Cancer Science. 2019;00:1–10.  | 1wileyonlinelibrary.com/journal/cas

Received:18February2019  |  Revised:21May2019  |  Accepted:4June2019DOI:10.1111/cas.14092

O R I G I N A L A R T I C L E

Monitoring of cancer patients via next‐generation sequencing of patient‐derived circulating tumor cells and tumor DNA

Kaoru Onidani1,2 | Hirokazu Shoji1,3  | Takahiko Kakizaki1 | Seiichi Yoshimoto4 | Shinobu Okaya1 | Nami Miura1 | Shoichi Sekikawa2 | Koh Furuta5 | Chwee Teck Lim6,7,8 | Takahiko Shibahara2 | Narikazu Boku3 | Ken Kato3 | Kazufumi Honda1,9

1DepartmentofBiomarkersforEarlyDetectionofCancer,NationalCancerCenterResearchInstitute,Tokyo,Japan2DepartmentofOralandMaxillofacialSurgery,TokyoDentalCollege,Tokyo,Japan3GastrointestinalMedicalOncologyDivision,NationalCancerCenterHospital,Tokyo,Japan4DepartmentofHeadandNeckSurgery,NationalCancerCenterCentralHospital,Tokyo,Japan5DivisionofClinicalLaboratory,KanagawaCancerCenter,Kanagawa,Japan6DepartmentofBiomedicalEngineering,NationalUniversityofSingapore,Singapore7BiomedicalInstituteofGlobalHealthResearchandTechnology,NationalUniversityofSingapore,Singapore8MechanbiologyInstitute,NationalUniversityofSingapore,Singapore,Singapore9JapanAgencyforMedicalResearchandDevelopment(AMED)CREST,Tokyo,Japan

ThisisanopenaccessarticleunderthetermsoftheCreativeCommonsAttribution-NonCommercialLicense,whichpermitsuse,distributionandreproductioninanymedium,providedtheoriginalworkisproperlycitedandisnotusedforcommercialpurposes.©2019TheAuthors.Cancer SciencepublishedbyJohnWiley&SonsAustralia,LtdonbehalfofJapaneseCancerAssociation.

OnidaniandShojicontributedequallytothiswork.

CorrespondenceKazufumiHonda,DepartmentofBiomarkersforEarlyDetectionofCancer,NationalCancerCenterResearchInstitute,5-1-1Tsukiji,Chuo-ku,Tokyo104-0045,Japan.Email:khonda@ncc.go.jp

Funding informationJapanAgencyforMedicalResearchandDevelopment,Grant/AwardNumber:18cm0106403h0003;JSPSKAKENHI,Grant/AwardNumber:16K19381;Grant-in-AidforScientificResearch(B)METXChallengingExploratoryResearchGrant

AbstractLiquidbiopsyofcirculatingtumorcells(CTC)andcirculatingtumorDNA(ctDNA)isgainingattentionasamethodforreal-timemonitoringincancerpatients.Conventionalmethodsbaseduponepithelialcelladhesionmolecule (EpCAM)expressionhaveariskofmissingthemostaggressiveCTCsubpopulationsduetoepithelial-mesenchy-maltransitionandmay,thus,underestimatethetotalnumberofactualCTCpresentinthebloodstream.Techniquesutilizingalabel-freeinertialmicrofluidicsapproach(LFIMA)enableefficientcaptureofCTCwithouttheneedforEpCAMexpression.Inthisstudy,weoptimizedamethodforanalyzinggeneticalterationsusingnext-gen-eration sequencing (NGS) of extracted ctDNA andCTCenriched using an LFIMAasafirst-phaseexaminationof30patientswithheadandneckcancer,esophagealcancer, gastric cancer and colorectal cancer (CRC). Seven patientswith advancedCRCwereenrolledinthesecond-phaseexaminationtomonitortheemergenceofal-terationsoccurringduringtreatmentwithepidermalgrowthfactorreceptor(EGFR)-specificantibodies.UsingLFIMA,weeffectivelycapturedCTC(mediannumberofCTC,14.5cells/mL)fromseveraltypesofcanceranddetectedmissensemutationsviaNGSofCTCandctDNA.Wealsodetectedtime-dependentgeneticalterationsthatappearedduringanti–EGFRtherapyinCTCandctDNAfromCRCpatients.The

2  |     ONIDANI et Al.

1  | INTRODUC TION

Geneticandphenotypicvariationsoccurbetweentumorsinvolvingdifferenttissuesandcelltypesaswellasbetweenindividualswiththe same tumor type.1-3 To enhance understanding of these phe-nomena,longitudinaltumorsamplingapproacheswillbeessentialtoelucidatethe impactoftumorheterogeneityoncancerevolution.4 Althoughmolecularprofilingdataobtainedfromtissuebiopsysam-plescanfacilitatedeliveryofprecisionmedicinebyenablingselec-tionofthemosteffectivechemotherapyapproachforanindividualpatient,tissuebiopsy is invasive,riskyandpainful. Inaddition,thebiological behavior of tumor cells can change with each passingmoment inresponsetoselectivepressuresassociatedwithcancertherapies.Therefore, is important todevelop real-timemonitoringmethodsthatareminimallyinvasivetoprofilethebiologicalbehaviorofanindividualtumor.Suchmethodscouldenhanceunderstandingofthemechanismsunderlyingcancerdiversityanddrugresistance.Indeed,addressingcancerdrugresistancewasakeyrecommenda-tionoftheBlueRibbonPanelthatadvisedtheBeauBidenCancerMoonshotinitiative.5

Liquidbiopsymethodshavegainedincreasingattentionastoolsfor real-time monitoring of cancer patients. Recent technologicaladvancesinthedetectionandcharacterizationofcirculatingtumorcells(CTC)andcirculatingtumorDNA(ctDNA)couldenabletheex-aminationofgenomicalterationsinminimallyinvasiveexaminationsandfacilitatetailoringoftreatmentsbasedonreal-timemonitoringoftumorevolution.AmajoradvantageofCTCprofilingistheeaseofobtainingsamplesformonitoringtumorevolutionandstudyingthemechanismofacquireddrugresistance.Indeed,genomicanalysesofCTCfromnon–smallcelllungcancerpatientsidentifiedtheT790Mgatekeepermutation,whichconfersresistancetoepidermalgrowthfactor receptor (EGFR) tyrosine kinase inhibitors.6,7 Furthermore,next-generation sequencing (NGS) of breast cancer CTC revealedsignificantinterpatientheterogeneitythatcouldbemonitoredovertime.8,9

Among current technologies for CTC detection, the only oneclearedbytheFDAforuseinclinicalsettingsistheCellSearchsys-tem. Inthissystem,CTCare isolatedwithanantibodyagainsttheepithelialcelladhesionmolecule(EpCAM)asabiomarker.However,conventionalmethodsbasedondetectionofEpCAMcarrytherisk

ofmissingthemostaggressiveCTCsubpopulationsduetoepithelial-mesenchymaltransition (EMT),potentially leadingtounderestima-tionofthetotalnumberofactualCTCpresentinthebloodstream.To overcome this drawback, a label-free inertial microfluidics ap-proach (LFIMA)wasrecentlydevelopedtoenrichCTCfrombloodsamples.10 This system enables efficient isolation of CTCwithoutaffinity purification of epithelial biomarkers, thereby avoiding un-derestimationofCTCsubpopulationsexhibitingdownregulationofEpCAMexpression.

