lung cancer in patients PCR system with digital advanced non-smallcell Plasma epidermal factor receptor growth mutation testing with achip-based Lung Cancer

ContentslistsavailableatScienceDirect

Lung Cancer

j o ur n a l ho me p ag e : w w w . e l s e v i e r . c o m / l o c a t e / l u n g c a n

Plasma epidermal growth factor receptor mutation testing with a chip-based digital PCR system in patients with advanced non-small cell lung cancer

Norimitsu Kasahara

, Hirotsugu Kenmotsu

, Masakuni Serizawa

,

Rina Umehara

, Akira Ono

, Yasushi Hisamatsu

, Kazushige Wakuda

, Shota Omori

, Kazuhisa Nakashima

, Tetsuhiko Taira

, Tateaki Naito

, Haruyasu Murakami

,

Yasuhiro Koh

, Keita Mori

, Masahiro Endo

, Takashi Nakajima

, Masanobu Yamada

, Masatoshi Kusuhara

, Toshiaki Takahashi

aDivisionofThoracicOncology,ShizuokaCancerCenter,Japan

bDrugDiscoveryandDevelopmentDivision,ShizuokaCancerCenterResearchInstitute,Japan

cClinicalTrialCoordinationOffice,ShizuokaCancerCenter,Japan

dDivisionofDiagnosticRadiology,ShizuokaCancerCenter,Japan

eDivisionofDiagnosticPathology,ShizuokaCancerCenter,1007Shimonagakubo,Nagaizumi-chou,Suntou-gun,Shizuoka,411-8777,Japan

fDepartmentofMedicalOncologyandHematology,OitaUniversityFacultyofMedicine,1-1Idaigaoka,Hasamamachi,Yufu,Oita,879-5593,Japan

gDivisionofRespiratoryMedicine,MinamiKyushuNationalHospital,1882Kida,Kajiki-chou,Aira,Kagoshima,899-5293,Japan

hThirdDepartmentofInternalMedicine,WakayamaMedicalUniversity,811-1Kimiidera,Wakayama,Wakayama,641-8509,Japan

iDepartmentofMedicineandMolecularScience,GunmaUniversityGraduateSchoolofMedicine,3-39-15,Showa-machi,Maebashi,Gunma371-8511, Japan

a r t i c l e i n f o

Articlehistory:

Received16July2016 Receivedinrevisedform 17December2016 Accepted2February2017

Keywords:

Liquidbiopsy

Epidermalgrowthfactorreceptormutation DigitalPCR

CirculatingfreeDNA Non-smallcelllungcancer

a b s t r a c t

Objectives:Epidermalgrowthfactorreceptor(EGFR)mutationtestingisacompaniondiagnostictodeter- mineeligibilityfortreatmentwithEGFRtyrosinekinaseinhibitors(EGFR-TKIs)innon-smallcelllung cancer(NSCLC).Recently,plasma-basedEGFRtestingbydigitalpolymerasechainreaction(dPCR),which enablesaccuratequantificationoftargetDNA,hasshownpromiseasaminimallyinvasivediagnostic.

Here,weaimedtoevaluatetheaccuracyofaplasma-basedEGFRmutationtestdevelopedusingchip- baseddPCR-baseddetectionof3EGFRmutations(exon19deletions,L858Rinexon21,andT790Min exon20).

Materials andmethods:Forty-nine patientswith NSCLC harboringEGFR-activating mutationswere enrolled,andcirculatingfreeDNAs(cfDNAs)wereextractedfromtheplasmaof21and28patients beforetreatmentandafterprogressionfollowingEGFR-TKItreatment,respectively.

Results:UsingreferencegenomicDNAcontainingeachmutation,thedetectionlimitofeachassaywas determinedtobe0.1%.Thesensitivityandspecificityofdetectingexon19deletionsandL858Rmuta- tions,calculatedbycomparingthemutationstatusinthecorrespondingtumors,were70.6%and93.3%, and66.7%and100%,respectively,showingsimilarresultscomparedwithpreviousstudies.T790Mwas detectedin43%of28cfDNAsafterprogressionwithEGFR-TKItreatment,butinnocfDNAsbeforethe startofthetreatment.

Conclusion:Thischip-baseddPCRassaycanfacilitatedetectionofEGFRmutationsincfDNAasaminimally invasivemethodinclinicalsettings.

©2017ElsevierB.V.Allrightsreserved.

