Transcranial Direct Current Stimulation Over the Motor or Frontal Cortex in Patients With Chronic Ankle Instability

Study Design

The present study implemented a double-blind parallel randomized controlled trial. Participants performed a 4-week intervention consisting of 8 rehabilitation sessions that emphasized muscle activation and motor planning. Participants were tested for outcome measures at week 0 (baseline), week 2 (midtraining), week 4 (posttraining), and week 6 (retention). Independent variables included group (motor, frontal, sham) and time (baseline, midtraining, posttraining, retention). Dependent variables included measures of neural excitability, dynamic postural control, lower-leg muscle activation and reaction times during a reactive hop test, and patient-reported outcome measures (modified Disablement in the Physically Active Scale [mDPAS], Foot and Ankle Ability Measure [FAAM], Tampa Scale of Kinesiophobia [TSK], and Global Rating of Change [GROC]). Participants were masked to group allocation, and both therapists and assessors were masked to whether participants received an active or sham current, although they were aware of electrode location (motor or frontal). This clinical trial was registered on clinicaltrials.gov (NCT06024720).

Participants

Forty-six individuals with CAI were recruited for the present study. Classification with CAI followed guidelines from the International Ankle Consortium,²² including experiencing their first ankle sprain more than 1 year before study enrollment and scoring above a 10 on the Identification of Functional Ankle Instability instrument.²³ Participants had no history of foot, ankle, or lower leg fractures or surgery or injuries restricting physical activity in the 3 months before study enrollment and reported no red-green color vision deficiency. Additionally, participants met criteria for the safe practice of transcranial magnetic stimulation (TMS) and tDCS, including no personal or immediate family history of seizure or epilepsy, metallic implants or medication infusion devices, skull abnormalities, frequent headaches or migraines, concussion within 6 months, or medications that raised the risk of seizure.^24–26 Sample size was based on preliminary data, which reported effect sizes of ${\text{η}}_p^2$ between 0.07 and 0.14.¹⁶ To achieve statistical power (1 − β) of .95 with the previously observed effect (f = 0.27) and a level of significance (α) at .05, 11 participants were required per group. To account for potential attrition of up to 25%, we aimed to recruit 45 total participants. All participants provided informed consent as approved by the Appalachian State University Institutional Review Board (22-0050).

Outcome Measures

Neural excitability was assessed using the Hoffmann reflex (H-reflex) and TMS for segmental and corticospinal excitability, respectively. For these measurements, participants were instrumented with electromyography (EMG) electrodes over the tibialis anterior (TA), peroneus longus, and soleus. The area over each muscle was palpated, shaved if necessary, cleaned with isopropyl alcohol, and abraded, and an active electrode connected to an amplifier (B&L Engineering) was placed along the muscle.²⁷ The H-reflex was assessed first with the patient in a prone position. A bar electrode connected to a constant current stimulator (DS7R; Digitimer LTD) set to 300 V was placed in the popliteal fossa. To assess H-reflex across all target muscles, the location of the sciatic nerve before its bifurcation was identified using 10-mA pulses and used for subsequent testing.²⁸ Pulses (1 millisecond duration) were applied every 10 seconds beginning at a stimulation of 0 and increasing by 2 mA each pulse until a plateau response was noted across all recorded muscles. The ratio of the maximal reflexive response (H_max), occurring 40 to 80 ms from stimulus, was compared with the maximal motor response (M_max), occurring from 10 to 30 ms from stimulus, to derive H_max : M_max for each muscle.²⁸

After H-reflex assessment, cortical excitability was assessed simultaneously across all muscles using TMS. Participants were seated and familiarized with TMS before the M1 was located by providing submaximal pulses and observing the location yielding the largest response in the TA.²⁹ The TA was selected due to its greater cortical representation relative to the peroneus longus and soleus.^30,31 Following these procedures, 40 to 50 stimuli, ranging from below the previously noted motor threshold to above a maximal response that would be expected from each muscle, were applied to obtain a stimulus-response curve from each muscle.³² The stimulus-response curve was used to estimate the resting motor threshold (RMT) and maximum motor evoked potential (normalized to M_max) for each muscle.²⁹ All neural outcomes were assessed in an electromagnetically-shielded laboratory.

