某醫(yī)藥公司產(chǎn)品組合優(yōu)化內(nèi)部資料

FromPipelinesToPortfolios

MinDing PennsylvaniaStateUniversity JehoshuaEliashberg UniversityofPennsylvaniaChoiceSymposiumJune4-8,2004OutlineOptimizingNewDrugPipeline–onebusinessopportunityDingandEliashberg,2002,ManagementScienceFromPipelinestoPortfolios–manybusinessopportunitiesSometheoreticalinsightsAnempiricaldemonstration--GlaxoSmithKlineJune4-8,20042TermsUsedinPresentationIndicationsAlsoknownasdiseaseEachindicationisconsideredasonebusinessopportunitye.g.,AIDSWeusetheterms“pipelineorproject”torepresenttheNPDstructureforagivenindicationWeuse“approach”torepresentaspecificway(e.g.,acompound,abiologicalmolecule)ofdrugdevelopmenttailoredforagivenindication.

TherapeuticalClassesAlsoknownastherapeuticalareas,groups,orcategoriese.g.,Anti-infectiveproductsEachclasshasmanydifferentindicationsWeusetheterm“portfolio”torepresenttheNPDstructureformanydifferentindications,withinoracrossdifferenttherapeuticalclasses.Aportfoliohasmanydifferentpipelines(orprojects).June4-8,20043OptimizingNewDrugPipeline

AnExampleBristol-MyersSquibbwantstodevelopapreventiveHIVvaccine...NPDinPharmaceuticalIndustryAlternativeapproaches: subunitvaccine,recombinantvectorvaccine, peptidevaccine,virus-likeparticlevaccine,plasmid DNAvaccine,“nakedDNA”vaccine,whole- inactivatedvirusvaccine,live-attenuatedvirus,etc.Question: Howmanyalternativeapproachesshould thefirmpursueateachNPDstage?Pre-clinicaltrialClinicalPhaseIClinicalPhaseIIClinicalPhaseIIIlaunch1June4-8,20044OptimizingNewDrugPipeline

ModelOutputForeachphaseindevelopment,wederivedtheexpectedreward,variance,andlikelihoodofhavingasuccessfuldrugcandidate.Forexample,forPhaseIII,theyare,respectively,WesolvefortheoptimalnumberofapproachestobesupportedateachphaseindevelopmentJune4-8,20045OptimizingNewDrugPipeline

OneKeyInsight-APatternofUnderspending?

(DingandEliashberg,ManagementScience,2002)June4-8,20046FirmsTargetMultipleBusinessOpportunities

–asurveyofnewdrugportfoliosin1450firmson12/31/2002Mostfirmsworkonmultipleindications.June4-8,20047FromPipelinestoPortfolios

AconventionalapproachtoportfolioconstructionPortfolioconstructionisthesumofpipelines.Process:Developa(notnecessarilyoptimal)pipelineforeachbusinessopportunityindependentlyandcalculatetheexpectedvalueRankordereachbusinessopportunityBuildaportfoliobysupportingthosebusinessopportunitieswhoseexpectedvaluesareaboveathreshold(e.g.,marginalreturn,IRR)Example:SmithKlineinmid-1990s(HBR,1998)June4-8,20048FromPipelinestoPortfolios

StateofArtPractice–GlaxoSmithKline,2003CardiovascularandUrogenitalMetabolicandViralDiseasesMicrobial,MusculoskeletalandProliferativeDiseasesNeurologyandGastrointestinalPsychiatryRespiratoryandInflammation

SixCentersofExcellence(TherapeuticClass)forDrugDiscovery(CEDDs)IndicationsSupportedInEachCEDD9(Stroke)and3(ED)5(Obesity)and7(HIV)7(Sepsis),2(osteoporosis)and9(cancers)11(Alzheimer's)and5(Irritablebowelsynd.)9(Depression)4(Asthma)and3(Rheumatoidarthritis)June4-8,20049FromPipelinestoPortfolios

