Two-Stage Adaptive Design for Bioequivalence: Methods and Application

• Activity Number: 60
• Type: Topic Contributed
• Date/Time: Sunday, August 1, 2010 : 4:00 PM to 5:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Yi-Lin Chiu*+ and Yannis Jemiai and Charles Locke and Pralay Senchaudhuri
• Companies: Abbott Global Pharmaceuticals Research and Development and Cytel Inc. and Abbott Laboratories and Cytel Inc.
• Keywords: Bioequivalence ; Simulations ; Power ; Adaptive ; Pharmacokinetics

Abstract

Pivotal bioequivalence (BE) studies can be critical for gaining approval. Poor characterization of the variability in the pharmacokinetics or existence of a true difference in performance can lead to underpowered studies and ultimately to failure to demonstrate equivalence and gain market access. Based on the work by Gould (1995), we develop a tool to adequately design, simulate, and power two-stage adaptive designs for BE. Small sample sizes commonly found in these settings require us to use Student's t-distribution to control the type 1 and 2 error rates, all the while permitting early stopping for equivalence or due to futility. The design was simulated and implemented in a pivotal BE study for submission. The data showed clear evidence for nonequivalence of the formulations for one of the pharmacokinetic variables and the study was successfully terminated for futility after stage 1.