Dynamic balance, muscle activation, and reaction times were assessed using a reactive hop. Participants were reinstrumented with EMG as described above using a system that allowed for free movement (Bagnoli-4; Delsys Inc). Participants were positioned in unipedal stance in the middle of 3 in-ground force plates (60 × 90 cm; Bertec), with a 5-cm vertical hurdle placed between each force plate. Three reactive lights (ROXProX; A-Champs) were placed on tripods surrounding the participant, with one 10 feet directly in front of the participant (memory light), and the other 2 placed at the front outside corners of the adjacent force plates (trigger lights). Participants were familiarized with the task, consisting of monitoring the memory light, which flashed 1 of 4 colors every 2 seconds, as participants were instructed to recall the previous 3 colors in order. At a random time in a 15-second interval, the trigger lights illuminated, and participants were instructed to hop toward the green-lit light. When participants were within 50 cm of the trigger light, it would deactivate and send a reaction time (ie, time between trigger light illumination and deactivation) to the linked software via Bluetooth connection (ROXPro Android application; A-Champs). Participants were instructed to land on the affected side, regain balance for 15 seconds, and recite the 3 colors that occurred before the trigger light. To offset fatigue, participants did 3 consecutive trials and then were provided a 1- to 2-minute rest period. Three trials were provided for familiarization and practice, and then a minimum of 5 hops to each side were collected. If participants did not successfully complete the task (eg, did not clear the hurdle, touched opposite limb down within 15 seconds, hopped in the wrong direction), an error was recorded on the data collection sheet and the trial was not included in the 5 that needed to be completed to each side. This method provided strong reliability for reaction time and color memory across multiple trials (intraclass correlation coefficient [1,k] between 0.707 and 0.828). Force plate and EMG data were collected in custom LabVIEW software (National Instruments) at 1000 Hz.

Dynamic balance was quantified from force plate data using dynamic postural stability indices (DPSIs), including anteroposterior (APSI), mediolateral (MLSI), and vertical (VSI) components.²⁰ Electromyography data were band-pass filtered (20–400 Hz), rectified, and low-pass filtered (10 Hz) to create a complete linear envelope, and normalized to peak activation across all trials. Average activation for each muscle was extracted in the 500 ms before and after landing on each force plate. Reaction times were recorded from the reactive lights using the ROXPro Android application, and color memory was written down noting the number of correct colors out of the 3 displayed. Trials were further stratified into medial and lateral hops, relative to the test leg.

Patient-reported outcome measures were collected in the Research Electronic Data Capture (REDCap) tools hosted by Appalachian State University.³³ Foot and ankle function was assessed using the FAAM, including Activities of Daily Living (ADL) and Sport subscales.³⁴ Health-related quality of life was assessed using the mDPAS.³⁵ Kinesiophobia related to sport activity was assessed using the TSK 17-item scale.³⁶ Lastly, global changes from baseline were assessed at midtraining, posttraining, and retention using the GROC, which consists of a 15-point scale, where 0 indicated no change (about the same) from baseline, 7 indicated a very great deal better and −7 indicated a very great deal worse.³⁷ As a performance-based metric of patient function, participants also performed a side-hop test, consisting of 10 side-to-side hops over 2 parallel lines placed 30 cm apart.³⁸ Participants were provided a full practice trial, and the time to complete a single trial was extracted. To test fidelity of masking efforts, at the conclusion of the retention visit, participants were asked whether they felt they received the real or sham current.

Intervention

After the baseline test session, participants were randomized into either motor, frontal, or sham groups using a block-randomization scheme (randomized block sizes of 3–9). To maintain masking, individuals were assigned a 6-digit code that was entered into a stimulator (1 × 1 CT; Soterix) that corresponded with an active current (2.0 mA provided over 18 minutes) or a sham current to mimic some sensation while maintaining participant blinding (2.0 mA provided over 1 minute).^16,19 The codes were accompanied by a stimulus location, instructing researchers where to place electrodes (frontal or motor location). This allowed for participants in the sham group to be near-equally allocated to have electrodes placed at motor or frontal locations.

