plasmid is capable to reach the zone where pDNA is protected (i.e. structures like T-tubules and caveolae [##REF##1487500##31##]), against the attack of serum and tissue specific nucleases [##REF##10543612##32##].</p>", "<p>Application of electroporation pulse leads to transient membrane disruption facilitating pDNA uptake. Generally, electroporation improves pDNA uptake and leads to several orders higher expression levels, as reviewed in [##REF##15161333##33##]. However, for further optimization of electroporation parameters for clinical application it is necessary to reduce a pain and potential muscle damage caused by this technique [##REF##11708877##34##, ####REF##16054757##35##, ##REF##15241788##36####15241788##36##]. The study published by Wang et al. [##REF##15880605##37##] determined, that critical parameters influencing electroporation are plasmid concentration, injection volume, concentration of saline media, size of plasmid DNA, repeated gene transfer. However, neither the influence of lag time between plasmid injection and electroporation nor the effect of the age of mice was observed. On the contrary, we detected the age-dependent differences (5-week old mice vs. 8-week old mice, Fig. ##FIG##1##2## vs. Fig. ##FIG##2##3##) of residual plasmid in muscles of mice vaccinated by naked pDNA or electroporated. This could be explained by the age-dependent changes of extracellular matrix structure, which might affect the permeation of pDNA and hence the efficiency of electroporation resulting in the decreased transfection efficacy in the older mice [##REF##9285779##38##]. This consideration is also confirmed by our data obtained with 5-week old mice, where the differences between the naked DNA and the electroporated group were more clearly pronounced (Fig. ##FIG##2##3##) and a slower clearance rate within the first 28 days was observed (compare Fig. ##FIG##1##2## and Fig. ##FIG##2##3##). Such important effect of extracellular matrix on local pDNA delivery was documented using the enzyme hyaluronidase that breaks down the components of extracellular matrix [##REF##11509960##39##, ####REF##12621454##40##, ##REF##16309967##41####16309967##41##]. Rapid plasmid decline in naked and electroporated group within the first 28 days (Fig. ##FIG##1##2##) could be also related to transfection of other cells than myocytes, e.g. endothelial cells, in which plasmid DNA is unstable and could be lost during mitosis. Relative stability of low plasmid level in muscle was observed within the period of the day 28 and 1 year after administration. pDNA is supposed to be located in the nucleus of myocytes, which can retain pDNA for a long time. Gradual decline of pDNA concentration could be explained by normal myonuclei turn-over in myocytes [##REF##10716774##42##]. For the exact evaluation, whether the plasmid is integrated into genomic DNA or presented in extrachromosomal state, a precise gel purification method would be necessary [##REF##11251381##4##,##REF##14724672##5##,##REF##11251382##12##,##REF##11438836##13##]. Furthermore, plasmid integration into genomic DNA is a very rare event, usually lower than the level of spontaneous mutation [##REF##11251381##4##]. Wang et al. [##REF##14724672##5##] reported that less than 0.2% of the intracellulary presented pDNA was integrated into genomic DNA after application of naked and electroporated plasmid, respectively. According to such calculations, plasmid integration into genomic DNA in our experiments would be mostly at the level below quantification limit or even undetectable.</p>", "<p>Cationic liposomes are mostly used as carries for intravenous systemic delivery, but novel lipid combinations might be suitable for i.m. delivery [##REF##16218776##2##,##REF##11228361##43##,##REF##11395213##44##] and they have been found to be well tolerated in both, animals and humans [##REF##12477411##45##]. When we compared the cationic liposomes with the standard method of i.m. delivery, i.e. the injection of naked pDNA without or with electroporation, plasmid levels retained in mouse muscles after 24 hours from pDNA:CDAN-DOPE group were even 100–1000× higher (between 7–11% of the initial amount). Generally, our data demonstrated a slower clearance of pDNA from the injection site of pDNA:CDAN-DOPE group within the period of day 1 and day 28 in comparison to both, the naked and electroporated groups (Fig. ##FIG##1##2## and Fig. ##FIG##2##3##). This data would support the consideration that pDNA in liposomal complex is more protected against the attack of nuclease. With regards to the observation of Hartikka et al[##REF##11228361##43##], who noticed that another cationic lipid formulation – Vaxfectin did not appear to increase transfection, we can suppose that high plasmid levels are located extracellulary. Using fluorescently labelled liposomes, histological analysis revealed that liposomal complexes were mostly distributed along the injection lane, forming a depot within muscle tissues even after 28 days (Fig. ##FIG##3##4##).</p>", "<title>Biodistribution and persistence of pDNA in beef cattle</title>", "<p>In order to facilitate further plasmid detection and potentially minimise a condemnation of whole consumable parts, coccygeus muscle was chosen as a suitable site for immunization. It is important to note that this small muscle, located closely to the root of the tail, is easy to reach and remove after slaughter. Having 10 animals available in experimental herd, we tested i.m. administration of pDNA vaccine and its various formulations intended for field vaccination trials. Unfortunately, we had not suitable electrodes for the electroporation of larger animals at the time of the experiments. Instead of electroporation we applied pDNA vaccine in combination with liposomal adjuvant B30-norAbu-MDP, which was proved to be effective in guinea pigs immunized by the same pDNA vaccine (unpublished results). Altogether, the performed QRTPCR assay revealed that pDNA persisted in ultra-low level at the injection site even 292 days after the second administration of pDNA. The highest amount of pDNA was detected in the group vaccinated by pDNA:cationic liposome complexes. These data are in good accordance with the results obtained in mice. The values of residual pDNA in the group injected by naked pDNA were mainly non-quantifiable. Combination of naked pDNA with the liposomal adjuvant B30-norAbu-MDP resulted in levels of residual pDNA close to quantification limit. It is important to emphasize that no plasmid was detected in distant muscle tissue, in draining lymph node or in the opposite muscle directly connected with these lymph nodes. The tissues located contralaterally to the injection sites could also be considered as negative controls for each vaccinated animal.</p>" ]

[ "<title>Conclusion</title>", "<p>Quantitative real-time PCR (QRTPCR) assay was developed to assess a residual pDNA vaccine pVAX-Hsp60 TM814 in mice and beef cattle. In beef cattle, ultra low residual level of pDNA vaccine was found only at the site of injection. According to rough estimation, consumption of muscles from the injection site represents almost an undetectable income of pDNA (400 fg/g muscle tissue) for the consumers. Residual plasmid in native state will hardly be found at measurable level following further meat. This study brings supportive data for animal and food safety and hence for further approval of pDNA vaccine field trials.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Application of plasmid DNA for immunization of food-producing animals established new standards of food safety. The addition of foreign products e.g. pDNA into the food chain should be carefully examined to ensure that neither livestock animals nor consumers develop unpredicted or undesirable side-effects.</p>", "<title>Methods</title>", "<p>A quantitative real-time PCR (QRTPCR) methodology was developed to study the biodistribution and persistence of plasmid DNA vaccine pDNAX (pVAX-Hsp60 TM814) in mice and beef cattle. The linear quantification range and the sensitivity of the method was found to be 10 – 10⁹copies per reaction (500 ng/gDNA) and 3 copies per reaction, respectively.</p>", "<title>Results</title>", "<p>Persistence of pDNAX in mice muscle tissue was restricted to injection site and the amount of pDNAX showed delivery formulation dependent (naked pDNA, electroporation, cationic liposome complexes) and mouse age-dependent clearance form injection site but pDNAX was still detectable even after 365 days. The QRTPCR analysis of various muscle tissue samples of vaccinated beef bulls performed 242–292 days after the last revaccination proved that residual pDNAX was found only in the injection site. The highest plasmid levels (up to 290 copies per reaction) were detected in the pDNAX:CDAN/DOPE group similarly to mice model. No pDNA was detected in the samples from distant muscles and draining lymph nodes.</p>", "<title>Conclusion</title>", "<p>Quantitative real-time PCR (QRTPCR) assay was developed to assess the residual pDNA vaccine pVAX-Hsp60 TM814 in mice and beef cattle. In beef cattle, ultra low residual level of pDNA vaccine was only found at the injection site. According to rough estimation, consumption of muscles from the injection site represents almost an undetectable intake of pDNA (400 fg/g muscle tissue) for consumers. Residual plasmid in native state will hardly be found at measurable level following further meat processing. This study brings supportive data for animal and food safety and hence for further approval of pDNA vaccine field trials.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>PO carried out development of QRTPCR, participated in quantification of pDNA, and participated in preparation of the manuscript. VK participated in preparation of cationic liposomes, carried out the histology experiments and electroporation. MR designed and prepared the plasmid for vaccination and participated in preparation of the manuscript. ADM designed and synthesised cationic lipids. ML designed and synthesised muramylglycopeptide adjuvans. JT conceived of the study, participated in its design and coordination, prepared and characterised liposomes, performed immunisation experiments and drafted the manuscript. All authors read and approved the final manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>This work was supported by grant NAZV QF 3115, the Ministry of Agriculture of the Czech Republic (grant No. MZE 0002716201) and MSM6198959223. We also thank IC-Vec Ltd, UK for support. Special thanks to Hana Kudláèková for assistance with animal handling and sampling, and to Jana Plocková for manuscript preparation.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Linearity analysis after QRTPCR amplification</bold>. Dilution series of pDNAX (10⁹– 3 × 10°copies) was amplified with 500 ng of mouse gDNA. Full squares represent Cp values (crossing point) recorded from three independent pDNAX dilutions. The strait line represents linear regression analysis with correlation coefficient (R²) greater than 0,99.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Levels of pDNAX detected by QRTPCR in calf muscle (at the injection site) after administration of 10 μg pDNAX in 8-week old BalB/C mice</bold>. The line connects the average levels of plasmid DNA detected by QRT-PCR in 500 ng of isolated DNA (MC/r) ± SD (four mice per time point). The straight line represents quantification limit of QRTPCR assay (10 pDNAX copies/reaction). The dotted line represent detection limit of QRTPCR assay (3 pDNAX copies/reaction). The data from control group were omitted (all control animals were negative). Routes of application: full circle denotes naked pDNAX; full triangle denotes pDNAX plus electroporation; full square denotes pDNAX:CDAN-DOPE complex.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Levels of pDNAX detected by QRTPCR in calf muscle (at the injection site) after administration of 10 μg pDNAX in 5-week old BalB/C mice</bold>. The line connects the average levels of plasmid DNA expressed in logarithm scale detected by QRTPCR in 500 ng of isolated DNA (MC/r) ± SD (four mice per time point). The straight line represent quantification limit of QRTPCR assay (10 pDNAX copies/reaction). The dotted line represent detection limit of QRTPCR assay (3 pDNAX copies/reaction). The data from control group were omitted (all control animals were negative). Routes of application: full circle denotes naked pDNAX; full triangle denotes pDNAX plus electroporation; full square denotes pDNAX:CDAN-DOPE complex.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p><bold>Expression of β-galactosidase activity and persistence of fluorescent liposome- pLacZ complexes in mice calf muscles</bold>. Mice calf muscles were histochemically stained for β-galactosidase activity at the day 1 (A) and at the day 28 (B) after i.m. injection of 10 μg pLacZ followed by electroporation. Histological detection of fluorescent liposome-pDNA complex (10 μg pLacZ/CDAN:PE-rh) in mice calf muscles at the day 1 (C) and 28 (D) after administration into young mice (the age of 5 weeks).</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Effect of various pDNAX formulations on its persistence in beef cattle after i.m. administration</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Beef cattle groups</td><td align=\"left\">Beef cattle ID code</td><td align=\"left\">Interval between 2^ndimmunisation and slaughter (days)</td><td align=\"left\">pDNA copies at the injection site/500 μg DNA (n = 5)</td><td align=\"left\">pDNA copies opposite -to- injection site muscle (n = 4)</td><td align=\"left\">pDNA copies distant muscle (n = 3)</td><td align=\"left\">pDNA copies DLN^atotal (n = 6)</td></tr></thead><tbody><tr><td align=\"left\">pDNA</td><td align=\"left\">20087</td><td align=\"left\">242</td><td align=\"left\">&lt; LQL (2); &lt; DL (2); Neg. (1)</td><td align=\"left\">0/4</td><td align=\"left\">0/3</td><td align=\"left\">0/6</td></tr><tr><td/><td align=\"left\">20105</td><td align=\"left\">277</td><td align=\"left\">73.97; 35.59; &lt; LQL (2); &lt; DL (1)</td><td align=\"left\">0/4</td><td align=\"left\">0/3</td><td align=\"left\">0/6</td></tr><tr><td/><td align=\"left\">20080</td><td align=\"left\">284</td><td align=\"left\">&lt; LQL (5)</td><td align=\"left\">0/4</td><td align=\"left\">0/3</td><td align=\"left\">0/6</td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"left\">DNA + B30-Nor-AbuMDP</td><td align=\"left\">20083</td><td align=\"left\">242</td><td align=\"left\">19.86; 16.23; 15.5; &lt; LQL (2)</td><td align=\"left\">0/4</td><td align=\"left\">0/3</td><td align=\"left\">0/6</td></tr><tr><td/><td align=\"left\">28504</td><td align=\"left\">270</td><td align=\"left\">92.78; 29.25; 28;92; 24.04; 23.75</td><td align=\"left\">0/4</td><td align=\"left\">0/3</td><td align=\"left\">0/6</td></tr><tr><td/><td align=\"left\">20090</td><td align=\"left\">291</td><td align=\"left\">13.48; 12.73; 10.87; &lt; LQL (1); Neg. (1)</td><td align=\"left\">&lt; DL(1/4)</td><td align=\"left\">0/3</td><td align=\"left\">0/6</td></tr><tr><td colspan=\"7\"><hr/></td></tr><tr><td align=\"left\">DNA:cationic liposome complex</td><td align=\"left\">20086</td><td align=\"left\">270</td><td align=\"left\">288; 220.08; 200.60; 30.07; &lt; LQL (1)</td><td align=\"left\">0/4</td><td align=\"left\">0/3</td><td align=\"left\">0/6</td></tr><tr><td/><td align=\"left\">20089</td><td align=\"left\">277</td><td align=\"left\">228.9; 169.90, 134.70; 39.57; 39.00</td><td align=\"left\">0/4</td><td align=\"left\">0/3</td><td align=\"left\">0/6</td></tr><tr><td/><td align=\"left\">3654</td><td align=\"left\">291</td><td align=\"left\">149.3; 64.79; 46.60; 18.88; 18.73</td><td align=\"left\">0/4</td><td align=\"left\">0/3</td><td align=\"left\">0/6</td></tr><tr><td/><td align=\"left\">20081</td><td align=\"left\">298</td><td align=\"left\">&lt; LQL (5)</td><td align=\"left\">0/4</td><td align=\"left\">0/3</td><td align=\"left\">0/6</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>The total amount 1000 μg of pDNAX in two equal doses was delivered into coccygeus muscle (injection site) as naked pDNAX, naked pDNAX + liposomal B30-norAbuMDP, and cationic liposome complex pDNAX:CDAN-DOPE. The level of pDNAX in the injection site, opposite-to-injection site, distant muscle tissues and draining lymph nodes was examined after 242–298 days after the second immunisation. Plasmid copies are expressed as mean plasmid copies per 500 ng of genomic DNA (MC/r) from duplicate QRTPCR assay. &lt; LQL: below linear limit of quantification (10 copies/reaction), &lt; DL: below detection limit (3 copies/reaction), Neg.: negative sample, ^aDLN- draining lymph nodes.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1479-0556-6-11-1\"/>", "<graphic xlink:href=\"1479-0556-6-11-2\"/>", "<graphic xlink:href=\"1479-0556-6-11-3\"/>", "<graphic xlink:href=\"1479-0556-6-11-4\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Introduction</title>", "<p>Chronic obstructive pulmonary disease (COPD) is characterized by inflammation in both central and peripheral airways [##REF##12882441##1##], which is dominated by neutrophils, macrophages, T lymphocytes (mainly CD8⁺cells), and B lymphocytes [##REF##15215480##2##, ####REF##11371392##3##, ##REF##16387936##4####16387936##4##]. In addition, some [##REF##9403729##5##,##REF##8173772##6##] but not all studies [##REF##10430750##7##,##REF##9769292##8##], have demonstrated higher numbers of mast cells in patients with COPD than in controls without airflow limitation. Theoretically, mast cells could play a role in the pathogenesis of COPD [##REF##18055703##9##] by inducing collagen production [##REF##8422426##10##] fibroblast proliferation [##REF##11929488##11##,##REF##10887338##12##], and release of various mediators, including the potent proteases tryptase and chymase. Tryptase, a mast cell specific protease, is present in all human mast cells, whereas chymase is present in a subset of mast cells.</p>", "<p>Balzar and co-workers recently evaluated the number of chymase and tryptase positive mast cells in central and peripheral airways of patients with severe asthma [##REF##15563633##13##]. They demonstrated that higher numbers of chymase positive mast cells in peripheral airways are associated with less severe airflow limitation. Therefore, they suggested that mast cells are protective for enhanced airway obstruction in patients with asthma.</p>", "<p>To our knowledge, the distribution of chymase and tryptase positive mast cells in central and peripheral airways, and their relation with lung function has not been investigated previously in patients with COPD. This study did so and compared this distribution with results in controls without airflow limitation.</p>" ]

[ "<title>Methods</title>", "<title>Subjects and tissue collection</title>", "<p>Tissue samples were collected from 29 individuals: 19 with COPD and 10 controls without airflow limitation (Table ##TAB##0##1##). COPD patients had no chronic bronchitis or α₁-antitrypsin deficiency (10 males, median [interquartile range] age 62 [54–72], pack-years 31 [27–40]) and they were former- or current smokers (14 and 5 patients, respectively). They had mild to very severe COPD according to the Global Initiative on Lung Disease (GOLD) criteria (stage I, II, III, IV in 3, 8, 1, and 7 patients, respectively) [##REF##17507545##14##]. Forced expiratory volume in one second (FEV₁) as percentage of predicted (% pred) was 56 [23–75], with an FEV₁/vital capacity (VC) ratio of 50% [30–59]. Thirteen out of 19 patients with COPD were treated with oral or inhaled corticosteroids or their combination. Controls without airflow limitation and without corticosteroid use (8 males, age 56 [47–70], pack years 30 [25–50]) were former (n = 5) or current (n = 4) smokers; smoking status of 1 control was unknown. Median [range] FEV₁% pred of the control group was 103 [90–106], and FEV₁/VC ratio 72% [72–75]. Lung function data were missing from one control, a 43-year-old male organ donor, from whom we obtained lung tissue that had not been used for transplantation because of logistic (unilateral transplantation) reasons. An experienced pulmonary pathologist (WT) found no signs of COPD or other significant pathology in lung tissue.</p>", "<p>Lung tissue samples from GOLD stage IV COPD patients were obtained from tissue remaining after standard pathology protocols in case of lung transplant procedures. The other tissue samples were obtained from subjects undergoing partial resection, (bi)lobectomy, or pneumonectomy for pulmonary carcinoma or metastasis. None of the patients had received chemotherapy. Lung tissue was taken as distant as possible from the tumor, or from non-involved lobes. The study protocol was consistent with national ethical and professional guidelines (\"Code of Conduct; Dutch Federation of Biomedical Scientific Societies\"; <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.federa.org\"/>).</p>", "<title>Tissue processing/immunohistochemistry</title>", "<p>Immunohistochemistry was performed on 3 μm formalin fixed, paraffin embedded lung tissue sections. Consecutive tissue sections were stained for tryptase (AA1, DAKO, Glostrup, Denmark) and chymase (CC1, Abcam, Cambridge, UK) using an immuno-alkaline phosphatase method with goat-anti-mouse immunoglobulin conjugated to alkaline phosphatase as a second step and Fast Red/Napthol AS-Mx as a chromogen providing a bright red reaction product. Double staining was performed to delineate the different compartments. After chymase or tryptase staining, slides were incubated with a mixture of mouse monoclonal antibodies detecting anti-alpha-smooth-muscle actin (Progen, Heidelberg, Germany; clone ASM-1) and pan-keratin (AE1-AE3, Boehringer Mannheim, Mannheim, Germany). As a detection step, slides were incubated with rabbit anti-mouse immunoglobulin conjugated to peroxidase. 3'3'di-amino-benzidin together with H₂O₂was used as a chromogen, providing a gold-brown reaction product. This double staining procedure with subsequent incubation steps allowed easy identification of mast cells with clear demarcation of smooth muscle area and delineation of bronchial epithelium, as well as outer limit of adventitia by (mainly type 2) alveolar epithelial cells (see figure ##FIG##0##1## for illustration). Appropriate isotype matched control sera were used for negative controls.</p>", "<title>Definition of lung regions and morphometric analysis</title>", "<p>We examined 1 central airway and, depending on the availability, 1 to 5 peripheral airways from each subject. Central airways were defined as cartilaginous airways, maximum from the third generation. Three central airway regions were analyzed separately: 1) epithelium; 2) subepithelial area (area between the bronchial epithelium and smooth muscle layer); and 3) airway smooth muscle.</p>", "<p>Peripheral airways were defined as airways with a circumferential size of maximally 6 mm, and a variation of maximal versus minimal diameter ≤ two-fold. Four peripheral airway regions were analyzed separately: 1) epithelium; 2) subepithelial area; 3) airway smooth muscle; and 4) adventitia (area outside smooth muscle to the connection points of the surrounding parenchyma).</p>", "<p>All areas were measured using the Leica QuantiMed morphometric system (Leica Microsystems, Lübeck, Germany) and chymase and tryptase positive, nucleated cells were counted manually within each area and in addition related to all these areas together.</p>", "<p>The percentage of degranulation was determined for all MC-C and MC-T mast cells, using the method of Carroll et al [##REF##12030728##15##]. In short, at a high power (x400) magnification, positively-stained nucleated mast cells were classified as intact if they were dense, compact, had unbroken cytoplasmic boundaries and did not have any surrounding positively stained granules. All other nucleated tryptase or chymase positive cells were classified as degranulated. Degranulated cells were expressed as a percentage of the total number of nucleated MC-T or MC-C-positive cells for the whole small or large airway section.</p>", "<p>A random selection of 10% of all tryptase sections was recounted to determine the intra-class correlation coefficient (ICC), reflecting level of agreement for repeat counts by the same observer. The same was done by two independent observers to determine inter-observer agreement.</p>", "<title>Statistical analysis</title>", "<p>Means and standard deviations or medians with interquartile ranges (IQR) of variables were calculated. Numbers of chymase and tryptase positive mast cells were expressed per mm length of basement membrane (epithelium) or mm²area (subepithelial area, airway smooth muscle, and adventitia). In case two or more peripheral airways from one subject were examined, mean cell concentration of all peripheral airways from this subject was used for statistical analysis. Differences between groups were analyzed using Chi-Square or Mann-Whitney U-tests, paired analyses within groups using Wilcoxon-rank-sum-tests, and correlations between cell concentrations and lung function using Spearman's rank correlation coefficient (r_s). SPSS 12.0 (SPSS Inc., Chicago, IL) software was used for statistical analysis. A p-value of 0.05 or less was considered significant.</p>" ]

[ "<title>Results</title>", "<title>Patient characteristics (Table ##TAB##0##1##)</title>", "<p>Patients with COPD had significantly lower FEV₁(% pred) and FEV₁/VC values than controls without airflow limitation (both p &lt; 0.001). Gender, smoking status, and pack-years smoking were similar in the two groups (p &gt; 0.05).</p>", "<title>Morphometric analysis</title>", "<title>Central airways</title>", "<p>Median [IQR] length of epithelium analyzed in the total study population was 7.5 mm [7.3–7.6], median tissue area 0.7 mm²[0.5–1.2] for the subepithelial area, and 0.6 mm²[0.4–0.9] for airway smooth muscle.</p>", "<title>Peripheral airways</title>", "<p>The length of epithelium per tissue section analyzed in the total study population was 2.3 mm [2.0–2.8] per peripheral airway. Median tissue area analyzed was 0.02 mm²[0.02–0.04] for the subepithelial area, 0.02 mm²[0.01–0.03] for airway smooth muscle, and 0.1 mm²[0.1–0.2] for the adventitia. Comparable tissue areas were analyzed in the two groups (p &gt; 0.05).</p>", "<title>Distribution of mast cells in central and peripheral airways in COPD and controls (Table ##TAB##1##2##, Figure ##FIG##0##1##))</title>", "<title>Tryptase positive mast cells (MC-T)</title>", "<p>In central airways, the concentration of MC-T was highest in the subepithelial area in both groups (Table ##TAB##1##2a##). In peripheral airways (Figure ##FIG##0##1##), the concentration of MC-T was highest in the adventitia followed by the subepithelial area in both groups.</p>", "<p>The concentration of MC-T was significantly lower in the subepithelial area of central airways from patients with COPD than from controls (53.7/mm²[25.7–120.8] versus 108.8/mm²[83.9–190.2], respectively, p = 0.04). Concentrations of MC-T in any of the other regions of central and peripheral airways were similar in both groups (p &gt; 0.05).</p>", "<p>MC-T numbers were significantly higher in subepithelial area in peripheral airways than in central airways in both COPD patients and controls (p = 0.01 and 0.03 respectively) without a difference in numbers in smooth muscle.</p>", "<title>Chymase positive mast cells (MC-C)</title>", "<p>Concentrations of MC-C were highest in the subepithelial area of central airways in both patients with COPD and controls (Table ##TAB##1##2b##). In peripheral airways, the concentration of MC-C was highest in the adventitia followed by the subepithelial area in both groups.</p>", "<p>The concentration of MC-C was significantly higher in the epithelium of central airways from patients with COPD than from controls, but with very low cell numbers (0.0 [0.0–0.0]/mm²versus 0.0/mm²[0.0–0.3], respectively, p = 0.05). Concentrations of MC-C in any other of the regions of both central and peripheral airways were similar in both groups.</p>", "<p>MC-C numbers were significantly higher in both subepithelial area and smooth muscle area of peripheral airways than in central airways in COPD patients (p = 0.03 and 0.05 respectively), but not in controls (p = 0.88 and 0.67 respectively).</p>", "<p>The percentage of degranulated mast cells was higher in peripheral than in central airways for MC-T as well as MC-C (both p &lt; .001), but was not significantly different between COPD and controls.</p>", "<p>Exclusion of 2 control subjects with missing lung function data did not affect any of the results. A subgroup analysis excluding all COPD patients treated with corticosteroids provided similar results as presented above. The inter-group difference in subepithelial MC-T in central airways was reduced to a trend (p = 0.09), probably due to loss of power.</p>", "<p>The ICC for the intra-observer variability was 0.96; the inter-observer variability for two independent observers resulted in an ICC of 0.98. This reflects excellent levels of agreement for repeat counts by either the same observer or two independent observers.</p>", "<title>Relation between mast cells and lung function in COPD patients</title>", "<title>Tryptase positive mast cells</title>", "<p>Higher numbers of MC-T in the epithelium and in the subepithelial area of peripheral airways correlated significantly with higher values of FEV₁/VC (r_s0.47, p = 0.05 and r_s0.48, p = 0.05, respectively; figure ##FIG##1##2a## and ##FIG##1##2b##), but not with FEV₁% pred (r_s0.29, p = 0.25 and r_s0.34, p = 0.17, respectively). Higher numbers of MC-T in the adventitia of peripheral airways tended to correlate with higher values of FEV₁/VC (r_s0.41, p = 0.09) and FEV₁% pred (r_s0.43, p = 0.08). A higher percentage of MC-T in the total airway wall area of peripheral airways also correlated significantly with higher values of FEV₁and FEV₁/VC (r_s0.52, p = 0.03 and r_s0.55, p = 0.02).</p>", "<p>No significant correlations existed between the number of MC-T in any of the other airway regions and lung function (either expressed as FEV₁% pred or FEV₁/VC).</p>", "<title>Chymase positive mast cells</title>", "<p>Higher numbers of MC-C in the smooth airway muscle of peripheral airways correlated significantly with higher values of FEV₁% pred (r_s0.57, p = 0.02; figure ##FIG##2##3##), but not with FEV₁/VC (r_s0.35, p = 0.17). Higher numbers of MC-C in the adventitia of peripheral airways tended to correlate positively with FEV₁% pred (r_s0.43, p = 0.08).</p>", "<p>We found no significant correlations between the number of MC-C in any of the other airway regions and lung function (either expressed as FEV₁% pred or FEV₁/VC).</p>" ]

[ "<title>Discussion</title>", "<p>In the present study, we demonstrate that <italic>the distribution </italic>of tryptase and chymase positive mast cells in the airways is similar in patients with COPD and controls without airflow limitation. In central airways, the concentration of both tryptase and chymase positive mast cells is highest in the subepithelial area. In peripheral airways, the concentration is highest in the adventitia, followed by almost similar high numbers in the subepithelial area. Interestingly, <italic>the numbers </italic>of mast cells differed between COPD and controls. Numbers of tryptase positive mast cells in the subepithelial area of central airways were lower in COPD than in controls, without other relevant significant differences. Finally, a higher number of tryptase and chymase positive mast cells in different regions and for tryptase also in the <italic>total </italic>wall area of peripheral airways is associated with less severe airflow limitation in COPD, relationships not observed in central airways.</p>", "<p>Previous studies on the role of mast cells in COPD have not investigated both central and peripheral airways nor the relation of mast cells and their specific granule contents with lung function [##REF##9403729##5##,##REF##15672853##16##]. We observed no difference in the number of MC-T in peripheral airways between patients with COPD and controls. In contrast, Grashoff <italic>et al</italic>. demonstrated higher numbers of MC-T in the epithelium of peripheral airways, (but not in the remainder of the airway wall) in patients with COPD than in controls [##REF##9403729##5##]. These seemingly incongruent findings may be due to the definition of COPD and controls used in both studies. We adhered to current guidelines [##REF##17507545##14##,##REF##11316667##17##] taking both FEV₁and FEV₁/VC into account, whereas Grashoff's study, conducted before these guidelines were introduced, selected groups solely based on FEV₁. Compatible with our findings, no correlation between MC-T in central airways and FEV₁was observed in another study investigating a heterogeneous population of 29 current smokers with and without airflow limitation [##REF##15672853##16##]. The number of MC-T in peripheral airways was not assessed in the latter study. Interestingly, our findings are in keeping with a recent study in severe asthma [##REF##15563633##13##], where higher numbers of MC-C in peripheral airways were associated with less severe airflow limitation [##REF##15563633##13##]. However, in contrast to our findings, the asthma study did not observe such a relation for the number of MC-T and lung function. Thus, the present study is the first to report the relation of both tryptase and chymase positive mast cells in both central and peripheral airways with lung function in patients with COPD.</p>", "<p>There are a few considerations when interpreting our results. Some studies [##REF##15672853##16##] but not all [##REF##16055612##18##], have shown higher numbers of airway mast cells in current than in ex- or never smokers. We observed a significantly lower number of MC-T in the subepithelial area of central airways in COPD than in controls. This might have been influenced by a higher percentage of current smokers in the control group, yet controls and COPD had similar smoking status. Furthermore, 13 of our 19 COPD patients were treated with corticosteroids versus none of the controls (p &lt; 0.001). Treatment with corticosteroids can reduce mast cell numbers in central airways of patients with COPD [##REF##12200525##19##,##REF##12070058##20##]. However, as indicated in the results section, a subgroup analysis excluding all COPD patients treated with corticosteroids provided similar results as presented above.</p>", "<p>Another point which might have had an impact on our results is that we used lung tissue obtained during surgery for lung cancer. The relation between tumour and mast cells has mainly been studied with respect to mast cells in the tumour tissue and/or effects of mast cells on the tumour [##REF##18632284##21##]. In fact the only effects reported so far are related to mast cells in the immediate environment of tumour-cells or effects of chemotherapy on mast cell number and function.[##REF##18632284##21##] To limit effects of the tumour on airflow limitation and inflammation, we only included patients, who had not received chemotherapy, with a peripherally localized tumour, and selected lung tissue as far as possible from the tumour. Taken together, we feel that the results we describe are valid.</p>", "<p>How to put the present results in perspective of the underlying pathophysiology of COPD? The significant relation of MC-T and MC-C with airflow limitation in COPD was present in peripheral and not central airways, compatible with the widely accepted notion that peripheral airways are the major site of airflow limitation in COPD [##REF##15215480##2##,##REF##5650164##22##]. In line with a previous study in severe asthmatics [##REF##15563633##13##], but yet to our surprise, a positive relation was found between mast cell numbers and lung function. This may either reflect a protective role of mast cells in the pathogenesis of COPD, or in contrast, it may represent a risk factor if lower numbers of MC-T and MC-C reflect increased degranulation of mast cells, leaving them undetected in an immunohistochemical approach [##REF##15778496##23##]. Previous findings in asthma showed that the proportion of degranulated mast cells is higher in peripheral airways from patients with asthma than in controls [##REF##12030728##15##]. In the present study, using the same protocol, we did not find a difference in degranulation in central or small airways between COPD patients and control subjects. For both groups we found more MC-T in peripheral than in subepithelial area of the central airways, whereas in COPD, but not in controls, there are less MC-C in the subepithelial area and more in the smooth muscle area in the peripheral airways than in the central airways. Moreover, for both groups we found a significantly higher percentage of MC-T and MC-C degranulation in peripheral than in central airways. Both numbers and percentage of degranulation of MC-T and MC-C were not different for COPD patients and control subjects. In the article and on-line supplement of Battaglia et al. [##REF##16917001##24##] a similar increase of mast cells in peripheral airways has been described for ex- and current smokers, with or without COPD. Apparently, the quite extensive presence and degranulation of mast cells in peripheral airways does not necessarily lead to airway obstruction and may have other physiologic effects.</p>", "<p>We realize that there are several arguments against a protective role for mast cells in COPD, since they secrete e.g. proteases, interleukin-8 [##REF##9212819##25##] and tumour necrosis factor (TNF). [##REF##18031566##26##,##REF##17259966##27##] These cytokines contribute to neutrophilic inflammation in COPD [##REF##8564092##28##, ####REF##9266891##29##, ##REF##12162440##30####12162440##30##], which is in turn associated with more severe airflow limitation [##REF##9769292##8##]. Indeed mast cells generally have detrimental effects in various inflammatory conditions [##REF##14760748##31##, ####REF##12215644##32##, ##REF##10704463##33####10704463##33##]. However, mast cells can have quite opposing effects in different situations, rather dependent on their micro-environment and the mast cell modulating factors that play a role at that particular microlocalization [##REF##18055703##9##]. Moreover, there are potential beneficial effects as well. Mast cells were associated with a protective role in e.g. the development of glomerulonephritis in mice models of renal inflammatory disease [##REF##16622030##34##]. Furthermore, they might protect against the detrimental smoke effects on lung function by the ability of tryptase to enhance epithelial cell proliferation [##REF##8598474##35##], which could improve epithelial integrity. In addition, chymase may inhibit smooth muscle proliferation and thus the associated increased peripheral airway resistance in patients with COPD [##REF##12097409##36##]. The proteases that are produced by mast-cells may limit the development of subepithelial fibrosis. Finally, mast cells can also have the opposite effect and contribute to fibroblast proliferation and collagen production [##REF##8422426##10##, ####REF##11929488##11##, ##REF##10887338##12####10887338##12##,##REF##11734467##37##]. This might increase extracellular matrix production in the adventitia, thereby counteracting effects of emphysematous destruction of peribronchial attachments.</p>", "<p>In conclusion, in our view, the strength of the current study is that it is the first to investigate the distribution of tryptase and chymase positive mast cells in central and peripheral airways, and their relation with airflow limitation in patients with COPD. It remains to be clarified whether our findings reflect a beneficial or a detrimental effect of mast cells, or even a combination of both, on lung function in patients with COPD.</p>" ]

[]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>In asthma, higher chymase positive mast cell (MC-C) numbers are associated with less airway obstruction. In COPD, the distribution of MC-C and tryptase positive mast cells (MC-T) in central and peripheral airways, and their relation with lung function, is unknown. We compared MC-T and MC-C distributions in COPD and controls without airflow limitation, and determined their relation with lung function.</p>", "<title>Methods</title>", "<p>Lung tissue sections from 19 COPD patients (median [interquartile range] FEV₁% predicted 56 [23–75]) and 10 controls were stained for tryptase and chymase. Numbers of MC-T and MC-C were determined in different regions of central and peripheral airways and percentage of degranulation was determined.</p>", "<title>Results</title>", "<p>COPD patients had lower MC-T numbers in the subepithelial area of central airways than controls. In COPD, MC-T numbers in the airway wall and more specifically in the epithelium and subepithelial area of peripheral airways correlated positively with FEV₁/VC (Spearman's rho (r_s) 0.47, p = 0.05 and r_s0.48, p = 0.05, respectively); MC-C numbers in airway smooth muscle of peripheral airways correlated positively with FEV₁% predicted (r_s0.57, p = 0.02). Both in COPD patients and controls the percentage of degranulated MC-T and MC-C mast cells was higher in peripheral than in central airways (all p &lt; 0.05), but this was not different between the groups.</p>", "<title>Conclusion</title>", "<p>More MC-T and MC-C in peripheral airways correlate with better lung function in COPD patients. It is yet to determine whether this reflects a protective association of mast cells with COPD pathogenesis, or that other explanations are to be considered.</p>" ]

[ "<title>Abbreviations</title>", "<p>ASM: alpha-smooth-muscle actin; COPD: Chronic Obstructive Pulmonary Disease; FEV: Forced Expiratory Volume; GOLD: Global Initiative on Lung Disease; IQR: interquartile ranges; MC-C: mast cell-chymase; MC-T: mast cell-tryptase; VC: Vital Capacity.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>MG carried out the data analysis and drafted the manuscript. DP, NHTtH and WT participated in the design of the original study, were responsible for clinical and histological patient data and contributed substantially to the manuscript. BR, ML, MS, and MAL carried out the immunohistology experiments and contributed to the manuscript. JV assisted with data analysis and interpretation, supervised statistics and contributed to the manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>Sources of support: Netherlands Organization for Scientific Research (NWO)</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>1a. (left) Double immunostaining: tryptase positive mast cells (bright red) in a small airway in a tissue sample from a patient with moderate COPD (GOLD stage 2); bronchial and alveolar type 2 epithelial cells and smooth muscle stain golden brown, magnification × 100; 1b. (right) larger magnification (× 200).</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>The relation between the concentration of tryptase positive mast cells and lung function in patients with COPD.</bold> 2a. MC-T in the epithelium of peripheral airways and FEV₁/VC (%). 2b. MC-T in the subepithelial area of peripheral airways with FEV₁/VC (%). FEV₁/VC = the ratio of the forced expiratory volume in one second and vital capacity; MC-T = tryptase positive mast cells.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>The relation between the concentration of chymase positive mast cells in peripheral airways smooth muscle with FEV₁% predicted in patients with COPD</bold>. FEV₁= forced expiratory volume in one second; % predicted = percentage of the predicted value; MC-C = chymase positive mast cells.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Characteristics of COPD patients and controls</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\"><bold>Controls</bold></td><td align=\"center\"><bold>COPD</bold></td></tr></thead><tbody><tr><td align=\"left\">N</td><td align=\"center\">10</td><td align=\"center\">19</td></tr><tr><td align=\"left\">Male/female</td><td align=\"center\">8/2</td><td align=\"center\">10/9</td></tr><tr><td align=\"left\">Current smoker (y/n/unknown)</td><td align=\"center\">4/5/1*</td><td align=\"center\">5/14/0</td></tr><tr><td align=\"left\">Age (median IQR)</td><td align=\"center\">56 (47–70)</td><td align=\"center\">62 (54–72)</td></tr><tr><td align=\"left\">Packyears (median (range))</td><td align=\"center\">30 (25–50) *</td><td align=\"center\">31 (27–40)</td></tr><tr><td align=\"left\">ICS (y/n/unknown)</td><td align=\"center\">0/9/1*</td><td align=\"center\">8/11/0</td></tr><tr><td align=\"left\">OCS (y/n/unknown)</td><td align=\"center\">0/9/1*</td><td align=\"center\">7/12/0</td></tr><tr><td align=\"left\">ICS and/or OCS (y/n/unknown)</td><td align=\"center\">0/9/1*</td><td align=\"center\">13/6/0</td></tr><tr><td align=\"left\">B2 agonist (y/n/unknown)</td><td align=\"center\">0/9/1*</td><td align=\"center\">10/9/0</td></tr><tr><td align=\"left\">Anticholinergics (y/n/unknown)</td><td align=\"center\">0/9/1*</td><td align=\"center\">10/9/0</td></tr><tr><td align=\"left\">COPD Severity Stage (GOLD 1/2/3/4)</td><td align=\"center\">-</td><td align=\"center\">3/8/1/7</td></tr><tr><td align=\"left\">FEV₁(% predicted)</td><td align=\"center\">103 (92–107)**</td><td align=\"center\">56 (23–75)</td></tr><tr><td align=\"left\">FEV₁/IVC (%)</td><td align=\"center\">72 (72–75)**</td><td align=\"center\">50 (30–59)</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\" colspan=\"3\"><bold>a</bold>. The concentration of tryptase positive mast cells (MC-T) per region for patients with COPD and controls</td></tr></thead><tbody><tr><td/><td align=\"center\"><bold>Controls</bold><break/><bold>(n = 10)</bold></td><td align=\"center\"><bold>COPD</bold><break/><bold>(n = 19)</bold></td></tr><tr><td colspan=\"3\"><hr/></td></tr><tr><td align=\"left\"><bold><italic>Central airways</italic></bold></td><td/><td/></tr><tr><td align=\"left\">Epithelium (cells/mm)</td><td align=\"center\">0.07 [0.0–0.9]</td><td align=\"center\">0.0 [0.0–0.1]</td></tr><tr><td align=\"left\">Subepithelial area (cells/mm²)</td><td align=\"center\">108.8 [83.9–190.2]</td><td align=\"center\">53.7 [25.7–120.8]^¶</td></tr><tr><td align=\"left\">Airway smooth muscle (cells/mm²)</td><td align=\"center\">32.5 [15.8–49.9]</td><td align=\"center\">20.1 [10.5–33.3]</td></tr><tr><td align=\"left\">% degranulated mast cells</td><td align=\"center\">35.9 [21.9–51.2]</td><td align=\"center\">28.0 [22.4–37.4]</td></tr><tr><td align=\"left\"><bold><italic>Peripheral airways</italic></bold></td><td/><td/></tr><tr><td align=\"left\">Epithelium (cells/mm)</td><td align=\"center\">0.0 [0.0–0.9]</td><td align=\"center\">0.0 [0.0–0.2]</td></tr><tr><td align=\"left\">Subepithelial area (cells/mm²)</td><td align=\"center\">247.9 [73.9–548.2]</td><td align=\"center\">184.6 [100.4–282.3]</td></tr><tr><td align=\"left\">Airway smooth muscle (cells/mm²)</td><td align=\"center\">12.6 [0.0–36.9]</td><td align=\"center\">18.4 [0.0–47.6]</td></tr><tr><td align=\"left\">Adventitia (cells/mm²)</td><td align=\"center\">279.4 [79.2–646.9]</td><td align=\"center\">214.9 [107.2–327.2]</td></tr><tr><td align=\"left\">% degranulated mast cells</td><td align=\"center\">68.6 [41.5–81.4]</td><td align=\"center\">67.5 [53.0–84.1]</td></tr><tr><td colspan=\"3\"><hr/></td></tr><tr><td align=\"left\" colspan=\"3\"><bold>b</bold>. The concentration of chymase positive mast cells (MC-C) per region for patients with COPD and controls</td></tr><tr><td colspan=\"3\"><hr/></td></tr><tr><td/><td align=\"center\"><bold>Controls</bold><break/>(n = 10)</td><td align=\"center\"><bold>COPD</bold><break/>(n = 19)</td></tr><tr><td colspan=\"3\"><hr/></td></tr><tr><td align=\"left\"><bold><italic>Central airway samples</italic></bold></td><td/><td/></tr><tr><td align=\"left\">Epithelium (cells/mm)</td><td align=\"center\">0.0 [0.0–0.0]</td><td align=\"center\">0.0 [0.0–0.3]^¶¶</td></tr><tr><td align=\"left\">Subepithelial area (cells/mm²)</td><td align=\"center\">18.1 [8.2–28.9]</td><td align=\"center\">11.0 [4.5–28.3]</td></tr><tr><td align=\"left\">Airway smooth muscle (cells/mm²)</td><td align=\"center\">10.6 [2.7–15.2]</td><td align=\"center\">3.5 [0.0–6.0]</td></tr><tr><td align=\"left\">% degranulated mast cells</td><td align=\"center\">33.3 [25.7–52.2]</td><td align=\"center\">40.0 [26.8–58.3]</td></tr><tr><td align=\"left\"><bold><italic>Peripheral airways</italic></bold></td><td/><td/></tr><tr><td align=\"left\">Epithelium (cells/mm)</td><td align=\"center\">0.0 [0.0–0.0]</td><td align=\"center\">0.0 [0.0–0.2]</td></tr><tr><td align=\"left\">Subepithelial area (cells/mm²)</td><td align=\"center\">19.4 [0.0–71.8]</td><td align=\"center\">54.5 [0.0–160.0]</td></tr><tr><td align=\"left\">Airway smooth muscle (cells/mm²)</td><td align=\"center\">0.0 [0.0–54.1]</td><td align=\"center\">0.0 [0.0–50.7]</td></tr><tr><td align=\"left\">Adventitial layer (cells/mm²)</td><td align=\"center\">128.0 [44.1–184.0]</td><td align=\"center\">68.5 [13.2–162.7]</td></tr><tr><td align=\"left\">% degranulated mast cells</td><td align=\"center\">66.8 [55.6–93.2]</td><td align=\"center\">63.9 [52.1–95.0]</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>Data are presented as median (IQR)</p><p>* n = 9; 1 subject (anonymous donor) with some missing data; ** n = 8, 2 subjects (incl. donor) with missing lung function data</p></table-wrap-foot>", "<table-wrap-foot><p>Data are presented as median [interquartile range]; ^¶p = 0.035 versus controls; ^¶¶p = 0.045 vs controls</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1465-9921-9-64-1\"/>", "<graphic xlink:href=\"1465-9921-9-64-2\"/>", "<graphic xlink:href=\"1465-9921-9-64-3\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Psychopaths are characterized by specific deficits in interpersonal relations, affective attributes and behavioral features [##UREF##0##1##, ####UREF##1##2##, ##UREF##2##3####2##3##]. The Psychopathy Checklist-Revised (PCL-R) [##UREF##1##2##] was designed to measure these attributes and is considered to be the gold standard for the diagnosis of psychopathy, as its diagnostic validity is well replicated [##REF##15914717##4##]. The PCL-R consists of 20 items, which are assessed on a three-step scale. Factor analyses of the PCL-R items found two factors: Factor 1 represents items which assess interpersonal and affective characteristics, and Factor 2 represents behavioral and lifestyle factors, such as impulsivity and antisocial behaviors. Other studies have even found a much debated three [##REF##11433793##5##] and a robust four factor model of psychopathy [##UREF##3##6##]. Even though the majority of studies examining the predictive validity of the PCL-R for criminal recidivism were conducted on mostly Caucasian North American prisoners, several studies were able to demonstrate the usefulness of this instrument for ethnically diverse populations in English-speaking countries [##REF##11113965##7##,##UREF##4##8##]. All in all, the PCL-R score can be considered a solid and accurate estimate for recidivism and especially violent recidivism in English-speaking countries, as well as in Europe [##REF##11113965##7##,##UREF##5##9##,##REF##10333757##10##]. In German-speaking countries, there have been few studies examining the validity of the PCL-R and the PCL:SV in predicting re-offending and they confirm, at the least, a moderate predictive validity of the instrument [##UREF##6##11##, ####REF##17031774##12##, ##REF##16955314##13####16955314##13##]. A study examining 428 forensically assessed offenders in Switzerland also found a moderate accuracy for the screening version of the PCL-R, the PCL:SV, with an AUC value of 0.65 (mean score in the PCL:SV: 9.4 points) [##UREF##7##14##].</p>", "<p>For many professionals working in the criminal justice system, the question arises whether the PCL-R is also able to predict violence during institutionalization (correctional facility, forensic hospital, psychiatric hospital). The relationship between psychopathy and inpatient disruptive behavior has been studied in numerous samples of – mostly North American – male prisoners and psychiatric patients [##REF##14680527##15##]. Early studies reported a significant association [##UREF##8##16##,##REF##6519865##17##], but their methodology was criticized by Cunningham and Reidy [##REF##9768465##18##] on the grounds that the criterion of interest had not been adequately defined. More recent studies among prison populations indicated only a moderate association between psychopathy and inmate misbehavior [##REF##10653992##19##, ####UREF##9##20##, ##UREF##10##21##, ##UREF##11##22####11##22##]. More specifically, research conducted with <italic>forensic psychiatric patients </italic>found at least a moderate correlation between violence in institutions and psychopathy [##REF##14680527##15##,##UREF##9##20##,##REF##8845529##23##,##REF##8934700##24##]. Rice, Harris and Cormier [##UREF##12##25##] compared high and low PCL-R scores in a Canadian forensic hospital and found an association between high PCL-R scores (PCL-R &gt; 25) and a higher level of behavioral problems during treatment, including more episodes of seclusion during the first and last year of treatment. Gray, Hill, McGleish, Timmons, MacCulloch, and Snowden, [##REF##12795569##26##] studied 34 mentally disordered offenders from a medium-security hospital in the U.K. The total PCL-R score had moderate predictive validity for damage to property and physical violence. In a sample of 218 male offenders (aged 17–71 years) admitted to a State Hospital with psychiatric disorders, Heilbrun et al. [##UREF##9##20##] found a significant correlation between the total number of aggressive incidents in the first 2 months of hospitalization and the total PCL score. Douglas et al. [##REF##15914717##4##] found a mean score of 13 for the Psychopathy Checklist Screening Version (PCL:SV) among 216 male forensic patients of a Swedish hospital [##UREF##13##27##]. The AUC of the PCL:SV in predicting aggression was 0.63. On the other hand, there are – interestingly – only few studies examining the relationship between PCL-R score and violent behavior among <italic>prison inmates</italic>. Edens et al. [##REF##10653992##19##] examined the discriminative validity of the PCL-R among an ethnically diverse, young, and randomly selected population of 50 male English-speaking inmates of a US prison. The data showed a generally modest but statistically significant correlation between the PCL-R and indexes of aggressive institutional behavior during the first year of incarceration: The correlation between PCL-R score and physical aggression was 0.18 (not significant) but 0.28 for verbal aggression (significant). Kroner and Mills [##UREF##14##28##] examined institutional misconduct in 97 violent offenders sentenced to 2 to 6 years in Canada over an 8 month period. The mean PCL-R score was 19.7 and the prevalence of major misconduct was 36%, including e.g. rioting, threatening, drug use and assault, however, only two incidents were concerned with threatening or attempting to assault staff. The authors found that the PCL-R was barely predictive (r = 0.14, AUC = .58). Edens, Buffington-Vollum, Colwell, Johnson and Johnson [##UREF##15##29##] examined 92 incarcerated sex offenders. They found that the sum score was predictive for physical and verbal aggressive infractions. This finding contrasts somewhat with the results of Buffington-Vollum, Edens, Johnson and Johnson [##UREF##16##30##] in a prospective study of 58 sex offenders incarcerated in the Texas Department of Criminal Justice. They reported base rates for physical and verbal aggression of 8% and 39%. The authors concluded that the PCL-R correlated moderately with verbal aggression, non-aggressive offenses, and 'any disciplinary offense', but not with physically aggressive offenses. Coid, Petruckevitch, Bebbington, Jenkins, Brugha, Lewis, et al. [##UREF##17##31##] examined psychopathy in a sample of 496 prisoners in England and Wales using the PCL-R and found that a higher proportion of inmates segregated due to disciplinary infractions scored 25 or higher on the PCL-R (OR = 3.08).</p>", "<p>There is some evidence for the usefulness of the PCL-R in predicting disciplinary infractions in institutions. It is unclear, though, whether the PCL-R is an accurate instrument to predict violent behavior in prisons, and results seem to indicate that the instrument is better suited for predicting verbal rather than physical aggression. The objective of this study was to examine the predictive validity of the PCL-R for physical and verbal aggression in a sample of imprisoned sex and violent offenders in Switzerland. In addition, the motives and consequences of the aggressive behavior were analyzed.</p>" ]

[ "<title>Methods</title>", "<title>Ethical approval</title>", "<p>The sample of the present study is a subsample of a large epidemiological study conducted on convicted offenders (inmates as well as offenders on probation) in the Canton of Zurich in the year 2000, which was approved as a whole by an external Ethics Committee, the Kantonale Ethikkommission Zürich (KEK – <ext-link ext-link-type=\"uri\" xlink:href=\"http://www.swissethics.ch/index.php?id=10\"/>). In agreement with the committee, no informed consent had to be obtained as there was no contact with any of the study subjects. All data was collected entirely from the subjects' files and anonymised before further analysis.</p>", "<title>Characteristics of the Swiss penitentiary</title>", "<p>The subjects examined in the present study were all inmates of the maximum security unit of the state penitentiary \"Pöschwies\" in the Canton of Zurich, Switzerland. The \"Pöschwies\" is the largest and most modern penitentiary in Switzerland and has space for 436 inmates. The penitentiary contains three prisons. The main prison is a <italic>maximum security unit </italic>and contains offenders, who have to serve at least a 2 years prison term. The second prison is a <italic>medium security unit </italic>and contains offenders who serve short prison sentences and the third prison is a <italic>minimum security unit </italic>which contains mostly first time offenders and offenders from the maximum security unit before they are discharged to a halfway house.</p>", "<p>In the maximum security unit every prisoner has his own cell and can rent a private television set, as well as a computer. The doors of the cells are locked for approximately 12 hours a day (during the night). The prisoners have to work 7 hours a day, from Monday to Friday, in the industrial workshops of the penitentiary. There are 11 different industrial workshops (e.g. print office, bakery, laundry, painter's shop, joiner's workshop, mechanic's workshop, bookbindery). Depending on their vocational qualification, the prisoners can earn up to 600 Swiss Francs (approximately 550$) per month and have the opportunity to enroll in a vocational education program. Inmates live in group homes, which house up to 20 persons each. All members of a group home have their meals in their own refectory, and have the possibility of spending time together in their own sitting room. Furthermore, in their spare time, the inmates can practice sports, participate in courses or spend time in their cell. Offenders who do not present an acute risk to others are allowed to see visitors once a week. Married offenders can use a private room for conjugal visits. Aside from the regular group homes there are special needs group homes: One for offenders with long-term prison sentences, one for high risk offenders, one for offenders with substance use problems and one group home for older prisoners. The inmates of the special needs group homes benefit from several privileges: Their cells are locked later in the evening, they don't have to go to work, they can cook their own meals etc.</p>", "<p>270 prison officers and master-workmen take care of 436 inmates. In case a prisoner needs medical care there are two general practitioners, two psychiatrists, two part time dentists, two part time physical therapists and four nurses available. The inmates can also apply for offense-oriented psychotherapy. There are currently nine fulltime forensic psychologists offering specialized treatment programs. Approximately 15% of the offenders (n = 70) take part in a single and/or group therapy with an intensity of 1 to 12 hours per week.</p>", "<title>Sample selection criteria</title>", "<p>Only the inmates of the maximum security unit were examined for inclusion in the study. The inclusion criteria were: (1) conviction due to a violent or sex offense, (2) sentenced to at least 10 months imprisonment, (3) administrated by the Zurich correctional and probation service in August 2000, and (4) existence of a psychiatric expert assessment in the offender's prison records. 123 offenders fulfilled these inclusion criteria. The exclusion of all subjects with more than four omitted items in the PCL-R reduced the sample to 113 subjects.</p>", "<p>All subjects were male, 57.5% (n = 65) were Swiss nationals and 12.3% (n = 14) originated from an EU country. Further countries of origin were Russia, China, the Philippines, Sri Lanka, Peru, the Dominican Republic, Jamaica, Croatia, Lebanon, and Turkey. The mean age at the beginning of the sentence was 36.3 years (SD = 8.9, range: 20–60). At the time of the offense, 15.2% (n = 17) of the offenders were married, 34.9% (n = 38) had a child, and 31.2% (n = 34) had lived in a foster home before the age of 15. 83.2% (n = 94) of the offenders had a criminal record prior to the index offense and 31.2% (n = 34) had previously been treated in an inpatient psychiatric facility. Psychiatric disorders according to ICD-10 [##UREF##18##32##], with 86.7% (n = 98), were very prevalent in our sample. Over half of the 114 offenders had been diagnosed with an affective disorder (56.6%; n = 64) and 16.8% (n = 19) with a personality disorder. Prevalence was also high for diagnoses of schizophrenia at 6.2% (n = 7). The mean time spent in the \"Pöschwies\" penitentiary at the time of the investigation was 55 months (SD = 34.98, range: 0.6–169). In nearly half of the sample (47.8%, n = 54) the index offense was murder or manslaughter. 9.8% (n = 22) of the offenders were convicted of rape, 10.6% (n = 12) of child abuse, 8.9% (n = 10) of armed robbery, 5.3% (n = 6) of physical assault, and 5.3% (n = 6) of arson. Three cases (2.7%) matched none of these categories.</p>", "<title>Procedure and measures</title>", "<p>The scoring of the PCL-R as well as the assessment of psychiatric, psychological, criminological, and socio-economic variables were performed based on file data. At no time was there any direct contact with the prisoners. The files contained extensive historical details on the subject, including criminal and medical/psychiatric history, exact type and circumstances of the offense, as well as a personality assessment. To assess psychopathy, no direct contact is necessary if enough collateral information can be gathered from files and expert psychiatric assessments [##UREF##1##2##]. The interrater reliability (of n = 10 cases) was assessed with Krippendorff's alpha [##UREF##19##33##]. The advantage of this coefficient is that it can be used to analyze the agreement of multiple raters, even if there are unequal sample sizes or missing data. Furthermore, it can be computed when the variables are nominal, ordinal, or continuous. For the PCL-R, Krippendorff's alpha was 0.89.</p>", "<p>In a first step PCL-R scores and the outcome variable were assessed. To prevent any bias, the PCL-R was scored before evaluating the outcome variable (physical and verbal aggression). Inmate behavior was assessed using the penitentiary's files. Physical aggression was defined as physical behavior that harmed or had the potential to harm others (staff members or other prisoners). Verbal aggression was defined as threats or gross verbal abuse. Damage to property was not considered an aggressive infraction. Further categories of infractions were illegal drug abuse and the possession of illegal drugs. In a second step, two psychologists coded the motives of the violent infractions, as well as their consequences, independently of each other. The motives were categorized as: (1) conflict with prison officer (threatening), (2) minor assault without physical harm, (3) starting a fight following a verbal conflict (reactive violence), (4) deliberate use of violence without prior verbal conflict (instrumental violence). The consequences of the violence used were categorized as (1) no harm, (2) minor harm, (3) moderate harm (outpatient medical treatment was necessary), and (4) severe harm (inpatient medical treatment was necessary). The interrater agreement was nearly perfect. Only in two instances the raters disagreed. After consulting the files, the raters were able to reach a consensus.</p>", "<title>Hypothesis and statistical analysis</title>", "<p>The authors hypothesized that the PCL-R would be a good instrument for predicting in-prison aggressive behavior. This assumption was tested by logistic regression analyses, where physical and verbal aggression were analyzed as function of the PCL-R sum score, employing a 5% level of significance. Additionally, stratified analyses were conducted for Factor 1 and 2 of the PCL-R.</p>", "<p>The results of all logistic regression analyses were controlled for time in prison by entering the natural log of time at risk, with its parameter fixed at 1 as a covariate, into the model.</p>", "<p>Predictive validity was estimated with ROC analyses. All models were computed with STATA SE 10.0.</p>" ]

[ "<title>Results and discussion</title>", "<title>Disciplinary infractions and classification of aggressive behavior</title>", "<p>83.2% (n = 94) of the inmates had been reported at least once for a disciplinary infraction during incarceration. The average number of incidents per prisoner was 5.1 (SD = 5.94, range: 0–26). With regard to the type of disciplinary infraction, 13.3% (n = 15) of inmates had been reported at least once for the use or possession of illegal drugs. 25.6% (n = 29) of offenders had been reported due to at least one verbally aggressive incident, and one third of the subjects (27.4%, n = 31) had been reported at least once due to physical aggression. 79.6% (n = 90) had been reported at least once for another non-violent infraction.</p>", "<p>For 28 of the 31 offenders behaving physically aggressive, the cause and consequences of the conflict were documented and allowed categorization: One of the offenders had a conflict with a prison officer, 7 committed a minor assault without physical harm, 17 started a fight following a verbal conflict (= reactive violence) and only 3 prisoners were deliberately violent without prior verbal conflict (= instrumental violence). Two thirds of the offenders behaving violently caused no physical harm (n = 21). Outpatient medical treatment was necessary due to 3 and inpatient medical treatment due to 4 offenders behaving violently.</p>", "<title>Differences between offenders with and without physically aggressive behavior</title>", "<p>Stratified analyses with bivariate logistic regression showed no significant difference between physically aggressive and physically non-aggressive inmates with respect to marital status (p ≤ 0.84), criminal record (p ≤ 0.23), index offense (p ≤ 0.65), age at the beginning of incarceration (p ≤ 0.46), time spent in the institution (p ≤ 0.07), and vocational education (p ≤ 0.75).</p>", "<title>PCL-R-scores in relation to physical and verbal aggression</title>", "<p>The PCL-R sum scores were not normally distributed. Both the mean score and the median of the PCL-R were 12 points (SD = 6.6), with scores ranging between 0 and 33.7, the 25^thpercentile was at 6.3 points and the 75^thpercentile at 17 points. Both the relationship between physical aggression and the sum score of the PCL-R, and the relationship between physical aggression and either of the two factors of the PCL-R, were not significant. However, both the sum score and Factor 1 predicted the occurrence of verbal aggression (AUC = 0.70 and 0.69), while Factor 2 did not (Table ##TAB##0##1##).</p>" ]

[ "<title>Results and discussion</title>", "<title>Disciplinary infractions and classification of aggressive behavior</title>", "<p>83.2% (n = 94) of the inmates had been reported at least once for a disciplinary infraction during incarceration. The average number of incidents per prisoner was 5.1 (SD = 5.94, range: 0–26). With regard to the type of disciplinary infraction, 13.3% (n = 15) of inmates had been reported at least once for the use or possession of illegal drugs. 25.6% (n = 29) of offenders had been reported due to at least one verbally aggressive incident, and one third of the subjects (27.4%, n = 31) had been reported at least once due to physical aggression. 79.6% (n = 90) had been reported at least once for another non-violent infraction.</p>", "<p>For 28 of the 31 offenders behaving physically aggressive, the cause and consequences of the conflict were documented and allowed categorization: One of the offenders had a conflict with a prison officer, 7 committed a minor assault without physical harm, 17 started a fight following a verbal conflict (= reactive violence) and only 3 prisoners were deliberately violent without prior verbal conflict (= instrumental violence). Two thirds of the offenders behaving violently caused no physical harm (n = 21). Outpatient medical treatment was necessary due to 3 and inpatient medical treatment due to 4 offenders behaving violently.</p>", "<title>Differences between offenders with and without physically aggressive behavior</title>", "<p>Stratified analyses with bivariate logistic regression showed no significant difference between physically aggressive and physically non-aggressive inmates with respect to marital status (p ≤ 0.84), criminal record (p ≤ 0.23), index offense (p ≤ 0.65), age at the beginning of incarceration (p ≤ 0.46), time spent in the institution (p ≤ 0.07), and vocational education (p ≤ 0.75).</p>", "<title>PCL-R-scores in relation to physical and verbal aggression</title>", "<p>The PCL-R sum scores were not normally distributed. Both the mean score and the median of the PCL-R were 12 points (SD = 6.6), with scores ranging between 0 and 33.7, the 25^thpercentile was at 6.3 points and the 75^thpercentile at 17 points. Both the relationship between physical aggression and the sum score of the PCL-R, and the relationship between physical aggression and either of the two factors of the PCL-R, were not significant. However, both the sum score and Factor 1 predicted the occurrence of verbal aggression (AUC = 0.70 and 0.69), while Factor 2 did not (Table ##TAB##0##1##).</p>" ]

[ "<title>Conclusion</title>", "<p>In this study only a moderate association between PCL-R scores and prison misconduct was observed. This finding corroborates the results of several other studies [##REF##14680527##15##,##UREF##9##20##,##REF##8845529##23##,##REF##8934700##24##,##UREF##20##34##]. More specifically, in our study the PCL-R was primarily predictive for verbal aggression (e.g. threats) but not for physical aggression (e.g. physical assaults). This finding replicates the results of the study by Edens et al. [##REF##10653992##19##] and Buffington-Vollum et al. [##UREF##16##30##], where high PCL-R scores were found to be related to behavioral problems (e.g. verbal aggression) but not to physical violence.</p>", "<p>Therefore, the question arises, how the weak relationship between PCL-R scores and physically aggressive inmate behavior can be explained. Some authors have discussed whether the low base-rate of violent infractions can be considered an explanation for the non-significant relation between PCL-R-score and violence [##REF##14680527##15##]. The prevalence of violence in our study, with 27%, was reasonably high. However, the base rate of severe violence was low, as is documented by the fact that medical treatment was only necessary in 7 of the 94 cases where offenders were reported for disciplinary infractions. Low prevalence effects can therefore not be excluded from this study as a possible explanation for the weak relationship between the PCL-R score and physically aggressive behavior. It can be argued, on the other hand, that the base rate of severe violence may well have been higher, given that in most cases of violent behavior prison staff will intervene before severe violence can develop. Future investigations in the field should assess the prison staff's influence on the severity of violent inmate behavior.</p>", "<p>There is another explanation aside from the low prevalence effect to be considered. When studying the association between psychopathic traits and violence, there is evidence suggesting that the motives of the perpetrators are important: Porter and Woodworth [##UREF##21##35##] differentiated between reactive and instrumental motivations. According to Woodworth and Porter [##REF##12150419##36##] most of the violence committed by psychopaths is instrumental. Even though the base rate for violence was not low in our sample, the base rate for instrumental violence was very low. Most of the violent infractions in the Zurich state penitentiary were reactive in nature as they resulted from verbal disagreements which led to marginal or moderate physical violence. Since the inmates examined in our study did not display an instrumental type of violence, our findings do not rule out the usefulness of the PCL-R to predict intramural violence. Furthermore, it has to be considered that in our sample the mean PCL-R score, with 12 points, was low – even though it was comparable to other studies from German speaking countries [##REF##16780950##37##]. Moreover, according to Buffington-Vollum et al. [##UREF##16##30##], restrictive environmental factors may inhibit the aggressive tendencies of people who might be violent in less restrictive settings. The Zurich state penitentiary is a highly controlled environment that is designed to prevent violence: The inmates have single cells – this measure increases the costs but allows privacy, which in turn lowers the prevalence of violent behavior through the seclusion of inmates for 12 hours a day. A workday is typically spent in small groups constantly supervised by staff members, who immediately respond to violence (verbal or physical) by separating the perpetrators and reporting them to the prison directorate. Furthermore, there are nine psychologists, two psychiatrists, and two general practitioners taking care of 316 inmates (of the maximum security unit). The ratio of mental health professionals to inmates of 1:24 can be considered high and could also be a relevant factor in reducing in-prison violence. In addition, many offenders receive offense-oriented forensic psychotherapy during imprisonment, which aims at reducing aggressive tendencies and trains empathy in offenders, both of which may help lower prevalence of aggressive incidences.</p>", "<p>Since most of the intramural violence observed could be classified as reactive, the usefulness of the PCL-R – especially when using only the sum score as predictor – is questionable – especially in highly controlled prison settings such as the Zurich state penitentiary. Another recent study from Switzerland showed that the Violence Risk-Appraisal Guide (VRAG) [##UREF##22##38##] also failed to predict violent infractions [##REF##17615429##39##], indicating that not only the PCL-R is problematic when it comes to predicting in-prison violence. Thus the development of a specific instrument or model to predict <italic>reactive </italic>institutional violence seems to be needed. Furthermore, future research should assess specifically, whether the PCL-R is useful to predict <italic>instrumental </italic>in-prison violence, rather than trying to assess if it can predict in-prison violence in general.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Research conducted with forensic psychiatric patients found moderate correlations between violence in institutions and psychopathy. It is unclear though, whether the PCL-R is an accurate instrument for predicting aggressive behavior in prisons. Results seem to indicate that the instrument is better suited for predicting verbal rather than physical aggression of prison inmates.</p>", "<title>Methods</title>", "<p>PCL-R scores were assessed for a sample of 113 imprisoned sex and violent offenders in Switzerland. Logistic regression analyses were used to estimate physical and verbal aggression as a function of the PCL-R sum score. Additionally, stratified analyses were conducted for Factor 1 and 2. Infractions were analyzed as to their motives and consequences.</p>", "<title>Results</title>", "<p>The mean score of the PCL-R was 12 points. Neither the relationship between physical aggression and the sum score of the PCL-R, nor the relationship between physical aggression and either of the two factors of the PCL-R were significant. Both the sum score and Factor 1 predicted the occurrence of verbal aggression (AUC = 0.70 and 0.69), while Factor 2 did not.</p>", "<title>Conclusion</title>", "<p>Possible explanations are discussed for the weak relationship between PCL-R scores and physically aggressive behavior during imprisonment. Some authors have discussed whether the low base rate of violent infractions can be considered an explanation for the non-significant relation between PCL-R-score and violence. The base rate in this study, however, with 27%, was not low. It is proposed that the distinction between reactive and instrumental motives of institutional violence must be considered when examining the usefulness of the PCL-R in predicting in-prison physical aggressive behavior.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>JE has made substantial contributions to the conception and design of the study, as well as the acquisition, analysis and interpretation of data and drafting of the manuscript. He has given his final approval for this version of the manuscript to be published. FU has made substantial contributions to the analysis and interpretation of data and has been involved in critically revising the manuscript. He has given his final approval for this version of the manuscript to be published. AL has made substantial contributions to the analysis and interpretation of data, has been involved in drafting and critically revising the manuscript. She has given her final approval for this version of the manuscript to be published. AR has made substantial contributions to the conception and design of the study, as well as the acquisition, analysis and interpretation of data and been involved in drafting the manuscript. She has given her final approval for this version of the manuscript to be published. SV has made substantial contributions to the conception and design of this study and has been involved in critically drafting and revising the manuscript. He has given his final approval for this version of the manuscript to be published.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-244X/8/74/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>We would like to thank Andreas Frischknecht for his contribution to data collection and Nicole Wetli for proof-reading the manuscript.</p>" ]

