Improved Prostate Cancer Biopsy Grading by Incorporation of Invasive Cribriform and Intraductal Carcinoma in the 2014 Grade Groups

Platinum

Priority

–

Prostate

Cancer

EditorialbyXXXonpp.x–yofthisissue

Improved

Prostate

Cancer

Biopsy

Grading

by

Incorporation

of

Invasive

Cribriform

and

Intraductal

Carcinoma

in

the

2014

Grade

Groups

Geert

J.L.H.

van

Leenders

*

,

Charlotte

F.

Kweldam

,

Eva

Hollemans

,

Intan

P.

Ku¨mmerlin

,

Daan

Nieboer

,

Esther

I.

Verhoef

,

Sebastiaan

Remmers

,

Luca

Incrocci

,

Chris

H.

Bangma

,

Theodorus

H.

van

der

Kwast

,

Monique

J.

Roobol

a_{DepartmentofPathology,ErasmusUniversityMedicalCentre,Rotterdam,TheNetherlands;}b_{DepartmentofUrology,ErasmusUniversityMedicalCentre,}

Rotterdam,TheNetherlands;c_{DepartmentofPublicHealth,ErasmusUniversityMedicalCentre,Rotterdam,TheNetherlands;}d_{DepartmentofRadiotherapy,}

ErasmusUniversityMedicalCentre,Rotterdam,TheNetherlands;e_{LaboratoryMedicineProgram,UniversityHealthNetwork,Toronto,Canada}

a v ai l a b l e a t w w w . s c i e n c e d i r e c t . c o m j o u r n al h o m e p a g e : w w w . e u r o p e an u r o l o g y . c o m Articleinfo Articlehistory: AcceptedJuly31,2019 AssociateEditor: MatthewCooperberg StatisticalEditor: MelissaAssel Keywords: Cribriform Gleasonscore Grade Intraductalcarcinoma Prostatecancer Abstract

Background: Gradegroups (GGs) are an important parameter for therapeutic decision makinginprostatecancer(Pca)patients.Invasivecribriformand/orintraductalcarcinoma (CR/IDC)hasanindependentprognosticvaluefordiseaseoutcome,butarenotincludedin theGGlimitingtheirclinicaluse.

Objective: Toperformaproof-of-principlestudyincorporatingCR/IDCinthecurrentGG. Design,setting,andparticipants: Allprostatebiopsiesof1031menwithscreen-detectedPca between1993and2000werereviewedforthecurrentGG(rangingfrom1to5)andCR/IDC. Thecribriformgrade(cGrade)wasequaltotheGGifCR/IDCwaspresentandGGminus1if not.GG1wascGrade1ifintraductalcarcinomawasabsent.

Intervention: BiopsyreviewforGGandCR/IDC.Atotalof406patientshadreceivedradical prostatectomy(RP),508radiotherapy(RT), 108surveillance,andeighthormonaltherapy,and thetreatmentwasunknownforonepatient.

Outcomemeasurementsandstatisticalanalysisdisease-specificsurvival(DSS), metastasis-freesurvival(MFS),andbiochemicalrecurrence_–freesurvival(BCRFS)after15.1yr (interquar-tilerange10.9–19.7yr)follow-upwerecomparedusingHarrell’sC-statistic.

Resultsandlimitations: ThebiopsyGGswere486GG1,310GG2,104GG3,64GG4,and 67GG5;cGradedistributionswere738cGrade1,102cGrade2,91cGrade3,58cGrade4,and 42cGrade5.ThecGradehadabetterdiscriminativevaluethantheGGforDSS(C-index0.79; 95%con_fidenceinterval0.74_–0.83vs0.76;0.71_–0.82)andMFS(0.79;0.74_–0.84vs0.77;0.72_– 0.82). Thediscriminativevalue for BCRFSafterRPandRT wassimilarfor both models. Differentdiagnostic,suchasuseofsextantbiopsies,andtherapeuticstrategiesinthe1990s arethelimitationsofthisstudy.

Conclusions: ThecGradeisasimplePcagrademodificationwithbetterdiscriminativevalues forDSSandMFSthantheGG,particularlyimpactingdecisionmakinginmenwithcurrent GG2Pca.

Patientsummary: Microscopicgradingisanimportantfactorfordecisionmakinginprostate cancer(Pca)patients.WeshowthatasimplegrademodificationbetterpredictsPcaoutcome andmightimprovetreatmentchoices.

©2019EuropeanAssociationofUrology.PublishedbyElsevierB.V.Allrightsreserved.