Participants reported to a separate laboratory and were instrumented with tDCS. For participants receiving the motor location, a 5 × 3 anode (EASYPad; Soterix) was placed at the C3/C4 location of the International 10-20 System, corresponding with the hemisphere contralateral to the test ankle, and the same-sized cathode was placed at the opposite supraorbital area (ipsilateral to the test ankle).¹⁶ For participants receiving the frontal location, a 5 × 7 EASYPad anode was placed over the F3/F4 location of the International 10-20 System, corresponding with the hemisphere contralateral to the injured ankle, and the same-sized cathode was placed at the opposite F3/F4 location.¹⁸ Electrodes were saturated with 6 to 8 mL of saline before placement on the individual, and electrodes were secured using an elastic fastener set provided by the manufacturer. Good electrode impedance was assured before beginning the stimulation. Participants were instructed to sit for 2 minutes after beginning stimulation, and then proceed through the exercise protocol.

Full details of the rehabilitation progression, including stages of progressions achieved by each group, are included in the Supplemental Material. First, participants performed hurdle walking, performed laterally and at oblique angles. Second, participants performed a unipedal stand-and-reach task with a go/no-go component, in which participants balanced on the test side and reached with the nontest side toward a light trigger placed at 50% of leg length in front of the participant. Lastly, participants performed 6 agility ladder exercises consisting of 3 rounds of lateral stepping and 3 rounds of a shuffle step (Figure 1). Treatment sessions were designed to take 18 to 20 minutes, aligning with the duration of tDCS administration. In all cases, tDCS remained through the entire 18-minute timer.

View Figure

• Download as Powerpoint
• Download Figure

Data Analysis

All data were assessed using linear mixed models to account for the nested nature of the data and allow for the inclusion of partial data in an intention-to-treat analysis. Descriptives were examined and baseline characteristics were compared for a group effect. Each variable was assessed for fixed effects of group (motor vs frontal vs sham), time (baseline, midtraining, posttraining, and retention), and group-by-time interaction effect. In the case of a significant effect on omnibus tests for fixed effects, we reported the marginal ( $R_M^2$) and conditional ( $R_C^{2 }$) R² values to estimate the contribution of fixed and random effects on the linear mixed models.³⁹ Further, for significant fixed effects, parameter estimates were examined post hoc to assess the sources of significant differences. To allow for understanding the magnitude of the observed changes, 95% CIs were reported surrounding mean differences (MDs). An a priori level of significance (α) was set at .05.

Demographics

Forty-six individuals were recruited for this study, with 37 completing all testing sessions and considered compliant (≥6 out of 8 training sessions completed) (Figure 2). There were no significant differences in attrition across groups ( ${\text{χ}}_2^2$ = 1.342, P = .511), with only the frontal group having retention below 80%. Demographic information for all groups is presented in Table 1. No significant differences were observed between groups for sex ( ${\text{χ}}_2^2$ = 5.004, P = .082), age (F_2,41 = 0.658, P = .523), height (F_2,41 = 0.963, P = .392), mass (F_2,41 = 0.254, P = .777), or Identification of Functional Ankle Instability scores (F_2,41 = 0.880, P = .422). A χ² analysis was conducted to explore whether individuals in each group perceived themselves to have received an active or sham current. In the motor, frontal, and sham groups, 72.7%, 75.0%, and 72.7% of participants thought they had received an active current, respectively ( ${\text{χ}}_2^2$ = 0.015, P = .992).

View Figure

• Download as Powerpoint
• Download Figure

Table 1.Participant Characteristics

View Fullscreen

Neural Excitability

Means, SDs, and omnibus effects for H_max : M_max, RMT, and maximum motor evoked potential are presented in Table 2. No significant main effects of time, group, or group-by-time interactions were observed for neural excitability variables within any muscle.

Table 2.Neural Excitability Means ± SDs Across Groups and Time Points and F Values for Main Effect of Time and Group-by-Time Interaction Effect for Each Variable

View Fullscreen

Dynamic Postural Control and Muscle Activation

Means, SDs, and omnibus effects for DPSI, APSI, MLSI, and VSI during medial and lateral hops are presented in Table 3. No significant group or group-by-time interaction effects were noted for any dynamic postural control variables. Significant main effects of time were observed for DPSI and VSI on the lateral hops (DPSI: F_3,100.2 = 4.389, P = .006, $R_M^2$= 0.120, $R_C^{2 }$= 0.559; VSI: F_3,100.2 = 4.225, P = .007, $R_M^2$= 0.118, $R_C^{2 }$= 0.555). Significant improvements in postural stability from baseline were noted at midtraining (DPSI: MD = −0.04; 95% CI = −0.06, −0.02; P < .001; VSI: MD = −0.04; 95% CI = −0.06, −0.02; P < .001), posttraining (DPSI: MD = −0.03; 95% CI = −0.05, 0.00; P = .020; VSI: MD = −0.03; 95% CI = −0.05, −0.00; P = .023), and retention (DPSI: MD = −0.03; 95% CI = −0.06, −0.01; P = .014; VSI: MD = −0.03; 95% CI = −0.05, −0.01; P = .017) for lateral hops.

