Exercise and disease progression in multiple sclerosis: can exercise slow down the progression of multiple sclerosis?

1. Disease progression evaluated with clinical measures

A number of studies evaluating structured exercise interventions lasting from 3 to 26 weeks have included clinical scales reflecting disease progression as an outcome measure. The applied clinical scales include the EDSS [Bjarnadottir et al. 2007; Dalgas et al. 2009; Fimland et al. 2010; Golzari et al. 2010; Petajan et al. 1996; Pilutti et al. 2011; Rodgers et al. 1999; Romberg et al. 2004; White et al. 2004], the MSFC [Pilutti et al. 2011; Romberg et al. 2005], the Guys Neurological Disability Scale (GNDS) [Kileff and Ashburn, 2005; van den Berg et al. 2006] and the Functional Independence Measure (FIM) [Romberg et al. 2005]. Studies applying the EDSS have generally not found any change after either endurance training [Petajan et al. 1996; Pilutti et al. 2011; Rodgers et al. 1999], resistance training [Dalgas et al. 2009; Fimland et al. 2010; White et al. 2004] or combined training interventions [Bjarnadottir et al. 2007; Romberg et al. 2004]. Only one study by Golzari and colleagues evaluating the effects of 8 weeks of combined training (3 days/week) reported an improvement in EDSS score [Golzari et al. 2010]. This finding was not confirmed in a long-term study (26 weeks) [Romberg et al. 2005] also evaluating the effects of combined training. In the study by Romberg and colleagues no effect on EDSS and FIM were found, but a small positive effect was seen in the MSFC. A few studies applied the GNDS with one reporting an improvement after 12 weeks of biweekly endurance training [Kileff and Ashburn, 2005] and one reporting no effects of 4 weeks endurance training completed 3 days a week [van den Berg et al. 2006].

In summary, structured exercise intervention studies of different exercise modalities lasting 3–26 weeks have generally found no effects on EDSS scores. A few exercise studies have shown positive effects when applying other clinical scales (MSFC and GNDS).

2. Disease progression evaluated with non-clinical measures

Two studies by Prakash and colleagues have evaluated the effects of cardiorespiratory fitness on brain function and structure by applying (functional) MRI [Prakash et al. 2007, 2009]. One study [Prakash et al. 2007] investigated the impact of cardiorespiratory fitness on cerebrovascular functioning of MS patients. Twenty-four female participants with relapsing–remitting MS were recruited for the study and all participants went through fitness assessment (VO₂ peak) and were scanned in a 3-T MRI system while performing the Paced Visual Serial Addition Test (PVSAT). Higher fitness levels were associated with faster performance during the PVSAT that could be related to greater recruitment of a specific region of the cerebral cortex (right inferior frontal gyrus [IFG] and middle frontal gyrus [MFG]) known to be recruited by MS patients during performance of PVSAT to purportedly compensate for the cognitive deterioration attributable to MS. In contrast, lower levels of fitness were associated with enhanced activity in the anterior cingulate cortex (ACC), thought to reflect the presence of a larger amount of conflict increasing the potential for error in lower fit MS participants. The authors interpreted the results as supporting aerobic training as an intervention to support the development of additional cortical resources in an attempt to counter the cognitive decline resulting from MS. Among a number of cognitive tests, only the Paced Auditory Serial Addition Test (PASAT) showed a weak correlation (p = 0.42) to VO₂ peak leading the authors to suggest that fitness does not have an influence on measures of general cognitive functioning.

In another study by Prakash and colleagues the relationship between cardiorespiratory fitness (VO₂ max) and measures of gray matter atrophy and white matter integrity (both of which have been associated with the disease process) were studied [Prakash et al. 2009]. A voxel-based approach to analysis of gray matter and white matter was applied on brainscans from a 3-T MRI system. More specifically it was examined whether higher levels of fitness in 21 female MS patients were associated with preserved gray matter volume and integrity of white matter. A positive association between cardiorespiratory fitness and regional gray matter volumes and higher focal fractional anisotropy values were reported. Both preserved gray matter volume and white matter tract integrity were associated with better performance on measures of processing speed. Recognizing the cross-sectional nature of the data, the authors suggested that fitness exerts a prophylactic influence on the structural decline observed early on, preserving neuronal integrity in MS, thereby reducing long-term disability.

