Bilateral contralaterally controlled functional electrical stimulation reveals new insights into the interhemispheric competition model in chronic stroke

Abstract

Background:

Upper-limb chronic stroke hemiplegia was once thought to persist because of disproportionate amounts of inhibition imposed from the contralesional upon the ipsilesional hemisphere. Thus, one rehabilitation strategy involves discouraging engagement of the contralesional hemisphere by only engaging the impaired upper-limb with intensive unilateral activities. However, this premise has recently been debated and has been shown to be task specific and/or applies only to a subset of the stroke population. Bilateral rehabilitation, conversely, engages both hemispheres and has been shown to benefit motor recovery. To determine what neurophysiologic strategies bilateral therapies may engage, we compared the effects of a bilateral and unilateral based therapy using transcranial magnetic stimulation.

Methods:

We adopted a peripheral electrical stimulation paradigm where participants received one-session of bilateral contralaterally controlled functional electrical stimulation (CCFES) and one-session of unilateral cyclic neuromuscular electrical stimulation (cNMES) in a repeated-measures design. Fifteen chronic stroke participants with a wide-range of motor impairment (Upper Extremity Fugl-Meyer score:15[severe]-63[mild]) underwent single 1-hr sessions of CCFES and cNMES. We measured whether CCFES and cNMES produced different effects on interhemispheric inhibition (IHI) to the ipsilesional hemisphere, ipsilesional corticospinal output, and ipsilateral corticospinal output originating from the contralesional hemisphere.

Results:

CCFES reduced IHI and maintained ipsilesional output when compared to cNMES. We found no effect on ipsilateral output for either condition. Last, the less impaired participants demonstrated a greater increase in ipsilesional output following CCFES.

Conclusions:

Our results suggest bilateral therapies are capable of alleviating inhibition upon the ipsilesional hemisphere and enhancing output to the paretic limb.

Introduction:

For stroke patients with hemiplegia, motor recovery typically plateaus by 6 months and about 65% of patients are unable to incorporate the paretic arm and hand into their daily activities.¹ The motor deficit is thought to persist, in part, because of an imbalance between the ipsilesional and contralesional hemispheres often described using the model of ‘interhemispheric competition.’ According to this model, paresis originates from the loss of corticospinal output to the impaired upper-limb and persists due to additional interhemispheric imbalances. The contralesional (non-lesioned) hemisphere exerts unopposed inhibition on the weak ipsilesional (lesioned) hemisphere, which further reduces output to the paretic limb.^2–4 Based on this model, one therapeutic strategy is to restrict activity of the non-paretic limb and encourage unilateral movement of the paretic limb; this is thought to limit interhemispheric inhibition (IHI) and improve the likelihood of functional recovery.

However, the interhemispheric competition model has recently been debated in the literature, where many studies have refuted its premise,^5–8 suggest that the model is task specific,⁹ or have demonstrated that the model only applies to a subset of the chronic stroke population.^4,10–13 Therefore, it is important to consider therapeutic approaches which may engage alternative neurophysiologic strategies. One such approach described in the literature is bilateral based therapies, where bilateral interventions have the potential to modulate the control and recovery of both the paretic and non-paretic limb and affect the recovery of both unilateral and bilateral tasks.

Several groups, including our own (for review please refer to Cunningham et al 2015¹⁴), have suggested that bilateral therapies may be a useful/feasible alternative for such patients who suffer from extensive ipsilesional damage. For example, some bilateral therapies, such as active passive bilateral training^15,16 and contralaterally controlled functional electrical stimulation^17–20 use movement of the non-paretic limb to drive/assist in the movement of the paretic limb. Movement of the non-paretic limb may help recruit viable, uncrossed pathways projecting from the contralesional hemisphere to the paretic limb, and this may support recovery of the paretic limb. Further, bilateral movements, in general, may invoke mutual disinhibition of both hemispheres, which may support corticospinal output from the ipsilesional hemisphere.^21–24 For example, Knutson and colleagues 2016 have shown that twelve weeks of training with contralaterally controlled functional electrical stimulation (CCFES), a bilateral therapy, reduces impairment and improves hand dexterity more than an equivalent dose of cyclic neuromuscular electrical stimulation (cNMES), a unilateral therapy

