Background
Many stroke survivors experience chronic gait impairments. After rehabilitation, 36% of stroke survivors cannot walk independently [
1] and walk ~ 50% slower than age-matched peers [
2], which is well below the speed required for safe community ambulation (1.06 m/s) [
3]. Walking endurance is also markedly reduced after stroke [
4]. Reduced walking speed and endurance are major barriers for community participation [
4,
5] and are associated with decreased physical activity [
6] and quality of life [
7,
8]. Gait rehabilitation has focused on development of interventions to restore ambulatory capacity, and there is a critical need to maximize benefits of current walking training interventions.
Because of its clinical and home accessibility, treadmill training has long been utilized as an effective and feasible method of walking training for post-stroke individuals [
9]. Recent studies have explored high-intensity interval training (HIIT) as a way to reduce training time and volume while maximizing intensity [
10]. HIIT involves alternating periods of walking at a high intensity and recovery intensity. Speed-based HIIT (HISTT) is one type of HIIT designed to improve an individual’s walking speed by training at the maximum tolerated treadmill belt speed. HISTT leads to greater improvements in overground walking speed than progressive treadmill training in chronic [
11,
12] and sub-acute [
13] stroke. The effectiveness of HISTT at improving clinical gait and neurophysiological outcomes post-stroke has not been thoroughly investigated.
Cortical priming with neurostimulation or movement is a promising adjuvant therapy to enhance effects of motor rehabilitation [
14]. The premise behind neural priming is that the brain retains its capacity to reorganize after stroke, and priming may improve the effect of associated motor training by correcting the imbalance in interhemispheric inhibition observed post-stroke and facilitating long-term potentiation and depression like mechanisms [
14]. One clinically translatable type of cortical priming is non-invasive transcranial direct current stimulation (tDCS). After stroke, tDCS has been used to correct interhemispheric imbalance in two primary ways: 1) anodal tDCS, in which the anode is placed over the ipsilesional hemisphere to increase ipsilesional cortical excitability and 2) cathodal tDCS, in which the cathode is placed over the contralesional hemisphere to decrease contralesional cortical excitability. Anodal tDCS has been shown to upregulate corticomotor pathways and improve motor learning and function [
14‐
16]. Although cathodal tDCS has also shown beneficial effects, suppressing the contralesional hemisphere may be maladaptive in individuals with limited neural resources in the ipsilesional hemisphere [
16,
17]. These findings support the role of tDCS to potentially enhance the effects of other types of motor training [
15]. Movement-based priming, another priming technique, often involves the performance of a repetitive movement, such as wrist flexion and extension, prior to performance of motor training [
18]. Like tDCS, movement-based priming may also increase corticomotor excitability (CME) and enhance the effects of motor training [
18,
19]. Thus, movement-based priming is also a potential adjuvant to enhance the effects of HISTT, and combining tDCS and movement-based priming may yield greater benefits than either type of priming in isolation.
Despite the potential benefits of tDCS and movement-based priming, there have been no systematic investigations on the effects of motor priming on HISTT-induced improvements in walking. Our lab recently found that a single session of HISTT paired with tDCS and movement priming increases excitability of the ipsilesional hemisphere and decreases excitability of the contralesional hemisphere, supporting the potential efficacy of priming for enhancing HISTT [
20]. In this controlled trial with stratification, our objective was to determine if motor priming can augment the effects of HISTT. As it is critical that the optimal priming technique be paired with gait rehabilitation, we compared the effects of three types of priming techniques on 4-weeks of HISTT: tDCS, movement-based priming, and both combined. Based on pilot data, we hypothesized that tDCS and movement-based priming would enhance the effects of HISTT on walking speed with corresponding changes in CME and combining both types of priming would lead to even greater improvement.
Methods
This study was approved by the institutional review board at the University of Illinois at Chicago (UIC), and all participants provided written informed consent. Trial registration:
ClinicalTrials.gov,
NCT03492229. Registered 10 April 2018 – retrospectively registered as data collection was initiated prior to the revised NIH clinical trial registration guidelines,
https://clinicaltrials.gov/ct2/show/NCT03492229.
Participants had sustained a single, monohemispheric stroke > 6 months prior, were 40–80 years old, had residual gait deficits but could walk without external aid for 5 min, had ≥5° active dorsiflexion in the paretic ankle necessary to perform movement priming, and had a Mini-Mental State Examination (MMSE) score of > 21 to ensure they could follow instructions. Participants were excluded if they had brainstem or cerebellar lesions, a score of ≥2 on the Modified Ashworth Scale, were taking uncontrolled spasticity medications, had major cardiorespiratory or metabolic diseases, or had contraindications to transcranial magnetic stimulation (TMS), including history of seizures, implanted metallic objects, and use of medications that alter cortical excitability.
