Vigorous cool room treadmill training to improve walking ability in people with Multiple Sclerosis using ambulatory assistive devices: A feasibility study.

Background: Aerobic training has the potential to restore function, stimulate brain repair, and reduce inflammation in people with Multiple Sclerosis (MS). However, disability, fatigue, and heat sensitivity are major barriers to exercise for people with MS. We aimed to determine the feasibility of conducting vigorous harness-supported treadmill training in a room cooled to 16°C (10 weeks; 3times/week) and examine the longer-term effects on markers of function, brain repair, and inflammation among those using ambulatory aids. Methods: Ten participants (9 females) aged 29 to 74 years with an Expanded Disability Status Scale ranging from 6 to 7 underwent training (40 to 65% heart rate reserve) starting at 80% self-selected walking speed. Feasibility of conducting vigorous training was assessed using a checklist, which included attendance rates, number of missed appointments, reasons for not attending, adverse events, safety hazards during training, reasons for dropout, tolerance to training load, subjective reporting of symptom worsening during and after exercise, and physiological responses to exercise. Functional outcomes were assessed before, after, and 3 months after the completion of training. Walking ability was measured using Timed 25 Foot Walk test and on an instrumented walkway at both fast and self-selected speeds (stance (%), swing (%), double support (%)). Fatigue was measured using standardized questionnaires (fatigue/energy/vitality sub-scale of 36-Item Short-Form (SF-36) Health Survey, Fatigue Severity Scale, modified Fatigue Impact Scale). Aerobic fitness (maximal oxygen consumption) was measured using maximal graded exercise test (GXT). Quality-of-life was measured using SF-36 Health Survey. Serum levels of neurotrophin (brain-derived neurotrophic factor) and cytokine (interleukin-6) were assessed before and after GXT. Results: Fast walking speed (cm/s), gait quality (double-support (%)) while walking at self-selected speed, fatigue (modified Fatigue Impact Scale), fitness (maximal workload achieved during GXT), and quality-of-life (physical functioning sub-scale of SF-36 Health Survey) improved significantly after training, and improvements were sustained after 3-months. Improvements in fitness (maximal respiratory exchange ratio and maximal

Conclusion: Vigorous cool room training is feasible and can potentially improve walking, fatigue, fitness, and quality-of-life among people with moderate to severe MS-related disability.

Background
Multiple Sclerosis (MS) is a chronic disease of the central nervous system (CNS), affecting approximately 2.3 million people worldwide (1). MS is characterized by acute inflammatory episodes in the CNS, often transitioning to a progressive neurodegenerative phase (1). About 80% of those who live with MS will develop the progressive form during their lifetime (1). Cellular mechanisms contributing to neurodegeneration in MS include lack of trophic support to neurons and glia, chronic microglial activation, and mitochondrial injury induced by oxidative stress (2). Several studies in animal models of MS suggest that exercise has direct protective and restorative effects by interacting with these mechanisms (3)(4)(5). Evidence suggests that aerobic training promotes neuroplasticity by upregulating neurotrophins such as brain-derived neurotrophic factor (BDNF) and insulin-like growth factor (IGF-1) (6)(7)(8)(9). Further, aerobic exercise could have direct effects on the neuro-immune axis in MS (7,10). A systematic review of evidence suggested that aerobic training significantly altered peripheral levels of cytokines, interleukin (IL) 6, IL-10, interferon-gamma, and tumor necrosis factoralpha (11). Whether aerobic training has the potential to affect multiple underlying targets such as enhancing markers of neuroplasticity by upregulating neurotrophins and attenuating neural inflammation by altering levels of cytokines is not clear (12,13). Since aerobic exercise performed on a treadmill also provides a high volume of task-specific practice, aerobic treadmill training has the potential to improve walking ability, fitness, and quality of life (12)(13)(14).
Although aerobic training is a promising rehabilitative strategy for MS, aerobic exercise increases metabolic rate by 5 to 15 times above resting state and heat produced by contracting muscles elevates core body temperature, which in turn acts as a barrier to exercise participation (15)(16)(17).
Paroxysmal or fleeting MS symptoms, such as pins and needles that persist for few seconds to minutes, often occur as a result of a temperature-dependent conduction block in demyelinated axons, triggered by an increase in body temperature (18). Impaired regulation of body temperature is a major barrier to exercise for people living with MS (19,20). In a cross sectional study, heat sensitive individuals with MS had simultaneous increase in core temperature and worsening of MS symptoms during aerobic exercise, when compared to resisted exercise (21). Furthermore, higher internal body temperature caused fatigue, even in trained participants during aerobic exercise sessions (22).
According to Allen et al. (23), people with MS experience attenuated sweating response, which may lead to a temporary worsening of disease symptoms and limit exercise tolerance under more thermally challenging conditions. Our previous research showed that cooling the exercise environment to 16°C, mitigated exercise-induced losses in central drive among people with MS who reported having heat sensitivity (24). Therefore, we aimed to determine the feasibility of conducting a vigorous aerobic walking training in a room cooled to 16°C using bodyweight supported treadmill (BWST) for people with MS who used ambulatory assistive devices, wheelchairs, and mobility scooters. We examined both the immediate and longer-term (at 3-month follow-up) impacts of training on walking speed, gait parameters, fatigue, aerobic fitness, and quality of life. We also examined in a preliminary way, whether the intervention would alter blood biomarkers of neuroprotection (BDNF) and inflammation (IL-6). BDNF and IL-6 were chosen as proxy indicators of neuroprotection and inflammation respectively because preliminary experiments showed that they were potential rehabilitative markers for people with MS having severe walking disability (25).