[]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Predictive validity of the PCL-R – Bivariate logistic regression analyses controlled for time of imprisonment</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"left\">PCL-R</td><td align=\"right\">OR</td><td align=\"right\">95% CI</td><td align=\"right\">AUC</td></tr></thead><tbody><tr><td align=\"left\">PA</td><td align=\"left\">Sum Score</td><td align=\"right\">1.026</td><td align=\"right\">0.963–1.094</td><td align=\"right\">0.613</td></tr><tr><td align=\"left\">VA</td><td/><td align=\"right\">*1.083</td><td align=\"right\">1.011–1.160</td><td align=\"right\">0.704</td></tr><tr><td align=\"left\">PA</td><td align=\"left\">Factor 1</td><td align=\"right\">1.040</td><td align=\"right\">0.911–1.188</td><td align=\"right\">0.610</td></tr><tr><td align=\"left\">VA</td><td/><td align=\"right\">*1.158</td><td align=\"right\">1.007–1.331</td><td align=\"right\">0.686</td></tr><tr><td align=\"left\">PA</td><td align=\"left\">Factor 2</td><td align=\"right\">1.032</td><td align=\"right\">0.920–1.158</td><td align=\"right\">0.614</td></tr><tr><td align=\"left\">VA</td><td/><td align=\"right\">1.075</td><td align=\"right\">0.955–1.210</td><td align=\"right\">0.672</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>Note:</p><p>PA = physical aggression</p><p>VA = verbal aggression, threats</p><p>AUC = Area under the curve. SE = Standard error. CI = Confidence interval. OR = Odds Ratio</p><p>*p &lt; .05.</p></table-wrap-foot>" ]

[]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Secreted frizzled related proteins (SFRPs) compose a family of soluble factors widely involved in the control of embryonic development and the homeostasis of adult tissues. Members of this family were independently isolated using a variety of approaches and immediately proposed as Wnt signalling inhibitors because of their ability to interfere with Wnt-induced embryonic axis duplication and forebrain development in <italic>Xenopus </italic>[##REF##9118218##1##,##REF##9118219##2##]. Many studies have thereafter confirmed that addition of SFRPs can block Wnt-mediated signalling activation in different experimental paradigms showing possible binding preferences between SFRP and Wnt pairs (reviewed in [##REF##18322270##3##]). Whether SFRP-mediated interference with Wnt signalling activation is the result of a single biochemical interaction between Wnt and SFRPs or instead reflects multiple binding mechanisms among SFRP, Wnt and their Frizzled (Fz) receptors is, however, a still unresolved issue.</p>", "<p>Indeed, SFRP molecules fold in two independent domains: an amino-terminal cysteine-rich domain (CRD) and a carboxy-terminal Netrin-related motif (NTR) [##REF##10660608##4##,##REF##11741940##5##]. The Sfrp_CRDcontains ten cysteines with a pattern of five disulfide bridges identical to that of the extracellular CRD of Fz [##REF##8824257##6##,##REF##9096311##7##]. Due to this structural relationship, it is generally assumed that Sfrp-mediated Wnt signalling inhibition results from the interaction between the ligand and Sfrp_CRD, which has been actually shown to immunoprecipitate with Wnt1 and Wnt2 [##REF##10347172##8##,##REF##9326585##9##]. However, Sfrp_CRDcan also form homo- and heterodimers with the CRD domain of Fz receptors [##REF##10347172##8##,##REF##11452312##10##], suggesting potential alternative mechanisms of action.</p>", "<p>The carboxy-terminal Sfrp_NTRis separated from the Sfrp_CRDby a linker region and is characterized by the presence of several conserved blocks of hydrophobic residues and a pattern of six conserved cysteines. NTR domains with similar features are found in a wide range of otherwise unrelated proteins, including Netrin-1, tissue inhibitors of metallo-proteinases (TIMPs), complement proteins and type I procollagen C-proteinase enhancer proteins (PCOLCEs) [##REF##10452607##11##]. Despite an initial suggestion that the Sfrp_NTRmay interact with Wnt ligands [##REF##10660608##4##], the participation of this domain in SFRP function has not been addressed.</p>", "<p>Here, we have combined biochemical studies, mutational analysis and functional assays in cell culture and medaka-fish embryos to test the functional relevance of the Sfrp_NTRin Wnt signalling modulation. We show that the Sfrp1_NTRmimics the function of the full-length Sfrp1, binds to Wnt ligands and prevents Wnt canonical signalling activation, effects shared by distantly related NTR domains such as that of Netrin-1. In contrast, Sfrp1_CRDfails to interact with Wnt but binds to Fz receptors, possibly explaining the potential that the CRD has to inhibit Wnt signalling. We thus conclude that SFRPs modulate Wnt signalling by interacting with both Wnt ligands and Fz receptors but through different domains of the molecule and propose possible models of SFRP function that may reconcile data available in the literature.</p>" ]

[ "<title>Materials and methods</title>", "<title>Whole-mount <italic>in situ </italic>hybridisation</title>", "<p>Whole-mount <italic>in situ </italic>hybridizations were performed in medaka embryos using digoxigenin- and fluorescein-labelled riboprobes. A minimum of 40 embryos were hybridized for each marker and condition. All embryos shown correspond to Iwamatsu stage 19–20 [##REF##15210170##40##].</p>", "<title>Construct generation</title>", "<p><italic>olSfrp1</italic>, <italic>mWnt8a</italic>, <italic>zWnt5 and zWnt1 </italic>expression constructs have been described [##REF##15210177##13##,##REF##15996547##36##,##REF##10601040##41##,##REF##12676324##42##]. z<italic>Sizzled </italic>was a kind gift of Dr Hibi and x<italic>Sizzled </italic>of Dr E De Robertis. Medaka <italic>Sfrp2 </italic>full length clone corresponds to the expressed sequence tag MF01SSA080C03, kindly provided by Dr. Takeda. <italic>zSfrp3 </italic>and <italic>olNetrin-1 </italic>where cloned by RT-PCR using specific primers. Full length, truncated and chimerical coding sequences of <italic>Sfrp1</italic>, <italic>Sfrp2</italic>, <italic>Sfrp3 </italic>and <italic>Netrin-1 </italic>where cloned by PCR into pCS2+. All chimerical constructs where designed so that the signal peptide of the corresponding protein was fused in frame with the linker region that precedes the NTR domain, ensuring proper secretion of the corresponding peptide (Figure ##FIG##0##1##). Cysteine to serine mutations were introduced into the NTR of both Sfrp1 and Netrin-1 by PCR. Given the structural similarity between serine and cysteine, this substitution is expected to disrupt di-sulphide bridge formation without altering the secondary structure of the peptide. Carboxy-terminal 3xHA tagged constructs of Sfrp1, Sfrp1_CRDand Sfrp1_NTRwere generated with linker oligos. All constructs were fully sequenced to ensure in-frame fusions.</p>", "<title>mRNA and morpholino injections</title>", "<p>pCS2 plasmids were linearised and transcribed <italic>in vitro </italic>using the SP6 Message mMachine kit (Ambion, Austin, TX, USA). The synthesized mRNA was purified and injected into two-cell stage embryos at different concentrations (titration curve: 50–300 ng/μl) and the severity of the induced phenotypes was dose dependent in all the cases. Injection solutions included 30 ng/ml of hGFP mRNA as a lineage tracer. Selected working concentrations correspond to equimolecular amounts of the different <italic>Sfrp </italic>mRNAs (full length, truncated and chimerical) to obtain equivalent protein levels (Tables 1 and 2). Mo studies were performed as previously described [##REF##15210177##13##] using the following tested Mo (Gene Tools, LLC, Philomath, OR, USA) designed against <italic>olSfrp1</italic>: 5'-CTGTGTTT GTAGGAACCTCGACTGG-3'. Mo were injected at the final concentration of 0.3 mM into one blastomere of embryos at the two-cell stage. For co-injection experiments, 60 ng of <italic>Sfrp1 </italic>or 30 ng of <italic>Sfrp1</italic>_{<italic>CRD </italic>}or 35 ng of <italic>Sfrp1</italic>_{<italic>NTR </italic>}mRNAs were used. At least three independent experiments were conducted for each marker and condition.</p>", "<title>Protein expression and immunoprecipitations</title>", "<p>To determine the efficiency of translation of the Sfrp1 and its derivatives, triply-HA tagged constructs were generated (see above) and their respective mRNAs were injected into medaka embryos in equimolecular amounts (<italic>Sfrp1-3HA</italic>, 200 ng/μl; <italic>Sfrp1</italic>_{<italic>CRD</italic>}-<italic>3HA</italic>, 100 ng/μl; and <italic>Sfrp1</italic>_{<italic>NTR</italic>}-<italic>3HA</italic>, 120 ng/μl) together with <italic>GFP </italic>mRNA as a tracer. For each construct, 30 embryos were treated with lysis buffer (150 mM NaCl; 1% NP40; 50 mM Tris pH 8; 10 μg/ml aprotinin; 10 μg/ml leupeptin and 1 mM phenylmethanesulphonylfluoride (PMSF). Lysates were precipitated with a polyclonal anti-HA (Sigma-Aldrich, St Louis, MI, USA) and Protein G-Sepharose for enrichment. The protein complex present in each of the pellets was re-suspended in 2 × SDS sample buffer containing 1 M urea. The proteins were resolved by SDS-PAGE blotted and the membranes probed with a monoclonal anti-HA (Sigma-Aldrich). Proteins from total cell extracts were subjected to SDS-PAGE, blotted and the membranes probed with an anti-GFP antibody (Molecular Probes, Invitrogen, Carlsbad CA, USA) and a secondary anti-rabbit-POD antibody.</p>", "<p>Sub-confluent HEK 293T cells were transiently and separately transfected with constructs encoding chick Wnt8c-HA, chick Sfrp1-myc or Sfrp1_CRD-mycor Sfrp1_NTR-mycin 2% fetal calf serum. After 2 days, the conditioned media were collected and clarified by centrifugation. The amount of protein present in the conditioned media was evaluated by western blot and similar amounts of peptides derived from each Sfrp1-myc present in the conditioned media were mixed with conditioned medium from Wnt8-HA or mock transfected for 2 hours. Sample volumes were adjusted to 600 μl with lysis buffer (as above). Proteins from conditioned media were precipitated with 3 μg of an anti-HA polyclonal antibody (Sigma-Aldrich) and Protein G-Sepharose. After four washes with lysis buffer, the protein complex was subjected to SDS-PAGE, blotted and the membranes probed with a monoclonal anti-myc antibody (9E10) and a secondary anti-mouse-POD antibody. Signal was detected with the Advanced ECL Western blotting detection Kit analysis (GE Healthcare Life Sciences, Pollards Wood, Buckinghamshire, UK). Reverse inmunoprecipitation experiments were performed using similar incubations of conditioned media. Proteins were precipitated with a polyclonal anti-myc antibody (SIGMA). The immunocomplexes were subjected to SDS-PAGE, blotted and the membranes probed with a monoclonal anti-myc antibody (9E10) and a secondary anti-mouse-POD antibody.</p>", "<p>For Fz2 and Fz5 immunoprecipitations, HEK 293T cells were transiently transfected with mouse <italic>Fz2-HA</italic>, chick-<italic>Sfrp1-myc </italic>or <italic>Sfrp1</italic>_{<italic>CRD</italic>-<italic>myc </italic>}or <italic>Sfrp1</italic>_{<italic>NTR</italic>-<italic>myc </italic>}or cotransfected with mouse <italic>Fz5-HA </italic>and chick-<italic>Sfrp1-myc </italic>or <italic>Sfrp1</italic>_{<italic>CRD</italic>-<italic>myc </italic>}or <italic>Sfrp1</italic>_{<italic>NTR</italic>-<italic>myc </italic>}expression constructs. After 2 days, cells were scraped in lysis buffer (as above). Immunoprecipitations were performed as previously described [##REF##16172602##24##].</p>", "<title>Reporter assays</title>", "<p>Dissociated cells from embryonic day (E)5 central retinas were prepared as described [##REF##12724355##29##], seeded in 24-well plates and transfected 3 hours later using the FuGENE HD Transfection Reagent (Roche, Nutley, NJ, USA). In each case the 700 ng/well of total DNA contained 200 ng of a plasmid containing a 4xLef-1 responsive luciferase reporter and 50 ng of pRL-TK (Promega, Madison, WI, USA) together with variable amounts of the effector plasmids or the empty vector. After 24 hours, luciferase activities were determined using a dual-luciferase assay system (Promega). The LEF-1 reporter luciferase activity was normalized with that of the <italic>Renilla </italic>luciferase to account for transfection efficiency. Data were statistically evaluated using the SPSS v15.0 software (SPSS Inc., Chicago, Illinois, USA) applying a one-way ANOVA test plus <italic>post hoc </italic>test (Dunnet test).</p>", "<title>Image acquisition</title>", "<p>Live embryos were visualized at room temperature under a Leica stereomicroscope equipped with a PLANAPO objective. Embryos processed for <italic>in situ </italic>hybridization were fixed with 4% paraformaldehyde (PFA) and equilibrated in 80% glycerol. After removal of the yolk, embryos were mounted and visualized under a Leica microscope. In all cases, images were captured with a Leica digital camera controlled by the Leica software.</p>" ]

[ "<title>Results</title>", "<title>Sfrp1_NTRmimics the effect of the full-length protein in the anterior neural plate</title>", "<p><italic>Sfrp1 </italic>is expressed in the anterior neural plate and is required to establish the prospective eye territory [##REF##11025229##12##,##REF##15210177##13##]. In line with this idea, <italic>Sfrp1 </italic>(Figure ##FIG##0##1##) over-expression in the medaka fish leads to a morphologically evident enlargement of the forebrain, posterior truncations and axial duplications (Figure ##FIG##1##2b##; Table ##TAB##0##1##). These defects correlate with the expansion of the expression domains of telencephalic, optic vesicle and diencephalic markers such as <italic>fgf8</italic>, <italic>rx3 </italic>and <italic>pax6 </italic>(Figure ##FIG##1##2f,J,n##), the alteration of the axial mesoderm marker <italic>foxa2 </italic>(Figure ##FIG##1##2j##) and the loss of the posterior domain of <italic>pax6 </italic>(arrow in Figure ##FIG##1##2n##). To determine whether the NTR domain of Sfrp1 contributed to this effect, we generated expression constructs encoding truncated (Sfrp1_CRD) or chimerical peptides (Sfrp1_NTR; harbouring its own signal peptide to ensure proper secretion; Figure ##FIG##0##1##) that comprised the two independent domains in which the protein has been shown to fold [##REF##11741940##5##]. Notably, injections of equimolar concentrations of <italic>Sfrp1</italic>_{<italic>NTR </italic>}mRNA led to the enlargement of the forebrain and the expansion of anterior markers (Table ##TAB##0##1##; Figure ##FIG##1##2d,h,l,p##), as observed after the over-expression of full-length <italic>Sfrp1</italic>. Although all peptides seemed to be produced at comparable levels (Figure ##FIG##2##3##; see below), higher concentrations of <italic>Sfrp1</italic>_{<italic>NTR </italic>}mRNA were necessary to induce posterior truncations or axial duplications (data not shown), suggesting a differential requirement of Sfrp1_NTRalong the antero-posterior axis. Alternatively, the peptide was less effective than the entire Sfrp1 protein, perhaps due to a difference in maturation and half-life or diffusion range. Another possible explanation is that monomeric Sfrp1_NTRis less effective than the full-length protein, since protein dimerization through the CRD motif has been previously described [##REF##10347172##8##,##REF##11452312##10##].</p>", "<p>Quite surprisingly, over-expression of Sfrp1_CRD, the domain postulated to mediate SFRP-Wnt interactions, did not result in comparable phenotypes (Table ##TAB##0##1##). Instead, <italic>Sfrp1</italic>_{<italic>CRD </italic>}mRNA injected embryos presented a small but appreciable reduction of the forebrain (Figure ##FIG##1##2c##), which was associated with a diminished expression of prosencephalic markers (Figure ##FIG##1##2g,k,o##). Forebrain reduction was more evident at earlier stages of differentiation even with lower doses of mRNA (data not shown), supporting that the Sfrp1_CRDgain-of-function phenotype did not reflect lower levels of peptide expression. Accordingly, Western blot analysis of embryos injected with haemagglutinin (HA)-tagged versions of the peptides indicated that <italic>Sfrp1 </italic>and <italic>Sfrp1</italic>_{<italic>NTR </italic>}mRNA were efficiently translated at comparable levels while the <italic>Sfrp1</italic>_{<italic>CRD </italic>}mRNA was produced in a larger amount, which existed in a monomeric and possibly a dimeric form (Figure ##FIG##2##3##).</p>", "<p>Morpholino (Mo)-based knock-down of <italic>Sfrp1 </italic>expression results in embryos with a reduced eye field associated, in the most affected embryos, with a shortening and widening of the antero-posterior axis [##REF##15210177##13##] (compare Figure ##FIG##3##4b,b'## with Figure ##FIG##3##4a,a'##). Low concentrations of <italic>Sfrp1 </italic>mRNA are sufficient to completely rescue this phenotype in a large part of the embryos [##REF##15210177##13##] (Figure ##FIG##3##4c,c',f##). If <italic>Sfrp1</italic>_{<italic>NTR </italic>}can mimic the effect of the entire molecule, it should also be able to rescue the effects of Mo interference. Supporting this hypothesis, co-injection of Mo-<italic>Sfrp1 </italic>and <italic>Sfrp1</italic>_{<italic>NTR </italic>}mRNA rescued the size of the eye field of the treated embryos with efficiency similar to that of <italic>Sfrp1 </italic>(Figure ##FIG##3##4e,e',f##). In contrast, <italic>Sfrp1</italic>_{<italic>CRD </italic>}mRNA did not counteract the Mo-<italic>Sfrp1 </italic>induced phenotype (Figure ##FIG##3##4d,d',f##) and even appeared to exacerbate it, in line with the over-expression studies.</p>", "<p>Together, these data suggested that the molecular events induced by the two domains of Sfrp1 were probably different in nature. The <italic>Sfrp1</italic>_{<italic>CRD</italic>}-induced phenotype was difficult to explain according to the generally accepted view that this domain binds Wnt ligands and antagonizes their activity. In contrast, the strong anteriorisation observed after <italic>Sfrp1 </italic>and <italic>Sfrp1</italic>_{<italic>NTR </italic>}over-expression could be easily explained as the result of an early and generalized antagonism of the canonical Wnt pathway, since inhibition of this pathway induces similar anteriorised and dorsalised phenotypes in both fish and <italic>Xenopus </italic>embryos [##REF##9118218##1##,##REF##9118219##2##,##REF##15210177##13##].</p>", "<p>To investigate this possibility, we next assayed whether injection of <italic>Sfrp1 </italic>and <italic>Sfrp1</italic>_{<italic>NTR </italic>}could alleviate the phenotypes caused by <italic>Wnt8</italic>-mediated activation of canonical Wnt signalling. As previously shown in other species [##REF##8422982##14##,##REF##7600994##15##], <italic>Wnt8 </italic>over-expression in medaka fish embryos led to a strong reduction of the forebrain associated with loss of the <italic>rx3</italic>-positive optic vesicles (Table ##TAB##1##2##; compare Figure ##FIG##4##5a,e## with Figure ##FIG##4##5b,f##). These anterior defects were similar to those observed after <italic>Sfrp1</italic>_{<italic>CRD </italic>}injections (Figures ##FIG##1##2c,k## and ##FIG##4##5c,g##; Table ##TAB##1##2##) but opposite to those induced by <italic>Sfrp1 </italic>or <italic>Sfrp1</italic>_{<italic>NTR </italic>}over-expression (Figures ##FIG##1##2b,d,j,l## and ##FIG##4##5d,h##; Table ##TAB##1##2##). Upon co-injection, <italic>Wnt8 </italic>and <italic>Sfrp1 </italic>mRNAs appeared to counteract each other's activity, resulting in mildly anteriorised embryos (Figure ##FIG##4##5i,l##; Table ##TAB##1##2##) that, however, still presented partial posterior truncations or axis duplications (Figure ##FIG##4##5i,l##). This suggests that, in the concentration range tested, <italic>Wnt8 </italic>cannot completely counteract <italic>Sfrp1</italic>-induced axial defects. In agreement with our previous observations, <italic>Sfrp1</italic>_{<italic>NTR </italic>}mRNA abrogated the <italic>Wn8</italic>-induced phenotype, restoring almost completely the size of the <italic>rx3 </italic>expression domain (Figure ##FIG##4##5k,n##; Table ##TAB##1##2##; compare to control embryos in Figure ##FIG##4##5a,b##). In contrast, <italic>Sfrp1</italic>_{<italic>CRD</italic>}, rather than counteracting, accentuated the reduction of the forebrain induced by <italic>Wnt8 </italic>(Figure ##FIG##4##5j,m##).</p>", "<p>Altogether, these results challenged the view that the CRD domain of the Sfrp1 protein plays an important role in Wnt antagonism. To exclude the possibility that inadequate folding or destabilization of the Sfrp1_CRDconstruct could mislead this interpretation, we designed an additional construct encoding the CRD and the entire linker region (Sfrp1_CRD2; Figure ##FIG##0##1##) to ensure proper folding of the Sfrp1 CRD domain [##REF##11741940##5##]. Over-expression of this new construct, <italic>Sfrp1</italic>_{<italic>CRD</italic>2}, caused phenotypes similar to those observed upon <italic>Sfrp1</italic>_{<italic>CRD </italic>}injection (Additional file ##SUPPL##0##1##). As an alternative explanation, the behaviour of the Sfrp1_CRDcould reflect a peculiarity of this specific member of the SFRP family. Therefore, the CRD domain of Sfrp3 (Sfrp3_CRD; Figure ##FIG##0##1##), the family member that diverges the most from <italic>Sfrp1 </italic>[##REF##15210177##13##], was also analyzed. Interestingly, over-expression of <italic>Sfrp3</italic>_{<italic>CRD </italic>}had no morphologically evident effects on embryonic development, even at high concentrations (Additional file ##SUPPL##0##1##; Table ##TAB##0##1##) and, in contrast to <italic>Sfrp1</italic>_{<italic>CRD</italic>}, failed to enhance <italic>Wnt8</italic>-induced phenotype (Additional file ##SUPPL##0##1##; Table ##TAB##1##2##).</p>", "<p>As a third possibility, we considered that our results could reflect differential affinities between SFRPs and this particular Wnt ligand [##REF##17462603##16##]. Therefore, co-injection studies were repeated using two different Wnts: Wnt1, another canonical Wnt that, like Wnt8, can induce posteriorisation of the embryos [##REF##15239957##17##], and Wnt5, which is thought to activate preferentially the non-canonical Wnt signalling pathway [##REF##12952939##18##]. As shown in Figure ##FIG##5##6##, injections of <italic>Sfrp1 </italic>and <italic>Sfrp1</italic>_{<italic>NTR </italic>}counteracted the phenotype caused by Wnt1-induced phenotype with efficiencies that were very comparable to those observed with Wnt8, while <italic>Sfrp1</italic>_{<italic>CRD </italic>}did not. <italic>Wnt5 </italic>over-expression in fish and <italic>Xenopus </italic>embryos leads to variable phenotypes [##REF##12952939##18##,##REF##8275867##19##], including defects in axial extension and reduction of the optic vesicle size, albeit less dramatic than those observed with <italic>Wnt8 </italic>(Additional file ##SUPPL##0##1##). Co-injection of Wnt5 with <italic>Sfrp1</italic>_{<italic>CRD </italic>}or <italic>Sfrp3</italic>_{<italic>CRD </italic>}did not rescue the <italic>Wnt5</italic>-induced phenotype (Additional file ##SUPPL##0##1##; Table ##TAB##1##2##), thus diminishing the relevance of the Sfrp_CRDas a Wnt ligand antagonist. In contrast, our results suggest a relevant role of Sfrp1_NTRin antagonizing Wnt activity.</p>", "<title>Sfrp1_NTReffects are shared by distantly related NTRs and require intact tertiary structure</title>", "<p>To explore this possibility further, we next investigated whether the relevance of the NTR domain in antagonizing Wnt ligands could be extended to other SFRP family members or even to distantly related NTR domains [##REF##10452607##11##]. According to phylogenetic analysis, the SFRP family is composed of three subfamilies: <italic>Sfrp1/2/5</italic>, <italic>Tlc/Sizzled </italic>and the very divergent <italic>Sfrp3/4 </italic>[##REF##15210177##13##]. We thus compared the activity of <italic>Sfrp1 </italic>and <italic>Sfrp1</italic>_{<italic>NTR </italic>}with equivalent constructs from <italic>Sfrp2 </italic>and <italic>Sfrp3 </italic>(Figure ##FIG##0##1##), close and a divergent members of the SFRP family, respectively. Furthermore, we also chose to analyze the NTR domain of Netrin-1 (Figure ##FIG##0##1##), a secreted protein involved in axon guidance where the NTR domain was first identified [##REF##10452607##11##,##REF##1329863##20##] as a distantly related module. When assayed for their ability to reproduce the <italic>Sfrp1 </italic>over-expression phenotype (Figures ##FIG##1##2b## and ##FIG##6##7b##), <italic>Sfrp2 </italic>and <italic>Sfrp2</italic>_{<italic>NTR </italic>}displayed a significant anteriorising activity almost identical to that of <italic>Sfrp1 </italic>and <italic>Sfrp1</italic>_{<italic>NTR</italic>}, respectively (Figure ##FIG##6##7c,f##; Table ##TAB##0##1##), while <italic>Sfrp3 </italic>and <italic>Sfrp3</italic>_{<italic>NTR </italic>}had a much weaker activity and expansion of anterior markers was only observed upon injection of high mRNA concentrations (Figure ##FIG##6##7d(inset),g##; Table ##TAB##0##1##). Intriguingly, <italic>Netrin-1</italic>_{<italic>NTR </italic>}mRNA injections led to a mild expansion of the forebrain at lower frequency than those of <italic>Sfrp1</italic>_{<italic>NTR </italic>}(Figure ##FIG##6##7i##; Table ##TAB##0##1##). These results indicate that, despite the evolutionary distance, this module can mimic SFRP function, presumably by binding to endogenous Wnt.</p>", "<p>We next asked whether the tertiary structure of Sfrp1_NTRwas important for its function. The NTR motif is, in general, poorly conserved and mainly defined by the presence of six conserved cysteine residues that form three disulfide bonds [##REF##11741940##5##,##REF##10452607##11##]. Mutations of the first two of these residues (Cys177 and Cys180) are predicted to disrupt two disulfide bonds, thus destabilizing the tertiary structure of the NTR domain. Indeed, over-expression of such a mutated construct (Sfrp1_{NTR-C177S;C180S}; Figure ##FIG##0##1##) did not alter medaka embryonic development (Figure ##FIG##6##7h##; Table ##TAB##0##1##), indicating that intact tertiary structure of the NTR motif is required for Sfrp1 activity. Notably, analogous mutations of the first two conserved cysteines of the Netrin-1_NTR(Netrin-1_{NTR-C471S;C475S}; Figure ##FIG##0##1##) also interfered with, but surprisingly not totally abolished, the anteriorising activity of this domain (Figure ##FIG##6##7j##; Table ##TAB##0##1##).</p>", "<p>Altogether, these data strongly support that the NTR domain has a relevant role in mediating SFRP function and that this role is conserved also in distantly related domains.</p>", "<title>Sfrp1_NTRand Sfrp1_CRDbind to Wnt8 and Frizzled, respectively, antagonizing canonical signalling</title>", "<p>In agreement with our finding that NTR domains of SFRPs are functionally relevant to Wnt signalling modulation, <italic>in vitro </italic>studies of the interaction between Sfrp1 and Wingless have mapped the relevant SFRP binding site to the carboxyl terminus of the protein [##REF##10660608##4##]. To assess whether a similar biochemical interaction between Wnt8 and Sfrp1_NTRcould explain our over-expression experiments in medaka fish embryos, we challenged Wnt8 interaction with the two Sfrp1 domains.</p>", "<p>To mimic the physiological conditions of the extracellular interaction between Wnt and SFRPs, we collected conditioned media derived from HEK 293T cells separately transfected with <italic>Wnt8-HA</italic>, <italic>Sfrp1-myc</italic>, <italic>Sfrp1</italic>_{<italic>NTR</italic>-<italic>myc </italic>}or <italic>Sfrp1</italic>_{<italic>CRD</italic>-<italic>myc</italic>}. The levels of proteins present in the conditioned media were carefully evaluated and equivalent amounts of Wnt8 (Figure ##FIG##7##8a(iii)##) were incubated with comparable quantities of either Sfrp1 or its derivatives (Figure ##FIG##7##8a(ii)##) and used for co-immunoprecipitation assays. Pull-downs with anti-HA IgG revealed that both Sfrp1-myc and Sfrp1_NTR-mycspecifically interacted with Wnt8-HA, while Sfrp1_CRD-mycdid not (Figure ##FIG##7##8a(i)##). Comparable levels of Sfrp1 and its derivatives were pulled down with anti-myc monoclonal antibodies (Figure ##FIG##7##8a(iv)##), minimising the possibility that the lack of Sfrp1_CRD-Wnt interaction might be due to a less efficient immunoprecipitation of the Sfrp1_CRD. Reverse pull-downs with a polyclonal anti-myc antiserum confirmed these results (Additional file ##SUPPL##1##2##).</p>", "<p>To further test the functionality of this interaction in β-catenin-mediated Wnt signalling and to compare it with that of other NTR domains, we performed TCF-luciferase reporter-based assays in embryonic retinal cells, where β-catenin-mediated transcriptional activity is physiologically low [##REF##11025229##12##]. We thus transfected retina cells with <italic>Fz5</italic>, a Wnt β-catenin associated receptor expressed in the anterior neural plate [##REF##11335120##21##] to ensure Wnt8-mediated signalling activation [##REF##9054360##22##]. Fz5 alone or in combination with Sfrp1, Sfrp1_CRDor Sfrp1_NTRdid not modify basal β-catenin activity (Additional file ##SUPPL##2##3##). Instead, co-transfection or addition of Sfrp1, Sfrp1_NTRor Netrin-1_NTRconditioned media strongly inhibited reporter activity induced by Wnt8 and Fz5 over-expression (Figure ##FIG##7##8b##; Additional file ##SUPPL##2##3##). Equivalent amounts of Sfrp3 or Sfrp3_NTRwere less effective (Figure ##FIG##7##8b##), in good agreement with what is observed in medaka fish embryos (Figure ##FIG##6##7##). In apparent contrast with immunoprecipitation experiments, co-transfection of <italic>Sfrp1</italic>_{<italic>CRD </italic>}also resulted in a significant decrease in reporter activity (Figure ##FIG##7##8b##). Notably, co-transfection with <italic>Sizzled </italic>or <italic>Sizzled</italic>_{<italic>CRD</italic>}, a SFRP family member that does not appear to interfere with Wnt signalling [##REF##16413488##23##], had a weaker activity (Additional file ##SUPPL##2##3##).</p>", "<p>Sfrp1 has been shown to form complexes with Fz6 [##REF##10347172##8##] and Fz2 [##REF##16172602##24##], while crystallographic studies have shown that Fz8_CRDand Sfrp3_CRDcan form dimers [##REF##11452312##10##]. It was possible, therefore, that Sfrp1_CRD-mediated inhibition of β-catenin transcriptional activity could result from Sfrp1_CRDbinding to the Fz5 receptor, thus preventing signal activation as previously proposed [##REF##10347172##8##]. To test this possibility, we performed co-immunoprecipitation studies using cell lysates from HEK 293T cells transfected with <italic>Fz5-HA</italic>, <italic>Sfrp1-myc </italic>or its derivatives or co-transfected with <italic>Fz2-HA</italic>, as a positive control [##REF##16172602##24##], and <italic>Sfrp1-myc </italic>or its derivatives. As shown in Figure ##FIG##7##8c##, both Sfrp1 and Sfrp1_CRD, but not Sfrp1_NTR, interacted with Fz5-HA, supporting the possibility that Sfrp1_CRDcould impede Fz5 activation in TCF-luciferase reporter-based assays by competing with Wnt8 for binding to the Fz receptor. A similar interaction was also observed between Fz2-HA and Sfrp1_CRD-mycas well as with the entire protein (Additional file ##SUPPL##1##2##), confirming and extending previous studies [##REF##16172602##24##].</p>" ]

[ "<title>Discussion</title>", "<p>Wnt signalling contributes to the regional specification of the anterior neural plate. Acquisition of diencephalic, eye and telecencephalic identities, however, require a differential contribution from canonical and non-canonical Wnt pathways, which are regulated by different Wnt antagonists, including <italic>Sfrp1 </italic>[##REF##16413771##25##]. Accordingly, Mo-based knock-down of <italic>Sfrp1</italic>, a Wnt antagonist broadly expressed in the anterior neural plate, strongly reduces the eye field size, concomitantly expanding the telencephalic but not the diencephalic or mesencephalic territories in the medaka fish [##REF##15210177##13##]. Conversely, <italic>Sfrp1 </italic>over-expression leads to expansion of the forebrain associated with posterior truncations and axial duplications [##REF##15210177##13##]. Taking advantage of these activities, we have shown here that the NTR domain of Sfrp1 mimics the function of the full-length protein, binds to Wnt8 and antagonizes Wnt-canonical signalling. This activity requires an intact tertiary structure and is shared by the distantly related Netrin-1_NTR. In contrast, the Sfrp1_CRDdoes not mirror the effects of Sfrp1 over-expression but interacts <italic>in vitro </italic>with Fz receptors and antagonizes Wnt8-mediated β-catenin transcriptional activity, indicating that Wnt signalling modulation may involve multiple and differential interactions among Wnt, Fz and SFRPs.</p>", "<p>These are somewhat surprising observations because it is generally accepted that Wnt-SFRP interaction takes place through the CRD domain due to its high degree of conservation with the extracellular portion of the Fz receptors [##REF##10347172##8##,##REF##9326585##9##]. Several studies in fact have provided convincing evidence that, when used in large amounts compared to Wnt protein concentration, SFRPs or their respective Sfrp_CRDcan efficiently block Wnt signalling in different contexts, such as in <italic>Xenopus </italic>axis formation [##REF##9118218##1##,##REF##9326585##9##], neural tube [##REF##16676313##26##], somites [##REF##10654605##27##] and heart formation [##REF##11159911##28##], although a certain specificity among SFRPs has been observed. Furthermore, studies using cell lysates from co-transfected cell lines have shown physical interactions between Wnt1 or Wnt2 and Sfrp3_CRD[##REF##10347172##8##,##REF##9326585##9##].</p>", "<p>In contrast with this view, we have provided evidence in favour of the relevance of the NTR domain in SFRP-Wnt interaction. Although our data suggest that Sfrp_CRDmore likely interacts with Fz receptors, there are several possibilities worth considering as to why we may have failed to observe a clear interaction between Sfrp1_CRDand Wnt. In the simpler scenario, the difference we have observed between the Sfrp1_NTRand Sfrp1_CRDdomains' abilities to mimic the effect of the entire molecule could have been related to a differential translation efficiency of their respective mRNA within the embryos. However, this possibility seems quite unlikely because western blot analysis of embryo lysates injected with equimolar amounts of tagged molecules indicated that the different peptides were produced with similar efficiency and, if any, the Sfrp1_CRDwas expressed at higher levels. Similarly, Sfrp1_CRD-mycwas retrieved at consistently higher levels in the culture medium from transfected cell lines [##REF##12724355##29##] and in primary cultures from retinal cells (unpublished observations). Furthermore, the reduction of the eye field observed after <italic>Sfrp1</italic>_{<italic>CRD </italic>}injections was observed even with low mRNA doses.</p>", "<p>A second possibility may relate to the stoichiometry of the Sfrp_CRD-Wnt interaction. It has been proposed that a dimer of the CRD Fz8 domain binds Wnt8 [##REF##18505162##30##] and dimerisation of the receptor may increase efficiency of signal transduction [##REF##12734397##31##]. If Sfrp1_CRDdimers form and bind Wnt8 more efficiently, it is possible that we may have missed this interaction since we noticed that we mostly immunoprecipitate the monomeric form (Figure ##FIG##7##8a(iv)##). This possibility, however, does not explain why in the reverse inmunoprecipitations (Additional file ##SUPPL##1##2##) the Wnt8-Sfrp1_CRDimmunocomplex was not observed. Similarly, it does not explain why Sfrp1_CRDcannot counteract Wnt1/5/8 function <italic>in vivo</italic>, where both monomers and possible dimers seem to be present in similar amounts (Figure ##FIG##2##3##).</p>", "<p>As a third possibility, failure of the Sfrp_CRDto antagonize Wnt signalling may reflect specificity of binding. Although we have shown that Sfrp_CRDfailed to interact with Wnt8 and did not counteract the effect of Wnt1, Wnt5 and Wnt8 overexpression, we cannot exclude that Sfrp1 might show selectivity of binding through the two domains with Wnts other than those we have tested.</p>", "<p>In agreement with our view of the importance of the Sfrp_NTRdomain in Wnt activity, several studies have provided indirect evidence in favour of the relevance of this domain. In <italic>Drosophila</italic>, the CRD motif of Dfz or Dfz2 is dispensable for Wg signal transduction and Frizzled proteins lacking the CRD can fully rescue the simultaneous loss of different Fz receptors or partially rescue the canonical signalling in <italic>fz/fz2 </italic>double mutants [##REF##15514021##32##]. Furthermore, a carrier function for the CRD has been suggested in studies where the CRD domain of the <italic>Drosophila </italic>fz receptor has been substituted with the structurally distinct Wnt-binding domain or with wingless itself [##REF##16163385##33##]. A recent study, aimed at demonstrating the interaction between Norrin and Fz4, failed to reveal a positive interaction between the CRD domain of all human SFRP family members and Xwnt8, which instead interacts with the CRD domain of Fz4, 5, 7 and 8 (see Figure ##FIG##1##2## in [##REF##17158104##34##]). Furthermore, <italic>in vitro </italic>analysis of the interaction between Sfrp1 and Wingless mapped the relevant SFRP binding site to the carboxyl terminus of the protein [##REF##10660608##4##]. Our biochemical and functional data are in line with this set of data, strongly supporting the proposal that the NTR domain has a relevant role in mediating Sfrp function. This role is conserved also in distantly related domains. Indeed, the NTR of Sfrp1, 2, and 5 shares a quite similar pattern of cysteine spacing, related to that of Netrin-1. Conformational similarities are, therefore, likely to explain why over-expression of Sfrp1_NTR, Sfrp2_NTRand Netrin-1_NTRresults in all cases in forebrain expansion and effective inhibition of Wnt8-induced β-catenin activation. In contrast, Sfrp3_NTRand Sfrp4_NTRdisplay a different cysteine spacing and, thus, a distinct pattern of disulphide bonds [##REF##11741940##5##], supporting that variations in the NTR structural features could underlie the differences in activities observed among the distinct subgroups of the family [##REF##11741940##5##,##REF##17462603##16##], as we have observed with Sfrp3_NTR.</p>", "<p>The crystallographic resolution of the structure of the mouse Sfrp3 and Fz8 CRD domains revealed the potential for the different CRDs to homo- or heterodimerise [##REF##11452312##10##]. This potential has also been demonstrated in biochemical studies where SFRPs and Fzs and/or their CRDs have been shown to form homo- and/or hetero-complexes [##REF##10347172##8##,##REF##16172602##24##,##REF##12734397##31##]. In line with these data, we have demonstrated a physical interaction between Sfrp1_CRDand Fz5 and Fz2. This binding may very well justify the potential of the Sfrp1_CRDto antagonize, albeit with lower efficiency, Wnt8-induced β-catenin activation, as we have observed in our experimental conditions mimicking the physiological extracellular interactions among Fz, Wnt and SFRPs. This interaction also provides a mechanism, based on functional inactivation of the receptor, to explain why, in many studies, addition of high levels of the CRD alone is sufficient to prevent Wnt signalling activation. The reason why, in our studies, <italic>Sfrp1</italic>_{<italic>CRD </italic>}over-expression in medaka fish embryos seems to synergize rather than prevent the effect of Wnt8 over-expression (Figure ##FIG##1##2##) is, however, unclear. As a tempting speculation, Sfrp1_CRDmay have higher affinity for Fz receptors that, like Fz2 [##REF##16480852##35##], are involved in mediating non-canonical signalling, which, in turn, has been shown to antagonize the Wnt canonical pathway during eye field specification [##REF##15996547##36##]. Alternatively, in the embryo, Sfrp1_CRDmay interfere with other cell signalling pathways, as demonstrated for the CRD of Sizzled, a related family member that binds and inhibits Tolloid/BMP1, metalloproteases that normally degrade the BMP inhibitor chordin, thereby promoting BMP signalling [##REF##16413488##23##,##REF##16518392##37##].</p>" ]

[ "<title>Conclusion</title>", "<p>We have provided functional and biochemical evidence that the NTR, but not the CRD, domain of Sfrp1 mimics the function of the entire molecule. These results challenge several reports implying that the CRD domain of SFRPs, due to its homology with the proposed Wnt binding region on Fz receptors, interferes with Wnt signalling by binding and sequestering the ligand [##REF##10347172##8##,##REF##9326585##9##]. These apparent contradictions can, however, be reconciled with two assumptions. First, SFRPs of different subgroups have different biochemical interactions with Wnt ligands. In support of this assumption, plasmon resonance binding studies using Sfrp1, 2, 3, 4 and Wnt3a and Wnt5 have shown that Wnt5 binds preferentially to Sfrp1 and 2, while Wnt3a binds at least two sites in Sfrp1, 2, 4 and one in Sfrp3 [##REF##17462603##16##]. Second, SFRP molecules interact with both Wnt and Fz in multiple ways and these interactions can modulate signal transduction in either a positive or negative manner. In this view, there are several possible mechanisms by which SFRPs can modulate Wnt signalling (Figure ##FIG##8##9##). SFRP could sequester Wnt ligands through the NTR domain, thus acting as antagonists (Figure ##FIG##8##9a##; this study) or act in a dominant-negative manner through the formation of inactive complexes with Fz receptors, preventing signal activation (Figure ##FIG##8##9b##; as proposed previously [##REF##10347172##8##], and this study). Alternatively, SFRPs could favour Wnt-Fz interaction by simultaneously binding to both molecules and, thus, synergizing with signal activation (Figure ##FIG##8##9c##), as reported previously [##REF##10660608##4##]. Finally, in the absence of Wnt ligands, Sfrp_CRD-Fz_CRDheterodimer formation could trigger signal transduction (Figure ##FIG##8##9d##), as proposed previously [##REF##16172602##24##]. Notably, the activation of the Fz receptors by a proposed ligand-antagonist is not unique to SFRP1, as Dickkopf2, which belongs to a different family of Wnt antagonists, can activate Wnt canonical signalling cooperating with at least three different Fzs [##REF##11137016##38##].</p>", "<p>Genetic manipulations selectively eliminating one or the other domain of SFRPs may provide further insights and help resolve the accuracy of these models. Additional studies characterizing the functionally relevant interactions among Sfrp_NTR-Wnt or Sfrp_CRD-Fz pairs are also undoubtedly needed. Interaction with additional components of the Wnt signalling cascade also needs to be addressed. Particularly relevant might be the contributions of proteoglycans, which are known to bind Wnts [##REF##15563523##39##] and may additionally interact with the Sfrp_NTR(PE, unpublished observations). An accurate establishment of SFRP mode of action is indeed particularly important given the growing interest in these molecules raised by the observations that their expression is altered in different type of cancers, bone pathologies, retinal degenerations and hypophosphatemic diseases, pointing to their potential value as therapeutic targets.</p>" ]

[ "<title>Background</title>", "<p>Secreted frizzled related proteins (SFRPs) are multifunctional modulators of Wnt and BMP (Bone Morphogenetic Protein) signalling necessary for the development of most organs and the homeostasis of different adult tissues. SFRPs fold in two independent domains: the cysteine rich domain (Sfrp_CRD) related to the extracellular portion of Frizzled (Fz, Wnt receptors) and the Netrin module (Sfrp_NTR) defined by homologies with molecules such as Netrin-1, inhibitors of metalloproteinases and complement proteins. Due to its structural relationship with Fz, it is believed that Sfrp_CRDinterferes with Wnt signalling by binding and sequestering the ligand. In contrast, the functional relevance of the Sfrp_NTRhas been barely addressed.</p>", "<title>Results</title>", "<p>Here, we combine biochemical studies, mutational analysis and functional assays in cell culture and medaka-fish embryos to show that the Sfrp1_NTRmimics the function of the entire molecule, binds to Wnt8 and antagonizes Wnt canonical signalling. This activity requires intact tertiary structure and is shared by the distantly related Netrin-1_NTR. In contrast, the Sfrp1_CRDcannot mirror the function of the entire molecule <italic>in vivo </italic>but interacts with Fz receptors and antagonizes Wnt8-mediated β-catenin transcriptional activity.</p>", "<title>Conclusion</title>", "<p>On the basis of these results, we propose that SFRP modulation of Wnt signalling may involve multiple and differential interactions among Wnt, Fz and SFRPs.</p>" ]

[ "<title>Abbreviations</title>", "<p>CRD: Cysteine-rich domain; E: Embryonic day; Fz: Frizzled; HA: Haemagglutinin; Mo: Morpholino; NTR: Netrin-related motif; SFRP: Secreted frizzled related protein; BMP: Bone morphogenetic protein: PMSF: Phenylmethylsulphonyl fluoride.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>JLR, PE and PB conceived, designed and discussed the study. JLR constructed most of the plasmids and performed and analysed the overexpression studies. PE constructed part of the plasmids, performed the reporter assays and immnoprecipitation studies and participated in the analysis of <italic>in vivo </italic>studies. JMR performed and analysed the Mo-rescue studies as well as Wnt1/Sfrp1 co-injection studies. JLR and PE wrote a draft of the manuscript. PB wrote the final version of the manuscript, which was approved by all authors.</p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>We are grateful to Drs F Cavodeassi, EM De Robertis, JL Gomez-Skarmetha, CP Heisenberg, M Hibi, and H Takeda for providing us with Wnt1, Xsizzled, Wnt8, Wnt5, zSizzled and Sfrp2 plasmids, respectively. We are also in debt to Drs E Cisneros, JR Martinez-Morales, G Nusspaumer, S Rodríguez de Córdoba, and R Zeller for critical reading of the manuscript and Dr K Heath for editorial assistance. This work was supported by grants from the Spanish MEC (BFU2004-01585 and BFU2007-61774), the Fundación la Caixa (BM04-77-0), the Fundación Mutual Madrileña (2006-0916), and Comunidad Autonoma de Madrid (CAM, P-SAL-0190-2006) to PB.</p>" ]

[ "<fig id=\"F1\" position=\"float\"><label>Figure 1</label><caption><p><bold>Schematic representation of the different constructs used in this study</bold>. Construct organization and generated mutations are indicated in the drawings. Light grey boxes, signal peptide (SP); light blue boxes cysteine e-rich domain (CRD); dark grey boxes, linker (L); yellow boxes, Netrin-related domain (NTR); green boxes, carboxy-terminal end of the protein.</p></caption></fig>", "<fig id=\"F2\" position=\"float\"><label>Figure 2</label><caption><p><italic><bold>Sfrp1</bold></italic>_{<italic><bold>NTR</bold></italic>}, <bold>but not </bold><italic><bold>Sfrp1</bold></italic>_{<italic>CRD</italic>},<bold> mimics the phenotype induced by the over-expression of full-length <italic>Sfrp1</italic>.</bold><bold>(a-p) </bold>All the panels are dorsal views of embryos at stage 19–20 (optic vesicle stage) injected with <italic>GFP </italic>mRNA alone (control) (a,e,i,m) or together with <italic>olSfrp1 </italic>(b,f,j,n), <italic>Sfrp1</italic>_{<italic>CRD </italic>}(c,g,k,o) or <italic>Sfrp1</italic>_{<italic>NTR </italic>}(d,h,l,p) mRNA. Embryos in (i-l) have been processed for double <italic>in situ </italic>hybridization with <italic>rx3 </italic>(red) and <italic>foxA2 </italic>(blue) probes. All other embryos were hybridized for one probe as indicated. Note how anterior markers are dramatically expanded in both the <italic>Sfrp1 </italic>and <italic>Sfrp1</italic>_{<italic>NTR </italic>}injected embryos (arrow in e-h), while over-expression of <italic>Sfrp1</italic>_{<italic>CRD </italic>}leads to a reduction of forebrain structures. <italic>Sfrp1 </italic>injected embryos also display axial duplications (j) and posterior truncations (b,j, arrow in n). See Table 1 for details. Scale bar: 0.1 mm.</p></caption></fig>", "<fig id=\"F3\" position=\"float\"><label>Figure 3</label><caption><p><italic><bold>Sfrp1</bold></italic>, <italic><bold>Sfrp1</bold></italic>_{<italic><bold>CRD </bold></italic>}<bold>and </bold><italic><bold>Sfrp1</bold></italic>_{<italic><bold>NTR</bold></italic>}<bold>mRNAs are translated at comparable levels when overexpressed <italic>in vivo</italic></bold>. Western blot (WB) analysis of lysates from embryos injected with equimolecular amounts of <italic>Sfrp1-3xHA</italic>, <italic>Sfrp1</italic>_{<italic>CRD </italic>}-<italic>3xHA </italic>or <italic>Sfrp1</italic>_{<italic>NTR</italic>-}<italic>3xHA </italic>mRNAs together with <italic>GFP </italic>mRNA as a tracer. Embryos were collected at St26 and their lysates were precipitated with a polyclonal anti-HA and blotted with monoclonal anti-HA. To account for possible variations in the amount of injected mRNA, the expression levels of Sfrp peptides were normalized against those of the co-injected EGFP protein. Note that the normalised density values of the three peptides (NDV) are very similar. Note also that Sfrp_CRDruns as a doublet that may represent monomeric and dimeric forms (arrows) or post-translational modifications. IP, immunoprecipitation.</p></caption></fig>", "<fig id=\"F4\" position=\"float\"><label>Figure 4</label><caption><p><italic><bold>Sfrp1</bold></italic>_{<italic><bold>NTR </bold></italic>}<bold>but not</bold><italic><bold>Sfrp1</bold></italic>_{<italic><bold>CRD</bold></italic>}, <bold>rescues the phenotype induced by knocking-down <italic>Sfrp1</italic></bold>. <bold>(a-e') </bold>All the panels are dorsal (a-e) and lateral (a'-e') views of embryos at stage 19–20 injected with <italic>GFP </italic>mRNA alone (a,a'), <italic>Mo-olSfrp1 </italic>alone (b,b') or co-injected with <italic>Sfrp1 </italic>(c,c'), <italic>Sfrp1</italic>_{<italic>CRD </italic>}(d,d') or <italic>Sfrp1</italic>_{<italic>NTR </italic>}(e,e') mRNAs as indicated in the panels. Embryos were hybridised for <italic>rx3 </italic>(eye field) and <italic>foxA2 </italic>(axial mesoderm) both visualised in blue. Optic vesicles fail to develop in embryos injected with Mo-<italic>Sfrp1</italic>, as judged by the reduction in <italic>rx3 </italic>expression (b,b'). This defect is reverted by the co-injection of <italic>Sfrp1 </italic>and <italic>Sfrp1</italic>_{<italic>NTR </italic>}mRNAs in 50% of the embryos (c,c',e,e',f) but not by that of <italic>Sfrp1</italic>_{<italic>CRD </italic>}(d,d',f) mRNA, where the reduction of the eye field is even more pronounced than that observed with the Mo-<italic>Sfrp1 </italic>alone. Note that <italic>Sfrp1 </italic>mRNA not only rescues the effect of Mo-<italic>Sfrp1 </italic>but also induces a partial over-expression phenotype (compare (c,c') with Figure 2a,j). <bold>(f) </bold>Quantification of the rescue efficiency in the different conditions. Scale bar: 0.2 mm.</p></caption></fig>", "<fig id=\"F5\" position=\"float\"><label>Figure 5</label><caption><p><italic><bold>Sfrp1</bold></italic>_{<italic><bold>NTR</bold></italic>}<bold>rescues the phenotype induced by <italic>Wnt8 </italic>over-expression</bold>. All the panels are dorsal <bold>(a-d; i-k) </bold>or lateral <bold>(e-h; l-n) </bold>views of embryos at stage 19–20 injected with <italic>GFP </italic>mRNA (a,e); <italic>GFP </italic>together with <italic>Wnt8 </italic>(b,f), <italic>Sfrp1</italic>_{<italic>CRD </italic>}(c,g), <italic>Sfrp1</italic>_{<italic>NTR </italic>}(d,h) or <italic>Wnt8 </italic>together with <italic>Sfrp1 </italic>(i,l), <italic>Sfrp1</italic>_{<italic>CRD </italic>}(j,m) or <italic>Sfrp1</italic>_{<italic>NTR </italic>}(k,n) mRNA as indicated. Optic vesicles fail to develop in embryos injected with <italic>Wnt8 </italic>mRNA, as judged by the reduction in <italic>rx3 </italic>expression (b,f). (i-n) This defect is reverted by <italic>Sfrp1 </italic>(i,l) and <italic>Sfrp1</italic>_{<italic>NTR </italic>}(k,n) but not by <italic>Sfrp1</italic>_{<italic>CRD </italic>}(j,m) co-expression. Note that <italic>Wnt8</italic>-induced forebrain reduction is somewhat enhanced in the presence of <italic>Sfrp1</italic>_{<italic>CRD</italic>}. Embryos were processed for double <italic>in situ </italic>hybridization with <italic>rx3 </italic>(red) and <italic>foxA2 </italic>(blue) probes. Arrows and arrowheads (i,l) indicate moderate expansion of anterior tissue and axial duplications induced by <italic>Sfrp1 </italic>over-expression. See Tables 1 and 2 for details. Scale bar: 0.18 mm (a-d,i-k); 0.25 mm (e-h;l-m).</p></caption></fig>", "<fig id=\"F6\" position=\"float\"><label>Figure 6</label><caption><p><italic><bold>Sfrp1</bold></italic>_{<italic><bold>NTR</bold></italic>}<bold>rescue ability is observed also with <italic>Wnt1</italic>, another canonical ligand</bold>. <bold>(a-e) </bold>Dorsal views of embryos at stage 19–20 injected with <italic>GFP </italic>mRNA (a), <italic>Wnt1 </italic>(b), <italic>or Wnt1 </italic>together with <italic>Sfrp1 </italic>(c), <italic>Sfrp1</italic>_{<italic>CRD </italic>}(d), or <italic>Sfrp1</italic>_{<italic>NTR </italic>}(e) mRNAs. <bold>(f-j) </bold>Lateral views of embryos processed for double <italic>in situ </italic>hybridization with <italic>rx3 </italic>(red) and <italic>foxA2 </italic>(blue) probes injected with the same mRNAs, respectively. The phenotype induced by <italic>Wnt1 </italic>mRNA injection is very similar to that observed with <italic>Wnt8</italic>: the optic vesicles fail to develop (b), with a reduction in <italic>rx3 </italic>expression (g). This defect is reverted by <italic>Sfrp1 </italic>(c,h) and <italic>Sfrp1</italic>_{<italic>NTR </italic>}(e,j) but not by <italic>Sfrp1</italic>_{<italic>CRD </italic>}(d,i) co-expression. <bold>(k) </bold>Percentage of embryos showing reduction in the size of the forebrain/optic vesicles upon injection of <italic>Wnt1 </italic>mRNA or together with equimolecular amounts of mRNAs encoding different variants of <italic>Sfrp1</italic>. Scale bar: 0.2 mm.</p></caption></fig>", "<fig id=\"F7\" position=\"float\"><label>Figure 7</label><caption><p><bold>Distantly related NTR domains mimic the activity of <italic>Sfrp1</italic>_{<italic>NTR </italic>}with different efficiencies</bold>. <bold>(a-j) </bold>Brightfield views of embryos injected with different full-length or chimerical mRNAs as indicated. Insets in (c,d,f,g) correspond to embryos processed for double <italic>in situ </italic>hybridization for <italic>rx3 </italic>(red) and <italic>foxA2 </italic>(blue). Note that injections of <italic>Sfrp1 </italic>(b), <italic>Sfrp2 </italic>(c, and inset) lead to similar expansion of anterior structures compared to control embryos (a), while <italic>Sfrp3 </italic>has a very weak anteriorizing effect (d) observed only in 4% of the injected embryos (Table 1; inset in (d) shows an embryo injected with a high dose (500 ng/μl) of <italic>Sfrp3 </italic>mRNA). Similarly, <italic>Sfrp3</italic>_{<italic>NTR </italic>}induces a weak anteriorisation at a low frequency (embryo shown in (g); Table 1), whereas the distantly related NTR motif from Netrin-1 (i) induces an expansion of the forebrain as observed with Sfrp1_NTR. Note that the functionality of the NTR domain depends on an intact tertiary structure, since cysteine to serine mutations in <italic>Sfrp1</italic>_{<italic>NTR</italic>-<italic>C</italic>177<italic>S</italic>;<italic>C</italic>180<italic>S </italic>}and <italic>Netrin-1</italic>_{<italic>NTR</italic>-<italic>C</italic>471<italic>S</italic>;<italic>C</italic>475<italic>S </italic>}constructs (h,j) induce a total or partial loss of the effect. See Table 1. Scale bar: 0.1 mm.</p></caption></fig>", "<fig id=\"F8\" position=\"float\"><label>Figure 8</label><caption><p><bold>Sfrp1_NTRand Sfrp1_CRDbind to Wnt8 and Frizzled-5, respectively, antagonizing canonical signalling</bold>. <bold>(a) </bold>HEK 293T cells were transiently transfected with Wnt8-HA, Sfrp1-myc, Sfrp1_CRD-mycor Sfrp1_NTR-mycexpression constructs. Conditioned media containing similar amount of each of the Sfrp1-myc derivates (ii) were mixed with conditioned media from Wnt8-HA (iii) or from mock transfected cells (Additional file ##SUPPL##1##2##). Proteins from mixed conditioned media were precipitated with a polyclonal anti-HA and blotted with a monoclonal anti-myc (i). In these conditions, both Sfrp1-myc and Sfrp1_NTR-myc(red asterisks) specifically co-immunoprecipitated with Wnt8-HA, while Sfrp1_CRD-mycdid not. Comparable levels of Sfrp1 and its derivatives were immunoprecipitated (iv). Note that Sfrp1_NTR-mycmigrates as a smear due to post-translational glycosylation. Sfrp1_CRD-myclikely suffers similar post-translational modifications and possibly forms dimers (arrow in (ii)) that do not completely dissociate. <bold>(b) </bold>Cells dissociated from E5 embryonic retinas were co-transfected with a reporter plasmid containing 4 × Lef-1 responsive element together with <italic>Wnt8</italic>, <italic>Fz5 (</italic>100 ng) in combination with the PCDNA plasmid alone (200 ng) or containing <italic>Sfrp1</italic>, <italic>Sfrp3</italic>, <italic>Sfrp1</italic>_NTR, <italic>Sfrp3</italic>_NTR, <italic>Netrin-1</italic>_NTRor <italic>Sfrp1</italic>_{<italic>CRD </italic>}constructs as indicated in the graph. <italic>Wnt8</italic>/<italic>Fz5 </italic>co-transfection activated the reporter expression 140-fold. This activation was strongly inhibited by the addition of Sfrp1, Netrin-1_NTR, Sfrp1_NTRor the combination of Sfrp1_NTRand Sfrp1_CRD. Equivalent amounts of Sfrp3, Sfrp3_NTRor Sfrp1_CRDalone were less effective. Data represent means ± standard error from three separate experiments performed in triplicates (*<italic>p </italic>&lt; 0.05; **<italic>p </italic>&lt; 0.01; ***<italic>p </italic>&lt; 0.001; Student's <italic>t</italic>-test). <bold>(c) </bold>HEK 293T cells were transiently co-transfected with plasmids encoding <italic>Sfrp1-myc</italic>, <italic>Sfrp1</italic>_{<italic>CRD</italic>-<italic>myc </italic>}or <italic>Sfrp1</italic>_{<italic>NTR</italic>-<italic>myc </italic>}together with <italic>Fz5-HA </italic>expression vector (a) or PCDNA vector (Additional file ##SUPPL##1##2##). Proteins from cell lysates were precipitated with anti-HA and then blotted with anti-myc antibody. Note that Sfrp1 and Sfrp1_CRD(red asterisks) interacted with Fz5 while the Sfrp1_NTRdid not. IP, immunoprecipitation; WB western blot.</p></caption></fig>", "<fig id=\"F9\" position=\"float\"><label>Figure 9</label><caption><p><bold>SFRP mode of action may rely on multiple interactions with Wnt ligands and/or Frizzled receptors</bold>. Schematic representation of possible mechanisms by which SFRPs could modulate Wnt/Frizzled signalling. <bold>(a) </bold>SFRPs can antagonize Wnt activity by directly binding to the ligand through its Netrin-related domain. <bold>(b) </bold>SFRPs could interact directly with Frizzled receptors through their corresponding CRD motifs and prevent signal transduction. <bold>(c) </bold>Frizzled, Wnt and SFRP molecules could form heterotrimeric complexes, where SFRP could present the Wnt ligand to the Frizzled receptor thanks to the differential interactions of the CRD and NTR domains. <bold>(d) </bold>In the absence of Wnt ligands, SFRPs can directly bind a Frizzled receptor and transduce a signal. See the text for further details.</p></caption></fig>" ]

