Aerobic Exercise Training and Depressive Symptoms in People With Multiple Sclerosis: Brief Report on Default-Mode Network Resting-State Functional Connectivity

Multiple sclerosis (MS) is a neuroinflammatory and neurodegenerative disorder of the central nervous system that affects more than 2.8 million people world-wide.^1,2  The disease process results in a heterogeneous, debilitating, and often progressive presentation of symptoms, including the onset of fatigue, mobility impairment, cognitive dysfunction, and poor mental health.^3,4  The presence of elevated depressive symptoms represents a particularly common, burdensome, and poorly managed symptom of MS.⁵ 

Depression is characterized by the absence of positive affect (eg, loss of interest in activities) and presence of negative affect (eg, sadness, hopelessness, guilt).⁶  The Diagnostic and Statistical Manual of Mental Disorders (Fifth Edition) (DSM-5) characterizes depression based on several subtypes, which can vary greatly in their severity, including major depressive disorder (MDD) or persistent depressive disorder (PDD). In many cases, people with MS present with depressive symptoms without meeting the DSM-5 criteria for MDD or PDD. Indeed, an estimated 25% to 54% of people with MS present with depressive symptoms compared with only 22% of the general population.^7,8  Such symptoms collectively have profound negative effects on quality of life, employment, social support, and medication adherence among people with MS.²  Moreover, suicide risk is twice as high in people with MS compared with the general population.⁹ 

Depressive symptoms are not optimally managed by clinicians who treat people with MS.¹⁰  For example, data show that first-line disease-modifying therapies (DMTs) can exacerbate depressive symptoms, but the results are ambiguous.¹¹  Antidepressant medication is frequently prescribed for the treatment of depression and elevated symptomatology, and there are both short- and long-term benefits of the medica tions.¹²  Antidepressants were also shown to relieve neuropathic pain and improve fatigue symptoms to some degree in those with MS.¹³  However, treatment with antidepressant medication can also yield physical and mental adverse effects, including sleeping disorders, weight gain, and sexual dysfunction.¹⁴  To that end, the overall efficacy of antidepressant medications in MS is unclear, as is the degree to which their benefit outweighs the risk of adverse effects.¹⁵ 

Psychotherapy is another option for treating depressive symptoms in MS. Though psychotherapy generally results in fewer adverse effects compared with antidepressants, the effects of psychotherapy alone are seemingly limited. Recovery rates via cognitive-behavioral therapy (CBT) are variable. Some studies report significant improvements after CBT in approximately 50% of patients with MS, whereas other studies report no significant changes in people with MS.¹⁶  Thus, psychotherapy effects might be enhanced when included as part of combinatory treatment.^17,18 

By comparison, exercise training has been identified as a highly promising approach for potentially mitigating depressive symptoms in people with MS.¹⁹  Four meta-analyses collectively report that exercise training is associated with small but significant reductions in depressive symptoms in this population.^8,20-22  The data further suggest that exercise interventions that meet or exceed public health guidelines for physical activity regarding frequency, duration, and intensity might exert larger reductions in depressive symptoms relative to exercise interventions that do not meet such public health guidelines.²¹  Nevertheless, there is limited evidence regarding the putative mechanisms for exercise training effects on depressive symptoms in MS. We have previously hypothesized that the processing and integration of multisensory input associated with exercise training represents the key to restoring function for people with MS based on adaptations in neural communication (ie, restingstate functional connectivity [RSFC]).²³  However, whether this hypothesis can be extended to explain exercise-related reductions in depressive symptoms in MS is unknown. The focus on underlying mechanisms is critical to provide a biological basis for exercise-related reductions in depressive symptoms in MS and thereby increase the likelihood of exercise training being included as part of the clinical treatment of MS-related depression.¹⁰ 

One approach to extend the application of the aforementioned hypothesis for exercise-related reductions in depressive symptoms in MS involves focusing on networks that are associated with depression. For example, neuroimaging outcomes of the default mode network (DMN) represent fairly well-established neural correlates of depression.²⁴  People with MS who present with elevated depressive symptoms demonstrate significantly greater RSFC in DMN regions compared with people with MS who do not present with elevated depressive symptoms.^25,26  Such evidence coincides with reports of exercise-related changes in RSFC of the DMN in people with MS and older adults from the general population.^27,28  Thus, it is plausible that observations of exercise-related reductions in depressive symptoms in MS may result from exercise-related improvements in RSFC in the DMN.