Methods 1.1 Design
This was a repeated measures feasibility study with a non-randomized single arm aimed to examine the feasibility and preliminary effects of the intervention. This study was approved by the instructed to walk two laps on the walkway at self-selected walking speed to measure speed and spatiotemporal parameters. For all walking assessments, participants were provided with standardized instructions and used their ambulatory devices. If the participants required additional assistance while walking, they were assisted using a gait belt by a member of the research team, who was a physiotherapist. Gait parameters (stance phase (%), swing phase (%), double support phase (%), and walking speed (cm/s)) were extracted from the walkway as previously described (40).

Fatigue
Fatigue was assessed using three methods:

Aerobic fitness
Fitness was assessed using maximal GXT on a seated recumbent stepper (48). All participants were asked to exercise at increasingly difficult levels while wearing a facemask to measure how much oxygen they consumed (Moxus Metabolic Systems; AEI Technologies, Inc.). A heart rate monitor was placed on their chest to measure maximal heart rate achieved during GXT. Participants were verbally encouraged to exercise as long as they could, and the workload was increased in ~20-watt increments every 2 minutes, starting from load level 3 (21 watts) until exhaustion (48

Serum analysis
Blood was collected from the median cubital vein at three testing time points (pre, post, and followup) immediately before and after GXT in two 5mL serum vacutainers (50). The samples were left to clot for 30-60 minutes, centrifuged at 2200g for 10 minutes, and the collected serum was stored frozen at −80 °C until assayed. Serum levels of neurotrophin (BDNF) and cytokine (IL-6) were measured using ELISA kits for human BDNF (R&D Systems Inc. Minneapolis, Minnesota, USA) and IL-6 (BD Biosciences, San Diego, California, USA) as per manufacturer's instructions.

Intervention
All participants underwent a personalized, progressively intense, moderate to vigorous intensity (40-65% heart rate reserve (HRR) (51)) training for ten weeks (3x/week) in a temperature-controlled room (16°C) starting at 80% self-selected walking speed on a BWST equipped with safety straps to prevent falls. Exercise intensity was estimated using resting and maximal heart rates measured before and during GXT respectively at baseline (Exercise heart rate = % target intensity (maximal heart rateresting heart rate) + resting heart rate) (49,52). Each training session lasted up to 40 minutes, including 5 minutes of warm-up and cool-down. A gradual progression of workload was undertaken to minimize muscle injury (53), starting with moderate intensity (40% HRR) at 80% self-selected speed, and progressing to vigorous intensity (65% HRR) at gradually increasing walking speed as tolerated (51), with simultaneous reduction of bodyweight support provided on the treadmill from 10% to 0% and increase of treadmill incline from 1% to 10%. For individuals with severe walking impairment, manual support was provided to advance the weaker lower extremity as required.

