Datasets:

bhargavi909

/

cancer_classification

like 0

This study was performed to explore the effective management of bleeding associated with radiofrequency ablation RFA of benign thyroid nodulesMethods Thirtyfive patients with benign thyroid nodules who were treated with ultrasoundguided RFA from July to December at the Third Affiliated Hospital of Sun YatsenUniversity were retrospectively reviewed The technique efficacy bleeding and other complications were assessed during the followup periodResults The mean technique efficacy was 06 at month and 06 at months after the procedure One case of an intranodular haematoma and two cases of voicechange month were observed All patients recovered with corresponding treatmentConclusion Although the incidence of haemorrhage is low serious haematomas are lifethreatening Therefore having a comprehensive understanding of the potential complicationsan accurate clinical strategy and adequate technical skills may prevent or help to properly managethese complicationsKeywordsRadiofrequency ablation benign thyroid nodules haemorrhage management haematomaultrasoundDate received January accepted June 1Department of Medical Ultrasound The Third AffiliatedHospital of Sun Yatsen University Guangzhou China2General Surgery Department The Third AffiliatedHospital of Sun Yatsen University Guangzhou ChinaThese authors contributed equally to this workCorresponding authorsBo Liu General Surgery Department The Third AffiliatedHospital of Sun Yatsen University Tianhe RoadGuangzhou City Guangdong Province China Jie RenDepartment of Medical Ultrasound The Third AffiliatedHospital of Sun Yatsen University Tianhe RoadGuangzhou City Guangdong Province ChinaEmails renjieguangzhou126com liubojake126comCreative Commons Non Commercial CC BYNC This is distributed under the terms of the CreativeCommons AttributionNonCommercial License creativecommonslicensesbync40 which permitsnoncommercial use reproduction and distribution of the work without further permission provided the original work is attributedas specified on the SAGE and Access pages ussagepubcomenusnam accessatsage 0cIntroductionAssociationfor AdultThyroid nodules are extremely commonand the associated morbidity rate rangesfrom to according to highresolution ultrasound US findings12Mostthyroid nodules are benign andrequire no intervention other than clinicalfollowup According to the AmericanThyroidManagementGuidelinesPatients withThyroid Nodulesand DifferentiatedThyroid Cancer thyroidstimulating hormone suppression therapy for benign thyroid nodules BTNs is not recommendedbecause the potential harm outweighs thebenefit3 Radioiodine therapy was historically an effective treatment for thyroid hotnodules and a possible alternative to surgery Howeverthis technique has beenproven to have uncertain efficacy andsome adverse effects such as hypothyroidrecurrence4“ Surgery may beism orconsideredgrowing BTNs withpressurerelated symptoms neck discomfort cosmetic concerns or decreased quality of life3 At present partialtotal thyroidsurgery is considered the gold standardtreatment Surgeryassociated withnumerous complications such as nerveinjury anaesthesiarelated problemslonghospital stays conspicuous scars haemorrhage and lesions ofthe parathyroidglands78 In addition hypothyroidism isinevitable after totalthyroidectomy andrequires lifelong hormone supplementationHenceincreasingly minimally invasivetherapeutic strategies are currently used totreat BTNs In most cases several thermalablation techniques such as laser ablationmicrowave ablation radiofrequency ablation RFA and highintensity focused UShave been shown to be effective in BTNsAmong these thermal ablation techniquesRFA is the most widely applied910forisRFA of thyroid diseases first reported in“ is considered efficacious and safeJournal of International Medical Researchfor treatment of BTNs1415 To date no lifethreatening complications related to RFAhave been reported Howeverseveralcases of haemorrhagerelated to fineneedle aspiration FNA or core needlebiopsy CNB have been reported16“Although a microinvasive procedure suchas FNA can result in massive uncontrolledbleeding resulting in upper airway respiratoryuncontrolledbleeding is a rare but lifethreatening complication of RFA Thus management ofbleeding associated with RFA of BTNs isof vitalimportance This study was performed to explore the effective managementof bleeding associated with RFA of BTNsobstructionsuchMaterials and methodsofThis study was approved by the EthicsCommitteethe Third AffiliatedHospital of Sun Yatsen University andwritten informed consent was obtainedfrom all patients prior to the performanceof USguided FNA or CNB and RFA Therequirement to obtain informed consent forpublication was waived because of the retrospective nature of the studyPatientsAll consecutive patients who underwentRFA of BTNs at our institution from July to December were analysed Thefollowing inclusion criteria were appliedconfirmation of benignancyBethesdaClass II by FNA cytology or CNB complaints of pressure symptoms compressivesymptoms neck discomfort orforeignbody sensation or cosmetic problems a2cm maximum diameter of the indexnodule anxiety about a malignancy unsuitability for surgery or unwillingness toundergo surgery and a normal serum thyrotropin concentration normal completeblood counts and normal blood coagulation test results The exclusion criteria 0cHu et alwere nodules showing malignant featuresie taller than wide spiculated marginmarked hypoechoic appearance or microcalcifications on US imaging19 abnormalthyroid function performance of othertreatments for the thyroid nodules within months before the procedure pregnancyand age of years For the presentstudy only patients with 15 months offollowup after the procedure were included Thirtyfive patients met the inclusioncriteriaPretreatment assessmentBefore the procedure conventional USfindings USguidedFNA findingscontrastenhanced US CEUS findingsand laboratory and clinical results wereevaluated Two radiologists TW and JR with and years of thyroid US experiencerespectively performed the USUSguided FNA and CEUS examinationsusing a Logiq E9 US device GE MedicalSystems Milwaukee WI USA equippedwithtransducerwith a MHzfrequency range “ MHz The USexamination included characterisation ofthe location shape size margins solidcystic proportions echogenicity calcification status and internal vascularity ofeach nodulefrequency ofan ML615centrelinearLaboratory tests included the levels ofthyroidstimulating hormone free triiodothyronine free thyroxin and thyrotropina complete blood cell count and a coagulation test prothrombin time and activatedpartial thromboplastin time The nodulevolume was calculated using the followingvolume¼ length 02 width 02equationdepth 02 In addition all patientsunderwent vocal cord function assessmentsby an experienced laryngologist before theablation procedure Atenrolment allpatients were asked to rate their pressuresymptoms on a 10cm visual analoguescale grade “ cm and the cosmeticgrading score was assessed by the physicianas described in the consensus statement20Procedures and equipmentpreventsignificantAll RFA procedures were performed by oneradiologist JR with years of experienceperforming RFA in an outpatient clinic Weused an RF generator VIVA RF SystemVR STARmed Gyeonggisi South Korea andan internally cooled 18gauge 70mmlength or 10mm activetip electrodeStar RF ElectrodeVR STARmed Localanaesthesia with lidocaine was appliedto the puncture site The hydrodissectiontechnique was used under US guidance glucose and norepinephrine weremixed and injected into the surroundingthyroid capsule which provided a safe distance between the needle tip and adjacentcritical structures During the procedurewe paid special attention to the preservation of surrounding important structurestocomplicationsTherefore two essential techniques werethe transisthmic approach andappliedtechnique2122 Ablationthe movingshotwas suspended when the index nodule wascovered by hyperechoic zones The technique efficacy TE was then evaluated byCEUS at to minutes after RFA untilthedisappearedTechnicalthechange of an entire nodule to a noenhancement zone on realtime CEUSFor nodules with an enhancement zonean additional ablation was performed todestroy the nodule as much as possibleComplications were monitored immediatelyafter the procedure and during the followup period Major and minor complicationsand adverse effects were defined accordingto the criteria established by the Society ofInterventional Radiology2324success was defined ashyperechoiczones 0cJournal of International Medical ResearchFollowup evaluationatandperformedserum thyroidAny specific complaints or concerns wererecorded for month Postproceduralfollowup wasand months after treatment At each followup visit a US examination CEUS examinationhormonemeasurements were performed pressuresymptoms and the cosmetic grading scorewere evaluated and the volume ofthetreated nodule was calculated The TE wascalculated using the following equationTE¼ final nodule volumeinitial nodule volume 02 Statistical analysisstatistical analyses were performedAllusing SPSS software version IBMCorp Armonk NY USA Continuousvariables are expressed as mean 06 standarddeviation Quantitative data for volumeand TE were analysed using a pairedttest A P value of 14 was consideredstatistically significantResultsThe patients™ characteristics are summarised in Table Thirtyfive patients underwent RFA including male and femalepatients mean age years The meanlargest BTN dimension was 06 mmrange “ mm and the mean BTNvolume was 06 mL Twentytwototal complications minor and majorcomplications were observed among thetreated patients None of these complications was lifethreatening and all occurredwithout sequelaeNodule volumeAfter treatment the overall volume of thesignificantly decreased 06nodules 06 mL at mL at month and 06 mL at monthsbaselineTable Patients™ baseline characteristics n¼ Characteristics 06 06 06 06 06 06 06 06 Age at treatment yearsMalefemale ratioBody weight kgBody height cmBody mass index kgm2Symptom score “Cosmetic score “Cosmetic score of Cosmetic score of Cosmetic score of Preablation serum FT4 level pmolLPreablation serum TSH level mIULIndex nodule on ultrasoundRight sideLeft sideLargest dimension mm to to 15Data are presented as mean 06 standard deviation ornumber of patientsFT4 free thyroxin TSH thyroidstimulating hormoneP and the TE was 06 at month and 06 at months P Table Figure shows the shrinkage of the nodules at and months after the procedure comparedwith baseline no hypoechoic blood supplywas observed within the area ofthenodulesBleeding complicationsTwelve patients developed bleeding complicationsincluding a perithyroidal haematoma minor complication in patientsand an intranodular haematoma majorcomplication in patient as shown inTable The haematomas were detectedby US scans which revealed gradualenlargement of a hyperechoic mass in oraround the nodules Figure For thepatient with intranodular haemorrhage 0cHu et alTable Changes in volume before RFA and at each followup visitParameterInitial month laterLargest diameter mm 06 “ 06 “Volume mLTechnique efficacy ”Data are presented as mean 06 standard deviation range 06 “ 06 “ 06 “ months laterP value 06 “ 06 “ 06 “ Figure a c e Ultrasound examination and b d f contrastenhanced ultrasound examination of a39yearold woman treated with radiofrequency ablation a b Ultrasound and contrastenhanced ultrasound revealed a cysticsolid nodule before ablation c d One month after ablation ultrasound showed ahypoechoic nodule with a decreased volume d e Six months after ablation the volume of the nodule haddecreased further and no blood supply was observed within the area of the nodule 0cJournal of International Medical ResearchTable Complications and adverse effects in patients who underwent RFA of thyroid nodulesComplication or adverse effectAdverse effectsFeverPainDizzinessSensation of heatMinorPerithyroidal haematomaVomitingnauseaOedemaswellingVoice change for monthMajorVoice change for monthIntranodular haemorrhageData are presented as n the haematoma was controlled throughtimely use of the ablation needle to coagulate the injured blood vessel and by injecting lyophilising thrombin powder into thehaematoma Figure Most of the perithyroidalseriesrequired only observation with or withoutcompression and disappeared within to weeks after the procedure None of the patientssubscapularhaematomahaematomasdevelopedthisinaOther complications and adverse effectsThe adverse effects of RFA included fevern¼ pain n¼ dizziness n¼ and a sensation of heatn¼ Minor complications includedoedemaswelling n¼ and a voicechange for month n¼ vomitingnausean¼ DiscussionImageguided thermal ablation techniquessuch as laser ablation ethanol ablationmicrowave ablation highintensity focusedUS and particularly RFA have recentlybecome more widely used to treat thyroidthe creation ofnodules Briefly the basic mechanism ofRFA involvesthermaldamage by friction and heat conductionwhich is generated from an oscillatinghighfrequency alternating electric currentproduced by the RFA generator and thentransferred through the electrode tip Theenergy of RFA is powerful and accurate2526 RFA is considered an effectiveand safe treatment for control of BTNsIn most cases the incidence of haemorrhage and other complications is low20However haemorrhage is sometimes lifethreatening because serious haematomasmay compress the upper airways Manyreports have described active bleedingduring FNA of thyroid nodules and RFAof hepatocellular carcinomas14““ andsome reports have described fatalities14ThusimportantcomplicationhaemorrhageanisThree types of haemorrhage may occurperithyroidal subcapsular and intranodular121427“ The mechanism of haemorrhage is thoughtto be related to themechanical or thermal injuries induced bythe RFA electrode tip3031 Thyroid nodulesreportedly have abundant capsular vesselsthat are usually anastomosed with vesselspenetrating into the core32 These numerousvessels are abnormal thinwalled and susceptible to rupture Large thyroid nodulesare another cause of haemorrhage becausemultiple insertions are often required totreat such nodules In additionthepatient cannot coordinate with the physician during the RFA procedure the perithyroidal orintrathyroidal vessels mayeasily be damaged by movement of theneedle tip or production of heat energyifIt is important to manage bleeding associated with RFA of BTNs Based on ourexperience we suggest several steps to preventshouldobtain a thorough medical history of eachpatient before the procedure All risk factorsdrugssuch bleeding Physiciansincludingbleedingfor 0cHu et alFigure Ultrasound examination and contrastenhanced ultrasound examination of patients with intranodular haemorrhage and perithyroidal haemorrhage a Ultrasound and contrastenhanced ultrasoundrevealed a hyperechoic mass lesion in the nodule b Ultrasound and contrastenhanced ultrasound revealedperithyroidal haemorrhagedrugsnonsteroidalantiplateletantiinflammatory drugs and anticoagulantsand diseases affecting coagulation shouldbe recorded33 In addition the patient™scoagulation function should be thoroughlyevaluated All patients with clinical coagulation disorders should be excluded Evenwhen coagulation indices are normalinpatients with high risk factors for bleedingsuch as liver cirrhosis endstage renal disease anticoagulant use or hypertension34sufficient preoperative preparation shouldbe emphasised A patient with active bleedingthesein the presentstudy metconditions Although his coagulation indices were normal he had a subclinical coagulation disorder due to endstage liverdisease Fresh frozen plasma or human prothrombin complex should be used inpatients with liver cirrhosis and anticoagulants should be withdrawn in these patientswhich will help to improve coagulation function before the procedure If a possibility ofbleeding exists Reptilase haemocoagulaseatrox forinjection Pentapharm BaselSwitzerland can be used preoperativelyDuring the RFA procedure an effectiveclinical strategy and adequate technical 0cJournal of International Medical ResearchFigure Ultrasound examination and contrastenhanced ultrasound examination of patients with intranodular haemorrhage or perithyroidal haemorrhage during ablation a Ultrasound revealed a hyperechoicmass lesion in the nodule b Ultrasound showed an ablation needle inserted into the nodule to coagulatebleeding vessels c After lyophilising thrombin powder was injected into the hematoma ultrasound andcontrastenhanced ultrasound showed disappearance of the hyperechoic mass lesion and microbubbleextravasation d Ultrasound showed a hyperechoic mass lesion around the thyroid and contrastenhancedultrasound showed no microbubble extravasation around the thyroid e After lyophilising thrombinpowder was injected into the haematoma no microbubble extravasation was observedskills are both essential Patient cooperationis the first requirement When the needle tipis in the patient™s body any uncooperativemotion of the patient may lead to injury ofvessels or other structures Most patientscan endure the procedure under local anaesthesia however anxious patients mayrequire general anaesthesia to achieve cooperation If possible smallbore electrodesshould be chosen to decrease the risk ofbleeding35 It is necessary to cauterise thesupplying vasculature of nodules to avoidrecurrence and residue Howeverthepuncture route should be carefully designedto avoid pericapsular vessels and the electrode tip should be closely monitoredActive bleeding during needle puncture isvisible as a rapidly expanding hypoechoicor anechoic signal Locating the haemorrhagic focus is not difficult with CEUSguidance The bleeding pointcan beblocked by RF electrode tip insertion anddirect ablation When the bleeding is toorapid to control with the RF electrode tipby increasing the power drug injection is asuitable alternative Lyophilised thrombin 0cHu et alpowder can be dissolved in normal salineand then injected at the bleeding pointthrough a syringe with US guidance Onereport also described haemorrhage treatedby local injection of hypertonic saline andepinephrine solution in a patient with hepatocarcinoma36 Mildbleeding whichappears as a hypoechoic layer can mostlybe controlled using ice and compression ofthe neck for several minutes after the procedure30 All bleeding can be controlled byconservative methodsthus no surgicalintervention is needed Ecchymosis can befound after the procedure and usually disappears in approximately to weeksPostprocedure CEUS is indispensablefor all patients regardless of whether bleeding occurs CEUS is an objective evaluationtool for active bleeding37 Close clinicalobservation for hours postoperatively isrecommended in our department becausemost bleeding occurs during the firstlobectomy38Observation of the neck can help to detecta haematoma early and may aid in preventing serious adverse effectsthyroidhoursafterConclusionAcute thyroid bleeding is one possible complication of RFA although rare it is potentially lifethreatening Proper selection ofpatients and sufficient preparation areessential During the RFA procedureboth an effective clinical strategy and adequate technical skills are indispensable Thephysician should trace the electrode tipusing realtime US and sufficiently managebleeding Mild bleeding has limited morbidity and can be easily controlled by compression Active bleeding tends to be rarehowever it may be disastrous if the operator is unaware or careless Direct ablationwith the RF electrode tip and drug injectioninto the bleeding focus are effective modalities for active bleeding CEUS and closeobservation are also recommended afterthe procedureto detect abnormalitiesearly RFA is an effective and relativelysafe alternative for selected patients withBTNs if performed by skilled physiciansAuthor contributionsI Conception and design Jie Ren and Bo LiuII Administrative support Jie RenIIIstudy materials or patientsProvision ofKunpeng Hu and Yufan Lian IV Collectionand assembly of dataJinfen Wang andWenchao Li V Data analysis and interpretation Wenchao Li and Zhicheng YaoVIManuscript writing All authors VII Finalapproval of manuscript All authorsData availabilityData regarding the patients™ characteristics usedto support the funding are shown in Table Declaration of conflicting interestThe authors declare that there is no conflict ofinterestFundingthe NaturalThis work was supported by the NationalNatural Science Foundation of China CNNoScienceFoundation of Guangdong ProvinceNo2016A030313200 the Science and TechnologyProject of Guangzhou City No the Hengrui Foundation of Hepatobiliary andPancreaticNoCXPJJH1180000120183331NaturalScience Foundation of Guangdong ProvinceNothe FundamentalResearch Funds for the Central UniversitiesSun Yatsen University No 17ykpy67 andthe Clinical Research Project of Sun Yatsen University No 2017A030313580theCancerResearchORCID iDKunpeng Huorcid00000001 0cReferences Guth S Theune U Aberle J et al Very highprevalence of thyroid nodules detected byhigh frequency MHz ultrasound examination Eur J Clin Invest “ Tan GH and Gharib H Thyroid incidentalomas management approaches to nonpalpable nodules discovered incidentally onthyroid imaging Ann Intern Med “ Haugen BR Alexander EK Bible KC et al AmericanThyroid AssociationManagement Guidelines for Adult Patientswith Thyroid Nodules and DifferentiatedThyroid Cancer The American ThyroidAssociation Guidelines Task Force onThyroid Nodulesand DifferentiatedThyroid Cancer Thyroid “ Ceccarelli C Bencivelli W Vitti P et alOutcome ofradioiodine131 therapy inhyperfunctioning thyroid nodules a years™ retrospective study Clin EndocrinolOxf “ Reiners C and Schneider P Radioiodinetherapy of thyroid autonomy Eur J NuclMed Mol Imaging S471“S478 Nieuwlaat WA Hermus AR SivroPrndeljF et al Pretreatment with recombinanthuman TSH changes the regional distribution of radioiodine on thyroid scintigramsof nodular goiters J Clin Endocrinol Metab “ LinosDEconomopoulosKPKiriakopoulos A et al Scar perceptionsafter thyroid and parathyroid surgery comparison of