Effects of lorazepam and baclofen on short‐ and long‐latency afferent inhibition

Abstract

Key points

• Short‐latency afferent inhibition (SAI) is modulated by GABA_A receptor activity, whereas the pharmacological origin of long‐latency afferent inhibition remains unknown.

• This is the first study to report that long‐latency afferent inhibition (LAI) is reduced by the GABA_A positive allosteric modulator lorazepam, and that both SAI and LAI are not modulated by the GABA_B agonist baclofen.

• These findings advance our understanding of the neural mechanisms underlying afferent inhibition.

Abstract

The afferent volley evoked by peripheral nerve stimulation has an inhibitory influence on transcranial magnetic stimulation induced motor evoked potentials. This phenomenon, known as afferent inhibition, occurs in two phases: short‐latency afferent inhibition (SAI) and long‐latency afferent inhibition (LAI). SAI exerts its inhibitory influence via cholinergic and GABAergic activity. The neurotransmitter receptors that mediate LAI remain unclear. The present study aimed to determine whether LAI is contributed by GABA_A and/or GABA_B receptor activity. In a double‐blinded, placebo‐controlled study, 2.5 mg of lorazepam (GABA_A agonist), 20 mg of baclofen (GABA_B agonist) and placebo were administered to 14 males (mean age 22.7 ± 1.9 years) in three separate sessions. SAI and LAI, evoked by stimulation of the median nerve and recorded from the first dorsal interosseous muscle, were quantified before and at the peak plasma concentration following drug ingestion. Results indicate that lorazepam reduced LAI by ∼40% and, in support of previous work, reduced SAI by ∼19%. However, neither SAI, nor LAI were altered by baclofen. In a follow‐up double‐blinded, placebo‐controlled study, 10 returning participants received placebo or 40 mg of baclofen (double the dosage used in Experiment 1). The results obtained indicate that SAI and LAI were unchanged by baclofen. This is the first study to show that LAI is modulated by GABA_A receptor activity, similar to SAI, and that afferent inhibition does not appear to be a GABA_B mediated process.

Key points

• Short‐latency afferent inhibition (SAI) is modulated by GABA_A receptor activity, whereas the pharmacological origin of long‐latency afferent inhibition remains unknown.

• This is the first study to report that long‐latency afferent inhibition (LAI) is reduced by the GABA_A positive allosteric modulator lorazepam, and that both SAI and LAI are not modulated by the GABA_B agonist baclofen.

• These findings advance our understanding of the neural mechanisms underlying afferent inhibition.

Introduction

The afferent volley evoked by peripheral nerve stimulation is capable of modifying the neural output of the primary motor cortex (M1), as assessed using transcranial magnetic stimulation (TMS). This phenomenon, known as afferent inhibition, occurs at short (i.e. short‐latency afferent inhibition, SAI) and long latencies (i.e. long‐latency afferent inhibition, LAI). SAI occurs when the nerve stimulus precedes a TMS pulse by an interval approximately equal to the time required for the sensory afferent input to reach the somatosensory cortex (∼20–25 ms) (Tokimura et al. 2000), whereas LAI occurs between 200 and 1000 ms (Chen et al. 1999). SAI is reduced in populations with cognitive deficits such as Alzheimer's disease (Nardone et al. 2006, 2008) and mild cognitive impairment (Nardone et al. 2012a; Yarnall et al. 2013). Abnormal LAI is seen in individuals with sensorimotor deficits, including Parkinson's disease (Sailer et al. 2003) and complex regional pain syndrome (Morgante et al. 2017). Furthermore, the magnitude of SAI and LAI declines with age (Young‐Bernier et al. 2012, 2014, 2015; Bhandari et al. 2016), possibly reflecting an age‐related decline in sensorimotor function (He et al. 2017). Therefore, measures of afferent inhibition can be used to assess the integrity of the sensorimotor system and probe sensorimotor function. A detailed review of this topic is provided by Turco et al. (2018a).

Pharmacological studies have provided insight into the neural genesis of SAI (Ziemann et al. 2015). SAI is reduced following iv injection of scopolamine, a muscarinic antagonist (Di Lazzaro et al. 2000b), leading to the conclusion that SAI is mediated by cholinergic transmission. This is further supported by findings of reduced SAI in disorders of cognition, which have underlying cholinergic deficits (Nardone et al. 2006, 2008). SAI is also reduced by lorazepam, a positive allosteric modulator of the GABA_A receptor, indicating a role for GABAergic neurotransmission in the genesis of SAI (Di Lazzaro et al. 2005a,b, 2007a). By contrast to SAI, the neural mechanisms that mediate LAI are less understood. LAI was unaltered by GABA_A agonists lorazepam and zolpidem in a double‐blinded, non‐placebo‐controlled study (Teo et al. 2009). However, in the present study, LAI was tested with an interstimulus interval of 100 ms when inhibition is not always present (Chen et al. 1999).

