PrognosticRelevanceofIntegratedGeneticProfilinginAcuteMyeloidLeukemia急性髓細胞白血病的整體遺傳特征輪廓預后相關性研究JayP.Patel,MithatG?nen,Ph.D.,MariaE.Figueroa,M.D.,HugoFernandez,M.D.,ZhuoxinSun,Ph.D.,JanisRacevskis,Ph.D.,PieterVanVlierberghe,Ph.D.,IgorDolgalev,B.S.,SabrenaThomas,B.S.,OlgaAminova,B.S.,KetyHuberman,B.S.,JaniceCheng,B.S.,AgnesViale,Ph.D.,NicholasD.Socci,Ph.D.,AdrianaHeguy,Ph.D.,AthenaCherry,Ph.D.,GailVance,M.D.,RodneyR.Higgins,Ph.D.,RhettP.Ketterling,M.D.,RobertE.Gallagher,M.D.,MarkLitzow,M.D.,MarcelR.M.vandenBrink,M.D.,Ph.D.,HillardM.Lazarus,M.D.,JacobM.Rowe,M.D.,SelinaLuger,M.D.,AdolfoFerrando,M.D.,Ph.D.,ElisabethPaietta,Ph.D.,MartinS.Tallman,M.D.,AriMelnick,M.D.,OmarAbdel-Wahab,M.D.,andRossL.Levine,M.D.AbstractBackgroundAcutemyeloidleukemia(AML)isaheterogeneousdiseasewithrespecttopresentationandclinicaloutcome.Theprognosticvalueofrecentlyidentifiedsomaticmutationshasnotbeensystematicallyevaluatedinaphase3trialoftreatmentforAML.急性髓細胞白血病是一種有關表達和臨床結果異構性疾病。最近發現的體細胞突變的預后價值尚用于急性髓細胞白血病治療的臨床三期試驗的系統評價。MethodsWeperformedamutationalanalysisof18genesin398patientsyoungerthan60yearsofagewhohadAMLandwhowererandomlyassignedtoreceiveinductiontherapywithhigh-doseorstandard-dosedaunorubicin.Wevalidatedourprognosticfindingsinanindependentsetof104patients.我們選擇年齡小于60歲的AML398例隨機分配到高劑量組或標準劑量組，接受柔紅霉素的誘導治療，治療后對18個基因作突變的分析。并另選擇104例患者進行驗證了預后結果ResultsWeidentifiedatleastonesomaticalterationin97.3%ofthepatients.WefoundthatinternaltandemduplicationinFLT3(FLT3-ITD),partialtandemduplicationinMLL(MLL-PTD),andmutationsinASXL1andPHF6wereassociatedwithreducedoverallsurvival(P=0.001forFLT3-ITD,P=0.009forMLL-PTD,P=0.05forASXL1,andP=0.006forPHF6);CEBPAandIDH2mutationswereassociatedwithimprovedover-allsurvival(P=0.05forCEBPAandP=0.01forIDH2).ThefavorableeffectofNPM1mutationswasrestrictedtopatientswithco-occurringNPM1andIDH1orIDH2mutations.WeidentifiedgeneticpredictorsofoutcomethatimprovedriskstratificationamongpatientswithAML,independentlyofage,white-cellcount,inductiondose,andpost-remissiontherapy,andvalidatedthesignificanceofthesepredictorsinanindependentcohort.High-dosedaunorubicin,ascomparedwithstandard-dosedaunorubicin,improvedtherateofsurvivalamongpatientswithDNMT3AorNPM1mutationsorMLLtranslocations(P=0.001)butnotamongpatientswithwild-typeDNMT3A,NPM1,andMLL(P=0.67).我們可以確定97.3%的患者至少有一個細胞發生了改變。在FLT3基因中存在串聯重復（FLT3-ITD），在MLL基因中存在部分串聯重復（MLL-PTD），ASXL1，PHF6突變與降低總體生存率相關(P=0.001forFLT3-ITD,P=0.009forMLL-PTD,P=0.05forASXL1,andP=0.006forPHF6)；CEBPA和IDH2突變與總生存率改善相關(P=0.05forCEBPAandP=0.01forIDH2)。NPM1突變的有利作用僅限于NPM1和IDH1或IDH2同時發生突變患者。我們確定的改進AML患者危險分層之間遺傳因素預測因子，該因子獨立于年齡、白細胞計數、誘導劑量和治療后緩解治療，并通過獨立的隊列驗證這些預測因子的重要性。ConclusionsWefoundthatDNMT3AandNPM1mutationsandMLLtranslocationspredictedanimprovedoutcomewithhigh-doseinductionchemotherapyinpatientswithAML.ThesefindingssuggestthatmutationalprofilingcouldpotentiallybeusedforriskstratificationandtoinformprognosticandtherapeuticdecisionsregardingpatientswithAML.