minimaland conventionalapproaches Surgery “ Jeannon JP Orabi AA Bruch GA et allaryngeal nervethyroidectomy a systematicDiagnosis ofpalsy afterreview Int J Clin Pract “recurrent Lang B Woo YC and Chiu KW Identifyingpredictive factors for efficacy in high intensity focused ultrasound HIFU ablationof benign thyroid nodules “ a retrospectiveInt J Hyperthermia analysis“ Mauri G Pacella CM Papini E et alImageguided thyroid ablation proposalforterminology andstandardization ofJournal of International Medical Researchreporting“criteria Thyroid Sato M Tateishi R Yasunaga H et alMortality and hemorrhagic complicationsassociated with radiofrequency ablation fortreatment of hepatocellular carcinoma inendstagepatients on hemodialysissurveyrenalJ“Gastroenterol Hepatolnationwidediseasefora Krokidis M Spiliopoulos S Jarzabek Met al Percutaneous radiofrequency ablationof small renal tumours in patients with asingle functioning kidney longterm resultsEur Radiol “ Lim HK Lee Dupuy DE Monchik JM Decrea C et alRadiofrequency ablation of regional recurrence from welldifferentiated thyroid malignancy Surgery “JH Ha EJalRadiofrequency ablation of benign nonfunctioning thyroid nodules 4year followup results for patients Eur Radiol “et Braga M Cavalcanti TC Collaco LM et alEfficacy of ultrasoundguided fineneedleaspiration biopsy in the diagnosis of complex thyroid nodules J Clin EndocrinolMetab “ Kakiuchi Y Idota N Nakamura M et alA fatal case of cervical hemorrhage after fineneedle aspiration and core needle biopsy ofthe thyroid gland Am J Forensic Med Pathol “ Donatini G Masoni T Ricci V et al Acuterespiratory distress following fine needleaspiration of thyroid nodule case reportand review of the literature G Chir “ Roh JL Intrathyroid hemorrhage and acuteupper airway obstruction after fine needleaspirationthyroidglandLaryngoscope “theofofAssociation Gharib H Papini E Garber JR et alAmericanClinicalEndocrinologists American College ofEndocrinology and Associazione MediciEndocrinologi medical guidelines for clinicalpractice for the diagnosis and managementthyroid nodules“ update EndocrofPract “ 0cHu et al Na DG LeeetJHJung SLalRadiofrequency ablation of benign thyroidnodules and recurrent thyroid cancers consensusstatement and recommendationsKorean J Radiol “ Ha EJ Baek JH and Lee JH Movingshotversus fixed electrode techniques for radiofrequency ablation comparison in an exvivo bovine liver tissue model Korean JRadiol “ Jeong WK Baek JH Rhim H et alRadiofrequency ablation of benign thyroidnodules safety and imaging followup in“patients Eur Radiol Cardella JF Kundu S Miller DL et alSociety of Interventional Radiology clinicalpractice guidelines J Vasc Interv Radiol S189“S191 Sacks D McClenny TE Cardella JF et alSociety of Interventional Radiology clinicalpractice guidelines J Vasc Interv Radiol S199“S202 Goldberg SN Radiofrequency tumor ablation principles and techniques Eur JUltrasound “ Rhim H Goldberg SN Dodd GR et alEssential techniques for successful radiofrequency thermal ablation of malignanthepatic tumors Radiographics Spec No S17“S35 S36S39 Korkusuz Y Erbelding C Kohlhase Ket al Bipolar Radiofrequency Ablation ofBenign Symptomatic Thyroid Nodules initial Experience Rofo “ Garberoglio R Aliberti C Appetecchia Met al Radiofrequency ablation for thyroidnodules which indications The first Italianopinion statement J Ultrasound “ Baek JH LeeJH Sung JYet alComplications encountered in the treatmentof benign thyroid nodules with USguidedradiofrequencya multicenterstudy Radiology “ablation Chen MH Dai Y Yan KalIntraperitoneal hemorrhage duringandafter percutaneous radiofrequency ablationof hepatic tumors reasons and managementChin Med J Engl “et Rhim H Dodd GR Chintapalli KN et alRadiofrequency thermal ablation of abdominal tumors lessons learned from complications Radiographics “ Terry WI Radium emanations in exophthalmic goiter”blood vessels of adenomas ofthyroid J Am Med Assoc “ Hor T and Lahiri SW Bilateral thyroidhematomas after fineneedle aspiration causing acute airway obstruction Thyroid “ Minami Y Hayaishi S and Kudo MRadiofrequency ablation for hepatic malignancies is needle tract cauterization necessary for preventing iatrogenic bleeding DigDis “ Baek JH Kim YS Lee D et al Benign predominantly solid thyroid nodules prospective study of efficacy of sonographicallyguided radiofrequency ablation versus control condition AJR Am J Roentgenol “ Koda M Murawaki Y Hirooka Y et alComplications of radiofrequency ablationfor hepatocellular carcinoma in a multicenter study an analysis of treated nodules in patients Hepatol Res “ Wiggermann P Wohlgemuth WA Heibl Met al Dynamic evaluation and quantificationof microvascularization during degradablestarch microspheres transarterial chemoembolisation DSMTACE of HCC lesionsusingultrasoundCEUS a feasibility study Clin HemorheolMicrocirc “enhancedcontrast Rosenbaum MA Haridas M and McHenryCR Lifethreatening neck hematoma complicating thyroid and parathyroid surgeryAm J Surg “ 0c'

Neurologic Manifestations of Systemic Disease D Lapides Section EditorNeurologic Manifestationsof Systemic Disease SleepDisordersEric M Davis MD1Chintan Ramani MBBS1Mark Quigg MD MSc2Address1Division of Pulmonary and Critical Care Department of Medicine University ofVirginia Charlottesville VA USAEmail emd9bvirginiaedu2Department of Neurology University of Virginia Charlottesville VA USA Springer ScienceBusiness Media LLC part of Springer Nature This is part of the Topical Collection on Neurologic Manifestations of Systemic DiseaseKeywords Sleep disorders I Sleep manifestations of systemic diseases I Sleep impacts on health I Sleep apnea IInsomniaAbstractPurpose of review Sleep is intimately involved in overall health and wellbeing We provide acomprehensive report on the interplay between systemic diseases and sleep to optimizethe outcomes of systemic disordersRecent findings Spanning the categories of endocrinologic disorders metabolictoxicdisturbances renal cardiovascular pulmonary gastrointestinal infectious diseases autoimmune disorders malignancy and critical illness the review highlights the prevalentcoexisting pathology of sleep across the spectrum of systemic disorders Although it is rarethat treating a sleep symptom can cure disease attention to sleep may improve quality oflife and may mitigate or improve the underlying disorder Recent controversies inassessing the cardiovascular relationship with sleep have called into question some ofthe benefits of treating comorbid sleep disorders thereby highlighting the need for anongoing rigorous investigation into how sleep interplays with systemic diseasesSummary Systemic diseases often have sleep manifestations and this report will help theclinician identify key risk factors linking sleep disorders to systemic diseases so as tooptimize the overall care of the patient 0c Page of IntroductionCurr Treat Options Neurol All Earth™s species maintain a solar 24h cycle of rest andactivity and disrupting the cycle affects adaptation andhomeostasis Sleep™s quotidian œnormalness meansthat analogous to fish not knowing about water until itis dry sleep is not commonly thought about until it isdisruptedFor example about of the adult populationcomplain of transient insomnia and about experience chronic insomnia that disrupts daytime function[] Patients with chronic insomnia experience less workproductivity more absenteeism more accidents andmore hospitalizations leading to direct treatment costsof approximately 60B annually [] Considering thepotential widespread reach of comorbid sleep disordersevaluating sleep in the neurological patient is importantThis review will introduce the accepted anizationof sleep disorders review important features in historytaking and evaluation and survey the systemic diseasesthat have important comorbidities with particular sleepdisordersGeneral considerationsClassification of sleep disordersAn abridged listing of sleep disorders from the American Academy of SleepMedicine Table provides an overview of the current classification []Insomnia is a chronic dissatisfaction with sleep duration and quality that isassociated with daytime dysfunction Although pharmacologic treatment isoften pursued for chronic insomnia management outcomes are often betteraddressing underlying factors with the early use of cognitivebehavioral therapyfor insomnia CBTi []Sleeprelated breathing disorders involve dysfunction of the respiratory systemduring sleep usually resulting in daytime hypersomnia Obstructive sleepapnea OSA central sleep apnea CSA and respiratory effort related arousalsare classified under this category Treatment options including continuouspositive airway pressure CPAP positional therapy mandibular advancementdevices healthy weight loss and even a novel cranial nerve stimulator whichprotrudes the tongue forward during sleep [4cid129cid129]Central hypersomnias are defined as a primary dysregulation of sleep resultingfrom dysfunction of the central nervous system that causes daytimehypersomnia Often treatment addresses the underlying cause and may includeuse of strategic napping and wakepromoting medicationsCircadian disorders consist of various lesions or external disruptions of thecircadian timing system that desynchronize the brain™s clock from the externalsolar lightdark cycle resulting in hypersomnia or insomnia in a clockdependent fashion Treatment of circadian rhythm disorders involves adjustinglife around the patient™s desired sleep time or augmenting factors that entrainthe body™s clockParasomnias represent disorders of faulty inhibition of waking behaviors thatarise inappropriately during sleep and are divided into those that occur duringnonREM sleep REM sleep or state transitions REM sleep behavior disorder is aparasomnia characterized by loss of muscle atonia during REM sleep thatusually occurs in patients with neurodegenerative disorders It is often treatedeffectively addressing other sleep disturbances and treating with clonazepam ormelatonin [] 0cCurr Treat Options Neurol Page of Table Abridged classification of the AASM sleep disordersInsomniaChronic insomnia disorderShortterm insomnia disorderExcessive time in bedShort sleeperSleeprelated breathing disordersObstructive sleep apneaCentral sleep apneaSleeprelated hypoventilation disordersSleeprelated hypoxemia disordersCentral disorders of hypersomnolenceNarcolepsy types and Idiopathic hypersomniaKleineLevin syndromeHypersomnia due to medical disorder medication substance psychiatric disorderInsufficient sleep syndromeCircadian rhythm sleepwake disordersDelayedAdvancedIrregularNon hShift workJet lagParasomniasNREM relatedArousal disordersConfusional arousalsSleepwalkingSleep terrorsSleeprelated eating disorderREM relatedREM sleep behavior disorderRecurrent isolated sleep paralysisNightmare disorderOtherExploding head syndromeSleeprelated hallucinationsEnuresisSleep talkingSleeprelated movement disorders 0cCurr Treat Options Neurol Page of Table ContinuedRestless legs syndromePeriodic limb movement disorderLeg crampsBruxismRhythmic movement disorderBenign sleep myoclonus of infancyPropriospinal myoclonus at sleep onsetNormal variantsSleep historySleeprelated movement disorders consist of fragmentary often repetitive bodymovements that can disrupt sleep or sometimes worse disturb the sleep of bedpartners Periodic limb movement disorder PLMD and restless legs syndromeRLS both fall under this category and are treated with repletion of iron storesand consideration of dopaminergic agonists []A sleep history helps a patient disclose sleep findings and helps the physiciananize it into categories of hypersomnia sleep habits and scheduling sleepcharacteristics environmental issues and sleep interrupters Table The Epworth Sleepiness Scale quantifies the degree of hypersomnia []Most adults require “ h of daily sleep [] and prefer it anized into eithera monophasic nocturnal schedule or in a biphasic pattern augmented with anafternoon œsiesta The sleep pattern characterizes the presence and severity ofsleeponset insomnia sleep maintenance insomnia or terminal insomnia insomnia distributed within the last half of the sleep period œCatchup sleep aphenomenon of prolonged sleep on a free day is a classic sign of sleepdeprivation Habitual earlyphase advances œmorning larks latephase delays œnight owls or a chaotic irregular schedule can be a sign of circadiandisorders One also must inquire about common sleep disruptors including legmovements snoring witnessed apneas and environmental factorsDiagnostic testing modalitiesSleep diaryPolysomnographyThe sleep diary often available through standardized forms or evenwebsites or smartphone apps consists of “ weeks of selfreported sleeptimesThe overnight polysomnography PSG is the goldstandard measurementof sleep architecture respiratory disorders such as OSA and parasomniasIn the case of OSA the unattended home sleep study has had an 0cCurr Treat Options Neurol Page of Table A categorical sleep historyHypersomniaEpworth Sleepiness Scale Considering the last weeks how likely would you fall asleep while doing each task not at all points slight moderate severe Normal ‰¤ pointsSitting and readingWatching TVSitting inactive in public lecture church ¦Car passenger for an hourLying down to rest in the afternoonSitting conversationSitting quietly alone after lunchDriving stopped in trafficSchedulesleep timeWorkday bedtime and out of bedtimeWeekday bedtime and out of bedtimeWhat is your estimated sleep latency If min what are you doing in bed before you fall asleepHow often do you awaken at night and whyDo you need an alarm clock to awaken in the morningHow many days of the week do you nap and for how longEnvironmentDo you have a bedroomDo you have a bedpartner TV Mobile phone or other electronicsWhat are you doing right before bedtimeHow much caffeine coffeeteasoda popenergy drinks and alcohol do you consume and when is the latest intakeInterruptersDo you have leg pain or restlessnessDo you have chronic pain that prevents or interrupts sleepDo you have daytime hallucinations or dreams severe or lucid nightmares sleep paralysis or cataplexyDo you snore or have witnessed apneasMultiple sleep latency testincreasing role as a diagnostic testing alternative to the traditional inlabPSG Concerns of other sleep disorders or those that may be presentcomorbidly with probable OSA require inlab PSG that can measure sleeparchitecture and sleepassociated movementsThe multiple sleep latency test MSLT consists of a series of daytime napsfrom which sleep onset is calculated The test in combination with PSGperformed the night before is the gold standard in measuringhypersomnia especially in the evaluation of narcolepsy 0c Page of ActigraphyPersonal devicesCurr Treat Options Neurol Wrist actigraphy provides measurements of longterm patterns of rest andactivity as proxies for sleep and wakefulness Such patterns can help tocorroborate histories of sleep duration and timingPopular smartphones and other ambulatory devices with physiologicalmonitoring capabilities may transform the evaluation of sleep However arecent comparison of different brands of activity trackers found that sleepwake measurements varied widely in comparison with sleep diaries orstandard PSG [] The overall conclusion is that at the beginning of wearable devices are not ready for reliable quantification of sleep acrossindividuals Although serial recordings confined to a single individual mayhold some value these measurements have yet to be validatedSleep comorbidities with systemic diseasesEndocrine disordersThyroid diseaseConsidering the various sleep disorders and diagnostic tools afforded by a goodsleep history and sleep testing understanding the relationship between sleepdisorders and systemic diseases has farreaching implications in optimizing thecare of the patient The following sections will address sleep manifestations ofvarious neurological disorders arising from systemic disease based on an systemAlmost half of the patients with hypothyroidism report at least one sleep complaint such as restless sleep choking hypersomnia or fatigue [] OSA is presentin approximately [] A unique mechanism of airway restriction in hypothyroidism is myxedematous mucoprotein deposition in the airway™s soft tissuesand dilator muscles even though myxedema can be absent [] Larger goiters canalso cause OSA by external compression of the airway []On the other side of the thyroid spectrum hyperthyroidism is most closelyassociated with insomnia occurring in of patients [] Arousaldisorders”specifically sleep walking”also occur especially in the setting ofthyrotoxicosis [] proposed to arise from frequent arousals and impairmentof attaining slowwave sleep as the direct result of thyroid hormoneBeyond the treatment of the specific sleep disorder sleep problems usuallyremit following appropriate treatment of the underlying thyroid disorder []Type diabetes mellitusSleep disorders affect high proportions of those with type diabetes mellitusDM surveys of patients with DM compared with those of controls show a 0cCurr Treat Options Neurol Page of nearly 2fold propensity for insomnia fourfold higher use of sedativehypnoticsand a 10fold higher rate of hypersomnolence [] OSA is highly prevalent inDM and many are undiagnosed [] Contributors to a multifactorial series ofsleep disruptors include periodic limb movements and restless legs syndromeRLS diabetic neuropathy and fluctuations in blood glucose []DM presents an excellent model by which to demonstrate the reciprocaleffects of sleep disruption on the primary disease First sleep disturbances affectthe regulation of the neuroendocrine control of appetite Sleep deprivationpromotes overeating through hyperactivity of orexin system [] and activatesthe hypothalamicpituitaryadrenal system to increase cortisol secretionresulting in impaired glucose tolerance [ ] These multiple mechanismssupport clinical observations that untreated OSA may be reason for the ineffective treatment of DM and that accordingly treatment with CPAP leads toimprovements in glycemic control in some patients []Sex hormones and gender affect the distribution and susceptibility to a varietyof sleep disorders Men on the basis of relative airway collapsibility haveapproximately a twofold increased risk of OSA compared with women “ in males and “ in females [] A potential side effect in thetreatment of hypoandrogenism is the facilitation of OSA given the impacttestosterone has on upper airway collapsibility []Testosterone levels may affect the propensity for chronic insomnia Menwith hypoandrogenism demonstrate reduced sleep efficiency increased nighttime awakenings and reduced deep sleep compared with the normaltestosteronelevel controls although it is not clear whether these features improve with testosterone therapy [] Women experience higher rates of chronicinsomnia risk ratio of for women versus men which becomes even morepronounced in the elderly [] Despite sleeping longer overall sleep quality isoften lower in women than men []The distribution of sleep disorders in women varies with reproductivelifespan Younger women are more susceptible to restless legs syndromeRLS mainly on the basis of mensesassociated irondeficiency During pregnancy women are at significantly increased risk for the development of RLSwith an overall prevalence exceeding of all pregnant patients [] Treatment of RLS in pregnancy involves iron supplementation with a goal ferritinlevel mcgl Often oral iron repletion is adequate although there arereports of intravenous iron therapy in severe cases of pregnancyrelated RLSand irondeficiency [] Pregnancy is also associated with an increased prevalence of OSA up to of pregnant patients during the third trimester whichis associated with increased risks of complications including gestational hypertension gestational DM and preeclampsia []Although not a particular systemic neurological disease pharmacological effectson sleep form an important aspect of neurological sleep medicine since manymedications that are used by neurologists may affect sleep Table showscommon medications that provoke insomnia hypersomnolence respiratorysuppression parasomnias and RLSperiodic limb movement disorderSex hormonesMedications 0c Page of Curr Treat Options Neurol Table Medication classes and specific examples that can cause sleep disturbancesInsomniaCentral nervous system stimulants methylphenidate amphetamines modafinilCaffeineAntidepressantsSelective serotonergic reuptake inhibitors fluoxetine sertralineSelective norepinephrine reuptake inhibitors venlafaxine duloxetineSecondary tricyclic antidepressants desipramine nortriptylineCardiovascularBeta2 agonists albuterolVasopressors epinephrine dopamineCorticosteroidsSympathetic amines phentermineHypersomniaBenzodiazepines alprazolam diazepamNonbenzodiazepine receptor agonists zolpidem eszopicloneOpioidsH1 antihistamines diphenhydramineAntiepileptic agents phenytoin levetiracetamAntidepressantsSelective serotonergic reuptake inhibitors paroxetine sertralineTertiary tricyclic antidepressants amitriptylineTypical and atypical antipsychotics haloperidol olanzapineDopaminergic agonists ropinirole carbidopalevodopaAnticholinergic medicationsCentrally acting α agonists clonidine dexmedetomidineRespiratory suppressionOpioids oxycodone morphineBenzodiazepines diazepam clonazepamAlcoholPhenobarbitalParasomniasAntidepressants clomipramine fluoxetine citalopramNonbenzodiazepine receptor agonists zolpidemCaffeineAlcohol withdrawalRestless legs syndrome and periodic limb movementsSelective serotonergic reuptake inhibitors fluoxetine mirtazapineAntipsychotics haloperidol risperidoneTricyclic antidepressants amitriptyline clomipramine 0cCurr Treat Options Neurol Page of Renal diseaseInfectious diseasesSleep disturbances are highly prevalent in patients with chronic kidney diseaseCKD spanning the broad spectrum of sleep disorders including hypersomniainsomnia sleeprelated breathing and RLSThe prevalence of OSA in CKD ranges from to rates that are notexplained solely by overlapping comorbidities common to both OSA and CKD[] The cooccurrence of both CKD and OSA is