2 个体化医学检测质量保证和标准化：问题与思考 李金明

个体化医学检测质量保证和标准化：问题与思考李金明卫生部临床检验中心全国EGFR突变室间质量评价调查：2012年7月-8月正式室间质评：2014年全国KRAS突变室间质量评价调查：2013年7月-8月正式室间质评：2014年全国法华林用药基因多态性室间质量评价调查：2014年5月正式室间质评：2015年全国Her2扩增检测室间质量评价调查：2014年7月-8月全国BRAF突变室间质量评价正式质评项目2015年室间质量评价2013年KRAS突变质评总体的检测情况结果全部正确的实验室35家（35/63）ZhangR,HanY,HuangJ,MaL,LiYulong,LiJ.ClinChemLabMed.2014Jun194Zhanget al.:PTforKRASmutationtestinginChinaidentifiedtheG13D,G12A,andG12Dmutations.Intotal,69errorswerereported;39.1%ofthesereportedmorethanonemutation,ofwhichonewascorrect.Therewerefre-quentinstancesoflaboratoriesreportingthewrongmuta-tion(27.5%)orreportingmutationinwild-typesamplesornomutationinmutantsamples(26.1%).Only7.2%oftheerrorswerecausedbydetectionfailure.ProficienciesinKRASmutationdetectionareshowninTable2.Only55.6%ofrespondentscorrectlyidentifiedallthemutations.Thebestresults(correctidentificationofallmutations)wereprovidedbylaboratoriesusingSangersequencing(73.7%),pyrosequencing(80%),andMALDI-TOF-MS(80%).Incontrast,only39.2%oflabo-ratoriesusingARMSand25%oflaboratoriesusingHRMgenerated100%proficientresults.Themajorityoferrorswerefalse-positiveor-negativeresults.Fifteenfalse-negativeresults[15/504,(3.0%)]werereported.Allmutationsweredetectedbymorethan90%ofthelaboratorieswithorwithoutfalse-positiveresults,withtheexceptionofKRAS-06,whichcarriestheG12Smutation[53/63,(84.1%)].Twenty-oneofthelaborato-riesreported45false-positiveresultsincluding52false-positivemutations;ofthese,16laboratoriesreportedfalse-positiveresultsinmorethanonesample(Table1).Twelveof14laboratoriesusingthecommercialAmoyDxtestreportedfalse-positives(Table3).False-positiveswerereportedbylaboratoriesusingARMS-AmoyDx,in-houseSangersequencing[10],andpyrosequencing[3].In-houseandcommercialARMSassays,whichwereperformedbyonlyoneorafewlaboratories,yieldednofalse-positiveresults.14.3%6.3%ABC4.8%7.9%7.3%7.3%2.4%4.5%22.7%9.1%27.3%31.8%4.5%51.2%31.7%66.7%QIAGENARMSARMSSangersequencingSangersequencingPyrosequencingPyrosequencingPCR-HRMPCR-HRMPCR-luminexMALID-TOFmassspectrometryPNA-PCROMEGATIANGENOthersNotreportedFigure1CommercialkitsusedforDNAextractionandmethodsusedforKRAStestinginhospitalsandthecommerciallaboratories(A)compositionofcommercialnucleicacidextractionkitsapplied;(B)themethodsutilizedbytheparticipantsinhospitals;(C)themethodsutilizedbythecommerciallaboratories.Table2Proficiencyresultsofeachassay.KRAStestingassaysNo.ofdatasetsNo.ofproficientdatasets100%proficient80%–99%proficient80%proficientNotproficientAllassays63351774ARMS-AmoyDx140860ARMS-ACCB54100ARMS-Yuanqi44000ARMS-MicroDiag20101ARMS-Gpmedical11000ARMS-QIAGEN11000In-houseARMS11000In-houseSangersequencing1914401Pyrosequencing54001MALDI-TOF-MS54100HRM-MicroDiag31011In-houseHRM10100PCR-luminex11000PNA10100100%proficient:allmutationsdetectedcorrectly.80%–99%proficient:80%–99%ofmutationsdetectedcorrectly.80%proficient:60%–80%ofmutationsdetectedcorrectly.Notproficient:60%ofmutationsdetectedcorrectly.KRAStestingperformanceWide-typesamplesKRAS-04andKRAS-10weremostoftencorrectlyidentified(95.2%and98.4%,respec-tively)(Table1).Morethan90%oflaboratoriescorrectlyBroughttoyouby|NewYorkUniversityAuthenticated|175.30.146.143DownloadDate|8/24/147:27AMZhanget al.:PTforKRASmutationtestinginChina5DiscussionArtificialFFPEsampleswerecreatedfromculturedcelllinestoconstructaproficiencypanelcontainingsevencommonKRASmutationsincodons12and13.Dijkstraetal.preparedsampleswith0%,2.5%,5%,10%,and15%mutatedcellsandassessedthereliabilityofaseriesofcommonlyusedmethods[13].However,ourstudyaimedprimarilytoevaluateKRASgenotyping;thus,onlymutatedcellswereusedandallthepercentagesofmutantKRASallelesweredeterminedtobe50%byMALDI-TOF-MS.ForDNAsequencing,thepercentageofmutantKRASallelesshouldbeatleast10%–15%[16].Inthiscontext,weconcludedthattheKRASmutationabundanceinPTsamplesshouldbehigherthanthedetectionlimitsforallmethods.ThePTresultssuggestthatonly55.6%oflaborato-riesarecapableofcorrectlyidentifyingallKRASmuta-tions.Undernormalconditions,false-negativeresultsmayoccurduetolowsensitivityforthechosenanalyticalmethod.Inthisstudy,afewfalse-negativeresults(3.0%)werereportedforspecimenswiththepercentageofmutantKRASalleles50%;theseresultsweregeneratedbyARMS,Sangersequencing,pyrosequencing,MALDI-TOF-MS,HRM,andPNAmethods,indicatingthatmethodsensitivitymightsignificantlydifferbetweenlaboratories.KRASmutationscanbemissedevenbythemostsensitivemethodsiftheproceduresarenotperformedcorrectly.Surprisingly,therewerealargenumberoffalse-positiveresults,withonly42of63datasetsbeing100%specific.Thefalse-positiveratewashigherthanreportedintheEQAsconductedinEurope[18],Italy[23],andtheUK[14].TheproficiencypanelcontainedonlytwoKRASmutation-negativesamples;however,eachKRAS-mutatedsamplecouldberegardedasthenegativecontrolforothermutations.Althoughsomefalse-positiveresultsmightnotchangetheclinicalconsequence,allfalse-positiveresultswerescoredasfailuresinthisstudy.Table3False-negativeand-positiveresultsbyassays.KRAStestingassaysNo.ofdatasetsNo.offalse-negativeresults/totalno.ofpositiveresults(%)No.ofmutationdetectionfailur