Structured exercise training is a cornerstone of MS treatment, but its effect on NEPA levels of PwMS is unclear. The present data support our hypothesis that PwMS change their NEPA in response to exercise. More specifically, a reduction in MVPA was observed after 5 and 10 months (-113 ± 31 min/week and − 95 ± 26 min/week), which approximately matched the weekly duration of the exercise sessions at those time points (114 ± 9 min/week and 105 ± 9 min/week). Furthermore, PwMS also increased their uninterrupted SB on days without exercise session compared to days with exercise session (+ 0.7 ± 0.3h). These NEPA/SB changes were not seen in HC, indicating that these behavioural changes are MS-specific. Previous systematic reviews also found that healthy adults and overweight/obese persons did not change their NEPA/SB in response to prescribed exercise[25, 26].
It is intriguing to speculate why the present NEPA changes seem to be MS-specific. Firstly, there were no associations between any of the MS characteristics (EDSS score, MS duration and walking mobility) and NEPA/SB changes. It is however important to note that this might also be due to the low variation within the MS characteristics, as the included PwMS were only mildly disabled and had a rather short MS duration. Secondly, fatigue, reported to be the most frequent and debilitating MS symptom[27], was proposed by King et al. as important explanatory factor for inter-individual variability in NEPA changes in healthy persons[28]. In the PwMS of the present study, a large variation in fatigue was measured (range MFIS questionnaire: 2–64 points), but no correlations were present between fatigue at any time point and NEPA or SB changes. Furthermore, fatigue significantly improved with 20.5% in the MS group. This implies that fatigue did not cause the NEPA/SB changes. However, although the MFIS questionnaire assesses multidimensional fatigue, it only captures the average fatigue of the last four weeks. This limits conclusions on acute fatigue and especially on exercise-induced fatigue/tiredness. Smith et al. reported that fatigue does not increase immediately nor 24h after exercise in PwMS, but only a single exercise session was included and the intensity was rather low (Borg scale 12–14/20)[29]. Moreover, an observational study on symptomatic fatigue in PwMS (n: 309) reported that 64% of participants reported worse fatigue after “moderate exercise”, while this was the case for 83% after “vigorous exercise”[30]. Although sub analyses of the current results showed that exercise intensity (HIIT versus MICT sessions) had no effect on the NEPA/SB changes, the overall demand of 2–3 exercise sessions per week at moderate-to-vigorous intensity might require necessary NEPA/SB compensations in order to conserve energy to adhere to the exercise intervention. Such energy conservation strategies are often taught to PwMS to prevent or treat fatigue[31]. However, it might also be true that PwMS implemented dysfunctional energy conservation strategies, which were not adapted accordingly to their improving fatigue and cardiorespiratory fitness levels. This was not investigated in the present study, but this indicates that fatigue and the implementation of energy conservation strategies might need to be monitored and possibly revised throughout an exercise intervention.
Another plausible explanation might be that the current PwMS could not maintain their baseline PA levels in addition to doing more exercise, because their baseline PA was already high (mean ± SD: 41 ± 23min MVPA/day and 9782 ± 2569 steps/day [data not shown]). However, PwMS in previous research had clearly lower baseline PA levels (mean ± SD: 26 ± 18min MVPA/day, 4488 ± 2251 steps/day and 6095 ± 2363 steps/day), but their total PA also did not increase while they self-reported to do more exercise[10, 11]. In a study of Keadle et al.[32], overweight/obese participants had even higher baseline MVPA (mean ± SD 50 ± 17min/day) than the present PwMS, and they were able to further increase this during an exercise intervention when they also received education on SB and self-monitoring of non-exercise MVPA. This might also be possible in PwMS when they receive a multicomponent intervention targeting both exercise PA and NEPA/SB and should be explored in future research.
The exercise intervention effectively improved the cardiorespiratory fitness, body composition, fatigue and resting HR in both groups, with no difference between groups. This is in line with previous periodised exercise interventions in PwMS and HC[13, 33]. However, this might seem contradictory, because PwMS changed their NEPA and SB whilst HC did not. Moreover, the non-exercise MVPA reduction approximately matched the exercise time, which would result in an unaffected net MVPA change. The fact that exercise effects were not constrained, might be explained by a difference in PA intensity. More specifically, the exercise intensity was probably higher (78.7 ± 6.1% of HRmax) compared to that of the reduced non-exercise MVPA, limiting the impact on exercise outcomes. Unfortunately, it is not possible to draw strong conclusions on the exact intensity of the reduced NEPA, because all walking and running activities with ≥ 100 steps/min are classified as MVPA by the activPAL algorithm. It might be relevant to include continuous heart rate monitoring in future research in order to measure the exact NEPA intensities.
Additionally, the present PwMS were only mildly disabled and already very physically fit at baseline (on average > 100% of their predicted cardiorespiratory fitness). It could be hypothesised that more disabled or less fit PwMS show larger NEPA/SB compensations in response to exercise, which may in turn have negative effects on the eventual exercise adaptations. Furthermore, other relevant physical exercise outcomes that were not measured, might have been impacted. In a study of Keadle et al.[34], the effect of changes in NEPA in overweight/obese participants was also assessed on fasting lipids and insulin sensitivity. Interestingly, insulin sensitivity only improved within participants who performed exercise and also reduced their SB (-5%; which was replaced with MVPA and LIPA), not in participants who only performed exercise (with no changes in NEPA). It might be especially relevant to monitor the effect of NEPA and SB changes on insulin sensitivity in PwMS in future research, because they already have a higher risk of developing insulin resistance (x2.48 compared to HC)[35], which is associated with a worsening of disability[8, 35, 36]. Furthermore, the current findings also show that PwMS significantly increased their uninterrupted SB in response to exercise. This has already been shown to be negatively associated with insulin sensitivity in a large sample of HC (n: 4935), independent of total MVPA and SB[37]. Hence, measures of insulin sensitivity should be included in future NEPA research in PwMS.
The present study has several strengths. Firstly, PA and SB were objectively assessed for seven consecutive days on three different time points throughout the study, and a distinction was made between days with and without exercise. Secondly, the Polar watches provided objective adherence rates in a home-based setting, which enabled a PP-analysis with participants that effectively trained. This allows us to conclude that the observed NEPA/SB changes occur when PwMS effectively perform the prescribed exercise sessions. Furthermore, a post-hoc power calculation showed a similar effect size for the observed improvement in cardiorespiratory fitness as what was a priori calculated (although the eventual sample size was smaller). Lastly, this is the first study that compared NEPA changes between PwMS and HC. An unfortunate limitation of the study is that T2 and T3 measures were conducted during the COVID-19 pandemic. Although the PA assessments at both time points occurred during periods with only minor COVID-related restrictions, PwMS might have been more afraid to get infected due to the autoimmune nature of their disease, and as such reduced their social and/or outdoor activities to a greater extent than HC. It is however important to note that our findings are in line with studies that were conducted before the COVID-19 pandemic. Furthermore, there was no randomisation in the present study, but our findings in probably very motivated PwMS highlight the relevance of further research to NEPA/SB changes with more rigorous study designs. Finally, the intervention consisted of a running program, which limits the generalisation of results to more disabled PwMS with walking impairments. Future research on NEPA changes in more deconditioned and/or disabled PwMS with other (non-weight bearing) intervention types such as (recumbent) cycling is warranted.