Table Of ContentSpringer Series in Pharmaceutical Statistics
Christy Chuang-Stein
Simon Kirby
Quantitative
Decisions
in Drug
Development
Second Edition
Springer Series in Pharmaceutical Statistics
SeriesEditors
ThomasPermutt,WalnutCreek,CA,USA
JoséPinheiro,Raritan,RJ,USA
FrankBretz,Basel,Switzerland
PeterMüller,Austin,TX,USA
Recent advances in statistics have led to new concepts and solutions in different
areas of pharmaceutical research and development. “Springer Series in Pharma-
ceutical Statistics” focuses on developments in pharmaceutical statistics and their
practicalapplicationsintheindustry.Themaintargetgroupsareresearchersinthe
pharmaceuticalindustry,regulatoryagenciesandmembersoftheacademiccommu-
nityworkinginthefieldofpharmaceuticalstatistics.Inordertoencourageexchanges
betweenexpertsinthepharmaindustryworkingonthesameproblemsfromdifferent
perspectives,anadditionalgoaloftheseriesistoprovidereferencematerialfornon-
statisticians. The volumes will include the results of recent research conducted by
the authors and adhere to high scholarly standards. Volumes focusing on software
implementation(e.g.inSASorR)areespeciallywelcome.Thebookseriescovers
different aspects of pharmaceutical research, such as drug discovery, development
andproduction.
Moreinformationaboutthisseriesathttp://www.springer.com/series/15122
·
Christy Chuang-Stein Simon Kirby
Quantitative Decisions
in Drug Development
Second Edition
ChristyChuang-Stein SimonKirby
Kalamazoo,MI,USA Ramsgate,UK
ISSN2366-8695 ISSN2366-8709 (electronic)
SpringerSeriesinPharmaceuticalStatistics
ISBN978-3-030-79730-0 ISBN978-3-030-79731-7 (eBook)
https://doi.org/10.1007/978-3-030-79731-7
1stedition:©SpringerInternationalPublishingAG2017
2ndedition:©TheEditor(s)(ifapplicable)andTheAuthor(s),underexclusivelicensetoSpringerNature
SwitzerlandAG2021
Thisworkissubjecttocopyright.AllrightsaresolelyandexclusivelylicensedbythePublisher,whether
thewholeorpartofthematerialisconcerned,specificallytherightsoftranslation,reprinting,reuse
ofillustrations,recitation,broadcasting,reproductiononmicrofilmsorinanyotherphysicalway,and
transmissionorinformationstorageandretrieval,electronicadaptation,computersoftware,orbysimilar
ordissimilarmethodologynowknownorhereafterdeveloped.
Theuseofgeneraldescriptivenames,registerednames,trademarks,servicemarks,etc.inthispublication
doesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfromtherelevant
protectivelawsandregulationsandthereforefreeforgeneraluse.
Thepublisher,theauthorsandtheeditorsaresafetoassumethattheadviceandinformationinthisbook
arebelievedtobetrueandaccurateatthedateofpublication.Neitherthepublishernortheauthorsor
theeditorsgiveawarranty,expressedorimplied,withrespecttothematerialcontainedhereinorforany
errorsoromissionsthatmayhavebeenmade.Thepublisherremainsneutralwithregardtojurisdictional
claimsinpublishedmapsandinstitutionalaffiliations.
ThisSpringerimprintispublishedbytheregisteredcompanySpringerNatureSwitzerlandAG
Theregisteredcompanyaddressis:Gewerbestrasse11,6330Cham,Switzerland
ToParents,Bruce,TerresaandIsaac
ChristyChuang-Stein
ToDad,Mum,Katie,Theo,Helen,Roband
Andy
SimonKirby
Preface
Byallcounts,2020wasanunusualyear.COVID-19lockdownsandpandemic-related
travelrestrictionshadplacedhugelimitsonourmobility.Manyclinicaltrialswere
adverselyimpacted,andsomewereplacedonholdduetouncertaintiessurrounding
patient recruitment and care delivery. On the other hand, the unprecedented speed
with which COVID-19 vaccines were developed and approved for emergency use
wastrulybreathtaking.Forthefirsttimeinmodernproductdevelopmenthistory,the
identification and the commercialization of a game-changing innovative medicine
occurredinlessthanoneyear.Whowouldhavethoughtthistobepossiblebefore
2020?
Confinedtoourrespectivelocationsanddeprivedoffrequentfamilyoutings,we
havedecidedtousethetimewiselyandrevisethefirsteditionofourbook.Thetiming
wasgoodsincealotofnewadvanceshavebeenmadesincethepublicationofour
firstedition.Therevisionalsoallowedustocorrectafewerrorsthatwentundetected
whenthefirsteditionwaspublishedandamistakespottedbyanobservantreader.
By and large, the second edition follows the structure of the first edition. We
havedecidedtodevoteanentirechapter(Chap.13)tothetopicofadaptivedesigns,
resultingin14chaptersintherevision.