To that end, the present study performed a secondary analysis of data from a randomized controlled trial (RCT) of aerobic exercise training effects on learning and memory and neuroimaging outcomes in 10 people with MS who had learning and memory impairment.²⁷  This secondary analysis examined a possible explanation of exercise-related reductions in depressive symptoms based on changes in RSFC. To that end, the present investigation may provide initial proof-of-concept data supporting a potential neural mechanism of the effects of exercise on depressive symptoms in MS that might warrant further inquiry in a subsequent, appropriately powered RCT of patients with elevated depressive symptoms or perhaps MDD.

This study (Exercise and Learning and Memory in Multiple Sclerosis, NCT03319771) represents a secondary analysis of functional MRI (fMRI) data from a 12-week, systematically developed, aerobic treadmill walking exercise training intervention for improving learning and memory in 11 fully ambulatory people with relapsing-remitting MS who demonstrated objective impairments in new learning.²⁷  Prospective participants needed to score at least 1.5 SDs below the age-adjusted normative score on the open-trial Selective Reminding Test to demonstrate objective impairment in learning new information as a study inclusion criterion. Participants were excluded from the study if they had uncontrolled clinical depression based on self-report. We did not make any inclusion or exclusion decisions based on baseline Hospital Anxiety and Depression Scale (HADS)²⁹  scores. Complete data on recruitment and inclusion and exclusion criteria associated with the current sample are provided elsewhere.²⁸  The study was approved by the University of Alabama at Birmingham Institutional Review Board, and all participants provided written informed consent.

Depressive symptoms were measured using the HADS, a 14-item measure of perceived symptoms of anxiety and depression for the previous 4 weeks. This involves two 7-item subscales on anxiety and depression, respectively. For this secondary data analysis, we focused on the depression subscale. After reverse scoring of a subset of items, the 7 items for the depression subscale were summed into total scores that ranged from 0 to 21; higher scores indicate more frequent depressive symptoms. Of note, a HADS depression score of 8 or higher can be interpreted as the respondent having elevated depressive symptoms; a score of 11 or higher can be interpreted as the respondent having moderate to severe depressive symptoms.²⁹  The HADS was administered at baseline and again at follow-up (ie, 12 weeks later) by treatment-blinded assessors.

Resting-state functional connectivity was measured using a 6-minute (180 acquisitions) resting-state scan on a Siemens 3T Prisma MRI scanner. Higher RSFC reflects better coherence of the blood oxygen level–dependent signal between 2 or more nonadjacent brain regions, such that those regions are considered to be functionally connected. A T2*-weighted pulse sequence was used to collect functional images (echo time (TE) = 30 ms; repetition time (TR) = 2000 ms; field of view (FOV) = 22 cm; flip angle = 90°; slice thickness = 3.99 mm; 33 slices; matrix = 92 × 92; in-plane resolution = 2.391 × 2.391 mm). A high-resolution magnetization prepared rapid gradient echo (MPRAGE) image was also acquired and used to normalize the functional data into standard space (TE = 3.41 ms; TR = 2100 ms, FOV = 256 mm; flip angle = 9°; slice thickness = 1 mm, number of excitations = 1, matrix = 256 × 256, in-plane resolution = 1 × 1 mm). These scans were performed at baseline and again at follow-up; data analyses were performed in a blinded fashion.

Details of the supervised, 12-week, treadmill walking exercise training intervention and stretching and range-of-motion activities control conditions are reported elsewhere.^27,30  Briefly, the intervention involved 12 weeks of aerobic walking exercise training on a motor-driven treadmill (3 days per week) that was supervised by trained exercise leaders. The intensity and duration of the intervention progressed throughout the study period according to American College of Sports Medicine guidelines. The stretching and range-of-motion activities control condition accounted for attention and social contact. This condition involved low-intensity stretches and range of motion exercises based on a National Multiple Sclerosis Society pamphlet. The control condition occurred on the same frequency and duration as the intervention condition.

We performed a 2-way, mixed analysis of variance to examine changes in HADS depression scores between the conditions. Time (pre-, post-) was included as a within-subjects factor, and condition (intervention, control) was included as a between-subjects factor. Regarding fMRI outcomes, the fMRI data were preprocessed using fMRIPrep (version 1.4.1).³¹  Preprocessing steps include slice timing correction, spatial realignment, coregistration to the T1 MPRAGE image for localization of activated areas, nonlinear warping into standard Montreal Neurological Institute (MNI) space, estimation of confounds such as framewise displacement (FD), and derivative of variance and global signals associated with white matter, cerebrospinal fluid, and a mask of the whole brain.³²  In addition, a set of physiological regressors was extracted to allow for component-based noise correction (CompCor).³³  The data were smoothed (6 mm full width at half maximum) to minimize anatomical differences and increase the signal-to-noise ratio, and then each voxel was scaled to the grand mean intensity of that voxel (across the acquisition run).

