Abstracts
Bayesian Statistics and FDA Regulatory Acceptability, Greg Campbell, PhD, Former Director of Biostatistics, U.S. Food and Drug Administration
In the United States Bayesian statistics has been used in regulatory submissions to the Food and Drug Administration (FDA) for confirmatory clinical trials medical devices for more than fifteen years. The Bayesian history and accomplishments for medical devices will be reviewed. Attention is then turned to the status and opportunities of Bayesian statistics for pharmaceutical drugs and biologicals. There are harbingers of change in the wind and these will be reviewed. Finally the challenges and the future of Bayesian statistics in the regulatory environment will be tackled.
Bayesian Dose-finding Designs for Modern Drug Development, Yuan Ji, Professor of Biostatistics, Department of Public Health Sciences, The University of Chicago
In this talk, I will introduce a variety and representative Bayesian designs for dose-finding trials. The topics to be covered include classical DLT-based dose-finding designs, designs with delayed toxicity using time-to-event endpoints, and designs for combination dose-finding trial. The key takeaways will be the illustration of Bayesian modeling and inference for dose-finding designs that utilize the concept of probability intervals and decision making. Examples and software packages will be provided to illustrate various methodologies.
Bayesian Dynamic Borrowing for Partial Extrapolation and Bridging Studies: Methods and Case Studies, Nicky Best, Head, Advanced Biostatistics and Data Analytics Centre of Excellence, GSK
In settings such as paediatric drug development or multi-regional clinical trials & bridging studies, a substantial body of evidence typically already exists regarding the drug efficacy and safety in other populations (e.g. adults or other regions). Bayesian dynamic borrowing methods offer a scientifically rigorous way to formally leverage this existing knowledge to better inform drug development and regulatory decision-making in these settings. In this webinar, I will introduce the key concepts and methodological details of Bayesian dynamic borrowing, focusing on the robust mixture prior method introduced by Schmidli et al (2014). Case studies will be presented to illustrate the application of this approach to real examples in a regulatory setting, and the benefits and challenges will be discussed.
Key Takeaways:
- Bayesian dynamic borrowing methods offer a scientifically rigorous way to formally leverage existing knowledge to better inform drug development and regulatory decision-making.
- Bayesian dynamic borrowing has been successfully used as a post-hoc analysis to support regulatory approval of a paediatric indication expansion for an approved adult treatment for systemic lupus erythematosus.
- Use of Bayesian borrowing designs as a pre-specified primary analysis requires clear communication and justification for the choice of prior and the study decision criteria, and consideration of a broad range of operating characteristics in addition to traditional frequentist control of type 1 error, in order to evaluate the benefits and risks of the proposed design.
Statistical Design and Conduct of Platform Trials, Jason Connor, President & Lead Statistical Scientist, ConfluenceStat
The talk will focus on the statistical design and conduct of Platform Trials. These are large trials, usually designing under a Master Protocol, that focus on a disease rather than on a therapy. The trial may study multiple drugs or devices simultaneously with the intention of being a perpetual design to serially test therapeutics for a specific disease area.
Bayesian Models for Precision Oncology Clinical Trials, Peter Mueller, Professor, Department of Mathematics and the Department of Statistics & Data Science, The University of Texas at Austin
We propose an adaptive Bayesian clinical trial design for patient allocation and subpopulation identification. We start with a decision
theoretic approach, including a utility function and a probability model across all possible subpopulation models. The main features of the proposed design and population finding methods are the use of a
flexible non-parametric Bayesian survival regression based on a random
covariate-dependent partition of patients, and decisions based on a
flexible utility function that reflects the requirement of the clinicians appropriately and realistically, and the adaptive allocation of patients to their superior treatments.
Recent Development on Bayesian Clinical Trial Designs Using Historical Data, Ming-Hui Chen, Professor and Head of the Department of Statistics, The University of Connecticut
This webinar starts with a brief overview of variations of the power prior for borrowing historical information and the recent development of the Bayesian methodology for the design of clinical trials. A general Bayesian methodology within the Bayesian decision rule framework for the design of clinical trials with a focus on controlling type I error and power will be introduced. Various special cases, including the posterior probability approach, the Bayesian factor approach, and the conditional borrowing approach, are discussed. In addition, various fitting priors such as power priors and hierarchical priors are constructed for the incorporation of historical data. Various properties of the Bayesian methods are examined and simulation-based computational algorithms are discussed. The Bayesian methods are then applied to the design of a non-inferiority medical device clinical trial with historical data from previous trials to demonstrate superiority of the Bayesian methods in sample size reduction.
Key takeaways:
- What is the power prior and how to determine the amount of historical data being borrowed?
- What is Bayesian sample size determination?
- How to control Bayesian type I error and increase Bayesian power?
- How to reduce sample size?
Leveraging external evidence in medical device decision-making. Dr. Ram C. Tiwari, Ph.D. Head of Statistical Methodology, Bristol Myers Squibb
The passage of the 21ST Century Cures Act has led to the increased use of external data in the planning and regulatory submission of medical device trials. Such use of external data, sometimes called real world evidence, requires the implementation of Bayesian techniques for the construction of single arm trials, comparator arms and other innovative designs. They can also strengthen the evidence in support of a clinical trial by demonstrating the relevance of data from other completed clinical trials.
The Medical Device Innovation Consortium External Evidence Methods (MDIC EEM) Framework is intended to help stakeholders navigate the use of such methods. Ram Tiwari will briefly present this Framework, and then demonstrate the use of propensity scoring methods for the integration of external data in a clinical trial for medical devices.