Human primary somatosensory cortex is differentially involved in vibrotaction and nociception

Whereas the role of the primary somatosensory cortex (S1) in vibrotaction is well established, its involvement in nociception remains strongly debated. By assessing, in healthy volunteers, the effect of high-definition transcranial direct current stimulation over S1, we demonstrate a differential involvement of S1 in vibrotaction and nociception.

Abstract

The role of the primary somatosensory cortex (S1) in vibrotaction is well established. In contrast, its involvement in nociception is still debated. Here we test whether S1 is similarly involved in the processing of nonnociceptive and nociceptive somatosensory input in humans by comparing the aftereffects of high-definition transcranial direct current stimulation (HD-tDCS) of S1 on the event-related potentials (ERPs) elicited by nonnociceptive and nociceptive somatosensory stimuli delivered to the ipsilateral and contralateral hands. Cathodal HD-tDCS significantly affected the responses to nonnociceptive somatosensory stimuli delivered to the contralateral hand: both early-latency ERPs from within S1 (N20 wave elicited by transcutaneous electrical stimulation of median nerve) and late-latency ERPs elicited outside S1 (N120 wave elicited by short-lasting mechanical vibrations delivered to index fingertip, thought to originate from bilateral operculo-insular and cingulate cortices). These results support the notion that S1 constitutes an obligatory relay for the cortical processing of nonnociceptive tactile input originating from the contralateral hemibody. Contrasting with this asymmetric effect of HD-tDCS on the responses to nonnociceptive somatosensory input, HD-tDCS over the sensorimotor cortex led to a bilateral and symmetric reduction of the magnitude of the N240 wave of nociceptive laser-evoked potentials elicited by stimulation of the hand dorsum. Taken together, our results demonstrate in humans a differential involvement of S1 in vibrotaction and nociception.

NEW & NOTEWORTHY Whereas the role of the primary somatosensory cortex (S1) in vibrotaction is well established, its involvement in nociception remains strongly debated. By assessing, in healthy volunteers, the effect of high-definition transcranial direct current stimulation over S1, we demonstrate a differential involvement of S1 in vibrotaction and nociception.

the role of the primary somatosensory cortex (S1) in vibrotaction is well established (Abraira and Ginty 2013). In contrast, its involvement in nociception remains elusive (Bushnell et al. 1999). For example, lesions of S1 markedly impair many aspects of tactile perception (Penfield and Boldrey 1937) but have little or no long-standing effect on the ability to perceive pain (Head and Holmes 1911). Similarly, focal seizures of S1 and direct electrical stimulation of S1 in awake patients undergoing surgery for epilepsy can generate vivid touch-related paresthesias but do not appear to elicit pain (Mazzola et al. 2012; Penfield 1947; Tuxhorn 2005). Also supporting the notion that S1 is involved differentially in the processing of touch and pain is the observation in animals that S1 is not the main projection site of nociceptive spinothalamic input, which instead projects predominantly to the insula, the secondary somatosensory cortex (S2), and the cingulate cortex (Dum et al. 2009). Nevertheless, functional neuroimaging studies using magnetic resonance imaging (MRI) or positron emission tomography (PET) have shown that nociceptive stimuli elicit a clear hemodynamic response in the contralateral S1, at a location corresponding to the somatotopic representation of the stimulated body site (Bushnell et al. 1999; Chen et al. 2011, 2012; Coghill et al. 1994; Hu et al. 2015). Electrophysiological studies using electroencephalography (EEG), magnetoencephalography (MEG), and intracerebral recordings have provided less consistent findings but still suggest that nociceptive stimuli elicit responses in the contralateral S1, at latencies compatible with the earliest stages of the cortical processing of nociceptive inputs (Kanda et al. 2000; Ploner et al. 1999, 2002; Tarkka and Treede 1993; Valentini et al. 2012).

