Cost-effectiveness of Pembrolizumab as Second-line Therapy for the Treatment of Locally Advanced or Metastatic Urothelial Carcinoma in Sweden

Cost-effectiveness

of

Pembrolizumab

as

Second-line

Therapy

for

the

Treatment

of

Locally

Advanced

or

Metastatic

Urothelial

Carcinoma

in

Sweden

Tushar

Srivastava

,

Vimalanand

S.

Prabhu

,

Haojie

Li

,

Ruifeng

Xu

,

Natalie

Zarabi

,

Yichen

Zhong

*

,

James

M.

Pellissier

,

Rodolfo

F.

Perini

,

Ronald

de

Wit

,

Ronac

Mamtani

a_{CompleteHEORSolutions,NorthWales,PA,USA;}b_{Merck&Co,Inc.,Kenilworth,NJ,USA;}c_{MerckSharp&Dohme,Stockholm,Sweden;}d_{ErasmusMCCancer} Institute,Rotterdam,TheNetherlands;e_{AbramsonCancerCenter,UniversityofPennsylvania,Philadelphia,PA,USA}

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 : e u o n c o l o g y . e u r o p e a n u r o l o g y. c o m Articleinfo Articlehistory: AcceptedSeptember25,2018 AssociateEditor: AshishKamat Keywords: Bladdercancer Cost-effectiveness Immunotherapy Pembrolizumab Urothelialcarcinoma Abstract

Background: Urothelialcarcinoma(UC)isthemostcommonsubtypeofbladder cancer.Therandomizedphase3KEYNOTE-045trialshowedthatpembrolizumab, usedassecond-line therapysignificantlyprolongedoverall survival withfewer treatment-relatedadverseeventsthanchemotherapyforadvancedUC. Pembro-lizumabhasbeenapprovedbytheEuropeanMedicinesAgencyforthetreatmentof locallyadvancedormetastaticUCinadultswhohavereceived platinum-contain-ing chemotherapy. Many European countries use cost-effectivenessanalysis to informreimbursementdecisions.

Objective: Toassessthecost-effectivenessofpembrolizumabassecond-line ther-apyforthetreatmentofadvancedUCfromaSwedishhealthcareperspective.

Design,setting, andparticipants: Wedevelopedapartitioned-survivalmodel to assess thecostsandeffectivenessofpembrolizumabcomparedwith vinflunine (basecase), paclitaxel,ordocetaxel monotherapyinpatientswith advancedUC overa15-yrtimehorizon.WeobtainedKaplan-Meierestimatesforsurvival end-points,adverseevents,andutilitydatafromKEYNOTE-045.

Outcomemeasurementsandstatisticalanalysis: Weperformedparametric extra-polationstoestimateoverallandprogression-freesurvivalbeyondtheclinicaltrial period.Swedishcostsandutilityweightswereusedtoestimatetotalcosts, quality-adjustedlifeyears(QALYs),andincrementalcost-effectivenessratios(ICERs).We performeddeterministic and probabilisticsensitivityanalyses to assess the ro-bustnessofthemodelresults.

Resultsand limitations: In thebase-caseanalysis,pembrolizumabresultedina meansurvivalgainof1.66years(1.38QALYs)atanincrementalcostofs69852and anICERofs50529/QALYgainedversusvinfluninemonotherapy.ICERsforother chemotherapiesweres81356/QALYforpembrolizumabversuspaclitaxelor doc-etaxel monotherapy, and s71924/QALY for pembrolizumab versus paclitaxel, docetaxel, orvinfluninemonotherapy. Long-termfollow-upfromKEYNOTE-045 andreal-worlddataareneededtovalidatetheextrapolations.

*Correspondingauthor.EconomicandDataSciences,CenterforReal WorldandObservational Studies,Merck&Co.Inc.,351N.SummneytownPike,NorthWales,PA19454,USA.

Tel.:+12673051282.

E-mailaddress:yichen.zhong@merck.com (Y.Zhong).

https://doi.org/10.1016/j.euo.2018.09.012

2588-9311/©2018MerckSharp&DohmeCorpandTheAuthors.PublishedElsevierB.V.onbehalfofEuropeanAssociationofUrology. ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Bladder cancer (BC) is the ninth most common cancer worldwide[1]andresultsinsignificantmortality, morbidi-ty,andcosts[2,3].ThevastmajorityofBCcases(90%)are urothelial carcinoma and occur in developed countries, with thehighestincidence in NorthAmerica and Europe

[1,4].In2015,theincidenceinSwedenwasapproximately 2700patientswithamedianageof73yr[5,6].In2012,the totalestimatedcostofBCwass4.9billioninEurope,with s2.9billionindirectcostsands2.0billioninindirectcosts associated with productivity losses due to mortality and morbidity [7,8]. In Sweden, BC accounted for s136.61 millionintotalcostsin 2012[7].Intensefollow-up,high recurrence rates, and preoperative and postoperative complicationswerethekeycontributorstothiseconomic burden[7].