In summary, (f)MRI studies suggesting a protective effect of cardiorespiratory fitness on brain function and structure in MS patients have started to emerge. However, the cross-sectional nature of the few existing studies limit conclusions regarding the existence of a causal relationship.

3. Disease progression evaluated with patient-reported measures

A number of studies have addressed the relationship between exercise or physical activity and disease progression in large-scale questionnaire studies applying patient-reported measures.

In a large descriptive longitudinal survey study, Stuifbergen and colleagues examined the correlations between the change in functional limitations, exercise behaviors and quality of life [Stuifbergen et al. 2006]. More than 600 MS patients completed a number of questionnaires every year for a period of 5 years. The self-reported longitudinal measures were analyzed by applying latent curve modeling. The Incapacity Status Scale provided a measure of functional limitations due to MS, whereas the Health Promoting Lifestyle Profile II provided a measure of exercise behavior. At the first test point (baseline test) cross-sectional data showed a significant negative correlation (r = −0.34) between functional limitations and exercise behaviors, suggesting that at the start of the study higher levels of functional limitations were associated with lower levels of exercise. Longitudinal data from the study showed that increasing rates of changes in functional limitations correlated with decreasing rates of change in exercise behaviors (r = −0.25). In other words these findings are suggesting that increases in exercise behaviors correspond with decreased rates of change in functional limitations. No correlation between the initial degree of limitation and continuing rate of exercise was found which led the authors to suggest that persons with MS with varied levels of limitations might slow the trajectory of increasing limitations over the long term with consistent exercise participation.

A series of studies from Motl and colleagues have addressed the relationship between physical activity, symptoms, functional limitations and disability in MS patients. In a cross-sectional study [Motl et al. 2006] in 196 MS patients, the number of symptoms within 30 days (MS-related Symptom Checklist) and physical activity (Godin Leisure-Time Exercise Questionnaire and 7-day accelerometer data) were collected. After modeling data a direct relationship between symptoms and physical activity were found (r = −0.24) indicating that a greater number of symptoms resulted in lower amounts of physical activity. However, the authors noted that the cross-sectional design precludes inferences about the direction of causality, and physical activity might affect symptoms as symptoms affect physical activity participation. When modeled this way a moderate inverse correlation between physical activity and symptoms was found (r = −0.42) indicating fewer symptoms when the physical activity level is high. This led the authors to suggest the existence of a bi-directional relationship between physical activity and symptoms.

In a following questionnaire study Motl and colleagues examined physical activity (Godin Leisure-Time Exercise Questionnaire and 7 day accelerometer data) and symptoms (Symptom Inventory and MS-related Symptom Checklist) as correlates of functional limitations and disability (Late-Life Function and Disability Inventory) in 133 MS patients [Motl et al. 2007, 2008b]. A model based on the disablement model proposed by Nagi (1976) was tested as the primary model and this showed that physical activity and symptoms were negatively correlated (r = −0.59) and those who were more physically active had better function (r = 0.4). Furthermore, those with better function had less disability (r = 0.63) which led the authors to conclude that the findings indicate that physical activity is associated with reduced disability (through an association with function) consistent with Nagi’s disablement model (Nagi 1976), but again the cross-sectional design limited definite conclusions on the direction of the relationships.

Motl and colleagues then published a longitudinal (case report) study examining the relationship between worsening of symptoms and the level of physical activity throughout a 3- to 5-year period [Motl et al. 2008a]. The study showed that worsening of symptoms (interview) was significantly associated with lower levels of self-reported physical activity (International Physical Activity Questionnaire [IPAQ]) in a group of 51 subjects with MS. The study supports symptoms as a possible explanation for the rate of physical inactivity among MS patients but the direction of the cause and effect relationship could still not be established. Based on the results the authors suggest that managing symptoms might be important for the promotion of physical activity, but also that symptoms may be both an antecedent and consequence of physical activity.