The present study aimed to explore the role of GABAergic neurotransmission in the genesis of LAI. In a double‐blinded, placebo‐controlled study, we assessed LAI in response to lorazepam (positive allosteric modulator of the GABA_A receptor) and baclofen (GABA_B agonist) and investigated SAI for comparison with other reports. Our novel findings indicate that LAI was reduced in the presence of lorazepam, which enhanced GABA_A transmission, although not by the GABA_B agonist, baclofen. Consistent with previous findings, SAI was also reduced in the presence of lorazepam (Di Lazzaro et al. 2005a,b, 2007a).

Methods

Ethical approval

The present study was approved by the Hamilton Integrated Research Ethics Board (HIREB 2731). The research conformed to the standards set by the latest revision of the Declaration of Helsinki, except for registration in a database. After explanation of the study protocol, the usual action and potential side‐effects of lorazepam and baclofen, all participants provided their written informed consent prior to participation.

Participants

Fourteen healthy, right‐handed males (mean age 22.7 ± 1.9 years) participated in Experiment 1 and 10 of these individuals (mean age 23.1 ± 1.7 years) returned to participate in Experiment 2. All participants were screened for contraindications to TMS, lorazepam and baclofen. Right‐hand dominance was confirmed by a modified handedness questionnaire (Oldfield, 1971). All pharmaceuticals and the randomization schedules were prepared by the McMaster University Medical Centre pharmacy. All experiments and data analyses were performed by experimenters who were blinded to the drug administered (CVT, JE and MBL).

Electromyography (EMG)

EMG was recorded with surface electrodes (9 mm Ag‐AgCl) placed over the first dorsal interosseous (FDI) muscle of the right hand. A wet ground was secured around the forearm. All EMG recordings were amplified 1000× (Model 2024F; Intronix Technologies Corporation, Bolton, ON, Canada) and band‐pass filtered between 20 Hz and 2.5 kHz. Data were digitized at 5 kHz using an analog‐to‐digital interface (Power 1401; Cambridge Electronics Design, Cambridge, UK) and analysed using commercial software (Signal, version 6.02; Cambridge Electronics Design).

TMS

TMS was performed with a customized figure‐of‐eight branding iron style coil (diameter 50 mm) connected to a Magstim 200² stimulator (Magstim, Whitland, UK). The TMS coil was positioned over the left M1 at the optimal location to elicit a motor‐evoked potential (MEP) in the right FDI muscle (i.e. motor hotspot). The coil was oriented at a 45° angle from the sagittal plane to induce a posterior–anterior current in the cortex. The location and orientation of the coil was registered digitally using Brainsight Neuronavigation (Rogue Research, Montreal, QC, Canada). Resting motor threshold (RMT) was taken as a measure of baseline cortical excitability (Siebner & Rothwell, 2003) before and after drug administration. RMT was determined using ML‐PEST, a systematic predictive algorithm that determines the next TMS intensity predicted to yield a 50% probability of generating a MEP (TMS Motor Threshold Assessment Tool; MTAT, version 2.0 (http://www.clinicalresearcher.org/software.html). A priori information was selected, and the starting TMS intensity was set to 37%. Twenty stimuli were delivered to accurately determine the RMT (Siebner & Rothwell, 2003; Ah Sen et al. 2017).

SAI and LAI

The latency of the N20 component of the somatosensory evoked potential (SEP) was first obtained in each individual. To record SEPs, electroencephalography electrodes were positioned on the scalp over the primary somatosensory cortex (S1) at C3′ located 2 cm posterior to C3 and referenced to Fz (International 10–20 system). Nerve stimulation was performed with a surface bar electrode positioned over the median nerve at the wrist (cathode proximal). A constant current stimulator (DS7AH; Digitimer, Welwyn Garden City, UK) delivered square wave pulses (200 μs pulse width) at the minimum intensity to evoke a visible twitch in the right abductor pollicis brevis (APB) muscle. Five hundred stimuli were delivered at a rate of 3 Hz and time‐locked averaged to determine the latency of the N20.

SAI was measured using two interstimulus intervals (ISIs) between nerve stimulation and the TMS pulse based on the latency of the N20 potential: N20 + 4 ms and N20 + 6 ms. Fifteen unconditioned MEPs were recorded (i.e. TMS alone) and randomized among 15 conditioned MEPs (nerve stimulation‐TMS) for each ISI with a 6 s inter‐trial interval. LAI was obtained at ISIs of 200, 400 or 600 ms. Similar to SAI, 15 unconditioned MEPs were randomized among 45 conditioned MEPs (15 each for ISI) and 6 s elapsed between trials. Trial sweeps recorded using the Signal software were 0.3 s and 1 s long for collection of SAI and LAI data, respectively. SAI and LAI were assessed pre‐ and post‐drug administration (Fig. 1). The intensity of TMS was set to evoke a MEP of ∼1 mV peak‐to‐peak amplitude prior to the collection of data before and after drug administration. For the collection of SAI and LAI, the nerve stimulation intensity was maintained at the minimum intensity to evoke a visible twitch in the right APB muscle.