Inpatientswithlocallyadvancedormetastaticurothelial carcinoma,platinum-basedcombinationtherapiesare recom-mended as the first-line standard of care (SoC) [9]. Most patientsexperiencerelapsefollowingfirst-linetherapy(40– 70%) and historically were faced with a choice between receiving second-line toxic chemotherapies with limited efficacy(vinflunineorataxane)orpalliativesymptomrelief withnoncurativeintent[10].Pembrolizumab,amonoclonal antibodyagainstPD-1,isthefirstimmunotherapyto demon-stratesuperioroverallsurvival(OS)comparedwith chemo-therapyinpatientswithadvancedurothelialcarcinomaafter failureofplatinum-basedtherapy[11].Theclinicalefficacyof pembrolizumabinadvancedurothelialcarcinomaafterfailure ofplatinum-basedtherapywasestablishedinKEYNOTE-045 (NCT02256436), a phase 3 multicenter global randomized clinicaltrialwith chemotherapycomparatorsthatincluded paclitaxel,docetaxel, andvinfluninemonotherapy.Thetrial demonstratedimprovementsinmedianOSwhencomparedto chemotherapy(10.3mointhepembrolizumabarmcompared with7.4mointhechemotherapyarm)[11].Onthebasisof KEYNOTE-045data,pembrolizumabwasapprovedinEurope forthetreatmentoflocallyadvancedormetastaticurothelial carcinomainadultswhohavepreviouslyreceived platinum-containingchemotherapy[12].

InmanyEuropeancountries,cost-effectiveness evalua-tionsplayacriticalroleinthedecisiontoreimbursehealth careprovidersforpharmaceuticals.IntheSwedishhealth caresystem,theDentalandPharmaceuticalBenefitsAgency

and the New Therapies Council recently recommended pembrolizumab for patients with advanced urothelial carcinomawhohavepreviouslyreceived platinum-contain-ingchemotherapyonthebasisofethicalconsiderationsand healtheconomicsevaluationofthistreatment[13].

Theobjectiveofthecurrentanalysiswastoevaluatethe cost-effectivenessofpembrolizumabcomparedwith che-motherapy as second-line therapy for the treatment of locally-advanced or metastatic urothelial carcinoma at a willingness-to-pay threshold of s100000 per quality-adjusted life year (QALY) from a Swedish health care perspective[14].

2. Patientsandmethods

2.1. Patientpopulation

Thetargetpopulationwaspatientswithadvancedurothelialcarcinoma who had received platinum-based chemotherapy and experienced diseaseprogressionduringorfollowingthatchemotherapy[11].

2.2. Comparators

Forthebase-caseanalysis,costsandeffectsforpembrolizumabwere comparedwiththoseforvin_flunine,theonlytherapyrecommendedas second-linetreatmentforadvancedurothelialcarcinomaaccordingto Swedish guidelines.Twoadditionalanalyses wereconsidered: pem-brolizumab compared with taxanes (paclitaxel or docetaxel mono-therapy),andpembrolizumabcomparedwithpaclitaxel,docetaxel,or vinfluninemonotherapy(thecontrolarminKEYNOTE-045).Thedose andfrequencyconsideredwereasforKEYNOTE-045:pembrolizumab, 200mg;vin_flunine,320mg/m2_{;paclitaxel,175mg/m}2_{;anddocetaxel,} 75mg/m2_{;eachtreatmentwasadministeredintravenouslyonceevery} 3week.

Costs and effectiveness were evaluated for pembrolizumab and comparatorsinthetargetpatientpopulationusingdatafrom KEYNOTE-045.Patientcharacteristicsinthepembrolizumabandcomparatorarms werecomparable.

2.3. Modelstructure

A partitioned-survival modelwas developed inExcel 2010 (Microsoft, Redmond,WA,USA)toprojectthecostsandeffectsforeachregimeninthe targetpopulation[15].Themodelincludesthreehealthstates(Fig.1): progression-free, progressive disease, and death. The progression-free healthstateincludespatientswhohaveprogressedonfirst-linetreatment withplatinum-basedchemotherapyandhavebegunsecond-linetreatment.

Conclusions: Theresultsindicatethatpembrolizumabimprovessurvival,increases QALYs,andiscost-effectiveassecond-linetherapyatawillingness-to-pay thresh-oldofs100000/QALYforthetreatmentofadvancedUC.

