All Times EDT
Keywords: Bayesian Hierarchical Model, non-inferiority, pediatric study
This presentation will discuss some case studies that Bayesian methods have been applied in regulatory decisions. How to obtain accurate subgroup treatment effect is always of high interest, especially when we need to present them in the drug trial snapshot for new molecular entity application. Using data from each individual subgroup separately may result in random high and random low subgroup treatment effect. In the first case study, Bayesian hierarchical modeling (BHM) is used to estimate more accurate subgroup treatment effect by borrowing data from other subgroups. Another question of interest in drug development is how to derive a reliable and feasible non-inferiority margin (NIM) in an active-control study. When the number of available studies is few for the active control, borrow information from studies for different drugs in the same class may be a good option. In the second case study, BHM is used to derive NIM for a new cardiovascular outcome trial (CVOT) using an active control. After considering uncertainty, the NIM based on the single completed study for the active control is too small to make the new study infeasible. By using BHM to borrow information from other drugs in the same class, the NIM derived is more reasonable than the margin using the single study for active control and it is also more drug-specific compared to the sponsor’s proposed margin from a random-effect meta-analysis. Due to limited number of subjects and other ethnic consideration, it is always challenging to designing and approving drug applications for pediatric population. In the third case study, we will discuss how to use Bayesian method to borrow patient information from historical controls to reduce the number of subjects to be randomized to the control arm in a new pediatric study. We will also discuss how we extrapolate the available adults’ information in supporting a regulatory approval for a pediatric indication.