Bayesian Methods in Phase 3 HCV Trials: Borrowing Historical Control Arm Data

• Activity Number: 75
• Type: Contributed
• Date/Time: Sunday, July 31, 2016 : 4:00 PM to 5:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Ran Liu* and Qi Tang and Martin King and Bo Fu and Sandra Lovell and Alan Hartford
• Companies: AbbVie and AbbVie and AbbVie and AbbVie and AbbVie and AbbVie
• Keywords: Bayesian method ; Historical data borrowing ; Type I error ; Power

Abstract

A Bayesian method was to be conducted to incorporate borrowing of data from historical trials of the control regimen for the primary analysis in a phase 3 trial to compare efficacy and safety of an experimental treatment to Sofosbuvir (SOF) co-administered with Ribavirin (RBV) in adults with chronic hepatitis C virus (HCV) genotype 3 (GT3) infection. In this method, data from historical trials of the control regimen were "borrowed" and combined with data from the control arm in the current trial, which could lead to more accurate point estimates, greater power, and reduced type I error. The extent of borrowing was controlled by the Bayesian hierarchical model being utilized based on the degree of similarity in the control arm between the historical and new trials. Simulations were performed to show that the borrowing of historical control data meaningfully improved the power of the study while maintaining appropriate control of the type I error rate. The proposed method thus provided a means to incorporate data from historical trials and improve the efficiency of the current trial while maintaining appropriate control of the type I error rate.