Patientsummary: Todate,pembrolizumabistheonlytreatmentassociatedwitha significantoverallsurvivalbenefitcomparedwithchemotherapyinarandomized controlledtrialassecond-linetherapyforadvancedurothelialcarcinoma.Our trial-basedcost-effectivenessanalysissuggeststhatpembrolizumabisacost-effective optionoverchemotherapyinpatientswithadvanced urothelialcarcinoma after platinum-basedtherapyinSweden.

©2018MerckSharp&DohmeCorpandTheAuthors.PublishedElsevierB.V.onbehalfof EuropeanAssociationofUrology.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

Transition to the progressive disease health state occurs once they experience disease progression. Progression is defined in line with KEYNOTE-045and assessed byblinded independentcentral review in accordancewithResponseEvaluationCriteriainSolidTumorsv1.1.Patients may alsodieineithertheprogression-freeorprogressivediseasehealthstate (transitiontothedeathstate).

Themodelestimatestheprobabilityofbeingineachofthethree healthstates atthe endof each week. Survival estimatesfrom the progression-freesurvival(PFS)curveprovidetheprobabilityapatientis intheprogression-freehealthstate.ThesurvivalestimatesfromtheOS distributionrepresentthepatient’slikelihoodofbeingdeadoraliveat eachparticulartimepoint.Finally,thedifferencebetweentheOSandPFS curvesyieldstheprobabilitythatthepatientisintheprogressivedisease health state. Kaplan-Meier survival curves based on clinical trial endpointsinKEYNOTE-045wereusedfortimepointsduringthetrial

period,whileparametricsurvivalfunctionsweredevelopedtoproject PFSandOSbeyondthetrialperiod(Table1).Lifeexpectancydatafrom Swedishlifetableswerealsoconsideredtoaccountforage-specific all-causemortality.Themodelassumedthatthecostandhealthbenefits accruedatweekly-spaceddiscretetimepointsoverthetimehorizonof 15yr.

2.4. Clinicalandutilityinputs

Clinicalinputssuchasadverseevent(AE)incidencerates,weight,body surface-area distributions,andutility inputswere drawn from KEY-NOTE-045withacutoffdateofJanuary18,2017.Weusedutilityvalues fortheprogression-freeandprogressivediseasehealthstates,andthe disutilityassociatedwithgrade3AEsusingEQ-5Dsurveyresponses fromKEYNOTE-045andSwedishutilityweights[16,17].

2.5. Costinputs

Thecostsforresourcesutilizedbypatientswereexpressedin2018euros (s) [18]. Weobtained drug acquisition costs from Pharmacy Heart (ApotekHjärtat),anationalSwedishpharmacy.Wealsoconsidereddrug administrationanddiagnostictestsinourmodel[19].Wemodeledthe costforsubsequenttherapyontheSwedishclinicalinputwhereby10%of patients in the vinflunine arm were subsequently administered gemcitabine and carboplatin for an average of 20 week [19]. We includedinouranalysisthediseasemanagementcostsincurredwhilein theprogression-freeandprogressivediseasehealthstates,hospiceand othercostsassociatedwiththelastmonthbeforedeath,andAE-related costs(Table1).

2.6. Statisticalanalysis

Kaplan-MeiersurvivalcurvesforPFS,OS,andtime-on-treatment(ToT) wereobtainedfromKEYNOTE-045.Because13.3%ofthepatientsinthe vin_fluninearmreceivedimmunotherapyafterdiscontinuation,itislikely thatOSwasoverestimatedinthecontrolarm[20].Toadjustforthisbias, weusedtheprespecifiedrank-preservingstructural-failuretime(RPSFT)

Fig.1–Diagramoftransitionsinthepartitionedsurvivalmodelusedto estimatehealtheconomicsoutcomes.

Table1–Modelinputs.

Pembrolizumab Vinflunine Datasource

Patientcharacteristics

Meanbodyweight,kg(SD) 73.58(17.23) 73.58(17.23) KEYNOTE-045

MeanBSA,m2_{(SD) 1.85(0.25) 1.85(0.25) KEYNOTE-045}

Costinputs Drugcost(s) 50-mgvial 1688 257 [27] 250-mgvial 1252 Administration(s/administration) 269 269 [18] Diseasemanagement

PF(s/week) 92 92 WeeklyHCRUandunitcosts

Progressivedisease(s/week) 215 215 WeeklyHCRUandunitcosts

Subsequenttreatment(s/week) 0 714 [19]

Terminalcare(monthbeforedeath) 7356 7356 [28]

Utilities MeanEQ-5Dutility(95%confidenceinterval)a _Datasource

Pembrolizumab Vin_flunine Pooled

PFwithgrade3AE 0.798(0.772–0.823) 0.822(0.796–0.848) 0.806(0.787–0.825) KEYNOTE-045