(FundedbytheNationalCancerInstituteandothers.)我們發現DNMT3A和NPM1突變和MLL易位預測高劑量誘導化療的AML患者的轉歸改善結果。研究結果表明，突變譜可能被用于AML患者的危險分層和預后預測和治療決策Previousstudieshavehighlightedtheclinicalandbiologicheterogeneityofacutemyeloidleukemia(AML).1-4。However,arelativelysmallnumberofcytogeneticandmolecularlesionshavesufficientrelevancetoinfluenceclinicalpractice.5Theprognosticrelevanceofcytogeneticabnormalitieshasledtothewide-spreadadoptionofriskstratification,withpatientsdividedintothreecytogeneticallydefinedriskgroupswithsignificantdifferencesinover-allsurvival.6Morerecently,FLT3,NPM1,andCEBPAmutationalanalysiswasshowntoimproveriskstratificationforpatientswhodonothavekaryotypicabnormalities.7AlthoughprogresshasbeenmadeindefiningprognosticmarkersforAML,asubstantialpercentageofpatientslackaspecificabnormalityofprognosticsignificance.Inaddition,thereisconsiderableheterogeneityintheoutcomeforindividualpatientsineachriskgroup.以前的研究都強調，急性髓細胞白血病的臨床和生物學異質性。然而，相對較少的細胞遺傳學和分子病變具有充分的關聯性而影響臨床實踐。細胞遺傳異常的預后相關性促使廣泛采用危險因素分層，患者根據細胞遺傳學定義的風險分為三組，其所有生存時間有顯著差異。最近，NPM1，FLT3，和CEBPA突變分析顯示沒有核型畸形患者改善危險分層。雖然在確定的預后標記的AML取得了進展，相當比例的患者缺乏特異性預后差異。此外，各風險組中存在有相當大的異質性結果。RecentstudieshaveidentifiednovelrecurrentsomaticmutationsinpatientswithAML.TheseincludemutationsinTET2,8,9ASXL1,10IDH1orIDH2,11-13DNMT3A,4,14andPHF6.15RetrospectiveanalysessuggestthatasubsetofthesemutationsmayhaveprognosticsignificanceinAML,4,14,16althoughthesefindingshavenotbeenvalidatedwithdetailedclinicalandmutationalannotationinlarge,homogeneouslytreatedcohortsofpatientswithAML.Inaddition,thequestionofwhethermutationalprofilingofalargersetofgenes,includingthesenoveldiseasealleles,improvesprognosticationinAMLhasnotbeeninvestigatedinaclinicaltrialcohort.最近的研究發現AML患者存在新的復發性體細胞突變。這些突變包括TET2，ASXL1，IDH1和IDH2，DNMT3A，4PHF6。回顧性分析表明，這些突變可能存在預后差異性，盡管這些研究沒有經過詳細的臨床和基因突變的進行驗證，也沒治療證實。此外，是否進行大量的基因輪廓分析，包括這些改善AML的預后疾病新的等位基因，這一問題尚未通過隊列研究進行臨床驗證。Arecentphase3clinicaltrial(E1900;ClinicalTnumber,NCT00049517)fromtheEasternCooperativeOncologyGroup(ECOG)showedthatinductiontherapywithcytarabineplus90mgofdaunorubicinpersquaremeterofbody-surfacearea,ascomparedwithcytarabineplus45mgofdaunorubicinpersquaremeter,improvedtheoutcomesinpatientswithnewlydiagnosedAMLwhowere17to60yearsofage17;asimilarstudyinpatientswhowereolderthan60yearsofageshowedthatdose-intensifieddaunorubicinimprovedoverallsurvivalinpatients60to65yearsofage.18Wehypothesizedthatintegratedmutationalanalysisofallknownmolecularalterationsoccurringinmorethan5%ofpatientswithAMLwouldallowustoidentifynovelmolecularmarkersofoutcomeinAMLandtoidentifymolecularlydefinedsubgroupsofpatientswhowouldbenefitfromdose-intensifiedinductionchemotherapy.最近的來自東部腫瘤協作組（ECOG）的3期臨床試驗表明，與阿糖胞苷加45毫克每平方米柔紅霉素相比，阿糖胞苷加90毫克每平方米體表面積的柔紅霉素誘導治療可改善17至60歲初診AML患者的預后。類似研究表明年齡60歲以上病人加大劑量的柔紅霉素治療可以提高60到65歲患者總體生存。我們假設，所有已知的分子改變發生超過5%的AML患者的基因突變分析，允許我們識別新的結果分子標記，識別受益于增強誘導化療劑量AML患者的亞組。