Characterizing the Extended Language Network in Individuals with Multiple Sclerosis

BackgroundLanguage dysfunction is increasingly recognized as a prevalent and early affected cognitive domain in individuals with MS. ObjectivesTo establish a network-level model of language dysfunction in MS. MethodsCognitive data and 3T functional and structural brain MRI were acquired from 54 MS patients and 54 matched healthy controls (HCs). Functional summary measures (anteriority, segregation, betweenness, within-ness) of the extended language network (ELN) were calculated and structural imaging metrics were derived. Group differences in ELN connectivity were evaluated. Associations between ELN connectivity and language performance were assessed; in the MS group, an unsupervised learning approach was used to assess relationships between multimodal neuroimaging features derived from language-related areas and performance on language tasks. ResultsThe MS group performed worse on semantic fluency and rapid automized naming tests (p<0.005) compared to HC. Regarding ELN measures, the MS group exhibited higher within-ELN connectivity than HCs (p<0.05). Principal component analysis (PCA) yielded a multimodal latent component that uniquely correlated with language performance (p<0.05). ConclusionWe identified network-level functional and structural measures to potentially characterize language dysfunction in MS. Further studies leveraging these features may reveal mechanisms and predictors of language dysfunction specific to MS.