Developing a clinical trial optimization tool for DMD using muscle MR biomarkers


Real World Data - Disease registries, natural history, post marketing surveillance

Poster Number: 165


Rebecca Willcocks, PhD, Sarah Kim, PhD, Stephan Schmidt, PhD, F.C.P., Glenn Walter, PhD, William Triplett, Sean Forbes, PhD, Dah-Jyuu Wang, PhD, William Rooney, MD, Krista Vandenborne, PT, PhD, Michael Daniels


1. University of Florida, 2. University of Florida, 3. University of Florida, 4. University of Florida/ Physiology and Functional Genomics, 5. University of Florida, 6. University of Florida/ Physical Therapy, 7. Children's Hospital of Philadelphia, 8. Oregon Health and Science University, 9. University of Florida, 10.

Background: Clinical trials in DMD face challenges in selecting optimal endpoints and study cohorts to detect therapeutic efficacy. Both mathematical modelling based approaches to trial design, and noninvasive magnetic resonance biomarkers have shown promise in this area. Combining these approaches to develop a clinical trial optimization tool utilizing magnetic resonance biomarkers could allow broader inclusion criteria without compromising statistical power and accelerate therapeutic development.
Objectives: The objective of this study is to describe the disease trajectory of the ImagingDMD study natural history data set using mathematical modelling.
Approach: ImagingDMD is a longitudinal natural history study which includes 1-9 year followup data in 175 boys with DMD. ImagingDMD participants visit one of 3 study sites annually for magnetic resonance imaging and spectroscopy measurements, as well as the collection of strength and functional data. Vastus lateralis (VL) muscle fat fraction data from 146 of these individuals were used to test our ability to quantify disease progression using a mathematical function. For a model validation purpose, the individuals were divided randomly into training and test datasets in a 4:1 ratio.
Results: Magnetic resonance imaging and spectroscopy measures of muscle fat infiltration significantly increased over time in ImagingDMD participants. The base model structure that best captured the course of the VL fat fraction score over time was a sigmoid Emax model. The model provides clinical interpretability, given the nature of the parameters describing the maximum increase in VL fat fraction score and the age at which the score is half of its maximum increase.
Conclusions: The longitudinal trajectory of VL fat fraction in DMD was well captured by a mathematical model, providing the opportunity to evaluate the effect of therapeutic intervention against the model-predicted disease trajectory and possibly broaden inclusion criteria for clinical trials in DMD.