Background: Multiple sclerosis (MS) is characterized by focal white matter damage, and when the brain is modeled as a network, lesions can be treated as disconnection events.
Objective: To evaluate whether modeling disconnection caused by lesions helps explain motor and cognitive impairment in MS.
Methods: Pathways connecting 116 cortical regions were reconstructed with magnetic resonance imaging (MRI) tractography from diffusion tensors averaged across healthy controls (HCs); maps of pathways were applied to 227 relapse-onset MS patients and 50 HCs to derive structural connectivity. Then, the likelihood of individual connections passing through lesions was used to model disconnection. Patients were grouped according to clinical phenotype (113 relapsing-remitting multiple sclerosis (RRMS), 69 secondary progressive multiple sclerosis (SPMS), 45 benign MS), and then network metrics were compared between groups (analysis of variance (ANOVA)) and correlated with motor and cognitive scores (linear regression).
Results: Global metrics differentiated RRMS from SPMS and benign MS patients, but not benign from SPMS patients. Nodal connectivity strength replicated global results. After disconnection, few nodes were significantly different between benign MS and RRMS patients. Correlations revealed nodes pertinent to motor and cognitive dysfunctions; these became slightly stronger after disconnection.
Conclusion: Connectivity did not change greatly after modeled disconnection, suggesting that the brain network is robust against damage caused by MS lesions.
Keywords: Multiple sclerosis; clinical phenotypes; diffusion MRI; fiber bundle transection; graph theory; simulation of disconnection.