Is there any improvement in clinical staging with 18F-FDG PET/CT compared to surgical staging in cases of lung cancer

Original

Article

Is

there

any

improvement

in

clinical

staging

with

18

_F-FDG

_PET/CT

compared

to

surgical

staging

in

cases

of

lung

cancer?

夽

Gonca

Kara

Gedik

,

Farise

Yilmaz

SelcukUniversity,FacultyofMedicine,DepartmentofNuclearMedicine,Konya,Turkey

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received15January2019 Accepted13May2019 Availableonline25June2019 Keywords: Clinical FDGPET/CT Lungcarcinoma Quantification Staging

a

b

s

t

r

a

c

t

Objective:18_{F-Fluorinefluorodeoxyglucosepositronemissiontomography/computedtomography(}18

F-FDGPET/CT)imagingisconsideredthestandardimagingmodalityforpatientswithnon-smallcelllung carcinoma.Theaimofthisstudywastocompareclinicalstaging(cTNM)performedwith18_F-FDGPET/CT

andsurgicalstaging(sTNM)inpatientswithnon-smallcellcarcinomatreatedwithsurgery.

Materialsandmethods:Weperformedaretrospectiveanalysisof99surgicalpatientswithnon-small cellcarcinomawhounderwent18_{F-FDGPET/CTexamination.Semiquantitativemeasureswere}

calcu-latedfromtheprimarylesionsandmediastinallymphnodes.FindingsofcTNMwerecomparedwith finalsurgical–pathologicalevaluation.Subjectsweredividedintotwogroups aspostsurgicalcTNM changedandcTNMunchanged.PatientsinthecTNMchangedgroupwerefurtherclassifiedas postsurgi-calupstaged(US)anddownstaged(DS).ResultsoftheUSpatientswerecomparedwiththeresultsofthe remainingpatientsconsistingofcTNMunchangedandDStoevaluatethepredictablerolesof semiquan-titativeparametersforpostsurgicalupstaging.Todeterminemediastinaltumoralinvolvement,cut-off valueswereobtainedfromcalculatedsemiquantitativeresultsofFDGuptakesinlymphnodes.Apvalue <.05wasconsideredstatisticallysignificant.

Results:Subjectswereaged40–82yearswithameanageof64.78±8.70years.Classificationagreement wasobservedin43patients(43%)andin57%,postsurgicalstagemigrationwasseen.Concurrenceof cTNMandsTNMwasmorepronouncedintheT1andN0subsetswhichwere84%and74%,respectively. ThelowestconcurrencewasobservedinN1classificationfollowedbyT4andN2(1%,50%and58%, respectively).ChangeinTstagingoccurredin20of56(36%),inNstaging22of56(39%)andchangeinT andNin14patients(25%).DistributionofUSandDSpatientsinthecTNMchangedgroupwas,43%(24 of56)and57%(32of56),respectively.Resultsofsemiquantitativemeasuresweresignificantlyhigherin USpatientsthantheresultsofthegroupconsistingofDSpatientsandcTNMunchangedpatients,forall parameters.Cut-offvaluecalculatedfrommediastinaluptakeswasmostspecificformetastasesinMTV (metabolictumorvolume)withanacceptablesensitivity(90%and67%,respectively).

Conclusions: TheconcordancebetweencTNMandsTNMwasbetterinstagingTcategorycompared toNstations.Semiquantitativemeasuresofprimarytumormayplayaroleinpredictingpostsurgical upstaging.TakingMTVintoconsiderationinthemediastinalregionmaybemorevaluablethanother parametersintheassessmentofnodalinvolvement.

©2019Sociedad Espa ˜noladeMedicinaNucleareImagenMolecular.PublishedbyElsevierEspa ˜na,S.L.U.Allrightsreserved.

¿Hay

alguna

mejora

en

la

estadificación

clínica

con

PET/TC

con

_F-FDG

_en

comparación

con

la

estadificación

quirúrgica

en

el

cáncer

de

pulmón?

Palabrasclave: Clínico FDGPET/TC Carcinomadepulmón Cuantificación Estadificación

r

e

s

u

m

e

n

Objetivo: La imagen detomografíapor emisión depositrones/tomografía computarizada con 18

F-fluordesoxiglucosa(18_{F-FDGPET/TC)seconsideraunamodalidaddeimagenestándarenpacientescon}

carcinomadepulmóndecélulasnopeque ˜nas.Elobjetivodeesteestudiofuecompararlaestadificación clínica(cTNM)realizadacon18_{F-FDGPET/TCylaestadificaciónquirúrgica(qTNM)enpacientescon}

carcinomadecélulasnopeque ˜nastratadosconcirugía.

Materialymétodos:Realizamosunanálisisretrospectivode99pacientesconcarcinomadecélulasno peque ˜nasintervenidosquirúrgicamentealosqueserealizóunestudio18_{F-FDGPET/TC.Serealizaron}

medidassemicuantitativasenlaslesionesprimariasylosganglioslinfáticosmediastínicos.Loshallazgos decTNMsecompararonconlaevaluaciónquirúrgica-patológicafinal.Lospacientessedividieronendos grupos,qTNMposquirúrgicomodificadoyqTNMnomodificado.Lospacientesenelgrupoconcambio deqTNMseclasificaronadicionalmentecomoupstaged(US)ydownstaged(DS)postquirúrgicos.Los

夽 Pleasecitethisarticleas:GedikGK,YilmazF.¿HayalgunamejoraenlaestadificaciónclínicaconPET/TCcon18F-FDGencomparaciónconlaestadificaciónquirúrgicaen elcáncerdepulmón?RevEspMedNuclImagenMol.2019;38:348–354.

∗ Correspondingauthor.