Data analysis
Variables were assessed if they met assumptions of non-parametric statistics. Statistical analyses (Friedman rank test followed by Wilcoxon matched-pairs signed-rank tests with Bonferroni alpha correction (0.05/3=0.01)) were then conducted to determine the effects of cool room BWST training.
Missing data were not imputed but were excluded pairwise due to small sample size and its concurrent higher variance. The relationships between the outcome measures were estimated by Spearman's rank correlation coefficient (r s ). When multiple correlations were conducted, a post-hoc correction was performed using the Bonferroni method (54,55). Clinically meaningful changes, both individual as well as a group, were determined for participants who attended all three testing time points using cut-off values published previously in the literature. passed PAR-Med-X, and were thus enrolled (n=10) in the study (Table 1). Recruitment was stopped prior to reaching enrollment goal (n=11) due to slow accrual and difficulty finding patients who were willing to participate in the training program. All participants (n=10) identified themselves as having fatigue and sensitivity to heat. Ten participants (9 females), aged 29 to 74 years, with EDSS ranging from 6.0 to 7.0 completed the baseline assessments following which, two dropped out of the study (after completing 2 and 7 sessions respectively), and eight participants continued to participate in the exercise training sessions (range, 24 to 30 sessions) ( Table 1). Eight participants (7 females) completed the 10-week exercise training and completed the assessments immediately after the training program. Three months after exercise training, seven participants (6 females) returned to complete the follow-up assessments.

Attendance rates and reasons for missed appointments
The attendance rates ranged from 80% to 100% among those who completed exercise training and the total number of missed appointments ranged from 1 to 6 per participant ( Table 1). The reasons for missing appointments were feeling tired or unwell (n=15), transportation issues (n=5), having medical appointments (n=4), personal scheduling conflict (n=4), leg pain and stiffness (n=3), inclement weather (n=3), recent fall (n=2), and forgot appointment (n=1). Participants rescheduled the missed appointments and continued to participate in the exercise sessions (Table 1).  The intervention was laboratory-based in a rehabilitation hospital setting; therefore, the researchers relied on physicians-on-call for emergencies. No adverse events (MS relapse, syncope, or medical emergencies) occurred during assessments and training sessions. One participant required electrocardiograph monitoring by the physician during GXT due to a history of arrhythmia. The GXT was terminated due to high systolic blood pressure (>220 mmHg); however, the participant was admitted into the study after clearance from the physician. Participants wore a safety harness during all training sessions and no safety hazards were identified.

Reasons for dropout
The participants were provided a clear option to drop out, if necessary, without having to provide any reason. One participant dropped out after attending two exercise training sessions as she complained of fatigue and felt unsafe to drive back home after exercise. Another participant dropped out after attending seven sessions as per physician's advice after beginning a new MS medication.

Training load and tolerance
All participants were able to perform progressively intense BWST training from moderate to vigorous intensity (40-65% HRR) (51), however participants did not have significant change in resting heart rate after training (p=0.289). Eight out of 10 participants were able to walk on the treadmill at 80% of their self-selected overground walking speed from the first exercise session onwards. One participant was able to start training at 60% and another at 40% of their respective self-selected overground walking speeds. All participants, but one, were able to walk on the treadmill with 10% body weight support from the first exercise session. Three participants were able to completely wean off to 0% body weight support over ten weeks. Three participants required manual assistance to advance their lower extremity during initial treadmill training sessions, which was weaned off gradually. The total time walked, and distance covered progressively increased while the total time required to rest decreased ( Figures 1A, B, C, and D). The participants were advised to take breaks in either sitting or standing position on the treadmill as required during training sessions ( Figure 1D). There was an overall increase in workload performed and oxygen consumed in both unilateral and bilateral walking aid users ( Figures 1E and F). Subjective reporting of symptoms and physiological response to exercise All participants were able to tolerate the cool room training with the air-conditioning set at 16°C. Two participants reported having mild symptoms, such as pins and needles sensations, that were fleeting for a few seconds or minutes during training sessions. Two participants reported having weak legs while walking on the treadmill. One participant complained of shoulder ache after bearing body weight through arms and requested greater body weight support. One participant had leg pain that resulted in the termination of one of the training sessions. None of the participants reported exacerbation of MS symptoms, such as the occurrence of motor weakness, ataxia of a limb, or any other MS symptoms, that lasted more than 24 hours after training sessions (56,57). Tympanic temperature, heart rate, and fatigue increased with exercise, while mean arterial pressure remained stable (Figure 2A, B, C, and D).   (Table 3). However, 4 out of 8 participants made a clinically meaningful change (>20%) after training ( Figure 3A) (58)(59)(60).
In regards to fast walking speed measured on the instrumented walkway (cm/s), speed increased by 15.5% (p=0.012), which was sustained at follow up compared to pre assessment (p=0.043) ( Table 3).

