Time‐Varying Brain Functional Reconfiguration Patterns Associated With Fatigue in Multiple Sclerosis

ABSTRACT Fatigue affects between 36.5% and 78% of people with multiple sclerosis (pwMS) and significantly impairs daily life. The neurobiological mechanisms underlying fatigue remain poorly understood and information about the time‐varying communication between brain regions and the networks they form may offer new insights into the complex pathology of MS‐related fatigue. Brain regions continuously reconfigure how they communicate within distinct networks (i.e., time‐varying reconfigurations) and aberrant time‐varying reconfigurations may contribute to the perception of fatigue in pwMS. This study aimed to explore if and how time‐varying reconfigurations are associated with fatigue in pwMS. In this cross‐sectional study, 155 pwMS (62% female; age = 39 ± 10 years; disease duration = 10 ± 8 years; median EDSS = 1.0 ± 2.0) and 48 healthy controls (HC) (71% female; age = 33 ± 10 years) underwent clinical, neuropsychological, and (resting‐state functional) MRI assessments. Fatigue was evaluated with the “Fatigue Scale for Motor and Cognitive Function”, comprising total, motor, and cognitive fatigue scores. Time‐varying connectivity was derived using a sliding‐window approach, with data‐driven assignment of brain regions to one of eight resting‐state networks for each window. Promiscuity (dispersion of reconfigurations), flexibility (frequency of reconfigurations), cohesion (joint reconfigurations), and disjointedness (independent reconfigurations) described the time‐varying reconfigurations of the whole brain and its networks. Among pwMS, 57% reported experiencing at least mild total fatigue (motor: 60%, cognitive: 57%). Higher total fatigue was correlated with greater global promiscuity ( r = 0.21, p = 0.032) and disjointedness ( r = 0.24, p = 0.008). Similarly, higher motor fatigue was associated with greater global promiscuity ( r = 0.25, p = 0.008), flexibility ( r = 0.21, p = 0.032), and disjointedness ( r = 0.28, p < 0.001). The associations with disjointedness remained significant even after controlling for demographics, clinical measures, and structural brain damage, such as lesion load and atrophy (total fatigue: adj. R 2 = 0.23, β = 0.17, p = 0.033; motor fatigue: adj. R 2 = 0.38, β = 0.16, p = 0.026). Network‐level analyses in pwMS revealed that higher total ( r = 0.25, p = 0.016) and motor ( r = 0.25, p = 0.016) fatigue were associated with greater limbic network promiscuity. No significant correlations were found for cognitive fatigue in pwMS, or for total, motor, and cognitive fatigue in HC (all p > 0.05). Elevated levels of fatigue, particularly motor fatigue, in pwMS were linked to more unstable network reconfigurations, particularly of regions in the limbic network, possibly reflecting dysfunctional reward processing. More frequent dispersion requires more energy and may therefore contribute to increased fatigue. Trial Registration This project was pre‐registered at ClinicalTrials.gov (registration number: NCT04892134)