Therefore, we sought to understand the differences in mechanisms of bilateral- versus unilateral based therapies by comparing the neurophysiologic effects of CCFES and cNMES. In addition, we investigated whether these effects vary as a function of severity of motor impairment. Patients with chronic stroke received single sessions of CCFES and cNMES in a randomized, crossover experiment. Transcranial magnetic stimulation (TMS) was used to assess IHI imposed upon the ipsilesional hemisphere, ipsilesional corticospinal output, and ipsilateral (uncrossed) output to the paretic limb. Reliability of assessments was also studied by having a cohort of patients undergo a control condition, during which they received no stimulation/training. We hypothesize that bilateral CCFES will result in reduced interhemispheric inhibition of the ipsilesional hemisphere and enhanced ipsilesional and ipsilateral corticospinal output to the paretic upper-limb.

Methods:

Subjects

A convenience sample of fifteen participants in the chronic phase of recovery (> 6 months) following a first-ever ischemic or hemorrhagic stroke were enrolled. We identified the pre-stroke dominant hand based on the Edinburgh Handedness Inventory assessment.²⁵ We included participants with wide-ranging upper limb impairment including those with just trace movement at fingers, thumb or wrist, as well as participants who were well recovered and still reported inadequate ability to use the paretic hand in daily life. Patients were excluded if they possessed any contraindications to the use of TMS, including cardiac pacemaker, metal implant in the head, history of seizures as an adult, or ongoing use of neuro- or psycho-active medications. The Institutional Review Board of the Cleveland Clinic approved the study. All participants provided written informed consent.

Clinical Assessment of impairment of the upper limb

The Upper-Extremity Fugl-Meyer (UEFM) scale was used to measure motor impairment of the paretic upper limb. The UEFM is a reliable and valid measure of post-stroke upper limb motor impairment.^26–31 The UEFM items take into account synergy patterns, isolated strength, coordination, and hypertonia. Volitional movement of the upper limb (shoulder, elbow, forearm, wrist, and hand) is examined in and out of synergies. Each item is graded on a 3-point ordinal scale (0, cannot perform; 1, perform partially; and 2, perform fully) and summed to provide a maximum score of 66.

Design

Participants partook in a randomized, crossover experiment, in which they received a one-hour session of unilateral cNMES and a one-hour session of bilateral CCFES, separated by ≥1-week interval to allow for washout of effects. Before and after each session, indices of neurophysiology were collected using TMS (Fig. 1). To establish within-session and between session reliability of TMS indices, a cohort of 12 of the 15 participants also underwent an additional session (‘control’ condition), during which they received no intervention for an hour but underwent TMS testing.

Figure 1.

Participants partook in 2 separate session separated by a minimum of 1 week: Cyclic Neuromuscular Electrical Stimulation (cNMES) and Contralaterally Controlled Functional Electrical Stimulation (CCFES). Transcranial magnetic stimulation (TMS) was delivered immediately before and immediately following each session. Abbreviations: rMT – resting motor threshold, aMT – active motor threshold, RC – recruitment curve, IHI – Ipsilateral silent period, iMEP – ipsilateral motor evoked potential.

Interventions

For both cNMES and CCFES, we positioned surface electrodes (2-inch square and/or 1.25-inch round) on the forearm to produce finger and thumb extension. A picture of the electrode placement was used to ensure consistent electrode placement between sessions. We targeted the extensor digitorum communis (EDC) and extensor pollicis longus muscles. No more than 2 independent monopolar channels (using a common anode) of stimulation were used for any participant in this study. The stimulators used in this study delivered biphasic rectangular current pulses at a frequency of 35 Hz and amplitude of 40 mA. For each electrode, a maximum intensity (pulse duration) of stimulation was determined as that which produced maximum visible hand opening without pain.