Abbreviations: ARMS,amplificationrefractorymutationsystem;BEAMing,beads,emulsions,amplificationandmagnetics;cfDNA,circulatingfreeDNA;CI,confidence interval;ddPCR,dropletdigitalpolymerasechainreaction;dPCR,digitalpolymerasechainreaction;EGFR,epidermalgrowthfactorreceptor;EGFR-TKI,epidermalgrowth factorreceptortyrosinekinaseinhibitors;JAK2,Januskinase2;NSCLC,non-smallcelllungcancer;PFS,progression-freesurvival;RR,responserate;SD,standarddeviation.

∗Correspondingauthorat:DivisionofThoracicOncology,ShizuokaCancerCenter,1007Shimonagakubo,Nagaizumi-chou,Suntou-gun,Shizuoka411-8777,Japan.

∗∗ Correspondingauthorat:DrugDiscoveryandDevelopmentDivision,ShizuokaCancerCenterResearchInstitute,1007Shimonagakubo,Nagaizumi-chou,Suntou-gun, Shizuoka411-8777,Japan.

E-mailaddresses:[email protected](H.Kenmotsu),[email protected](M.Serizawa).

1 Bothofthecorrespondingauthorsequallycontributedtothisstudy.

http://dx.doi.org/10.1016/j.lungcan.2017.02.001 0169-5002/©2017ElsevierB.V.Allrightsreserved.

1. Introduction

Epidermalgrowthfactor receptor (EGFR)mutationtesting is essentialfortreatmentdecisionsforpatientswithadvancednon- smallcell lung cancer(NSCLC). EGFRtyrosine kinaseinhibitors (EGFR-TKIs),includinggefitinib,erlotinib,andafatinib,areeffective againstpatientswithNSCLCharboringEGFR-activatingmutations, suchasexon19deletionsortheL858Rmutationinexon21[1–6].

However,mostpatientswithNSCLCtreatedwithEGFR-TKIseven- tuallyacquireresistance. TheT790MmutationinEGFR exon20 causesapproximately50%ofacquiredresistancetoEGFR-TKIsin patients[7,8].Recently,thethird-generationEGFR-TKIosimertinib, whichspecificallytargetsEGFRT790M,wasapprovedforusein somecountriesincludingtheUSandJapan,andtheimportance ofserialbiopsiesfordetectingT790Misincreasing,assuchinfor- mationdeterminestheappropriatenessofosimertinibtreatment [9,10].However,repeatedbiopsiesaresometimeshighlyinvasive andcanbedifficulttoperformwithoutcomplications[11].

Recently, some studies have reported the efficacy of EGFR- activatingmutation analysiswith circulating free DNA (cfDNA) extractedfromtheplasmaofpatientswithNSCLC[12–21].More- over, EGFR-activating mutation analysis with cfDNA was also approvedasacompaniondiagnosisforselectingpatientseligible fortreatmentwithgefitinibandosimertinibintheEuropeanUnion.

However,sometechnicallimitationsfordetectingEGFRmutations withcfDNAhave beenreported.For example,thequantityand qualityofcirculatingtumor-derivedDNAvarieswidelybetween patients[22].Moreover,thedetectablepercentageofthetumor- derivedDNAfractionincfDNAcanreachaslowas0.01%[23].The digitalpolymerasechainreaction(dPCR),whichenablesaccurate copy-numberquantificationoftarget moleculesfromlow-input DNA,isthoughttobeapromisingtechnologyforovercomingthe abovelimitationsinmutationtestingwithcfDNA[24].Thehigh performanceofdPCRisachievedbycompartmentalizinga sam- pleatthelevelofasingleDNAmoleculebydistributingasample intothousandsof separatePCRreactions. Accordingly,thetotal copynumberoftargetscanbedeterminedbycountingthepos- itiveandnegativepartitions.dPCRplatformsareclassifiedinto3 typesbasedonthecompartmentalizationmethodused,including dropletdigital PCR (ddPCR)[20,25–30],BEAMing (beads,emul- sions,amplificationandmagnetics)PCR[19,20,31],andchip-based dPCR[30,32,33].Plasma-basedEGFRmutationtestingwithddPCR hasbeenevaluatedinmanyinstitutes.However,chip-baseddPCR hasnotbeenwidelyevaluated.

