Bayesian Trial Designs: Drug Case Study

1
Bayesian Trial Designs Drug Case Study

• Donald A. Berry
• dberry_at_mdanderson.org

2
Outline

• Some history
• Why Bayes?
• Adaptive designs
• Case study

3
(No Transcript)
4
2004 JHU/FDA WorkshopCan Bayesian Approaches
to Studying New Treatments Improve Regulatory
Decision-Making?

• www.prous.com/bayesian2004
• www.cfsan.fda.gov/frf/ bayesdl.html

5
Upcoming in 2005

• Special issue of Clinical Trials
• Bayesian Clinical TrialsNature Reviews Drug
  Discovery

6
Selected history of Bayesian trials

• Medical devices (30)
• 200 at M.D. Anderson (Phase I, II, I/II)
• Cancer Leukemia Group B
• Pharma
• ASTIN (Pfizer)
• Pravigard PAC (BMS)
• Other
• Decision analysis (go to phase III?)

7
Why Bayes?

• On-line learning (ideal for adapting)
• Predictive probabilities (including modeling
  outcome relationships)
• Synthesis (via hierarchical modeling, for example)

8
PREDICTIVE PROBABILITIES

• Critical component of experimental design
• In monitoring trials

9
Herceptin in neoadjuvant BC

• Endpoint tumor response
• Balanced randomized, H C
• Sample size planned 164
• Interim results after n 34
• Control 4/16 25
• Herceptin 12/18 67
• Not unexpected (prior?)
• Predictive probab of stat sig 95
• DMC stopped the trial
• ASCO and JCOreactions

10
ADAPTIVE DESIGNS Approach and Methodology

• Look at the accumulating data
• Update probabilities
• Find predictive probabilities
• Use backward induction
• Simulate to find false positive rate and
  statistical power

11
Adaptive strategies

• Stop early (or late!)
• Futility
• Success
• Change doses
• Add arms (e.g., combos)
• Drop arms
• Seamless phases

12
Goals

• Learn faster More efficient trials
• More efficient drug/device development
• Better treatment of patients in clinical trials

13
ADAPTIVE RANDOMIZATIONGiles, et al JCO (2003)

• Troxacitabine (T) in acute myeloid leukemia (AML)
  combined with cytarabine (A) or idarubicin (I)
• Adaptive randomization to IA vs TA vs TI
• Max n 75
• End point Time to CR (lt 50 days)

14
Adaptive Randomization

• Assign 1/3 to IA (standard) throughout (until
  only 2 arms)
• Adaptive to TA and TI based on current
  probability gt IA
• Results ?

15
(No Transcript)
16
Drop TI
Compare n 75
17
Summary of results

• CR lt 50 days
• IA 10/18 56
• TA 3/11 27
• TI 0/5 0
• Criticisms . . .

18
Consequences of Bayesian Adaptive Approach

• Fundamental change in way we do medical research
• More rapid progress
• Well get the dose right!
• Better treatment of patients
• . . . at less cost

19
CASE STUDY PHASE III TRIAL

• Dichotomous endpoint
• Q P(pE gt pSdata)
• Min n 150 Max n 600
• 11 randomize 1st 50, then assign to arm E with
  probability Q
• Except that 0.2 P(assign E) 0.8

Small company!
20
Recommendation to DSMB to

• Stop for superiority if Q 0.99
• Stop accrual for futility if P(pE pS lt
  0.10data) gt PF
• PF depends on current n . . .

21
PF
22
Common prior density for pE pS

• Independent
• Reasonably non-informative
• Mean 0.30
• SD 0.20

23
(No Transcript)
24
Updating

• After 20 patients on each arm
• 8/20 responses on arm S
• 12/20 responses on arm E

25
Q 0.79
26
Assumptions

• Accrual 10/month
• 50-day delay to assess response

27
Need to stratify. But how?

• Suppose probability assign to experimental arm
  is 30, with these data . . .

28
(No Transcript)
29
One simulation pS 0.30, pE 0.45
Superiority boundary
Final Std 12/38 19/60
20/65 Exp 38/83 82/167 87/178
30
One simulation pE pS 0.30
Futility boundary
9 mos. End Final Std 8/39
15/57 18/68 Exp 11/42 32/81 22/87
31
Operating characteristics
32
FDA Why do this? Whats the advantage?

• Enthusiasm of patients investigators
• Comparison with standard design . . .

33
Adaptive vs tailored balanced design w/same
false-positive rate power (Mean number patients
by arm)
34
FDA

• Use flat priors
• Error size to 0.025
• Other null hypotheses
• We fixed all willing to modify as necessary

35
The rest of the story

• PIs on board
• CRO in place
• IRBs approved
• FDA nixed!

36
Outline

• Some history
• Why Bayes?
• Adaptive designs
• Case study