Finally, studies have attempted to assess the differential involvement of S1 in touch and pain by characterizing the effect of repetitive transcranial magnetic stimulation (rTMS) over S1 on the perception and event-related brain potentials (ERPs) elicited by nonnociceptive and nociceptive somatosensory stimuli delivered to the ipsilateral and contralateral hemibody (Poreisz et al. 2008a; Torta et al. 2013). This approach is highly relevant because, unlike studies based on sluggish hemodynamic responses sampled with functional neuroimaging techniques, studies based on the direct sampling of cortical activity using ERPs have the temporal resolution required to tease out S1 responses related to the early stages of the cortical processing of ascending somatosensory input from late responses triggered by reentrant feedback projections to S1 originating from higher-order cortical areas. Unfortunately, the results of these studies were largely inconclusive, mainly because they failed to demonstrate any clear and reproducible effect of rTMS on the excitability of S1. For instance, using various protocols of theta-burst stimulation (TBS, a special form of rTMS), Poreisz et al. (2008a) showed that rTMS delivered over S1 reduces the magnitude of the N240 wave of nociceptive laser-evoked brain potentials elicited by stimulation of the contralateral hand, but they did not compare directly this effect to the effect on the responses elicited by stimulation of the ipsilateral hand. Such a direct comparison was performed by Torta et al. (2013). In that study, they assessed the effects of TBS delivered over the primary motor cortex (M1) and S1 on the ERPs elicited by both nonnociceptive and nociceptive stimuli delivered to the two hands but failed to disclose any specific effect of TBS on the responses elicited by stimulation of the contralateral hand. These inconstant findings could be related to the increasingly acknowledged large interindividual variability of the effects of rTMS delivered over the sensorimotor cortex. For example, Hamada et al. (2013) and Huang and Mouraux (2015) recently showed that continuous TBS delivered over M1 decreases motor excitability in some individuals whereas it increases motor excitability in a similar number of other individuals.

Here, in a first experiment we attempt to determine whether S1 is involved differentially in the processing of touch and pain, using another noninvasive technique to modulate the excitability of the human sensorimotor cortex: cathodal high-definition transcranial direct current stimulation (HD-tDCS). The frequently proposed mechanism of cathodal tDCS is that neuronal populations located below the cathode become hyperpolarized, thereby reducing their excitability (Datta et al. 2009; Nitsche et al. 2008; Nitsche and Paulus 2000). The cathode electrode was placed over the expected hand representation of the left or right S1, surrounded by four return anode electrodes placed on a 5-cm-radius circle. Previous studies have shown that the neuromodulation induced by this 4 × 1 ring HD-tDCS montage is much more focal than the neuromodulation induced by the conventional tDCS configuration consisting of two large rectangular electrodes (Datta et al. 2009; Edwards et al. 2013; Kuo et al. 2013; Villamar et al. 2013) and comparable to that of TMS delivered with a 75-mm figure-of-eight coil (Edwards et al. 2013). This allowed us to compare, within subjects, the aftereffects of HD-tDCS applied for 20 min over S1 on the perception and ERPs elicited by nonnociceptive and nociceptive stimuli delivered to the ipsilateral and contralateral hands relative to the hemisphere onto which HD-tDCS was applied.

Transcutaneous electrical stimulation of the median nerve at the level of the wrist was used to compare the effects of HD-tDCS on the early-latency response of S1 to nonnociceptive somatosensory input originating from the contralateral and ipsilateral hands and, thereby, to confirm the specific neuromodulatory effect of HD-tDCS on S1. Within the same experimental sessions, mechanical vibrotactile stimuli and thermal nociceptive stimuli were delivered to the left and right hands to assess whether modulation of the state of the sensorimotor cortex exerts a differential effect on the cortical processing of nonnociceptive and nociceptive somatosensory inputs. We hypothesized that if S1 constitutes an obligatory relay for the cortical processing of somatic input originating from the contralateral hemibody, HD-tDCS delivered over the hand representation of the sensorimotor cortex would affect differently the ERPs elicited by stimulation of the contralateral hand vs. the ipsilateral hand.

Finally, we conducted a second experiment in another group of participants. This experiment was identical to the first experiment except for the fact that cathodal HD-tDCS was applied for only 30 s. Comparison of the aftereffects of real HD-tDCS (HD-tDCS experiment) and sham HD-tDCS (sham experiment) allowed us to test whether the effects on the processing of nonnociceptive and nociceptive somatosensory input were due to a true neuromodulatory effect of 20 min of HD-tDCS or to unrelated time-dependent effects.