In all cases, the data were checked for excessive motion (a shift of more than 3.5 mm, or 1° of angular motion) and data acquisition runs with excessive motion were discarded. The motion parameters and their derivatives from the realignment step were used as regressors of no interest in the deconvolution, as were FD and the first 6 components from CompCor. The residuals from the deconvolution were saved and subsequently concatenated to perform a group spatial Individual Component Analysis (ICA) using the MELODIC tool from the FMRIB Software Library. We chose this approach as it is data driven and allows signals acquired from the whole brain to be separated into the spatially independent neural components. The resulting maps were then used to perform a dual regression to identify significant neural networks for each individual. These maps were then entered into a linear mixed-effects (LME) model with the factors of Group (treatment vs control), Time (pre- vs postintervention), with depression entered as a quantitative variable (covariate) and subject as a random variable. Based on our a priori hypothesis regarding the potential effect of exercise training on the relationship between depressive symptoms and DMN connectivity, we opted for an LME approach. Accordingly, the results of the LME were corrected for multiple comparisons by using an individual voxel probability threshold of P < .005 and a cluster threshold of 10 voxels (voxel dimension = 2.4 × 2.4 × 4 mm). Monte Carlo simulations, using 3dClustSim (version AFNI_21.3.04; compile date: October 20, 2021), showed that this combination resulted in a corrected α level of P < .05 for a region of interest confined to the DMN.

Demographic information and clinical characteristics of the sample are provided in TABLE 1. Six participants were randomly assigned into the intervention condition and 5 into the control condition. One participant in the intervention condition was excluded from the present data analyses based on corruption of the fMRI data. Overall, the sample was predominantly female and highly educated and had mild MS disability. Baseline and follow-up HADS data are presented in TABLE S1. The sample did not demonstrate elevated depressive symptomology based on a mean baseline HADS depression score of 5.2 (SD = 3.5). Of note, baseline HADS depression scores were similar between the conditions (t = 0.53, P = .61).

Overall, there was a non–statistically significant condition × time interaction on depressive symptoms (F(1, 8) = 2.701, P = .14, n_p² = 0.25; d = 0.98). Although nonsignificant based on null-hypothesis significance testing, this interaction was large in magnitude based on effect size estimates and suggested that the treadmill walking condition was associated with large reductions in HADS depression scores compared with minimal change for the control condition (Table S1; FIGURE S1). The individual-level changes in HADS depression scores based on condition are provided in FIGURE S2. Of participants who were randomly assigned into the intervention condition, no participants demonstrated worsening of depressive symptoms; 3 of 5 participants demonstrated reductions in HADS depression scores and the other 2 participants demonstrated no change. By comparison, in the control condition, 2 of 5 participants demonstrated worsening of symptoms, and the remaining 3 participants demonstrated relatively small reductions in depressive symptoms across the 12-week study period.

Baseline and follow-up RSFC data are provided in FIGURE 1. The dual regression/ICA approach identified significant connectivity within the DMN in an area in the ventromedial prefrontal cortex (vmPFC) consisting of 15 voxels with maximal connectivity located at x = 47.5, y = 37.8 and z = 17.8 (x² = 18.71) (FIGURE 1A). At baseline (time 1), there was a nonsignificant relationship between DMN RSFC and HADS depression scores for the control group (coefficient = −0.00353, P = .72) and a significant positive relationship between DMN RSFC and HADS depression scores for the intervention group (coefficient = 0.05213, P = .001) (FIGURE 1B). However, at follow-up (time 2), there was a statistically significant negative relationship between DMN RSFC and HADS depression scores in the intervention condition, whereby higher RSFC was associated with lower HADS depression scores (coefficient= −0.05685, P < .015). By comparison, at follow-up, the control condition was associated with a statistically significant, positive relationship between DMN RSFC and HADS depression scores, whereby higher RSFC was associated with higher (ie, worse) HADS depression scores (coefficient= 0.02493, P < .025) (Figure 1B).

