Motor unit discharge characteristics and walking performance of individuals with multiple sclerosis

Abstract

Walking performance of persons with multiple sclerosis (MS) is strongly influenced by the activation signals received by lower leg muscles. We examined the associations between force steadiness and motor unit discharge characteristics of lower leg muscles during submaximal isometric contractions with tests of walking performance and disability status in individuals who self-reported walking difficulties due to MS. We expected that worse walking performance would be associated with weaker plantar flexor muscles, worse force steadiness, and slower motor unit discharge times. Twenty-three individuals with relapsing-remitting MS (56 ± 7 yr) participated in the study. Participants completed one to three evaluation sessions that involved two walking tests (25-ft walk and 6-min walk), a manual dexterity test (grooved pegboard), health-related questionnaires, and measurement of strength, force steadiness, and motor unit discharge characteristics of lower leg muscles. Multiple regression analyses were used to construct models to explain the variance in measures of walking performance. There were statistically significant differences (effect sizes: 0.21–0.60) between the three muscles in mean interspike interval (ISI) and ISI distributions during steady submaximal contractions with the plantar flexor and dorsiflexor muscles. The regression models explained 40% of the variance in 6-min walk distance and 47% of the variance in 25-ft walk time with two or three variables that included mean ISI for one of the plantar flexor muscles, dorsiflexor strength, and force steadiness. Walking speed and endurance in persons with relapsing-remitting MS were reduced in individuals with longer ISIs, weaker dorsiflexors, and worse plantar flexor force steadiness.

NEW & NOTEWORTHY The walking endurance and gait speed of persons with relapsing-remitting multiple sclerosis (MS) were worse in individuals who had weaker dorsiflexor muscles and greater force fluctuations and longer times between action potentials discharged by motor units in plantar flexor muscles during steady isometric contractions. These findings indicate that the control of motor unit activity in lower leg muscles of individuals with MS is associated with their walking ability.

INTRODUCTION

Multiple sclerosis (MS) is a debilitating neurological disorder that typically results in ~50% of afflicted individuals requiring walking aids within 15 yr of disease onset (Baert et al. 2014; LaRocca 2011; Motl et al. 2011). Gait analyses indicate that persons with MS prefer slower walking speeds, spend more time in the double-support phase, walk with wider strides, and consume more oxygen to walk at the same speed as healthy control subjects (Motl et al. 2012; Remelius et al. 2012). The adaptations in gait exhibited by individuals with MS are associated with declines in performance on both short (25 ft, 10 m, and 30 m)- and long(400 m, 2 min, and 6 min)-distance walk tests (Baert et al. 2014; Goldman et al. 2013; Kieseier and Pozzilli 2012; Pilutti et al. 2013).

The reductions in walking performance experienced by persons with MS are associated with decreases in several aspects of motor function (Kjølhede et al. 2015; Wagner et al. 2014; Wetzel et al. 2011). For example, maximal voluntary contraction (MVC) torque for the plantar flexors was the most consistent predictor of the variance in 25-ft walk time (gait speed), 6-min walk distance (walking endurance), and self-perceived limitations in walking [12-item MS Walking Scale (MSWS-12)] for 42 individuals with MS (Wagner et al. 2014). Similarly, MVC torque for the knee flexors, but not the knee extensors, emerged as a significant predictor for the variance in 25-ft time and 2-min distance for 52 persons with MS (Broekmans et al. 2013). However, MVC torques for the knee extensors and knee flexors did not explain any of the variance in 6-min walk distance for 24 individuals with MS (Hansen et al. 2014). Rather, the muscles that cross the ankle joint are critical in determining the walking ability of persons with MS (Manca et al. 2017; McLoughlin et al. 2015; Mount and Dacko 2006; Wagner et al. 2014).

Some evidence also suggests that decreases in the ability to control force during submaximal contractions are associated with declines in mobility for persons with MS. For example, Davies et al. (2017) found significant correlations between accuracy when matching a sinusoidal target by performing isometric contractions with the plantar flexor muscles and several gait variables when individuals with MS walked at a preferred speed. Participants who were less accurate during the force-matching task exhibited a lesser net torque about the ankle joint at toe-off, shorter step lengths, and slower walking speeds. Similarly, greater force fluctuations during steady contractions with the plantar flexors, when matching a target of 20% MVC torque, were correlated with slower walking speed, slower step rate, and shorter step length in persons with MS (Arpin et al. 2016).

These findings suggest that force steadiness—force fluctuations during submaximal isometric contractions (Galganski et al. 1993)—may be related to walking performance in individuals with MS. In a seminal study on force steadiness, Negro et al. (2009) demonstrated that 74% of the fluctuations in force during a steady contraction with a hand muscle at 10% of MVC could be explained by low-frequency modulation (≤10 Hz) of the discharge times of motor neuron action potentials. The measurement of force steadiness therefore provides an index of the variability in the common input signal received by the motor neurons engaged for a specific task (Enoka and Duchateau 2017; Farina et al. 2012). Moreover, force steadiness decreases during fatiguing contractions (Castronovo et al. 2015), can be improved with practice (Christou et al. 2003; Marmon et al. 2011a), and is significantly associated with performance on some functional tasks (Carville et al. 2007; Hyngstrom et al. 2014; Kobayashi et al. 2014). For example, force steadiness as measured by the coefficient of variation for force during submaximal isometric contractions can explain significant amounts of the variance in the time it takes young, middle-aged, and old adults to complete an index of manual dexterity, the grooved pegboard test (Almuklass et al. 2016; Hamilton et al. 2017; Marmon et al. 2011b).