Thenewchip-baseddPCRsystem,QuantStudio3D(QS3D)Digi- talPCRSystem(ThermoFisherScientific,Waltham,MA,USA),was launched.Theconventionalchip-basedBioMarkdPCRsystem(Flu- idigm,SanFrancisco,CA,US)compartmentalizesDNAinto9,180 micropores[32,33],whereastheQS3DDigitalPCRsystemcompart- mentalizesDNAinto20,000micropores,whichiscomparabletothe compartmentalizationabilityoftheddPCRplatform,QX100/QX200 Droplet DigitalPCR System(Bio-Rad, Hercules, CA,US)broadly usedinstudiesforliquidbiopsy[25–29].Moreover,bothPCRand detectionprocessescanbeconsecutivelyperformedwithinaher- meticallysealedreactionchamberintheQS3DDigitalPCRsystem;

however,thisisnotpossiblewiththeddPCRsystem.Thisfeatureof theQS3DDigitalPCRsystemofferstheadvantageofrequiringfewer pipettingprocess thanthose neededforddPCR.Together,these propertiescontributetothereducedriskofcross-contamination.

TheQS3DDigitalPCRsystemrunseachchipindividually,indicat- ingsuitabilityoftheQS3DDigitalPCRsystemforclinicalsettings insmallinstitutionswithoutvastamountsofsamples.In previ- ouslystudies,thisQS3DDigitalPCRsystemhasbeenusedtodetect DNAmutationsinJanuskinase2(JAK2)[34]andcoagulationfac- torVgenes[35],microRNAs[36],andviralDNA[37],butnotEGFR mutations.

Therefore, this study aimed to evaluate plasma-based EGFR mutationanalysisinpatientswithadvancedNSCLC,usingthisnew chip-baseddPCRsystem,QS3DDigitalPCRSystem.

2. Materialandmethods 2.1. Patients

Allpatientsamplesusedinthisstudywerecollectedfromour previousprospectivestudytoevaluateplasmaEGFRmutationtest- ingwiththeRNaseH-dependentPCRandblockingoligo-dependent PCR methods[38].Thepresent studywasdesigned toevaluate theperformanceoftheQS3DDigitalPCRSysteminplasma-based EGFRmutationanalysis,usingcfDNAsamplesarchivedinthatpre- viousstudy[38].To beeligibleforourpreviousstudy,patients needed tohave beendiagnosed with advanced-stage NSCLCor post-operativerecurrenceandtohaveanEGFR-activatingmutation intheirtumors,asconfirmedbytheScorpionARMS(amplification refractorymutationsystem)method[39].In ourpreviousstudy [38],49patientswereenrolledfromOctober2013toMarch2014, andtumorstagingwasevaluatedaccordingtotheSeventhEdi- tionof TNMinLungCancer[40].Clinicalfactorsat thetimeof bloodsamplingwereobtainedfromthepatients’medicalrecords.

Basedonthetimingofbloodsampling,patientsweredividedinto 2groups.Onegroupincluded21patientsenrolledbeforethestart ofEGFR-TKItreatment,andtheothergroupincluded28patients whoshoweddiseaseprogressionafterEGFR-TKItreatment.Writ- teninformedconsentwasobtainedfromallpatients.Thisstudy wasconductedinaccordancewiththeprovisionsoftheDeclara- tionofHelsinkiandwasapprovedbytheInstitutionalReviewBoard ofShizuokaCancerCenter(approvalnumber#20-50-27-2-3).

2.2. Chip-baseddPCR

PlasmasamplesandDNAextractionaredescribedintheSupple- mentaryMaterialsandMethods.dPCRwasperformedontheQS3D DigitalPCRSystem(ThermoFisherScientific),whichwascomposed ofaProFlexPCRsystem,aQS3DdigitalPCRchiploader,andaQS3D chipscanner(allfromThermoFisherScientific).

TodetectthemostcommondeletionsinEGFRexon19,apair ofprimersand2TaqManprobesweredesigned,basedonYung’s report[32](Supplementary Fig.1).Asareferenceprobe,a VIC- labeledprobewasdesignedagainstaregionneighboringtheexon 19deletionsitesthatlacksreportedmutations.Awild-type-specific FAMprobewasdesignedagainsttheregionwheremostexon19 deletionsoccur.Insamplesharboringexon19deletions,theFAM probecannotanneal;thus,onlyVICfluorescenceisdetected.The sequencesoftheamplificationprimersandTaqManprobesused aregiveninSupplementaryTable1.TodetecttheL858RandT790M mutations,predesignedTaqManprobeandprimersets,AHRSRSV (EGFR6224)andAHRSROS(EGFR6240),respectively,werepur- chasedfromThermoFisherScientific.