Figure 1. Study design.

Study design for (A) Experiment 1 and (B) Experiment 2. Timeline for experimental sessions with two time points corresponding to baseline (T0) and the peak plasma concentration of baclofen/lorazepam (T1).

Experimental design

Experiment 1 (Fig. 1 A) was double‐blinded and placebo‐controlled. Participants were tested in three sessions, each separated by a minimum of 1 week. Within a session, participants were administered either 2.5 mg of lorazepam, 20 mg of baclofen or a placebo. The dosage of lorazepam was chosen because it has been previously shown to reduce SAI (Di Lazzaro et al. 2005a,b, 2007a). The dosage for baclofen that has been previously shown to alter TMS measures of intracortical inhibition is 50 mg (McDonnell et al. 2006). However, to minimize possible risks associated with this dosage, we delivered 20 mg in Experiment 1. Dependent measures were acquired prior to (T0) and at 1 h 45 min (T1) following drug administration. The timing of T1 was based on the peak plasma concentrations of both lorazepam (1.5–2.5 h) and baclofen (1–2 h) (Kyriakopoulos et al. 1978; Ziemann et al. 1996).

In Experiment 1, we did not observe any influence of baclofen on sedation levels or physiological measures. We considered that our dosage of baclofen may have been insufficient to observe an effect. Therefore, in Experiment 2 (Fig. 1 B) (double‐blinded and placebo‐controlled), participants were administered either 40 mg of baclofen or a placebo, in two sessions separated by a minimum of 1 week. All TMS measures were acquired prior to (T0) and at 1 h 30 min (T1) following drug administration, based on the timing of the peak concentration of baclofen (1–2 h) (Kyriakopoulos et al. 1978; Ziemann et al. 1996).

For both experiments, the order of dependent measures (SAI, LAI) was pseudo‐randomized across participants using a William's square design. To evaluate the sedative effects of lorazepam and baclofen, a measure of sedation was performed independently by both experimenters present at T1 using a visual analogue scale (VAS) (Di Lazzaro et al. 2005a,b). This scale consisted of a 100 mm line, with 0 mm indicating that the participant was ‘alert’ and 100 mm indicated ‘very sedated’.

Statistical analysis

To avoid contamination of the MEP by background muscle activity, EMG trials were discarded if the peak‐to‐peak amplitude of the signal 50 ms before the TMS artefact was greater than 50 μV.

The following analyses were performed on data obtained from both Experiment 1 and Experiment 2. Normality for all variables was assessed using the Shapiro–Wilks test. If normality was not reached, a square root transformation was applied to the data.

Paired t tests were used to assess changes in RMT following drug administration. For SAI and LAI, the mean peak‐to‐peak MEP amplitude was obtained for the conditioned and unconditioned stimuli separately. Two‐tailed paired t tests were used to compare the conditioned MEP amplitude with the unconditioned MEP amplitude to determine whether significant SAI and LAI were obtained at T0 and T1. Next, inhibition was calculated as a ratio of the mean conditioned to mean unconditioned MEP.

$$\mathrm{SAI}/\mathrm{LAI}=\frac{\mathrm{ME}{\mathrm{P}}_{\mathrm{nerve}\text{-}\mathrm{TMS}}}{\mathrm{ME}{\mathrm{P}}_{\mathrm{TMS}}}=\frac{\mathrm{ME}{\mathrm{P}}_{\mathrm{conditioned}}}{\mathrm{ME}{\mathrm{P}}_{\mathrm{unconditioned}}}$$

A two‐way ANOVA was performed on SAI and LAI data separately for each drug condition using the within‐subject factors TIME (2 levels: T0, T1) and ISI (N20 + 4 ms, N20 + 6 ms for SAI or 200, 400, 600 ms for LAI). Post hoc testing was performed with Tukey's honestly significant difference.

The Mann–Whitney U test was used to compare VAS scores between raters, and the Wilcoxon signed rank test was used to compare VAS scores between drugs. To determine whether changes in afferent inhibition were related to changes in sedation, the percentage change in SAI/LAI was correlated with the VAS scores using Spearman's rho. For all analyses, P < 0.05 was considered statistically significant. Effect sizes were calculated using Cohen's d.

Results

Experiment 1

No serious adverse events were observed following administration of any drug. One participant experienced nausea and vomiting ∼2.5 h following lorazepam ingestion but recovered fully by the next morning. For SAI and LAI, <1% of the total number of trials were removed as a result of excessive EMG activity. Table 1 shows all group‐averaged data from Experiment 1.

Table 1.

