PT-002 - OPTIMIZING CLINICAL TRIALS FOR DUCHENNE MUSCULAR DYSTROPHY USING MULTIVARIATE DISEASE PROGRESSION MODELS THAT BRIDGE A FUNCTIONAL ENDPOINT AND MRI RELAXOGRAPHY OF FIVE LEG MUSCLES.

D. Yoon¹, M. Daniels¹, J. Morales¹, R. Willcocks¹, W. Triplett¹, A. Barnard¹, S. Forbes¹, G. Walter¹, W. Rooney², K. Vandenborne¹, S. Kim¹; ¹University of Florida, Gainesville, FL, USA, ²Oregon Health & Science University, Portland, OR, USA.

Background: To inform the use of muscle imaging biomarkers in clinical trials for Duchenne muscular dystrophy (DMD), we developed five multivariate disease progression models that quantify the longitudinal associations of a widely used functional endpoint, 6-minute walk distance (6MWD), and transverse relaxation time constant (T2) for each of five leg muscles: biceps femoris long head (BFLH), gracilis (GRA), medial gastrocnemius (MG), soleus (SOL) and vastus lateralis (VL).

Methods: The data from 111 individuals enrolled in the multi-center natural history ImagingDMD study were used (ClinicalTrials.gov: NCT01484678). After separately modeling the longitudinal trajectory of each measure as a function of age at each assessment, the selected univariate models were combined using correlation random effects. Nonlinear mixed effects modeling was performed in Monolix (2021R1), and the full model approach was applied for covariate analysis.

Results: The sigmoid Imax and Emax models best captured the longitudinal trajectories of 6MWT and T2, respectively. In all multivariate models, random effects were included to correlate IC50 and EC50 (age at which the measure is half of its maximum change) and S06MWD (extrapolated 6MWD when time is 0). Baseline 6MWD, baseline T2 and steroid use were significant covariates. The estimates of EC50 varied across muscles, reflecting different rates of muscle replacement with fatty/fibrous infiltrate (BFLH < VL < MG < GRA < SOL).

Conclusion: The five final multivariate models connecting 6MWD and T2 successfully quantified how individual characteristics alter disease trajectories and predicted observed values with reasonable precision and accuracy. The developed models will guide drug developers in using the MRI T2 biomarkers most efficiently in clinical trials.