PS25: Casting a Wide Net: Statistical Methods for Measuring Multiple Domains in Genetic Disease Trials

Recent advances in genetic medicines like antisense oligonucleotides, gene therapy and enzyme replacement offer transformative potential for people living with rare genetic diseases. These novel interventions directly target the primary disease mechanisms, enabling the capacity for disease modifying or even curative therapeutic benefits. However, there are many challenges when designing clinical trials to study these therapies. Of particular concern is the selection of a primary endpoint, which can be difficult for genetic diseases which are multi-symptomatic. Even monogenic diseases, which have a well-characterized genetic cause, can manifest a broad range of symptoms. Some examples include impaired cognition, lack-of-motor function, behavioral challenges, and / or seizures. Selecting a single primary endpoint for these kinds of diseases can be difficult because improvements in any of these areas would be valuable to patients. The choice of endpoints is further complicated by lack of clinical precedent and a limited understanding of the natural course of the disease. Finally, even transformational therapies may show only modest improvements in any single measure due to disease heterogeneity, potentially leading to underpowered clinical trials in a setting where recruitment is challenging and sample sizes are constrained. Rather than relying solely on a single clinical measure, clinical trials of disease-modifying interventions should consider strategies that 'cast a wide net,' utilizing endpoints and analyses that measure multiple clinical outcomes. This session will focus on recent methods for evaluating clinical benefit in multiple functional domains. The first talk will discuss methods for creating composite endpoints, like the Multiple Domain Responder index, that create a single measure of changes across multiple functional areas. The second talk provides a complementary perspective, showing how different domains can be analyzed as multiple-primary endpoints using Bayesian joint hierarchical models which naturally address the issue of multiple testing. Finally, the regulatory perspective will discuss considerations around endpoint selection in rare diseases and considerations when testing multiple endpoints. Throughout, emphasis will be placed on practical guidance for choosing and implementing these methods in trials of genetic medicines.