A Bayesian Adaptive Design for Phase 2b Dose-Finding Using a Composite Endpoint

• Activity Number: 45
• Type: Contributed
• Date/Time: Sunday, August 9, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Norman Bohidar* and Kyle Wathen
• Companies: and Johnson & Johnson
• Keywords: Adaptive ; Bayesian ; Dose

Abstract

In the development of therapeutic agents where validated surrogate markers are not available, it is not uncommon to want to bypass the Phase 2b study as the need to study multiple doses with hard clinical outcomes as the primary endpoint can be both cost and time prohibitive. Of course, initiating a large Phase 3 trial in the absence of understanding the probability of success is not sound from a scientific, regulatory, or business perspective. We will present a Bayesian adaptive 2b design to potentially alleviate these issues. We will discuss how an expanded composite (event based) endpoint can be utilized to improve internal and external interpretation of the study findings, while avoiding Phase 3 sample size levels. In addition, use of prior information which can be quantified and incorporated easily in the beta-binomial framework will also assist in further sample size savings. The process for dropping doses, as well as the general operating characteristics will be described. Lastly, issues regarding the interpretation of Type 1 Error control and the challenges of combining information from this study with the pivotal, confirmatory trial will also be explored.