Muscle Imaging Patterns and Variant-Specific Progression in DNM2-Related Centronuclear Myopathy.

INTRODUCTION/AIMS: DNM2-related centronuclear myopathy (DNM2-CNM) exhibits characteristic imaging patterns, but imaging-based staging and variant-specific severity pattern remain incompletely defined. We aimed to (1) characterize muscle involvement patterns, (2) establish imaging-based severity stages and identify muscles that best distinguish severity groups, and (3) model variant- and muscle-specific age-related score patterns, with comparison to adult-onset MTM1-related CNM. METHODS: We retrospectively analyzed whole-body muscle imaging (computed tomography/magnetic resonance imaging) in 30 genetically confirmed DNM2-CNM patients. Forty-five muscles were scored using the modified Mercuri score (mMs 0-4). Unsupervised uniform manifold approximation and projection (UMAP) and hierarchical clustering defined severity groups and involvement patterns. Violin plots characterized the distribution of mMs scores across severity stages. Linear regression and generalized additive models characterized variant and muscle-specific age-associated mMs patterns. Modal mMs values were used to generate comparative schematics for DNM2-CNM and MTM1-related CNM. RESULTS: UMAP identified three severity groups. Hierarchical clustering revealed a distal-predominant pattern with early severe involvement of posterior calf and paraspinal muscles, followed by posterior thigh and gluteal muscles. Variant modeling demonstrated distinct age-associated mMs patterns: PH domain variants (A618T/S615L/S619L) showed earliest onset and steepest age-associated rate of mMs increase (slope 10.1, p = 0.003), middle-domain variants (E368K/R369G/R369Q) showed intermediate rate of mMs increase (slope 2.86), R465W was milder (slope 2.56), and R369W was slowest (slope 0.74). DNM2-CNM showed distal-to-proximal involvement, contrasting with the proximal-predominant pattern of MTM1-related CNM. DISCUSSION: Muscle imaging defines reproducible severity stages and variant-specific age-associated severity patterns, providing a quantitative imaging framework for DNM2-CNM staging, monitoring, and risk stratification.