PFwithoutgrade3AE 0.865(0.857–0.874) 0.830(0.812–0.848) 0.859(0.851–0.866) KEYNOTE-045

Disutilityduetograde3AE(bysubtraction) 0.067 0.008 0.053 KEYNOTE-045

Progressivedisease 0.821(0.808–0.834) 0.764(0.736–0.792) 0.813(0.801–0.824) KEYNOTE-045

AE=adverseevent;BSA=bodysurfacearea;HCRU=healthcareresourceutilization;PF=progression-free;SD=standarddeviation.

a

methodinthebase-caseanalysis[21].Two-stageadjustmentwasnot feasibleforthebase-casecomparatorowingtothesmallsamplesize. Two-stageadjustmentwasappliedasabettermethodinaddressingthe OSbiasthantheRPSFTmethodforcomparisonofpembrolizumabwith paclitaxel/docetaxel/vin_flunine and with paclitaxel/docetaxel in the sensitivityanalysis[21].

Survival projections beyond the trial period were developed to estimatelong-termOSandPFSinaccordancewithNationalInstitutefor Health and Care Excellence (NICE) guidelines [22]. We applied a piecewiseextrapolationtechniquetoaccountforstructuraldifferences observedintheinitialpartsoftheKaplan-Meiercurvesfrom KEYNOTE-045.Wedividedthetimehorizonintotwoparts,withthecutoffpoint determinedaccordingtothestatisticalmethodrecommendedbyNICE, for both treatment arms[22].The first partuses the Kaplan-Meier estimates for OS and PFS, while the subsequent part provides extrapolation based on the best fit among exponential, Weibull, Gompertz,log-logistic,log-normal,andgeneralizedgammaparametric distributions.Weusedvisualinspection oftheKaplan-Meiercurves, goodness-of-fit statistics, and clinical plausibility to determine the parametricdistributionwiththebestfit.

SimilartotheapproachforOSandPFS,ToTsforpembrolizumaband comparatorarmswereexplicitlymodeledasanoutcomeandestimated onthebasisofKEYNOTE-045data.WeextrapolatedandmodeledToT datausingone-pieceparametriccurves(Supplementarymaterial).

2.7. Outputs

Weestimatedtheproportionofpatientsineachhealthstateforeach cycleover15yrbymodelingOSandPFS.Costsandutilityweightswere subsequentlyassignedtothehealthstates.Wesummedthecostsand utilitiesoverthecyclestoestimatetheprojecteddirectmedicalcosts, totalcosts,lifeyears,andQALYsforpembrolizumabandtheselected comparator. The incremental costs and QALYs were then used to estimatetheincrementalcost-effectivenessratio(ICER;incrementalcost perQALYgained).Weusedanannualdiscountrateof3%todetermine the present value of costs and health outcomes in this economic evaluation.

Wedepictedone-waysensitivityanalysesusingatornadodiagramto examinetheimpactofchangesineachoftheinputparametersonthe results.Weconductedaprobabilisticsensitivityanalysisusinga second-order Monte Carlo simulationwith 1000 iterations to examine the robustnessofthemodeloutputsandtheuncertaintyofallparameters takentogether.

3. Results

3.1. Base-caseresults

WeconsideredKaplan-Meiercurvesbeforeweek15forPFS andweek 32for OS andusedthebest-fitting parametric distributions,whichwerealog-logisticdistributionforthe pembrolizumabarmandanexponentialdistributionforthe vinfluninearm,fortimeafterthesepoints(Fig.2).ForToT, Weibullandexponentialdistributionsgavethebestfitfor thepembrolizumabandvinfluninearms,respectively.

Resultsfor thebasecasearepresentedinTable2.The projected discounted direct-treatment costs/per patient were s98354 and s28501 in the pembrolizumab and vinflunine arms, respectively. We further estimated that patientstreatedwithpembrolizumabandvinfluninewould experienceamean-discountedlifeexpectancyof2.40and 0.73yr,respectively.Pembrolizumab-treatedpatientshada

discountedQALYgainof1.38QALYswhencomparedwith vinflunine(1.99vs0.61QALYs).

Cost-effectiveness analysis revealedan overall ICERof s50529/QALYgainedwithpembrolizumabovervinflunine. ICERsofs81356/QALYands71924/QALYwereattainedfor pembrolizumab compared with taxanes (paclitaxel or docetaxel monotherapy) and with the pooled treatment withpaclitaxel,docetaxel,orvinfluninemonotherapy.