For the cNMES session, the stimulator was programmed to automatically and repetitively ramp the pulse durations from zero to the predetermined maximum intensity which produces functional hand opening in 1 second, maintain the stimulation at maximum intensity for 7 seconds, and then ramp down the stimulation to zero in 1 second, repeating this cycle every 24 seconds. Auditory and visual cues indicated when the stimulation ramped on and off, and the participants were instructed to attempt to volitionally open their paretic hand when the stimulation turned on and to not move the non-paretic hand. For the CCFES session, the stimulator was programmed to modulate the pulse duration from each stimulation channel from zero to maximum in proportion to the amount of opening of an instrumented glove worn on the non-paretic hand. The glove consisted of an assembly of 3 bend sensors in cloth sheaths attached to the dorsal side of the index, middle, and ring fingers (Fig. 1 right). The stimulator issued auditory and visual cues to tell the participant when and for how long to attempt to open both hands during an exercise session. The visual cues instructed participants to open and then relax both hands with timing that was identical to cNMES. Participants were verbally encouraged to volitionally open both hands in response to the audio and visual cues. For both therapies, participants performed stimulation-mediated hand opening exercises for three 15-minute sets separated by 3-minute breaks. Both cNMES and CCFES resulted in 112 repetitions. During the control session, participants sat quietly with their hands resting on a flat surface for the same amount of time as sessions that involved training with cNMES and CCFES.

TMS recordings

Using TMS, we measured three parameters: 1) inhibition imposed from the contralesional hemisphere upon the ipsilesional primary motor cortex (iM1), abbreviated “IHI upon iM1”, 2) output of residual pathways from the ipsilesional hemisphere to the paretic extensor digitorum communis (EDC), termed “ipsilesional output” and 3) output of pathways from the contralesional hemisphere to the paretic EDC (i.e., via ipsilateral motor pathways), termed “ipsilateral output”. The procedures for these measurements are described below.

TMS was delivered to the primary motor cortex (M1) via a figure-of-eight coil (70 mm) connected to a monophasic Magstim device (200² Magstim Co., Whitland, Dyfed, UK). The coil was placed tangential to the scalp with the handle oriented backward and laterally at a 45° angle to the midsagittal axis. MRI-based frameless stereotaxy was used to guide TMS application (BrainSight, v2, Rogue Research Inc., Montreal, Canada).¹² The participants MRI was used for stereotactic navigation if it was available; otherwise, a template MRI was used instead. Motor evoked potentials (MEPs) were recorded in the EDC muscle using surface EMG (PowerLab 4/25T, AD instruments, Colorado Springs, CO). The amplified (X500) EMG signals were filtered (bandpass, 10 Hz – 2 kHz, and notch filtered at 60 Hz), sampled at 4 kHz and stored on a personal computer for offline analysis (LabChart, version 7.3, AD Instruments Inc. Colorado Springs, CO).

The TMS ‘hotspot’ was identified as the scalp site eliciting reliable criterion MEP ≥50μV in at least 5/10 trials at the lowest stimulator intensity. The lowest TMS intensity that elicited criterion-level MEPs while the subject was instructed to keep the paretic EDC relaxed was termed as the resting motor threshold (rMT) (given as % Maximum Stimulator Output, MSO). Since resting state MEPs cannot be elicited in many participants with moderate-to-severe impairment, we also defined active motor threshold (aMT) as the lowest stimulator intensity required to elicit reliable criterion MEPs ≥200µV in at least 5/10 trials in the slightly-contracted EDC (~15% of maximum voluntary contraction (MVC)) (given as %MSO). MVC was collected during maximum isometric finger extension while EMG was recorded from the EDC. Root mean square (RMS) value of the EMG was calculated and displayed in real time on a monitor. Participants performed three trials of MVC lasting 3–5 seconds each where and the maximum RMS was averaged. Participants were provided with a visual target to help them maintain adequate force production.

IHI upon iM1.

IHI was measured using the ipsilateral silent period (iSP). ISP is defined as the suppression of ongoing volitional EMG contraction with TMS given to the ipsilateral motor cortices.^32,33 Here, we measured iSP during 50% MVC contraction of the paretic EDC. Supra-threshold TMS pulses were delivered to the hotspot in the contralesional hemisphere at an intensity of 150–200% rMT. Onset of iSP was defined when paretic muscle rectified EMG fell below the mean pre-stimulus EMG (defined as the activity recorded 100 ms prior to TMS) for 5ms; the termination of iSP was defined when EMG returned to the mean for 5ms.³⁴ Trials in which iSPs could not be elicited through visual inspection were discarded from the analysis. On average stimulator output of 180%± 40% rMT was required to generate iSPs and on average 19±5 out of 25±4 trails were retained for analysis. The same stimulator output was used before and after each session. IHI was computed as percentage decrease in mean rectified EMG during iSP relative to pre-stimulus rectified EMG (equation 1). ISP was used rather than the other commonly used bi-hemispheric TMS technique for measurement of IHI because the bi-hemispheric technique is contingent upon the availability of MEPs in the paretic muscle. But moderate-to-severely impaired patients cannot reliably elicit MEPs. Therefore, many groups including our own, have chosen to use iSPs as it has been shown to be a relevant index of IHI.^12,32