*Correspondingauthor.DepartmentofPathology,ErasmusUniversityMedicalCentreRotterdam, Postbus2040,3000CARotterdam,TheNetherlands.Tel.+31107043901;Fax:+31107038430. E-mailaddress:g.vanleenders@erasmusmc.nl(GeertJ.L.H.vanLeenders).

https://doi.org/10.1016/j.eururo.2019.07.051

1. Introduction

TheGleasongradingsystemisoneofthemostimportant parametersforclinicaldecisionmakinginprostatecancer (Pca) patients. Gleason score (GS) is entirely based on pathological assessment of tumour growth patterns and takesintoaccounttumourheterogeneity.Whilethesystem wasfirstdevelopedinthelate1960s,ithasbeenmodified severaltimesovertheyearsandiscurrentlybasedonthe 2014consensusconferenceoftheInternationalSocietyof UrologicPathology(ISUP)[1,2].

Althoughthe Gleason grading system has been estab-lishedforseveraldecades,ithasafewweaknesses.Firstly,the GSrangesbetween2and10,whichmightgivepatientswith GS6afalseimpressiontohaveatumourofintermediate grade.Secondly,theclinicallyimportantdistinctionbetween GS3+4andGS4+3isnotclearlyreflectedbyGS7[3].To overcometheseweaknesses, Pierorazioandcolleagues[4]

introduced “grade groups” (GGs), which are defined as follows:GG1(GS6),GG2(GS3+4=7),GG3(GS4+3=7), GG4(GS8),andGG5(GS9–10).TheGGswereendorsedbythe ISUPandtheWorldHealthOrganization(WHO),andshould bereportedinconjunctionwiththeGS[1,2].

In recent years, much evidence was obtained that invasivecribriformand/orintraductalcarcinoma(CR/IDC) arepathologicalfeatureswithstrongindependent predic-tivevaluesfordisease-specificsurvival(DSS),biochemical recurrence–freesurvival(BCRFS),andadversepathological parametersat radicalprostatectomy(RP) [5–10]. Patients with biopsy GG2 Pca with presence of CR/IDC show significantly shorter DSS and BCRFS than those without

[8,9].On thecontrary, DSSand BCRFSofbiopsyGG2 Pca

patientswithoutCR/IDCarenotstatisticallydifferentfrom thoseofmenwithGG1Pca[8,9].Furthermore,patientswith biopsy GG1 Pcawith concomitant intraductal carcinoma often show aggressive features at subsequent RP with extraprostaticexpansioninupto64%ofcases[11].Although invasive cribriform Gleason pattern 4 and intraductal carcinoma are strictlytwo separatepathological entities, they show significant morphological overlap and often coincide. Despite the accumulating evidence of their additional prognostic impact, presence of CR/IDC is not included in thecurrent grading system. This limits their comprehensiveand simpleincorporation inclinical deci-sionmodels.

Inthecurrentproof-of-principlestudy,weinvestigatea modifiedPcagrading model forprostate biopsies —cribri-form grade (cGrade)—which incorporates CR/IDC in the establishedGGsystem.

2. Patientsandmethods

2.1. Patientselection

Weidentifiedall1078menfromthefirstscreeningroundof the Dutch part of the European Randomized Study of Screening for Prostate Cancer (ERSPC), who had been diagnosed with Pcabetween November 1993and March 2000attheErasmusUniversityMedicalCentre,Rotterdam,

TheNetherlands[12].TheERSPCisanongoingmulticentre randomisedscreening trialthatwasinitiatedintheearly 1990s to evaluate the effect of screening with prostate-specificantigen(PSA)testingondisease-specificmortality rates.Exclusioncriteriaofthepresentstudywere unavail-abilityof slidesorparaffinblocksfor review(n=24)and presenceoflymphnodeordistantmetastasisatthetimeof diagnosis (n=23), leaving1031 patients for analysis. The studywasapprovedbytheinstitutionalMedicalResearch Ethics Committee (MEC-2018-1614) and in concordance withtheDeclarationofHelsinki.

2.2. Pathologicalevaluation

Threeinvestigators(C.K.,I.K.,andG.v.L.),whowereblinded to patientinformationand outcome, revisedall available histopathological slides (n=1031) according to the 2014 WHO/ISUP recommendations [2]. For each biopsy core, we recorded GS, GG, presence of CR/IDC, and percentage Gleason pattern 4 [13]. Invasive cribriform structures were defined as both small and expansive malignantepithelialproliferationswithintercellular lumi-na,inwhichthemajorityoftumourcellsdidnotcontact surroundingstromaandwhichspannedatleasthalfofthe glandular lumen. Intraductal carcinoma was diagnosed according tothe WHOcriteria[1]. Basalcell immunohis-tochemistrywasperformedtodifferentiateinvasive cribri-form from intraductal carcinoma in case this was not evident by histopathological parameters alone [9]. Since invasivecribriformandintraductalcarcinomashow signif-icant morphologicaloverlapandbothhaveadded clinical values,wecombinedbothpatternsintoonegroup(CR/IDC). TheworstbiopsyGGwasusedforstatisticalanalyses.