Here,weestablishedamethodusingLFIMAwithNGSfor theanalysisofgenomicalterations inCTC isolated frompatientswithheadandneckcancer (HNC)orgastrointestinal (GI)cancer. Inad-dition, we carried out blood-basedmolecular profiling to identifyactionable drug targets,monitor drug resistance, and track tumordynamicsusingCTCandctDNAfrompatientswithmetastaticcol-orectalcancer(CRC).

2  | MATERIAL S AND METHODS

2.1 | Patients and peripheral blood samples

A total of 37 patients diagnosed at the National Cancer CenterHospital with HNC or GI cancer (esophageal cancer [OC], gastriccancer[GC]orCRC)betweenJune2013andJuly2016wereenrolledforthisprospectivestudy.Allparticipantsprovidedsignedinformedconsentpriortosamplecollection.Peripheralbloodwascollectedin5-mLEDTAvacutainers(TERUMO)andprocessedwithin24hours.The studywas approvedby the ethics committeeof theNationalCancerCenterandregisteredwiththeUniversityHospitalMedicalInformationNetworkClinicalTrialsRegistry(ID:UMIN000014095).

In the second-phase examination focusing on CRC, patientsreceived irinotecan plus panitumumab or cetuximab (anti–EGFRantibody selection was at the physician's discretion) until diseaseprogressionorunacceptabletoxicitywasnoted.Cetuximab(MerckKGaA)wasadministeredinitiallyatadoseof400mg/m2,followedbyweeklyinfusionsof250mg/m2.Panitumumab(Takeda)wasad-ministeredatadoseof6mg/kgevery2weeks.Thedoseofirino-tecan was selected by each physician according to the patient,basedonpriorsymptomsoftoxicityexperiencedwithtwice-weeklyirinotecan.

resultsofNGSanalysesindicatedthatalterationsinthegenomicprofilerevealedbytheliquidbiopsycouldbeexpandedbyusingacombinationofassayswithCTCandctDNA.ThestudywasregisteredwiththeUniversityHospitalMedicalInformationNetworkClinicalTrialsRegistry(ID:UMIN000014095).

K E Y W O R D S

circulatingtumorcell,circulatingtumorDNA,gastrointestinalcancer,headandneckcancer,liquidbiopsy

     |  3ONIDANI et Al.

2.2 | Circulating tumor cell enrichment

AClearCellFXsystem(Biolidics[previouslyClearbridgeBiomedics],Singapore) compatible with the LFIMA was used to capture andenrichCTCfromperipheralbloodsamplesaccordingtothemanu-facturer'sprotocol.Atotalof5mLofbloodwasmixedwith15mLof redbloodcell lysisbuffer (G-Biosciences) at room temperaturefor10minutes.After incubation, the sampleswerecentrifugedat500 g for 10 minutes followed by aspiration of the supernatant,withfinaladditionof4.3mLofsuspensionreagentsuppliedbythemanufacturer.10

2.3 | Immunofluorescence cytochemistry

Circulatingtumorcellslideswerepreparedusingacyto-spindeviceandstoredat−80°C.Cellswerefixedwith4%paraformaldehydefor10minutesatroomtemperatureandpermeabilizedwith0.1%TritonX-100(Sigma-Aldrich).Thecellswerethenincubatedwithanti–panCK rabbit polyclonal antibody (NICHIREI BIOSCIENCE), followedbyincubationwithanti–rabbitIgGAlexaFluor488(ThermoFisherScientific).11

2.4 | DNA extraction and quantification

DNAfromCTCwaspreparedimmediatelyafterisolation,andwhole-genomeamplification(WGA)wascarriedoutusingaREPLI-gSingleCellKit (QiagenGmbH).TheamplifiedDNAwaspurifiedusinganAgencourtAMPureXPsystem(BeckmanCoulter).FragmentationoftheoutputDNAoftheWGAreactionwasassessedusingaTaqMan

RNasePDetectionReagentsKitandFFPEDNAQCassay(ThermoFisherScientific).

Afive-pointstandardcurvewaspreparedusinghumancontrolgenomicDNA(includedintheTaqManRNasePDetectionReagentsKit),andabsoluteDNAconcentrationsweredeterminedagainstthestandardcurveusingreal-timePCR.ThedegreeofDNAfragmenta-tionwasestimatedusingtheDNAratio(relativequantification,RQ)oflongamplicons(256bp)toshortamplicons(87bp).ctDNAwasex-tractedfrom2mLofplasmausingaQIAmpCirculatingNucleicAcidKit(Qiagen)orMaxwellRSCccfDNAPlasmakit(Promega)accordingtothemanufacturer'sinstructions.HumanbloodgenomicDNAwaspurifiedfrom250to1000μLofbuffycoatusingaQIAmpDNAMiniKit(Qiagen)orMaxwellRSCBuffyCoatDNAkit(Promega)toserveasastandard.TheextractedctDNAandhumanbloodgenomicDNAwere purified using an Agencourt AMPure XP system (BeckmanCoulter).DNAwasquantifiedusingaQubit2.0fluorometeraccord-ingtothemanufacturer'sinstructions.

2.5 | Targeted NGS

Librarypreparationwasperformedusing20ngofCTCDNA,ctDNA,and10ngbuffycoatDNAusinganIonAmpliSeqsLibraryKitPlusandIonAmpliSeqCancerHotspotPanel v2 (ThermoFisher Scientific).TheAmpliSeqCancerHotspotPanelwasdesigned toamplify207ampliconscoveringapproximately2790COSMICmutationsfrom50oncogenesandtumorsuppressorgenes(TableS1).EmulsionPCRwasperformedusingtheIon510&Ion520&Ion530Kit–ChefandIonChefsystem(ThermoFisherScientific).SequencingwasperformedonanIonS5XLSystemusinga530chip(ThermoFisherScientific).

