Statistical methods for enriched clinical trials with applications to Alzheimer's disease research

PROJECT SUMMARY / ABSTRACT With the rising prevalence of Alzheimer’s disease (AD) in the U.S. and worldwide, there is a crucial need for preventative and disease-modifying treatments. Randomized controlled clinical trials (RCTs) serve as the gold standard to determine whether a candidate treatment has a favorable benefit-to-risk ratio for a pre-specified

target patient population. However, heterogeneity of treatment effects across subpopulations (e.g., due to health disparities) may yield medical interventions that are not one-size-fits-all. Enrichment strategies are commonly employed in RCTs to identify the target populations most likely to benefit from a candidate treatment

and/or have the outcome of interest during the course of the trial. Enrichment in AD RCTs aligns with the National Plan to Address AD Strategy 1.B to expand research to develop disease-modifying treatments. Currently, there is a gap in the understanding of RCTs using enrichment and adaptations to the randomized

treatment assignment allocations (response-adaptive enrichment), especially for RCTs with a repeated measures (longitudinal, e.g., changes in activities of daily living scores) or censored (time-to-event, e.g., time to dementia) primary outcome. Application of standard statistical methods to enrichment designs may,

however, result in bias (tendency to systematically over- or under- estimate treatment effects). Biased estimates can lead to approval of less effective therapies, in the best case, and approval of potentially harmful or ineffective therapies or missing an effective therapy, in the worst case, as a consequence of over- or under-

estimating treatment effects. Our conjecture is that the bias induced in a fixed enrichment pre-post (only two assessments; one pre- and one post-randomization) RCT will be exacerbated when using response-adaptive enrichment in longitudinal or time-to-event RCTs. The applicant’s long-term objective as a collaborator on

RCTs and independent researcher is to provide well-calibrated and valid statistical inference for complex innovative designs to facilitate drug development in AD and other diseases. This F31 proposal aims to quantify the impact of enrichment (e.g., on bias), and as needed, develop novel statistical methods to obtain valid

inference in enriched RCTs with a longitudinal primary outcome (Aim 1) and a time-to-event primary outcome (Aim 2). Simulation studies using data from completed, large phase 3 NIA- and industry-sponsored mild cognitive impairment and AD trials will be used to empirically validate the newly developed theory and methods

in real-world settings. To provide resources for trialists, freely-available and user-friendly software based on Aims 1-2 will be developed (Aim 3) as an extension to the existing RCTdesign (www.rctdesign.org) R package, co-authored by the sponsor of this application. Research findings from Aims 1-2 will be disseminated via

conference presentations and peer-reviewed publications. Successful completion of Aims 1-3 will provide a framework and tools for trialists to be well-informed when designing enriched RCTs for any disease.