All Times EDT
Keywords: Bayesian Hierarchical Model, Power Priors, Animal Cellular Therapies, Information Borrowing
There are numerous challenges in establishing a predictable and efficient pathway to evaluate animal cellular therapies such as stem cell products. One such challenge is that cellular therapies have high inherent variability in biologic activity, which results in non-uniform safety and effectiveness profiles even under a well-controlled manufacturing process. Additionally, there is a finite number of doses in each batch produced according to a single manufacturing order during the same cycle of manufacture. Thus, new batches have to be continuously produced and evaluated for approval. While it may be unrealistic to conduct a large-scale study to confirm the effectiveness for each batch separately, it is intuitive to consider borrowing information across batches to increase study power. We investigated Bayesian Hierarchical Models and Bayesian Power Prior Models in 2 scenarios : 1) borrowing treatment effects across different batches in a single study; and 2) borrowing information from batches that have demonstrated effectiveness to evaluate the performance of new batches. In each scenario, we conducted simulations to explore the operating characteristics of the Bayesian approaches and compared them to a frequentist approach. In this talk, we will present the findings from our simulations and discuss the potential utility and challenges in using Bayesian models to evaluate data from animal cellular therapy studies.