associated with increasedcardiovascular events and allcause mortality [“] Usually OSA develops inpatients with CKD independent of underlying renal dysfunction but someevidence shows that CKD can cause or exacerbate OSA and central sleep apneaProposed mechanisms for this causal relationship include uremic neuropathyaltered chemosensitivity and hypervolemia [] Accordingly renal replacement therapy and fluid removal [] may improve obstructive or central sleepapnea Conversely treatment of sleep apnea with PAP may improve renalfunction in those with borderline renal impairment []RLS is a common and debilitating symptom in patients with CKD occurringin up to of patients on hemodialysis compared with that in approximately of the general population [] Although RLS symptoms generally follow acircadian rhythmicity with increased symptoms occurring at night RLS symptoms can occur during the long periods of daytime inactivity during hemodialysis [] Treatment is primarily focused on ensuring adequate iron stores thenconsidering medical therapy as per routine care of RLSSleep disorders and infectious diseases have few specific associations In general acute infection is associated with mild encephalopathy that masquerades ashypersomnolence and fatigue Proinflammatory cytokines are implicated inthe development of these constitutional symptoms Some infections howeverdirectly affect regulatory centers of the sleepwake systemEncephalitis lethargica is a historical pandemic cause of hypersomnolence ofrenewed interest since this review is being written in the middle of the COVID pandemic Also known as Von Economo™s encephalitis it occurred inassociation with the Spanish flu pandemic of [] An estimated millionwere affected worldwide The most common subtype the somnolentophthalmoplegic form developed after flulike symptoms of fever and malaiseand consisted of subsequent ophthalmoplegia accompanied by long periods ofhypersomnia Despite the appearance of deep sleep patients could be easilyawoken and sometimes maintained memories of activities that had transpiredaround them while œasleep This state of acute akinetic psuedosomnulencecould be followed by the development of chronic postencephaliticparkinsonismThe pandemic associated with the severe acute respiratory syndrome coronavirus SARSCoV2 ie COVID19 occurring during the writing of thisreview features evolving literature The first reports centered on respiratorysymptoms Although the involvement of the nervous system now appearsprevalent [] sleep disorders have yet to be specifically reported Howeverthe psychological responses to social distancing change in schedules and otherfeatures of an active pandemic have caused a wave of anxiety and depressionwhich in turn have been associated with poor sleep quality For example a 0c Page of Curr Treat Options Neurol survey of Chinese health care workers showed prevalences of depressionat anxiety at and insomnia at []Postinfectious or postvaccination narcolepsy is rare but is important in developing overall hypotheses in the etiology of idiopathic narcolepsy In certainvaccinations in Europe for the H1N1 pandemic caused narcolepsy at a risk of in pediatric patients [] Fortunately the risk of postvaccinationnarcolepsy appeared confined to specific vaccine formulations The incidenthowever has led to ongoing research in the immunological etiology ofnarcolepsyAfrican trypanosomiasis or sleeping sickness remains important in the developing world It is a parasitic infection spread by the tsetse fly that is endemic insubSaharan Africa The first symptoms include fever headaches and lymphadenopathy Once the parasite enters the central nervous system disorderedfragmented sleep ensues often with inversion of the circadian sleepwake cycleThe World Health anization outlines treatment with a regimen of antiparasitic medications once symptoms have started []Nonalcoholic fatty liver disease NAFLD consists of idiopathic hepatic steatosiswith a prevalence of to of the general population with increasedfrequency in individuals with obesity or DM [] Given these coassociationsOSA is common Untreated OSA may exacerbate liver injury because of oxidative stress and systemic inflammation [] and is a risk in conversion fromNAFLD to liver fibrosis [] Trials with CPAP have shown inconsistent resultsin markers of liver injury following treatment of OSA []The symptoms of gastroesophageal reflux disease GERD worsen during sleepparticularly if sleep occurs soon after a meal [] The lower esophageal sphincter that normally prevents reflux may be compromised by the increase inthoracic pressure in the setting of the upper airway obstruction [] Patientswith symptoms of GERD should be screened for OSA and conversely interruption of sleep in absence of OSA may improve with treatment with a protonpump inhibitor PPI [] or by simply elevating the head of the bedInflammatory bowel disease IBD has bilateral interactions with sleep []Given the relationship between sleep deprivationfragmentation on cytokineregulation and immune dysfunction it is hypothesized that poor sleep qualityworsens overall symptoms of IBD [ ] Additionally the proinflammatorystate disrupts the circadian rhythm [] Subjective and objective measurementsof sleep quality and timing should be considered in patients with IBD particularly in those who have frequent inflammatory flares despite otherwise adequate management An algorithmic approach to sleep assessment in IBD patients has been proposed by Canakis et al []Systemic lupus erythematosus and rheumatoid arthritis serve as the prototypical diseases of this group of disorders with a prevalence of sleep disturbancesof greater than [] The mechanisms of sleep disturbances as well as thereciprocal relationship in the contribution of poor sleep to worse autoimmunestatus are thought to be similar to those described above with IBD [ ] Thespecific sleep disorders prevalent in this group are OSA and periodic limbGastrointestinal systemAutoimmune disorders 0cCurr Treat Options Neurol Page of Pulmonarymovement disorder PLMD both with greater than prevalence [ ]As seen above hypersomnolence and activitylimiting fatigue arise from specificsleep disorders pain and medication side effects well as the primary effects ofthe primary proinflammatory status [ ] Often treating the underlyingautoimmune disorder improves associated fatigue However if sleepiness persists then evaluating for a comorbid sleep disorder such as obstructive sleepapnea is indicatedOne syndrome with possible autoimmune origins is chronic fatigue syndromeSleep disturbances insomnia and unrefreshing sleep are common symptoms yetpatients rarely report relief despite appropriate identification and treatment ofcomorbid sleep disorders [] Cognitivebehavioral therapy CBT and gradedexercise therapy are commonly pursued treatment approaches []Obstructive lung diseases most commonly asthma chronic obstructive pulmonary disease COPD and less common disorders such as cystic fibrosisCF or bronchiolitis obliterans may affect nocturnal ventilation OSA andCOPD often overlap given shared body habitus and other mutual risk factorsestimates of comorbid OSA and COPD range from to [] Patients withsevere COPD treated with nocturnal noninvasive ventilation NIPPV a moreadvanced form of positive airway pressure experience an absolute risk reduction of of the risk of hospital readmission or death at months compared with those treated with standard care and without NIPPV [64cid129]Insomnia is another common complaint among patients with COPD Circadian bronchial constriction may cause nocturnal wheezing dyspnea or othersymptoms of asthma prompting the patient to awaken [] In addition thehyperadrenergic response to beta agonist inhalers used in treatment for acutedyspnea impairs sleep onset see Table The growing success in treatments for CF patients means that sleep disordersarising from their intrinsic obstructive lung disease are now coming to theattention of caregivers Many factors contribute to sleep disruption includingchronic cough frequent infections abdominal discomfort reflux frequentstools medication side effects and psychological disease [] In addition tosleep disruption patients with CF are susceptible to hypoventilation thatworsens with disease progression Use of NIPPV in highrisk patients withhypercapnia has been shown to improve physiologic parameters and at timescan positively impact symptoms particularly in patients who have severedisease while awaiting lung transplant []Restrictive lung diseases defined by a reduced total lung capacity includethose with parenchymal damage such as idiopathic fibrosis hypersensitivitypneumonitis or other interstitial pneumonias Alternatively lung parenchymais normal in restrictive diseases such as obesity hypoventilation syndromehemidiaphragm paresis or neuromuscular disorders muscular dystrophiesamyotrophic lateral sclerosis Restrictive lung disease patients as seen abovewith obstructive disease patients are susceptible to nocturnal hypoventilationsubsequent CO2 retention and compensatory sleep fragmentation Use ofNIPPV in patients with severe restrictive lung disease spanning obesityhypoventilation syndrome to muscular dystrophies and ALS has had positiveimpacts on survival and quality of life [ ] 0c Page of CardiacCurr Treat Options Neurol Over of patients with congestive heart failure CHF have comorbid OSAmainly on the basis of mutual risk factors of DM hypertension obesity andolder age [ ] In addition insomnia in those with CHF may arise from avariety of factors including diuretic medications and subsequent nocturiapositional heart failure symptoms increased adrenergic status or psychosocialfactors [] Treatments addressing comorbid OSA and insomnia improve sleepquality but demonstrate mixed results in terms of longterm cardiovascularoutcomes [ ]Patients with acute myocardial infarction AMI experience both acute andchronic sleep disorders Due to the circadian variability of adrenergic hormonesand cardiac and systemic vasculature [] the timings of AMI sudden cardiacdeath and arrhythmia occur with increased frequency at night [] Cardiacischemia may present a series of nocturnal symptoms including paroxysmaldyspnea chest pain agitation or insomnia Surviving patients are at risk forchronic sleep disorders such as insomnia and sleepdisordered breathing withor without the cooccurrence of anxiety or depression []Retrospective longitudinal data demonstrate that those with OSA and whoare adherent with CPAP experience improved cardiovascular morbidity andmortality over nonadherent patients [] However these findings have notbeen clearly supported by prospective randomized trials The Sleep ApneacardioVascular Endpoints Trial SAVE Trial has called into question the causallink between the treatment of OSA and cardiovascular outcomes With a meanfollowup of years those randomized to PAP experienced no significantimprovements in study endpoints of death from cardiovascular causes AMIstroke and hospitalization for unstable angina CHF or transient ischemicattack compared with controls [78cid129cid129] Because of possible insufficient CPAPuse and because of the lack of main indications for CPAP treatment such assevere sleepiness interpretation of the findings of this large trial remainscontroversial In practice these authors often pursue CPAP treatment for patients with OSA and cardiovascular risk factors even in the absence of sleepiness at least for a trial period to assess adherence to treatment and to determineif there are subjective and objective improvements to sleep qualityWith a prevalence range of “ OSA is common in patients with atrialfibrillation and other arrhythmias [] Accordingly the Sleep Heart HealthStudy showed a two to fivefold higher risk of arrhythmia in patients with severeOSA compared with that in controls [] Retrospective series show that inpatients with atrial fibrillation and untreated OSA the risk of atrial fibrillationrecurrence following cardioversion is compared with in patients whoare adherent to CPAP [] However a prospective randomized control trialcalled retrospective findings into question [] Similar in design to the SAVETrial patients with atrial fibrillation were randomized to CPAP versus usualtherapy from a cohort in which sleepiness was specifically excluded This smalltrial total assessed the primary outcome of time to arrhythmia recurrenceBoth arms had recurrence rates of Although the trial showed that CPAPitself provides no specific benefit to those with atrial fibrillation the outcomesfor treatment of those with both disorders remain unclearAlthough the above studies centered on associations between cardiac diseaseand OSA patients with CHF AMI and atrial fibrillation experience high rates of 0cCurr Treat Options Neurol Page of CancerCritical illnesscentral sleep apnea CSA as well exceeding in patients with mild symptomatic CHF as an example [] CheyneStokes respiration a cyclical form ofCSA results when circulatory impairment perturbs the normal responsivenessin respiratory control resulting in alterations in œthe loop gain in modulatingchanges in carbon dioxide and oxygen levels in the bloodstream [] analogous to overly aggressive adjustments to a thermostat in response to changingtemperature The presence of CSA has been considered a marker of increasedmortality in patients with CHF although aims to resolve the treatment of CSAwith CPAP or more advanced modalities have not clearly demonstrated animprovement in cardiovascular outcomes []Estimates of the prevalence of sleep disturbances across cancer patients range widelyfrom to [ ] Insomnia is the most common disorder with prevalencelevels ranging from to [ ] Patients with cancer who undergo PSGhave shorter total sleep times longer times in bed low sleep efficiency andproportionately less deep sleep than controls [] Insomnia in patients with canceris driven by a multitude of factors including preexisting socioeconomic andpsychiatric disorders fatigue age RLS pain and medication effects [ ]Treatment follows that for the general population Although sedativehypnoticsare most commonly prescribed no evidence exists for specific pharmacologicinterventions for sleep disturbances in this population [] Cognitivebehavioraltherapy is currently the recommended firstline treatment for chronic insomnia[] Because the rarity of trained psychologists makes finding a provider difficult insome circumstances the electronic delivery of cognitivebehavioral therapy hasbeen sought as an alternative to facetoface therapy [ ]The bilateral interactions between sleep and critical illness form a rapidlychanging area of investigation which is made particularly challenging giventhe difficulties in measuring sleep in critically ill patients [ ] Lack ofsleep”or its encephalopathic analog”may affect outcomes in critical illnessesFor example a lack of scorable REM sleep correlates with longer ventilatorweaning time compared with controls with intact REM [] Failure rates onnoninvasive ventilation are impacted by sleep continuity [] Delirium acommon neurobehavioral syndrome seen in upwards of of patients inthe ICU [ ] is associated with significantly worse outcomes i

high throughput methods in biological and biomedical fields acquire alarge number of molecular parameters or omics data by a single experimentcombining these omics data can significantly increase the capability for recoveringfinetuned structures or reducing the effects of experimental and biological noise indataresultsfhclust for identifying patient subgroups from different omics information eg geneexpression mirna expression methylation in particular hierarchical structures of patientdata are obtained in each omic or view and finally their topologies are merged byconsensus matrix one of the main aspects of this methodology is the use of ameasure of dissimilarity between sets of observations by using an appropriate metricfor each view a dendrogram is obtained by using a hierarchical clustering based on afuzzy equivalence relation with łukasiewicz valued fuzzy similarity finally a consensusmatrix that is a representative information of all dendrograms is formed by combiningmultiple hierarchical agglomerations by an approach based on transitive consensusmatrix construction several experiments and comparisons are made on real data egglioblastoma prostate cancer to assess the proposed approachs fuzzy logic allows us to introduce more flexible data agglomerationtechniques from the analysis of scientific literature it appears to be the first time that amodel based on fuzzy logic is used for the agglomeration of multiomic data theresults suggest that fhclust provides better prognostic value and clinical significancecompared to the analysis of singleomic data alone and it is very competitive withrespect to other techniques from literaturekeywords multiomics data data integration hierarchical clustering fuzzy similarityfuzzy aggregation the authors open access this is licensed under a creative commons attribution international licensewhich permits use sharing adaptation distribution and reproduction in any medium or format as long as you give appropriatecredit to the original authors and the source provide a link to the creative commons licence and indicate if changes weremade the images or other third party material in this are included in the ™s creative commons licence unlessindicated otherwise in a credit line to the material if material is not included in the ™s creative commons licence and yourintended use is not permitted by statutory regulation or exceeds the permitted use you will need to obtain permission directlyfrom the copyright holder to view a copy of this licence visit httpcreativecommonslicensesby40 the creativecommons public domain dedication waiver httpcreativecommonspublicdomainzero10 applies to the data madeavailable in this unless otherwise stated in a credit line to the data 0cciaramella bmc bioinformatics 21suppl page of nowadays high throughput methods in biological and biomedical fields acquire a largenumber of molecular parameters by a single experiment in particular such measuredparameters are collected in œomics datasets eg genomics transcriptomics methylomics among multiple measured parameters dna genome sequence rna expressionand dna methylation are representative instances for individually analysing suchdata several methodologies have been introduced in literature even though recentlya number of studies pointed out the best performance coming from the integration ofmultiomics data for instance analysing each omic or view in the machine learningjargon set separately fundamental patterns can be detected from data however somefinetuned structures such as cancer subtypes can be highlighted by both gene expression and dna methylation information so that multiomics analysis can reduce theeffects of experimental and biological noise in data from literature three kinds ofintegration methodologies emerge¢ early integration builds a single featurebased matrix by concatenating each omic¢ intermediate integration builds a joint representation of data given the views¢ late integration each omic is analysed separately and the solutions are integratedin general late integration methods and in particular clustering are preferred whenthe analysis combines continuous and discrete data together for a review of integrationapproaches and their comparisons the reader may refer to in this work a multiviewclustering methodology named fhclust is introduced see fig for its general schemafor identifying patient subgroups from different omics information or datasets eg geneexpression mirna expression methylation specifically for each omic dataset a fuzzybased hierarchical clustering approach is adopted and finally the results are mergedby consensus matrix the idea behind the proposed approach comes from observingthat a hierarchical clustering dendrogram can be associated with a łukasiewicz fuzzydataset ie view and applies a singleomic analysisfig proposed approach a data preparation b data normalization and feature selection c multiomicshierarchical agglomerations d data integration e clustering and visualization 0cciaramella bmc bioinformatics 21suppl page of similaritybased equivalence relation so that a consensus matrix that is the representative information of all dendrograms is derived by combining multiple hierarchicalagglomerations following an approach based on transitive consensus matrix constructionmethodscluster analysis or clustering is an unsupervised technique that aims at agglomerating aset of patterns in homogeneous groups or clusters [ ] hierarchical clustering hc isone of several different available techniques for clustering which seeks to build a hierarchyof clusters and it can be of two types namely agglomerative where each sample starts inits own cluster and pairs of clusters are merged as one moves up the hierarchy or divisivewhere all samples start in one cluster and splits are performed recursively as one movesdown the hierarchy thus hc aims at grouping similar objects into a cluster and werethe endpoint is a set of clusters where each cluster is distinct from each other and theobjects within each cluster are broadly similar to each other hc can be performed eitheron a distance matrix or raw data agglomerative hc starts by treating each observationas a separate cluster and it repeatedly executes the following two steps identifies thetwo clusters that are closest together and merges the two most similar clusters thisprocess continues until all the clusters are merged togetherthe main output of hc is a dendrogram which shows the hierarchical relationshipbetween the clusters distances many distance metrics have been developed and thechoice should be made based on theoretical concerns from the domain of studylater on it is necessary to determine how the distance is computed eg singlelinkagecompletelinkage averagelinkage as with distance metrics the choice of linkage criteria should be based on theoretical considerations from the application domainin nonfuzzy clustering or hard clustering data is divided into distinct clusters