Online Program

Return to main conference page

All Times EDT

Friday, September 25
Fri, Sep 25, 2:00 PM - 3:15 PM
Virtual
Trial Design and Analysis Considerations in the Product Development for Rare Diseases

Bayesian Frameworks for Rare Disease Clinical Development Programs (301163)

View Presentation

Brad Carlin, Counterpoint Statistical Consulting, LLC 
*Freda Cooner, Amgen 
Forrest Williamson, Eli Lilly 

Keywords: Bayesian Statistics, Rare Diseases, Historical Data, Robust Mixture Prior

Clinical development of orphan products in rare diseases poses unique clinical and statistical challenges, mainly surrounding the small population issue. Bayesian statistics naturally are adopted as a tool for both more efficient trial designs and more reliable treatment effect estimates. We will start this presentation with a summary of the global regulatory background that fosters clinical research in rare diseases, and which in turn has stimulated statistical innovations including the use of Bayesian methodologies in clinical trials. The use of natural history studies, long-term safety evaluation following marketing approval, and real-world data utilization will be briefly discussed, with specific considerations in orphan products' clinical settings. The main focus of this presentation is the potential implementation of Bayesian approaches in registrational trials for orphan drug regulatory approvals. We will briefly review trial designs that could incorporate Bayesian statistics and existing Bayesian approaches (e.g. two-step and power priors), and also introduce robust mixture priors. It will then be followed by a case study in rare disease development to illustrate the robust mixture approach with historical data. Advantages and caveats when using Bayesian statistics will be discussed throughout, focusing on the rare disease clinical development environment. The presentation will conclude with some encouraging recent regulatory updates and our current thinking on rare disease clinical development, with emphasis on the Bayesian framework.