E-mailaddress:[email protected](G.K.Gedik).

resultadosdelospacientesUSsecompararonconlosresultadosdelospacientesrestantes,queconsistían enqTNMsincambiosyDS,paraevaluarelposiblepapelpredictivodelosparámetrossemicuantitativos enlaestadificaciónpostoperatoria.Paradeterminarlaafectaciónmediastínicatumoral,seobtuvieron valoresdecortedelosresultadossemicuantitativosdecaptacióndeFDGenlosganglioslinfáticos.Un valorp<0,05fueconsideradoestadísticamentesignificativo.

Resultados:Lospacientesteníanentre40y82a ˜nos,conunaedadmediade64,78±8,7a ˜nos.Seobservó unacuerdodeclasificaciónen43pacientes(43%),yen57%seobservómodificaciónenlaestadificación posquirúrgica.LaconcurrenciadecTNMyqTNMfuemásdestacadaenlaclasificaciónT1yN0,convalores de84%y74%,respectivamente.LaconcurrenciamásbajaseobservóenlaclasificaciónN1seguidadeT4 yN2(1%,50%y58%,respectivamente).ElcambioenlaestadificaciónTocurrióen20de56pacientes (36%);elcambioenlaestadificaciónN,en22de56(39%);yelcambioenTyN,en14pacientes(25%).La distribucióndepacientesdeUSyDSenelgrupoconcambiodecTNMfuede43%(24de56)y57%(32 de56),respectivamente.Losresultadosdelasmedidassemicuantitativasfueronsignificativamentemás altosenlospacientesdeUSquelosresultadosdelgrupocompuestoporpacientesconDSyqTNMsin cambios,paratodoslosparámetros.Elvalordecortecalculadoapartirdelascaptacionesmediastínicasfue másespecíficoparametástasisenelMTV(volumenmetabólicotumoral)conunasensibilidadaceptable (90%y67%,respectivamente).

Conclusiones: LaconcordanciaentreTNMyqTNMesmejorenlaclasificacióndelacategoríaTen com-paraciónconlacategoríaN.Lasmedidassemicuantitativasenlostumoresprimariospuedenjugarun papelenlaprediccióndelUSposquirúrgico.ElMTVenlaregiónmediastínicapuedetenerunmayorvalor queotrosparámetrosenlaevaluacióndelaafectaciónganglionar.

©2019SociedadEspa ˜nolade MedicinaNucleareImagenMolecular.PublicadoporElsevierEspa ˜na,S.L.U.Todoslosderechosreservados.

Introduction

Lungcarcinomaistheleadingandthesecondmostcommon causeofcancerdeathamongmenand women.1 _Accurate stag-ing, which includes delineation of the extent of disease, both loco-regionallyandatdistantsites,isimportantforselectingthe appropriatetherapeuticregimen.Moreover,staging,whichisbased onprimarytumor(T),regionallymphnodesand(N)andmetastases (M)system,isthesinglemostimportantprognosticfactorfor pre-dictingtheoutcomesofbothsurgicalandnonsurgicalpatientswith lungcarcinoma.2

Currently, two primary lung cancer staging procedures are available: clinical staging, which is provided by noninvasive imaging procedures, or minimally invasive approaches, such as mediastinoscopy.3 _{For the pathologic staging, information is} obtainedfrominvasivesurgeryandthepathologicevaluationofthe excisedtumor.Clinicalstagingcandonewithpreoperative imag-ingtechniquesincludingcomputedtomography(CT)andpositron emissiontomography/computedtomography(PET/CT)performed with18_{Fluorinefluorodeoxyglucose(}18_{F-FDG).Withthesizebased} criteriaformalignancy,CThasalimitedsensitivityandspecificity, especially in nodal staging of mediastinum.4 _{Mediastinoscopy,} whichisconsideredtobegoldstandardformediastinallymphnode staging,isunreliablein<10%ofcasesforlymphnodesmeasuring <1cm.5_{Surgicalstaging,ontheotherhand,makesavailablegreater} certaintyaboutprognosisandisusefulinmakingdecisionsabout adjuvanttherapyaftersurgery.6_{AstudycomparingimagingTNM} andpathologicalTNMinstaginglungcarcinomawithCTpointed outthelowdiagnosticaccuracyofclinicalstagingascomparedwith pathologicalstaging.6,7

18_{F-FDGPET/CTisincreasinglybeingusedfordiagnosis,staging} andforevaluationoftherapyresponseinvarioustumorsincluding lungcarcinoma.Improvementinstagingofpatientswithlung car-cinomahasbeenshownwithintegrated18_{F-FDGPET/CTsystems} comparedtoPETsystemsalone.8,9_{Besideshavingtheabilityof} ana-lyzingcellmetabolism,18_{F-FDGPET/CTalsoofferssemiquantitative} metabolicparametersofthetumorwhichgiveinformationabout metastaticpotentialandriskofrecurrence.Althoughmaximum standarduptakevalue(SUVmax)isthemostcommonlyusedone inclinicalpractice,theroleofmetabolictumorburdenasmeasured withmetabolictumorvolume(MTV)ortotallesionglycolysis(TLG)

hasbeenintroducedandhavebeensuggestedasyielding prognos-ticinformationonsurvivalandgivingbetterprognosticmeasures thanSUVmaxinpatientswithnonsmallcelllungcancer.2,10–12

Theaimofthisstudywastocompareclinicalstagingperformed with18_{F-FDGPET/CTandsurgicalstaginginpatientswithlung} car-cinomatreatedsurgically.Theroleofsemiquantitativeparameters derivedfromprimarytumorandmediastinallymphnodes,as pre-dictorsofpostoperativeupstagingandmediastinalinvolvement, havealsobeenevaluated.

