A P ti A P ti A Perspective o A Perspective o Environment for A Environment for A Tri Tri Robert T. O Robert T. O Senior Statist Senior Statist Office of Translational Office of Translational For presentation at the 2 nd Annual East U MA ; October 12, 2012 th C t th C t on the Current on the Current Adaptive Clinical Adaptive Clinical ials ials ’Neill, PhD ’Neill, PhD tical Advisor tical Advisor l Sciences, CDER, FDA l Sciences, CDER, FDA ser Group Meeting, Cambridge, 1 6 | Horizontal | Short | Landscape | Yes | Landscape
52
Embed
Eugm 2012 oneill - perspective on the current environment for adaptive clinical trials - 2012 eugm
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
A P ti A P ti A Perspective oA Perspective oEnvironment for AEnvironment for A
TriTriRobert T. ORobert T. O
Senior StatistSenior Statist
Office of TranslationalOffice of Translational
For presentation at the 2nd Annual East UMA ; October 12, 2012
th C t th C t on the Current on the Current Adaptive Clinical Adaptive Clinical ppials ials ’Neill, PhD’Neill, PhD
Philosophy of FPhilosophy of FPhilosophy of FPhilosophy of F
Understandable to a wiUnderstandable to a wiclinicians and statisticiaclinicians and statisticiaclinicians and statisticiaclinicians and statisticia
Cautionary but encouraCautionary but encourastudies and for confirmstudies and for confirm
ddwarrantedwarranted
Terminology and definiTerminology and definiSeamless Phase 2 /PhaseSeamless Phase 2 /PhaseSeamless Phase 2 /PhaseSeamless Phase 2 /Phaseexploration with confirmexploration with confirm
We are still learning, exWe are still learning, exl dl d
ggevolving , as is methodoevolving , as is methodo
FDA Guidance FDA Guidance FDA Guidance FDA Guidance
de audience, including de audience, including ans and othersans and othersans, and othersans, and others
age use for exploratory age use for exploratory matory studies, if and when matory studies, if and when
itions are important: itions are important: e 3 terminology confuses e 3 terminology confuses e 3 terminology confuses e 3 terminology confuses mationmation
xamples are increasing and xamples are increasing and ll
Definition andDefinition andd id iadaptiveadaptive(FDA Gu(FDA Gu
For the purposes of this guFor the purposes of this guclinical study is defined as clinical study is defined as
ti l l d d ti l l d d prospectively planned and sprospectively planned and spotential for modification opotential for modification oaspects of the study design aspects of the study design
l i f d t f bjl i f d t f bjanalysis of data from subjecanalysis of data from subjecof the accumulating study dof the accumulating study dprospectively planned poinprospectively planned poinb f d i b f d i f ll blif ll blibe performed in a be performed in a fully blinfully blinblinded mannerblinded manner, and may, at, and may, atoccur with or without formoccur with or without formt ti ( d i i t l ) t ti ( d i i t l ) testing (eg. decision tools). testing (eg. decision tools).
d concept of an d concept of an d i d ie designe design
uidance)uidance)idance, an adaptive design idance, an adaptive design a study that includes a a study that includes a
ifi d difi tiifi d difi ti specified modificationspecified modification or or of one or more specified of one or more specified and hypotheses, and hypotheses, based on based on t i th t dt i th t d A l A l cts in the studycts in the study. Analyses . Analyses
data are often performed at data are often performed at nts within the study, may nts within the study, may
d d i d d i nded manner or in a nonnded manner or in a non--t an interim time point, t an interim time point,
mal statistical hypothesis mal statistical hypothesis
hh “ i ”“ i ” hhThe The term “prospective”term “prospective” hereherewas planned (and details spewas planned (and details speexamined in an unblinded foexamined in an unblinded foinvolved in planning for theinvolved in planning for theinvolved in planning for theinvolved in planning for theplans that are introduced or plans that are introduced or has commenced if confidenchas commenced if confidencthe personnel involved is unthe personnel involved is unthe personnel involved is unthe personnel involved is unand documented when the mand documented when the mproposed. The documentatioproposed. The documentatiounequivocal assurance of bliunequivocal assurance of bli
l hil d i l hil d i personnel while a study is onpersonnel while a study is onto discuss with FDA. to discuss with FDA. ChangChangoccurring after an interim anoccurring after an interim andata and that were not prospdata and that were not prospdata and that were not prospdata and that were not prospwithin the scope of this guidwithin the scope of this guid
rificationsrificationsid )id )uidance)uidance)
h h d ih h d ie means that the adaptation e means that the adaptation ecified) before data were ecified) before data were orm by any personnel orm by any personnel
e revision This can include e revision This can include e revision. This can include e revision. This can include finalized after the study finalized after the study
ce in the blinded state of ce in the blinded state of nequivocally maintained nequivocally maintained nequivocally maintained nequivocally maintained modification plan is modification plan is on that can support on that can support inding for the necessary inding for the necessary
i b i i b i ngoing may be important ngoing may be important ges in study design ges in study design nalysis of unblinded nalysis of unblinded study study pectively planned are not pectively planned are not pectively planned are not pectively planned are not dance. dance.
