Planning from Pilot Study to Confirmatory Trial Using a Bayesian Approach

• Activity Number: 449
• Type: Contributed
• Date/Time: Wednesday, August 10, 2005 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Hong Amy Xia*+ and Bin Yao and Moraye Bear and Alan Forsythe
• Companies: Amgen, Inc. and Amgen, Ltd., Cambridge and Amgen, Inc. and Amgen, Inc.
• Keywords: Bayesian method ; Sample size ; Clinical significance ; Decision-making

Abstract

A single-arm pilot study may be intended to provide an initial assessment of drug efficacy before expanding to a larger controlled comparative clinical trial. A decision rule for the pilot study is needed for determining whether to move forward with the confirmatory trial. We adopt a Bayesian approach in such a setting to facilitate the decisionmaking process by calculating the probability of a "go" decision for the next trial based on a range of decision rules and sample size scenarios for the first trial; for the second trial, the unconditional probability of achieving either statistical or clinical "significance" over the prior distribution, based on information from the first trial. We show how sample sizes can be determined explicitly by a trade-off between how stringent the decision rule of the first study is set and the expected sample size of the second trial. A final decision will need to take into account input from medical experts, the stage of the product life cycle, and the commercial and regulatory risks. The interplay between statistics and other functions in drug development needs to be emphasized.