Altered corticospinal function during movement preparation in humans with spinal cord injury

Abstract

Key points

• In uninjured humans, transmission in the corticospinal pathway changes in a task‐dependent manner during movement preparation. We investigated whether this ability is preserved in humans with incomplete chronic cervical spinal cord injury (SCI).

• Our results show that corticospinal excitability is altered in the preparatory phase of an upcoming movement when there is a need to suppress but not to execute rapid index finger voluntary contractions in individuals with SCI compared with controls. This is probably related to impaired transmission at a cortical and spinal level after SCI.

• Overall our findings indicate that deficits in corticospinal transmission in humans with chronic incomplete SCI are also present in the preparatory phase of upcoming movements.

Abstract

Corticospinal output is modulated in a task‐dependent manner during the preparatory phase of upcoming movements in humans. Whether this ability is preserved after spinal cord injury (SCI) is unknown. In this study, we examined motor evoked potentials elicited by cortical (MEPs) and subcortical (CMEPs) stimulation of corticospinal axons and short‐interval intracortical inhibition in the first dorsal interosseous muscle in the preparatory phase of a reaction time task where individuals with chronic incomplete cervical SCI and age‐matched controls needed to suppress (NOGO) or initiate (GO) ballistic index finger isometric voluntary contractions. Reaction times were prolonged in SCI participants compared with control subjects and stimulation was provided ∼90 ms prior to movement onset in each group. During NOGO trials, both MEPs and CMEPs remained unchanged compared to baseline in SCI participants but were suppressed in control subjects. Notably, during GO trials, MEPs increased to a similar extent in both groups but CMEPs increased only in controls. The magnitude of short‐interval intracortical inhibition increased in controls but not in SCI subjects during NOGO trials and decreased in both groups in GO trials. These novel observations reveal that humans with incomplete cervical SCI have an altered ability to modulate corticospinal excitability during movement preparation when there is a need to suppress but not to execute upcoming rapid finger movements, which is probably related to impaired transmission at a cortical and spinal level. Thus, deficits in corticospinal transmission after human SCI extend to the preparatory phase of upcoming movements.

Key points

• In uninjured humans, transmission in the corticospinal pathway changes in a task‐dependent manner during movement preparation. We investigated whether this ability is preserved in humans with incomplete chronic cervical spinal cord injury (SCI).

• Our results show that corticospinal excitability is altered in the preparatory phase of an upcoming movement when there is a need to suppress but not to execute rapid index finger voluntary contractions in individuals with SCI compared with controls. This is probably related to impaired transmission at a cortical and spinal level after SCI.

• Overall our findings indicate that deficits in corticospinal transmission in humans with chronic incomplete SCI are also present in the preparatory phase of upcoming movements.

Abbreviations

AMT

active motor threshold

CMEP

cervicomedullary motor evoked potential

CS

conditioning stimulus

FDI

first dorsal interosseous

GO

execution of ballistic index finger isometric voluntary contractions

MEP

motor evoked potential

MEP‐max

maximal MEP size

M‐max

a maximal muscle response evoked by peripheral nerve electrical stimulation

MVC

maximal voluntary contraction

NOGO

suppression of ballistic index finger isometric voluntary contractions

RT

reaction time

SCI

spinal cord injury

SICI

short‐interval intracortical inhibition

TMS

transcranial magnetic stimulation

TS

test stimulus

Introduction

Transmission in the corticospinal pathway is altered in humans with incomplete spinal cord injury (SCI) (for reviews see Ellaway et al. 2007; Oudega & Perez, 2012). Studies showed that during small levels of voluntary activity corticospinal responses elicited by transcranial magnetic stimulation have delayed latencies, higher thresholds and decreased amplitudes in people with SCI compared with control subjects (Davey et al. 1998; Perez, 2012). SCI participants also show a decreased ability to modulate corticospinal excitability during a motor behaviour in a task‐dependent manner compared with controls (Barry et al. 2013; Bunday et al. 2014). Notably, corticospinal output does not only change during the execution of movements but also undergoes distinct modulation in the preparatory phase of upcoming movements (Chen et al. 1998; Leocani et al. 2000). At present, it is not known whether this ability is preserved in humans with SCI.

In uninjured humans, the contribution of the corticospinal pathway to movement preparation has been extensively studied using a GO/NOGO protocol (Hoshiyama et al. 1996; Leocani et al. 2000; Yamanaka et al. 2002; Fujiyama et al. 2011, 2012; Picazio et al. 2014). Here, individuals are instructed to rapidly initiate a movement after the appearance of a ‘go signal’ (GO trials) or suppress a movement after the appearance of a ‘no go signal’ (NOGO trials). Most studies agree that corticospinal excitability measured about 90 ms prior to movement onset increases in GO trials and decreases in NOGO trials, which is thought to be related to mechanisms at a cortical level (Waldwogel et al. 2000; Sohn et al. 2002; Fujiyama et al. 2012) although a spinal contribution has also been proposed (Van Boxtel et al. 1996). The goal of our study was to examine changes in corticospinal excitability during a GO/NOGO protocol in humans with chronic incomplete cervical SCI.