MethodsPatientsWeperformedmutationalanalysisondiagnosticsamplesobtainedfrompatientsintheECOGE1900trial.Allpatientsprovidedwritteninformedconsent.Thetestcohort(398patients)comprisedallpatientsintheE1900trialforwhomviablyfrozencellswereavailableforDNAextractionandmutationalprofiling.Thevalidationcohort(104patients)comprisedasecondsetofpatientsforwhomsampleswerebankedinTrizolreagent(Invitrogen),whichwasusedtoextractDNAformutationalstudies.Theclinicalcharacteristicsofthepatientswestudied,ascomparedwiththecompleteE1900trialcohort,areprovidedinTableS1intheSupplementaryAppendix,availablewiththefulltextofthisarticleatNEJM.org.Themedianfollow-uptimeforthepatientsincludedintheanalysis,calculatedfromthetimeofrandomizationforinductiontherapy,was47.4months.Cytogeneticanalysis,fluorescenceinsituhybridization,andreverse-transcriptase–polymerase-chain-reaction(RT-PCR)assaysforrecurrentcytogeneticlesionswereperformedasdescribedinitiallybySlovaketal.6andasusedpreviously,17withcentralreviewbytheECOGCytogeneticSubcommittee.我們對從腦電圖e1900試驗患獲得的診斷樣本進行了突變分析。所有患者提供書面知情同意書。試驗組（398例）包括在e1900所有患者可用冷凍細胞允許DNA提取和突變分析。驗證組（104例）組成的第二組病人的樣品儲存在Trizol試劑（Invitrogen公司），也是用來提取DNA作突變分析。我們研究的患者的臨床特征，如與完整的e1900試驗相比，在附錄中有表S1可瀏覽詳細信息。納入分析患者的中位隨訪時間為從隨機的誘導治療的時間計算，為47.4個月。如最初由斯洛伐克等人所進行的研究，由皮層細胞遺傳學中心審查委員會審查，通過染色體核型分析，熒光原位雜交，逆轉錄-聚合酶鏈反應（RT-PCR）進行細胞遺傳學損傷檢測。MutationalAnalysisThesourceoftheDNAwasbonemarrowinthecaseof55.2%ofthesamples(277of502)andperipheralbloodinthecaseof44.8%(225of502).WesequencedtheentirecodingregionsofTET2,ASXL1,DNMT3A,CEBPA,PHF6,WT1,TP53,EZH2,RUNX1,andPTENandtheregionsofpreviouslydescribedmutationsforFLT3,NPM1,HRAS,KRAS,NRAS,KIT,IDH1,andIDH2.ThegenomiccoordinatesandsequencesofalltheprimersusedinthisstudyareprovidedinTableS2intheSupplementaryAppendix.PairedremissionDNA(i.e.,DNAfrompatientswhohadacompletere-missionafterinductionchemotherapy)wasavail-ablefrom241ofthe398participantsinthetestcohortandfrom65ofthe104inthevalidationcohort.DataonvariantsthatcouldnotbevalidatedasbonafidesomaticmutationsowingtounavailableremissionDNAandtheabsenceofreportsofthemutationsinthepublishedliteratureofsomaticmutationswerecensoredwithrespecttomutationalstatusforthatspecificgene.FurtherdetailsofthesequencingmethodsareavailableintheSupplementaryAppendix.DNA來自樣本中的55.2%例骨髓（277/502）和44.8%例的外周血（225/502）。我們測序整個編碼區的TET2，ASXL1，DNMT3A，CEBPA，PHF6，WT1基因，TP53，EZH2，RUNX1，和PTEN和先前描述的突變NPM1，FLT3，HRAS，KRAS，NRAS，KIT，IDH1和IDH2基因。基因組坐標和所有本研究中所用的引物序列在附錄中表S2詳細列出。對配對緩解DNA（即，有完整的誘導緩解后化療患者DNA）是從241的398的參與者在測試隊列和從65的104在驗證隊列獲取。對變異不能由無法緩解的DNA體細胞突變驗證，在已發表的文獻中的基因突變的報道認為沒有通過對特定基因的突變狀態方面的數據。進一步測序方法的細節在附錄中詳細列舉。