Abstract:
|
Research & clinical trials involving rare diseases are more complicated & difficult due to the small cohort of people affected. We often lack sufficient natural history of the disease, the pharmacokinetics (PK) & pharmacodynamics (PD) of the drugs we wish to test, & the ideal sample size to demonstrate a statistically significant treatment effect. Rare diseases that are fatal or primarily affect children present even greater challenges. In this talk, we look at several recent advances in hierarchical Bayesian methods that use all available information to design & analyze rare disease trials. We begin with a look at hierarchical Bayesian methods proposed for PK/PD analysis & a corresponding adaptive design for a Phase IIa clinical trial of Lorenzo's oil, an available treatment for the rare & fatal pediatric disease x-ALD. We will then look at some existing & some novel Bayesian methods as statistical frameworks for incorporating information from all sources, including older studies to strengthen our trial, & their appropriateness for extrapolating from adult to pediatric populations. We conclude by illustration our approaches using examples taken from recent pharmaceutical trials.
|