$$\text{\%}\hspace{0.17em}\text{Inhibition=}\frac{\text{Mean}\hspace{0.17em}\text{pre}\hspace{0.17em}\text{stimulus}\hspace{0.17em}\text{rectified}\hspace{0.17em}\text{EMG}-\text{iSP}\hspace{0.17em}\text{EMG}}{\text{Mean}\hspace{0.17em}\text{pre}\hspace{0.17em}\text{stimulus}\hspace{0.17em}\text{rectified}\hspace{0.17em}\text{EMG}}\times 100$$ Equation 1:

Ipsilesional output.

In participants who could elicit MEPs in the actively contracting paretic muscle, ipsilesional output was defined based on an input/output curve. During slight v(~15% MVC) contraction of the paretic EDC, TMS pulses were delivered to the ipsilesional hotspot at incremental intensities (10%) between 90% and 150% aMT, while amplitude of resultant MEPs was noted. Ten pulses were delivered at each TMS stimulus intensity. Order of intensities was randomized and changed after every 5 pulses. The same intensities were used before and after each session. Several periods of rest were given to avoid muscle fatigue. Peak-to-peak MEP amplitudes recorded at each intensity level were ranked in ascending order of size. The top 20% and bottom 20% of the ranked trials (4 out of 10) were then discarded from the analysis.^35,36 This method has been shown to give an accurate representation of centrality.³⁶ We normalized all MEP amplitudes (pre and post) to the size of the maximum MEP acquired at baseline (MEP_Maxima). From this normalized recruitment curve, area under the curve (AUC) and percent change in MEP amplitude at each intensity was calculated.³⁷ Since MEPs could only be evoked in paretic EDC in 9 out of 15 participants, and of those high aMT thresholds precluded testing at intensities higher than 100% MSO in 2 participants, input/output curve measurements were completed in 7 out of 15 individuals.

Ipsilateral output.

Output of ipsilateral pathways dedicated to the paretic hand was studied as size of the ipsilateral MEPs (iMEPs) evoked in the contracting paretic EDC (50% MVC) with supra-threshold TMS (150–200% rMT) given to the contralesional hemisphere. The presence or absence of ipsilateral MEPs (iMEPs) was determined in responses collected over 25 TMS pulses. Based on the work of Tazoe and Perez,³⁸ we first visually inspected iMEPs, and those with amplitudes that exceeded mean pre-stimulus EMG amplitude were labeled as iMEPs. Then an average of all iMEP waveforms was created. The onset of the averaged iMEP was defined as the time point at which its amplitude exceeded the mean rectified pre-stimulus EMG activity by 1SD. Termination of the averaged iMEP was defined as the time point at which its amplitude fell below 1SD of the mean rectified pre-stimulus EMG. Area (size) of the iMEP was then calculated using Equation 2:

$$\text{iMEP}\hspace{0.17em}\text{area=}\frac{\text{area}\hspace{0.17em}\text{of}\hspace{0.17em}\text{rectified}\hspace{0.17em}\text{EMG}\hspace{0.17em}\text{of}\hspace{0.17em}\text{iMEP}}{\text{mean}\hspace{0.17em}\text{prestimulus}\hspace{0.17em}\text{EMG}\hspace{0.17em}\text{*}\hspace{0.17em}\hspace{0.17em}\text{duration}\hspace{0.17em}\text{of}\hspace{0.17em}\text{iMEP}}$$ Equation 2:

Statistics

We tested normality of the data using a Shapiro-Wilk test. Within-session reliability was ascertained by comparing data collected before and after the control condition using the intra-class coefficient (ICC). ICC was conducted with absolute agreement at a p-value of .05, where an ICC value of ≥ 0.6 indicates good to excellent reproducibility. Between-session reliability was also defined using ICC based on data collected at baseline for each session (CCFES, cNMES and control (N = 12)).

