EMEA-EFPIA Workshop on Adaptive Designs in Confirmatory Clinical Trials

London, December 14 , 2007

When is it appropriate to combine phases

Vlad DragalinStatistical Research and Applications Global Biostatistics & Programming

V. Dragalin | EMEA-EFPIA Workshop on Adaptive Designs | Dec 14, 2007 | London 2

Critical Path Initiative

Calls for “New tools to get

n fundamentally better answers about how the safety and effectiveness of new products can be demonstrated,

n in faster time frames,n with more certainty, n and at lower costs”

Seamless Phase II/III designs offer great potentialfor achieving these objectives

V. Dragalin | EMEA-EFPIA Workshop on Adaptive Designs | Dec 14, 2007 | London 3

PhRMA WGAD White Paper

V. Dragalin | EMEA-EFPIA Workshop on Adaptive Designs | Dec 14, 2007 | London 4

Seamless Designs: Definition

Seamless designn A clinical trial design that combines into a single trial

objectives which are traditionally addressed in separate trials (operationally seamless)

Adaptive Seamless designn A seamless trial in which the final analysis will use

data from patients enrolled before and after the adaptation (inferentially seamless)

V. Dragalin | EMEA-EFPIA Workshop on Adaptive Designs | Dec 14, 2007 | London 5

Faster: Operationally SeamlessTraditional Phase II + Phase III trials

A

B

C

Placebo

B

Placebo

Data

Analysis

Planning

Phase III

Development Timeline

Operationally Seamless Phase II/III trials

A

B

C

Placebo

B

Placebo

Phase II

Phase II

End of Phase III

End of Phase III

Confirmatory Analysis

V. Dragalin | EMEA-EFPIA Workshop on Adaptive Designs | Dec 14, 2007 | London 6

At lower costs: Inferentially Seamless

Development Timeline

Inferentially Seamless Phase II/III trials

A

B

C

Placebo

B

Placebo

Phase II End of Phase III

Confirmatory Analysis

Operationally Seamless Phase II/III trials

A

B

C

Placebo

B

Placebo

Phase II End of Phase III

Confirmatory Analysis

V. Dragalin | EMEA-EFPIA Workshop on Adaptive Designs | Dec 14, 2007 | London 7

At lower costs: Inferentially Seamless

n Combining data from the ‘learning’ stage with data from the confirmation stage l adds efficiency in using patient datal draws stronger conclusions with the same number of exposed

patientsl reduces development time

n 2nd Stage data and the relevant groups from 1st Stage data should be combined in a way thatl Guarantees the Type I error rate for the comparison with

controll Produces efficient unbiased estimates and confidence intervals

with correct coverage probability

V. Dragalin | EMEA-EFPIA Workshop on Adaptive Designs | Dec 14, 2007 | London 8

Better: Adaptive Seamless

Development Timeline

D

E

Placebo

Adaptive Seamless Phase II/III trials

A

B

C

IA IA IA IA

Dropped for toxicity

Dropped for futility

V. Dragalin | EMEA-EFPIA Workshop on Adaptive Designs | Dec 14, 2007 | London 9

Better: Adaptive Seamless

n More doses than in traditional Phase II can be usedn Phase II part uses target population n Treatment selection may be based on a short-term

endpoint X, while confirmation stage uses a long-term (clinical) endpoint Y

n Complex relationship between X and Y may improve decision makingl Include direct effects of treatment on Y not mediated through Xl And effects of patient covariates on both Y and X

n Adaptive model-based designs in Stage I may improve selection of the target dose for Stage II

V. Dragalin | EMEA-EFPIA Workshop on Adaptive Designs | Dec 14, 2007 | London 10

More Certainty: Adaptive Dose Ranging

0

10

20

30

2 4 6 8

ANOVA5 doses

Dopt5 doses

2 4 6 8

GADA5 doses

MCPMod5 doses

2 4 6 8

MTT5 doses

BMA5 doses

2 4 6 8

LOCFIT5 doses

ANOVA7 doses

Dopt7 doses

GADA7 doses

MCPMod7 doses

MTT7 doses

BMA7 doses

0

10

20

30

LOCFIT7 doses

0

10

20

30

ANOVA9 doses

2 4 6 8

Dopt9 doses

GADA9 doses

2 4 6 8

MCPMod9 doses

MTT9 doses

2 4 6 8

BMA9 doses

LOCFIT9 doses

Dose selected

% T

rials

PhRMA ADRS WG

White Paper

JBS 2007, v 17(6)

Adaptive dose-ranging methods lead to:

• gains in power to detect DR

• precision to select target dose, and

• to estimate DR

V. Dragalin | EMEA-EFPIA Workshop on Adaptive Designs | Dec 14, 2007 | London 11

Adaptive Seamless Designs

Primary objective – combine “dose selection” and “confirmation” into one trial

n Although dose is most common Phase IIb objective, other choices could be made, e.g. population

n After dose selection, the only change is to new enrollments (patients are generally not re-randomized)

n Patients on terminated treatment groups could be followed

n All data from the chosen group and comparator is used in the final analysis. Appropriate statistical methods must be used

V. Dragalin | EMEA-EFPIA Workshop on Adaptive Designs | Dec 14, 2007 | London 12

Potential Benefitsn More efficacy/dose information prior to triggering Phase III

l by response-adaptive treatment allocation, dropping treatment arms

n Reduced development timelines and costl by cutting out the ‘white space’ between Ph II and Ph III, using the

same operational infrastructure

n More safety information on longer term patient exposurel by allowing for longer follow-up on patients enrolled in the 1st Stage

n Higher chance of patients within the trial to be treated with efficacious and safe dosesl by making timely use of available information in the interest of

patient benefit

V. Dragalin | EMEA-EFPIA Workshop on Adaptive Designs | Dec 14, 2007 | London 13

New Challenges

n Drug supply and drug packaging could be more challenging

n Need the final formulation available at the start of the seamless trial

n Challenges regarding data review at the end of ‘learning’ stage

n Decision process may require additional expertise and/or perspective not usually represented on DMCs

n Sponsor involvement in decision makingn Information inferable from the selection decision may

impact the operational bias