Transcutaneous electrical nerve stimulation enhances locomotor adaptation savings in people with multiple sclerosis

Abstract

Locomotor adaptation on a split-belt treadmill can improve gait symmetry across various clinical populations, including people with multiple sclerosis (PwMS). As many PwMS experience sensory impairments, mobility interventions relying on sensory prediction errors may be less effective. Transcutaneous electrical nerve stimulation (TENS) has been shown to amplify sensorimotor function in PwMS and healthy controls, but its influence on motor learning remains unexplored. This randomized crossover trial investigated the effects of TENS on locomotor adaptation and cortical activation in PwMS. In total, 28 PwMS and 20 age- and sex-matched healthy controls completed two locomotor adaptation sessions, one with active TENS and one with inactive TENS. Locomotor adaptation was evaluated using step length asymmetry, quantified across four outcome metrics: adaptation magnitude, early change, after-effect and savings. Functional near-infrared spectroscopy recorded cortical activation, and linear mixed-effect models assessed group, visit and TENS condition effects on behavioural and cortical activation outcomes. PwMS exhibited reduced adaptation magnitude compared with healthy controls. TENS did not influence early change (representing adaptation rate) but significantly improved adaptation savings for PwMS who received TENS during their second visit only (initial savings: adj-P = 0.005, d = 1.35; early savings: adj-P = 0.014, d = 1.13). Additionally, both PwMS and healthy controls exhibited decreased cortical activation during locomotor adaptation with TENS, particularly in the dorsal premotor cortex for PwMS (adj-P = 0.019, d = 0.84). These findings indicate that TENS promotes the retention of prior locomotor adaptation, enhancing the efficiency of relearning. Additionally, reduced cortical activation with TENS in both groups indicates reduced cortical reliance during adaptation. Together, these effects suggest that TENS could have broader utility for enhancing motor learning in populations with sensory impairments, potentially leading to amplified retention and automaticity during motor rehabilitation paradigms.

Keywords: functional near-infrared spectroscopy; locomotor adaptation; motor learning; multiple sclerosis; transcutaneous electrical nerve stimulation.

Graphical Abstract

Figure 1

Study design. (A) Crossover design and randomization of transcutaneous electrical nerve stimulation (TENS) for each visit. The first group received TENS ON during Visit 1 and TENS OFF during Visit 2, while the second group had the reverse order. A 4-week washout period separated the visits. (B) Locomotor adaptation paradigm. Participants first completed baseline overground walking at preferred and fast speeds, followed by tied-belt treadmill walking at their preferred speed. During the split-belt adaptation phase (10 min), the more affected limb was set to the overground fast speed, while the less affected limb was set to half the overground fast speed. TENS was active during adaptation phases only. The protocol concluded with tied-belt treadmill walking (10 min) at the preferred speed. (C) Functional near-infrared spectroscopy (fNIRS) imaging block design. Following a 60-s standing rest period, pairs of 30-s blocks were averaged to quantify cortical activation for each treadmill walking timepoint in the baseline, adaptation and deadaptation phases.

Figure 2

Consort diagram. Flowchart illustrating trial recruitment, randomization and retention of participants. A total of 55 participants were assessed for eligibility, with four excluded (three did not meet inclusion criteria, one declined to participate). The remaining 51 participants (31 people with multiple sclerosis (PwMS) and 20 healthy controls (HC)) were randomized into the two transcutaneous electrical nerve stimulation (TENS) conditions. During the trial, two PwMS were lost to follow-up and one had poor data quality, resulting in 28 PwMS and 20 HC participants in this analysis.

Figure 3

Step length asymmetry (SLA) adaptation. (A) SLA adaptation curves and boxplots for people with multiple sclerosis (PwMS) and healthy controls (HC) averaged across all visits and conditions. While early change remained similar, PwMS exhibited reduced adaptation magnitude compared with HCs (t(44) = −2.39, adj-P = 0.042, d = 0.81). (B) SLA adaptation curves and boxplots for PwMS for each visit and condition. Transcutaneous electrical nerve stimulation (TENS) did not significantly affect early change during adaptation; however, at Visit 2, TENS ON (purple) significantly enhanced adaptation savings at Initial Adapt (t(44) = 3.56, adj-P = 0.005, d = 1.35) and Early Adapt (t(44) = 2.98, adj-P = 0.014, d = 1.13) compared with TENS OFF (green). During deadaptation, PwMS showed reduced early change at Visit 2, while savings did not differ between TENS conditions. (C) SLA adaptation curves and boxplots for HC for each visit and condition. TENS had no significant impact on any adaptation variable. However, early change was greater at Visit 2 compared with Visit 1 during adaptation. Similar to PwMS, reduced early change was observed during deadaptation at Visit 2. Each data point in all adaptation curves (A–C) represents the group-average SLA for a single gait cycle (x-axis) with different marker symbols and colours indicating experimental conditions, totalling 1000 data points per condition. For visual purposes SLA curves were interpolated, but all analyses were based on the uninterpolated data. Linear mixed-effects model results of all adaptation outcomes are reported in Table 2.

Figure 4

Functional near-infrared spectroscopy (fNIRS) results. (A) fNIRS channel locations are shown relative to the 10–10 system. Sources are represented by red circles, detectors by blue circles and short-distance detectors by blue rings. Each source-detector line indicates an individual channel, with the thick coloured lines highlighting channels assigned to a specific region of interest (ROI) based on their Brodmann areas (BA). ROIs included dorsal and ventral premotor areas (PMd, PMv), primary motor and somatosensory areas (M1, S1) and superior and inferior parietal lobules (SPL, IPL). (B) Cortical activation with transcutaneous electrical nerve stimulation (TENS) OFF and TENS ON. Group level t-score maps oxyhaemoglobin (HbO) beta change for the Early Adapt—Baseline contrast across premotor, sensorimotor and posterior parietal regions for people with multiple sclerosis (PwMS) (N = 28) and healthy controls (HC) (N = 20). With TENS OFF, both groups exhibited channel-wide increases in activation, including significant activation of the dorsal premotor cortex (PMd) during Early Adapt compared with Baseline. In contrast, with TENS ON, channel-wide activation was significantly reduced compared with TENS OFF for both groups, with no increase from Baseline to Early Adapt. Among only PwMS, this reduced activation with TENS ON was observed specifically in the PMd, primary motor (M1), primary sensory (S1) and superior parietal lobule (SPL) regions of interest (ROIs). Linear mixed-effects model results of all ROIs are reported in Table 3.

Figure 5

Associations between cortical activation and savings. Correlation plots showing relationships between step length asymmetry savings and cortical activation, quantified as oxyhaemoglobin (HbO) beta, at Early Adapt. (A) In people with multiple sclerosis (PwMS), a strong negative correlation was observed between savings and HbO beta in a dorsal premotor cortex (PMd) channel, with a potential correlation in a superior parietal lobule (SPL) channel (N = 13). (B) With transcutaneous electrical nerve stimulation (TENS) ON, these correlations were no longer present in PwMS (N = 15). In healthy controls (HCs), no correlations were observed at these channels under either (C) TENS OFF (N = 9) or (D) TENS ON (N = 11) conditions. Pearson correlations were used for all associations shown. Each data point represents an individual participant's average step length asymmetry savings and corresponding HbO beta in the specified channel.