The purpose of our study was to examine the associations between strength, force steadiness, and motor unit discharge characteristics of lower leg muscles with clinical assessments of walking performance and disability status in individuals who self-reported walking difficulties due to MS. We hypothesized that the time to walk 25 ft and the distance that could be walked in 6 min by individuals with relapsing-remitting MS would be worse for those with weaker plantar flexor muscles, elevated force steadiness, and slower motor unit discharge times.

MATERIALS AND METHODS

The study enrolled 23 volunteers diagnosed with relapsing-remitting MS (53.4 ± 7.3 yr; descriptors in Table 1). Key exclusion criteria included history of seizure disorder, implanted biomedical devices or metal, and skin disease. Most of the data in the present report were obtained from 21 participants who attended multiple experimental sessions during a randomized clinical trial (ClinicalTrials.gov identifier: NCT02152085) that comprised a 6-wk intervention with neuromuscular electrical stimulation (NMES) applied to lower leg muscles (Almuklass et al. 2017). These individuals completed 3 evaluation sessions and 18 treatment sessions distributed over 10 wk. Each evaluation session comprised 2 days of testing. The evaluation sessions were performed 1 wk before beginning the 6-wk intervention, within 1 wk after completing the intervention, and ~4 wk after the end of the intervention. The outcomes produced by the intervention are reported in detail elsewhere (Almuklass et al. 2017) but included clinically significant increases in the distance walked in 6 min (walking endurance) and decreases in the time it took to walk 25 ft (maximal gait speed). Given that the purpose of the study was to explain the variance in walking performance, each evaluation session for these 21 individuals was treated independently in the present study. Because of time constraints, two participants completed only the initial evaluation session. All participants provided written informed consent, and the Institutional Review Board at the University of Colorado Boulder approved the protocol (No. 13-0720).

Table 1.

Descriptive statistics for the 23 participants

Sample size/women	23/14
Age, yr	56.0 ± 7.3†
6-min walk, m	358 ± 135
25-ft walk, s	6.0 ± 15.0†
Grooved pegboard, s	97 ± 40†
PDDS	3.5 ± 1.0†
MFIS	42 ± 20
MSWS-12	49 ± 11†
Plantar flexor MVC torque, N·m
More affected leg	21 ± 12
Less affected leg	23 ± 11
Dorsiflexor MVC torque, N·m
More affected leg	10.1 ± 6.2
Less affected leg	14.5 ± 6.7*
Plantar flexor steadiness, %
10% MVC	3.9 ± 2.6†
20% MVC	3.0 ± 1.9†
Dorsiflexor steadiness, %
10% MVC	4.7 ± 5.5†
20% MVC	3.5 ± 5.2†

Values are reported as means ± SD or

^†medians ± SD. Steadiness values are coefficient of variation for force (%).

^*P < 0.05 relative to the more affected leg.

On the first day of each evaluation session, participants performed walking tests (25-ft walk test and 6-min walk test) and a manual dexterity test (grooved pegboard) and completed three health-related questionnaires [Patient Determined Disease Steps (PDDS), Modified Fatigue Impact Scale (MFIS), and MSWS-12]. The grooved pegboard test and the questionnaires provided information about the disability status of each participant; time to complete a pegboard test is strongly associated with disability status of persons with MS (Goodkin et al. 1988; Kierkegaard et al. 2012; Koch et al. 2014; Yozbatıran et al. 2006). Maximal walking speed was measured as the time it took an individual to walk 25 ft as quickly as possible. Based on the recommendation of the National MS Society, the average of two trials was used as the measure of maximal walking speed. Walking endurance was characterized as the distance walked in 6 min around a 160-m track. Participants were encouraged to walk briskly, and the distance covered at 1, 2, 4, and 6 min was recorded. Manual dexterity was quantified as the time taken to complete the grooved pegboard test, which required participants to place 25 pegs into holes on a pegboard as quickly as possible (Almuklass et al. 2016).

On the second day of each evaluation session, muscle strength and force steadiness were measured while the subject lay in a supine position with the hip, knee, and ankle joints at neutral angles, which maximized the isolated contraction of each muscle group in a comfortable position. Muscle strength was quantified as the peak torque (N·m) achieved by the dorsiflexor and plantar flexor muscles of each leg when participants gradually increased muscle torque up to maximum and sustained it briefly. Subjects received verbal instructions to increase force gradually during the maximal contractions. Participants performed two to five MVC trials. When two MVC torques were within 10% of each other, the greater value was designated as the MVC torque. Participants then performed submaximal isometric contractions with the dorsiflexor and plantar flexor muscles to match target forces of 10% and 20% MVC torque with the leg each participant identified as experiencing fewer symptoms. The task was to match the target force displayed on a monitor (visual angle = ~0.25°) as steadily as possible during two 30-s trials at each target force with one muscle group (dorsiflexors or plantar flexors) at a time. Force fluctuations during each steady contraction were quantified as the coefficient of variation for force of the low-pass-filtered signal and used as a measure of force steadiness.