This was a single-center, multi-arm trial with four parallel groups. Participant group was selected with minimization and an allocation ratio of 1:1:1:1. Minimization required a similar number of participants in each group within age, speed, and paretic Fugl Meyer lower extremity assessment (FMLE) score ranges. Age ranges: ≤50, 51–60, 61–70, and 71–80 years; speed ranges: 0.1–0.4, > 0.4–0.8, and > 0.8 m/s; and FMLE ranges: < 20, 21–25, 26–30, > 30. When multiple groups could achieve minimization, participant group was assigned randomly. A study coordinator performed minimization, enrollment, and allocation. Group assignment was concealed from investigators performing pre-, post-, and 3-month follow-up assessments. Testing and training were performed in the Brain Plasticity Laboratory at UIC.
A power analysis based on pilot data indicated that 20 participants per group were required (α = 0.05, 1-β = 0.80) to detect between group differences in change in walking speed (akin to a group X time interaction effect) of 0.3 m/s. An analysis performed after testing ~ 20 participants per group found no group differences for change in comfortable walking speed (largest conditional power, Z
k = 0.08). A futility analysis was performed to determine whether testing 5 additional participants per group would likely yield a significant between group difference [
21]. The probability of rejecting a false null hypothesis after the addition of these participants was 0.006% (futility: 0.9994), indicating that there is almost no chance that testing these additional participants would yield a significant between group difference.
Motor priming
Participants were assigned an intervention group dictating the priming received prior to HISTT: 1) control–15 min of rest, 2) tDCS–15 min of stimulation-based priming with tDCS only, 3) AMT–15 min of movement-based priming with ankle motor tracking (AMT) and sham tDCS, and 4) tDCS+AMT–15 min of concurrent priming with tDCS and AMT.
The AMT and tDCS+AMT groups performed a skilled visuomotor ankle motor control task with the paretic leg, analogous to movement-based priming approaches in the upper limb [
18]. Participants were seated comfortably with the paretic leg secured to a custom ankle tracking device and were instructed to perform dorsiflexion and plantarflexion to match ankle position to a moving sinusoid displayed on a screen [
22,
23]. The sinusoid had a random frequency (0.2–0.4 Hz) and amplitude (60–80% of each participant’s comfortable range of motion). Two 60-s practice trials and twelve 60-s trials were performed with 60 s of rest after every four trials. In the tDCS+AMT group, tDCS was applied concurrently with ankle tracking. In the AMT group, tDCS electrodes were affixed, but sham stimulation was applied. Ankle position was sampled at 1000 Hz with Spike2 (CED, UK).
Facilitatory tDCS was applied with a constant current stimulator (Chattanooga Iontophoresis, DJO Global, CA, USA). A saline-soaked sponge electrode (anode: 5 × 2.5 cm) was placed over the leg representation of the ipsilesional motor cortex (M1) identified during the pre-test. A carbonized dispersive electrode (cathode: 4.5 × 5.5 cm) was placed over the contralesional supraorbital region. After a 30-s ramp-up, 1 mA current was applied for 15 min. We have demonstrated focality and effectiveness of lower limb tDCS with low current intensities and a small active electrode (1 mA with a ~ 8-cm
2 electrode vs. standard 2 mA with a 35-cm
2 electrode). The current density (0.08 mA/cm
2) and total charge (0.072 C/cm
2) are comparable to other studies and within safety limits [
16]. For sham stimulation in the AMT group, the stimulator was turned off after the 30-s ramp-up.
High-intensity speed-based treadmill training (HISTT)
Before training, participants received ~ 10 min of lower limb stretching. Participants performed 12 sessions (4 weeks, 3 sessions/week) of HISTT [
24]. To start each week, maximal walking speed was assessed with the 10-m walk test (10MWT). Treadmill sessions started (warmup) and ended with 5 min of walking at 50% of the weekly maximal walking speed. After warmup, high-intensity intervals were performed. For each interval, treadmill speed was increased over a 2-min period up to the peak speed that was safe and tolerable. Peak speed was held for 10 s. After each interval, participants walked at their warm-up, recovery speed until heart rate (HR) was within 5 bpm of warmup. If HR did not decrease within 4 min, the treadmill was stopped, and participants stood until HR reached the required level. At the end of intervals and recovery periods, Rating of Perceived Exertion (RPE; Borg 10-point scale) was assessed. Total walking time per session was 40 min, less time during which the treadmill was paused. Peak HR, speed, and distance walked were recorded. For safety, participants wore a harness without body-weight support and were allowed to hold onto handrails. Minimal manual assistance with hip and knee flexion at toe off was provided as needed.