Experiment 1 group‐averaged measures (mean ± SD)

	Placebo		Baclofen		Lorazepam
Measure	T0	T1	T0	T1	T0	T1
RMT (%MSO)	41.6 ± 7.1	42.4 ± 7.6	40.6 ± 6.3	40.6 ± 6.6	41.9 ± 7.6	40.0 ± 7.2
Somatosensory evoked potentials
APB motor threshold (mA)	12.6 ± 3.3		11.3 ± 2.6		11.3 ± 3.3
N20 latency (ms)	19.4 ± 0.7		19.7 ± 0.6		19.5 ± 0.6
Short‐latency afferent inhibition
APB motor threshold (mA)	12.2 ± 3.4	11.9 ± 2.7	12.0 ± 3.1	11.9 ± 2.9	11.4 ± 3.3	11.0 ± 2.9
1 mV MEP (%MSO)	51.9 ± 8.9	53.3 ± 9.7	52.1 ± 10.2	50.3 ± 8.7	54.8 ± 7.6	53.2 ± 9.8
N20 + 4 ms	0.63 ± 0.18	0.59 ± 0.24	0.58 ± 0.24	0.64 ± 0.39	0.62 ± 0.20	0.71 ± 0.24
N20 + 6 ms	0.80 ± 0.26	0.84 ± 0.31	0.82 ± 0.26	0.94 ± 0.43	0.75 ± 0.26	0.92 ± 0.30
Averaged ISI	0.71 ± 0.21	0.71 ± 0.25	0.70 ± 0.23	0.79 ± 0.39	0.69 ± 0.22	0.81 ± 0.23
Long‐latency afferent inhibition
APB motor threshold (mA)	12.3 ± 3.6	12.2 ± 2.9	11.9 ± 2.8	12.0 ± 2.7	11.5 ± 3.3	11.0 ± 3.1
1 mV MEP (%MSO)	52.7 ± 8.6	52.4 ± 9.5	53.2 ± 11.2	50.2 ± 9.7	54.4 ± 9.5	53.3 ± 8.6
200 ms	0.57 ± 0.22	0.76 ± 0.55	0.56 ± 0.31	0.55 ± 0.34	0.60 ± 0.31	0.86 ± 0.38
400 ms	0.58 ± 0.22	0.66 ± 0.30	0.60 ± 0.31	0.60 ± 0.29	0.55 ± 0.28	0.77 ± 0.30
600 ms	0.66 ± 0.22	0.72 ± 0.32	0.70 ± 0.28	0.63 ± 0.26	0.62 ± 0.24	0.85 ± 0.47
Averaged ISI	0.61 ± 0.19	0.72 ± 0.35	0.62 ± 0.27	0.60 ± 0.25	0.59 ± 0.25	0.83 ± 0.33

MSO, maximum stimulator output.

RMT was not significantly modified by lorazepam, baclofen or placebo (two‐tailed paired t tests, all P > 0.05), as reported elsewhere (Di Lazzaro et al. 2005a,b, 2007a; McDonnell et al. 2006; Teo et al. 2009). VAS rating of sedation was not different between raters following baclofen, lorazepam or placebo (Mann–Whitney U test, P > 0.05); therefore, the VAS score was averaged across raters. The mean VAS score was significantly greater following lorazepam (60.8 ± 16.5) compared to baclofen (23.5 ± 17.7) and placebo (26.2 ± 11.6) (Wilcoxon signed rank test, all P < 0.05).

Table 2 displays the statistics from the ANOVAs performed on normalized LAI and SAI data from Experiment 1. A two‐way ANOVA using within‐subject factors of ISI and TIME revealed a main effect of TIME (F _1,13 = 6.190, P = 0.027) for LAI in the lorazepam condition, such that LAI was significantly reduced by lorazepam (40.40% reduction) (Fig. 2 A and B). Two‐way ANOVAs for the baclofen and placebo conditioned revealed no main effects or interactions for LAI (Fig. 2 C and D). No correlation between VAS scores and percentage change in LAI following lorazepam was observed (Spearman's rho, r = 0.297, P > 0.05), indicating that sedation was not associated with the reduction in LAI. Individual effects of lorazepam on LAI are shown in Fig. 3 A. Ten individuals (shown with asterisk) demonstrate a reduction of LAI following lorazepam, whereas the remainder show an increase (n = 3) or no LAI at baseline (n = 1). Figure 3 B and C shows individual responses to baclofen and placebo, respectively. For all drug conditions, LAI was present at T0 (MEP_conditioned vs. MEP_{unconditioned}, two‐tailed paired t test, all P < 0.001) and T1 (MEP_conditioned vs. MEP_{unconditioned}, two‐tailed paired t test, all P < 0.05). In summary, the data indicate that LAI is reduced by lorazepam and not baclofen.

Table 2.

