Reliability and accuracy of skeletal muscle imaging in limb-girdle muscular dystrophies

Objective: To evaluate the reliability and accuracy of skeletal muscle CT to correctly identify different muscular dystrophies manifesting with limb-girdle weakness. Methods: Four evaluators assessed scans from 118 patients with limb-girdle muscular dystrophy (LGMD) caused by mutations in 7 different genes and from 32 controls. The conditions studied were scans of genetically confirmed cases of Becker muscular dystrophy (BMD) (n = 28), LGMD2C-F (sarcoglycanopathies) (n = 11), LGMD2I (n = 4), LGMD1B (n = 26), LGMD2A (n = 24), Bethlem myopathy (n = 14), and LGMD2L (n = 11). The control group (n = 32) consisted of patients with neuromuscular disorders manifesting with limb-girdle weakness in which the aforementioned muscular dystrophies were excluded. The scans were compared with the characteristic patterns described in literature. Results: The overall interobserver agreement was poor (kappa = 0.27), with markedly higher scores for BMD (kappa = 0.51) and Bethlem myopathy (kappa = 0.59). The sensitivity to detect selective patterns in relation to the genetic diagnosis was 40% if all LGMDs were taken together. The specificity was 58%, positive predictive value (PPV) 77%, and 1 - negative predictive value (NPV) 79%. Markedly better scores were observed for BMD (sensitivity 91%, PPV 66%, 1 - NPV 3%) and Bethlem myopathy (sensitivity 90%, PPV 69%, 1 - NPV 1%). Conclusions: Our findings suggest that muscle CT might be an adjunct to the clinical diagnosis of BMD and Bethlem myopathy. However, pattern recognition was cumbersome in the other LGMDs. Neurology (R) 2012; 79: 1716-1723