FGFR1 肺癌

FGFR1 肺癌

VirchowsArch(2014)465:547–558DOI10.1007/s00428-014-1634-2

ORIGINALARTICLE

FGFR1amplificationisassociatedwithpoorprognosisandsmokinginnon-small-celllungcancer

AnNaSeo&YanJin&HeeJinLee&Ping-LiSun&HyojinKim&SanghoonJheon&KwhanmienKim&Choon-TaekLee&Jin-HaengChung

Received:20March2014/Revised:9July2014/Accepted:10July2014/Publishedonline:3August2014#Springer-VerlagBerlinHeidelberg2014

AbstractFGFR1amplificationhasbeenidentifiedrecentlyasanimportanttherapeutictargetinnon-small-celllungcan-cer(NSCLC),particularlysquamouscellcarcinoma(SqCC).However,datafrompreviousstudiesontheclinicalimplica-tionsofFGFRamplificationinNSCLCareinconsistent.WeevaluatedFGFR1genecopynumber(GCN)in369casesofsurgicallyresectedNSCLCusingfivepreviouslyreportedcriteriaandinvestigatedassociationsbetweenclinicopatho-logicparametersandFGFR1amplification.FGFR1amplifi-cationwasfoundin32/369(8.7%)ofNSCLCandwasmore

ElectronicsupplementarymaterialTheonlineversionofthisarticle(doi:10.1007/s00428-014-1634-2)containssupplementarymaterial,whichisavailabletoauthorizedusers.

A.N.Seo

DepartmentofPathology,KyungpookNationalUniversityHospital,KyungpookNationalUniversitySchoolofMedicine,Jung-guDaegu,RepublicofKorea

Y.Jin:P.<L.Sun:J.<H.Chung(*)

DepartmentofPathology,SeoulNationalUniversityBundangHospital,300Gumi-dong,Bundang-gu,Seongnam-si,Gyeonggi463-707,RepublicofKoreae-mail:chungjh@snu.ac.kr

H.J.Lee

DepartmentofPathology,AsanMedicalCenter,UniversityofUlsanCollegeofMedicine,Seoul,RepublicofKorea

H.Kim:J.<H.Chung

DepartmentofPathology,SeoulNationalUniversityCollegeofMedicine,Jongno-guSeoul,RepublicofKorea

S.Jheon:K.Kim

DepartmentofThoracicandCardiovascularSurgery,SeoulNationalUniversityBundangHospital,Seongnam-si,Gyeonggi,RepublicofKorea

C.<T.Lee

DepartmentofInternalMedicine,SeoulNationalUniversityBundangHospital,Seongnam-si,Gyeonggi,RepublicofKorea

frequentinSqCC(18.0%inSqCC,3.0%inadenocarcinoma;p<0.001)andinsmokers(p<0.001).Onunivariateanalysis,FGFR1amplificationwassignificantlyassociatedwithshorteroverallsurvival(OS,58.6vs80.0months;p=0.033)andshorterdisease-freesurvival(DFS,58.5vs80.0months;p=0.042)inpatientswithSqCC,butthiswasnotstatisticallysignificantonmultivariateanalysis(OS:hazardratio[HR]=1.79,95%confidenceinterval[CI]=0.83–3.87,p=0.139;DFS:HR=1.73,95%CI=0.93–3.21,p=0.081).Thecorrela-tionbetweenFGFR1amplificationandproteinexpressionwaspoor(rho=0.08;p=0.123).TheseresultssuggestthatFGFR1amplificationisassociatedwithsmokinghistoryandsquamouscellcarcinomahistologyandmightindicatepoorprognosis.

KeywordsNon-small-celllungcancer.Squamouscell

carcinoma.Adenocarcinoma.Smoking.Prognosis.FGFR1

Introduction

Thefibroblastgrowthfactorreceptor(FGFR)tyrosinekinasefamilyconsistsoffourmembers,FGFRs1–4[1–3],anditssignalingpathwayisimplicatedinnormaldevelopmentalandphysiologicalprocesses,suchasproliferation,anti-apoptosis,stemness,andangiogenesis[3,4].DysregulatedFGFRsig-nalingthroughgeneamplification,chromosomaltransloca-tion,andpointmutationpromotesoncogenesisinvarioushumanmalignancies[2,4–6].Inparticular,amplifiedoracti-vatedFGFR1hasbeenreportedinseveralhumancancers,suchasoralandtonguesquamouscellcarcinoma(SqCC)[4],esophagealSqCC[7,8],breastcancer[9,10],prostatecancer[11],andrhabodomyosarcoma[12,13].Recently,Weissetal.[14]reportedamplificationofFGFR1in155primarylungSqCCspecimens.Furthermore,inhibitionoftheFGFR1path-waywithFGFRinhibitorssignificantlydecreasedtumorsize