[ "<table-wrap id=\"T1\" position=\"float\"><label>Table 1</label><caption><p>Anteriorised phenotypes induced by over-expression of different <italic>Sfrp </italic>variants</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><th align=\"left\">Injected mRNA</th><th align=\"center\">Percentage of embryos<break/> showing an enlarged forebrain</th></tr></thead><tbody><tr><td align=\"left\"><italic>Sfrp1 </italic>(200 ng/μl; n = 70)</td><td align=\"center\">91</td></tr><tr><td align=\"left\"><italic>Sfrp1</italic>_{<italic>CRD </italic>}(100 ng/μl; n = 162)</td><td align=\"center\">0 (55)*</td></tr><tr><td align=\"left\"><italic>Sfrp1</italic>_{<italic>CRD</italic>-2}(100 ng/μl; n = 86)</td><td align=\"center\">0 (48)*</td></tr><tr><td align=\"left\"><italic>Sfrp1</italic>_{<italic>NTR </italic>}(120 ng/μl; n = 158)</td><td align=\"center\">65</td></tr><tr><td align=\"left\"><italic>Sfrp1</italic>_{<italic>NTR</italic>-<italic>C</italic>177<italic>S</italic>;<italic>C</italic>180<italic>S </italic>}(120 ng/μl; n = 48)</td><td align=\"center\">13</td></tr><tr><td/><td/></tr><tr><td align=\"left\"><italic>Sfrp2 </italic>(200 ng/μl; n = 62)</td><td align=\"center\">93</td></tr><tr><td align=\"left\"><italic>Sfrp2</italic>_{<italic>NTR </italic>}(120 ng/μl; n = 47)</td><td align=\"center\">47</td></tr><tr><td/><td/></tr><tr><td align=\"left\"><italic>Sfrp3 </italic>(200 ng/μl; n = 51)</td><td align=\"center\">4 (42; n = 40)^†</td></tr><tr><td align=\"left\"><italic>Sfrp3</italic>_{<italic>NTR </italic>}(120 ng/μl; n = 38)</td><td align=\"center\">3 (27; n = 56)^†</td></tr><tr><td align=\"left\"><italic>Sfrp3</italic>_{<italic>CRD </italic>}(100 ng/μl; n = 36)</td><td align=\"center\">0 (0; n = 75)^†</td></tr><tr><td/><td/></tr><tr><td align=\"left\"><italic>Netrin-1</italic>_{<italic>NTR </italic>}(120 ng/μl; n = 61)</td><td align=\"center\">56</td></tr><tr><td align=\"left\"><italic>Netrin-1</italic>_{<italic>NTR</italic>-<italic>C</italic>471<italic>S</italic>;<italic>C</italic>475<italic>S </italic>}(120 ng/μl; n = 40)</td><td align=\"center\">42</td></tr></tbody></table></table-wrap>", "<table-wrap id=\"T2\" position=\"float\"><label>Table 2</label><caption><p>Antagonistic interaction between <italic>Sfrp </italic>variants and <italic>Wnt8/Wnt5</italic></p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><th/><th align=\"center\" colspan=\"2\"><italic>Wnt8 </italic>(50 ng/μl)</th><th align=\"center\" colspan=\"2\"><italic>Wnt5 </italic>(50 ng/μl)</th></tr><tr><th/><th colspan=\"2\"><hr/></th><th colspan=\"2\"><hr/></th></tr><tr><th align=\"left\">Co-injected mRNA</th><th align=\"center\">n</th><th align=\"center\">Percentage of<break/>embryos showing a<break/> reduced forebrain</th><th align=\"center\">n</th><th align=\"center\">Percentage of<break/>embryos showing a<break/> reduced forebrain</th></tr></thead><tbody><tr><td align=\"left\">None (<italic>Wnt8</italic>/<italic>Wnt5 </italic>alone)</td><td align=\"center\">107</td><td align=\"center\">88</td><td align=\"center\">81</td><td align=\"center\">86</td></tr><tr><td align=\"left\"><italic>Sfrp1 </italic>(200 ng/μl)</td><td align=\"center\">81</td><td align=\"center\">0 (30)</td><td align=\"center\">78</td><td align=\"center\">0 (90)</td></tr><tr><td align=\"left\"><italic>Sfrp1</italic>_{<italic>CRD </italic>}(100 ng/μl)</td><td align=\"center\">68</td><td align=\"center\">96</td><td align=\"center\">72</td><td align=\"center\">83</td></tr><tr><td align=\"left\"><italic>Sfrp1</italic>_{<italic>NTR </italic>}(120 ng/μl)</td><td align=\"center\">110</td><td align=\"center\">20 (14)</td><td align=\"center\">90</td><td align=\"center\">7 (60)</td></tr><tr><td align=\"left\"><italic>Sfrp3 </italic>(200 ng/μl)</td><td align=\"center\">96</td><td align=\"center\">87</td><td align=\"center\">98</td><td align=\"center\">69</td></tr><tr><td align=\"left\"><italic>Sfrp3</italic>_{<italic>CRD </italic>}(120 ng/μl)</td><td align=\"center\">117</td><td align=\"center\">92</td><td align=\"center\">81</td><td align=\"center\">93</td></tr><tr><td align=\"left\"><italic>Sfrp3</italic>_{<italic>NTR </italic>}(100 ng/μl)</td><td align=\"center\">94</td><td align=\"center\">60</td><td align=\"center\">89</td><td align=\"center\">72</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional File 1</title><p>SFRP_CRDpeptides cannot rescue the <italic>Wnt8</italic>- or <italic>Wnt5</italic>-induced over-expression phenotype. All the panels are dorsal views of embryos at stage 19–20 (optic vesicle stage) injected with <italic>GFP </italic>mRNA alone or combined with <italic>Wnt8</italic>, <italic>Wnt5</italic>, <italic>Sfrp1</italic>_{<italic>CRD</italic>-2}, or <italic>Sfrp3</italic>_{<italic>CRD </italic>}mRNA as indicated. Note that <italic>Sfrp1</italic>_{<italic>CRD</italic>-2}(b) behaves as <italic>Sfrp1</italic>_{<italic>CRD </italic>}(Figure ##FIG##0##1c##) in over-expression assays, while <italic>Sfrp3</italic>_{<italic>CRD </italic>}has no evident effect even at high concentrations (c; 300 ng/μl). Consistently, neither <italic>Sfrp1</italic>_{<italic>CRD </italic>}nor <italic>Sfrp3</italic>_{<italic>CRD </italic>}can rescue the phenotype induced upon <italic>Wnt8 </italic>(_{<italic>D</italic>-<italic>F</italic>}) or <italic>Wnt5 </italic>(_{<italic>G</italic>-<italic>I</italic>}) over-expression. See Tables ##TAB##0##1## and ##TAB##1##2## for details. Scale bar: 0.1 mm.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S2\"><caption><title>Additional File 2</title><p>Wnt8 binds to Sfrp1 and Sfrp1_NTRwhile Sfrp1_CRDbinds to Frizzled 2. <bold>(a) </bold>Mixed conditioned media used in Figure ##FIG##7##8a## (see legend) were precipitated with a polyclonal anti-myc and blotted with a monoclonal anti-HA (upper panel). Controls for inputs (middle and lower panels) are the same as those described in Figure ##FIG##7##8a(ii and iii)##. Wnt8-HA co-immunoprecipitated with both Sfrp1-myc and Sfrp1_NTR-mycwhile Sfrp1_CRD-mycdid not. <bold>(b) </bold>HEK 293T cells were transiently co-transfected with <italic>Fz2-HA </italic>constructs together with <italic>Sfrp1</italic>-_{<italic>myc</italic>}, <italic>Sfrp1</italic>_{<italic>CRD</italic>-<italic>myc </italic>}or <italic>Sfrp1</italic>_{NTR-<italic>myc</italic>}. Proteins from cell lysates were precipitated with anti-HA and then blotted with anti-myc antibody. Note that Sfrp1 and Sfrp1_CRD(red asterisks) interact with Fz2 while the Sfrp1_NTRdoes not. <bold>(c) </bold>Conditioned media from mock transfected cells were mixed with Sfrp1-myc, Sfrp1_NTR-mycor Sfrp1_CRD-mycconditioned media (as above). Addition of anti-HA polyclonal antibodies did not cause unspecific immunoprecipitations as revealed by western blotting with anti-Myc monoclonal antibody. <bold>(d) </bold>Addition of anti-HA polyclonal antibodies did not cause unspecific immunoprecipitations in cell lysates from mock and Sfrp1-myc, Sfrp1_NTR-mycor Sfrp1_CRD-mycco-transfected cells as revealed by western blots with anti-Myc antibody.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S3\"><caption><title>Additional File 3</title><p>Wnt8/Fz5 mediated activation of β-catenin transcriptional activity in dissociated embryonic retinal cells is inhibited by soluble Sfrp1 and Sfrp1_NTRas well as by the Sfrp1_CRD. <bold>(a) </bold>E5 embryonic chick retinal cells were dissociated and co-transfected with a reporter plasmid containing 4xLef-1 responsive element, the control plasmid pRLTK and the effector plasmids for each condition. In retinal cells, endogenous β-catenin transcriptional activity is low and barely modified by transfection of <italic>Fz5 </italic>alone or by the co-transfection of <italic>Fz5 </italic>with <italic>Sfrp1</italic>, <italic>Sfrp1</italic>_{<italic>CRD </italic>}or <italic>Sfrp1</italic>_{<italic>NTR</italic>}. In contrast, strong reporter activation is observed upon <italic>Fz5 </italic>and <italic>Wnt8 </italic>co-transfection. <bold>(b) </bold>HEK 293T cells grown in 2% fetal calf serum were transfected with <italic>Sfrp1-myc</italic>, <italic>Sfrp1</italic>_{<italic>CRD</italic>-<italic>myc</italic>}, or <italic>Sfrp1</italic>_{<italic>NTR</italic>-<italic>myc</italic>}. Two days later the conditioned media were collected and similar amounts of proteins were added to dissociated retinal cell cultures co-transfected with a reporter plasmid (as above), <italic>pRLTK</italic>, <italic>Wnt8 </italic>and <italic>Fz5</italic>. TCF-luciferase activity was measured after 24 hours of incubation. Note how the conditioned media strongly inhibit reporter activities. Data represent means ± standard error from three separate experiments performed in triplicates. <bold>(c) </bold>Cells dissociated from E5 embryonic retinas were co-transfected with a reporter plasmid containing 4xLef-1 responsive element together with <italic>Wnt8</italic>, <italic>Fz5 </italic>(100 ng) in combination with the PCDNA plasmid alone (200 ng) or containing <italic>Sfrp1</italic>, <italic>zSizzled</italic>, <italic>Sfrp1</italic>_{<italic>CRD </italic>}or <italic>zSizzled</italic>_{<italic>CRD </italic>}as indicated in the graph. <italic>Wnt8</italic>/<italic>Fz5 </italic>co-transfection activated the reporter expression. This activation was significantly inhibited by the addition of Sfrp1 and Sfrp1_CRDwhile with less efficiency by the addition of <italic>zSizzled </italic>or <italic>zSizzled</italic>_{<italic>CRD</italic>}. Similar results were obtained with the <italic>Xenopus </italic>sizzled constructs. *<italic>p </italic>&lt; 0.05; **<italic>p </italic>&lt; 0.01; ***<italic>p </italic>&lt; 0.001).</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><p>Percentage of embryos showing an anteriorised phenotype upon injection of equimolecular amounts of mRNAs encoding different variants of Sfrp or Netrin-1 proteins, as shown in Figures 1 and 3 and Additional file ##SUPPL##1##2##. The anteriorised phenotype was scored by an evident morphological expansion of the prosencephalic tissue at late neurula stages. The percentage in brackets marked with asterisks represent the frequency of embryos showing a reduction in the size of the forebrain (instead of an increase; see text for details). The percentages in brackets marked with a dagger represent the frequency of appearance of the phenotype at higher concentration: 500 ng/μl for <italic>Sfrp3 </italic>and 300 ng/μl for <italic>Sfrp3</italic>_{<italic>NTR </italic>}and <italic>Sfrp3</italic>_{<italic>CRD</italic>}.</p></table-wrap-foot>", "<table-wrap-foot><p>Percentage of embryos showing a size reduction of the forebrain/optic vesicles upon injection of equimolecular amounts of mRNAs encoding <italic>Wnt8 </italic>or <italic>Wnt5 </italic>together with different variants of <italic>Sfrp1 </italic>and <italic>Sfrp3 </italic>mRNAs. Representative embryos are shown in Figure 2 and Additional file ##SUPPL##1##2##. <italic>Wnt8</italic>-induced forebrain reduction is much more severe (optic vesicles are completely absent), than that observed upon <italic>wnt5 </italic>over-expression, where the optic vesicles are, in general, significantly reduced in size but still visible. In the case of <italic>Wnt </italic>and <italic>Sfrp1 </italic>and <italic>Sfrp1</italic>_{<italic>NTR </italic>}co-injections, the number shown in brackets represents the frequency of appearance of the anteriorised phenotype (enlarged forebrain tissue), which is reduced compared to the over-expression of the given <italic>Sfrp </italic>construct alone (Table 1).</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1749-8104-3-19-1\"/>", "<graphic xlink:href=\"1749-8104-3-19-2\"/>", "<graphic xlink:href=\"1749-8104-3-19-3\"/>", "<graphic xlink:href=\"1749-8104-3-19-4\"/>", "<graphic xlink:href=\"1749-8104-3-19-5\"/>", "<graphic xlink:href=\"1749-8104-3-19-6\"/>", "<graphic xlink:href=\"1749-8104-3-19-7\"/>", "<graphic xlink:href=\"1749-8104-3-19-8\"/>", "<graphic xlink:href=\"1749-8104-3-19-9\"/>" ]

[ "<media xlink:href=\"1749-8104-3-19-S1.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1749-8104-3-19-S2.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1749-8104-3-19-S3.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Although different subtypes of motoneurons of invertebrate species use different neurotransmitters to activate muscle [##REF##6145757##1##,##REF##17500093##2##], all vertebrate motoneurons activate muscle via release of acetylcholine (ACh) [##REF##9914246##3##]. Historically vertebrate motoneurons have been considered exclusively cholinergic. However, several recent studies provide evidence that mammalian spinal motoneurons release both ACh and glutamate from collaterals within the spinal cord that synapse with inhibitory interneurons known as Renshaw cells, although ACh is still thought to be the only neurotransmitter that mediates motoneuron activation of skeletal muscle [##REF##15379996##4##, ####REF##15781854##5##, ##REF##15883359##6####15883359##6##]. It is unknown how two distinct neurotransmitters are differentially regulated within these motoneurons. But the importance of appropriate regulation is underscored by a recent study showing that forced expression of neurotransmitters other than ACh in frog motoneurons causes inappropriate expression of non-cholinergic receptors at the neuromuscular junction [##REF##17190810##7##].</p>", "<p>Expression of the correct neurotransmitter is crucial for normal nervous system function, although the mechanisms that establish appropriate neurotransmitter expression are not well understood. Interneurons in the chick spinal cord can be induced to express ACh inappropriately by forced expression of MNR2, Lhx3, or Islet1 transcription factors [##REF##9778248##8##,##REF##12150931##9##]. However, forced expression of these transcription factors causes the interneurons to initiate a program of motoneuron differentiation [##REF##9778248##8##,##REF##12150931##9##] for which ACh is the appropriate neurotransmitter, suggesting that neurotransmitter expression is established by programs that specify cell fate. On the other hand, it is well-known that at least some neural crest-derived neurons of the peripheral nervous system normally change their neurotransmitter phenotypes during development, and that this is regulated by environmental signals [##REF##6108833##10##,##REF##1699321##11##]. These studies show that under some conditions, neurotransmitter expression is altered in response to the environment after cell fate is specified. Consistent with this idea, changing calcium-mediated neural activity can regulate neurotransmitter expression in neurons in culture [##REF##15175743##12##] and in vivo [##REF##17190810##7##] without affecting expression of markers of cell fate specification [##REF##17190810##7##]. Together these studies suggest that regulation of neurotransmitter phenotype is complex and involves both intrinsic factors that regulate differentiation programs as well as responses to environmental signals.</p>", "<p>One environmental signal known to affect neurotransmitter phenotype in motoneurons is Hepatocyte growth factor (HGF; also known as Scatter factor). Axotomy of adult hypoglossal motoneurons leads to a dramatic loss of mRNA and protein of the ACh synthetic enzyme, choline acetyltransferase (ChAT); this loss can be prevented by administration of HGF [##REF##10583502##13##]. HGF has also been shown to stimulate choline acetyltransferase activity in motoneurons <italic>in vitro </italic>[##REF##9030587##14##]. HGF acts through the Met receptor tyrosine kinase [##REF##14685170##15##], which is expressed in motoneurons and has been shown to be important for their development. For example, HGF acts through Met as an axonal attractant and survival factor for some populations of mammalian and avian motoneurons [##REF##9030587##14##,##REF##8982163##16##, ####REF##9247333##17##, ##REF##10725250##18##, ##REF##10627610##19##, ##REF##12106690##20##, ##REF##12533619##21##, ##REF##16884699##22####16884699##22##] and has also been shown to be required to recruit a subpopulation of motoneurons to a specific motor pool [##REF##12948444##23##].</p>", "<p>Experiments carried out using a variety of cell types have shown that activation of Met can initiate intracellular signaling through several different downstream cascades, including mitogen activated protein kinase (MAPK), phosphatidylinositol 3-kinase (PI3K) and p38 and Akt pathways [##REF##16361255##24##,##REF##11134526##25##]. These cascades can act independently or can be stimulated simultaneously, and there can be crosstalk among them [##REF##10576742##26##, ####REF##11042204##27##, ##REF##11259586##28####11259586##28##]. In addition to activation via Met, these intracellular signaling pathways can also be activated by other receptors [##REF##18061162##29##, ####REF##17077290##30##, ##REF##11278435##31##, ##REF##9779982##32##, ##REF##11601879##33####11601879##33##]. To elucidate the roles of these pathways in cellular events, a number of specific pathway inhibitory reagents have been developed, including LY294002, which inhibits the PI3K pathway [##REF##8106507##34##], U0126, which inhibits the MAPK pathway by inhibiting MEK1/2 [##REF##9660836##35##], and SB203580, which inhibits Akt [##REF##10702313##36##] and p38 MAP kinase [##REF##7750577##37##] (Figure ##FIG##0##1##). It is currently unknown which of these signaling cascades is activated by HGF-mediated Met signaling in motoneurons, and whether different cascades affect different aspects of HGF-mediated Met function in these cells.</p>", "<p>In the present study, we have taken advantage of the experimental tractability of embryonic zebrafish to investigate the role of Met in motoneuron differentiation. Development of zebrafish spinal motoneurons has been well-characterized [##REF##12880634##38##]. Zebrafish have two waves of motoneuron differentiation: primary and secondary motoneurons. Primary motoneurons (PMNs) constitute a small set of segmentally reiterated cells generated during gastrulation [##REF##12880634##38##,##REF##3960108##39##]. Each PMN is individually identified based on its morphology and gene expression pattern. Within each spinal hemisegment, CaP has the most caudally located cell body, RoP has the most rostrally located cell body, and MiP has a cell body located between CaP and RoP. Some spinal hemisegments have an additional PMN, called VaP, which is essentially a duplicated CaP that typically dies [##REF##2299397##40##]. PMN axons pioneer nerve pathways followed later by axons of secondary motoneurons (SMNs) [##REF##3746410##41##,##REF##1618146##42##]. SMNs are born later than PMNs and are more numerous [##REF##12880634##38##]. SMNs are born [##REF##4056102##43##] and extend axons [##REF##1618146##42##] over a protracted period of development; several studies suggest that there are distinct subsets of SMNs [##REF##3746409##44##, ####REF##15680372##45##, ##REF##12952942##46##, ##REF##16943272##47####16943272##47##], although this has not yet been studied in detail. In addition to extensive characterization of their development, many recent studies have characterized the physiological properties of zebrafish motoneurons and how motoneurons are driven by interneurons to activate various types of behavior (see [##REF##12880634##38##]). The neurotransmitters of zebrafish spinal interneurons have been extensively characterized. Specific subsets of interneurons have been shown to be glycinergic or glutamatergic [##REF##15515020##48##,##REF##15515025##49##]. In addition, several types of interneurons have been shown to express the neurotransmitter gamma-amino butyric acid (GABA) [##REF##15515020##48##, ####REF##15515025##49##, ##REF##1479075##50####1479075##50##].</p>", "<p>A previous study showed that zebrafish <italic>met </italic>is a specific marker of CaP and VaP [##REF##11395004##51##]. We report here that a few hours later, <italic>met </italic>is also expressed in MiP and RoP, and that even later in development <italic>met </italic>is expressed in at least some, perhaps all, SMNs. We used morpholino antisense oligonucleotides (MOs) to knock down Met function and found that this had distinct effects on SMNs and PMNs. Many SMNs required Met for their differentiation and the SMN population was significantly reduced in <italic>met </italic>MO-injected embryos. In contrast, PMN differentiation appeared normal in <italic>met </italic>MO-injected embryos. However, some PMNs had truncated peripheral axons or developed interneuron-like processes within the spinal cord. In addition, in the absence of Met many PMNs inappropriately expressed GABA. Whether vertebrate motoneurons ever normally express GABA is controversial, and we return to this point in the discussion. To learn whether distinct Met-activated signaling cascades are responsible for the different phenotypes we observed following Met knock down, we used inhibitors that affect different pathways downstream of Met activation. Our results suggest that the p38 and/or Akt cascade may be required for SMN differentiation, whereas the MEK1/2 cascade may be required for appropriate neurotransmitter expression and to prevent formation of interneuron-like axons in PMNs. Together our results suggest that Met acts through different pathways to affect different aspects of motoneuron development.</p>" ]

[ "<title>Materials and Methods</title>", "<title>Animal husbandry and lines</title>", "<p>Zebrafish embryos were obtained from natural spawning of AB or AB/TU wild-type or <italic>mn2Et </italic>(also referred to as <italic>parg</italic>^{<italic>mn2Et</italic>}) [##REF##15347431##52##], <italic>Tg(gata2:GFP) </italic>[##REF##9177206##53##] and <italic>Tg(pax2a:GFP) </italic>[##REF##12070097##54##] transgenic lines. Fish were staged by hours post-fertilization at 28.5°C (hpf) [##REF##8589427##55##].</p>", "<title>Cloning of zebrafish chat and met</title>", "<p>We amplified two fragments, 1,400 bp and 900 bp from cDNA from a mixture of 24 hpf and 48 hpf embryos; the 1,400 bp fragment was amplified using primers CAT3 and CAT6 and the 900 bp fragment was amplified using primers CAT5 and CAT6. The fragments were cloned into TOPO-TA vector, sequenced and tested for specific expression patterns. Primer sequences were:</p>", "<p>CAT3, 5'-ACAGGTTAGCACTACCTGTC-3';</p>", "<p>CAT5, 5'-CTGAATGACAGCAACAGACG-3';</p>", "<p>CAT6: 5'-TGGTCCGTCTGAGGATTGTAG-3'.</p>", "<p>We cloned a 1.1 kb fragment of zebrafish <italic>met </italic>from cDNA from a mixture of 24 hpf and 48 hpf embryos using primers zfmetE1-1 and zfmetE1-2. The fragment was cloned into PCRII-TOPO and verified by sequence and expression pattern. Primer sequences were:</p>", "<p>zfmetE1-1, 5'-ATGTGAGGAACCAATAGAAAGC-3';</p>", "<p>zfmetE1-2, 5'-CAGATCCTGGAAAGTGACGG-3'.</p>", "<title>Downregulation of Met</title>", "<p>To knock down Met activity, we used three MOs (GeneTools; Philomath, OR, USA): CM1a and CM2 were designed to block <italic>met </italic>(ENSDART00000104456; NCBI Entrez GeneID 492292) translation; these are the same MOs used by Haines and colleagues [##REF##15342468##56##], thus, we repeated their experiments, showing the absence or reduction of <italic>myoD </italic>RNA expression in fin myoblasts in MO-injected embryos at 48 hpf, to verify that these MOs worked. CM1a was designed to anneal to ATG-containing sequences and CM2 was targeted against the 5' untranslated region of <italic>met</italic>. We also used an additional MO, E6I6, which was designed to block <italic>met </italic>mRNA splicing, leading to a deletion of exon 6 and a truncation of the Met protein. We determined that the E6I6 MO blocked <italic>met </italic>mRNA splicing by RT-PCR (Figure ##FIG##0##1a##). As a control, we used an MO similar to CM1a, but containing a 5 bp mismatch (CM1a-5 mm). About 2–4 nl of MO, diluted in water, were injected into the cytoplasm of one-cell-stage embryos. For most experiments, CM1a and CM2 were injected together, at concentrations at which each MO alone had no effect (CM1a, 0.6 mM; CM2, 0.8 mM); throughout this paper, embryos injected with a combination of CM1a and CM2 are referred to as <italic>met </italic>MO-injected embryos. Injection of E6I6 resulted in essentially the same phenotypes as injection of CM1a, CM2 or a mixture of the two translation blockers. MO sequences and concentrations were:</p>", "<p>CM1a, 5'-ATAGTGAATTGTCATCTTTGTTCCT-3', 0.7–1.0 mM;</p>", "<p>CM2, 5'-CTGTAAAATAAAGACACCTGTCGGA-3', 0.9–1.2 mM;</p>", "<p>E6I6, 5'-GATTTGTGATGACTCTTACCACAAA-3', 0.7–1.0 mM;</p>", "<p>CM1a-5 mm, 5'-ATA<underline>C</underline>T<underline>C</underline>AATT<underline>C</underline>TCAT<underline>G</underline>TTT<underline>C</underline>TTCCT-3', 1.0 mM;</p>", "<p>underlines represent mismatches.</p>", "<p>In some experiments we co-injected mouse <italic>Met </italic>mRNA (<italic>mMet </italic>in pSP64; generous gift of G. Vande Woude, Van Andel Research Institute, Grand Rapids, MI, USA) [##REF##14685170##15##] together with the MOs. As a further control to test whether RNA diluted the MO to a non-effective concentration, we performed a set of control experiments in which we injected a similar amount of <italic>lacZ </italic>mRNA, <italic>mMet </italic>RNA, together with the MOs. Injection of <italic>lacZ </italic>mRNA alone did not affect neuronal development. In contrast, injection of <italic>lacZ </italic>mRNA plus MOs resulted in a phenotype similar to injection of MOs alone, showing that any effects of <italic>mMet </italic>RNA plus MO injections were specific to the <italic>mMet </italic>mRNA.</p>", "<title>Blocking Met downstream effectors with pharmacological inhibitors</title>", "<p>Met activates a number of different signaling pathways [##REF##14685170##15##,##REF##17667909##57##,##REF##11430831##58##] and inhibitors that block these pathways have been previously used to test Met function (see Figure ##FIG##0##1b##) [##REF##16361255##24##]. We used the following inhibitors to test Met function in zebrafish: U0126 (InvivoGen; San Diego, CA, USA), which blocks MEK1 and MEK2 [##REF##9660836##35##], LY294002 (InvivoGen), which blocks PI3K [##REF##8106507##34##], and SB203580 (InvivoGen) which blocks p38 and Akt [##REF##10702313##36##,##REF##7750577##37##]. Because LY294002 had severe effects on overall development, we did not pursue its effects on motoneuron differentiation. For the inhibitor studies, embryos were dechorionated and incubated in embryo medium. Cell permeable inhibitor was added at 16 hpf; embryos remained in the inhibitor solution until further processed at either 26 hpf or 48 hpf. We performed dose-response experiments to determine the optimal inhibitor concentrations to use for experiments. We tested concentrations between 10 and 120 μM. In both cases we found that concentrations below 50 μM had no effect and concentrations above 80 μM were deleterious to embryonic development. Therefore, we used both U0126 and SB203580 at 60 μM.</p>", "<title>RNA <italic>in situ </italic>hybridization and immunohistochemistry</title>", "<p>RNA <italic>in situ </italic>hybridization and immunohistochemistry were carried out according to standard protocols [##REF##7940754##59##].</p>", "<title>RNA <italic>in situ </italic>hybridization</title>", "<p>The following antisense RNA probes were used: <italic>islet2 </italic>(<italic>isl2</italic>) [##REF##8575312##60##], <italic>glutamate decarboxylase 1 </italic>(<italic>gad1</italic>, also known as <italic>gad67 </italic>[##REF##15515020##48##]), <italic>met </italic>(1.1 kb fragment spanning 1–1,165 bp of NCBI sequence AY687384; Entrez GeneID: 492292) and <italic>chat </italic>(900 bp fragment spanning 1,099–1,967 of NCBI sequence XM682602; Entrez GeneID: 559274).</p>", "<title>Immunohistochemistry</title>", "<p>The following primary antibodies were used: JL-8 mouse anti-GFP (Chemicon; Temecula, CA, USA) was used at 1:200; zn1 (University of Oregon) was used at 1:150; znp1 (University of Oregon) was used at 1:750; rabbit anti-GABA (Sigma; St. Louis, MO, USA) was used at 1:1,000; anti-Alcam (Alcam was previously known as DM-GRASP, Neurolin, zn5 antigen, and zn8 antigen; University of Oregon) was used at 1:4,000; F59 [##REF##3943663##61##] was used at 1:10; 4D9 [##REF##2570637##62##] was used at 1:50. Primary antibodies were revealed using secondary antibodies coupled to Alexa Fluor568 (goat anti-rabbit 1:1,000; Invitrogen-Molecular Probes; Eugene, OR, USA); Alexa Fluor488 (goat anti-mouse 1:1,000; Invitrogen-Molecular Probes); Alexa Fluor546 (goat anti-mouse IgG1, 1:1,000; Invitrogen-Molecular Probes); Alexa Fluor488 (goat anti-mouse IgG2a, 1:1,000; Invitrogen-Molecular Probes); Alexa Fluor546 (goat anti-mouse IgG2b, 1:1,000; Invitrogen-Molecular Probes).</p>", "<p>To reveal ACh receptor (AChR) clusters, embryos were fixed at 4°C for 4 h, rinsed in phosphate-buffered saline, incubated in 5 μg/ml α-bungarotoxin (αBTX-546; Invitrogen-Molecular Probes) in incubation buffer [phosphate buffered saline plus 0.1% Tween/20 (PBT) + 1% dimethyl sulfoxide (DMSO) + 5% normal goat serum (NGS)] for 30 minutes at room temperature, rinsed in PBT, and then followed by a regular immunohistochemistry protocol (adapted from [##REF##15136152##63##] with minor modification).</p>", "<p>Embryos were scored and photographed with a Zeiss Axioplan microscope and photographed using a Nikon Coolpix 995 digital camera or imaged using a Zeiss LSM5 confocal microscope.</p>", "<title>Acridine orange staining</title>", "<p>For characterization of cell death, embryos were stained according to Williams and Holder [##REF##10771194##64##], with minor modifications. Briefly, embryos were incubated for 20 minutes in 5 mg/ml Acridine Orange (Sigma) in embryo medium, washed three times for 5 minutes in embryo medium and observed under fluorescence microscopy with an fluoro-iso-thio-cyanate (FITC) filter.</p>", "<title>Behavioral assay</title>", "<p>We used high-speed imaging to monitor trunk movements evoked by touch [##REF##15136152##63##]. In our experiment, the head of the embryo was embedded in agarose in such a way that the trunk was not restricted in its movements. The embryo was stimulated with an insect pin mounted on a micromanipulator (Narishige; East Meadow, NY, USA). The stimulus was repeated 5 times at 1 s intervals and recorded using a high-speed digital video camera (Pixelink; Ottawa, ON, Canada) at 100 frames per second. Movies were analyzed using Quicktime (Apple) and individual frames were assembled for presentation in Adobe Photoshop.</p>" ]

[ "<title>Results</title>", "<title><italic>met </italic>is expressed in a subset of zebrafish spinal motoneurons</title>", "<p><italic>met </italic>expression began in 1–2 cells in the ventral region of each spinal hemisegment at 22 hpf (Figure ##FIG##1##2a##), but within a few hours it was expressed in many more cells (Figure ##FIG##1##2c##). To learn which cells were <italic>met</italic>-positive, we first asked whether <italic>met </italic>is co-expressed with <italic>islet2</italic>, a specific marker of CaP and VaP at 22 hpf [##REF##8575312##60##]. We found, as previously reported [##REF##11395004##51##] that at this stage <italic>met </italic>was specifically expressed in CaP and VaP (Figure ##FIG##1##2b##), but not in other PMNs. However, by 26 hpf, <italic>met </italic>was expressed in all PMNs (Figure ##FIG##1##2c##). Later, by 48 hpf, <italic>met </italic>was detectable in clusters of four to eight cells in the ventral region of each spinal hemisegment (Figure ##FIG##1##2d##). To determine whether these <italic>met</italic>-positive cells were SMNs, we used transgenic embryos that express green fluorescent protein (GFP) under the control of the <italic>gata2 </italic>promoter [<italic>Tg(gata2:GFP)</italic>], which has been shown to be expressed in ventrally-projecting SMNs [##REF##15347431##52##] (Figure ##FIG##1##2e##) and in some interneurons [##REF##9177206##53##], and is routinely used to study SMN development [##REF##15456722##65##,##REF##16943272##47##,##REF##12952942##46##,##REF##17761655##68##]. We also used embryos of the <italic>mn2Et </italic>line that express GFP under the control of the <italic>parg </italic>promoter [##REF##15347431##52##] (Figure ##FIG##1##2f##). GFP has been previously reported to be expressed specifically in CaP in the <italic>mn2Et </italic>line [##REF##15347431##52##]. However, we find that, in this line, GFP is expressed in all zebrafish PMNs (Figure ##FIG##1##2g–j##), and in at least some zebrafish SMNs (Figure ##FIG##1##2j##). At 48 hpf, in <italic>mn2Et </italic>embryos all <italic>met</italic>-positive cells co-expressed GFP, but not all GFP-positive cells expressed <italic>met </italic>(Figure ##FIG##1##2f##) suggesting that <italic>met </italic>is expressed in at least some SMNs. We also found that many, but not all, <italic>met</italic>-positive cells co-expressed <italic>gata2</italic>:<italic>GFP </italic>(Figure ##FIG##1##2e##); cells expressing <italic>met </italic>mRNA but not GFP in <italic>Tg(gata2:GFP) </italic>embryos are probably PMNs and dorsally projecting SMNs that are <italic>gata2:GFP</italic>-negative. Together these results provide evidence that <italic>met </italic>is initially expressed only in CaPs and VaPs, but soon after it is expressed in other PMNs and later it is expressed in SMNs. These results do not allow us to conclude whether or not <italic>met </italic>is ever expressed by dorsally-projecting SMNs. In addition, because SMNs are born over a protracted period of development [##REF##4056102##43##], we also cannot conclusively determine whether <italic>met </italic>is expressed by all ventrally-projecting SMNs or only by a subset of these cells.</p>", "<title>Met is required for normal touch-evoked behavior</title>", "<p>Expression of <italic>met </italic>in PMNs and at least some SMNs prompted us to ask whether Met function is necessary for proper regulation of motoneuron-mediated behaviors. Zebrafish embryos begin to exhibit spontaneous muscle contractions that result in coiling movements shortly after PMN axons first extend out of the spinal cord [##REF##9315900##69##, ####REF##9858263##70##, ##REF##14706693##71####14706693##71##]; spontaneous coiling requires functional PMNs [##REF##14706693##71##]. <italic>met </italic>MO-injected embryos had normal spontaneous coiling, suggesting that some PMNs were present and functional.</p>", "<p>At later stages of development, embryos stop coiling spontaneously and instead respond to touch on the head or tail [##REF##14706693##71##]. Therefore, we compared motility of control and <italic>met </italic>MO-injected embryos at 28–30 hpf, a stage at which control embryos are touch-responsive but no longer coil spontaneously. Control embryos responded to tail touch with a stereotyped bend of the trunk (see [##REF##9858263##70##]); the average time for completing the stereotyped movement was about 50 ms (Figure ##FIG##2##3a,b##). <italic>met </italic>MO-injected embryos had impaired tail touch-evoked motility (Figure ##FIG##2##3a,b##). They moved slower, with an average time for completing the movement of 190 ms, and often had spasmodic movements instead of the smooth bending seen in controls. The time-course of the touch-evoked behavioral response was significantly different between <italic>met </italic>MO-injected and control embryos, but was not significantly different between <italic>met </italic>mismatch MO-injected and control embryos (Figure ##FIG##2##3d##), showing that normal Met function is required for proper tail touch-evoked movements.</p>", "<p>Even though <italic>met </italic>MO-injected embryos responded significantly slower to tail touches than did control embryos, their ability to move suggested that muscle function was normal and that neuromuscular junctions were present. We verified this by labeling 26 hpf control and <italic>met </italic>MO-injected embryos with zn1 or znp1 antibodies that recognize zebrafish PMNs [##REF##9315900##69##,##REF##2344406##72##] and antibodies that recognize specific muscle cell types, including anti-Engrailed that recognizes a specific subset of slow muscle fibers [##REF##2570637##62##,##REF##1682127##73##] (Figure ##FIG##3##4a,b##), and F59, that recognizes fast muscle fibers (Figure ##FIG##3##4c,d##) [##REF##8951054##74##]. PMNs were present in <italic>met </italic>MO-injected embryos, although in some cases CaP axons were truncated (see below). Engrailed and F59 labeling were both present in <italic>met </italic>MO-injected embryos and appeared similar to controls. We also examined the localization of AChR clusters using αBTX [##REF##1315852##75##]. AChR clusters were present in <italic>met </italic>MO-injected embryos and had a similar distribution as in controls (Figure ##FIG##3##4e,f##). Together these observations raise the possibility that Met function is not required for formation of muscles or neuromuscular junctions and that the impaired tail touch-evoked motility of <italic>met </italic>MO-injected embryos resulted from a requirement for Met for normal differentiation of PMNs, SMNs or both.</p>", "<title>Met plays a role in formation of ventral motor nerves</title>", "<p>To learn whether Met is required for normal differentiation of PMNs and/or SMNs, we labeled 26 hpf and 48 hpf embryos with the znp1 antibody, which reveals primary and secondary motor axons [##REF##10821763##76##]. By 26 hpf, PMN axons had extended both ventrally and dorsally in control embryos (Figure ##FIG##4##5a,a'##). In <italic>met </italic>MO-injected embryos, dorsally projecting MiP axons appeared normal, but some ventrally extending CaP axons were truncated (Figure ##FIG##4##5b,b'##). However, only about 25% of CaP axons were affected by Met knock down (Figure ##FIG##4##5e##).</p>", "<p>By 48 hpf, SMNs had formed both dorsal and ventral motor nerves (Figure ##FIG##4##5c–d'##). In <italic>met </italic>MO-injected embryos the ventral nerves appeared thinner than in control embryos (Figure ##FIG##4##5c,d##). Because znp1 labels both PMNs and SMNs, and in 75% of hemisegments CaP axons had extended normally, this result suggests that at least some SMN ventral axons were truncated or failed to extend when Met was knocked down. In addition, there were some truncated ventral nerves. Previous studies showed that CaP is unnecessary for extension of SMN ventral axons [##REF##1618146##42##]. Thus, the truncated ventral nerves could represent truncated CaP axons in segments in which SMN axons failed to extend. Alternatively, they could represent a combination of truncated CaP and truncated SMN axons. Together these observations suggest that Met is important for normal ventral axon extension by both PMNs and SMNs. We provide further tests of this hypothesis below.</p>", "<title>Normal secondary motoneuron differentiation requires Met signaling</title>", "<p>To learn whether Met was required for normal SMN development, we examined SMNs in <italic>met </italic>MO-injected <italic>Tg(gata2:GFP) </italic>embryos at 48 hpf. We also labeled these embryos with an antibody to Alcam, a cell surface protein expressed on floor plate and transiently on the somata and fasciculated segments of SMN axons [##REF##10102505##77##]. In 48 hpf <italic>Tg(gata2:GFP) </italic>embryos, most of the GFP-positive SMN somata were also Alcam-positive (Figure ##FIG##5##6a,a'##). However, there were many more Alcam-positive somata than GFP-expressing somata, consistent with the observation that <italic>gata2</italic>-driven GFP is expressed in ventrally projecting SMNs but not in dorsally-projecting SMNs [##REF##9177206##53##,##REF##12482713##66##,##REF##17065443##67##]. In addition, <italic>Tg(gata2:GFP) </italic>embryos had some dorsally-located, GFP-expressing somata that were Alcam-negative (Figure ##FIG##5##6a,a'##). These cells could be SMNs that have down-regulated Alcam expression or, alternatively, they might be GFP-expressing interneurons [##REF##9177206##53##]. In contrast to control <italic>Tg(gata2:GFP) </italic>embryos, in <italic>met </italic>MO-injected <italic>Tg(gata2:GFP) </italic>embryos the number of ventrally projecting SMN somata and axons was severely reduced (Figure ##FIG##5##6b,b'##; Table ##TAB##0##1##) and some of the GFP-positive somata projected interneuron-like axons within the spinal cord rather than peripheral axons (Figure ##FIG##5##6c,c'##; Table ##TAB##0##1##). These cells might be interneurons that normally express GFP in this transgenic line. Alternatively, they could be SMNs that have developed as interneurons. Consistent with the latter possibility, the number of GFP-positive somata just dorsal of the SMN soma domain (more than three cell diameters dorsal of the floor plate) was increased in <italic>met </italic>MO-injected embryos (Table ##TAB##0##1##). It is clear that in <italic>met </italic>MO-injected embryos there are several types of GFP-positive interneurons, because some of them have ascending axons (Figure ##FIG##5##6c##) whereas others have descending axons (Figure ##FIG##5##6c'##). To learn whether Met knock down caused SMNs to die, we labeled embryos with acridine orange to reveal dying cells [##REF##17530925##78##] at 28, 36, and 48 hpf, but saw no significant difference between the spinal cords of <italic>met </italic>MO-injected embryos and controls (data not shown). In addition, the overall cellular structure of the spinal cord appeared entirely normal in <italic>met </italic>MO-injected embryos (data not shown). Together these results suggest that Met is required for formation of at least some SMNs and raise the possibility that when Met is knocked down, at least some SMNs develop as interneurons.</p>", "<title>Met may mediate secondary motoneuron development through activation of p38 and/or Akt</title>", "<p>Activation of the Met receptor can initiate signaling through several different intracellular cascades (Figure ##FIG##0##1##) [##REF##14685170##15##,##REF##17667909##57##,##REF##11430831##58##]. To investigate which of these downstream pathways is involved in SMN differentiation, we treated embryos with compounds designed to inhibit specific intracellular signaling cascades activated by Met and other receptors, and assayed SMN development at 48 hpf. Specifically, we used U0126, which has been shown to inhibit MEK1/2 (Figure ##FIG##0##1##) [##REF##9660836##35##], and SB203580, which has been shown to inhibit Akt and p38 (Figure ##FIG##0##1##) [##REF##8106507##34##,##REF##10702313##36##]. The MEK1/2 inhibitor U0126 had no effect on SMN development (Table ##TAB##1##2## and data not shown), indicating that the MEK1/2 pathway is not involved. In contrast, the Akt and p38 inhibitor SB203580 dramatically affected SMN development (Table ##TAB##1##2##). In fact, SMN development was even more severely affected by blocking p38 and Akt signaling than by knocking down Met function with MOs: inhibitor-treated embryos were missing essentially all SMN somata and axons (Figure ##FIG##5##6d–g##). There is some correctly spliced <italic>met </italic>mRNA following MO injection (Figure ##FIG##0##1##), suggesting that the more severe effect of the inhibitor could at least partially result from a more complete knock down of Met signaling than is achieved in <italic>met </italic>MO-injected embryos. This result also raises the possibility that in addition to Met, other pathways act upstream of p38 and/or Akt during SMN formation. Future experiments to learn the identities of these other pathways will help elucidate the mechanisms required for normal SMN development. In addition, zebrafish has two <italic>p38 </italic>genes, <italic>p38a </italic>and <italic>p38b</italic>, that are broadly expressed at the stages of development we have studied [##REF##16774848##79##], raising the possibility that some effects on SMNs could be cell non-autonomous. Thus, to test whether SB203580 had a general effect on ventral spinal cord neurons, we treated <italic>Tg(pax2a:GFP) </italic>embryos [##REF##12070097##54##] with SB203580 and examined their spinal cords. Spinal cord architecture appeared essentially normal in SB203580-treated <italic>Tg(pax2a:GFP) </italic>embryos (Figure ##FIG##5##6d–g##) suggesting that this inhibitor specifically affected SMNs. Together these results suggest that Met may act through the Akt and/or p38 cascades to promote formation of zebrafish SMNs.</p>", "<p>Surprisingly, although SB203580-treated embryos lacked most SMNs, at 30 hpf their tail touch-evoked motility was essentially indistinguishable from control embryos (data not shown). Only a few SMNs have projected axons by this stage [##REF##1618146##42##]; thus, this result suggests that at this stage most of the touch response is mediated by activity of PMNs, rather than by activity of SMNs, although this has not yet been tested directly. If this is the case, then the movement defects we observed at 30 hpf in <italic>met </italic>MO-injected embryos would have to arise from a requirement for Met in PMN differentiation, rather than from the decrease in SMN number. To test whether this was the case, we examined the effects of Met knock down on PMN development.</p>", "<title>Met prevents CaPs from extending interneuron-like axons within the spinal cord</title>", "<p>Despite the motility defects in <italic>met </italic>MO-injected embryos, three quarters of the CaP axons looked essentially normal as judged by their morphology (Figure ##FIG##4##5##), suggesting that Met might be required for proper development of CaP characteristics other than the peripheral axon. To learn whether Met was required for differentiation of other features of CaP morphology, we examined CaPs in <italic>met </italic>MO-injected <italic>mn2Et </italic>embryos and we labeled embryos with zn1 and znp1 antibodies. Surprisingly, we found that many CaPs had an interneuron-like process within the spinal cord, in addition to the normal peripheral axon (Table ##TAB##2##3##; Figure ##FIG##6##7a–a\"##). To learn whether there was a correlation between CaPs with truncated axons and those with interneuron-like axons, we analyzed morphology of a subset of CaPs (Table ##TAB##3##4##). There did not appear to be a correlation, as we found that both CaPs with normal-appearing peripheral axons and CaPs with truncated peripheral axons had interneuron-like axons. These results suggest that Met is important in maintaining CaP morphology by preventing CaP from forming an interneuron-like central process.</p>", "<title>Met prevents CaPs from expressing an interneuron-specific neurotransmitter</title>", "<p>Based on their morphology, many CaPs in <italic>met </italic>MO-injected embryos appeared to have a hybrid identity that combined features of both motoneurons and interneurons. To learn whether this hybrid identity extended to molecular features, we assayed two aspects of CaP: expression of Islet proteins and neurotransmitter phenotype. We found that CaPs in <italic>met </italic>MO-injected embryos had normal expression of Islet proteins (Figure ##FIG##6##7b,c##), suggesting that they retained motoneuron identity. To learn whether CaPs expressed normal cholinergic properties or expressed interneuron-specific neurotransmitters, we tried several antibodies to ChAT, the ACh synthetic enzyme, but none of them worked in our hands. Because the <italic>chat </italic>sequence is highly conserved within vertebrates, we used the goldfish <italic>chat </italic>sequence [##REF##15951213##80##] to blast zebrafish databases (NCBI) and found a hypothetical sequence with high homology to all vertebrate <italic>chat </italic>sequences. We amplified <italic>chat </italic>from zebrafish cDNA, verified the sequence and found that it was expressed in cells with the correct position and morphology to be PMNs (Figure ##FIG##6##7d##). We confirmed <italic>chat </italic>expression in CaPs using double fluorescent <italic>in situ </italic>hybridization with <italic>islet2 </italic>(Figure ##FIG##6##7e##), and also confirmed that CaPs co-express <italic>met </italic>and <italic>chat </italic>(Figure ##FIG##6##7g##). We analyzed <italic>chat </italic>expression in CaPs in <italic>met </italic>MO-injected embryos at 24 hpf and found that it was unaltered relative to controls (Figure ##FIG##6##7e,f##), showing that this aspect of CaP identity did not depend on Met function.</p>", "<p>We previously showed that in the absence of Islet1, zebrafish PMNs develop interneuron-like axons rather than their normal peripheral projections and that some of these cells express the interneuron-specific neurotransmitter GABA [##REF##16672347##81##]. Therefore, we asked whether CaPs expressed GABA when Met function is knocked down. We found that in embryos injected with either translation blocking or splice blocking <italic>met </italic>MOs, the majority of CaPs and VaPs expressed GABA (Figure ##FIG##7##8a–c\"##) in contrast to control embryos, in which no CaPs or VaPs expressed GABA (Table ##TAB##4##5##). We confirmed this by showing that in contrast to controls, in <italic>met </italic>MO-injected embryos CaPs and VaPs co-expressed GABA and Islet (Figure ##FIG##7##8h,i##). We assayed the time-course of GABA expression and found that it was first visible at about 24 hpf, shortly after the time at which <italic>met </italic>is expressed in PMNs. Because the GABA antibody cross-reacts with an antigen in the muscle (see Figure ##FIG##7##8a,b##), we confirmed PMN expression of GABA using a riboprobe to <italic>gad1</italic>, the synthetic enzyme for GABA [##REF##15515020##48##,##REF##9634146##82##] (Figure ##FIG##7##8d–e\"##). Expression of <italic>gad1 </italic>and GABA in PMNs of <italic>met </italic>MO-injected embryos is in stark contrast to control embryos, in which PMNs never express GABA (Figure ##FIG##7##8a–e\"##; see also [##REF##16672347##81##]). To confirm that expression of GABA in CaPs required Met function, we coinjected <italic>mMet </italic>mRNA with <italic>met </italic>translation blocking MOs (Figure ##FIG##7##8f–f\"##) and found that this abolished GABA expression in CaPs, whereas embryos coinjected with <italic>lacZ </italic>mRNA and <italic>met </italic>MOs expressed GABA in CaPs (Figure ##FIG##7##8g–g\"##). These results show that not only do CaPs lacking Met function have a motoneuron/interneuron hybrid axonal morphology, they also express both motoneuron and interneuron neurotransmitters.</p>", "<title>GABA expression in CaP may be regulated by the MEK1/2 but not by the p38 or Akt pathways</title>", "<p>We used pharmacological inhibitors to learn which Met-activated intracellular signaling pathway(s) transduced the signal required to prevent GABA expression in CaP. Embryos were exposed to inhibitors of these pathways from 16–26 hpf, then fixed and examined for expression of GABA in CaPs (Table ##TAB##4##5##). Exposure to the MEK1/2 signaling inhibitor U0126 resulted in GABA-positive CaPs, similar to results observed in <italic>met </italic>MO-injected embryos (Figure ##FIG##8##9a–c\"##). In addition, some CaPs in U0126-treated embryos had both peripheral and central axons, similar to <italic>met </italic>MO-injected embryos (Table ##TAB##2##3##; Figure ##FIG##8##9e–g##). Together these results suggest that the CaP axonal and neurotransmitter phenotypes seen in the absence of Met function are likely to result from lack of Met activation of MEK1/2. We also treated embryos with SB203580, which inhibits the Akt and p38 pathways and blocked SMN development. In contrast to the effect of U0126, CaPs in embryos exposed to SB203580 had the same neurotransmitter (Figure ##FIG##8##9d–d\"##; Table ##TAB##4##5##) and axonal (Figure ##FIG##8##9h##; Table ##TAB##2##3##) phenotype as control CaPs, suggesting that signaling via p38 and Akt is not involved in regulating CaP neurotransmitter and axon phenotype.</p>" ]

[ "<title>Discussion</title>", "<p>We report two key findings about the role of the Met receptor tyrosine kinase in motoneuron development. First, Met is required for formation of some zebrafish SMNs. Our experiments suggest that this role of Met acts through the p38 and/or Akt signaling cascade. Second, Met is required to prevent CaP motoneurons from co-expressing features of motoneurons and interneurons, including axon pathway and neurotransmitter phenotype. Our experiments suggest that this role of Met acts through the MEK1/2 signaling cascade. We discuss each of these observations in turn.</p>", "<title>Met is necessary for formation of some zebrafish secondary motoneurons</title>", "<p>In chick, mouse, and rat, <italic>Met </italic>is expressed in a subset of spinal motoneurons and the Met ligand HGF is important for the differentiation of these cells [##REF##9030587##14##,##REF##8982163##16##,##REF##9247333##17##,##REF##10627610##19##]. <italic>Met </italic>appears to be expressed primarily in limb-innervating lateral motor column motoneurons and HGF acts as a chemoattractant for the axons of these cells <italic>in vitro </italic>and <italic>in vivo</italic>, as well as promoting their survival through the period of normal programmed cell death when tested <italic>in vitro </italic>[##REF##8982163##16##,##REF##9247333##17##,##REF##10627610##19##].</p>", "<p>In zebrafish, <italic>met </italic>is also expressed in a subset of spinal motoneurons, in this case in all primary and at least some secondary motoneurons of the medial motor column. In the case of SMNs, Met appears to be required for formation of these cells, as their number is significantly reduced when Met activity is knocked down; whether or not this is the case for the HGF-dependent limb-innervating motoneurons of chick, mouse, or rat has not been reported. The decrease in SMN number when Met is knocked down might result from death of SMNs or their progenitors. We did not see increased cell death, suggesting an alternative possibility that SMNs differentiated as interneurons, as we have previously seen in the absence of Islet1 [##REF##16672347##81##] or Nkx6 proteins [##REF##15456722##65##,##REF##17376808##83##]. Consistent with this possibility, at 48 hpf there were more interneuron-like axons within the spinal cord and somata in positions consistent with spinal interneurons in <italic>Tg(gata2:GFP) </italic>embryos injected with <italic>met </italic>MOs than in <italic>Tg(gata2:GFP) </italic>control embryos.</p>", "<p>To begin to learn which intracellular signaling pathways may be involved in Met-mediated SMN formation, we exposed embryos to inhibitors that act on specific pathways downstream of Met activation. Our results suggest that the p38 and/or Akt pathways are required for normal development of SMNs. However, a caveat to this interpretation is that the phenotype was much more severe when these pathways were blocked than when Met was knocked down using MOs. One possible explanation is that Met was incompletely knocked down by our MOs but completely blocked by the pharmacological inhibitor. Alternatively, because these pathways are activated by other receptors in addition to Met, pharmacological blockade may lead to more widespread effects, including effects that are non cell-autonomous (see below). In the future it will be important to learn which other receptors activate these pathways in SMNs and how this is related to activation of these pathways by Met. Finally, the expression of both zebrafish <italic>p38 </italic>genes is widespread throughout early development [##REF##16774848##79##], raising the possibility that p38 activated pathways could have both cell-autonomous and non cell-autonomous effects on SMN development.</p>", "<title>Met prevents primary motoneurons from expressing interneuron-like properties</title>", "<p>We have previously found that knocking down function of several transcription factors expressed in PMNs, and in some cases in their progenitors, results in these cells expressing interneuron-like properties. Thus, in the absence of Islet1, PMNs develop interneuron-like axons within the spinal cord, rather than peripheral axons, and many of these cells express the interneuron neurotransmitter GABA [##REF##16672347##81##]. Similarly, in the absence of Nkx6 transcription factors, MiPs develop a hybrid phenotype in which they have both peripheral axons that innervate muscle and central axons that extend within the spinal cord, although these cells do not express GABA [##REF##17376808##83##]. Here we report that knocking down Met function causes CaPs to express a hybrid phenotype in which many of them have both a peripheral axon innervating muscle and a central axon extending within the spinal cord. In addition, these cells co-express cholinergic and GABAergic properties. Despite expression of GABA, PMNs are still able to activate muscle in the absence of Met function. However, the touch response is significantly slower than in control embryos at developmental stages at which this behavior is likely to be mediated primarily by PMNs. Thus, this slower response is probably a result of impairment in PMN function.</p>", "<p>The ability of PMNs to develop a hybrid phenotype in the absence of Met reveals a degree of plasticity not previously reported for motoneurons. Previous studies showed that the absence of specific transcription factors in motoneuron progenitors [##REF##12150931##9##,##REF##10482234##84##, ####REF##10970877##85##, ##REF##12052910##86##, ##REF##10482235##87##, ##REF##11567614##88##, ##REF##18427115##89####18427115##89##] or in newly post-mitotic motoneurons [##REF##17376808##83##] allows these cells to co-express motoneuron and interneuron properties. These studies reveal the importance of specific transcription factors in preventing motoneurons from developing interneuron properties early in their development. Other studies have shown that postmitotic motoneurons can change their identity from one motoneuron subtype to another in response to environmental cues [##REF##1708527##90##] and that environmental signals can override genetic programs and cause motor axons to extend along aberrant pathways [##REF##10952312##91##]. These studies reveal plasticity in motoneuron subtype specification. In contrast, here we show that motoneurons are able to express interneuron-like properties at late stages of development. Zebrafish <italic>met </italic>is expressed in PMNs about 6–8 hours after they initially extend growth cones and after their axons have extended to and innervated their specific target muscles. Thus, our studies raise the question of whether the ability to develop interneuron characteristics long after their peripheral axons innervate their muscle targets is a general feature of motoneurons. It is well-known that at least some neural crest-derived peripheral neurons have long-lived phenotypic plasticity [##REF##6108833##10##,##REF##1699321##11##], but it is not typically believed that this is the case for central neurons. This issue is particularly important because a recent study has shown that forcing motoneurons to release neurotransmitters other than ACh causes their muscle targets to express receptors to the motoneuron-expressed neurotransmitters [##REF##17190810##7##], potentially leading to inappropriate muscle responses to motoneuron activation.</p>", "<p>The late expression of <italic>met </italic>in PMNs raises the interesting question of what prevents CaPs from extending interneuron-like axons and expressing GABA at stages prior to <italic>met </italic>expression. Although we do not have an answer to this question, we hypothesize that many different factors are required to prevent motoneurons from expressing interneuron-like properties. Consistent with this hypothesis, several transcription factors, including Islet1 [##REF##16672347##81##], Nkx6 [##REF##10970877##85##,##REF##11567614##88##,##REF##15456722##65##,##REF##17376808##83##], Lhx3 [##REF##12150931##9##], Hb9 [##REF##12150931##9##] and AML1/Runx1 [##REF##18427115##89##] have been shown to prevent various types of motoneurons from adopting interneuron-like properties. It is not yet clear, but will be exciting to learn, the identities of the downstream targets of these transcription factors and how they regulate different aspects of interneuron development. We predict that different downstream targets prevent motoneurons from expressing different interneuron properties at different developmental stages. We have proposed that zebrafish PMNs have a high propensity to develop into motoneuron/interneuron hybrids because, as has been postulated from studies in mammals [##REF##12052910##86##], zebrafish motoneurons are closely-related to specific types of interneurons [##REF##17376808##83##]. In zebrafish and chick, both motoneurons and interneurons can arise from a single progenitor [##REF##8149908##92##, ####REF##15539490##93##, ##REF##17442701##94####17442701##94##]; whether this is also the case in mammals is unknown because single progenitor labeling experiments have not been reported.</p>", "<title>GABA expression in developing motoneurons</title>", "<p>Although vertebrate motoneurons are generally considered exclusively cholinergic, several recent studies provide evidence that mammalian spinal motoneurons can release both ACh and glutamate at central synapses on Renshaw cells [##REF##15379996##4##, ####REF##15781854##5##, ##REF##15883359##6####15883359##6##]. However, ACh is still thought to be the only neurotransmitter that mediates motoneuron activation of skeletal muscle [##REF##15379996##4##, ####REF##15781854##5##, ##REF##15883359##6####15883359##6##]. Thus, it is surprising that during early development, frog muscles express not only AChRs at the nascent neuromuscular junction (NMJ) but also several other types of neurotransmitter receptors, including glutamate receptors, glycine receptors and GABA receptors [##REF##17190810##7##]. From experiments in which they altered motoneuronal neurotransmitter expression, Borodinsky and Spitzer [##REF##17190810##7##] have argued that the final complement of receptors at the NMJ results from matching the neurotransmitter released by motoneurons with the receptors on muscle cells. In their studies, they never saw GABA expression by motoneurons under control conditions. However, several earlier studies reported transient expression of GABA in motoneurons in chick, monkey [##REF##2259443##95##] and rat [##REF##1460116##96##]. GABA may act not only as a neurotransmitter, but also as a trophic factor during development [##REF##11545256##97##,##REF##12495613##98##], and it may be important for integrating developing neurons into circuits [##REF##17590449##99##]. These features might explain early transient expression in neurons that do not normally use GABA as a neurotransmitter. However, neither we nor others have reported GABA expression in zebrafish spinal motoneurons, and the issue of whether transient GABA expression is a common feature of vertebrate spinal motoneurons remains unresolved.</p>" ]

[ "<title>Conclusion</title>", "<p>It has been known for many years that environmental signals can alter subtype specification in newly post-mitotic motoneurons [##REF##1708527##90##]. Here we show that motoneurons retain the ability to develop interneuron-like characteristics, including both axon trajectory and neurotransmitter phenotype, long after they have innervated their muscle targets. In zebrafish, motoneurons and some types of interneurons are generated from the same progenitor domain [##REF##8149908##92##, ####REF##15539490##93##, ##REF##17442701##94####17442701##94##], and previous studies showed that in the absence of Notch signaling motoneurons are the preferred fate of cells within that domain [##REF##17442701##94##,##REF##9425133##100##]. Here we suggest that despite this, motoneurons may require continuous signaling to prevent them from developing interneuron-like properties. Our current results also show that motoneurons that co-express interneuron-like properties can still innervate target muscle. In addition, we suggest that the Met receptor tyrosine kinase acts through different intracellular signaling cascades to affect distinct aspects of development in different motoneuron subtypes.</p>" ]

[ "<p>This is an open access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Expression of correct neurotransmitters is crucial for normal nervous system function. How neurotransmitter expression is regulated is not well-understood; however, previous studies provide evidence that both environmental signals and intrinsic differentiation programs are involved. One environmental signal known to regulate neurotransmitter expression in vertebrate motoneurons is Hepatocyte growth factor, which acts through the Met receptor tyrosine kinase and also affects other aspects of motoneuron differentiation, including axonal extension. Here we test the role of Met in development of motoneurons in embryonic zebrafish.</p>", "<title>Results</title>", "<p>We found that <italic>met </italic>is expressed in all early developing, individually identified primary motoneurons and in at least some later developing secondary motoneurons. We used morpholino antisense oligonucleotides to knock down Met function and found that Met has distinct roles in primary and secondary motoneurons. Most secondary motoneurons were absent from <italic>met </italic>morpholino-injected embryos, suggesting that Met is required for their formation. We used chemical inhibitors to test several downstream pathways activated by Met and found that secondary motoneuron development may depend on the p38 and/or Akt pathways. In contrast, primary motoneurons were present in <italic>met </italic>morpholino-injected embryos. However, a significant fraction of them had truncated axons. Surprisingly, some CaPs in <italic>met </italic>morpholino antisense oligonucleotide (MO)-injected embryos developed a hybrid morphology in which they had both a peripheral axon innervating muscle and an interneuron-like axon within the spinal cord. In addition, in <italic>met </italic>MO-injected embryos primary motoneurons co-expressed mRNA encoding Choline acetyltransferase, the synthetic enzyme for their normal neurotransmitter, acetylcholine, and mRNA encoding Glutamate decarboxylase 1, the synthetic enzyme for GABA, a neurotransmitter never normally found in these motoneurons, but found in several types of interneurons. Our inhibitor studies suggest that Met function in primary motoneurons may be mediated through the MEK1/2 pathway.</p>", "<title>Conclusion</title>", "<p>We provide evidence that Met is necessary for normal development of zebrafish primary and secondary motoneurons. Despite their many similarities, our results show that these two motoneuron subtypes have different requirements for Met function during development, and raise the possibility that Met may act through different intracellular signaling cascades in primary and secondary motoneurons. Surprisingly, although <italic>met </italic>is not expressed in primary motoneurons until many hours after they have extended axons to and innervated their muscle targets, Met knockdown causes some of these cells to develop a hybrid phenotype in which they co-expressed motoneuron and interneuron neurotransmitters and have both peripheral and central axons.</p>" ]