Neural processes during movement preparation influence aspects of movement execution (for review see Cisek & Kalaska, 2010). Evidence showed that reaction times in response to a signal to move are prolonged in people with SCI compared with controls (Labruyére et al. 2011, 2013) and prolonged reaction times in humans with SCI have been related to an altered ability to synchronize corticospinal descending volleys at the spinal cord (Cirillo et al. 2016). Transmission in the corticospinal (Chen et al. 1998; Leocani et al. 2000) and spinal cord (Requin, 1969; Eichenberger & Ruegg, 1984; Komiyama & Tanaka, 1990; Bonnet et al. 1991) pathways is modulated in a task‐dependent manner at early stages of movement preparation in control subjects. Notably, activity in some of the spinal pathways involved in movement preparation in controls (Komiyama & Tanaka, 1990; Bonnet et al. 1991) is altered when tested during a voluntary motor behaviour in humans with SCI (Yang et al. 1991; Woolacott & Burne, 2006). Therefore, we used information gathered during voluntary motor output in SCI subjects to make inferences about changes in corticospinal transmission during movement preparation. We hypothesized that humans with incomplete SCI will have a decreased ability to modulate corticospinal excitability in the preparatory phase of a GO/NOGO task, which will be related to impairments at a spinal level.

Methods

Subjects

Eighteen individuals with SCI (mean age ± SD = 52.2 ± 13.7 years, 2 female) and 13 age‐matched controls (mean age = 55.0 ± 14.9 years, 3 female, P = 0.89) participated in the study. All subjects gave informed consent to experimental procedures, which were approved by the local ethics committee at the University of Miami. The experiments were carried out following guidelines laid down in the Declaration of Helsinki. SCI subjects had a chronic (≥1 year) cervical injury (C2–C7), an intact or impaired, but not absent, innervation in dermatome C6 during light touch and pin prick stimulus using the International Standards for Neurological Classification of Spinal Cord Injury sensory scores and residual hand motor function (Table 1). Nine SCI participants took baclofen as part of their daily drug therapy for 5.4 ± 2.7 years and nine had never taken baclofen since their diagnosis. Since we found no differences in the electrophysiological outcomes tested in the preparatory phase of movement between subjects taking and those not taking baclofen, SCI participants were grouped together. All SCI subjects were able to exert isometric voluntary contractions by moving their index finger into abduction against resistance. At the start of the experiment, subjects were instructed to perform three brief maximal voluntary contractions (MVCs, 3–5 s) with the index finger into abduction, separated by ∼60 s of rest. During index finger abduction, subjects were asked to press with the index finger against a custom lever into the abduction direction with the forearm pronated and the wrist restrained by straps. The maximum electromyographic (EMG) activity in any of the three trials by each subject was used for analysis. EMG activity during index finger abduction MVC was greater in control (0.68 ± 0.14 mV) than in SCI (0.37 ± 0.25 mV, P < 0.001) subjects.

Table 1.

Spinal cord injury participants

SCI subjects	Age (years)	Sex	Level	ASIA score	Aetiology	Time since injury (years)	FDI MVC (N)	Maximum FDI MVC (mV)	Reaction time (ms)
1	66	M	C4	D	T	8	24.15	0.59	291.9
2	53	M	C5	C	T	14	36.71	0.45	295.5
3	62	F	C5	C	T	16	12.06	0.23	320.9
4	72	M	C4	D	T	4	18.95	0.17	262.9
5	59	M	C5	D	T	2	5.97	0.51	312.2
6	54	M	C3	C	T	2	30.59	0.38	250.5
7	44	M	C5	D	T	12	9.92	0.09	302.4
8	62	M	C4	C	T	2	29.72	0.31	258.1
9	45	F	C7	A	T	13	5.04	0.17	348.4
10	40	M	C5	D	T	3	20.40	0.25	364.5
11	68	M	C3	D	T	3	16.86	0.37	257.8
12	49	M	C5	C	T	28	30.07	0.98	258.1
13	50	M	C3	C	T	10	27.06	0.88	285.6
14	42	F	C3	C	T	8	16.14	0.20	271.5
15	19	M	C2	C	T	3	16.21	0.14	231.5
16	67	M	C5	D	T	6	13.50	0.43	312.2
17	55	F	C4	D	T	5	19.74	0.24	294.7
18	32	M	C3	D	T	9	22.90	0.20	285.2

SCI = spinal cord injury, ASIA = American Spinal Cord Injury association impairment scale, FDI = first dorsal interosseous, MVC = maximal voluntary contraction, M = male, F = female, C = cervical, T = traumatic.