Self-selected walking speed
There was no significant change in self-selected walking speed (cm/s) (measured on an instrumented walkway) at both post and follow up (p values, 0.674 and 0.063 respectively) (Table 3). However, 6 out of 8 participants made a clinically meaningful change of more than 12% beyond the benchmark accepted for walking assessments in MS ( Figure 3B) (61,62).
There was no significant change in stance and swing phases (%) while walking at self-selected speed (p values, 0.093 and 0.093 respectively), however total double support phase (%) was significantly reduced at post compared to pre (p=0.036) ( Table 3). Duration of the stance phase (%), swing phase (%), and total double support phase (%) while walking at self-selected speed improved significantly at follow up compared to pre (p values, 0.018, 0.018, and 0.018) ( Table 3).

Fatigue
Participants rated three aspects of fatigue, (1)  Severity of fatigue reported on FSS (total score) was not significantly different at post or at follow up compared to pre (p values, 0.123 and 0.345 respectively) ( Table 3). However, 4 out of 8 participants achieved a change of 1.9 or more points on mean FSS scores at post, a minimal detectable clinically meaningful change for people with MS ( Figure 3E) (45).
Impact of fatigue reported on mFIS was significantly less (p=0.017) at post, which was sustained at follow up compared to pre (p=0.034) ( Table 3) ( Figure 3F). However, only 1 out of 8 participants had a clinically meaningful change beyond the accepted benchmark of 20.2 points at post, and two at follow up ( Figure 3F) (45).

Aerobic fitness
There was no statistically significant change in maximal VȮ 2 and maximal heart rate achieved during GXT at post compared to pre (p values, 0.484 and 0.078 respectively) ( Table 3) ( Figure 3C). However, the participants were able to achieve a greater workload during GXT at both post and follow up compared to pre values (p values, 0.012 and 0.043, respectively) ( Table 3). The oxygen uptake efficiency slope, a measure of the cardiorespiratory reserve, significantly increased at post (p=0.049), which was sustained during follow up (p=0.735) ( Table 3) (64,65).
In terms of indicators of achievement of a maximal GXT, four out of 10 participants achieved two or more criteria for test termination at pre, 3 out of 8 at post, and 3 out of 7 at follow up (48,49

Blood markers
We collected blood samples from participants immediately before and after GXT, conducted during pre, post, and follow up testing time points. We were unable to draw blood samples from three participants on 6 out of 52 occasions. All serum BDNF levels were within the detectable ranges.
Serum IL-6 levels were not detectable in seven participants on 22 out of 46 occasions.

Neurotrophins
In terms of serum BDNF, there were no significant differences in resting and exercise-induced levels (After minus Before GXT) measured at pre ( Figure 4A), post ( Figure 4B), and at follow up ( Figure 4C) (p values, 0.223 and 1.0 respectively) ( Table 3). There was a significant decrease in serum BDNF after GXT compared to before GXT levels, both at pre ( Figure 4A) and follow up ( Figure 4C) (p values, 0.036 and 0.028 respectively), but not at post-training (p=0.310) ( Figure 4B).

