Author(s)
Goemans N1, Wong B2, Muntoni F3, McDonald C4, Mercuri E5, Manzur A3, Signorovitch J6, Sajeev G6, Wong H6, Hossain I6, Jenkins M7, Ward SJ8
1University Hospitals Leuven, Leuven, Belgium, 2University of Massachusetts Medical School, Worcester, MA, USA, 3Dubowitz Neuromuscular Centre UCL Great Ormond Street Institute of Child Health & Great Ormond Street Hospital, London, UK, 4University of California, Sacramento, CA, USA, 5Catholic University, Rome, Italy, 6Analysis Group, Inc., Boston, MA, USA, 7Analysis Group, Inc., London, UK, 8Collaborative Trajectory Analysis Project (cTAP), Cambridge, MA, USA
OBJECTIVES: Use of natural history (NH) controls in DMD drug evaluations is of high interest; however, the heterogeneity among patients or outcome assessments could potentially bias comparisons between NH and clinical trials, especially for performance-based outcomes. As a follow-up to our previous work using the 6-minute walk distance (6MWD), we aimed to assess this concern by comparing outcomes between NH data sources and clinical trial placebo arms using the NSAA, a commonly used primary or secondary outcome measure. METHODS: Placebo arm data from phase III trials of tadalafil and ataluren in DMD was used. NH data came from PRO-DMD-01 (provided by CureDuchenne) and UZ Leuven. Sensitivity analyses incorporated data from Cincinnati Children’s Hospital Medical Center (CCHMC) and the NorthStar UK Clinical Network (NSUK). Change in NSAA total score over ~48-week intervals was studied in boys aged 6-18 years with NSAA >12 at baseline. RESULTS: Multivariable regression was used to compare NSAA changes between NH and placebo arms adjusting for baseline prognostic factors. Primary analyses included 187 intervals (187 patients) from placebo arms, and 317 intervals (180 patients) from NH data. The unadjusted difference in mean ~48-week change in NSAA between NH and placebo was -1.36 (p=0.001). Adjusting for baseline characteristics (age, baseline NSAA, steroid use, timed function tests, height, weight and BMI) decreased the difference to -0.04 units (p=0.9). Results were similar in sensitivity analyses incorporating the CCHMC (unadjusted: -1.67; p <0.001, adjusted: -0.19; p=0.6) and NSUK databases (unadjusted: -1.14; p=0.004, adjusted: -0.28; p=0.5). Without adjustment for baseline prognostic factors, changes in NSAA total score differed slightly between NH and placebo. After adjustment, there were no significant differences between NH and placebo. CONCLUSIONS: These findings align with our previous research using 6MWD, and further demonstrate the potential of NH controls to augment or replace placebo arms, in DMD drug evaluations.
Conference/Value in Health Info
2019-11, ISPOR Europe 2019, Copenhagen, Denmark
Code
PRO125
Topic
Clinical Outcomes, Methodological & Statistical Research
Topic Subcategory
Confounding, Selection Bias Correction, Causal Inference, Modeling and simulation, Performance-based Outcomes
Disease
Musculoskeletal Disorders, Neurological Disorders, Rare and Orphan Diseases
