Quantitative Muscle MRI Fat Fraction as a Biomarker of Disease Severity in Mitochondrial Myopathies

BACKGROUND: Quantitative muscle MRI is increasingly used to assess structural muscle damage in inherited myopathies, but its application in primary mitochondrial myopathies (PMM) has not been systematically evaluated in large cohorts. Because PMM are clinically and genetically heterogeneous, objective imaging biomarkers are needed to quantify skeletal-muscle involvement and define meaningful subgroups. We assessed whether MRI-derived proton density fat fraction (PDFF) captures genotype- and phenotype-specific patterns of fatty replacement and reflects disease severity using motor function outcomes and circulating biomarkers. METHODS: We performed a cross-sectional analysis of 49 adults with genetically confirmed PMM enrolled in a prospective natural history study. Participants underwent standardised lower-limb MRI, including T1-weighted sequences and mDIXON-QUANT fat fraction imaging. Muscle involvement was assessed semi-quantitatively with the modified Mercuri scale and quantitatively by PDFF measurement in selected pelvic girdle and thigh muscles. A total fat fraction score (FF sum) was calculated. PDFF was compared across genotypes and phenotypes and against MRC score, NSAA, 6-min walk test, 100-m run test, CK, serum creatinine, GDF15 and disease duration. RESULTS: Mean age at onset was 26.3 ± 15.1 years, mean age at MRI was 50.0 ± 12.3 years, and median disease duration was 23 years (IQR 15-31). Genetic diagnoses included mtDNA variants in 29 patients (59.1%; single large-scale deletions, n = 19; point mutations, n = 10), nuclear variants in POLG (n = 8), TK2 (n = 8) and TWNK (n = 4). Phenotypes were exercise intolerance without overt weakness (n = 7), isolated progressive external ophthalmoplegia (PEO; n = 16), PEO-plus (n = 16) and progressive myopathy (n = 8). Highest median PDFF values were observed in tensor fasciae latae (31%), gluteus maximus (30%), sartorius (25%) and gracilis (20%). FF sum correlated with MRC score (r = -0.567, p < 0.0001), NSAA (r = -0.731, p < 0.0001), 100-m run test time (r = 0.629, p < 0.0001), 6-min walk distance (r = -0.311, p = 0.0476) and serum creatinine (r = -0.715, p < 0.0001), but not with CK or GDF15. TK2 deficiency showed the highest fat replacement, particularly in gluteus maximus and gracilis (p < 0.001). Progressive myopathy showed greater fatty replacement than other phenotypes (p < 0.001), while exercise intolerance showed higher fat fraction than isolated PEO (p = 0.033). CONCLUSIONS: Quantitative muscle MRI provides objective and clinically meaningful measures of muscle involvement in PMM. MRI-derived fat fraction reflects disease severity, discriminates between genotypes and phenotypes, and detects subclinical muscle damage. These cross-sectional findings support PDFF as a candidate imaging biomarker for patient stratification and disease characterisation; prospective longitudinal studies will be required to establish its sensitivity to change and its validity as an outcome measure for therapeutic trials.