andeach data point can only belong to exactly one cluster in fuzzy clustering data pointscan potentially belong to multiple clusters for example in hard clustering given someparameters a œsymptom can be in a mutually exclusive way present or absent red orblue whereas in fuzzy clustering that œsymptom could simultaneously be of somegrade red and some other grade blue in fig a comparison between hard and fuzzycategorisation is shown the reader can refer to for a recent comparison betweenhard and fuzzy clustering in this work we introduce a data integration methodologybased on fuzzy concepts in particular we associate a dendrogram to a fuzzy equivalencerelation ie łukasiewicz valued fuzzy similarity so that a consensus matrix in a multiview clustering that is the representative information of all dendrograms can be obtainedfrom multiple hierarchical agglomerations [ ] the main steps of fuzzy agglomerationcan be summarised as follows characterisation of membership functions computation of a fuzzy similarity matrix or dendrogram for all models at a giventime construction of a consensus matrix for all hierarchical agglomerationsmembership functionswhen dealing with clustering tasks fuzzy logic fl permits to obtain a soft clusteringinstead of an hard clustering of data specifically data points can belong to more 0cciaramella bmc bioinformatics 21suppl page of fig hard vs fuzzy in symptom risk example a hard categorization b fuzzy categorizationthan one cluster simultaneously the fundamental concept in fl upon which all thesubsequent theory is constructed is the notion of fuzzy set a generalisation of a crisp setfrom classical set theorya fuzzy set generalises a crisp set by allowing its characteristic function ie itsmembership function assuming values in the interval [ ] rather than in the set in this way a given item belongs to the fuzzy set with a degree of truthranging from do not belong at all ie its membership function assumes value to completely belong ie the membership function assumes value in fl applications fuzzy sets make it possible to represent qualitative nonnumeric values ielinguistic variables such as high medium low for approximate reasoninginference or fuzzy control systems linguistic variables can be represented by fuzzy setsthrough a transformation step called fuzzification and it is achieved by using different types of membership functions representing the degree of truth to whicha given input sample belongs to a fuzzy set see œmembership functions sectionin supplementary material 0cciaramella bmc bioinformatics 21suppl page of fuzzy similarity matrixa measure of similarity or dissimilarity defines the resemblance between two samples orobjects similarity measure is a significant means for measuring uncertain informationfuzzy similarity measure is a measure that depicts the closeness among fuzzy sets and hasbeen used for dealing issues of pattern recognition and clustering analysisa binary fuzzy relation that is reflexive symmetric and transitive is known as a similarity relation fuzzy similarity relations are the generalisation of equivalence relationsin binary crisp relations to binary fuzzy relations in details a fuzzy similarity relationcan be considered to effectively group elements into crisp sets whose members are similar to each other to some specified grade and it is a generalization of classical equivalencerelation as described in œfuzzy similarity section in supplementary material in orderto introduce the fuzzy similarity in the following we focus on the properties of thełukasiewicz tnorm tl and the biresiduum in this way we obtain a fuzzy equivalencerelation that can be used for building dendrogram for more details in the derivation ofthese results see œfuzzy similarity section in supplementary materialdendrogram and consensus matrixif a similarity relation is mintransitive ie t min then it implies the existence ofthe dendrogram see œdendrogram and consensus matrix section in supplementarymaterial for details the mintransitive closure of a relation matrix r can be easilycomputed and the overall process is described in algorithm the last ingredient to accomplish an agglomerative clustering is a dissimilarity relationhere we considered the following result lemma letting r be a similarity relation with the elements rcid2x ycid3 ˆˆ[ ] and lettingd be a dissimilarity relation which is obtained from r bydx y ˆ’ rcid2x ycid3then d is ultrametric iif r is mintransitivein other words we have a onetoone correspondence between mintransitive similaritymatrices and dendrogram and between ultrametric dissimilarity matrices and dendrograms finally after the dendrograms have been obtained each time a consensus matrixie the representative information of all dendrograms is obtained by combining thetransitive closures ie maxmin operation the overall approach is described inalgorithm the overall workflow of the proposed approach is summarised in fig in particular for each omic data set xi a fuzzification step is adopted for obtaining thenew data set yi see supplementary material successively adopting a fuzzy similaritymeasure the similarity matrix si is computed and to guarantee the transitive closure ofthe matrix a new matrix ci is computed see algorithm finally all the ci matricesare collected for obtaining the consensus matrix a and the overall final dendrogram seealgorithm in fig we show an example that summarize a realistic agglomeration result we plotin figs 4abc three input hierarchies obtained on datasets that should be combinedin this case four sequences of patients are considered namely s1 s2 s3 and s4 respectively in fig 4d we show the final result by agglomerating dendrograms we observe that 0cciaramella bmc bioinformatics 21suppl page of fig workflow of the fuzzy based hierarchical clusteringthe output hierarchy contains clusters s1 s2 s3 and s1 s2 s3 s4 at different levels andeach of these clusters eg s1 s2 s3 are repeated at least in two out of the three inputdendrograms moreover it is worth stressing that the proposed approach based on theagglomeration of dendrograms can also be applied with commonly used metrics egeuclidean distance in fig we show a comparison between the dendrograms obtainedby using an euclidean metric and a similarity based approach ie łukasiewicz tnormrespectively in this realistic example we simulate three omic data sets with rows ienumber of patients and columns ie features we split the single datasets in twopartitions or clusters such that the first rows are random samples from a standard normal distribution with variance and the other rows have the same distribution withfig combination algorithm abc input dendrograms d combined hierarchy 0cciaramella bmc bioinformatics 21suppl page of fig crisp hierarchical clustering vs fuzzy based hierarchical clustering a dendrogram of euclidean basedhierchical clustering b dendrogram of similarity based hierachical clusteringalgorithm mintransitive closure input relation si output transitive relation ci sti elaborate compute sˆ— if sˆ—ielse ci sti si ˆª si —¦ sicid8 si replace si with sˆ—i sˆ—i and go to step i and the algorithm terminatesvariance obtaining a sort of an overlap we observe that both methods find two separated clusters but the similarity based approach in fig 5b permits to obtain a perfectseparation of the source partitionsresults and discussionin the following we describe the behaviour of the proposed methodology on multiomicsbenchmark datasetsalgorithm combination of dendrograms input ci ‰¤ i ‰¤ l l input similarity matrices dendrograms output similarity matrix dendrogram a aggregate the similarity matrices to a final similarity matrixa aggregate c1 c2 cla let aˆ— be the identity matrixb for each ci calculate e aˆ— aˆ— ˆª aˆ— —¦ cic if aˆ— is not changed a aˆ— and goto step else goto step 1b create the final dendrogram from aomics datasetswe consider multiomics cancer datasets available from the cancer genome atlastcga tcga is a large multiomic repository of data on thousands of cancerpatients all datasets contain three omics gene expression mirna expression and 0cciaramella bmc bioinformatics 21suppl page of table datasets description three omics are provided for each dataset respectively dna geneexpression mirna and methylationcases dnaoridatasetamlbiccoadgbmkirclihcluscskcmovsarclnrfmirnaorilnmethyorilnmultiomicsorilnrfrfrfthe number of features at each variable selection method is shown ori original variable dimension ln logarithm andnormalisation and rf random forest based on mean decrease gini indexdna methylation1 in table are summarised the main properties of the datasetsnamely acute myeloid leukemia aml breast invasive carcinoma bic colon adenocarcinoma coad glioblastoma multiforme gbm kidney renal clear cell carcinoma kirc liver hepatocellular carcinoma lihc lung squamous cell carcinomalusc skim cutaneous melanoma skcm ovarian serous cystadenocarcinoma ovsarcoma sarc the number of patients ranges from for aml to for bicmultiview clustering algorithmsfor validating the effectiveness of our model we compared it against several categories ofmultiview clustering algorithms2¢ kmeans and spectral clustering techniques ¢ lracluster it is a lowrank approximation based integrative probabilistic model¢ pins perturbation clustering for data integration and disease subtyping pinsis able to address subtype discovery as well as integration of multiple data types thealgorithm is built upon the resilience of patient connectivity and cluster ensembles toensure robustness against noise and biasto fast find the shared principal subspace across multiple data types¢ snf similarity network fusion snf allows for discovery of disease subtypesthrough integration of several types of highthroughput data on a genomic scale snfcreates a fused network of patients using a metric fusion technique and thenpartitions the data using spectral clustering snf appears to be the state of the art inthis area and has proven to be very powerful however the unstable nature ofkernelbased clustering makes the algorithm sensitive to small changes in molecularmeasurements or in its parameter settings¢ mcca multi canonical correlation analysis mcca which extends theapplication of canonical correlation analysis cca to more than two views is oneof the most widely used dimension reduction method for finding linear relationsbetween two or more multidimensional random variables1row data are available at httpacgtcstauacilmulti_omic_benchmarkdownloadhtml2httpsgithubcomshamirlabmultiomicscancerbenchmark 0cciaramella bmc bioinformatics 21suppl page of evaluation metricsin order to assess the performance of each method we adopt three evaluation metricsthat are the logrank test the enrichment of clinical labels in the clusters and the methods runtime the logrank test assumes that if clusters of patients have significantlydifferent survival they are different in a biologically meaningful way for the enrichmentof clinical labels in clusters six clinical labels are considered gender age at diagnosispathologic tumor pathologic metastases pathologic lymph nodes and pathologic stagethe four latter parameters are discrete pathological parameters measuring the progression of the tumor metastases and cancer in lymph nodes and the total progressionpathologic stage enrichment for discrete parameters was calculated using the χ2 testfor independence and for numeric parameters using kruskalwallis test not all clinicalparameters were available for all cancer types so a total of clinical parameters wereavailable for testing to derive a pvalue for the logrank test the χ2 test for independence the kruskalwallis test and the statistic for these three tests is assumed to have χ2distribution preprocessingtcga datasets were preprocessed as follows patients and features with more than missing values were removed and missing values were imputed using knearest neighborimputation the sequence features were logtransformed the features with highestvariance from geneexpression and methylation omics were selected in the mirna omicfeatures with zero variance were filtered all features were then normalized to have zeromean and standard deviation for methylation we selected the features with maximal variance in each dataset and also adopted the standard pipeline proposed in whose procedure filters out the probes from the x and y chromosomes or probes that areknown to have common snps at the cpg sitea further unsupervised variable selection step has been performed by using the meandecrease gini based on random forest the mean decrease in gini is the average of a variable total decrease in node impurity weighted by the proportion of samplesfig mean performance of the algorithms on ten multiomics cancer datasets the xaxis measures thedifferential survival between clusters mean log10 of logrank™s test pvalue and the yaxis is the meannumber of clinical parameters enriched in the clusters 0cciaramella bmc bioinformatics 21suppl page of fig performance of the algorithms on ten multiomics cancer datasets for each plot the xaxis measuresthe differential survival between clusterslog10 of logrank™s test pvalue and the yaxis is the number ofclinical parameters enriched in the clusters red vertical lines indicate the threshold for significantly differentsurvival pvalue cid2 reaching that node in each individual decision tree in the forest this is effectively a measure of how important a variable is for estimating the value of the target variable acrossall of the trees that make up the forest a higher mean decrease in gini indicates highervariable importance therefore the most important variables to the model is the highestin the plot with the largest mean decrease in gini values conversely the least importantvariable is the lowest in the plot with the smallest mean decrease in gini values by following this strategy we cutoff all those variables whose importance is zero the numberof variable cutoff at each step is summarised in table experimental resultsin the experiments for all methods the number of searched clusters is selected in therange [ ˆ’ ] to determine the number of clusters for a method we used the œelbowmethod to automatically pick out the optimal elbow rather than choose it manually weused as approximation the second derivative of a vector vv [i ] v [i ˆ’ ] ˆ’ 2v[ i] in particular we consider the index i that brings this expression to a maximum or minimum depending on whether v increases or decreases for all methods we adhered tothe guidelines for usage and parameter selection given by the developers in some casestable performance on ten multiomics number of clinical parameters enriched in the clustersfhclustcrisp hclustkmeansspectrallraclusterpinssnfmccaamlbiccoadgbmkirclihcluscskcmovsarcmeans 0cciaramella bmc bioinformatics 21suppl page of table performance on ten multiomics differential survival between clusters log10 of logrank™stest pvaluefhclustcrisp hclustkmeansspectrallraclusterpinssnfmccaamlbiccoadgbmkirclihcluscskcmovsarcmeanswhere no information was provided by the authors we devised parameter selection methods we performed the same process pipeline used in for evaluating the performanceof our method all methods were run on a multicore intelr xeonr cpu e52620v3 240ghz with gb ram in the following the obtained experimental results aredescribedfigure shows the average performance for multiomics data and for each singleomicseparately across all cancer types and fig shows the performance on the different cancer datasets all algorithms show quite similar performance in either differential survivalor enriched clinical parameters with respect to survival our fhclust method achievedthe overall best prognostic value sum of ˆ’log10 pvalues while pins and mcca ranked second and third respectivelyin table the differential survival between clusters mean ˆ’log10 of logrank™s testpvalue are reported spectral achieved the highest total number of significant clinical parameters with parameters fhclust along with lracluster and kmeansplaced themselves second with parameters snf achieved the third position with parameterswith respect to survival table fhclust outperformed its competitors achieving parameters mcca pins and snf have achieved good results with and enriched parameters respectivelywe also counted the number of datasets for which a method solution obtains significantly different survival these results are reported in table all methods that weredeveloped for multiomics data had at least four cancer types with significantly differentsurvival in this case fhclust and pins had different cancer subtypes for which itsclustering had significantly different prognosis fhclust spectral clustering and mccahad enrichment in cancer typeson average fhclust pins and mcca had better prognostic value but found lessenriched clinical labels as compared to spectral clustering methodtable for each benchmarked algorithm the number of cancer subtypes for which its clusteringhad significantly different prognosis first row and had at least one enriched clinical label secondrow are shownsignificant different survivalsignificant clinical enrichmentfhclustkmeansspectrallraclusterpinssnfmcca 0cciaramella bmc bioinformatics 21suppl page of table number of clusters chosen by the benchmarked algorithms on ten multiomics cancerdatasetsfhclustcrisp hclustkmeansspectrallraclusterpinssnfmccaamlbiccoadgbmkirclihcluscskcmovsarcmeansthe number of clusters found for each dataset are presented in table ranging from to because of the good methods performance in the previous analysis partitioning thedata into a relatively high number of clusters could indicate that clustering cancer patientsinto more clusters improves prognostic value and clinical significanceconcerning with methods computational burden their run times are reported intable fhclust takes on average seconds per dataset while spectral and snf gotlower timing the worst method takes roughly minutes per dataset see fig finally fig shows the benchmarked methods performance for singleomic datamoreover for each dataset and method the single omic that gave the best results forsurvival and clinical enrichment are also shown these results suggest that fhclust provides better prognostic value and clinical significance on multiomics data compared tothe analysis of singleomic data used separately nevertheless the interested reader mayrefer to the supplementary material for details on additional results concerning singleomics we also stress that the proposed method differently from other methods suchas snf does not need any hyperparameter tuning moreover clustering is embeddedin the data integration and vice versa and the use of fuzzy concepts ie tnormsfrom one hand permits to obtain a generalisation of the clustering approaches whereason the other hand gives the possibility to apply an inference system eg mamdanifor a quantitative and qualitative measure eg œhigh œmedium œlow in cancer riskassessmentsin this work we proposed a multiview clustering methodology for identifying patientsubgroups from different omics data in biological and biomedical fields combining theseomics data can significantly increase data mining capabilities one of the main aspects oftable runtime in seconds of the algorithms on ten multiomics cancer datasetsovfhclustcrisp hclustkmeansspectrallraclusterpinssnfmccacoadskcmbicamlgbmkirclihcluscsarcmeans 0cciaramella bmc bioinformatics 21suppl page of fig computational time comparisonsthis methodology is the use of a measure of dissimilarity between sets of observations byusing an appropriate metric and a consensus matrix that is a representative agglomerateinformation of all the dendrograms as emerged from the analysis of the scientific literature to the best of our knowledge our work concerns for the first time a model based onfuzzy logic used for the agglomeration of multiomic data the use of fuzzy logic allowsus to introduce more flexible data mining features also related to approximate reasoningseveral experiments and comparisons have been made on real data eg glioblastomaprostate cancer to assess the proposed methodology the results suggest that fhclustprovides better prognostic value and clinical significance compared to analysis of singleomic data alone fuzzy logic concepts and in particular membership functions permitsfig summarized performance of the algorithms across ten cancer datasets for each plot the xaxismeasures the total differential prognosis between clusters sum across all datasets of “log10 of logrank™s testpvalue and the yaxis is the total number of clinical parameters enriched in the clusters across all cancertypes a“c results for singleomic datasets d results when each method uses the single omic that achievesthe highest significance in survival e same with respect to enrichment of clinical labels 0cciaramella bmc bioinformatics 21suppl page of to develop a fuzzy inference model ie mamdani fuzzy cognitive maps for easilyobtaining a model for a quantitative and qualitative risk assessment of the cancer themodel based on approximate reasoning can be particularly useful for embedded devicesin future work it could be possible to improve results for multiomics analysis ina number of ways for instance more accurate feature selection algorithms couldbe adopted for improving the overall performance on one hand the integration oflabelled data could improve the feature selection step on the other hand some specific feature extraction strategies could be adopted indeed approaches based on thesignal analysis of gene expression data eg nonlinear principal component analysis compressive sensing could possibly further improve the performance [ ]in future it is possible to foresee a different weight for each omic data in order toobtain a more robust similarity and parametric similarity measures can be adoptedeg uninorm for generalizing the concept of and and or connections betweenclusterssupplementary informationsupplementary information accompanies this paper at httpsdoi101186s12859020035676additional file supplementary materialabbreviationsfhclust fuzzyhierarchical clustering dna deoxyribonucleic acid rna ribonucleic acid hierarchical clustering hccrisp hc crisp hierarchical clustering fl fuzzy logic tcga the cancer genome atlas aml acute myeloid leukemia bicbreast invasive carcinoma coad colon adenocarcinoma gbm glioblastoma multiforme kirc kidney renal clear cellcarcinoma lihc liver hepatocellular carcinoma lusc lung squamous cell carcinoma skcm skim cutaneousmelanoma ov ovarian serous cystadenocarcinoma sarc sarcoma pins perturbation clustering for data integrationand disease subtyping lracluster low rank approximation based multiomics data clustering snf similarity networkfusion mcca multi canonical correlation analysisabout this supplementthis has been published as part of bmc bioinformatics volume supplement proceedings from the 13thbioinformatics and computational biology international conference bbcc2018 the full contents of the supplement areavailable online at httpsbmcbioinformaticsbiomedcentralcomssupplementsvolume21supplement10authors™ contributionsac originally designed the methodology ac and dn worked on the developing of the method and the design of theexperiments ac dn and as contributed for interpreting and for analysing the results all authors contributed forwriting the manuscript read and approved the final manuscriptfundingpublication costs are funded by a grant from the dipartimento di scienze e tecnologie università degli studi di napoliœparthenope tecniche di machine learning e soft computing per l™elaborazione di dati multivariati softmulan piciaramellaavailability of data and materialscode and data of the proposed approach are available on multiomicscancerbenchmark github repositoryethics approval and consent to participateno ethics approval was required for the studyconsent for publicationnot applicablecompeting intereststhe authors declare that they have no competing interestsauthor details1dipartimento di scienze e tecnologie università degli studi di napoli œparthenope centro direzionale c4 island naples italy 2hitachi rail sts via argine naples italypublished august 0cciaramella bmc bioinformatics 21suppl page of referencescamastra f di taranto md staiano a statistical and computational methods for genetic diseases an overviewcomput math meth med 20152015 id “serra a fratello m fortino v raiconi g tagliaferri r greco d mvda a multiview genomic data integrationmethodology bmc bioinformatics “rappoport n shamir r multiomic and multiview clustering algorithms review and cancer benchmark nucleicacids res “reddy ck aggarwal cc data clustering boca raton chapman and hallcrc camastra f ciaramella a son lh riccio a staiano a fuzzy similaritybased hierarchical clustering for atmosphericpollutants prediction lncs “ciaramella a staiano a on the role of clustering and visualization techniques in gene microarray data algorithmsbora dj gupta ak int j emerg trends technol comput sci “napolitano f pinelli m raiconi g tagliaferri r ciaramella a staiano a miele g clustering and visualizationapproaches for human cell cycle gene expression data analysis int j approx reason “ciaramella a cocozza sand clustering of genomic data neural netw “iorio f miele g napolitano f pinelli m raiconi g tagliafer

