Muscle MRI Fat Fraction Emerges as a Powerful Biomarker of Mitochondrial Myopathy Severity

Mitochondrial myopathies have long frustrated clinicians trying to answer a deceptively simple question: how sick is this patient, really? These rare genetic muscle disorders, driven by faults in mitochondrial DNA or in nuclear genes that maintain the mitochondrial genome, vary enormously from person to person. Two patients carrying mutations in the same gene can present with completely different symptoms, rates of progression and degrees of disability. That heterogeneity has made it notoriously difficult to stratify patients, compare outcomes and design clinical trials. Now a prospective study of 49 adults with primary mitochondrial myopathies, conducted at a Spanish reference centre for neuromuscular and mitochondrial disease and published in the Journal of Cachexia, Sarcopenia and Muscle, offers a compelling solution: a quantitative magnetic resonance imaging measure that turns the invisible burden of muscle degeneration into a reproducible number.

The measure in question is the proton density fat fraction, or PDFF, derived from chemical shift encoding-based water-fat MRI, specifically a six-echo three-dimensional Dixon sequence acquired on a 1.5 Tesla scanner. PDFF expresses the ratio of mobile protons belonging to triglycerides to the total mobile protons from both fat and water within a tissue, yielding a standardised percentage that reflects how much of a muscle has been replaced by fat. Unlike conventional visual scoring, PDFF is largely independent of scanner settings and acquisition parameters, which makes it attractive as a biomarker that could be compared across centres and over time. In many inherited muscle disorders, from dystrophinopathies to limb-girdle muscular dystrophies, quantitative fat mapping has already proven its worth. Mitochondrial myopathies, by contrast, had been something of a blind spot, with only scattered case reports and small series suggesting that limb muscle MRI might be uninformative, particularly in patients whose dominant symptom is progressive external ophthalmoplegia.

The new study set out to change that perception with unusually rigorous methodology. Between May 2023 and May 2025, the researchers imaged 49 adults whose diagnoses rested on pathogenic variants in mitochondrial DNA or in the nuclear genes POLG, TK2 and TWNK. Each participant underwent a dedicated MRI protocol of the pelvis, thighs and lower legs, combining conventional T1-weighted and STIR sequences with quantitative Dixon-based fat fraction mapping of the thighs. Two experienced radiologists, blinded to genotype and clinical data, independently graded the same fifteen muscles using the semi-quantitative Mercuri visual scale, a four-point system that estimates the percentage of muscle volume replaced by fat and connective tissue. In parallel, the team manually drew regions of interest on fat fraction maps at three standardised anatomical levels, quantifying individual muscle fat fractions and summing them into a composite score reflecting the overall structural burden of disease.

The first striking result was the near-perfect agreement between the old visual method and the new quantitative one. Across all analysed muscles, fat fraction values correlated strongly with Mercuri grades, with correlation coefficients exceeding 0.7 and p values below 0.00001, and intraclass correlation coefficients reached 1.00 in the smaller genotypic subgroups. This matters because it anchors the quantitative technique to a well-established clinical standard while offering something the visual scale cannot: a continuous, fine-grained variable capable of detecting subtle change. The cohort itself was deeply characterised. Patients ranged widely in age at onset, from childhood to late adulthood, with a median disease duration of 23 years, and their phenotypes spanned isolated exercise intolerance, pure progressive external ophthalmoplegia, ophthalmoplegia plus additional weakness, and progressive myopathy.

Quantitative imaging revealed a highly selective pattern of fatty replacement that had been underappreciated in mitochondrial disease. The tensor fasciae latae showed the highest median fat fraction at 31 percent, followed by the gluteus maximus at 30 percent, the sartorius at 25 percent and the gracilis at 20 percent. At the other extreme, the quadriceps components were comparatively spared, with the vastus medialis at just 11 percent and the vastus intermedius at 13 percent. These values dwarf published reference ranges for healthy adults of comparable age, in whom quadriceps fat fractions typically sit between 2 and 4 percent, indicating that the fatty infiltration observed reflects genuine disease-related degeneration rather than ordinary ageing.

