Global Rearrangement of Degree Centrality Reflects Cognitive Impairment and Fatigue in Multiple Sclerosis

Abstract

Background and Objectives: The aim of this secondary data analysis was to determine whether multiple sclerosis (MS) is associated with changes in global degree rank order disruption index (kD), a graph theory-based functional connectivity measure representing shift in overall distribution of nodal degree centrality. Additionally, we tested the relationship between kD and MS symptoms (cognitive and motor impairment, fatigue, and global disability). Methods: Global kD was computed in a pre-existing cross-sectional fMRI dataset and compared between patients with MS (PwMS) and healthy controls (HCs). Group differentiation was tested against other known biomarkers in MS (regional degree centrality, structural MRI with volumetry, diffusion-weighted imaging, lesion mapping) using receiver operating characteristic and logistic regression analysis. Associations between kD and cognitive processing speed (Symbol Digit Modalities Test), fatigue (Fatigue Scale for Motor and Cognitive Functions), gait (Timed Up and Go Test), and disability (Expanded Disability Status Scale [EDSS]) were evaluated using Spearman correlation coefficient and ordinal regression adjusted for structural imaging, age, sex, and disease duration. Results: Analysis included 56 PwMS and 58 HCs (35/27 women, median age 45.1/40.5 years). Global kD was lower in PwMS (median −0.30, inter-quartile range [IQR] 0.55) than in HCs (median −0.06, IQR 0.54; p = 0.009, Mann-Whitney U test). kD yielded acceptable differentiation between groups (area under curve 0.64), but did not improve such differentiation on top of structural imaging. Both kD and regional degree in medial prefrontal cortex (MPFC) were correlated with cognitive decline (kD: Spearman's ρ = 0.32, p = 0.019; MPFC: ρ = −0.45, p = 0.001, n = 55), while kD was also correlated with fatigue (ρ = −0.34, p = 0.010, n = 56), but not with EDSS (ρ = −0.06, p = 0.674, n = 56) or gait (ρ = −0.18, p = 0.211, n = 52). kD significantly explained cognitive impairment (χ2 = 4.49, p = 0.034) and fatigue (χ2 = 7.18, p = 0.007). Discussion: Our data provide evidence that kD is a potential biomarker of cognitive decline and fatigue. Further cross-validations are required to assess its generalizability.