2.3. Assessmentofthedetectionlimitforeachmutation

Thedetectionlimitofeachassaywasdefinedasthelowesttarget concentrationthatcouldbespecificallydetected(i.e.,withnofalse- positivedetectionintheabsenceofthetarget)andwasdetermined using20ngsampleoneachof2chipscontainingHDxReference StandardDNA(HorizonDiscovery,Cambridge,UK),whichwasval- idatedusingaddPCRsystemofBio-Rad.Thereferencestandard DNAsusedincludedHD251E746-A750(registeredintheCOS- MICdatabaseasthemostfrequentexon19deletion[http://cancer.

sanger.ac.uk/cancergenome/projects/cosmic/]),HD254forL858R, andHD258forT790M.EachreferencemutantDNAcontainseach mutantsequenceatafrequencyof50%.EachreferencemutantDNA

Table1

Patientcharacteristics.

Allpatients (n=49)(%)

Patientsbefore startofEGFR-TKI treatment (n=21)(%)

PatientswithPD afterEGFR-TKI treatment (n=28)(%) Age,years

median 67 74 65

range 42–89 50–89 42–81

Sex

Male 20(41) 7(33) 13(46)

Female 29(59) 14(67) 15(54)

Smokinghistory

Never 28(57) 16(76) 12(43)

Current/Former 21(43) 5(24) 16(57)

Histology

Adenocarcinoma 49(100) 21(100) 28(100)

EGFRmutation

Ex19del 34(69) 14(67) 20(71)

L858R 15(31) 7(33) 8(29)

Stage

IIIB/IV 41(84) 18(86) 23(82)

Postoperativerecurrence8(16) 3(14) 5(18) EGFR-TKI,epidermalgrowthfactorreceptortyrosinekinaseinhibitor;PD,progres- sivedisease;Ex19deletion,exon19deletionmutation;L858R,L858Rmutationin exon21.

wasseriallydilutedwiththecorrespondingwild-typeEGFRrefer- enceDNA(HD709,HorizonDiscovery)to50%,10%,5%,1%,0.5%, 0.1%,0.05%,or0.01%. Weregarded theperformance ofdPCRas adequateonlyiftheseriallydilutedDNAcopiesweredetected,as expected.Theaveragecopynumberdetectedinthepresenceof wild-typeEGFRreferenceDNA(i.e.,negativecontrol)plus3stan- darddeviations(SD)wasusedasthecut-offvalueforeachmutation site.

2.4. Statisticalanalysis

Thesensitivityand specificityofthechip-baseddPCRassays indetectingEGFRmutationswereevaluatedbycomparingresults obtainedwiththeScorpionARMSmethodusingthecorresponding tumorsamples(thereferencemethod).The95%confidenceinter- vals(CI)forthesensitivityandspecificitywerecomputedusing theClopperandPearsonmethod[41].Fisher’sexacttestwasused toexaminewhetherthetestresultscorrelatedwithclinicopatho- logicalcharacteristics.Progression-freesurvival(PFS)wasdefined asthedurationbetweentheinitiationofEGFR-TKItherapyandthe onsetofprogressivediseaseordeathfromanycause.PFSwascalcu- latedaccordingtotheKaplan–Meiermethodandcomparedusing thelog-ranktest.Correlation coefficients werecalculated using Spearman’srankanalysis.P<0.05wasconsideredstatisticallysig- nificant.StatisticalanalyseswereperformedusingJMPsoftware, version12.0.1for Windows(SAS Institute, Cary,NC,US)and R software,version3.1.3(http://www.r-project.org).

3. Results

3.1. Patientcharacteristics

Thepatientcharacteristicsare shownin Table1.Thirty-four patientswithNSCLCharboredanexon19deletionand15patients presentedtheL858Rmutationintheirtumorsamples,asdetected using the Scorpion ARMS method. Twenty-one blood samples werecollectedfrompatientswhohad never beentreatedwith EGFR-TKIs,and 28 blood sampleswere collectedfrompatients with progressive disease after treatment with EGFR-TKIs. The patientshadamedianageof67years (range42–89years),and 29patients(59%)werefemale.Twenty-eightpatients(57%)were never-smokers,and 21 (43%) had a smoking history. Forty-one

Table2

ComparisonofEGFRexon19deletionandL858RstatusincfDNAandpairedtumor samples.

Tumortissue

PlasmacfDNA + – Sensitivity Specificity Concordance Exon19del

+ 24 1 70.6% 93.3% 77.6%

− 10 14

L858R

+ 10 0 66.7% 100.0% 89.8%

− 5 34

cfDNA,circulatingfreeDNA;Ex19del,exon19deletionmutation;L858R,L858R mutationinexon21;M+,mutationpositive;M-,mutationnegative.

patients(84%)hadstageIIIBorstageIVdisease,and8patients(16%) hadpost-operativerecurrence.Alllungtumorswereclassifiedhis- tologicallyasadenocarcinoma.

