Bayesian monitoring of a phase 2 chemoprevention trial in high-risk cohorts for prostate cancer

Abstract

The objective of phase 2 cancer chemoprevention trials is to evaluate whether a chemopreventive agent will cause significant modulation of intermediate endpoint biomarkers (IEB) in patients at high risk for the disease. A phase 2 chemoprevention trial of 4-hydroxyphenyl retinamide (4-HPR) versus placebo was conducted in men with a histologic diagnosis of early prostate cancer and scheduled to have radical prostatectomy. A Bayesian monitoring method was used to sequentially monitor this trial for evidence of biological activity or ineffectiveness based on a single IEB variable. Different prior distributions were used and posterior distributions were obtained to calculate the probability that treatment differences are greater than or less than a predetermined clinically significant effect. The interim analysis of transforming growth factor–α expression indicated a high probability of insufficient biological activity of 4-HPR on this IEB. This study demonstrates the potential utility of Bayesian methods in the decision-making process in the conduct of phase 2 chemoprevention trials.