R i i d dR i i d dRevisions made or proposedRevisions made or proposedanalysisanalysis raise major concernraise major concernpotential introduction of biapotential introduction of biad fi d d f ll i ld fi d d f ll i ldefined and carefully impledefined and carefully impleirresolvable uncertainty in tirresolvable uncertainty in tIn contrast, In contrast, revisions based revisions based
f d t ( t f d t ( t of data (e.g., aggregate evenof data (e.g., aggregate evenrates, baseline characteristicrates, baseline characteristicbias to the study or into subbias to the study or into subth l C t ith l C t ithe same personnel. Certainthe same personnel. Certainsuch as sample size revisionsuch as sample size revisionor variance, are advisable pror variance, are advisable pr
b i l l b i l lcan be prospectively plannecan be prospectively planne
rificationsrificationsid )id )uidance)uidance)
d ft bli d d i t id ft bli d d i t id after an unblinded interimd after an unblinded interimns about study integrity (i.e., ns about study integrity (i.e., as) and need to be prospectively as) and need to be prospectively
t d t id i ki t d t id i ki emented to avoid risking emented to avoid risking the interpretation of study results. the interpretation of study results. on blinded interim evaluations on blinded interim evaluationst t i di ti ti t t i di ti ti t rates, variance, discontinuation t rates, variance, discontinuation
cs, etc.) usually do not introduce cs, etc.) usually do not introduce bsequent study revisions made by bsequent study revisions made by
bli d d bli d d d t b d h d t b d h n blindedn blinded--data based changes, data based changes, ns based on aggregate event rates ns based on aggregate event rates rocedures for most studies, and rocedures for most studies, and d d ed. ed.
Exploratory studies Exploratory studies primary basis for effprimary basis for effprimary basis for effprimary basis for effwell designed to addwell designed to addexplore dose, populaexplore dose, populaconditions, PD endpconditions, PD endp
y studies and y studies and ry studiesry studiesry studiesry studies
es : otherwise known as es : otherwise known as t ll d ll l t t ll d ll l t controlled, usually later controlled, usually later
es that have prees that have pre--stated stated
: not intended as : not intended as ficacy evaluations yet ficacy evaluations yet ficacy evaluations yet ficacy evaluations yet dress dose selection or dress dose selection or ations, optimal study ations, optimal study points, markers, etc.points, markers, etc.
The intersection oThe intersection ob i hb i hbayesian approachesbayesian approaches
We are not there yet, buWe are not there yet, buin progressin progressin progressin progress
The NIH/FDA ADThe NIH/FDA AD
F th di iF th di i Further discussionFurther discussiondemonstrate familydemonstrate familycontrol for multiplcontrol for multipldesignsdesignsdesignsdesigns
Promotion to all when oPromotion to all when oissues issues –– a potential to raa potential to ra
ddpp
contribute to disappoincontribute to disappoin
of frequentist and of frequentist and t d ti d i t d ti d is to adaptive designss to adaptive designs
ut interesting initiatives are ut interesting initiatives are
APTAPT--IT projectIT project
i l ti t i l ti t n on simulations to n on simulations to y wise strong type 1 y wise strong type 1 le hypotheses in K stage le hypotheses in K stage
only a few understand the only a few understand the aise expectations and aise expectations and
l dl dpp
ntments ntments –– it can be solvedit can be solved
• “To illustrate and exploreTo illustrate and explore clinical trial designs to impdrugs and medical devicesdrugs and medical devicesmethods to characterize aopinions and concerns ofopinions, and concerns of and after the developmen
ectiveective
how best to use adaptivehow best to use adaptive prove the evaluation of s and to use mixeds and to use mixed and understand the beliefs, f key stakeholders duringf key stakeholders during nt process.”