F I G U R E 1  Flowdiagramoftheoptimizedprotocolfordetectinggenomicalterationsincirculatingtumorcells(CTC)andctDNA,andimmunofluorescencecytochemistryofisolatedCTC.A,Bloodfrompatients(upto2×5mLofperipheralblood)wascollectedusingEDTAvacutainers.Onecollectiontubeofhemolyzedwholebloodwasdiluted3-fold.CTCwereisolatedfromthebloodusingalabel-freeinertialmicrofluidicsapproach(LFIMA).ctDNAandbuffycoatDNAwereisolatedfromtheothercollectiontube.Targetednext-generationsequencingwasperformedusingtheextractedDNA.B,FluorescenceimageofisolatedCTCstainedforcytokeratin(green).NGS,next-generationsequencing;WGA,whole-genomeamplification

Isolated CTCWhole blood

Buffy coat

Plasma

Diluted 3x

ctDNA extracted

Buffy coat DNA extracted as control

WGA

Detection of CTC by immunofluorescence

Targeted NGS

(A)

(B)

4  |     ONIDANI et Al.

2.6 | Sequencing data analysis and variant calling

SequencingdatawereassessedusingTorrentSuitesoftware,version5.6.Variantswerecalledusingion-plugin-coverageAnalysis,version5.6.0.1,andion-plugin-variantCaller,version5.6.0.4.Singlenucleo-tidevariants,insertionsanddeletionswereannotatedusingtheIonReportersoftware,version5.6(ThermoFisherScientific).Theallelefrequencythresholdwassetto5%,andminimumcoveragepertar-getampliconwassetto250×toreportdenovomutations.

3  | RESULTS

3.1 | Development of efficient analytical methods using DNA extracted from liquid biopsy samples from patients in the first‐phase examination

The studyworkflow is summarized in Figure 1A. To develop effi-cientmethodsfortheanalysisofmutations inCTCcollectedfrompatientbloodsamples,weoptimizedtheCTCcapturemethodusinganLFIMAfromthefirstexaminationphase,whichconsistedof30patientswithHNC,OC,GCandCRC.

Baselinecharacteristicsofthe10patientswithHNCareshowninTable1.Themedianagewas70years (range,42-80years).Thetumorlocationswereasfollows:5intheoralcavity(50%),1inthesalivaryglands(50%),3 inthepharynx(30%)and1 inthecervicalesophagus(10%).Ninepatientshadsquamouscellcarcinoma,and1patienthadadenoidcysticcarcinoma.Thenumberofpatientsatclin-icalstageII,IIIandIVwas3(30%),1(10%)and6(60%),respectively.

Baseline characteristics of the 20 patients with advanced GIcancers,which consistedof 8 (40%) patientswithOC, 1withGC(5%)and11withCRC(55%),areshowninTable2.Themedianagewas61.5years(range,46-73years).EasternCooperativeOncologyGroupperformancestatusatconsentof0,1and2were9(45%),10(50%)and1(5%),respectively.Ofthesepatients,14hadrecurrence.Thenumberofpatientswithpriorchemotherapylineswas3(15%)with1line,5(25%)with2linesand3(15%)withgreaterthan4lines.

3.2 | Genomic profiling of circulating tumor cells and ctDNA using next‐generation sequencing in the first‐phase examination

TodeterminethenumberofCTC,wecarriedoutimmunofluorescentanalyseswithanti–pankeratinantibody(Figure1B).Ofthepatientsenrolledinthefirst-phaseexamination,thenumberofCTCwasde-terminedfor27patients(Figure2AandS1).ThemediannumberofCTCwas14.5/mL(range,3-133/mL).

The resultsofgenomicprofilingbyNGSofHNCCTCsamplesare shown in Figure 2A.Details regarding CTCmutation profiles,allele frequencies and coverages are shown inTableS2.Missensemutationsweredetectedin4outof10(40%)HNCpatients;theseincludedmutations inEGFR and SMAD4 (n=1),TP53 (n=1),RB1 (n=1) andCDKN2A (n=1).Themissensemutations inEGFR and SMAD4were detected in the same case ofHNC. In contrast, themissensemutationsinTP53, RB1 and CDKN2Aweredetectedindif-ferentcases.

Themost frequentmissensemutations inCTC fromGIcancerweredetectedinALK, GNAQ, RB1 and SMAD4,andthesemutationsoccurred in 4 cases (20%).Moreover,missensemutations inAPC, EGFR, RET and SMARCB1weredetectedin3cases(15%).Themostfrequentmissensemutations in casesofOCandCRCoccurred inSMARCB1(3/8,37.5%)andRB1(3/11,27.3%),respectively.Themis-sensemutationsanalyzedinthisfirst-phaseexaminationofCTCoc-curredin4casesofOC(4/8,50%),1caseofGC(1/1,100%)and4casesofCRC(4/11,36.4%).

Of the30patients enrolled in first-phaseexamination, ctDNAwasobtainedfrom28patientstoconfirmmutations incellscircu-lating intheplasmausingNGS.Details regardingctDNAmutationprofiles,allelefrequenciesandcoveragesareshowninTableS2.ThemutationprofileisshowninFigure2B.MissensemutationsinALK and METweredetectedin1caseofHNC;however,thesemissensemutationscouldnotbedetectedintheremaining9casesofHNC.Themost frequentmissensemutations in GI cancers occurred inTP53,andtheseweredetectedin8casesofGIcancer(44.4%).Themost frequentmissensemutations in bothOC andCRCoccurredin TP53,andtheseweredetectedin4casesofOC(4/8,50%)and4casesofCRC(4/9,44.4%).NonsensemutationsinAPC were de-tectedin2casesofCRC.AframeshiftdeletioninAPCwasdetectedin1caseofCRC;inaddition,aframeshiftinsertioninAPC was de-tectedinacaseofOC.Nomissense/nonsenseorframeshift-inser-tion/-deletionmutationsweredetectedin3casesofOC(3/8,37.5%)and2casesofCRC(2/9,22.2%),respectively.

TA B L E 1  Clinicopathologiccharacteristicsofheadandneckcancerpatientsinthefirstexaminationphase

Female Male Total (%)

Age,years,median(range)

75(59-77) 67(42-80) 70(42-80)

Sex 5 5 10

Primarytumorsite

Oralcavity 3 2 5(50)

Salivarygland 1 0 1(10)

Pharynx 0 3 3(30)

Cervicalesophagus

1 0 1(10)

Histology

Squamouscellcarcinoma

4 5 9(90)

Adenoidcysticcarcinoma

1 0 1(10)

Stagea

II 2 1 3(30)

III 0 1 1(10)

IV 3 3 6(60)

aAccordingtotheInternationalUnionAgainstCancer(UICC)TNMClassificationofMalignantTumours,7thedition(2010).

     |  5ONIDANI et Al.