EMG recordings

EMG was recorded from the dominant first dorsal interosseous (FDI) muscle in control subjects and from the less affected hand in individuals with SCI through surface electrodes secured to the skin over the belly of each muscle (Ag–AgCl, 10 mm diameter). The signals were amplified, filtered (20–1000 Hz) and sampled at 2 kHz for off‐line analysis (CED 1401 with Signal software, Cambridge Electronic Design, Cambridge, UK). During MVC, force exerted at the proximal interphalangeal joint of the index finger and thumb was measured by load cells (Honeywell Ltd, Morris Plains, NJ, US, range, ±498.1 N, voltage, ±5 V, high‐sensitivity transducer 0.045 V N⁻¹). Force was sampled at 200 Hz and stored on a computer for off‐line analysis.

Experimental setup

Subjects were seated in a custom chair with both arms flexed at the elbow by 90 deg. The tested hand was placed on a custom platform with the index finger positioned next to a lever connected to a force transducer with the forearm pronated and the wrist restrained by straps. Subjects completed a GO/NOGO task during a small level of voluntary activity (∼5% of MVC into index finger abduction: GO trials: Controls = 4.0 ± 3.0%, SCI = 5.0 ± 3.9%; NOGO trials: Controls = 3.9 ± 2.9%, SCI = 5.1 ± 4.0%; Baseline trials: Controls = 4.2 ± 3.2%, SCI = 5.0 ± 4.2%; F _(1,66) = 1.3, P = 0.3). Each trial started with the presentation of a fixation cross that remained on the centre of a computer screen for a variable delay (0.5–1.5 s). Then, an imperative stimulus consisting of a green (GO trial) or a red (NOGO trial) square appeared for 0.2 s (Fig. 1). During GO trials, participants were required to perform an index finger abduction as fast as possible, activating the FDI muscle above the 5% MVC background level of activity. During NOGO trials subjects were instructed to maintain the 5% MVC background level of activity at all times. The inter‐trial interval was between 3.5 and 4.5 s. EMG activity from the FDI muscle was displayed continuously on an oscilloscope and verbal feedback was provided to subjects to ensure that physiological measurements were acquired at similar levels of EMG activity at all times. A familiarization block consisting of 20 GO and 20 NOGO trials was conducted at the beginning of testing to determine reaction time (RT = defined as the time point in which EMG activity exceeded ±4 SD of the average contracting mean rectified EMG, measured 100 ms before the stimulus artifact; Controls = 267.7 ± 24.8 ms, SCI = 294.0 ± 42.0 ms, P = 0.002; Fig. 2). Stimulation during GO and NOGO trials was provided ∼90 ms before the RT recorded during trials without stimulation (Controls = 84.2 ± 19.5 ms, SCI = 84.1 ± 33.7 ms, P = 1.0). We selected this interval because previous studies showed that Motor Evoked Potentials (MEPs) are modulated during both GO and NOGO trials within this time interval prior to movement execution (Hoshiyama et al. 1996; Yamanaka et al. 2002; Fujiyama et al. 2012). Stimulation was also provided ∼90 ms after the fixation cross (referred to as ‘baseline trials’). After familiarization, subjects completed four blocks of 30 trials. In each block, 10 GO, 10 NOGO and 10 baseline trials were tested in a randomized order. The probability of GO and NOGO signals was kept constant at 50% (Eimer, 1993; Nieuwenhus et al. 2003). Of the total of all trials in which mean rectified EMG activity exceeded ±2 SD of the average mean rectified EMG measured 100 ms before the stimulus artifact, 7.6 ± 6.6% were excluded from analysis (Cirillo et al. 2016). Trials with late (>700 ms), no response in GO and false responses in NOGO were also excluded from analysis (2.2 ± 2.1%; Leocani et al. 2000).

Figure 1. Schematic representation of the experimental setup during the GO/NOGO task.

A fixation cross (white cross) appeared in the centre of the computer screen at the beginning of all trials, followed by an imperative visual signal instructing subjects to execute (GO = green square) or suppress (NOGO = red square) ballistic index finger abduction isometric voluntary contractions. Raw traces show the time at which motor evoked potentials (MEPs) and other physiological measurements were tested during GO (green), NOGO (red) and baseline (black) trials. Baseline MEPs were recorded ∼90 ms after the fixation cross and GO and NOGO MEPs were recorded ∼90 ms before the reaction time (RT) recorded during trials without stimulation.

Figure 2. Reaction times.