Wheredoesthestateoftheartpracticesucceedandfail?Succeeded:Correctlyrecognizedtheimportanceofcross-projecteffectsinstructuringportfolioDevelopedthenecessaryorganizationalstructure(e.g.,3tierstructureatGSK)Failed:Riskhasnotreceivedsufficientattention,ENPVisthemostcommonlyusedmetric.Norigorousportfolioevaluationprocessandoptimizationmodelexist.June4-8,200410FromPipelinestoPortfoliosResearchObjectivesUnderstandhowriskandexpectedrewardchangeasthenatureofapipelinechanges.Today: HowPhaseIIIvarianceandexpectedrewardchangeasthedegreeofcorrelationchangesIndicationlevelArealevelDevelopoptimalportfolio,givenacertainmanagerialobjectiveToday: AnapplicationforGlaxo-Smith-Kline(GKS)’sPhaseIIIsubportfolioJune4-8,200411FromPipelinestoPortfoliosIndicationLevel--ScenarioScenario: Oneindicationwithtwoapproaches,whosesuccessisconditionalonovercomingacommonobstacleIndication1Approach12Approach11p1: jointprobabilityofovercomingthecommonobstacleinPhaseIIIANDotherobstaclesrelatedtoPhaseIIIp1i: probabilityofovercomingthecommonobstaclerelatedtoPhaseIIIJune4-8,200412FromPipelinestoPortfoliosIndicationLevel––SomeInsights(Risk-Reward)VarianceExpectedRewardVarianceExpectedRewardCase1.Lowp1Case2.Highp1Asp1iincreases,bothvarianceandexpectedrewardatPhaseIIIincreaseAsp1iincreases,variancedecreasesandexpectedrewardincreasesatPhaseIIIJune4-8,200413FromPipelinestoPortfoliosAreaLevel----ScenarioScenario: Oneareawithtwoindications,eachindicationwithtwoapproaches.Thesuccessofanyapproachisconditionalonovercomingcommonobstaclesatbothareaandindicationlevel.Indication1AreaIndication2Approach22Approach21Approach12Approach11June4-8,200414FromPipelinestoPortfoliosAreaLevel–SomeInsights(Risk-Reward)Case. Lowp1,highp1aVarianceExpectedRewardAsp1iincreases,bothvarianceandexpectedrewardatPhaseIIIincrease.Wehavealsocharacterizedsituationswherehigherp1ileadstolowervarianceandhigherexpectedrewardatPhaseIIIJune4-8,200415Work-In-Progress:EvaluatingPortfoliosBasedonotherRiskMetrics(PerspectivesfromFinancialLiterature)Value-at-Risk(VaR)Measuresthelikelihoodthatagivenportfolio'slosseswillexceedacertainamount.ExamplesExpectedtailloss(expectedshortfall)Semi-varianceJune4-8,200416FromPipelinestoPortfoliosAccommodatingRicherEnvironmentManagerswant/needtoknowtheentiredistributionofreward;Managerswant/needtooptimize,inadditiontoevaluateportfolio;Managersneedtorelaxassumptionsusedearlier,forexample,AlloweachapproachineachindicationtohaveuniqueprobabilityofsuccessSolution: AgeneralmodelJune4-8,200417FromPipelinestoPortfoliosAnApplicationoftheGeneralModelWhatweshowhereAnoptimizationexercisebasedonGlaxoSmithKlinePhaseIIISubportfolio(December2003)ThepredicteddistributionofmarketpayoffsofitscurrentportfoliobasedonourportfoliomodelIfGSKwereinterestedinlicensingtwoadditionalPhaseIIIcompounds,whichcompoundsitshouldaddtoitsexistingportfolio,under3different2-tierobjectivesWhatwedonotshowhere(butcouldbedonesimilarly)OptimizationofearlierstagesubportfoliosOptimizationbasedonothertypesofstrategicobjectives,e.g.,Discontinuingsomecurrentprojects(portfolioaudit)Optimalbudget(determinebudgetbasedonstrategicobjectives)June4-8,200418GSKPhaseIIISubportfolioFromGSKWebsiteA,B,…Jonlydenotethesequenceofindicationsinagivenarea,thesameletterrepresentsdifferentindicationindifferentarea.June4-8,200419GSKPhaseIIISubportfolioANumericMatrixRepresentationA,B,…Jonlydenotethesequenceofindicationsinagivenarea,thesameletterrepresentsdifferentindicationindifferentarea.ThenumberinacellrepresentsthenumberofapproachessupportedtargetingthatparticularindicationJune4-8,200420DistributionofMarketPayoff

GivenGSK’’sCurrentPhaseIIISubportfolioJune4-8,200421OptimizationScenariosGSKwantstolicenseintwocompoundstoaddtoitsexistingPhaseIIIsubportfolio,whichcompoundsshouldbelicensedundereachofthethreedifferent2-tierstrategicob