StatisticalAnalysisThemutualexclusivityofpairsofmutationswasevaluatedwiththeuseoftwo-by-twocontingencytablesandFisher’sexacttest.Theassociationbetweenmutationsandcytogeneticriskclassificationwastestedwiththeuseofthechi-squaretest.HierarchicalclusteringwasperformedwiththeuseoftheLance–Williamsdissimilarityformulaandthecomplete-linkagealgorithm.Survivaltimewasmeasuredfromthedateofrandomizationtothedateofdeathforpatientswhodiedandtothedateofthelastfollow-upforthosewhowerealiveatthetimeoftheanalysis.SurvivalprobabilitieswereestimatedwiththeuseoftheKaplan–Meiermethodandwerecom-paredbetweenpatientswithamutationandthosewithoutmutantallelesbymeansofthelog-ranktest.MultivariateanalyseswereconductedwiththeuseoftheCoxmodelwithforwardse-lection.Wecheckedtheproportional-hazardsassumptionbytestingforanonzeroslopeinaregressionofthescaledSchoenfeldresidualsonfunctionsoftime(TableS3intheSupplementaryAppendix).Whennecessary,suchasintheanalysesperformedinvarioussubsets,theresultsoftheunivariateanalyseswereusedtoselectthevariablestobeincludedintheforwardvariablesearch.Finalmultivariatemodelsinformedthedevelopmentofnovelrisk-classificationrules.Whensoindicated,Pvalueswereadjustedtocontrolthefamily-wiseerrorratewiththeuseofthecompletenulldistributionapproximatedbyresamplingobtainedthroughthePROCMULTTESTprograminSASorthemulttestlibraryinR.19Theonlyexceptionwastheadjustmentintestsoftheeffectofmutationsontheresponsetotheinductiondose,forwhichastep-downHolmprocedurewasusedtocorrectformultipletesting.AllanalyseswereperformedwiththeuseofSASsoftware,version9.2(),andtheRstatisticalpackage,version2.12()兩對共同雙突變分析由2*2列聯表和Fisher精確檢驗進行分析。基因突變和細胞遺傳學風險分類之間的關聯性采用卡方檢驗進行分析。層次聚類采用Lance-Williams法和complete-linkage算法進行。生存時間測定日期的采用隨機抽取患者死亡日期和最后隨訪的日期之差計算取得。生存概率采用Kaplan-Meier法進行估計和患者的突變和未突變的等位基因的比較采用log-rank檢驗。多變量分析采用Cox模型的前進法進行。我們Schoenfeld殘差非零時間函數回歸規模對比例風險進行假設檢驗(附錄中表S3)。必要時，如對不同亞組進行處理，通過單因素分析選擇有意義的變量。最后多變量模型采用新的風險分類規則進行分析。當如此表示，P值進行調整以控制總Ⅰ類錯誤率與通過SAS的PROCMULTTEST程序或R數據庫的multtest程序的重復采樣完整近似零假設分布的。ResultsFrequencyofGeneticAlterations基因修飾頻數Somaticalterationswereidentifiedin97.3%ofthepatients.Figure1showsthefrequencyofsomaticmutationsintheentirecohortandtheinterrelationshipsamongthevariousmutations,asrepresentedvisuallywiththeuseofaCircosplot.DataforallmolecularsubsetsareprovidedinFiguresS1andS2andTablesS4andS5intheSupplementaryAppendix.Inparticular,mutationalheterogeneitywasgreaterinpatientswithintermediate-riskAMLthaninpatientswithfavorable-riskorunfavorable-riskriskAML(P=0.01)(Fig.S2DintheSupplementaryAppendix).97.3%的病人發生體細胞修飾。根據Circos圖，圖1表明整個研究隊列的體細胞突變頻數以及各種突變間的相互關系。所有分子亞組的數據由附錄中圖S1、S2和表S4、S5進行表述。尤其是，中度風險AML病人的突變異質性比低度風險和惡性風險病人高很多(P=0.01)(附錄中圖S2D)。MutationalComplementationGroups突變互補基因群Integratedmutationalanalysisallowedustoidentifyfrequentlyco-occurringmutationsandmutationsthatweremutuallyexclusiveintheE1900patientcohort(TableS6intheSupplementaryAppendix).Inadditiontonotingfrequentco-occurrenceofKITmutationswithcore-binding–factoralterationst(8;21)andinv(16)/t(16;16),wefoundsignificantco-occurrenceofIDH1andIDH2mutationswithNPM1mutationsandofDNMT3AmutationswithNPM1,FLT3,andIDH1alleles(P<0.001forallcomparisons)(TableS7intheSupplementaryAppendix).WerecentlyreportedthatIDH1andIDH2mutationsweremutuallyexclusivewithTET2mutations20;detailedmutationalanalysisrevealedthatIDH1andIDH2mutationswerealsomutuallyexclusivewithWT1mutations(P<0.001)(Fig.S3andTableS8intheSupplementaryAppendix).WealsoobservedthatDNMT3AmutationsandMLLtranslocationsweremutuallyexclusive(P<0.01).