Experiment 1 two‐way ANOVA statistics

Dependent measure	ANOVA
Long‐latency afferent inhibition
Lorazepam	ISI2,26 = 0.769, P = 0.474 TIME1,13 = 6.190, P = 0.027 ISI × TIME2,26 = 0.118, P = 0.889
Baclofen	ISI2,26 = 2.435, P = 0.107 TIME1,13 =  = 0.110, P = 0.745 ISI × TIME2,26 = 0.273, P = 0.763
Placebo	ISI2,26 = 0.793, P = 0.463 TIME1,13 = 1.252, P = 0.283 ISI × TIME2,26 = 0.638, P = 0.536
Short‐latency afferent inhibition
Lorazepam	ISI1,13 = 20.634, P = 0.001 TIME1,13 = 5.233, P = 0.040 ISI × TIME1,13 = 0.495, P = 0.494
Baclofen	ISI1,13 = 41.920, P < 0.001 TIME1,13 = 1.116, P =  = 0.310 ISI × TIME1,13 = 0.532, P = 0.479
Placebo	ISI1,13 = 19.710, P = 0.001 TIME1,13 = 0.000, P = 0.989 ISI × TIME1,13 = 1.713, P = 0.213

Figure 2. Experiment 1, LAI.

A, mean ± SD LAI expressed as a ratio of the conditioned MEP (nerve stimulation preceding TMS) to the unconditioned MEP (TMS alone) before (T0) and after (T1) administration of lorazepam. A main effect of TIME is shown on the right, where LAI was significantly reduced following lorazepam administration (significance indicated by an asterisk). B, LAI in one participant before (T0, continuous line) and after (T1, dashed line) administration of placebo, lorazepam or baclofen. Traces show the time‐locked averaged conditioned MEP for participant 14. C, mean LAI (±SD) before and after administration of baclofen. D, mean ± SD LAI before and after administration of placebo.

Figure 3. Experiment 1, individual LAI.

LAI (averaged across ISIs) in individual participants before (T0) and after (T1) (A) lorazepam intake, (B) baclofen intake or (C) placebo intake. Asterisks indicate individuals who demonstrated reduction in LAI following lorazepam intake, reflecting the TIME main effect found in the two‐way ANOVA.

For SAI in the lorazepam condition (Fig. 4 A and B), a two‐way repeated measures ANOVA revealed a main effect of ISI (F _1,13 = 20.634, P = 0.001), such that the magnitude of SAI (mean ± SD) was stronger at N20 + 4 ms (0.67 ± 0.20) than N20 + 6 ms (0.84 ± 0.22). A main effect of TIME was also revealed (F _1,13 = 5.233, P = 0.040), such that SAI was significantly reduced by lorazepam (18.73% reduction). For SAI in the baclofen condition (Fig. 4 C), a two‐way ANOVA showed a main effect of ISI (F _1,13 = 41.920, P < 0.001), such that the magnitude of SAI (mean ± SD) was stronger at N20 + 4 ms (0.61 ± 0.26) than N20 + 6 ms (0.88 ± 0.32). Finally, for SAI in the placebo condition (Fig. 4 D), a two‐way ANOVA revealed a main effect of ISI (F _1,13 = 19.710, P = 0.001), such that the magnitude of SAI (mean ± SD) was stronger at N20 + 4 ms (0.61 ± 0.18) than N20 + 6 ms (0.82 ± 0.25). The reduction of SAI was unrelated to sedation caused by lorazepam (Spearman's rho, r = 0.024, P > 0.05). Individual data for lorazepam effects show that, at N20 + 4 ms (Fig. 5 A, left), 10 individuals showed a reduction following lorazepam, whereas others showed no change (n = 1) or an increase (n = 3). SAI at N20 + 6 ms (Fig. 5 A, right) was reduced in eight participants (indicated by an asterisk), whereas others showed no change (n = 1), an increase (n = 3) or no SAI at baseline (n = 2). Figure 5 B and C shows individual responses to baclofen and placebo, respectively. For all drug conditions, SAI was present at T0 (MEP_conditioned vs. MEP_{unconditioned}, two‐tailed paired t test, all P < 0.001). At T1, SAI was present following lorazepam (MEP_conditioned vs. MEP_{unconditioned}, two‐tailed paired t test, P < 0.01) and placebo (MEP_conditioned vs. MEP_{unconditioned}, two‐tailed paired t test, P < 0.001) but not following baclofen (MEP_conditioned vs. MEP_{unconditioned}, two‐tailed paired t test, P = 0.06). Of note, although the unconditioned MEP is not different from the conditioned MEP at T1 following baclofen, the ANOVA shows no main effect of TIME, indicating that SAI is not significantly modulated by baclofen. In summary, the data indicate that SAI is reduced by lorazepam and not baclofen.

Figure 4. Experiment 1, SAI.

A, mean ± SD SAI expressed as a ratio of the conditioned MEP (nerve stimulation preceding TMS) to the unconditioned MEP (TMS alone) before (T0) and after (T1) administration of lorazepam. Main effects of ISI and TIME are shown on the right, where SAI was stronger for N20 + 4 ms compared to N20 + 6 ms, and SAI was reduced by lorazepam (significance indicated by an asterisk). B, SAI in one participant before (T0, continuous line) and after (T1, dashed line) administration of placebo, lorazepam or baclofen. Traces show the time‐locked averaged conditioned MEP for a participant 9. C, mean ± SD SAI before and after administration of baclofen. A main effect of ISI is shown on the right, where SAI was significantly stronger for N20 + 4 ms compared to N20 + 6 ms. D, mean ± SD SAI before and after administration of placebo. A main effect of ISI is shown on the right, where SAI was significantly stronger for N20 + 4 ms compared to N20 + 6 ms.