Materialsandmethods

Patients

Thefilesof691patientswithlungcarcinoma,whounderwent 18_{F-FDGPET/CTbetweenSeptember2012andDecember2018in} SelcukUniversityMedicalFacultyDepartmentofNuclearMedicine, wereretrospectivelyreviewed.Patientswithhistologicallyproven lung cancer, who had undergone an 18_{F-FDG PET/CT scanning} for stagingand had beensurgicallytreated afterimaging, were selected.Patientswithsmallcellcarcinoma,patientswhose patho-logicaldataofthesurgicalspecimencouldnotbetherecovered andcasesinwhominductiontherapywasdeliveredafter18_F-FDG PET/CT,wereexcludedfromthestudy.Thisstudywasapprovedby thelocalethicscommitteeofourfaculty(meetingdate:24.10.2018, decisionnumber:2018/366)andwritteninformedconsentform wasobtainedinallpatients.

18_{F-FDGPET/CTscanning}

18_{F-FDG PET/CTimaging wascarried outwith anintegrated} scanner(BiographmCT,Siemens,Germany)in8or9bedpositions withanacquisitiontimeof2minperbedpositionfromskullbase tothemidthigh.Allpatientswereaskedtofastatleast6hbefore theexaminationandbloodsugarlevelswereconfirmedtobeless than200mg/dLbeforetheinjectionof370MBqof18_F-FDG.Sixty minutesaftertheadministrationoftheradiopharmaceutical,alow doseunenhancedCTusingthe16sliceCTwithacquisition parame-tersof190mA,5mmslicethicknessand140kVwasperformedfor attenuationcorrection,whichwasfollowedbythePETacquisition.

Imageinterpretationandsemiquantification

18_{F-FDGPET/CTimageswereinterpretedandsemiquantitative} analysisof18_{F-FDGuptakewascalculatedonaSiemensSyngo.via} PET/CTworkstation.Ineachpatient,a1.0cmdiameterregionof interest(ROI)wasmanuallyplacedovertheprimarylunglesion andsuspiciouslymphnodestationsontheattenuationcorrected 18_{F-FDGPETimagesontheslicewiththehighest}18_F-FDGuptake. Theactivity concentrationwithin theROI was determined and expressedasSUVwhichwascalculatedusingtheformulabelow: activityintheregionofinterest(MBq/mL)/(injectedactivity/body size).From these ROIs,SUVmax and peak standardized uptake value(SUVpeak)werecalculated,bothofwhichwerenormalized tobodyweight.Fordeterminingmediastinalandhilarlymphnode status,werandomlyselected50patients,drewa1.0cmdiameter ROIinthecenterofsuperiorvenacavaandcalculatedthemeanof theirSUVmaxtorepresentbloodpoolSUV,whichwascalculatedas 3.30.ForMTVandTLGcalculation,40%oftheSUVmaxwasadopted asthethresholdtodelineatethetumor;MTVandTLGoftheprimary tumorandmediastinallymphnodewereautomaticallymeasured bythesoftware.

Dataanalysis

Ineachpatientsizeoftheprimarytumor,presenceof medi-astinal and/or hilar lymph nodes and distant metastases were evaluatedandpatientswereclinicallystaged(cTNM)accordingto theeightheditionofTNMclassificationforlungcancer,takingthe findingsofthe18_{F-FDGPET/CTexaminationintoconsideration.}1 Patientsinwhomnolymphnodesinthemediastinalregionwere observedortheoneswithnodaluptakeofSUVmaxbelow3.30; werestagedasN0.Anynodaluptakeoverthevalueof3.30was recordedandconsideredaspositive fortumoralinvolvementof mediastinaland/orhilarlymphnodes.Incasesofmultiplelymph nodeshaving18_{F-FDGuptakeaboveourthresholdinN2stations,} theresultofthehighest18_{F-FDGuptakewasusedinanalysis.} Semi-quantitativemeasurementsoflunglesionsandnodaluptakeswere noted.

Pathologicandsurgicaldataforallpatientswereobtainedfrom theelectronicmedicalrecordsystemofourhospital.Forsurgical staging(sTNM),thesameclassificationsystemwasused. Compar-isonofcTNMandsTNM wasperformedregarding sTNMasthe goldstandard.AftercomparingthecTNMandsTNM,subjectswere dividedinto2groups:cTNMchangedandcTNMunchanged post-operatively.PatientsofcTNMchangedpostoperativelygroup,were furtherclassifiedinto2groupsasupstaged(US)anddownstaged (DS).Besideschangesbetweendifferentstages,changeswithinthe samestage(e.g.:changefromstageIA3toIA2orfromstageIIA toIIB)wasalsotakenintoaccountandregardedasdownstaged orupstaged,respectively.Inordertoevaluatethepredictiveroleof semiquantitativeparameteresforpostsurgicalupstaging;resultsof patientsinUSgroupwascomparedwiththeresultsofthe remain-ingpatientsconsistingofcTNMunchangedandDS.

Statisticalanalysis

ThesoftwarepackageSPSSv16.0wasusedtocarryoutthe sta-tisticalanalysis.ResultsofSUVmax,SUVpeak,MTVandTLGwere expressedasmean±SDormedian.Mann–WhitneyUtestwasused tocomparetheresultsofsemiquantitativeanalysisoftheprimary tumorindifferenthistologicaltypesandalsotocomparetheresults ofpatientsinUSandtheresultsofthegroupofpatients contain-ingcTNMunchangedandDS.ResultsofUSandDSpatientswere alsocomparedbyMann–WhitneyUtest.Fromsemiquantitative measuresobtainedfrommediastinallymphnodes,cut-offvalues werecalculatedfortheirabilitytopredictnodalmalignancy,by

ROCcurveanalysis.Thep-valueswereconsideredstatistically sig-nificantif<0.05.