整合的基因突變輪廓分析讓我們多次辨別共同發生的突變和在E1900病人隊列中發生的相互排斥的突變（附錄表S6）。另外，為了記錄core-binding–factor修飾t(8;21)和inv(16)/t(16;16)的KIT突變的經常發生的共出現頻數，我們發現IDH1andIDH2的突變和NPM1突變具有顯著的共同出現差異，DNMT3A突變與NPM1,FLT3,andIDH1等位基因有顯著的共同出現差異（與所有對照比較P<0.001）（附錄表S7）。最近我們報到了IDH1和IDH2突變與TET2相互排斥，進一步的突變分析證實IDH1與IDH2突變也與WT1突變相互排斥（P<0.001）（附錄圖S3和表S8）.也同時發現DNMT3A與MLL翻譯相互排斥（P<0.001）。MolecularDeterminantsofOverallSurvival總生存的分子決定因子Univariateanalysisrevealed,aspreviouslydescribed,21,22thatFLT3internaltandemduplication(FLT3-ITD)mutationsandMLLpartialtandemduplication(MLL-PTD)mutationswereassociatedwithreducedoverallsurvival(P=0.001forFLT3-ITDandP=0.009forMLL-PTD)(TableS9intheSupplementaryAppendix),whereasCEBPAmutationsandcore-binding–factoralterationst(8;21)andinv(16)/t(16;16)wereassociatedwithimprovedoverallsurvival(P=0.05forCEBPAandP<0.001forthecore-binding–factoralterations).2,23Inaddition,PHF6andASXL1mutationswereassociatedwithreducedoverallsurvival(P=0.006forPHF6andP=0.05forASXL1)(Fig.S4intheSupplementaryAppendix).IDH2mutationswereassociatedwithanimprovedrateofoverallsurvivalintheentiretestcohort(3-yearrate,66%;P=0.01)(Fig.S5intheSupplementaryAppendix).ThefavorableeffectofIDH2mutationswasfoundexclusivelyinpatientswithIDH2R140Qmutations(P=0.009)(Fig.S5intheSupplementaryAppendix).Allthefindingsintheunivariateanalysiswerealsosignificantinthemultivariateanalysis(P<0.05,withadjustmentforage,white-cellcount,transplantationstatus[didvs.didnotundergostem-celltransplantation],andcytogeneticcharacteristics)(TableS9intheSupplementaryAppendix),withtheexceptionofthefindingsforMLL-PTD,PHF6,andASXL1mutations.KITmutationswereassociatedwithreducedoverallsurvivalamongpatientswhowerepositiveforthet(8;21)core-binding–factoralteration(P=0.006)butnotamongpatientswiththeinv(16)/t(16;16)alteration(P=0.19)(Fig.S6intheSupplementaryAppendix).正如先前描述，單變量分析顯示，FLT3內在銜接復制(FLT3-ITD)突變和MLL內在銜接復制(MLL-PTD)突變與減少總生存相關(P=0.001forFLT3-ITDandP=0.009forMLL-PTD)(附錄表S9)，而CEBPA突變和core-binding–factor修飾t(8;21)andinv(16)/t(16;16)與改善總生存相關(P=0.05forCEBPAandP<0.001forthecore-binding–factor修飾)。另外，PHF6和ASXL1突變與減少總生存相關(P=0.006forPHF6andP=0.05forASXL1)(附錄圖S4)，IDH2突變與整個檢驗隊列提高總生存率相關(3年生存率66%;P=0.01)(附錄圖S5)。IDH2突變的有利效應僅在有IDH2R140Q突變的病人出現(P=0.009)(附錄圖S5)。除了MLL-PTD,PHF6,andASXL1的突變結果，單變量分析中所有的結果再多因素分析中也是有統計學意義(P<0.05,調整年齡、白細胞計數、移植情況[有vs沒有干細胞移植],、細胞遺傳特征)（附錄表S9）。在出現core-binding–factor改變的陽性病人中，KIT突變與降低總生存相關(P=0.006)，而與nv(16)/t(16;16)改變的病人無相關性(P=0.19)（附錄圖S6）PrognosticValueofMolecularAlterationsinIntermediate-RiskAMLAML中度風險患者的分子修飾的預后價值Amongpatientswithintermediate-riskAMLasdefinedbycytogeneticanalysis(TableS10intheSupplementaryAppendix),FLT3-ITDmutationswereassociatedwithreducedoverallsurvival(P=0.008),afindingthatisconsistentwiththeresultsofpreviousstudies.21ASXL1andPHF6mutationswereassociatedwithreducedsurvival,andIDH2R140Qmutationswithimprovedsurvival,amongpatientswithintermediate-riskAML(TableS10intheSupplementaryAppendix),aneffectsimilartothatintheentirecohort.Inaddition,wefoundthatTET2mutationswereassociatedwithreducedoverallsurvivalamongpatientswithintermediate-riskAML(P=0.007)(Fig.S7intheSupplementaryAppendix).