Held on October 15, 2009 at the Marriot Renaissance Chicago North Shore Hotel in Northbrook, IL.
The Program will consist of two presentations:
- Adaptive Design Considerations for Evaluating Cardiovascular Risk among Patients with Type-2 Diabetes Craig Wilson, Ph.D., Takeda Pharmaceutical
- Bayesian Adaptive Randomization Design - A Case Study and More Yili Pritchett, PhD., & Shu Han, PhD., Abbott Laboratories
Adaptive Design Considerations for Evaluating cardiovascular Risk among Patients with Type-2 Diabetes by Craig Wilson, PhD., Takeda Pharmaceutical
Biographical Background Dr. Craig Wilson is a Principal Statistician at Takeda Global R&D in Lake Forest , Illinois . Since receiving his PhD from Oklahoma State University in 1998, Craig has served as study statistician and lead statistician on a variety of drugs for treatment of T1DM and T2DM. During his time in industry, Craig has had extensive discussions with the FDA regarding the design and analysis of diabetes trials, including trial requirements for satisfying the CV guidance.”
Abstract In December, 2008, the FDA released a final Guidance for Industry for evaluating CV risk in subjects with T2DM. This guidance established criteria for assessing the risk ratio of an investigational drug relative to control in premarketing applications. In particular, if sufficient data are available to rule out a risk ratio of 1.8, then an investigational drug may be approved with a postmarketing commitment; if a risk ratio of 1.3 may also be ruled out, then a postmarketing requirement to assess CV risk may not be required. For sponsors with investigational drugs for treatment of T2DM currently under development, one approach to satisfy this guidance is to design a single stand-alone CV trial which may sequentially rule out risk ratios of 1.8 and 1.3. This presentation will focus on design considerations for such a trial including adaptive design issues. Discussion of the pros/cons of Bayesian vs. group sequential design will be provided.
Bayesian Adaptive Randomization Design - A Case Study and More by Yili Pritchett, PhD. and Shu Han, PhD., Abbott laboratories
Biographical Background Dr. Shu Han is currently a research statistician at Abbott Laboratories, where he has played active role in the designs and implementations of multiple clinical trials using Bayesian adaptive designs. Prior to joining Abbott in 2006, Shu worked for Guidant/Boston Scientific Corporation, where he collaborated with the FDA to design an adaptive seamless exploratory-confirmatory clinical trial evaluating heart failure diagnostic medical devices. Shu also worked for the Quantitative Science Division of M.D. Anderson Cancer Center as a Research Assistant to Dr. Donald A. Berry from 2003 to 2005. Shu received his Ph.D. in statistics from a joint doctoral program at M.D. Anderson Cancer Center and Rice University, after receiving his Master's Degree in statistics from Columbia University. He is currently pursuing a MBA degree at the University of Chicago.
Yili L. Pritchett is a Research Fellow and a Director of Clinical Statistics in Global Pharmaceutical R&D at Abbott Laboratories. She is responsible for statistical aspects of Phase II – IV drug development in Neuroscience, Pain Care, and Renal Care therapeutic areas. Before joining Abbott in April 2006, Dr. Pritchett was a Research Advisor at Eli Lilly and Company where she provided statistical leadership at various levels for the development and approval of several brands. Dr. Pritchett obtained her Ph.D. in Statistics from the University of Wisconsin–Madison in 1994. Dr. Pritchett is an active member of PhRMA Adaptive Trial Design Working Group. She has championed the use of adaptive design in Abbott, and led the efforts of delivering a number of protocols with different types of adaptive design. Dr. Pritchett authored or co-authored 46 peer-reviewed manuscripts or book chapters, and made over 100 presentations at statistical or medical conferences.
Abstract In recent years, Bayesian adaptive randomization design has been increasingly applied to clinical trails. In this presentation, we will use a real case study to go through three major mathematical components of the design: the modeling of dose-response relationship, the algorithm of computing updated randomization ratio, and the longitudinal model that predicts endpoint using partial observations. In addition, the decision rules that allow the study to stop early due to efficacy or futility will be explained, and the operating characteristics of the design and the results of sensitivity analyses for the key parameters will be illustrated. Simulation procedures via MCMC (Markov Chain Monte Carlo) method will also be described. Lastly, other real clinical trial cases where Bayesian adaptive designs were used to gain efficiency and effectiveness in drug and medical device developments will be shared.