Integrating Machine Learning and Magnetic Resonance Imaging Radiomics for Trajectory Modeling in Duchenne Muscular Dystrophy: Explainable Artificial Intelligence to Inform Clinical Trial Design and Personalized Intervention

2026-01-01

<p xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" dir="auto" id="d626081e99"> <b>Objectives:</b> Duchenne muscular dystrophy (DMD) is a rare pediatric condition characterized by progressive muscle degeneration. This study aimed to inform subgroup selection for clinical trial designs and personalized interventions by quantifying the individual-level rate and severity of DMD progression trajectories using machine learning (ML) models. We leveraged previously developed disease progression models [ <a class="xref-link" href="#r1">1</a>] and magnetic resonance imaging (MRI) derived radiomic features collected at screening visits. <p xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" dir="auto" id="d626081e107"> <b>Methods:</b> This study utilized a subset of data from the ImagingNMD study (NCT01484678), which included 48 individuals with DMD assessed using Dixon MRI sequences. Radiomic features were extracted using PyRadiomics [ <a class="xref-link" href="#r2">2</a>] from segmented soleus in 3 axial slices around the landmark slice, generating 1,130 features per individual, including shape-based, first-order, and texture-based characteristics. Baseline functional test assessments and demographics were also incorporated as inputs. Feature pre-selection was performed using Pearson correlation to remove collinear covariates, followed by the LASSO regularization to remove coefficients below 0.03. Multiple ML models were then trained and compared with 5-fold grid search cross-validation to evaluate their predictive performance for individual parameters describing the increase in fat fraction in the soleus as DMD progresses: DPT50 (age at which the measure is half of its maximum increase), DPmax (extent), γ (rate), and S0 (extrapolated measure when age is 0). Key radiomic biomarkers were identified through impurity-based feature importance and post-statistical analysis. Unsupervised clustering of key radiomic biomarkers was further applied to derive distinct subgroups. <p xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" dir="auto" id="d626081e115"> <b>Results:</b> Fifty-nine baseline features (38 texture-based, 14 first-order, 4 functional assessments, BMI, age, and steroid usage) were pre-selected to train ML models. The tree-based XGBoost regression model demonstrated the best performance in predicting individualized DMD progression parameters. In addition, these evaluations fall within the typical 6-month to 1-year interval of clinical screening for individuals with DMD, with a mean absolute error of ±0.64 years for DPT50 (R²=0.79), emphasizing the model’s relevance to standard monitoring practices. Feature importance analysis revealed potential radiomic biomarkers including texture-based nonuniformity, emphasis, gray-level correlation, as well as first-order mean and interquartile range. Unsupervised clustering identified distinct subgroups with mean values of DPT50=11.66 and 15.01 years and γ=8.196 and 7.857 in fast (n=17) and slow (n=30) disease progression, respectively. <p xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" dir="auto" id="d626081e120"> <b>Conclusions:</b> These findings highlight the potential of ML approaches with baseline MRI-derived radiomic features to accurately predict disease progression trajectories in individuals with DMD. The AI-assisted feature importance analysis effectively identified potential imaging biomarkers, based on their strong association with disease progression. Clustering analysis revealed clinically meaningful subgroups with interpretable decision criteria that may inform clinical trial designs and personalized intervention strategies.

Optimizing Duchenne Muscular Dystrophy Clinical Trial Design by Modeling and Simulation: Identifying Characteristics of Magnetic Resonance Imaging Fat Fraction and Timed Function Test Measures and Their Longitudinal Association