[ "<title>Abbreviations</title>", "<p>ACh: Acetylcholine; AChR: ACh receptor; áBTX: ábungarotoxin; ChAT: Choline acetyltransferase; GABA: Gamma-amino butyric acid; GFP: Green fluorescent protein; HGF: Hepatocyte growth factor; MAPK: Mitogen activated protein kinase; MO: Morpholino antisense oligonucleotide; NMJ: neuromuscular junction; PI3K: Phosphatidylinositol 3-kinase; PMN: Primary motoneuron; SMN: Secondary motoneuron.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>AT carried out all of the experiments described in this paper and helped draft the manuscript. JSE participated in the conception and design of the study and helped draft the manuscript. All authors read and approved the final manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>We thank Keith Beadle for help with motility assays, G Vande Woude for the mouse <italic>Met </italic>mRNA, Peter Currie for a <italic>met </italic>MO sample and a partial <italic>met </italic>DNA, Shin-Ichi Higashijima and Joe Fetcho for <italic>gad1 </italic>mRNA, Steve Ekker for the <italic>mn2Et </italic>line, Michael Brand for the <italic>Tg(pax2a:GFP) </italic>line, Shuo Lin for the <italic>Tg(gata2:GFP) </italic>line, Chris Doe, Phil Washbourne and Monte Westefield for critical reading of the manuscript, and Amanda Lewis, Joy Murphy, Jacob Lewis and the staff of the UO Zebrafish Facility for animal husbandry. Supported by NIH grants NS23915 and HD22486 and AHA postdoctoral fellowship 0420027Z.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Blocking Met function</bold>. <bold>(a) </bold>RT-PCR showing that the E6I6 MO blocks <italic>met </italic>mRNA splicing. <bold>(b) </bold>Activation of Met can initiate intracellular signaling through several different downstream cascades, including MEK1/2 and PI3K [##REF##16361255##24##,##REF##11134526##25##]. This diagram shows the cascade and where specific inhibitors act. See text for details.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Zebrafish <italic>met </italic>is expressed in developing spinal motoneurons</bold>. All photographs are dorsal to the top and anterior to the left; this is also the case for subsequent figures except where noted. <bold>(a) </bold>A 22 hpf embryo showing <italic>met </italic>RNA expression in one cell of a spinal hemisegment (asterisk). <bold>(b) </bold>A 22 hpf embryo showing co-expression of <italic>met </italic>(green) and <italic>islet2 </italic>(red) in CaP and VaP. <bold>(c) </bold>A 26 hpf embryo showing <italic>met </italic>expression in all four PMNs in this spinal hemisegment. <bold>(d) </bold>A 48 hpf embryo showing <italic>met </italic>expression in eight cells (asterisks) in this hemisegment. <bold>(e) </bold>A 48 hpf <italic>gata2:GFP </italic>transgenic embryo showing GFP expression in ventrally projecting SMNs (green) and <italic>met </italic>expression (red) in a subset of these cells. The axon of the SMN labeled with an asterisk is shown as it projects out of the spinal cord toward its ventral muscle target. The inset to the left shows the same SMN, also marked with an asterisk, in only the red channel, clearly revealing that the SMN expresses <italic>met </italic>RNA. <bold>(f) </bold>A 48 hpf <italic>mn2Et </italic>transgenic embryo showing GFP expression (green) in PMNs and SMNs; <italic>met </italic>(red) is expressed in a subset of these cells. <bold>(g-j) </bold><italic>mn2Et </italic>transgenic embryos at 24 hpf showing GFP expression in motoneurons: posterior segment showing GFP expression in CaP (g); posterior segment showing GFP expression in CaP and VaP (h); a more anterior segment showing GFP expression in CaP and MiP (i); an even more anterior segment showing GFP expression in CaP, VaP, MiP, RoP and several SMNs (j). Note that even as late as 5 days post-fertilization, in <italic>mn2Et </italic>embryos GFP-positive cells in the spinal cord all appear to have peripheral axons and no interneuron-like cells express GFP, suggesting that in the <italic>mn2Et </italic>line GFP is expressed exclusively in motoneurons. Scale bars, 10 μm.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Normal touch-evoked movements require Met function</bold>. Embryos are oriented anterior to the top and viewed from dorsal. <bold>(a) </bold>Control embryo (top panel), head embedded in agarose, responded to touch by bending away from the probe (asterisk); times indicated are milliseconds. A <italic>met </italic>MO-injected embryo (bottom panel) responded to touch significantly more slowly than the control. Data from the control and <italic>met </italic>MO-injected embryo are shown on the same time scale. <bold>(b) </bold>The same embryos as in (a), but for each embryo the entire time-course of the movement is shown. The entire touch response took about 50 ms in the control embryo, but took about 190 ms in the MO-injected embryo, which remained in the coiled position for about 60 ms. Red box indicates maximal bending. <bold>(c) </bold>Black arrows display the time span of one touch response, as shown in (b), for control and <italic>met </italic>MO-injected embryos; the red section indicates the time the embryo stayed in a coiled position. <bold>(d) </bold>Time frame in which control and injected embryos stayed in the coiled position [red in (b,c)]. The differences between control and control MO-injected embryos were not significantly different, but they were both significantly different from <italic>met </italic>MO-injected embryos (<italic>p </italic>&lt; 2.6 × 10^-13, n = 46 touch-evoked responses of 8 control embryos and 32 touch-evoked responses of 8 <italic>met </italic>MO-injected embryos; <italic>p </italic>&lt; 1.07 × 10^-7, n = 22 touch-evoked responses of 4 control MO-injected embryos and 32 touch-evoked responses of 8 <italic>met </italic>MO-injected embryos.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p><bold>Met appears unnecessary for muscle and neuromuscular junction formation</bold>. <bold>(a,b) </bold>Engrailed antibody (Eng, red) labeling showing muscle pioneer cells and znp1 antibody labeling showing motor axons (green). In <italic>met </italic>MO-injected embryos, some CaP axons are truncated (asterisk). <bold>(c,d) </bold>F59 antibody (red) labeling showing fast muscle fibers and znp1 antibody staining showing motor axons (green). F59 labeling appears the same in control (c) and <italic>met </italic>MO-injected (d) embryos, which have some truncated CaPs (asterisk). <bold>(e-f) </bold>αBTX (red) labeling showing AChRs and znp1 antibody labeling showing motor axons (green). The distribution of AChRs appears the same in control (e) and <italic>met </italic>MO-injected embryos (f) that have some truncated CaP axons (asterisks); however, it appears that the number of AChRs may be decreased at the myoseptal varicosity (arrows) by MO injection. For each experiment, 8 spinal hemisegments plus somites were examined in each of 21–33 <italic>met </italic>MO-injected embryos and 8 spinal hemisegments plus somites in each of 15 controls. Scale bar, 20 μm.</p></caption></fig>", "<fig position=\"float\" id=\"F5\"><label>Figure 5</label><caption><p><bold>Met is required for normal CaP axons. <bold>(a-b') </bold>Embryos at 26 hpf showing PMN axons labeled with znp1 antibody</bold>. <italic>met </italic>MO-injected embryos have normal MiP axons [arrow in (b')] compared to controls [arrow in (a')]. However, some CaP axons are truncated in <italic>met </italic>MO-injected embryos [asterisk in (b)] compared to controls (a). <bold>(c-d') </bold>Embryos at 48 hpf showing PMN and SMN axons labeled with znp1 antibody. As at 26 hpf, dorsally projecting axons in <italic>met </italic>MO-injected embryos appear normal [arrow in (d')] compared to controls [arrow in (c')]. However, some ventrally projecting axons are truncated in <italic>met </italic>MO-injected embryos [asterisk in (d)], compared to controls (c). <bold>(e) </bold>Average percentage of truncated axons in control and <italic>met </italic>MO-injected embryos at 26, 36 and 48 hpf. n (26 hpf) = 8 somites in each of 12 embryos; n (36 hpf) = 8 somites in each of 10 embryos; n (48 hpf) = 8 somites in each of 11 embryos. Scale bar, 20 μm.</p></caption></fig>", "<fig position=\"float\" id=\"F6\"><label>Figure 6</label><caption><p><bold>Met is required for secondary motoneuron formation</bold>. <bold>(a-b') </bold><italic>gata2:GFP </italic>(green) transgenic embryos at 48 hpf labeled with antibody to Alcam (red). <italic>gata2:GFP </italic>is expressed in ventrally projecting SMNs, some of which also express Alcam (a); (a') shows only the GFP. <italic>met </italic>MO-injected embryos (b,b') show a severe decrease in SMNs, but Alcam labeling of the floor plate appears normal. A small number of SMNs form and make ventral projections (arrow). (b) Both markers shown; (b') only GFP shown. <bold>(c,c')</bold>. Some of the GFP-positive cells in <italic>met </italic>MO-injected embryos show interneuron-like axons. (c) An interneuron with an ascending axon; (c') an interneuron with a descending axon. The two GFP-positive cells to the right in (c') are SMNs. <bold>(d-g) </bold><italic>gata2:GFP </italic>transgenic embryos (green) at 48 hpf labeled with Alcam antibody (red). Exposure to SB203580 (f) severely decreased the formation of SMNs relative to controls (d), but the overall architecture of the spinal cord appeared normal based on expression of GFP driven by the <italic>pax2a </italic>promoter (e,g). Scale bar, 20 μm in all panels except (c,c'), which is 10 μm.</p></caption></fig>", "<fig position=\"float\" id=\"F7\"><label>Figure 7</label><caption><p><bold>Met is required for aspects of CaP identity</bold>. <bold>(a) </bold>A <italic>met </italic>MO-injected <italic>mn2Et </italic>(green) transgenic embryo labeled with znp1 antibody (red) showing that CaP has both a peripheral axon (arrow) and a central axon (asterisk). <bold>(a') </bold>The same cell showing only GFP. <bold>(a\") </bold>The same cell showing only znp1 labeling. <bold>(b,c) </bold>Islet1/2 antibody (Isl; red) and zn1 plus znp1 antibodies (green) labeling showing that CaPs in <italic>met </italic>MO-injected embryos (c) co-express these markers as they do in controls (b). For each condition we examined eight spinal hemisegments per embryo. n = 56 <italic>met </italic>MO-injected embryos and 32 control embryos. <bold>(d) </bold><italic>chat </italic>is expressed in somata in the normal location of PMNs; gray lines show segment boundaries. <bold>(e,f) </bold>Expression of <italic>chat </italic>(green) and the CaP and VaP-specific marker <italic>islet2 </italic>(red) shows that in <italic>met </italic>MO-injected embryos (f) CaPs express <italic>chat</italic>, as they do in controls (e). For each condition we examined eight hemisegments per embryo. n = 24 <italic>met </italic>MO-injected embryos and 15 control embryos. <bold>(g) </bold>Co-expression of <italic>chat </italic>(red) and <italic>met </italic>(green) in PMNs. Scale bars, 10 μm.</p></caption></fig>", "<fig position=\"float\" id=\"F8\"><label>Figure 8</label><caption><p><bold>Met is required to prevent CaPs from expressing an inappropriate neurotransmitter</bold>. For each condition we examined eight spinal hemisegments per embryo. <bold>(a-c\") </bold>Embryos at 26 hpf labeled with antibodies to GABA (red) and zn1 plus znp1 (green); (a',b',c') only the green channel is shown; (a\",b\",c\") only the red channel of the micrographs shown in (a,b,c) is shown. CaPs expressed GABA in embryos injected with translation blocking (b-b\") or splice blocking (c-c\") <italic>met </italic>MOs, but not in controls (a-a\"). Note that some CaPs in <italic>met </italic>MO-injected embryos also have ectopic axons within the spinal cord (arrows). <bold>(d-e\") </bold>Embryos at 26 hpf labeled with riboprobes to <italic>gad1 </italic>(green) and <italic>islet2 </italic>(red). (d',e') Only the green channel is shown; (d\",e\") only the red channel of the micrographs shown in (d,e) is shown. In <italic>met </italic>MO-injected embryos, <italic>islet2</italic>-positive CaPs also express <italic>gad1 </italic>(e-e\"), whereas these two genes are not co-expressed in CaPs of control embryos (d-d\"). <bold>(f-g\") </bold>Embryos at 26 hpf labeled with antibodies to GABA (red) and zn1 plus znp1 (green). (f',g') Only the green channel is shown; (f\",g\") only the red channel of the micrographs shown in (f,g) is shown. CaPs do not express GABA in embryos co-injected with <italic>met </italic>MO and <italic>mMet </italic>mRNA (f-f\"), but do express GABA in embryos co-injected with <italic>met </italic>MO and <italic>lacZ </italic>mRNA (g-g\"). <bold>(h,i) </bold>Embryos at 26 hpf labeled with antibodies to GABA (red) and Islet (green). CaPs co-express Islet and GABA in <italic>met </italic>MO-injected embryos (i) but not in controls (h). n = 80 embryos in (a); n = 128 embryos in (b); n = 24–32 embryos each for (c,e,f,g); n = 20 embryos for (h); n = 35 embryos for (i). The phenotype was seen in 70% of embryos injected with the splice-inhibitor MO and 80% of embryos injected with translation-blockers; in affected embryos all segments showed the phenotype. Rescue was seen in 90% of embryos injected with <italic>mMet </italic>mRNA and 0% of embryos injected with <italic>lacZ </italic>mRNA. Scale bars, 10 μm.</p></caption></fig>", "<fig position=\"float\" id=\"F9\"><label>Figure 9</label><caption><p><bold>Met signaling may influence CaP axonal and neurotransmitter phenotypes through the MEK1/2 pathway</bold>. All panels show <italic>mn2Et </italic>transgenic embryos. <bold>(a-d\") </bold>Embryos at 26 hpf showing GFP (green) and GABA (red); asterisks indicate CaPs or CaP/VaP pairs. (a',b',c',d') Only the green channel is shown; (a\",b\",c\",d\") only the red channel of the micrographs shown in (a,b,c,d) is shown. CaPs express GABA in U0126-treated embryos (c-c\") and <italic>met </italic>MO-injected embryos (b-b\") but not in controls (a-a\") or in SB203580-treated embryos (d-d\"). <bold>(e-h) </bold>Some CaPs have both peripheral (arrows) and central (asterisks) axons in U0126-treated embryos (g) and <italic>met </italic>MO-injected embryos (f); CaPs have only have peripheral axons in controls (e) and in SB203580-treated embryos (h). Scale bars, 10 μm.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>The number of second motoneurons is reduced following Met knock down</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\">Control</td><td align=\"center\"><italic>met </italic>MO</td><td align=\"center\">SB203580</td></tr></thead><tbody><tr><td align=\"left\">Number of embryos</td><td align=\"center\">18</td><td align=\"center\">20</td><td align=\"center\">16</td></tr><tr><td align=\"left\">Percent of segments with ventral nerve</td><td align=\"center\">98</td><td align=\"center\">61</td><td align=\"center\">39</td></tr><tr><td align=\"left\">GFP⁺somata within 3 cell diameters of floor plate</td><td align=\"center\">20 ± 0.47</td><td align=\"center\">9 ± 0.61</td><td align=\"center\">6 ± 0.40</td></tr><tr><td align=\"left\">GFP⁺somata 3 cell diameters dorsal of floor plate</td><td align=\"center\">5 ± 0.39</td><td align=\"center\">10 ± 0.52</td><td align=\"center\">8 ± 0.68</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>SB203580 affects SMN development</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\">Number of embryos</td><td align=\"center\">Percent of segments<break/>with reduced SMN<break/> somata</td></tr></thead><tbody><tr><td align=\"left\">Control</td><td align=\"center\">75</td><td align=\"center\">0</td></tr><tr><td align=\"left\"><italic>met </italic>MO</td><td align=\"center\">96</td><td align=\"center\">80</td></tr><tr><td align=\"left\">U0126</td><td align=\"center\">20</td><td align=\"center\">0</td></tr><tr><td align=\"left\">SB203580</td><td align=\"center\">28</td><td align=\"center\">90</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Met knockdown results in CaPs with both peripheral and central axons</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\">Number of <break/>embryos</td><td align=\"center\">Number of CaPs with<break/> only a peripheral axon</td><td align=\"center\">Number of CaPs with<break/> peripheral and central axons</td></tr></thead><tbody><tr><td align=\"left\">Control</td><td align=\"center\">48</td><td align=\"center\">139</td><td align=\"center\">5</td></tr><tr><td align=\"left\"><italic>met </italic>MO</td><td align=\"center\">56</td><td align=\"center\">139</td><td align=\"center\">29</td></tr><tr><td align=\"left\">U0126</td><td align=\"center\">45</td><td align=\"center\">117</td><td align=\"center\">18</td></tr><tr><td align=\"left\">SB203580</td><td align=\"center\">27</td><td align=\"center\">81</td><td align=\"center\">0</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Whether CaP has an interneuron-like axon appears uncorrelated with whether it has a truncated peripheral axon</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"2\">Control <break/>23 CaP axons in 6 embryos</td><td align=\"center\" colspan=\"2\"><italic>met</italic><break/>MO 33 CaP axons in 10 embryos</td><td align=\"center\" colspan=\"2\">U0126<break/> 55 CaP axons in 8 embryos</td></tr><tr><td/><td colspan=\"2\"><hr/></td><td colspan=\"2\"><hr/></td><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\">Peripheral axon<break/> extended to</td><td align=\"center\">No interneuron-<break/>like axon</td><td align=\"center\">Interneuron-<break/>like axon</td><td align=\"center\">No interneuron-<break/>like axon</td><td align=\"center\">Interneuron-<break/>like axon</td><td align=\"center\">No interneuron-<break/>like axon</td><td align=\"center\">Interneuron-<break/>like axon</td></tr></thead><tbody><tr><td align=\"left\">Ventral edge of muscle</td><td align=\"center\">22</td><td align=\"center\">1</td><td align=\"center\">17</td><td align=\"center\">8</td><td align=\"center\">36</td><td align=\"center\">10</td></tr><tr><td align=\"left\">Horizontal myoseptum</td><td align=\"center\">1</td><td align=\"center\">0</td><td align=\"center\">1</td><td align=\"center\">1</td><td align=\"center\">5</td><td align=\"center\">2</td></tr><tr><td align=\"left\">Just out of neural tube</td><td align=\"center\">0</td><td align=\"center\">0</td><td align=\"center\">4</td><td align=\"center\">2</td><td align=\"center\">2</td><td align=\"center\">0</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T5\"><label>Table 5</label><caption><p>CaPs express GABA in embryos in which Met is knocked down</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\">Number of<break/> embryos</td><td align=\"center\">Percent of <break/>GABA⁺CaPs</td></tr></thead><tbody><tr><td align=\"left\">Control</td><td align=\"center\">135</td><td align=\"center\">0</td></tr><tr><td align=\"left\"><italic>met </italic>MO</td><td align=\"center\">88</td><td align=\"center\">70</td></tr><tr><td align=\"left\">U0126</td><td align=\"center\">56</td><td align=\"center\">85</td></tr><tr><td align=\"left\">SB203580</td><td align=\"center\">88</td><td align=\"center\">0</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>GFP-positive SMN axons (three segments/embryo) and somata (two segments/embryo) were analyzed in 48 hpf <italic>Tg(gata2:GFP) </italic>embryos. Soma counts of treatment groups are all significantly different from controls (<italic>p </italic>&lt; 0.00015).</p></table-wrap-foot>", "<table-wrap-foot><p>SMN somata were analyzed in three segments of each embryo at 48 hpf. In all cases, the three segments had the same phenotype; in other words, within an embryo all segments had reduced SMN somata or no segments had reduced SMN somata.</p></table-wrap-foot>", "<table-wrap-foot><p>CaPs were analyzed in 26 hpf <italic>mn2Et </italic>embryos; 3 hemisegments were examined per embryo.</p></table-wrap-foot>", "<table-wrap-foot><p>Axons were analyzed in <italic>mn2Et </italic>embryos at 26 hpf.</p></table-wrap-foot>", "<table-wrap-foot><p>CaPs were analyzed in four segments of each embryo. In all cases, CaPs in all four segments had the same GABA phenotype; they were either all GABA⁺or all GABA^-.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1749-8104-3-18-1\"/>", "<graphic xlink:href=\"1749-8104-3-18-2\"/>", "<graphic xlink:href=\"1749-8104-3-18-3\"/>", "<graphic xlink:href=\"1749-8104-3-18-4\"/>", "<graphic xlink:href=\"1749-8104-3-18-5\"/>", "<graphic xlink:href=\"1749-8104-3-18-6\"/>", "<graphic xlink:href=\"1749-8104-3-18-7\"/>", "<graphic xlink:href=\"1749-8104-3-18-8\"/>", "<graphic xlink:href=\"1749-8104-3-18-9\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Improving quality of primary care is a key focus of health policy both nationally and internationally [##REF##12609914##1##,##UREF##0##2##]. Quality improvement can take a number of forms. The approach adopted in the United Kingdom has placed a large focus on the clinical quality of care. Financial incentives in the Quality and Outcomes Framework (QOF) are provided on the basis of achieving certain quality indicators, which at the time of introduction in 2004 included 10 clinical domains of care (76 in total), 56 in organisational areas, four assessing patients' experience, and a number of indicators for additional services [##REF##12609914##1##].</p>", "<p>The emphasis on clinical care within the QOF reflects a professional conceptualisation of quality. Patients consistently report that a key priority (alongside technical competence) is the interpersonal skills of their physician [##REF##9823053##3##, ####REF##7804476##4##, ##REF##7899933##5####7899933##5##]. The importance of interpersonal care is supported by the fact that most patient complaints centre around issues with doctors' manner, attitude or communication skills [##UREF##1##6##, ####UREF##2##7##, ##REF##14744713##8####14744713##8##]. Communication skills are central to effective clinical practice such as diagnosis [##UREF##3##9##], and impact on certain health outcomes [##REF##7728691##10##].</p>", "<p>Given the importance of interpersonal skills, the question arises of how best to rectify any deficiencies. Two main methods of quality improvement have been proposed.</p>", "<p>Feedback of patient-based surveys has been suggested as a cost-effective quality improvement method [##REF##10487981##11##]. Physicians in both the United Kingdom and the United States are currently remunerated to varying degrees to assess the views of their patient populations. However, the effectiveness of the feedback of patient assessments in improving the quality of interpersonal care is unclear. The controlled trial literature on audit and feedback (defined as being any summary of clinical performance of healthcare over a specified period of time) suggests that it is effective as a strategy to improve professional practice [##UREF##4##12##,##REF##16625533##13##]. However, such reviews have concentrated on clinical aspects of care such as guideline implementation. It cannot be assumed that feedback of patient assessments will have the same effect as feedback of clinical indicators, as physicians may place greater emphasis on professionally-based audit measures compared with patient assessments [##UREF##3##9##].</p>", "<p>Continuing medical education has been the traditional approach to improve clinical performance, and can range from passive, didactic, large group presentations to highly interactive learning methods, such as workshops, small groups and individualised training sessions. Systematic reviews of the literature have found that medical education can improve clinical performance with the most effective methods being interactive educational meetings, outreach events and strategies that involve multiple educational interventions (e.g. outreach plus reminders) [##REF##1501333##14##, ####REF##7585368##15##, ##REF##10478694##16####10478694##16##]. A recent systematic review of the education literature indicated a positive impact on clinical performance when education was coupled with feedback [##REF##16707292##17##].</p>", "<p>This study therefore uses systematic review techniques to assess the efficacy of (i) the feedback of patient assessments, (ii) brief training, and (iii) interventions combining both feedback and brief training (i.e. (i) and (ii) together), on the interpersonal skills of primary care physicians.</p>" ]

[ "<title>Methods</title>", "<title>Inclusion and exclusion criteria</title>", "<p>Studies eligible for inclusion were:</p>", "<p>1) Randomised controlled trials (RCTs) published in English</p>", "<p>2) Based on primary care practitioners and their patients. Primary care practitioners were defined as medical health care professionals providing first contact and on-going care to patients, regardless of the patient's age, gender or presenting problem, and included other relevant specialties such as general internists, family practitioners, paediatricians and obstetricians working in primary care settings. Medical students were excluded from the review. There were no restrictions on age, gender, ethnicity or health condition of patients included in the review.</p>", "<p>3) Utilising one of the following interventions:</p>", "<p>a) feedback of the assessments of real patients on the interpersonal skills of individual physicians. These assessments (i.e. patient satisfaction scores) were provided to physicians outside the consultation (e.g. written reports);</p>", "<p>b) 'brief' (up to one working week in length) training focussed on the improvement of interpersonal care.;</p>", "<p>c) interventions combining (a) and (b).</p>", "<p>4) Utilising a patient based assessment of change in interpersonal skills as an outcome.</p>", "<p>We considered interpersonal care in the broadest sense and included generic interpersonal skills (e.g. listening, providing information) and more specific areas (e.g. shared decision making skills, responding to the patient agenda). We did exclude feedback or training interventions that were specific to a particular disease.</p>", "<title>Search Strategy</title>", "<p>A list of initial search terms and synonyms was formulated by SCS on the basis of the population (primary care physicians and their patients), the interventions (patient based feedback and brief education) and the outcome (patient assessment of interpersonal care). Relevant published studies and reviews e.g[##REF##16625533##13##,##REF##8945689##18##] were reviewed for additional keywords. These searches were combined with the Cochrane Highly Sensitive Search Strategy (HSSS) for randomised controlled trials [##REF##11868444##19##]. The search strategy itself was built by grouping the individual free text and MeSH terms into categories and then combining those components Each search strategy was adapted to each database to ensure the appropriateness of the MeSH terms.</p>", "<p>Three electronic databases were searched during April–May of 2007. The primary search was of the CENTRAL register of controlled trials from the Cochrane Library. Subsequently, Medline and Embase searches were conducted limited to the years 2004–2007 in order to capture any articles of relevance that may at that point, due to volume and time constraints, not have been entered into the CENTRAL database. This approach was chosen on the basis of recent published evidence which shows that when searching specifically for RCTs, exhaustive searching of multiple electronic databases is not necessary due to the comprehensive nature of the CENTRAL database [##REF##16042789##20##]. References lists of included articles and of existing published reviews were searched for other relevant articles as well as utilising a citations tracker in order to identify any new relevant literature. The full search strategy is available for the interested reader (see additional file ##SUPPL##0##1##).</p>", "<p>One reviewer (SCS) applied the inclusion criteria to all the titles and abstracts identified by the electronic searches. Full text copies of all articles judged to be potentially relevant were retrieved for further investigation. Two reviewers (SCS and PB) then independently assessed these articles against the inclusion and exclusion criteria. Any disagreements were resolved via discussion at a series of face to face meetings.</p>", "<title>Data Abstraction</title>", "<p>For each included study, the two reviewers independently performed the data extraction using a modified version of the data collection checklist used by the Effective Practice and Organisation of Care group of the Cochrane Collaboration. Any discrepancies in data extraction or quality assessment were discussed and resolved by consensus. Due to time constraints, it was not possible to contact relevant authors for missing data.</p>", "<p>Methodological rigour was assessed by rating individual study criteria as indicators of trial quality [##REF##15817526##21##]. The following eight criteria were assessed:</p>", "<p>1) Allocation concealment</p>", "<p>2) Power calculation</p>", "<p>3) Sample size</p>", "<p>4) Follow-up of professionals</p>", "<p>5) Baseline comparability</p>", "<p>6) Published outcome measurement instrument</p>", "<p>7) Protection against contamination</p>", "<p>8) Unit of analysis issues</p>", "<p>The criteria were assessed in the following way:</p>", "<title>1) Allocation concealment</title>", "<p>The reviewers assessed this criterion as being 'done' where the unit of allocation was described explicitly and there was some form of centralised randomisation or an adequately concealed method (e.g. sealed opaque envelopes), 'not done' if the allocation was transparent before assignment, or 'not clear' where there was insufficient detail about the allocation method.</p>", "<title>2) Power calculation and 3) Sample size</title>", "<p>The reviewers assessed power calculations as being 'done' where there was evidence of a power calculation being conducted, 'not clear' if it was not reported and 'not done' if the authors specifically report that the study was under-powered. For sample size, the number of participants reported as being randomised was recorded.</p>", "<title>4) Follow-up of professionals</title>", "<p>The reviewers assessed this criterion as being 'done' where &gt; = 80% of the professionals randomised had been followed-up, 'not done' if outcome measures had been obtained for less than 80% and 'not clear' if it was not specified within the paper.</p>", "<title>5) Baseline comparability</title>", "<p>For baseline comparability of intervention and control group participants, we recorded this as being 'done' where the authors had done an analysis of baseline comparability and reported finding no significant differences that may affect the study results, 'not done' if there are significant baseline differences and 'not clear' where no evidence of any analysis of baseline comparability was reported. As statistical significance testing of baseline characteristics is flawed when sample sizes are small, we also report separately any cases that recorded a &gt; 10% differences in measured characteristics at baseline.</p>", "<title>6) Published outcome measurement instrument</title>", "<p>As we were interested in patient assessments of interpersonal care, it was important that the instruments that study authors had used to measure such assessments were valid and reliable. Although it would have been preferable to investigate the psychometric properties of the measurement tools in more detail, this was not possible due to resource constraints. Therefore, a proxy code was used, concerning whether the instruments were published in a peer reviewed journal or not.</p>", "<title>7) Protection against contamination</title>", "<p>The reviewers recorded this as 'done' where the physicians were the unit of allocation and 'not done' if patients rather than professionals were randomised.</p>", "<title>8) Unit of analysis</title>", "<p>Studies which randomize at the level of the clinician or practice but assess the effects on patients need to be analysed correctly, taking into account the difference between the unit of allocation and the unit of analysis. Failure to consider this may lead to inappropriate statistical testing [##REF##9794847##22##]. Appropriate analysis in cluster randomized trials was assessed as 'done' where appropriate adjustment was made for clustering or analysis was conducted at the cluster level, 'not done' when authors were explicit that no adjustment was made, and 'not clear' where there was insufficient detail about adjustment.</p>", "<p>For the analysis, study results were analysed by intervention type (i.e. feedback, brief training, and their combination).</p>" ]

[ "<title>Results</title>", "<p>The electronic search identified 20,840 citations (see figure ##FIG##0##1##). Screening of the titles and abstracts reduced this to 103 after excluding 20,737 ineligible articles. In addition to the electronic search, hand-searching located two further studies. These 105 studies were retrieved and reviewed by both SCS and PB. In total, nine studies were found to meet all four inclusion criteria, comprising two feedback and seven training studies. No studies combining both feedback with training were identified. Two of the included studies were reported across two separate publications respectively [##UREF##5##23##, ####REF##11743151##24##, ##REF##10065061##25##, ##REF##11083035##26####11083035##26##]. For an overview of the characteristics of included studies see additional file ##SUPPL##1##2##.</p>", "<title>Participants and settings</title>", "<p>General practitioners and their adult patients were the focus of four of the included studies [##UREF##5##23##,##REF##11743151##24##,##REF##11489102##27##, ####REF##3426975##28##, ##REF##16707508##29####16707508##29##], whereas a mixture of primary care physicians were included in the remainder. Three studies [##REF##11489102##27##,##REF##1861939##30##,##REF##3339486##31##] reported on interventions involving trainee physicians such as internal medicine residents, whereas the remaining studies utilised experienced practicing physicians. Five studies were conducted in the United States [##REF##10065061##25##,##REF##11083035##26##,##REF##1861939##30##, ####REF##3339486##31##, ##REF##8667091##32##, ##REF##10610626##33####10610626##33##]. The remaining four studies were conducted in the United Kingdom [##REF##16707508##29##]. The Netherlands [##UREF##5##23##,##REF##11743151##24##] and Australia [##REF##11489102##27##,##REF##3426975##28##]. Further details of physician and patient participants can be found in additional file ##SUPPL##2##3## and additional file ##SUPPL##3##4## respectively.</p>", "<title>Interventions</title>", "<title>Feedback</title>", "<p>Both studies used written feedback as the intervention. The frequency of feedback however did vary, with the first study, providing the intervention five times within a two year period (at three month intervals) [##REF##11489102##27##], whereas in the second study the intervention was only provided once within the fifteen month study period [##UREF##5##23##,##REF##11743151##24##]. An overview of the feedback interventions is given in additional file ##SUPPL##4##5##.</p>", "<title>Training</title>", "<p>Four studies investigated the effects of communication skills training [##REF##3426975##28##,##REF##1861939##30##,##REF##8667091##32##,##REF##10610626##33##] with the remaining studies investigating interventions that target specific interpersonal skills to increase trust [##REF##10065061##25##,##REF##11083035##26##], medical interviewing skills [##REF##3339486##31##] and increasing awareness of the patient agenda [##REF##16707508##29##]. Both individual and group settings were used to deliver training. The control groups in five studies received no training [##REF##10065061##25##,##REF##11083035##26##,##REF##3426975##28##,##REF##16707508##29##,##REF##3339486##31##,##REF##10610626##33##], whereas two studies gave equivalent training except for the specific educational content which was unrelated to interpersonal skills [##REF##1861939##30##,##REF##8667091##32##]. Details relating to the specific content of the training is provided in additional file ##SUPPL##5##6##.</p>", "<title>Methodological quality</title>", "<p>Of the nine included studies, only one study reported adequate methods of allocation concealment [##UREF##5##23##,##REF##11743151##24##]. Seven studies reported taking account of the clustered nature of the data in their analyses [##UREF##5##23##, ####REF##11743151##24##, ##REF##10065061##25##, ##REF##11083035##26####11083035##26##,##REF##16707508##29##, ####REF##1861939##30##, ##REF##3339486##31##, ##REF##8667091##32##, ##REF##10610626##33####10610626##33##]. Over half the studies performed power calculations [##UREF##5##23##, ####REF##11743151##24##, ##REF##10065061##25##, ##REF##11083035##26##, ##REF##11489102##27####11489102##27##,##REF##16707508##29##,##REF##10610626##33##] and all studies reported following up at least 80% of the professional participants. Sample sizes ranged from 19 to 210 physicians. In terms of baseline comparability of physicians, all studies reported no significant differences in measured characteristics at baseline although in some cases the differences were relatively large. All studies had ensured protection against contamination by randomising the physicians and one study reported specifically asking intervention group physicians not to discuss the intervention with control group physicians [##REF##3339486##31##]. Three of the studies used measurement tools that were not published in peer reviewed journals [##REF##3426975##28##,##REF##8667091##32##,##REF##10610626##33##]. In summary only one study met all the quality criteria [##UREF##5##23##,##REF##11743151##24##]. A summary of the methodological quality of included studies is given in additional file ##SUPPL##6##7##.</p>", "<title>Outcomes</title>", "<p>Meta-analysis was not possible as a minority of studies provided useable data. In addition, there was significant heterogeneity among studies in terms of interventions, which might make a pooled analysis difficult to interpret. Therefore a narrative description of the outcomes is presented.</p>", "<title>Feedback</title>", "<p>In the first of the feedback studies, three groups of trainee physicians were studied over their two year GP vocational training. The two intervention groups both received feedback at five points, with one of the groups also receiving preceptor discussions with their supervisors at two of the intervention points. The control group only received feedback at the first and fifth intervention points. Both intervention groups were found to have significant increases in patient satisfaction compared to control, particularly towards the earlier stages of training, however there were no significant differences between the two intervention groups [##REF##11489102##27##].</p>", "<p>In the second study, experienced physicians in the intervention group received an individual written feedback report on patients' evaluations of care whereas the control group received no feedback. The intervention group did not show any significant improvements in patient satisfaction scores, despite physicians reporting making changes in line with feedback [##UREF##5##23##,##REF##11743151##24##]. Further details of the analysis and results can be found in additional file ##SUPPL##7##8##.</p>", "<title>Training</title>", "<p>Only one training study reported a significant positive effect for the communications training intervention [##REF##3426975##28##]. The intervention consisted of two, three hour seminars using both written and oral methods, whereas the control group received no training. The remaining six studies reported that the training interventions had no significant positive effect [##REF##10065061##25##,##REF##11083035##26##,##REF##16707508##29##,##REF##3339486##31##,##REF##10610626##33##]. In one study, the control group actually showed greater improvements in the reported average satisfaction scores than in the intervention group [##REF##10610626##33##]. Further details of the results of training studies can be found in additional file ##SUPPL##8##9##.</p>" ]

[ "<title>Discussion</title>", "<p>The aim of this review was to determine the effectiveness of patient based feedback, brief training and their combination on the interpersonal skills of physicians. Only a small number of trials were identified and thus any conclusions about the effectiveness of these interventions is preliminary.</p>", "<title>The effectiveness of patient based feedback</title>", "<p>The search identified only two feedback studies, which have been identified previously in a review of instruments and feedback methods for the assessment of physicians using patient surveys [##REF##17264071##34##]. It is unclear from this review whether patient based feedback is an effective quality improvement tool for changing physicians' interpersonal care behaviour. The study involving trainees showed a significant positive effect for patient feedback on patient satisfaction scores [##REF##11489102##27##], whereas the study involving experienced physicians showed no effect [##UREF##5##23##,##REF##11743151##24##]. A study of feedback excluded from the review (conducted in a hospital setting) also reported a significant positive effect on the interpersonal skills of trainee internal medicine residents [##REF##3794839##35##]. Clinically experienced physicians may have more enduring interpersonal care behaviours that have developed over many years of practice, whereas trainees may be more easily able to adapt their behaviours in line with feedback.</p>", "<p>An alternative explanation for the difference in results may relate to the intensity of the feedback, with the study reporting positive effects applying the intervention at five time points (3 months apart) over a two year period [##REF##11489102##27##]. In contrast the study reporting no effect, only gave patient feedback at one time point (3–6 months after the start of the study) within the 15 month study period [##UREF##5##23##,##REF##11743151##24##].</p>", "<p>The use of patient feedback assumes that patients can judge the quality of interpersonal care and that the current assessment technology is capable of capturing patient views. Although doctors and patients have been shown to disagree about what constitutes technical quality of care [##REF##15528286##36##, ####REF##16793783##37##, ##REF##9463981##38####9463981##38##], it could be argued that no one is better placed than patients to judge interpersonal performance. There is evidence that patients are able to detect improvements in the quality of the physician-patient interaction [##REF##9527298##39##].</p>", "<p>A recent systematic review examined instruments designed to evaluate patients' experiences with individual practicing physicians and whether they are able to provide performance feedback at the individual level [##REF##17264071##34##]. Although many had some evidence of validity, it was generally limited, and it was not clear how well they correlated with other measures of doctor performance. One particular problem with using patient assessment instruments is the so called 'ceiling effect' due to the majority of patients express high levels of satisfaction with care i.e. there is little variation in responses [##UREF##6##40##, ####REF##8477901##41##, ##REF##7973865##42####7973865##42##]. The failure of these instruments to capture negative feedback is another issue that may reduce their effectiveness.</p>", "<p>Studies of feedback are unique in that the intervention is very similar to the outcome assessment (i.e. both use patient assessments, although only in the former is the data fed back). If the mere act of measurement (without feedback) were sufficient to change behaviour, then these studies may underestimate the effect of the intervention.</p>", "<p>Finally, the authors of the review discussed above [##REF##17264071##34##] found that the aim of feedback was often vague, the exact procedures to be used lacked specificity, and there was a lack of detail about the mechanism by which feedback was expected to lead to improvement, beyond an implicit suggestion of the impact of normative comparisons. The format of feedback may also be important. There was limited detail about the exact form of feedback given in the two studies, although one fed back data on individual questions and nine dimensions of care, with individual data for the GP and reference figures for all GPs [##UREF##5##23##]. Studies suggest that the style and content of feedback is important [##REF##15375098##43##], and there may be potential in testing different methods of presenting the data and the use of qualitative information from patients to complement quantitative data. Further work on the 'active ingredients' of feedback is clearly required.</p>", "<title>The effectiveness of brief training</title>", "<p>Brief training has previously been found to be effective in changing physician behaviour in general [##REF##10478694##16##,##REF##10824348##44##] and reviews focussing specifically on training for interpersonal skills, have also suggested that training can be effective [##REF##11687181##45##]. For example, a Cochrane review of training to improve patient-centredness reported positive effects on a number of outcomes [##REF##11687181##45##]. The difference between the results of the Cochrane review and the current study may reflect differences in outcome measures. The Cochrane review included multiple measures, including process measures of physician behaviour and health outcomes. When restricted to the seven trials using a patient based assessment of interpersonal care skills (the inclusion criteria for the present review) only two of the seven studies in the Cochrane review showed a positive significant effect, a result not substantively different from the results reported here [##REF##3794839##35##,##REF##7646751##46##].</p>", "<p>The only positive study was the oldest of all the training studies. This may be due to lower baseline levels of physician interpersonal skills. The medical training undertaken by the participating physicians (whose average age was 41.7 years when the study was undertaken and published in 1987) may have placed less of an emphasis on teaching interpersonal skills as trainees and practitioners were assumed to acquire interpersonal skills incidentally, simply via interacting with patients [##UREF##7##47##]. Physicians in more recent studies would have undertaken more formal instruction and assessment.</p>", "<p>Although the review was restricted to primary care physicians, findings from the wider literature on communication skills for health professionals may be informative in developing more effective interventions. Reviews suggest that effective interventions require combinations of didactic components with practice rehearsal and feedback from peers [##REF##15586302##48##]. Interventions may also need to focus on attitudes that may clash with the interpersonal skills being taught [##REF##15258839##49##]. Another key issue is the length of training. The study used a maximum of one week training as an inclusion criterion, based on discussions with GP colleagues as to what was likely to be feasible in relation to practising GPs. The limited effects of training may simply reflect the limited duration of the interventions, and reflect the paradox that in primary care, effective training may be unfeasible, whereas feasible training may be ineffective [##REF##10642207##50##].</p>", "<title>Limitations of the study</title>", "<p>We offer several cautions about the interpretation of these results, over and above caveats concerning the number of identified studies. Firstly, as in all such reviews there is the potential for publication bias. Such bias can lead to an overestimation of an intervention's effect on the outcomes i.e. a false positive [##REF##1672966##51##,##UREF##8##52##]. Secondly, due to time constraints, we were unable to contact authors for additional information therefore we included only published data. A second consequence of time constraints excluded searching via other means e.g. hand-searching of journals and conference proceedings etc. Thirdly, if studies showing an intervention to be effective are more likely to be published in English, then any summary of only the English language reports retrieved through a database search may result in an overestimate of effectiveness due to a language bias [##REF##7853041##53##,##REF##9251637##54##].</p>", "<p>Both of the included studies indicating positive effects did not adjust for clustering. There is a risk of inflating statistical significance when analysing patient level data without adjusting for clustering [##REF##9794847##55##].</p>", "<p>The study included trials where the outcome measure was a patient assessment. This criteria was used because interventions that change in patient assessments are likely to be of greater interest to policy makers. However, it should be noted that it may be more appropriate to use a range of assessment technologies (such as process measures of behaviour in the consultation) as well as patient outcomes [##REF##15586302##48##].</p>", "<p>The current review was restricted to primary care physicians as they currently provide the majority of care in this setting [##UREF##9##56##]. Future reviews into these interventions should take into account the potential shift towards increased nurse-led delivery of primary care [##REF##18700082##57##].</p>", "<title>Implications for research</title>", "<p>Although the trials identified in the review were of reasonable quality, their limited number means that confident conclusions about the efficacy of these interventions must await the publication of new studies. A more substantial evidence base is also required to explore the various factors that may affect the efficacy of patient based feedback. Such factors may include the frequency, content and style of feedback and training, and physician and patient characteristics.</p>", "<p>The theoretical basis of feedback and training interventions was sometimes unclear. More explicit statements of theory underlying interventions and qualitative research conducted as part of the trials may provide insights into why these interventions succeed or fail.</p>", "<p>Thirdly, the effectiveness of patient based feedback in combination with other interventions should be investigated (e.g. the combination of patient based feedback with brief training, or with financial incentives). Financial incentives are known to be effective (external) motivators [##REF##10162128##58##]. This type of arrangement is already utilised in the United States. General practice in the United Kingdom has become accustomed to conducting patient surveys on an annual basis for financial incentives, but the current incentive structure pays physicians primarily on the basis of conducting the survey rather than making changes.</p>", "<p>Finally, the cost effectiveness of these interventions need to be assessed. The National Health Service in the United Kingdom has already made a significant financial investment in the process of patient assessments in primary care, and it is critical that this investment can be proven to be a good use of resources compared to other competing priorities.</p>" ]

[ "<title>Conclusion</title>", "<p>There is limited evidence available on the effects of patient based feedback. There is reasonable evidence that brief training as currently delivered is not effective, although the evidence is not definitive, because of the small number of trials and the variation between them in terms of their training methods and goals. Further research into both feedback and brief training is required. The interventions to be tested in future research should consider using insights from the wider literature on communication outside primary care, might benefit from a clearer theoretical basis, and should examine the use of combined brief training and feedback to improve physicians' interpersonal skills.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Improving quality of primary care is a key focus of international health policy. Current quality improvement efforts place a large focus on technical, clinical aspects of quality, but a comprehensive approach to quality improvement should also include interpersonal care. Two methods of improving the quality of interpersonal care in primary care have been proposed. One involves the feedback of patient assessments of interpersonal care to physicians, and the other involves brief training and education programmes. This study therefore reviewed the efficacy of (i) feedback of real patient assessments of interpersonal care skills, (ii) brief training focused on the improvement of interpersonal care (iii) interventions combining both (i) and (ii)</p>", "<title>Methods</title>", "<p>Systematic review of randomised controlled trials. Three electronic databases were searched (CENTRAL, Medline and Embase) and augmented by searches of the bibliographies of retrieved articles. The quality of studies was appraised and results summarised in narrative form.</p>", "<title>Results</title>", "<p>Nine studies were included (two patient based feedback studies and seven brief training studies). Of the two feedback studies, one reported a significant positive effect. Only one training study reported a significant positive effect.</p>", "<title>Conclusion</title>", "<p>There is limited evidence concerning the effects of patient based feedback. There is reasonable evidence that brief training as currently delivered is not effective, although the evidence is not definitive, due to the small number of trials and the variation in the training methods and goals. The lack of effectiveness of these methods may reflect a number of issues, such as differences in the effectiveness of the interventions in experienced practitioners and those in training, the lack of theory linking feedback to behaviour change, failure to provide sufficient training or to use a comprehensive range of behaviour change techniques. Further research into both feedback and brief training interventions is required before these interventions are routinely introduced to improve patient satisfaction with interpersonal care in primary care. The interventions to be tested in future research should consider using insights from the wider literature on communication outside primary care, might benefit from a clearer theoretical basis, and should examine the use of combined brief training and feedback.</p>" ]

[ "<title>Competing interests</title>", "<p>PB has been involved in the development of a patient assessment questionnaire which is currently recommended for use by GPs in the United Kingdom as part of their contract. The questionnaire is free to use for NHS staff, but commercial companies selling patient evaluation services are required to pay a license fee. Funds derived from selling the questionnaire in these instances are received by the organization and used to fund research and administration. PB does not gain financially from its use.</p>", "<title>Authors' contributions</title>", "<p>The study was conducted as part of a Masters dissertation by SC–S. Both authors developed the idea for the study. SC–S devised the search strategy, reviewed identified articles, extracted data, conducted the analysis and drafted the manuscript. PB assisted with identification of studies and data extraction and aided in the drafting of the manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1472-6963/8/179/prepub\"/></p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>We thank Rosalind McNally from NPCRDC for her expertise with databases and her input into the search strategy, and the reviewers for their helpful comments on the initial draft.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Overview of searching process.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional file 1</title><p>full CENTRAL search strategy.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S2\"><caption><title>Additional file 2</title><p>Overview of Included Studies.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S3\"><caption><title>Additional file 3</title><p>Population Characteristics – Physicians.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S4\"><caption><title>Additional file 4</title><p>Population Characteristics – Patients.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S5\"><caption><title>Additional file 5</title><p>Intervention – Feedback.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S6\"><caption><title>Additional file 6</title><p>Interventions – Training.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S7\"><caption><title>Additional file 7</title><p>Quality of Included Trials.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S8\"><caption><title>Additional file 8</title><p>Results – feedback.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S9\"><caption><title>Additional file 9</title><p>Results – training.</p></caption></supplementary-material>" ]

[]

[ "<graphic xlink:href=\"1472-6963-8-179-1\"/>" ]

[ "<media xlink:href=\"1472-6963-8-179-S1.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1472-6963-8-179-S2.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1472-6963-8-179-S3.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1472-6963-8-179-S4.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1472-6963-8-179-S5.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1472-6963-8-179-S6.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1472-6963-8-179-S7.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1472-6963-8-179-S8.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1472-6963-8-179-S9.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Data from recent epidemiological studies suggest that depressive disorders exist on a continuum, rather than in separate categories [##REF##9268772##1##,##REF##15023482##2##]. As a consequence, research has begun to accumulate on the clinical relevance and public health significance of depressive symptoms not meeting diagnostic criteria, variously labelled subthreshold, subclinical, subsyndromal, mild, or minor depression. Here, we use the term subthreshold depression. Subthreshold depression is prevalent [##REF##16612172##3##], increases the risk of developing major depressive disorder [##REF##15049768##4##], and has considerable economic costs [##REF##17302623##5##]. At the individual level, disability from subthreshold depression is lower than for depressive disorders; however, the burden of disability for the population as a whole is substantial for subthreshold depression because of its greater prevalence [##REF##12462855##6##]. Given that unipolar depressive disorders were the leading cause of disability burden globally in 2001 [##UREF##0##7##], depressive symptoms falling short of a disorder are of major public health significance.</p>", "<p>Several trials have investigated treatments for milder depressive states, with some success [##REF##16612172##3##,##REF##17498154##8##]. However these treatments, which include antidepressant medication and brief psychotherapy, involve the participation of health professionals. An approach that does not further burden clinical resources is preferable, as there is already a large group of people with major depression who do not receive treatment [##REF##10885163##9##], and treating these people deserves priority over those with subthreshold symptoms. An alternative approach is self-help that can be applied by the individuals affected without the need for professional guidance.</p>", "<p>Self-help approaches for depression are commonly used, particularly for milder forms of depression [##REF##14982135##10##,##UREF##1##11##], and are perceived as helpful by the public [##REF##16004615##12##]. However, some self-help methods in common use are probably self-defeating (for example, substance use). If effective informal self-help methods could be identified, they could be used as a cost-effective way of reducing subthreshold depressive symptoms. Health promotion campaigns on other major sources of disease burden, such as heart disease and cancer, routinely include information on actions that can be taken to reduce risk. Jorm and Griffiths [##REF##16356291##13##] called for this approach to be extended to self-help interventions for depression, with the aim of reducing subthreshold depressive symptoms and the risk of progressing to a depressive disorder. If applied successfully, such an approach would have the potential to reduce the distribution of symptoms across the whole population. However, due to the risk of suicide and detriment to functioning if symptoms deteriorate or do not improve, such an approach would also need clear guidelines on when to seek professional help rather than relying on self-help strategies.</p>", "<p>If a health promotion approach were to be applied, the first step is to identify a small number of self-help actions that are likely to be effective and that can be applied easily by many people at low cost. A number of reviews have examined the evidence for self-help or complementary therapies for depression [##REF##12065003##14##, ####REF##17144787##15##, ##REF##17562566##16##, ##REF##16926053##17##, ##REF##16449696##18##, ##REF##17014404##19####17014404##19##]. These have found reasonable evidence for St John's wort, S-adenosylmethionine, exercise, bibliotherapy, and light therapy. Although these reviews are informative, we decided to undertake our own systematic review of the evidence because prior reviews were either outdated (in a rapidly growing research area), only reviewed treatments for depressive disorders and not subthreshold symptoms, or they focused solely on complementary and alternative therapies rather than other self-help strategies.</p>" ]

[ "<title>Methods</title>", "<title>Selection of treatments to review</title>", "<p>Treatments were identified from previous systematic reviews of complementary and self-help treatments for depression [##REF##12065003##14##,##REF##17014404##19##]. Not all of these treatments were included for review here as some required the assistance of another person (for example, LeShan distance healing) or a visit to a practitioner (for example, acupuncture).</p>", "<title>Search methodology</title>", "<p>PubMed, PsycINFO and the Cochrane Database of Systematic Reviews were searched using the following terms: name-of-treatment (and synonyms) AND (depressi* OR dysthym* OR affective OR mood), limited to English and humans (see Additional file ##SUPPL##0##1## for search details). Most searches were carried out of literature up to March 2007, however a few treatments found in the course of the review were searched up to September 2007. Reference lists and citations of included studies were also checked. Treatments with no relevant studies to review are listed in Table ##TAB##0##1##. Studies were reviewed by one author and the accuracy of each review was checked by a second.</p>", "<title>Inclusion/exclusion criteria</title>", "<p>Studies were included for review if they evaluated the treatment's effects on depression symptoms or depressed mood, using a reliable and valid scale for depression or depressed mood. In contrast with the previous reviews which only included studies with individuals selected to have a depressive disorder or a high level of depressive symptoms, in this review we also included studies with participants not selected for depression, as they may have had subthreshold or mild depression symptoms. Studies were grouped as involving depressive disorders (participants with a depressive disorder or a high level of depressive symptoms) or non-clinically depressed (participants not selected for depression). The scope of the review was limited to randomised controlled trials with sufficiently large samples that had the power to detect a standardised mean difference (d) of 1. Studies with independent groups were rejected if they had less than 17 participants per group (this gives 80% power to detect an effect size of d = 1 in independent groups with alpha set at 0.05) and crossover studies were rejected if there were less than 10 participants (assuming a correlation of 0.5 between ratings, this gives 80% power to detect an effect size of d = 1 with alpha set at 0.05). Trials without an appropriate control intervention or with uninterpretable findings were also excluded. No age restrictions were applied, but studies with children/adolescents, adults, or older adults were reviewed separately where appropriate. Preference was given to reviewing recent meta-analyses or systematic reviews where they were available. As we were interested in interventions that could be applied by most individuals with depression, and without recourse to a professional, studies were excluded from the review if:</p>", "<p>• the self-help treatment was in addition to an antidepressant or psychotherapy (adjunctive or augmentation studies);</p>", "<p>• participants had a comorbid medical or mental illness with depression secondary;</p>", "<p>• participants were primarily bipolar patients;</p>", "<p>• they investigated premenstrual syndrome/premenstrual dysphoric disorder, postpartum depression, or hormone-related depression (for example, in menopausal women);</p>", "<p>• the depression symptoms were caused by a clear underlying nutritional deficiency (for example, magnesium) or underlying medical condition (for example, coeliacs disease or mitochondrial disorder).</p>" ]

[ "<title>Results</title>", "<p>There were 38 interventions with relevant evidence to review. For convenience, interventions have been grouped under the categories of herbal remedies or dietary supplements, substances, dietary methods, psychological methods, lifestyle changes, and physical and sensory methods. For some interventions, no evidence regarding effects on depression was available (see Table ##TAB##0##1##).</p>" ]

[ "<title>Discussion</title>", "<p>The self-help interventions with the best evidence of efficacy for depressive disorders are S-adenosylmethionine, St John's wort, bibliotherapy, computerised interventions, distraction, relaxation training, exercise, pleasant activities, sleep deprivation, and light therapy. With the possible exception of St John's wort, these interventions have been less researched than standard treatments provided by a professional such as antidepressants or cognitive behaviour therapy. Preliminary evidence also appears promising for a number of other interventions; however these have received less research attention. These include borage, carnitine/acetyl-<sc>L</sc>-carnitine, saffron, autogenic training, yoga, massage, and negative air ionisation.</p>", "<p>There were fewer interventions with good or preliminary evidence in non-clinically depressed samples. Promising interventions with immediate effects on depressed mood include distraction, exercise, humour, music, negative air ionisation, and singing. Autogenic training, light therapy, omega 3 fatty acids, pets, and prayer may have helpful longer-term effects over days or weeks. The mechanism of action for many interventions is unclear, and for some with promising effects the mechanism is completely unknown, for example, negative air ionisation. Studies in non-clinically depressed samples may include participants with varying degrees of depressive symptoms, from none through to symptoms at the threshold for major depression. The context behind depressive symptoms is also unknown: symptoms could be residual following resolution of a major depressive episode, prodromal to a major depressive episode, ongoing, or reactions to life stresses or bereavement. As such, it is not surprising that fewer studies had positive results than studies in participants with depressive disorders, even though it is probable that some self-help interventions are effective in reducing depressive symptoms within specific ranges of symptom severity and in particular contexts. Another problem is measuring change in symptoms in populations near the normal end of the depression spectrum. A lack of instrument sensitivity to small changes in symptoms may be responsible for no significant changes detected in many trials with non-clinically depressed participants. As prodromal symptoms of depressive disorders appear to involve anxiety and irritability [##UREF##49##181##], it may be more appropriate to measure general psychological distress in these populations with lower levels of depressive symptoms, for example by using instruments such as the K10 questionnaire [##REF##12214795##182##].</p>", "<p>Although some interventions appear promising, there remains much to be learned about active ingredients and mechanisms, the specification of activities, behaviours and intervention content (for example, for exercise, the type and ideal dosage), as well as possible side effects and safety issues. Also, interventions were conducted in ideal conditions with at least some degree of professional involvement. Whether these effects generalise to conditions of informal self-help, where there is no professional involvement, is yet to be evaluated. Many of the trials were poor quality, suffering from short durations with no follow-up, little information on attrition, possible blinding issues, or had yet to be replicated by other research groups. Differential effects across age groups have not received much attention either.</p>", "<p>For the majority of interventions searched, there were no trials available to review, and for some interventions there was a lack of research on their use as monotherapy. Many self-help strategies for depressive symptoms are not feasible or ethical to evaluate in RCTs, such as taking time off work, and may require alternative approaches to evaluating evidence. One approach is to ask individuals who have experienced depression what they find personally helpful [##UREF##50##183##]. This approach found that exercise, yoga/meditation, massage, and relaxation were rated highly and as strongly as professionally recommended strategies such as CBT and SSRIs. Another approach is to develop consensus of experts on what works best. We are currently undertaking just such a project using the Delphi [##REF##11095242##184##,##REF##7640549##185##] method of consensus, by gathering the views of expert clinicians and consumers on what self-help strategies are likely to be most helpful for subthreshold depressive symptoms.</p>", "<p>Some interventions with good or reasonable evidence are very feasible to implement by an individual and would fit well into a promotional campaign. Others may be less feasible due to the need to purchase expensive equipment or supplements, or require an investment in time or effort to learn. As there may be no support or monitoring from professionals, the risk/benefit ratio would need to be low as well.</p>", "<p>There are a number of limitations to this review, including a search restricted to articles written in English; restricting the reporting of mood effects to that of depressed mood only, rather than including other possible relevant mood variables such as energy level, fatigue or anxiety; and the number of interventions reviewed which precluded a more detailed analysis of each intervention.</p>" ]

[ "<title>Conclusion</title>", "<p>A number of self-help interventions have promising evidence for reducing subthreshold depressive symptoms, although a larger evidence base is needed. Promotion of effective self-help strategies for subthreshold depressive symptoms could fit within a clinical staging model for depressive disorders. A clinical staging model allows for different intervention approaches at different stages of illness development. Intervening early during prodromal or subthreshold symptoms with benign but effective techniques could delay or prevent onset of depressive illness, reduce functional impairment, and prevent progression to other undesirable outcomes such as harmful use of substances [##REF##18560251##186##]. The present review has identified a number of self-help interventions that could usefully be evaluated for prevention and early intervention with depressive symptoms.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Research suggests that depressive disorders exist on a continuum, with subthreshold symptoms causing considerable population burden and increasing individual risk of developing major depressive disorder. An alternative strategy to professional treatment of subthreshold depression is population promotion of effective self-help interventions that can be easily applied by an individual without professional guidance. The evidence for self-help interventions for depressive symptoms is reviewed in the present work, with the aim of identifying promising interventions that could inform future health promotion campaigns or stimulate further research.</p>", "<title>Methods</title>", "<p>A literature search for randomised controlled trials investigating self-help interventions for depressive disorders or depressive symptoms was performed using PubMed, PsycINFO and the Cochrane Database of Systematic Reviews. Reference lists and citations of included studies were also checked. Studies were grouped into those involving participants with depressive disorders or a high level of depressive symptoms, or non-clinically depressed participants not selected for depression. A number of exclusion criteria were applied, including trials with small sample sizes and where the intervention was adjunctive to antidepressants or psychotherapy.</p>", "<title>Results</title>", "<p>The majority of interventions searched had no relevant evidence to review. Of the 38 interventions reviewed, the ones with the best evidence of efficacy in depressive disorders were S-adenosylmethionine, St John's wort, bibliotherapy, computerised interventions, distraction, relaxation training, exercise, pleasant activities, sleep deprivation, and light therapy. A number of other interventions showed promise but had received less research attention. Research in non-clinical samples indicated immediate beneficial effects on depressed mood for distraction, exercise, humour, music, negative air ionisation, and singing; while potential for helpful longer-term effects was found for autogenic training, light therapy, omega 3 fatty acids, pets, and prayer. Many of the trials were poor quality and may not generalise to self-help without professional guidance.</p>", "<title>Conclusion</title>", "<p>A number of self-help interventions have promising evidence for reducing subthreshold depressive symptoms. Other forms of evidence such as expert consensus may be more appropriate for interventions that are not feasible to evaluate in randomised controlled trials. There needs to be evaluation of whether promotion to the public of effective self-help strategies for subthreshold depressive symptoms could delay or prevent onset of depressive illness, reduce functional impairment, and prevent progression to other undesirable outcomes such as harmful use of substances.</p>" ]