The less affected leg, as self-determined by the participant, was confirmed with the measurements of MVCs on both legs (Table 1). To minimize the influence of day-to-day variation in symptom intensity across the clinical trial, the measurements of force steadiness and motor unit discharge characteristics were performed with the less affected leg. Motor unit activity was recorded during eight steady contractions for each participant: two muscle groups (plantar flexors and dorsiflexors) × two target forces × two trials for each target force. Motor unit activity was recorded for medial gastrocnemius, lateral soleus, and tibialis anterior muscles with a high-density surface electromyography (EMG) system (4 × 8 detection points with interelectrode distance of 10 mm). The electrodes were attached over the belly of each muscle in the following locations: medial gastrocnemius, 2–4 cm distal from the popliteal fossa and ~1 cm proximal to the gastrocnemius aponeurosis; lateral soleus, 2–4 cm from the gastrocnemius aponeurosis; tibialis anterior, ~1 cm lateral to the tibial prominence.

The surface of the skin was prepared with an abrasive gel and isopropyl rubbing alcohol, and the high-density array electrodes were placed on the skin over the three muscles (Fig. 1). The EMG electrodes were attached with adhesive pads and tape. Motor unit action potentials were discriminated off-line from the EMG recordings with a custom decomposition algorithm (Holobar et al. 2010; Holobar and Zazula 2007).

Fig. 1.

A and B: locations of the surface grid electrodes over medial gastrocnemius and soleus (A) and tibialis anterior (B) muscles. C: selected motor unit action potentials (MUAPs), with each column representing a 30-ms recording. D: measurement of force steadiness at 10% MVC.

The force exerted by the limb during the strength and steadiness tasks was measured with a strain gauge transducer (MLP-300; Transducer Techniques, Temecula, CA). The force signal was low-pass filtered (≤50 Hz; Coulbourn Instruments, Allentown, PA), recorded on a computer, and digitized at 1,000 samples/s. All force data were obtained with Spike2 data acquisition software (version 5.20; Cambridge Electronic Design, Cambridge, UK) and stored on a computer for off-line analysis. The recorded force signals were filtered with a 20-Hz low-pass, second-order Butterworth filter to quantify the force fluctuations.

Data analysis.

Because of limitations in automatic decomposition algorithms, the decomposed signals comprised both motor unit activity and transient waveforms. Before decomposition each trial was visually inspected by individuals who are familiar with these recordings, and trials that were deemed to contain too many artifacts were discarded. Each decomposed signal (n = 4,143) was examined, and data were deemed artifacts and excluded on the basis of the following criteria, performed in order, based on our previous work on single-motor unit recordings (Barry et al. 2007; Moritz et al. 2005; Pascoe et al. 2014): 1) discard any ISI < 25 ms or > 400 ms; 2) exclude any motor unit with a coefficient of variation for ISI < 8% or > 55%; 3) reject any motor unit with an ISI distribution that had a skewness value in the range of ±0.5; 4) eliminate any motor unit that had a coefficient of variation for ISI < 10% and a distribution skewness < 1. In addition, the waveforms of those motor units with coefficients of variation for ISI in the ranges of 8–10% and 50–55% were visually inspected to ensure that the identified waveforms were consistent with expected shapes for motor unit action potentials. As a result, 911 decomposed signals were discarded and the analysis was performed on the recordings of 3,232 motor units obtained from the two trials at each target force for the three muscles of the 23 participants.

Given that the goal of our present study was to assess the association between motor unit discharge characteristics and force steadiness with the two measures of walking performance, the measurements at each time point were treated as independent outcomes. To ensure the appropriateness of this approach, we considered the extent to which the same motor unit might be present in multiple recordings. Differences in the average discharge characteristics (mean ISI, coefficient of variation for ISI, skewness, and kurtosis) across the two trials at each target force indicated that different sets of motor units were recorded during each trial. Nonetheless, some motor units were likely recorded during both trials at each target force (Martinez-Valdes et al. 2017) and perhaps at the two target forces. The potential influence of duplicate recordings on the average discharge characteristics was estimated by iteratively removing 40% (Martinez-Valdes et al. 2017) of the motor units recorded in the first trial that contributed to an overall mean from all the motor unit means from both trials and comparing it with the mean obtained without the removal of motor units. The procedure was performed on a subset of the motor units (n = 481) that contributed to the overall mean values for each condition (2 target forces × 3 muscles). Removal of 40% of the motor units had a minimal influence on the overall mean values. For example, the maximal change in overall mean ISI after 100 iterations of removing 40% of the motor units was 5.77 ± 2.64%. These results suggest that duplicate recordings across trials would not have confounded the outcomes of the regression analysis.

Moreover, the regression analysis was based on the average discharge characteristics for all the motor units identified in a single session for each muscle. After the exclusion procedures, the average number of motor units recorded in each session (2 trials × 2 target forces) was 38 for tibialis anterior, 18 for medial gastrocnemius, and 27 for soleus, which resulted in 244 averages that were used in the regression analyses.

The Kolmogorov-Smirnov (>50 samples) and Shapiro-Wilk (<50 samples) tests were used to assess normality. The Kruskal-Wallis test was used to compare motor unit characteristics between the three muscles. Post hoc comparisons with the Kruskal-Wallis test were performed with Dunn’s nonparametric comparisons, and significance values were adjusted by Bonferroni corrections for multiple comparisons. The motor unit discharge characteristics for each muscle were compared at the two target forces (10% and 20% MVC) with the Mann-Whitney U-test. The muscle strength data were normally distributed; hence the paired t-test was used to compare strength of the more and less affected legs. The effect size for the Kruskal-Wallis test was quantified as φ, which was obtained from $\left(\sqrt{\frac{x^2}{N}}\right)$ and from $\left(\frac{z}{\sqrt{n1\text{+}n2}}\right)$ for the Mann-Whitney U-test. Effect sizes of 0.1 were considered small, and those ≥ 0.5 were deemed large.