After that Motl and colleagues published a cross-sectional study examining the correlation between physical activity and neurological impairment and disability in a group of 80 MS patients [Motl et al. 2008c]. Physical activity (7-day accelerometer day), impairment and disability (Symptom Inventory and self-reported EDSS) was measured and significant correlations were found between physical activity and both EDSS (r = −0.60) and Symptom Inventory (r = −0.56). The authors concluded that physical activity was associated with reduced neurological impairment and disability, but also stated that no causal relationship could be established due to the cross-sectional nature of the study.

Motl and McAuley then published a large-scale longitudinal questionnaire study examining the changes in physical activity (Godin Leisure-Time Exercise Questionnaire and 7-day accelerometer data) and symptoms (Symptom Inventory and MS-related Symptom Checklist) as correlates of changes in functional limitations and disability (Late-Life Function and Disability Inventory) [Motl and McAuley, 2009]. A total of 292 MS patients were followed for 6 months. Again a model based on the disablement model proposed by Nagi (1976) was tested as the primary model and this showed that change in physical activity was associated with residual change in function (r = 0.22) and change in function was associated with residual change in disability (r = 0.20). This led the authors to conclude that the findings indicate that change in physical activity is associated with change in disability (through an association with function) consistent with Nagi’s disablement model, but other models may be applied during analysis and a causal interpretation, therefore, still could not be adopted.

In a 6-month longitudinal study Motl and colleagues then tested the hypothesis that a change in physical activity (Godin Leisure-Time Exercise Questionnaire and International Physical Activity Questionnaire) would be inversely associated with a change in walking impairment (Multiple Sclerosis Walking Scale-12) in patients with relapsing–remitting MS [Motl et al. 2011a]. Data from 263 MS patients were analyzed using linear panel analysis and covariance modeling. Findings showed that a standard deviation unit change of 1 in physical activity was associated with a standard deviation unit residual change of 0.16 in walking impairment. These findings, therefore, support physical activity as an important approach, when trying to avoid walking impairments.

Finally, Motl and McAuley published a paper on longitudinal data (6 months) from 292 MS patients evaluating the relationship between a change in physical activity (7-day accelerometer data) and change in disability progression (Patient Determined Disease Steps Scale) [Motl and McAuley, 2011]. Panel analysis showed that a change in physical activity was associated with a change in disability progression (path coefficient: –0.09). This led the authors to conclude that a reduction in physical activity is a behavioral correlate (but not necessarily a cause) of short-term disability progression in persons with MS.

Recently, Tallner and colleagues evaluated the relationship between sports activity (Baecke Questionnaire – sports index) and MS relapses during the last 2 years (based on self-reports) in 632 German MS patients [Tallner et al. 2011]. Patients were divided into four groups based on their sports index. The study showed no overall differences between the four groups concerning the number of relapses within the last 2 years. However, the most active group had the lowermost mean and standard deviation of all groups. Consequently, these data suggest that exercise does not negatively influence relapse rate and the data further indicate that exercise actually reduce relapse rate.

In summary, patient-reported measures of the association between exercise or physical activity and disease progression (expressed as symptoms, functional limitations or disability) or activity (relapse rate) provide evidence of an association with more physical activity providing protection. However, due to the nature of the studies the causality of this association has not been established.

4. Disease progression evaluated in animal studies

Some obvious methodological difficulties exists in designing a human study clarifying whether or not exercise has an impact on disease progression in MS patients. Therefore, the question has been addressed in the EAE animal model of MS.