Over the 4-week training period, peak treadmill speed was continuously increased. If the participant could safely maintain the peak speed achieved during an interval, treadmill speed was increased for the subsequent interval. At speeds below 3.3 mph (1.48 m/s), each increase in peak speed was 10% of the previous peak speed. To avoid transition to running, increases in peak speed were relatively smaller when increasing above 3.3 mph. If participants displayed foot dragging, needed excessive manual assistance, displayed other signs of instability, or had an excessive increase in HR during an interval, peak speed was decreased by 10% for the subsequent interval.
Outcomes
Outcomes were assessed at pre-, post-, and 3-month follow-up assessments. Pre- and post-assessments were performed within 2 days of the first/last training session; the 3-month follow-up was performed ~ 90 days after the last training session to investigate retention. A single laboratory member performed all assessments and was blinded to group assignment.
Primary outcome: walking speed
Participants performed two trials of the 10MWT walking at a “normal comfortable” speed. Time to complete trials was recorded with a stopwatch, and walking speed was computed from the mean across trials. Walking was performed without assistive devices whenever possible. Participants also performed two 10MWT trials “as fast as safely possible,” to determine a secondary outcome (maximal walking speed).
Secondary outcome: corticomotor excitability (CME)
Muscle activity was recorded from the tibialis anterior (TA) of both legs with surface EMG. The skin was shaved and prepped. A reference electrode was placed over C7. EMG data were sampled with an amplifier system (Bagnoli 8, Delsys, MA, USA; frequency: 2000 Hz, gain: 1000, band pass filter: 20–450 Hz) and recorded with Spike2. Participants performed 3 maximal voluntary isometric contractions (MVIC) of the ankle dorsiflexors. Contractions were held for ~ 5 s and performed bilaterally. Visual feedback and verbal encouragement were provided.
TMS was used to assess CME using established procedures [
22,
25,
26]. Participants performed isometric ankle dorsiflexion at 10% MVIC with both limbs. Stimulation was applied from a single-pulse stimulator (Magstim 200, Magstim Inc., MN, USA) through a double-cone coil oriented in the posterior-anterior direction. The TMS coil was systematically moved to identify the “hotspot,” the location with the largest, consistent contralateral motor evoked potentials (MEPs). The active motor threshold was determined as the minimum stimulus intensity eliciting MEPs with a peak-to-peak amplitude of ≥0.1 mV in 4 of 8 trials. Recruitment curves were generated by applying stimulation at seven intensities (6 stimuli at each of 80, 90, 100, 110, 120, 130, and 140% of active motor threshold). MEPs were rectified and the area was calculated from MEP onset to offset. The relation between average MEP area and the corresponding stimulation intensity (% active motor threshold) was described with a linear function. CME was characterized as the slope of this function.
As not everyone responds with upregulation to tDCS [
27,
28], we subcategorized participants in the tDCS and tDCS+AMT groups into up-regulators and down-regulators. After completion of recruitment curves during the pre-assessment, tDCS was applied for 15 min at rest. CME was re-assessed after tDCS with TMS at 120% AMT, and the percent change in MEP area was calculated.
Other secondary outcomes
-
Walking endurance, 6-min walk test (6mWT). Participants walked at their comfortable pace for 6 min, and the distance covered was recorded.
-
Static balance and fall risk, Berg Balance Scale (BBS). Participants completed 14 static and dynamic balance activities. Maximal score: 56.
-
Health status after stroke, Stroke Impact Scale (SIS). Participants completed a 59-item questionnaire. Strength, hand, ADL/IADL, and mobility domain scores were combined as the SIS-16.
Exploratory outcomes
-
Mobility, Timed Up and Go Test (TUG). Participants rose from a chair, walked 3 m, turned around a cone, walked back, and sat down. Mean time to complete two trials was recorded.
-
Dynamic balance, Mini Balance Evaluation Systems Test (miniBESTest). Participants completed 14 dynamic balance activities. Maximal score: 28.
-
Motor impairment, FMLE. Participants were scored on 7 movement categories. Maximal score: 34.
