PS2C - New Statistical Approaches for Leveraging Auxiliary and Real-World Data in Rare Disease Research

Room: Salon E 

It is estimated that more than 30 million people in the U.S. are impacted by acknowledged rare diseases, which now number over 7000. Sadly, the development of clinical trials to study such diseases has been hampered by the inherently small patient populations available for study, as well as poor understanding of the natural history of such diseases. Since August 2018, the use of Bayesian and other nontraditional approaches in this area has been fostered by the FDA's Complex Innovative Trial Design (CID) Pilot Meeting Program. More recently, in May 2022, the FDA Center for Drug Evaluation and Research (CDER) launched its Accelerating Rare disease Cures (ARC) Program, which seeks to speed and increase the development of effective and safe treatment options addressing the unmet needs of patients with rare diseases. While Bayesian and other novel statistical methods have previously been suggested in rare disease research, programs like CID and ARC offer new alternate regulatory pathways to marketing approval. Moreover, since the actions of regulators in Europe and elsewhere around the world are often influenced by developments at FDA, this is an important practical example of statistical thinking and innovation having a global impact.

This "two-speaker-plus-small-panel" session will discuss the use of novel methods in rare disease drug development, highlighting recent methodological advances. First, we begin with an overview the FDA ARC program and some of the new methodologic opportunities it affords, especially related to reducing the need for sample size calculations, incorporating highly sequential trials, and formal analysis for multiple and compound endpoints. A secondary (and particularly novel) goal of the program is to encourage the use of statistical methods that are in line with formal decision making, rather than allowing slavish devotion to controlling the Type I error probability to dominate all other inferential concerns and goals. Next, we will hear about several key technical issues in implementing modern Bayesian adaptive trials, especially regarding the many approaches to adaptively borrow strength from historical and other auxiliary data. Recent developments include the elastic prior, which uses a margin corresponding to a practically significant difference in the treatment effect to help scope the degree of borrowing. An even more recent proposal is the self-adapting mixture (SAM) prior, a form of the robust mixture prior in which the mixture weight is data-driven and self-adapting. Both approaches estimate the similarity of the historical and concurrent data (as power and commensurate priors do), but in more flexible ways that can be informed by discussions with regulatory authorities.

Finally, we will invite the panel to extend our discussion beyond clinical data to the use of real world evidence (RWE), an increasingly important area in the small-sample size world of rare disease. This discussion will include related recent developments in small n, sequential multiple assignment randomized trial (snSMART) design and analysis, as well as the use of historical or synthetic controls as comparators in single arm studies. In both cases, causal inference tools (say, propensity scores or instrumental variables) can be used to adjust for the bias resulting from the lack of randomization. We close with a brief discussion of what can and cannot be achieved regulatorily with RWE, broadening our focus on its use in clinical studies and their execution to applications in early discovery, marketing application, product launch, and life cycle management.

Bayesian adaptive clinical trial

causal inference

covariate adjustment

complex innovative design (CID)

real world evidence (RWE) 

Pediatric, Small Population, Rare Disease

ASA Biopharmaceutical Section Regulatory-Industry Statistics Workshop 2023

Presenting Author

Dionne Price, Food and Drug Administration

Presenting Author

Ying Yuan, University of Texas, MD Anderson Cancer Center

CoAuthor(s)

Lei Nie, The US FDA
Jonathon Vallejo

Presenting Author

Kelley Kidwell, University of Michigan

Presenting Author

Matthew Psioda, GSK