决策树是通过递回分割.ppt

AFive-GeneSignatureandClinicalOutcomeinNonSmall-CellLungCancer,From:nengljmed356;1january4,2007By:Hsuan-YuChen,M.Sc.,Sung-LiangYu,Ph.D.,etalReporter:R6謝廣宇,Background,Currentstagingmethodsareinadequateforpredictingoutcomedevelopeda5-genesignaturenonsmall-celllungcancer(NSCLC)mostcommoncauseofdeathfromcancerworldwiderelapseratewithearly-stageNSCLCis40%within5yearsafterpotentiallycurativetreatment,決策樹是通過遞迴分割（recursivepartitioning）建立而成，遞迴分割是一種把資料分割成不同小的部分的疊代過程,cDNAmicroarrayschema,cDNA晶片製造原理,Microarrayanalysis,OperationPrinciple:Samplesaretaggedwithflourescentmaterialtoshowpatternofsample-probeinteraction(hybridization)Microarraymayhave60Kprobe,Gene-expressionprofilingbymeansofmicroarraysandreverse-transcriptasepolymerasechainreaction(RT-PCR)examinedgeneexpressionin125surgicalspecimensofNSCLC,usingmicroarraysandreal-timeRT-PCRtoidentifyagenesignaturecorrelatedwithclinicaloutcome,Methods,computer-generatedrandomnumberstoassignspecimensfrom185consecutivepatientsformicroarrayanalysisfrozenspecimensoflung-cancertissuefrom125randomlyselectedpatientswhounderwentsurgicalresectionofNSCLCatTaichungVeteransGeneralHospitalbetweenDecember1999andDecember2003-five-generisk-predictionmodelusinganindependentcohortof60randomlyselectedpatients,125specimens-60wereadenocarcinomas,52weresquamous-cellcarcinomas,and13wereothertypesofcancernotreceivedadjuvantchemotherapy672genesassociatedwithinvasiveactivity,identifiedinapreviousstudy-rearrayedinduplicateonanylonmembrane4goftotalRNAfromeachspecimen,validatelevelsofexpressionofgenesfoundonmicroarrayanalysisRT-PCRwasperformedon16genesandacontrolgeneforTATA-boxbindingprotein(TBP),withuseofspecificTaqManprobesandprimersets-transcriptswereamplifiedwithreagent(TaqManOne-StepRT-PCRMasterMixReagent,AppliedBiosystems)andasequencedetectionsystem(ABIPrism7900HT,AppliedBiosystems),Toreducebackgroundnoisebackgroundintensityvaluesoflessthan3000wereassignedvalueof3000log-transformedtoabase-2scaleGeneswithcoefficientsofvariationoflessthan3%wereexcludedfromfurtheranalysesthegene-expressionintensityvaluesweretransformedtoordinalcodingvaluesbyrankingoflevelofgeneexpressionamongthe485genesin125patients(60,625observations),intensityvaluewascodedas1-2-3-4accordingto0-25%-50%-75%-100%HazardratiosfromunivariateCoxregressionanalysistodeterminewhichgenesassociatedwithdeathfromanycauseorrecurrenceofcancerProtectivegenesweredefinedwithahazardratiofordeathof1,univariateCoxproportional-hazardsregressionanalysisGenessignificantlycorrelatedwithsurvival-alinearcombinationofgene-expressioncodingvaluesweightedbyregressioncoefficientstocalculateariskscoreforeachpatientpatientsriskscorewascalculatedassumoflevelsofexpressionofeachgene,asmeasuredbymicroarrayanalysis,multipliedbycorrespondingregressioncoefficients,high-riskgenesignatureoralow-riskgenesignature,with50thpercentile(median)oftheriskscoreasthethresholdvalue(median,4.9;range,1.3to21.9)-thresholdvaluetoreflectalmosthalfofpatientswithearlystageNSCLCrelapsewithin5yearsafterpotentiallycurativesurgery,eliminateeffectofextremevalueslevelsofexpressionof16genesconfirmedbyRT-PCRandindexedbySpearmansrank-correlationtestfurtheridentified5genessignificantlyassociatedwithsurvival,recursive-partitioningdecisiontreethecontrolgenewasTBP(X54993_s_at)maximumfollow-uptimeforsurvivalanalysisinourstudywas62months,weused5-yearsurvivaldatafor86patients,Results,16genescorrelatedwithdeathfromanycause:4wereprotectivegenes(hazardratio1),dual-specificityphosphatase6(DUSP6)monocyte-to-macrophagedifferentiationassociatedprotein(MMD)signaltransducerandactivatoroftranscription1(STAT1)v-erb-b2avianerythroblasticleukemiaviraloncogenehomolog3(ERBB3)lymphocyte-specificproteintyrosinekinase(LCK),5-genesignaturestronglyassociatedwithoverallsurvival(sensitivity,98%;specificity,93%;positivepredictivevalue,95%;negativepredictivevalue,98%;andoverallaccuracy,96%)AmongpatientswithstageIIdisease,overallsurvivaldidnotdiffersignificantlybetweenhigh-riskandlow-riskgenesignatures,probablyowingtosmallnumberofpatients,Discussion,NSCLCisaheterogeneousdiseasebasedonclinicalandpathologicalfindingsmayhavelimitofusefulnessforpredictingoutcomes,butmolecularmethodsaddvalueuseofmicroarraysinclinicalpracticelimitedbythelargenumberofgenesintheanalysis,complicatedmethods,lackofreproducibilityandindependentvalidationofresults,andneedforfresh-frozentissue,RT-PCRinvolvingasmallnumberofgenesallowsforaccurateandreproduciblequantificationofresultsforRNAobtainedfromsmallamountsofparaffin-embeddedspecimens125frozentumorspecimensfrompatientswithNSCLCrandomlydividedintoatrainingset(63specimens)andatestingset(62specimens),selectionofgenesinmicroarraytrainingsetwasvalidatedinmicroarraytestingset,andpatternsofgeneexpressionfoundonmicroarrayanalysiswerevalidatedbyRT-PCRresultsinourChinesepatientswerealsovalidatedwithuseofasetofpublishedNSCLCmicroarraydatafrompatientsfromaWesternpopulationwithNSCLC,proposewhohavetumorswithahigh-riskgenesignaturecouldbenefitfromCisplatin-basedadjuvantchemotherapy,thosewithalow-riskgenesignaturecouldbespared-large-scale,multicenterstudiesarenecessarytotestthisidea,STAT1arrestedgrowthandapoptosisinmanytypesofcancercellsbyinducingexpressionofp21WAF1andcaspase.MMDexpressedinmaturemacrophages,macrophageactivationpromotescancermetastasis,BUTfunctionofMMDproteinisunknownDUSP6inactivatesextracellularsignal-regulatedkinase2(ERK2)(alsoknownasmitogen-activatedproteinkinase1MAPK1),resultingintumorsuppressionandapoptosis,ERBB3memberofepidermalgrowthfactorreceptorfamilyoftyrosinekinases,canshortencellsurvivalLCKmemberoftheSrcfamilyofproteintyrosinekinases,isexpressedmainlyinTcellsandisoneoffirstsignalingmoleculesdownstreamoftheT-cellreceptor,akeyrolenotonlyindifferentiationandactivati