-
Confidence in performing ambulatory activities without falling, Activities-Specific Balance Confidence Scale (ABC). Participants completed a 16-item questionnaire.
Statistical analysis
We performed intention-to-treat analyses with linear mixed modeling, which accounts for all data and assumes that data is missing at random. Random intercepts accounted for repeated observations, and a compound symmetry repeated covariance type was used. To predict outcomes, linear mixed models used independent factors of time (pre, post, 3-month: 2 degree of freedom (df) test to determine if the 3 time points are equal) and group (control, tDCS, AMT, tDCS+AMT: 3df test). Primary focus of linear mixed modeling was on group X time interactions (6df test). Significant contrasts are reported. To predict variables from treadmill training sessions (average weekly value) and weekly maximal walking speed, linear mixed models used independent factors of week (1–4: 4df test) and group (3df test). Post-hoc pairwise comparisons were performed with Bonferroni correction. Baseline characteristics were compared between groups with 1-way ANOVA and chi-square analysis. Post-hoc group comparisons were performed with Tukey HSD. Cohen’s d was used for effect sizes. Statistical analyses were performed with SPSS Statistics (IBM, NY, USA), with a P value considered statistically significant at 0.05.
Discussion
This is the first study to examine the effects of priming paired with a speed-based, high-intensity gait training paradigm in chronic stroke. Four weeks of HISTT led to improvements in walking speed and endurance, which were partially maintained 3 months after training. Priming with tDCS, AMT, or both did not enhance the effects of HISTT on walking speed, but tDCS+AMT enhanced ipsilesional CME, which was retained at follow-up. Responsiveness to tDCS influenced the effects of priming on HISTT.
HISTT was feasible and effective, improving walking speed and endurance in all groups, with retention 3 months later. These results suggest that HISTT leads to significant and long-term improvements in walking. Consistent with our findings, small-scale studies using aerobic HIIT show improvements in aerobic capacity, walking speed, and walking endurance after stroke [
29,
30]. The current study and these previous studies all found ~ 10% improvement in walking speed and endurance following high-intensity training. Improvements in walking speed exceed those from progressive treadmill training [
11‐
13] with less training time, making HISTT a more efficient approach to rehabilitation after stroke. Mobility, balance, and health status also improved with HISTT. Our findings also suggest that longer duration training may be beneficial. After 4 weeks of HISTT, improvements in walking speed did not plateau. From week 3–4, maximal walking speed improved by 0.04 m/s. At this rate, an additional improvement of 0.16 m/s (MCID) could be expected after 8 weeks of HISTT. Studies involving up to 6 months of progressive treadmill training also have not found a plateau in improvements [
31,
32]. Future studies should investigate the effects of longer duration HISTT.
After stroke, cortical excitability is often suppressed in the ipsilesional hemisphere and enhanced in the contralesional hemisphere [
33], which may be secondary to imbalanced interhemispheric inhibition [
34,
35]. We expected tDCS and motor priming to enhance the effects of HISTT because both have been shown to modulate cortical excitability [
18,
33,
36‐
38]. Furthermore, tDCS has been associated with a variety of functional improvements in the lower limb, including enhanced strength, motor control, mobility, and balance [
23,
38‐
41]. However, in the current study, we did not find evidence that combining 4 weeks of HISTT with tDCS, AMT, or both enhances improvements in walking speed or secondary behavioral outcome measures. We may have failed to demonstrate the efficacy of motor priming because HISTT is a strong intervention that masks the effects of priming. Many in the control group had meaningful improvement in comfortable walking speed, indicating that HISTT led to improvements in speed that may have saturated the response potential or masked the effects of motor priming. Additionally, our motor priming interventions were 15 min, while HISTT was 40 min—over 2.5X longer. Motor priming may be more effective when: 1) applied for longer durations, 2) paired with shorter bouts of HISTT, 3) paired with a less intensive intervention, or 4) applied concurrently with or subsequent to HISTT.
Currently, there are 7 published gait studies that used repeated sessions of tDCS paired with walking training for stroke [
42‐
48]. Some of these studies found that tDCS enhanced endurance when paired with robotic gait training [
45] and enhanced walking speed when paired with body-weight-supported treadmill training [
46]. However, others found that tDCS paired with robotic gait training did not improve walking speed more than gait training alone [
42‐
45,
48]. Similarly, pairing tDCS with standard physical therapy did not enhance improvements in walking for acute or subacute stroke survivors [
47,
49,
50]. The study that found a significant effect of tDCS on speed [
46] may have done so because the intervention paired with tDCS yielded minimal change in speed (0.04 m/s), allowing identification of priming effects. Although Seo et al. 2017 found a significant effect of tDCS on endurance, they failed to find an effect on speed [
45]. Our study expands on these prior studies by pairing tDCS with a higher-intensity intervention (HISTT) and by investigating a larger sample size, and we show no additional benefit of tDCS.