• “To illustrate and exploreTo illustrate and explore clinical trial designs to impdrugs and medical devicesdrugs and medical devicesmethods to characterize aopinions and concerns ofopinions, and concerns of and after the developmen
ectiveective
how best to use adaptivehow best to use adaptive prove the evaluation of s and to use mixeds and to use mixed and understand the beliefs, f key stakeholders duringf key stakeholders during nt process.”
Most of FDA’s experience is witMost of FDA’s experience is witnetwork, funding streams, studynetwork, funding streams, study
These clinical studies are dealing with These clinical studies are dealing with solutions to date solutions to date –– ie. Study failures ie. Study failures -- itityycontribution to refining the issues, studcontribution to refining the issues, stud
The project has an evaluation of the proThe project has an evaluation of the pro
t i ADAPTt i ADAPT ITITrest in ADAPTrest in ADAPT--ITIT
R, CDER ) in the collaboration and R, CDER ) in the collaboration and tical) in the experiencetical) in the experiencetical) in the experiencetical) in the experience
mples) with adaptive designsmples) with adaptive designs
th the industry and not the NIH th the industry and not the NIH y sectionsy sections
complex issues that have not had good complex issues that have not had good t is an opportunity to determine if new t is an opportunity to determine if new pp ypp ydy designs, questions, etc.dy designs, questions, etc.
ocess and an assessment of participantsocess and an assessment of participants
A guidance on enricA guidance on enricsoon be out for commsoon be out for commsoon be out for commsoon be out for commcommitment from Pcommitment from P
designs and designs and eir relationship eir relationship ve approachesve approachesve approachesve approaches
hment designs will hment designs will ment ment –– another another ment ment another another DUFA IVDUFA IV
Early reportingEarly reportingEarly reportingEarly reportingunblinding, preservunblinding, preservcritical for adaptivecritical for adaptivecritical for adaptivecritical for adaptive
g of trial results g of trial results g of trial results, g of trial results, ving trial integrity ving trial integrity ––e designs to succeede designs to succeede designs to succeede designs to succeed
A BA B P d t f tP d t f tA ByA By--Product of tProduct of t
Better up front planning andBetter up front planning andclinical trials, fixed or adapticlinical trials, fixed or adapticlinical trials, fixed or adapticlinical trials, fixed or adapti
Consideration of routine useConsideration of routine useadvance how trial assumptioadvance how trial assumptiounexpected features will impunexpected features will impunexpected features will impunexpected features will imp
My concern !My concern !
Not enough time to doNot enough time to do Not enough time to doNot enough time to dofast fast –– this is a return tothis is a return to
th AD i iti tith AD i iti tithe AD initiativethe AD initiative
d design choices for all late phase d design choices for all late phase ive ive –– the choice is not obviousthe choice is not obviousive ive the choice is not obviousthe choice is not obvious
e of simulations to play out in e of simulations to play out in ons as well as unknown or ons as well as unknown or pact results pact results NOT THAT EASYNOT THAT EASYpact results pact results –– NOT THAT EASYNOT THAT EASY
o the planning o the planning –– desire to move too desire to move too o the planning o the planning –– desire to move too desire to move too o failure in my mindo failure in my mind
Disease progression basDisease progression basfollowfollow--up data: clinical up data: clinical d d i i l d l fi i l d l fdata data -- empirical model fempirical model f
SimulSimulMany options aMany options aMany options aMany options aUsed for planniUsed for planni
To evaluate what you obTo evaluate what you obsuspected suspected -- sensitivity asensitivity asuspected suspected sensitivity asensitivity a
To evaluate what you hTo evaluate what you hdata approaches data approaches -- fill infill in
i h i bi h i bpoints where it was obspoints where it was obs
To predict what you didTo predict what you didapproaches with properapproaches with properapproaches with properapproaches with properuncertainty associated wuncertainty associated wprediction itselfprediction itself
l fl f To evaluate performancTo evaluate performancbias, achieved dose levebias, achieved dose leve
lationslationsand applicationsand applicationsand applicationsand applicationsing and analysising and analysis
bserved, but where bias is bserved, but where bias is analysisanalysisanalysisanalysis
have not observed have not observed -- missing missing n data between design n data between design
ddservedserved