[ "<title>Herbal remedies or dietary supplements</title>", "<title>Borage (Borago officinalis or Echium amoenum)</title>", "<title>Description and rationale</title>", "<p>Borage is a herb originating in Syria. The flowers of the plant can be used in herbal teas. Although the plant is used in traditional Iranian medicine for mood enhancement, its antidepressant mechanism is unclear.</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>There has been one small randomised controlled trial (RCT) [##REF##16309809##20##]. A total of 35 adults with mild to moderate major depressive disorder received either placebo or 375 mg of aqueous extract of borage flowers daily for 6 weeks. By week 4 there was a small significant difference in levels of depression symptoms between the two groups, with lower levels in the borage group. Results at week 6 were similar but no longer statistically significant.</p>", "<title>Conclusion</title>", "<p>There is preliminary evidence that borage flower extract may be helpful for depression. Longer trials with larger samples are needed to confirm these results. There is no evidence on the effects of borage in non-clinically depressed people.</p>", "<title>Carnitine/acetyl-<sc>L</sc>-carnitine</title>", "<title>Description and rationale</title>", "<p>Carnitine is a nutrient involved in energy metabolism. It is produced in the body and is available in food such as meat and dairy products. Acetyl-<sc>L</sc>-carnitine is an ester of carnitine that readily enters the brain. Carnitine supplements are available from pharmacies and health food shops. The antidepressant mechanism is unknown. Possible mechanisms include an inhibitory effect on the hypothalamic-pituitary-adrenal axis activity resulting in a reduction of cortisol levels [##REF##10608918##21##] or effects on membrane phospholipid metabolism and membrane physical/chemical properties [##REF##11126392##22##].</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>Three RCTs have evaluated acetyl-<sc>L</sc>-carnitine supplementation in individuals with dysthymia [##REF##2099360##23##, ####UREF##2##24##, ##REF##16316746##25####16316746##25##]. Of these trials, 2 were in 46 or 52 older adults (aged 60 to 80 years) and compared 3 g daily doses of acetyl-<sc>L</sc>-carnitine with placebo over 60 days. Those taking acetyl-<sc>L</sc>-carnitine showed significantly improved depression symptoms compared with those taking placebo. The other trial compared 1 g daily dosage of acetyl-<sc>L</sc>-carnitine with 50 mg amisulpride in 193 participants with dysthymia and found both groups had improved depression symptoms over 3 months and there was no significant difference in improvement between groups [##REF##16316746##25##].</p>", "<title>Non-clinically depressed</title>", "<p>A double-blind RCT evaluated the effect over days of carnitine on depressed mood [##REF##14760514##26##]. A total of 400 adult females received either a placebo, 2 g carnitine, 1.6 g lecithin, or both lecithin and carnitine for 3 days. Carnitine supplementation had no effect on depressed mood.</p>", "<title>Conclusion</title>", "<p>Preliminary evidence suggests acetyl-<sc>L</sc>-carnitine may be helpful for dysthymia, particularly in older adults. From the limited evidence available carnitine does not appear to be effective in non-clinically depressed adults.</p>", "<title>Chromium</title>", "<title>Description and rationale</title>", "<p>Chromium is an essential trace mineral involved in carbohydrate, fat and protein metabolism. Chromium is available in food or as a supplement from health food shops, usually in the form chromium picolinate. The antidepressant mechanism is unknown but could involve increased insulin sensitivity resulting in enhanced central noradrenergic and serotonergic activity [##UREF##3##27##].</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>A trial of 113 adults with atypical depression who took either 400 to 600 μg chromium picolinate or placebo for 8 weeks found no significant difference in the reduction of depression symptoms or rates of response [##REF##16184071##28##]. However, a subgroup analysis found that adults who had high carbohydrate craving showed a greater response to chromium than placebo.</p>", "<title>Conclusion</title>", "<p>Limited evidence suggests chromium supplementation is not helpful for depressive disorders, although there is tentative evidence that it may be helpful for a subgroup of atypically depressed adults with high levels of carbohydrate craving. There is no evidence on the effects of chromium in non-clinically depressed people.</p>", "<title>Ginkgo biloba</title>", "<title>Description and rationale</title>", "<p>Extracts from the leaves of the Ginkgo biloba (maidenhair) tree are available in tablet form from health food shops. Its antidepressant mechanism is proposed to be a reduction in the production of stress hormones [##UREF##4##29##]. Ginkgo may also be effective for the treatment of impaired cerebral circulation in the elderly, one symptom of which is depressed mood [##REF##10442441##30##].</p>", "<title>Review of efficacy</title>", "<title>Non-clinically depressed</title>", "<p>Two RCTs in 104 healthy young adults and 93 older adults of 120 mg ginkgo daily for 12 weeks showed no effect on depressed mood [##REF##16329161##31##].</p>", "<title>Conclusion</title>", "<p>From the limited evidence available, ginkgo does not appear effective for depressed mood in non-clinically depressed adults. There is no evidence on the effects of ginkgo on depressive disorders.</p>", "<title>Korean ginseng (Panax ginseng)</title>", "<title>Description and rationale</title>", "<p>Korean ginseng is a herb native to Korea and China. Extracts from the root of the plant are available as supplements from health food shops. The major active constituents are thought to be ginsenosides which may increase resistance to stress through their action on the hypothalamic-pituitary-adrenal axis [##REF##12895687##32##].</p>", "<title>Review of efficacy</title>", "<title>Non-clinically depressed</title>", "<p>One RCT has examined ginseng's effects on mood over months in healthy adults. In all, 83 participants took either 200 mg ginseng, 400 mg ginseng or placebo daily for 60 days [##REF##11424544##33##]. Ginseng supplementation had no effect on depressed mood.</p>", "<title>Conclusion</title>", "<p>From the limited evidence available, ginseng does not appear to be effective for depressed mood in non-clinically depressed individuals. There is no evidence on the effects of ginseng on depressive disorders.</p>", "<title>Lavender (Lavandula angustifolia)</title>", "<title>Description and rationale</title>", "<p>Lavender is a traditional herbal remedy that is thought to 'strengthen the nervous system' [##REF##12551734##34##] and may aid sleep and relaxation. Extracts are obtained from the flowering tops.</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>One small double-blind RCT has compared lavender with an antidepressant in adults with depressive disorders [##REF##12551734##34##]. A total of 45 adults with major depression participated in a 4-week trial where they received 60 drops of a lavender tincture plus placebo tablet, 100 mg imipramine plus placebo drops, or lavender plus imipramine. Although depression symptoms improved significantly in all groups, the lavender group improved significantly less than the imipramine group, and there was no placebo control group to rule out placebo effects.</p>", "<title>Conclusion</title>", "<p>There is insufficient evidence to determine whether lavender may be helpful for depressive disorders. There is no evidence on the effects of lavender in non-clinically depressed people.</p>", "<title>Lecithin</title>", "<title>Description and rationale</title>", "<p>Lecithin is a mixture of phospholipids and is a major component of cell membranes. Lecithin is found in foods such as eggs and soy beans, but is also available as a supplement from health food shops. Choline, a component of lecithin, is a precursor to acetylcholine, which is needed for normal brain functioning.</p>", "<title>Review of efficacy</title>", "<title>Non-clinically depressed</title>", "<p>One double-blind RCT has examined the effect over days of lecithin on depressed mood [##REF##14760514##26##]. A total of 400 adult females received either a placebo, 1.6 g lecithin (phosphatidylcholine), 2 g carnitine, or both lecithin and carnitine for 3 days. Lecithin supplementation had no effect on depressed mood.</p>", "<title>Conclusion</title>", "<p>From the limited evidence available lecithin does not appear to be effective for depressed mood in non-clinically depressed individuals. There is no evidence on the effects of lecithin on depressive disorders.</p>", "<title>Melatonin</title>", "<title>Description and rationale</title>", "<p>Melatonin is a hormone involved in the regulation of sleep/wake cycles. Over the counter supplements are available in some countries. The mechanism is unclear, but research suggests melatonin production is disturbed in depressed people, and that a dysfunction in the timing of melatonin production is a possible cause of seasonal affective disorder [##REF##16861139##35##].</p>", "<title>Review of efficacy</title>", "<title>Non-clinically depressed</title>", "<p>An RCT of 53 adults with subsyndromal SAD and/or weather-associated syndrome who took 2 mg slow-release melatonin in the evening for 3 weeks found no significant difference in atypical depression symptoms between melatonin and placebo [##REF##12729938##36##].</p>", "<title>Conclusion</title>", "<p>Limited evidence suggests that melatonin has no effect on depressive symptoms in non-clinically depressed individuals. There is no evidence on the effects of melatonin on depressive disorders.</p>", "<title>Omega 3 fatty acids (fish oils)</title>", "<title>Description and rationale</title>", "<p>Omega 3 fatty acids are long-chain polyunsaturated fatty acids. The two most important for depression are eicosapentanoic acid (EPA) and docosahexanoic acid (DHA), which are found in fish or are made in the body from alpha-linolenic acid (another omega 3 fatty acid, found in flaxseed, walnuts and canola oil). Omega 3 supplements (containing EPA and DHA) are available from health food shops and pharmacies. Several lines of evidence suggest a link between omega 3 fatty acids and depression. An increase in rates of depression in Western countries has paralleled a change in diet to one favouring omega 6 over omega 3 fatty acids; across countries there is a strong negative association between fish consumption and depression; and lower concentrations of omega 3 have been found in the blood of depressed people. Possible mechanisms include omega 3's effects on the fluidity of cell membranes, which leads to changes in signalling within and between brain cells; and omega 3's anti-inflammatory effects, as depression may be caused by an overactive inflammation response.</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>Although there have been several reviews of omega 3 fatty acids for depression [##REF##16741195##37##,##REF##16650900##38##], only one study has evaluated omega 3 as a single treatment for depression in sufficient participants [##REF##12727707##39##]. A double-blind RCT of 35 depressed adults with low fish intake who took 2 g DHA or placebo daily for 6 weeks found that omega 3 supplementation was no better than placebo in reducing depression symptoms.</p>", "<title>Non-clinically depressed</title>", "<p>A single RCT of 49 healthy adults examined the effect on depressed mood of supplementation of 4 g fish oil (containing 1,600 mg EPA and 800 mg DHA), or placebo for 35 days [##REF##16269019##40##]. Depressed mood reduced significantly in the omega 3 group but not in the placebo group.</p>", "<title>Conclusion</title>", "<p>The only trial to qualify for inclusion in the review found that omega 3 fatty acids were not helpful for depressive disorders. Preliminary evidence suggests omega 3 fatty acids for depressed mood in non-clinically depressed individuals may be beneficial, but requires replication in further trials.</p>", "<title>S-Adenosylmethionine</title>", "<title>Description and rationale</title>", "<p>S-Adenosylmethionine (SAMe) is a compound that is manufactured in the body, is a major methyl donor in the brain and is involved in many biochemical reactions. Supplements are available in a number of countries from pharmacies and health food shops. The antidepressant mechanism of SAMe is unknown, but may involve its effects on the fluidity of neuronal membranes or its involvement in serotonin, dopamine and norepinephrine synthesis.</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>Both a recent systematic review [##REF##16021987##41##] and a meta-analysis [##UREF##5##42##] have found SAMe helpful for depressive disorders. The systematic review was restricted to trials that passed a quality assessment. Those included were five uncontrolled trials and two RCTs. Despite differences in doses, route of administration (oral, intramuscular, intravenous) and comparison or control treatments, SAMe had a consistent positive effect over weeks or months. An additional RCT was included after the review was completed, which found that the efficacy of 1,600 mg/day oral SAMe or 400 mg/day intramuscular SAMe was not significantly different from 150 mg/day of imipramine. The meta-analysis included 28 trials and found greater improvement with SAMe than with placebo (global effect size ranging from 17% to 38% depending on definition of response), and no difference in outcomes between treatment with SAMe and standard tricyclic antidepressants.</p>", "<title>Conclusion</title>", "<p>There is consistent evidence that SAMe may be helpful for depressive disorders in adults. Further large, longer-term RCTs are needed to clarify questions regarding optimum dosage, safety and comparison with newer antidepressants. An RCT in children and adolescents is warranted. There is no evidence on the effects of SAMe in non-clinically depressed people.</p>", "<title>Saffron (<italic>Crocus sativus </italic>L.)</title>", "<title>Description and rationale</title>", "<p>Saffron is the world's most expensive spice, made from the stigma of the flower of the <italic>Crocus sativus</italic>. Both the stigma and the petal (which is much cheaper) have been used for the treatment of depression. Saffron is used for depression in Persian traditional medicine. Its mechanism is unclear, but it has been proposed that two components of saffron, crocin and safranal, inhibit reuptake of dopamine, norepinephrine and serotonin [##REF##15341662##43##].</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>Four double-blind RCTs have examined the effect of saffron (stigma) or Crocus sativus petals on depressed adults. Two trials each with 40 adults compared saffron stigma (30 mg daily), with fluoxetine (20 mg) [##REF##15707766##44##] or with placebo [##REF##15852492##45##] for 6 weeks. Saffron significantly reduced depression symptoms more than placebo, and there was no significant difference in efficacy between saffron and fluoxetine. Similarly, 30 mg extracts from the petals of Crocus sativus have also shown efficacy similar to 20 mg fluoxetine [##REF##17174460##46##] and greater efficacy than placebo [##REF##16979327##47##] in trials of 40 adults.</p>", "<title>Conclusion</title>", "<p>Evidence for the efficacy of saffron in adults with depressive disorders is promising. The results need to be replicated by other research groups in larger trials with longer durations. There is no evidence on the effects of saffron in non-clinically depressed people.</p>", "<title>Selenium</title>", "<title>Description and rationale</title>", "<p>Selenium is an essential trace element although it can be toxic in high doses. Selenium is found in high protein foods, or is available as a supplement from health food shops. Although it is preferentially retained in the brain during times of deficiency, no mechanism has been proposed for how it might affect mood.</p>", "<title>Review of efficacy</title>", "<title>Non-clinically depressed</title>", "<p>Two trials have examined selenium intake and depressed mood in non-depressed adults. A double-blind crossover trial found daily supplementation of 100 μg selenium in 50 adults significantly improved depressed mood over 5 weeks (compared to placebo) [##REF##2096413##48##,##REF##1873372##49##] and a RCT found no effect of a range of dosages of selenium supplementation in 448 older adults over 6 months [##REF##16181615##50##].</p>", "<title>Conclusion</title>", "<p>Evidence for selenium's effect on depressed mood in non-clinically depressed adults is inconsistent. Although one trial found an effect, the larger and better designed study did not. There is no evidence on the effects of selenium on depressive disorders.</p>", "<title>St John's wort (<italic>Hypericum perforatum</italic>)</title>", "<title>Description and rationale</title>", "<p>St John's wort is a traditional herbal remedy for depression. It is widely available as a supplement from health food shops, pharmacies and supermarkets. The most important active compounds are believed to be hypericin and hyperforin, but other compounds may also play a role. How it works is still not entirely clear, however it may inhibit the uptake of serotonin, norepinephrine, and dopamine [##REF##17105697##51##].</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>Several systematic reviews and meta-analyses examining St John's wort for depression have been published in recent times. A systematic review of these reviews [##REF##17105697##51##] concluded that although review methodologies have varied, St John's wort has consistently been found to be beneficial for mild to moderate depression compared to placebo, although the degree of benefit has varied between reviews. Comparisons against antidepressants have usually found no difference in benefit. The most recent Cochrane review of St John's wort for depression, published after the above mentioned review was completed, paints a more complex picture [##REF##15846605##52##]. The review was restricted to double blind RCTs of at least 4 weeks duration in adults with depressive disorders. A total of 37 trials involving 4,925 participants met inclusion criteria, and the majority were judged reasonable to good quality. Pooled results from 24 trials found that St John's wort was overall superior to placebo (response rate ratio 1.55, 95% confidence interval (CI) 1.42 to 1.70), and pooled results from 13 trials found no difference between St John's wort and older or newer antidepressants (response rate ratio 1.01, 95% CI 0.93 to 1.10). However, results from the studies comparing St John's wort to placebo were heterogeneous, with metaregression analyses leading to the conclusion that St John's wort showed greater benefits for individuals with mild depression. A variety of preparations of St John's wort were used and daily doses ranged from 240 mg to 1,800 mg. St John's wort caused fewer negative side effects than older antidepressants, and may have caused slightly fewer negative side effects than newer antidepressants. Use of St John's wort is not without risk however, as it has the potential to make other medications (such as immune suppressants, oral contraceptives and anticoagulants) less effective by increasing their rate of metabolism, and can also interact with selective serotonin reuptake inhibitors to cause a toxic reaction [##REF##17105697##51##].</p>", "<title>Conclusion</title>", "<p>St John's wort for depressive disorders has been well researched and there is evidence that it is helpful for mild depression. Consumers should be aware of risks involved when taken with other medications, and the possibility of variable quality of extracts in different brands and batches. There is no evidence on the effects of St John's wort in non-clinically depressed people.</p>", "<title>Vitamins</title>", "<title>Description and rationale</title>", "<p>Vitamins may play a role in depression or depressed mood because the brain depends on a constant supply to function effectively, and subclinical deficiencies are relatively common [##REF##12042457##53##]. Thiamine is required for the synthesis of acetylcholine. Vitamin B₆is a cofactor for the decarboxylases involved in the synthesis of neurotransmitters GABA, dopamine, norepinephrine, serotonin and histamine [##UREF##6##54##]. Folic acid and vitamin B₁₂are coenzymes for catechol-O-methyl transferase important in the breakdown of catecholamines. Vitamin C is necessary for the synthesis of dopamine and norepinephrine [##REF##7477807##55##]. As vitamin D levels decrease during winter due to reduced sunlight exposure, low levels of vitamin D may play a role in winter depression (seasonal affective disorder).</p>", "<title>Review of efficacy</title>", "<title>Vitamin B₁(thiamine)</title>", "<title>Non-clinically depressed</title>", "<p>A double-blind RCT in 117 healthy young adult females of 50 mg thiamine or placebo daily for 2 months found that supplementation had no effect on depressed mood [##REF##9122365##56##].</p>", "<title>Vitamin B₆</title>", "<title>Depressive disorders</title>", "<p>Although a systematic review has examined vitamin B₆for depression [##REF##15964874##57##], all trials evaluated vitamin B₆in combination with another treatment or used it only with hormone-related depression.</p>", "<title>Non-clinically depressed</title>", "<p>A single double-blind RCT has been carried out in 211 young, middle-aged and older female adults of 75 mg vitamin B₆, 750 μg folate, 15 μg vitamin B₁₂or placebo for 5 weeks. Vitamin B₆supplementation had no effect on depression symptoms or depressed mood [##REF##12042457##53##].</p>", "<title>Vitamin B₁₂</title>", "<title>Non-clinically depressed</title>", "<p>Two double-blind RCTs have tested the effect of supplementation of vitamin B₁₂on depression symptoms in healthy adults. A weekly injection of 1 mg B₁₂for 4 weeks in 134 elderly adults who showed signs of vitamin deficiency did not reduce depression symptoms significantly more than placebo [##REF##15337331##58##]. Similarly, B₁₂had no effect on depression symptoms or depressed mood when taken daily in a dose of 15 μg for 5 weeks in a double-blind RCT of 211 young, middle-aged and older female adults [##REF##12042457##53##].</p>", "<title>Folate</title>", "<title>Depressive disorders</title>", "<p>Although a systematic review examined folate for depression [##REF##12804463##59##] no RCTs were included that examined folate on its own as a treatment for people who were depressed without other medical conditions.</p>", "<title>Non-clinically depressed</title>", "<p>A double-blind RCT in 211 young, middle-aged and older female adults of 750 μg folate, 15 μg vitamin B₁₂, 75 mg vitamin B₆or placebo for 5 weeks found folate supplementation had no effect on depression symptoms or depressed mood [##REF##12042457##53##].</p>", "<title>Vitamin C</title>", "<title>Non-clinically depressed</title>", "<p>A double-blind RCT in 81 healthy young adults who took 3,000 mg sustained-release vitamin C or placebo for 14 days found depression symptoms significantly decreased in the vitamin C but not the placebo group [##REF##12208645##60##]. However, the decrease was very small.</p>", "<title>Vitamin D</title>", "<title>Non-clinically depressed</title>", "<p>Three RCTs have examined vitamin D supplementation in healthy adults. In all, 250 middle-aged and older adult females took 377 mg calcium plus 400 IU vitamin D daily, or 377 mg calcium on its own for a year [##REF##8140183##61##]. Depressed mood was assessed four times over the year, with vitamin D showing no effect. A 5-day trial in 44 adults of either vitamin D (400 IU or 800 IU) plus vitamin A (9,000 IU or 8,000 IU) versus 10,000 IU vitamin A only, found that vitamin D improved positive mood, but did not change negative mood [##REF##9539254##62##]. Finally, a large 6-month trial of 2,117 women aged over 70 years compared supplementation with vitamin D (800 IU) plus calcium (1,000 mg) with no supplementation. No significant difference was found in depressed mood between the two groups [##REF##16554952##63##].</p>", "<title>Multivitamins</title>", "<title>Non-clinically depressed</title>", "<p>Seven double-blind RCTs have examined the effects of multivitamin supplementation on depressed mood or symptoms in healthy non-depressed adults. A total of 120 young adults took a placebo or a multivitamin that contained 10 times the US recommended daily amount except for vitamin A (3,334 IU vitamin A, 14 mg B₁, 16 mg B₂, 180 mg B₃, 22 mg B₆, 2 mg B₇, 0.03 mg B₁₂, 600 mg vitamin C, 100 mg vitamin E and 4 mg folate), for 12 months [##REF##7477807##55##]. Supplementation had no effect on depressed mood after 3 or 12 months. A similar trial in 126 older adults of supplementation of the same multivitamin combination and dosage for 24 weeks also had no effect on depressed mood [##UREF##6##54##]. A trial in 95 adults of Pharmaton capsules (a supplement containing vitamins, minerals, trace elements and ginseng) for 8 weeks showed no effect on depressed mood [##UREF##7##64##]. A larger follow-up trial in 313 adults of the same supplement for 8 weeks also showed no effect on depressed mood. However, a subgroup analysis found that participants who were dieting had a greater improvement in depressed mood if they were taking the supplement than if they were taking the placebo [##UREF##8##65##]. A trial in 77 adult males of Berocca Performance supplementation (containing vitamins B₁, B₂, B₃, B₅, B₆, B₇, B₁₂, folate, C, and calcium, magnesium, zinc) for 28 days found that Berocca was no better than placebo at reducing depression symptoms [##REF##10907676##66##]. Finally, a trial of antioxidant supplementation (consisting of 12 mg/day β-carotene, 400 mg/day α-tocopherol, and 500 mg/day vitamin C) in 185 older adults for 12 months also showed no effect on depressed mood [##UREF##9##67##].</p>", "<title>Conclusion</title>", "<p>The limited evidence suggests that thiamine, vitamin B₆, vitamin B₁₂and folate supplementation are not helpful for depressed mood or symptoms in non-clinically depressed individuals. The evidence for vitamin D in non-clinically depressed individuals is inconsistent, but the larger, longer trials suggest it is not helpful. The evidence is more conclusive that multivitamins are not helpful for depressed mood in non-clinically depressed people. However, limited evidence suggests that vitamin C may be helpful in non-clinically depressed individuals, but these results require replication.</p>", "<title>Substances</title>", "<title>Caffeine</title>", "<title>Description and rationale</title>", "<p>Caffeine is a central nervous system stimulant that blocks adenosine receptors, which causes an increase in the levels of several neurotransmitters including dopamine and serotonin [##REF##15732884##68##]. Caffeine consumption is associated with depression symptoms. This may be because depressed individuals self-treat with caffeine [##UREF##10##69##]. However, large doses can produce anxiety symptoms.</p>", "<title>Review of efficacy</title>", "<title>Non-clinically depressed</title>", "<p>A review of studies, including several RCTs that evaluated caffeine consumption in healthy adults generally concluded that caffeine temporarily improves feelings of wellbeing, energy and mood [##UREF##10##69##]. However, caffeine use is widespread, study participants are typically not allowed caffeine before the experiment, and withdrawal from caffeine often involves depressed mood [##REF##15448977##70##]. Therefore some argue that the mood benefits are due to a reversal of withdrawal symptoms [##REF##16001109##71##]. Others disagree with this interpretation and argue that positive effects of caffeine on mood have been found when participants were not in caffeine withdrawal [##UREF##10##69##].</p>", "<title>Conclusion</title>", "<p>Although consumption of caffeine appears to improve depressed mood in non-clinically depressed individuals, it is still unclear whether this is caused by a reversal of withdrawal symptoms or is a true effect. There is no evidence on the effects of caffeine on depressive disorders.</p>", "<title>Dietary methods</title>", "<title>Carbohydrate-rich, protein-poor meals</title>", "<title>Description and rationale</title>", "<p>It has been suggested that a meal rich in carbohydrates but low in protein lifts mood, and that some depressed people (particularly those with seasonal affective disorder) could increase their carbohydrate intake in order to relieve depressive symptoms. The proposed mechanism is that a meal almost exclusively carbohydrate increases the level of tryptophan transported into the brain, where it is then converted into serotonin. However, most high-carbohydrate meals contain sufficient protein to block this mechanism [##REF##12034132##72##].</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>A crossover trial has compared the effects on depressed mood over hours of a carbohydrate-rich, protein-poor meal with a protein-rich, carbohydrate-poor meal in 16 adults with seasonal affective disorder and 16 non-depressed adults [##REF##2720016##73##]. Participants ate each meal on separate days, with the order of meals randomised. Results are difficult to interpret due to order effects. Both types of meals reduced depressed mood when eaten first, but when they were eaten second, the carbohydrate-rich meal decreased depressed mood while the protein-rich meal increased it.</p>", "<title>Non-clinically depressed</title>", "<p>Three RCTs have examined the effect over minutes or hours of a carbohydrate-rich, protein-poor meal on depressed mood. One trial found depressed mood decreased across all participants after a carbohydrate-rich meal [##UREF##11##74##], one trial found depressed mood did not increase under stressful conditions in high-stress prone individuals after the consumption of a carbohydrate-rich meal compared with a protein-rich meal [##REF##11006432##75##], and one trial found no significant difference in depressed mood between a carbohydrate-rich and a protein-rich meal [##UREF##12##76##]. Studies varied in the type of participants (young adults, older adults, obese adults), time of day when meal was eaten (lunch time, early or mid afternoon), type of meal (such as cake or liquid) and the carbohydrate, fat and protein proportions of meals classified carbohydrate-rich and protein-rich.</p>", "<title>Conclusion</title>", "<p>Studies have varied methodologies and inconsistent results, making it difficult to determine whether a carbohydrate-rich, protein-poor meal improves depressed mood in people with or without a depressive disorder. Given that the proposed mechanism is unlikely to account for any effect, another mechanism, such as palatability, may be behind any effects found. In any case, the strategy would only be helpful for short-term use, as a diet low in protein would reduce the dietary source of tryptophan.</p>", "<title>Psychological methods</title>", "<title>Autogenic training</title>", "<title>Description and rationale</title>", "<p>Autogenic training is the regular practice of simple mental exercises in body awareness which aim to promote relaxation and stress relief. The exercises involve passive concentration on breathing, heartbeat and warmth and heaviness of body parts. Books and websites that teach autogenic training are available. Autogenic training may be helpful for depression because it aims to teach self-regulation of autonomic nervous system processes.</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>A quasi-RCT compared autogenic training with psychotherapy and delayed treatment in 55 adults with depressive disorders [##UREF##13##77##]. Participants undertook weekly group autogenic training sessions plus twice-daily practice or weekly individual psychotherapy for 10 weeks. Depression symptoms in the autogenic training group improved significantly more than in the delayed-treatment control group, but significantly less than in the psychotherapy group.</p>", "<title>Non-clinically depressed</title>", "<p>One RCT allocated 134 adults with minor psychological problems to 3 months of individual autogenic training with a therapist plus twice-daily practice or a delayed-treatment group [##UREF##14##78##]. The autogenic training group had significantly improved depressed mood after 3 months, whereas the control group showed no change.</p>", "<title>Conclusion</title>", "<p>Preliminary evidence for autogenic training appears promising. However, these results have been achieved under the guidance of a therapist, and the helpfulness of self-taught autogenic therapy has not been evaluated.</p>", "<title>Bibliotherapy</title>", "<title>Description and rationale</title>", "<p>Bibliotherapy is a form of self help that uses structured written materials, such as books. The books present a treatment program, usually based on cognitive behaviour therapy, which encourage the reader to make changes leading to improved self-management. Two self-help books for depression that have been evaluated in trials and are available in bookstores are <italic>Feeling good </italic>[##UREF##15##79##] and <italic>Control your depression </italic>[##UREF##16##80##]. Other similar books that have not been evaluated specifically, but may be helpful [##REF##15904559##81##], are <italic>Mind over mood </italic>[##UREF##17##82##], <italic>Overcoming depression </italic>[##UREF##18##83##], and <italic>Overcoming depression: a five areas approach </italic>[##UREF##19##84##].</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>Several meta-analyses have evaluated the helpfulness of bibliotherapy for depression. A recent meta-analysis pooled results from 17 trials (16 RCTs) which compared bibliotherapy with a delayed treatment control group [##UREF##20##85##]. Trial participants varied in age from adolescents to the elderly and usually had mild to moderate depression without other physical or mental health problems. Trials lasted for 7 weeks on average. The meta-analysis found bibliotherapy more effective than controls (d = 0.77, 95% CI 0.61 to 0.94). Another meta-analysis of six RCTs that used the book <italic>Feeling good </italic>also found a large difference in depression over 4 weeks in favour of bibliotherapy over delayed treatment (standardised mean difference = -1.36, 95% CI -1.76 to -0.96) [##REF##15904559##81##]. However, the trials were small and of limited quality. An earlier meta-analysis of four RCTs, which compared bibliotherapy with individual therapy, found no significant difference in depression [##REF##9194011##86##]. The trials used different kinds of bibliotherapy, had small samples, and lasted between 6 and 11 weeks.</p>", "<title>Conclusion</title>", "<p>Evidence suggests that bibliotherapy is helpful for depressive disorders. However a number of caveats should be noted. The trials have not evaluated the use of bibliotherapy in the absence of any professional involvement. Also, not everyone may benefit from bibliotherapy; there are those who may lack the concentration or motivation required, have insufficient reading skills, or not be suited for personality reasons. There is no evidence on the effects of bibliotherapy in non-clinically depressed people.</p>", "<title>Computerised interventions</title>", "<title>Description and rationale</title>", "<p>Computerised interventions consist of the presentation of information via the internet or computerised cognitive behaviour therapy (CBT), which is the provision of structured sessions of CBT via computer. The delivery method can be over the internet or via interactive CD-ROM, and the level of professional involvement can vary from none to substantial. Although some computerised CBT packages are only available through a health professional, there are some which are freely available on the internet [##UREF##21##87##, ####UREF##22##88##, ##UREF##23##89##, ##UREF##24##90####24##90##].</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>A meta-analysis of 5 RCTs examined the effects of internet-based CBT on depression over weeks or months in a total of 1,982 adults recruited from a mix of clinical and community sources [##REF##17112400##91##]. The meta-analysis showed an overall small difference in depression between the internet CBT and control groups (fixed effects analysis d = 0.27, 95% CI 0.15 to 0.40; mixed effects analysis d = 0.32, 95% CI 0.08 to 0.57). The trials were of reasonable to good quality and had no professional involvement in four. Similarly, another review of eight RCTs found that computerised CBT without professional involvement had a small effect on depression, but that computerised CBT with professional involvement had a bigger effect, similar to that achieved in face to face CBT [##UREF##25##92##]. The author proposed that the smaller effect on depression of unsupervised computerised CBT could be due to low completion rates caused by the absence of a motivating professional. Only one RCT has examined the use of a depression information website [##REF##14742346##93##]. This intervention was found to produce significantly greater change in depression than a control condition and was not significantly different from web-based CBT.</p>", "<title>Non-clinically depressed</title>", "<p>A controlled trial in 59 adolescent males of MoodGYM, an internet-based CBT program, compared 5 weekly sessions of in-class use of MoodGYM with the usual personal development class scheduled at that time [##REF##16500776##94##]. There was no significant difference in change in depression symptoms between the two groups. However, compliance was low, with only 40% completing at least half of the MoodGYM program.</p>", "<title>Conclusion</title>", "<p>The evidence for computerised interventions for depressive disorders appears promising, particularly if a professional is involved. Pure self-help computerised CBT is not as helpful, but is a potentially beneficial option for those who are sufficiently motivated to complete the program on their own. There is insufficient evidence to determine the helpfulness of computerised interventions in those without depressive disorders.</p>", "<title>Distraction</title>", "<title>Description and rationale</title>", "<p>Distraction is directing attention away from the symptoms of depression and its possible causes and consequences and towards pleasant or neutral thoughts and actions. Response styles theory [##REF##1757671##95##] proposes that rumination in response to depressed mood worsens and prolongs it, whereas distraction reduces the intensity and duration of depressed mood. Depressed people tend to ruminate on their depression and depressed mood, in the belief that this will lead to greater understanding and better problem solving. However, ruminating whilst in a depressed mood is likely to lead to more negative thinking and make depression symptoms seem more prominent. Distraction may interfere with rumination and its distortions in thinking and allow better problem solving once the depressed mood has improved.</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>A number of experiments have been conducted on the effects of distraction on depressed mood over minutes, in both clinically depressed people and people with a high score on a depression symptom scale [##REF##8366423##96##, ####REF##7643299##97##, ##REF##9686457##98##, ##UREF##26##99##, ##UREF##27##100##, ##REF##15169614##101##, ##REF##15225341##102##, ##REF##16719984##103####16719984##103##]. A number of distraction tasks have been used, such as thinking about and visualising a series of neutral external things (for example, the shape of the African continent or the layout of a typical classroom), describing pictures, playing a board game, or thinking about broad social issues. Many of these experiments have compared distraction to a rumination task which involves focusing on current feelings and personal characteristics, such as 'your feelings right now and why you are feeling this way'. The generally consistent finding has been that rumination increases or maintains depressed mood, whereas distraction reduces depressed mood. The few studies that compared distraction with a control (such as sitting quietly or receiving no instructions from experimenters) also show that distraction is better at reducing depressed mood [##UREF##28##104##, ####REF##6697030##105##, ##REF##3602236##106##, ##REF##16797484##107####16797484##107##].</p>", "<title>Non-clinically depressed</title>", "<p>Other studies have experimentally induced depressed mood in non-clinically depressed participants before applying distraction [##REF##16797484##107##, ####UREF##29##108##, ##REF##2324941##109##, ##UREF##30##110##, ##REF##15122938##111##, ##UREF##31##112##, ##REF##17291694##113####17291694##113##] and have also typically found that distraction reduces depressed mood.</p>", "<title>Conclusion</title>", "<p>There is good evidence that distraction (in the form of thinking or visualising pleasant or neutral thoughts) is helpful for temporarily alleviating depressed mood, particularly if the alternative is ruminating on the causes and consequences of it. Other strategies may be required once the mood has lifted to prevent the recurrence of depressed mood.</p>", "<title>Meditation</title>", "<title>Description and rationale</title>", "<p>Meditation refers to a variety of self-regulation practices that focus on training attention and awareness. Different forms may emphasise concentration on something (such as an inner sound or the breath) as in transcendental meditation, or awareness of thoughts without judgement, as in mindfulness meditation or vipassana. Although meditation is often undertaken to achieve spiritual or religious goals, this is not a requirement of practice, and it has even been combined with Western treatments, such as mindfulness-based stress reduction, and mindfulness-based cognitive therapy. Meditation aims to reduce anxiety and promote relaxation. Additionally, mindfulness meditation may be helpful for depression because it leads to a distancing of self from negative thoughts and reduces rumination.</p>", "<title>Review of efficacy</title>", "<title>Non-clinically depressed</title>", "<p>Five RCTs have evaluated the effects of meditation on depressed mood or symptoms in non-clinically depressed individuals, with inconsistent results. An RCT in 73 elderly of transcendental meditation versus other mental relaxation or concentration tasks or waitlist found no significant difference in depression between groups after 12 weeks [##REF##2693686##114##]. An RCT in 42 young adults that compared mindfulness meditation with guided visual imagery for 3 weeks found that neither intervention had an effect on depressed mood [##REF##17324679##115##]. In contrast to these two trials, three RCTs found an effect. An RCT in 150 adults who participated in a week-long Buddhist meditation retreat found that the meditation group had significantly reduced depression symptoms compared with the delayed treatment control group [##UREF##32##116##]. An RCT in 61 adults who were assigned to 1 of 2 meditation groups or a control group, found that those assigned to the group using an Indian Vedic mantra (hypothesised to be particularly helpful for depression) had a significantly greater reduction in depression symptoms after 28 days of meditating, compared with either the control group or the group using a mantra composed of meaningless Sanskrit syllables [##UREF##33##117##]. Finally, an RCT that induced a depressed mood in 177 young adults found that a short mindfulness meditation significantly improved mood more than a distraction or rumination task [##UREF##31##112##].</p>", "<title>Conclusion</title>", "<p>For non-clinically depressed individuals, the evidence for meditation is inconsistent, with some trials showing benefit and others not. There is no evidence on the effects of meditation on depressive disorders.</p>", "<title>Relaxation training</title>", "<title>Description and rationale</title>", "<p>This review concerns relaxation training based primarily on progressive muscle relaxation, which involves teaching a person to relax voluntarily by tensing and relaxing specific muscle groups. Relaxation training may be helpful for depression because it improves a person's ability to deal with anxiety, and anxiety may lead to depression.</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>Nine RCTs have evaluated progressive muscle relaxation training in adolescents or adults with depressive disorders [##REF##15715034##118##, ####REF##8970662##119##, ##UREF##34##120##, ##REF##389965##121##, ##REF##8559866##122##, ##REF##3534032##123##, ##REF##7073647##124##, ##REF##8894955##125##, ##REF##6737208##126####6737208##126##]. The number of participants undergoing relaxation training varied between 8 and 43, the number of relaxation sessions varied from 5 to 40, and the training was delivered by trained persons in 7 of these trials. An unpublished meta-analysis of four of these trials [##UREF##34##120##,##REF##3534032##123##,##REF##7073647##124##,##REF##6737208##126##] that compared relaxation training with wait-list or minimal treatment control groups found significantly lower depression scores overall after treatment in the relaxation group (d = -0.66, 95% CI -1.07 to -0.25). However, a meta-analysis of results from six trials [##UREF##34##120##, ####REF##389965##121##, ##REF##8559866##122##, ##REF##3534032##123##, ##REF##7073647##124##, ##REF##8894955##125####8894955##125##] comparing relaxation with psychological treatment found that relaxation was significantly less helpful in reducing depression than psychological therapy (d = 0.52, 95% CI 0.25 to 0.79).</p>", "<title>Non-clinically depressed</title>", "<p>Five RCTs have compared progressive muscle relaxation training with a placebo or non-treatment control in non-depressed individuals [##REF##3303097##127##, ####UREF##35##128##, ##UREF##36##129##, ##REF##6387631##130##, ##REF##7870559##131####7870559##131##]. All found that relaxation was no better than the control in reducing depression symptoms or depressed mood. These trials varied in the age of participants (from children to older adults), duration of the intervention (6 to 11 weeks), length of relaxation sessions (5 min to 1.5 h), and whether the training was administered in a group or by the participant at home.</p>", "<title>Conclusion</title>", "<p>Research suggests that progressive muscle relaxation training may be helpful for those with depressive disorders, although it may not be as helpful as psychological treatment. It does not appear beneficial for depression in non-clinically depressed individuals.</p>", "<title>Lifestyle changes</title>", "<title>Exercise</title>", "<title>Description and rationale</title>", "<p>The two main types of exercise are aerobic (exercises the heart and lungs, such as in jogging) or anaerobic (strengthens muscles, such as in weight training). The antidepressant mechanism is unclear. Proposed mechanisms include physiological factors, such as effects on sleep regulation or serotonin and endorphins. Proposed psychological mechanisms include the interruption of negative thoughts that may prolong or worsen depression, or an increase in perceived coping ability. Exercise is also incompatible with inactivity and withdrawal, which are common unhelpful coping strategies for depression [##UREF##37##132##].</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>The most recently published meta-analysis of exercise for depression restricted included trials to those with adults or older adults who were clinically depressed [##UREF##37##132##]. Results were pooled from 11 RCTs involving 513 participants that compared exercise to a control condition (waitlist, placebo, low-intensity exercise or health education). Exercise interventions varied in frequency from between two to four times weekly, in duration between 20 and 45 min, and in intensity between unspecified and 70–85% maximum heart rate, for up to 12 weeks. The meta-analysis found an overall very large difference in depression between the two groups, with exercise being more effective (d = 1.42, 95% CI 0.92 to 1.93). Preliminary subgroup analyses indicated that anaerobic exercise may be as effective as aerobic exercise. A systematic review examining exercise specifically in older adults found 5 RCTs involving 318 older adults with depression (diagnosed or high level of symptoms), varying between 6 and 16 weeks in duration [##REF##16676285##133##]. Compared to controls, depression symptoms were significantly lower in the exercise condition (both aerobic and anaerobic) in four of the five trials, although trials were not of high quality. Exercise has also been systematically reviewed as an intervention in children and young people with depressive disorders [##REF##16856055##134##]. Three small trials were found, involving 81 participants. These were of low to moderate quality and all indicated no significant difference in outcome between exercise and various control conditions. The authors concluded that the evidence base was too scarce to determine the effect of exercise on depression in children and young people.</p>", "<title>Non-clinically depressed</title>", "<p>A non-systematic review of trials evaluating the effect on depression symptoms of aerobic exercise in non-depressed adults found that older lower quality studies had mixed results, but more recent RCTs generally find no reduction in depression symptoms [##UREF##38##135##]. A systematic review of exercise for depression in non-depressed older adults found 5 RCTs involving 766 participants [##REF##16676285##133##]. Trials lasted between 12 weeks and 12 months. Three trials comparing aerobic exercise with control interventions had mixed results, whereas two trials comparing anaerobic exercise with controls found no significant difference in reduction of depression symptoms. A systematic review of exercise for depression in non-depressed young people found exercise more effective than no intervention in 5 low quality trials involving 145 participants [##REF##16856055##134##]. Conversely, there was no effect of exercise found in 2 low quality trials comparing exercise with low intensity exercise (182 participants) and 2 low to moderate quality trials comparing it to psychosocial interventions (161 participants). Researchers have also investigated the effects over minutes of a single session of exercise on depressed mood. A selective review found 17 trials in non-depressed adolescents or adults where a variety of exercise (aerobic dance, yoga, jogging, rock climbing, swimming, tai chi and walking), ranging in duration from 10 to 80 min, had made improvements to depressed mood in participants [##UREF##39##136##]. The review did not indicate whether trials were controlled. The authors noted that positive effects depend upon complex interactions between participant characteristics (such as whether they found the exercise enjoyable), exercise mode (such as whether it is competitive or non-competitive), and exercise practice conditions (such as intensity and duration).</p>", "<title>Conclusion</title>", "<p>There is good evidence that exercise is helpful for reducing depression symptoms in adults with depressive disorders. It also appears to be helpful for older adults with depressive disorders; however there is insufficient evidence to determine the helpfulness in children and young people. Results from studies in non-clinically depressed individuals are mixed, perhaps reflecting the reduced room for improvement in these individuals. However, there is some evidence that single sessions of exercise may improve depressed mood in non-clinically depressed individuals. Research has yet to clarify the most appropriate dose and type of exercise required for an effect.</p>", "<title>Humour</title>", "<title>Description and rationale</title>", "<p>Laughter has similar physiological effects as vigorous exercise, such as reducing stress hormones, relieving tension, and releasing endorphins into the brain [##UREF##40##137##,##UREF##41##138##]. Responding to a stressful situation with humour may also help depression by causing a shift in thinking, promoting objectivity and distance from the threat or problem [##UREF##40##137##,##UREF##41##138##].</p>", "<title>Review of efficacy</title>", "<title>Non-clinically depressed</title>", "<p>Two RCTs have examined humour's effect on depressed mood after exposure to stress or a negative mood induction. A trial with 38 young adults who underwent a depressed mood induction found that only listening to a humorous tape restored mood to pre-experimental levels, compared with a neutral tape or no tape [##REF##2377684##139##]. An RCT in 80 males found that those who produced a humorous narration instead of a serious narration to a stressful silent film had significantly lower levels of depressed mood, although the effect did not last beyond 15 min [##UREF##41##138##]. The only trial to evaluate effects lasting longer than minutes was an RCT in 61 nursing home residents that examined the effect of humorous weekly group sing-a-longs on depression symptoms [##UREF##42##140##]. Compared to residents in control homes who received no intervention, those in the sing-a-long groups had significantly reduced depression symptoms after 4 weeks on one measure, but not on another. However, it is not clear whether the humour or other aspects of the intervention (such as social interaction or singing) were responsible for the effect.</p>", "<title>Conclusion</title>", "<p>Limited evidence suggests that exposure to humour (such as by watching a humorous video) temporarily improves depressed mood. Longer-term effects have not been adequately investigated. There is no evidence on the effects of humour in depressive disorders.</p>", "<title>Pets</title>", "<title>Description and rationale</title>", "<p>Spending time with pets might improve relaxation levels in their owners, provide companionship and a buffer against loneliness, and strengthen a sense of responsibility and self respect.</p>", "<title>Review of efficacy</title>", "<title>Non-clinically depressed</title>", "<p>Two RCTs have examined the effects of live-in birds assigned individually to older adults. Half of 40 older adult residents of skilled rehabilitation units received a caged budgerigar in their room for 10 days [##REF##8839325##141##]. The group who received a bird had significantly reduced depression scores at the end of the trial, although the authors noted this result might have been caused by an increase in human visitors to see the bird. A larger, better designed trial also had a similar result. A total of 144 nursing home residents were given a canary, a plant or nothing to look after in their rooms for 3 months [##REF##16191447##142##]. Depression symptoms significantly improved in the group assigned canaries, but not in the other two groups.</p>", "<title>Conclusion</title>", "<p>Studies in non-clinically depressed elderly nursing home residents suggest a positive effect of live-in pets on depression symptoms. These results may not generalise to the broader population. There is no evidence on the effects of pets on depressive disorders.</p>", "<title>Pleasant activities</title>", "<title>Description and rationale</title>", "<p>Depressed people engage in pleasant activities less often and find fewer activities pleasant compared with other people. Increasing engagement in pleasant activities can be performed informally or included as part of activity scheduling in cognitive behaviour therapy (CBT). Increasing the frequency of pleasant activities is thought to improve depressed mood by increasing opportunities for the reinforcement of healthy (non-depressed) behaviour and countering avoidance, withdrawal and inactivity.</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>A meta-analysis of RCTs of activity scheduling for depression in adults found clear indications that it is effective [##REF##17184887##143##]. A total of 10 studies compared activity scheduling with a control (usually a delayed treatment) and there was an overall large difference in depression, favouring activity scheduling (d = 0.87, 95% CI 0.60 to 1.15). A total of 14 studies compared activity scheduling with other psychological treatments (usually cognitive therapy), and overall there was no difference in depression after treatment (d = 0.13, 95% CI -0.05 to 0.30). Trials were generally small and not of the highest quality. However, these trials only examined activity scheduling as a treatment from a professional. Other factors, such as the therapeutic relationship, ritual of the therapy, or even other treatment components in particular trials, such as social skills training, may play a role in treatment outcome. Therefore, it is difficult to generalise these findings to a self-help method of activity scheduling.</p>", "<title>Non-clinically depressed</title>", "<p>An RCT of 65 non-depressed young adults had 3 groups: a monitor only control group, who monitored their daily activities and mood; a behaviour group, who additionally increased the number of activities found pleasurable; and a cognitive/behaviour group, who in addition to increasing pleasurable activities, focused on the positive aspects of the pleasant activities and the benefits of participating in them [##REF##3958195##144##]. After 2 weeks, depression scores significantly decreased for the monitor group and the cognitive/behaviour group, but not for the behaviour group. This finding was interpreted as support for the view that cognitive processing is required in addition to activity scheduling for an antidepressant effect, but this does not account for the decrease in depression shown in the monitor group.</p>", "<title>Conclusion</title>", "<p>There is reasonably good evidence that professional treatment involving activity scheduling is helpful for depression. This effect may not be applicable to a depressed person who independently attempts to increase pleasant activities. The evidence for the helpfulness of activity scheduling is inconclusive in non-clinically depressed individuals.</p>", "<title>Prayer</title>", "<title>Description and rationale</title>", "<p>Prayer has traditionally been used in times of illness and is often used by the public to help cope with mental health problems.</p>", "<title>Review of efficacy</title>", "<title>Non-clinically depressed</title>", "<p>One RCT in 88 Christian adults found practicing the Jesus prayer ('Lord Jesus Christ, have mercy on me') for 10 min daily for 30 days lowered depression scores significantly more than a non-treatment control group [##UREF##43##145##].</p>", "<title>Conclusion</title>", "<p>Limited evidence suggests prayer may be helpful for depressive symptoms in Christians who are not clinically depressed. There is no evidence on the effects of prayer on depressive disorders.</p>", "<title>Qigong</title>", "<title>Description and rationale</title>", "<p>Qigong is a 3,000-year-old Chinese self-training method involving meditation, breathing exercises and body movements. Qigong regulates the flow of qi (energy) throughout the body, removing imbalances or blockages, which cause emotional disturbances or physical symptoms.</p>", "<title>Review of efficacy</title>", "<title>Non-clinically depressed</title>", "<p>One crossover trial with order randomised has evaluated the effects over minutes of qigong on depressed mood [##REF##17090803##146##]. A total of 15 older adults recruited from existing qigong classes participated in a session of both qigong and brisk walking. Level of depressed mood did not significantly change after either session.</p>", "<title>Conclusion</title>", "<p>There is insufficient evidence to determine whether qigong is helpful for depressed mood in non-clinically depressed individuals. There is no evidence on the effects of qigong on depressive disorders.</p>", "<title>Sleep deprivation</title>", "<title>Description and rationale</title>", "<p>Total sleep deprivation is staying awake for a whole night and the following day, without napping. Partial sleep deprivation is restricting sleep to either the early or latter part of the night and remaining awake for the remainder of the night. Although the antidepressant mechanism is poorly understood, many have been proposed, such as normalisation of metabolic activity within the limbic system, or effects on serotonin functioning [##REF##12531127##147##].</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>Reviews of the efficacy of sleep deprivation for depressive disorders conclude that about 60% of depressed individuals improve after sleep deprivation [##REF##12531127##147##,##REF##10459393##148##]. The degree of symptom change ranges from complete remission to worsening in a minority. The effect is delayed in some individuals who only respond following sleep the next day. The evidence is unclear, but partial sleep deprivation may be as effective as total sleep deprivation [##REF##12531127##147##]. Although the antidepressant effect is rapid, it typically does not last, with 50–80% of responders suffering a complete or partial relapse following recovery sleep. Researchers have attempted to prevent relapse with other treatments such as antidepressant drugs, shifting of sleep time or light therapy, which show promise for reducing the risk of relapse.</p>", "<title>Non-clinically depressed</title>", "<p>One RCT in 40 males found a significant increase in depressed mood after 24 h of wakefulness as compared to controls who had a typical night's sleep [##REF##759101##149##].</p>", "<title>Conclusion</title>", "<p>Evidence is consistent that sleep deprivation is helpful for many individuals with depressive disorders, although the effects are typically temporary. In non-clinically depressed individuals, sleep deprivation may cause an increase in depressed mood.</p>", "<title>Tai Chi</title>", "<title>Description and rationale</title>", "<p>Tai chi is a type of moving meditation that originated in China as a martial art. It involves slow, purposeful movements and focused breathing and attention. In traditional Chinese medicine, tai chi is thought to benefit health through the effects of stereotyped hand and foot movements on important acupoints and visceral channels [##REF##2724196##150##]. Tai chi could affect depression because it is a form of moderate exercise or because it is a relaxing distraction from anxiety and stress.</p>", "<title>Review of efficacy</title>", "<title>Non-clinically depressed</title>", "<p>Two RCTs have compared tai chi with different forms of exercise or relaxation. A total of 96 healthy adult tai chi practitioners participated in 1 h of tai chi, brisk walking, tai chi meditation or neutral reading after being subject to experimentally induced emotional or mental stress [##REF##1593511##151##]. All activities significantly improved depressed mood. Another trial compared a 16-week program of tai chi against low or moderate intensity walking, low intensity walking with relaxation, and no treatment in 135 adults [##REF##7674883##152##]. Depressed mood significantly improved in women in the tai chi group compared to those in the control, but changes in depressed mood in men did not differ significantly between the different groups.</p>", "<title>Conclusion</title>", "<p>There is insufficient evidence to determine whether tai chi is helpful for depressed mood in non-clinically depressed individuals. There is no evidence on the effects of tai chi on depressive disorders.</p>", "<title>Yoga</title>", "<title>Description and rationale</title>", "<p>Yoga exercises the mind and body through physical postures, breathing techniques and meditation. Each posture is held for a period of time and synchronised with breathing. Yoga is thought to relieve stress and improve relaxation, but it may also be effective due to feelings of mastery from learning difficult postures, or improvements in body image from greater bodily awareness and control.</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>A recent systematic review of randomised controlled trials of yoga for depression found five trials to review [##REF##16185770##153##]. The studies varied in the type of yoga studied (such as Iyengar, Shavasana, and Sudarshan Kriya Yoga), severity of depression (mild to severe), number of participants per yoga group (10–25), and length of intervention (3 days to 5 weeks), and participants were all under 50 years. Nevertheless, the authors concluded that yoga for depressive disorders is potentially beneficial, but that further investigation is needed.</p>", "<title>Non-clinically depressed</title>", "<p>Three RCTs have evaluated the effects over months of yoga classes on depressive symptoms or depressed mood, with inconsistent results. Two trials in older adults of 60–90 min yoga sessions once or twice a week for 16 weeks to 6 months found no effect on depressed mood or symptoms relative to controls [##REF##2768768##154##,##REF##16454146##155##]. However one shorter trial of 6 weeks in adults found depressed mood significantly improved in the yoga group compared with a wait-list control group [##UREF##44##156##].</p>", "<title>Conclusion</title>", "<p>Initial evidence suggests that yoga may be beneficial for depressive disorders. The evidence is inconsistent for effects in non-clinically depressed individuals.</p>", "<title>Physical and sensory methods</title>", "<title>Aromatherapy</title>", "<title>Description and rationale</title>", "<p>Aromatherapy is the therapeutic use of essential oils, which are highly concentrated extracts of plants. Essential oils can be diluted in carrier oils and absorbed through the skin via massage, or heated and vaporised into the air. Essential oils said to have antidepressant effects include bergamot, geranium, jasmine, lavender and Egyptian rose [##REF##16599645##157##]. They are available from health food shops or pharmacies. The antidepressant mechanism is unclear, but may be due to the odour either being perceived as pleasant or triggering memories and emotions that affect mood. Alternatively, the oil's chemical constituents may be absorbed into the blood stream and have pharmacological effects [##REF##16599645##157##].</p>", "<title>Review of efficacy</title>", "<title>Non-clinically depressed</title>", "<p>One RCT has examined aromatherapy's effects over minutes on depressed mood in non-clinically depressed adults. A total of 73 adults were exposed to water or essential oils of lavender or rosemary for 10 min whilst they completed a stressful mental task. At the end of the task there was no significant difference in depressed mood between groups [##REF##15587240##158##].</p>", "<title>Conclusion</title>", "<p>There is insufficient evidence to determine whether aromatherapy is helpful for depressed mood in non-clinically depressed individuals. There is no evidence on the effects of aromatherapy in depressive disorders.</p>", "<title>Hydrotherapy</title>", "<title>Description and rationale</title>", "<p>Hydrotherapy includes hot air and steam baths or saunas, wet packings, and various kinds of warm and cold baths [##REF##12779182##159##]. Hydrotherapy was a popular historical treatment for depression and was thought to promote relaxation [##REF##12779182##159##].</p>", "<title>Review of efficacy</title>", "<title>Non-clinically depressed</title>", "<p>An RCT in 40 adults found no effect on depressed mood of a 10-min immersion in a spa bath with either the whirlpool motor on or off [##REF##2348454##160##].</p>", "<title>Conclusion</title>", "<p>Limited evidence suggests that hydrotherapy is not effective for the relief of depression symptoms or depressed mood. There is no evidence on the effects of hydrotherapy on depressive disorders.</p>", "<title>Light therapy</title>", "<title>Description and rationale</title>", "<p>Light therapy is exposure of the eyes to bright light for an appropriate duration, often in the morning. The light is emitted from a box or lamp which the person sits in front of. Several manufacturers make their own versions of light therapy devices, some of which have not been evaluated in clinical trials. These can be bought over the internet. Different devices may use different parts of the light spectrum, at different intensities of illumination. Light therapy was originally used to treat seasonal affective disorder (SAD), by advancing the phase delay in circadian rhythms caused by exposure to less sunlight in winter. It has now been extended to treat non-seasonal depression and therefore the phase advance is probably not the only mechanism [##REF##16041296##161##].</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>Two recent meta-analyses have been carried out. The first examined light therapy for depression and SAD in non-geriatric adults over days or weeks [##REF##15800134##162##]. Included studies were RCTs of a reasonably high standard, with light therapy groups receiving adequate doses of light exposure. Three studies were included in the meta-analysis of light therapy for depression, eight for light therapy for SAD, and five for dawn simulation for SAD. The meta-analyses revealed a large effect of light therapy (d = 0.84, 95% CI 0.60 to 1.08) and dawn simulation on SAD (d = 0.73, 95% CI 0.37 to 1.08), compared with placebo, and a medium-sized effect of light therapy on depression (d = 0.53, 95% CI 0.18 to 0.89). Another meta-analysis looked exclusively at light therapy for non-seasonal depression and applied broader inclusion criteria for trials [##REF##15106233##163##]. A total of 18 RCTs were analysed, however only 2 used light therapy as the only treatment. Pooling results from these two studies showed that light therapy was beneficial when using a fixed effects analysis, but the result did not reach significance using a random effects model. A systematic review of light therapy for depressed children and adolescents also concluded that limited evidence suggests it is helpful for winter depression, but not for non-seasonal depression [##REF##17014404##19##].</p>", "<title>Non-clinically depressed</title>", "<p>Four RCTs have evaluated light therapy in non-clinically depressed individuals over days or weeks, with mixed findings. Trials with participants who experienced winter difficulties (subsyndromal SAD) found light therapy helpful for depressed mood or symptoms [##REF##2774849##164##,##REF##2139066##165##]. Light therapy doses were 2,500 lux for 2–5 h in the morning or split over both morning and evening, Trials with participants who had no winter difficulties generally did not find light therapy helpful [##REF##2774849##164##, ####REF##2139066##165##, ##REF##8516421##166##, ##REF##16756690##167####16756690##167##] and some of these participants experienced negative effects, such as irritability, after light therapy [##REF##8516421##166##].</p>", "<title>Conclusion</title>", "<p>There is good evidence that light therapy is effective for SAD (winter depression). It also appears to be helpful for non-seasonal depressive disorder, but the evidence is not as strong and the effect is smaller. It may also be helpful for non-clinically depressed individuals who experience mild symptoms of SAD.</p>", "<title>Massage</title>", "<title>Description and rationale</title>", "<p>Massage is thought to work by stimulating vagal activity, leading to a reduction of stress hormones and physiological arousal; or by influencing body chemistry, such as increasing serotonin or releasing endorphins [##REF##14717648##168##].</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>Two RCTs have evaluated the effects over minutes of massage in people with depressive disorders. One trial was of a 30-min massage in children or adolescents with dysthymia [##REF##1537763##169##], and one was of a 20-min massage in depressed pregnant women [##REF##15715034##118##]. Both trials compared massage with a form of relaxation and found massage significantly reduced depressed mood compared with relaxation. These trials also evaluated the effects of multiple doses of massage. Changes in depressed mood or symptoms were examined over the duration of each trial (lasting 5 days and 16 weeks respectively), with both finding a significant reduction in the massage group compared with controls.</p>", "<title>Non-clinically depressed</title>", "<p>Four RCTs have evaluated the effects over minutes of massage on non-depressed adults compared to some control intervention. The results have been variable. One RCT found massage produced greater effects than reading [##REF##14501546##170##], however two RCTs found no significant difference from relaxation [##REF##8884390##171##,##REF##9210776##172##] and one found no difference from resting [##UREF##45##173##]. Massages were restricted to the upper body and were 10 to 30 min long. One RCT has evaluated multiple doses of massage, but only against relaxation therapy. A total of 50 adults were given a 15-min massage or were told to tighten and relax their muscles twice a week for 5 weeks. Depressed mood in both the massage group and the relaxation control group significantly improved [##REF##8884390##171##].</p>", "<title>Conclusion</title>", "<p>Preliminary evidence suggests massage may have immediate and longer-term effects on depressed mood and symptoms in those with depressive disorders. The evidence for immediate and longer-term effects of massage in those who are not depressed is inconsistent. It should be noted that in virtually all studies massage was given by trained massage therapists, and the effects of massage given by self or non-trained professionals has not been evaluated.</p>", "<title>Music</title>", "<title>Description and rationale</title>", "<p>Music has been called the 'language of emotions' and appears to activate emotional systems in the brain. It is unclear to what degree the emotional impact is caused by specific attributes of the music itself (such as rhythm and melody), or cultural context and memories [##REF##12426066##174##].</p>", "<title>Review of efficacy</title>", "<title>Non-clinically depressed</title>", "<p>Listening to music (both classical and modern) has frequently been used in experimental settings to induce particular moods in participants. A meta-analysis of these studies concluded that music is a moderately effective method of inducing temporary depressed or elated mood in experimental settings with non-depressed individuals [##UREF##46##175##]. Only one RCT has examined the effects of listening to music over weeks [##UREF##47##176##]. A total of 102 adult female nurses participated in a 6-week trial where they were instructed to listen to 20 min of music self-selected to be stress reducing three times a week, perform 20 min of self-selected aerobic exercise three times a week or maintain their usual exercise and stress-reduction activities. Listening to music did not reduce depression symptoms significantly more than the control group.</p>", "<title>Conclusion</title>", "<p>Listening to music can be an effective method of lifting mood in the short term (less than an hour) in non-clinically depressed individuals, but there is no evidence that music can reduce depression over periods of days or weeks. There is no evidence on the effects of music on depressive disorders.</p>", "<title>Negative air ionisation</title>", "<title>Description and rationale</title>", "<p>A negative air ioniser is a device that uses high voltage to electrically charge air particles. The antidepressant mechanism is unknown but may involve effects on both central and peripheral serotonergic activity [##REF##16045061##177##].</p>", "<title>Review of efficacy</title>", "<title>Depressive disorders</title>", "<p>One RCT in participants with seasonal affective disorder found significantly greater reduction in depression symptoms over weeks with 30 min daily exposure to high density negative ionisation (4.5 × 10¹³/s flow rate) compared with a placebo of low density negative ionisation [##REF##9783557##178##].</p>", "<title>Non-clinically depressed</title>", "<p>One RCT [##REF##16756690##167##] compared exposure to high density (4.5 × 10¹⁴/s) negative ion generators, bright light, music or low density (1.7 × 10¹¹/s) negative ion generators (placebo) for 30 min in 118 non-depressed adults. The high density negative ions significantly decreased depressed mood compared with placebo. Another study of depressed mood in non-depressed adults found that exposure to high concentrations of negative ions significantly decreased depressed mood compared with exposure to low concentrations of negative ions. However, depressed mood was significantly increased in those who were experimentally manipulated to feel angry and exposed to high negative ion concentrations, compared with those who were also experimentally manipulated to feel angry but were exposed to low negative ion concentrations [##REF##3989668##179##].</p>", "<title>Conclusion</title>", "<p>A small number of studies suggest exposure to high density negative ions (at least 2.7 × 10⁶/cm³) is helpful for seasonal affective disorder and depressed mood in non-clinically depressed individuals.</p>", "<title>Singing</title>", "<title>Description and rationale</title>", "<p>Music elicits strong emotional responses in humans, however singing may also improve mood through changes in breathing patterns, the expression of emotion, or through the content of lyrics [##UREF##48##180##].</p>", "<title>Review of efficacy</title>", "<title>Non-clinically depressed</title>", "<p>Effects over weeks were examined in an RCT involving 61 nursing home residents who participated in humorous weekly group sing-a-longs [##UREF##42##140##]. Compared to residents in control homes who received no intervention, those in the sing-a-long groups had significantly reduced depression symptoms after 4 weeks on one measure, but not on another. However, it is not clear whether the singing or other aspects of the intervention (such as humour or social interaction) were responsible for the effect.</p>", "<title>Conclusion</title>", "<p>Limited research suggests that group singing may improve depressed mood or depression symptoms in non-clinically depressed individuals, however these results require replication. The effects of singing have not been examined in individuals with depressive disorders.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>AM contributed to aspects of study design, carried out the literature searches, and drafted the reviews and manuscript. AJ conceived the study, contributed to the reviews and helped to draft the manuscript. Both authors read and approved the final manuscript.</p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>Claire Kelly provided feedback on the reviews. Funding was provided by the National Health and Medical Research Council and the Colonial Foundation. These funding sources had no further role in study design; in the collection, analysis, and interpretation of data; in the writing of the manuscript; and in the decision to submit the manuscript for publication.</p>" ]

[]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Self-help methods with no relevant trials to review</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Category</bold></td><td align=\"left\"><bold>Treatment</bold></td></tr></thead><tbody><tr><td align=\"left\">Medicines/herbs/dietary supplements</td><td align=\"left\">5-hydroxytryptophan, American ginseng (<italic>Panax quinquefolius</italic>), ashwaganda (<italic>Withania somnifera</italic>), astragalus (<italic>Astragalus membranaceous</italic>), Bach flower remedies, basil (<italic>Ocimum spp</italic>.), black cohosh (<italic>Actaea racemosa </italic>and <italic>Cimicifuga racemosa</italic>), brahmi (<italic>Bacopa monniera</italic>), California poppy (<italic>Eschscholtzia californica</italic>), catnip (<italic>Nepeta cataria</italic>), cat's claw (<italic>Uncaria tomentose</italic>), chamomile (<italic>Anthemis nobilis</italic>), chaste tree berry (<italic>Vitex agnus castus</italic>), chocolate, choline, clove (<italic>Eugenia caryophyllata</italic>), coenzyme Q₁₀, combined preparations (Empowerplus (Truehope Nutritional Support Ltd.); euphytose; Mindsoothe Jr. (Native Remedies); Sedariston; Worry Free), cowslip (<italic>Primula veris</italic>), damiana (<italic>Turnera diffusa</italic>), dandelion (<italic>Taraxacum officinale</italic>), flax seeds (linseed) (<italic>Linum usitatissimum</italic>), Gamma-aminobutyric acid (GABA), ginger (<italic>Zingiber officinale</italic>), gotu kola (<italic>Centella asiatica</italic>), glutamine, hawthorn (<italic>Crataegus laevigata</italic>), hops (<italic>Humulus lupulus</italic>), hyssop (<italic>Hyssopus officinalis</italic>), inositol, Kava (<italic>Piper methysticum</italic>), lemon balm (<italic>Melissa officinalis</italic>), lemongrass leaves (<italic>Cymbopogon citrates</italic>), liquorice (<italic>Glycyrrhiza glabra</italic>), magnesium, milk thistle (<italic>Silybum marianum</italic>), mistletoe (<italic>Viscum album</italic>), motherwort (<italic>Leonurus cardiaca</italic>), natural progesterone, nettles (<italic>Urtica dioica</italic>), oats (<italic>Avena sativa</italic>), painkillers/over the counter medicines, para-aminobenzoic acid (PABA), passionflower (<italic>Passiflora incarnata</italic>), peppermint (<italic>Mentha piperita</italic>), phenylalanine, potassium, purslane (<italic>Portulaca oleracea</italic>), rehmannia (<italic>Rehmannia glutinosa</italic>), Rhodiola rosea, rosemary (<italic>Rosmarinus officinalis</italic>), sage (<italic>Salvia officinalis</italic>), schizandra (<italic>Schizandra chinensis</italic>), Siberian ginseng (<italic>Eleutherococcus senticosus</italic>), skullcap (<italic>Scutellaria lateriflora</italic>), spirulina (<italic>Arthrospira platensis</italic>), St Ignatius bean (<italic>Ignatia amara</italic>), taurine, tension tamer, thyme (<italic>Thymus vulgaris</italic>), tissue salts, tyrosine, valerian (<italic>Valeriana officinalis</italic>), vervain (<italic>Verbena officinalis</italic>), vitamin B₂, vitamin B₃, vitamin B₅, vitamin B₇, vitamin E, vitamin K, wild yam (<italic>Dioscorea villosa</italic>), wood betony (<italic>Stachys officinalis</italic>; <italic>Betonica officinalis</italic>), yeast, zinc, zizyphus (<italic>Zizyphus spinosa</italic>)</td></tr><tr><td align=\"left\">Dietary methods</td><td align=\"left\">Avoiding barley, rye, sugar, wheat, or dairy foods, ketogenic diet</td></tr><tr><td align=\"left\">Substances</td><td align=\"left\">Drinking or reducing alcohol consumption, using cannabis or quitting cannabis, smoking a cigarette or quitting smoking</td></tr><tr><td align=\"left\">Lifestyle changes</td><td align=\"left\">Adequate sleep, holiday or vacation, pilates, recreational dance, shopping</td></tr><tr><td align=\"left\">Physical and sensory methods</td><td align=\"left\">Crystal healing or charm stone, fragrance, reflexology</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional file 1</title><p><bold>Search strategy</bold>. Microsoft Word document of literature search strategy used.</p></caption></supplementary-material>" ]

[]

[]

[ "<media xlink:href=\"1744-859X-7-13-S1.doc\" mimetype=\"application\" mime-subtype=\"msword\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>The interaction between tuberculosis (TB) and human immunodeficiency virus (HIV) infection is complex. In the individual patient, HIV infection weakens the immune system and increases the susceptibility to TB. HIV increases the likelihood of reactivation, reinfection and progression of latent TB infection to active disease. It also alters the clinical presentation of TB, complicates the follow up and compromises the response to anti-TB treatment [##REF##11367039##1##].</p>", "<p>In a population, the lifetime risk of developing active TB once infected, in absence of HIV infection, is about 10% [##REF##11362673##2##]. However, it increases tenfold in HIV infected individuals. This has resulted in a large increase in the number of TB cases [##REF##16899146##3##,##REF##16870527##4##]. The proportion of smear-negative pulmonary TB (PTB) and extrapulmonary TB (EPTB) is higher among HIV co-infected TB patients [##UREF##0##5##].</p>", "<p>At TB control programme level, an increase in the TB burden leads to increased need of trained staff, diagnostic facilities and patient care. The number of smear positive PTB cases registered has been used as the basis for procurement and distribution of drugs and supplies [##UREF##1##6##]. However, changes in the proportion of smear negative PTB and EPTB due to HIV co-infection may require adjustments. In Ethiopia, ten per cent of HIV infected people require antiretroviral therapy and the need is more among TB patients co-infected with HIV [##UREF##2##7##]. Therefore, knowledge about the rate of HIV infection in TB patients might help in planning and resource allocation. Regular surveillance of HIV infection in TB patients and the general population would also help in understanding the spread of the dual infections and monitoring the performances of TB and HIV control activities [##UREF##3##8##,##UREF##4##9##].</p>", "<p>However, knowledge about the prevalence of HIV infection in the general population and its correlation with the rate of HIV infection in TB patients is limited in Ethiopia. The aim of this study was to determine the rate of HIV infection in TB patients and its correlation with the rate HIV infection in pregnant women attending antenatal care (ANC) in Southern Ethiopia.</p>" ]

[ "<title>Methods</title>", "<title>Study area and population</title>", "<p>This study was conducted in the Southern Nations, Nationalities and Peoples' Region (SNNPR) of Ethiopia. The region has 13 administrative zones and an estimated population of 14 million, of which 93% live in rural areas. Only half of the population live within two-hour walking distance from a public health institution. The Regional Health Bureau has adopted the World Health Organization recommended directly observed short course treatment strategy for TB control since 1995. The first round HIV survey among TB patients and pregnant women was conducted in 2002 [##REF##15370654##10##]. In this study, the number of surveillance sites was increased to include more urban and rural communities to represent all zones of the region.</p>", "<title>Study design and site selection</title>", "<p>This is a cross-sectional study carried out from September 2004 to April 2005.</p>", "<title>TB-HIV co-infection survey</title>", "<p>Health institutions were selected based on their capacity to diagnose and treat TB patients. The diagnostic services included direct sputum microscopy, routine blood tests and x-rays. Ten health institutions (Figure ##FIG##0##1##) were randomly selected. All TB patients were consecutively enrolled at their first visit to the treatment units.</p>", "<title>ANC – based HIV sentinel survey</title>", "<p>Health institutions that deliver ANC, had an adequate client volume, collect blood samples for routine tests such as haemoglobin determination and syphilis testing and facilities to maintain cold chain were identified of which twelve health institutions (Figure ##FIG##0##1##) were randomly selected. All pregnant women attending ANC were consecutively enrolled at their visit to health institutions [##UREF##5##11##].</p>", "<p>In both surveys, TB patients and pregnant women referred from other health institutions or coming for the second visit during the survey period were excluded to avoid repetition. In six of the study sites, both surveys were conducted in the same health institutions providing health service to TB patients and pregnant women from the same districts. However, in the remaining sites, the surveys were conducted in health institutions providing health service to the population in the nearby districts.</p>", "<title>Diagnosis of TB</title>", "<p>The diagnosis of TB was based on the recommendations of the National TB and Leprosy Control Programme [##UREF##1##6##]. Briefly, patients presenting with symptoms suggestive of PTB who had productive cough for three weeks or more with at least two positive sputum smears or one positive smear and x-ray findings consistent with active PTB were classified as smear-positive PTB cases. Patients presenting with cough of three weeks or more with initial three negative smears and no clinical response to a course of broad-spectrum antibiotics, three negative smear results after a course of broad-spectrum antibiotics, x-ray findings consistent with active PTB and decided by a clinician to be treated with anti-TB chemotherapy were classified as smear-negative PTB cases. Patients presenting with dry cough of three weeks or more were diagnosed based on strong clinical evidence and x-ray findings consistent with active TB. Patients presenting with symptoms suggestive of TB other than the lungs, which did not respond to a course of broad-spectrum antibiotics and decided by a clinician to be treated with anti-TB chemotherapy were classified as EPTB cases. In children, TB was diagnosed if there were symptoms and signs suggestive of TB, contact history with a known TB patient and x-ray findings consistent with active TB.</p>", "<title>Data and specimen collection</title>", "<p>Trained laboratory technicians and health workers from TB and ANC functions collected the data using pretested questionnaires. The main variables were age, sex, residence and survey site for all participants, and disease classification and category for TB patients. 5 ml of blood samples were collected from TB patients and pregnant women. Routine blood tests except for HIV were done locally and reported to the attending health workers. The remaining serum samples were stripped off individual identifying markers and were assigned unique codes. They were kept at 4°C, transported to the regional Centre for Health Research Laboratory (CHRL) and stored at – 20°C until analysis. The serum samples were anonymously tested for HIV using ELISA test (Vironostica ^®Uniform II Ag/Ab BIOMÉRIEUX). All the samples were sent to the Ethiopian Health and Nutrition Research Institute (EHNRI) to repeat ELISA test using Enzygnost Anti-HIV1/2 Plus (Dade Behring, Germany) and quality control. ELISA reactive specimens at CHRL and EHNRI were considered positive and discordant specimens were retested using similar tests [##UREF##5##11##,##UREF##6##12##].</p>", "<title>Data analysis</title>", "<p>We used SPSS 14.0 (SPSS Inc, Chicago, IL, USA) for data entry and analysis. We determined the rate of HIV infection in TB patients and pregnant women. Univariate and multivariate logistic regression analysis were used to determine the risk factors for HIV infection in TB patients and pregnant women. Socio-demographic variables that were significant by univariate analysis were included in the model to calculate adjusted odds ratio and 95% confidence interval by HIV status in TB patients. We also did linear regression analysis to determine the variation of HIV infection in TB patients explained by the prevalence of HIV infection among pregnant women from all study sites and then for the study participants from the same health institutions. P-value &lt; 0.05 was considered as statistically significant.</p>", "<title>Ethical clearance</title>", "<p>Ethical Review Committee of the Regional Health Bureau approved the study. Oral informed consent was obtained for all study participants. The study participants who wanted to know their HIV status were advised to go to voluntary counselling and testing service located within the health institutions or nearby.</p>" ]

[ "<title>Results</title>", "<p>1308 TB patients and 4199 pregnant women were included in the study. Of the TB patients, 729 (56%) were men and 569 (44%) were women. 309 (24%) patients came from urban and 978 (76%) patients from rural areas. Their mean age was 28.4 years. 544 (42%) patients had smear-positive PTB, 449 (34%) smear-negative PTB and 308 (24%) EPTB. The rate of HIV infection in TB patients was 18% (226/1261) [95%CI: 15.8–20.0] ranging from 8.3% (in Silte zone) to 35.3% (in South Omo zone). The rate of HIV infection in TB patients was similar for men and women (OR = 1.00, 95%CI: 0.75 – 1.34). There was no difference in the rate of HIV infection by TB disease classification: the rate of HIV infection among smear-positive PTB cases 17.5% (92/526) was similar to smear-negative PTB 18.1% (78/432) [OR = 1.048, 95%CI: 0.723–1.519] and EPTB cases 18.2% (54/297) [OR = 1.009, 95%CI: 0.687–1.480]. The rate of HIV infection was higher in TB patients from urban (24.5%, 73/298) than rural areas (15.8%, 149/945) [AOR = 1.78, 95%CI: 1.27–2.48] as shown in Table ##TAB##0##1## &amp;##TAB##2##3##.</p>", "<p>Of the 4199 pregnant women attending ANC, 3097 (74%) came from rural and 1096 (26%) from urban areas. Their mean age was 25.7 years. The prevalence of HIV infection among the pregnant women was 3.8% (155/4091) [95%CI: 3.2–4.4] ranging from 1.5% (in Gamo Goffa zone) to 10.5% (in Wolaita zone). The rate of HIV infection was higher among women from urban 7.5% (80/1066) than rural 2.5% (75/3025) areas [OR = 3.19, 95% CI: 2.31–4.41] (Table ##TAB##1##2## &amp;##TAB##2##3##).</p>", "<p>In all survey sites, where both surveys were conducted in the same as well as in different health institutions, we found no correlation between the rate of HIV infection among pregnant women and TB patients (R²= 0.034). Briefly, South Omo zone with the highest TB-HIV co-infection rate did not have higher rate of HIV infection among pregnant women whereas Silte zone that had the lowest rate of TB-HIV co-infection did not have the lowest rate of HIV infection among pregnant women (Table ##TAB##2##3##).</p>", "<p>In contrast, in the six study sites where the two surveys were conducted in the same health institutions, there was a strong correlation between the rate of HIV infection among pregnant women and TB patients (R²= 0.732). Upon further analysis by residence, the magnitude of correlation was stronger for study participants from urban (R²= 0.998) than rural areas (R²= 0.546) as shown in Table ##TAB##3##4## and Figure ##FIG##1##2##. From a linear regression analysis, we found the equation, prevalence of HIV among pregnant women = -6.22 + 0.89* the rate of HIV infection in TB patients. Each per cent increase of HIV seroprevalence in TB patients corresponded to an increase in seroprevalence of 0.89% among pregnant women.</p>" ]

[ "<title>Discussion</title>", "<p>In the recent decades, the number of TB cases has increased by several folds especially in sub-Saharan African countries. HIV infection is considered the main risk factor for the increase in the number TB patients and the proportion of smear-negative and EPTB cases [##REF##16899146##3##,##REF##15042183##14##,##REF##16045459##15##]. The information about the rate of HIV infection among different groups of a community is important to understand the extent of the problem and to implement appropriate prevention and control measures.</p>", "<p>In a large representative survey of TB patients in southern Ethiopia, less than a fifth of them were HIV infected similar to other reports from the region [##UREF##4##9##,##REF##9360249##16##]. Higher TB-HIV co-infection rates, as high as 47% was reported from Ethiopia [##UREF##8##17##,##UREF##9##18##]. These studies however were hospital-based and were conducted in few major towns where the prevalence of HIV infection in the general population was much higher.</p>", "<p>In our study, there was no difference in the rate of HIV infection among TB patients by gender, TB classification and category. Unlike several other studies which reported higher rates of HIV infection among smear-negative and EPTB cases compared to smear-positive cases [##REF##16899146##3##,##UREF##0##5##,##REF##15370654##10##], we did not find difference in the rate of HIV infection among different TB classifications. This could be due to the relatively low prevalence of HIV infection in the region [##UREF##7##13##]. Another possible explanation could be under diagnosis or referral of some smear-negative and EPTB suspects with a potentially higher risk of HIV infection due to limited diagnostic facilities.</p>", "<p>Although the ANC-based HIV sentinel surveillance has weaknesses as the results may be affected by low attendance of ANC, exclusion of private clinics, the rate of contraceptive use and provides no information about men, it has been used as a proxy for HIV prevalence in the general population [##UREF##4##9##]. In our study, the prevalence of HIV infection among pregnant women attending ANC was 3.8%. This was similar to the previous reports from the region [##REF##15370654##10##] but lower than the reports of sentinel surveillance from other parts of the country [##UREF##2##7##] and sub-Saharan African countries [##UREF##10##19##,##REF##16643653##20##]. As expected, the prevalence of HIV among pregnant women was higher in urban areas than rural areas; this could be due to the difference in the risk and rate of HIV infection in urban and rural communities [##UREF##11##21##,##REF##7563263##22##].</p>", "<p>In our study, the rate of HIV infection in TB patients strongly correlated with the rate of HIV infection among pregnant women. This was because HIV is the main risk factor fuelling TB epidemic. Similarly, countries with high HIV prevalence in the general population had higher incidence of TB and relatively higher rates of TB-HIV co-infection.</p>", "<p>In southern and eastern Africa, reports have shown an increase in TB notification rate of 13 cases per 10⁵population per year for each 1% increase in HIV prevalence in countries with high prevalence of HIV infection [##REF##16870527##4##]. In a generalized HIV epidemic, the rate of HIV infection among TB patients is an indicator of the maturity of the HIV epidemic and predicts the occurrence of new TB cases at country level [##UREF##4##9##]. A six per cent increase in the number of TB cases and high rates of HIV infection among TB patients over the last two decades were reported from sub-Saharan Africa. This was shown by a strong correlation between adult HIV prevalence and TB case notification in a community; and a higher prevalence of HIV infection in pregnant women was accompanied by high rate of HIV infection in TB patients [##UREF##10##19##]. Similarly, a strong correlation (R²= 0.77) was reported from Europe [##UREF##12##23##].</p>", "<p>In our study, the correlation between the seroprevalence in pregnant women and TB patients coincided with the spread and stage of HIV epidemic in a community. This was reflected by the higher rate of HIV infection among TB patients and pregnant women in urban areas. This could be because of matured HIV epidemic in urban areas that led to an increased number of TB cases and number of HIV infected TB patients [##REF##11125503##24##,##REF##12076570##25##]. In rural areas, we found lower correlation possibly due to the low HIV prevalence in the rural communities [##UREF##7##13##] and a lag period between the spread of HIV infection and maturity of the epidemic. In Zimbabwe, an increase in TB incidence occurred four to five years after the spread of HIV infection in the community [##REF##16870527##4##] and a lag period of seven years was reported from Kenya [##REF##16200083##26##]. Generally, HIV prevalence surveys in Africa, Asia and Pacific showed HIV prevalence in TB patients to be many times higher than that was seen in the general population [##REF##16511773##27##, ####REF##12549845##28##, ##REF##15520482##29####15520482##29##]. Similar to the report from Cameroon, surveillance of HIV infection in TB patients could be used as an estimate of the rate of HIV infection in the general population [##REF##15139474##30##].</p>" ]