2.3. Cribriformgrade

The cGrade was conceived as a modification of the GG systemandwasdeterminedasfollows(SupplementaryFig. 1).ForPcapatientswithGG2–5,thecGradeissimilartothe GGifCR/IDCispresent.ThecGradeisequaltotheGGminus 1,ifCR/IDCisabsent.FormenwithGG1,thecGradeis1.In the rare case of GG1 Pca with concomitant intraductal carcinoma,theassignedcGradeis2.

2.4. Clinicalfollow-up

Afterdiagnosisandinitialtreatment,patientswere moni-tored semiannuallybya chartreviewto assess potential progressionandsecondarytreatments.DSSwasdefinedas thetimefromdiagnosistodeathattributedtoPcaandwas evaluated by anindependent causes of deathcommittee

[12].Metastasis-freesurvival(MFS)wasdefinedasthetime fromdiagnosistothedevelopmentofdistantmetastasisor censoring. Biochemical recurrence was defined as a PSA levelof0.2ng/mlassessedattwoconsecutivetimepoints >3moapartafterRPoranyPSAincreaseof>2ng/mlhigher thanthePSAnadirvalueafterradiotherapy(RT)[14].BCRFS was defined as the time from RP or the start of RT to biochemicalrecurrenceorcensoring.

EUROPEAN UROLOGYXXX(2019)XXX–XXX

2

2.5. Statisticalanalysis

For DSS and BCRFS analyses, both GG and cGrade were included in five categories (GG1–5 and cGrade1–5), and percentageGleasonpattern4wasincludedasacontinuous variable. Hazard ratios (HRs) for survival time were calculated using univariable Cox proportional hazard regression.Toadjustfordifferenttreatments,weincluded a stratified baseline hazard for treatment and estimated commonHRforGGandcGrade.Survivalprobabilitieswere plottedusing Kaplan-Meier curves.Discriminative ability forgradingmodelswasquantifiedusingHarrell’sC-index. AllstatisticalanalyseswereperformedinSPSSversion24 (IBM, Chicago, IL, USA) and R version 3.2.2 (R, Vienna,

Austria). Two-sided p values of <0.05 were considered statisticallysignificant.

3. Results

3.1. Patientcharacteristics

Our study cohort contained 1031 screen-detected Pca patients(Table1).Theirmedianageatthetimeofdiagnosis was 67yr (interquartile range [IQR] 62–71yr) and the medianPSAlevelwas5.6ng/ml(IQR3.9–8.8ng/ml).Intotal, 549 (53.3%)menreceivedsix biopsies,460(44.6%)seven biopsies,and22(2.1%)eightbiopsies.Fourhundredandsix (39.2%)patientsunderwentRP,508(49.0%)underwentRT,

Fig.1–Kaplan-MeiersurvivalprobabilitiesofPcapatientsstratifiedforthepresenceorabsenceofCR/IDCin(A)gradegroup(GG)2,(B)GG3,(C)GG4, and(D)GG5.

108 (10.5%) were under watchful waiting/active surveil-lance,andeight(0.8%)receivedhormonaltherapy.Primary treatmentwasunknowninone(0.1%)patient.Themedian follow-upofpatientsbeingaliveatthelastfollow-upwas 15.1yr(IQR10.9–19.7yr).

The biopsygradedistribution wasasfollows:486 Pca casesweregradedasGG1(GS6),310casesasGG2(GS 3+4=7), 104casesasGG3(GS4+3=7),64casesasGG4(GS 8),and67casesasGG5(GS9or10).Atotalof566(54.9%) mendiedduringfollow-up,90(8.7%)ofwhomdiedfrom Pca. Disease-specific death occurred in 2.1% (10/486) of patients diagnosed with biopsy GG1, 6.5% (20/310) with GG2,22.1%(23/104)withGG3,25.0%(16/64)withGG4,and 31.3%(21/67)withGG5Pca.

3.2. Cribriformgrowthand/orintraductalcarcinomaandDSS

Four (0.8%) GG1 (GS6) Pca patients had concomitant intraductalcarcinomaoftheprostate.CR/IDCwasobserved in54(17.4%),60(57.7%),33(51.6%),and42(62.7%)patients withGG2,GG3,GG4,andGG5Pca,respectively(Fig.1).In multivariable analysis including the five GG categories, biopsyCR/IDChadadditionalpredictivevaluesforDSS(HR 3.8; 95%confidence interval[CI]2.3–6.3;p< 0.001) and MFS (HR 3.7; 95% CI 2.2–6.2; p< 0.001). Adding an interaction between GG and CR/IDC did not lead to a significantlybettermodelfitforDSS(p= 0.8)orMFS(p= 0.6).Ofinterest,DSSofmenwithbiopsyGG2Pcawithout CR/IDCwasnotdifferentfromtheDSSofthosewithGG1 Pca(HR 1.9; 95% CI 0.8–4.7; p= 0.15). Whilepercentage Gleasonpattern4hasalsobeenrecognisedasaparameter forclinicalstratificationofGG2Pcapatients,itdidnothave apredictivevalueforDSS(HR1.5;95%CI0.03–73.7;p=0.8) orMFS(HR1.4;95%CI0.03–64.1;p= 0.9)asacontinuous variableinourcohort[1,5,15].

