A Novel Bayesian Approach to Designing Dose-Ranging Clinical Trials: A More Efficient Alternative to Traditional Approaches

• Activity Number: 178
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Jianjun Gan*+ and Amy Cutrell
• Companies: GlaxoSmithKline and GlaxoSmithKline
• Keywords: Phase IIb ; Bayesian Methods ; Dose Ranging ; Predictive ; Clinical Trials

Abstract

The goal here is to present a Bayesian study design that provides the information needed for dose ranging decision-making in a highly efficient manner. This will be accomplished through the examination of a simulated case study: A three-arm, dose ranging trial employing a Bayesian design to compare the primary endpoint (virologic response) of two dose arms of new antiviral Compound+ backbone to current standard of care. The weak informative prior is used for the distribution of the response rate for two tested dose arms. For standard of care, a strong informative prior for the response rate allows for a reduction in the sample size for the control arm while maintaining overall power and controlling type I error for the primary comparisons. Also the Bayesian predictive probability will be used to determine whether one test dose arm needs to be dropped after approximately 50% subjects are enrolled. Simulation results demonstrating the properties of the design will be shown as well.