Dose-Finding Using Bayesian E-Max Model to Find Minimum Effective Dose

• Activity Number: 178
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Yukiko Imai*+
• Companies: GlaxoSmithKline
• Keywords: Dose finding ; Bayesian estimation ; Emax model ; Minimum effective dose

Abstract

Traditional dose finding using pairwise comparison could sometimes lead to poor understanding and large uncertainty for the dose-response curve. In pharmaceutical industries, dose finding is important phase to find the right dose for the successful drug development. In order to allow some flexibility to find the dose-response curve, we will use Bayesian E-max model with power (Hill) parameter applying to dose and ED50 parameter. We will estimate several sets of E-max model parameter by using Monte Carlo Markov Chain's (MCMCs) and evaluate the conversion by utilizing diagnostics such as Gelman-Rubin, Geweke, Raftery and Lewis, and visual inspection will be performed as well. The primary focus for this research is to find the minimum effective dose (MED). The primary endpoint for the study is continuous variable. The setting of the study is to assess the active arms with low and two to three medium and high doses over the placebo. MED is defined as smallest dose which is clinically relevant effect over placebo response. We will set the criteria of MED and discuss the performance.