Cytokines
In terms of serum IL-6, there were no significant differences in resting and exercise-induced levels (After minus Before GXT) measured at pre ( Figure 4D), post ( Figure 4E), and at follow up ( Figure 4F) (p values, 0.282 and 0.368) ( Table 3). There was no significant change in serum IL-6 after GXT compared to before GXT levels, at pre ( Figure 4D   The improvement in fast walking speed was associated with reduced fatigue measured using physical subcomponent score of mFIS (Spearman's rank correlation coefficient, r s =-0.847, p=0.008) ( Figure   5A). The improvement in fatigue measured using total mFIS score was related to higher maximal respiratory exchange ratio achieved during GXT (r s =-0.810, p=0.015) ( Figure 5B). The improvement in maximal respiratory exchange ratio achieved during GXT was associated with an increase in resting serum BDNF (r s =0.786, p=0.036) ( Figure 5C). The improvement in fitness measured using maximal VȮ 2 was associated with a decrease in resting serum IL-6 (r s =-0.757, p=0.049) ( Figure 5D). However, after correcting for multiple correlations (0.05/7=0.007) (54,55), none of the relationships were statistically significant.

Discussion
We tested a novel cool room intensive treadmill training for people with moderate to severe walking disability in the progressive phase of MS or transitioning to the progressive phase. We found that this intervention was feasible, and most participants achieved clinically meaningful improvements in walking, fatigue, fitness, and quality of life.

Feasibility of vigorous cool room training in MS
In 1890, Uhthoff (68) reported that patients with MS had exercise-induced amblyopia, a phenomenon later discovered to be due to an increase in body temperature (18). Nearly 60 years later, Watson  (75). In fact, a recent meta-analysis concluded that while precooling lowered the finishing core temperature, concurrent cooling methods that affected a large body surface area contributed to a large positive effect on exercise performance (76). Our study marks the first attempt to examine the effects of the whole-body concurrent cooling method as therapy for people living with MS having barriers to exercise participation such as walking disability, fatigue, and heat sensitivity.
There is an urgent need to develop new therapies and exercise-based rehabilitation treatments that could potentially stabilize or even improve MS symptoms, especially among people who have accumulated substantial disability. In our study, we have demonstrated the feasibility of conducting a progressively intense (moderate to vigorous) aerobic walking training strategy with concurrent cooling (16°C cool room) using BWST for people with MS requiring ambulatory assistive devices, wheelchairs, and mobility scooters. We measured physiological responses across training days ( Figure   2), to characterize exercise-induced changes during training sessions (77). We noted an exerciseinduced increase in tympanic temperature, heart rate, and fatigue; however, mean arterial pressure remained stable during all training sessions (Figure 2). At the end of the training, our participants experienced improved energy levels measured using SF-36 (36.4%) when compared to the conventional benchmark (11% to 20%) (63) which was higher than that previously reported following robot-assisted gait training (16%) (78). Furthermore, participants who had greater improvements in fatigue walked faster ( Figure 5A) and also achieved a higher maximal respiratory exchange ratio during GXT after training ( Figure 5B). There were no adverse events (MS relapse, syncope, or medical emergencies) in our cohort, except for fleeting symptoms of neurologic origin, such as pain, pins and needles, and weak legs, which lasted less than 24 hours. Whether such fleeting symptoms occurred due to heat (exercise-induced increase in core temperature) sensitivity in our participants is unknown, and further research is required to confirm the physiological mechanisms of worsening MS symptoms during training (79). It was interesting to note that although participants began their training at 80% of their walking speed, three required manual assistance within just a few minutes during initial sessions, supporting that fatigue and leg weakness are major impediments to effective exercise interventions (80, 81).  (84). Furthermore, the gains in fast walking speed were sustained at 3-month follow-up assessment (Table 3), whereas previous examination of robot-assisted gait training showed that training gains (when measured using 10meter walk test) were lost three months later (85). Lastly, our participants walked considerably faster at self-selected speed overground at 3-month follow up compared to post-training assessment (Table   3). Although there is no consensus as to what value constitutes a clinically meaningful change in selfselected overground walking speed, a change of 12% to 20% in related walking tests is indicative of a meaningful change in MS (61). Considering this criterion, 6 out of 8 participants made more than 12% improvement on self-selected walking speed post-training, which was sustained in 3 out of 7 participants during 3-month follow up (7.41 cm/sec increase) ( Figure 3B) ( Table 3). These findings were in contrast with robot-assisted gait training in which participants had a 7.0 cm/s increase immediately after training but returned to pre levels three months after training (86). Additionally, spatiotemporal gait parameters at self-selected pace were improved well beyond previous reports employing non-gait specific training methods (legs and trunk resistance training twice a week for eight weeks) (87), supporting that gait quality also improved.