3.3 | Combination analysis of genomic mutation profiles obtained from circulating tumor cells and ctDNA

The results of genomic mutation profiling of CTC and ctDNA(Figure 2A,B) suggested that the genetic mutational concordancebetweenprofiles ofCTCand ctDNAwasnot high.As tumorhet-erogeneity suggested that theCTC and ctDNA samples exhibiteddifferentprofiles,weconductedacombinationanalysis(Figure2C).

The combination analysis improved the rate of genomic alter-ation detection compared to the assays of CTC or ctDNA alone.Thecombinationanalysisdetectedmissensemutationsin5casesofHNC(5/10,50%)and15casesofGIcancer(15/18,83.3%).Thesameaminoacidchangesweredetected in6of28cases inwhichbothCTCandctDNAwereanalyzed(TablesS2andS3).Detailsofassoci-ationsbetweengenomicalterationsdetectedinCTCandctDNAareshowninFigure2D(HNC)andE(GIcancer).

3.4 | Second‐phase examination: Analysis of genomic alterations in circulating tumor cells and ctDNA over time during anti–epidermal growth factor receptor therapy for metastatic colorectal cancer

WeevaluatedthefragmentationofCTCDNAinthesecond-phaseexamination.RepresentativeRQvaluesforassessmentofCTCDNA

fragmentationareshowninFigureS2.CTCDNAwasunfragmented.ThemeanamountofamplifiedandpurifiedCTCDNAwas249ng(range,93-444ng).

Theclinicalmanagementof individualswhodevelopresistancetoanti–EGFRtherapythroughtheemergenceofgenemutationsre-mainschallenging.Toaddress this issue,wemonitoredchanges ingeneticprofilesofCTCandctDNAovertimeinpatientsundergoinganti–EGFR therapy formetastatic CRC (Figure 3A).We examinedbloodsamples from7patientswithmetastaticCRCwho receivedanti–EGFRtherapy(TableS4).Thesepatientsweremonitoredat2timepoints:before initiationof anti–EGFR therapyandatdiseaseprogression.Details regardingmutationprofiles,allele frequenciesandcoveragesareshowninTableS5.Inpatients1and3,genomical-terationsemergedinbothCTCandctDNAwithdiseaseprogression.Patient1developedanewmissensemutationinSMARCB1(p.T72K)inCTCandmissensemutationsinFGFR3(p.N644D),RB1(p.I680T),RB1 (p.L670P) and SMAD4 (p.V354L), and an intronic mutationin EGFR in ctDNA. Inpatient3, amissensemutation inSMARCB1 (p.T72K) in CTC and several other missense mutations were de-tectedduringirinotecanandcetuximabtreatment.

In patients 2, 4 and 6, genomic alterations with disease pro-gression were detected only in ctDNA. In patient 2, new mis-sense mutations in ERBB4 (p.N620_C621 delinsKS) and NRAS (p.Q61L) were detected. Although patient 4 exhibited amissensein SMARCB1 (p.T72K) before beginning treatment with irinotecanand panitumumab, the mutation disappeared as the disease pro-gressed.InctDNA,newmissensemutationsinATM (p.R337C)andKRAS (p.Q61H)weredetected inaddition toanonsensemutationin TP53presentbeforetreatment.Patient6developedanewmis-sensemutationinTP53(p.R282W)andanonsensemutationinAPC (p.Q1097*)asdiseaseprogressed.

In patient 7, a nonsensemutation inPIK3CA (p.D84*)was de-tected in theCTCanalysisduringprogressionafter irinotecanandpanitumumabtherapy.

Interestingly, genomic alterations were detected in DNA ex-tractedfromCTCand/orctDNAinallcasesexamined.RepresentativebaselineanddiseaseprogressionCTimagesareshowninFigure3B.

4  | DISCUSSION

Inthisprospectiveobservationalstudy,weevaluatedwhetherCTCcanbeusedtomonitormolecularchangesinrealtimethroughouttheclinicalmanagementofpatientswithadvancedcancer.Weef-fectivelycapturedCTCfromseveraltypesofcancersusingLFIMAandperformedtargetedsequencingofCTCandctDNAusingNGStechnology.Furthermore,wedesignedastrategycombininganaly-ses of genomic mutation profiles of CTC and ctDNA to identifyuniquemutations that arise during anti–EGFR therapy in patientswithmetastaticCRC.

Circulating tumor cells released into the bloodstream fromprimary tumors andmetastasesmay reflect current tumor status.Genomic alterations inCTC are of growing interest because their

TA B L E 2  Clinicopathologiccharacteristicsofgastrointestinalcancerpatientsinthefirstexaminationphase

Female Male Total (%)

Age,years,me-dian(range)

61.5(63-73) 59.5(46-67) 61.5(46-73)

Sex 4 16 20

Primarytumorsite

Esophagus 1 7 8(40)

Stomach 0 1 1(5)

Colonandrectum

3 8 11(55)

ECOGperformancestatus

0 3 6 9(45)

1 0 10 10(50)

2 1 0 1(5)

Diseasestatusa

StageIV 1 5 6(30)

Recurrence 3 11 14(70)

Numberofpriorchemotherapylines

0 1 8 9(45)

1 1 2 3(15)

2 1 4 5(25)

≥4 1 2 3(15)

aAccordingtotheInternationalUnionAgainstCancer(UICC)TNMClassificationofMalignantTumours,7thedition(2010).

6  |     ONIDANI et Al.

Missense Nonsense Synonymous Intronic MNV

Frameshift deletion Frameshift insertion

Missense Nonsense Synonymous Intronic0

20406080

100120140

12 12 126 4

18 7 1328 18

*

77

10 15 17* * *

53

204

133

16 239 17 14

3 6

30

Frameshift deletion Frameshift insertion

Missense Nonsense Synonymous Intronic MNV

SMAD4RB1

EGFR

KDR

PDGFRACSF1R

MET

ctDNA

ALK

KITABL1

TP53

CDKN2AERBB2

ERBB4

SMAD4

RB1

EGFR

APC

ALK

ctDNA

TP53

GNAQSMARCB1 PIK3CAFBXW7

FGFR3

MET

KDR

KITFGFR1 KRAS

RET

ERBB2

ATM

ABL1

PDGFRA

SMO

BRAF

HNF1A

Head and neck Esophageal Gastric Colorectal

Gastrointestinal cancer

Head and neck Esophageal Gastric Colorectal

Gastrointestinal cancer

Head and neck Esophageal Gastric Colorectal

Gastrointestinal cancer

Num

ber o

f CTC

s/m

L1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 20 21 22 23 24 25 26 27 28 29 30

