Optimizing rare neurological disease trials: Bayesian frameworks and hierarchical models for improved efficiency in...

While traditionally used in early-phase oncology trials, we explore the benefits of Bayesian methods in clinical trial design, emphasizing their application in rare neurological disease trials. Focusing in this report on a potential redesign of the Phase II trial of SAR445088 in chronic inflammatory demyelinating polyneuropathy (CIDP), we compare the performance of the original design to two potential optimization alternatives. First, we retained continuous outcomes to preserve information. In the second modification, we propose the use of information borrowing through Bayesian hierarchical models (BHM). BHM for subgroup analysis increase precision in individual subpopulations by borrowing information from the whole study population and pooling individual estimates towards the overall estimate. Such models are used in basket trials which aim to evaluate a single targeted therapy across multiple settings that share a common characteristic. We found that the proposed optimizations demonstrate notable power gains. We advocate enhancing the efficiency of neurological disorders research through the implementation of this optimized Bayesian approach.