A, traces showing rectified raw electromyographic activity in the first dorsal interosseous (FDI) muscle in a representative control and spinal cord injury (SCI) subject during GO trials without transcranial magnetic stimulation. Each waveform represents the average of 40 trials in a control (black traces) and an SCI (grey traces) participant. B and C, group data (Controls, n = 13; SCI, n = 18) showing reaction times in milliseconds (ms) during GO trials without TMS (B) and the time at which TMS was applied during GO trials with TMS (C) in both groups. Error bars indicate SEM. ^* P < 0.05.

Transcranial magnetic stimulation (TMS)

Transcranial magnetic stimuli were delivered from a Magstim 200 stimulator (Magstim Company, Whitland, UK) through a figure‐of‐eight coil (loop diameter, 7 cm; type number SP15560) with a monophasic current waveform. We determined the optimal position for eliciting an MEP in the FDI muscle (hot spot) by moving the coil, with the handle pointing backward and 45 deg away from the midline, in small steps along the hand representation of the motor cortex. The hot spot was defined as the region where the largest MEP in the FDI could be evoked with the minimum intensity (Rothwell et al. 1999). With this coil position the current flowed in a posterior–anterior direction and probably produced D and early I wave activation (Sakai et al. 1997). The TMS coil was held to the head of the subject with a custom coil holder, while the head was firmly secured to a headrest by straps. TMS measurements included active motor threshold (AMT), MEPs, maximal MEP size (MEP‐max), and short‐interval intracortical inhibition (SICI). Based on our previous results, we determined that values 10% above or below the mean result found in baseline trials were considered to display an effect (Bunday & Perez, 2012; Bunday et al. 2014; Cirillo et al. 2016). Note that individual data and group data for each main physiological outcome tested in the study is also presented in the figures.

MEPs

AMT was defined as the minimal stimulus intensity required to induce MEPs ≥ 200 μV peak‐to‐peak amplitude above the background EMG in 5/10 consecutive trials in the contracting muscle (Rothwell et al. 1999). MEPs were tested during ∼5% of MVC at intensities of 120% of AMT (Controls = 122.1 ± 6.7%, SCI = 122.7 ± 15.0%, P = 0.9). The MEP‐max was defined at rest by increasing stimulus intensities in 5% steps of maximal device output until the MEP amplitude did not show additional increases (Controls = 5.0 ± 1.9 mV, SCI = 2.0 ± 1.4 mV, P < 0.001). Forty MEPs were averaged in GO, NOGO, and baseline trials in each subject.

Short‐interval intracortical inhibition (SICI)

SICI was tested using a previously described method (Kujirai et al. 1993) during ∼5% of index finger abduction in controls (n = 12) and SCI (n = 12) subjects. A conditioning stimulus (CS) was set at an intensity needed to elicit ∼50% of SICI, which corresponded to ∼70% of AMT (Controls = 71.2 ± 7.4, SCI = 71.0 ± 5.9%, P = 0.9). This low‐intensity stimulus allowed us to assess SICI independently of the effects on short‐intracortical facilitation at low contraction levels (Ortu et al. 2008). The same stimulation intensity was used for the CS across conditions. The test stimulus (TS) was set at an intensity needed to elicit an MEP ∼50% of the MEP‐max, which corresponded to ∼120% of AMT (Controls = 121.0 ± 15.2% of AMT, SCI = 115.1 ± 12.3%, P = 0.4). The CS was delivered 2 ms before the TS. Previous evidence showed that the size of a test MEP can affect the magnitude of SICI (Illic et al. 2002). Because MEP size increased during GO trials and decreased during NOGO trials compared with baseline we also tested SICI by adjusting the test MEP size to match MEP amplitudes produced at baseline. This was accomplished by measuring SICI again and now decreasing (in GO trials) or increasing (in NOGO trials) the intensity of the TS as needed in order to elicit a test MEP as close as possible to the baseline MEP. SICI was calculated by expressing the size of the conditioned MEP (CS at ∼70% of AMT + TS at ∼120% of AMT) as a percentage of the test MEP size (TS at ∼120% of AMT) using the following calculation: (conditioned MEP × 100)/(test MEP). Twenty test MEPs and 20 conditioned MEPs were tested in each condition.

Cervicomedullary MEPs (CMEPs)

The corticospinal tract was stimulated at the cervicomedullary level (Controls, n = 7; SCI, n = 6) by a high voltage electrical current (100 μs duration, Digitimer, DS7AH, Welwyn Garden City, UK DS7AH) passed between adhesive Ag–AgCl electrodes fixed to the skin behind the mastoid process (Taylor & Gandevia, 2004). We used the same test timing as in the TMS experiment. The stimulation intensity was set to elicit a CMEP in the FDI muscle of ∼10% of the maximal motor response (M‐max, a maximal muscle response evoked by direct activation of the efferent fibres by supramaximal electrical stimulation of the ulnar nerve at the wrist) during ∼5% of MVC in baseline trials (CMEP as % of M‐max: Controls = 8.2 ± 2.2%, SCI = 8.5 ± 3.1%, P = 0.9). Fifteen to twenty CMEPs were tested for each condition.