通過細胞遺傳學分析定義的AML中度風險患者，FLT3-ITD突變與減少總生存相關(P=0.008)，該發現與早期研究一致。在具有中度風險的AML患者中，ASXL1和PHF6突變與降低生存相關，IDH2R140Q與改善生存相關（附錄表S10），改結果與整個隊列研究結果相似。MultivariateanalysisrevealedthatFLT3-ITDmutationsconstitutedtheprimarypredictorofoutcomeinpatientswithintermediate-riskAML(adjustedP<0.001).AsubsequentmultivariateanalysisaccordingtoFLT3-ITDstatusshowedthatinpatientswithwild-typeFLT3-ITD,mutationsinTET2,ASXL1,PHF6,andMLL-PTDwereindependentlyassociatedwithanadverseout-come.Patientswithintermediate-riskAMLwhohadbothNPM1andIDH1orIDH2mutationshadanimproved3-yearrateofoverallsurvival,ascomparedwithpatientswhohadmutantNPM1andbothwild-typeIDH1andwild-typeIDH2(89%vs.31%,P<0.001)(Fig.S8intheSupplementaryAppendix).Wethenclassifiedpatientswithintermediate-riskAMLwhohadwild-typeFLT3-ITDintothreecategories,withmarkeddifferencesinthe3-yearrateofoverallsurvival(adjustedP<0.001):patientswithIDH1orIDH2mutationsandNPM1mutations(overallsurvival,89%);patientswithTET2,ASXL1,PHF6,orMLL-PTDmutations(overallsurvival,6.3%);andpatientswithwild-typeTET2,ASXL1,PHF6,andMLL-PTD,withoutco-occurringIDHorNPM1mutations(overallsurvival,46.2%)(Fig.2A).Similarresultswereobtainedwhentheanalysiswasrestrictedtopatientswithanormalkaryotype(Fig.S9AintheSupplementaryAppendix).多因素分析證實FLT3-ITD突變構成AML中度風險患者結局的主要因素(調整P<0.001)。依據FLT3-ITD狀況，一些列多因素分析表明，在FLT3-ITD野生型的患者中，TET2,ASXL1,PHF6,和MLL-PTD的突變分別與不利結局相關。與具有NPM1突變和同時具有IDH1野生型和IDH2野生型的患者比較，同時具有NPM1與IDH1或IDH2突變的AML中度風險患者具有改善的三年生存期(89%vs.31%,P<0.001)(附錄圖S8)。然后，我們對具有FLT3-ITD野生型AML中度風險患者分成三類，其三年生存率具有顯著差異(調整P<0.001)：同時具有IDH1或IDH2突變和NPM1突變患者（總生存率89%）；具有TET2,ASXL1,PHF6,orMLL-PTD突變患者（總生存率6.3%）；具有野生型TET2,ASXL1,PHF6,andMLL-PTD,而非同時出現IDH和NPM1的患者（總生存率46.2%）。與正常核心患者的分析結果相似（附錄圖S9A）。Inpatientswithintermediate-riskAMLwhohadmutantFLT3-ITD,wefoundthatCEBPAmutationswereassociatedwithanimprovedout-comeandthattrisomy8andTET2,DNMT3A,andMLL-PTDmutationswereassociatedwithanadverseoutcome.Weusedthesedatatoclassifypatientswithintermediate-riskAMLwhohadmutantFLT3-ITDintothreecategories.Thefirstcategoryincludedpatientswithtrisomy8orTET2,DNMT3A,orMLL-PTDmutations,whichwereassociatedwithanadverseoutcome(3-yearrateofoverallsurvival,14.5%);thisrateofsurvivalwassignificantlylowerthantheratesamongpatientsinthesecondcategory,thosewithwild-typeCEBPA,TET2,DNMT3A,andMLL-PTD(overallsurvival,35.2%;P<0.001),andpatientsinthethirdcategory,thosewithCEBPAmutations(overallsurvival,42%;P<0.001)(Fig.2B).Therateofsurvivalamongpatientswithintermediate-riskAMLwhohadmutantFLT3-ITDandwild-typeCEBPA,TET2,DNMT3A,andMLL-PTDdidnotdiffersignificantlyfromtherateamongpatientswithmutantFLT3-ITDandmutantCEBPA(P=0.34),suggestingthatthepresenceofmutationsassociatedwithanunfavorable-riskprofilemorepreciselyidentifiespatientswithmutantFLT3-ITDwhowillhaveadverseoutcomesofAMLthandoestheabsenceofCEBPAmutationsalone.ThesesamethreeriskcategoriesalsohadsignificantprognosticvalueinpatientswithAMLwhohadmutantFLT3-ITDandanormalkaryotype(Fig.S9BintheSupplementaryAppendix).