3.3. Cribriformgrademodification

Distribution of cGrade among the 1031 patients was as follows:738patientsweregradedashavingcGrade1,102as cGrade2,91ascGrade3,58ascGrade4,and42ascGrade5

Pca(Table2).IncorporationofCR/IDCinthecGradehadthe

mostprominenteffectinGG2tumours,ofwhich256/310 (82.6%)wereclassifiedascGrade1.Reclassificationoccurred

in42.3%(44/104),48.4%(31/64),and37.3%(25/67)ofGG3, GG4, and GG5 Pca cases, respectively. Only four out of 486patientswithGG1Pca(0.8%)wereupgradedtocGG2 duetothepresenceofintraductalcarcinoma.

3.4. DSSandMFS

Bothgradingsystemswerestronglyassociatedwithclinical outcome(Table3).ThediscriminativevalueofthecGrade (C-index 0.79; 95% CI 0.74–0.83) in predicting DSS was significantly(p= 0.029)higherthanthatoftheGG(C-index 0.76;95%CI0.71–0.82;Fig.2).ThecGrademodel(C-index 0.79;95%CI0.74–0.84)alsoshowedbetterdiscrimination thantheGGmodel(C-index0.77;95%CI0.72–0.82)forMFS, although this did not meet conventional measures of statisticalsignificance(p= 0.062).

3.5. Biochemicalrecurrence–freesurvival

Ofthe406patientswhohadundergoneRP,216(53%)had GG1,129(32%)GG2,33(8%)GG3,14(3%)GG4,and14(3%) GG5 Pca.Conversionto theproposedcGradedistribution resulted in 326 patients (80%) having cGrade1, 34 (8%) cGrade2,27(7%)cGrade3,12(3%)cGrade4,andseven(2%) cGrade5 Pca. In total, 85 (20.9%) patients experienced biochemicalrecurrence afteramedianof4.5yr(IQR2.3– 9.3yr).Bothgradingmodelswerestronglyassociatedwith postoperative BCRFS (Table 4). There was no significant difference(p=0.4)betweenthediscriminativevaluesofthe cGrademodel(C-index0.65;95%CI0.58–0.71)andtheGG model(C-index0.65;95%CI0.59–0.71).

Intotal,223outof508patientswhoreceivedRT(43.9%) experienced biochemical recurrence after a median

Table1–Patientcharacteristicsofthestudypopulation. GS6GG1 (n=486) GS3+4=7GG2 (n=310) GS4+3=7GG3 (n=103) GS8GG4 (n=64) GS9–10GG5 (n=67)

Age,median(IQR) 66(61–70) 67(62–71) 69(65–72) 69(66–72) 67(64–71)

PSA,median(IQR) 4.7(3.5–6.9) 5.8(4.0–9.0) 7.4(4.7–14.9) 11.0(6.1–17.4) 9.4(5.3–16.3)

CR/IDC,n(%) 4(0.8) 54(17.4) 60(57.7) 33(51.6) 42(62.7) Radicalprostatectomy,n(%) 216(44.4) 129(41.6) 33(32.0) 14(21.9) 14(20.9) Radiationtherapy,n(%) 188(38.7) 154(49.7) 66(64.1) 48(75.0) 52(77.6) Activesurveillance/WW,n(%) 80(16.5) 23(7.4) 3(2.9) 1(1.6) 1(1.5) Hormonaltherapy,n(%) 2(0.4) 3(1.0) 2(1.9) 1(1.6) 0 Unknown,n(%) 0 1(0.3) 0 0 0

CR/IDC=invasivecribriformand/orintraductalcarcinoma;GG=gradegroup;GS=Gleasonscore;IQR=interquartilerange;PSA=prostate-specificantigen; WW=watchfulwaiting.

Table2–Distributionandrelationofgradegroups(GGs)and cribriformgrade(cGrade).

GG1 GG2 GG3 GG4 GG5 Total cGrade1 482 256 – – – 738 cGrade2 4 54 44 – – 102 cGrade3 – – 60 31 – 91 cGrade4 – – – 33 25 58 cGrade5 – – – – 42 42 Total 486 310 104 64 67 1031 EUROPEAN UROLOGYXXX(2019)XXX–XXX 4

follow-upof4.8yr(IQR2.7–8.2yr).Accordingtothecurrent gradingsystem,188(37%)menhadGG1,154(30%)GG2,66 (13%)GG3,48(9%)GG4,and52(10%)GG5Pca, correspond-ing to 306 (60%) cGrade1, 64 (13%) cGrade2, 59 (12%)

cGrade3, 44(9%)cGrade4,and35(7%)cGrade5 (Table4). ThebiopsycGrade(C-index0.71;95%CI0.68–0.75)hada comparablediscriminativevalue(p=0.2)forBCRFSafterRT withtheGG(C-index0.72;95%CI0.69–0.75).