RB1SMAD4

ALKEGFRGNAQAPCRET

SMARCB1FBXW7TP53ATMKIT

CDKN2AFGFR3MET

ERBB2ABL1KDR

CSF1RERBB4FGFR1PDGFRA

SMOAKT1BRAFCDH1

CTNNB1EZH2FGFR2FLT3

GNA11GNASHNF1AHRASIDH1JAK2JAK3IDH2KRASMLH1MPL

NOTCH1NPM1NRASPIK3CAPTEN

PTPN11SRC

STK11VHL

19

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 23 24 25 26 27 28 29 30RB1

SMAD4ALK #EGFRGNAQAPC #RET

SMARCB1FBXW7TP53ATMKIT

CDKN2AFGFR3MET

ERBB2ABL1KDR #

CSF1RERBB4FGFR1PDGFRA

SMOAKT1BRAFCDH1

CTNNB1EZH2FGFR2FLT3

GNA11GNASHNF1AHRASIDH1JAK2JAK3IDH2KRASMLH1MPL

NOTCH1NPM1NRASPIK3CAPTEN

PTPN11SRC

STK11VHL

# #19 20 21 22

# ##

## #

#

# ## #

#

## #

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 21 23 24 25 26 27 28 29 30RB1

SMAD4ALKEGFRGNAQAPCRET

SMARCB1FBXW7TP53ATMKIT

CDKN2AFGFR3MET

ERBB2ABL1KDR

CSF1RERBB4FGFR1PDGFRA

SMOAKT1BRAFCDH1

CDKN2AEZH2FGFR2FLT3

GNA11GNASHNF1AHRASIDH1JAK2JAK3IDH2KRASMLH1MPL

NOTCH1NPM1NRASPIK3CAPTEN

PTPN11SRC

STK11VHL

20 22

(A) (B)

(C)(D)

(E)

CTC

CTC

     |  7ONIDANI et Al.

identification could aid in the development of targeted therapies.However,CTCareextremelyrareintheblood,with1-100CTC/mLamongmillionsofwhitebloodcellsandbillionsofredbloodcells.12 ToovercomethechallengeofisolatingCTC,multipleplatformshavebeendevelopedforenrichmentanddetection.WeusedtheLFIMAto enrichCTC rather than the conventionalmethodbasedon theidentificationofcirculatingcellsexpressingepithelialmarkers,mostnotablyEpCAM.10UsingtheLFIMA,wecoulddetecta largerCTCpool(independentofEpCAMexpression)toidentifyCTCsubpopu-lationsorEMT-derivedCTC.Enrichedsamplesfromatleast28pa-tientsenrolledinthefirst-phaseexaminationofourstudycontained3-133CTC/mLofblood.InpreviousreportsdescribingisolationofCTCfromHNCandCRCpatientsusingtheCellSearchsystem,only10of80cases(12.5%)ofHNC13and7of20cases(35%)ofCRC14 harboredCTCintheperipheralblood.Inourpreviousstudy,wecom-paredthenumberofCTCinsamplesfrommetastaticCRCpatientsusing theClearCell FX andCellSearch systems. Themedian num-berofCTCwashigherwiththeClearCellFXsystem(14cells[range3-26]/mL)comparedwiththeCellSearchsystem(0cells/7.5mL).15

BecausethenumberofCTCinthebloodstreamisextremelylowincomparisonwithothernormalhematopoieticcells,WGAmeth-odsaregenerallyrequiredforanalyzingNGSdataforDNAobtainedfrom CTC.WGA strategies are known to introduce artifacts anderrorsinvariationdetectionstudies.16Recently,theresultsofNGSanalysesofCTCusingWGAmethodshavebeenreported.MultipledisplacementamplificationtechnologywasdevelopedforanalyzinggeneticalterationsfromextremelysmallamountsofDNAextractedfromsinglecells.Theaccuracyofmultipledisplacementamplifica-tiontechnologyhasbeenevaluatedintermsofitsreliabilityinWGAusingCTC.12,17ToconfirmthereliabilityofWGA,weassessedthefragmentation of amplified CTC DNA. No fragmentation of CTCDNAwasdetected.Thus,experimentsinvolvingWGAweredeemedreliable because the method was the same for all CTC analyses.WeoptimizedtheefficiencyofamethodforNGSanalysisofCTCDNAusingpatientsampleswithmultipledisplacementamplificationtechnology.

Molecularandcellularheterogeneityarehallmarksofcancerthathaveimportantimpactsonthediagnosisandtreatmentoftumors.18 Apreviousstudyshowedthatthedegreeofintratumorheterogene-itycanbehighlyvariable,with0to>8000codingmutationsfoundtobeheterogeneouswithinprimarytumorsorbetweenprimaryandmetastaticorrecurrentsites.19Liquidbiopsyisanessentialtoolfornon-invasivereal-timemonitoringofcancerandalsoenableschar-acterization of tumor heterogeneity because blood carries DNAderivedfromcancercellslocatedatdistinctmetastaticsites,incon-trast to single-tissue biopsies.20 CTC and ctDNAmay represent a

molecularproxyoftheoveralldisease.Inthisstudy,whencomparingmutationsdetectedinCTCandctDNAfrompatientswithHNCandCRC,wefoundthatinsomebloodsamples,CTCexhibitedmutationsthatwerenotdetectedinctDNA,whereasinothers,ctDNAexhib-itedmutations thatwere not detected inCTC. This suggests thatCTCandctDNAexhibitheterogeneity,andtherefore,bothmustbeevaluatedintheclinicalsettingtoenableoptimalsurveillanceofdis-easeprogressionand treatment selection. In this study,NGSdatarevealedthatthesamegeneticalterationcouldbedetectedindataobtainedfromCTCandctDNAusingmultipledisplacementampli-ficationtechnology.However,wefoundthatthegeneticalterationprofiles were not perfectly correlated between CTC and ctDNA.ThesedatasuggestthatCTCandctDNAexhibituniquegenetical-terationprofiles.Inotherwords,usingacombinationassayinvolvingCTCandctDNAcouldincreasethesensitivityofdetectinggeneticalterationswithoutdecreasing thespecificity, thuscontributing totheestablishmentofprecisionmedicineforcancer.

Considerablerecentattentionhasfocusedonthebiologicalhet-erogeneity of CTC.However, in this study,we did not assess theheterogeneity of CTC because the technique utilizing a label-freeinertialmicrofluidicsapproachenrichedCTCinbulkaccordingtocellsize,andwecarriedoutWGAimmediatelyafter isolation. Inaddi-tion,topredictthesensitivityoftumorstomoleculartherapies,themutation statusofamajorityof the tumorsmustbeknown.NGSanalysisof thegenomeofbulkCTCcould facilitatebetterpredic-tions of the efficacyof personalizedmolecular targeted therapiescomparedwithanalysesofsingleCTC.