3.2. Sensitivityanalyses

The one-waysensitivityanalysis results are presentedin

Figure3.Thedoseintensityofpembrolizumab, extrapola-tion ofOS for pembrolizumab,and the discount rate for healthoutcomes hadthegreatestimpact on ICERvalues, while the costs for AEs, terminal care, and subsequent treatmenthadtheleastimpact.

ProbabilisticsensitivityanalysisgaveanexpectedICERof s50845/QALY, suggesting that pembrolizumab has a 99%chanceofbeingcost-effectivecomparedtovinflunine at a willingness-to-pay threshold of s100000/QALY [14]

(Supplementarymaterial). 4. Discussion

Treatmentofadvancedurothelialcarcinomaposessignificant clinical and economic burdens on health care systems

[2].Despitetheavailabilityofplatinum-basedchemotherapy foradvancedurothelialcarcinoma, themajority ofpatients experiencediseaseprogression[23].Untilrecently,noglobally accepted SoC was available in the second-line setting. In KEYNOTE-045,pembrolizumabwassuperiorto chemothera-py interms of OS, safety, and qualityof life[11,24]. Thus, pembrolizumab received approval from global regulatory agenciesforthetreatmentofpatientswithlocallyadvancedor metastaticurothelialcarcinomawhohavepreviouslyreceived platinum-containingchemotherapy.Ourmodel-based analy-sis showed that pembrolizumab is projected to be cost-effectivecomparedwithvinflunineamongpatientsinSweden whenevaluatedatacost-effectivenessthresholdofs100000/ QALYfromaSwedishhealthcareperspective[14]. Pembroli-zumabwasalsocost-effectivewhencomparedwithpaclitaxel ordocetaxelmonotherapy,orthepooledarmforpaclitaxel, docetaxel, or vinflunine monotherapy. In both scenarios, pembrolizumabhadbetterLYsandQALYs.

Medical costs associated with pembrolizumab were primarilydrivenbydrugacquisitionanddisease manage-mentcosts.Thetotaldrugacquisitioncostsfor pembroli-zumab were sensitive to the pembrolizumab treatment duration.AE-relatedcostsonlyaccountforapproximately 2% of the non–drug costs for pembrolizumab, compared with18%forvinflunine.

AnanalysisbasedonKaplan-MeiercurveswithoutRPSFT adjustmentforpatientsreceivingsubsequenttherapyinthe vinflunine armrevealedaslightincrease intheICER,but pembrolizumabwasstillconsideredcost-effective(Table2). Sarfaty et al. [25] recently reported on the cost-effectiveness of pembrolizumab as second-line therapy for advanced urothelial carcinoma. However, their study

Fig.2–Estimationofoverallsurvival(OS)andprogression-freesurvival(PFS).KM=Kaplan-Meier;KM15+LogLogistic=KMupto15weekand log-logisticmodelthereafter;KM15+Exponential=KMupto15weekandexponentialmodelthereafter;KM32+LogLogistic=KMupto32weekand log-logisticmodelthereafter;KM32+Exponential=KMupto32weekandexponentialmodelthereafter.OSforvinfluninewasadjustedusingthe rank-preservingstructural-failuretimemethod.

Table2–Discountedresultsforthecost-effectivenessofpembrolizumabcomparedtochemotherapy.

Comparator Total Incremental ICER

Costs(s) LYs QALYs Costs(s) LYs QALYs (s/QALY)

Basecase:Pembrovsvinflunine(withRPSFTadjustment)

Vinflunine 28501 0.73 0.61

Pembrolizumab 98354 2.40 1.99 69852 1.66 1.38 50529

Pembrovsvinflunine(noadjustment)

Vinflunine 28844 0.76 0.63

Pembro 98354 2.40 1.99 69510 1.63 1.36 51215

Pembrovspaclitaxel/docetaxel/vinflunine(withtwo-stageadjustment)a

P/D/V 25054 1.18 0.97

Pembro 98208 2.40 1.99 73154 1.22 1.02 71924

Pembrovspaclitaxel/docetaxel(withtwo-stageadjustment)a

P/D 25182 1.33 1.09

Pembro 98348 2.40 1.99 73166 1.07 0.90 81356

ICER=incrementalcost-effectivenessratio;RPSFT=rank-preservingstructuralfailuretime;LY=lifeyears;QALY=quality-adjustedlifeyears;P/D/=paclitaxel/ docetaxel;P/D/V=paclitaxel/docetaxel/vinflunine.

a

FortheP/D/VandP/Dcontrolarms,two-stageadjustmentandalog-logisticdistributionforfittingofoverallsurvivalwasthemostappropriatetechnique accordingtostatisticalanalysisoftheKEYNOTE-045data.TheRPSFTadjustmentmethodwasthesecondbestmethodinthisscenario.

wassubjecttolimitationsofdataavailabilityand subopti-mal modeling methods for survival, utility, and costs