In a preliminary study by Le-Page and colleagues four groups of EAE rats were followed from day 1 to day 10 after injection with an agent inducing EAE [Le-Page et al. 1994]. The injection resulted in three different disease courses in the rats, namely acute (rats rapidly developed serious clinical signs and died without signs of recovery), monophasic (rats developed only one bout of disease followed by complete recovery) and chronic relapsing (CR-EAE, more than one bout of disease followed by remission). The CR-EAE disease course is characterized by the development of an initial acute paralytic attack 10–20 days after immunization with neuroantigens and the development of spontaneous relapses thereafter. A female and a male group of rats exercised and a female and male group served as control. Exercise consisted of running on a treadmill from day 1 to day 10 after injection. The protocol was progressively adjusted with the duration increasing from 60 min towards 120 min and the running speed increasing from 15 to 30 m/min. The study showed that in the exercised CR-EAE rats of both sexes the onset of the disease was significantly delayed compared with the onset in control CR-EAE rats. Also, the duration of the first relapse was significantly reduced in exercised CR-EAE rats compared with control rats whereas no effect was seen on the peak severity of the disease. No effects of exercise were observed in the acute and monophasic EAE rats. The authors concluded that endurance exercise during the phase of induction of EAE diminished lightly one type of EAE (CR-EAE) but also that exercise did not exacerbate the disease.

In a complementary study Le-Page and colleagues conducted further four experiments in the monophasic EAE model [Le-Page et al. 1996]. Experiments 1 and 2 showed that 2 consecutive days of intensive exercise (250–300 min/day) performed just after injection had a lowering effect on the course of the clinical signs of disease as compared with control rats. Also, the onset of the disease and the day of maximal severity were both delayed in the exercising rats, whereas no change was observed in disease duration. When the 2 consecutive days of exercise were performed before injection no effects were observed. In experiments 3 and 4 it was tested how 5 days of more moderate exercise at either constant (15–25 m/min for 2 hours) or variable speed (3 min at 2 m/min and then 2 min at 35 m/min for a total of 1 hour) affected the course of the disease and the clinical parameters. No effects were observed on the disease course and on the clinical parameters. The authors concluded that severe exercise contrary to more moderate exercise slightly influenced the effector phase of monophasic EAE, and confirmed that physical exercise performed before onset of EAE did not exacerbate the clinical signs.

More recently, Rossi and colleagues further explored the effects of physical activity on disease progression in the CR-EAE mice model [Rossi et al. 2009]. In this study one group of mice had their cage equipped with a running wheel on the day of immunization, while the control group had no running wheel. The amount of physical activity was not controlled and it was therefore the amount of voluntary physical activity in the running wheel that constituted the intervention. In a further experiment EAE mice in standard cages were compared with EAE mice in cages equipped with a blocked wheel. This was done to dissect the role of physical activity from that of sensory enrichment caused by the wheel itself, and showed not to influence the clinical course of the disease. During the initial phase (13 days after injection) of the disease the exercising mice ran spontaneously an average of 760 turns/day in the running wheel which dropped to 18 turns/day when motor impairment peaked (20–25 days after injection). The study showed that the severity of EAE-induced clinical disturbances was attenuated in both acute and chronic phases of EAE in the physically active mice, who consistently exhibited less severe neurological deficits compared with control EAE animals during a time period of 50 days after EAE induction. Furthermore, it was shown that both synaptic and dendritic defects caused by EAE were attenuated by physical activity.

In summary, aerobic exercise (or voluntary physical activity) has the potential to influence the clinical course of the disease in the EAE animal model of MS.

MS disease progression

Some major methodological problems arise when trying to measure MS disease progression. The ideal MS outcome measure would quantify irreversible sustained disease progression, but in MS this has proven difficult. The pleiotropic expression of MS makes it challenging to measure all facets of the disease and it may be necessary to focus on specific symptoms. Furthermore, great patient heterogeneity, population variability in the disease course and tempo of progression, subclinical MRI changes of uncertain impact on delayed disability progression, multifaceted neurological deficits with varied abilities for individual patients to compensate and patient comorbidities complicate things further [Goldman et al. 2010].