Chapter1offersahigh-leveloverviewofclinicaltestingandregulatoryreviewof
apharmaceuticalproduct.Chapter2reviewstheFrequentistapproachtothetesting
of hypotheses and in particular the two-action decision problem. In the context of
drug development, the two actions correspond to progressing or not progressing a
drugforfurtherdevelopment.Wediscussmultiplicityandselectiveinferenceaswell
as how their indiscriminate uses have contributed to the replication crisis and the
P-valuecontroversy.Chapter3discussesthemetricscommonlyusedtocharacterize
theperformanceofadiagnostictest.Chapter4drawsananalogybetweensuccessive
trialsconductedduringtheclinicaltestingofaninvestigationalproductandaseries
ofdiagnostictests.Underthisanalogy,theconditiontodiagnosebyaclinicaltrial
is the existence of a clinically meaningful effect for the investigational product.
Wehavefoundthisanalogyparticularlyusefultoexplaintoourclinicalcolleagues
whyreplicationissuchanimportantconceptindrugdevelopmentandtoshowwhy
replicationisnotaseasyasmanypeoplemighthope.
vii
viii Preface
The predictive power of a diagnostic test depends on the existing information
concerningtheprevalenceoftheconditiontobediagnosedinarelevantpopulation.
Similarly,thepredictivepowerofaclinicaltrialdependsonavailablepriorknowledge
concerningtheinvestigationalproduct.Articulatingsuchpriorknowledgeisthetopic
ofChap.5.Inthepast10years,wehavewitnessedanincreasinguseofhistorical
controldatainthedesignandanalysisofclinicaltrials,bothwhenconcurrentcontrol
data are available and when they are completely absent. We expanded Chap. 5 to
include this topic and include an example on the challenge and use of historical
control data when the trial providing evidential basis for an orphan drug approval
included no concurrent control. In Chap. 6, we describe metrics that are useful to
evaluatedesignsandassociateddecisionrulesforefficacyassessmentatthevarious
stagesofpre-marketingdevelopment.Thefocusonefficacyisduetothegenerally
well-definedendpointstodecideuponthebeneficialeffectofanewdrug.Chapter7
covers the proof-of-concept stage, while Chaps. 8 and 9 cover the dose–response
and confirmatory stage, respectively. Throughout Chaps. 7–9, we have added new
exampleswheretheprimaryendpointisbinary.
Chapter10focusesonassessingthedesignofatrialforcomparativeeffectiveness
assessment. By comparative effectiveness, we mean the comparison of different
active treatments to determine which treatment works best. This focus reflects the
increasingimportanceofthesecomparisonsinthemarketplaceduetotheneedto
justifythepriceandtoqualifyforreimbursement.
ThemetricsusedinChaps.7–10donotincludeanycostconsiderationexplicitly.
But, cost is an integral part of drug development strategy optimization. Incorpo-
rating cost into design consideration is the topic of Chap. 11 with two example
approaches.Thefirstoneoptimizesabenefit–costefficiencyscorethatmeasuresthe
cost-effectivenessofaproof-of-concepttrialdesign.Thesecondapproachcombines
costs and potential commercial returns to assess drug development options. The
chapterincludesadetaileddiscussiononthecalculationoftheexpectednetpresent
value which could be of interest to readers without much exposure to product
valuation.
InChap.12,weexaminethebiasthatcanbeproducedbyuseofPhase2resultsthat
havebeenselectedbecauseofafavorableoutcome.Wehavehintedatthissourceof
biasinearlierchaptersandhavededicatedChap.12tothisissue.Wehaveexpanded
thischaptertoincludeanapproachbasedonthemaximumlikelihoodestimationof
atruncatedNormaldistributionsandnewapproacheswepublishedintheliterature
inrecenttimes.Wecompareddifferentadjustmentmethodswithrespecttobiasand
othermeasuressuchastheprobabilityoflaunchingaPhase3trialandtheaverage
statisticalpowerofthelaunchedtrial.Weoffersomerecommendationsbasedonthe
comparativeresults.
Chapter 13 is new, starting with classes of adaptive designs discussed in the
finalizedguidanceonadaptivedesignissuedbytheFoodandDrugAdministration
(FDA) in the United States (USA). We have included four examples of adaptive
designsfor,agroupsequentialtrialwithaninterimanalysisforfutility,anadaptive
dose–response study, a Phase 3 trial with a pre-planned sample size re-estimation
pluspopulationenrichmentandaseamlessPhase2/3trial.
Preface ix
In the final chapter of the book, we include selected topics that affect design
anddecisionchoicesatallstagesofdrugdevelopment.Examplesincludesequences
of trials with a correlated treatment effect, benefit–risk and economic assessment.
Theseareallactiveresearchareas.Eventhoughweoffersomereferences,itisnot
ourintenttocovertheseareasindetailinthisbook.