[ "<title>Conclusion</title>", "<p>The rate of HIV infection in TB patients was associated with the prevalence of HIV infection among pregnant women in the general population. The seroprevalence information for TB patients and pregnant women could be valuable for planning, monitoring and evaluation of joint prevention and control activities. The trend and level of interaction of HIV infection in TB patients and pregnant women need further study.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>A complex interaction exists between tuberculosis (TB) and human immunodeficiency virus (HIV) infection at an individual and community level. Limited knowledge about the rate of HIV infection in TB patients and the general population compromises the planning, resource allocation and prevention and control activities. The aim of this study was to determine the rate of HIV infection in TB patients and its correlation with the rate HIV infection in pregnant women attending antenatal care (ANC) in Southern Ethiopia.</p>", "<title>Methods</title>", "<p>All TB patients and pregnant women attending health institutions for TB diagnosis and treatment and ANC were consecutively enrolled in 2004 – 2005. TB diagnosis, treatment and HIV testing were done according to the national guidelines. Blood samples were collected for anonymous HIV testing. We used univariate and multivariate logistic regression analysis to determine the risk factors for HIV infection and linear regression analysis to determine the correlation between HIV infection in TB patients and pregnant women.</p>", "<title>Results</title>", "<p>Of the 1308 TB patients enrolled, 226 (18%) (95%CI: 15.8 – 20.0) were HIV positive. The rate of HIV infection was higher in TB patients from urban 25% (73/298) than rural areas 16% (149/945) [AOR = 1.78, 95%CI: 1.27–2.48]. Of the 4199 pregnant women attending ANC, 155 (3.8%) [95%CI: 3.2–4.4] were HIV positive. The rate of HIV infection was higher in pregnant women from urban (7.5%) (80/1066) than rural areas (2.5%) (75/3025) [OR = 3.19, 95% CI: 2.31–4.41]. In the study participants attending the same health institutions, the rate of HIV infection in pregnant women correlated with the rate of HIV infection in TB patients (R²= 0.732).</p>", "<title>Conclusion</title>", "<p>The rate of HIV infection in TB patients and pregnant women was higher in study participants from urban areas. The rate of HIV infection in TB patients was associated with the prevalence of HIV infection in pregnant women attending ANC.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>DGD, LTC and LEK supervised data collection and laboratory testing. DGD, MAY and BL analysed, interpreted the findings and prepared the drafts. All authors contributed to the final manuscript.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2458/8/266/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>We would like to thank the SNNPR Health Bureau for providing financial, technical and material support for the study. We are also grateful to the staff working in the ANC, TB units and laboratories of the participating health institutions. Our special thanks go to TB patients and pregnant women who consented to take part in the survey.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Map of the Southern Nations, Nationalities and Peoples' Region of Ethiopia showing the survey sites, 2004 – 2005</bold>. Zonal boundary. Regional boundary. International boundary. Regional capital, Awassa. Lake. TB-HIV survey sites. ANC-based sentinel survey sites. Overlapping sites.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>The association of HIV infection among TB patients and pregnant women attending antenatal care in southern Ethiopia, 2004 – 2005. Urban. Rural. Fit line for total.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Socio-demographic characteristics and HIV status of TB patients, southern Ethiopia, 2004 – 2005</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Variables</bold></td><td/><td align=\"left\"><bold>TB Patients without HIV </bold><break/><bold>(N = 1035), n (%)</bold></td><td align=\"left\"><bold>TB patients with HIV </bold><break/><bold>(N = 226), n (%)</bold></td><td align=\"left\"><bold>OR (95%CI)</bold></td><td align=\"left\"><bold>P-value</bold></td><td align=\"left\"><bold>AOR (95%CI)</bold></td><td align=\"left\"><bold>P-value</bold></td></tr></thead><tbody><tr><td align=\"left\">Age</td><td align=\"left\">Mean (SD)</td><td align=\"left\">29.24 (9.85)</td><td align=\"left\">28.29 (13.77)</td><td/><td/><td/><td/></tr><tr><td align=\"left\">Gender</td><td align=\"left\">Male</td><td align=\"left\">581(82.1)</td><td align=\"left\">127 (17.9)</td><td align=\"left\">1</td><td/><td/><td/></tr><tr><td/><td align=\"left\">Female</td><td align=\"left\">445 (82.1)</td><td align=\"left\">97 (17.9)</td><td align=\"left\">0.99 (0.75 – 1.34)</td><td align=\"left\">0.985</td><td/><td/></tr><tr><td align=\"left\">Residence</td><td align=\"left\">Rural</td><td align=\"left\">796 (84.2)</td><td align=\"left\">149 (15.8)</td><td align=\"left\">1</td><td/><td/><td/></tr><tr><td/><td align=\"left\">Urban</td><td align=\"left\">225 (75.5)</td><td align=\"left\">73 (24.5)</td><td align=\"left\">1.73 (1.26 – 2.38)</td><td/><td align=\"left\">1.77 (1.28 – 2.46)</td><td align=\"left\">0.001</td></tr><tr><td align=\"left\">Age group</td><td align=\"left\">0 – 14</td><td align=\"left\">109 (90.8)</td><td align=\"left\">11 (9.2)</td><td align=\"left\">1</td><td/><td/><td/></tr><tr><td/><td align=\"left\">15 – 24</td><td align=\"left\">344 (88.0)</td><td align=\"left\">47 (12.0)</td><td align=\"left\">1.35 (0.68 – 2.70)</td><td/><td align=\"left\">2.01 (0.54 – 7.49)</td><td align=\"left\">0.301</td></tr><tr><td/><td align=\"left\">25 – 34</td><td align=\"left\">267 (73.4)</td><td align=\"left\">97 (26.6)</td><td align=\"left\">3.60 (1.86 – 6.98)</td><td/><td align=\"left\">2.54 (0.76 – 8.46)</td><td align=\"left\">0.129</td></tr><tr><td/><td align=\"left\">35 – 44</td><td align=\"left\">153 (76.9)</td><td align=\"left\">46 (23.1)</td><td align=\"left\">2.98 (1.48 – 6.01)</td><td/><td align=\"left\">7.10 (2.17 – 23.26)</td><td align=\"left\">0.001</td></tr><tr><td/><td align=\"left\">45 – 54</td><td align=\"left\">113 (86.9)</td><td align=\"left\">17 (13.1)</td><td align=\"left\">1.76 (0.78 – 3.93)</td><td/><td align=\"left\">5.78 (1.72 – 19.38)</td><td align=\"left\">0.005</td></tr><tr><td/><td align=\"left\">≥ 55</td><td align=\"left\">57 (95.0)</td><td align=\"left\">3 (5.0)</td><td align=\"left\">0.52 (0.14 – 1.95)</td><td/><td align=\"left\">3.34 (0.94 – 11.93)</td><td align=\"left\">0.063</td></tr><tr><td align=\"left\">TB classification</td><td align=\"left\">PTB +ve</td><td align=\"left\">434 (82.5)</td><td align=\"left\">92 (17.5)</td><td align=\"left\">1</td><td/><td/><td/></tr><tr><td/><td align=\"left\">PTB -ve</td><td align=\"left\">354 (81.9)</td><td align=\"left\">78 (18.1)</td><td align=\"left\">1.04 (0.75 – 1.45)</td><td align=\"left\">0.82</td><td/><td/></tr><tr><td/><td align=\"left\">EPTB</td><td align=\"left\">243 (81.8)</td><td align=\"left\">54 (18.2)</td><td align=\"left\">1.05 (0.72 – 1.52)</td><td align=\"left\">0.803</td><td/><td/></tr><tr><td align=\"left\">TB category</td><td align=\"left\">New</td><td align=\"left\">956 (82.6)</td><td align=\"left\">202 (17.4)</td><td align=\"left\">1</td><td/><td/><td/></tr><tr><td/><td align=\"left\">RFDO</td><td align=\"left\">32 (74.4)</td><td align=\"left\">11 (25.6)</td><td align=\"left\">1.61(0.79 – 3.24)</td><td align=\"left\">0.184</td><td/><td/></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Socio-demographic characteristics and HIV status of pregnant women attending ANC, Southern Ethiopia, 2004 – 2005</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Variables</bold></td><td/><td align=\"left\"><bold>ANC attendants without HIV </bold><break/><bold>(N = 3936), n (%)</bold></td><td align=\"left\"><bold>ANC attendants with HIV </bold><break/><bold>(N = 155), n (%)</bold></td><td align=\"left\"><bold>OR (95%CI)</bold></td><td align=\"left\"><bold>P – value</bold></td></tr></thead><tbody><tr><td align=\"left\">Age</td><td align=\"left\">Mean (SD)</td><td align=\"left\">25.45 (5.25)</td><td align=\"left\">25.72 (5.19)</td><td/><td/></tr><tr><td align=\"left\">Age group</td><td align=\"left\">15 – 24</td><td align=\"left\">1547 (96)</td><td align=\"left\">64 (4.0)</td><td align=\"left\">1</td><td/></tr><tr><td/><td align=\"left\">25 – 34</td><td align=\"left\">2077 (96.4)</td><td align=\"left\">77 (3.6)</td><td align=\"left\">0.89 (0.64 – 1.26)</td><td align=\"left\">0.525</td></tr><tr><td/><td align=\"left\">≥ 35 – 44</td><td align=\"left\">312 (95.7)</td><td align=\"left\">14 (4.3)</td><td align=\"left\">1.09 (0.60 – 1. 96)</td><td align=\"left\">0.788</td></tr><tr><td align=\"left\">Residence</td><td align=\"left\">Rural</td><td align=\"left\">2950 (97.5)</td><td align=\"left\">75 (2.5)</td><td align=\"left\">1</td><td/></tr><tr><td/><td align=\"left\">Urban</td><td align=\"left\">986 (92.5)</td><td align=\"left\">80 (7.5)</td><td align=\"left\">3.19 (2.31 – 4.41)</td><td align=\"left\">0.0001</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>The rate of HIV infection among TB patients and pregnant women attending antenatal care in southern region of Ethiopia 2004 – 2005</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Survey sites by zones*</bold></td><td align=\"left\"><bold>ANC attendants with HIV % (N)</bold></td><td align=\"left\"><bold>TB patients with HIV % (N)</bold></td><td align=\"left\"><bold>R^2†</bold></td><td align=\"left\"><bold>Adjusted R²</bold></td><td align=\"left\"><bold>P-value^‡</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Urban survey sites</bold></td><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"> Sidama zone</td><td align=\"left\">9.48 (29/306)</td><td align=\"left\">17.84 (38/213)</td><td/><td/><td/></tr><tr><td align=\"left\"> Wolaita zone</td><td align=\"left\">10.53 (26/247)</td><td align=\"left\">13.79 (12/87)</td><td/><td/><td/></tr><tr><td align=\"left\"> Gedeo zone</td><td align=\"left\">9.46 (21/222)</td><td align=\"left\">18.11 (23/127)</td><td/><td/><td/></tr><tr><td align=\"left\"> Bench Maji zone</td><td align=\"left\">2.25 (8/360)</td><td align=\"left\">32.5 (66/203)</td><td/><td/><td/></tr><tr><td align=\"left\"> South Omo zone</td><td align=\"left\">1.72 (7/408)</td><td align=\"left\">35.29 (12/34)</td><td/><td/><td/></tr><tr><td align=\"left\"> Kaffa zone</td><td align=\"left\">2.45 (8/326)</td><td align=\"left\">26.23 (16/61)</td><td/><td/><td/></tr><tr><td align=\"left\"><bold>Rural survey sites</bold></td><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"> Hadiya zone</td><td align=\"left\">2.7 (7/259)</td><td align=\"left\">9.17 (21/229)</td><td/><td/><td/></tr><tr><td align=\"left\"> Gurage zone</td><td align=\"left\">4.5 (18/400)</td><td align=\"left\">13.14 (23/175)</td><td/><td/><td/></tr><tr><td align=\"left\"> Gamo Goffa zone</td><td align=\"left\">1.48 (6/405)</td><td align=\"left\">10.61 (7/66)</td><td/><td/><td/></tr><tr><td align=\"left\"> Silte zone</td><td align=\"left\">1.95 (8/411)</td><td align=\"left\">8.33 (4/48)</td><td/><td/><td/></tr><tr><td align=\"left\"> Sheka zone</td><td align=\"left\">2.31(8/346)</td><td/><td/><td/><td/></tr><tr><td align=\"left\"> Kambata Tembaro zone</td><td align=\"left\">2.24 (9/401)</td><td/><td/><td/><td/></tr><tr><td align=\"left\"><bold>All survey sites</bold></td><td/><td/><td align=\"left\">0.034</td><td align=\"left\">0.034</td><td align=\"left\">&lt; 0.001</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>The rate of HIV infection among TB patients and pregnant women attending antenatal care in the same health institutions of southern region of Ethiopia 2004 – 2005</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\"><bold>Survey sites by zones*</bold></td><td align=\"left\"><bold>ANC attendants with HIV % (N)</bold></td><td align=\"left\"><bold>TB patients with HIV % (N)</bold></td><td align=\"left\"><bold>R^2†</bold></td><td align=\"left\"><bold>Adjusted R²</bold></td><td align=\"left\"><bold>P – value</bold></td></tr></thead><tbody><tr><td align=\"left\"><bold>Urban survey sites</bold></td><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"> Sidama zone</td><td align=\"left\">9.48 (29/306)</td><td align=\"left\">17.84 (38/213)</td><td/><td/><td/></tr><tr><td align=\"left\"> Wolaita zone</td><td align=\"left\">10.53 (26/247)</td><td align=\"left\">13.79 (12/87)</td><td/><td/><td/></tr><tr><td align=\"left\"> Gedeo zone</td><td align=\"left\">9.46 (21/222)</td><td align=\"left\">18.11 (23/127)</td><td/><td/><td/></tr><tr><td align=\"left\"> All urban sites</td><td/><td/><td align=\"left\">0.998</td><td align=\"left\">0.998</td><td align=\"left\">&lt; 0.001</td></tr><tr><td align=\"left\"><bold>Rural survey sites</bold></td><td/><td/><td/><td/><td/></tr><tr><td align=\"left\"> Hadiya zone</td><td align=\"left\">2.7 (7/259)</td><td align=\"left\">9.17 (21/229)</td><td/><td/><td/></tr><tr><td align=\"left\"> Gurage zone</td><td align=\"left\">4.5 (18/400)</td><td align=\"left\">13.14 (23/175)</td><td/><td/><td/></tr><tr><td align=\"left\"> Gamo Goffa zone</td><td align=\"left\">1.48 (6/405)</td><td align=\"left\">10.61 (7/66)</td><td/><td/><td/></tr><tr><td align=\"left\"> All rural sites</td><td/><td/><td align=\"left\">0.547</td><td align=\"left\">0.546</td><td align=\"left\">&lt; 0.001</td></tr><tr><td align=\"left\"><bold>All survey sites</bold></td><td/><td/><td align=\"left\">0.732</td><td align=\"left\">0.732</td><td align=\"left\">&lt; 0.001</td></tr></tbody></table></table-wrap>" ]

[ "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>", "<inline-formula></inline-formula>" ]

[]

[]

[]

[]

[]

[ "<table-wrap-foot><p>TB = Tuberculosis, HIV = Human immunodeficiency virus, OR = odds ratio, CI = confidence interval, AOR = adjusted OR for age and residence</p><p>SD = standard deviation, PTB +ve = smear positive pulmonary TB, PTB -ve = smear negative pulmonary TB, EPTB = extrapulmonary TB, R = relapse,</p><p>F = failure, D = return after default, O = others. Missing variables: age – 15 (1.2%), sex – 12(0.9%), address – 21(1.6%), disease classification – 7(0.5%), disease category – 58(4.4%), HIV result – 47(3.6%), sex &amp; HIV result – 58(4.4%), age group and HIV – 61(4.7%), address and HIV – 21(1.6%), disease classification and HIV – 53(4.1%) and age category and HIV- 61(4.7%).</p></table-wrap-foot>", "<table-wrap-foot><p>ANC = antenatal care, HIV = human immunodeficiency virus, OR = odds ratio, CI = confidence interval, SD = standard deviation</p><p>Missing variables: age – 6(0.1%), address – 6(0.1%), HIV result – 108(2.6%), age group &amp; HIV – 108(2.6%), address &amp; HIV – 108(2.6%)</p></table-wrap-foot>", "<table-wrap-foot><p>*The survey sites were areas where we conducted the two surveys in the same and different health institutions.</p><p>^†R²-coefficient of determination weighed for the number of study participants</p><p>^‡P-value for adjusted R²HIV = Human immunodeficiency virus, ANC = Antenatal care</p></table-wrap-foot>", "<table-wrap-foot><p>*The survey sites were areas where we conducted the two surveys in the same health institutions in a district.</p><p>^†R²-coefficient of determination weighed for the number of study participants</p></table-wrap-foot>" ]

[ "<inline-graphic xlink:href=\"1471-2458-8-266-i1.gif\"/>", "<inline-graphic xlink:href=\"1471-2458-8-266-i2.gif\"/>", "<inline-graphic xlink:href=\"1471-2458-8-266-i3.gif\"/>", "<inline-graphic xlink:href=\"1471-2458-8-266-i4.gif\"/>", "<inline-graphic xlink:href=\"1471-2458-8-266-i5.gif\"/>", "<inline-graphic xlink:href=\"1471-2458-8-266-i6.gif\"/>", "<inline-graphic xlink:href=\"1471-2458-8-266-i7.gif\"/>", "<inline-graphic xlink:href=\"1471-2458-8-266-i8.gif\"/>", "<graphic xlink:href=\"1471-2458-8-266-1\"/>", "<inline-graphic xlink:href=\"1471-2458-8-266-i9.gif\"/>", "<inline-graphic xlink:href=\"1471-2458-8-266-i10.gif\"/>", "<inline-graphic xlink:href=\"1471-2458-8-266-i11.gif\"/>", "<graphic xlink:href=\"1471-2458-8-266-2\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Finding strategies to deal with chronic medical conditions is a priority for improving the health of the world population [##UREF##0##1##]. Studies [##UREF##1##2##, ####REF##9619963##3##, ##REF##15928225##4####15928225##4##] in various countries demonstrate that many people have multimorbidity, defined as the co-occurrence of two or more chronic diseases [##UREF##2##5##]. Persons with multimorbidity require more frequent appointments and hospitalizations [##UREF##1##2##,##REF##2046132##6##, ####REF##11252208##7##, ##REF##10210233##8##, ##REF##8911426##9####8911426##9##] and are at greater risk for drug interactions, acute deteriorations, disability, and mortality [##REF##11735757##10##, ####REF##12623886##11##, ##REF##15043173##12##, ##REF##3339381##13##, ##REF##3132042##14####3132042##14##]. They also have higher psychological distress [##REF##17003141##15##] and their quality of life is affected [##REF##16411033##16##,##REF##15380021##17##]. Among the strategies for improving the health of people with chronic diseases, physical activity is extensively supported in the published literature [##REF##11838650##18##, ####REF##12424867##19##, ##REF##8667571##20####8667571##20##]. Physical activity improves psychological well-being, and decreases stress, anxiety, and the feelings associated with depression [##REF##8410742##21##,##UREF##3##22##]. Physical activity also decreases pain during the treatment of painful conditions [##REF##12946287##23##], favors resistance and vigor of people with chronic diseases [##REF##11838650##18##], and decreases their risk of developing functional incapacities [##REF##12424867##19##].</p>", "<p>However, sixty percent of the world population and more than 50% of American adults do not attain the minimum of physical activity levels recommended by the American College of Sports Medicine and the Centers for Disease Control and Prevention, namely, 30 minutes of moderate physical activity at least five days a week [##REF##7823386##24##]. Levels of physical activity decrease with age, and women are less physically active than men [##UREF##1##2##]. Physical activity levels are lower among populations with low income or low education [##REF##11701302##25##,##REF##11722684##26##], and among people with functional incapacities or limited ability to be active [##REF##11701302##25##,##REF##15595285##27##]. Studies also suggest a relationship between employment [##REF##16230835##28##,##REF##15941745##29##], perceived health status or psychological distress, and physical activity levels [##REF##11701302##25##,##REF##8447857##30##, ####REF##16800402##31##, ##REF##15530578##32####15530578##32##].</p>", "<p>Considering the numerous preventive and curative effects of physical activity, health care professionals should get involved in promoting physical activity among their patients with chronic diseases and multimorbidity [##REF##11838650##18##]. We already know that patients with multimorbidity are more likely to have limited activity levels [##REF##15353000##33##], psychological distress [##REF##17003141##15##], and perception of a lower health status [##REF##10880774##34##]. Kaplan et al. examined predictors of frequent and infrequent self-reported physical activity lasting more than 15 min in community-dwelling people ≥ 65 yr of age. Using logistic regression analyses, they found that the absence of chronic conditions was associated with more frequent physical activity in late life [##REF##11701302##25##]. The relationship between multimorbidity and physical activity levels remains however poorly documented among non geriatric people. A better understanding of this relationship would facilitate development of more appropriate counseling strategies among this clientele.</p>", "<p>The purpose of the present exploratory study was to examine the relationship between a) multimorbidity and physical activity levels; and b) long-term limitations on activity, self-rated general health, psychological distress, and physical activity levels for both sexes in adults, after age, education, income, and employment factors have been controlled for. Our hypothesis, suggested by Kaplan's study, was that multimorbidity may be associated with a decreased physical activity level in adults.</p>" ]

[ "<title>Methods</title>", "<title>Study design and data Source</title>", "<p>This study was a secondary analysis of the Quebec Health Survey 1998 (QHS). QHS was approved by the Ethics Committee of Santé Québec. Detailed descriptions of the methods and variables used for the QHS have been described elsewhere [##UREF##1##2##]. Briefly, the survey was a multistage random probability sample of Quebec residential households designed to collect data about the health and well-being of the Quebec population, and the needs and priority areas for intervention and the allocation of resources. In total, 15,330 households were surveyed. Information about the members of the household was obtained from the person most knowledgeable about the health of family members. The survey, which was in two parts, was conducted in a cascade fashion: one administered by an interviewer; the other, self-administered. The self-administered survey was conditional on having replied to the interviewer survey. The response rate for the interviewer-administered survey was 82.1%. The response rate for the self-administered survey was 69%.</p>", "<title>Participants</title>", "<p>We included respondents who were 18–69 yr of age at the time of the survey. Respondents who had cerebral palsy, amputation or long-term limitations on activity after an accident were excluded so that we included only those people with limitations related to chronic health problems. In total, 16,782 persons met these inclusion criteria.</p>", "<title>Independent Variables</title>", "<p>The set of independent variables were multimorbidity, long-term limitations on activity, self-rated general health, and psychological distress.<italic>Multimorbidity </italic>(used in the interviewer survey) was conceptualized in two different ways. First, given the exploratory nature of this study, we examined different cut-offs of multimorbidity: starting with 0 as a reference level, followed by 1 chronic condition, then two, three, four and five chronic diseases or more, based on the self-reported diagnosis of conditions from the following list: anemia, dermatologic disease, allergies, back disease, arthritis, serious articular, muscular or tendinous condition, cancer, diabetes, pulmonary condition, mental deficiency, depression, anxiety, psychosis, epilepsy, hypertension, cardiac disease, renal disease, ulcer or other gastric problem, thyroid disease, frequent headache, stroke sequelae, cognitive deficit, obesity, and ocular disease. Second, we also conceptualized multimorbidity as two chronic diseases or more in the same list used by Kaplan: asthma, arthritis or rheumatism, back disease, hypertension, chronic bronchitis or emphysema, diabetes, cardiac disease, stroke sequelae, cognitive deficit and ocular disease. Kaplan's list included also bowel disease but this disease was not evaluated in the survey database. <italic>Long-term limitations on activity </italic>(used in the interviewer survey) was a dichotomous variable that assessed the absence or presence of any self-reported limitation on activity because of a chronic disease or health problem, in comparison with the activity level of people of the same age. <italic>Self-rated general health </italic>(used in the self-administered survey) provided a global assessment of the person's perception of his or her overall health status compared with that of others of the same age. Responses were grouped into two categories: good-to-excellent and poor-to-average. <italic>Psychological distress </italic>(used in the self-administered survey) was measured with a modified version of the Psychiatric Symptom Index of Ilfeld [##REF##8969117##35##]. A high score indicated a greater level of reported psychological distress. The scores were grouped into two categories: low-to-average and high. All independent variables were used as dummy variables with the first category as the reference category. Finally, <italic>socio-demographic characteristics </italic>– age, education, income, and employment – that might account for part of the variance in physical activity levels [##UREF##1##2##,##REF##11701302##25##,##REF##16230835##28##,##REF##15941745##29##] were included as covariates in the analyses.</p>", "<title>Dependent Variables</title>", "<p><italic>Physical activity level </italic>(used in the self-administered survey) represented the number of leisure physical activity sessions of 20–30 min in the last three months [##REF##2614536##36##]. Responses were grouped into three categories: none, less than three times a week, and more than three times a week.</p>", "<title>Statistical Analyses</title>", "<p>All analyses were conducted with SUDAAN software version 9.0 that helped account for the complex survey design and household clustering. Assessment of the associations between the independent variables and the dependent variable was done with a generalized multinomial logit model. Regression parameters were then estimated with the use of generalized estimating equation with a robust variance estimation. The Wald F Test of significance was used with a <italic>P </italic>value set at 0.05.</p>", "<p>Data for men and women were analyzed separately. All covariates (age, education, income, and employment) were the first introduced into the generalized logit equations and were maintained as control variables in all models. After the main effects of the independent variables were calculated, the interactions between multimorbidity, long-term limited mobility, self-rated general health, and psychological distress were calculated. Two separate multinomial regressions were run for each of the two forms of conceptualization of multimorbidity (as different levels up to 5 or more chronic conditions, and based on Kaplan et al study's set of diseases). Results were expressed as odds ratios with their <italic>P </italic>values and 95% confidence intervals.</p>", "<p>Using a simulation algorithm in SUDAAN software, the power to detect a value ≤ -0.30 (Odds ratios ≤ 0,74) for any parameter for multimorbidity (0, 1, and ≥ 2 diseases) was at least 70.6% for men and 89.2% for women.</p>" ]

[ "<title>Results</title>", "<p>Table ##TAB##0##1## presents the characteristics of the study sample: 16,782 persons (46% men) were included in the study. Of these men (n = 7,776), 60% (n = 4,525) were 18–44 yr and 40% (n = 3,251) were 45–69 yr of age. Of the women (n = 9,006), 59% (n = 5,385) were 18 to 44 yr and 41% (n = 3,621) were 45–69 yr of age.</p>", "<p>None of the interactions that we tested achieved significance in this study. In the following sections, we present the models without the interaction terms. Multimorbidity was not associated with physical activity levels either for men or women when age, education, income, employment, long-term limitations on activity, self-rated general health, and psychological distress were controlled for, regardless of how multimorbidity was conceptualized (Tables ##TAB##1##2##, ##TAB##2##3##, ##TAB##3##4## and ##TAB##4##5##). However, compared with men and women with no physical limitations on activity, men and women with long-term limitations on activity were less likely to be physically active. Men and women who rated their general health as poor to average were less likely to exercise than men and women who rated their general health as good to excellent. Finally, no relation between psychological distress and physical activity was found for men. However, women with high levels of psychological distress were less likely to be physically active than women with low-to-average levels of psychological distress.</p>" ]

[ "<title>Discussion</title>", "<p>To our knowledge, this is the first study to explore the relationship between multimorbidity and physical activity in this age group. We found that none of the cut-offs of multimorbidity (number of chronic conditions) that we used was associated with physical activity levels when long-term limitations on activity, psychological distress, perceived health status, age, sex, education, income, and employment were controlled for.</p>", "<p>Our results differ from those of Kaplan's study [##REF##11701302##25##]. Kaplan et al. used data for 12,611 community-dwelling people ≥ 65 yr of age from the 1996–1997 Canadian National Population Health Survey to examine predictors of frequent and infrequent self-reported physical activity lasting more than 15 min, including many independent and confounding variables (geographic location, psychological distress measured with the Generalized Distress Scale, age, gender, educational level, marital status, perceived social support, chronic medical conditions, physical limitations due to injury, functional limitations, smoking behavior and body mass index). Using logistic regression analyses, they found that the absence of chronic conditions was associated with more frequent physical activity in late life. Our two studies shared many similarities: 1) Population Health Surveys; 2) Larges samples; 3) Comparable lists of chronic conditions; 4) Comparable self-reported questionnaires for physical activity; 5) Logistic regression analysis; 6) Separated analysis for males and females. However, Kaplan et al documented a relationship between the absence of chronic conditions and an increased level of physical activity, a relationship that we could not ascertain in our sample. A few hypotheses may be raised to explain different results. First, the ages of our study populations were different. Kaplan et al. [##REF##11701302##25##] recruited people ≥ 65 yr of age, whereas we deliberately excluded people &gt; 69 yr so that we could evaluate a non-geriatric population. Relationship between multimorbidity and physical activity may be different in these two groups. Second, we don't know if the independent \"chronic conditions\" was considered continuous or categorical in Kaplan's study. This choice may have impact on results. Finally, they include different variables in their model but did not include self-rated general health that is significant in our models. Addition of this variable may have changed their results.</p>", "<p>Neither study used a comorbidity or multimorbidity index that accounted for the severity of any single disease or the involvement of a system. One of our previous studies [##REF##17003141##15##] showed that the count of chronic diseases was not associated with an outcome such as psychological distress, whereas chronic disease was associated with psychological distress when an index that takes severity into account was used.</p>", "<p>Even if multimorbidity does not seem to be related to physical activity levels, it is related to other variables such as long-term limitations on activity [##REF##15353000##33##], psychological distress [##REF##17003141##15##], and perception of a lower health status [##REF##10880774##34##] that can influence physical activity. Our findings for these variables are in accordance with those of previous studies that demonstrated a relationship between functional incapacities or long-term limitations on activity [##REF##15595285##27##,##REF##12831658##37##, ####REF##15931024##38##, ##REF##7698044##39####7698044##39##], perceived health status, and physical activity level [##REF##8447857##30##,##REF##16800402##31##]. We found a significant relationship between psychological distress and physical activity levels only among women. Other studies [##REF##11701302##25##,##REF##11722684##26##,##REF##15530578##32##] demonstrated a relationship between psychological distress and physical activity, but did not distinguish between the sexes.</p>", "<title>Implications for research and practice</title>", "<p>Keeping in mind the exploratory nature of our study, results suggest some clues to clinicians regarding the counseling that they can do among patients with multimorbidity. For example, when addressing barriers to physical activity, it may be more appropriate to go beyond the number of medical chronic conditions to evaluate and address long-term limitations on activity, and perception of a lower health status. Effective strategies could also be different for males and females in presence of psychological distress.</p>", "<p>Other studies, ideally longitudinal, and using a validated comorbidity or multimorbidity index that takes the severity of diseases into account should be conducted to confirm the relationship between multimorbidity and physical activity levels. Further evaluation is also needed to examine the relationship between psychological distress and physical activity levels of men and women.</p>" ]

[ "<title>Conclusion</title>", "<p>Multimorbidity was not associated with physical activity levels for either sex in adults when age, education, income, and employment factors were controlled for. Long-term limitations on activity and poor-to-average self-rated general health status seem to be related to lower physical activity levels for both sexes, whereas psychological distress was associated with lower physical activity levels only among women.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Abundant literature supports the beneficial effects of physical activity for improving health of people with chronic diseases. The relationship between multimorbidity and physical activity levels, however, has been little evaluated. The purpose of the current exploratory study was to examine the relationship between a) multimorbidity and physical activity levels, and b) long-term limitations on activity, self-rated general health, psychological distress, and physical activity levels for each sex in adults, after age, education, income, and employment factors were controlled for.</p>", "<title>Methods</title>", "<p>Data from the Quebec Health Survey 1998 were used. The sample included 16,782 adults 18–69 yr of age. Independent variables were multimorbidity, long-term limitations on activity, self-rated general health, and psychological distress. The dependent variable was physical activity levels. Links between the independent and dependent variables were assessed separately for men and women with multinomial regressions while accounting for the survey sampling design and household clustering.</p>", "<title>Results</title>", "<p>About 46% of the participants were men. Multimorbidity was not associated with physical activity levels for either men or women. Men and women with long-term limitations on activity and with poor-to-average self-rated general health were less likely to be physically active. No relationship between psychological distress and physical activity was found for men. Women with high levels of psychological distress were less likely to be physically active.</p>", "<title>Conclusion</title>", "<p>Multimorbidity was not associated with physical activity levels in either sex, when age, education, income, and employment factors were controlled for. Long-term limitations on activity and poor-to-average self-rated general health seem related to a reduction in physical activity levels for both sexes, whereas psychological distress was associated with a reduction in physical activity levels only among women. Longitudinal studies using a comorbidity or multimorbidity index to account for severity of the chronic diseases are needed to replicate the results of this exploratory study.</p>" ]

[ "<title>Limitations of the study</title>", "<p>The following limitations of this study should be kept in mind. In our analyses we used the variables available from the QHS. First, despite the good quality of the data from the QHS 1998 survey (it had a large sample, four data collections to take seasonal variations into account, and a good response rate), it is possible that recall bias and social desirability may have influenced the respondents' answers. Second, our measure of multimorbidity was limited to a number of chronic diseases. All of these chronic diseases had equal weight in the analysis without any assessment of their severity. Third, although self-efficacy and physical activity levels are strongly related [##UREF##4##40##,##REF##1738853##41##], the original QHS survey did not measure the relationship between self-efficacy and physical activity prospectively, so we could not include this variable in our model. We were not able to include other variables such as cholesterol, blood glucose, medication and renal function as they were not available in the population survey. Also, other characteristics such as diet, body mass index and smoking may have been included. However, we wanted this exploratory study to be as focused as possible and decided not to include these as covariables in the analysis. Finally, since the data from this study are cross-sectional, we cannot assign any causality to the relationships that were identified. Despite these limitations, our study remains one of the first to explore the topic.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>CH participated in the conception and design of the study, supervised analysis, drafted the manuscript and provided funding. HS participated in the conception and design of the study, helped supervise analysis, drafted the manuscript, and provided funding. MF participated in the design of the study, critically reviewed the manuscript and provided funding. All authors gave their final approval for the manuscript submitted for publication.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2458/8/304/prepub\"/></p>" ]

[ "<title>Acknowledgements</title>", "<p>We are grateful to the Quebec Statistics Institute and the CADRISQ for providing the data and the informatics resources to do the analyses. Our special thanks go to Maxime Boucher for his help with the statistical analyses. We are also grateful for the support of the Department of Family Medicine at the University of Sherbrooke. We thank Sharon Nancekivell, medical editor, for editorial assistance with preparing and revising this paper.</p>" ]

[]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Characteristics of the study population</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"2\"><bold>No. (%)</bold></td></tr><tr><td/><td colspan=\"2\"><hr/></td></tr><tr><td align=\"left\"><bold>Characteristic</bold></td><td align=\"center\"><bold>Men </bold>(n = 7,776)</td><td align=\"center\"><bold>Women </bold>(n = 9,006)</td></tr></thead><tbody><tr><td align=\"left\"><bold>Age</bold></td><td/><td/></tr><tr><td align=\"left\">18–44 yr</td><td align=\"center\">4525 (60.3)</td><td align=\"center\">5385 (59.0)</td></tr><tr><td align=\"left\">45–69 yr</td><td align=\"center\">3251 (39.7)</td><td align=\"center\">3621 (41.0)</td></tr><tr><td align=\"left\"><bold>Relative education</bold></td><td/><td/></tr><tr><td align=\"left\">Lowest</td><td align=\"center\">1808 (19.8)</td><td align=\"center\">1852 (20.9)</td></tr><tr><td align=\"left\">Low</td><td align=\"center\">1565 (18.7)</td><td align=\"center\">2006 (22.6)</td></tr><tr><td align=\"left\">Average</td><td align=\"center\">1371 (17.8)</td><td align=\"center\">1868 (21.1)</td></tr><tr><td align=\"left\">High</td><td align=\"center\">1666 (22.8)</td><td align=\"center\">1531 (17.3)</td></tr><tr><td align=\"left\">Highest</td><td align=\"center\">1198 (20.9)</td><td align=\"center\">1602 (18,1)</td></tr><tr><td align=\"left\"><bold>Income in Canadian dollars</bold></td><td/><td/></tr><tr><td align=\"left\">&lt; 6000</td><td align=\"center\">809 (11.2)</td><td align=\"center\">2397 (28.9)</td></tr><tr><td align=\"left\">6000–19,999</td><td align=\"center\">1797 (24.8)</td><td align=\"center\">2829 (34.2)</td></tr><tr><td align=\"left\">20000–39,999</td><td align=\"center\">2543 (35.0)</td><td align=\"center\">2180 (26.3)</td></tr><tr><td align=\"left\">≥ 40,000</td><td align=\"center\">2106 (29.0)</td><td align=\"center\">874 (10.6)</td></tr><tr><td align=\"left\"><bold>Employment</bold></td><td/><td/></tr><tr><td align=\"left\">Part- or fulltime</td><td align=\"center\">6160 (79.3)</td><td align=\"center\">5431 (60.3)</td></tr><tr><td align=\"left\">Not working</td><td align=\"center\">1612 (20.7)</td><td align=\"center\">3574 (39.7)</td></tr><tr><td align=\"left\"><bold>Number of chronic diseases</bold></td><td/><td/></tr><tr><td align=\"left\"> 0</td><td align=\"center\">3962 (51.0)</td><td align=\"center\">3183 (35.3)</td></tr><tr><td align=\"left\"> 1</td><td align=\"center\">1975 (25.4)</td><td align=\"center\">2319 (25.7)</td></tr><tr><td align=\"left\"> 2</td><td align=\"center\">890 (11.4)</td><td align=\"center\">1436 (15.9)</td></tr><tr><td align=\"left\"> 3</td><td align=\"center\">443 (5.7)</td><td align=\"center\">874 (9.7)</td></tr><tr><td align=\"left\"> 4</td><td align=\"center\">247 (3.2)</td><td align=\"center\">518 (5.8)</td></tr><tr><td align=\"left\"> ≥ 5</td><td align=\"center\">259 (3.3)</td><td align=\"center\">676 (7.5)</td></tr><tr><td align=\"left\"><bold>Long-term limitations on activity</bold></td><td/><td/></tr><tr><td align=\"left\">Not limited</td><td align=\"center\">7139 (91.8)</td><td align=\"center\">8046 (89.4)</td></tr><tr><td align=\"left\">Limited</td><td align=\"center\">635 (8.2)</td><td align=\"center\">958 (10.6)</td></tr><tr><td align=\"left\"><bold>Self-rated general health</bold></td><td/><td/></tr><tr><td align=\"left\">Good to excellent</td><td align=\"center\">6580 (89.9)</td><td align=\"center\">7731 (89.5)</td></tr><tr><td align=\"left\">Poor to average</td><td align=\"center\">743 (10.1)</td><td align=\"center\">905 (10.5)</td></tr><tr><td align=\"left\"><bold>Psychological distress</bold></td><td/><td/></tr><tr><td align=\"left\">Low to average</td><td align=\"center\">6290 (83.8)</td><td align=\"center\">6724 (77.4)</td></tr><tr><td align=\"left\">High</td><td align=\"center\">1218 (16.2)</td><td align=\"center\">1960 (22.6)</td></tr><tr><td align=\"left\"><bold>Physical activity</bold></td><td/><td/></tr><tr><td align=\"left\">None</td><td align=\"center\">2280 (30.0)</td><td align=\"center\">2310 (26.2)</td></tr><tr><td align=\"left\">&lt; 3 times a week</td><td align=\"center\">3422 (45.1)</td><td align=\"center\">4329 (49.0)</td></tr><tr><td align=\"left\">≥ 3 times a week</td><td align=\"center\">1890 (24.9)</td><td align=\"center\">2194 (24.8)</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Odds ratios, <italic>P </italic>values, and 95% confidence intervals from multinomial logit model linking different levels of multimorbidity, long-term limitations on activity, self-rated health status, and psychological distress levels to physical activity among men.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"3\"><bold>Odds ratio (95% confidence intervals)</bold></td></tr><tr><td/><td colspan=\"3\"><hr/></td></tr><tr><td align=\"left\"><bold>Independent variables</bold></td><td align=\"center\"><bold>Physical activity &lt; 3 times per week</bold></td><td align=\"center\"><bold>Physical activity ≥ 3 times per week</bold></td><td align=\"center\"><bold><italic>P </italic>value</bold></td></tr></thead><tbody><tr><td align=\"left\">Number of chronic diseases</td><td/><td/><td/></tr><tr><td align=\"left\"> 0</td><td align=\"center\">1.00</td><td align=\"center\">1.00</td><td/></tr><tr><td align=\"left\"> 1</td><td align=\"center\">0.96 (0.78–1.18)</td><td align=\"center\">1.02 (0.81–1.29)</td><td/></tr><tr><td align=\"left\"> 2</td><td align=\"center\">1.06 (0.79–1.41)</td><td align=\"center\">0.93 (0.67–1.29)</td><td/></tr><tr><td align=\"left\"> 3</td><td align=\"center\">0.98 (0.68–1.42)</td><td align=\"center\">1.11 (0.74–1.66)</td><td/></tr><tr><td align=\"left\"> 4</td><td align=\"center\">1.11 (0.71–1.73)</td><td align=\"center\">0.85 (0.49–1.48)</td><td/></tr><tr><td align=\"left\"> ≥ 5</td><td align=\"center\">0.88 (0.52–1.51)</td><td align=\"center\">0.74 (0.41–1.33)</td><td align=\"center\">0.9273</td></tr><tr><td align=\"left\">Long-term limitations on activity</td><td/><td/><td/></tr><tr><td align=\"left\">Absent</td><td align=\"center\">1.0</td><td align=\"center\">1.0</td><td/></tr><tr><td align=\"left\">Present</td><td align=\"center\">0.61 (0.43–0.87)</td><td align=\"center\">0.68 (0.46–1.00)</td><td align=\"center\">0.0204</td></tr><tr><td align=\"left\">Self-rated general health status</td><td/><td/><td/></tr><tr><td align=\"left\">Good to excellent</td><td align=\"center\">1.0</td><td align=\"center\">1.0</td><td/></tr><tr><td align=\"left\">Poor to average</td><td align=\"center\">0.71 (0.53–0.96)</td><td align=\"center\">0.52 (0.36–0.76)</td><td align=\"center\">0.0018</td></tr><tr><td align=\"left\">Psychological distress level</td><td/><td/><td/></tr><tr><td align=\"left\">Low to average</td><td align=\"center\">1.0</td><td align=\"center\">1.0</td><td/></tr><tr><td align=\"left\">High</td><td align=\"center\">1.20 (0.98–1.48)</td><td align=\"center\">0.99 (0.76–1.28)</td><td align=\"center\">0.0977</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T3\"><label>Table 3</label><caption><p>Odds ratios, <italic>P </italic>values and 95% confidence intervals from multinomial logit model linking different levels of multimorbidity, long-term limitations on activity, self-rated health status, and psychological distress levels to physical activity among women.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"3\"><bold>Odds ratio (95% confidence intervals)</bold></td></tr><tr><td align=\"left\"><bold>Independent variables</bold></td><td align=\"center\"><bold>Physical activity &lt; 3 times per week</bold></td><td align=\"center\"><bold>Physical activity ≥ 3 times per week</bold></td><td align=\"center\"><bold><italic>P </italic>value</bold></td></tr></thead><tbody><tr><td align=\"left\">Number of chronic diseases</td><td/><td/><td/></tr><tr><td align=\"left\"> 0</td><td align=\"center\">1.0</td><td align=\"center\">1.0</td><td/></tr><tr><td align=\"left\"> 1</td><td align=\"center\">1.03 (0.84–1.26)</td><td align=\"center\">1.06 (0.84–1.35)</td><td/></tr><tr><td align=\"left\"> 2</td><td align=\"center\">1.10 (0.87–1.39)</td><td align=\"center\">1.00 (0.75–1.31)</td><td/></tr><tr><td align=\"left\"> 3</td><td align=\"center\">0.98 (0.75–1.30)</td><td align=\"center\">0.96 (0.69–1.33)</td><td/></tr><tr><td align=\"left\"> 4</td><td align=\"center\">1.04 (0.75–1.45)</td><td align=\"center\">1.17 (0.81–1.68)</td><td/></tr><tr><td align=\"left\"> ≥ 5</td><td align=\"center\">0.94 (0.66–1.33)</td><td align=\"center\">0.89 (0.59–1.35)</td><td align=\"center\">0.9796</td></tr><tr><td align=\"left\">Long-term limitations on activity</td><td/><td/><td/></tr><tr><td align=\"left\">Absent</td><td align=\"center\">1.0</td><td align=\"center\">1.0</td><td/></tr><tr><td align=\"left\">Present</td><td align=\"center\">0.68 (0.50–0.93)</td><td align=\"center\">0.72 (0.51–1.01)</td><td align=\"center\">0.0424</td></tr><tr><td align=\"left\">Self-rated general health status</td><td/><td/><td/></tr><tr><td align=\"left\">Good to excellent</td><td align=\"center\">1.0</td><td align=\"center\">1.0</td><td/></tr><tr><td align=\"left\">Poor to average</td><td align=\"center\">0.66 (0.50–0.87)</td><td align=\"center\">0.51 (0.36–0.70)</td><td align=\"center\">0.0002</td></tr><tr><td align=\"left\">Psychological distress level</td><td/><td/><td/></tr><tr><td align=\"left\">Low to average</td><td align=\"center\">1.0</td><td align=\"center\">1.0</td><td/></tr><tr><td align=\"left\">High</td><td align=\"center\">0.82 (0.68–0.98)</td><td align=\"center\">0.73 (0.59–0.90)</td><td align=\"center\">0.0105</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T4\"><label>Table 4</label><caption><p>Odds ratios, <italic>P </italic>values, and 95% confidence intervals from multinomial logit model linking multimorbidity based on Kaplan et al study's set of diseases, long-term limitations on activity, self-rated health status, and psychological distress levels to physical activity among men.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"3\"><bold>Odds ratio (95% confidence intervals)</bold></td></tr><tr><td/><td colspan=\"3\"><hr/></td></tr><tr><td align=\"left\"><bold>Independent variables</bold></td><td align=\"center\"><bold>Physical activity &lt; 3 times per week</bold></td><td align=\"center\"><bold>Physical activity ≥ 3 times per week</bold></td><td align=\"center\"><bold><italic>P </italic>value</bold></td></tr></thead><tbody><tr><td align=\"left\">Number of chronic diseases</td><td/><td/><td/></tr><tr><td align=\"left\"> 0</td><td align=\"center\">1.0</td><td align=\"center\">1.0</td><td/></tr><tr><td align=\"left\"> 1</td><td align=\"center\">1.02 (0.84–1.22)</td><td align=\"center\">0.78 (0.54–1.12)</td><td/></tr><tr><td align=\"left\"> ≥ 2</td><td align=\"center\">1.01 (0.81–1.26)</td><td align=\"center\">0.88 (0.59–1.31)</td><td align=\"center\">0.6851</td></tr><tr><td align=\"left\">Long-term limitations on activity</td><td/><td/><td/></tr><tr><td align=\"left\">Absent</td><td align=\"center\">1.0</td><td align=\"center\">1.0</td><td/></tr><tr><td align=\"left\">Present</td><td align=\"center\">0.65 (0.46–0.92)</td><td align=\"center\">0.66 (0.44–0.97)</td><td align=\"center\">0.0298</td></tr><tr><td align=\"left\">Self-rated general health status</td><td/><td/><td/></tr><tr><td align=\"left\">Good to excellent</td><td align=\"center\">1.0</td><td align=\"center\">1.0</td><td/></tr><tr><td align=\"left\">Poor to average</td><td align=\"center\">0.73 (0.54–0.99)</td><td align=\"center\">0.51 (0.36–0.74)</td><td align=\"center\">0.0014</td></tr><tr><td align=\"left\">Psychological distress level</td><td/><td/><td/></tr><tr><td align=\"left\">Low to average</td><td align=\"center\">1.0</td><td align=\"center\">1.0</td><td/></tr><tr><td align=\"left\">High</td><td align=\"center\">1.20 (0.98–1.48)</td><td align=\"center\">0.98 (0.76–1.28)</td><td align=\"center\">0.0913</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T5\"><label>Table 5</label><caption><p>Odds ratios, <italic>P </italic>values and 95% confidence intervals from multinomial logit model linking multimorbidity based on Kaplan et al study's set of diseases, long-term limitations on activity, self-rated health status, and psychological distress levels to physical activity among women.</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"3\"><bold>Odds ratio (95% confidence intervals)</bold></td></tr><tr><td align=\"left\"><bold>Independent variables</bold></td><td align=\"center\"><bold>Physical activity &lt; 3 times per week</bold></td><td align=\"center\"><bold>Physical activity ≥ 3 times per week</bold></td><td align=\"center\"><bold><italic>P </italic>value</bold></td></tr></thead><tbody><tr><td align=\"left\">Number of chronic diseases</td><td/><td/><td/></tr><tr><td align=\"left\"> 0</td><td align=\"center\">1.0</td><td align=\"center\">1.0</td><td/></tr><tr><td align=\"left\"> 1</td><td align=\"center\">1.05 (0.88–1.26)</td><td align=\"center\">1.05 (0.85–1.29)</td><td/></tr><tr><td align=\"left\"> ≥ 2</td><td align=\"center\">0.95 (0.71–1.27)</td><td align=\"center\">0.84 (0.59–1.18)</td><td align=\"center\">0.7332</td></tr><tr><td align=\"left\">Long-term limitations on activity</td><td/><td/><td/></tr><tr><td align=\"left\">Absent</td><td align=\"center\">1.0</td><td align=\"center\">1.0</td><td/></tr><tr><td align=\"left\">Present</td><td align=\"center\">0.68 (0.51–0.92)</td><td align=\"center\">0.73 (0.53–1.01)</td><td align=\"center\">0.0359</td></tr><tr><td align=\"left\">Self-rated general health status</td><td/><td/><td/></tr><tr><td align=\"left\">Good to excellent</td><td align=\"center\">1.0</td><td align=\"center\">1.0</td><td/></tr><tr><td align=\"left\">Poor to average</td><td align=\"center\">0.66 (0.50–0.87)</td><td align=\"center\">0.52 (0.37–0.72)</td><td align=\"center\">0.0003</td></tr><tr><td align=\"left\">Psychological distress level</td><td/><td/><td/></tr><tr><td align=\"left\">Low to average</td><td align=\"center\">1.0</td><td align=\"center\">1.0</td><td/></tr><tr><td align=\"left\">High</td><td align=\"center\">0.81 (0.68–0.97)</td><td align=\"center\">0.72 (0.59–0.90)</td><td align=\"center\">0.0084</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[]

{ "acronym": [], "definition": [] }

[]

[]

[]

[]

[]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>" ]

[ "<title>Editorial</title>", "<p>The Signal Transduction Society (STS) [##UREF##0##1##] is delighted to join BioMed Central with an open access journal. Over the last years, our society members have increasingly appreciated that access to scientific information generated by publicly funded academic research must not be restricted by commercial interests. With overwhelming support of the society members, the presidial council and advisory board of the STS are therefore now taking action and moving from an access-restricted print journal to online open access publishing.</p>", "<p>We are convinced that unduly limiting the flow of scientific knowledge has a negative impact on the development of benefits for mankind and believe that BioMed Central is a highly valuable platform that is vital not only to scientists but to society in general and hence deserves our strong support.</p>", "<p>The need for free information flow within the academic community is especially true for rapidly evolving fields such as cell signaling, which is now entering a new and exciting era of development. The foundations in our molecular understanding, laid by numerous colleagues over the last decades, are now beginning to bear fruit in the form of an increasing number of targeted, signal transduction-modulating drugs entering clinical use. Over a dozen examples can be found in current cancer therapies alone, with more than a hundred further anti-cancer drug candidates currently undergoing clinical evaluation. Nevertheless, much still needs to be done to improve the outcomes for countless severely suffering patients.</p>", "<p><italic>Cell Communication and Signaling </italic>(<italic>CCS</italic>) and the STS are committed to playing an active part in the signal transduction research community by providing a stimulating and collegial forum for the speedy publication of state-of-the art research and discussions.</p>", "<p><italic>CCS </italic>intends to cover all aspects of cell signaling in health and disease, from ultrastructural and molecular studies with atomic resolution to signaling pathways, molecular machines and signaling networks and further on to various forms of inter-organismal signaling as, for example, exerted by bacterial and viral pathogens. We will accept timely reviews and commentaries as well as original articles with high-quality data in several formats (further details below). Supported by the journal's editorial board members and solicited expert reviewers, the editors will make it a priority to ensure that publications are not unduly delayed. It has become an aggravating fact of life for many scientists that even well-prepared manuscripts are rarely accepted without requests to perform multiple additional experiments. Some manuscripts undergo four or five rounds of resubmission, often resulting in publication delays for several years. This does not go down well with many funding bodies and puts especially young investigators at a disadvantage. The friendly reviewers' comment \"This is a great paper well worth publishing as soon as possible provided the authors can submit further supporting data from a mouse knock-in and a genome-wide gene expression array analysis of no less than 500 patient samples\" has become a standing joke at many scientific gatherings. In far too many cases, delaying manuscripts has less to do with improving manuscript quality and more with conflicting interests of various kinds.</p>", "<p><italic>CCS </italic>will not be a part of this. We recognise that science is a continuous process that is never finished and that real proof comes primarily from independent confirmations by other researchers. This is often not possible unless a study, albeit incomplete, is publicly accessible.</p>", "<p>If an original manuscript's data are novel, important and technically sound, if the conclusions drawn are not severely flawed, the quality of results and presentation are state-of-the-art and the ethics acceptable, the work will be published. The editors do, however, reserve the right to add an editorial comment if they feel that this is important for the scientific community. Authors may be invited to comment on these. Reviewers will be requested to provide fair and constructive criticism that helps the authors to improve their manuscripts.</p>", "<p>We intend to publish the following types of contributions: original research, reviews, commentaries, debate articles, hypotheses, methodology articles and short reports. Further details regarding these types of article is available via the <italic>Cell Communication and Signaling </italic>about page [##UREF##1##2##].</p>", "<p>As many readers will already know, it is not possible to make high-quality open access publishing a reality without a cost contribution of the authors. As such, <italic>Cell Communication and Signaling </italic>levies an article-processing charge (APC) for each accepted manuscript. There are currently over 300 institutions with BioMed Central membership [##UREF##2##3##] whereby the APC will be fully or partially covered by the membership, and a number of funding bodies allow their grants to be used to cover the APC [##UREF##3##4##]. Authors can request a waiver of the APC through the submission system and these will be considered where there is a genuine inability to pay.</p>", "<p>With the support of the STS, of our editorial board members and of our valued colleagues around the globe, we believe that <italic>CCS </italic>will achieve its goal, which is to become an appreciated place to quickly disseminate exciting findings from and to the signaling community and beyond. We are looking forward to working with you all.</p>", "<p>With warm regards,</p>", "<p>Stephan M. Feller, Editor</p>", "<p>Ralf Hass, President STS &amp; Associate Editor</p>", "<p>Ottmar Janssen, Presidial Council STS &amp; Associate Editor</p>", "<p>Karlheinz Friedrich, Presidial Council STS</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Krüppel-like factor 9 (KLF9), previously referred to as <underline>B</underline>asic <underline>T</underline>ranscription <underline>E</underline>lement <underline>B</underline>inding (BTEB) Protein 1, is an evolutionarily well-conserved member of the Krüppel-like factor (KLF) family of transcriptional regulators, so named for the presence of <italic>Drosophila </italic>Krüppel-like DNA-binding domain(s) in the protein's C-terminal region [##REF##15820306##1##]. KLFs function as transcriptional activators or repressors, depending upon cellular context and presence of partner co-regulators [##REF##15820306##1##,##REF##17508399##2##]. Individual KLF proteins affect cell proliferation, differentiation, apoptosis, DNA damage, and stress responses [##REF##17508399##2##]. Members of this family have been implicated in stem cell renewal, maintenance of pluripotency, lineage determination, organogenesis, and oncogenesis [##REF##17508399##2##, ####REF##18035408##3##, ##REF##18264089##4##, ##REF##18157115##5####18157115##5##], underscoring their wide-ranging regulatory roles in development.</p>", "<p>KLF9 was first isolated and identified as trans-repressor of the rat liver CYP1A1 gene and inducer of SV40 early and HIV-1 long terminal repeat promoters [##REF##1356762##6##,##REF##8257632##7##]. Subsequent work demonstrated KLF9 activation of the liver CYP7A gene [##REF##9875228##8##]. Albeit trans-activation and trans-repression functions of KLF9 are mediated by formation of different transcriptional protein complexes, they occur through KLF9 binding to highly similar GC/GT boxes [##REF##17508399##2##,##REF##1356762##6##,##REF##8257632##7##]. KLF9 mRNA is expressed at highest levels in rat brain, kidney, lung and testis [##REF##1356762##6##]. Potential brain functions for KLF9 have been recently elucidated. This transcription factor is induced in rat brain by 3, 5, 3'-triiodothyronine (T₃) and mediates effects of T₃on neuronal process development [##REF##10438482##9##, ####REF##12395077##10##, ##REF##12021188##11####12021188##11##]. Mice lacking KLF9 exhibited deficits in motor learning, motor coordination and fear-conditioning [##REF##12640131##12##]. Recently, KLF9 was shown to regulate crypt-villus cell renewal in mice [##REF##17379758##13##] and secondary antibody responses of human splenic B cells [##REF##17673551##14##].</p>", "<p>The uterus is a complex organ that exhibits hormonally (estrogen, progesterone) driven changes in the expression of myriad genes and gene products during the estrous cycle and pregnancy. Morphologically, the uterus is comprised of epithelium (glandular and luminal), stromal fibroblasts, immune cells, and myometrium. Regulatory actions of KLF9 in the uterus during development and pregnancy have been partially elucidated by our laboratories. Mice null for KLF9 exhibited uterine hypoplasia, smaller litter size, reduced numbers of implantation sites, partial progesterone resistance in the uterus, and delayed parturition [##REF##15117941##15##,##REF##18305227##16##]. The sub-fertility phenotype of the KLF9 knockout mouse was shown to correlate with aberrant timing of uterine stromal and epithelial proliferation during the peri-implantation period, suggesting an out-of-phase uterus relative to blastocyst development as potentially causative for sub-fertility of KLF9 null females [##REF##15917344##17##]. KLF9 facilitated progesterone-inductive effects on uterine gene expression by its co-recruitment with the progesterone receptor [##REF##12672823##18##,##REF##16384861##19##] and inhibited estrogen receptor α trans-activity by promoting this receptor's estrogen-induced down-regulation [##REF##17717078##20##].</p>", "<p>In previous studies, analysis of estrogen receptor-negative human endometrial carcinoma (HEC-1-A) cells that were genetically engineered to over- or under-express KLF9 identified serum-dependent mitogenic functions for this nuclear protein, based on its influence on cell phenotype and gene expression changes that support increased proliferation [##REF##11476943##21##,##REF##11953011##22##]. HEC-1-A cells with increased KLF9 expression grew as flat monolayers, whereas those with less KLF9 tended to round up on plastic, formed multi-layers, and exhibited dome formation [##REF##11476943##21##]. Because the repertoire of genes identified in that study was limited and did not fully explain the potentially opposing functions of KLF9 on cell adhesion and proliferation, defects of which may underscore cell survival, migration, invasion, and tumorigenesis, we have now performed global expression profiling of the HEC-1-A sub-lines that over- or under-express KLF9. Results demonstrate prominent effects of KLF9 on genes encoding basement membrane and ECM proteins, cell stress response and detoxification pathway members, uterine endometrial steroid- and estrous cycle-regulated proteins, and membrane- and nuclear-associated receptors. These findings correlate with the attenuated expression of KLF9 with high endometrial tumor grade, thereby suggesting the potential involvement of KLF9 dys-regulation in both pregnancy failure and endometrial pathogenesis.</p>" ]

[ "<title>Methods</title>", "<title>Cell lines</title>", "<p>The generation and initial characterization of the cell lines used in this study were previously described [##REF##11476943##21##,##REF##11953011##22##]. The 4S and 9S sub-lines were derived from the original human HEC-1-A endometrial carcinoma cell line and contain a stably transfected expression plasmid encoding full-length rat KLF9, whereas the 2AS and 3AS sub-lines contain a stably transfected plasmid encoding an antisense (AS) RNA to KLF9. The primary sequence of human and rat KLF9 coding regions are 98% identical [##REF##8291025##23##]. All cell lines were derived concurrently [##REF##11476943##21##]. The 'sense' (S) and 'antisense' (AS) cell lines differed in relative levels (S &gt; AS) of KLF9 protein and activity [##REF##11476943##21##]. S sub-lines expressed ~2.4 fold greater immunoreactive KLF9 than AS sub-lines [##REF##11476943##21##]. The 4S/2AS and 9S/3AS sub-lines were propagated separately to near confluence and RNA was harvested using TRIzol reagent (Invitrogen, Carlsbad, CA).</p>", "<title>Affymetrix GeneChip technology and bio-informatics</title>", "<p>RNA was purified using the RNeasy Mini Kit (QIAGEN, Valencia, CA) followed by on-column DNA digestion with RNase-Free DNase (QIAGEN). RNA integrity was confirmed using the RNA 6000 Nano LabChip (Agilent Biotechnologies, Palo Alto, CA). Two replicate cRNA targets were generated from each HEC-1-A sub-line RNA preparation; all eight targets were made at the same time. Total cellular RNA (8 ug) was converted to cDNA using a T7-(deoxythymidine)₂₄primer and Superscript II (Life Technologies, Inc., Gaithersburg, MD). Resulting cDNA was used with the ENZO BioArray High Yield RNA Transcript labeling kit (ENZO, Farmingdale, NY) to synthesize biotin-labeled cRNA; the latter was purified on an RNeasy spin column (QIAGEN) and chemically fragmented to a size range of 35 to 200 bp. cRNAs were concurrently hybridized to HG-U133A GeneChips (Affymetrix, Santa Clara, CA). Hybridizations were performed for 16 hours, followed by incubations with streptavidin-conjugated phycoerythrin, and polyclonal anti-streptavidin antibody coupled to phycoerythrin. GeneChips were scanned using an Agilent GeneArray laser scanner and images analyzed using Affymetrix MAS 5.0 software. Bacterial sequence-derived probes on the arrays served as external controls for hybridization, whereas the housekeeping genes β-actin and GAPDH served as endogenous controls and for monitoring the quality of the RNA targets.</p>", "<p>Unsupervised nearest-neighbor hierarchical clustering (Spotfire DecisionSite, Somerville, MA) identified a significant effect of culture condition/date of RNA collection on overall gene expression profiles. Therefore, to identify candidate KLF9 gene targets, the following was separately performed on the combinations of 4S/2AS and 9S/3AS. Intensity values of probe sets were imported into GeneSpring Gx 7.3 software for analysis. Values were processed using the Robust Multiarray Analysis algorithm for background adjustment, normalization and log₂-transformation of perfect match values. Data were subjected to per-chip and per-gene normalization and analyzed for differences between cell lines (fold-change, S relative to AS, value of 1.3 or higher; <italic>P </italic>&lt; 0.05, Student's <italic>t </italic>test). Transcripts that passed these filters for both the 4S/2AS and 9S/3AS cell line combinations comprised the final KLF9-regulated gene lists and were compared for gene overlaps. The final gene list (comprised only of overlapping transcripts) was annotated using NETAFFX [##UREF##0##24##] and NCBI <italic>Entrez </italic>[##UREF##1##25##]. The microarray data have been deposited in Gene Expression Omnibus [##UREF##2##26##] as series GSE11855.</p>", "<title>Quantitative real-time RT-PCR (qRT-PCR)</title>", "<p>One μg of total cellular RNA from each cell line was reverse-transcribed using random hexamers and MultiScribe Reverse Transcriptase in a two-step RT-PCR reaction (Applied Biosystems, Foster City, CA). Primers (Table ##TAB##0##1##) were designed using 'Primer Express' (Applied Biosystems) to yield a single amplicon; this was verified by dissociation curves. SYBR Green real-time PCR was performed with an ABI Prism 7000 Sequence Detector or Bio-Rad MyiQ Real-Time PCR Detection System. Thermal cycling conditions included pre-incubation at 50°C for 2 min, 95°C for 10 min followed by 40 PCR cycles at 95°C for 15 sec and 60°C for 1 min. Relative transcript levels were calculated using the relative standard curve method (User Bulletin #2, Applied Biosystems) and results were normalized to 18 S rRNA. Data are reported as mean ± SEM and were analyzed by one-way ANOVA (SigmaStat; Systat Software, Inc., Point Richmond, CA). <italic>P </italic>&lt; 0.05 was considered to represent a significant difference.</p>", "<p>A TissueScan™ endometrial cancer qPCR array was purchased from OriGene Technologies, Inc. (Rockville, MD). This panel was comprised of a series of normalized cDNAs prepared from pathologist-verified human endometrium and endometrial tumors. Ages of tissue donors ranged from 30 to 87 years.</p>" ]

[ "<title>Results</title>", "<title>Gene expression profiling</title>", "<p>The 4S/2AS and 9S/3AS pairs of sub-lines were grown on separate occasions to ~80% confluence. Total cellular RNAs from the four sub-lines were concurrently subjected to global gene expression profiling, with each RNA sample analyzed in duplicate. Unsupervised hierarchical clustering of gene expression profiles for all transcripts indicated a significant effect of time of experiment (i.e., when sub-lines were cultured) (data not shown). We therefore performed bio-informatic comparisons of 2AS vs. 4S and 3AS vs. 9S separately, and then searched for overlap in the differentially expressed transcripts between each paired comparison. There were more genes expressed at higher levels in KLF9 sense (S) than antisense (AS) sub-lines among the differentially expressed transcripts identified (Figure ##FIG##0##1##). However, whereas the numbers of KLF9-induced (160 and 149 for 4S and 9S, respectively) and KLF9-suppressed (76 and 93 for 2AS and 3AS, respectively) transcripts for each pair of sub-lines were comparable, less than half of these were found to overlap in both final gene lists for S (total of 60) and AS (24) sub-lines. As a conservative estimate, these overlapping transcripts were taken to reflect KLF9 activity and/or expression levels in the clonal sub-lines. Unsupervised hierarchical clustering of these signature transcripts demonstrated the close relatedness of the S sub-lines expression profiles (Figure ##FIG##1##2##). As expected, the expression profiles of the AS sub-lines differed from the S sub-lines; however, they also differed from each other (Figure ##FIG##1##2##).</p>", "<p>Transcripts that varied in expression across the two pairs of sub-lines were annotated for their known or putative function(s), fold-change, and cellular locations (additional file ##SUPPL##0##1##: Table 1 and additional file ##SUPPL##1##2##: Table 2). KLF9 under-expression in HEC-1-A cells (i.e., AS sub-lines) caused induction in relatively few (twenty four) genes. Six of these genes/transcripts were examined by qRT-PCR and all were confirmed to be induced in the AS sub-lines (Table ##TAB##1##2##). In general, for a given transcript/gene there was excellent agreement between the fold-change calculated from microarray and that from qRT-PCR (Table ##TAB##1##2##). These KLF9-suppressed (directly or indirectly) mRNAs encode proteins which are important participants in: aldehyde metabolism (AKR7A2, ALDH1A1), regulation of the actin cytoskeleton and cell motility (e.g., ANK3, ITGB8), cellular detoxification (SULT1A1, ABCC4), cellular signaling (e.g., ACBD3, FZD5, RAB25, CALB1), and transcriptional regulation (PAX2, STAT1). In addition, two hypothetical proteins (C12orf29, C1orf186) are encoded by mRNAs whose levels were increased with KLF9 suppression.</p>", "<p>Sixty mRNAs were more abundant in 4S and 9S than 2AS and 3AS sub-lines (Figure ##FIG##0##1##). We examined a subset of these (fourteen in total) by qRT-PCR and confirmed their differential expression (Table ##TAB##1##2##). KLF9-induced mRNAs encode proteins which participate in: serine biosynthesis (e.g., PSAT1, a major progesterone-induced protein of the rabbit uterus); regulation and function of the actin cytoskeleton (COTL1, FSCN1, FXYD5, MYO10); cell adhesion, extracellular matrix and basement membrane formation (e.g., AMIGO2, COL4A1, COL4A2, LAMC2, NID2); transport (CLIC4); cellular signaling (e.g., BCAR3, MAPKAPK3); transcriptional regulation [e.g., KLF4, NR3C1 (glucocorticoid receptor), RXRα]; growth factors/cytokine actions (SLPI, BDNF); or are membrane-associated proteins and receptors (e.g., CXCR4, PTCH1); and metallothioneins (Table ##TAB##1##2##; additional file ##SUPPL##1##2##: Table 2). In addition, two mRNAs that encode hypothetical proteins (C10orf38, C9orf167) were increased in abundance with KLF9 over-expression.</p>", "<title>Novel human membrane proteins expressed in HEC-1-A cells</title>", "<p>We confirmed the reduction in abundance of the mRNA encoding C1orf186 in the KLF9 S sub-lines by qRT-PCR (Table ##TAB##1##2##). C1orf186 protein contains 172 amino acid residues and has a predicted mol wt of 19,405 and an isoelectric point of 4.53. Its sequence bears hallmarks of a single pass trans-membrane protein with multiple potential serine, threonine and tyrosine phosphorylation sites in its intracellular face (data not shown). In humans, this transcript is highly expressed in the uterus where its abundance is regulated by stage of the menstrual cycle [##UREF##1##25##,##UREF##2##26##]. A second novel membrane protein-encoding mRNA identified from our microarray results was that for C9orf167. In contrast to C1orf186 mRNA, C9orf167 mRNA was induced by KLF9 in HEC-1-A cells. The protein encoded by C9orf167 is comprised of 422 amino acids in the human and is an uncharacterized member of the clpA/clpB family, Torsin subfamily of proteins. The paralogous protein in Chimpanzee is 99.8% similar, while the mouse protein is 84.9% similar to that for the human [##UREF##1##25##]. This transcript is ubiquitous in human and mouse tissues with abundant expression in the uterus; albeit not as dramatic as for C1orf186, uterine expression of C9orf167 also is regulated by stage of menstrual cycle in women [##UREF##1##25##,##UREF##2##26##].</p>", "<title>KLF9 mRNA expression in human uterine tumors</title>", "<p>Endometrial cancer is associated with unopposed estrogen activity. In Ishikawa endometrial cancer cells, KLF9 opposed ER-α activity and decreased this protein's expression [##REF##17717078##20##]. Given that KLF9 over-expression increased proliferation [##REF##11476943##21##], and induced the expression of transcripts encoding multiple ECM and basement membrane components for increased adhesion to substratum (additional file ##SUPPL##0##1##: Table 1 and additional file ##SUPPL##1##2##: Table 2) in HEC-1-A cells, it was of interest to examine human endometrial tumors for expression of KLF9 mRNA. A reduction in KLF9 mRNA abundance in combined stages II through IV tumors, compared to normal endometrium and stage I tumors was observed (Fig. ##FIG##2##3##).</p>" ]