Effects of CCFES and cNMES on TMS metrics were compared using a two-way repeated measures ANOVA [Intervention (cNMES, CCFES) X Time] (N = 15). The percent pre to post difference was compared between CCFES and cNMES using paired t-tests for post-hoc analysis. Bonferroni correction was applied to adjust p-value for multiple comparisons. Secondarily, the influence of baseline impairment severity on changes in TMS parameters was explored for each condition using bivariate Pearson’s correlation analysis. All tests were 2-tailed conducted at α-level of significance= .05.

Results

Participant characteristics are presented in Table 1. Mean (± SD) UEFM was 42±16.13 (range: 15 to 63). Lesions were primarily subcortical (9), followed by mixed cortical/subcortical (2), pontine (2), cortical (1). We were unable to determine lesion location for one participant because imaging was unavailable. Fourteen participants were classified as right handed and nine participants had a stroke that affected their dominant hand.

Table 1:

Participant Demographics

	Age	Gender	Time Since Stroke (years)	Dominant Hand	Side of Paresis	Lesion Location	MEP Present (Rest/Active)	UEFMtotal (out of 66)
	60	F	18.6	R	R	Cortical/Subcortical (TH, F)	N/Y	15
	58	M	22.0	R	L	Cortical/Subcortical (ALIC, PLIC, CN, T, F, P, O, Putamen)	N/N	21
	70	F	5.9	R	L	Subcortical (BG, Nucleus Lentiformus)	Y/Y	22
	56	M	4.0	R	R	Subcortical (CR, PLIC, putamen)	N/N	25
	55	M	3.8	R	R	Subcortical (CR, PLIC, TH, putamen)	N/N	26
	51	M	2.8	R	R	Subcortical (PLIC, posterior lentiform nucleus)	Y/N	34
	77	M	7.0	R	R	Subcortical (CR, CN, PLIC, putamen)	Y/Y	42
	56	M	1.9	R	R	Subcortical (Centrum Semi-ovale, BG, CN, Frontal Nucleus Atrophy)	Y/Y	45
	69	M	4.0	R	R	Subcortical (CR, TH, PLIC, putamen)	N/N	47
	69	M	1.6	R	L	Cortical (Mesial frontal, superior frontal)	Y/Y	48
	51	F	1.8	R	R	Subcortical TH, PLIC	Y/Y	52
	73	M	7.0	R	L	Pontine	Y/Y	59
	83	F	3.2	R	L	Unknown	Y/Y	59
	62	M	0.9	R	L	Pontine	Y/Y	60
	57	M	1.1	L	L	Subcortical (CN, TH, PLIC, putamen)	Y/Y	63
	63.1		5.7					41.33
Mean	9.78		6.27					16.16
SD	83		264					63
Max	51		11					15
Min

Abbreviations: UEFM – Upper Limb Fugl-Meyer, N – No, Y – Yes, BG – Basal Ganglia, PLIC – Posterior Limb of Internal Capsule, CR – Corona Radiata, CN – Caudate Nucleus, ALIC – Anterior Limb of Internal Capsule, TH- Thalamus, F – Frontal, PF – Prefrontal. Stroke locations were determined through chart reviews and magnetic resonance imaging if available. MEPs were determined based on resting motor thresholds and active motor threshold described in the methods.

Within-session Reliability of TMS indices:

The control condition demonstrated there was excellent reliability of ipsilesional rMT and aMT (rMT ICC = .985, p <.001; aMT ICC = .996, p <.001) and MEP_maxima used for normalization of MEPs collected from the input/output curve measurements (ICC = .989, p < .001). AUC for the input/output curve also showed good reliability (ICC = .675, p = .09). IHI and size of iMEP showed excellent reliability (ICC = .982, p < .001 and ICC = .778, p = .023, respectively).