[20,25]. Our analysis has several strengths that explain thedifferenceinresults.WemodeledOSandPFSusinga partitionedsurvivalapproachthatdirectlyusestrialresults tomodelsurvivalandiscommonlyappliedwhenevaluating cost-effectivenessinadvancedoncologyindications[15].A piecewiseapproachwasusedforextrapolation,andsurvival wasadjustedforpatientsinthecontrolarmwhoreceived immunotherapy after chemotherapy. These analyses allowedmoreaccuratepredictionofOS.Wealsoadopted a timehorizon longenough to capture long-term health outcomes andcosts.This wasessential, asthetreatment effectofimmunotherapymaylastevenafter discontinua-tionandleadtolong-termsurvivalgains[26].Ourmodel usesEQ-5Dutilityvaluesestimatedbyhealthstatedirectly from KEYNOTE-045 trial. In addition to drug costs, administration costs, and AE costs as used by Sarfaty etal.[25],wemodeleddiseasemanagementcostsbyhealth stateaswellasterminalcarecosts.Ourcomparatorsinclude vinflunine, taxanes (paclitaxel and docetaxel), and the entire KEYNOTE-045control arm,thus providinga com-prehensive evaluation of the cost-effectiveness of pem-brolizumab in comparison to different comparators as second-linetherapyforadvancedurothelialcarcinomaafter platinum-basedchemotherapy.

Overall, pembrolizumab remained cost-effective com-pared with vinflunine, taxanes (paclitaxel or docetaxel monotherapy),andwiththepooledtreatment(paclitaxel, docetaxel, orvinflunine monotherapy)over severalinput parameters as second-line therapy for patients with

advanced urothelial carcinoma who had received plati-num-based chemotherapy, and experienced disease pro-gression during or following that chemotherapy. The evidence for thecost-effectivenessof pembrolizumabfor advancedurothelialcarcinomaasobservedforSwedenin ouranalysisisgenerallyrepresentativeofseveralEuropean countrieswithsimilarhealthcaresystems.

4.1. Limitations

Owing to the lack of long-term PFS and OS data, extrapolations performed in themodel make the results sensitive to distributional assumptions. The model used statistical methodology and objective criteria recom-mendedbyNICEto extrapolateKaplan-Meierdatafor OS andPFSoverthetimehorizonofthemodel[22].However, actual survival rates may differand long-term follow-up fromclinicaltrialsorreal-worlddataisneededtovalidate the results.A sensitivityanalysis showed thatthe model resultsarerobusttoinputvariations.Inaddition,theICER results are only interpretable for the health system evaluated in this study or for countries with similar resourceutilization,treatmentcosts,andAEmanagement costs.However,themodelingframeworkandmethodsused inthisstudyaregeneralizabletotheothercountries. 5. Conclusions

The model results suggest thatpembrolizumab improves survival,increasesQALYs,andiscost-effectiveassecond-line therapy comparedwithvinflunineforthetreatmentoflocally

Fig.3–Tornadodiagramshowingtheimpactofchangesineachoftheinputparametersontheincrementalcost-effectivenessratio(ICER)results. AE=adverseevent;RPSFT=rank-preservingstructuralfailuretime;OS=overallsurvival;PFS=progression-freesurvival;ToT=timeontreatment; Llogistic=log-logistic.Theinterceptandlog(scale)areestimatedparametersfortherespectiveparametricsurvivalmodels.

advancedmetastaticurothelialcarcinomainSwedishadults whohavereceivedpriorplatinum-containingchemotherapy atawillingness-to-paythresholdof100000s/QALYgained. Pembrolizumabisalsocost-effectivewhencomparedwith paclitaxel ordocetaxel monotherapy, and with paclitaxel, docetaxel,orvinfluninemonotherapy.

Authorcontributions:YichenZhonghadfullaccesstoallthedatainthe studyandtakes responsibilityfor the integrityof the dataandthe accuracyofthedataanalysis.

Study concept and design: Li, Mamtani, Pellissier, Perini, Prabhu, Srivastava.

Acquisitionofdata:Prabhu,Srivastava,Zarabi.

Analysisandinterpretationofdata:Li,Mamtani,Pellissier,Perini,Prabhu, deWit,Srivastava,Xu,Zarabi,Zhong.

Draftingofthemanuscript:Prabhu,Srivastava,Zhong.

Criticalrevisionofthemanuscriptforimportantintellectualcontent:Li, Mamtani,Pellissier,Perini,Prabhu,deWit,Srivastava,Xu,Zarabi,Zhong. Statisticalanalysis:Prabhu,Srivastava.

Obtainingfunding:Zhong.

Administrative,technical,ormaterialsupport:None. Supervision:None.