Table 3.Means ± SDs for Balance Variables on Medial and Lateral Hops Across Groups and Time Points and F Values for Main Effect of Time and Group-by-Time Interaction Effect for Each Variable

View Fullscreen

Means, SDs, and omnibus effects for muscle activation during medial and lateral hops are presented in Table 4. No significant main effects of time, group, or group-by-time interaction effects were observed for average activation within any muscle before or after landing.

Table 4.Means ± SDs for Electromyography Variables on Medial and Lateral Hops Across Groups and Time Points and F Values for Main Effect of Time and Group-by-Time Interaction Effect for Each Variable; All Units Are Percentages of Ensemble Peak

View Fullscreen

Reaction Times and Cognitive Performance

Reaction times and cognitive performance are presented in Table 5. No significant changes in reaction times on medial or lateral hops were observed for the main effects of group, time, or group-by-time interaction effect. Improvements were observed for the number of colors correctly recalled over time (F_3,108.0 = 9.83, P < .001, $R_M^2$= 0.146, $R_C^{2 }$ = 0.478), with significant differences from baseline noted at midtraining (MD = 0.2; 95% CI = 0.1, 0.3; P < .001), posttraining (MD = 0.2; 95% CI = 0.1, 0.4; P < .001), and retention (MD = 0.3; 95% CI = 0.2, 0.4; P < .001).

Table 5.Means ± SDs for the Side Hop Test, Choice Reaction Time, and Cognitive Performance; F Values for the Main Effect of Time and the Group-by-Time Interaction Effect for Each Variable

View Fullscreen

Side Hop Test

Significant improvements were observed in side-hop test performance over time (F_3,104.5 = 11.004, P < .001, $R_M^2$= 0.096, $R_C^{2 }$ = 0.674), but no group or group-by-time interaction effect was observed (Table 5). Significant differences from baseline indicating improvement performance were noted at midtraining (MD: −3.2; 95% CI = −4.7, −1.6, P < .001), posttraining (MD: −3.9; 95% CI = −5.5, −2.3; P < .001), and retention (MD: −4.2; 95% CI = −5.9, −2.6; P < .001).

Patient-Reported Outcomes

Means, SDs, and omnibus test effects are presented in Table 6. The FAAM-ADL revealed no significant group or group-by-time interaction effects, but the time effect was significant (F_3,105.7 = 6.775, P < .001, $R_M^2$= 0.036, $R_C^{2 }$= 0.846). Parameter estimates were significant at posttraining (MD = 3.3; 95% CI = 1.3, 5.3; P = .002) and retention (MD = 4.36; 95% CI = 2.3, 6.4; P < .001) but not at midtraining (MD = 1.7; 95% CI = −0.3, 3.7; P = .091). The FAAM-Sport revealed no significant group or group-by-time interaction effects, but the time effect was significant (F_3,106.5 = 13.58, P < .001, $R_M^2$= 0.062, $R_C^{2 }$ = 0.830). Parameter estimates were significant at midtraining (MD = 3.8; 95% CI = 0.7, 6.9; P = .019), posttraining (MD = 8.1; 95% CI = 4.9, 11.3; P < .001) and retention (MD = 9.35; 95% CI = 6.1, 12.6; P < .001). Both scales followed the same pattern of increased perceived function over time.