In the current study, we broadly characterized the effects of HISTT and motor priming on walking by quantifying changes in walking speed. Stroke is associated with a number of other changes in walking (e.g. spatiotemporal kinematics, kinetics, and postural stability) which may have been affected by our study interventions. There are no studies evaluating the effect of HISTT on all these walking characteristics, but high-intensity stepping does result in changes in both kinematics and kinetics [
51,
52]. Additionally, the application of tDCS over 4 weeks improves postural control as assessed with the Tinetti test [
53]. In contrast, other studies suggest that our interventions would not have affected other walking characteristics. A single session of tDCS does not alter spatiotemporal kinematics or kinetics during walking [
54‐
56]. Moreover, other studies have found that combining tDCS with walking training does not alter walking spatiotemporal kinematics [
42,
48]. Understanding the impact of HISTT and motor priming on specific walking impairments (e.g. spatiotemporal kinematics, kinetics, and postural stability) is an important area for future research.
The premise of previous work and the current study is that cortical priming will enhance neuroplastic changes in the brain that occur in response to walking training. However, it is well known that spinal central pattern generators are important for walking in humans [
57]. Consequently, cortical priming strategies may have a limited benefit, and spinal priming strategies may be more effective. Several studies have investigated this possibility by applying trans-cutaneous spinal direct current stimulation (tsDCS) in conjunction with robotic gait training [
48,
58]. These studies found that combining tsDCS with tDCS or trans-cutaneous cerebellar direct current stimulation is more effective than any type of stimulation in isolation for improving walking endurance. These studies suggest that tsDCS may be an important component of future priming strategies.
In our study, the tDCS+AMT group had decreased motor threshold and increased recruitment curve slope in the ipsilesional hemisphere after training, both indicators of increased CME. These effects were retained 3 months after the end of training and are consistent with our finding that a single session of tDCS and HISTT increases CME [
20]. Overall, these results suggest that motor priming paired with HISTT leads to long-term enhancement of CME. We may have failed to find other significant differences between groups for TMS measures because MEPs were only available in higher-functioning individuals, presenting a ceiling effect for outcomes derived from TMS. Studies evaluating tDCS modulation of CME have provided mixed results, with some raising questions about the variability and reliability of tDCS [
59]. Nonetheless, a recent metanalysis concluded that anodal tDCS significantly improves CME [
60]. Interestingly, low currents (≤1 mA) administered for > 10 min (charge of > 0.029 mA/cm
2) had a greater effect on CME than higher currents. It is still unclear whether these changes in CME benefit functional activities.
We further noticed that 53% of participants receiving tDCS (tDCS and tDCS+AMT groups) had an increase in CME following tDCS. Variability in response to tDCS was expected and is similar to other studies of tDCS responsiveness [
27,
28]. These up-regulators had 300% greater improvements in comfortable and maximal walking speed than participants who had a decrease in CME following tDCS (down-regulators). These findings suggest that, in some stroke survivors, applying anodal tDCS to the ipsilesional hemisphere may paradoxically decrease CME, limiting the effectiveness of priming with tDCS. Other investigators have also suggested that applying anodal tDCS to the ipsilesional hemisphere may not aid recovery in all individuals and that neuromodulatory interventions should be individually tailored [
61]. Thus, response to tDCS may help identify individuals who would benefit from motor priming. Alternatively, responsiveness to tDCS may also reflect a generalized potential for neuroplasticity and functional improvement. If true, greater improvements in walking speed in tDCS up-regulators may reflect greater responsiveness to HISTT, and not (or in addition to) greater responsiveness to tDCS.
A recent review suggests that tDCS is safe in individuals with stroke, with minimal side effects [
62]. Additionally, most studies applying tDCS after stroke have used TMS-based exclusion criteria, which is conservative for tDCS. Thus, it is likely that a large portion of the population could receive this intervention safely. Similarly, several studies support the safety of HIIT in individuals with chronic stroke [
63,
64]. Our study is in line with these findings because we found minimal side effects from tDCS or HISTT. Beyond safety, the feasibility of home-based tDCS [
65] and HIIT interventions [
66] in chronic stroke suggests that these interventions could easily translate to clinical practice.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.