Other:None.

Financialdisclosures:YichenZhongcertifiesthatallconflictsofinterest, includingspecificfinancialinterestsandrelationshipsandaffiliations relevanttothesubjectmatterormaterialsdiscussedinthemanuscript (eg,employment/affiliation,grantsorfunding,consultancies,honoraria, stockownershiporoptions,experttestimony,royalties,orpatents_filed, received, or pending), are the following: Tushar Srivastava is an employeeofCompleteHEORSolutions,whoreceivedconsulting fees fromMerckandCo.Inc.fortheproductionofthismanuscript.Ronaldde WithasaconsultancyrolewithMerck,Lilly,andRoche,hasreceived speakerfeesfromSanofi,andhasreceivedresearchgrantsfromSanofi andBayer.RonacMamtanihasreceivedpersonalfeesfromRocheoutside thesubmittedworkandhasreceivedanNIHgrant(K23CA187185). HaojieLi,JamesM.Pellissier,VimalanandS.Prabhu,RuifengXu,Yichen Zhong,andRodolfoF.PeriniareemployeesofMerck.NatalieZarabiisan employeeofMerckSharp&Dohme.HaojieLi,VimalanandS.Prabhu,and YichenZhongownrestrictedstockoptionsand/orstocksinMerck. Funding/Supportandroleofthesponsor:Thisworkwassupportedby Merck&Co.Inc.,Kenilworth,NJ,USA.Thesponsorplayedaroleinthe designandconductofthestudy;datacollection,management,analysis, and interpretation; and preparation, review, and approval of the manuscript.

Acknowledgments:Theauthorswouldliketoacknowledgethefollowing fortheircontributionstothestudy:JohnCook,TaniaGulati,RheaGupta, andManikantaDasariformodeling supportandmanuscript review; AnkitGulatiforanExcelcheckonthemodelquality;JaneLiaoandLiMa fortheirdedicatedworkondataanalyses;andTaraFrenklforacritical reviewofthestudyprotocolandmanuscript.

AppendixA. Supplementarydata

Supplementarydata associatedwiththisarticlecanbe found,intheonlineversion,atdoi:10.1016/j.euo.2018.09.012. References

[1] WongMC,FungFD,LeungC,etal. Theglobalepidemiology of bladdercancer:ajoinpointregressionanalysisofitsincidenceand

mortalitytrendsandprojection.SciRep2018;8:1129.http://dx.doi. org/10.1038/s41598-018-19199-z.

[2]KamatAM,HahnNM,EfstathiouJA,etal.Bladdercancer.Lancet 2016;388:2796–810. http://dx.doi.org/10.1016/S0140-6736(16) 30512-8.

[3]SullivanR,PeppercornJ,SikoraK,etal.Deliveringaffordablecancer care in high-income countries. Lancet Oncol 2011;12:933–80.

http://dx.doi.org/10.1016/S1470-2045(11)70141-3.

[4]AntoniS,FerlayJ,SoerjomataramI,etal.Bladdercancerincidence and mortality: a global overview and recent trends. Eur Urol 2017;71:96_–108.http://dx.doi.org/10.1016/j.eururo.2016.06.010. [5]RegionalaCancercentrumiSamverkan.Cancer iurinblåsa,

njur-bäcken, urin-ledare och urinrör Nationellt vårdprogram 2015.

www.cancercentrum.se/globalassets/cancerdiagnoser/urinvagar/ urinblase–och-urinrorscancer/vardprogram/natvp_cancer_ urinvagar_25aug15.pdf.

[6]RegionalaCancercentrumiSamverkan.Nationelltkvalitetsregister förurinblåsecancerKvalitetsregisterrapport,diagnosår2015.www. cancercentrum.se/globalassets/cancerdiagnoser/urinvagar/ urinblase–och-urinrorscancer/rapport2015final.pdf.

[7]LealJ,Luengo-FernandezR,SullivanR,WitjesJA.Economicburden

of bladder cancer across the European Union. Eur Urol

2016;69:438_–47.http://dx.doi.org/10.1016/j.eururo.2015.10.024. [8]Luengo-FernandezR,LealJ,GrayA,SullivanR.Economicburdenof

canceracrosstheEuropeanUnion:apopulation-basedcost analy-sis. Lancet Oncol 2013;14:1165–74.http://dx.doi.org/10.1016/ S1470-2045(13)70442-X.

[9]NationalInstituteforHealthCareExcellence.Bladdercancer: diag-nosis and management 2015. www.nice.org.uk/guidance/ng2/ resources/bladder-cancer-diagnosis-and-management-pdf-51036766405.