Anti–EGFRtherapy,eitheraloneorincombinationwithchemo-therapy, isthestandardtreatmentforpatientswithRASwild-typemetastaticCRC.21-25IthasbecomeapparentthatRASmutationsarecorrelatedwithresistancetoanti–EGFRtherapy,andthepresenceofRASmutationsaccountsforapproximately50%-60%ofpatientswithmetastaticCRCrefractorytoanti–EGFRtherapy.26InadditiontomutationsinRAS,mutationsinBRAF and PIK3CA can induce con-stitutiveactivationoftheEGFRandsubsequentintracellularsignal-ing,ultimately leadingtodrugresistance.27-29SeveralstudieshavedetectedthesemutationsinCTCandctDNAisolatedfrompatientswithCRC.30-32WealsodetectedgenomicalterationsinKRAS, NRAS,and PIK3CAinCTCandctDNAusingtargetedNGSinpatientsresis-tanttoanti–EGFRtherapy,asdescribedinpreviousreports.InliquidbiopsyofCTCandctDNA,codon61mutationsinKRAS and NRAS thatweredetectedinourstudyaremorefrequentlyobservedafterCRCpatientshaveacquired resistance toanti–EGFR therapy thanbeforestartingtheanti–EGFRtherapy.20

More than80%ofmutations detected inPIK3CA havebeenreported in 2 hotspots, the helicase domain of exon 9 and the

F I G U R E 2  Targetednext-generationsequencingandcombinationanalysisofgenomicalterationsusingcirculatingtumorcells(CTC)andctDNA.A,GenomicalterationsinCTCfrompatientswithheadandneckcancer(HNC),esophagealcancer(OC),gastriccancer(GC)andcolorectalcancer(CRC).ThenumberofCTCisshowninthecolumns.*ThenumberofCTCcouldnotbedeterminedfor4patients.B,GenomicalterationsinctDNAfrompatientswithHNC,GC,andCRC.ctDNAcouldnotbeextractedfrom2patientswithCRC.C,GenomicalterationsinCTCandctDNAinpatientswithHNC,GCandCRC.#ThesameaminoacidchangesweredetectedinbothCTCandctDNA.D,CombinationanalysisofgenomicalterationsusingCTCandctDNAfrompatientswithHNC.E,CombinationanalysisofgenomicalterationsusingCTCandctDNAfrompatientswithCRC

8  |     ONIDANI et Al.

F I G U R E 3  A,Clinicalcourseincolorectalcancerpatientswhoreceivedanti–epidermalgrowthfactorreceptor(EGFR)therapyandgenomicalterationsincirculatingtumorcells(CTC)andctDNA.A,MonitoringgenomicprofilesofCTCandctDNAduringanti–EGFRtherapy.BV,bevacizumab;CAPOX,capecitabine,oxaliplatin;Cmab,cetuximab;CPT,irinotecan;FOLFIRI,folinicacid,fluorouracilandirinotecan;FOLFOX,folinicacid,fluorouracilandoxaliplatin;Pmab,panitumumab;RT,radiationtherapy;SIRB,tegafur/gimeracil/oteracil,irinotecanandbevacizumab;TAS102,trifluridineandtipiracil.B,RepresentativeCTimages.(a)Growthoflungmetastasesandincreasedpleuraleffusionwereobservedduringdiseaseprogressioninpatient1.(b)Growthofliverandlungmetastasesandincreasedpleuraleffusionwereobservedduringdiseaseprogressioninpatient5

First diagnosis

20162012

FOLFIRI + BV CPT + Pmab

2017

Surgery

2015

Recurrence Diseaseprogression

Disease progression

RIP

No. 5

No. 5-1CTC : -ctDNA : TP53 p.R248W

No. 5-2CTC : ATM p.G3023D; EGFR p.G724D;

FBXW7 p.G499SctDNA : -

No. 6

2015

First diagnosisLung and liver metaSurgery

20172016

CPT + Pmab

RIP

No. 6-1CTC : -ctDNA : -

FOLFOX + BV Clinical trial TAS102

No. 6-2CTC : -ctDNA : TP53 p.R282W; APC p.Q1097*

FOLFIRI + BV

Disease progression

Disease progression

Disease progression

Disease progression

Disease progression

Surgery

No. 7

2006

First diagnosis

20172009

CPT + Pmab

No. 7-1CTC : -ctDNA : SMARCB1 p.T72K

FOLFOX + BV

No. 7-2CTC : PIK3CA p.D84* ctDNA: -

Recurrence

2011

XELIRI + BV

2012

Disease progression

Diseaseprogression

FOLFOX + BV

Disease progression

SurgerySurgery

No. 1

2007

First diagnosis

Liver meta

2010

FOLFOX + BV

20162009

Recurrence

2014

Diseaseprogression

FOLFIRI + BV CPT + Pmab

Disease progression Disease progression

RIP2008

No. 1-1CTC : -

ctDNA : -

No. 1-2CTC : SMARCB1 p.T72KctDNA : FGFR3 p.N644D; RB1 p.I680T;

RB1 p.L670P; SMAD4 p.V354L; EGFR-AS1

20162014

Disease progression

2015

First diagnosis

Surgery

CAPOX + BV FOLFIRI + BV

Disease progression

CPT + Pmab

No. 2-1CTC : -ctDNA: -

No. 2

Disease progression

No. 2-2CTC : -ctDNA: ERBB4 p.N620_C621delinsKS;

NRAS p.Q61L

RIP

MissenseNonsenseIntronicFrameshift deletion

No. 3

First diagnosis

20162015 20172014

No. 3-1CTC : -ctDNA: TP53 p.R249T; APC p.E1309fs

No. 3-2CTC : SMRCB1 p.T72KctDNA : ALK p.G1184E; EZH2 p.I633T; FBXW7 p.T482A;

GNAQ p.T224N; IDH1 p.P118L; MET p.A179T; MTET p.I367V; PDGFRA p.H654P; RB1 p.I680T; SMAD4 p.V354L; SMAD4 p.G256L; TP53 p.R249T

CAPOX + BV

Disease progression

CPT-11 + Cmab

Diseaseprogression

TAS102CPT + BV

Disease progression

Surgery

No. 4

2013

First diagnosis

2016

SIRB

2015 2017

Clinical trial

2014

No. 4-1CTC : SMARCB1 p.T72KctDNA: TP53 p.Q104*

No. 4-2CTC : -ctDNA: ATM p.R337C; KRAS p.Q61H;

TP53 p.Q104*

CAPOX + BV

Disease progression

CPT + Pmab

Disease progression Disease progression

RT

Disease progression

RIP

Recurrence

(A)

No. 1

Disease progression

No. 5

Baseline

Baseline Disease progression

(a)

(b)

(B)

     |  9ONIDANI et Al.

kinasedomaininexon20,suggestingthatthesemutationscanbeusedtopredictresponsetotreatmentwithanti–EGFRtherapy.27 Theexon1mutationwedetectedinCTCwasveryinfrequent,anditsimportanceandfunctionremainincompletelyunderstood.Asforothergenes,mutationsinFBXW7 and SMAD4werefrequentlydetected,andthesemutations,whicharelocatedinthesamedo-main as themutationwe reported in this study, are involved inacquired resistance to anti–EGFR therapy.33,34 Our data, whichwereobtainedfromarelativelysmallnumberofsamples,showedthatfurtherstudyisneededtodeterminewhethergeneticinfor-mation forCRCcellsobtained from liquidbiopsyofpatients re-sistanttoanti–EGFRtherapycouldenhanceeffortstoovercomedrugresistance.