Table3–Coxproportionalhazardmodelsforprostatecancerdisease-specificsurvivalandmetastasis-freesurvivalstratifiedbygradegroups (GGs)andcribriformgrade(cGrade).

Disease-speci_ficsurvival Metastasis-freesurvival

Group Gradegroup cGrade Gradegroup cGrade

HR(95%CI) pvalue HR(95%CI) pvalue HR(95%CI) pvalue HR(95%CI) pvalue

1 ref – ref – ref – ref –

2 3.1(1.4–6.6) 0.003 4.6(2.3–9.1) <0.001 3.8(1.7–8.4) <0.001 3.7(1.8–7.7) <0.001

3 10.3(4.8–22.2) <0.001 8.4(4.5–15.8) <0.001 10.4(4.6–23.3) <0.001 9.3(5.0–17.2) <0.001

4 11.9(5.3–26.7) <0.001 11.2(5.9–21.4) <0.001 16.7(7.3–38.3) <0.001 12.1(6.4–23.0) <0.001

5 16.4(7.5–35.8) <0.001 17.9(9.1–35.2) <0.001 18.0(7.9–40.9) <0.001 15.6(7.8–31.2) <0.001

CI=confidenceinterval;HR=hazardratio;ref=reference.

Table4–Coxproportionalhazardmodelsforbiochemicalrecurrence-freesurvivalafterradicalprostatectomyandradiationtherapy stratifiedbyGradegroups(GG)andcribriformGrade(cGrade).

BCRFSafterradicalprostatectomy BCRFSafterradiationtherapy

Group Gradegroup cGrade Gradegroup cGrade

HR(95%CI) pvalue HR(95%CI) pvalue HR(95%CI) pvalue HR(95%CI) pvalue

1 ref – ref – ref – ref –

2 2.1(1.2–3.5) 0.006 2.2(1.1–4.4) 0.02 1.9(1.3–2.8) 0.001 2.5(1.7–3.7) <0.001

3 4.1(2.1–7.8) <0.001 4.4(2.4–7.9) <0.001 4.4(2.9–6.7) <0.001 4.4(3.0–6.5) <0.001

4 5.0(2.2–11.6) <0.001 4.5(1.9–10.4) 0.001 5.9(3.8–9.3) <0.001 5.7(3.8–8.5) <0.001

5 5.2(2.2–11.9) <0.001 4.4(1.6–12.2) 0.005 7.5(4.9–11.5) <0.001 7.4(4.8–11.5) <0.001

BCRFS=biochemicalrecurrence–freesurvival;CI=confidenceinterval;HR=hazardratio;ref=reference.

Fig.2–Kaplan-MeiersurvivalprobabilitiesofPcapatientsgradedaccordingtothe(A)currentgradegroupmodeland(B)proposedcGrademodel. cGrade=cribriformgrade;GG=gradegroup;Pca=prostatecancer.

3.6. Eligibilityforactivesurveillance

The mostprominenteffectof thecGrade model overthe currentGGmodelisthereclassificationofmenpotentially being eligible for active surveillance. In total 272/1031 (26.4%)menfulfilledtheProstateCancerResearch Interna-tional Active Surveillance (PRIAS) criteria for active surveillance (GG1, positive biopsies 2, cT2, PSA<10 ng/ml,andPSAdensity<0.2ng/ml2_{).IncasecGrade1was} usedinsteadofGG1,thetotalnumberofpatientsfulfilling thePRIAScriteriaincreasedby84(8.1%)to356/1031(34.5%) of the entire population. The effect was larger if less stringenteligibilitycriteriawereapplied.Ifthenumberof positivebiopsieswasnottakenintoaccount,asincreasingly happens in current practice with larger numbers of systematicaswellasmultiparametricmagneticresonance imaging (mpMRI)-targeted biopsies, the total number of patientsbeingeligibleforactivesurveillanceincreasedby 151 (14.6%), from361(35.0%) forGG1 to 512(49.7%) for cGrade1.

4. Discussion

Pathological grading by GS and GG is one of the most important parameters for predicting Pca outcome and guiding clinical decision making. In the past decade, intraductal carcinoma and invasive cribriform carcinoma havebothbeenrecognisedasnovelpathologicalfeatures with independent predictive values for adverse Pca outcome.IncorporationoftheseparametersinPcagrading is,however,neededtofacilitatewidespread implementa-tion into clinical practice. In the current study, we demonstratethatasimplemodificationoftheGGsystem withrespecttothepresenceorabsenceofCR/IDC outper-forms current Pca grading. In a set of over a thousand diagnostic prostate biopsies, we found that the cGrade showedastrongerdiscriminativevalueforthepredictionof DSSandMFSthancontemporaryPcagrading.