Figure 5. SAI at N20.

SAI at N20 + 4 ms (left) and N20 + 6 ms (right) observed in individual participants before (T0) and after (T1) (A) lorazepam intake, (B) baclofen intake or (C) placebo intake. Asterisks indicate individuals who demonstrated reduction in SAI following lorazepam intake, reflecting the TIME main effect found in the two‐way ANOVA.

Experiment 2

No serious adverse events were observed following administration of 40 mg of baclofen. For SAI and LAI, <1% of trials were removed as a result of excessive EMG activity. Table 3 displays all group‐averaged data from Experiment 2. RMT was not significantly modified by baclofen or the placebo (two‐tailed paired t test, both P > 0.05). VAS rating of sedation was not different between raters (Mann–Whitney U test, P > 0.05) and the mean VAS scores did not differ following baclofen (23.4 ± 9.9) and placebo (24.5 ± 12.5) (Wilcoxon signed rank test, P > 0.05).

Table 3.

Experiment 2 group‐averaged measures (mean ± SD)

Measure	Placebo		Baclofen
	T0	T1	T0	T1
RMT (%MSO)	43.3 ± 7.7	41.6 ± 8.0	44 ± 9.2	43.4 ± 9.7
Somatosensory evoked potentials
APB motor threshold (mA)	8.8 ± 2.4		9.3 ± 3.1
N20 latency (ms)	19.3 ± 1.1		19.5 ± 0.9
Short‐latency afferent inhibition
APB motor threshold (mA)	9.4 ± 2.5	9.7 ± 3.2	9.1 ± 3.2	9.0 ± 3.1
1 mV MEP (%MSO)	54.0 ± 12.7	55.1 ± 12.5	57.2 ± 14.0	52.2 ± 10.4
N20 + 4 ms	0.80 ± 0.46	0.72 ± 0.35	0.75 ± 0.37	0.66 ± 0.46
N20 + 6 ms	0.79 ± 0.36	0.81 ± 0.30	0.84 ± 0.23	0.82 ± 0.39
Averaged ISI	0.79 ± 0.39	0.77 ± 0.31	0.79 ± 0.28	0.74 ± 0.32
Long‐latency afferent inhibition
APB motor threshold (mA)	9.6 ± 2.4	9.9 ± 3.3	9.1 ± 3.2	9.2 ± 3.0
1 mV MEP (%MSO)	55.6 ± 12.6	55.4 ± 13.1	54.6 ± 13.2	53.5 ± 10.7
200 ms	0.81 ± 0.47	0.73 ± 0.58	0.60 ± 0.30	0.66 ± 0.45
400 ms	0.78 ± 0.54	0.58 ± 0.30	0.51 ± 0.26	0.56 ± 0.26
600 ms	0.75 ± 0.29	0.73 ± 0.26	0.53 ± 0.24	0.68 ± 0.28
Averaged ISI	0.78 ± 0.39	0.68 ± 0.28	0.55 ± 0.22	0.64 ± 0.27

MSO, maximum stimulator output.

Table 4 shows the statistics from the ANOVAs performed on the normalized LAI and SAI data from Experiment 2. Two‐way ANOVAs with factors of ISI and TIME were performed on LAI data separately for each drug, and no main effects or interactions were found (Fig. 6 A). For SAI, two‐way ANOVAs with factors of ISI and TIME were performed for each drug separately and showed no main effects or interactions (Fig. 6 B).

Table 4.

Experiment 2 ANOVA statistics

Dependent measure	ANOVA
Long‐latency afferent inhibition
Baclofen	ISI1,9 = 0.631, P = 0.543 TIME2,18 = 1.497, P = 0.252 ISI × TIME2,18 = 1.125, P = 0.346
Placebo	ISI1,9 = 0.509, P = 0.610 TIME2,18 = 0.896, P = 0.369 ISI × TIME2,18 = 0.548, P = 0.588
Short‐latency afferent inhibition
Baclofen	ISI1,9 = 1.299, P = 0.284 TIME1,9 = 0.884, P = 0.372 ISI × TIME1,9 = 0.279, P = 0.610
Placebo	ISI1,9 = 0.297, P = 0.599 TIME1,9 = 0.093, P = 0.767 ISI × TIME1,9 = 2.557, P = 0.144

Figure 6. Experiment 2, LAI and SAI.