具有FLT3-ITD突變型基因AML中度風險患者，我們研究發現CEBPA突變與改善的結局相關，三體性8的TET2,DNMT3A與MLL-PTD突變與不利結局相關。利用這些數據，我們把具有FLT3-ITD突變型的AML中度風險患者分為三類。第一類包括三倍性8或TET2,DNMT3A或MLL-PTD突變型，與不利結局相關(三年生存率14.5%)；這一生存率明顯低于第二類、第三類患者的生存率，第二類表現為野生型CEBPA,TET2,DNMT3A,andMLL-PTD(總生存率35.2%;P<0.001)，第三類患者表現為CEBPA突變(總生存率42%;P<0.001)(圖2B)。AML中度風險患者中具有突變株FLT3-ITD和野生型CEBPA,TET2,DNMT3A,和MLL-PTD與具有突變型FLT3-ITD和CEBPA的患者明顯沒有差別(P=0.34)，表明。對于具有FLT3-ITD突變型和正常核型AML患者，同樣的三類危險因素分類有顯著的預后價值（附錄圖S9B）。PrognosticSchemawithIntegratedMutationalandCytogeneticProfiling整體突變和細胞遺傳分析的預后模式Theseresultsallowedustodevelopaprognosticschemathatintegratedourfindingsfromthecomprehensivemutationalanalysiswithcytogeneticdatatoidentifythreeriskgroups:agroupwithafavorable-riskprofile(mediansurvival,notreached;3-yearrateofoverallsurvival,64%),agroupwithanintermediate-riskprofile(mediansurvival,25.4months;3-yearrateofoverallsurvival,42%),andagroupwithanadverse-riskprofile(mediansurvival,10.1months;3-yearrateofoverallsurvival,12%)(Fig.3Aand3B,andTableS11intheSupplementaryAppendix).Inmultivariateanalysis,themutationalprognosticschemapredictedtheoutcomeindependentlyofage,white-cellcount,inductiondose,andtransplantationstatus(adjustedP<0.001).Ourclassificationheldtrueregardlessofthetypeofpost-remissiontherapy(autologousorallogeneictransplantationorconsolidationchemotherapyalone)(Fig.S10intheSupplementaryAppendix).這些研究結果讓我們形成預后模式，根據細胞遺傳數據的綜合突變分析的整合結果，可以分為三類危險因素組：一是有利風險輪廓組(中位生存期，沒有達到;三年生存率64%)，二是中位風險輪廓組(中位生存期25.4月;三年生存率42%)，三是不利風險輪廓組(中位生存期10.1月，三年生存率12%)（附錄圖3A,3B，表S11）。多元統計分析，突變預測模式預測結局的獨立因素，包括年齡、白細胞計數、誘導計量、移植狀態(調整P<0.001)。不管術后緩解治療如何，我們的分類是真實有效的（自體或異體移植或單獨強化化療）（附錄圖S10）。Giventhenumberofvariablesinourprognosticclassification,wetestedthereproducibilityofthispredictorinanindependentcohortof104patientsfromtheECOGE1900trial.MutationalanalysisofthevalidationcohortconfirmedthereproducibilityofourprognosticschemaforpredictingtheoutcomeinpatientswithAML(adjustedP<0.001)(Fig.3C).Thepredictivevalueofthemutationalprognosticschemawasindependentofriskwithrespecttotreatment-relateddeath(definedasdeathwithin30daysafterinitiationoftreatment)orlackofresponsetoinductionchemotherapy(i.e.,lackofacompleteremission)inthetestcohortandinthecombinedtestandvalidationcohorts(TableS12intheSupplementaryAppendix).根據預后變量的數目，從ECOGE1900試驗患者獨立選擇104名患者的數據，我們檢驗了這些預測指標的重現性。有效隊列的突變分析證實用來預測AML患者結局的預后預測模式(調整P<0.001)(圖.3C)。，突變預后模式的預測值獨立于有關危險因素，包括治療相關死亡（誘導治療三十天內死亡）或再檢驗隊列中缺少誘導治療響應（如，缺少緩解信息）或綜合試驗或合法隊列（附錄表S12）。