2026-01-01

<p xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" dir="auto" id="d612609e129"> <b>Objectives:</b> This study aims to support the clinical usage of magnetic resonance (MR) imaging biomarkers (i.e., fat fraction (FF)) for Duchenne muscular dystrophy (DMD) clinical trials [ <a class="xref-link" href="#r1">1</a>][ <a class="xref-link" href="#r2">2</a>] by quantitatively analyzing the longitudinal relationship between commonly used timed function tests (TFTs) (i.e., 10-meter walk run (TMW), supine to stand (STS), climb 4 stairs (CFS)) and FF measures from two leg muscles (i.e., soleus (SOL) and vastus lateralis (VL)). Additionally, the research aims to identify characteristics of imaging and functional measures, including the sensitive age range and relationships with clinically relevant covariates. <p xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" dir="auto" id="d612609e140"> <b>Methods:</b> Six multivariate disease progression models linking the 3 TFT velocity endpoints and 2 FF measures were developed using natural history data of ImagingNMD study (NCT01484678). The models were validated with placebo data of 3 clinical trials. [ <a class="xref-link" href="#r3">3</a>] Distributions of the simulated model parameters of the entire data population were compared across measures. To intuitively explore the combined effects of covariates, 16 virtual individual profiles were generated by pairing each profile with a counterpart differing in a single covariate. Each individual was simulated 500 times and the median and 40th–60th percentile bands of TFT velocity and FF disease progression trajectories were compared. <p xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" dir="auto" id="d612609e148"> <b>Results:</b> Multiplication of Chapman-Richards growth and Imax function and sigmoid Emax function provided the best fit as a structural model for TFT velocity and FF measures, respectively. DPmax_FF,i (maximum change in FF) was positively correlated with γ_TFT,i (steepness of the velocity curve) and negatively with Gmax_TFT,i (maximum possible velocity) in TMW and CFS velocity models linked to FFSOL (0.36/0.35 and -0.31/-0.52), but not in STS velocity. DPT50_FF,i (age at which the change is half of its maximum) showed positive correlations with Gmax_TFT,i for all 3 models linked to FFVL (0.40, 0.47, 0.57) and negative with γ_TFT,i (-0.18 and -0.4) except for STS velocity. STS velocity models linked to both FFSOL and FFVL showed strong positive correlations between DPT50_TFT,i and DPT50_FF,i (0.79 and 0.82). The peaks of DPT50,i distribution from the population-based simulations were 12, 10 and 11.1 (years) for TMW, STS and CFS velocity, and 12.5, 10.6 (years) for FFSOL and FFVL, respectively. Baseline TFT velocity was the most influential covariate for TMW velocity progression, masking other covariate effects. Meanwhile, steroid use had the strongest effect for STS and CFS velocity progression, followed by baseline TFT velocity, and baseline age. In altering FF trajectories, baseline FF was the key covariate for both, followed by steroid use in SOL and baseline age in VL. <p xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" dir="auto" id="d612609e153"> <b>Conclusions:</b> The models highlighted FF measures as reliable biomarkers, with significant correlations to the TFT velocity measures. Model simulation results demonstrated that the models capture the sensitive age range of each measure and identify key covariates that alter disease progression trajectories. Accordingly, the multivariate models have potential to serve as informative quantitative tools for guiding biomarker selection and the choice of inclusion/exclusion criteria, further optimizing DMD clinical trial design.

A New Algorithm for Sampling Parameters in a Structured Correlation Matrix With Application to Estimating Optimal Combinations of Muscles to Quantify Progression in Duchenne Muscular Dystrophy

Statistics in Medicine · 2025-09-01

The goal of this paper is to estimate an optimal combination of biomarkers for individuals with Duchenne muscular dystrophy (DMD), which provides the most sensitive combinations of biomarkers to assess disease progression (in this case, optimal with respect to standardized response mean (SRM) for 4 muscle biomarkers). The biomarker data is incomplete (missing and irregular) multivariate longitudinal data. We propose a normal model with structured covariance designed for our setting. To sample from the posterior distribution of parameters, we develop a Markov Chain Monte Carlo (MCMC) algorithm to address the positive definiteness constraint on the structured correlation matrix. In particular, we propose a novel approach to compute the support of the parameters in the structured correlation matrix; we modify the approach from [1] on the set of the largest possible submatrices of the correlation matrix, where the correlation parameter is a unique element. For each posterior sample, we compute the optimal weights of our construct. We conduct data analysis and simulation studies to evaluate the algorithm and the frequentist properties of the posteriors of correlations and weights. We found that the lower extremities are the most responsive muscles at the early and late ambulatory disease stages, and the biceps brachii is the most responsive at the nonambulatory disease stage.

275PMagnetic resonance imaging based radiomic feature extraction for bone quality assessment in corticosteroid treated boys of Duchenne muscular dystrophy

Neuromuscular Disorders · 2025-09-01

Quantitative Magnetic Resonance Imaging of the Forearm in Myotonic Dystrophy Type 1

Tomography · 2025-12-05

Introduction: Myotonic dystrophy type 1 is the most prevalent muscular dystrophy in adults, characterized by weakness, impaired functional abilities, and myotonia. However, little is known about the relationship between quantitative MRI measures (fat fraction and T2 relaxation time) and clinical findings of the upper extremity. This study assessed forearm muscle structure in patients with myotonic dystrophy using quantitative MRI and correlated these measures with strength, function, and handgrip myotonia. Materials and Methods: Eighteen adults with myotonic dystrophy type 1 underwent MRI using three-point Dixon and T2 spin echo imaging of the forearm. Results: The average fat fraction and T2 relaxation time were greatest in the flexor digitorum profundus (26.7% and 55.6 ms, respectively). Correlations were found between quantitative MRI values and clinical tests of strength (r = −0.61 to −0.92, p &lt; 0.01), function (r = −0.64 to −0.83, p &lt; 0.01), and handgrip myotonia (r = 0.48, p &lt; 0.05). Overall, the anterior forearm fat fraction values showed higher correlations with strength and function compared to those of the posterior forearm. Discussion: Our results support the use of quantitative MRI measures to assess forearm disease pathology and show potential to monitor the effectiveness of therapeutic treatments in patients with myotonic dystrophy type 1.