Ability to perform GXT and improvements in cardiorespiratory reserve
We showed that patients with high levels of disability were able to complete GXT in 10 out of 25 occasions (40% success rate). To our knowledge, we are the first to assess the feasibility of conducting GXTs on a recumbent stepper among patients with high levels of MS-related disability (28, 78,85,86,[88][89][90][91][92][93][94]. We found that our participants with high disability achieved 12.2% greater maximal workload during GXT as a result of training, despite small increases in maximal heart rate (1.8%), maximal respiratory exchange ratio (2.2%), and maximal VȮ 2 (6.6%). A meaningful change in maximal VȮ 2 due to an exercise training has been estimated at 0.540 L/min (18.9%) in healthy individuals (95). Our participants obtained 0.061 L/min increase on average; a 6.6% increase after training, which further increased to 8% at 3-month follow up. We note that the meaningful change criterion for maximal VȮ 2 estimated for healthy individuals may not be relevant for our participants with severe MS-related disability (95). Furthermore, there is no known clinically meaningful change benchmark for maximal VȮ 2 applicable to people with MS having severe deconditioning (96). For patients with severe MS-related disability (EDSS 6.5), the measurement of oxygen uptake efficiency slope is an alternative (sub-maximal) method to express cardiorespiratory fitness when maximal exercise testing is not feasible (64,65). Originally, this relative measure of cardiorespiratory work during GXT was validated in individuals with low, mild, and moderate disability (EDSS <6.0) who reached 90% of their age-predicted maximal heart rate during GXT (64,65). In our study (EDSS 6.0 to 7.0), only about 50% of our participants achieved 90% of their age-predicted maximal heart rates during GXT performed at pre, post, and follow-up. Heine and colleagues (64,65), determined the concurrent validity of oxygen uptake efficiency slope (how well a new test correlates with a previously validated measure) in those with EDSS <6.0 and reported a significant correlation with maximal VȮ 2 and maximal workload achieved during GXT (p values, <0.05). Similarly, we noted that the oxygen uptake efficiency slopes were higher in those who had higher maximal VȮ 2 and maximal workload achieved during GXT (p values (not reported), <0.05) at all three testing time points. In our study, the increase in oxygen uptake efficiency slope during GXT both immediately (12.7%) and 3-month after training (13.5%), could likely be attributed to a combined improvement of cardiovascular, musculoskeletal, and respiratory functions (64,97). Future studies should examine the links between improvements in cardiorespiratory reserve, walking speed, and health-related quality of life following training in people with advanced MS (96).

Improved health-related quality of life
Overall, we noted a clinically meaningful improvement in the quality of life (i.e., more than a 3-point increase in all SF-36 domains except social functioning immediately after training compared to baseline) (Table 3) (66,67). Furthermore, improvements in physical functioning, bodily pain, and perception about health compared to last year were significantly improved after training compared to baseline, which was sustained 12 weeks after the intervention ceased (Table 3). When comparing our results to others, robot-assisted gait training of shorter duration (6 weeks, 2 sessions/week) failed to improve SF-36 subcomponents, physical functioning, and bodily pain after training (78). Likewise, robot-assisted gait training for four weeks (3 sessions/week; 12 sessions) made no change in physical and mental health measured using SF-36 at post, 3-month, and 6-month follow up assessments (85).
Although Vaney,Gattlen (93) reported that nine sessions of 30-minute robot-assisted training added to intensive strengthening exercises resulted in significant improvements in quality of life, the benefits were not sustained at follow up. One study examining a longer duration program (12 weeks) of BWST training reported improvements in both physical and mental health composites measured on MS Quality of Life-54 scale, but the sustainability of benefits were not examined at follow-up (88).
Sustained improvement in quality of life (physical functioning), like that observed in our study, suggests that the benefits gained with vigorous cool room BWST training resulted in improved walking ability even after cessation of training, rather than simply short-term performance enhancement, which was meaningful for the participants.