GeneticPredictorsofResponsetoInductionChemotherapy誘導治療反應的基因預測RecentstudieshaveshownthatmutantDNMT3AwasassociatedwithadverseoutcomesinpatientswithAML.4,14However,wefoundthatDNMT3AmutationswerenotassociatedwithadverseoutcomesintheECOGE1900cohort(Fig.4A)(P=0.15).IntheECOGE1900trial,patientswererandomlyassignedtoinductiontherapywithcytarabinepluseither45mgofdaunorubicinpersquaremeteror90mgofdaunorubicinpersquaremeter.17Wethereforehypothesizedthathigh-dosedaunorubicinimprovedtheoutcomesinpatientswithdaunorubicinwhohadDNMT3Amutations.Indeed,wefoundthatDNMT3Amutationalstatushadasignificanteffectontheout-comewithdose-intensivechemotherapy(Fig.4B)(P=0.02).WethenassessedtheeffectsofDNMT3Amutationalstatusontheoutcomeaccordingtotreatmentgroupandfoundthathigh-dosedaunorubicinwasassociatedwithanimprovedrateofsurvivalamongpatientswithmutantDNMT3A(P=0.04)(Fig.S11AintheSupplementaryAppendix)butnotamongpatientswithwild-typeDNMT3A(P=0.15)(Fig.S11BintheSupplementaryAppendix).Inaddition,univariateanalysisrevealedthatdose-intensifiedinductiontherapywasassociatedwithanimprovedoutcomeinpatientswithAMLwhohadMLLtranslocations(P=0.01;P=0.06withadjustmentformultipletesting)(Fig.S11CandS11DintheSupplementaryAppendix)andinthosewhohadNPM1mutations(P=0.01;P=0.10withadjustmentformultipletesting)(Fig.S11EandS11FandTableS13intheSupplementaryAppendix).BecausetheadjustedPvaluesforNPM1mutationsandMLLtranslocations(P≤0.10)areclosetostatisticalsignificance,theyshouldbestudiedfurtherinprospectivetrials.最近的研究表明，突變型DNMT3A與AML患者不利結局相關。但是，我們也發現DNMT3A的突變與ECOGE1900隊列中的不利結局不相關(圖4A)(P=0.15)。在ECOGE1900隊列試驗中，患者隨機分配接受兩種誘導治療，一種是阿糖胞苷加45mg柔紅霉素/平方米，一組是阿糖胞苷加90mg柔紅霉素/平方米。因此，我們假設高劑量柔紅霉素組能夠改善帶有DNMT3A突變患者生存。確實，我們發現DNMT3A突變狀態對劑量敏感性的化療結局確實有影響(圖4B)(P=0.02)。我們根據治療分組結局，評價DNMT3A的突變狀態效果，發現高劑量柔和霉素和具有DNMT3A突變患者改善的生存率相關(P=0.04)(附錄圖.S11A)，而和具有DNMT3A野生型患者改善的生存率不相關(P=0.15)(附錄圖.S11B)。另外，單因素分析顯示，劑量強化誘導治療與發生混合性白血病淋巴轉移AML患者的結局改善相關(P=0.01;P=0.06調整的多重分析)(附錄圖S11C、S11D)，與NPM1突變的AML患者的結局改善相關(P=0.01;P=0.10調整的多重分析)(附錄圖S11E、S11F和表S13)。因為NPM1突變和MLL轉移的調整P值接近檢驗水準，必須在將來的后續試驗中做進一步研究。Wethenseparatedthepatientsinourcohortintotwogroups:patientswithmutationsinDNMT3AorNPM1orwithMLLtranslocationsandpatientswithwild-typeDNMT3AandNPM1andnoMLLtranslocations.Dose-intensiveinductiontherapywasassociatedwithamarkedimprovementintherateofsurvivalamongpatientswhowerepositiveforDNMT3AorNPM1mutationsorMLLtranslocations(P=0.001)(Fig.4C)butnotamongpatientswithwild-typeDNMT3AandNPM1andnoMLLtranslocations(P=0.67)(Fig.4D).Thisfindingwasindependentoftheclinicalcovariatesofage,white-cellcount,andstatuswithrespecttotransplantation,treatment-relateddeath,andresponsetochemotherapy(adjustedP=0.008andP=0.34forpatientswithmutantandwild-typegenes,respectively),suggestingthathigh-doseanthracyclinechemotherapyprovidesabenefitingeneticallydefinedsubgroupsofpatientswithAML.我們把進入隊列的患者分成兩組：一組是具有DNMT3A或NPM1突變，或具有MLL轉移的患者，一組是又有DNMT3A和NPM1野生型，沒有發生MLL轉移的患者。強化劑量誘導治療與DNMT3A陽性或者NPM1突變或MLL轉移患者生存率的顯著改善相關(P=0.001)(圖.4C)，與有DNMT3A和NPM1野生型，沒有發生MLL轉移的患者沒有相關(P=0.67)(Fig.4D)。這一發現獨立于一些臨床協變量，如年齡、白細胞計數、移植反應狀態、治療相關死亡、化學療法的反應（調整P=0.008、P=0.34分別和突變與野