FGFR1 肺癌

548insmoking-relatedlungcancer[5,15].Baseduponthesedata,FGFR1inhibitorsareconsideredapotentialtherapeuticop-tioninFGFR1-amplifiedlungSqCCandearlyclinicaltrialsastheirefficacyhavebeenconducted[16–18].FGFR1am-plificationwasobservedin10–22%oflungSqCCbutinonly0–5.2%oflungadenocarcinoma[2,3,6,14,16,19].Recent-ly,Kimetal.[2]reportedthatinlungSqCC,FGFR1ampli-ficationisanindependentnegativeprognosticfactorandisassociatedwithsmokinghistory.However,datafromotherstudiesontheclinicalimplicationsofFGFRamplificationinlungSqCCareinconsistent[3,19,20].Severalfactorscouldberesponsibleforthesediscrepancies,includingdifferencesbetweenpatientcohortsanddifferentcriteriaforFGFR1amplificationscoring[16].Therefore,criticalcomparativeevaluationofscoringcriteriaofFGFR1amplificationisim-perative.Inthisstudy,wecomparedscoringcriteriaofFGFR1amplificationinnon-small-celllungcancer(NSCLC)usedinpreviousstudies.WealsoinvestigatedassociationsofFGFR1amplificationwithclinicopathologicalcharacteristicsanditsprognosticsignificance.

MaterialsandmethodsPatientsandsamples

Weincludedinthisretrospectivestudy369NSCLCpatientswhounderwentsurgicalresectionforprimaryNSCLCatSeoulNationalUniversityBundangHospitalbetweenMay2003andJune2008.Allpatientswerepreoperativechemotherapyandradiotherapynaive.Caseswithadiagnosisotherthanadeno-carcinomaorSqCCwereexcludedduetotheirsmallnumber.Hematoxylinandeosin(H&E)-stainedslideswerereviewedbytwopathologists(A.N.S.andJ.H.C.).Incaseofdiscordantdiagnosis,aconsensusdiagnosiswasmadeusingadditionalimmunohistochemicalstainingandrepeatedexamination.Thehistopathologicalclassificationofthetumorswasbasedonboththe2004WHOclassificationsystem[21]andthenewadenocarcinomaclassificationbyIASLC/ATS/ERS[22].Clin-icopathologicalinformationwasobtainedfromthepatient’smedicalrecordsandsurgicalpathologyreports.TumorstagewasdeterminedbyUICC/AJCC(7thedition)forthelung[23].Disease-freesurvival(DFS)andoverallsurvival(OS)werecalculatedasthenumberofmonthsbetweenthedateofsurgeryandthedateoftheinitialtumorrelapseandthedateofdeath,respectively.ThisstudywasapprovedbyanInstitutionalRe-viewBoardofSeoulNationalUniversityBundangHospital,andinformedconsentwaswaived.Tissuemicroarray

One2-mmcorewastakenfromarepresentativetumorareaofaselectedparaffinblockofeachtumorandarrangedintoa

VirchowsArch(2014)465:547–558

recipienttissuemicroarray(TMA)block,asdescribedprevi-ously[24].

Fluorescentin-situhybridization

TodeterminateFGFR1status,fluorescentin-situhybridiza-tion(FISH)wasperformedonTMAsectionsusingtheZytoLightSPECFGFR1/CEN8DualColorProbe(ZytoVision,Bremerhaven,Germany),accordingtotheman-ufacturer’sprotocol(Fig.1).TheFGFR1genewasdetectedasagreensignalandthepairedchromosome8(CEP8)copynumbercontrolasanorange/redsignal.Aftervisualscanningoftheslideat×600magnification,atleast60malignantnon-overlappingnucleipercorewereselected,inwhichthenum-berofFGFR1andCEP8signalswascountedat×1,000mag-nification.Smallorlargeclustersfoundinsometumornucleiwereconsideredtohave6or12signals,respectively,follow-ingtheinterpretationguidefortheVentanaINFORM?HER2DNAprobestainingofbreastcarcinoma(VentanaMedicalSystems,Tucson,AZ,USA).Immunohistochemistry