As the data were not normally distributed, Spearman correlations were used to compare walking performances (25-ft test, 6-min test), clinical assessments (PDDS, MFIS, MSWS-12, grooved pegboard test), and neuromuscular measurements (MVC torque, force steadiness, motor unit discharge characteristics). The significantly correlated outcomes were entered into a stepwise, linear multiple regression analysis to identify the neuromuscular characteristics that were most strongly associated with performance on the two tests of walking performance. Multicollinearity was estimated with variance inflation factor (VIF). Normality tests were performed on the residuals, and Cook’s distance criteria were used to identify and remove outliers (Cook 1977). All statistical procedures were performed with SPSS (version 24.0; SPSS, Chicago, IL) with the α set at 0.05 and adjusted with Bonferroni corrections as required.

RESULTS

The data were obtained from 65 experimental sessions that involved 456 successful trials in which force steadiness was measured and 488 successful trials in which motor units were discriminated during steady isometric contractions. Because of technical limitations, the number of time points (weeks 0, 7, and 11) from which acceptable motor unit recordings were obtained from the 23 participants ranged from 1 to 3 for the three muscles (Table 2).

Table 2.

Number of subjects from whom acceptable motor unit recordings were obtained

	Week 0	Week 7	Week 11
Tibialis anterior	8	10	5
Gastrocnemius	9	6	6
Soleus	9	7	7

Values are numbers of subjects from whom acceptable motor unit recordings were obtained for the 3 lower leg muscles at 3 time points (weeks 0, 7, and 11) during the 6-wk intervention.

Motor unit characteristics.

The 6-wk NMES intervention had a relatively minor influence on the discharge characteristics of the identified motor units in the three lower leg muscles (Table 3). There were no statistically significant effects on the mean ISI and only two statistically significant reductions in the coefficient of variation for ISI at the 20% target force at week 11. There were also two statistically significant effects on the ISI distributions: skewness for the medial gastrocnemius motor units was increased at week 7 for the 20% target, and kurtosis for tibialis anterior motor units was increased at week 11 for the 10% target.

Table 3.

Motor unit numbers and discharge characteristics for the three test muscles across the 6-wk NMES intervention

	Motor units, n		Mean ISI, ms		CV for ISI, %		ISI Skewness		ISI Kurtosis
	10%	20%	10%	20%	10%	20%	10%	20%	10%	20%
Tibialis anterior
Week 0	196	187	110 ± 24	106 ± 27	31 ± 13	32 ± 12	2.0 ± 1.2	2.3 ± 1.4	11 ± 13	13 ± 19
Week 7	287	276	112 ± 26	99 ± 19	30 ± 13	31 ± 13	2.1 ± 1.3	2.0 ± 1.1	11 ± 12	11 ± 11
Week 11	331	357	109 ± 20	101 ± 20	28 ± 13	29 ± 12*†	2.2 ± 1.5	2.1 ± 1.2	14 ± 17*	11 ± 10
Effect size [P value]			0.03 [0.68]	0.07 [0.11]	0.09 [0.05]‡	0.12 [0.003]	0.07 [0.13]	0.03 [0.69]	0.11 [0.01]	0.05 [0.42]
Gastrocnemius
Week 0	74	124	138 ± 49	136 ± 30	31 ± 15	31 ± 13	1.8 ± 1.4	1.6 ± 1.3	10 ± 13	9 ± 15
Week 7	91	112	132 ± 30	131 ± 40	32 ± 14	32 ± 12	1.7 ± 1.4	1.9 ± 1.1*	9 ± 11	9 ± 8
Week 11	125	170	142 ± 34	138 ± 30	32 ± 13	31 ± 13	1.5 ± 1.0	1.6 ± 1.3†	7 ± 9	8 ± 9
Effect size [P value]			0.14 [0.07]	0.12 [0.05]	0.03 [0.84]	0.03 [0.83]	0.11 [0.18]	0.14 [0.02]	0.14 [0.07]	0.09 [0.22]
Soleus
Week 0	96	133	156 ± 34	146 ± 34	30 ± 12	31 ± 12	1.4 ± 1.2	1.5 ± 1.0	7 ± 9	7 ± 7
Week 7	125	163	148 ± 31	142 ± 36	30 ± 12	32 ± 13	1.6 ± 1.1	1.5 ± 1.2	8 ± 10	7 ± 9
Week 11	167	218	147 ± 33	146 ± 33	29 ± 13	28 ± 13†	1.7 ± 1.6	1.7 ± 1.6	10 ± 15	10 ± 14
Effect size [P value]			0.13 [0.05]	0.08 [0.22]	0.05 [0.60]	0.12 [0.03]	0.10 [0.13]	0.02 [0.91]	0.10 [0.16]	0.07 [0.30]

Values are means ± SD. CV,  coefficient of variation.

^*P < 0.05 relative to week 0.

^†P < 0.05 relative to week 7.

^‡Kruskal-Wallis test suggested a significant difference between groups, but pairwise comparison with Bonferroni adjustments failed to detect the difference.