Inthepastdecade,theindependentpredictivevalueof additionalhistopathologicalfeatureshasbeenrecognised. Intraductalcarcinomaisaproliferationofmalignantcellsin pre-existentprostateglands,whichhasbeenrelatedtoan adverse disease outcome but which is not included in GleasongradingortheGG[10,11,13,16].Invasivecribriform growth is a Gleason pattern 4 subgroup associated with worse DSS than the other Gleason 4 growth patterns

[5,6,9,17,18].Invasivecribriformandintraductalcarcinoma

show overlapping morphological features, and might be difficult to differentiate without the use of basal cell immunohistochemistry.Therefore,bothpatternsareoften groupedasoneentity[13,19].PercentageGleasonpattern 4hasalsobeenshowntohaveanadditiveprognosticvalue for Pca outcome [5,15]. However, we did not find any additivepredictivevalueofGleasonpattern4.This discor-dance might beexplained by our analysis of percentage Gleasonpattern4asacontinuousinsteadofadichotomised variable.

OneofthemostimportantweaknessesofGleasongrading isitssubstantialinterobservervariability[20].While

interob-servervariabilitymighthaveminortherapeuticimplications inpatientswithGG3–5Pca,ithassignificantimpactonmen withGG1andGG2diseasesincetheirdistinctioncommonly determines eligibility for active surveillance. Grading dis-cordancesparticularlyoccurindifferentiatingpoorlyformed and fused Gleason pattern 4 from tangentially sectioned pattern 3 [20,21]. Interobserver agreement is, however, excellent for cribriform pattern 4, which is generally not confusedwithGleasonpattern3[20,21].Thefactthatmen withbiopsyGG2PcawithoutCR/IDChadcomparableclinical outcome to those with GG1 disease,and the existence of substantial interobserver variability in distinguishing be-tweenGG1andGG2Pcaconstituteanimportantrationalefor coalescing both groups into cGrade1 [8,9]. This is also supported by the low rate of lymph node metastases in menwith GG2Pca. Diolombiand Epstein [22] found0.6% pelviclymphnodemetastasisamong3235menwithGG2Pca at RP withoutstratificationfor CR/IDC.In anotherGG2 RP cohort,lymphnodemetastasiswerepresentin12/228(5.3%) patientswithCR/IDCbutinnoneof192menwithoutCR/IDC

[23].Studiescomparingtheinterobservervariabilityofboth models,however,stillhavetobeperformed.

Inthecurrentstudy,256outof310menwithbiopsyGG2 diseasewerereclassifiedascGrade1.Theproportionofmen fulfillingthePRIAScriteriaforactivesurveillanceincreased from26.4%to34.5%ifcGrade1wasusedinsteadofGG1.This numberfurtherincreasediflessstringenteligibilitycriteria were used. Since upgrading is the most important parameter for surveillance discontinuation, applying the cGradeinsteadofGGwouldleadnotonlytoalargernumber ofpatientsmeetingtheinclusioncriteria,butalsotofewer dropoutsduetogradeprogressionfromGG1toGG2[24]. Gleason gradinghasbeenaworkinprogressformore thanhalfacentury.TheWHO/ISUPsignificantlymodified the Gleason grading systemin 2005andachieved minor modifications in 2014 [2,25]. While the cGrade model outperformedthecurrentGGmodel,wefeelthatthereis stillroomforimprovement.Forinstance,wenotedinour cohortthatmenwithGG4diseasewithoutCR/IDChadDSS ratescomparabletothosewithGG2andGG3disease.Thisis inlinewiththefindingsofHarding-Jacksonetal[27],who also found relatively good outcome in GG4 Pca patients withoutcribriformpattern.WiththeintroductionoftheGG system, all patients with GS8 are classified as GG4 Pca patientsirrespectiveoftheGSbeing3+5,4+4,or5+3.Van den Bergh et al found that men with GS 3+5=8 had significantlybetterBCRFSthanGS8patientswithanyofthe othergradecombinations,althoughsuchadifferencecould not beconfirmedbyothers[26-28].Detailedanalyses for thepresenceofCR/IDCinthisheterogeneouspopulationof GG4PcapatientsmightfurtherimprovePcagrading.Atthis moment, we do not advocate to modify Pca grading. Additional improvements, reproducibility studies, and independentvalidationsarewarrantedbeforegrade mod-ifications canbeimplementedwith broadsupport ofthe scientificcommunity.