" tumor microenvironment tme plays an important role in malignant tumors our study aimed toinvestigate the effect of the tme and related genes in osteosarcoma patientsmethods gene expression profiles and clinical data of osteosarcoma patients were downloaded from the targetdataset estimate algorithm was used to quantify the immune score then the association between immune scoreand prognosis was studied afterward a differential analysis was performed based on the high and lowimmunescores to determine tmerelated genes additionally cox analyses were performed to construct two prognosticsignatures for overall survival os and diseasefree survival dfs respectively two datasets obtained from the geodatabase were used to validate signaturesresults eightyfive patients were included in our research the survival analysis indicated that patients with higherimmune score have a favorable os and dfs moreover genes were determined as tmerelated genes theunsupervised clustering analysis revealed two clusters were significantly related to immune score and t cells cd4memory fraction in addition two signatures were generated based on three and two tmerelated genesrespectively both two signatures can significantly divide patients into low and highrisk groups and were validatedin two geo datasets afterward the risk score and metastatic status were identified as independent prognosticfactors for both os and dfs and two nomograms were generated the cindexes of os nomogram and dfsnomogram were and respectively tme was associated with the prognosis of osteosarcoma patients prognostic models based on tmerelated genes can effectively predict os and dfs of osteosarcoma patientskeywords tumor microenvironment osteosarcoma prognosis immune features nomogram osteosarcoma is the most common bone tumor especiallyin children and adolescents it was reported that approximately of patients are between and yearsold and osteosarcoma is considered as the second leadingcause of death in this age group currently surgery and correspondence 407404159qqcom4wenzhou medical university wenzhou chinafull list of author information is available at the end of the chemotherapy are still major treatments for osteosarcomapatients and these therapies are constantly improving inrecent years however due to the susceptibility of localaggressiveness and lung metastasis in osteosarcoma patients the prognosis of osteosarcoma remains unfavorable previous studies indicated that the 5years survivalrates were and in metastatic and nonmetastaticpatients respectively thereforeit is necessary to the authors open access this is licensed under a creative commons attribution international licensewhich permits use sharing adaptation distribution and reproduction in any medium or format as long as you giveappropriate credit to the original authors and the source provide a link to the creative commons licence and indicate ifchanges were made the images or other third party material in this are included in the 's creative commonslicence unless indicated otherwise in a credit line to the material if material is not included in the 's creative commonslicence and your intended use is not permitted by statutory regulation or exceeds the permitted use you will need to obtainpermission directly from the copyright holder to view a copy of this licence visit httpcreativecommonslicensesby40the creative commons public domain dedication waiver httpcreativecommonspublicdomainzero10 applies to thedata made available in this unless otherwise stated in a credit line to the data 0chu bmc cancer page of investigate the mechanism of pathogenesis and progressionof osteosarcoma and accurately classify the risk of patientsrecently an increasing number of diagnostic and prognostic biomarkers of osteosarcoma patients have beenidentified for example chen reported that tumorsuppressor p27 is a novel biomarker for the metastasis andsurvival status in osteosarcoma patients moreover huang discovered that dysregulated circrnas serve asprognostic and diagnostic biomarkers in osteosarcomapatients and the relative potential mechanism mainly attributes to the regulation of downstream signaling pathwaysby sponging microrna in addition lncrna microrna and many clinical data were also identified asprognostic biomarkers for osteosarcoma patients however osteosarcoma is one of the malignant cancers entitiescharacterized by the high level of heterogeneity in humanstherefore it is necessary to find accurate biomarkers forosteosarcomain recent years researchers have paid more and moreattention to the role of the tumor microenvironmenttme in malignant tumors the function of tme inthe tumorigenesis progression and therapy of tumorshave been initially understood [ ] more importantly estimation of stromal and immune cells in malignant tumor tissues using expression data estimate an algorithm to quantify the score of immune cellsand stromal cells by analyzing the gene expression datawas developed in based on the algorithm theprognostic value of immune and stromal cells in bladdercancer acute myeloid leukemia gastric cancer cervicalsquamous cell carcinoma adrenocortical carcinomaclear cell renal cell carcinoma hepatocellular carcinomathyroid cancer and cutaneous melanoma have beenreported [“] generally the above research indicatedthat tme can serve as the prognostic biomarker in tumorsand many tmerelated genes were determined as the prognostic genes however the role of tme and tmerelatedgenes in osteosarcoma patients remains unclearin the present study gene expression data and corresponding clinicopathologic data were obtained from thetherapeutically applicable research to generate effectivetreatments target dataset then the estimatealgorithm was performed to quantify the immune score ofosteosarcoma and the tmerelated genes were identifiedby the differential expression analysis subsequently theprognostic value of tme and tmerelated genes weredetermined by a series of bioinformatics methodsmethodsgene expression datasetslevel data of gene expression profiles and correspondingclinical data of osteosarcoma patients were downloadedfrom target dataset ocgcancergovprogramstarget accessed on oct the correspondingclinicopathologic data included in the present study wereage gender race ethnicity tumor site and metastaticstatus after data were extracted from the public domainthe estimate an algorithm inferring tumor puritystromal score and immune cell admixture from expression data was performed to evaluate the immune score byusing the estimate package in r software version meanwhile the messenger rnamrna expressionprofiles and clinical data ofincludinggse21257 and gse39055 were obtained fromthe gene expression omnibus as external validationcohortstwo cohortssurvival analysis and correlation analysisafter scores were obtained patients were divided intohighscore group and lowscore group according to themedian of the immune score the kaplanmeier survivalanalysis with logrank test was performed to estimatethe differences of overall survival os and diseasefreesurvival dfs between high and lowscore cohorts inaddition the association between clinicopathologic dataand tme score was also studied mannwhitney signedrank test was performed to compare the differences ofimmune score between each clinical group all statisticalanalyses in the present study were performed using rsoftware except for the special instructions p value twoside was identified as statistically significantin the present studydegexpressed genedifferentially expressed gene analysisdifferentiallyanalysis wasperformed by comparing the proteincoding genesexpression between the lowimmune score group andthe highimmune score group the limma package in rsoftware was used to perform the differential analysisand genes with log fc and adjusted pvalue qvalue were identified as degs to further understand the function of degs identifiedin the present study gene ontology goincludingbiological processes bp molecular functions mf andcellular componentscc and kyoto encyclopedia ofgenes and genomes kegg analysis were performedby clusterprofiler package in r software evaluation of association with immune cellsto further investigate the association between degs andimmune cells the cibersort package was used toestimate the relative proportions of types of immunecells meanwhile the œconsensusclusterplus package was used to cluster in an unbiased and unsupervisedmanner based on the overlapping degs cumulative distribution function cdf and relative change inarea under the cdf curve were used to determine theoptimal number of clusters k then mannwhitney 0chu bmc cancer page of signedrank test was performed to study the differenceof immune cells proportion between the clusters and theviolin plot was established to show the differences ofimmune cells among clusters survival analysis of degsbased on the degs the univariate cox analysis was performed to determine the prognostic value of immunerelated genes then the osrelated genes were validatedin the gse21257 dataset while the dfsrelated geneswere validated in the gse39055 dataset only genes successfully validated were selected for further analysis afterward based on the validated genes the multivariate coxanalysis was performed to establish the prognostic signature for predicting the prognosis of osteosarcoma patientsthe risk score for each patient was calculated based onthe coefficient from the multivariate cox analysis and thecorresponding gene expression meanwhile all patientswere divided into the high and lowrisk groups accordingto the median of the risk score the survival receiver operating characteristic roc curve was used to show the discrimination of signatures and the kaplanmeier survivalcurve with the logrank test was generated to show thedifferences of os and dfs between high and lowriskgroups in addition the risk score of patients in the validation cohort was also calculated according to the aforementioned risk signature the kaplanmeier survivalcurve and survival roc curve were generated to show thepredictive ability of the signature in the validation cohortdevelopment of a nomogram for osteosarcoma patientsnomogram is a tool to visualize the predictive model andconvenient for clinical practice therefore we attemptedto develop a nomogram based on the tmerelated genessignature and clinicopathologic data to predict the prognosis of osteosarcoma patients firstlythe univariatecox analysis was performed to filter prognostic variableswhich will be further included in the multivariate coxanalysis secondly based on independent prognostic variables two nomograms were established for predicting theos and dfs respectively the cindex was used to assessthe discriminatory performance of the nomogram whichrange from to a cindex of means agreement by chance and a cindex of represents perfectdiscriminatory performance the higher value of the cindex the better performance of the nomogram is furthermore the calibration curves of and 3year weredeveloped to evaluate the effectiveness of nomogramsresultsimmune significantly associated with the prognosis ofosteosarcoma patients osteosarcoma patients were included in the presentstudy including males and females the immunescore of the cohort range from ˆ’ to tostudy the relationship between the immune score and theprognosis of osteosarcoma patients patients wereincorporated into the lowimmune score group while theremaining patients were incorporated into the highimmune score group the survival analysis indicated thatpatients with higher immune score had a favorable osand dfs fig 1a and b after adjusted age tumor siteand metastatic status the immune score still was a prognostic variable for both os and dfsfig 1a and b inaddition the relationship between immune score and clinical features was also investigated however there was nosignificant relationship between immune score and clinicalvariables supplementary figure 1a1cdifferential expression analysisaccording to the median of the immune score patients were divided into highscore n and lowfig association between immune score and prognosis in osteosarcoma patients a kaplanmeier survival analysis of overall survival for patientswith high vs low immune score b kaplanmeier survival analysis of diseasefree survival for patients with high vs low immune score 0chu bmc cancer page of score group n there were differentiallyexpressed genes between two groups which include upregulated genes and downregulated genesfig 2a b and supplementary table to furtherunderstand the function of degs go analysisand kegg analysis were performed the top significant results of go analysis among three types wereillustrated in fig 2c interestingly we can find that theresults of go analysis are mostly associated with immunity which further verify that the immunerelated degsare associated with immune features in addition the results of kegg also confirmed it such as œphagosomeœautoimmune thyroid disease œantigen processing andpresentation œb cell receptor signaling pathway œintestinal immune network for iga production œinflammatorybowel disease œprimary immunodeficiency œth1 andth2 cell differentiation œth17 cell differentiation œnatural killer cell mediated cytotoxicity and œnfˆ’kappa bsignaling pathway fig 2dconsensusunsupervisedevaluation of degs and immune cellsto further understand the molecular heterogeneity ofosteosarcomaanalysis wasperformed to divide patients into subgroups to explorewhether immunerelated genes presented discernable patterns based on the consensus matrix heat map patientswere clearly divided into two clustersfig 3a in additionby comprehensively analyzing the relative change in areaunder the cumulative distribution function two clusterswere determined fig 3bc the immune score betweentwo clusters was significantly different fig 3d in additionthe proportion of types of immune cells in osteosarcomapatients was illustrated in a barplot fig 3e interestinglywe can see that the t cells cd4 memory activated ofcluster is significantly higher than cluster fig 5fprognostic value of tmerelated genesprevious studies indicated that tmerelated genes canserve as the prognostic biomarker for tumor patientsfig differentially expressed genes with the immune score in osteosarcoma patients a heatmap of significantly differentially expressed genesbased on immune score b the volcano figure to show the upregulated and downregulated genes c go analysis of differentially expressedgenes d kegg of differentially expressed genes go gene ontology kegg kyoto encyclopedia of genes and genomes 0chu bmc cancer page of fig the immune landscape of the tumor microenvironment ac unsupervised clustering of all samples based on the overlapping degs dcomparison of immune score between two clusters e the distribution of types of immune cells in osteosarcoma patients f the comparisonof types of immune cells between clusters deg differentially expressed genehence we performed the univariate cox analysis toidentify prognostic degs the results showed that and genes were identified as os and dfsrelateddegs respectively supplementary table and afterward five osrelated genes were successfully validated inthe gse21257 data set and five dfsrelated genes were successfully validated in the gse39055 cohort furthermoremultivariate cox analysis was performed and two prognostic signatures were generated for predicting the os anddfs respectively the risk score for predicting the os wasasrisk score fcgr2b0766 gfap0702 mpp70387 in addition the risk score for predicting thedfs was as follows risk score cyp2s10574 icam3 the auc values of osrelated signature were follows 0chu bmc cancer page of and in and 3year respectively fig 4aand the auc values of dfsrelated signature were and in and 3year respectively fig 5amoreover survival curves showed that patients in the highrisk group had worse os and dfs compared with the lowrisk patients figs 4b and 5b heat maps risk score plotsand survival status were generated to show the distinctionbetween highrisk patients and lowrisk patients figs 4ceand 5ce then both signatures were validated in independent cohorts for os signature the auc values ofvalidation cohort were and at and3year fig 4f for dfs signature the auc values ofvalidation cohort were and at and3year fig 5f additionallyin both validation cohortssurvival curves showed that lowrisk patients were favorableprognosis than highrisk patients figs 4g and 5gheat maps risk score plots and survival status of validation cohorts were also generated to show the distinction between highrisk patients and lowrisk patientsfigs 4hj and f 5hjdevelopment of a nomogram for osteosarcoma patientsto generate a nomogram for clinical use the cox analysiswas performed to select the clinical prognostic variables infig establishment and validation of the prognostic model for overall survival based on significant degs a receiver operating characteristiccurves of prognostic signature in the training cohort b the survival curve showed the different overall survival status between high and lowriskpatients c the heat map showed the expression of prognostic genes in the training cohort d the risk curve of each sample reordered by riskscore e the scatter plot showed the overall survival status of osteosarcoma patients in the training cohort f receiver operating characteristiccurves of prognostic signature in validation cohort g the survival curve showed the different overall survival status between high and lowriskpatients h the heat map showed the expression of prognostic genes in the validation cohort i the risk curve of each sample reordered by riskscore j the scatter plot showed the overall survival status of osteosarcoma patients in the validation cohort 0chu bmc cancer page of fig establishment and validation of the prognostic model for diseasefree survival based on significant degs a receiver operatingcharacteristic curves of prognostic signature in the training cohort b the survival curve showed the different diseasefree status between highand lowrisk patients c the heat map showed the expression of prognostic genes in the training cohort d the risk curve of each samplereordered by risk score e the scatter plot showed the diseasefree status of osteosarcoma patients in the training cohort f receiver operatingcharacteristic curves of prognostic signature in validation cohort g the survival curve showed the different diseasefree status between high andlowrisk patients h the heat map showed the expression of prognostic genes in the validation cohort i the risk curve of each sample reorderedby risk score j the scatter plot showed the diseasefree status of osteosarcoma patients in the validation cohortthe univariate cox analysis risk score and metastatic status were identified as both os and dfsrelated variablesfig 6a and e afterward risk score and metastatic statuswere determined as both independent os and dfsrelated variables in the multivariate cox analysis fig 6band f based on independent variables