There were modest differences in the discharge characteristics of cumulative set motor units at the two target forces (Table 4). Mean ISI was briefer at the greater target force (20% MVC) for both tibialis anterior (102 ± 22 and 110 ± 23 ms; effect size = 0.20, P < 0.001) and soleus (145 ± 34 and 150 ± 33 ms; effect size = 0.07, P = 0.04) but not medial gastrocnemius (138 ± 36 and 135 ± 34 ms; effect size = 0.06, P = 0.09). However, the ISI distributions for each muscle, as characterized by the coefficient of variation, skewness, and kurtosis, were not significantly different (effect sizes ≤ 0.06, P values > 0.05) across target forces for any of the three muscles (Table 4).

Table 4.

Motor unit characteristics in tibialis anterior, gastrocnemius, and soleus during steady isometric contractions at the two target forces

	Tibialis Anterior	Gastrocnemius	Soleus	Effect Size [P value]
Motor units, n	1,634	696	902
Mean ISI, ms
10%	110 ± 23	138 ± 36*	150 ± 33*†	0.56 [<0.001]
20%	102 ± 22‡	135 ± 34*	145 ± 34*†‡	0.60 [<0.001]
Effect size [P value]	0.20 [<0.001]	0.06 [0.09]	0.07 [0.04]
Coefficient of variation for ISI, %
10%	30 ± 13	32 ± 14	30 ± 12	0.06 [0.07]
20%	30 ± 13	31 ± 13	30 ± 13	0.04 [0.23]
Effect size [P value]	0.03 [0.14]	0.02 [0.69]	0.01 [0.8]
ISI distribution skewness
10%	2.1 ± 1.3	1.6 ± 1.3*	1.6 ± 1.3*	0.26 [<0.001]
20%	2.1 ± 1.2	1.7 ± 1.3*	1.6 ± 1.3*	0.24 [<0.001]
Effect size [P value]	0.02 [0.45]	0.05 [0.19]	0.01 [0.9]
ISI distribution kurtosis
10%	12 ± 15	8 ± 11*	9 ± 12*	0.24 [<0.001]
20%	11 ± 13	8 ± 11*	8 ± 11*	0.21 [<0.001]
Effect size [P value]	0.02 [0.37]	0.06 [0.10]	0.01 [0.8]

Values are means ± SD.

^*P < 0.05 relative to tibialis anterior.

^†P < 0.05 relative to gastrocnemius.

^‡P < 0.05 relative to 10%.

In contrast, there were more substantial differences in discharge characteristics between muscles at the two target forces. Mean ISI was longer at both target forces [effect sizes: 0.56 (P < 0.001) and 0.60 (P < 0.001)] for medial gastrocnemius and soleus relative to tibialis anterior and for soleus relative to medial gastrocnemius (Table 4). Similarly, the ISI distributions differed across muscles as indicated by lesser skewness and kurtosis values [0.21–0.26 (P < 0.001)] for medial gastrocnemius and soleus relative to tibialis anterior (Table 4). In contrast, the coefficient of variation for ISI (~30%) did not differ across muscles at either of the target forces [0.06 (P < 0.07) and 0.04 (P < 0.23)].

Associations between walking performance and neuromuscular characteristics.

The distance walked in 6 min was negatively correlated with mean ISI for both soleus (r = −0.46, P = 0.007) and medial gastrocnemius (r = −0.53, P = 0.002) during steady isometric contractions with the calf muscles at 10% MVC force (Table 5). The negative correlations indicate that briefer mean ISIs were associated with a greater distance walked in 6 min. In addition, 6-min walk distance was positively correlated with the coefficient of variation for ISI of motor units in medial gastrocnemius during steady isometric contractions with the plantar flexor muscles at 10% MVC (r = 0.47, P = 0.006). This association indicates that individuals with greater ISI variability were able to walk farther in 6 min (Table 5). The distance walked in 6 min was also significantly correlated with MVC torque for the dorsiflexors in the more affected leg (r = 0.43, P = 0.001) and force steadiness of the less affected leg for both the dorsiflexors (r = −0.39, P = 0.004) and the plantar flexors (r = −0.45, P < 0.001) during isometric contractions at 20% MVC force (Table 5). These associations indicate that the 6-min walk distance was longer for participants with stronger dorsiflexors (more affected leg) and longer for participants with less force variability during dorsiflexion and plantar flexion (less affected leg).

Table 5.

Correlation coefficients between walking tests (6-min walk and 25-ft walk) and neuromuscular characteristics

	6-min Walk, m	25-ft Walk, s
Soleus
10% MVC
Mean ISI, ms	−0.46	0.44
Coefficient of variation for ISI, %	0.28	−0.21
20% MVC
Mean ISI, ms	−0.29	0.29
Coefficient of variation for ISI, %	0.28	−0.27
Gastrocnemius
10% MVC
Mean ISI, ms	−0.53	0.53
Coefficient of variation for ISI, %	0.47	−0.44
20% MVC
Mean ISI, ms	−0.22	0.19
Coefficient of variation for ISI, %	0.16	−0.16
Tibialis anterior
10% MVC
Mean ISI, ms	0.00	0.00
Coefficient of variation for ISI, %	0.15	−0.14
20% MVC
Mean ISI, ms	−0.07	0.08
Coefficient of variation for ISI, %	0.07	−0.03
Plantar flexor MVC torque, N·m
More affected leg	0.22	−0.34
Less affected leg	0.02	−0.17
Dorsiflexor MVC torque, N·m
More affected leg	0.43	−0.50
Less affected leg	0.22	−0.34
Plantar flexor force steadiness, %
10% MVC	−0.21	0.27
20% MVC	−0.45	0.46
Dorsiflexor force steadiness, %
10% MVC	−0.26	0.22
20% MVC	−0.39	0.30

Steadiness values are coefficient of variation for force (%). ISI, interspike interval. Bold font indicates P < 0.05.

