Resting-State EEG Microstates as Dynamic Biomarkers of Network Dysfunction and Cognitive Impairment in Patients With...

Multiple sclerosis (MS) is characterized by large-scale brain network dysfunctions. EEG microstate analysis may track these dynamics. We investigated whether microstates could distinguish patients with MS from healthy volunteers, and cognitively impaired from cognitively preserved (CP) patients.

For this cross-sectional observational study, we recruited participants from the Policlinico Umberto I Multiple Sclerosis Center (Rome). We compared microstates of patients with relapsing-remitting MS (Expanded Disability Status Scale [EDSS] score <3, no relapses within 30 days) with those of age-matched and sex-matched healthy volunteers. We also compared cognitively impaired and preserved patients. Group comparisons were performed on microstate temporal parameters and topographies (topographic analysis of variance). Classification used stepwise linear discriminant analysis (LDA).

We enrolled 88 patients with MS (60 women, mean age 42.75 years) and 46 healthy volunteers (23 women, mean age 33.7 years). Compared with healthy volunteers (n = 45, 23 women, mean age 33.7 years), patients with MS (n = 46, 25 women, mean age 34.6 years) showed greater explained variance (p < 0.001), occurrence (p < 0.001), and coverage (p < 0.001) of class B (visual network) and reduced mean duration (MD) of classes F (salience network, p = 0.014) and G (sensorimotor network, p = 0.009). Topographic differences were found between cognitively impaired (n = 57, 20 women, mean age 35.5 years) and preserved (n = 31, 40 women, mean age 46.7 years) patients for classes C, F, and G (p < 0.05). Patients with cognitive impairment exhibited significantly reduced explained variance of class F (p = 0.003). Class A was associated with longer disease duration (explained variance, p = 0.006; MD, p = 0.042; coverage, p = 0.007), higher EDSS scores (MD, p = 0.013), and poorer ambulation (explained variance, p = 0.049). Class D inversely correlated with EDSS scores (occurrence, p = 0.011; coverage, p = 0.036), whereas class B inversely correlated with information processing speed (explained variance, p = 0.025; occurrence, p = 0.018; coverage, p = 0.028). Stepwise LDA identified classes B and D as discriminators between patients with MS and healthy volunteers (77% accuracy, p < 0.001), and class F as a discriminator between CP and impaired participants (64.8% accuracy, p = 0.007).

EEG microstates reveal behaviorally relevant alterations in patients with MS, supporting their utility as sensitive markers of large-scale functional network reorganization associated with clinical disability.