3.2. Determinationofassayperformance

Afterpreparing7 serialdilutionsof genomicDNA harboring an EGFR mutant in genomic DNA with wild-type EGFR (range:

50%–0.01%mutantEGFR), westably detectedthetargetedEGFR mutationsatafrequencyaslowas0.1%(SupplementaryFig.2).

Experimentswererepeatedover3separatedays.Todeterminethe cut-offvalue,referenceDNA withwild-typeEGFRwasanalyzed 8times usingamplificationprimers andeach mutation-specific probe.Theaveragecopynumbersforthewild-typereferenceDNA (plus3SDs),whichwereusedasthecut-offvalues,were1.6,2.1, and0copies/assayforexon19deletions,L858R,andT790M,respec- tively. Supplementary Fig.3A shows comparisons betweenthe expectedmutationfrequencyin20ngofDNAloadedinto2separate chipsvs.theobservedcopynumbersofEGFRmutations.Mutant alleleswerestablydetectedusingthesecut-off values,downto aprevalenceof0.1%(SupplementaryFig.3A).Wealsoevaluated thecopynumberofEGFRmutationsinplasmasamplesusingthese determinedcut-offvalues.Apositivelinearrelationshipbetween theexpected and observedmutation frequencies wasobserved overarangeof0.1–50%duringtestingforallmutantsexamined.

Regressionanalysisshowedcoefficientsofdetermination(R²)of 0.9976,0.9952,and0.9961withdetectionoftheexon19deletion mutant,L858R,andT790M,respectively(SupplementaryFig.3B).

3.3. EGFRmutationtestinginclinicalplasmacfDNAsamples

Among49 patients, themedian concentrationof cfDNA iso- latedpermlofplasmaacrossallsampleswas35.1ng/ml(range 12.2–139.5).Twenty-fiveplasmasampleswereidentifiedaspos- itiveforanexon19deletion,basedonthecut-offcopynumber (>1.6),and10wereidentifiedaspositiveforL858R(cut-offcopy number>2.1),asshowninFig.1andSupplementaryTable2.Only 1plasmasamplewaspositiveforbothanexon19deletionand L858R,althoughonlytheL858Rmutationwasdetectedinthecorre- spondingtumortissue.ThesensitivityandspecificityofeachdPCR assaycalculatedbycomparisonwiththecorrespondingtumorsam- pleswere70.6%(24/34;95%CI,52.5%–84.9%)and93.3%(14/15;

95%CI,68.1%–99.8%)forexon19deletion,and66.7%(10/15;95%

CI,38.4%–88.2%)and100%(34/34;95%CI,89.7%–100%)forL858R (Table2).Theoverallconcordanceratebetweentheplasmaand tumortissueswas77.6%(38/49;95%CI,63.4%–88.2%)forexon19 deletionand89.8%(44/49;95%CI,77.8%–96.6%)forL858R.

Twelvesampleswereidentifiedaspositive(cut-offcopynum- ber>0)fortheT790Mmutation(Fig.1andSupplementaryTable 2).AllplasmasampleswiththeT790Mmutationwereobtained frompatientsshowingprogressionafterEGFR-TKItreatment(43%;

Fig.1.DistributionofEGFRmutant-allelecopynumbersperassayinplasmasam- ples.Intheexon19deletionassay,thecopynumbersdetectedinplasmasamples rangedfrom1.0to610.7copies/assay,and25plasmasampleswerescoredasposi- tiveusingacut-offvalueof1.6copies/assay.IntheL858Rassay,thecopynumbers detectedinplasmarangedfrom4.6to5380.4copies/assay,and10sampleswere scoredaspositiveforL858Rusingacut-offvalueof2.1copies/assay.IntheT790M assay,thecopynumbersmeasuredinplasmasamplesrangedfrom4.6to267.2 copies/assay,and12plasmasampleswerescoredaspositiveusingacut-offvalue of0copies/assay.Thedashedlineshowstheaveragecopynumbersfornegativesam- plesplus3×SDasthecut-offvalue.Exon19del,exon19deletion;L858R,L858R mutationinexon21;T790M,T790Mmutationinexon20.

12/28),withtheT790Mmutationnotbeingobservedinpatient samplesobtainedbeforeEGFR-TKItreatment.