Results

Our retrospective study included 99 patients (83 male, 16 female, age range: 40–82 years, mean age: 64.78±8.70). All patientsweretreatedsurgically.Systematiclymphnodedissection wascarriedoutinstations2,4,7,8,9,10inrightsidedtumorsand in5,6,7,8,9,10forleftsidedtumors.RemainingN1nodeswere resectedaspartofthesurgicalspecimen.Histologicalanalysisof theprimarytumorrevealedadenocarcinomain44patients, squa-mouscellcarcinoma(SCC)in46,largecellneuroendocrinetumor in6,carcinoidtumorin2andlargecellcarcinomain1patient.

Inallpatients,SUVmaxoftheprimarytumorwasdetermined.In 12patients,inwhichthe18_{F-FDGPET/CTstudieswerenotavailable,} thevaluesofSUVmaxofprimarytumorsandmediastinallymph nodeswererecordedfromthereportsofthepatients;theseresults wereincludedin thereview.However,SUVpeak, TLGandMTV oftheprimarytumorandmediastinallymphnodescouldnotbe calculated.ThemeanandmedianSUVmaxvalueswerehigherin patientswithSCCthanintheoneswithadenocarcinoma,which werecalculated as16.43±10.54and 14.62and 10.19±7.4 and 8.43, respectively(p=0.0001).Among the87 patientsinwhom allsemiquantitativeparameterscouldbecalculated(39patients withadenocarcinoma,39withSCC,9withotherhistologicaltypes), meanandmedianSUVpeak,TLGandMTVvalueswereallhigher inpatientswithSCCthaninpatientswithadenocarcinomaand thedifferenceswerestatisticallysignificant(p<0.05).Theresults ofsemiquantitativemeasurementsofprimarytumorsofdifferent histopathologictypesaredepictedinTable1.

Outof99patients,cTNMrevealedthat1patientwasinstageIA1, 10wereinIA2,20wereinIA3.Therewere20patientsincIB,7in cIIA,19incIIBandfinally15incIIIA,7incIIIB.Noneofthepatients werenotedtohavedistantmetastases.cTNMchanged postoper-ativelyin56patients(57%)andunchangedintheremaining43 (43%,Fig.1).sTNMdemonstratedthat,7patientswereinstageIA1, 13instageinIA2and16wereinsIA3.Therewere20patientsin sIBgroup,8insIIAand19insIIBcategory.SurgicalIIIAandIIIB classesincluded13and 3patients,respectively.Change in sub-stageoccurredin19of56patients(34%),whereasin37patients (66%)achangeofcTNMwasobservedbetweenstages.Comparison ofcTNMandsTNMissummarizedinTable2.

DiagnosticaccuracyincTstagingwashigherthanofcNstaging whichwerecalculatedas72%and62%,respectively.Concurrence ofcTNMandsTNMwasmorepronouncedinT1andN0subsets whichwere84%and74%,respectively.Thelowestconcurrencewas observedinN1classificationfollowedbyT4andN2(1%,50%and 58%,respectively).ChangeinTstagingoccurredin20of56(36%), inNstaging22of56(39%)andchangeinTandNin14ofthem (25%).Among56patientswhosecTNMwerechanged,upstaging wasobservedin24of56(43%,Fig.2)anddownstagingin32(57%,

Fig.3).InpatientsofUSgroup,Tupstagingwasobservedin10of 24(42%),Nupstagingin10of24(42%)andTandNupstagingin 4ofthem(16%).InpatientsofDSgroup,downstageinTstagewas presentin10of32(31%),downstaginginNstageoccuredin12of 32(38%)andTandNdownstagingwasobservedin10of32patients (31%).ComparisonsofcTNMandsTNMinTandNcategoriesare outlinedinTables3and4.

Theanalyses which werecarried out toinvestigatethe role ofsemiquantitativemeasuresinpredictingpostsurgical upstag-ing revealed that all parameteres were significantly higher in US patientsthan theresults ofthe groupconsisting of DS and cTNMunchangedpatients(p<0.05).Themeanandmedian val-uesofSUVmaxoftheprimarytumorinUSandDSgroupswere

Table1

SemiquantitativeresultsofprimarytumorsinofdifferenthistopathologicaltypesofNSCLC.

SUVmax SUVpeak TLG* _MTV*

Adenocarcinoma(mean/median) 10.19±7.4/8.43 7.71±6.43/5.98 21.85 5.72

SCC(mean/median) 16.43±10.54/14.62 12.43±8.71/11.58 74.30 8.76

pValue 0.0001 0.001 0.001 0.0039

NSCLC:non-smallcelllungcarcinoma,SCC:squamouscellcarcinoma.

*_{Onlymedianvaluesareexpressedbecauseofhighlevelsofstandarddeviationinmeanvalues.}

Fig.1.TransaxialPET(a),CT(b),andfused(c)imagesof68yearoldmalepatientshowedaprimarytumorinthelowerlobeoftherightlungwithalongestdiameterof 61mm,whichwasdiagnosedassquamouscellcarcinoma(arrow,c)InthetransaxialPET(d),CT(e),andfused(f)imagesofthesamepatient,subcarinallymphnodeswitha SUVmaxof16.47wereobserved,whichwerefinallyconfirmedasmetastatic(arrow,f).Calculatedsemiquantitativevaluesforlymphnodeswereallabovethecut-oflevels determinedforthem,includingMTV.ThispatientwasclinicallyandsurgicallystagedasT3N2.

