PS08: Constructing the Counterfactual: Natural History Comparators for Single-Arm Gene Therapy Trials

The use of natural-history study (NHS) comparators has become increasingly important for gene therapy programs; in recent years more than half of FDA's gene therapy approvals have incorporated non-randomized evidence into the efficacy assessment. With more than 200 gene therapy products now in clinical development, reliance on external controls is expected to grow further, making NHS-based comparisons the norm rather than the exception. This shift underscores a pressing need for principled statistical approaches that strengthen the credibility, interpretability and regulatory acceptability of NHS-derived evidence. Key challenges include non-comparability between trial and NHS populations, ambiguity in defining a meaningful baseline in longitudinal NHS data, and quantifying uncertainty in the estimated treatment effect. As reliance on NHS-derived evidence grows, so does the need for improved analytical methods. Methodological advances, such as Bayesian disease-progression modeling and target trial emulation promise more robust estimators of treatment effect, addressing common concerns about NHS and trial comparability. This session will feature three presentations. The first speaker will discuss target trial emulation as a framework for bringing advances from causal inference and comparative effectiveness research into external-control comparisons. The second speaker will describe how Bayesian disease-progression models can be used to improve comparability of NHS and trial data by aligning subjects based on a latent, underlying disease age rather than calendar time. The final speaker will provide regulatory perspectives on the use of natural-history data as external controls in gene therapy submissions, using two case studies to highlight key regulatory considerations.