two nomogramswere established for predicting the os and dfs in osteosarcoma patients respectively fig 6c and g the cindexvalues were and in os nomogram and dfsnomogram respectively the results of cindex mean thatboth two nomograms have good discrimination meanwhile to evaluate the calibration of nomograms six calibration curves were generated and the results showed thatthe predictive curves were close to the ideal curve fig 6dand h which indicated a good calibrationdiscussionthe relationship between tme and tumor have beenwidely studied in recent years in the present study estimate algorithm was utilized to quantify the immunescore based on gene expression profiles in osteosarcomapatients from target database we confirmed that thetme is significantly associated with the prognosis ofosteosarcoma patientsinadditionfunctional enrichment analyses of tmerelated genes indicated that immunerelated processesincluding os and dfs 0chu bmc cancer page of fig nomograms based on the tumor microenvironment related genes for osteosarcoma patients a univariate cox analysis of overall survivalrelated variables b multivariate cox analysis of overall survivalrelated variables c nomogram for predicting the overall survival in osteosarcomapatients d1 and 3year calibration curves of overall survival nomogram e univariate cox analysis of diseasefree survivalrelated variables fmultivariate cox analysis of diseasefree survivalrelated variables g nomogram for predicting the diseasefree survival in osteosarcoma patientsh1 and 3year calibration curves of diseasefree survival nomogramknown to contribute to tumor progression more importantly degs based on the tme were identified asimportant prognostic biomarkers for osteosarcoma patients and two nomograms were developed for predicting the os and dfs of osteosarcoma patientsrespectivelyin recent years an increasing number of studiesfocused on the carcinogenesis and progression of tumorsbased on the tme and the estimate algorithm is oneof the most important quantitative tools for this researchfield based on the estimate algorithm the association between the prognosis and tme has been initially 0chu bmc cancer page of elucidated in some tumors such as cervical squamouscell carcinoma gastric cancer cutaneous melanomaacute myeloid leukemia bladder cancer and clear cellrenal carcinoma [ “] however previousstudies indicated that tme scores serve as a differentrole in different tumors for example for hepatocellularcarcinoma gastric cancer acute myeloid leukemiabladder cancer and clear cell renal carcinoma patientswith high immune score have a worse prognosis [ “] however for cervical squamous cell carcinoma adrenocortical carcinoma and cutaneous melanoma patients with high immune score have a favorableprognosis [ ] therefore we can find great heterogeneity among different tumors from the perspectiveof tme for osteosarcoma patients the present studyindicated that patients with higher immune score had abetter os and dfs hence the present study indicatedthat immune cells infiltrating tumor tissue may play animportant role in suppressing tumor progressionin our research tmerelated genes were identified by comparing the highscore and lowscore osteosarcoma patients the functional enrichment includinggo and kegg analyses showed that tmerelated geneswere mainly involved in the immune features such asregulation of leukocyte activation mhc protein complex mhc protein and complex binding more importantly the unsupervised cluster analysis based on degswas performed and all patients were divided into twoclusters immune score and t cell cd4 memory activated fraction were significant difference between twoclusters which further elucidated the relationship between degs and immune featuresdue to the poor prognosis of osteosarcoma patientsidentifying robust prognostic biomarker is very importantthe tumor immune microenvironment is closely relatedto the prognosis of bone tumor patients emilie etal performed the first genomewide study to describe therole of immune cells in osteosarcoma and found thattumorassociated macrophages are associated with reduced metastasis and improved survivalin highgradeosteosarcoma recently the prognostic signature based ontmerelated genes have been established for many tumors [ ] but only one study focused on osteosarcoma patients compared with the study performedby zhang we think that our research have someadvantages firstly our signatures were established basedon several validated genes and both two signatures weresuccessfully validated in independent cohorts secondlythe outcome of dfs was not reported in the previousstudy as reported in published studies tumor recurrenceis a terrible medical problem for osteosarcoma patientsand the 5year survival rate for osteosarcoma patients withmetastasis or relapse remains disappointing [ ]hence the dfs nomogram can improve the managementof osteosarcoma patients finally two nomograms incorporated tmerelated signature and clinical variables wereestablished in our research which further facilitated theclinical application of our findingsin our research five genes were incorporated into thefinal prognostic signatures fcgr2b gfap and mpp7were identified and validated as osrelated biomarkerswhile cyp2s1 and icam3 were dfsrelated biomarkersthe role of these genes in tumor prognosis had beenwidely reported in previous studies [“] fcgr2bhas been confirmed as an immunerelated gene previously although the relationship between fcgr2band prognosis in sarcoma patients had not been reported the prognostic value of fcgr2b had been widelyconfirmed in other cancerssuch as hepatocellularcarcinoma and glioblastoma [ ] in addition newm etal demonstrated that mpp7 is novel regulatorsof autophagy which was thought to be responsible forthe prognosis of pancreatic ductal adenocarcinomacyp2s1 described as cytochrome p450 family subfamily s member was reported significantly associatedwith colorectal cancer in primary colorectal cancercyp2s1 was present at a significantly higher level ofintensity compared with normal colon more importantly the presence of strong cyp2s1 immunoreactivity was associated with poor prognosis the roleof icam3 in cancer was also widely reported in published studies and the akt pathway plays an importantrole in the impact of icam3 on tumors yg kim etal reported that icam3 can induce the proliferationof cancer cells through the pi3kakt pathway additionally jk park etal showed that the icam3 can enhancethe migratory and invasive potential of human nonsmall celllung cancer cells by inducing mmp2 andmmp9 via akt pathway showed that the icam3can enhance the migratory and invasive potential ofhuman nonsmall celllung cancer cells by inducingmmp2 and mmp9 via akt pathwayalthough the role of tme and tmerelated genes inosteosarcoma patients have been initially studied by bioinformatic and statistical analyses in our research somelimitations should be elucidated firstly the treatmentinformation cannot be obtained from the target database which may influence the prognosis of osteosarcomapatients secondly two nomograms were generated andshowed good performance in our study however externalvalidation by a large cohort is needed thirdly many independent prognostic genes for osteosarcoma patients wereidentified in the present study but the potential mechanism to influence osteosarcoma remains unclear finally inthe training cohort and degs were identified asos and dfsrelated degs respectively however onlyfive os and five dfsrelated genes were identified in thevalidation cohort the different age structures smaller 0chu bmc cancer page of sample sizes and the platform covering only part of thegenes may contribute to this resultreceived february accepted july in tme plays an important role in osteosarcoma patients and related with the progression of thetumor moreover tmerelated genes can serve as prognostic biomarkers in osteosarcoma patients howeverfurther researches are needed to study the potentialmechanism and validate the nomogram that developedin our present studysupplementary informationsupplementary information accompanies this paper at doi101186s12885020072162additional file additional file additional file additional file abbreviationstme tumor microenvironment deg differentially expressed genesos overall survival dfs diseasesfree survival roc receiver characteristiccurve estimate estimation of stromal and immune cells in malignanttumor tissues using expression data target therapeutically applicableresearch to generate effective treatments go gene ontology bp biologicalprocesses mf molecular functions cc cellular components kegg kyotoencyclopedia of genes and genomes cdf cumulative distribution functionacknowledgementsnoneauthors™ contributionsc h l y sq t c l and yh w conceived of and designed the study c h r sand c l performed literature search r s l y and b c generated the figuresand tables l y hl r x y and jy l analyzed the data c h wrote themanuscript and sq t and l y critically reviewed the manuscript l ysupervised the research all authors have read and approved the manuscriptfundingwe received no external funding for this studyavailability of data and materialsthe data of this study are from target and geo databaseethics approval and consent to participatethe research didn™t involve animal experiments and human specimens noethics related issuesconsent for publicationnot applicablecompeting intereststhe authors declare that they have no competing interestsauthor details1department of joint surgery the affiliated hospital of qingdao universityqingdao china 2department of medical oncology the first hospital ofchina medical university shenyang china 3department of nursing sir runrun shaw hospital affiliated to zhejiang university hangzhou china4wenzhou medical university wenzhou chinareferencesjaffe n bruland os bielack s pediatric and adolescent osteosarcoma vol new york springer science business media vander rg osteosarcoma and its variants orthopedic clin north am “biermann js adkins d benjamin r brigman b chow w conrad eu 3rdfrassica d frassica fj gee s healey jh bone cancer j natl comprcancer netw “simpson s dunning md de brot s grauroma l mongan np rutland cscomparative review of human and canine osteosarcoma morphologyepidemiology prognosis treatment and genetics acta vet scand chen x cates jm du yc jain a jung sy li xn hicks jm man tkmislocalized cytoplasmic p27 activates pak1mediated metastasis and is aprognostic factor in osteosarcoma mol oncol “huang x yang w zhang z shao z dysregulated circrnas serve as prognosticand diagnostic markers in osteosarcoma by sponging microrna to regulatethe downstream signaling pathway j cell biochem “liu m yang p mao g deng j peng g ning x yang h sun h long noncoding rna malat1 as a valuable biomarker for prognosis in osteosarcoma asystematic review and metaanalysis int j surg “xu k xiong w zhao s wang b microrna106b serves as a prognosticbiomarker and is associated with cell proliferation migration and invasionin osteosarcoma oncol lett “zheng w huang y chen h wang n xiao w liang y jiang x su w wens nomogram application to predict overall and cancerspecific survival inosteosarcoma cancer manag res kahlert c kalluri r exosomes in tumor microenvironment influence cancerprogression and metastasis j mol med “ binnewies m roberts ew kersten k chan v fearon df merad m coussenslm gabrilovich di ostrandrosenberg s hedrick cc understanding thetumor immune microenvironment time for effective therapy nat med“ yoshihara k shahmoradgoli m martínez e vegesna r kim h torresgarcia wtreviño v shen h laird pw levine da inferring tumour purity and stromaland immune cell admixture from expression data nat commun yang s liu t nan h wang y chen h zhang x zhang y shen b qian pxu s comprehensive analysis of prognostic immunerelated genes inthe tumor microenvironment of cutaneous melanoma j cell physiol “ deng z wang j xu b jin z wu g zeng j peng m guo y wen z miningtcga database for tumor microenvironmentrelated genes of prognosticvalue in hepatocellular carcinoma biomed res int zhao k yang h kang h wu a identification of key genes in thyroid cancermicroenvironment med sci monit xu wh xu y wang j wan fn wang hk cao dl shi gh qu yyzhang hl ye dw prognostic value and immune infiltration of novelsignatures in clear cell renal cell carcinoma microenvironment agingalbany ny chen b chen w jin j wang x cao y he y data mining of prognosticmicroenvironmentrelated genes in clear cell renal cell carcinoma a studywith tcga database dis markers li x gao y xu z zhang z zheng y qi f identification of prognostic genesin adrenocortical carcinoma microenvironment based on bioinformaticmethods cancer med “ pan xb lu y huang jl long y yao ds prognostic genes in the tumormicroenvironment in cervical squamous cell carcinoma aging albany ny wang h wu x chen y stromalimmune scorebased gene signature aprognosis stratification tool in gastric cancer front oncol huang s zhang b fan w zhao q yang l xin w fu d identification ofprognostic genes in the acute myeloid leukemia microenvironment agingalbany ny yan h qu j cao w liu y zheng g zhang e cai z identification ofprognostic genes in the acute myeloid leukemia immunemicroenvironment based on tcga data a

" gastric neoplasms containing neuroendocrine carcinoma nec components are rare malignancieswith highly aggressive behavior and a poor prognosis and include pure nec and mixed tumors containing neccomponents we aimed to investigate whether there is a distinct difference in overall survival os between gastricneoplasms containing nec components and gastric adenocarcinomamethods surgically resected gastric neoplasms containing nec components n and gastricadenocarcinomas n from january to december at peking university cancer hospital wereretrospectively analysed patients were categorized into a surgical group and a neoadjuvant group and adjustedusing propensity score matching in the two groups gastric neoplasms containing nec components were dividedinto pure nec and mixed tumors with less than ghminen between and ghminen andmore than ghminen neuroendocrine carcinoma components os was compared between thesegroups and the gastric adenocarcinoma groupresults the os of gastric neoplasms containing neuroendocrine nec components was poorer than that of gastricadenocarcinomas in the surgical group regardless of whether the percentage of neuroendocrine cancercomponents was less than between and more than or cox multivariable regressionanalysis suggested that tumor category neoplasms containing nec components or gastric adenocarcinoma wasan independent risk factor for prognosis interestingly among patients receiving neoadjuvant therapy thedifference was not significantcontinued on next page correspondence buzhaodecjcrcn jijiafuhscpkueducn jiahui chen anqiang wang and ke ji contributed equally to this workdepartment of gastrointestinal surgery key laboratory of carcinogenesisand translational research ministry of education peking university cancerhospital institute no fucheng road haidian district beijing china the authors open access this is licensed under a creative commons attribution international licensewhich permits use sharing adaptation distribution and reproduction in any medium or format as long as you giveappropriate credit to the original authors and the source provide a link to the creative commons licence and indicate ifchanges were made the images or other third party material in this are included in the 's creative commonslicence unless indicated otherwise in a credit line to the material if material is not included in the 's creative commonslicence and your intended use is not permitted by statutory regulation or exceeds the permitted use you will need to obtainpermission directly from the copyright holder to view a copy of this licence visit httpcreativecommonslicensesby40the creative commons public domain dedication waiver httpcreativecommonspublicdomainzero10 applies to thedata made available in this unless otherwise stated in a credit line to the data 0cchen bmc cancer page of continued from previous pages gastric neoplasms containing any proportion of nec components had poorer overall survival thangastric adenocarcinoma in patients treated with surgery directly indicating that these neoplasms are moremalignant than gastric adenocarcinoma among the patients receiving neoadjuvant therapy the difference inoverall survival was not significant which was in sharp contrast with the results of the surgery group suggestingthat neoadjuvant therapy may have a good effect in the treatment of these neoplasmskeywords neuroendocrine carcinoma gastric adenocarcinoma overall survival gastric neoplasms containing neuroendocrine carcinomanec components are a heterogeneous subgroup ofgastric cancer with highly aggressive behavior and poorprognosis and include pure necs and mixed tumorscontaining nec components every yearthere areapproximately million new cases of gastric cancerworldwide and gastric neoplasms containing nec components account for approximately “ of thesecases [ ] given the low incidence there is little comprehensive basic and clinical research to systematicallyguide the treatment of these gastric neoplasms makingthe prognosis of these tumors unsatisfactory [“]according to the world health anizationwho digestive neuroendocrine tumor classificationneuroendocrine neoplasm nen can be divided intothree categories based on ki67 levels and mitotic counts— hpf grade g1 ki67 ‰ mitoses grade g2 ki67 ‰ ‰ mitoses‰ grade g3ki67 mitoses meanwhile the americanjoint committee on cancer ajcc defines highly differentiated nen as a neuroendocrine tumor net and thepoorly differentiated nen as a neuroendocrine carcinoma nec based on the degree of tumor cell differentiation generally g1 g2 and rare welldifferentiated g3nens belong to the nets while poorly differentiatedg3 nens belong to necs[ ] gastric mixedneuroendocrinenonneuroendocrineneoplasm gminen is a special type of gastric nen that is definedas containing more than of both neuroendocrineand nonneuroendocrine components accountingfor approximately of all gnens and of gastricneuroendocrine carcinomas gnecs [“] for thosemixed tumors with less than or more than neuroendocrine carcinoma components there is no uniform definition consideringthe heterogeneity ofminen and the malignancy degree of the different components in the tumor la rosa [ ] proposeddividing minen into three categories highgradeintermediategrade and lowgrade highgrade minenconsists of nec and carcinomaadenoma intermediategrade mimen consists of net and carcinoma and lowgrade minen consists of net and adenoma thereforein this study gastric highgrade mixed neuroendocrinenonneuroendocrine neoplasm ghminen was defined as gastric cancer containing more than of bothneuroendocrineadenocarcinomacomponentscarcinomaandgenerally the prognosis of mixed tumors is largely determined by the most malignant component kim found that gnec has shorter progressionfree survival pfs than gastric adenocarcinoma huang found that the prognosis of patients with more than of neuroendocrine cancer components is significantly poorer than that of patients with less than components all of these studies provide evidence thattumors containing neuroendocrine cancer componentsmay contribute to a worse prognosis therefore wehypothesized that a mixed tumor containing neuroendocrine carcinoma components would have a worse prognosis than pure adenocarcinoma alone we sought tofind studies on the overall survival os comparison between ghminen and gastric adenocarcinoma butfailed thus we think that a study of the comparison ofthe os of ghminen and gastric adenocarcinoma willprovide a valuable supplement to current research on ghminen to overcome the bias caused by the differences between the covariates in the comparison we usedpropensity score matching psm to match importantfactors such as age gender tumor location tumor sizepathological staging and adjuvant chemotherapy between the two groups making the research results morereliablemethodspatient selectionwe retrospectively collected patients diagnosed withgastric nens and underwent radical resection at pekinguniversity cancer hospital beijing from january to december the inclusion criteria were as follows pathologically confirmed pure nec or tumorcontaining nec components no other tumors werediagnosed before the operation complete clinicopathological information and survival information thatcould be obtained through followup patients diagnosedwith cm1 or ct4b before surgery or died from perioperative complications were excluded from the study 0cchen bmc cancer page of patients with gastric adenocarcinomas undergoing radical surgery were randomly selected for psm analysesperformed the chisquared test and mannwhitney utest were used to further verify the matching resultsfollowupwe followed the patients at least twice a year serumtumor markers test gastroscope and computed tomography ct scans were used to reexamine patients aftersurgery depending on the patients™ status magneticresonance imaging mri and positron emission tomography computed tomography petct were alsoconsidered for patients who cannot regularly visit ourcenter for postoperative examination we use telephonefollowup to obtain survival informationdiagnosis and classificationwe reevaluated the diagnosis and classification of ghminen mixed tumors with less than or morethan neuroendocrine carcinoma components werealso included in this study which were defined as ghminen and ghminenrespectively atumor consisting of nec is defined as pure necall neuroendocrine tumors were identified diagnosedand classified by two independent pathologists in accordance with the who classification of tumors neuroendocrine components were identified byhistological features and immunohistochemical specificity marks such as synaptophysin syn chromogranina cga and neuro cell adhesion molecule cd56 orncam the tumor staging described in the study wasbased on the ajcc 8th edition tnm staging guidelines all possible disagreements were discussed in ourstudy groupdefinition of variables and groupsin this study patients