d not observe d not observe -- prediction prediction r accounting for stochastic r accounting for stochastic r accounting for stochastic r accounting for stochastic with model selection and with model selection and
ce characteristics; power, ce characteristics; power, elsels
The current eThe current efor protocofor protocofor protocofor protoco
Simulate the performance chaSimulate the performance chadesigns , fixed or adaptive, befordesigns , fixed or adaptive, beforexercise to engage the entire deveexercise to engage the entire deve
made and the expectations suchmade and the expectations suchmade and the expectations suchmade and the expectations sucheveryone understand variability, everyone understand variability, uncertainties involved uncertainties involved –– that shouthat shou
aracteristics of candidate study aracteristics of candidate study re they are carried out and use the re they are carried out and use the lopment team in the choices to be lopment team in the choices to be h choices present h choices present it will help it will help h choices present h choices present –– it will help it will help sources of heterogeneity, and the sources of heterogeneity, and the uld reduce failure rates hopefullyuld reduce failure rates hopefully
Interactions with FDAInteractions with FDAof an adaptiveof an adaptivepp
possibly dupossibly du
Early and Middle PerioEarly and Middle Perio
Late stages of drug deveLate stages of drug deve
Special Protocol AssessSpecial Protocol Assesspp
FDA does not wish to bFDA does not wish to bdesign decisions, as witdesign decisions, as witdecisionsdecisions
However, it may be timHowever, it may be timi t i ht di t i ht dinspects, oversights, audinspects, oversights, audcompliance with SOP’s compliance with SOP’s
A during the planning A during the planning e design study and e design study and g yg yuring its conducturing its conduct
d of Drug developmentd of Drug development
elopmentelopment
mentsments
be involved in any adaptive be involved in any adaptive th not being aware of DMC th not being aware of DMC
me to consider how one me to consider how one dit d dit d dits, and assures dits, and assures for firewalls for firewalls
A lot of literature coming ouA lot of literature coming oublinded access to interim datblinded access to interim dat
Others suggest new ways to Others suggest new ways to designsdesigns
V l t i i t i l t V l t i i t i l t Value to up sizing a trial to mValue to up sizing a trial to mzone’zone’
Interesting to see and hear NInteresting to see and hear Nresponse and/or reactions response and/or reactions nnresponse and/or reactions response and/or reactions –– nnthisthis
Also interesting in ADAPTAlso interesting in ADAPT--IIinteractions and iterative feeinteractions and iterative feeinteractions and iterative feeinteractions and iterative feeexperienceexperience
h thi k ?h thi k ?hers think ?hers think ?
ut ut –– some do not even agree that some do not even agree that ta protects type 1 errorta protects type 1 errorp ypp yp
i t i i t i th ‘ i i th ‘ i i maintain power maintain power –– the ‘promising the ‘promising
NIH network statisticians NIH network statisticians not clear where they are with not clear where they are with not clear where they are with not clear where they are with
IT to observe the clinical team IT to observe the clinical team edback edback –– part of the learning part of the learning edback edback –– part of the learning part of the learning
So what have wSo what have wthe last 5 ythe last 5 ythe last 5 ythe last 5 y
There has been an extensive literatuThere has been an extensive literatu
Regulators have helped focus the pRegulators have helped focus the p Regulators have helped focus the pRegulators have helped focus the pfuture of these designs (Europe andfuture of these designs (Europe and
A lot of education remains A lot of education remains –– but it ibut it i
Several companies now specializinSeveral companies now specializin p pp p
FDA feedback and constructive advFDA feedback and constructive advare viewed and usedare viewed and used
Methodology is not the issue Methodology is not the issue –– impimp
Over promotion; naïve unblOver promotion; naïve unbl
Continued need for communicationContinued need for communication
we learned over we learned over years or soyears or soyears or soyears or so
ure developed on AD’s with good ideas ure developed on AD’s with good ideas
purpose utility and framework for the purpose utility and framework for the purpose, utility and framework for the purpose, utility and framework for the d US)d US)
is happening; conferences, ADAPTis happening; conferences, ADAPT--ITIT
ng in advice and designng in advice and designg gg g
vice is very important to how the designs vice is very important to how the designs
plementation isplementation is
linding can be a problemlinding can be a problem
n among disciplines and experiencen among disciplines and experience