Table2

Comparisonofclinicalandsurgicalstaginginwholepatients.

cTNM sIA1 sIA2 sIA3 sIB sIIA sIIB sIIIA sIIIB Total

IA1 1 1 IA2 3 6 1 10 IA3 3 9 5 3 20 IB 2 11 6 1 20 IIA 4 2 1 7 IIB 1 2 3 2 1 6 4 19 IIIA 2 2 2 1 3 4 1 15 IIIB 1 1 3 2 7 Total 7 13 16 20 8 19 13 3 99 Agreement(c/p) 14% 46% 56% 55% 50% 32% 31% 67% 43%

Fig.2. IntransaxialPET(a),CT(b),andfused(c)imagesofa68yearmalepatientwhowasclinicallystagedasT2aN0,aprimarytumorlocatedintheupperlobeofleftthe lungwasobserved(arrow,c).Inthehistopathologicalanalysis,besidesmaintumor,asatellitenoduleinthesamelobeconsistentwithadenocarcinomawasalsoreported. ThepatientwasfinallystagedassT3N0.

calculatedas14.52±6.02and13.91intheUSgroupcomparedto 13.05±11.92and9.47intheDSgroup,respectivelyandthe dif-ferencebetweenthemwasnotstatisticallysignificant(p=0.061). InthecTNMchanged group;SUVpeak, TLGandMTV couldbe calculatedin47patients,22wereUSand25wereDS.Meanand

medianvaluesofSUVpeakwere11.49±5.32and10.51inUSgroup and10.13±10.56and7.05inDSgroup,respectively.LikeSUVmax, thedifferenceofmedianvaluesbetweenUSandDSgroupswas notfoundasstatisticallysignificant(p=0.105).However,whenwe lookedattheTLGandMTVvaluesofprimarytumorsinUSandDS

Table3

Comparisonofclinicalandsurgicaltumorstaging.

cTNM sT1 sT2 sT3 sT4 Total T1 38 8 – – 46 T2 7 25 5 – 37 T3 – 2 8 3 13 T4 – – – 3 3 Total 45 35 13 6 99 Agreement(c/p) 84% 71% 62% 50% 72% Table4

Comparisonofclinicalandsurgicalnodalstaging.

cTNM sN0 sN1 sN2 sN3 Total N0 55 13 1 – 69 N1 7 2 1 – 10 N2 11 2 4 – 17 N3 1 1 1 – 3 Total 74 18 7 – 99 Agreement(c/p) 74% 1% 58% – 62% Table5

ResultsofsemiquantitativeparametersofprimarytumorsincTNMchangedgroup.

SUVmax SUVpeak* _TLG* _MTV*

US(mean/median) 14.52±6.02/13.91 10.51 94.44 10.67 DS(mean/median) 13.05±11.92/9.47 7.05 22.05 4.08

pValue 0.061 0.105 0.013 0.006

US:upstaged,DS:downstaged.

* _{Onlymedianvaluesareexpressedbecauseofhighlevelsofstandarddeviation} inmeanvalues.

groups,medianTLGwascalculatedas94.44inUSgroupand22.05 inDSgroup.MedianvalueofMTVinUSgroupwasagainhigherthan thevalueinDSgroup(10.67vs.4.08),andthestatisticalanalysis showedsignificantdifferenceforbothparameters(p=0.013and p=0.006,respectively).Theresultsofsemiquantitativeparameters oftheprimarytumorsforUSandDSpatientsareshowninTable5. Topredictmediastinallymphnodemetastases,acut-offvalue of5.05wascalculatedwiththesensitivityandspecificityof60% and64%,respectivelyforSUVmax.ForSUVpeak,cut-offvaluewas foundas3.84with67%sensitivityand75%specificity.Areasunder curveswere0.632forSUVmaxand0.731forSUVpeak.Ontheother hand,forvolumetrictumorparametersthecut-offvaluewas4.23 withasensitivityof67%andspecificityof90%forMTVand12.97 forTLGwithsensitivityandspecificityof78%and75%,respectively. Areasunderthecurveswerefoundas0.786forMTVand0.772for TLG.

Discussion

Studiescomparingclinicalandpathologicstaginginlung carci-nomaunderlinetheinaccuracyofclinicalstagingand,furthermore pooledresultssupportthatthetrueextentofintrathoracic dis-easeisunderoroverestimatedinspiteofthetechnologicadvances inimagingmodalities.4,6,7,13_{Gdeedoetal.reportedthat,imaging} TNMprovidesnoaccurateinformation,CTisinadequatefor medi-astinallymphnodestagingandT3andT4determinationsofCTare unreliableinlungcancer.7_{ThelimitedroleofCTin} demonstrat-ingmediastinalinvolvementandinT4definitioninpatientswith non-smallcarcinomawerealsoemphasizedinanotherstudyby C¸etinkayaetal.,later.4_{AnotherstudybyLopez-Encuentraetal.,} compared clinicalagainst pathological staging in 2994cases of lungcancer,foundlowdiagnosticaccuracyofclinicalstagingand finalizedtheirreportwiththehopesofintegratedPETsystemsto improveclinicalTNstaging.6_{Vermaetal.,lookedforananswer} forthisquestionrecently;reported42%TNMupstagedpatients

postoperativelyandfoundoutthathighprimarytumorSUVmax ofadenocarcinomamightpredictpostsurgicalupstaging.14

Wefoundaclinicalandsurgicalconcurrenceof43%and post-operativestagechangingwasobservedinmorethanhalfofthe patientsinwhomdownstagingwasmoreprominent.Inthe con-ductedstudiescomparingclinicalstagingwithsurgicalstagingin lungcarcinomabeginningfromtheearly1990s;56%,15_35%7 _and

47%4,6,13_{agreementlevelswerereportedinpatientsmajorityof}

whomunderwentCT.Althoughdifferentclinicalstagingmodality andaTNMstagingsystemconsistingofmoresubstageswasused, ourresultwassimilarwiththelevelof47%whichwasreported bymostoftheexperiences.Weobservedhigheraccuracyinthe determinationofTstatusthanNstations(72%and62%, respec-tively;Tables 3and4)whichwasalsoasimilarresultwiththe otherstudies.4,7,11,13,15

