Introduction: We sought to define the whole-body MRI (WB-MRI) fingerprint of muscle involvement in pediatric LMNA-related dystrophy (LMNA-RD) and to compare it with SEPN1-related myopathy (SEPN1-RM).
Methods: Signal abnormality and atrophy in 109 muscles were scored by semiquantitative scales in 8 children with LMNA-RD and represented by heatmaps. These features were compared with those from 9 SEPN1-RM patients by random forests.
Results: LMNA-RD showed predominant signal abnormalities in erector spinae, serratus anterior, subscapularis, gluteus medius and minimus, vastii, adductor magnus and longus, semimembranosus, medial gastrocnemius, and soleus muscles. Psoas, sternocleidomastoid, gracilis, and sartorius muscles often had normal signal but showed atrophy. Cranial, flexor digitorum longus, and tibialis posterior muscles were spared. According to random forests, atrophied semimembranosus in SEPN1-RM was the most relevant feature to distinguish these patients from LMNA-RD.
Conclusions: A selective pattern in WB-MRI for pediatric LMNA-RD exists and can be differentiated from SEPN1-RM by machine learning. Muscle Nerve 54: 192-202, 2016.
Keywords: LMNA; SEPN1; child; laminopathy; pattern recognition; whole-body MRI.
© 2016 Wiley Periodicals, Inc.