were divided into a surgical groupand a neoadjuvant group based on whether they had received neoadjuvant therapy before surgery patients inthe surgery group were assessed by the ptnm stagingsystem while patients in the neoadjuvanttreatmentgroup were assessed by the yptnm staging system osrefers to the time from surgery to the last followup thetime of death or the end ofloss offollowup or other cause of deathfollowup egpropensity score matchingto accurately compare the prognosis of ghminenand gastric adenocarcinoma we employed psm to balance the differences between the two groups psm wasperformed through the pamatching plugin in spss software logistic regression models were used toestimate propensity scores based on gender age tumorlocation tumor size and pathological staging given a caliper width nearest neighbor matching wasstatistical analysisall statistical analyses were performed using spss statisticalsoftware ibm united states the chisquared test and mannwhitney u test were used forstatistical analysis of categorical variables and continuous variables respectively kaplanmeier method wasused for the comparison of os the logrank test wasused to compare survival rates multivariable cox proportional hazards models were used to identify predictors of survival outcome p was regarded as thethreshold of significanceresultspatient selection and psm resultsbetween and among the patients treated atthe gastrointestinal cancer center of peking universitycancer hospital a total of patients with gastric neoplasms containing nec components met the inclusioncriteria for the study including cases of pure necand cases of mixedtype of these patients a total of patients received neoadjuvant therapy nec ghminen ghminen ghminen while the remaining patients receivedsurgery directly nec ghminen ghminen ghminen there were aninsufficient number of patients in group ghminen group to conduct effective statistical analysisso we combined the ghminen group with thenec group for further analysis we also randomly selected patients with gastric adenocarcinoma whounderwent radical surgery among them patientsreceived neoadjuvant therapy and the remaining patients were treated with surgery directly fig immunohistochemical specificity markers were utilizedto identify the neuroendocrine components fig 2asyn was expressed in almost all neoplasms containingnec components while the positive rates ofcga and cd56 were much lower and respectively no significant difference in the positiverate of syn and cga was observed between pure nec ghminen ghminen and ghminenfig 2b c only the positive rate of cd56 was found tobe higher in the pure nec group than that in the ghminen group fig 2dtherefore priorto os comparison psm wasperformed to ensure that there were no significant differences in patient gender age tumor location tumorsize pathological staging and adjuvant chemotherapybetween the two groups 0cchen bmc cancer page of fig flow chart of patient enrolmentcomparison of os between all patients with neccomponents and patients with gastric adenocarcinoma inthe surgical group and neoadjuvant groupbefore psm we compared the survival curves between all patients with nec components and patientswith gastric adenocarcinoma by the kaplanmeiermethod fig apparently patients with nec components had a poorer os than those with gastricadenocarcinoma fig 3a p in the surgicalgroup in contrast no significant difference was observed between the patientsreceiving neoadjuvanttherapy fig 3b p according to the proportion of nec components patients were classifiedinto pure nec ghminen ghminenand ghminen the os was also comparedbetween patients with adenocarcinomaand thesegroups and the results were similar to the overallcomparison fig 3c dfig illustrations of immunohistochemical staining patterns in gastric neoplasms containing nec components a an overview of the expressionof syn cga and cd56 in tumors containing nec components b syn expression in different nec component groups c cga expression indifferent nec component groups d cd56 expression in different nec component groups cd56 neuro cell adhesion molecule cgachromogranin a nec neuroendocrine carcinoma syn synaptophysin pvalue 0cchen bmc cancer page of fig see legend on next page 0cchen bmc cancer page of see figure on previous pagefig comparison of os between gastric neoplasms containing nec components and gastric adenocarcinoma a os comparison betweengastric neoplasms containing nec components and gastric adenocarcinoma before psm in the surgical group b os comparison between gastricneoplasms containing nec components and gastric adenocarcinoma before psm in the neoadjuvant group c os comparison between differentnec content groups pure nec ghminen ghminen and ghminen and gastric adenocarcinoma before psm in the surgicalgroup d os comparison between the different nec content groups and gastric adenocarcinoma before psm in the neoadjuvant group e oscomparison for patients in the surgical group after psm f os comparison for patients in the neoadjuvant group after psm nec neuroendocrinecarcinoma os overall survival psm propensity score matchingbefore psm significant differences between the baseline characteristics were observed in the surgical groupand the neoadjuvant group table table to balance the clinicopathological differences between the twogroups psm was performed to ensure that there wereno significant differences in patient gender age tumorlocation tumor size pathological staging and adjuvantchemotherapy between the two groups the detailedclinicopathological characteristics before and after psmare shown in table and table as a result patients with nec components and patients with gastric adenocarcinoma were matchedin the surgical group table patients with nec components also had a poorer os than those with gastricadenocarcinoma fig 3e p multivariable analysis showed that adjuvant therapy tumor category andtnm stage werefactorstable independent prognosticto investigate whether neoadjuvant therapy had an effect on os patients with nec components and patients with gastric adenocarcinoma were matched inthe neoadjuvant group table interestingly kaplanmeier analysis showed that among patients receivingneoadjuvant therapy there was still no significant difference in os between the two groups fig 3f p comparison of os between patients with differentproportions of nec components and patients with gastricadenocarcinomato investigate whether the level of nec componentshad an effect on os in the surgical group ghminen ghminen pure nec and pure nec plus ghminen were compared with gastric adenocarcinoma after psm the results showed that even thegroup with the lowest proportion of nec componentsthe ghminen group had a poorer os thanadenocarcinoma fig 4a p as expected theghminen pure nec and pure nec plus ghminen groups each with relatively high proportionsof nec components had worse os than the gastricadenocarcinoma group fig 4bd p detailed clinical information after matching isshown in additional file tables s1s4psm was also performed in the neoadjuvant group incontrast to the results of the surgery group in the purenec group containing the highest proportion ofnec componentstill no significantdifference in os from gastric adenocarcinoma fig5d the other three groups with lower nec contentwere also notfrom gastricadenocarcinoma in terms of os fig 5ac detailedclinicopathologicaland afterpsm are shown in additional file tables s5s8characteristics beforethere wassignificantly differentdiscussionamong gastric neuroendocrine neoplasms the tumorcontaining nec components is a special type includingpure nec and mixed tumor containing nec components the incidence of these tumors is extremely lowbut they are more invasive and have a poorer prognosisthan welldifferentiated gnens [ ]received neoadjuvantin previous study kim found that in patientschemotherapywho had notprogressionfree survivalpfs of pure gnec waspoorer than that of gastric adenocarcinoma while thepfs of mixedtype tumors was not significantly differentin kim™sfrom that of gastric adenocarcinoma study the mixed type was defined as net mixed withgastric cancer rather than nec net is much less malignant than nec [ ] this may be the reason whythere was no significant difference in os between mixedtype and gastric adenocarcinomas in addition mixed tumors with less than or more than of nec components were not included in that study which webelieve was a deficit of the study pfs is an important indicator for evaluating prognosis in many cases it can reflect the trend of os based on kim™s research resultswe regarded tumors containing nec components as awhole and found that the os of these tumors was poorerthan that of adenocarcinoma in the surgical group inthe comparison of os between mixed tumors with different proportions of nec components and gastricadenocarcinoma the results for pure nec cases wassimilar to kim™s while the os of mixed tumors was alsopoorer than that of gastric adenocarcinoma whether theproportion of neuroendocrine cancer components wasless than between and or more than which was not mentioned in kim™s study cox multivariable regression analysis showed thattumor categoryneoplasm with nec component or adenocarcinoma 0cchen bmc cancer page of table comparison of clinicopathological characteristics before and after psm in surgical grouppatient characteristicsunmatched comparisonpatients with neccomponents n p valuematched comparisonpatients with neccomponents n age year mean ± sdgender malefemalebmi mean ± sdadjuvant therapyyesnotumor locationupper thirdmiddle thirdlower thirdentiretumor size cm‰¥ cmtype of gastrectomytotal gastrectomydistal gastrectomy ± ± proximal gastrectomy surgical procedureopenlaparoscopict staget1t2t3t4n stagen0n1n2n3m stagem0m1ptnm stageiiiiiiiv gastricadenocarcinoman ± ± ± ± p value gastricadenocarcinoman ± ± bmi body mass index minen mixed neuroendocrinenonneuroendocrine neoplasm nec neuroendocrine carcinoma psm propensity score matchingpatients with nec components nec high grade minen high grade minen and high grade minen 0cchen bmc cancer table comparison of clinicopathological characteristics before and after psm in neoadjuvant groupmatched comparisonpatient characteristicsunmatched comparisonpatients with neccomponents n age year mean ± sdgender malefemalebmi mean ± sdadjuvant therapyyesnotumor locationupper thirdmiddle thirdlower thirdentiretumor size cm‰¥ cmtype of gastrectomytotal gastrectomydistal gastrectomyproximal gastrectomysurgical procedureopenlaparoscopict staget0t1t2t3t4n stagen0n1n2n3m stagem0m1yptnm stageiiiiiiiv ± ± gastricadenocarcinoman ± ± p valuepatients with neccomponents n ± ± page of p valuegastricadenocarcinoman ± ± bmi body mass index minen mixed neuroendocrinenonneuroendocrine neoplasm nec neuroendocrine carcinoma psm propensity score matchingpatients with nec components nec high grade minen high grade minen and high grade minen 0cchen bmc cancer page of table univariate and multivariate analyses of survival after psm in surgical grouppatient characteristicsunivariate analysishr ci“multivariate analysishr cip valueage yeargendermale vs femalebmiadjuvant therapyyes vs notumor size‰¥ cm vs cmtumor categorycarcinoma with nec component vsgastric adenocarcinoma vstype of gastrectomytotal gastrectomydistal gastrectomyproximal gastrectomysurgical procedurelaparoscopic vs opentnm stageiiiiiiivp value“““ ““““““““““ “ ““““““““tumor size and tnm staging were independent risk factors for prognosis this suggests that the prognosis ofgastric neoplasms with nec components is substantiallydifferent from that of gastric adenocarcinoma and evena small percentage of nec components can alsoimpair prognosis which challenges the current cutoffvalue of the proportion of each component that must theoretically be greater than was set in andsince who has also adopted this standard to define minen this largely avoids the overdiagnosisof minen in tumors with only focal neuroendocrinemarker expression and no corresponding morphologicalchanges in additionit also prevents clinicians fromdealing with these rare neoplasms too often withoutguidelines nevertheless it is now being questionedby an increasing number of scholars the componentsin mixed tumors are not evenly distributed for large tumorsthe randomness of biopsy and postoperativepathological sampling causes the proportion of eachcomponent to fluctuate greatly making it difficult to describe the proportion of each component precisely park compared the os between tumors with morethan nec components and gastric adenocarcinomawith or without less than nec and they found thattumors with an nec composition of more than hada worse prognosis this suggests that even a small proportion of malignant components can affect prognosis while in park™s study for unknown reasons the authors did not compare the prognosis of mixed tumorswith nec components less than with gastricadenocarcinomas directly nor did they compare allneccontaining tumors as a whole with gastric adenocarcinoma which we believe was a deficit of the studyin our study we regarded tumors containing neccomponents as a whole and found that the os of thesetumors was poorer than that of adenocarcinoma in thesurgical group in addition we also found that the os ofmixed tumors with less than between and more than nec components or pure nec wasworse than that of gastric adenocarcinoma analysis ofimmunohistochemical markers show that there was nosignificant difference in the positive rate of syn and cgabetween different nec content groups only the positiverate of cd56 was found to be higher in the pure necgroup than that in the ghminen group therole of cd56 in the diagnosis of nec is still controversial however syn and cga are two wellrecognized 0cchen bmc cancer page of fig comparison of os between gastric neoplasm with different proportions of nec and gastric adenocarcinoma in the surgical group aoverall survival comparison between ghminen and gastric adenocarcinoma b overall survival comparison between ghminen andgastric adenocarcinoma c overall survival comparison between ghminen plus pure nec and gastric adenocarcinoma d overall survivalcomparison between pure nec alone and gastric adenocarcinomamarkers therefore from the results of immunohistochemistry we believed that there was no significantlydifference in tumors containing nec componentsstudies on the molecular mechanism of pathogenesisshow that nec components and adenocarcinoma components have similar genomic abnormalities similarlosses of heterozygosity loh and mutations at multiple loci and key oncogenes such as tp53 apc and rbgenes all these results imply that the two componentsin the mixed tumor may have a common origin and acquire biphenotypic differentiation during carcinogenesis[“] moreoverin the who definition of mixedneuroendocrine and nonneuroendocrine neoplasms ofother ans ie lung and thyroid no minimumpercentage for either ingredient is established thereforewe believe that mixed tumors containing nec components are actually of the same origin have similar biological characteristics and are differentfrom gastricadenocarcinoma we propose considering mixed tumorscontaining nec components as a whole rather than defining them based on the definition for both tumorcomponents which has not been raised by other studiespreviously many studies have confirmed the efficacyof neoadjuvant chemotherapy in gastric adenocarcinoma[ ] in a retrospective study involving patientsma found that neoadjuvant chemotherapy improves the survival of patients with nec and hminenof the stomach van der veen reported that 0cchen bmc cancer page of fig comparison of os between gastric neoplasm with different proportions of nec components and gastric adenocarcinoma in theneoadjuvant group a overall survival comparison between ghminen and gastric adenocarcinoma b overall survival comparisonbetween ghminen and gastric adenocarcinoma c overall survival comparison between ghminen plus pure nec and gastricadenocarcinoma d overall survival comparison between pure nec and gastric adenocarcinomaneoadjuvant chemotherapy could not benefit the survivalof patients with mixed tumors containing nec components however because only eight patients wereincluded in the neoadjuvant group van™s results arequestionable in our study among patients receivingneoadjuvanttherapy no significant difference in osbetween mixed tumor and gastric adenocarcinoma wasobserved even for the pure nec group with the highestnec contentthere was no significant differencesuggesting that neoadjuvant therapy may have a positiveeffect on these neoplasmsalthough this is only a singlecenter retrospectivestudy the sample we reported is considerable for thisrare disease which can provide new ideas for clinicaland basic research in addition we proposed treatingall gastric neoplasms containing nec components asa whole and found that neoadjuvanttherapy mayhave a good effect on these neoplasms in the futurewe will conduct more genomics studies to confirmour ideas this study also has its limitations due tothe lack of recurrence and detailed chemotherapy information we were unable to compare progressionfree survival and analyse the effects of differentchemotherapy regimens as a retrospective study despite our performing psm in advance selection biascannot be completely avoided in addition since theexact proportion of each componentin the mixedtumor could not be obtained we could not determine 0cchen bmc cancer page of whether there is a cutoff value for the diagnosis ofthe mixed tumor with nec componentless than so we could only treat all mixed tumors withnec component as a wholesour study demonstrated that gastric neoplasms withnec components regardless of the proportion of components have poorer overall survival than gastric adenocarcinomaindicating a higher degree of malignancythan gastric adenocarcinoma among the patients receiving neoadjuvant therapy the difference in overallsurvival was not significant which was in sharp contrastwith the results of the surgery group suggesting thatneoadjuvant therapy may have a good effect on theprognosis of these malignancies therefore for this typeof malignancy we should adopt more aggressive andpowerful treatments than those used for gastric adenocarcinoma to improve the prognosis of patients neoadjuvant chemotherapy may be a good way to improve theefficacy offor these tumors at advancedstagestreatmentsupplementary informationsupplementary information accompanies this paper at httpsdoi101186s12885020072817additional file table s1 comparison of clinicopathologicalcharacteristics before and after psm of 30ghminen patients insurgical group table s2 comparison of clinicopathologicalcharacteristics before and after psm of ghminen patients in surgicalgroup table s3 comparison of clinicopathological characteristics beforeand after psm of 70ghminen plus pure nec patients in surgicalgroup table s4 comparison of clinicopathological characteristics beforeand after psm of pure nec patients in surgical group table s5 comparison of clinicopathological characteristics before and after psm of 30ghminen patients in neoadjuvant group table s6 comparison ofclinicopathological characteristics before and after psm of ghminen patients in neoadjuvant group table s7 comparison of clinicopathologicalcharacteristics before and after psm of 70ghminen plus pure necpatients in neoadjuvant group table s8 comparison of clinicopathological characteristics before and after psm of pure nec patients in neoadjuvant groupabbreviationsajcc american joint committee on cancer ct computed tomography ghminen gastric highgrade mixed neuroendocrinenonneuroendocrineneoplasm gnec gastric neuroendocrine carcinoma hpf high power fieldminen mixed neuroendocrinenonneuroendocrine neoplasmnec neuroendocrine carcinoma nen neuroendocrine neoplasmnet neuroendocrine tumor mri magnetic resonance imaging os overallsurvival petct positron emission tomography computed tomographypfs progressionfree survival psm propensity score matching who worldhealth anizationacknowledgmentsthanks to dr zhongwu li of the department of pathology peking universitycancer hospital and his colleagues for their assistance in pathologicaldiagnosis and review thanks to all colleagues in the department ofgastrointestinal surgery of peking university cancer hospital and dr jianghong from the statistics department for their assistance in this studyauthors™ contributionsall authors contributed to the study conception and design jc performeddata collection and wrote the manuscript aw wrote and t revised hemanuscript kj helped with statistical analysis and prepared the illustrationszb edited the manuscript jj conceived the study and reviewed themanuscript all authors read and approved the final manuscriptfundingthis work was supported by the national science foundation for youngscientists of china beijing youth talent plan qml20191101 andscience foundation of peking university cancer hospital “ thefunders had no role in study design data collection and analysis decision topublish or preparation of the manuscriptavailability of data and materialsthe datasets used andor analysed during the current study are availablefrom the corresponding author on reasonable requestethics approval and consent to participatethe study was approved by the ethics committee of peking universitycancer hospital and the patients™ written consent was also obtained writteninformed consent for publication was obtained and stored in pekinguniversity cancer hospitalconsent for publicationnot applicablecompeting intereststhe authors declare that they have no competing interestsreceived may accepted august referencesbray f ferlay j soerjomataram i siegel rl torre la jemal a global cancerstatistics globocan estimates of incidence and mortality worldwidefor cancers in countries ca cancer j clin “ matsubayashi h takagaki s otsubo t iiri t kobayashi y yokota t advanced gastric glandularendocrine cell carcinoma with 1year survivalafter gastrectomy gastric cancer “park jy ryu mh park ys park hj ryoo by kim mg prognosticsignificance of neuroendocrine components in gastric carcinomas eur jcancer “la rosa s inzani f vanoli a klersy c dainese l rindi g histologiccharacterization and improved prognostic evaluation of gastricneuroendocrine neoplasms hum pathol “ishida m sekine s fukagawa t ohashi m morita s taniguchi h neuroendocrine carcinoma of the stomach morphologic andimmunohistochemical characteristics and prognosis am j surg pathol“rayhan n sano t qian zr obari ak hirokawa m histological andimmunohistochemical study of composite neuroendocrineexocrinecarcinomas of the stomach j med investig ““jiang sx mikami t umezawa a saegusa m kameya t okayasu i gastriclarge cell neuroendocrine carcinomas a distinct clinicopathologic entityam j surg pathol “ohike nan la rosa s who classification of tumours of endocrineans 4th ed lyon iarc press amin mb edge sb ajcc cancer s