MATERIALS AND METHODS

Participants

Fourteen healthy right-handed volunteers were included in a first experiment assessing the effect of cathodal HD-tDCS over the sensorimotor cortex on the perception and ERPs elicited by nonnociceptive and nociceptive stimuli delivered to the ipsilateral and contralateral hands relative to the hemisphere onto which HD-tDCS was applied [HD-tDCS experiment: 12 women, 2 men; 23.2 ± 1.1 yr (mean ± SD; range 21–25 yr)]. Fourteen other participants took part in a second experiment in which real HD-tDCS (20 min of stimulation) was replaced by sham HD-tDCS (30 s of stimulation) [sham experiment: 8 women, 6 men; 27.3 ± 4.7 yr (range 22–35 yr)]. All participants were blinded to the aims of the study. Because participants took part in one or the other experiment, all subjects were equally naive to the procedures when coming to their first and only session. Handedness was assessed with the Flinders Handedness survey (FLANDERS) (Nicholls et al. 2013). Because the skin reflectance, absorption, and transmittance of the infrared radiation generated by the neodymium:yttrium-aluminum-perovskite (Nd:YAP) laser used to deliver nociceptive stimuli (wavelength: 1.34 µm) are highly dependent on skin pigmentation, only participants with light skin were recruited. They were recruited among students and staff of the university. All participants were screened by a neurologist for contraindications to tDCS (Nitsche et al. 2008). None of them had any history of psychiatric or neurological disorders including epilepsy or family history of seizure. The experimental procedures were approved by the Ethics Committee (Commission d’Éthique Biomédicale Hospitalo-Facultaire) of the Université catholique de Louvain (B403201316436), and all participants provided written informed consent.

Experimental Design

HD-tDCS experiment.

Subjects were comfortably seated on a reclining chair during the entire experiment, consisting of two successive EEG recording sessions, immediately before and immediately after application of cathodal HD-tDCS over the left or right sensorimotor cortex (Fig. 1). The side of stimulation was counterbalanced across participants. The second recording always began within 5 min and ended within 25 min of the end of HD-tDCS. Each EEG recording consisted of four blocks whose order was counterbalanced across subjects. In two separate blocks, transcutaneous electrical nerve stimuli were delivered to the left or right median nerve to characterize the early-latency N20 wave, i.e., the first cortical response to nonnociceptive somatosensory input ascending through lemniscal pathways. A total of 500 stimuli were delivered to each hand, with a constant 0.25-s interstimulus interval. In a third block, nonnociceptive vibrotactile stimuli were applied to the left and right index fingertips to elicit ERPs related to the selective activation of tactile mechanoreceptors. In a fourth block, nociceptive laser stimuli were applied to the left and right hand dorsum to elicit ERPs related to the selective activation of heat-sensitive Aδ-fiber nociceptors. In these blocks, the stimuli were delivered randomly to the left or right hand with a random 5- to 7-s interstimulus interval. A total of 25 stimuli were delivered to each hand. After each stimulus, participants were asked to verbally report the intensity of perception using a numerical rating scale (NRS) ranging from 0 (no sensation) to 10 (most intense sensation), 5 marking the border between nonpainful and painful domains of sensation.

Fig. 1.

In 2 separate groups, we assessed the effects of 20 min of HD-tDCS vs. sham HD-tDCS over the sensorimotor cortex on the perception and ERPs elicited by nonnociceptive and nociceptive stimuli delivered to the ipsilateral and contralateral hands. The 2 experiments consisted of 2 EEG recording sessions, immediately before and immediately after 20 min of real HD-tDCS (HD-tDCS experiment) or sham HD-tDCS (sham experiment) of the left or right sensorimotor cortex. During each EEG session, ERPs elicited by nonnociceptive and nociceptive stimuli delivered to the ipsilateral and contralateral hands were recorded. Nonnociceptive stimuli were transcutaneous electrical stimuli delivered to the median nerve at the level of the wrist and vibrotactile stimuli delivered to the index fingertip. Nociceptive heat stimuli were laser pulses delivered to the hand dorsum. The second recording session always began within 5 min after the end of HD-tDCS or sham stimulation and was completed within 25 min.

Sham experiment.

The sham experiment was identical to the HD-tDCS experiment except for the fact that participants received sham cathodal HD-tDCS instead of real HD-tDCS over the left or right sensorimotor cortex. As in the HD-tDCS experiment, the stimulation procedure lasted 20 min. However, the duration of actual current stimulation lasted <2 min (Fig. 1).

