Brain MRI of Dystroglycanopathy Patients in a Chinese Pediatric Cohort: Genotype-Specific Neuroimaging Signatures and Neurodevelopmental Correlates

Objective To delineate clinical-genetic-neuroimaging correlations in a Chinese cohort of patients with dystroglycanopathy (DGP). Methods This observational cohort study enrolled 90 genetically confirmed DGP patients. Brain MRI was available for 63 patients; 38 high-quality scans underwent blinded phenotyping. Patients were classified into Group 1 (combined supratentorial and infratentorial malformations), Group 2 (isolated infratentorial malformations), and Group 3 (normal/non-specific findings). Clinical assessments included motor milestones, intellectual disability, expressive language, receptive language, and epilepsy. Spearman’s correlation and network analyses were performed. Results The genetic spectrum included POMGNT1 (12), CRPPA (7), POMT1 (6), FKRP (5), FKTN (4), POMT2 (2), and B3GALNT2 (2). Commonest MRI abnormalities were cerebellar dysplasia (71%), brainstem hypoplasia (63%), and cerebellar microcysts (58%). Group 1 \(\:\text{n=19}\) had severe intellectual disability (100%) and global motor delay; 63% were preverbal/early verbal, 32% had limited phrases, and only 5% achieved basic conversation; epilepsy occurred in 53% and mortality in 16%. Group 2 \(\:\text{n=9}\) had milder intellectual (56%) and language impairment (44%); 67% achieved basic conversation, 89% comprehended simple commands, and epilepsy occurred in 33%. Group 3 \(\:\text{n=10}\) had the mildest phenotype; no epilepsy or mortality occurred, 80% achieved basic conversation, and 100% comprehended simple commands. Imaging severity correlated strongly with clinical phenotype \(\:\text{ρ=0.756,P<0.001}\). Expressive language was the most connected clinical hub, but partial correlations adjusting for supratentorial lesions rendered infratentorial associations non-significant (all P>0.05). Receptive language deficits correlated specifically with lissencephaly \(\:\text{ρ=0.62,P<0.001}\); cerebellar dysplasia independently correlated with epilepsy (partial r=0.351, P=0.033). Genotype dictated distinct correlation spectra: POMGNT1 almost exclusively drove Group 1 (83%); CRPPA uniformly showed cerebellar dysplasia; POMT1/POMT2 exhibited “functional-structural dissociation”; FKRP showed the broadest heterogeneity and weakest imaging-clinical alignment. Conclusions Receptive language deficit is highly specific and strongly associated only with lissencephaly, whereas expressive language is the most widespread and sensitive clinical indicator. Genotype-guided neuroimaging assessment is essential for precise prognostication.