Table 6.Means ± SDs for Patient-Reported Outcome Measures and F Values for the Main Effect of Time and the Group-by-Time Interaction Effect for Each Variable

View Fullscreen

The mDPAS Physical subscale revealed no significant group or group-by-time interaction effect, but a significant effect of time was observed (F_3,106.9 = 8.049, P < .001, $R_M^2$= 0.045, $R_C^{2 }$= 0.791). Parameter estimates showed significant differences from baseline at posttraining (MD = −3.8; 95% CI = −4.8, −1.8; P < .001) and retention (MD = −4.7; 95% CI = −5.7, −2.6; P < .001), but not at midtraining (MD = −1.8; 95% CI = −3.8, 0.2; P = .083). The mDPAS Mental subscale also demonstrated no significant group or group-by-time interaction effect, but did demonstrate a significant effect of time (F_3,107.5 = 3.43, P = .020, $R_M^2$= 0.045, $R_C^{2 }$= 0.791). Parameter estimates showed significant differences at posttraining (MD = −0.8; 95% CI = −1.5, −0.1; P = .034) and retention (MD = −0.9; 95% CI = −1.6, −0.2; P = .014), but not at midtraining (MD = −0.0; 95% CI = −0.7, 0.6; P = .919). A similar pattern was observed for the TSK, as there was no significant group or group-by-time interaction effect, but a significant effect of time was observed (F_3,107.7 = 2.719, P = .048, $R_M^2$= 0.040, $R_C^{2 }$= 0.772). However, post hoc parameter estimates revealed no significant differences at any given time for the TSK.

The GROC followed a similar trend, demonstrating no significant group or group-by-time interaction effect, but a significant effect of time (F_2,69.6 = 20.046, P < .001, $R_M^2$= 0.129, $R_C^{2 }$= 0.747). Parameter estimates showed significant differences at posttraining (MD = 1.3; 95% CI = 0.7, 1.8; P < .001) and retention (MD = 1.6; 95% CI = 1.1, 2.1; P < .001).

Functional Improvements After the Intervention

Across all groups, participants displayed improvements in patient-reported outcome measures, side-hop test performance, and dynamic balance during lateral hops. These improvements were observed after the 4-week intervention, regardless of tDCS application, suggesting improvements were tied to the exercises performed. The exercises in the rehabilitation intervention—consisting of stepping over obstacles, reactive semidynamic balance exercises, and agility—were selected to demand activation of motor-execution areas of the cortex while simultaneously challenging planning areas, which may be less efficient in individuals with CAI.⁷ Implementing motor-learning strategies and dual-task focused interventions has recently grown in patients with musculoskeletal injury,^40,41 with improvements generally observed in balance performance.⁴² Given deficits in motor planning in patients with CAI, it is possible that incorporating motor-planning–focused exercises into rehabilitation was sufficient to improve patient-reported function, supporting the implementation of these exercises in rehabilitation efforts for CAI.⁴⁰

Although the observed improvements in patient-reported function were statistically significant, caution should be exercised when interpreting the clinical significance of these data. Many of the outcome measures did not exceed published minimum clinically important differences (MCIDs). Improvements did not exceed the MCID for the FAAM-ADL (MCID = 8),⁴³ TSK (MCID = 4),⁴⁴ and mDPAS (MCID = 9),³⁵ with only the FAAM-Sport (MCID = 9)⁴³ having confidence intervals that exceeded this threshold. For the GROC, indicating participants’ perceived improvement from the intervention, scores observed were in line with the MCID of 2; however, a score of 5 has been recommended to gauge the success of an intervention.⁴⁵ Although we observed improvements from our intervention, it would be important to compare an intervention emphasizing motor planning with a more typical rehabilitation protocol that may emphasize balance and strength training.⁴⁶

Participants were recruited from a general population rather than those specifically seeking care for CAI. As such, although similar across groups, these individuals may have had higher baseline function, limiting the magnitude of potential improvements. However, the observed increases in patient-reported function represent a potentially meaningful change. Clinically, this suggests that bringing patients who exhibit impairments with few participation restrictions through continued rehabilitation emphasizing motor planning and dual tasking may improve function. The responsiveness to the intervention seen across groups may be evidence of unfulfilled potential in traditional rehabilitation, where patients often resume activity and discontinue care for a variety of reasons before advanced, dynamic rehabilitation strategies can be implemented. However, the observed small effect sizes for these clinical improvements may be due to a lack of concurrent neural changes. That is, although individuals gained confidence in their ankle over the course of the intervention, the absence of neural changes (eg, improvements in neural excitability) potentially limited the magnitude of these effects and may reflect a lack of durability to these changes. Overall, these results suggest a degree of effectiveness to the intervention, but a failure to optimally restore patient function.