The time to walk 25 ft was positively correlated with mean ISI for both soleus (r = 0.44, P = 0.011) and medial gastrocnemius (r = 0.53, P = 0.001) during steady isometric contractions with the calf muscles at 10% MVC force (Table 5). The correlations indicate that briefer mean ISIs were associated with faster times to walk 25 ft. In addition, 25-ft walk time was negatively correlated with the coefficient of variation for ISI of motor units in medial gastrocnemius during steady isometric contractions with the plantar flexor muscles at 10% MVC (r = −0.44, P = 0.01). This association indicates that individuals with greater variability in medial gastrocnemius ISIs were able to walk 25 ft more quickly (Table 5). The time to walk 25 ft was also significantly correlated with MVC torque for the dorsiflexors in the more affected (r = −0.50, P < 0.001) and less affected (r = −0.34, P = 0.007) legs, MVC torque of the plantar flexors in the more affected leg (r = −0.34, P = 0.007), and force steadiness for the dorsiflexors (r = 0.30, P = 0.03) at 20% MVC force and the plantar flexors at both 10% (r = 0.27, P = 0.04) and 20% (r = 0.47, P < 0.001) MVC force (Table 5). These associations indicate that the 25-ft walk time was faster for participants with stronger dorsiflexors (both legs) and plantar flexors (more affected leg) and lesser force fluctuations during steady isometric contractions with both the dorsiflexors and the plantar flexors.

The 6-min distance was negatively correlated with the three measures of disability status: the PDDS score (r = −0.66, P < 0.001), the MSWS-12 score (r = −0.52, P < 0.001), and time to complete the grooved pegboard test (r = −0.29, P = 0.02) (Table 6). These associations indicate that participants who could walk farther in 6 min had lower self-reported levels of disability (PDDS and MSWS-12) and completed the grooved pegboard test more quickly. Similarly, the 25-ft walk time was also correlated with the same three measures of disability status, but the correlations were positive: PDDS score (r = 0.58, P < 0.001), MSWS-12 score (r = 0.44, P = 0.001), and time to complete the grooved pegboard test (r = 0.35, P = 0.001) (Table 6). These associations indicate that participants who walked the 25 ft more quickly self-reported lower levels of disability (PDDS and MSWS-12) and took less time to complete the grooved pegboard test.

Table 6.

Correlation coefficients between walking tests (6-min walk and 25-ft walk) and clinical measures

	PDDS	MFIS	MSWS-12	Grooved Pegboard
6-min walk, m	−0.66	−0.01	−0.52	−0.29
25-ft walk, s	0.58	−0.06	0.44	0.35

Bold font indicates P < 0.05.

Regression models.

Based on the correlation results (Tables 5 and 6), a stepwise, multiple regression analysis was used to construct models that explained significant amounts of the variance in the 6-min walk distance and 25-ft walk time. The significant correlations between the outcome variables and the two measures of walking performance that were included in the regression models are shown in Fig. 2. Neuromuscular characteristics were able to explain 40% of the variance in the 6-min walk distance with two predictor variables (Fig. 3A): mean ISI for medial gastrocnemius when the plantar flexors performed a steady isometric contraction at 10% MVC force (partial r = −0.48, VIF = 1.1, P = 0.006; Fig. 2A) and MVC torque for the dorsiflexors of the more affected leg (partial r = 0.37, VIF = 1.1, P = 0.04; Fig. 2B).

Fig. 2.

Strength of the correlations between the 2 tests of walking performance and the predictor variables that were included in the regression models. The data that contributed to the regression models comprised 3 individuals with measurements from 3 time points each, 8 subjects with measurements from 2 time points, and 7 subjects with a single time point. A: 6-min distance and mean ISI for motor units in medial gastrocnemius during a steady contraction at 10% MVC force. B: 6-min distance and the MVC torque for the dorsiflexors of the more affected leg. C: 25-ft time mean ISI for motor units in soleus during a steady contraction at 10% MVC force. D: 25-ft time and MVC torque for the dorsiflexors of the more affected leg. E: 25-ft time and force steadiness for the plantar flexors during an isometric contraction at 20% MVC force.

Fig. 3.

Associations between observed and predicted times to complete the 2 walking tests based on neuromuscular characteristics. A: 6-min walk test (n = 32). The predictor variables (R² = 0.40) were mean ISI for motor units in medial gastrocnemius during a steady contraction at 10% MVC force and the MVC torque for the dorsiflexors of the more affected leg. B: 25-ft walk test (n = 31). The predictor variables (R² = 0.47) were mean ISI for motor units in soleus during a steady contraction at 10% MVC force, MVC torque for the dorsiflexors of the more affected leg, and force steadiness for the plantar flexors during an isometric contraction at 20% MVC force.