Data analysis

Normal distribution was tested by the Shapiro–Wilk's test and Mauchly's test was used to test sphericity. When normal distribution could not be assumed data were log transformed. When sphericity could not be assumed the Greenhouse–Geisser correction statistic was used. Repeated‐measures ANOVAs were performed to determine the effect of TRIAL (GO, NOGO, baseline) and GROUP (Controls, SCI) on MEPs, SICI and mean EMG activity. A repeated‐measures ANOVA was performed to determine the effect of STIMULATION (magnetic, electrical) and GROUP on MEP latency. Non‐parametric repeated‐measures ANOVAs were performed to determine the effect of TRIAL and GROUP on CMEPs. Bonferroni post hoc analysis was used to test for pairwise comparisons. Significance was set at P < 0.05. Group data are presented as mean ± SD in the text.

Results

MEPs

Figure 3 A and B illustrates MEPs in the FDI muscle from representative participants. Note that during NOGO trials, MEP size decreased compared with baseline in the control subject but remained unchanged in the SCI participant. In contrast, during GO trials, MEPs increased compared to baseline in both participants.

Figure 3. Motor evoked potentials (MEPs).

A and B, raw traces of MEPs elicited by transcranial magnetic stimulation in the first dorsal interosseous (FDI) muscle in a representative control and spinal cord injury (SCI) subject during NOGO (A) and GO (B) trials. Each waveform represents the average of 20 MEPs at baseline and during GO (Control = black; SCI = grey) and NOGO (Control = black; SCI = grey) trials. Amplitude scales are different for the SCI and control representative participants to better show changes across task. C and D, group data (Controls, n = 13, black circles; SCI, n = 18, grey squares) show MEPs during NOGO (C) and GO (D). The horizontal dotted line shows the FDI MEP size at baseline. Data from individual subjects are shown in controls (open circles) and SCI subjects (open squares). Error bars indicate SEM. ^* P < 0.05, comparison between groups; ^¥ P < 0.05, comparison between baseline and GO or NOGO trials.

Repeated‐measures ANOVA showed a significant effect of TRIAL (F _(1.6,47.3) = 51.1, P < 0.001), GROUP (F _(1,29) = 4.6, P = 0.04) and in their interaction (F _(1.6,47.3) = 6.4, P = 0.003) on MEP size. Post hoc testing showed that in NOGO trials MEP size decreased compared with baseline in controls (66.5 ± 17.0%, P < 0.001) but not in individuals with SCI (93.7 ± 22.6%, P = 0.6; Fig. 3 C). Note that MEPs during NOGO trials were suppressed in the majority of control subjects (11/13) but not in SCI subjects (7/18). We also found that MEP size increased during GO trials compared with baseline in controls (127.4 ± 24.7%, P < 0.005) and SCI (125.2 ± 22.9%, P < 0.001; Fig. 3 D) subjects. The increase in MEP size was similar in both groups (P = 0.8). Also, note that the majority of control (11/13) and SCI (15/18) participants showed increases in MEP size during GO trials compared with baseline. Mean background rectified EMG activity in the FDI was similar across conditions (F _(2,58) = 2.1, P = 0.1) and groups (F _(1,29) = 2.9, P = 0.1).

SICI

Figure 4 A and B illustrates raw data from SICI measurements in representative subjects. The magnitude of SICI in NOGO trials increased in the control subject compared with baseline but remained unchanged in the subject with SCI, whereas during GO trials, SICI was decreased compared with baseline in both participants.

Figure 4. Short‐interval intracortical inhibition (SICI).

A and B, SICI tested in the FDI muscle during NOGO (A) and GO (B) trials. Each waveform represents the average of 20 rectified test (black) and conditioned (grey) MEPs. Arrows indicate the test and conditioned MEP. Amplitude scales are different for the SCI and control representative participants to better show changes across task. C and D, group data (Controls, n = 12, red circle NOGO and green circle GO; SCI, n = 12, pink square NOGO and light green square GO) show SICI during NOGO (C) and GO (D) trials. The horizontal dotted line shows SICI at baseline. Data from individual subjects are shown in controls (open red circles, NOGO; open green circles, GO) and SCI subjects (open pink squares, NOGO; open light green squares, GO). Error bars indicate SEM. ^* P < 0.05, comparison between groups; ^¥ P < 0.05, comparison between baseline and GO or NOGO trials.