Efficacy of tDCS

Our a priori hypotheses were that the implementation of tDCS would enhance rehabilitation by driving mechanistic changes that would subsequently improve function. However, nearly all measures tied to mechanistic changes in individuals with CAI did not demonstrate improvements over the course of the intervention. Notably, improvements in segmental or corticospinal excitability—a consistent finding in tDCS-related research—were not observed in this study.⁴⁷ Similarly, no improvements were observed in muscle activation or reaction times during the reactive hop task.

These data are in contrast to previous studies implementing tDCS in individuals with musculoskeletal injuries.^15,16,48 In individuals with CAI, tDCS has been implemented with eccentric strengthening exercises,¹⁶ foot intrinsic muscle strengthening,¹⁵ and balance training,⁴⁸ yielding improvements in neural excitability, muscle activation, and dynamic balance outcomes as well as patient function when compared with a similar sham group.^15,16 Although no evidence exists for the use of DLPFC stimulation in individuals with musculoskeletal injury, improvements have been seen in motor performance of complex walking in healthy individuals.⁴⁹ Alternatively, stimulation over the supplementary motor area, similarly used to improve motor planning, has not demonstrated improvements on motor-planning outcomes in individuals with CAI.⁵⁰ A key difference between the current investigation and these previous studies is the intervention paired with tDCS. Here, the intervention was selected to create both motor execution and planning demands, allowing for all participants to do the same exercise sets despite the intention of their tDCS application (motor execution versus planning). However, our results and prior investigations suggests that the exercise selection may be of utmost importance in the treatment of joint instability. Transcranial direct current stimulation over the M1 may need to be implemented with direct strengthening exercises that have disinhibitory effects (eg, eccentric exercise, plyometrics), whereas frontal cortex stimulation may require interventions that more intentionally create demand on the DLPFC, such as motor imagery or action-observation interventions.⁷

Although early evidence has supported the use of M1 stimulation in patients with CAI, the use of DLPFC stimulation has been less explored.^15,16 The aim of frontal cortex stimulation would be to improve neural efficiency in feed-forward motor planning by improving the brain’s ability to anticipate outcomes; however, patients with CAI may have increased dependence on these frontal areas and facilitation here may not be warranted.¹⁰

Limitations

Certain limitations in study design limit our ability to draw finite conclusions based on our intervention. All participants received exercises for CAI, potentially generating biases to improve scores on patient-reported outcome measures, reflecting a potential Hawthorne effect.⁵¹ A true control group receiving no intervention could have alleviated this concern. Similarly, physical and cognitive performance improvements, such as that on the side-hop test, VSI, or cognitive correct letters, may have been tied to learning effects that may explain some of the improvements we observed. Alternatively, as approximately 75% of participants across all groups felt they had received an active tDCS current, a placebo effect may have aided outcome scores. Inclusion of longer follow-up might have provided a better indication of whether the improvements in patient-reported outcomes were durable beyond 2 weeks after cessation of training. Although efforts were made to control for as many variables at baseline, there is of course a great deal of variability among the deficits observed in patients with CAI that could have been mitigated by controlling for factors such as sex and baseline function before randomization; however, doing so often limits the clinical applicability of the data observed.

Clinical and Research Implications

Recent findings related to the use of brain stimulation in treating musculoskeletal injuries reflect the need for neuromodulatory interventions in the treatment of joint instability. Although it is unclear whether the increase in brain stimulation research in sports medicine has yielded its increased use in clinical practice, the results of this study provide some pause toward tDCS implementation. Placing this investigation in the context of the existing evidence, it appears that the impacts of tDCS are optimized when motor cortex stimulation is combined with direct motor interventions such as eccentric strengthening or foot intrinsic strengthening.^7,15 Conversely, the evidence does not support the use of tDCS over motor-planning areas in these patients.

Our rehabilitation intervention, emphasizing motor planning and dual-tasking components, did yield improvements in function. This emphasizes the need to incorporate these components into rehab, especially in continuing rehabilitation for individuals with few participation restrictions. These findings can be further enhanced by considering individualized changes to mechanical, sensorimotor, and psychological function, reflecting the varied clinical presentation of CAI.³