Between-session Reliability of TMS indices:

There was excellent between-session reliability as well for ipsilesional rMT, aMT, MEP_maxima and input/output curve AUC, (ICC = .842, p = .001; ICC = .968, p <.001; ICC = .931, p < .001; ICC = .909, p = .001, respectively). IHI also showed excellent between-session reliability (ICC = .909, p <.001). iMEPs however showed poor between-session reliability (ICC = −.922, p = .935).

CCFES vs. cNMES:

There was no change in MVC following either intervention (CCFES Pre: 0.18±.03 mV Post :0.18±03 mV, p = .84; cNMES Pre: 0.2±.03 mV Post: 0.2±.04 mV, p = .87). Two-way ANOVA revealed there was a significant time*intervention interaction for IHI upon iM1 (F(1,14) = 8.44, p = .012)). IHI was significantly reduced following CCFES compared to cNMES (−9.5±−3.7% vs. +4.5±5.2%) (t = −2.2, p = .049) (figure 2a–b). Further, ipsilesional output was significantly preserved following CCFES compared to cNMES. There was a significant time*intervention interaction for AUC (F(1,6) = 9.18, p = .023). AUC was significantly reduced after cNMES compared to CCFES (−15.3 ±6.67% vs. +4.2±5.62%) (t = 3.2, p = .018) (figure 3a). CCFES also resulted in larger increases in normalized MEP amplitudes collected at 110% aMT and 120% aMT compared to cNMES (t = 3.3, p = .016; t = 2.52, p = .045, respectively) (figure 3b). We were able to identify iMEPs in ~7±3 (out of 25±4 ) trials before and after each session (pre: 6.8±2.7 (SD); post: 7.1±3.3 (SD); p = .45). iMEPs latency were significantly later than contralateral MEPs evoked in the non-paretic EDC (5.75±6.3 ms (SD), p = .001). We observed a non-significant time*intervention interaction in iMEP amplitude (F(1,14) = 2.59, p = .13).

Figure 2.

a) Interhemispheric inhibition to ipsilesional M1 decreased following CCFES, but not after cNMES. b) One patient representation of pre and post ipsilateral silent period (pre = black; post = red). * p < .05.

Figure 3.

a) Ipsilesional output (recruitment curve area under the curve) decreased following cNMES when compared to CCFES. b) The percent change in recruitment curves differed between CCFES and cNMES at 110, 120 and at 150% aMT. c) c) Mean normalized ipsilesional recruitment curve for all participants’ pre and post cNMES and CCFES. Error bars are SE. * p < .05, ǂ p < .1.

Effect of Severity of Impairment:

We studied how the change in TMS metrics observed with each session varied as a function of baseline severity of impairment. The changes in TMS metrics following cNMES did not vary as a function of impairment; however, change in IHI and ipsilesional output following CCFES did correlate with baseline impairment. Participants with more severe (i.e., lower UEFM) motor impairment scores trended towards greater reduction in IHI (r = .503, p = .056) (figure 4a). In addition, participants with higher motor impairment scores showed the largest increases in ipsilesional output (AUC) (r = .78, p = .04) (figure 4b). There was no correlation between baseline impairment and change in ipsilateral MEP following CCFES, though participants with greater motor impairment tended to have higher ipsilateral MEP amplitudes at baseline than participants with less impairment (r = −.730, p = .002) (figure 4c–d).

Figure 4.

Neurophysiology as a function of motor impairment. a) Patients with more severe motor impairments demonstrated the greatest reduction of inhibition upon the affected hemisphere following CCFES. b) Patients with more mild motor impairments showed the greatest increase in ipsilesional output following CCFES. c) We found no intervention effect of iMEP and its relationship to motor severity; however, patients with greater motor impairment showed greater iMEP output at baseline. d) Representation of 2 patients (more impaired (UEFM: 21) vs. less impaired (UEFM: 63) and their respective impaired upper-limb iMEP.

Discussion

This study has provided preliminary evidence that bilateral CCFES and unilateral cNMES have different neurophysiologic mechanisms in chronic stroke participants. Unlike unilateral cNMES, IHI from the contralesional motor cortices decreased and ipsilesional output to the paretic limb was maintained following bilateral CCFES. Further, we found that the less impaired participants showed greater increase in ipsilesional output and participants with more severe impairments showed a trend towards having greater reduction in IHI. However, we found no evidence that 1-hr of bilateral CCFES or unilateral cNMES movements differed in ipsilateral output to the paretic hand. Overall, our findings suggest that bilateral CCFES can recruit cooperative potential between the contralesional and the ipsilesional motor cortices across different ranges of impairment.