Overestimationorunderestimationofthestagingvariesamong studies and this is mostly caused by false negative or positive interpretationsofmediastinalnodalassessments.Withtheclinical introductionofofintegratedFDGPET/CTclinically,improvement initsdiagnosticpowerandaccuracyofstaginginnonsmallcell car-cinomacomparedtoPETsystemsalonehasbeenwellestablished. TheeffectivenessofFDGPET/CTsystemsindecreasingthenumber offutilethoracotomiesandalsoshowingunexpectedmetastatic diseaseisdemonstratedaswell.16,17_{IntegratedFDGPET/CTisalso} usefulforpredictingnegativityofthemediastinallymphnode sta-tus,howeverfalsepositiveresultscausedbyinflammatoryprocess hasbeenreportedinmediastinalstationswhichwasalsoobserved byus.Elevenofthe17patients(65%)whowereclassifiedascN2 and7of 10patients(70%) ofcN1werefinallystagedassN0in ourstudy(Table4).Thereisnotanyconsensusintheliterature forcut-offvaluesofSUVmaxinmediastinallymphnodesinorder toclarifytheonescontainingtumorcellsandrecommended val-uesrangefrom2.5–5.3.18_{Becauseofthiswidespectrumofcut-off} values,wecalculatedthemeanoftheSUVmaxofbloodpoolfrom randomlyselectedpatientsassuggestedintheliterature.18 How-everoureffortsofcalculatingameanvalueinsteadofregardingan arbitrarycut-offcouldnothelpusandtheunreliabilityof mediasti-nalnodalassessmentstillexistedinourstudy.False-positivelymph nodesexceedingourthresholdvalueof3.30wererelatedto inflam-matoryreactionincludingallof3patientswhowereclassifiedas cN3whofirstgonethroughmediastinoscopyandlatertoracotomy whenmetastaticinvolvementwasruledout(Figs.3and4).Onthe othersideofthecoin,13of18patients(72%)ofsN1category,were falsenegativelystagedascN0andonly2ofthem(1%)couldbe cor-rectlyclassified(Table4).Thisinaccuracyofpredictinghilarlymph nodeinvolvementwasalsoobservedbyC¸etinkayaetal.andLin etal.,4,18_{Al-Sarrafetal.,showedthatthedistinctionbetween} pri-marytumorandadjacentlymphnodeinvasioncanbedifficultin casesofhilartumors.19_{Inourstudy,7of13(54%)patientshadhilus} locatedtumorallesionswhichsupportsthishypothesis;howeverin 6,thelocationsoftumorswereoutsidethehilum,butwewerestill unabletodiscriminateperibronchialandintrapulmonarylymph nodeinvolvement,theresolutionofcurrentFDGPET/CTsystems.

In terms of cTstaging, thehighest diagnosticaccuracy was observedinT1assessments(84%)andthelowestinT4(50%).When welookedatthesubstagesofT1category;FDGPET/CTwasable todiscriminateT1ctumorsbetterthanothercategoriesandwas worstatdefinitionofT1a(65%and44%clinicalandsurgical con-cordance,respectively).Inthe8theditionTNMstagingsystem,7 categoriespersizeascomparedwith6thecategoriesexistwith1 moresubdivisioninT1categorywhichmakesdifficulttoresolve thetumorsofsmallerthan1cmstandinginT1acategorywhichwas alsomentionedbyKandathiletal.20_{Althoughthespatialresolution} hasimprovedfrom25mmto5mmwithimprovementin scan-nerdesignandperformance,21 _{inaccuraciesexistinmicroscopic} determinationoftumormarginswhichwasalsoemphasizedby

Fig.3. TransaxialPET(a),CT(b),andfused(c)imagesofa68yearoldfemalepatientdepictedtheprimarytumorintheupperlobeoftherightlung(arrow,c).Mediastinal windowoftransaxialPET(d),CT(e),andfused(f)imagesofthesamepatientshowedlymphnodewithelevatedFDGuptakeintherightparatrachealregion(arrow,f). SUVmaxofthelymphnodewascalculatedas9.51,thepatientwasclinicallystagedascIIIA.Histopathologicevaluationrevealedinflammatoryreactioninthelymphnode andconfirmedadenocarcinomaintheprimarytumorsothefinalstageofthepatientwassIA2.ThecalculatedMTVlevelfromthislymphnodewasfoundas3.98.

Fig.4.IntransaxialPET(a),CT(b),andfused(c)imagesofa70yearoldmalepatient,theprimarytumorlocatedinthelefthilium(arrow,c)whichthenprovedas adenocarcinoma,wasseen.TransaxialPET(d),CT(e),andfused(f)imagesofsamepatientshowedincreasedFDGuptakewithcalculatedresultsof7.70and3.88forSUVmax andMTVrespectivelyinrighthilarlymphnodes(arrow,f).BeingclinicallystagedascIIIB,thefinalstageturnedouttosIIBwithinflammatoryreactionintherighthilar lymphnodeswhichwereremovedbymediastinoscopy,however,metastaticinvolvementintheleftonesinthefinalpathologicevaluationofthesurgicalspecimen.