Whilethemajorityofthebookisdedicatedtotrialplanningandsettingupdecision
rules, we have also included the analyses of completed trials to share insight and
lessonslearned.Exampleswiththisobjectiveincludetwodose–responsestudiesof
tofacitinibformoderateandsevererheumatoidarthritisinChap.8andthreeadaptive
designsinChap.13.
WehaveincludednumerousguidancespublishedbytheUSFDA,theEuropean
MedicinesAgency(EMA)andtheInternationalCouncilforHarmonisation(ICH).
Instead of providing URL links to these guidances which could become obsolete
overtime,weareofferingpathstolocatetheseguidances.Forguidancesissuedby
the US FDA, readers can use the “Search All FDA Guidance” option at the path
of www.fda.gov -> Drugs -> Guidance, Compliance and Regulatory Information.
Guidances issued by the ICH could be reached by selecting the “Work Products”
tabfollowedby“AllGuidelines”optionattheICHhomepagewww.ich.org.Asfor
the EMA guidelines, readers can use the “Search for scientific guidelines” link in
the middle of the page reached by paging through www.ema.europa.eu -> Human
Regulatory->ResearchandDevelopment->ScientificGuidelines.
As we stated in the Preface for the first edition, developing a new drug is a
high-riskandhigh-rewardenterprise.Thehighriskisreflectedbythegenerallylow
successrateofturninganewmolecularentityintoanapproved drug.Thesuccess
rate has fluctuated over time and has also varied across therapeutic areas. While
thesuccessratehasimprovedinrecentyearsforcancerdrugsduetotheadventof
targetedtherapies,theratehasbeendisappointinglylowforcertaindisorderssuch
asAlzheimer’sdisease.
In addition to the high risk, the cost of developing a new drug has increased at
a pace faster than inflation. Tuft’s Center for the Study of Drug Development has
publishedaseriesofreportsexaminingtheaveragepre-taxindustrycosttobringa
new medicine to market. The most recent report, published in 2016, estimated an
averagecostaround$2.56billionUSDin2013money.Bycomparison,in2003,the
costwasabout$1.04billionin2013dollars,basedonthesamemethodofcalculation.
While some researchers have questioned these figures, these reports nevertheless
showasubstantialincreaseinthecostofdrugdevelopmentoverafewdecades.
The low success rate and the high cost have motivated many pharmaceutical
companiestolookforbettermethodstomakeportfoliodecisionsincludingwhether
toinvestinaparticularnewmolecularentityandhowtomakeGo/No-Godecisions.
Sincethemajorityofdevelopmentprogramsarelikelytofail,itisimportanttobe
abletoterminateaprogramwithalowprobabilityofsuccessasearlyaspossible.
At Pfizer, where both of us worked for many years, the journey to quantitative
decisionsbeganduringthefirstdecadeofthetwenty-firstcentury.Theimplementa-
tionbeganwithproof-of-conceptstudies.Teamsdesigningtheseearlystudieswere
required to present, to a technical review committee, the operating characteristics
x Preface
oftheirtrials/decisionruleswithrespecttothetargetproductprofile.Themoveto
assesstheprobabilityofsuccessinthelate-stagetrialswasfirmlyinplacebytheyear
2010 with the establishment of a Probability of Technical and Regulatory Success
(PTRS)Council.
Manystatisticiansandscientistsplayedacriticalroleintheabovejourney.The
inputfromcommercialcolleagueshelpedsolidifytheneedtoquantitativelyincor-
poratethetargetproductprofilewhendesigningatrialandsettingupthesubsequent
decisionrule.WehavelearnedagreatdealfromtheearlypioneeradvocatesatPfizer.
Theirworkinspiredustowritethisbook.WeareparticularlyindebtedtoMikeBrown,
AlanClucas,VladDragalin,WayneEwy,BradleyMarchant,KenKowalski,Mike
K.Smith,JonathanFrench,CyrusHoseyni,RichardLalonde,ScottMarshall,Peter
Milligan,MohanBeltangady,PhilWoodward,JoannaBurke,NealThomasandLiam
Ratcliffefortheirscientificandorganizationalleadership.
The book is written for readers with a broad range of responsibilities in drug
development. While the book contains a lot of technical details for quantitative
scientists,italsocontainsplentyofconceptspresentedinaunifiedframeworkwhich,
webelieve,canhelplessquantitativereadersmakemorequantitativedecisions.
Wehopeyouwillenjoyreadingthebookasmuchaswedidwritingandrevising
it. Try as we may to ensure that the contents of the book are correct, there is
always the possibility that some mistakes have gone undetected. If you spot one
of these mistakes, we would be grateful if you could let us know by mailing us
[email protected](ChristyChuang-Stein)[email protected]
(SimonKirby).Alternatively,wecanbereachedatwww.linkedIn.com.
Toclose,wehopethatwewillallcomeoutofthepandemicwithnewinsightson
trial conduct and what public–private partnerships could accomplish in expediting
thedevelopmentoflife-savingmedicines.
Kalamazoo,USA ChristyChuang-Stein
Ramsgate,UK SimonKirby
March2021