The current study is a secondary analysis of data on RSFC and depressive symptom outcomes from an RCT that originally focused on the effect of walking exercise on learning and memory in memory-impaired persons with MS. Our secondary analysis affords the opportunity to investigate a potential neural mechanism of the effect of aerobic walking exercise on depressive symptoms in individuals with MS. Despite the small sample size, the results indicated that aerobic walking exercise training was associated with a large but nonsignificant reduction in depressive symptoms, whereas there was no change in participants in the active control condition. This effect may be explained by changes in within-DMN RSFC, whereas at follow-up, higher RSFC was associated with lower HADS depression scores in participants in exercise training compared with higher HADS depression scores in participants in the active control condition. To that end, the overall pattern of results provides initial proof-of-concept data supporting a potential extension of a published hypothesis to explain exercise-related improvements in MS symptoms as well as a putative neural mechanism of aerobic walking exercise–related reductions in depressive symptoms in fully ambulatory people with MS.²³  As the present study was a secondary analysis of RCT data, such an extension of that hypothesis and potential neural mechanism warrants further focal consideration in subsequent RCT research.

The current pattern of results is consistent with previous data on exercise-related reductions in depressive symptoms in persons with MS^8,20-22 : twelve weeks of aerobic treadmill walking exercise training was associated with large reductions in HADS depression scores. Although not statistically significant, likely based on a small sample size, the large effect exceeded the mean overall effect sizes from recent meta-analyses on exercise-related reductions in depressive symptoms in MS.^8,20-22  Of note, 2 meta-analyses reported that interventions that exceeded public health guidelines with at least 3 days per week of training may result in larger reductions in depressive symptoms among persons with MS.^21,20  The final month of the present intervention consisted of 3 days per week of vigorous intensity walking exercise, exceeding the exercise volume physical activity guidelines for adults with MS.^27,30,34  Although the RCT was designed to improve learning and memory among cognitively impaired persons with MS, such a secondary observation sets the stage for subsequent RCT research involving aerobic treadmill walking exercise as a potential treatment for MS-related depression in those who have a definitive diagnosis of MDD (ie, those who present with the problem being studied).

Regarding neural mechanisms of exercise-related effects on depressive symptoms in persons with MS, the current secondary analysis of data identified RSFC within the DMN as a network that coincided with the aforementioned reductions in depressive symptoms. At follow-up, increases in RSFC were associated with reductions in depressive symptoms for those who underwent 12 weeks of supervised aerobic walking exercise training. DMN RSFC has also been linked with depressive symptoms as well as changes in depressive symptoms in the general adult population.^24-26,35-37  Of particular importance, the DMN has been identified as a network that serves as an integration center for incoming extrinsic information with established intrinsic information.³⁸  Such converging evidence provides support for the current pattern of results such that the processing and integration of multisensory input required for the physiological regulation of aerobic treadmill walking exercise training resulted in DMN adaptations and downstream reductions in depressive symptoms in fully ambulatory persons with MS.²³  Such a hypothesis is also consistent with previous research on exercise-related changes in DMN RSFC in general and in persons with MS.^27,28  Nevertheless, subsequent RCTs of exercise training for reducing depressive symptoms in MS might consider focusing on the RSFC within the DMN as a potential target to maximize effects.

This study has several noteworthy limitations. The primary limitation involves the small sample size and the secondary analysis of data from an RCT that was designed for another purpose. The small sample, mostly women with relatively mild MS disability (ie, an inclusion criterion of the original trial^27,30 ), limits generalizability of results. Furthermore, because the original trial focused on MS-related cognitive impairment, people who reported uncontrolled clinical depression were screened out as a potential confound of cognitive impairment.^27,30  Thus, only a few people in the reported sample had clinically significant depression. This restricted the range on HADS depression scores, reducing our ability to detect depression-related effects based on null hypothesis significance testing. Considering the large observed effect size, the fact that we nevertheless found a relationship between depressive symptoms and DMN connectivity suggests that this relationship may be robust. Additionally, there were baseline differences in the relationships between DMN RSFC and HADS depression score between the 2 conditions. However, the small sample size precluded us from controlling for these differences given the loss of degrees of freedom resulting from lack of statistical power.

Future studies should consider potential baseline differences in the relationship between DMN RSFC and HADS depression scores by performing stratified randomization and/or examining intervention effects on the follow-up DMN RSFC/ HADS depression score correlation and controlling for baseline correlation as an a priori analytic approach. Another important avenue for future research involves evaluating potential differences in exercise-related reductions in depressive symptoms based on various aerobic exercise modalities (eg, aerobic walking vs cycling vs arm ergometry) in fully ambulatory persons with MS. It is possible that aerobic exercise activities that involve different neurophysiological demands require different patterns of processing and integrating multisensory input that might differentially result in adaptations within the DMN. Nevertheless, the pattern of results warrants future consideration in the design and conduct of future, appropriately powered RCTs that aim to treat MS-related depression with aerobic exercise training in people with MS who are prescreened for clinical depression.