[10] FrantziM,LatosinskaA,FlüheL,etal.Developingproteomic bio-markersforbladdercancer:towardsclinicalapplication.NatRev Urol2015;12:317.http://dx.doi.org/10.1038/nrurol.2015.100. [11] BellmuntJ,DeWitR,VaughnDJ,etal.Pembrolizumabas

second-line therapy for advanced urothelial carcinoma. N Engl J Med 2017;376:1015_–26.http://dx.doi.org/10.1056/NEJMoa1613683. [12] Eurpoean MedicinesAgency.Keytruda(pembrolizumab). www.

ema.europa.eu/ema/index.jsp?.curl=pages/medicines/human/ medicines/003820/human_med_001886.jsp&mid=WC0b01ac 058001d124.

[13] Tandvårds-ochläkemedelsförmånsverket.Keytruda (pembrolizu-mab) www.tlv.se/download/18.2863d50f15f9b0b125bacaef/ 1510665873724/bes170904_underlag_keytruda.

[14] SvenssonM,NilssonFO,ArnbergK.Reimbursementdecisionsfor pharmaceuticalsinSweden:theimpactofdiseaseseverityandcost effectiveness.Pharmacoeconomics2015;33:1229–36.http://dx.doi. org/10.1007/s40273-015-0307-6.

[15] WoodsB,SiderisE,Palmer S,LatimerN,SoaresM.Partitioned survivalanalysisfordecisionmodellinginhealthcare:acritical review.NICEDSU technicalsupportdocument 19.http://scharr. dept.shef.ac.uk/nicedsu/wp-content/uploads/sites/7/2017/06/ Partitioned-Survival-Analysis-final-report.pdf.

[16] BurströmK,SunS,GerdthamU-G,etal.Swedishexperience-based valuesetsforEQ-5Dhealthstates.QualLifeRes2014;23:431_–42.

http://dx.doi.org/10.1007/s11136-013-0496-4.

[17] DolanP.ModelingvaluationsforEuroQolhealthstates.MedCare 1997;35:1095–108.

[18]Södraregionsvårdsnämnden.Regionalapriserochersättningarför södrasjukvårdsregionen2017[Regionalpricelist].

[19] FASS.Vardpersonal. Gemzarand Carboplatin. www.fass.se/LIF/ product?nplId=20080410000054&userType=0.

[20] SonpavdeG,DranitsarisG,NecchiA.Improvingthecostefficiency of PD-1/PD-L1 inhibitors for advanced urothelial carcinoma: a

majorroleforprecisionmedicine?EurUrol2018;74:63–5.http:// dx.doi.org/10.1016/j.eururo.2018.03.015.

[21]LatimerN.Treatmentswitchinginoncologytrialsandthe acceptabil-ityofadjustmentmethods.ExpertRevPharmacoeconOutcomesRes 2015;15:561.http://dx.doi.org/10.1586/14737167.2015.1037835. [22] LatimerNR.Survivalanalysisforeconomicevaluationsalongside

clinicaltrials—extrapolationwithpatient-leveldata: inconsisten-cies, limitations, and a practical guide. Med Decis Making 2013;33:743–54.http://dx.doi.org/10.1177/0272989X12472398. [23] FisherM,Shenolikar R,MillerPJ,etal. Treatmentpatternsand

outcomesinmetastaticbladder cancerincommunityoncology settings.JClinOncol2017;35(6Suppl):396.http://dx.doi.org/10. 1200/JCO.2017.35.6_suppl.396.

[24] DeWitR,BajorinDF,BellmuntJ,etal.Health-relatedqualityoflife (HRQoL)ofpembrolizumab(pembro)vschemotherapy(chemo)for

previouslytreatedadvancedurothelialcancer(UC)in KEYNOTE-045. JClin Oncol 2017;35(15Suppl):4530. http://dx.doi.org/10. 1200/JCO.2017.76.9562.

[25] SarfatyM,HallPS,ChanKK,etal.Cost-effectivenessof pembroli-zumab in second-line advanced bladder cancer. Eur Urol 2018;74:57_–62.http://dx.doi.org/10.1016/j.eururo.2018.03.006. [26] RobertC,RibasA,HamidO,etal.Durablecompleteresponseafter

discontinuation of pembrolizumab in patients with metastatic melanoma. J ClinOncol2018;36:1668–74.http://dx.doi.org/10. 1200/JCO.2017.75.6270. [27] http://www.apotekhjartat.se/soksida/?query=Javlor+; www. apoteket.se/produkt/keytruda-pulver-till-koncentrat-till-infusionsvatska-losning-50-mg-1-styck-injektionsflaska-306900/. [28] www.tlv.se/Upload/Halsoekonomiska_bedomningar/ halsoekonomiskt-kunskapsunderlag-yervoy_forsta_linjen.