The strongest point ofthis studyisthe useof a well-characterised population of Pca patients with long-term follow-upandstrongclinicalendpoints.Thediagnosticand

EUROPEAN UROLOGYXXX(2019)XXX–XXX

6

treatmentmodalitiesinthe1990s,however,were signifi-cantlydifferentfromcontemporarystrategiesthatincludea largernumberofbiopsies,applicationofmpMRI,andactive surveillance.Therefore,validationofthecurrentfindingsin morecontemporarycohortsisimportantbeforeitsroutine adoption.Weexpect thatthe performanceof thecGrade modelwillbeevenbetterincontemporarycohorts,sincea largernumberofsystematicandtargetedbiopsiesdecrease biopsysamplingartefacts.

5. Conclusions

Inthisproof-of-principle study,we demonstrate thatthe biopsycGradeisasimplemodification ofthe GGsystem havingbetterdiscriminativevaluesforDSS andMFSthan current Pcagrading. Reclassificationof GG2 Pca patients with favourable outcome as cGrade1 might allow more patientstobeconsideredforactivesurveillance.

Authorcontributions:GeertJ.L.H.vanLeendershadfullaccesstoallthe datainthestudyandtakesresponsibilityfortheintegrityofthedataand theaccuracyofthedataanalysis.

Studyconceptanddesign:vanLeenders,Roobol.

Acquisitionofdata:Kweldam,Kümmerlin,vanLeenders,vanderKwast. Analysis and interpretation of data: van Leenders, Nieboer, Incrocci, Bangma,vanderKwast,Roobol.

Draftingofthemanuscript:vanLeenders.

Criticalrevisionofthemanuscriptforimportantintellectualcontent:van Leenders, Kweldam, Hollemans, Kümmerlin, Nieboer, Verhoef, Remmers,Incrocci,Bangma,vanderKwast,Roobol.

Statisticalanalysis:vanLeenders,Nieboer. Obtainingfunding:None.

Administrative, technical, or material support: Hollemans, Verhoef, Remmers.

Supervision:vanLeenders,Roobol. Other:None.

Financialdisclosures:GeertJ.L.H.vanLeenderscertifiesthatallconflicts ofinterest,includingspeci_{ficfinancial}interestsandrelationshipsand af_filiationsrelevanttothesubjectmatterormaterialsdiscussedinthe manuscript(eg,employment/affiliation,grantsorfunding, consultan-cies,honoraria,stockownershiporoptions,experttestimony,royalties, orpatentsfiled,received,orpending),arethefollowing:None.

Funding/Supportandroleofthesponsor:None.

Con_flictsofinterest:Theauthorshavenothingtodisclose.

AppendixA. Supplementarydata

Supplementary material related to this article can be found,intheonlineversion,atdoi:https://doi.org/10.1016/j.

eururo.2019.07.051.

References

[1] MochH,HumphreyPA,UlbrightTM,ReuterVE.WHOclassi_fication

oftumoursoftheurinarysystemandmalegenitalorgans.Lyon,

France:InternationalAgencyforResearchonCancer(IARC);2016.

p.135–67.

[2]EpsteinJI,EgevadL,AminMB,etal.The2014InternationalSocietyof

Urological Pathology (ISUP)consensus conferenceon Gleason grading

ofprostaticcarcinoma:definitionofgradingpatternsandproposalfor

anewgradingsystem.AmJSurgPathol(40):2016;244–52.

[3]StarkJR,PernerS, StampferMJ,etal. Gleasonscore andlethal

prostatecancer:does3+4=4+3?JClinOncol2009;27:3459_–64.

[4]PierorazioPM,WalshPC,PartinAW,EpsteinJI.PrognosticGleason

gradegrouping:databasedonthemodifiedGleasonscoring

sys-tem.BJUInt2013;111:753–60.

[5]ChoyB,PearceSM,AndersonBB,etal.Prognosticsignificanceof

percentageandarchitecturaltypesofcontemporaryGleason

pat-tern4prostatecancerinradicalprostatectomy.AmJSurgPathol

2016;40:1400–6.

[6]DongF,YangP,WangC,etal.Architecturalheterogeneityand

cribri-formpattern predictadverseclinicaloutcome forGleasongrade

4prostaticadenocarcinoma.AmJSurgPathol2013;37:1855–61.

[7]IczkowskiKA,PanerGP,VanderKwastT.Thenewrealizationabout

cribriformprostatecancer.AdvAnatPathol2018;25:31_–7.

[8]KweldamCF,KummerlinIP,NieboerD,etal.Prostatecancer

out-comesofmenwithbiopsyGleasonscore6and7withoutcribriform

orintraductalcarcinoma.EurJCancer2016;66:26–33.

[9]KweldamCF,KummerlinIP,NieboerD,etal.Disease-specific

sur-vivalofpatientswithinvasivecribriformandintraductalprostate

canceratdiagnosticbiopsy.ModPathol2016;66:26_–33.