In28 patientswithprogressive diseaseaftertreatmentwith EGFR-TKIs, 11 out of 12 patients with NSCLC and detectable EGFRT790Mintheplasmaunderwentre-biopsy,5ofwhichalso presentedtheT790Mmutationintheirre-biopsysamples(Supple- mentaryTables2and3).Inaddition,9outof16patientswithNSCLC lackingdetectableT790Mintheplasmaunderwentre-biopsy,and 2patientspresentedT790Mintheirre-biopsysamples(Supple- mentaryTables2and3).

Amongthe25patientswithanexon19deletionintheplasma, thepercentmutationrateintheplasmarangedfrom0.3%to45.4%

(median2.3%).Amongthe10patientswithdetectableL858Rinthe plasma,thepercentmutationrateintheplasmarangedfrom0.5%

to70.8%(median2.4%),including1patientwhohadbothanexon 19deletionandtheL858Rmutation.Amongthe12patientswith detectableT790Mintheplasma,thepercentmutationrateinthe plasmarangedfrom0.5%to11.6%(median1.4%).Nostatistically significantrelationshipsbetweentheamountofinputcfDNAand thecopynumbersofeachEGFRmutationwereobserved(Supple- mentaryFig.4).

3.4. AssociationbetweenEGFRmutationdetectionand extra-thoracicmetastaticdisease

Amongallpatients,therelationshipbetweenthedetectionof EGFRmutationsincfDNAandtheexistenceofextra-thoracicmetas- taseswasevaluated.EGFR-activatingmutationstendedtobemore frequentlydetectedintheplasmaofpatientswithextra-thoracic metastaticdisease(M1b)thaninpatientswithdiseaseconfined tothethoraciccavity(M1a/M0;p=0.053;Table3).Nodifference inthedetectionoftheEGFRT790Mmutationintheplasmawas observedbetweenpatientswithM1a/M0andM1bdiseaseclassi- fications(p=0.29).

Table3

RelationshipbetweenEGFR-activatingmutationsandNSCLCclassification.

Diseaseclassification

PlasmacfDNA M1a/M0 M1b Total

EGFR-activatingmutations + 8 26 34

– 8 7 15

Total 16 33 49

P=0.053 cfDNA,circulatingfreeDNA.

Fig.2. Progression-freesurvival(PFS)of20patientswithNSCLCwhoenrolledin thisstudybeforethestartofEGFR-TKItreatment.ThemedianPFSwas9.7months inpatientswithdetectableEGFR-activatingmutationsintheplasma(dashedline), and11.7monthsinpatientswithoutdetectableEGFR-activatingmutationsinthe plasma(solidline;p=0.60).

3.5. PlasmaEGFRmutationstatusatpre-treatmentvs.clinical outcomes

Amongthe21patientswhoenrolledbeforethestartofEGFR-TKI treatment,20patientsweretreatedwithEGFR-TKIsafterplasma sampling.Nostatisticallysignificantdifferencewasobservedinthe responserate(RR)andPFSfollowingEGFR-TKItreatmentbetween patientswithdetectable(n=12) and undetectable(n=8)EGFR- activatingmutationsintheplasma(RR: 87.5vs.66.7%,p=0.60;

medianPFS:11.7vs.9.7months,p=0.60;Fig.2).

4. Discussion

Toourknowledge,thisisthefirststudytoevaluatetheper- formance of detection assays for exon 19 deletion, L858R, and T790MinplasmasamplesusingtheQS3DDigitalPCRSystem.EGFR mutationswerestablydetectedataprevalenceaslowas0.1%,indi- catingconsistencywithdatafromapreviousstudyfordetection ofJAK2V617Fmutationinmyeloproliferativeneoplasmswiththe QS3DDigitalPCRsystem[34].ThedetectionlimitsofddPCR,using theQX100/QX200DropletDigitalPCRSystemwereestimatedas 0.01–0.04%[25–29].Similarly,thedetectionlimitofBEAMingPCR wasestimatedas0.01%[31],andthatofchip-baseddPCRusing theBioMarksystemwas0.03–0.1%[32,33](Table4).Asshownin Table4,weobservedsimilarsensitivityandspecificitycomparedto theotherdPCRassays.Theseresultsindicatedthatthelowerdetec- tionlimitdidnotcorrelatewiththesensitivityandspecificityin detectingEGFRmutationsinplasma.Thisdiscrepancymighthave resultedfromtheuseofdifferenttypesofDNA,suchasplasmid andhumangenomicDNA,whendeterminingthelowerdetection limit.Moreover,itispossiblethatdifferencesinthedesignofthe primersandprobes,aswellasthecut-offvalueselected,alsocaused thedifferencesintheresults.