Similarly, neuromuscular characteristics were able to explain 47% of the variance in the 25-ft walk time with three predictor variables (Fig. 3B): mean ISI for soleus when the plantar flexors performed a steady isometric contraction at 10% MVC (partial r = 0.51, VIF = 1.06, P = 0.005; Fig. 2C), MVC torque for the dorsiflexors of the more affected leg (partial r = −0.43, VIF = 1.05, P = 0.02; Fig. 2D), and force steadiness when the plantar flexors performed a steady isometric contraction at 20% MVC (partial r = 0.39, VIF = 1.04, P = 0.037; Fig. 2E).

DISCUSSION

The main findings of our study were that moderate amounts of variance in the walking performance of persons with MS could be explained by the strength of the dorsiflexor muscles in the more affected leg and the force fluctuations and ISI durations during submaximal isometric contractions with the plantar flexor muscles in the less affected leg. The predictive power of these outcomes for both walking tests (25 ft and 6 min) was greater for the two measures of muscle activation (force steadiness and mean ISI) than the strength of the dorsiflexor muscles.

Muscle strength.

In contrast to our hypothesis, it was the strength of the dorsiflexors and not the plantar flexors that was more strongly associated with performance on the two walking tests. Moreover, it was the dorsiflexor muscles in the more affected leg that were associated with walking performance. The hypothesis was primarily based on the report by Wagner et al. (2014) in which the strength of the plantar flexor muscles and not the dorsiflexor muscles explained significant amounts of the variance in both the 25-ft and 6-min tests of individuals with mild levels of disability [average Expanded Disability Status Scale (EDSS) score = 3]. However, their analysis was based on a measure of muscle weakness, which they defined as the minimal peak value achieved during MVCs.

Conversely, 6 wk of strength training in the dorsiflexors of the more affected leg by persons with relapsing-remitting MS (EDSS < 6) increased muscle strength (Cohen’s d = 0.6), increased 6-min walk distance (d = 0.5), and decreased 10-m walk time (d = 0.8) (Manca et al. 2017). Similarly, the wearing of a dorsiflexion-assist orthosis reduces the physiological cost and attenuates the declines in muscle strength and balance when persons with MS perform the 6-min test (McLoughlin et al. 2015). Moreover, increases in the 25-ft walk time with progression of the disease are accompanied by significant decreases in the strength of the dorsiflexor muscles (Zackowski et al. 2015). Among the four significant correlations listed in Table 5 between the strength of lower leg muscles and walking performance, the correlations for MVC torque of the dorsiflexor muscles in the more affected leg were the greatest. The results of our study, therefore, underscore the critical role of dorsiflexor strength in constraining walking performance of persons with MS (Benedetti et al. 1999; Martin et al. 2006; Matsuda et al. 2011; Thickbroom et al. 2008).

Muscle activation.

More of the variance in each of the walking tests was explained by one or two muscle activation measures (motor unit discharge characteristics or force steadiness) than by the strength of the dorsiflexors in the more affected leg. The predictor variable with the largest partial r values in the two regression models (6-min and 25-ft walk) was the mean ISI for one of the lower leg muscles during steady isometric contractions at 10% MVC force. The involved muscles were medial gastrocnemius for the 6-min walk (partial r = −0.48) and soleus for the 25-ft walk (partial r = 0.51), which indicates that longer mean ISIs were associated with worse walking performance: less distance walked in 6 min and longer time to walk 25 ft.

The mean ISIs observed in our study during steady isometric contractions may be a consequence of the decrease in peak discharge rate that has been reported for the vastus lateralis muscles in persons with MS (Rice et al. 1992). Mean ISIs in lower leg muscles of healthy young subjects during submaximal (~25% MVC) isometric contractions are briefest (65 ± 14 ms) for tibialis anterior (Connelly et al. 1999), intermediate (101 ± 29 ms) for medial gastrocnemius (Kirk et al. 2016), and longest (114 ± 22 ms) for soleus (Dalton et al. 2009). In comparison, the ISIs at 20% MVC in our study were longer for all three muscles (tibialis anterior = 102 ± 22 ms, medial gastrocnemius = 135 ± 34 ms, soleus = 145 ± 34 ms; Table 5). The central effects of the disease appear to compromise the capacity of motor neurons to generate action potentials, which may be related to the motor unit remodeling and decline in motor unit numbers in these individuals (Vogt et al. 2009).

In contrast to the significant association of dorsiflexor strength with walking performance, it was the mean ISIs—an index of muscle activation—for the two plantar flexors and not the dorsiflexor (tibialis anterior) that emerged as significant predictors of walking ability. Moreover, the regression model included the mean ISIs for different muscles for each walking test: soleus for the 25-ft test and medial gastrocnemius for the 6-min test. The result was unexpected, as the 6-min test is a measure of walking endurance (Kieseier and Pozzilli 2012; Sandroff et al. 2015), but our findings indicate that the distance walked in 6 min was inversely related to mean ISI for motor units in medial gastrocnemius during an isometric contraction with the plantar flexors of the less affected leg at 10% MVC force. Longer distances for the 6-min test, therefore, were presumably associated with faster contractile speed for medial gastrocnemius. Conversely, slower speeds for the 25-ft test were associated with slower contractile speed—longer mean ISIs—for soleus.