Our most important finding is that bilateral CCFES reduces IHI to ipsilesional M1 across the majority of participants (mild to severely impaired) (figure 2a and 4a). This outcome would not be predicted by the interhemispheric competition model, which instead predicts that activity in the contralesional hemisphere will increase inhibition of ipsilesional M1. However, several groups have challenged the premise of the interhemispheric competition model.^4,11,39 According to a more recent hypothesis called the ‘bimodal balance-recovery model’ by Di Pino et al. 2015, the degree of IHI imposed by the contralesional hemisphere varies based on the extent of damage to corticospinal output to the impaired upper-limb. Our work only shows a trend for a relationship between change in IHI following CCFES and extent of baseline motor severity (more reduced IHI in those with more severe impairment) (figure 4a). Therefore, we are not able to confirm the bimodal hypothesis likely due to limited sub-sample size and insufficient power. Importantly, however, we have built upon the premise of the bimodal balance-recovery model where we have shown that the adaptive role of the contralesional hemisphere may be invoked using bilateral CCFES across the moderate-to-severely impaired as well as the mildly impaired participants. Future studies should incorporate multiple sessions of CCFES treatment to better understand the relationship between the individual neurophysiologic mechanisms and degree of recovery seen across a wide range of motor severity.

Another important finding is that ipsilesional output stayed relatively the same after CCFES but decreased after cNMES. Peripheral nerve stimulation (PNS) has been shown to promote recovery and either increase or decrease corticospinal excitability in stroke volunteers.⁴⁰ PNS has been applied via electrodes over the skin overlying peripheral nerves or via a mesh-glove for whole-hand electrical stimulation. These changes appear to be dependent on lesion location, stimulus intensity, frequency and duration.^40–43 In our study, we showed an interaction between our two conditions, even though stimulus intensity, frequency and duration were the same. One possible reason may be due to central fatigue among participants as a result of the 1-hour therapy. Kotan et al.⁴⁴ have observed a significant reduction in corticospinal excitability after muscle fatigue induced by combined neuromuscular electrical stimulation and voluntary contraction, but not with neuromuscular electrical stimulation alone. The authors suggest their findings may be due to an increase in excitability of inhibitory networks. Central fatigue has been recorded in many studies that show that maintaining submaximal muscle contraction for an extended period of time results in what is known as “post-exercise depression” of output devoted to fatigued muscles.^45,46 In fact, Knorr et al.⁸ has shown that neuromuscular induced central fatigue in the chronic stroke population is more prevalent in the paretic limb when compared to the non-paretic limb. These results, combined with Kotan et al, inform our findings because we observed reduction in ipsilesional output following cNMES and control conditions but no such decrease after CCFES. One possible reason may be that CCFES being a bilateral condition induced ‘disinhibition’, a mechanism characteristic of synchronous bilateral movements,^21–24 as evidenced by reduction in IHI after training. This disinhibition may have helped overcome fatigue-related diminution of ipsilesional output otherwise observed after other conditions, an interpretation that remains to be tested in future studies.

Still, it is intriguing that relative to cNMES, CCFES resulted in CCFES enhanced ipsilesional output to the impaired upper-limb when compared to cNMES. In fact, participants with mild motor impairment (UEFM >50) in our sample demonstrated the greatest gains in ipsilesional output following CCFES (figure 4b). These results suggest that bilateral CCFES may not align with the interhemispheric competition model and even the bimodal balance recovery model because movement of the non-paretic limb (in conjunction with the paretic) resulted in increases in ipsilesional output in mildly impaired patients. We are unable to corroborate evidence for the more severely impaired given their absence of paretic MEPs.