However,thisstudyhassomelimitations.First,ourstudycon-sistedof a smallnumberof samples.The results thusneed tobevalidated in a prospective manner with a sufficient sample size.Second, CTC were defined as cytokeratin-positive in this study.Recently,CTChavebeendistinguishedfromothercellspresentinthebloodbasedonbeing:(i)nuclearpositive;(ii)cytokeratinposi-tive;and(iii)CD45negative.35WepreviouslystainedcellsforthesemarkerstoidentifyCTC;however,itisnowpossibletodistinguishCTCmorphologically.Indeed,wereportedtheuntargetedmolecu-larprofilingofsingleCTCobtainedfrompatientswithgastriccan-cerandcolorectalcancerusing livesingle-cellmassspectrometryintegratedwithmicrofluidics-based cell enrichment techniques.36 Inthatstudy,wedetectedCTCbasedonCD45stainingandmor-phologyanddemonstratedcleardifferences in themetabolomicsprofilesofCTCand leucocytes.Thepresentstudywasnotasin-gle-cell analysis of CTC but rather amutational analysis of CTC.Therefore,weemployedimmunostainingwithananti–cytokeratinantibodyratherthananti–CD45antibody.Third,wedidnotassesstheconcordanceofgenemutationsbetweenprimarytumortissuesand liquidbiopsysamplebecausealmostallcaseswereadvancedstagesorrecurrences,andtheacquisitionofbiopsyspecimensfrompatientswas difficult in the clinical setting at the timelymanner.However,whetherthetumormutationprofileobtainedfromtumorbiopsy samples truly reflects tumor heterogeneity is unclear,37,38 andinthecaseofsurgicalspecimens,itmaybedifficulttocomparethegenemutationstatusofliquidbiopsysampleswiththatofsurgi-calspecimens,giventhatthebiologicalbehavioroftumorcellscanchangemomenttomomentinresponsetoselectivepressuresas-sociatedwithcancertherapies,andtheclonalrevolutionoccurredin primary tumors.39 Therefore, by comparing the genemutationstatusofliquidbiopsysamplesandresponsivenesstospecificcan-certherapies,weaimtoestablishbiomarkersthatpredictrespon-sivenessbasedonthegenemutationprofileobtainedfrom liquidbiopsysampleswithoutbeinginfluencedbythemutationprofileoftheprimarytumors.

Insummary,weoptimizedtheefficiencyofaplatformforcaptur-ingCTCusinganLFIMAandrevealedtheimportanceofbothCTCand ctDNA as diagnostic tools. In addition, our data suggest thatbothCTCandctDNAcanbeusedtocloselymonitortheemergenceofmolecularchangesinpatientswithmetastaticCRC.

ACKNOWLEDG MENTS

Wethankalloftheparticipants,physicians,nursesandstaffmem-bersinvolvedinthisstudy.

DISCLOSURE

Theauthorshavenoconflictsofinteresttodeclare.

ORCID

Hirokazu Shoji https://orcid.org/0000-0002-8922-5227

Kazufumi Honda https://orcid.org/0000-0003-1321-5345

R E FE R E N C E S

1. Vogelstein B, Papadopoulos N, Velculescu VE, Zhou S, Diaz LAJr, Kinzler KW. Cancer genome landscapes. Science. 2013;339: 1546-1558.

2. LawrenceMS,StojanovP,PolakP,etal.Mutationalheterogeneityincancerandthesearchfornewcancer-associatedgenes.Nature. 2013;499:214-218.

3. Jamal-HanjaniM,WilsonGA,McGranahanN, et al. Tracking theevolution of non-small-cell lung cancer.N Engl J Med. 2017;376: 2109-2121.

4. BurrellRA,McGranahanN,BartekJ,SwantonC.Thecausesandconsequencesofgeneticheterogeneityincancerevolution.Nature. 2013;501:338-345.

5. JacksT, JaffeeE, SingerD.CancerMoonshotBlueRibbonPanelReport2016(NationalCancerAdvisoryBoard)).

6. SundaresanTK,SequistLV,HeymachJV,etal.DetectionofT790M,theacquiredresistanceEGFRmutation,bytumorbiopsyversusnon-invasiveblood-basedanalyses.Clin Cancer Res.2016;22:1103-1110.

7. Maheswaran S, Sequist LV, Nagrath S, et al. Detection of mu-tations in EGFR in circulating lung-cancer cells. N Engl J Med. 2008;359:366-377.

8. YuM,BardiaA,AcetoN,etal.Cancertherapy.Exvivocultureofcirculatingbreasttumorcellsforindividualizedtestingofdrugsus-ceptibility.Science.2014;345:216-220.

9. DeLucaF,RotunnoG,SalviantiF,etal.Mutationalanalysisofsinglecirculatingtumorcellsbynextgenerationsequencinginmetastaticbreastcancer.Oncotarget.2016;7:26107-26119.

10. HouHW,WarkianiME,KhooBL,etal.Isolationandretrievalofcir-culatingtumorcellsusingcentrifugalforces.Sci Rep. 2013;3:1259.

11. MiuraN,KamitaM,KakuyaT,etal.Efficacyofadjuvantchemo-therapyfornon-smallcell lungcancerassessedbymetastaticpo-tentialassociatedwithACTN4.Oncotarget.2016;7:33165-33178.

12. LiuHE,TribouletM,ZiaA,etal.WorkflowoptimizationofwholegenomeamplificationandtargetedpanelsequencingforCTCmu-tationdetection.NPJ Genom Med.2017;2:34.

13. GrobeA,BlessmannM,HankenH,etal.Prognostic relevanceofcirculating tumor cells in blood and disseminated tumor cells inbonemarrowofpatientswithsquamouscellcarcinomaoftheoralcavity.Clin Cancer Res.2014;20:425-433.