In contrast to the significant associations for mean ISI, the coefficient of variation for ISI during the steady contractions, which depends on synaptic noise and its interaction with the time course of the afterhyperpolarization phase (Calvin and Stevens 1968; Matthews 1996), was not included in either of the regression models. The coefficient of variation in our study was ~30% for all three muscles (Table 2), which is greater than nominal values for tibialis anterior (15.9 ± 9.61%; Jesunathadas et al. 2012), soleus (15.7 ± 4.7%; Mochizuki et al. 2007), and vastus lateralis (~12%; Vila-Chã et al. 2010). Despite significant correlations between the coefficient of variation for ISI of the medial gastrocnemius motor units during the steady contraction at 10% MVC force and both walking tests (Table 3), it did not explain any of the variance in 6-min distance or 25-ft time. The elevated values observed in our study indicate a disease-related disturbance in the processes responsible for the coefficient of variation for ISI that was correlated with walking ability, but these changes were statistically less influential than those underlying the lengthening of the mean ISIs during the submaximal isometric contractions.

Similar to the influence of mean ISI on walking performance, the other measure of muscle activation (force steadiness) was also significantly correlated with 6-min distance and 25-ft time. Moreover, it was force steadiness at 20% MVC force for the plantar flexors of the more affected leg and not the dorsiflexors that was included in the regression model for 25-ft time. This association (partial r = 0.39) indicated that faster 25-ft times were related to lesser force fluctuations (lower coefficients of variation for force) during the isometric contraction, which likely indicates less variability in the common modulation of discharge times for motor neurons innervating the plantar flexor muscles during the steady contraction (Farina and Negro 2015; Feeney et al. 2017; Negro et al. 2009). This result is consistent with a previously observed correlation for persons with MS between greater force fluctuations during steady contractions with the plantar flexor muscles and slower walking speeds (Arpin et al. 2016). Thus both measures of muscle activation (mean ISI and force steadiness) for the plantar flexors were significantly associated with 25-ft time, but only mean ISI for medial gastrocnemius was associated with 6-min distance.

Disability status.

The findings in our study underscore the associations between self-reported measures of disability (PDDS and MSWS-12) and quantitative measures of motor function (25-ft test, 6-min test, grooved pegboard test; Table 6). The data reported in this article were obtained in our clinical trial in which individuals with relapsing-remitting MS received 18 treatment sessions distributed across 6 wk of NMES applied to the lower leg muscles (Almuklass et al., unpublished observations). In addition to the expected improvements in walking performance and balance, the intervention reduced self-reported levels of fatigue (MFIS) and walking limitations (MSWS-12) and decreased the time it took the participants to perform the grooved pegboard test. These findings demonstrate the broad central effects of NMES treatments applied to lower leg muscle (Mang et al. 2011) and further emphasize the utility of pegboard tests as measures of disease status in individuals with MS (Goodkin et al. 1988; Kierkegaard et al. 2012; Koch et al. 2014; Yozbatıran et al. 2006).

As an example of how central adaptations can be exploited to improve motor function in persons with MS, Manca et al. (2016) found that 6 wk of strength training of the dorsiflexors in the less affected leg decreased 10-m time and increased the strength of the dorsiflexors in the more affected leg. Moreover, the magnitude of the increase in strength due to the crossed effect was greater than that achieved by explicitly strength training the dorsiflexors of the more affected leg (Manca et al. 2017). Such crossed effects appear to be mediated by an increase in the capacity of the motor cortex to activate the involved muscles (Hendy and Lamon, 2017; Lee et al. 2009) and indicate that the compromised nervous system of individuals with MS is responsive to a training stimulus.

The walking limitations exhibited by persons with MS differ for short- and long-distance tests (Baert et al. 2014), as indicated by the different predictor variables for the two walking tests in our study. In addition, the performance measure was distributed continuously for the 6-min test (Fig. 2A) but appeared to cluster into two groups for the 25-ft test (Fig. 2B). Although both sets of data were derived from the same individuals, the different distributions suggest that each test was constrained by different deficits.

One of the limitations of our study was that the regression models only explained moderate amounts of the variance in the two walking tests. Perhaps more of the variance could have been explained if we had recorded motor unit activity and measured force steadiness in the lower leg muscles of the more affected leg. Although we focused on the muscles that cross the ankle joint, as suggested by other studies, some reports indicate that the strength of the knee extensors may contribute to walking performance in persons with MS.

In summary, our findings indicate that moderate amounts of the variance in walking performance of persons disabled by MS were explained by mean ISIs of action potentials by motor units in plantar flexor muscles during steady submaximal contractions, the strength of the dorsiflexor muscles in the more affected leg, and force steadiness during a steady isometric contraction. Furthermore, walking performance was worse in the participants who exhibited greater levels of disability as indicated by clinical assessments. These findings indicate, for the first time, that the discharge characteristics of motor units innervating the plantar flexor muscles contribute significantly to differences in walking performance among individuals with relapsing-remitting MS.

GRANTS

The Eunice Kennedy Shriver National Institute of Child Health and Human Development provided support for the project (R03 HD-079508). A. M. Almuklass was supported by a scholarship from the King Saud bin Abdulaziz University.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

A.M.A. and R.M.E. conceived and designed research; A.M.A. and L.D. performed experiments; A.M.A., L.D., L.D.H., T.M.V., A.B., and R.M.E. analyzed data; A.M.A. and R.M.E. interpreted results of experiments; A.M.A., T.M.V., A.B., and R.M.E. prepared figures; A.M.A. and R.M.E. drafted manuscript; A.M.A., L.D., L.D.H., T.M.V., A.B., and R.M.E. edited and revised manuscript; A.M.A., L.D., L.D.H., T.M.V., A.B., and R.M.E. approved final version of manuscript.