Immunohistochemistry(IHC)analysiswasperformedusinganFGFR1antibody(1:100dilution;Bioworld,St.LouisPark,MN),asdescribedpreviously[25].Theintensityofthemem-branousand/orcytoplasmicstainingofthetumorcellswasscored(0–3:0=nostaining,1=weakstaining,2=moderatestaining,and3=strongstaining)aswellasthepercentageofcellsperlevelofintensity(0–100%).Thesumoftheproductsofintensityandpercentageresultedinanoverallscorerangingfrom0to300.Themedianvalueoftheobtainedscoreswaschosenasthecutoffvalueforthedivisionintolowandhighproteinexpression.Statisticalanalyses

AllstatisticalanalyseswereperformedusingtheSPSSVer-sion18.0(Chicago,IL).Thechi-squaretestorFisher’sexacttest(forthecategoricalvariables)andtheKruskal-WallistestorMann-WhitneyUtest(forthecontinuousvariables)wereusedtoevaluatetherelationshipbetweenFGFR1statusandclinicopathologicalparameters.Toidentifythecutoffpointwithwhichtodiscriminatesignificantsurvivaldifferences,receiveroperatingcharacteristic(ROC)curvesforOSwereused,whichallowedustocalculatesensitivity,specificity,andtheareaundertheROCcurve(AUC)ascriteria.Theassoci-ationbetweenFGFR1amplificationandsmokingpackyearhistorywasevaluatedusinglinearbylinearassociation.TheSpearmancorrelationtestwasusedtoassessthecorrelationbetweenFGFR1genecopynumber(GCN)andproteinex-pression.SurvivalcurveswereplottedusingtheKaplan-Meiermethod,andsignificancewasdeterminedusingthe

FGFR1 肺癌

VirchowsArch(2014)465:547–558549

log-ranktest.MultivariateanalysiswasperformedusingtheCoxproportionalhazardregressionmodelwithstepwisemod-elselection.Thehazardratio(HR)andits95%confidenceinterval(CI)werecalculatedforeachfactor.Statisticalsignif-icancewasdefinedasp<0.05.

SettingthecutoffpointforFGFR1GCN

AsmostFGFR1amplificationswereobservedinpatientswithSqCC,thecutoffpointwascalculatedonSqCC.ThroughtheROCcurveforOS,6.2forFGFR1GCNwerechosenasthecutoffpoint(sensitivity=28.6%,specificity=85.6%,andAUC=0.559).

1.1–12.3)inadenocarcinoma.Inaddition,themedianCEP8copynumberwas2.3(range1.2–7.0)inallNSCLCcases:2.5(range1.2–7.0)forSqCCand2.2(range1.3–6.8)inadeno-carcinoma.ThemedianratioforFGFR1/CEP8was1.0(range0.2–10.5)inallNSCLCcases:1.1(range0.4–10.5)inSqCCand1.0(range0.2–5.0)inadenocarcinoma.Thesedifferenceswerestatisticallysignificant(Fig.2).

EvaluationofthecriteriausedinthepreviousreportsBecauseastandardizeddefinitionforFGFR1amplifica-tionhasnotyetbeenestablished,weevaluatedfivecriteriapreviouslyreportedbyvariousinvestigators[2,3,6,14,16,19].ThedefinitionofcriteriaforeachscoreisprovidedinTable2.FGFR1amplificationwasscoredashighwhenapplyingscoresAandEin21.6%(30/139)oftheSqCCcasesand3.9%(9/230)oftheadenocarci-noma.ForscoreB,thiswas8.6%(12/139)oftheSqCCand0.9%(2/230)oftheadenocarcinomacases,forscoreC,30.2%(42/139)oftheSqCCand4.8%(11/230)oftheadenocarcinomacases,andforscoreD,13.7%(19/139)oftheSqCCand1.7%(4/230)oftheadenocarci-nomacases.