Clinical outcome measures

EDSS, MSFC and relapse rate are the standard clinical outcome measures for MS therapeutic trials and the most widely used measure of disease progression is the EDSS [Goldman et al. 2010]. Our literature review shows that exercise studies (resistance, endurance and combined training) applying EDSS generally do not report any change after an exercise intervention. In medical studies applying EDSS, large sample sizes and interventions lasting 2–3 years are typically required to measure changes in exacerbation rates between treatment and placebo [Bates, 2011]. This corresponds poorly to the short intervention periods (3–26 weeks) and the small sample sizes applied in most exercise studies. This is due to the overall low responsiveness and sensitivity to change of the EDSS as reported in a number of studies (for references see Goldman et al. [2010]). Also, the EDSS have been criticized for its noninterval scaling, emphasis on ambulation status and absence of adequate cognitive and visual components [Balcer, 2001]. Despite the emphasis on ambulation and that a recent meta-analysis concluded that exercise impacts walking positively [Snook and Motl, 2009], no changes were seen in the EDSS in most of the reviewed studies, indicating low scale responsiveness towards exercise interventions. In clinical trials the MSFC is claimed to be more sensitive to change than the EDSS [Goldman et al. 2010]. This suggestion is supported by the finding from one exercise study applying both the EDSS and the MSFC. In this long-term study (26 weeks) [Romberg et al. 2005] the effects of combined training on EDSS and MSFC were evaluated. Only the MSFC showed a significant effect which led the authors to conclude that the MSFC was more sensitive than the EDSS in the detection of improvement of functional impairment as a result of combined exercise. In future exercise studies evaluating disease progression it should therefore be considered to add the MSFC as a clinical outcome measure.

In addition to low scale responsiveness, short-term interventions and small sample sizes other explanations for the general lack of effects on clinical outcome measures can be hypothesized. Despite no clear pattern in the existing data, the type of exercise (e.g. endurance versus resistance training) may influence the effect captured by clinical scales. Also, most studies have evaluated mild to moderately impaired (EDSS <6) MS patients. Perhaps the clinical scales would be more sensitive to change in more severely impaired patients. Finally, findings can be biased if it is generally more physically fit patients that accept to be enrolled in exercise studies. If so, the baseline fitness level may be above average in these patients further lowering the possibility of a change on clinical scales with low responsiveness.

Only a few studies [Bjarnadottir et al. 2007; Petajan et al. 1996; Romberg et al. 2004; White et al. 2004] present clear data on relapse rate but due to the short intervention periods and the small sample sizes in most studies changes in the relapse rate, would not be expected to be evident. However, Romberg and colleagues found a total of 11 relapses (five in the combined training group and six in the control group) during a 6-month intervention period [Romberg et al. 2004]. Similarly, Petajan and colleagues (endurance training group four relapses and control group three relapses) [Petajan et al. 1996] and Bjarnadottir and colleagues (combined training group one relapse and control group one relapse) [Bjarnadottir et al. 2007] reported identical relapse rates in exercise and control groups. In the study by White and colleagues no participants experienced relapses during the 8-week intervention evaluating resistance training [White et al. 2004]. Recently, Tallner and colleagues collected self-report questionnaires on relapse rates and physical activity from MS patients to examine the relationship of different levels of sports activity and relapses [Tallner et al. 2011]. Based on these data the authors concluded that exercise had no significant influence on clinical disease activity. Taken together the few existing data do not indicate that any type of exercise increases relapse rate among MS patients. However, these data should be interpreted with caution due to the small number of participants (not stratified according to disease type or severity) and the short intervention periods in most studies. Consequently, future long-term studies with a large number of participants should, therefore, include relapse rate as an outcome measure.

Nonclinical measures

Application of MRI has revolutionized the diagnosis and management of patients with MS [Bar-Zohar et al. 2008]. In regard to clinical trials, MRI offers several advantages over the accepted clinical outcome measures for MS, including an increased sensitivity to disease activity and a better association with histopathology findings. Also, MRI provides highly reproducible measures on ordinal scales, and the assessment of MRI can be performed at the highest degree of blinding [Bar-Zohar et al. 2008]. Consequently, a surrogate MRI measure reflecting disease progression such as lesion activity (gadolinium-enhanced lesions and new or enlarged T2-hyperintense lesions) or disease severity (total T2-hyperintense lesion volume, total T1-hypointense lesion volume and whole-brain atrophy) [Bermel et al. 2008] may reduce the required sample sizes needed to evaluate the effects of exercise therapy on disease progression considerably. Until now only two cross-sectional studies have evaluated the effects of exercise (expressed as the current cardiorespiratory fitness level) on different MRI measures limiting the conclusions that can be drawn from this type of study. However, the promising findings do encourage the inclusion of MRI as an outcome measure, in future longitudinal trials evaluating the effects of exercise on disease progression.