Our findings may have broad implications for bilateral and unilateral movements in general. One reason bilateral CCFES may not completely align with the interhemispheric competition model or the bimodal balance recovery model is that both theories are based on principles of unilateral movement. Bilateral movements, however, invoke neurophysiologic mechanisms that are unique compared to unilateral movements. Studies have shown that by symmetrically moving both limbs for a common purpose, both hemispheres become coupled.^47,48 As a result of coupling, Mudie and Matyas⁴⁷ explain that the contralesional hemisphere may offer a ‘template’ of motor network recruitment to the ipsilesional hemisphere, allowing the impaired upper-limb to learn from the non-impaired upper-limb. In fact, studies with fMRI show that bilateral movements elicit unique and greater activation of bilateral primary sensorimotor, premotor and supplementary motor cortices in comparison to unilateral movements, and that these distinctions amplify with therapy.⁴⁹ However, an important caveat, is that previous studies have demonstrated that stroke motor deficits are hemisphere specific and thus differ between non-dominant vs. dominant hemisphere lesion locations. For example, Schaefer et al,⁵⁰ have shown patients with left hemisphere lesions experience reduce modulation of motor acceleration amplitude, where participants with right hemisphere lesions experience greater final motor position errors. Therefore, the ‘template’ provided by the contralesional hemisphere is likely hemisphere specific as well.

Coupling between hemispheres could involve mutual disinhibition of both hemispheres that is characteristic of bilateral movements.^21–24 It is important to note, however, that coupling between hemispheres appears to be more apparent during synchronous bilateral movements than during asynchronous movements. For example, bilateral synchronous movements result in a decrease in intracortical inhibition within M1,^15,16,51, enhanced activation and connectivity between hemispheres,²² and as demonstrated in this study, a decrease in IHI between M1s. A similar phenomenon has not been shown to occur during bilateral asynchronous movements.⁵¹Release of inhibition during synchronous movements could have contributed to the enhanced output of the ipsilesional M1 for the mildly impaired. Further, the release of inhibition could have enhanced spared pathways, such as the higher motor areas, for patients with greater corticomotor damage (i.e. no MEP present) not investigated in this study.^{12,15,48,49,52,53} Thus, it is possible that synchronous somatosensory feedback in bilateral motion, and a single set of motor commands linking bilateral movements may help upper limbs to become functionally coupled, and both hemispheres to release their inhibition upon one another to allow transfer from the contralesional to the ipsilesional hemisphere regardless of motor severity. In addition to the advantage of bilateral movements for the ipsilesional hemisphere, bilateral movements could engage the adaptive role of the contralesional hemisphere, especially for patients with greater corticomotor damage. Whitall et al. 2011 showed that eighteen 1-hour sessions of bilateral therapy results in greater contralesional hemisphere activation and the enhancement of the contralesional hemisphere was associated with improved arm function. In the present study, however, we were not able to find greater ipsilateral output to the paretic hand after CCFES when compared to cNMES (p = .13) as originally hypothesized; this may be the result of our small sample size and large distribution of motor impairment. Still, we do show that the ipsilateral output to the paretic-upper limb is primed in patients with moderate-to-severe motor impairment; results that are in-line with Schwerin et al. 2008 (figure 4c). Future studies will benefit from a more systematic investigation of the ipsilateral motor pathways originating from the contralesional hemisphere in patients with moderate-to-severe motor impairment.

Limitations and Conclusion:

First, we were unable to quantify the volitional finger extension effort of the paretic-limb during CCFES and cNMES. Ideally, we would have been able to record EMG; however, the electrical stimulus artifact makes this a challenging outcome to record. Second, while a crossover design was a strength of the study allowing for robust within-subject comparisons, applying only 1-hour of therapy would not likely result in any functionally relevant or impairment based outcomes and is more likely to induce fatigue. Without functional and impairment related outcomes and an assessment of the longevity of the effects on the TMS outcomes, we can only speculate that the changes measured in this study may help explain improvements in outcomes like hand dexterity following CCFES, as shown by Knutson et al. 2016. Future studies should incorporate multiple sessions of treatment to better understand the relationship between the individual neurophysiologic mechanisms and patient recovery.

In conclusion we show that bilateral CCFES may help recruit the recovery potential of the contralesional motor cortices by reducing competitive inhibition. Further, bilateral CCFES may support residual ipsilesional corticospinal output in those with less extensive damage and may not align with the interhemispheric competition model and the bimodal balance recovery model. Future studies will expand upon these findings by including multiple sessions of CCFES in order to make direct comparisons between the neurophysiologic benefit and the capacity to maximize recovery.