[10] VanderKwastT,AlDaoudN,ColletteL,etal.Biopsydiagnosisof

intraductalcarcinomaisprognosticinintermediateandhighrisk

prostate cancer patients treated by radiotherapy. Eur J Cancer

2012;48:1318–25.

[11] KhaniF,EpsteinJI.ProstatebiopsyspecimenswithGleason3+3=6

and intraductal carcinoma: radical prostatectomy _findings and

clinicaloutcomes.AmJSurgPathol2015;39:1383_–9.

[12] SchroderFH,HugossonJ,RoobolMJ,etal.Screeningand

prostate-cancermortalityinarandomizedEuropeanstudy.NEnglJMed

2009;360:1320–8.

[13] KweldamCF,WildhagenMF,SteyerbergEW,BangmaCH,vander

KwastTH,vanLeendersGJ.Cribriformgrowthishighlypredictive

forpostoperativemetastasisanddisease-speci_ficdeathinGleason

score7prostatecancer.ModPathol2015;28:457–64.

[14] Roach3rdM,HanksG,ThamesJrH,etal.Definingbiochemical

failurefollowingradiotherapywithorwithouthormonaltherapyin

menwithclinicallylocalizedprostatecancer:recommendationsof

theRTOG-ASTROPhoenixConsensusConference.IntJRadiatOncol

BiolPhys2006;65:965–74.

[15] SauterG,SteurerS,ClauditzTS,etal.Clinicalutilityofquantitative

Gleasongradinginprostatebiopsiesandprostatectomyspecimens.

EurUrol2016;69:592–8.

[16] GuoCC,EpsteinJI.Intraductalcarcinomaoftheprostateonneedle

biopsy:histologicfeaturesandclinicalsignificance.ModPathol

2006;19:1528_–35.

[17] IczkowskiKA,TorkkoKC,KotnisGR,etal.Digitalquanti_ficationof

fivehigh-gradeprostatecancerpatterns,includingthecribriform

pattern,andtheirassociationwithadverseoutcome.Am JClin

Pathol2011;136:98–107.

[18]Trudel D,DownesMR,SykesJ,KronKJ,TrachtenbergJ,vander

KwastTH.Prognosticimpactofintraductalcarcinomaandlarge

cribriformcarcinomaarchitectureafterprostatectomyina

contem-porarycohort.EurJCancer2014;50:1610–6.

[19] ChuaMLK,LoW,PintilieM,etal.Aprostatecancer"nimbosus":

genomicinstabilityandSChLAP1dysregulationunderpin

aggres-sion of intraductal and cribriform subpathologies. Eur Urol

2017;72:665_–74.

[20] EgevadL,AhmadAS,AlgabaF,etal.StandardizationofGleason

grading among 337 European pathologists. Histopathology

[21]KweldamCF,NieboerD,AlgabaF,etal.Gleasongrade4prostate

adenocarcinoma patterns: an inter-observer agreement study

amonggenitourinarypathologists.Histopathology2016;69:441–9.

[22] DiolombiML,Epstein JI.Metastaticpotentialtoregionallymph

nodeswithGleason score_</=7,including tertiarypattern5,at

radicalprostatectomy.BJUInt2017;119:872_–8.

[23] HollemansE,VerhoefEI,BangmaCH,etal.Largecribriformgrowth

patternidentifies ISUPgrade 2prostate cancerathighriskfor

recurrenceandmetastasis.ModPathol2019;32:139–46.

[24]SimpkinAJ,TillingK,MartinRM,etal.Systematicreviewand

meta-analysisoffactorsdeterminingchangetoradicaltreatmentinactive

surveillanceforlocalizedprostatecancer.EurUrol2015;67:993_–1005.

[25] EpsteinJI,AllsbrookJrWC,AminMB,EgevadLL,CommitteeISUP

Grading.The 2005InternationalSocietyofUrologicalPathology

(ISUP)consensusconferenceonGleasongradingofprostatic

carci-noma.AmJSurgPathol2005;29:1228–42.

[26] vandenBerghRC,vanderKwastTH,deJongJ,etal.Validationofthe

novelInternationalSocietyofUrologicalPathology2014five-tier

Gleason grade grouping: biochemical recurrence rates for 3+5

diseasemaybeoverestimated.BJUInt2016;118:502_–5.

[27] Harding-JacksonN,KryvenkoON,WhittingtonEE,etal.Outcomeof

Gleason3+5=8prostate cancerdiagnosedonneedlebiopsy:

prognostic comparison with Gleason 4 + 4 = 8. J Urol

2016;196:1076–81.

[28] MahalBA,MuralidharV,ChenYW,etal. Gleasonscore5+3=

8 prostate cancer: much more like Gleason score 9? BJU Int

2016;118:95–101.

EUROPEAN UROLOGYXXX(2019)XXX–XXX

8