A Test Based on Monte Carlo Simulations for Biomarker-Adaptive Threshold Design

• Activity Number: 653
• Type: Contributed
• Date/Time: Thursday, August 8, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Yafeng Zhang*+ and Glen Laird
• Companies: UCLA and Bristol-Myers Squibb
• Keywords: Adaptive threshold design ; Biomarker ; Monte Carlo test

Abstract

Under the background of personalized medicine, identifying a subset of sensitive patients who benefit from a treatment is crucial to the success of the treatment. Jiang et al. (2007) proposed a biomarker-adaptive threshold phase III design to find such a subset and test the treatment effect in the identified subgroup. However, this design relies on either a permutation test or an asymptotic approximation of P value (Miller and Sigmund 1982). The permutation test is time consuming, and the asymptotic approximation of P value is shown to be too conservative by simulation studies. Therefore, we proposed a test based on Monte Carlo simulations to test the treatment effect in the identified subset of patients. Simulations studies showed that the proposed test is fast and reliable, controls type I error rate and has similar power with permutation test.