1.5
Vigorously aerobic cool room training might have the potential to affect multiple underlying mechanisms Aerobic exercise is an intervention that has both neuroprotective and anti-inflammatory benefits (98,99). Evidence suggests that progressively intense, aerobic training performed 2 or 3 times per week for at least 8 to 9 weeks could improve walking ability as well as result in a trend towards an increase in resting BDNF levels in people with MS (12) and in other neurological disorders such as stroke (100).
Similarly, our participants with MS experienced statistically significant improvement in walking ability, and 6 out of 7 participants had an increase in resting levels of serum BDNF after training. Further investigation is required to determine whether a simultaneous increase in walking ability and resting serum BDNF levels would result in clinically meaningful restoration of function in MS. With regards to resting serum IL-6 measurements in our study, there were fewer number of data points above minimum detectable limits to glean any meaningful trends ( Figure 5D). However, the improvement in maximal VȮ 2 was associated with a decrease in resting serum IL-6 levels after training ( Figure 5D). Kosaka, Sugahara (101) demonstrated that whole-body cold air exposure significantly suppressed inflammation, specifically the pro-inflammatory cytokines, IL-6 and IL-1beta. Given the preliminary data from our study, further studies examining the synergistic effects of cold air exposure and treadmill training on suppression of inflammation are necessary to understand the molecular mechanisms of recovery.

Limitations
Despite the fact that this is the first report of vigorous cool room BSWT training among people with progressive MS, there are several limitations to consider. First of all, our study had a very small sample size, thus limiting statistical power to obtain conclusive results. Seven of the 37 potential participants we contacted did not wish to participate in such an exercise program, and two of the 10 participants dropped out. This suggests that the vigorous cool room treadmill training method is not acceptable to about 20% of people who are eligible. Secondly, we were unable to complete blood draws in some subjects, and it appeared that hypo-hydration could have been a factor. Although we did not determine whether participants had bladder problems, about 80 to 100% of people with progressive MS have bladder insufficiency (102,103). Future trials should consider the issue of hydration during exercise.

Conclusions
A vigorous cool room walking training is feasible for people with MS using ambulatory assistive devices. We did not identify any adverse events or safety hazards during the training. The total time walked, and distance covered progressively increased while total resting time decreased. People with MS walked significantly faster after cool room training with better gait quality, which was sustained at three months follow up suggesting that there was long-term improvement of function and not simply short-term performance enhancement. Fatigue (SF-36 fatigue/energy/vitality and mFIS), fitness (maximal workload and cardiorespiratory reserve), and quality of life measures (physical functioning, bodily pain, and health compared to last year) improved significantly after training, and improvements on fatigue (mFIS), fitness (maximal workload), and quality of life (physical functioning) were sustained 12 weeks after completion of the program. Vigorous training in a cool room using BWST has the potential to be an effective treatment option for improving walking ability, fatigue, fitness, and quality of life in people with MS using walking aids, which provides a strong rationale for a future clinical trial. There were associations between improvements in walking, fatigue, fitness, and blood markers (serum BDNF and IL-6) that are worthy of further evaluation. given at least one week to review the study protocol and the consent form. All participants signed and provided written informed consent to the principal investigator prior to entering the study.

Consent for publication Not Applicable
Availability of data and materials The data supporting this study are available at request from the corresponding author at the Memorial University of Newfoundland, Canada.

Competing interests
The authors declare no competing interests.         Relationship between outcomes. A: correlation between modified fatigue impact scale physical subcomponent change score and fast walking speed change score (rs=-0.847, p=0.008); B: correlation between modified fatigue impact scale total change score and maximal respiratory exchange ratio change score (rs=-0.810, p=0.015); C: correlation between resting serum brain-derived neurotrophic factor change score and maximal respiratory exchange ratio change score (rs=0.786, p=0.036); D: correlation between resting serum interleukin-6 change score and maximal oxygen consumption change score (rs=-0.757, p=0.049). rs, Spearman correlation coefficient; p, significance.