2021 CMSC Annual Meeting

Functional Neuroimaging Metrics Predict Processing Speed and Correlate with Disease Burden in Relapsing-Remitting Multiple Sclerosis


Background: Currently, it is hypothesized that the presentation of cognitive impairment observed in people with multiple sclerosis (PwMS) is dependent on a complex interaction between multifactorial structural damage of the central nervous system and the resiliency of functional brain networks supporting cognition. These networks, believed to support communication between disparate brain regions, may be able to reflect the impact of structural damage even before cognitive symptoms emerge. However, to date, the development of data-driven, whole-brain functional neuromarkers is limited, and their usefulness above and beyond structural measures of disease burden has not been sufficiently investigated. Objectives: The current study utilized connectome-based predictive modeling to derive resting-state functional neuromarkers of processing speed and working memory performance in individuals with relapsing-remitting MS (RRMS), and assessed the variance in overall disease burden accounted for by metrics derived from these models. Methods: 33 individuals with RRMS completed the WAIS-IV Working Memory and Processing Speed Indices (WMI, PSI) as well as a 14-minute resting-state fMRI scan. Functional connectivity was calculated as the correlation between brain regions as defined by a whole-brain atlas. We derived functional connectivity models to predict PSI and WMI separately, and then regressed model-derived metrics, cognitive scores, and total lesion volume onto Expanded Disability Status Scale (EDSS) scores to assess the variance accounted for by each predictor variable. Results: We successfully derived a model of processing speed (rs= .41, p= .03), but not working memory, perhaps due to our low disease severity sample (EDSS < 5.5) and heterogenous presentation of functional changes. Model derivation revealed functional connections primarily within the cerebellar network that were associated with slower processing speed and higher EDSS. Additionally, fMRI-derived PSI metrics uniquely accounted for 13.19% of the variance in EDSS. In contrast, behavioral performance on PSI and WMI, and total lesion volume did not explain a significant amount of variance in EDSS. Conclusions: Our study illustrates that models of MS disease burden may benefit from functional inputs in addition to structural and behavioral measures. Given individual variation in functional brain reserve, functional networks may be better able to capture the net impact of structural damage on disease burden. Future investigations should continue to utilize whole-brain approaches that include cerebellar and brainstem network coverage.