14. GermanoG,MauriG,SiravegnaG,etal.Parallelevaluationofcircu-latingtumorDNAandcirculatingtumorcellsinmetastaticcolorec-talcancer.Clin Colorectal Cancer.2018;17:80-83.

15. Kato K, Shoji H, Kakizaki F, et al. Next generation sequencingof circulating tumor cells isolated from the peripheral blood ofpatients with gastrointestinal cancer. Circle-1 trial. Ann Oncol. 2014;25(suppl_4):iv558.

10  |     ONIDANI et Al.

16. MacaulayIC,VoetT.Singlecellgenomics:advancesandfutureper-spectives.PLoS Genet.2014;10:e1004126.

17. HouY,WuK,ShiX,etal.Comparisonofvariationsdetectionbe-tween whole-genome amplification methods used in single-cellresequencing.GigaScience.2015;4:37.

18. Wang Y, Waters J, Leung ML, et al. Clonal evolution in breastcancer revealed by single nucleus genome sequencing. Nature. 2014;512:155-160.

19. JohnsonBE,MazorT,HongC,etal.Mutationalanalysisrevealstheorigin and therapy-driven evolution of recurrent glioma. Science. 2014;343:189-193.

20. BettegowdaC,SausenM,LearyRJ,etal.DetectionofcirculatingtumorDNAinearly-andlate-stagehumanmalignancies.Sci Transl Med.2014;6:224ra24.

21. BokemeyerC,BondarenkoI,HartmannJT,etal.EfficacyaccordingtobiomarkerstatusofcetuximabplusFOLFOX-4asfirst-linetreat-mentformetastaticcolorectalcancer:theOPUSstudy.Ann Oncol. 2011;22:1535-1546.

22. VanCutsemE,PeetersM,SienaS,etal.Open-labelphaseIIItrialofpanitumumabplusbestsupportivecarecomparedwithbestsup-portivecarealoneinpatientswithchemotherapy-refractorymeta-staticcolorectalcancer.J Clin Oncol.2007;25:1658-1664.

23. Heinemann V, von Weikersthal LF, Decker T, et al. FOLFIRIplus cetuximab versus FOLFIRI plus bevacizumab as first-line treatment for patients with metastatic colorectal cancer(FIRE-3): a randomised, open-label, phase 3 trial. Lancet Oncol. 2014;15:1065-1075.

24. Venook AP, Niedzwiecki D, Lenz HJ, et al. Effect of first-linechemotherapy combined with cetuximab or bevacizumab onoverall survival in patients with KRAS wild-type advanced ormetastatic colorectal cancer: a randomized clinical trial. JAMA. 2017;317:2392-2401.

25. Price T, Kim TW, Li J, et al. Final results and outcomes by priorbevacizumabexposure, skin toxicity, andhypomagnesaemia fromASPECCT: randomized phase 3 non-inferiority study of panitu-mumabversuscetuximabinchemorefractorywild-typeKRASexon2metastaticcolorectalcancer.Eur J Cancer.2016;68:51-59.

26. MisaleS,DiNicolantonioF,Sartore-BianchiA,SienaS,BardelliA.Resistancetoanti-EGFRtherapyincolorectalcancer:fromhetero-geneitytoconvergentevolution.Cancer Discov.2014;4:1269-1280.

27. DeRoockW,ClaesB,BernasconiD,etal.EffectsofKRAS,BRAF,NRAS, and PIK3CAmutations on the efficacy of cetuximab pluschemotherapy in chemotherapy-refractory metastatic colorec-tal cancer: a retrospective consortium analysis. Lancet Oncol. 2010;11:753-762.

28. Sanz-GarciaE,ArgilesG,ElezE,TaberneroJ.BRAFmutantcolorec-talcancer:prognosis,treatment,andnewperspectives.Ann Oncol. 2017;28:2648-2657.

29. TherkildsenC,BergmannTK,Henrichsen-SchnackT,LadelundS,NilbertM.ThepredictivevalueofKRAS,NRAS,BRAF,PIK3CAand PTEN for anti-EGFR treatment in metastatic colorectalcancer: a systematic review and meta-analysis. Acta Oncol. 2014;53:852-864.

30. GaschC,BauernhoferT,PichlerM,etal.Heterogeneityofepider-malgrowthfactorreceptorstatusandmutationsofKRAS/PIK3CAin circulating tumor cells of patients with colorectal cancer.Clin Chem.2013;59:252-260.

31. HeitzerE,AuerM,GaschC,etal.Complextumorgenomesinferredfromsinglecirculatingtumorcellsbyarray-CGHandnext-genera-tionsequencing.Can Res.2013;73:2965-2975.

32. XuJM,WangY,WangYL,etal.PIK3CAmutationscontributetoac-quiredcetuximabresistanceinpatientswithmetastaticcolorectalcancer. Clin Cancer Res.2017;23:4602-4616.

33. LupiniL,BassiC,MlcochovaJ,etal.Predictionofresponsetoanti-EGFR antibody-based therapies by multigene sequencing in col-orectalcancerpatients.BMC Cancer. 2015;15:808.

34. MehrvarzSarshekehA,AdvaniS,OvermanMJ,etal.Associationof SMAD4 mutation with patient demographics, tumor charac-teristics, and clinical outcomes in colorectal cancer. PLoS ONE. 2017;12:e0173345.

35. Yagi S, KohY,AkamatsuH, et al.Development of an automatedsize-basedfiltrationsystemforisolationofcirculatingtumorcellsinlungcancerpatients.PLoS ONE.2017;12:e0179744.

36. AbouleilaY,OnidaniK,AliA,etal.Livesinglecellmassspectrom-etryrevealscancer-specificmetabolicprofilesofcirculatingtumorcells. Cancer Sci.2019;110:697-706.

37. NormannoN,RachiglioAM,LambiaseM,et al.HeterogeneityofKRAS,NRAS,BRAFandPIK3CAmutationsinmetastaticcolorectalcancerandpotentialeffectsontherapy intheCAPRIGOIMtrial.Ann Oncol.2015;26:1710-1714.

38. McGranahan N, Favero F, de Bruin EC, Birkbak NJ, Szallasi Z,SwantonC.Clonalstatusofactionabledrivereventsandthetim-ing of mutational processes in cancer evolution. Sci Transl Med. 2015;7:283ra54.

39. Yachida S, Jones S, Bozic I, et al. Distant metastasis occurslate during the genetic evolution of pancreatic cancer. Nature. 2010;467:1114-1117.

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle.   

How to cite this article:OnidaniK,ShojiH,KakizakiT,etal.Monitoringofcancerpatientsvianext-generationsequencingofpatient-derivedcirculatingtumorcellsandtumorDNA.Cancer Sci. 2019;00:1–10. https://doi.org/10.1111/cas.14092