Repeated‐measures ANOVA showed a significant effect of TRIAL (F _(2,44) = 18.2, P < 0.001), not GROUP (F _(1,22) = 3.6, P = 0.07) but in their interaction (F _(2,44) = 5.9, P = 0.005) on SICI. Our results indicate that during NOGO trials, SICI increased compared with baseline in controls (83.4 ± 15.6%, P = 0.02) but not in SCI participants (109.9 ± 28.5%, P = 0.97; Fig. 4 C). The increase in SICI was present in the majority of control subjects (8/12) but not in SCI subjects (4/12). In GO trials, SICI decreased compared with baseline in SCI (125.3 ± 32.6%, P = 0.01) and controls (117.9 ± 14.6%, P = 0.04; Fig. 4 D) subjects. Note that here the decrease in SICI was similar in both groups (P = 0.7). Also note that SICI decreased in the majority of SCI (8/12) and control (8/12) participants. Similar results were obtained when we adjusted the size of the test MEP to match baseline MEP amplitude (TRIAL, F _(2,44) = 16.7, P = 0.001; GROUP, F _(1,22) = 2.9, P = 0.1; interaction, F _(2,44) = 5.2, P = 0.009). Here, we found that SICI increased compared to baseline in NOGO trials in control but not in SCI subjects (Controls = 84.9 ± 18.0%, P = 0.04; SCI = 112.2 ± 21.8%, P = 0.1; Fig. 4 C), whereas SICI decreased in both groups to a similar extent compared to baseline in GO trials (Controls = 123.4 ± 23.6%, P = 0.02; SCI = 123.7 ± 28.4%, P = 0.02; Fig. 4 D). Mean background rectified EMG activity in the FDI remained similar across conditions (F _(2,96) = 2.7, P = 0.1) and groups (F _(1,46) = 0.8, P = 0.4).

CMEPs

Figure 5 A and B illustrates subcortically evoked CMEPs in representative participants. See here that CMEP size decreased in NOGO trials and increased in GO trials compared with the baseline in the control subject but remained unchanged in both tasks in the SCI participant. Importantly, we found that CMEPs had shorter latencies than MEPs (Controls = 18.9 ± 1.3 ms, 22.3 ± 2.0 ms, respectively; SCI = 21.8 ± 3.3 ms, 25.7 ± 2.0 ms, respectively; F _(1,11) = 61.5, P < 0.001), indicating that the stimulation activated corticospinal axons bypassing the motor cortex.

Figure 5. Cervicomedullary MEPs (CMEPs).

A and B, CMEP elicited in FDI muscle by electrical stimulation at the cervicomedullary junction in a representative control and SCI subject during NOGO (A) and GO (B) trials. Each waveform represents the average of 10 rectified CMEPs at baseline (black), during NOGO (Control = red; SCI = light red), and GO (Control = green; SCI = light green) trials. Amplitude scales are different for the SCI and control representative participants to better show changes across task. C and D, group data (Controls, n = 7; SCI, n = 6) showing CMEPs expressed as % of M‐max in control (C) and SCI (D) participants during GO (Controls: green circle; SCI: light green square), NOGO (Controls: red circle, SCI: pink square) and baseline trials (Controls: black circle; SCI: black square). Data from individual subjects are shown in controls (circles) and SCI subjects (squares). Error bars indicate SEM. ^* P < 0.05, comparison between groups; ^¥ P < 0.05, comparison between baseline and GO or NOGO trials.

Repeated‐measures ANOVA showed a significant effect of TRIAL (F _(2,22) = 12.2, P < 0.001), not GROUP (F _(1,11) = 0.5, P = 0.5) but in their interaction (F _(2.22) = 5.0, P = 0.02) on CMEP size. We found in NOGO trials that the size of CMEPs decreased compared with baseline in controls (85.4 ± 18.0%, P = 0.04; Fig. 5 C) but not in SCI subjects (105.2 ± 30.4, P = 0.6; Fig. 5 D). The decrease in CMEP size was present in 5/7 control subjects and in 2/6 SCI participants. During GO trials, CMEPs increased compared with baseline trials in controls (127.4 ± 16.6%, P = 0.03; Fig. 5 C) but remained unchanged in SCI participants (105.5 ± 36.8%, P = 1; Fig. 5 D). The increase in CMEP size was present in 6/7 control and in 2/6 SCI subjects. Mean background rectified EMG activity in the FDI remained similar across conditions (F _(2,22) = 0.8, P = 0.4) and groups (F _(1,11) = 1.5, P = 0.2).