A, mean ± SD LAI expressed as a ratio of the conditioned MEP (nerve stimulation preceding TMS) to the unconditioned MEP (TMS alone). LAI was not significantly modified by baclofen (left) or placebo (right). B, mean ± SD SAI expressed as a ratio of the conditioned MEP (nerve stimulation preceding TMS) to the unconditioned MEP (TMS alone). SAI was not significantly modified by baclofen (left) or placebo (right).

Discussion

The present study examined the pharmacological influence of GABA_A and GABA_B receptor modulators on SAI and LAI. We report the novel finding that LAI is reduced by lorazepam but not by baclofen, suggesting that LAI is GABA_A but not GABA_B receptor‐modulated. We support previous research indicating that SAI is reduced by benzodiazepines that are positive allosteric modulators of the GABA_A receptor (Di Lazzaro et al. 2005a,b, 2007a; Teo et al. 2009) and extend this knowledge to indicate that SAI is not modulated by the GABA_B agonist baclofen. We discuss these findings and their putative neural mechanisms below.

In the present study, we observed a ∼40% decrease in LAI following administration of lorazepam. One study examined the effect of 2.5 mg of lorazepam and 10 mg of zolpidem (a benzodiazepine) and observed no change in LAI (Teo et al. 2009). Of note, we did not test LAI at the same ISI (100 ms) used by Teo et al. (2009) as a result of the low level of inhibition that they observed at baseline (∼15%). Our LAI data revealed ∼41% inhibition at baseline in accordance with previous work (Chen et al. 1999), allowing for a greater opportunity for the reduction in LAI should it occur following drug ingestion. Next, we observed a ∼19% decrease in SAI following lorazepam. This reduction is consistent with previous findings showing an SAI reduction ranging from ∼15 to 40% following lorazepam or zolpidem administration (Di Lazzaro et al. 2005a,b, 2007a; Teo et al. 2009). SAI is only modulated by benzodiazepines that target GABA_A receptors bearing the α1 subunit, including zolpidem and lorazepam (Di Lazzaro et al. 2007a).

How does lorazepam reduce SAI and LAI?

Lorazepam appears to reduce inhibition in S1 while at the same time increasing inhibition in neighbouring M1. Lorazepam reduces inhibition in S1 as indicated by a decrease in the paired pulse suppression of the SEP components recorded from S1 (Huttunen et al. 2008; Stude et al. 2016). By contrast, lorazepam reduces late I‐waves recorded epidurally following TMS over M1 (Di Lazzaro et al. 2000a), an outcome consistent with increasing the inhibitory effect of GABAergic interneurons within M1. The opposing effects of lorazepam observed in S1 vs. M1 may simply be a result of the differing composition of these two cortical areas. Inhibition plays a large role in S1 with respect to modulating receptor response profiles, where networks of inhibitory interneurons shape the spatial and temporal profiles of excitatory pyramidal neurons (DiCarlo & Johnson, 2000; Wood et al. 2017). M1 is governed by a balance of excitation and inhibition, with excitation mainly governing motor output (Werhahn et al. 2007).

What then are the potential mechanisms by which SAI and LAI are reduced by lorazepam? Although lorazepam acts globally within the cortex, lorazepam may increase GABA_A receptor transmission on the dense inhibitory interneuron population within S1 that ultimately acts to disinhibit pyramidal neurons (DiCarlo & Johnson, 2000; Wood et al. 2017). Disinhibition of S1 pyramidal neurons would allow for excitation of M1 pyramidal neurons via long‐range connections throughout layers II/III (Amassian et al. 1987) and V (Ferezou et al. 2007; Aronoff et al. 2010). In M1, however, lorazepam would be expected to inhibit MEPs relative to baseline by increasing the inhibitory influence of GABAergic interneurons and, although this may be the case, the net influence from the arrival of the afferent volley in M1 is to increase the output of corticospinal pyramidal neurons.

This is the first report to examine the effect of the GABA_B agonist baclofen on LAI and SAI and we did not observe any induced effects using a single 20 or 40 mg dose. Although SAI and LAI are not influenced by baclofen, other TMS evoked circuits are modulated by baclofen. Baclofen increases long‐interval intracortical inhibition (McDonnell et al. 2006; Müller‐Dahlhaus et al. 2008) and reduces intracortical facilitation (Ziemann et al. 1996). For short‐interval intracortical inhibition, baclofen does not change (McDonnell et al. 2007), reduces (McDonnell et al. 2006) or increases (Ziemann et al. 1996) inhibition. Although it is unclear why GABA_B receptors are modulators of the aforementioned circuits but not SAI and LAI, the obvious difference relates to the transmission of the afferent volley that is essential for afferent inhibition.