Table4

PerformancecomparisonwithotherPCRmethodsforevaluatingEGFRmutationsin plasma.

Method Detection

limit(%)

Sensitivity (%)

Specificity (%) Allele-specificPCR(Cobas)[18–21] 0.1 60.7–84.3 96.4–97.0

BEAMingPCR[19,20,30] 0.01 72.7–87.0 96.8–100

DropletdPCR[25–29] 0.01–0.04 66.7–84.3 95.8–100

Chip-baseddPCR[31,32] 0.03–0.1 78.9 100

Thisstudy 0.1 66.7–70.6 93.3–100

dPCR,digitalPCR.

EGFR-activating mutations tended to be more frequently detectedintheplasmaofpatientswithextra-thoracicmetastatic disease (M1b) than in patients with limited thoracic disease (M1a/M0; p=0.053; Table3).These resultsare consistentwith thoseofpreviousstudiesandsuggestthatthetumorburdenor metastaticstatusmayinfluencethedetectionofEGFRmutations intheplasma[15,19,20,42].Althoughnostatisticallysignificant differenceswereobservedintheRRandPFSfollowingEGFR-TKI treatment between patients with or without detectable EGFR- activatingmutationsintheirplasma,patientswithEGFR-activating mutationsintheplasmaappearedtohaveinferiorPFS(Fig.2).These resultsmighthavebeeninfluencedbythemetastaticstatus.

WhilealowfrequencyofdenovoT790Mmutationsintumor sampleswasreportedin onestudy[43],anotherstudyshowed theexistenceoftheT790Mmutationinapproximatelyhalfofthe tumorsamplestestedafterprogressionwithEGFR-TKItreatment [8].In thepresentstudy,T790M mutationswereonlydetected inplasmasamplesofpatientsshowingdiseaseprogressionafter EGFR-TKItreatment, and thefrequency of T790M detection in plasma samples was 43%, similar to data from previous dPCR studies showing that 28.6%–66.7% of patients withNSCLC had detectableT790Mintheplasma[19,20,26,28,29,33].Twoprevious reportsshowedthattheclinicalresponseratesforthird-generation EGFR-TKIs in patients positive for the T790M mutation in the plasmawerenearlyidenticaltothatinpatientswithpositivetis- suesamples[19,20].TheseresultssuggestthatplasmaT790Mcan serveasabiomarkerfortheuseofthird-generationEGFR-TKIs;

thus,itwillbecomemoreimportanttotestfor thepresenceof plasmaT790Minfutureclinicalsettings.

Thisstudyhassomelimitations.Firstly,wecouldnotperform thecomparisonassaywiththesamecfDNAsamplesbetweenthe QS3DDigitalPCRsystemandotherdPCRplatforms.Instead,inthis study,weperformedtheassessmentofassayperformanceofthis QS3DDigitalPCRassaywithcommerciallyavailablecertifiedref- erenceDNA materialswhose mutant allelecopy numberswere validatedusingaddPCRsystemofBio-Rad(SupplementaryFigs.

2and3).Secondly,becauseenoughmatchingtumorsampleswere notavailabletoevaluatetheQS3DDigitalPCRassay,thedetec- tionofEGFR-mutationsbydPCRwasperformedonlywithplasma samplesand notwithtissue samples.Thesensitivityand speci- ficityoftheassayswerecalculatedaccordingtotheEGFRmutation statusoftumorsamples,asfirstdeterminedusingtheScorpion ARMSmethod.Thirdly,differencesoccurredinthetimingbetween bloodsamplingandtumortissuesamplingforsomepatients.The prevalenceof EGFR-activating mutations in the plasma can be influencedbytreatmentwithEGFR-TKIsorcytotoxicchemother- apies[25]. Thetime course mayalsohave an influenceonthe levelofEGFR-activatingmutationsdetectedintheplasma.Accord- ingly,wecouldnotevaluatetheconcordanceofT790Mmutations betweenplasmaandtissuesamples.TheleveloftheT790Mmuta- tionpresentinplasmamaychangefollowingEGFR-TKItreatment [25].In thisstudy,T790Mmutationswereonlydetectedinthe plasmaofpatientsshowingprogressionafterEGFR-TKItreatment;

thus,theT790Mmutationstatusinthecorrespondingtumorsam-

plesatthetimeofbloodsamplingisunclear.However,thisisalso generallyalimitationinclinicalsettings.Theselimitationsmight alsobe associated withthecausesowing towhich the perfor- manceoftheQS3DDigitalPCRsystemwassimilartothatofthe conventionalchip-basedBioMarkdPCRsystem(Fluidigm),which hasfewermicroporescomparedtotheQS3DDigitalPCRsystem [32],althoughapreviousstudyreportedthatthenumberofparti- tionsimpactsthemeasurementprecisionanddynamicrange[44].