Discussion

The novel finding in our study is that humans with incomplete chronic cervical SCI have an altered ability to modulate corticospinal excitability during movement preparation when there is a need to suppress but not to execute upcoming rapid finger movements. This is probably related to impaired transmission at a cortical and spinal level. Specifically, we demonstrate that when the instruction was to suppress an upcoming movement (NOGO trials), the size of MEPs and CMEPs remained unchanged compared with baseline in individuals with SCI but was suppressed in control subjects. In contrast, when the instruction was to execute an upcoming movement (GO trials), the size of MEPs increased compared to baseline to a similar extent in SCI and control subjects but CMEPs increased only in controls. Short‐interval intracortical inhibition increased in controls but not in SCI subjects during NOGO trials and decreased in both groups during GO trials. Thus, deficits in corticospinal transmission in humans with SCI precede movement onset.

Does corticospinal output change during movement preparation after SCI?

As in previous studies, we found in control subjects that MEP amplitudes decreased during movement preparation in NOGO trials (Hoshiyama et al. 1996; Leocani et al. 2000; Waldwogel et al. 2000; Yamanaka et al. 2002; Nakata et al. 2006; Fujiyama et al. 2011, 2012). Since cortical and subcortical influences can both contribute to changes in MEP size (Burke & Pierrot‐Deseilligny, 2010), it is possible that both sources influenced our results. Responses to cervicomedullary stimulation of corticospinal axons (CMEPs) can help to elucidate cortical versus spinal contributions. Evidence showed that cervicomedullary stimulation activates axons of pyramidal tract neurons in the subcortical white matter and that CMEPs probably reflect changes in the efficacy of cortico‐motoneuronal synapses or motoneuron excitability (Ugawa et al. 1991; Gandevia et al. 1999; Taylor & Gandevia, 2004). Therefore, the suppression of CMEPs during NOGO compared to baseline trials in control subjects suggests a spinal origin for this effect. One might expect that if mechanisms at the spinal level contribute to NOGO processes, spinal damage will affect the modulation of ongoing physiological responses when individuals perform the task. Indeed, this is consistent with the lack of changes in CMEP size in individuals with SCI during NOGO trials. This agrees with findings showing that humans with SCI showed an impaired ability to modulate transmission in subcortically mediated pathways during ongoing motor behaviours (Barry et al. 2013; Bunday et al. 2014). Also, previous evidence showed that spinal interneurons of non‐human primates change their activity level at early stages of movement preparation when animals needed to withhold the execution of an upcoming movement (Prut & Fetz, 1999).

It is also possible that cortical mechanisms contributed to our results. We found in control subjects that SICI increased during NOGO trials compared to baseline, which is consistent with previous findings (Sohn et al. 2002; Fujiyama et al. 2012). In humans, intracortical inhibition is commonly assessed by a paired‐pulse TMS protocol referred as SICI (Kujirai et al. 1993). Here, a subthreshold TMS pulse probably activates intracortical inhibitory circuits and reduces the size of an MEP which is elicited 2–5 ms later. Studies using pharmacology and epidural recordings strongly indicate that SICI is due to activation of γ‐aminobutyric acid (GABA) inhibitory circuits (GABA_A) in the motor cortex (Hanajima et al. 1998; Di Lazzaro et al. 2000; Ziemann et al. 2001). Therefore, the lack of increments in SICI in SCI participants compared with controls during NOGO trials suggests that the motor cortex is another possible site that could have contributed to our results. This agrees with studies showing that SICI at rest is reduced in humans with SCI compared with controls (Shimizu et al. 2000; Roy et al. 2011). However, this result needs to be interpreted with caution since MEPs used for SICI quantification are also based on the summation of descending volleys at spinal motoneurons. Evidence showed that paired‐pulse TMS protocols possibly reflecting the ability to summate descending volleys at the spinal level (Cirillo & Perez, 2015) are affected in humans with incomplete SCI (Cirillo et al. 2016).

An intriguing question is why corticospinal responses tested during GO trials were modulated to a similar extent in SCI and control subjects? As previously shown, we found that MEP amplitudes increased during GO trials in control subjects (Hoshiyama et al. 1996; Leocani et al. 2000; Yamanaka et al. 2002; Nakata et al. 2006; Fujiyama et al. 2011, 2012; Picazio et al. 2014). In this group, the increases in MEP size were accompanied by decreases in intracortical inhibition and increases in the size of CMEPs, suggesting that both cortical and spinal mechanisms might have contributed to these results. This is consistent with findings showing that intracortical inhibition decreased during GO trials in control subjects (Sohn et al. 2002; Fujiyama et al. 2012). Furthermore, previous evidence showed that the size of H‐reflexes (Requin, 1969; Eichenberger & Ruegg, 1984; Bonnet et al. 1991) and stretch reflexes (Komiyama & Tanaka, 1990) increases around ∼90 ms before movement initiation in response to an external signal to move in control subjects. We found that individuals with SCI retained their ability to increase corticospinal excitability during GO trials, regardless of the lack of modulation of CMEPs. This is interesting considering that the spinal cord is the final common pathway of the system to shape inputs to muscles. One possibility is that corticospinal transmission in GO trials in humans relies to a larger extent on the activity in cortical pathways. This is supported by the similar reduction in SICI in GO trials in SCI and control participants. This also agrees with previous evidence showing that brain regions such as the dorsal premotor cortex, which sends projections to the motor cortex (Tokuno & Nambu, 2000) and down to spinal motoneurons (Dum & Strick, 1991), show higher activation patterns in the preparatory phase of GO compared with NOGO trials (Kalaska & Crammond, 1995). Regardless of the mechanisms contributing to the modulation of SICI after SCI, our results indicate that cortical pathways represent a critical mechanism contributing to the modulation of corticospinal drive prior to an upcoming movement during GO trials in human subjects.

