Impact of Mis-Specified Prior in Combination Dose Finding

• Activity Number: 620 - Dose-Finding for Monotherapy and Combination Therapy in Oncology and Other Complex Studies
• Type: Contributed
• Date/Time: Thursday, August 3, 2017 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Lixia Pei*
• Companies: Janssen Research & Development
• Keywords: dose finding ; oncology ; combination therapy ; prior ; bayesian ; logistic

Abstract

In Oncology, combination therapies are often used to treat cancer patients and the dose finding includes two steps. First is to conduct a monotherapy dose finding trial. As a new molecular entity (NME), this step is often required to understand the safety and efficacy property of the NME. In this monotherapy trial, instead of identifying one single recommended phase 2 dose level (RP2D) for the NME, now trials are designed to identify a range of active dose levels for further combination dose finding. Second is to conduct a combination dose finding trial. Bayesian logistic regression model (BLRM) could be used to incorporate the monotherapy dose finding information into the combination dose finding trials. This paper will present the operation characteristics of BLRM method in combination dose finding. How to appropriately set the priors for the Bayesian parameters? How much weight should the prior monotherapy trial information be used in the combination dose finding study? What is the impact of a mis-specified prior?