Descriptive characterization of ambulatory health states in Duchenne muscular dystrophy: Motor function trajectories and times to loss of ambulation

Journal of Neuromuscular Diseases · 2025-09-23

We described ambulatory Duchenne muscular dystrophy (DMD) progression, across multiple functional measures, via previously established prognostic groups for loss of ambulation (LoA) and health states. Patients closer to vs. farther from LoA had greater declines in some measures (e.g., 6-min walk distance) and less change in others (e.g., timed rise from floor velocity) due to floor effects. Patients in the late vs. early ambulatory health state were concordantly shifted towards higher LoA risk. Findings further characterize health states and prognostic factors in ambulatory DMD and highlight the importance of multiple measures of function to fully characterize disease progression.

678VPVastus lateralis fat fraction is associated with functional efficacy endpoints in patients with Duchenne muscular dystrophy treated with givinostat

Neuromuscular Disorders · 2025-09-01

Givinostat in Duchenne Muscular Dystrophy: Natural history comparison applying propensity score matching

Nervenheilkunde · 2025-03-01

Long‐Term Evaluation of Givinostat in Duchenne Muscular Dystrophy, and Natural History Comparisons

Annals of Clinical and Translational Neurology · 2025-08-19 · 8 citations

OBJECTIVES: This ongoing, open-label extension study is evaluating the long-term safety, tolerability, and efficacy of givinostat, a Class I and II histone deacetylase inhibitor, in patients with Duchenne muscular dystrophy (DMD). METHODS: The recruited patients completed one of two prior clinical studies (one Phase 2 and one Phase 3 [EPIDYS]), receiving givinostat or placebo, or were successfully screened but not randomized into EPIDYS. All receive givinostat oral suspension open-label at a flexible, weight-based dose in addition to systemic corticosteroids, and attend visits every 4 months. RESULTS: A total of 194 patients are included in the current analyses, with a mean duration of givinostat exposure (excluding use in prior studies) of 559.6 days (SD 373.0); when including use in the prior studies, the maximum exposure to givinostat was > 8 years. Although the majority of patients reported ≥ 1 adverse event (169/194 [87.1%]), most were mild/moderate in severity, and the safety profile of givinostat was consistent with prior studies. Post hoc comparisons with natural history datasets (ImagingDMD and CINRG) suggest, in propensity matched populations, givinostat added to systemic corticosteroids significantly delayed the loss of the ability to rise from the floor, the loss of the ability to complete the 4-stair climb test, and the loss of ambulation (by medians of 2.0-3.3 years; all nominal p < 0.05). INTERPRETATION: Overall, the safety and tolerability of long-term administration of givinostat in patients with DMD was consistent with previous studies. Comparisons with natural history data suggest that givinostat delays the occurrence of major disease progression milestones. TRIAL REGISTRATION: EudraCT number: 2017-000397-10; ClinicalTrials.gov identifier: NCT03373968.

Large-scale serum protein biomarkers discovery associated with function and clinical milestones in Duchenne muscular dystrophy

Nature Communications · 2025-10-13 · 4 citations

Duchenne muscular dystrophy (DMD) is characterized by progressive muscle wasting and weakness. Serum proteins may offer insight into disease processes and clinical decline. This observational study uses the 7 K SomaScan® assay to discover serum proteins associated with muscle function and disease milestones. In total 702 serum samples from 153 male patients, collected across two centers (2009–2022), are analyzed. Using linear mixed effects modelling, we evaluate age and corticosteroid use as covariates affecting protein levels and assess protein correlations with longitudinal clinical function. Here we show 318 aptamers (294 proteins) significantly associated with motor performance across the two sites, with most associations found with lower limb functional tests (NSAA, 10MRW, and 6MWT). Thirty-six proteins are associated with milestones including RGMA, ART3, ANTXR2, and DLK1. These proteins show promise as prognostic biomarkers, and could potentially be used for patient stratification in clinical trial design and for monitoring interventions. Duchenne muscular dystrophy (DMD) is the most common muscular dystrophy. In this study, the authors identified proteins in blood that correlate with disease progression, opening the possibility to monitor disease trajectories using non-invasive blood testing.