Data-driven analysis shows robust links between fatigue and depression in early multiple sclerosis

Yuan-Ting Chang, Patrick K.A. Kearns, Alan Carson, David C. Gillespie, Rozanna Meijboom, Agniete Kampaite, Maria del C. Valdés Hernández, Christine Weaver, Amy Stenson, Niall MacDougall, Jonathan O’Riordan, Margaret Ann Macleod, Francisco Javier Carod-Artal, Peter Connick, Adam D. Waldman, Siddharthan Chandran, Peter Foley

Research output: Working paperPreprint


Fatigue is common and disabling in multiple sclerosis, yet its mechanisms are poorly understood. In particular, overlap in measures of fatigue and depression complicates interpretation. A clearer understanding of relationships between fatigue and key clinical, neuropsychiatric and imaging variables including depression could yield clinically relevant mechanistic insight. We applied a data-driven multivariate network approach to quantify relationships between fatigue and other variables in early multiple sclerosis.

Data were collected from Scottish patients with newly diagnosed, immunotherapy-naïve, relapsing-remitting multiple sclerosis at baseline and month 12 follow-up in FutureMS, a nationally representative multicentre cohort. Subjective fatigue was assessed using the validated Fatigue Severity Scale. Detailed phenotyping included measures assessing physical disability, affective disorders, objective cognitive performance, subjective sleep quality, and structural brain imaging. Bivariate correlations between fatigue and other variables were calculated. Network analysis was then conducted to estimate partial correlations between variables, after accounting for all other included variables. Secondary networks included individual depressive symptoms, to control for overlapping symptom items in measures of fatigue and depression.

Data from 322 participants at baseline, and 323 at month 12, were included. At baseline, 49.5% of the cohort reported clinically significant fatigue. Bivariate correlations confirmed that fatigue severity was significantly correlated with all included measures of physical disability, affective disturbance (anxiety and depression), cognitive performance (processing speed and memory/attention), and sleep quality, but not with structural brain imaging variables including normalized lesion and grey matter volumes. In the network analysis, fatigue showed strong correlations with depression, followed by Expanded Disability Status Scale. Weak connections with walking speed, subjective sleep quality and anxiety were identified. After separately controlling for measurement of “tiredness” in our measure of depression, some key depressive symptoms (anhedonia, subjective concentration deficits, subjectively altered speed of movement, and appetite) remained linked to fatigue. Conversely, fatigue was not linked to objective cognitive performance, white matter lesion volume, or grey matter volumes (cortical, subcortical or thalamic). Results were consistent at baseline and month 12. Depression was identified as the most central variable in the networks. Correlation stability coefficients and bootstrapped confidence intervals of the edge weights supported stability of the estimated networks.

Our findings support robust links between subjective fatigue and depression in early relapsing-remitting multiple sclerosis, despite absence of links between fatigue and either objective cognitive performance, or structural brain imaging variables. Depression, including specific depressive symptoms, could be a key target of treatment and research in multiple sclerosis-related fatigue.
Original languageEnglish
Publication statusPublished - 13 Jan 2022


Dive into the research topics of 'Data-driven analysis shows robust links between fatigue and depression in early multiple sclerosis'. Together they form a unique fingerprint.

Cite this