Stopping Rules for Bayesian Dose-Finding Oncology Trials

• Activity Number: 514
• Type: Contributed
• Date/Time: Wednesday, August 4, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Yanqiong Zhang*+ and Yi He and Guillaume Marchand and Pierre Mancini and Sandrine Micallef
• Companies: Sanofi-Aventis and sanofi-aventis and sanofi-aventis and sanofi-aventis and sanofi-aventis
• Keywords: Maximum tolerated dose (MTD) ; Dose-limiting toxicity (DLT) ; Stopping rule ; Bayesian approach

Abstract

The main goal of oncology dose-finding trials is to find the maximum tolerated dose (MTD) for a new agent. MTD corresponds to the dose with an acceptable dose-limiting toxicity (DLT) level. Recently model-based Bayesian approach is increasingly being used due to its flexibility and accuracy in detecting the MTD. When to stop the trial? From a scientific and ethical standpoint, stopping rules are essential component of dose-finding methods to claim that the MTD is determined and stop accrual. In this talk, different Bayesian stopping rules based on the posterior probability distribution summaries will be explored. These rules aim at early stopping of trials with certain confidence level of MTD accuracy. Comparisons will be performed through simulations based on several underlying dose-response relationships in terms of sample size, probability of finding correct MTD and estimated bia.