Dose-Response Modeling in Phase II Dose-Finding Studies

• Activity Number: 688
• Type: Contributed
• Date/Time: Thursday, August 8, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Jiahui Li and Ming-Chun Hsu and Yaung-Kaung Shen*+
• Companies: Janssen R&D, LLC and Janssen R&D, LLC and Centocor
• Keywords: dose-response modeling ; dose estimation ; active comparator

Abstract

Identifying the correct dose is one of the most important and difficult goals in drug development. Multiple comparison procedure and model-based approach are two major strategies, while the statistical analysis based on the latter one has been a popular topic recently. We focus on two papers on the modeling approach. One is F. Bretz, et al. (2005, Biometrics), and the other is B. Klingenberg (2009, Statistics in Medicine). In both papers, a set of candidate models is specified, each model must be fit to the observed data, and the model that is best supported by the data from among the candidate set is identified. The selected model is then used to produce inference on adequate doses, employing the model-based approach. A thorough comparison of these two methods is made with the data from a completed phase 2/3 study. One of the methods is chosen to be applied to one of our on-going phase 2 studies. Prediction of maintenance is also carried out by the aid of the selected model. Furthermore, a Bayesian method for target dose estimation is developed with an active comparator, not merely with a placebo arm as seen in these papers.