[ "<title>Discussion</title>", "<p>The present study extended our initial characterization of HEC-1-A sub-lines which were genetically engineered to have enhanced or reduced expression of KLF9, relative to the parental HEC-1-A cell line [##REF##11476943##21##]. Concordant with KLF9's dual function as a transcriptional activator and transcriptional repressor, a range of genes involved in distinct signal transduction pathways was induced or repressed by KLF9. Using the unbiased microarray approach, we found significantly more KLF9-induced than KLF9-repressed transcripts in HEC-1-A sub-lines. Although the biological significance of this finding is unclear, it raises the important question of whether cellular context defines in part, the transcriptional direction of KLF9 activity, and whether this difference may be underscored by distinct interactions of KLF9 with various co-activators and co-repressors also present. The degree of overlap of transcripts noted between the two pairs of compared sub-lines was less than anticipated and may reflect the arbitrary cut-offs used for bio-informatic analysis, the effects of small differences in culture conditions on mRNA expression, and the non-specific effects of the process of clonal sub-line derivation.</p>", "<p>In previous studies, we showed that the S lines with higher KLF9 expression and mitotic index grew as monolayers, whereas the AS lines with lower KLF9 expression and mitotic index tended to round up on plastic, formed multi-layers, and exhibited dome formation [##REF##11476943##21##], the latter potentially indicative of enhanced tumor invasive capability and metastasis. Our current findings identified putative KLF9-regulated genes whose proteins are known to contribute to basement membrane formation, ECM formation, cell adhesion, and cytoskeletal organization and regulation. Of note was the KLF9 up-regulated expression of the LAMC2 gene, encoding the basement membrane laminin gamma 2 protein, which is important for epithelial anchoring to substratum. While regulation of other laminin genes by other KLF family members has been previously reported [##REF##12034813##27##,##REF##14634001##28##], our findings constitute the first linkage of KLF9 and laminin gene expression for any cell type. Basement membranes and extracellular matrix are dynamic entities that do more than just passively anchor cells to surfaces; they are, for example, important modifiers of cellular growth, differentiation, apoptosis, <italic>anoikis </italic>and migration. Thus, loss of KLF9 which regulates secreted and membrane-associated proteins such as laminin, might constitute an early event in the epithelial-mesenchymal transition, leading to invasion and subsequent metastasis. Consistent with this, endometrial tumors of higher grades (and with greater invasion and metastasis) had decreased KLF9 gene expression compared to normal endometrium and stage I tumors. While this association is not particularly strong and requires additional confirmation with increased sample size and evaluation at the protein level, our data mirrors a recent report that decreased KLF9 expression also is a feature of increased grade of colon tumors [##REF##18477211##29##].</p>", "<p>Previously, we observed increased KLF5 mRNA abundance in HEC-1-A cells after KLF9 over-expression, with no differences in mRNA level for Sp1, the best characterized transcriptional activator of the Sp/KLF family [##REF##11953011##22##]. In the present study, this positive association with KLF9 was similarly noted for KLF4. In colon crypt epithelium, KLF4 mediates growth arrest and induction of the colonic epithelial phenotype, whereas KLF5 stimulates proliferation of crypt progenitor cells [##REF##8702718##30##,##REF##11005769##31##]. By contrast, KLF13, the most closely related family member to KLF9 [##REF##15820306##1##] was found to be inversely associated with KLF9 expression in uteri of KLF9 null mice, possibly reflective of partial functional compensation [##REF##15117941##15##]. While the roles of KLF4, KLF5 and KLF 13 in normal and tumorigenic uterine epithelium and their regulation by KLF9 are as yet unknown, the presence of all three members in the uterus suggests a KLF network where these proteins' cooperative, opposing, or compensatory functions in the physiological contexts of uterine remodeling, embryo implantation, tumorigenesis, and stem cell regulation may be manifest. Indeed, functional KLF networks have been reported in the maintenance of the intestinal epithelial cell phenotype [##REF##12087155##32##], self renewal of embryonic stem cells [##REF##18264089##4##], and secondary antibody response in memory B cells [##REF##17673551##14##].</p>", "<p>Previous work from our laboratories has implicated KLF9 in progesterone and estrogen actions in the uterus [##REF##15117941##15##,##REF##15917344##17##,##REF##16384861##19##,##REF##11751593##33##]. The present results further support these linkages. PSAT1, an <italic>in vivo </italic>progesterone-induced gene in rabbit uterus [##REF##3651428##34##], was more highly expressed in the KLF9 over-expressing HEC-1-A cells. We also found repression of ALDH1 gene expression and increased levels of CXCR4 and SLPI gene transcripts in KLF9 S cells. In ovariectomized mice, uterine ALDH1A1 mRNA was rapidly induced by progesterone, whereas CXCR4 transcripts were down-regulated by chronic progesterone treatment [##REF##15845616##35##]. SLPI is a progesterone-induced, secreted anti-protease with anti-microbial and mitogenic activities for uterine epithelial cells [##REF##10411515##36##, ####REF##12023969##37##, ##REF##15642791##38####15642791##38##]. Expression of the uterine SLPI gene/protein is increased around the time of implantation in rodents, pigs, primates and humans in response to estrogen and progesterone [##REF##9475401##39##, ####REF##14521952##40##, ##REF##12855598##41##, ##REF##15239838##42##, ##REF##15044260##43####15044260##43##]. Given the distinct patterns of expression of CXCR4, ALDH1A1 and SLPI genes with KLF9 in vitro and with progesterone in vivo, results suggest the functional contribution of KLF9 to progesterone receptor signaling. In this regard, a recent study from our laboratories demonstrated the progesterone-dependent co-recruitment of KLF9 and the progesterone receptor to the SLPI promoter, concomitant with induction of this gene's expression, in Ishikawa endometrial cancer cells [##REF##16384861##19##].</p>", "<p>We observed that Patched 1 (PTCH1) transcripts were induced in KLF9 S sub-lines. This gene encodes the receptor for the morphogen Sonic Hedgehog (SHH), a known stimulator of endometrial cell proliferation <italic>in vitro </italic>[##REF##17332280##44##]. The related molecule, Indian Hedgehog (IHH), is a progesterone-induced molecule in the mouse uterus and mediates effects of progesterone on uterine epithelial-stromal cell interactions essential for implantation [##REF##16951680##45##]. CXCR4 encodes a chemokine receptor that is up-regulated during the window of implantation in human endometrium and which may facilitate blastocyst adhesion to the uterine surface [##REF##12651900##46##, ####REF##12709027##47##, ##REF##17021345##48##, ##REF##18077318##49####18077318##49##]. The CXCR4 ligand, namely stromal cell-derived factor-1α (SDF-1α), promotes proliferation of HEC-1-A cells <italic>in vitro </italic>through Akt and ERK1/2 signaling pathways [##REF##16884765##50##]. Thus, up-regulated expression of PTCH1 and CXCR4 mRNAs with over-expression of KLF9 may explain the increased mitogenesis of KLF9 S cell lines observed <italic>in vitro </italic>[##REF##11476943##21##,##REF##11953011##22##]. CXCR4 is also a co-receptor for HIV-1, and uterine expression of HIV receptors/co-receptors may underlie HIV transmission to women [##REF##12709027##47##]. Moreover, KLF9 is itself a transcriptional inducer of the HIV-1 LTR [##REF##8257632##7##]. The latter data raise the provocative question of whether KLF9 affects uterine and vaginal uptake and replication of HIV-1 <italic>in vivo</italic>, and if so, whether KLF9 might constitute a genetic modifier of HIV risk. Further studies are required to evaluate this possibility as well as the mode of regulation of CXCR4 by KLF9 in the female reproductive tract.</p>", "<p>The power of microarrays is that they often reveal new participants and novel functional connections in specific physiological contexts. We found several interesting examples of these in the present work. Brain-derived neurotrophic factor (BNDF) was observed to be up-regulated in KLF9 S sub-lines, suggesting BNDF as a downstream effector of KLF9 signaling. Recently, this soluble factor was implicated as a mediator of cyclic changes in sympathetic innervation in the uterus and in uterine-embryo communication [##REF##14656325##51##,##REF##17880937##52##]. We also found that the Frizzled gene FZD5, which encodes the receptor for the secreted glycoprotein Wnt 5B, normally expressed in human endometrium [##REF##16306079##53##] is suppressed by KLF9 in HEC-1-A cells; suggesting the contribution of KLF9 to the control of this growth factor-receptor pathway.</p>", "<p>We noted the induction of several metallothionein genes (MT1H, MTH1X, and MT2A) with KLF9 over-expression. Metallothioneins are expressed in cyclic fashion in human endometrium with most abundant expression during the secretory phase, where they may play protective roles during cell stress [##REF##16306079##53##,##REF##10878548##54##]. The hyaluronic acid receptor CD44 mRNA also was increased in abundance with KLF9 over-expression. This molecule is regulated in human endometrium by stage of menstrual cycle (maximal during mid and late secretory phases) [##REF##7543442##55##]. Versican mRNAs are up-regulated in human endometrium during the mid-secretory phase [##REF##16306079##53##]; accordingly, we noted increased expression of this gene with KLF9 over-expression. Thus, with these genes, KLF9 transactivation was correlated with secretory phase uterine expression. However, the apparent repression of SULT1A1 gene by KLF9 in HEC-1-A cells contrasted with the increased abundance of this gene during the secretory relative to the proliferative phase of the human endometrium [##REF##10541560##56##].</p>", "<p>A prominent role for KLF9 in cellular response to chemotherapeutic agents and other cell stressors is suggested from the present results. As noted above, metallothioneins, which were KLF9-induced, likely have a protective role during endometrial cell stress [##REF##10878548##54##]. KLF9 over-expressing HEC-1-A cells had increased mRNA expression of BCAR3 and PSAT1, proteins previously implicated in the acquisition of tamoxifen resistance by breast cancer cells [##REF##17616674##57##,##REF##15899800##58##]. PSAT1 also is involved in the process by which colon cancer cells acquire chemotherapy resistance [##REF##18221502##59##]. Expression of the KLF9 gene target SLPI is induced in endometria of women receiving tamoxifen [##REF##16322341##60##]. Thus, KLF9 through several of its downstream targets may participate in the pathway(s) and network(s) by which tamoxifen becomes estrogenic and thereby, tumorigenic for a subset of pre-disposed uterine endometrial epithelial cells.</p>", "<p>KLF9 over-expressing HEC-1-A cells had increased abundance of MAPKAPK3, TRIB3 and ELK3 mRNAs; the corresponding proteins participate in cell stress-response pathways [##REF##8943323##61##]. Interestingly, MAPKAPK3, a Ser-Thr kinase, is activated by serum and other growth inducers and is a point of convergence of ERK, p38MAP kinase and Jun N-terminal kinase (JNK) signaling pathways [##REF##8943323##61##]. By inducing expression of MAPKAPK3, KLF9 may enhance basal and/or overall activity of one or more of these important pathways. Consistent with these findings, we previously demonstrated that KLF9 over-expression caused: 1) HEC-1-A cells to become more mitogenically responsive to serum and TGF-β1 [##REF##11476943##21##], and 2) the IGFBP2 gene and its promoter to be potently induced by serum [##REF##11953011##22##]. Moreover, curcumin, an inhibitor of JNK, blocked the inductive effect of KLF9 on serum-stimulation of the IGFBP2 gene promoter [##REF##11953011##22##]. Another group has reported that acetaldehyde induced KLF9 expression in rat hepatic stellate cells via activation of the JNK pathway [##REF##10733585##62##]. Altogether, these observations invoke KLF9 as a nuclear participant in the JNK pathway and the cellular responses that it subserves. Thus, in keeping with a functional KLF network in endometrial epithelial cells, KLF9 joins its nuclear brethren, KLF4 and KLF5, as potentially important mediators of cellular stress response [##REF##17508399##2##].</p>", "<p>Results implicate KLF9 in the transcriptional regulation of key pathways that sub-serve proliferation, adhesion, migration, stress, response to chemotherapeutic agents, embryo attachment and implantation, and uterine remodeling. Coupled with the ability of KLF9 to interact with other members of the Sp/KLF family, progesterone receptors, and receptor co-activators and co-repressors, our data suggest expanded possibilities for KLF9 to affect distinct physiological processes as well as the genesis of endometrial tumors. Given that KLF9 null mice are viable with a relatively mild uterine phenotype, further study of KLF9 is warranted to clearly define its unique functions in normal uterine biology and in endometrial neoplasia distinct from those potentially compensated by closely related KLFs.</p>" ]

[ "<title>Conclusion</title>", "<p>Microarray profiling of HEC-1-A sub-lines, differing in relative expression of KLF9, identified a network of genes downstream of this transcription factor. The nature of the participants in this network implicates KLF9 in control of endometrial cell adhesion to substratum, tumor cell migration and invasion, cell stress responses, embryo attachment to endometrium, and uterine remodeling. Additional studies of the complex, multi-factorial role of KLF9 in the uterus are therefore warranted.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Krüppel-like factor 9 (KLF9) is a transcriptional regulator of uterine endometrial cell proliferation, adhesion and differentiation; processes essential for pregnancy success and which are subverted during tumorigenesis. The network of endometrial genes controlled by KLF9 is largely unknown. Over-expression of KLF9 in the human endometrial cancer cell line HEC-1-A alters cell morphology, proliferative indices, and differentiation, when compared to KLF9 under-expressing HEC-1-A cells. This cell line provides a unique model for identifying KLF9 downstream gene targets and signaling pathways.</p>", "<title>Methods</title>", "<p>HEC-1-A sub-lines differing in relative levels of KLF9 were subjected to microarray analysis to identify differentially-regulated RNAs.</p>", "<title>Results</title>", "<p>KLF9 under-expression induced twenty four genes. The KLF9-suppressed mRNAs encode protein participants in: aldehyde metabolism (AKR7A2, ALDH1A1); regulation of the actin cytoskeleton and cell motility (e.g., ANK3, ITGB8); cellular detoxification (SULT1A1, ABCC4); cellular signaling (e.g., ACBD3, FZD5, RAB25, CALB1); and transcriptional regulation (PAX2, STAT1). Sixty mRNAs were more abundant in KLF9 over-expressing sub-lines. The KLF9-induced mRNAs encode proteins which participate in: regulation and function of the actin cytoskeleton (COTL1, FSCN1, FXYD5, MYO10); cell adhesion, extracellular matrix and basement membrane formation (e.g., AMIGO2, COL4A1, COL4A2, LAMC2, NID2); transport (CLIC4); cellular signaling (e.g., BCAR3, MAPKAPK3); transcriptional regulation [e.g., KLF4, NR3C1 (glucocorticoid receptor), RXRα], growth factor/cytokine actions (SLPI, BDNF); and membrane-associated proteins and receptors (e.g., CXCR4, PTCH1). In addition, the abundance of mRNAs that encode hypothetical proteins (KLF9-inhibited: C12orf29 and C1orf186; KLF9-induced: C10orf38 and C9orf167) were altered by KLF9 expression. Human endometrial tumors of high tumor grade had decreased KLF9 mRNA abundance.</p>", "<title>Conclusion</title>", "<p>KLF9 influences the expression of uterine epithelial genes through mechanisms likely involving its transcriptional activator and repressor functions and which may underlie altered tumor biology with aberrant KLF9 expression.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>FAS and RCMS conceived and designed the study. YS performed bioinformatics, gene annotation, and qRT-PCR data analyses. RX conducted the microarray experiment. ZZ performed the human endometrial tumor gene expression study. FAS prepared the initial draft of the manuscript. All co-authors provided inputs during final manuscript preparation.</p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>We thank Renea Eason for assistance with the qRT-PCR studies. This research was supported by NIH Grant RO1 HD21961 (RCMS) and by funds from the Arkansas Children's Hospital Research Institute and the Arkansas Biosciences Institute (FAS, RCMS).</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Differentially expressed transcripts of HEC-1-A sub-lines</bold>. Venn diagrams summarize the number of differentially expressed genes noted between 2AS and 4S sub-lines, between 3AS and 9S sub-lines, and those in common for both comparisons (final annotated gene lists are presented in additional file ##SUPPL##0##1##: Table 1 and additional file ##SUPPL##1##2##: Table 2). There were greater numbers of genes induced than repressed in concert with relative KLF9 expression.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Hierarchical clustering of differentially expressed RNAs</bold>. The microarray data for mRNAs that were identified to be differentially expressed between S and AS sub-lines (additional file ##SUPPL##0##1##: Table 1 and additional file ##SUPPL##1##2##: Table 2) were subjected to hierarchical clustering. The transcript profiles were very similar for both S sub-lines (each run in duplicate); whereas the two AS sub-lines differed from each other and from the S sub-lines. Lower case letters signify duplicate microarrays for each sub-line.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Quantitative RT-PCR of KLF9 mRNA in human endometrium and endometrial tumors</bold>. A normalized cDNA panel of human endometrial tumors was obtained from OriGene Technologies, Inc. This panel was comprised of cDNAs from n = 6 of normal (N) endometria, n = 9 of Stage I tumors, n = 8 of Stage II tumors, n = 19 of Stage III tumors, and n = 6 of Stage IV tumors. Shown are box plots (median, upper and lower quartiles, minimum and maximum data values) of relative abundance of KLF9 mRNA for tumors (delineated by stage and tumor type: endometrioid, serous). Sample numbers are indicated. ANOVA indicated no differences in mRNA abundance between any of the individual stages. However, a significant difference between combined stages (normal plus stage I vs. stages II, III, and IV) was noted by the Mann-Whitney Rank Sum test.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Primers used in quantitative real-time RT-PCR</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"center\">Gene Name</td><td align=\"center\">Upstream Primer (5'-3')</td><td align=\"center\">Downstream Primer (5'-3')</td><td align=\"center\">PCR product size (bp)</td></tr></thead><tbody><tr><td align=\"left\">18S</td><td align=\"left\">TCTTAGCTGAGTGTCCCGCG</td><td align=\"left\">ATCATGGCCTCAGTTCCG A</td><td align=\"center\">150</td></tr><tr><td align=\"left\">AKR7A2</td><td align=\"left\">AACTGGACACGGCCTTCATG</td><td align=\"left\">CCTTGGTGGCAATTTTCACTCTG</td><td align=\"center\">109</td></tr><tr><td align=\"left\">ALDH1A1</td><td align=\"left\">ACCCCAGGAGTCACTCAAGG</td><td align=\"left\">ACTGTGGGCTGGACAAAGTAG</td><td align=\"center\">149</td></tr><tr><td align=\"left\">BCAR3</td><td align=\"left\">CCTGGAAATGCCACAGATCAC</td><td align=\"left\">CTTCATGCAGGAGTTTGCTGAA</td><td align=\"center\">124</td></tr><tr><td align=\"left\">C1orf186</td><td align=\"left\">TAGCTTGGATAGCTCCTGCAGTTC</td><td align=\"left\">CATTTTTTAGTTCTCCAGGGTCAGA</td><td align=\"center\">101</td></tr><tr><td align=\"left\">CLIC4</td><td align=\"left\">TCACCAAAACACCCAGAATCAA</td><td align=\"left\">ACCCCTCTCCAGTGCTTCATTA</td><td align=\"center\">108</td></tr><tr><td align=\"left\">COL4A1</td><td align=\"left\">CACGGGTACTCTTTGCTCTACGT</td><td align=\"left\">AAGGGCATTGTGCTGAACTTG</td><td align=\"center\">101</td></tr><tr><td align=\"left\">COL4A2</td><td align=\"left\">CATGCCCTTCCTGTACTGCAA</td><td align=\"left\">GATGTACTTGATCTCGTCCT</td><td align=\"center\">133</td></tr><tr><td align=\"left\">CXCR4</td><td align=\"left\">CATCAGTCTGGACCGCTACC</td><td align=\"left\">GCAAAGATGAAGTCGGGAATAGTC</td><td align=\"center\">138</td></tr><tr><td align=\"left\">ER-α</td><td align=\"left\">CGGCATTCTACAGGCCAAATT</td><td align=\"left\">AGCGAGTCTCCTTGGCAGATT</td><td/></tr><tr><td align=\"left\">FZD5</td><td align=\"left\">GCTACCAGCCGTCCTTCAGT</td><td align=\"left\">GAAGCGTTCCATGTCGATGAG</td><td align=\"center\">128</td></tr><tr><td align=\"left\">KLF4</td><td align=\"left\">CTGCGGCAAAACCTACACAAA</td><td align=\"left\">GAATTTCCATCCACAGCCGT</td><td align=\"center\">106</td></tr><tr><td align=\"left\">KLF9</td><td align=\"left\">TGGCTGTGGGAAAGTCTATGG</td><td align=\"left\">CTCGTCTGAGCGGGAGAACT</td><td/></tr><tr><td align=\"left\">LAMC2</td><td align=\"left\">GATGGCATTCACTGCGAGAAG</td><td align=\"left\">TCGAGCACTAAGAGAACCTTTGG</td><td align=\"center\">105</td></tr><tr><td align=\"left\">LAPTM5</td><td align=\"left\">CATCTTTTCCATCGCCTTCATCAC</td><td align=\"left\">TCCACCGAGTTCATGCACTTG</td><td align=\"center\">102</td></tr><tr><td align=\"left\">MAPKAPK3</td><td align=\"left\">TCCCACCCTTCTACTCCAACA</td><td align=\"left\">TTCAACAGGAGGCGGATCA</td><td align=\"center\">141</td></tr><tr><td align=\"left\">NR3C1</td><td align=\"left\">AGAGGAGGAGCTACTGTGAAGG</td><td align=\"left\">ACTGAGCCTTTTGGAAAATCAACC</td><td align=\"center\">109</td></tr><tr><td align=\"left\">PAX2</td><td align=\"left\">CCCAGAGTGGTGTGGACAGTTT</td><td align=\"left\">GTAGGAAGGACGCTCAAAGACC</td><td align=\"center\">101</td></tr><tr><td align=\"left\">PSAT1</td><td align=\"left\">ACGCCTCCATGTTTCAGCAT</td><td align=\"left\">TGAGATTTGATGGAGCTAAGCTTCT</td><td align=\"center\">104</td></tr><tr><td align=\"left\">PTCH1</td><td align=\"left\">GTCGAGCTGTTCGGCATGAT</td><td align=\"left\">AGCAACGTGAACGGTGAACTC</td><td align=\"center\">111</td></tr><tr><td align=\"left\">RAB25</td><td align=\"left\">GGAGCTCTATGACCATGCTGAA</td><td align=\"left\">CCAGGAAGAGCAGTCCATTGTT</td><td align=\"center\">125</td></tr><tr><td align=\"left\">RXRA</td><td align=\"left\">AGGACTGCCTGATTGACAAGC</td><td align=\"left\">GACTCCACCTCATTCTCGTTCC</td><td align=\"center\">141</td></tr><tr><td align=\"left\">SLPI</td><td align=\"left\">GCTGTGGAAGGCTCTGGAAA</td><td align=\"left\">TGCCCATGCAACACTTCAAG</td><td align=\"center\">298</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>qRT-PCR confirmation of differential expression of selected transcripts^a</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td/><td align=\"center\" colspan=\"2\"><bold>4S <italic>vs. </italic>2AS (fold)</bold></td><td align=\"center\" colspan=\"2\"><bold>9S <italic>vs. </italic>3AS (fold)</bold></td></tr></thead><tbody><tr><td align=\"center\"><bold>Gene Symbol</bold></td><td align=\"center\"><bold>qRT-PCR</bold></td><td align=\"center\"><bold>Microarray</bold></td><td align=\"center\"><bold>qRT-PCR</bold></td><td align=\"center\"><bold>Microarray</bold></td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td/><td align=\"center\" colspan=\"4\"><bold>Up-regulated</bold></td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td align=\"left\">BCAR3</td><td align=\"left\">3.907</td><td align=\"left\">2.488</td><td align=\"left\">1.171</td><td align=\"left\">2.611</td></tr><tr><td align=\"left\">CLIC4</td><td align=\"left\">31.636</td><td align=\"left\">21.46</td><td align=\"left\">24.750</td><td align=\"left\">11.710</td></tr><tr><td align=\"left\">COL4A1</td><td align=\"left\">3.387</td><td align=\"left\">3.584</td><td align=\"left\">1.768</td><td align=\"left\">2.179</td></tr><tr><td align=\"left\">COL4A2</td><td align=\"left\">2.967</td><td align=\"left\">7.692</td><td align=\"left\">1.834</td><td align=\"left\">2.725</td></tr><tr><td align=\"left\">CXCR4</td><td align=\"left\">5.000</td><td align=\"left\">2.762</td><td align=\"left\">13.037</td><td align=\"left\">3.876</td></tr><tr><td align=\"left\">KLF4</td><td align=\"left\">4.532</td><td align=\"left\">2.110</td><td align=\"left\">1.816</td><td align=\"left\">3.185</td></tr><tr><td align=\"left\">LAMC2</td><td align=\"left\">1.633</td><td align=\"left\">3.003</td><td align=\"left\">2.276</td><td align=\"left\">3.676</td></tr><tr><td align=\"left\">LAPTM5</td><td align=\"left\">22.135</td><td align=\"left\">24.331</td><td align=\"left\">37.060</td><td align=\"left\">5.917</td></tr><tr><td align=\"left\">MAPKAPK3</td><td align=\"left\">3.665</td><td align=\"left\">3.861</td><td align=\"left\">7.056</td><td align=\"left\">3.831</td></tr><tr><td align=\"left\">NR3C1</td><td align=\"left\">7.111</td><td align=\"left\">2.841</td><td align=\"left\">12.500</td><td align=\"left\">4.386</td></tr><tr><td align=\"left\">PSAT1</td><td align=\"left\">1.394</td><td align=\"left\">2.278</td><td align=\"left\">2.876</td><td align=\"left\">6.897</td></tr><tr><td align=\"left\">PTCH1</td><td align=\"left\">3.572</td><td align=\"left\">4.310</td><td align=\"left\">3.008</td><td align=\"left\">2.242</td></tr><tr><td align=\"left\">RXRα</td><td align=\"left\">10.953</td><td align=\"left\">5.405</td><td align=\"left\">5.750</td><td align=\"left\">2.445</td></tr><tr><td align=\"left\">SLPI</td><td align=\"left\">26.636</td><td align=\"left\">9.524</td><td align=\"left\">4.788</td><td align=\"left\">11.779</td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td/><td align=\"center\" colspan=\"4\"><bold>Down-regulated</bold></td></tr><tr><td colspan=\"5\"><hr/></td></tr><tr><td align=\"left\">ALDHIAI</td><td align=\"left\">0.222</td><td align=\"left\">0.174</td><td align=\"left\">0.109</td><td align=\"left\">0.068</td></tr><tr><td align=\"left\">AKR7A2</td><td align=\"left\">0.647</td><td align=\"left\">0.491</td><td align=\"left\">0.264</td><td align=\"left\">0.387</td></tr><tr><td align=\"left\">C1orf186</td><td align=\"left\">0.366</td><td align=\"left\">0.155</td><td align=\"left\">0.278</td><td align=\"left\">0.265</td></tr><tr><td align=\"left\">FZD5</td><td align=\"left\">0.365</td><td align=\"left\">0.440</td><td align=\"left\">0.552</td><td align=\"left\">0.380</td></tr><tr><td align=\"left\">PAX2</td><td align=\"left\">0.145</td><td align=\"left\">0.247</td><td align=\"left\">0.041</td><td align=\"left\">0.382</td></tr><tr><td align=\"left\">RAB25</td><td align=\"left\">0.012</td><td align=\"left\">0.076</td><td align=\"left\">0.002</td><td align=\"left\">0.082</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional file 1</title><p>Table ##TAB##0##1##: RNAs that are more highly expressed in KLF9 AS HEC-1-A sub-lines.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S2\"><caption><title>Additional file 2</title><p>Table ##TAB##1##2##: RNAs that are more highly expressed in KLF9 S HEC-1-A sub-lines.</p></caption></supplementary-material>" ]

[ "<table-wrap-foot><p>^aAll fold change values were statistically significant (<italic>P </italic>&lt; 0.05); n = 2 replicates per cell line.</p></table-wrap-foot>" ]

[ "<graphic xlink:href=\"1477-7827-6-41-1\"/>", "<graphic xlink:href=\"1477-7827-6-41-2\"/>", "<graphic xlink:href=\"1477-7827-6-41-3\"/>" ]

[ "<media xlink:href=\"1477-7827-6-41-S1.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1477-7827-6-41-S2.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Obesity is perhaps the most significant public health problem facing the United States and the Western world today. Each year, an estimated 300,000 Americans die from obesity-related illnesses [##UREF##0##1##]. The latest National Health and Nutrition Examination data show that the prevalence of obesity with body mass index (BMI) ≥ 30 kg/m²has increased from 22.9% in 1994 to 30.5% in 2000. The prevalence of morbid obesity (BMI ≥ 40 kg/m²) also significantly increased, from 2.9% to 4.7% [##REF##12365955##2##]. This increase has affected most surgical practices, as surgeons are operating on obese patients in increasing numbers [##REF##9404886##3##,##REF##11280562##4##].</p>", "<p>Perioperative morbidity, mortality, and prolonged hospital stays are particularly common in obese patients, because these patients often have preexisting cardiac and respiratory disease [##REF##9404886##3##,##REF##10511607##5##]. Moreover, epidemiologic studies have shown that obesity and diabetes are frequently associated with nonalcoholic fatty liver disease, which includes a spectrum of liver disorders that may progress to hepatocellular carcinoma (HCC) [##REF##12105866##6##,##REF##12076877##7##]. Although several studies have analyzed the impact of obesity on patients after major surgical procedures, including liver transplantation [##REF##11280562##4##,##REF##8910182##8##,##REF##17175347##9##], there are, to our knowledge, no data on the outcome of major liver resection for HCC in morbidly obese patients.</p>", "<p>In this report, we discuss the treatment of a large HCC in a morbidly obese patient with a BMI greater than 50 kg/m². We also discuss the current literature on surgical complications in obese patients, and we make some general recommendations about treating HCC in such patients.</p>" ]

[]

[]

[ "<title>Discussion</title>", "<p>Several studies have found that obesity increases the risk of complications and length of hospital stay and is independently associated with increased mortality after elective abdominal surgery [##REF##16957821##10##, ####REF##17103267##11##, ##REF##17481540##12##, ##REF##11997837##13####11997837##13##]. In contrast, a prospective study of 6336 patients who underwent elective noncardiac surgery at a university hospital found that obesity alone was not a risk factor for postoperative complications [##REF##12814714##14##,##REF##15027438##15##]. However, these findings were probably due to the unusually low prevalence of major comorbidities in the obese patients in these studies.</p>", "<p>In a large study of 18,172 adult patients, including 3877 obese patients, who underwent LT in the US between 1988 and 1996, the rates of primary graft nonfunction and of 1- and 2-year mortality were significantly higher in the morbidly obese patients than in the other patients. The authors of that study recommended that morbid obesity (BMI &gt; 35 kg/m²) be considered a relative contraindication for LT [##REF##11786965##16##].</p>", "<p>With regard to our morbidly obese patient (BMI, 56 kg/m²) with a large HCC, during the initial surgical evaluation, she was considered a high-risk candidate for extended right hepatectomy because of her markedly abnormal pulmonary function test results and the insufficient volume of the left lateral segment of her liver. We believe that the neo-adjuvant treatment protocol we implemented prevented tumor progression during the aggressive weight-reduction program that the patient was instructed to follow. This program was instituted because pulmonary function test results and respiratory drive parameters have been found to improve markedly after weight loss [##REF##16354856##17##].</p>", "<p>The locoregional therapy protocol we implemented was chosen on the basis of evidence that combination therapy achieves a higher response rate than repeated TACE alone in large HCCs [##REF##11923602##18##,##REF##11012432##19##]. Yttrium-90 microsphere injection is a novel form of transarterial radiotherapy that has been used increasingly for HCC as a single agent, and it has produced a good response rate [##REF##7947110##20##,##REF##16371529##21##]. To our knowledge, no study has evaluated the use of radioembolization in conjunction with other treatment modalities for any type of malignant disease. However, evidence suggests that doxorubicin hinders the repair of radiation-induced DNA damage in HCC; thus, these treatments may have a synergistic therapeutic effect [##REF##1967294##22##].</p>", "<p>As we anticipated, the tumor was found to be receiving its blood supply from both branches of the hepatic artery. To prevent ischemic injury to segments II and III of the left lobe, we avoided injecting the embolization particles through the left hepatic artery that supplied the lateral aspect of the tumor. This might explain the tumor's failure to respond despite repeated treatments. On the other hand, selective injection into the middle and right hepatic arteries might have spared segments I, II, and III the adverse effects of chemoradiation treatment that were seen in non-tumorous segments of the right lobe.</p>", "<p>Preoperative portal vein embolization is becoming a standard technique for inducing compensatory hypertrophy of the remaining liver and improving the safety and rate of resectability in patients with small-for-size remnant livers [##REF##12560779##23##,##REF##17189109##24##]. Furthermore, sequential preoperative arterial and portal venous embolization can induce tumor necrosis and hypertrophy of the normal liver, which allow safe resection and longer recurrence-free survival [##REF##15249411##25##,##REF##16779884##26##].</p>", "<p>We would have continued the locoregional therapy had there been evidence of tumor response. On the other hand, if the tumor had progressed, we would have added systemic therapy, such as administering the multikinase inhibitor sorafenib, to the treatment protocol. The decision to proceed with surgical resection was based on the tumor's lack of response and, more importantly, on the improved pulmonary function and reduced metabolic syndrome that resulted from the successful weight-reduction program the patient followed during locoregional treatment.</p>" ]

[ "<title>Conclusion</title>", "<p>To reduce the risks that major liver resection poses in morbidly obese patients with significant comorbidity, we suggest implementing a dietary weight-reduction and exercise program to improve the performance status of these patients before resection. While this program is underway, regional therapy can be implemented to prevent the tumor from progressing to the point of inoperability. Portal vein embolization may be required before resection to increase the volume of the remnant liver and to reduce the risk of liver failure and other postoperative complications. We believe that further studies that include large numbers of patients are needed to determine the upper limit of BMI for performing extensive liver resection safely in morbidly obese patients.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Morbid obesity strongly predicts morbidity and mortality in surgical patients. However, obesity's impact on outcome after major liver resection is unknown.</p>", "<title>Case presentation</title>", "<p>We describe the management of a large hepatocellular carcinoma in a morbidly obese patient (body mass index &gt;50 kg/m²). Additionally, we propose a strategy for reducing postoperative complications and improving outcome after major liver resection.</p>", "<title>Conclusion</title>", "<p>To our knowledge, this is the first report of major liver resection in a morbidly obese patient with hepatocellular carcinoma. The approach we used could make this operation nearly as safe in obese patients as it is in their normal-weight counterparts.</p>" ]

[ "<title>Case presentation</title>", "<p>A 41-year-old woman presented with a 2-month history of pruritus. Her medical history included morbid obesity (BMI, 56 kg/m²), hypertension, and type II diabetes. Her initial liver function tests showed moderately elevated total bilirubin and alkaline phosphatase levels and a normal alpha-fetoprotein (AFP) level (Table ##TAB##0##1##). A computed tomography scan (CT-scan) revealed a large (14-cm), hypervascular mass that involved segment IV of the left lobe and segments V and VIII of the right lobe of the liver, partially occluding the proximal part of the common bile duct and causing moderate dilatation of the intrahepatic biliary system (Figure ##FIG##0##1##). Percutaneous biopsy of the tumor confirmed well-differentiated HCC. In addition, biopsy of segment II of the left lobe revealed mild hepatitis with no evidence of steatosis. Volumetric measurement showed that segments I, II, and III accounted for less than 20% of the total liver volume and less than 0.45% of the patient's total body weight.</p>", "<p>Surgical resection was initially ruled out because of a small-for-size remnant liver and abnormal pulmonary function tests that suggested a combination of restrictive and peripheral airway diseases (Table ##TAB##1##2##). After discussing with the patient the risk of complications and potential liver failure associated with extensive liver resection, we elected to pursue locoregional therapy consisting of hepatic transarterial chemo/radioembolization with doxorubicin and yttrium-90 (Y-90) microspheres (Sirtex Medical Limited, Lake Forest, IL, USA). The patient was also placed on a weight-reduction program based on a hypocaloric Mediterranean diet, which has been proven effective for weight loss. Protein intake was calculated as 1 g/kg of body weight. The patient was also instructed to enroll in an aerobic and resistance exercise program in an attempt to improve her metabolic syndrome.</p>", "<p>The treatment protocol consisted of 6 weekly injections of doxorubicin mixed with ethiodized oil, followed by 500- to 700-micron Embospheres (Biosphere Medical Inc, Rockland, MA, USA) alternated with Y-90 microspheres injected selectively into the right and middle hepatic arteries by interventional radiologists. The patient underwent 5 cycles of treatment; side effects were minimal and were related to postembolization effects. The total cumulative doses of doxorubicin and Y-90 were 200 mg and 40.4 mCi, respectively.</p>", "<p>After 7 months of treatment, a follow-up CT scan of the abdomen showed no significant change in the size and enhancement pattern of the tumor. However, the patient's weight had decreased from 159 kg to 136 kg (so that BMI decreased from 56 to 48 kg/m²). This change was accompanied by improvements in most pulmonary function parameters (Table ##TAB##1##2##) and reductions in the dosage of the patient's antihypertensive and antidiabetic medications. At that time, the decision was made to proceed with extended right hepatectomy to remove segments IV, V, VI, VII, and VIII after right portal vein embolization (PVE) to allow compensatory hypertrophy of segments II and III. A volumetric study performed 8 weeks after PVE showed that the caudate lobe and segments II and III accounted for 33% of the total liver volume.</p>", "<title>Surgical technique</title>", "<p>The patient underwent an extended right hepatectomy. She was positioned on a bariatric operating table (Maquet surgical table; Getinge AB, Getinge, Sweden). Exploratory laparotomy was performed through bilateral subcostal incisions with upper midline extensions. A bariatric Thompson self-retaining retractor (Thompson Surgical Instruments, Inc., Traverse City, MI, USA) was used to elevate the costal margins and facilitate exposure. Despite extensive locoregional therapy, there was minimal inflammatory reaction and adhesions between the liver and adjacent organs. Intraoperative ultrasound was used to confirm the previously defined anatomic relation of the tumor with the intrahepatic vasculature. Hilar dissection and mobilization of the right lobe of the liver were carried out in standard fashion for extended right hepatectomy. Parenchymal transaction was performed with a dissecting sealer (TissueLink Medical, Inc., Dover, NH, USA). The total operative time was 630 min. Estimated blood loss was 720 mL. No transfusion of blood products was required.</p>", "<p>The patient's postoperative course was uneventful, despite the long operative time and the technical difficulties encountered during mobilization of the liver because of the compensatory hypertrophy of the left lateral segment and the tumor's large size. The patient remained in the intensive care unit for 2 days and was discharged from the hospital on postoperative day 6. However, superficial wound dehiscence developed that involved the skin and the subcutaneous tissue. This was treated with vacuum-assisted closure (with the VAC Therapy system; KCI, Inc, San Antonio, TX, USA), which facilitated wound healing by secondary intention in 8 weeks.</p>", "<p>Histopathologic examination of the excised tumor and portion of the normal liver revealed a well-differentiated 11-cm HCC. There were focal areas of necrosis and hemorrhage from previous chemoradiation therapy, but there was no evidence of microvascular invasion. In the normal liver parenchyma, there was evidence of postembolization effects, mainly focal areas of foreign body giant cell reaction, but minimal fibrosis and no steatosis. All lymph nodes were negative for malignancy. Currently, the patient is doing well, with no evidence of recurrence 17 months after tumor resection.</p>", "<title>List of abbreviations</title>", "<p>AFP: Alpha-Fetoprotein; BMI: Body Mass Index; CT: Computed Tomography; HCC: Hepatocellular Carcinoma; LT: Liver Transplantation; PVE: Portal Vein Embolization.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>OB: Performed the operation, devised the therapeutic plan, and wrote the manuscript. MS: Performed the TACE; helped in drafting the manuscript. BT: Performed the TACE and Y-90 Sir-Sphere treatment, and helped in drafting the manuscript. JF: Performed the portal vein embolization and TACE, and helped in drafting the manuscript. CFO: Helped in drafting the manuscript. RPW: Co-surgeon during the operation; helped in designing the therapeutic plan, and proofread the manuscript.</p>", "<title>Consent</title>", "<p>Written informed consent was obtained from the patient for publication of this case report and any accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal.</p>" ]

[ "<title>Acknowledgements</title>", "<p>Stephen N. Palmer, PhD, ELS, contributed to the editing of this manuscript. Dr. Palmer is an employee of the Texas Heart Institute at St. Luke's Episcopal Hospital.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>A triple-phase helical CT scan shows a 14-cm hypervascular mass involving the right lobe and the medial segment of the left lobe of the liver.</p></caption></fig>" ]

[ "<table-wrap position=\"float\" id=\"T1\"><label>Table 1</label><caption><p>Preoperative liver function tests and alpha-fetoprotein (AFP) level</p></caption><table frame=\"hsides\" rules=\"groups\"><tbody><tr><td align=\"left\">Total bilirubin (mg/dL)</td><td align=\"center\">1.8</td></tr><tr><td align=\"left\">Alkaline phosphatase (IU/L)</td><td align=\"center\">280</td></tr><tr><td align=\"left\">Alanine aminotransferase, ALT (IU/L)</td><td align=\"center\">80</td></tr><tr><td align=\"left\">Aspartate aminotransferase, AST (IU/L)</td><td align=\"center\">81</td></tr><tr><td align=\"left\">Albumin (g/dL)</td><td align=\"center\">3.8</td></tr><tr><td align=\"left\">AFP (ng/mL)</td><td align=\"center\">3</td></tr></tbody></table></table-wrap>", "<table-wrap position=\"float\" id=\"T2\"><label>Table 2</label><caption><p>Pulmonary function test results at initial evaluation and immediately before and after operation</p></caption><table frame=\"hsides\" rules=\"groups\"><thead><tr><td align=\"left\">Parameter</td><td align=\"center\">Initial value</td><td align=\"center\">Postoperative value</td></tr></thead><tbody><tr><td align=\"left\">Forced vital capacity (L)</td><td align=\"center\">3.38</td><td align=\"center\">3.73</td></tr><tr><td align=\"left\">Forced expiratory volume in 1 second (L)</td><td align=\"center\">1.96</td><td align=\"center\">2.93</td></tr><tr><td align=\"left\">Maximum voluntary ventilation (L/min)</td><td align=\"center\">42</td><td align=\"center\">75</td></tr><tr><td align=\"left\">Vital capacity (L)</td><td align=\"center\">2.6</td><td align=\"center\">3.7</td></tr><tr><td align=\"left\">Total lung capacity (L)</td><td align=\"center\">4.1</td><td align=\"center\">5.9</td></tr><tr><td align=\"left\">Functional residual capacity (L)</td><td align=\"center\">1.6</td><td align=\"center\">2.3</td></tr><tr><td align=\"left\">Expiratory reserve volume (L)</td><td align=\"center\">0.02</td><td align=\"center\">0.06</td></tr></tbody></table></table-wrap>" ]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1477-7819-6-100-1\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Although carcinoid tumors can be found throughout the body, 90% occur within the gastrointestinal tract [##REF##9737302##1##]. They preferentially metastasize to the liver and occasionally (&lt; 10%) cause the carcinoid syndrome by secretion of serotonin and its precursors, as well as other vasoactive substances [##UREF##0##2##]. Primary carcinoid tumors of the liver are exceedingly rare, with only about 60 cases reported in the current literature. Meticulous follow-up is necessary to rule out an occult extrahepatic malignancy with hepatic metastasis to confirm the primary nature of hepatic carcinoids.</p>" ]

[]

[]

[ "<title>Discussion</title>", "<p>A total of sixty cases of primary hepatic carcinoid have been reported, with the largest series being eight patients [##REF##14745329##3##], with long-term follow-up ranging from two to eleven years. Of the reported cases, there is a wide range of age at presentation and there does not seem to be gender predominance. There is no apparent association with cirrhosis or preexisting liver disease.</p>", "<p>Primary hepatic carcinoid tumors may be an incidental finding or can present with severe symptoms including abdominal pain, jaundice, palpable right upper quadrant mass, carcinoid syndrome [##REF##8835904##4##], carcinoid heart disease [##REF##16258211##5##], and Cushing's Syndrome [##REF##17556871##6##]. Less than 10% of gastrointestinal carcinoids present with the carcinoid syndrome and when the syndrome is present it is almost always associated with hepatic metastasis. Interestingly, the syndrome is rarely present in primary hepatic carcinoid tumors, with only two reported cases [##REF##8835904##4##,##REF##16258211##5##].</p>", "<p>Imaging studies of any hepatic mass should begin with ultrasound and a triple-phase CT scan. One report supports the use of contrast-enhanced ultrasound, although there is limited experience with that modality [##REF##16134160##7##]. MRI is increasingly being used, with improved visualization of carcinoid tumors on T2-weighted images [##REF##9862161##8##]. Additional information can be gained from nuclear medicine imaging scans, specifically utilizing Technetium-99m isotopes, as was done with our patient [##REF##15965342##9##]. Finally, if carcinoid is diagnosed postoperatively on histopathology, workup for a primary gastrointestinal site should continue with upper and lower gastrointestinal endoscopy, if these were not performed preoperatively.</p>", "<p>The differentiation between primary and secondary NETs of the liver is not possible by histology alone, although a centrally located solitary tumor may suggest a primary [##REF##14745329##3##]. Additionally, some epithelial tumors (e.g. well-differentiated hepatocellular carcinoma, adenocarcinomas and other neoplasms) may exhibit a NET-like morphology. In such cases, immunohistochemical staining for neuroendocrine markers (e.g. chromogranin, synaptophysin, CD56) should be performed to establish the cell of origin. However, it should be noted that most laboratories, including our own, use chromogranin <italic>A </italic>monoclonal antibody to stain for NETs, therefore NETs expressing chromogranin <italic>B </italic>may be non-reactive with this antibody, as was the case with our specimen.</p>", "<p>All neuroendocrine tumors have malignant potential. As such, some authors recommend using the terms \"low-grade neuroendocrine tumor,\" \"well-differentiated neuroendocrine tumor,\" \"well-differentiated endocrine tumor\" or \"grade I neuroendocrine carcinoma\" instead of \"carcinoid tumor\" to emphasize their biologic behavior. The value of the term \"neuroendocrine tumor\" reflects a particular phenotype that may respond to specific targeted therapies [##UREF##1##10##].</p>", "<p>Despite the classic low-grade cytoarchitectural morphology present in this patient's tumor, its large size (5.2 cm in greatest dimension) and focally infiltrative border are worrisome. As a general principle, NETs smaller than 1.0 cm, in any anatomic location, usually behave in an indolent fashion with only rare recurrences or distant spread while those larger than 2.0 cm are usually more aggressive [##UREF##2##11##]. However, this parameter may not be as important in primary hepatic NETs when it is noted that some of the reported cases have tumors ranging in size from 3.0–16 cm and six of eight patients have remained disease-free after follow-up of more than three years [##REF##14745329##3##]. On the other hand, there are reported cases with sizes ranging from 8.2–26 cm where three of five patients died as early as seven months post-operatively [##REF##10365851##12##]. The large size of the tumors in this particular series was surmised to be the cause of the unfavorable outcome.</p>", "<p>After the appropriate workup of a hepatic mass, initial management is surgical resection when possible. Extent of resection is determined by location and size of the tumor(s), with multicentric bilobar disease often precluding resection. When this is the case, alternative therapies include radiofrequency ablation [##REF##16134160##7##], hepatectomy with transplantation [##REF##14745329##3##], selective hepatic artery embolization [##REF##10522038##13##], regional or systemic chemotherapy, and intravenous octreotide infusion for symptomatic relief. The limited experience with this disease entity makes current recommendations of management difficult. Traditional approaches to hepatic tumors are employed at the discretion of the treating surgeons, gastroenterologists, interventional radiologists, and oncologists.</p>", "<p>The rigorous follow-up and frequent monitoring of patients with hepatic carcinoid also serves as screening for recurrent disease. Recurrences have been reported as early as one year postoperatively and as late as thirteen years, and can occur in the liver or in regional lymph nodes [##REF##16506372##14##, ####REF##12201876##15##, ##REF##15879630##16####15879630##16##]. Distant metastasis without primary hepatic recurrence has not been reported.</p>" ]

[ "<title>Conclusion</title>", "<p>Carcinoid tumors involving the liver are common, but primary hepatic carcinoid tumors are rare. Classification as a primary hepatic tumor requires extensive workup and prolonged follow-up. Regardless of their size, location, and degree of differentiation, NETs have an inherent malignant potential that must be recognized. Management remains surgical resection, with several alternative options available for non-resectable tumors and severe symptoms. The risk of recurrence of primary hepatic carcinoid tumors after resection remains unknown.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Primary carcinoid tumors of the liver are uncommon and rarely symptomatic. The diagnosis of primary hepatic etiology requires rigorous workup and continued surveillance to exclude a missed primary.</p>", "<title>Case Presentation</title>", "<p>We present a case of a 62-year-old female with a primary hepatic carcinoid tumor successfully resected, now with three years of disease-free follow-up. We present a review of the current literature regarding the diagnosis, pathology, management, and natural history of this disease entity.</p>", "<title>Conclusion</title>", "<p>Primary carcinoid tumors of the liver are rare, therefore classifying their nature as primary hepatic in nature requires extensive workup and prolonged follow-up. All neuroendocrine tumors have an inherent malignant potential that must be recognized. Management remains surgical resection, with several alternative options available for non-resectable tumors and severe symptoms. The risk of recurrence of primary hepatic carcinoid tumors after resection remains unknown.</p>" ]

[ "<title>Case presentation</title>", "<p>EG is a 62-year-old female who presented with right upper quadrant abdominal pain, intermittent in timing and dull in nature, not related to oral intake and not associated with nausea or vomiting. Her past medical history included hypertension, irritable bowel syndrome, osteoarthritis, and a history of recurrent bilateral lower extremity deep venous thrombosis on Warfarin. On physical exam there were no abdominal scars, normal bowel sounds on auscultation, minimal right upper quadrant tenderness to palpation, no rebound tenderness or guarding, no hepatomegaly and a negative Murphy's sign. Her laboratory studies were significant for a GGT of 162 U/L (normal 5–80 U/L), with otherwise normal liver function tests. Tumor markers were negative, with an AFP of 3.1 ng/ml.</p>", "<p>Diagnostic imaging included an abdominal ultrasound (Figure ##FIG##0##1##) which revealed a heterogeneous solid mass in the lateral segment of the left hepatic lobe measuring 6.3 × 5.3 × 5.0 cm. A CT scan with intravenous contrast was obtained which revealed a 4.9 × 4.9 cm enhancing, poorly marginated mass in segment II of the liver, with no other intra-abdominal masses or lymphadenopathy (Figure ##FIG##1##2##).</p>", "<p>A CT-guided biopsy was performed which yielded scant tissue with poorly cohesive cells arranged in papillae. PAS-D stain showed focal, small mucin droplets in some cells. Immunohistochemistry was positive for CEA and CK-7 and negative for calretinin, CDX-2, CK-20, Muc-2 and Muc-6. The limited sample was diagnosed as papillary adenocarcinoma, favoring metastasis, on the basis of morphology, special stain results and immunoprofile. However, a second panel of immunohistochemical stains for synaptophysin, CD56 and chromogranin were performed on the biopsy specimen. The tumor cells were negative for chromogranin but expressed synaptophysin and CD56, consistent with the immunoprofile of a neuroendocrine tumor (NET).</p>", "<p>Further workup for a primary tumor or other metastatic sites included a negative CT scan of the chest, upper and lower gastrointestinal endoscopy, and a Technetium-99m bone scan. The decision was made to resect the hepatic tumor.</p>", "<p>An uncomplicated left lateral segmentectomy (II &amp; III) and cholecystectomy were performed. No peritoneal carcinomatosis was noted upon exploration. The postoperative course was uneventful and she was discharged home on the fourth postoperative day.</p>", "<p>Grossly, the tumor measured 5.2 × 5.0 × 5.0 cm and had a tan gray, soft, fish-fleshy cut surface (Figure ##FIG##2##3##). Although there was a focal infiltrative edge, it was well-circumscribed and located 5.9 cm away from the resection margin. The tumor consisted of approximately 40% solid areas and 60% hemorrhagic and cystic degenerative areas. There were no satellite nodules. Surgical margins were negative for malignancy, including the left hepatic artery, vein, duct, gallbladder and portal lymph nodes.</p>", "<p>Microscopically, the tumor consisted predominantly of solid sheets and organoid nests of uniform, intermediate-sized, polyhedral cells (Figure ##FIG##3##4A##) in a vascular stroma. Other areas showed a trabecular arrangement of these cells with focal stromal hyalinization (Figure ##FIG##3##4B##); cystic areas were also present. Cytologically, the tumor cells had a moderate amount of eosinophilic cytoplasm with perinuclear eosinophilic inclusions and round to oval nuclei with vesicular to finely granular chromatin. There were no areas of necrosis, and mitoses were infrequent.</p>", "<p>Immunohistochemistry was consistent with the immunoprofile of the biopsy specimen, i.e. positive staining for synaptophysin (Figure ##FIG##4##5A##) and CD56 (Figure ##FIG##4##5B##) and negative staining for chromogranin. Additionally, there was immunoreactivity for epithelial markers CK-7, CAM 5.2 and pancytokeratin AE1/AE3. There was negative staining for HEPT, CA19.9 and TTF-1, thus ruling out hepatocellular carcinoma, metastatic carcinoma from the gastrointestinal tract and metastatic lung carcinoma, respectively. The histomorphologic features coupled with the immunohistochemical results supported the diagnosis of a carcinoid tumor/low grade NET.</p>", "<p>Follow-up over the subsequent three years included CT scans of the abdomen at six month intervals. To date, no recurrent or metastatic disease has been identified. She remains symptom free and in good health.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>GS drafted the case presentation and literature review sections of this manuscript. AC and HG reviewed the specimens and drafted the review of the pathological findings associated with this disease entity. FA and JO were the primary physicians diagnosing, treating, and currently following the referenced patient.</p>" ]

[ "<title>Acknowledgements</title>", "<p>Written informed consent was obtained from the patient for publication of this case report and the accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Ultrasound of the abdomen; Ultrasound of the abdomen depicting a 6.3 × 5.3 × 5.0 heterogenous solid mass in the lateral segment of the left lobe of the liver</bold>.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>CT scan of the abdomen and pelvis; CT scan of the abdomen and pelvis with IV contrast demonstrates a 4.9 × 4.9 cm enhancing, poorly marginated mass in segment II of the liver.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p>Gross image of the specimen; The specimen was measured at 5.2 × 5.0 × 5.0 cm and had a tan gray, soft, fish-fleshy cut surface.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p>Microscopic image of the specimen; The tumor consisted of solid sheets and organoid nests of uniform, intermediate-sized, polyhedral cells in a vascular stroma (Image A) as well as areas of trabecular arrangement with focal stromal hyalinization (Image B).</p></caption></fig>", "<fig position=\"float\" id=\"F5\"><label>Figure 5</label><caption><p>Immunohistochemistry of the resected specimen; Immunohistochemistry was positive for synaptophysin (Image A) and CD56 (Image B), consistent with a NET.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1477-7819-6-91-1\"/>", "<graphic xlink:href=\"1477-7819-6-91-2\"/>", "<graphic xlink:href=\"1477-7819-6-91-3\"/>", "<graphic xlink:href=\"1477-7819-6-91-4\"/>", "<graphic xlink:href=\"1477-7819-6-91-5\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Cutaneous metastasis arising from colorectal malignancy is a rare occurrence though well reported in literature. It is a pointer to more widespread disease and usually indicates a poor prognosis. Cutaneous metastasis can be the presenting feature before the primary is diagnosed e.g. Sister Mary Joseph's nodules. They can also occur late after the primary has been completely excised and can present either as cutaneous rash or as subcutaneous nodules in proximity to previous operative scars (abdominal or perineal) or on the abdominal wall skin. This is an unusual presentation of a large fungating peristomal metastasis without any visceral involvement following abdomino-perineal excision of a large T4 rectal cancer done 14 years earlier.</p>" ]

[]

[]

[ "<title>Discussion</title>", "<p>Stomal recurrence following abdomino-perineal resection of a rectal cancer would indicate a metachronous tumour arising from the colonic mucosa. This is well reported in world literature with at least 10 cases described [##REF##12051539##1##, ####REF##17567928##2##, ##REF##16263022##3####16263022##3##]. This may present as a stomal stricture, peristomal rash or ulceration, or as a peristomal mass. The time of presentation of the metachronous tumour can be 4 to 30 years following the original resection of the primary [##REF##12051539##1##]. In the present case, initially the clinical and CT appearances seemed to fit in with a metachronous tumour, explaining the reason for doing a completion colectomy and wide local excision of the mass <italic>en bloc</italic>. However, the histopathology of the specimen clearly shows normal colonic mucosa from caecum to the stoma with only some ulceration at the stoma site, precluding a metachronous tumour. This suggests a true isolated metastasis to the peristomal skin, as the pre-operative staging CT scan and the operative findings did not show any evidence of other visceral metastasis.</p>", "<p>The incidence of cutaneous metastases associated with all cancers in both sexes is 0.7 to 9.0% [##REF##7622642##4##]. In Lookingbill's study of metastatic disease, the incidence of cutaneous metastasis in colorectal cancer was 4.4% [##REF##8335743##5##]. Colorectal metastases usually occur within 2 years of the primary tumour resection and the common organs involved are liver, peritoneum, pelvis, lung and bone in decreasing frequency [##REF##15221607##6##]. The clinicopathological risk factors for skin metastasis are primary tumours that extend transmurally through the wall of the colon or rectum, lymph node metastases at presentation and perforated primary tumour [##REF##10826422##7##]. The commonest site of cutaneous metastasis in colorectal cancer is the abdominal wall skin [##REF##15221607##6##]. They can arise in previous surgical scars (abdominal or perineal) including colostomy reversal scar [##REF##17567928##2##]. Rarely this may occur in the skin of the lower limbs, face or back [##REF##11016477##8##,##REF##11452828##9##]. The gross appearance of skin metastasis is usually of small subcutaneous nodules, which are under 5 cms in size, or as superficial cutaneous papules. It can present rarely as an inflammatory rash which can be confused with primary skin conditions [##REF##14605930##10##,##REF##15237593##11##].</p>", "<p>Rarely colorectal metastasis may present as a large cutaneous or subcutaneous mass (&gt; 5 cms). Alexandrescu <italic>et al </italic>described 2 cases of large cutaneous and subcutaneous metastases (11 cms and 5 cms) presenting in the abdominal incision scars occurring 5 and 3 years respectively after the primary tumour resection [##REF##16150230##12##]. At the time of reporting, the patient had survived for 22 and 12 months respectively after wide local excision, without any other metastatic involvement. Tan <italic>et al. </italic>described inflammatory subcutaneous mass &gt; 10 cms diameter appearing 22 months after primary resection, growing in size rapidly on the scapular region within a documented period of 1 month duration [##REF##17006588##13##]. This was successfully excised but no information is available regarding the survival period. Greenberg <italic>et al </italic>reported a case of peristomal erythema, ulceration and induration appearing 6 months after abdomino-perineal resection which on shave biopsy turned out to be metastatic disease and a tiny primary focus was found in the resected stump suggesting a true synchronous tumour (&lt; 1 year of primary resection) with peristomal cutaneous metastasis [##REF##17043208##14##].</p>", "<p>Our case is unique due to the size (&gt; 10 cms), location (proximity to colostomy) and the time interval after resection of the primary (14 years). One of the possible mechanisms of spread of the tumour from the rectum to the colostomy site could be via the lymphatic channels in continuity with the loop colostomy that was constructed prior to the patient undergoing the abdomino-perineal excision. The second possibility is for micro-metastasis left behind in lymph nodes along the inferior mesenteric artery pedicle at the time of the abdomino-perineal resection. No specific mention was made in the initial operative notes of the curative or palliative intent of the operation, and if a high tie was carried out (ligation the inferior mesenteric artery flush with the aorta). This could lead to lymph node metastasis developing at a later date and could break through the skin presenting as a fungating mass. Histopathology did not reveal any evidence of lymphoid tissue either in the resected mass or in the remaining mesocolon of the completion colectomy to support these possible lymphatic routes. The third possible mechanism is by iatrogenic implantation of the colorectal cancer cells at the time of initial surgery. This is a well recognised problem and can present as metastatic recurrence in the midline incisions, perineal wounds, port sites and drain sites following surgery. The incidence of wound recurrence following open resection of colorectal cancer was 0.8% to 1.5% in two large series of total of 3314 patients and 80% of these recurrences occurred within 12 months of the initial surgery [##REF##6223795##15##,##REF##8620788##16##]. The trauma of surgery results in an inflammatory response which has been shown to enhance the successful implantation of exfoliated tumour cells in animal models. This may be a consequence of enhanced tumor cell adhesion or transient generalised immune suppression following surgery [##REF##16231370##17##]. Tumour cell implantation could be the likely cause of metastatic recurrence in our case.</p>", "<p>The survival period following diagnosis of cutaneous colorectal metastasis is 1 to 34 months. Lookingbill <italic>et al </italic>found a mean survival period of 18 months in 18 patients with skin metastases from colorectal carcinoma [##REF##8335743##5##]. In the series by Saeed <italic>et al</italic>. the survival period was 2 to 4.5 months in 5 cases of colonic cancer [##REF##15186430##18##]. There are case reports of patients surviving 4 years and 10.5 years following treatment of isolated metastsasis [##REF##15221607##6##]. The prognosis of cutaneous metastasis would depend on the presence of concomitant metastasis elsewhere at the time of diagnosis and also on the surgical clearance achieved if found to be true isolated metastasis. Tumour differentiation and lymphovascular invasion are also important factors in altering the prognosis.</p>", "<p>Wide local excision of the cutaneous or subcutaneous lesion is the preferred treatment option in isolated lesions. There are no clear guidelines for the optimum chemotherapeutic regimen for a non-resectable recurrence of colorectal cancer. Focal radiotherapy has been tried for cutaneous rash with poor response [##REF##11016477##8##,##REF##15237593##11##]. The chemotherapy treatments described include 5-fluorouracil [##REF##15655588##19##], capecitabine [##REF##15237593##11##], irinotecan [##REF##15221607##6##], oxaliplatin [##REF##15221607##6##,##REF##15237593##11##,##REF##17043208##14##] and cisplatin [##REF##15655588##19##]. The chemotherapy drugs have evolved from a single agent 5-Fluorouracil (5-FU) as the first line agent to the current combination of drugs. The combination of irinotecan to bolus 5-FU has increased median survival from 12 months to 14.8 months [##REF##11006366##20##]. The combination of infusional 5-FU and leucovorin(LV) with oxaliplatin (FOLFOX) [##REF##14665611##21##] or infusional 5-FU/LV with irinotecan (FOLFIRI) [##REF##15939923##22##] has increased this survival figure to above 20 months. Treating patients sequentially with FOLFIRI followed by FOLFOX, or with FOLFOX followed by FOLFIRI, has increased the median survival times to 21.5 months and 20.6 months, respectively [##REF##14657227##23##]. Our patient was offered chemotherapy but she declined it. We intend to follow her up at 6 monthly intervals with CEA level estimations and cross sectional liver imaging.</p>" ]

[ "<title>Conclusion</title>", "<p>Cutaneous metastasis following colorectal cancer resection is a well-recognised entity though rare. Any unusual skin lesions especially on the abdominal wall skin, previous incision scars or near the stoma should be biopsied early to rule out metastatic disease and systematic work-up should be carried out to rule out any metachronous tumour or metastasis elsewhere in the body. For isolated cutaneous or subcutaneous metastasis, wide local excision would be the preferred surgical option followed by adjuvant chemotherapy depending on the histopathological status.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Cutaneous metastasis from colorectal cancer after excision of the primary is a rare occurrence and presents as cutaneous or subcutaneous nodules or as a rash commonly on the anterior abdominal wall.</p>", "<title>Case presentation</title>", "<p>This is a case description of the management of a large fungating peristomal cutaneous metastasis occurring 14 years after abdomino-perineal excision of the primary cancer. The gross appearance initially suggested possibility of a true metachronous cancer with peristomal spread. But histopathology of the resected specimen showed no colonic mucosal involvement suggesting a true large cutaneous peristomal metastasis which has not been reported previously. Literature review of presentation, management and prognosis of cutaneous metastasis from colorectal cancer is described</p>", "<title>Conclusion</title>", "<p>Cutaneous metastasis following colorectal cancer resection is a well-recognised entity though rare. Any unusual skin lesions especially on the abdominal wall skin, previous incision scars or near the stoma should be biopsied early to rule out metastatic disease and systematic work-up should be carried out to rule out any metachronous tumour or metastasis elsewhere in the body.</p>" ]

[ "<title>Case presentation</title>", "<p>A 61 year old lady was referred by the General Practitioner as an emergency with a rapidly enlarging fungating mass around her end-colostomy site. She had undergone an abdomino-perineal resection 14 years earlier for a Duke's B rectal cancer. She had initially presented 14 years ago with features of large bowel obstruction secondary to a large rectal tumour. She had a defunctioning loop colostomy constructed followed by adjuvant radiotherapy to shrink the rectal tumour. At the time of the Abdomino-perineal Resection, the tumour was adherent to the uterus and she underwent hysterectomy with bilateral salpingo-oopherectomy with partial vaginectomy to achieve full surgical clearance. Histopathology of the original specimen revealed a T4 tumour with clear margins and no lymph node involvement. No chemotherapy was given after surgery.</p>", "<p>She was lost to follow-up till she re-presented with this peristomal mass. On clinical examination, she was found to have a large 12 × 12 cms mass, eroding through the skin in two areas, around the stoma, which appeared stenosed but was functioning normally (Figure ##FIG##0##1##). There was no history of loss of weight or appetite and the mass had reportedly grown rapidly over a period of four weeks. Trucut biopsy of the mass gave a diagnosis of adenocarcinoma of colonic origin. Her serum CEA level was elevated at 11.1 μg/L and her Haemoglobin was 9.9 g/dL (MCV – 82.8 fL). A staging CT scan of the chest, abdomen and pelvis demonstrated the 8 × 5 cms mass adjacent to the stoma lying mainly in the subcutaneous fat with no invasion into the muscles (Figure ##FIG##1##2##). There was no evidence of any metastatic disease elsewhere.</p>", "<p>At operation, the mass and the adjacent colostomy were excised wide with generous margins and placed in a bowel bag to avoid any tumour seeding (figure ##FIG##2##3##). A midline Laparotomy was then performed. There was no evidence of any intra-abdominal metastasis and a curative resection of the mass with <italic>en bloc </italic>completion colectomy was performed (Figure ##FIG##3##4##). The large 17 × 14 cms defect was covered with V.A.C^®(KCI) dressing and an end ileostomy was constructed in the right iliac fossa (figure ##FIG##4##5##). She made an uneventful post-operative recovery and after 19 days of V.A.C^®therapy a meshed split skin graft was harvested from the anterior thigh and used to resurface the abdominal wound.</p>", "<p>The histopathology of the specimen showed complete excision of the subcutaneous mass with clear margins (figure ##FIG##5##6##) and the microscopic examination showed extensively necrotic and inflamed, well-differentiated colonic adenocarcinoma invading into the subcutaneous fat. There was no mucosal abnormality in the resected colon and there was no evidence of lymphovascular invasion. She was offered adjuvant chemotherapy but she declined it.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>CV prepared the draft manuscript. SN helped in literature search and preparation of manuscript. MJC conceived the idea and edited the final version for its scientific content. All authors read and approved the final manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>Written informed consent was obtained from the patient for publication of this case report.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>Clinical photograph showing peristomal cutaneous metastasis.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>CT scan of the mass, showing abdominal wall metastasis.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p>Wide local excision of the mass with the stoma.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p>Completed excision wound with end ileostomy.</p></caption></fig>", "<fig position=\"float\" id=\"F5\"><label>Figure 5</label><caption><p>V.A.C^®dressing on the open wound.</p></caption></fig>", "<fig position=\"float\" id=\"F6\"><label>Figure 6</label><caption><p>Specimen after resection.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "<graphic xlink:href=\"1477-7819-6-96-1\"/>", "<graphic xlink:href=\"1477-7819-6-96-2\"/>", "<graphic xlink:href=\"1477-7819-6-96-3\"/>", "<graphic xlink:href=\"1477-7819-6-96-4\"/>", "<graphic xlink:href=\"1477-7819-6-96-5\"/>", "<graphic xlink:href=\"1477-7819-6-96-6\"/>" ]