"The second patient with EGFR mutation achieved the longest PFS and OS (727 and 1249 days respectively). .0087629.g007 Kaplan-Meier estimates of PFS and OS. No statistically significant difference (P?=?0.007) in PFS was observed between metabolic non-progressive (mNP) patients (median PFS 292 days ; range 190“727) and metabolic (mP) progressive patients (median PFS 64 days ; range: 37“216). Improved PFS in non-progressive patients was associated with prolonged OS (mNP; n?=?4; median OS: 1031 days ; 296 to 1249 days versus mP; n?=?8 ; 337 5 days ; 71 to 734 days) (HR 0.34; 95% CI 0.06 to 0.84; P?=?0.03). Discussion Despite the widespread use of [18F]FDG-PET/CT in NSCLC staging a large-scale study recently failed to confirm an overall survival gain in NSCLC patients.[17] This result highlights the value of [18F]FDG-PET/CT in unmet clinical needs such as prediction of residual NSCLC after surgery[18] neoadjuvant therapy[19] or antineoplastic therapy.[20] Prediction of response to antineoplastic therapies would appear to be particularly adapted to targeted therapies that do not induce rapid tumor shrinkage. NSCLC preclinical models have validated this hypothesis with both gefitinib[21] and erlotinib.[22] This original method could compensate for the weakness of RECIST criteria and has led to the proposal of evaluation of new criteria by addition metabolic evaluation by FDG-PET to CT scan.[23] The value of PET in evaluation of response to new targeted therapies emerged in the early 2000 s with the first reports on the efficacy of imatinib mesylate in Gastro Intestinal Stromal Tumor (GIST). Subsequently many studies have confirmed that PET is able to identify very early (i.e. only 24 hours after initiation of treatment) a decrease in glucose metabolism which is correlated with overall survival and progression-free survival of patients with GIST.[24] [25] In the present exploratory study a decrease in SUVmax of at least 21.6% soon after starting therapy (9±3 days) was able to discriminate progressive from non-progressive patients and was associated with improved PFS and OS. This result confirms the results of Mileshkin et al. who showed in a series of 51 patients receiving second- or third-line treatment with erlotinib that an early (14 days) [18F]FDG-PET partial metabolic response was associated with improved PFS and OS even in the absence of subsequent RECIST response.[26] Evaluation of response by [18F]FDG-PET can be performed semi-quantitatively for instance by establishing a SUV cut-off to discriminate metabolic progressive patients from non-metabolic progressive patients. This patient classification (mP/mNP) seems to be more appropriate to assess response to cytostatic therapy that is designed to stabilize disease rather than achieve complete response. The main difficulty of this approach is the overlap of SUV changes between mP and mNP patients. Furthermore different cut-off variations can be expected depending on the types of SUV measured the types of drugs used and the types of tumors which increase the difficulty of establishing a reliable SUV cut-off. However despite the absence of consensus on the most appropriate cut-off value it is generally admitted that the rationale for metabolic response or non-progression of tumor is decreased [18F]FDG tumor uptake or at least stability of tumor uptake over time respectively. Another limitation of semi-quantitative analysis of FDG-PET is that it does not take into account the development of new lesions. However PET detection of new lesions early in the course of therapy has been reported to be a strong independent predictive factor of OS in NSCLC patients treated by EGFR inhibitor.[27] Our findings are consistent with this observation as new lesions occurred in 2/8 patients correctly classified as progressive on PET2 and in 4/5 patients correctly classified as progressive on PET3. One patient (patient #7) was reclassified as mP on PET3 due to the appearance of a new lesion despite a decrease of SUVmax to below the cut-off value. As in our study previous studies failed to demonstrate any difference between SUVmax and SUVpeak.[22] [28] However SUVmax remains the standard for semi-quantitative [18F]FDG-PET assessment probably because is a parameter that can be reliably reproduced by independent operators. It is noteworthy that in our study no significant difference in mean SUV values was observed between PET1 PET2 and PET3 which can be explained by the nature of the cytostatic therapy. 11/12 patients were correctly classified (P versus NP) by PET2 and 10/10 were correctly classified by PET3 by applying the SUV cut-off determined by ROC analysis. In 9/10 patients PET3 revealed response information concordant with PET2. The only patient with discordant [18F]FDG-PET findings was classified by SUV analysis as progressive on PET2 and non-progressive on PET3. Blood glucose injected dose or uptake time were normal and/or not significantly different between PET2 and PET3 (1.16 and 1.4 g/l; 261 and 262 MBq; 60 and 75 min respectively) excluding any to methodology-related error. A flare-up phenomenon could be proposed as described on several occasions on [18F]FDG-PET during cytotoxic treatments for squamous cell carcinoma in prostate cancer patients with bone metastases[29]“[33] and particularly NSCLC patients treated with erlotinib presenting an osteoblastic bone flare-up response mimicking disease progression.[34] Benz et al also described a case of flare-up on early PET in a NSCLC patient treated by erlotinib.[27] Another explanation is that the P/NP classification probably increases mismatches of response assessments related to a discordant outcome of patients with stable disease.[27] Our results suggest that therapeutic efficacy PFS and OS of erlotinib therapy can be predicted 2 weeks after starting erlotinib. These data are consistent with the data of a retrospective study recently published by Kobe et al.[26] [35] At the present time anticancer therapy is currently monitored in the context of hormone-sensitive cancers by regular assay of tumor markers (such as prostate-specific antigen in prostate cancer). The efficacy of hormonal therapy is reflected by a decrease in blood levels of the marker. When the marker remains elevated hormonal therapy is considered to be ineffective and is therefore stopped. Repeated PET imaging can be considered to be a promising approach to evaluate cancer therapy such as targeted therapies that do not induce tumor shrinkage. This new approach appears to be supported by the results of recent clinical trials. The ˜Tarceva Versus Docetaxel or Pemetrexed for Second Line Chemotherapy of Advanced Stage NSCLC™ (TITAN) trial failed to demonstrate an improvement in OS with erlotinib compared to chemotherapy in unselected NSCLC patients receiving second-line treatment (HR?=?0.96; 95% CI 0.78“1.19; p?=?0.73).[36] In a similar group of NSCLC patients the results of the TAILOR trial indicated a highly significant increase of PFS in favor of docetaxel (HR?=?0.71; 95% CI 0.53“0.95; p?=?0.02) versus erlotinib.[37] We consider that evaluation of the metabolic response to erlotinib could provide useful information to rapidly identify patients in whom erlotinib therapy is ineffective especially in EGFR patients without EGFR-activating mutations or unknown status. [18F]FDG-PET could also become a theranostic tool for clinicians. By stopping ineffective therapy earlier physicians can rapidly propose other drugs to a larger proportion of patients with better performance status. This approach could increase the number of patients included in early trials and accelerate drug development. However no medico-economic study has been conducted to determine whether the additional costs induced by [18F]FDG-PET are compensated by the decreased costs of drug (erlotinib) and medical care induced by side effects. Our study highlights the need for more prospective and randomized studies to evaluate the theranostic use of [18F]FDG-PET for management of erlotinib therapy in NSCLC including medico-economic considerations. Conclusion [18F]FDG-PET performed within two weeks of starting erlotinib therapy (9±3 days) appears to be able to predict morphologic response at 2 months according to RECIST criteria. [18]FDG-PET may be clinically useful for early evaluation of targeted therapies as a theranostic tool. Nathalie BAIZE MD Université d'Angers CHU Angers Pôle des Spécialités Médicales et Chirurgicales Intégrées Département de Pneumologie Angers France References 1 FerlayJ ParkinDM Steliarova-FoucherE (2010) Estimates of cancer incidence and mortality in Europe in 2008. Eur J Cancer46: 765“78120116997 2 JemalA BrayF CenterMM FerlayJ WardE et al (2011) Global cancer statistics. CA Cancer J Clin61: 69“9021296855 3 Chemotherapy in non-small cell lung cancer: a meta-analysis using updated data on individual patients from 52 randomised clinical trials. Non-small Cell Lung Cancer Collaborative Group. BMJ311: 899“909 4 SchillerJH HarringtonD BelaniCP LangerC SandlerA et al (2002) Comparison of four chemotherapy regimens for advanced non-small-cell lung cancer. N Engl J Med346: 92“9811784875 5 LynchTJ BellDW SordellaR GurubhagavatulaS OkimotoRA et al (2004) Activating mutations in the epidermal growth factor receptor underlying responsiveness of non-small-cell lung cancer to gefitinib. N Engl J Med350: 2129“213915118073 6 MokTS WuYL ThongprasertS YangCH ChuDT et al (2009) Gefitinib or carboplatin-paclitaxel in pulmonary adenocarcinoma. N Engl J Med361: 947“95719692680 7 RosellR CarcerenyE GervaisR VergnenegreA MassutiB et al (2012) Erlotinib versus standard chemotherapy as first-line treatment for European patients with advanced EGFR mutation-positive non-small-cell lung cancer (EURTAC): a multicentre open-label randomised phase 3 trial. Lancet Oncol13: 239“24622285168 8 ShepherdFA Rodrigues PereiraJ CiuleanuT TanEH HirshV et al (2005) Erlotinib in previously treated non-small-cell lung cancer. N Engl J Med353: 123“13216014882 9 CappuzzoF CiuleanuT StelmakhL CicenasS SzczesnaA et al (2010) Erlotinib as maintenance treatment in advanced non-small-cell lung cancer: a multicentre randomised placebo-controlled phase 3 study. Lancet Oncol11: 521“52920493771 10 MillerAB HoogstratenB StaquetM WinklerA (1981) Reporting results of cancer treatment. Cancer47: 207“2147459811 11 TherasseP ArbuckSG EisenhauerEA WandersJ KaplanRS et al (2000) New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer National Cancer Institute of the United States National Cancer Institute of Canada. J Natl Cancer Inst92: 205“21610655437 12 EisenhauerEA TherasseP BogaertsJ SchwartzLH SargentD et al (2009) New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer45: 228“24719097774 13 FischerBM MortensenJ HojgaardL (2001) Positron emission tomography in the diagnosis and staging of lung cancer: a systematic quantitative review. Lancet Oncol2: 659“66611902536 14 LardinoisD WederW HanyTF KamelEM KoromS et al (2003) Staging of non-small-cell lung cancer with integrated positron-emission tomography and computed tomography. N Engl J Med348: 2500“250712815135 15 VansteenkisteJF StroobantsSG DupontPJ De LeynPR VerbekenEK et al (1999) Prognostic importance of the standardized uptake value on (18)F-fluoro-2-deoxy-glucose-positron emission tomography scan in non-small-cell lung cancer: An analysis of 125 cases. Leuven Lung Cancer Group. J Clin Oncol17: 3201“320610506619 16 BerghmansT DusartM PaesmansM Hossein-FoucherC BuvatI et al (2008) Primary tumor standardized uptake value (SUVmax) measured on fluorodeoxyglucose positron emission tomography (FDG-PET) is of prognostic value for survival in non-small cell lung cancer (NSCLC): a systematic review and meta-analysis (MA) by the European Lung Cancer Working Party for the IASLC Lung Cancer Staging Project. J Thorac Oncol3: 6“1218166834 17 DinanMA CurtisLH CarpenterWR BiddleAK AbernethyAP et al (2012) Stage migration selection bias and survival associated with the adoption of positron emission tomography among medicare beneficiaries with non-small-cell lung cancer 1998-2003. J Clin Oncol30: 2725“273022753917 18 VelazquezER AertsHJ OberijeC De RuysscherD LambinP (2010) Prediction of residual metabolic activity after treatment in NSCLC patients. Acta Oncol49: 1033“103920831492 19 AukemaTS KappersI OlmosRA CodringtonHE van TinterenH et al (2010) Is 18F-FDG PET/CT useful for the early prediction of histopathologic response to neoadjuvant erlotinib in patients with non-small cell lung cancer? J Nucl Med51: 1344“134820720059 20 plannedT SchefflerM NogovaL KobeC Engel-RiedelW et al (2011) Early prediction of nonprogression in advanced non-small-cell lung cancer treated with erlotinib by using [(18)F]fluorodeoxyglucose and [(18)F]fluorothymidine positron emission tomography. J Clin Oncol29: 1701“170821422426 21 SuH BodensteinC DumontRA SeimbilleY DubinettS et al (2006) Monitoring tumor glucose utilization by positron emission tomography for the prediction of treatment response to epidermal growth factor receptor kinase inhibitors. Clin Cancer Res12: 5659“566717020967 22 UllrichRT ZanderT NeumaierB KokerM ShimamuraT et al (2008) Early detection of erlotinib treatment response in NSCLC by 3?-deoxy-3?-[F]-fluoro-L-thymidine ([F]FLT) positron emission tomography (PET). PLoS One3: e390819079597 23 WahlRL JaceneH KasamonY LodgeMA (2009) From RECIST to PERCIST: Evolving Considerations for PET response criteria in solid tumors. J Nucl Med50 Suppl 1122S“150S19403881 24 StroobantsS GoeminneJ SeegersM DimitrijevicS DupontP et al (2003) 18FDG-Positron emission tomography for the early prediction of response in advanced soft tissue sarcoma treated with imatinib mesylate (Glivec). Eur J Cancer39: 2012“202012957455 25 Van den AbbeeleAD (2008) The lessons of GIST“PET and PET/CT: a new paradigm for imaging. Oncologist13 Suppl 28“1318434632 26 MileshkinL HicksRJ HughesBG MitchellPL CharuV et al (2011) Changes in 18F-fluorodeoxyglucose and 18F-fluorodeoxythymidine positron emission tomography imaging in patients with non-small cell lung cancer treated with erlotinib. Clin Cancer Res17: 3304“331521364032 27 BenzMR HerrmannK WalterF GaronEB ReckampKL et al (2011) (18)F-FDG PET/CT for monitoring treatment responses to the epidermal growth factor receptor inhibitor erlotinib. J Nucl Med52: 1684“168922045706 28 KahramanD SchefflerM ZanderT NogovaL LammertsmaAA et al (2011) Quantitative analysis of response to treatment with erlotinib in advanced non-small cell lung cancer using 18F-FDG and 3?-deoxy-3?-18F-fluorothymidine PET. J Nucl Med52: 1871“187722065872 29 BjurbergM HenrikssonE BrunE EkbladL OhlssonT et al (2009) Early changes in 2-deoxy-2-[18F]fluoro-D-glucose metabolism in squamous-cell carcinoma during chemotherapy in vivo and in vitro. Cancer Biother Radiopharm24: 327“33219538055 30 MessiouC CookG ReidAH AttardG DearnaleyD et al (2011) The CT flare response of metastatic bone disease in prostate cancer. Acta Radiol52: 557“56121498309 31 KrupitskayaY EslamyHK NguyenDD KumarA WakeleeHA (2009) Osteoblastic bone flare on F18-FDG PET in non-small cell lung cancer (NSCLC) patients receiving bevacizumab in addition to standard chemotherapy. J Thorac Oncol4: 429“43119247091 32 BiersackHJ BenderH PalmedoH (2004) FDG-PET in monitoring therapy of breast cancer. Eur J Nucl Med Mol Imaging31 Suppl 1S112“11715112111 33 MortimerJE DehdashtiF SiegelBA TrinkausK KatzenellenbogenJA et al (2001) Metabolic flare: indicator of hormone responsiveness in advanced breast cancer. J Clin Oncol19: 2797“280311387350 34 LindJS PostmusPE SmitEF (2010) Osteoblastic bone lesions developing during treatment with erlotinib indicate major response in patients with non-small cell lung cancer: a brief report. J Thorac Oncol5: 554“55720357621 35 KobeC SchefflerM HolsteinA ZanderT NogovaL et al (2012) Predictive value of early and late residual 18F-fluorodeoxyglucose and 18F-fluorothymidine uptake using different SUV measurements in patients with non-small-cell lung cancer treated with erlotinib. Eur J Nucl Med Mol Imaging39: 1117“112722526960 36 CiuleanuT StelmakhL CicenasS MiliauskasS GrigorescuAC et al (2012) Efficacy and safety of erlotinib versus chemotherapy in second-line treatment of patients with advanced non-small-cell lung cancer with poor prognosis (TITAN): a randomised multicentre open-label phase 3 study. Lancet Oncol13: 300“30822277837 37 GarassinoMC MartelliO BrogginiM FarinaG VeroneseS et al (2013) Erlotinib versus docetaxel as second-line treatment of patients with advanced non-small-cell lung cancer and wild-type EGFR tumours (TAILOR): a randomised controlled trial. Lancet Oncol14: 981“98823883922 Nucleic Acids Res Nucleic Acids Res nar nar Nucleic Acids Research 0305-1048 1362-4962 Oxford University Press 24970867 4117748 10.1093/nar/gku489 15 Methods Online Integrated RNA and DNA sequencing improves mutation detection in low purity tumors Wilkerson Matthew D. 1 2 * Cabanski Christopher R. 1 3 Sun Wei 2 4 Hoadley Katherine A. 1 2 Walter Vonn 1 Mose Lisle E. 1 Troester Melissa A. 1 5 Hammerman Peter S. 6 7 Parker Joel S. 1 2 Perou Charles M. 1 2 Hayes D. Neil 1 8 * 1Lineberger Comprehensive Cancer Center University of North Carolina at Chapel Hill Chapel Hill NC 27599 USA 2Department of Genetics University of North Carolina at Chapel Hill Chapel Hill NC 27599 USA 3The Genome Institute at Washington University St. Louis MO 63108 USA 4Department of Biostatistics University of North Carolina at Chapel Hill Chapel Hill NC 27599 USA 5Department of Epidemiology University of North Carolina at Chapel Hill Chapel Hill NC 27599 USA 6Department of Medical Oncology Dana-Farber Cancer Institute Boston MA 02215 USA 7Broad Institute of Harvard and MIT Cambridge MA 02142 USA 8Department of Internal Medicine Division of Medical Oncology Multidisciplinary Thoracic Oncology Program University of North Carolina at Chapel Hill Chapel Hill NC 27599 USA *To whom correspondence should be addressed. Tel: +1 919 966 3098; Fax: +1 919 966 1587; Email: mwilkers@med.unc.edu Correspondence may also be addressed to D. Neil Hayes. Tel: +1 919 966 3786; Fax: +1 919 966 1587; Email: hayes@med.unc.edu 01 9 2014 26 6 2014 26 6 2014 42 13 e107 e107 15 5 2014 22 4 2014 14 10 2013 © The Author(s) 2014. Published by Oxford University Press on behalf of Nucleic Acids Research. 2014 This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0/) which permits unrestricted reuse distribution and reproduction in any medium provided the original work is properly cited. Identifying somatic mutations is critical for cancer genome characterization and for prioritizing patient treatment. DNA whole exome sequencing (DNA-WES) is currently the most popular technology; however this yields low sensitivity in low purity tumors. RNA sequencing (RNA-seq) covers the expressed exome with depth proportional to expression. We hypothesized that integrating DNA-WES and RNA-seq would enable superior mutation detection versus DNA-WES alone. We developed a first-of-its-kind method called UNCeqR that detects somatic mutations by integrating patient-matched RNA-seq and DNA-WES. In simulation the integrated DNA and RNA model outperformed the DNA-WES only model. Validation by patient-matched whole genome sequencing demonstrated superior performance of the integrated model over DNA-WES only models including a published method and published mutation profiles. Genome-wide mutational analysis of breast and lung cancer cohorts (n = 871) revealed remarkable tumor genomics properties. Low purity tumors experienced the largest gains in mutation detection by integrating RNA-seq and DNA-WES. RNA provided greater mutation signal than DNA in expressed mutations. Compared to earlier studies on this cohort UNCeqR increased mutation rates of driver and therapeutically targeted genes (e.g. PIK3CA ERBB2 and FGFR2). In summary integrating RNA-seq with DNA-WES increases mutation detection performance especially for low purity tumors. cover-date 2014 INTRODUCTION Somatically acquired sequence mutations (nucleotide substitutions insertions and deletions) fuel the initiation and progression of cancer (1). Knowledge of mutations in patient specimens informs therapeutic management (23) and in large patient cohorts provides the basis to assess recurrently altered genes that may drive molecular pathogenesis (14“5). DNA whole exome sequencing (DNA-WES) is currently the popular technology to sequence cancer genomes and has led to an abundance of discoveries in many cancer types (46“8). However detecting somatic mutations by DNA-WES with high sensitivity and specificity remains a challenge (79“10) as evidenced by validation rates of 73% in repeated sequencing and by large inter-rater disagreement among different groups analyzing the same sequencing data (710). The biggest challenge is high quality mutation detection in low purity tumors "

"\" exosomes are extracellular vesicles containing a variety of biological molecules including micr(...TRUNCATED)

"\" the development of a safe effective reversible nonhormonal contraceptive method for men hasbeen(...TRUNCATED)

" Hydrophobicity drives receptormediateduptake of heatprocessed proteins by THP1macrophages and dend(...TRUNCATED)

"pulmonary disease COPD is due to structural changes and narrowing of small airways and parenchymald(...TRUNCATED)