PS19: Breakthroughs in Data Integration for Drug Development in Rare Outcomes and Diseases

The traditional paradigm of using randomized clinical trials (RCTs) to demonstrate effectiveness of experimental treatments, has become increasingly challenging in developing novel therapeutics for rare diseases, e.g., unethical, or infeasible to conduct RCTs, slow patient recruitment and prolonged development time. Some of these challenges also apply to clinical trials in common diseases with rare outcomes. Moreover, even if an experimental treatment received regulatory approval, it could still face the questions on its long-term clinical and economic benefits to the healthcare system. Those challenges, if left unattended, would lead to delayed even rejected access to patients. In response, there is a growing interest on utilizing data beyond clinical trials, including real-world data (RWD), and innovative data Integration techniques, to generate evidence needed for stakeholders such as regulators, payors, and health care professionals. This session will highlight recent breakthroughs in data integration methods that go beyond traditional approaches e.g., propensity scores and power or meta-analytic priors. Those breakthroughs include proximal inference methods, and Bayesian data selection approach. Using case studies in HIV prevention and gene therapies in Hemophilia B, we will demonstrate how those breakthroughs could enhance our understanding of treatment effects for rare conditions in a faster and more efficient way. In addition, an expert panel-including leaders from academia, industry, and the FDA-will discuss the challenges and opportunities associated with integrating these approaches into regulatory practices, clinical development, and reimbursement decisions. The session will start with two talks (each 15 min): 1. Presentation: Using Proximal Inference to Borrow Historical Data for HIV Prevention Trials In this talk, Ting Ye (University of Washington) will introduce the concept of negative controls in the context of PDUFA VII and how it can be applied to borrow data from previous HIV prevention trials to improve current trial designs. 2. Presentation: A Bayesian Approach for Selecting Relevant External Data (BASE): Application to a Study of Long-Term Outcomes in a Hemophilia Gene Therapy Trial In this talk, Xiang Zhang (CSL Behring) will discuss a Bayesian data selection method for leveraging external data to infer long-term outcomes for a gene therapy. This innovative approach enables more informed and precise estimations of long-term treatment effects, which are essential for diseases where assessing long-term effectiveness is critical for reimbursement decisions. The session will also feature a 45-min panel discussion following the talks. Topics will include regulatory considerations on using data integration methods, the fit-for-purpose consideration of external data including RWD, the discussion on traditional and recent data integration methods, and the potential of innovative design and analysis approaches to accelerate the access of promising treatments to patients in need. The panel includes: Academia: Miguel Hernán (Harvard) and Ting Ye (University of Washington) Industry: Aloka Chakravarty (Eli Lilly) and Xiang Zhang (CSL Behring) FDA: John Scott and Hana Lee