[]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Recent events have demonstrated that the capability to assess exposure and infection of individuals by biological threat agent (BTA) well in advance of onset of illness or at various stages post-exposure could offer important diagnostic and therapeutic benefits. Direct pathogen identification can be elusive since many pathogens sequester in tissues initially. Direct pathogen methods include the classical culture methods, immunoassay, and gene amplification of the pathogen. Although these methods are being improved for incredibly greater sensitivity [##REF##11777824##1##, ####REF##11572145##2##, ##REF##12843075##3####12843075##3##], the efficiency of any diagnostic approach for direct pathogen assessment depends upon the presence of agent in a small specimen matrix. By the time detectable levels of pathogen are reached, it is frequently too late to halt the progression of the intractable illness [##UREF##0##4##,##REF##11863216##5##]. Host responses that occur rapidly after exposure to specific pathogenic agents could provide the needed information for defense against the biothreat agents at time periods when clinical signs might include general malaise or flu-like symptoms that would not differentiate among diverse pathogenic agents (Additional file ##SUPPL##0##1##).</p>", "<p>Correlation of the course of the infection and the disease progression with molecular responses provides opportunities to understand pathogenesis resulting from biological agent exposure. Symptoms of exposure to biological toxins such as <italic>Vibrio cholerae </italic>toxin (CT), <italic>Clostridium botulinum </italic>toxin A (BoNT-A), and Staphylococcal enterotoxin B (SEB) include violent reactions within hours of exposure and may lead to death in a few days. In contrast, <italic>Brucella </italic>infection <italic>(B. melitensis </italic>16 M) produces late-onset of mild symptoms that persist over a long period of time. Other bacterial pathogens, such as those causing anthrax (<italic>B. anthracis</italic>) or plague (<italic>Yersinia pestis</italic>), induce systemic, flu-like symptoms initially and progress to lethal shock and death days later. In the case of viral threat agents, Venezuelan equine encephalitis (VEE) initially causes serious illness, associated with severe malaise and an extended recovery time. Dengue (DEN-2) patients usually recover fully within days (&lt; 1% mortality), but some individuals, especially those with prior dengue exposure, may develop dengue hemorrhagic fever (DHF). Therefore, the time when an individual is exposed, the incubation period, and the time of manifestation of the illness are crucial to designing diagnostic and therapeutic strategies.</p>", "<p>In the post-genomic period, it is no longer unrealistic to hope that the examination of host responses, by interrogating large numbers of genes, could reveal unique responses to the various BTA. However, there are a number of obstacles yet to be overcome. Various pathogens may affect different tissues or cells that may not be available for diagnostic purposes. For example, botulinum toxins target neuromuscular synapses and cholera toxin aims for intestinal epithelial cells. As a first step toward the goal, we focused on gene expression changes in human peripheral blood mononuclear cells (PBMC) since they could be readily obtained from an exposed individual, thus taking advantage of their \"reconnaissance\" role. We carried out <italic>in vitro</italic> exposure to each of 8 biological threat or infectious agents. We confirmed the <italic>in vitro</italic> results by using peripheral blood mononuclear cells (PBMC) from nonhuman primates (NHP) exposed to a bacterial pathogen (<italic>Bacillus anthracis</italic>) or, separately, to a toxin (SEB) at various time points post-exposure to compare findings <italic>in vitro</italic> to those seen <italic>in vivo</italic>. We have identified host gene expression patterns that can discriminate exposure to various BTA, even at early time periods when flu-like symptoms occur.</p>" ]

[ "<title>Methods</title>", "<p>For the <italic>in vitro</italic> studies, the bacterial pathogens used were <italic>Bacillus anthracis</italic>, <italic>Yersinia pestis</italic>, <italic>Brucella melitensis</italic>; toxins: SEB, CT, BoNT-A; viruses: VEE and DEN-2.</p>", "<p>We isolated RNA from human lymphoid cells after exposure and analyzed the gene expression patterns induced by these agents using cDNA arrays. We have not amplified our message for gene array analysis in order to avoid PCR-mediated bias.</p>", "<title>Bacterial strain and growth conditions</title>", "<title>Y. pestis</title>", "<p><italic>Y. pestis </italic>KIM5 <italic>Pgm </italic>negative mutant bacteria (Laboratory stock) were grown on Brain Heart Infusion (BHI, Invitrogen, Rockville, MD) agar plates 30°C for 48 h. Pre-culture of <italic>Y. pestis </italic>bacteria were grown in BHI broth at 26°C overnight in shaker incubator at 180 rpm and were diluted 1/25 in fresh BHI medium. The organism was then grown at 26°C for 5 h (OD 600 ~0.5) and used for infection. To determine the MOI we counted serial dilutions of the bacterial culture by plating on (BHI) agar plates followed by incubation at 30°C for 48 h.</p>", "<title>B. anthracis</title>", "<p>Spores were prepared from <italic>B. anthracis </italic>Ames strain (pXO1+, pXO2+). Briefly, 5% sheep blood agar (SBA) plates were inoculated with <italic>B. anthracis </italic>Ames spores and incubated overnight at 35°C. Several isolated colonies were transferred to a sterile screw capped tube containing 5 ml of sterile PBS. NSM plates (New sporulation medium: per liter added Tryptone; 3 grams. Yeast extract; 3 grams. Agar; 2 grams. Lab Lemco agar; 23 grams. 1% MnCl₂·4H₂O; 1 ml) (150 × 25 mm Petri plate) was inoculated with 200 μl of prepared cell suspension. These plates were incubated for 48 hrs at 35°C, then checked for sporulation progress by microscopic examination. Continued incubation at room temperature was performed until free refractive spores constituted 90–99% of total suspension. Spores were then harvested from plates using 5 ml of sterile water. Spores were then washed 4 times in sterile water. Spores were checked for purity by plating 10 μl in triplicate onto 5% SBA plates and incubating overnight @ 35°C. Enumerations of spores were calculated via CFU/ml (determination of viable spores) and also for actual spores/ml using Petroff Hauser chamber.</p>", "<title>B. melitensis</title>", "<p><italic>B. melitensis </italic>16 M strain was grown in shaker flasks overnight in <italic>Brucella </italic>broth at 37°C, plated on <italic>Brucella </italic>agar for 48 h at 37°C to obtain confluent growth. Bacteria were scraped from plates in pyrogen-free, sterile 0.9% NaCl for irrigation (saline), washed twice, concentration adjusted in saline to the appropriate OD₆₀₀.</p>", "<title>Venezuelan Equine Encephalitis Virus (VEE)</title>", "<p>The Trinidad (Trd) strain used in these studies is a virulent virus of the epizootic IA/B variety of VEEV and was originally isolated from the brain of a donkey. Virus was diluted to an appropriate concentration in Hank's buffered saline solution containing 1% fetal bovine serum.</p>", "<title>Dengue 2 Virus</title>", "<p>DV-2 was grown and propagated in mycoplasma-free Vero cell lines. The viral titer was determined by limiting dilution plaque assays on Vero cells. All virus stocks and culture supernatants used in the present study were free from LPS and mycoplasma [##REF##17056582##6##].</p>", "<title>Isolation of cells from Human PBMC using elutriation methods</title>", "<p>Leukopheresis units were obtained from volunteer donors using the procedures outlined in our approved human use protocol, reviewed by the established Institutional Review Board at WRAIR. The written informed consent document was provided to the volunteers in advance of the procedure.</p>", "<p>We obtained PBMC (74 blood draws over a period of ~2.5 years, and collected from ~8–10 AM to minimize variability) from healthy human volunteers who had been screened to be HIV and Hepatitis B negative, were from 19–61 years of age and both male and female. Human monocytes and lymphocytes of peripheral blood mononuclear cells were purified from leukopacks of healthy donors by centrifugation over lymphocyte separation medium (Organaon Tecknika, NC). Monocytes and lymphocytes were then further purified by counter flow centrifugation-elutriation with pyrogen-free, Ca²⁺- and Mg²⁺-free phosphate-buffered saline as the eluant. The resulting monocytes and lymphocyte preparations had greater than 95% viability. Monocytes and lymphocytes were mixed in the ratio 1:4 and were used immediately. Cell cultures were maintained in RPMI media at 37°C.</p>", "<title>Exposure of monocytes and lymphocytes to the various pathogenic agents</title>", "<p>Lymphoid cells were then exposed to a pathogenic agent under conditions (dose, exposure time) deemed optimal for biological activity for each agent by the pathogen specialist. The multiplicity of infection (MOI) of the bacteria or viruses to lymphoid cells was as follows: anthrax spores (1 or 3), VEE (1 or 3), DEN-2, (0.2 or 1), <italic>Brucella </italic>(2), and plague (20). Following a 30-min infection period cells were washed once with Hanks Balance Salt Solution (HBSS, Invitrogen, Rockville, MD). Both uninfected and infected cells were maintained in RPMI Medium 1640 with 10% human AB serum at 37°C 5% CO2 for the different time periods post exposure. Cells were harvested and total RNA isolated using Trizol reagent (Invitrogen, Carlsbad, CA).</p>", "<p>The bacterial toxins, CT (3 nM), SEB (100 ng/mL), and BoNT-A (1 nM), were added to newly plated cells in flasks for the time period specified. Cells, incubated in the absence and presence of the toxins, were collected by centrifugation. Trizol™ (Invitrogen, Carlsbad, CA,) was added to the cells for RNA isolation and the cells were frozen at -70°C until use.</p>", "<title>RNA isolation and cDNA arrays</title>", "<p>RNA was isolated according to the Trizol method (Invitrogen, Carlsbad, California) followed with DNAse digestion [##REF##17429414##7##]. The custom cDNA slides contained ~10,000 genes (Additional file ##SUPPL##1##2##). Stratagene reference RNA was labeled with Cy3 and used to compare with RNA (Cy5) from either control or exposed samples. RNA was labeled using Micromax-TSA labeling kits (Perkin Elmer, Boston, MA), hybridized and scanned in an Axon scanner. GenePix 3.0 (Axon) was used to analyze the scanned image. For studies using Human cDNA membranes (Clontech Laboratories, Palo Alto, California), RNA samples were labeled with radioactive ³³P. After washing, the blots were exposed to Kodak screen and scanned in a BIORAD multifluor scanner. Atlas Image software (BD Biosciences Clontech) was used for spot alignment and normalization of the scanned arrays.</p>", "<title>Statistical analyses and data scrutiny</title>", "<p>We have adhered to \"MIAME\" (minimum information about microarray experiments) for all our studies. For each pathogen, 3–6 successive time periods were studied and for each time period, data from 2–4 separate experiments were obtained and, using the data from these multiple experiments, 2-way ANOVA analysis were carried out. GeneSpring version 5.0 (Silicon Genetics, San Carlos, CA) and Partek Pro 5.0 (St. Charles, MO) were used to visualize and analyze the data. Welch's ANOVA (p &lt; 0.05) was performed followed by Benjamini Correction [##UREF##1##8##] for various sets of data, to find genes that varied significantly across samples and to identify patterns of gene regulation in PBMC exposed to various pathogens. For custom microarrays, we used the scatter plot smoother, Lowess algorithm [##REF##11842121##9##], to normalize for dye bias among samples. We filtered the array data at 2 steps. In the first step, data filtration allowed only elements for which intensities in both channels were above twice background intensity. In the second step, elements that had intensities below twice the background intensity in one channel only were set at twice background levels. Last, to identify patterns of gene expression among different pathogens, <italic>k</italic>-means and self-organizing map clustering analyses were performed. Complete linkage hierarchical clustering of an uncentered Pearson correlation similarity matrix was also applied using the Eisen Cluster software [##REF##9843981##10##], and the results were visualized with the program TreeView. We have used the major dataset (data from Figure ##FIG##0##1b##) as a training set to apply a class prediction method (GeneSpring 6.1) that uses the k-nearest neighborhood algorithm to classify blinded samples used as test sets.</p>", "<title>Feature selection, computation and classification</title>", "<p>Extracting discrete data: We applied the Greedy algorithm approach (46).</p>", "<p>For each of the 8 pathogens at each time point evaluated (29) (where a condition is the combination of a pathogen and a time interval), we compute for each gene a regulation type which is a nonempty subset of {U, D, S}. For a given condition and gene, the regulation type contains U (respectively D, S) if for at least one array for that condition, using either sum or median normalization techniques and either difference or ratio criteria, the gene appears to be up regulated (respectively down regulated, stay the same). Thus we model both variation between donors and experimental error. The regulation type {U, D} is treated as if it were {U, D, S}; i.e. we consider it to provide no information.</p>", "<p>Ordering the genes: For each value of n, we would like to select the n genes which best distinguish between the 29 conditions. Since this is an intractable task, we compute an approximation by ordering the genes according to a heuristic, and for each n, choose the first n genes from the list. There are two straightforward greedy approaches to generating such a list. In the grow approach we start with an empty list and at each step add the gene that gives the new list with the best discrimination power. In the shrink approach we generate the list in reverse order by starting with the full set of genes and removing the one that that leaves the remaining set with the best discrimination power.</p>", "<p>We estimate the discrimination power of a set of genes by computing an integer vector of length 812, containing an entry for each of the 29 × 28 ordered pairs of distinct conditions, which itself is a sum computed over all genes in the set. The values summed are the number of elements in the regulation type for the first condition of the pair that do not occur in the regulation type for the second condition. This vector is then sorted least element first. To compare the discrimination power of two gene sets, their vectors are computed and compared lexicographically, with larger vector considered to correspond to the gene set with better discrimination power. The first 50 genes to be ranked by this method are listed in the Additional file ##SUPPL##2##3##.</p>", "<title><italic>In vivo</italic> anthrax exposure</title>", "<p>Animal work was conducted in compliance with the Animal Welfare Act and other federal statutes and regulations relating to animals and experiments involving animals and adheres to principles stated in the Guide for the Care and Use of Laboratory Animals, NRC Publication, 1996 edition. Nine anthrax-naïve rhesus macaque NHPs were exposed to approximately 8 LD₅₀of B. anthracis spores (Ames strain) by a head-only aerosol exposure system. Blood was drawn before exposure to determine baseline values. After exposure, 3 animals were euthanized at each 24, 48 and 72 h and blood taken for various analyses, including gene response patterns. A full necropsy was performed to collect biological samples for use in B. anthracis diagnostic assay development.</p>", "<title>Primer set design</title>", "<p>Primer sets were designed for selected genes for expression profile confirmation. Additional file ##SUPPL##3##4## lists the accession numbers and sequences for both sense and antisense primers.</p>", "<title>Real-time PCR</title>", "<p>Total RNA from the all the pathogenic agent studies was reverse-transcribed simultaneously using the same master mix. The cDNA was then used to perform real-time PCR using BIORAD I cycler and the light cycler DNA master SYBR green I kit (Roche Diagnostics, Indianapolis, Indiana). The 18S gene was used as an endogenous control to normalize the HIF-1, GBP, and C5AR genes. Serial 10-fold dilutions of lymphoid cDNA were used to determine the PCR efficiency of each primer set. The slope value was applied to the formula E = 10^-1/m- 1 where m = slope value. The Ct (threshold cycle) values for all the genes were converted to fold change using the formula (1 + E)^ΔCt, where E denotes the efficiency of the primer set for a gene. ΔCt denotes the difference between the Ct values of control and treated samples of a given gene. (Personal Communication, C. Baker, National Institutes of Health)</p>" ]

[ "<title>Results</title>", "<title>Host gene expression <italic>in vitro</italic></title>", "<p>Microarray analysis was carried out at 3–6 time periods post exposure of PBMC to each pathogen or vehicle. Prior studies [##REF##12578971##11##] showed specific gene sets related to sex, age and other parameters, therefore it was important to first identify genes that are normally variant among healthy humans. Data from only the control samples of these healthy donors were subjected to ANOVA (p =&lt; 0.05) and 6% of the genes varied widely among the individuals who were healthy human donors. These genes that showed inconsistent expression profiles were excluded from further comparisons among the data sets from both control and exposed samples. This provided a baseline to confidently identify transcriptional responses induced by bacteria (anthrax, plague, <italic>Brucella</italic>), toxins (CT, SEB, BoNTA), or viruses (Dengue, VEE). Expression ratios of 10,000 genes on the custom array (accession numbers of which are listed in Additional file ##SUPPL##1##2##) and genes (Additional file ##SUPPL##4##5##) in Human Atlas 1.2 were determined by comparing the levels of mRNA in control and pathogen-treated PBMC paired for each exposure time frame. Each measurement was carried out at least 3 times.</p>", "<title>Consistency of responses</title>", "<p>We used PBMC from at least 3 different donors, exposing cells to pathogen or vehicle for specified periods of time. Figure ##FIG##0##1a## is a cluster analysis of exposures to B. anthracis for 2, 4, 8 and 24 h exposures. The result from the 3 different donors (male, ages 61, 27, 41) is closely replicated among the donors.</p>", "<title>Unique gene patterns induced by BTAs</title>", "<p>The gene responses were dissected to identify sets of genes that will differentiate one agent from another based on the patterns of host gene induction. The GeneSpring (Silicon Genetics, California) clustering diagram illustrates gene expression patterns that can discriminate among the various pathogenic agents (Fig. ##FIG##0##1b##) by identification of sets of genes where up regulation (red) and down regulation (blue) is seen for specific pathogens. The combination of these selected genes can be the foundation for designing specific diagnostic assays for exposure to one or more agents. For example, gene sets A and B (column labels at the bottom of the dendogram), differentiate <italic>B. anthracis </italic>from all other pathogenic agents except Dengue; gene sets F and G readily illustrate expression patterns that differentiate gene responses to these 2 pathogenic agent. Similarly, gene sets D and E and H show host responses to SEB that are distinguishing; BoNTA and VEE show similar patterns with gene sets A and B, but are readily separated by gene set C. Principal component analysis (PCA) (Fig. ##FIG##1##2a##) illustrates clustering relationships that show marked differences in overall gene patterns among the 3 toxins used in this study. Additionally, gene patterns for the earliest exposure for SEB or CT clustered less closely with the later exposure times (Fig. ##FIG##1##2b##), but when observed relative to all pathogens, the four exposure time periods for SEB were relatively closely clustered. A striking observation (Fig. ##FIG##1##2b##) is that for all pathogens <underline>except</underline> SEB, the longest exposure times differ markedly from the clusters of the early time periods. For <italic>B. anthracis, Y. pestis, B. melitensis</italic>, and CT, those late exposure times cluster together for these various pathogens. This loss of pathogen-specific responses <italic>in vitro</italic> after lengthy exposure was not seen for the <italic>in vivo</italic> studies.</p>", "<title>Use of training and test data sets for classifying test exposures</title>", "<p>To determine whether the microarray data obtained in this study can be used to predict the exposure type of an uncharacterized sample or condition, we applied a supervised learning method for class prediction (GeneSpring) that uses the k-nearest neighbor algorithm. When algorithm was applied on the data set (training set) to predict the exposure type of a data set obtained from an exposure to <italic>Y. pestis </italic>(test set), we were able to correctly predict the type of exposure with a p &lt; 0.02. We previously reported that a set of predictor genes was identified when samples from exposures of piglets to SEB were used as test sets [##REF##15522586##12##,##REF##15522843##13##].</p>", "<title>Functional classification of genes differentially regulated</title>", "<p>Gene ontological analysis was carried out for the genes that were differentially expressed. Comparison of gene responses, based on functional similarities, not surprisingly, showed many up regulated genes coding for inflammatory mediators (Fig. ##FIG##2##3##). We clustered and sorted the differentially expressed genes by their functional classification. These functional classifications are depicted in Figure ##FIG##2##3##. For gene group (<italic>i</italic>) \"Growth Factor, Cytokines &amp; Chemokines,\" anthrax, <italic>Brucella </italic>and SEB showed major up regulation of most genes coding for inflammatory mediators; the other 5 agents had mixed or modest effects. Similarly, categories (<italic>iii</italic>) \"Interleukins and Interferon Receptors\" and (<italic>iv</italic>) \"Interleukins\" showed up regulation by most pathogens, notable exceptions being the viruses. Down regulated genes, though seen extensively throughout the study, displayed functional clustering for each pathogenic agent such as (<italic>ii</italic>) \"Homeostasis &amp; detoxification,\" (<italic>v</italic>) \"Ligand-gated ion channels,\" and so forth. Plague induced high levels of interleukin-6, macrophage inflammatory protein-1 beta, tumor necrosis factor-alpha (TNF-α), and granulocyte macrophage colony stimulating factor (GM-CSF) when compared with <italic>Brucella </italic>and anthrax. Not surprisingly, the superantigen SEB displayed kinetic patterns for over expression of interferon-γ, IL-2, IL-6, MIP-1α, and GM-CSF (Fig. ##FIG##2##3##). There are major differences in expression of death receptors, homeostasis, and caspases, examples of which include defensins and certain oxidases (homeostasis) that are down regulated by plague and SEB (Fig. ##FIG##2##3##, bullet <italic>ii</italic>). A large number of transcription factors are down regulated by anthrax, <italic>Brucella</italic>, and SEB, but plague consistently down regulated the widest range of these genes.</p>", "<title>Gene responses induced by BTAs <italic>in vivo</italic>; comparison with <italic>in vitro</italic> changes</title>", "<p>To determine gene changes induced by BTAs in an animal model, NHP were exposed to <italic>B. anthracis </italic>spores by aerosol challenge. This model has been characterized previously to mimic inhalation anthrax in humans. Blood samples were collected 24 h, 48 h, and 72 h post exposure (by 72 h the NHP were beginning to show signs of the illness, which progresses very rapidly to lethality). The gene expression profiles for <italic>in vitro</italic> exposure of PBMC to anthrax spores were compared with those found in isolated PBMC at various time periods from NHP. Even by 24 h, a robust response was observed (Fig. ##FIG##3##4a##), showing up regulation of genes coding for proteases; proteosome components c2, c3, c5; various cytokines; pro-apoptotic genes; cyclic adenosine monophosphate (cAMP)-related kinases, cAMP regulated transcription factors; and hypoxia inducible factor-1 (HIF-1). Down regulated genes included tyrosine kinases, cytokine receptors, growth factors, and adenosine diphosphate (ADP) ribosylation factors. Comparison of the <italic>in vivo</italic> results with the <italic>in vitro</italic> changes induced by anthrax (Fig. ##FIG##3##4b##), showed remarkable similarities in gene patterns. Clearly many more changes were observed <italic>in vivo</italic> than <italic>in vitro</italic>. Certain surface antigens showed significant alteration that was unique to anthrax exposure. Diagrams were constructed to identify sets of genes that were up regulated (Fig. ##FIG##4##5##) at either 24 (blue) or 72 h (red); other gene sets showed up regulation at both time periods (center of graph, genes in both blue and red). Similarly, certain gene sets showed unique and common down-regulation patterns (far right, Fig. ##FIG##4##5##).</p>", "<p>A few genes were selected that showed changes induced by <italic>B. anthracis </italic>exposure were confirmed by RT-PCR, and the level of expression was compared both <italic>in vitro</italic> and <italic>in vivo</italic> after anthrax exposure. The <italic>in vivo</italic>/<italic>in vitro</italic> trends were very similar for many genes including IL-6 (Fig. ##FIG##5##6a##) and Transducin beta-1 subunit (GNB1) (Fig. ##FIG##5##6b##). Altered regulation of that G-protein was not seen with the other pathogenic agents. In an experiment of SEB exposure to NHP (Fig. ##FIG##6##7##), IL-6 and guanylate binding protein GBP-2 were up regulated (6- and 65-fold, respectively) by 30 min post-exposure and the increased expression persisted through 24 h (the last time point tested, data not shown). Among all the pathogens studied so far, SEB was found to be the only pathogen to dramatically alter GBP-2.</p>", "<title>Evaluating the gene selections</title>", "<p>In order to evaluate the quality of the set of genes selected by the two methods above, a set of 10 simulated samples was generated for each condition by randomly choosing a set of values consistent with the conditions regulation types. Sets of samples in which values had a 10%, 20%, 30%, 40% and 50% chance of being chosen at random were also generated. Each test sample was then classified using the chosen set of genes by scoring it against each of the 29 conditions. A condition scored 1 for every gene on which its regulation type was consistent with the sample. The test sample was considered to be classified correctly if the condition from which it was originally generated had the unique highest score. The results using the list of genes obtained by the \"grow\" (respectively \"shrink\") method are shown in Figure ##FIG##7##8##.</p>", "<p>Comparison of gene selection approaches: Because there is overlap between the gene types for some pairs of conditions, there must exist a sample consistent with some condition that cannot be classified correctly as that condition. Nevertheless for the generated data, both approaches obtained perfect classification at the 20% noise level for selections of between 150 and 775 genes. Even at the 30% noise level the grow method achieved perfect classification with 297–305 genes and the shrink method achieved perfect classification with 340–342 genes. Interestingly, in the added noise cases, gene selections above a certain size start to perform worse. This is because removing genes that play little role in distinguishing between conditions removes added noise without removing much discrimination power.</p>", "<p>We have used simulated random noise perturbing the input data samples, since we had no prior basis to assume any specific bias in noise (direction or subset of genes). If noise levels on a given sample were for some reason biased toward the identified patterns of some other disease for that subject, our technique will not perform as well as it does against random noise. Since at this point possible bias or coloration of the noise is unknown, we experimented with a range of noise levels, including levels well higher than the total noise experienced in modern gene expression platforms.</p>", "<title>Gene profiles to discriminate control from infected lymphoid cells</title>", "<p>To identify common gene profiles that existed among all these individual donors, we used GeneSpring to identify shared baseline expression levels of genes examined for 75 control samples. As these control datasets were subjected to various analyses, after excluding normally varying genes, we noticed genes that were expressed at low levels in the control samples but were significantly overexpressed in response to one or more pathogens. We selected genes that showed a dramatic change in expression level and could be used to discriminate among the pathogens. Gene profiles from two pathogens are shown with BoNT-A (Fig. ##FIG##8##9a##) or <italic>B. melitensis </italic>(Fig. ##FIG##8##9b##) in which the indicated genes readily differentiated it from the other 7 BTAs. Although some of these genes were slightly up regulated by one or more of the other pathogens, even those few genes illustrate the possibilities of distinguishing BoNT-A or <italic>B. melitensis </italic>from each of the other 7 pathogens.</p>", "<title>Confirmation of gene changes by real-time PCR</title>", "<p>To further confirm the expression levels of a few selected genes, we performed semi-quantitative real-time PCR using the same RNA samples isolated from lymphoid cells that had been exposed to the various pathogenic agents/vehicle and used to carry out the microarray experiments. Gene expression for 3 selected genes are compared based on real-time PCR along with gene array results. Expression of GBP-2, which was significantly up regulated by exposure of lymphoid cells to SEB (~10-fold by this technique vs. 6-fold from the microarray analysis), was not altered by any other pathogenic agents studied. Also up regulation was observed for HIF-1 and C5AR by BoNTA and CT and there was good agreement with data from the microarray studies (Fig. ##FIG##9##10##).</p>" ]

[ "<title>Discussion</title>", "<p>The objective of this study was to use host gene expression responses to aid in detection of exposure to biological threat agents, we aimed to a) discriminate among various pathogenic agents that start with similar flu-like symptoms [##REF##9244332##14##] yet lead to severe illness by various routes (Additional file ##SUPPL##0##1##) select sets of genes that can convincingly be used to differentiate normal from exposed samples c) identify sets of genes that could be used for very early detection of exposure or estimating stage of illness post-exposure and d) continue to characterize similarities and differences in host responses for <italic>in vitro</italic> exposures to show some predictability for host gene profiles in NHP models that replicate the illness induced in humans. In another study, we found sets of genes from SEB exposures <italic>in vitro</italic> also were predictive of <italic>in vivo</italic> responses in a piglet model of SEB-induced lethal shock [##REF##15522586##12##].</p>", "<p>The approach depends on circulating lymphoid cell mRNA responses reflecting a historical record (by their mRNA responses) of encounters with pathogenic agents. Use of unique gene patterns for diagnosis has been shown for certain cancers [##REF##12067990##15##, ####REF##11290540##16##, ##REF##11553815##17####11553815##17##]; another study showed that various pathogens induced unique gene responses in lymphocytes [##REF##11805339##18##] and in dendritic cells [##REF##11679675##19##]. To minimize variability as reported [##REF##12578971##11##] in baseline gene expression among donors, in our study blood was collected at the same time of the day for all donors. Of the 8 pathogenic agents in this study, most have been characterized as having rapid effects on lymphoid cells [##REF##9244332##14##,##REF##10943532##20##]. For BoNT-A and VEE, the primary target tissues are inaccessible, although VEE has been shown to interact with lymphoid cells [##REF##10623754##21##]. The gene expression data suggest that pathogen binding to specific receptors on the human PBMC initiates a series of events that contribute to the ultimate illness, producing host responses indicative of a particular pathogenic agent (Additional file ##SUPPL##5##6##) or showing common responses typical of a severe inflammatory reaction (Fig. ##FIG##2##3##).</p>", "<p>The kinetics of the course of the infection and the disease process is essential in the study of gene changes induced in the host by these pathogens (Additional file ##SUPPL##5##6##). Upon inhalation of <italic>Y. pestis </italic>by NHP, infected alveolar macrophages migrate to the liver and spleen [##UREF##2##22##,##REF##11863216##23##] where they proliferate rapidly (~24 h). It is thought that LPS from the bacterial cell wall of <italic>Y. pestis </italic>(Gram-negative) induces circulatory collapse and widespread organ failure, leading to death within days [##REF##10807389##24##]. SEB, CT, and LPS can induce rapid onset of illness (even less than 1 h), involving loss of regulation of vascular tone, vascular leakage and end-stage organ failure, in 1 to 3 days [##REF##9244332##14##,##REF##4825613##25##,##UREF##3##26##]. <italic>B. anthracis </italic>(Gram-positive) is transmitted as spores, which, upon inhalation, become engulfed by macrophages and transported to lymph nodes [##REF##10943532##20##,##REF##11747719##27##, ####REF##11700539##28##, ##REF##11173038##29##, ##REF##11700539##30##, ##REF##2507648##31####2507648##31##]. Upon production of a sufficient bacterial load, flu-like symptoms occur followed by sudden onset of respiratory distress, progressive shock, and death. As recent events have demonstrated, with cases of inhalation anthrax, treatments initiated after patients became seriously symptomatic can be marginally successful [##REF##11747719##27##]. In this study, host gene expression responses in NHP to <italic>B. anthracis </italic>exposure were seen at the earliest time point examined, 24 h (Fig. ##FIG##3##4b##). In contrast, in these same NHP, a sufficient pathogen load to be identified by culture techniques did not occur until day 3 and clinical diagnosis would not be possible until ~ day 4–5, much too late to initiate reliably effective treatments.</p>", "<p>Studies of pathogenesis show up regulated cytokine genes as a common response for many pathogens [##REF##9244332##14##,##UREF##4##32##] and that would not, necessarily, distinguish among them (Fig ##FIG##2##3##). Therefore, we focused on host responses that may potentially identify stage-specific targets and can also serve as early diagnostic markers. The regulation of certain early genes is transient and may relate to factors that participate in recruitment of monocytes to the sites of infection. The genes that are expressed late relate to DNA damage-inducing proteins, hypoxia-inducible proteins, and proteases. Characterization of apoptosis as a result of exposure was reported for SEB, DEN-2 [##REF##11549879##33##], plague [##REF##10922034##34##], and anthrax [##REF##10943532##20##,##REF##11747719##27##,##REF##11353060##35##] in numerous cell types, including those of lymphoid origin. We observed induction of apoptotic genes by these agents in our studies (Additional file ##SUPPL##5##6##). In contrast, <italic>Brucellae </italic>is known to inhibit apoptosis in their mononuclear phagocytic host cells [##REF##10603407##36##] and we also observed pro-apoptotic genes to be down regulated by Brucella. In regard to the findings with plague exposure, it is important to note that these infections occurred under conditions that limit the ability of <italic>Yersinia </italic>outer proteins (YOP) to alter host cell physiology by down regulation of cytokines [##REF##10922034##34##] and oxygen radicals [##REF##3965398##37##] in calcium-free cultures of macrophages. We observed up regulation of certain cytokines after 1–2 hr of exposure to <italic>Y. pestis </italic>and a pattern of gene expression that can explain reduced synthesis of oxygen radicals. We suspect that different biochemical mechanisms contribute to pathogenesis when YOPs are produced [##REF##11713916##38##,##REF##11020378##39##].</p>", "<p>In contrast to common infectious diseases, human cases of exposure to some of the biological threat agents are rare. Indeed, for certain of these pathogens, there is no appropriate animal model that replicates the illness as it appears in humans. Furthermore, dose effects and other variations suggest the need to investigate <italic>in vitro</italic> approaches that show some correlations to <italic>in vivo</italic> findings. For <italic>in vivo</italic><italic>B. anthracis </italic>studies, PBMC from the spore-exposed animals was collected 24, 48, and 72 h. The <italic>in vitro</italic> study utilized PBMC (from healthy donors) exposed to spores for various time periods. We found many more gene expression changes <italic>in vivo</italic> than <italic>in vitro</italic>, perhaps because <italic>in vivo</italic> changes include both primary and secondary responses. Comparisons of <italic>in vitro</italic>/<italic>in vivo</italic> results showed similarities in genes that code for lymphoid receptors/signaling pathways that, when taken as a group, showed a pattern specific for <italic>B. anthracis </italic>and include a G-protein Transducin beta subunit (GNB1), cAMP related genes, Calmodulin regulated genes, cytokines and MAPKK (mitogen activated protein kinase kinase), some of which had previously been reported in response to anthrax exposures [##REF##11104681##40##]. Not unexpectedly, genes coding for cytokines showed similarities <italic>in vitro</italic> vs <italic>in vivo</italic> (Fig ##FIG##4##5## and ##FIG##5##6##). Our microarray data were in accordance with recent reports by Pickering et al. that showed up regulation of some of the cytokines in response to infection by <italic>B. anthracis </italic>spores including TNF-α, IL-8, IL-1β, GM-CSF, IFN-γ and IL-6 [##REF##15501768##41##]. However, these genes, alone, would not necessarily distinguish anthrax from other pathogens. In general, the genes expressed by 4 h <italic>in vitro</italic> and 72 h <italic>in vivo</italic> were similar and correlated to the symptoms that appear after progression of inhalation anthrax in NHP.</p>", "<p>Aerosol challenge of SEB in NHPs up regulated (65-fold) the gene coding for interferon-regulated GBP-2 (Fig. ##FIG##6##7##) but <italic>in vitro</italic> (Figure ##FIG##9##10##) it was up-regulated 10-fold. <italic>In vivo</italic>, GBP-2 upregulation occurred by 15 min post exposure. Other pathogens showed minor or no effects on the expression of that particular G-protein (Figure ##FIG##9##10##). This may not be surprising in light of seminal studies showing that CT and pertussis toxin work through different guanine triphosphate (GTP)-binding proteins to regulate intracellular cAMP levels [##REF##8815789##42##,##REF##10895075##43##]. This study confirmed gene expression responses induced by CT, anthrax, and <italic>Brucella </italic>that are known participants in regulation of adenylyl cyclase as well as those relating to ADP ribosylation factor [##REF##7927706##44##, ####REF##11579087##45##, ##REF##9193671##46####9193671##46##] (Additional file ##SUPPL##5##6##). We observed a down regulation of the host adenylyl cyclase but an up regulation of cAMP-related genes upon anthrax exposure in NHP samples (Figure ##FIG##3##4a##). Because <italic>B. anthracis </italic>has its own adenylyl cyclase, it may be playing a role in affecting cAMP-related genes of the host [##REF##11173038##29##,##REF##8135809##47##].</p>", "<p>Since most biothreat pathogen exposures start with flu-like symptoms, discriminating them from common pathogenic illnesses for early diagnosis at a treatable stage is one of the critical issue. Even though these 8 pathogens initially cause similar symptoms, such as malaise, fever, headache, and cough, unique sets of genes are induced by each and can be related to the course of illness [##REF##9244332##14##] (Additional files ##SUPPL##0##1## and ##SUPPL##5##6##). Using these signature gene profiles to assess possible exposure to pathogenic agents or to differentiate them from non-lethal illnesses, when the classical identification of a pathogen is not conclusive, has the potential to fill a gap in the arsenal of diagnostic tools.</p>" ]

[ "<title>Conclusion</title>", "<p>Rapid detection, before the symptoms appear or even at various stages of illness offers the opportunity to initiate stage-specific therapeutic approaches to ameliorate the devastating results of these pathogenic agents. The use of host genomic markers offers an option to differentiate classes of pathogen exposure, gauge severity of impending illness and apply appropriate therapeutic strategies.</p>", "<p>¹Genes were selected and their expression profiles compared with gene array and real-time PCR. 18S was used as a reference gene for comparison of these 3 test genes.</p>", "<p>²GBP-2 was a gene that was identified as being massively up regulated by SEB using differential display PCR (C. Mendis, et al).</p>", "<p>nd = not determined</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Effective prophylaxis and treatment for infections caused by biological threat agents (BTA) rely upon early diagnosis and rapid initiation of therapy. Most methods for identifying pathogens in body fluids and tissues require that the pathogen proliferate to detectable and dangerous levels, thereby delaying diagnosis and treatment, especially during the prelatent stages when symptoms for most BTA are indistinguishable flu-like signs.</p>", "<title>Methods</title>", "<p>To detect exposures to the various pathogens more rapidly, especially during these early stages, we evaluated a suite of host responses to biological threat agents using global gene expression profiling on complementary DNA arrays.</p>", "<title>Results</title>", "<p>We found that certain gene expression patterns were unique to each pathogen and that other gene changes occurred in response to multiple agents, perhaps relating to the eventual course of illness. Nonhuman primates were exposed to some pathogens and the <italic>in vitro</italic> and <italic>in vivo</italic> findings were compared. We found major gene expression changes at the earliest times tested post exposure to aerosolized <italic>B. anthracis </italic>spores and 30 min post exposure to a bacterial toxin.</p>", "<title>Conclusion</title>", "<p>Host gene expression patterns have the potential to serve as diagnostic markers or predict the course of impending illness and may lead to new stage-appropriate therapeutic strategies to ameliorate the devastating effects of exposure to biothreat agents.</p>" ]

[ "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>RD and RH drafted the manuscript, performed the genomic analysis, data mining and the apoptosis studies. GVL; BK; XH; CP; MJ; LS; NK; DH and LL carried out the exposures to the various pathogens. SE and PL participated in the statistical analysis of the microarray data. PR; AD; AM; CM; CC; AR; SP and RN participated in the microarray studies for the different pathogens. MJ conceived of the study, and participated in its design and coordination. GVL; DY and EH participated in the design and coordination on the study. All authors read and approved the final manuscript.</p>", "<title>Disclaimer</title>", "<p>Material has been reviewed by the Walter Reed Army Institute of Research. There is no objection to its presentation and/or publication. The opinions or assertions contained herein are the private views of the author, and are not to be construed as official, or as reflecting true views of the Department of the Army or the Department of Defense.</p>", "<title>Pre-publication history</title>", "<p>The pre-publication history for this paper can be accessed here:</p>", "<p><ext-link ext-link-type=\"uri\" xlink:href=\"http://www.biomedcentral.com/1471-2334/8/104/prepub\"/></p>", "<title>Supplementary Material</title>" ]

[ "<title>Acknowledgements</title>", "<p>This work was supported by the Defense Advanced Research Projects Agency (DARPA) and by the Defense Threat Reduction Agency, Project Number: G0020_04_WR_B.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p>(a) <italic>B. anthracis </italic>exposure to PBMC from 3 different donors. Data shown are from exposures at 2, 4, 8 and 24 h. Data from each exposure time period were separately evaluated in order to identify common trends among the three donors (males, ages 61, 41, and 27 years old (respectively) with diverse ethnicity). (b). Comparisons of gene profiles for 8 pathogenic agents. Human PBMC were exposed to each of these pathogenic agents for at least 3 appropriate time periods. RNA was isolated, and the reverse transcript hybridized to cDNA arrays. Unique gene patterns were induced by BTAs. Cluster diagram of gene expression patterns use Gene Spring analysis to illustrate groups of genes that show discriminatory patterns for various threat agents. These genes were compared for their expression patterns across all agents and time points. Red is up regulated, blue is down regulated and black is no change compared to the control sample. The expression patterns illustrate how one can differentiate pathogenic agents by selection of sets of gene expression patterns for examination. (Gene accession ID numbers, rather than gene names, are all provided legibly in the graphs of Additional files).</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p>PCA relational analysis to show how the gene profiles (various exposure times) cluster for each toxin (a) and the relationship among the various pathogens (b). Human PBMC were exposed to each of these pathogenic agents for at least 3 appropriate time periods. RNA was isolated, and the reverse transcript hybridized to cDNA arrays.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p>Functional categories of genes that show similarities and differences between these pathogens. Accession numbers are shown associated with Figure 2, Additional files. Human PBMC were exposed to each of these pathogenic agents for at least 3 appropriate time periods. RNA was isolated, and the reverse transcript hybridized to cDNA arrays.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p>Comparison of host gene responses in vivo and in vitro exposures to anthrax. Gene expression profiles in PBMC from healthy human donors exposed to anthrax spores in vitro were compared with gene expression patterns obtained in PBMC taken at 24, 48 and 72 hr after exposure of NHP to anthrax spores by aerosol challenge. (a) Gene cluster analysis of significantly altered genes in vivo. (b) comparison of gene expression profile between in vivo and in vitro exposures.</p></caption></fig>", "<fig position=\"float\" id=\"F5\"><label>Figure 5</label><caption><p>Clustered sets of genes to illustrate stage-specific vs. commonly expressed genes for in vivo exposures of NHP at 24, 72 h or at both time periods.</p></caption></fig>", "<fig position=\"float\" id=\"F6\"><label>Figure 6</label><caption><p>Confirmation of selected gene changes by RT-PCR with in vitro and in vivo samples for IL-6 (a) and Transducin beta-1 subunit, GNB1, (b).</p></caption></fig>", "<fig position=\"float\" id=\"F7\"><label>Figure 7</label><caption><p>Expression of GBP-2 and IL-6 genes after in vivo SEB exposure of NHP for 30 min. Gene expression profiles in PBMC from NHP exposed to SEB for 30 min. RNA samples were isolated and used in the PCR assays using primers specific for IL-6 and GBP-2.</p></caption></fig>", "<fig position=\"float\" id=\"F8\"><label>Figure 8</label><caption><p>Ordered genes and resulting percentage correct classifications. For a given number of genes, a set of genes most able to discriminate between disease states is selected. Simulations of noisy readings of patient gene expression levels are performed for varying levels of noise. Each colored line plots the percentage of correct classifications versus the number of genes used to make the classification for one particular percentage of random values in the simulated readings. With no noise, very few genes are required to discriminate perfectly. With high noise levels (say, 50%), even 1000 genes cannot reliably discriminate well.</p></caption></fig>", "<fig position=\"float\" id=\"F9\"><label>Figure 9</label><caption><p>Expression patterns of genes that were at baseline levels in all controls and showed unique expression patterns in (a) BoNT-A exposures or (b) <italic>B. melitensis </italic>exposures compared to all 8 pathogens.</p></caption></fig>", "<fig position=\"float\" id=\"F10\"><label>Figure 10</label><caption><p>Real-time PCR determination of gene expression in response to each of 8 pathogenic agents. Primers were designed for these 3 genes and 18S, which was used as a reference gene for comparison of these 3 test genes. GBP-2 was a gene that was identified as being massively up regulated by SEB using differential display PCR (Mendis, et al) and was of particular interest to us.</p></caption></fig>" ]

[]

[]

[]

[]

[]

[]

[ "<supplementary-material content-type=\"local-data\" id=\"S1\"><caption><title>Additional file 1</title><p>Comparison of time course of progression of illness for selected BTAs and pathogens: Graph courtesy of COL George Korch, USAMRIID, Ft. Detrick, Maryland. Hatched marks indicate onset of flu-like symptoms (fever, headache, chills) for each agent, <italic>X </italic>indicates time frame in which death usually occurs (case fatality rate, CFR), and the number of <italic>X</italic>s suggest the degree of lethality if untreated early in the course of illness. (i) The toxins (yellow bars) cause onset of acute illness within a few hours, but side effects can persist for many weeks. (ii) Bacterial BTAs (pink bars) follow different time course after infection ranging from days to weeks. Many bacterial infections begin with flu-like symptoms, but proceed to respiratory distress and lethal shock within ~ a week. In contrast, a prolonged illness is common in brucellosis, which is caused by Gram-negative bacteria (<italic>B. melitensis. B. suis</italic>, and <italic>B. abortus</italic>) that are highly infectious via aerosol route. (iii) Viruses (green bars) VEE and DEN infections each progress differently because VEE can proceed to the meninges of the brain, developing into encephalitis. In the case of dengue, the incubation period is 3 to 15 days, with the acute febrile illness lasting 3–5 days. The period of mortality is associated with cessation of the febrile illness or with secondary complications in DHF.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S2\"><caption><title>Additional file 2</title><p>A list of genes on the custom array. This table shows a list of the genes that are present on the microarrays used in this study.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S3\"><caption><title>Additional file 3</title><p>List of the first 50 genes that were selected using the Grow/Shrink method. This table lists the genes that passed the statistical analysis using the Grow/Shrink method.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S4\"><caption><title>Additional file 4</title><p>List of the accession numbers and sequences of the primer sets used in this study.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S5\"><caption><title>Additional file 5</title><p>Lists the genes of the Clontech 1.2 human array.</p></caption></supplementary-material>", "<supplementary-material content-type=\"local-data\" id=\"S6\"><caption><title>Additional file 6</title><p>Comparisons of gene profiles for 8 pathogenic agents. Cluster analysis of gene expression profiles of PBMC exposed to the 8 pathogens. Human PBMC were exposed to each of these pathogenic agents for appropriate time periods, and the results are sorted based on functional responses, rather than clustering of similar gene expression patterns. RNA was isolated, reverse transcribed and hybridized to cDNA arrays. Red bars indicate up regulation, green bars show down regulation of the genes, and selection of genes for significance is defined in methods. The numbered bullets (right margin of the figure) indicate families of genes showing similarities or unique properties. Gene accession numbers are shown to the left of the figure.</p></caption></supplementary-material>" ]

[]

[ "<graphic xlink:href=\"1471-2334-8-104-1\"/>", "<graphic xlink:href=\"1471-2334-8-104-2\"/>", "<graphic xlink:href=\"1471-2334-8-104-3\"/>", "<graphic xlink:href=\"1471-2334-8-104-4\"/>", "<graphic xlink:href=\"1471-2334-8-104-5\"/>", "<graphic xlink:href=\"1471-2334-8-104-6\"/>", "<graphic xlink:href=\"1471-2334-8-104-7\"/>", "<graphic xlink:href=\"1471-2334-8-104-8\"/>", "<graphic xlink:href=\"1471-2334-8-104-9\"/>", "<graphic xlink:href=\"1471-2334-8-104-10\"/>" ]

[ "<media xlink:href=\"1471-2334-8-104-S1.eps\" mimetype=\"application\" mime-subtype=\"postscript\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1471-2334-8-104-S2.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1471-2334-8-104-S3.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1471-2334-8-104-S4.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1471-2334-8-104-S5.pdf\" mimetype=\"application\" mime-subtype=\"pdf\"><caption><p>Click here for file</p></caption></media>", "<media xlink:href=\"1471-2334-8-104-S6.eps\" mimetype=\"application\" mime-subtype=\"postscript\"><caption><p>Click here for file</p></caption></media>" ]

{ "acronym": [], "definition": [] }

[ "<title>Background</title>", "<p>Antarctic Krill are ancestors of clayfish and prawn. They have a slow evolutionary speed, and are not good at swimming. They are distributed in Vancouver saigan sea area, Russia, Ukraine and so on. They have the largest amount of protein among all organisms so far, over 16% in wet weight while over 65% in dry weight. Antarctic KO contain more than 30% of essential eicosapentaenoic acid (EPA, C:20:5, n-3) and docosahexaenoic acid (DHA, C:22:6, n-3) as well as astaxanthin (provitamin E) in concentrations of 200 – 400 ppm [##UREF##0##1##]. Besides, they also have abundant phospholipids, flavonoids, vitamin A, Alpha-linolenic acid (ALA), astacin and other nutrients [##UREF##1##2##].</p>", "<p>Cardiovascular disease (CVD) and colon cancer incidence are known to be closely related to dietary factors [##REF##5635018##3##]. As the modern medical science indicated, CVD had become the first killer of people's health. And diabetes, CVD and hypertension often cluster [##REF##11223440##4##,##UREF##2##5##]. All kinds of dangerous factors, including change of life style, dietetic habit, indicate that the morbidity of coronary heart disease in China will rise quickly in the future. The latest survey of nutritional and health conditions of the citizens showed, prevalence of abnormal lipids among adults in China was 18.6%. Several recommendations based on experimental, epidemiological or nutritional data had shown that the incidence of CVD was positively correlated with saturated fatty acids intake and negatively with unsaturated fatty acids intake [##REF##10545672##6##]. Controlled intervention trails with encapsulated fish oil supplements containing EPA plus DHA have established their triglyceride-lowering effects and modifying influence on other CVD risk factors independent of blood lipid-lowering [##REF##8900453##7##,##UREF##3##8##]. Cancer chemoprevention has emerged as an important way to control cancer by using dietary agents capable of blocking neoplastic inception or delaying disease progression [##UREF##4##9##]. Research indicated that this strategy was promising on reducing the morbidity of cancer in high-risk population and in the common group [##REF##16204691##10##]. In the recent years, the role of nutrients as chemopreventive agents has been the focus of current world. The idea of using natural extracts for the chemoprevention of cancer was obtained from animal models, clinical trials, human epidemiologic studies and cell studies [##REF##12354359##11##]. Colon cancer is the second most common cause of cancer deaths in the USA, with an estimated annual incidence of 104,950 and mortality of 56,290 in 2005. The incidence of colon cancer is also constantly increasing in Asia [##UREF##5##12##]. A lot of effective methods and measures had been on service to reduce the mortality of colon cancer. Using naturally compounds present in dietary sources for chemoprevention is considered as a practical hopeful approach.</p>", "<p>Some compounds have undergone clinical trials against colon cancer based on this idea [##REF##2403415##13##,##REF##10750661##14##]. The anti-cancer properties of KO are currently under investigation to see whether it can block or delay the malignant progression of transformed cells by modulating cell proliferation or differentiation. The cytostatic effect could be attained by the ingestion of KO on human colon carcinoma cell. Therefore, we studied whether KO had potential to slow the proliferation rates via programmed cell death. At present, studies are mainly focusing on the distribution, biological and physiological characteristics of Krill resources, few on the functional characteristics of KO. Therefore, the aim of this study was to evaluate the antilipidemic and anti-cancer effects of KO.</p>" ]

[ "<title>Methods</title>", "<title>Preparation of KO</title>", "<p>The krill oil was purchased from local market. The oil was placed at the refrigerator with the temperature of -20°C. Distilled water was added to Ryoto sugar ester (S-1170F, Mitsubishi-Kagaku Foods Corporation) to make the concentration 5 g/L, and then the solution was mixed for 10 minutes at the speed of 1500 rpm using magnetic whisk mix. The final solution was used as the solvent for KO to make desired samples for rats.</p>", "<title>Cell lines</title>", "<p>Human colon cancer cells, SW480 were purchased from Stem Cell Bank, Chinese Academy of Sciences, and were propagated in the recommended medium. And the cells were cultured in RPMI 1640 medium with 10% FBS and 1% penicillin streptomycin (GIBCO) at 37°C in a incubator of humidified atmosphere with 5% CO₂.</p>", "<title>Animal study</title>", "<p>Sixty adult male SD rats, weighing 180 ~ 190 gram each, were purchased from Zhejiang Academy of Medical Science, and divided into six groups. Ten rats of each group were housed in one cage and fed in the center for experimental animals of Zhejiang University with the room temperature (25 ± 2)°C and humidity (63 ± 2)%. The HFD was composed of 78.8% common feed, 1% cholesterol, 10% yolk powder, 10% lard and 0.2% cholate. Body weights and feed intakes were recorded every three days during the next seven weeks. Following was the animal study procedure: SD rats were kept in SPF animal experiment lab for 1 week with fundament diet. Serum samples were taken from rats' tails and centrifuged at 3000 rpm for 15 min at 4°C for the analysis of TC, TG, HDL-C and LDL-C levels. Then rats were fed with HFD for 2 weeks to establish the hyperlipidemia model according to TC, TG, HDL-C and LDL-C levels. And then the rats were divided into 6 groups according to the TC levels randomly. The low dose group was given HFD with 16.65 g/L of KO, and the mid dose, the high dose and the higher dose groups were feed by HFD supplemented with 33.3 g/L, 99.9 g/L and 199.8 g/L of KO, respectively. The high lipid control was given HFD with the same volume solvent. The dosages of KO were 0.5 ml/100 g body weight of experimental rats. And the medication group was treated with lovastatin 100 mg/kg/day. Serum from 12 h-fasting rats' tails was taken after 1 w, 3 w and 7 w. The analysis of TC, TG, HDL-C and LDL-C were measured by automatic hitachi-7170 analyzer.</p>", "<title>Cell proliferation assay</title>", "<p>The effect of KO on the viability of cells was determined by methyl thiazolyl tetrazolium (MTT) (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl tetra zoliumbromide) assay. In short, the cells were plated at 1 × 104 cells per well in 200 μl of complete culture medium containing 0, 2.5, 5, 10, 15, and 20 μg/ml concentrations of KO in 96-well microtiter plates. The KO solutions were dissolved in anhydrous ethanol and mixed with aseptic PBS to achieve the desired final concentration. Each concentration of KO was repeated in ten wells. After incubation at 48 h, 72 h, 120 h in the incubator, cell viability was determined. Twenty microlitres MTT (5 mg/ml in phosphate-buffered saline stock, diluted to working strength 1 mg/ml with media) was added to each well and incubated for 4 h. After that, the MTT solution was removed from the wells by careful aspiration. Then, 200 μl of buffered DMSO was added to each well and plates were well-mixed. The absorbance was recorded on a microplate reader at the wavelength of 490 nm. The effect of KO on growth inhibition was evaluated as percent cells viability where vehicle KO-treated cells were considered as 100% survival.</p>", "<title>Statistical analysis</title>", "<p>All data were handled with SPSS package program version 11.0 (SPSS, Chicago, IL). The results were listed in the way of mean ± std ( ± s), and the significance between the control and treated groups was performed using ANOVA Statistical analysis. Differences were considered significant at P &lt; .05.</p>" ]

[ "<title>Results</title>", "<title>Establishment of hyperlipidemia rats model</title>", "<p>The results showed that the total TC, TG, and LDL-C of serum of rats significantly increased (P &lt; 0.05) after 2 weeks of HFD, while high HDL-C decreased significantly (P &lt; 0.05) compared with pre-feeding. It indicated the hyperlipemia models were established successfully. Thus, the following experiment was feasible.</p>", "<title>Effects of KO on body weights</title>", "<p>Our study showed that, there were no significant differences, in body weight levels in other groups compared with the control group before administration (Figure ##FIG##0##1##). On the other hand, after four weeks' treatment with KO and lovastatin, diversity was observed. Except for Higher dose group, the body weight was decreased in each group obviously compared with control (p &lt; 0.05). Among them, the difference between medication and control group was up to super remarkable level (p &lt; 0.01). In addition, compared with medication group, there were no diversities in low, mid and high dose groups.</p>", "<title>Effects on the lipid levels</title>", "<p>TG, TC, HDL-C and LDL-C levels in serum before and after administration of KO were analysed. Significant changes in serum lipids were observed after 4 week consumption of KO. The intake of KO for 4 weeks had a lowering effect on TG, TC and LDL-C, while no changes were observed in HDL-cholesterol. All doses of KO could significantly decrease the TC (Figure ##FIG##1##2##), TG (Figure ##FIG##2##3##) and LDL-C (Figure ##FIG##4##5##) level of serum (P &lt; 0.05) after 4 weeks of KO; high and higher doses of KO could significantly increase the HDL-C (Figure ##FIG##3##4##) levels. And compared with the hyperlipidemia control group, TC and LDL-C of all dose groups, TG of low and mid dose groups decreased significantly (p &lt; 0.05), and no remarkable differences were observed. It indicated that KO could decrease TC, TG and LDL-C levels of serum significantly. On the other hand, it had little influence on serum HDL-C.</p>", "<title>Anti-cancer effect</title>", "<p>As was shown in Figure ##FIG##5##6##, treatment of SW480 cells with 2.5, 5, 10, 15, 20 μg/ml concentrations of KO for 48 h, resulted in 15.2%, 20.6%, 22.0%, 24.6% and 29.9% inhibition of cell growth respectively, compared with vehicle treated control. Exposure of SW480 cells to KO resulted in slight decrease in cell viability at the highest concentration of 50 μg/ml (data not shown). Treatment of colon cancer cells with KO also resulted in time-dependent inhibition of cell growth, and the effect was more pronounced at 72 h post-treatment. The inhibition rates were 16.7%, 23.4%, 26.7%, 27.2% and 30.9%, respectively. While in 120 h, the numbers were 19.7%, 24.2%, 28.0%, 30.8% and 33.5%, respectively.</p>" ]

[ "<title>Discussion</title>", "<p>It suggested that administration of KO caused significant change in body weight and the KO had good weight lose effect similar to the lipid-reducing medicine. It has also been reported that different types of fatty acids have different effects on body weight gain and insulin resistance. Saturated fatty acids (SFAs) produce more weight gain and insulin resistance than polyunsaturated fatty acids (PUFAs) in some studies [##REF##8901785##15##,##REF##9572150##16##]. Of all the PUFAs, EPA and DHA exert more favorable influence on body weight [##REF##1991575##17##]. It was reported body weight of the diet group supplement with fish oil increased by 25% [##UREF##6##18##]. Fatty acids derived from fish oil had been shown to alter proinflammatory cytokine production and acute-phase protein (APP) synthesis in vitro. The presence of APP has been suggested to contribute to weight loss. The administration of PUFAs to hepatocytes had suggested that EPA may have direct effects on the modulation of APP production [##REF##8998126##19##]. The mechanism of the KO used in the present study and the relative contributions of its components requires further study.</p>", "<p>The present study proved that KO could significantly decrease the TC, TG and LDL-C levels of serum, while slightly increase the HDL-C levels as well. It was reported the long-term intake of egg-white hydrolysed (hEW) for 20 weeks had a lowering effect on TG and TC, while no changes were observed in HDL-C [##REF##17245440##20##]. It was also found that some natural active substances could significantly decrease the TC, TG and LDL-C levels, but there were no significance in HDL-C [##UREF##7##21##,##UREF##8##22##]. A great deal of findings indicated that EPA and DHA could reduce body blood TG, TC and LDL-C. It had been reported that fish oil had cholesterol-lowering effect because it increased DHA content in the membrane and could improve membrane fluidity. Thus, it increased removal rate of VLDL-C and LDL-C particles from plasma by improving hepatic microsomal membrane fluidity, and thus, it decreased the TG, TC, LDL-C significantly. But it didn't increase the HDL-C level. Unlike fish oil and KO, corn oil significantly increased plasma HDL-C levels and reduced the risk factor for CVD since HDL-C had the important role of reversing cholesterol transport [##UREF##9##23##]. The reason may be the difference of various food composition, operation mechanism, or different animal study itself (great individual differences), the complex of metabolism of HDL-C and so on.</p>", "<p>As far as we know, this is the first report showing the anti-cancer effect of KO on human colon cancer cells. However, more detailed studies are required to determine the exact mechanism(s). KO contains four constituents which may reduce the risk of developing the colon cancer: EPA, DHA, ALA, sphingolipid. Many reports indicated that EPA and DHA could inhibit the growth of some cancer cells, such as breast, prostate cancer [##REF##11857389##24##,##REF##10584888##25##]and so on. Fish oil was known to reduce growth of certain tumors, such as colon, breast and prostate cancers [##REF##15570047##26##]. These effects were attributed to the content of PUFAs of the n-3 family in fish oil, in particular EPA and DHA, which regulate cellular signaling paths. Several PUFAs are known regulators of the ligand-activated transcription factors known as peroxisome proliferators-activated receptors (PPARs) [##REF##11748246##27##]. Originally implicated in the regulation of lipid metabolism and adipocyte differentiation, the PPARs have also been implicated in cell differentiation, cell proliferation, and in inflammatory responses [##UREF##2##5##,##REF##10584888##25##]. However, the mechanism(s) underlying of the anti-cancer effects were not fully understood. ALA alters the fatty acid composition of cell membranes in crucial ways and inhibits the release of pro-inflammatory eicosanoids, which control the growth and invasiveness of tumor cells and modulate the cycle of cell apoptosis among the many factors [##UREF##10##28##].</p>", "<p>As was known, sphingomyelin which existed in cell plasma membrane of most mammalian was the main component of myelin sheath. It was rich in cell membrane of brain and nerve. Several studies had demonstrated altered total sphingolipid composition, both increases and decreases, in cancer cells. It is unclear what effect changes in sphingolipid composition and metabolism(s) have on the sensitivity of cancer cells to therapy. It was reported that dietary sphingomyelin protected against apoptosis and hyperproliferation caused by the hydrophobic bile salt deoxycholate potential implications for colon cancer. They thought the use of sphingomyelin to boost the chemotherapy response of cancer cells could have a significant impact on treatment outcome [##UREF##11##29##]. Numerous epidemiologic studies have provided a gist for the development cancer chemoprevention protocols using bio-active dietary agents capable of eliminating pre-malignant or malignant cells. The results of this study suggested KO had the potential to affect the steady state cell population.</p>" ]

[ "<title>Conclusion</title>", "<p>The main finding of this study was that the treatment with KO reduced the body weight and serum TG, TC and LDL-C levels significantly. The results showed hypolipidemic properties and thus, the consumption of KO may provide benefit to control serum lipid levels in certain diseases. In addition, KO affected the steady state cell population to inhibit growth of colon cancer and thus may be a good candidate for development as a chemopreventive and/or therapeutic agent against colon cancer. In a word, it is hopeful for KO to be one kind of potential functional food.</p>" ]

[ "<p>This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<ext-link ext-link-type=\"uri\" xlink:href=\"http://creativecommons.org/licenses/by/2.0\"/>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.</p>", "<title>Background</title>", "<p>Cardiovascular disease (CVD) and colon cancer incidence are known to be closely related to dietary factors. This article evaluated effects of krill oil (KO) on serum lipids of hyperlipidemia rats and human colon cancer cells (SW480). Serum lipids of rats fed with high fat diet (HFD) and different doses of KO were measured by automatic analyzer. Effect of KO on viability of cells was determined by methyl thiazolyl tetrazolium (MTT) assay.</p>", "<title>Results</title>", "<p>Except for higher dose group, body weights decreased significantly. Total cholesterol (TC), LDL-cholesterol (LDL-C) of all dose groups, Triglycerides (TG) of low and mid dose groups descended significantly, while there were no significant differences of HDL-cholesterol (HDL-C), compared with control group. Treatment of colon cancer cells with KO also resulted in time-dependent inhibition of cell growth.</p>", "<title>Conclusion</title>", "<p>Our findings indicated that the consumption of KO may provide benefits to control serum lipid levels in certain diseases and inhibit growth of colon cancer cells. Therefore, KO may be a good candidate for development as a functional food and nutraceutical.</p>" ]

[ "<title>List of abbreviations used</title>", "<p>ALA: alpha-linolenic acid; CVD: cardiovascular disease; DHA: docosahexaenoic acid; EPA: essential eicosapentaenoic acid; HFD: high fat diet; KO: krill oil; MTT: methyl thiazolyl tetrazolium; PPARs: peroxisome proliferators-activated receptors.</p>", "<title>Competing interests</title>", "<p>The authors declare that they have no competing interests.</p>", "<title>Authors' contributions</title>", "<p>JJZ participated in the design of the study and carried out the statistical analysis. JHS performed the animal and cell studies, participated in preparing the test materials and writing of the manuscript. WBQ participated in the design of the cell study. ZZC participated in the revision of the manuscript. DL conceived of the study, and took part in its design and coordination. All authors read and approved the final manuscript.</p>" ]

[ "<title>Acknowledgements</title>", "<p>The authors express their appreciation to Chun-Mei Yang, Yang Yang and Hui Liu for assistance in cell study. The authors also gratefully acknowledge Zhi-Guo Zhang, Hua-Li and Gao-Feng Yuan for animial study.</p>" ]

[ "<fig position=\"float\" id=\"F1\"><label>Figure 1</label><caption><p><bold>Effects of KO on body weights of rats</bold>. The body weight of experimental rats ( ± s, g, n = 10) in the third (3rd w) and seventh (7th w) weeks. *, p &lt; 0.05, compared with high lipid model control group; **, p &lt; 0.01, compared with high lipid control group; a, p &lt; 0.05, compared with medication group.</p></caption></fig>", "<fig position=\"float\" id=\"F2\"><label>Figure 2</label><caption><p><bold>Effects of KO on Total Cholesterol content</bold>. Serum Lipids levels obtained after treatments with KO and lovastatin. The rats were fed for 4 weeks continuously. Data were mean values ± std for 10 rats.</p></caption></fig>", "<fig position=\"float\" id=\"F3\"><label>Figure 3</label><caption><p><bold>Effects of KO on Triglycerides content</bold>. Serum Lipids levels obtained after treatments with KO and lovastatin. The rats were fed for 4 weeks continuously. Data were mean values ± std for 10 rats.</p></caption></fig>", "<fig position=\"float\" id=\"F4\"><label>Figure 4</label><caption><p><bold>Effects of KO on HDL-Cholesterol content</bold>. Serum Lipids levels obtained after treatments with KO and lovastatin. The rats were fed for 4 weeks continuously. Data were mean values ± std for 10 rats.</p></caption></fig>", "<fig position=\"float\" id=\"F5\"><label>Figure 5</label><caption><p><bold>Effects of KO on LDL-Cholesterol content</bold>. Serum Lipids levels obtained after treatments with KO and lovastatin. The rats were fed for 4 weeks continuously. Data were mean values ± std for 10 rats.</p></caption></fig>", "<fig position=\"float\" id=\"F6\"><label>Figure 6</label><caption><p><bold>Effect of KO on cell viability time-dependent in human colon cancer SW480 cells</bold>. The cells were exposed to the specified concentration of KO for 48 h, 72 h and 120 h, and viability of cells were determined by MTT assay. Cell viabilities were described as percentages; vehicle-treated cells were regarded as 100% viable. Details were described in methods.</p></caption></fig>" ]

[]

[ "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M1\" name=\"1476-511X-7-30-i1\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mtext>x</mml:mtext><mml:mo>¯</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>", "<inline-formula><mml:math xmlns:mml=\"http://www.w3.org/1998/Math/MathML\" id=\"M2\" name=\"1476-511X-7-30-i1\" overflow=\"scroll\"><mml:semantics><mml:mover accent=\"true\"><mml:mtext>x</mml:mtext><mml:mo>¯</mml:mo></mml:mover></mml:semantics></mml:math></inline-formula>" ]

[]

[]