Patient-reported measures

Patient-reported measures of the association between exercise or physical activity and disease progression (expressed as symptoms, functional limitations or disability) provide evidence of an association with more physical activity providing protection. However, the nature of the studies does not allow conclusions on the causality of this association. In the group of studies applying patient-reported measures we decided to include not only measures of exercise, but also measures of physical activity. It is acknowledged that a measure of physical activity is not necessarily a surrogate measure of exercise, but the many interesting findings from particularly the group of Motl and colleagues caused this. In a recent paper, based on their own studies, Motl and colleagues concludes that recent research has identified physical activity as a behavioral correlate of disability in MS. This made the authors suggest, that physical activity might attenuate the progression of what they call ‘mobility disability’ by improving physiological function in persons with MS, particularly those who have achieved a benchmark of irreversible disability (EDSS >4) [Motl, 2010]. It might be more cost effective to offer the more disabled (EDSS >4) MS patients exercise therapy, but it must be noted that most exercise studies do not indicate that a relationship between the degree of training adaptation and neurological disability exist. In fact, studies indicate that MS patients with an EDSS score below 4.5 experience the largest improvements after a period of endurance training as compared with more disabled MS patients [Ponichtera-Mulcare et al. 1997; Schapiro et al. 1988] or that no differences exists [Petajan et al. 1996]. It must be noted that none of these studies were powered to evaluate the effects of exercise in MS patients with different levels of disability. However, a recent study by Filipi and colleagues specifically evaluated whether 6 months of resistance training improves strength in MS patients with different levels of disability (EDSS 1–8) and concluded that all individuals with MS, despite different disability levels, showed parallel improvement in muscle strength [Filipi et al. 2011]. This leads to the suggestion, that exercise may be equally important during the early phases of the disease, also in regard to impact on disease progression.

An important advantage of applying patient-reported measures is the opportunity to collect data from large sample sizes in longitudinal studies. Furthermore, it seems important to collect data on patient perspective when evaluating the effects of exercise on disease progression. Future studies including patient-reported measures should also include clinical and/or nonclinical outcome measures if possible.

Animal studies

Our review showed that aerobic exercise (or activities) has the potential to influence the clinical course of the disease in the EAE animal model of MS. The obvious question is whether or not the findings from the EAE animal model of MS can be extrapolated to humans. At the moment no clear answer can be given to this question. A recent review summarized whether the current disease-modifying treatments are justified on the basis of the results of EAE studies. Here it was concluded that although EAE is certainly an imperfect mirror of MS, many clinical, immunopathological and histological findings are impressively replicated by animal models, making EAE invaluable in elucidating the basic immunopathological mechanisms of MS and providing a testing ground for novel therapies [Farooqi et al. 2010]. Consequently, a direct transfer of findings into human subjects cannot be made, but testing of difficult hypotheses can start here. Also, it should be noted that in EAE you cannot control the relative exercise intensity since no maximal exercise test (such as a VO₂ max test) can be performed. As a consequence the applied relative exercise intensity may differ between animals. This is also why it is very difficult to evaluate the effects of aerobic exercise on aerobic capacity in EAE. Nonetheless, the EAE model offers a number of advantages compared to human studies. In addition lower costs, easy control with adherence to the intervention and controlled environmental and genetic factors the EAE model also allows evaluation of possible mechanisms located in the central nervous system (CNS), which should have attention in future studies. Another review stated that the genetic heterogeneity, which is so critical in the MS population, is only reflected when multiple different models of EAE are studied in parallel [Gold et al. 2006]. This aspect should also be incorporated in future studies.

Possible mechanisms

Several mechanisms have been proposed as a possible link between exercise and disease status in MS. Some of the most promising candidates include cytokines and neurotrophic factors [White and Castellano, 2008a].