Predicting Trajectories of Ambulatory Function in Duchenne Muscular Dystrophy (DMD)


Translational Research

Poster Number: Virtual


Francesco Muntoni, MD, Great Ormond Street Institute of Child Health, London, UK, James Signorovitch, PhD, Analysis Group, Inc., Boston, MA, USA, Nathalie Goemans, MD, PhD, University Hospitals Leuven, Leuven, Belgium, Adnan Manzur, FRCPCH, Great Ormond Street Institute of Child Health, London, UK, Nicolae Done, PhD, Analysis Group, Inc., Boston, MA, USA, Gautam Sajeev, ScD, Analysis Group, Inc., Boston, MA, USA, Susan J. Ward, PhD, The Collaborative Trajectory Analysis Project, Cambridge, MA, Craig McDonald, MD, UC Davis Health

The North Star Ambulatory Assessment (NSAA) is a validated assessment that serves as an endpoint in multiple DMD clinical trials. Accurately predicting NSAA trajectories is important for understanding effects of novel therapies, via creation of individualized controls for treated patients, especially over longer-term (> 18 month) follow-up for which placebo controls are infeasible. We used longitudinal NSAA data for boys with DMD aged 4-10 years and starting with NSAA >12 at baseline to develop prognostic models for trajectories of ambulatory function in DMD for up to 4 years of follow-up. Data were drawn from four natural history databases: UZ Leuven, PRO-DMD-01 (provided by CureDuchenne), the North Star Clinical Network, and iMDEX. NSAA total score trajectories were fit using mixed effects models with age, steroid use, height, weight, and ambulatory function as baseline predictors. Predictive accuracy was evaluated in a held-out sample. Among N=261 subjects, mean age at baseline was 6.8 years, mean NSAA score was 25.6; 53% were receiving prednisone, 23% deflazacort, and 24% no steroid. Almost all initiated steroids within the first year of follow-up. The average patient had 6 post-baseline NSAA assessments over 2-4 years of follow-up. Important predictors of greater decline in NSAA identified in the training sample (N=208) included older age, longer rise-from-floor times, higher NSAA (more room to decline), greater height, and greater body mass index. The best-fitting model explained 27% of variation in post-baseline NSAA with average prediction errors of ±5.6 units (root mean squared error) in the held-out sample (N=53). An increasing likelihood of missing NSAA data was evident as boys approached worse function. In conclusion, trajectories of ambulatory function in DMD can be well-predicted using baseline characteristics. However, missing NSAA assessments may bias predictions, most likely towards better-than-actual outcomes. Additional validation in larger samples and assessments of missing data are warranted.