PWI-003 - USE OF QUANTITATIVE MAGNETIC RESONANCE IMAGING BIOMARKERS IN CLINICAL TRIALS FOR DUCHENNE MUSCULAR DYSTROPHY: MULTIVARIATE DISEASE PROGRESSION MODELS BRIDGING TIMED MOTOR FUNCTION TESTS AND FAT FRACTION

D. Yoon¹, M. Daniels², R. Willcocks³, J. Morales⁴, W. Triplett³, R. Belfiore-Oshan⁴, G. Walter⁵, W. Rooney⁶, K. Vandenborne³, S. Kim⁷; ¹Center for Pharmacometrics & Systems Pharmacology College of Pharmacy, University of Florida, ²Department of Statistics, University of Florida, ³Department of Physical Therapy, University of Florida, ⁴Critical Path Institute, ⁵Department of Physiology and Aging, University of Florida, ⁶Advanced Imaging Research Center, Oregon Health & Science University, ⁷Center for Pharmacometrics and Systems Pharmacology, College of Pharmacy, University of Florida.

Background: This research aims to inform the use of quantitative magnetic resonance (qMR) muscle imaging biomarkers in clinical trials for Duchenne muscular dystrophy (DMD) by quantifying the longitudinal associations between widely used timed motor functions tests (i.e. velocities of 10-meter walk/run test (TMW), stand from supine test (STS) and climb 4 stair test (CFS)) and fat fractions of leg muscles (i.e. Soleus (SOL) and Vastus Lateralis (VL)) measured by magnetic resonance spectrometry.

Methods: We developed six multivariate disease progression models using natural history ImagingDMD study (NCT01484678) with 118 individuals and validated the model using the data from three clinical studies with a total of 87 individuals. After separately quantifying the longitudinal trajectory of each measure as a function of age at each assessment, the selected univariate models were combined by allowing correlations between certain individual-specific parameters. Since Bayesian inference was used, we were able to include published prior information for the timed motor function tests. The Bayesian nonlinear mixed effects modeling was conducted in Monolix (2023R1), and the full model approach was applied for covariate analysis.

Results: The sigmoid Emax function and the product of the Chapman-Richards growth and sigmoid Imax functions best captured the longitudinal dynamics of fat fractions and timed motor function tests, respectively (Table 1). The final models included correlation between TMW/CFS and SOL via the following dependent pairs of parameters: Gmax and DPmax,SOL, γTMW/CFS and DPmax,SOL, and Gmax and γTMW/CFS; while the correlation between TMW/CFS and VL was induced by the following dependent pairs of parameters Gmax and DPT50,VL, γTMW/CFS and DPT50,VL, and Gmax and γTMW/CFS. The correlations between STS and SOL/VL included DPT50,STS and DPT50,SOL/VL, and Gmax and DPT50,SOL/VL, and Gmax and DPT50,SOL/VL. Baseline scores of each baseline age group and steroid use were statistically and clinically significant covariates.

Conclusion: The final models adequately capture longitudinal disease progression of DMD. These models will guide drug developers in using the qMR imaging biomarkers most efficiently along with the widely used timed motor function tests in DMD clinical trials.