Vermaetal.,14_{Inourstudy,agreementinT3determinationwas} higherthaninT4(62%vs.50%,respectively)butwasstilllower thanT1andT2stages.AmongsT4patients,1ofthemwasfalsely classifiedascT3becauseofaseparatelesionindifferentlobeof thesamelungwasnotedintheexcisedpathologyspecimenand thiswasnotobservedbyPETscan.Intheremaining2ofsT4and 5of13ofsT3patientswhowerefalselyclassifiedascT3andcT2 respectively,thelongestdimensionofthetumorwasmeasured smallerthantheactualsizewhichmaybesecondaryto consolida-tivechangesaroundthetumorwhichwerenotregardedaspart ofmalignancy.OurresultofobservingthesuccessofcTNMinT1

andT2stagingandunreliabilityindeterminationofT3andT4 sta-tusisconcordantwiththeresultsofotherstudiesreportedinthe literature.6,7

Althoughthehighestdiagnosticaccuracyof67%seemstobe observedincIIIBstageinourstudywhichisdiscordantwiththe otherresultsreportedbefore,wethoughtthatthisfindingevolved secondarytothelownumberpatientsinsIIIBgroupwhich con-sistedonly3patients(Table2).FollowedbyclinicalIIIBstage,cIA3 tookthesecondhighestclinicalandsurgicalconcurrenceratioof 56%whichwasconcordantwiththehighagreementlevelinstage IobservedbyLopez-Encuentraetal.6

Oneof the main advantages of PET imaging is its ability of quantifyingthemetabolicactivityofthetumorwhichisunique forthismodality.Besidesusingfordifferentiatebetweenbenign andmalignant lesions,semiquantitativemeasurements arealso valuableforevaluation prognosisand assessingtheresponseto treatmentin oncologyarmamentarium. Amongthese,SUVmax, whichrepresentsthehighest voxelvaluewithintheROI, isthe mostcommonly used one becauseof its observer-independent natureandbeingavailableincommercialworkstations.However, SUVmaxcanbesubstantiallyaffectedbyimagenoise,thiscanbe eliminatedbycalculatingSUVpeak,whichwassuggestedamore robustalternativethanSUVmax.22_{DuanXYetal.reportedthatthe} histopathologictypeandgradeoftumorwererelatedtoSUVmax inpatientswithlungcancer.23_{Vermaetal.reportedthatSUVmax} oftheprimarytumorwasfoundasapredictorforupstaging in patientswithadenocarcinomaoflungcancerwhichmayserveas asurrogateforpreoperativenodalsamplingoflymphnodeswith equivocalFDGuptake.14_{Inourstudy,allofthesemiquantitative} parameterscalculatedfromtheprimarytumorweresignificantly higherinsquamouscellcarcinomathanadenocarcinomawhich demonstratesthedifferenceinmetabolismaccordingtohistologic subtypeswhichwasalsoreportedbyKimetal.24_{Concordantwith} theresultsofVermaetal.,14_{resultsofSUVmaxandaswellasother} semiquantitativeparameters,wereallsignificantlyhigherin US patientscomparedtootherpatients.However,onthecontraryto studiessupportingSUVasarespresentativeparameteroftumor aggressiveness,SUVmaxandSUVpeakvaluescalculatedfromthe primarytumorwerenotsignificantlydifferentinpatientsbetween USandDSinourstudy.ThismaybebecauseSUVmaxshowingthe highestintensityofFDGuptakewithintheregionofinterestcannot representtotaltumoruptakefortheentiretumormass.Resultsof MTVandTLGderivedfromprimarytumorontheotherhand,were significantlyhigherinUSpatientsthanDSpatientswhichmay sup-portthehypothesisoftheirbeingbetterprognosticmeasuresfor diseaseprogressionthanSUVmaxreportedforsurgicaland nonsur-gicalpatientswithnon-smallcellcarcinomabefore.2_{Forpredicting} malignancyinmediastinallymphnodes,theareaundercurvein ROCanalysiswasthelargestforMTVandcalculatedcut-offlevel forthisparameterwasfoundasthemostspecificonefor predict-inglymphnodemetastaseswithanacceptablesensitivityinour study.Buttheproblemofwhetherthesemiquantitativedataof onereportcanbetranslatablefromonecentertoanotherisstill unansweredandadoptingstandardizedtechniquesfor measure-mentsoftheprimarytumorandaswellofthelymphnodeshave beenrecommendedbyBryantetal.,25_{Wethinkthat,determining} centerspecificcut-offvaluesforsemiquantitativeparametersmay helpdeterminingpostsurgicalupstagingandclarifylymphnodes harboringtumoralinvolvement.

Thelimitationsofourstudyareitsretrospectiveoriginand rel-ativelylownumberofpatientsinsubgroups.Furtherstudieswith largernumberofpatientsareneededtovalidatetheroleofMTVin predictingmedistinallymphnodemetastases.Excludingpatients withmetastasesbecauseofstudydesignalsolimitedevaluatingthe roleofFDGPET/CTinidentifyingunexpectedmetastaticdisease.

Inconclusion,clinicalstagingofTstatusismoreaccuratethan nodalstaging with FDGPET/CT in patientswithnon-small cell carcinoma.ThequantitativenatureofFDGuptakeoftheprimary tumorplayroleinpostsurgicalupstaging.VolumedependentPET parametersmaybemorevaluablethanSUVmaxinpredictingnodal involvement.Adoptingcenterspecificcut-offvaluesforMTVfor interpretationofFDGuptakesin mediastinal lymphnodesmay improvethediagnosticaccuracyofFDGPET/CTinnodal assess-mentswhichenablescorrectclinicalstaging.

Conflictofinterest

Theauthorshavenoconflictsofinteresttodeclare.

References

1.KayFU,KandathilA,BatraK,SabooSS,AbbaraS,RajiahP.RevisiontotheTumor, NodeMetastasisstagingoflungcancer(8thedition):rationale,radiologic find-ingsandclinicalimplications.WorldJRadiol.2017;9:269–79.

2.ZhangH,WroblewskiK,LiaoS,KampalathR,PenneyBC,ZhangY,etal.Prognostic valueofmetabolictumorburdenfrom18_{F-FDGPETinsurgicalpatienytswith}

non-smallcelllungcancer.AcadRadiol.2013;20:32–40.