Crucially, the imaging numbers tracked closely with how patients actually functioned. The summed fat fraction correlated inversely with global muscle strength on the Medical Research Council scale, with a Spearman coefficient of −0.567, and even more strongly with the North Star Ambulatory Assessment, at −0.731. Higher fat burden also predicted longer completion times on the 100-metre run test, with a positive correlation of 0.629, and showed a weaker inverse relationship with six-minute walk distance. Among circulating biomarkers, the strongest association was with serum creatinine, a surrogate of muscle mass, which fell sharply as fat fraction rose, with a coefficient of −0.715. Creatine kinase and growth differentiation factor 15, by contrast, showed no significant relationship with the imaging measure, a biologically coherent finding: creatine kinase reflects ongoing muscle damage, GDF15 signals systemic mitochondrial stress, while fat fraction captures the chronic structural endpoint of fibre loss.

Genotype left a clear fingerprint on the images. Patients with TK2 variants, which impair the replication and maintenance of mitochondrial DNA within muscle nuclei, displayed by far the most severe involvement, with median gluteus maximus fat fractions of 65 percent, semitendinosus values of 54 percent and gracilis values of 44.5 percent, alongside the lowest serum creatinine levels of any group. This quantitative signature confirms and extends earlier qualitative descriptions of late-onset TK2 deficiency as a disorder with a distinctive radiological pattern centred on the pelvic girdle and antero-medial thigh. Patients with single large-scale mitochondrial DNA deletions, who most often presented with pure external ophthalmoplegia and the earliest average age at onset, showed the mildest limb muscle involvement, while POLG and TWNK cases occupied intermediate, overlapping territory.

Perhaps the most clinically provocative finding concerned patients who appeared, by conventional standards, only mildly affected. Those presenting with exercise intolerance but no overt weakness carried significantly more muscle fat than patients with pure external ophthalmoplegia, demonstrating that quantitative MRI can expose subclinical structural damage invisible to manual strength testing. Heatmap-based hierarchical clustering and exploratory multiple correspondence analysis reinforced this picture, arranging patients along a severity gradient that mirrored both phenotype and genotype. Notably, disease duration correlated with none of the outcome measures, suggesting that the extent of fatty replacement is dictated more by the underlying genetic defect than by the simple passage of time, an argument for imaging-based rather than time-based staging in this heterogeneous population.

The authors are careful about limits. The cohort was modest, subgroups were small, the design was cross-sectional, and no prospectively imaged healthy control group was scanned on the same platform, so literature-derived normal values cannot serve as formal thresholds. Multivariable adjustment for age, sex and genotype was not feasible, and volumetric segmentation was impractical without dedicated software. Longitudinal studies will be needed to establish whether fat fraction is sensitive to change and predictive of future decline, the properties that would qualify it as a validated trial endpoint. Yet the direction of travel is unmistakable. As disease-modifying therapies for mitochondrial disorders, particularly nuclear-encoded ones, move toward the clinic, a reproducible imaging biomarker that quantifies structural muscle loss, discriminates genotypes, detects silent involvement and mirrors strength and function is exactly the tool the field has been waiting for. Quantitative muscle MRI, long dismissed as uninformative in mitochondrial disease, may now claim a central place in patient stratification and trial readiness.

Subject of Research: Quantitative muscle MRI fat fraction as an imaging biomarker of disease severity in adult mitochondrial myopathies

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

Article References: Bermejo‐Moriñigo, A., Martín‐Jiménez, P., González‐Méndez, V., Bermejo‐Guerrero, L., Ochoa, L. E., Martín‐Arriscado, C., Alcalá‐Galiano, A., Casado‐Pérez, C., Navarro‐Riquelme, M., Garrido‐Moraga, R., González Quintana, A., Blázquez, A., & Domínguez‐González, C. (2026). Quantitative Muscle MRI Fat Fraction as a Biomarker of Disease Severity in Mitochondrial Myopathies. Journal of Cachexia, Sarcopenia and Muscle, 17(5), Article e70380. https://doi.org/10.1002/jcsm.70380

Image Credits: AI Generated

DOI: 10.1002/jcsm.70380

Keywords: mitochondrial myopathy, muscle MRI, fat fraction, PDFF, biomarker, TK2, POLG, TWNK, mtDNA deletion, Mercuri scale, serum creatinine, clinical trials