Regardlessofwhichcriterionwasapplied,inNSCLC,FGFR1amplificationwasassociatedwithsmokinghistory,SqCC,andmalegender.However,FGFR1amplificationwasnotcorrelatedwithpatientoutcomebyunivariateanalysis,regardlessofthecriteriaused.InNSCLC,FGFR1amplifica-tionbyscoreA,scoreC,andscoreEwassignificantlycorrelatedwithhightumor-node-metastasis(TNM)stage(p=0.009,p=0.010,andp=0.002,respectively).Inaddition,FGFR1amplificationbyscoreCandscoreEcorrelated

with

Results

Patients’clinicopathologiccharacteristics

Ofthe369patients,251(68.0%)weremaleand118(32.0%)werefemale;139(37.7%)weresquamouscellcarcinomasand230(62.3%)wereadenocarcinomas.Themedianagewas65years(range21–84).ThemajorityofthepatientswithSqCCweresmokers(130/139,93.5%),whereasthosewithadenocarcinomawerepredominantlynon-smokers(133/230,57.8%;p<0.001)(Table1).FGFR1GCN

ThemedianFGFR1GCNwas2.0(range1.1–20.9)inalltheNSCLCcases:2.5(range1.1–20.9)forSqCCand1.9(range

Fig.1Representativefibroblastgrowthfactorreceptor(FGFR1)andchromosome8(CEP8)copynumberpatternsbyfluorescentinsituhybridization(FISH).anormal;bFGFR1amplification

FGFR1 肺癌

550

Table1ClinicopathologiccharacteristicsandFGFR1andchromosome8statusCharacteristic

AllpatientsNo.(%)369

65(59.0to71.0)21to84251(68.0)118(32.0)142(38.5)227(61.5)2.9(2.2to4.5)0.4to16135(36.6)160(43.4)74(20.1)205(55.6)164(44.4)161(43.6)103(27.9)96(26.0)9(2.4)221(59.9)148(40.1)211(57.2)158(42.8)311(84.3)58(15.7)2.0(1.7to2.8)1.1to20.92.3(1.9to2.9)1.2to7.01.0(0.9to1.1)0.2to10.5

SquamouscellcarcinomaNo.(%)139

67(62.0to72.0)39to84131(94.2)8(5.8)9(6.5)130(93.5)3.5(2.4to5.0)0.4to14.540(28.8)64(46.0)35(25.2)66(47.5)73(52.5)43(30.9)63(45.3)32(23.0)1(0.7)96(69.1)43(30.9)85(61.2)54(38.8)112(80.6)27(19.4)2.4(1.9to4.8)1.1to20.92.5(2.0to3.3)1.2to7.01.1(0.9to1.4)0.4to10.5

VirchowsArch(2014)465:547–558

AdenocarcinomaNo.(%)230

64(56.0to70.0)21to82120(52.2)110(47.8)133(57.8)97(42.2)2.8(2.1to4.0)0.5to16.095(41.3)96(41.7)39(17.0)139(60.4)91(39.6)118(51.3)40(17.4)64(27.8)8(3.5)125(54.3)105(45.7)126(54.8)104(45.2)199(86.5)31(13.5)1.9(1.7to1.9)1.1to12.32.2(1.9to2.7)1.3to6.81.0(0.8to1.0)0.2to5.0

No.ofpatientsAge(years)Median(IQR)RangeSexMaleFemale

SmokinghistoryNon-smokerSmoker

Tumorsize(cm)Median(IQR)RangepTstageT1T2T3

pNstageN0

N1,2,and3pTNMstageI

IIIIIAIIIB

PleuralinvolvementAbsentPresent

LymphaticinvasionAbsentPresent

VenousinvasionAbsentPresent

FGFR1genecopynumberMedian(IQR)Range

CEP8copynumberMedian(IQR)Range

FGFR1/CEP8ratioMedian(IQR)Range

FGFR1fibroblastgrowthfactorreceptor1,IQRinterquartilerange,CEP8chromosome8

nodalmetastasis(p=0.028andp=0.031,respectively)inad-enocarcinoma.Incontrast,FGFR1amplificationbyscoreBcorrelatedwithhighTNMstage(p=0.014)inSqCCandlymphaticinvasion(p=0.039)inadenocarcinoma.