Functional relevance of afferent inhibition

The reduction of both SAI and LAI by lorazepam provides evidence that they are more similar than originally assumed. This is consistent with a study reporting that chronic subthalamic nucleus deep brain stimulation normalized both SAI and LAI in Parkinson's disease (Wagle Shukla et al. 2013). However, the functional relevance of these two phenomena may be entirely different. SAI is impaired in a variety of clinical populations (Turco et al. 2018b). Most often shown is reduced SAI in disorders of cognition such as Alzheimer's disease (Di Lazzaro et al. 2002, 2006, 2007b, 2008, Nardone et al. 2006, 2008, 2013; Sakuma et al. 2007; Martorana et al. 2009; Celebi et al. 2012; Marra et al. 2012; Di Lorenzo et al. 2013; Terranova et al. 2013; Yarnall et al. 2013) and in those with mild cognitive impairment (Tsutsumi et al. 2012; Nardone et al. 2012a). In those with REM sleep behaviour disorder, SAI is positively correlated with greater executive function, verbal memory and visuospatial abilities (Nardone et al. 2012b, 2013). Furthermore, in healthy individuals, SAI has been shown to be enhanced only during the retrieval phase and not the encoding or consolidation phase of memory (Bonni et al. 2017). Therefore, it is clear that SAI plays a role in various aspects of human cognition. Previous work also shows that SAI can be used to quantify neurophysiological changes. SAI is reduced in individuals with chronic incomplete spinal cord injury, reflecting impaired transmission of afferent input to M1 (Bailey et al. 2015). Furthermore, following ischaemic stroke, individuals showing greater reductions in SAI also show greater improvement in symptoms 6 months post‐injury (Di Lazzaro et al. 2012). Therefore, SAI may potentially be used as a biomarker of functional recovery following neurological injury; however, further research is necessary to confirm this notion.

There is a paucity of research investigating LAI in relation to human behaviour. In clinical populations, LAI is most often abnormal in those displaying deficits in sensorimotor abilities such as Parkinson's disease (Sailer et al. 2003), complex regional pain syndrome (Morgante et al. 2017) and focal hand dystonia (Pirio Richardson et al. 2009). It is not known whether, similar to SAI, LAI is also related to human cognition. Next, it is not known whether afferent inhibition is related to basic aspects of sensation and movement. However, it is commonly assumed that both SAI and LAI are indirect assessments of sensorimotor integration based on the cortical loci that are targeted by the afferent signal, mainly S1 and M1, as well as reports that afferent inhibition is modulated during movement and movement planning (Voller et al. 2005, 2006; Richardson et al. 2008; Ni et al. 2011; Asmussen et al. 2013, 2014; Cho et al. 2016). We recently reported that there is no significant relationship between afferent inhibition and tactile or motor performance (Turco et al. 2018b). Further research is needed to expose behavioural correlates of afferent inhibition to improve our understanding of this phenomenon.

Limitations and future considerations

Our sample size was determined based on estimates from previous literature (Di Lazzaro et al. 2005a,b, 2007a). However, we note that post hoc power analyses of our data reveal a power of 0.64 and 0.51 for the reduction in LAI and SAI by lorazepam, respectively. Therefore, to achieve a higher power of 0.8, we would need to test 26 participants. Baclofen did not demonstrate significant sedative effects as assessed with the VAS at either the 20 or 40 mg dosages. It is possible that a higher baclofen dosage would induce sedative effects and alter SAI/LAI, although higher dosages were beyond the safety limitations of the research. We only examined the effect of GABAergic modulators on afferent inhibition. Future studies should consider other neuromodulators that may play a role in shaping afferent inhibition. SAI is modulated by cholinergic drugs (Di Lazzaro et al. 2000b); however, the role of ACh in LAI is unknown. Serotonin, a neuromodulator that excites GABAergic interneurons (Abi‐Saab et al. 1999) and reduces the responsiveness of neurons in the somatosensory cortex to afferent input (Waterhouse et al. 1986), may also modulate afferent inhibition.

Conclusions

We have shown for the first time that LAI is modulated by GABA_A receptor activity. SAI was reduced by lorazepam, confirming previous studies reporting that GABA_A receptors modulate SAI. Furthermore, LAI and SAI are not influenced by baclofen, suggesting that GABA_B receptor activity does not modulate these phenomena. These findings advance our understanding of the pharmacological basis of afferent inhibition in humans.

Additional information

Competing interests

The authors declare that they have no competing interests.

Author contributions

CVT, JE, RC, SB and AJN conceived and designed the study. CVT, JE, MBL, RC and AJN analysed and interpretated data. CVT, JE, MBL, RC, SB and AJN drafted the work or revised it critically for important intellectual content. All authors have approved the final version of the manuscript and agree to be accountable for all aspects of the work. All persons designated qualify for authorship, and all those who qualify for authorship are listed. Experiments were conducted at McMaster University.

Funding

This work was supported by funding from the Natural Sciences and Engineering Research Council (NSERC) of Canada to AJN (RGPIN 2015).

Biography

Claudia Turco is a PhD candidate at McMaster University. She uses a combination of peripheral nerve stimulation and transcranial magnetic stimulation to investigate the sensorimotor phenomenon known as afferent inhibition. Her primary research interest is to uncover the neural mechanisms that underlie afferent inhibition and improve our understanding of afferent inhibition in relation to human behaviour.