Analytical Framework for Assessment of Long-Term Efficacy of Therapies for Duchenne Muscular Dystrophy Using External Controls

Topic:

Clinical Trials

Poster Number: P86

Author(s):

Eugenio Mercuri, MD, Catholic University and Nemo Pediatrico, Fondazione Policlinico Gemelli IRCCS, Rome, Italy, Damon Asher, PhD, Sarepta Therapeutics Inc., Cambridge, MA, USA, Kai Ding, PhD, Sarepta Therapeutics, Inc., Cambridge, MA, USA, Matthew Furgerson, PhD, Sarepta Therapeutics Inc., Cambridge, MA, USA, Jianhua Jin, Sarepta Therapeutics Inc., Cambridge, MA, USA, Jing Li, Sarepta Therapeutics Inc., Cambridge, MA, USA, Xiao Ni, PhD, Sarepta Therapeutics Inc., Cambridge, MA, USA, James Signorovitch, PhD, Analysis Group, Inc., Linda Lowes, PhD, Sarepta Therapeutics, Jacob S Elkins, MD, Sarepta Therapeutics, Inc., Cambridge, MA, USA

Assessing the long-term efficacy of therapies for Duchenne muscular dystrophy (DMD) presents significant ethical and methodological challenges. Although randomized, placebo-controlled trials remain the gold standard for short-term studies, prolonged placebo use in patients with DMD is fundamentally unethical. Therefore, to evaluate and benchmark long-term treatment effects, researchers must rely on external controls (ECs) from natural history and real-world observational studies.

As patients with DMD age, their functional decline accelerates. This is especially pronounced at the late ambulatory stage, typically between 8 and 12 years of age. Loss of function limits the ability to administer certain tests, causing endpoints such as the North Star Ambulatory Assessment (NSAA) and timed function tests (TFTs) to become increasingly susceptible to the floor effect (ie, inability to capture the full treatment effect because of the aggregation of data around the value indicating lowest performance). Notably, even utilizing velocity measurements for TFTs fails to circumvent the floor effect, while the inherent non-normal distribution of measurements affected by the floor effect introduces further statistical complexity.

Here, we outline an analytical framework for comparing clinical trial data with EC data, including the selection of EC data sources, assessment of potential selection bias through the comparison with independent data sources, imputation of “missing” data attributable to loss of function, matching or adjustment based on baseline prognostic factors, and the analysis of NSAA and TFTs by means of mixed-effects models for repeated measures.

In addition, we critically examine the limitations of analyzing functional endpoints as continuous variables at the late ambulatory stage of DMD, such as the floor effect resulting from progressive functional loss, and explore alternative approaches to mitigate these constraints. Finally, we outline the application of nonparametric approaches when comparing clinical trial data with ECs and statistical inference strategies for managing non-normal data distribution. These methods provide an approach to predict the course of DMD for a given patient population and assess long-term treatment effects.