Dynamic Functional Connectivity Better Predicts Disability Than Structural and Static Functional Connectivity in...

Area of Science:

• Neuroimaging
• Neurology
• Biomedical Engineering

Background:

• Advanced MRI techniques reveal changes in brain networks (structural connectivity - SC, functional connectivity - FC) in people with multiple sclerosis (pwMS).
• No prior study has directly compared SC, static FC, and dynamic FC (dFC) for distinguishing pwMS from healthy controls (HC) or classifying disability within pwMS.
• Understanding these network changes is crucial for mapping disability and compensatory mechanisms in pwMS.

Purpose of the Study:

• To compare the classification performance of SC, static FC, and dFC in differentiating HC from pwMS.
• To compare the classification performance of SC, static FC, and dFC in categorizing pwMS based on disability status.
• To identify specific brain regions and connectome measures most influential in these classification tasks.

Main Methods:

• 100 pwMS and 19 HC underwent diffusion MRI for SC and resting-state fMRI for FC/dFC analysis.
• Disability was assessed using the Expanded Disability Status Scale (EDSS); pwMS with EDSS<2 were classified as having no disability.
• Logistic regression with ridge regularization was employed, incorporating demographic/clinical data and connectome matrices (SC, FC, dFC) for classification performance evaluation via AUC.

Main Results:

• Regional SC models achieved the highest accuracy (median AUC 0.89) in distinguishing HC from pwMS.
• Regional dFC and dFC metrics models significantly outperformed others (median AUC 0.65 and 0.61) in classifying pwMS by disability status.
• Key regions for HC vs. pwMS classification included dorsal attention, subcortical, and cerebellar networks (SC). For pwMS disability, altered dFC in dorsal attention, visual, frontoparietal, and cerebellar networks was significant.

Conclusions:

• Structural connectivity is the most accurate connectomic measure for differentiating between healthy controls and people with multiple sclerosis.
• Dynamic functional connectivity metrics are superior for assessing disability levels in pwMS, potentially indicating functional compensation strategies.
• These findings highlight the distinct roles of SC and dFC in MS pathophysiology and disability progression.