A Hierarchical Bayes Model for Biomarker Subset Effects in Clinical Trials

• Activity Number: 527
• Type: Contributed
• Date/Time: Wednesday, August 1, 2012 : 10:30 AM to 12:20 PM
• Sponsor: SSC
• Author(s): Bingshu Chen*+ and Wenyu Jiang and Dongsheng Tu
• Companies: Queen's University and Queen's University and Queen's University
• Keywords: Biomarker ; Clinical Trials ; Gibbs Sampling ; Hierarchical Bayes Model ; Markov Monte Calo ; Survival Analysis

Abstract

Translational and clinical trials research in oncology increasingly emphasize the importance of biomarker development and validation. In this paper, we develop novel hierarchical Bayes method for estimating and making statistical inference for biomarker-defined sensitive subset of patient population, in which the treatment and the biomarker interactively affect clinical outcomes of patients. Here we consider the threshold as a random variable with certain probability distribution. By applying the hierarchical Bayes model, we are able to make use of the observed data to construct the prior distribution for the threshold parameter such that the posterior distribution is less depend on the prior assumption. Compared to the existing approaches such as the profile likelihood method, which makes inference about the threshold parameter using bootstrap, the proposed Bayes method provides better finite sample properties in term of biases for parameters estimation and coverage probabilities for the 95\% confidence intervals. The proposed method are applied to a clinical trial of prostate cancer with the serum prostatic acid phosphatase (AP) biomarker.