Reduced Network Dynamics on Functional MRI Signals Cognitive Impairment in Multiple Sclerosis

Background: Previous studies have demonstrated extensive functional network disturbances in patients with multiple sclerosis (MS), showing a less efficient brain network. Recent studies indicate that the dynamic properties of the brain network show a strong correlation with cognitive function. Purpose: To investigate network dynamics on functional MRI in cognitively impaired patients with MS. Materials and Methods: In secondary analysis of prospectively acquired data, with imaging performed between 2008 and 2012, differences in regional functional network dynamics (ie, eigenvector centrality dynamics) between cognitively impaired and cognitively preserved participants with MS were investigated. Functional network dynamics were computed on images from functional MRI (3 T) by using a sliding-window approach. Cognitively impaired and preserved groups were compared by using a clusterwise permutation-based method. Results: The study included 96 healthy control subjects and 332 participants with MS (including 226 women and 106 men; median age, 48.1 years 6 11.0). Among the 332 participants with MS, 87 were cognitively impaired and 180 had preserved cognitive function; mildly impaired patients (n = 65) were excluded. The cognitively impaired group included a higher proportion of men compared with the cognitively preserved group (35 of 87 [40%] vs 48 of 180 [27%], respectively; P = .02) and had a higher mean age (51.1 years vs 46.3 years, respectively; P , .01). The clusterwise permutation-based comparison at P less than .05 showed reduced centrality dynamics in default-mode, frontoparietal, and visual network regions on functional MRI in cognitively impaired participants versus cognitively preserved participants. A subsequent correlation and hierarchical clustering analysis revealed that the default-mode and visual networks normally demonstrate negatively correlated fluctuations in functional importance (r = 20.23 in healthy control subjects), with an almost complete loss of this negative correlation in cognitively impaired participants compared with cognitively preserved participants (r = 20.04 vs r = 20.14; corrected P = .02). Conclusion: As shown on functional MRI, cognitively impaired patients with multiple sclerosis not only demonstrate reduced dynamics in default-mode, frontoparietal, and visual networks, but also show a loss of interplay between default-mode and visual networks.