3.KligermanS,AbbottG.AradiologicreviewofthenewTNMclassificationsystem. AmJRoentgenol.2010;194:562–73.

4.CetinkayaE,TurnaA,YildizP,DodurgaliR,BedirhanMA,GürsesA,etal. Com-parisonofclinicalandsurgical–pathologicalstagingofpatientswithnon-small celllungcarcinoma.EurJCardiothoracSurg.2002;22:1000–5.

5.PerigaudC,BridjiB,RousselJC,SaganC,MugniotA,DuveauD,etal.Prospective preoperativemediastinallymphnodestagingbyintegratedpositronemission tomography-computerisedtomographyinpatientswithnon-smallcelllung cancer.EurJCardiothoracSurg.2009;36:731–6.

6.Lopez-EncuentraA,Garcia-LujanR,RivasJJ,Rodriguez-RodriguezJ,Torres-Lanza J,Varela-SimoG,etal.Comparisonbetweenclinicalandpathologicalstagingin 2994casesoflungcancer.AnnThoracSurg.2005;79:974–9.

7.GdeedoA,SchilPV,CorthoutsB,MieghemFV,MeerbeeckJV,MarckEV,etal. ComparisonofTNM[(i)TNM]andpathologicalTNM[pTNM]instagingof bron-chogeniccarcinoma.EurJCardiothoracSurg.1997;12:224–7.

8.CerfolioRJ,OjhaB,BryantAS,RaghuveerV,MountzJM,BartolucciAA.The accu-racyofintegratedPET-CTcomparedwithdedicatedPETaloneforthestagingof patientswithnonsmallcelllungcancer.AnnThoracSurg.2004;78:1017–23.

9.LardinoisL,WederW,HanyTF,KamelEM,KoromS,SeifertB,etal.Stagingof non-small-celllungcancerwithintegratedpositron-emissiontomographyand computedtomography.NEngJMed.2003;348:2500–7.

10.RahimMK,KimSE,SoH,KimHJ,CheonGJ,LeeES,etal.Recenttrendsin PETimageinterpertationsusingvolumetricandtexture-basedquantification methodsinnuclearoncology.NuclMedMolImaging.2014;48:1–15.

11.LeeP,BazanJG,LavoriPW,WeerasuriyaDK,QuanA,LeQT.Metabolictumor volumeisanindependentprognosticfactorinpatientstreateddefinitivelyfor non-small-celllungcancer.ClinLungCancer.2012;13:52–8.

12.LiaoS,PenneyBC,WroblewskiK,ZhangH,SimonCA,KampalathR,etal. Prog-nosticvalueofmetabolictumorburdenon18_{F-FDGPETinnonurgicalpatients}

withnon-smallcelllungcancer.EurJNuclMedMolImaging.2012;39:27–38.

13.FernandoHC,GoldstrawP.Theaccuracyofclinicalevaluativeintrathoracic staginginlungcancerasassessedbypostsurgicalpathologicstaging.Cancer. 1990;65:2503–6.

14.VermaS,ChanJ,ChewC,SchultzC.PET-SUVmaxandupstagingoflungcancer. HeartLungCirc.2019;28:436–42.

15.BülzebruckH,BoppR,DringsP,BauerE,KrysaS,ProbstG,etal.Newaspectsin thestagingoflungcancer.Cancer.1992;70:1102–10.

16.VanTinterenH,HoekstraOS,SmitE,vandenBerghJH,SchreursAJ,Stallaert RA,etal.Effectivenessofpositronemissiontomographyinthepreoperative assessmentofpatientswithsuspectednon-smallcelllungcancer:thePLUS multicentretrial.Lancet.2002;359:1388–93.

17.WederW,SchmidRA,BruchhausH,HillingerS,SteinertHC,vonSchulthessGK, etal.Detectionofextrathoracicmetastasesbypositronemissiontomography inlungcancer.AnnThoracSurg.1998;66:86–92.

18.LinWY,HsuWH,LinKH,WangSJ.RoleofpreoperativePET-CTinassessingin earlystagelungcancer.JChinMed.2012;75:203–8.

19.Al-SarrafN,AzizR,GatelyK,LuceyJ,WilsonL,McGovernE,etal.Patternand predictorsofoccultmediastinallymphnodeinvolvementinnon-smallcelllung cancerpatientswithnegativemediastinaluptakeonpositronemission tomog-raphy.EurJCardiothoracSurg.2006;82:417–22.

20.KandathilA,KayFU,ButtYM,WachsmannJW,SubramaniamRM.RoleofFDG PET/CTintheeighteditionofTNMstagingofnon-smallcelllungcancer. Radio-graphics.2018;38:2134–49.

21.ZanzonicoP. Positronemission tomography:a reviewof basicprinciples, scannerdesign and performance, and current systems. Semin NuclMed. 2004;34:87–111.

22.PakK,KimSJ.WhatdowemeasureinOncologyPET?NuclMedMolImaging. 2017;51:212–6.

23.DuanXY,WangW,LiM,LiY,GuoYM.Predictivesignificanceof standard-izeduptakevalueparametersofFDG-PETinpatientswithnon-smallcelllung carcinoma.BrazJMedBiolRes.2015;48:267–72.

24.KimE,WuHG,KeamB,KimTM,KimDW,PaengJC,etal.Significanceof 18F-FDGPETparametersaccordingtohistologicsubtypeinthetreatmentoutcomeof stageIIInon-small–celllungcancerundergoingdefinitiveconcurrent chemora-diotherapy.ClinLungCancer.2019;20:e9–23.

25.BryantAS,CerfolioRJ,KlemmKM,OjhaB.Maximumstandarduptakevalue ofmediastinallymphnodesonintegratedFDG-PET-CTpredictspathologyin patientswithnon-smallcelllungcancer.AnnThoracSurg.2006;82:417–23.