Toovercometheabove-mentionedlimitations,prospectivestudies withtime-matchedbloodandtissuesamplesobtainedfrommore patientswithlungadenocarcinomaarenecessaryforfurthereval- uationoftheperformanceoftheEGFRmutation-detectionsystem developedinthisstudy.

5. Conclusion

We evaluated dPCR assays for detecting EGFR mutations in plasmasamplesofpatientswithNSCLCusingtheQS3DDigitalPCR System.ThissystemshowedsimilarperformancetootherdPCR assays.OurresultsindicatedthatEGFRmutationtestingwithcfDNA usingchip-baseddPCRcanbeusefulasaminimallyinvasivemon- itoringmethodinclinicalsettings.

Conflictofinterest

Theauthorshavethefollowingconflictsofinteresttodeclare:H.

K.,Y.H.,M.Y.,andT.T.receivedresearchgrantsfromAstraZeneca.

K.K.H.K.,H.M.,Y.K.,M.Y.,andT.T.receivedspeakingfeesfrom AstraZeneca.K.K.H.K.,Y.K.,andM.Y.receivedresearchgrants fromBoehringerIngelheim,Japan.H.K.,H.M.,Y.K.,M.Y.,andT.

T.receivedspeakingfeesfromBoehringerIngelheim,Japan.Y.H., Y.K.,M.Y.,andT.T.receivedresearchgrantsfromChugaiPharma- ceuticalCo.,Ltd.H.K.,A.O.,S.O.,H.M.,M.Y.,andT.T.received speakingfeesfromChugaiPharmaceuticalCo.,Ltd.H.K.,A.O.,and S.O.receivedspeakingfeesfromTaihoPharmaceuticalCo.,Ltd.T.

T.receivedresearchgrantsfromEliLilly JapanK.K.H.K.andA.

O.receivedspeakingfeesfromEliLillyJapanK.K.T.T.received researchgrantsfromPfizerJapan,Inc.

Acknowledgement

ThisworkwassupportedbytheJapanSocietyforthePromotion ofScienceKAKENHIGrantNumber15K21688(A.O.).

AppendixA. Supplementarydata

Supplementarydataassociatedwiththisarticlecanbefound,in theonlineversion,athttp://dx.doi.org/10.1016/j.lungcan.2017.02.

001.

References

[1]M.Maemondo,A.Inoue,K.Kobayashi,S.Sugawara,S.Oizumi,H.Isobe,A.

Gemma,M.Harada,H.Yoshizawa,I.Kinoshita,Y.Fujita,S.Okinaga,H.Hirano, K.Yoshimori,T.Harada,T.Ogura,M.Ando,H.Miyazawa,T.Tanaka,Y.Saijo,K.

Hagiwara,S.Morita,T.Nukiwa,North-EastJapanStudyG,Gefitinibor chemotherapyfornon-small-celllungcancerwithmutatedEGFR,NewEngl.

J.Med.362(2010)2380–2388.

[2]T.Mitsudomi,S.Morita,Y.Yatabe,S.Negoro,I.Okamoto,J.Tsurutani,T.Seto, M.Satouchi,H.Tada,T.Hirashima,K.Asami,N.Katakami,M.Takada,H.

Yoshioka,K.Shibata,S.Kudoh,E.Shimizu,H.Saito,S.Toyooka,K.Nakagawa, M.Fukuoka,Gefitinibversuscisplatinplusdocetaxelinpatientswith non-small-celllungcancerharbouringmutationsoftheepidermalgrowth factorreceptor(WJTOG3405):anopenlabel,randomisedphase3trial,Lancet Oncol.11(2010)121–128.

[3]R.Rosell,E.Carcereny,R.Gervais,A.Vergnenegre,B.Massuti,E.Felip,R.

Palmero,R.Garcia-Gomez,C.Pallares,J.M.Sanchez,R.Porta,M.Cobo,P.

Garrido,F.Longo,T.Moran,A.Insa,F.DeMarinis,R.Corre,I.Bover,A.Illiano, E.Dansin,J.deCastro,M.Milella,N.Reguart,G.Altavilla,U.Jimenez,M.