FGFR1 肺癌

VirchowsArch(2014)465:547–558551

Fig.2aFGFR1orbchromosome8copynumbercratio(FGFR1/CEP8)indifferentlunghistologicalsubtypesrevealedsignificantdifference

ClinicalimplicationsofFGFR1amplification

FGFR1amplificationwasfoundin8.7%(32of369)ofNSCLCcases(25of139SqCCcases[18.0%]and7of230adenocarcinomacases[3.0%]).ThefrequencyofFGFR1amplificationwassignificantlyhigherinmales(p<0.001),smokers(p<0.001),andpatientswithSqCC(p<0.001).Intheadenocarcinomasubgroup,FGFR1amplificationwassig-nificantlyassociatedwithmalegender(p=0.015)andsmokinghistory(p=0.044),whereasintheSqCCsubgroup,ithadnostatisticallysignificantassociationwithanyclinico-pathologicalparameter(Table3).

Onunivariateanalysis,FGFR1amplificationwasassociatedwithpoorOS(58.6vs80.0months;p=0.033;Fig.3a)andshorterDFS(58.5vs80.0months;p=0.042;Fig.3b)inpatientswithSqCC.Incontrast,itwasnotcorrelatedwithOSandDFSinpatientswithNSCLCnorinthosewithadenocarcinoma.ForpatientswithSqCCwithinthesamestagesubgroup,FGFR1amplificationwasnotassociatedwithoutcome(datanotshown).MultivariateanalysiswasperformedusingtheCoxproportionalhazardmodeladjustedforpleurainvasion,lymphaticinvasion,venousinvasion,andpathologicalTNM(pTNM)stage.Whilevenousinvasion(p=0.013)andpTNMstage(p=0.010)wereproventobeindependentprognosticfactors,FGFR1amplifica-tionfailedtoshowindependencyforOSandDFS(OS:HR=1.79,95%CI=0.83–3.87,p=0.139;DFS:HR=1.73,95%CI=0.93–3.21,p=0.081;SupplementaryTableS1).

AssociationbetweenFGFR1amplificationandcigarettesmokinghistory

ThefrequencyofFGFR1amplificationwassignificantlyhigherincurrentorformersmokersthaninneversmokersinNSCLC(p=0.002)andadenocarcinoma(p=0.031)(Fig.4).InNSCLC,thefrequencyofFGFR1amplificationwas50.0%(16/32),50.0%(16/32),and6.3%(2/32)incurrent,former,andneversmokers,respectively.InNSCLCand

adenocarcinoma,thefrequencyofFGFR1amplificationsig-nificantlyincreasedwithcigarettesmokingpackyearhistory(p<0.001andp=0.021,respectively).InSqCC,nostatisticallysignificantcorrelationwasfoundbetweenFGFR1amplifica-tionandsmokingpackyearhistory(p=0.636).FGFR1proteinexpression

FGFR1proteinexpressionwasstudiedbyIHCon358cases.Weakcytoplasmicand/ormembranousexpressionwasfoundinnormalbronchialepithelialcellsbutnotinalveolarpneumocytes.Somealveolarmacrophages,necroticexudates,andstromalcellsshowedweakcytoplasmicand/ornuclearstaining.HomogeneousFGFR1expressionwasdetectedinthecytoplasmandonthecellmembraneoftumorcells.FGFR1expressionwasmorecommonlyobservedinadeno-carcinoma(120/221,54.3%)thaninSqCC(56/137,40.9%)(p=0.046).Ofthe32NSCLCcaseswithFGFR1amplifica-tion,17(53.1%)wereFGFR1-IHC-positive:13(52.0%)SqCCand4(66.7%)adenocarcinomas.FGFR1expressionwassignificantlycorrelatedwithadenocarcinoma(p=0.017)andlymphaticinvasion(p<0.001)butnotwithotherclinico-pathologicalparametersinNSCLC(SupplementaryTableS2).IntheSqCCsubgroup,FGFR1expressionwasrelatedtosmokinghistory(p=0.032)andabsenceofnodalmetastasis(p=0.015),whileintheadenocarcinomasubgroup,itcorrelatedwithlymphaticinvasion(p<0.001).However,FGFR1expressiondidnotshowanyrelationshipwithpatientoutcome.Inaddition,FGFR1GCNgaindidnotcorrelatewithproteinexpression(rho=0.08;p=0.123).

Discussion

Thekeyfindingsofourstudyareasfollows:(1)FGFR1amplificationismorecommoninsmokersandinpatients

with