Functional considerations

Many of our daily motor behaviours encounter situations where we rapidly need to execute or suppress prepared movements. Thus, our results during a GO/NOGO protocol might be relevant for goal‐oriented motor behaviours that are part of our daily living. Our findings indicate that humans with chronic incomplete SCI show an altered ability to suppress corticospinal excitability when there is a need to rapidly stop upcoming finger movements, which is probably related to impaired transmission at a cortical and spinal level. What are the possible implications of these results in humans with SCI? On one side, inhibition during movement preparation may contribute to increase the signal‐to‐noise ratio needed for the proper selection of actions (Hasbroucq et al. 1997). On the other side, inhibitory mechanisms engaged during response selection, when there is a need to withhold an upcoming movement, have been associated with the need to prevent the occurrence of premature responses (Duque et al. 2005). Thus, in SCI subjects the lack of suppression of subcortical responses (CMEPs) during movement preparation in NOGO trials, when there is a need to withhold an upcoming movement, agrees with the view that spinal pathways contribute to involuntary motor responses after SCI (Gorassini et al. 2004; Zijdewind et al. 2014). It is important to consider that TMS applied over the left or right motor cortex can differentially affect RTs during a choice–RT task (Schluter et al. 1998). This might be relevant since we tested the less affected side in SCI participants, regardless of whether this was the left or the right side. However, RTs in a contralateral hand were more affected when TMS was applied ∼300 ms after an imperative stimulus and we delivered TMS ∼200 ms after a GO or NOGO task, a time window where RTs were less affected by the stimulation (Schluter et al. 1998). Note that the main effects reported in our study were present in the majority but not in all subjects. The mechanism for this phenomenon is not clear. However, this is consistent with previous results in other physiological studies using similar TMS outcomes (Barry et al. 2013; Bunday et al. 2014; Cirillo et al. 2016) and with the inter‐individual differences found in the response to TMS outcomes in human subjects (Wasserman, 2002; Yi et al. 2014). Previous evidence also showed that intake of baclofen affects corticospinal excitability measured during tonic voluntary activity in individuals with chronic cervical SCI (Barry et al. 2013; Bunday et al. 2014). Although we did not find differences in the electrophysiological outcomes measured during movement preparation between subjects taking and those not taking baclofen, we have to note that a few individuals in the group who never took baclofen showed some MEP suppression during NOGO trials. The effects of baclofen on voluntary motor output remain largely unknown and the precise duration and dose of baclofen use needed for these changes to occur remains to be tested.

Neural processes involved in the ability to rapidly execute or suppress upcoming movements can act at multiple sites within the CNS (for review see Cisek & Kalaska, 2010). It is well established in non‐human primates and humans that preparatory processes involved in executing or suppressing upcoming movements during GO/NOGO protocols involve cortical regions in the frontal and prefrontal cortex (Waldwogel et al. 2000; Sohn et al. 2002; Duann et al. 2009; Picazio et al. 2014). Although a spinal contribution has been proposed during GO/NOGO protocols (Van Boxtel et al. 1996), the majority of studies proposed that cortical regions achieve inhibitory and/or excitatory control during the preparatory phase by acting on subcortical structures such as basal ganglia and thalamo‐cortical outputs (for reviews see Cisek & Kalaska, 2010; Aron et al. 2014). Thus, our findings extend previous results by highlighting the contribution of spinal mechanisms when there is a need to suppress and initiate upcoming motor responses in intact humans.

Additional information

Competing interests

None declared.

Author contributions

P.F. and M.A.P. contributed to all aspects of the study. Both authors have approved the final version of the manuscript and agree to be accountable for all aspects of the work. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

This work was supported by the NIH National Institute of Neurological Disorders and Stroke (NINDS) (grant nos 1R01NS076589‐01 and 1R01NS090622‐01 to M.A.P.).