Predictive Probability Approach for the Design and Analysis of Response and Toxicity Data in Cancer Clinical Trials

• Activity Number: 552
• Type: Invited
• Date/Time: Wednesday, August 4, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): J. Jack Lee*+ and Guosheng Yin and Nan Chen
• Companies: MD Anderson Cancer Center and The University of Hong Kong and M D Anderson Cancer Center
• Keywords: Bayesian predictive probability ; efficiency ; ethical consideration ; response adaptive randomization ; operating characteristics ; early stopping

Abstract

Predictive probability (PP) computes the probability of a future target event given the interim observed data assuming current trend continues. If PP for efficacy or toxicity is very low or very high, the trial can be stopped early. For safety analysis, patients are not assigned to the arm(s) with high PP of toxicity. For efficacy analysis, coupling with response adaptive randomization, more patients are assigned to more effective arm(s) until high PP for concluding superiority of equivalence is reached. Design parameters including prior, criteria for declaring superiority/equivalence, threshold for early stopping, and maximum sample size can be chosen to control both type I and type II error rates. PP can also be applied to planned or unplanned interim analysis. Operating characteristics for PP, posterior probability approach and frequentist designs are compared and contrasted.