A Conditional Bayesian Logistic Regression Model (BLRM) with Overdose Control

• Activity Number: 128
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Lu-May Chiang* and Adarsh Joshi and David Ohlssen and Jyotirmoy Dey+
• Companies: Novartis Pharmaceuticals Corporation and Eli Lilly and Company and Novartis Pharmaceuticals Corporation and Novartis Pharmaceuticals Corporation
• Keywords: Phase I trial ; dose escalation ; BLRM ; overdose control ; maximum tolerated dose ; combination therapy

Abstract

Mathematical models for the rate of dose limiting toxicities (DLT) are sometimes used in phase I Oncology trials to estimate maximum tolerated dose with single agent or combination of agents. They are also used to guide dose-escalation while controlling the chance of a study subject experiencing DLT below a pre-specified limit. In this context, we propose a "conditional" Bayesian model in a setting where an experimental agent, anti-cancer or palliative, is given to patients in combination with a fixed approved dose of standard anti-cancer therapy. The proposed model accounts for the fixed dose of standard therapy, is parsimonious compared to others used in this setting and shows similar performance in simulation studies. It also handles the setting where DLT related only to the experimental agent is collected.