Cathodal HD-tDCS

Cathodal HD-tDCS was delivered for 20 min with a 4 × 1 ring montage of five Ag-AgCl sintered ring electrodes (B10 EasyCap) inserted in a ring electrode adapter to increase the area of contact between the electrode, gel, and skin (electrode-gel contact area: 100 mm²; gel-skin contact area: >1.5 cm²) (Minhas et al. 2010). The 4 × 1 ring montage consisted of one cathode electrode placed over International 10-20 System position C3 or C4, surrounded by four return anode electrodes placed on a circle of ~5-cm radius around the cathode (FC5, FC1, CP1, CP5 or FC6, FC2, CP2, CP6). Edwards et al. (2013) demonstrated with electric field modeling and by comparing the effects on motor excitability of HD-tDCS delivered at adjacent positions relative to M1 that the focal aspect of HD-tDCS delivered with this 4 × 1 montage is comparable to that of TMS delivered with a 75-mm figure-of-eight coil. Impedances between the cathode electrode and each anode electrode were kept below 5 kΩ. The stimulation was generated with a constant-current electrical stimulator (Eldith, NeuroConn). In the HD-tDCS experiment, the current was ramped up from 0 to 1 mA during the first 40 s of stimulation and was then maintained constant for 20 min. At the end of these 20 min, the current was ramped down in 40 s. In the sham experiment, the stimulation protocol also lasted 20 min. However, the actual duration of the stimulation was set to 30 s. As in previous studies (Borckardt et al. 2012; Minhas et al. 2010), both real HD-tDCS and sham HD-tDCS elicited a moderate tingling and itching sensation at the site of stimulation. Because this sensation faded within a couple of minutes even when HD-tDCS was maintained for 20 min, the sensations generated by real HD-tDCS and sham HD-tDCS were highly similar and, most probably, indistinguishable.

Nonnociceptive and Nociceptive Somatosensory Stimuli

Nonnociceptive transcutaneous electrical stimulation.

Nonnociceptive transcutaneous electrical stimulation of the median nerve was used to assess in a reliable fashion the early-latency S1 response elicited by nonnociceptive somatosensory input ascending through the lemniscal pathway (i.e., the N20 wave). The stimuli consisted of nonpainful constant-current square-wave (0.5 ms) electrical pulses generated with a DS7 stimulator (Digitimer, Letchworth, UK) and delivered with a pair of 24-mm-diameter adhesive electrodes (Covidien Kendall Disposable Surface EMG/ECG/EKG, Mansfield, MA), separated by a 2-cm interelectrode distance, placed over the median nerve at the level of the wrist. The cathode electrode was positioned proximal relative to the anode electrode. The intensity of stimulation (7.1 ± 1.8 mA) was set to elicit a consistent and visible twitch of the thumb. The same intensity of stimulation was used before and after HD-tDCS, and the adhesive electrodes were not displaced.

Nonnociceptive vibrotactile stimuli.

Nonnociceptive vibrotactile stimuli were short-lasting (50 ms) mechanical vibrations (300 Hz) delivered on the index fingertip with a vibrotactile transducer (length: 2.8 cm, width: 1.2 cm, Haptuators; Tactile Labs, Montreal, ON, Canada). The index fingertip was chosen because of the important density of Pacinian mechanoreceptors in that skin area (Abraira and Ginty 2013). During vibrotactile stimulation, white noise was played through headphones to avoid any auditory response to the sound produced by the transducers.

Nociceptive heat stimuli.

Nociceptive heat stimuli were short-lasting (5 ms) pulses of radiant heat delivered on the hand dorsum with an Nd:YAP laser (wavelength: 1.34 µm; ElEn Group, Firenze, Italy). The hand dorsum was chosen to avoid issues related to the conduction of heat within the thicker skin of the fingertip. Beam diameter at target site was set to 5 mm. The energy of the stimulus (2.0 ± 0.2 J) was adjusted individually to elicit a clear pinprick sensation detected with a reaction time shorter than 650 ms, i.e., a reaction time compatible with the conduction velocity of Aδ fibers (Mouraux et al. 2003; Plaghki et al. 1994; Towell et al. 1996). The same energy was used before and after HD-tDCS. The target of the laser stimulus was slightly displaced after each trial in order to avoid nociceptor habituation and/or sensitization.

EEG Recording

The EEG was recorded at a sampling rate of 4,000 Hz with an average reference (32-channel ASA-LABORATORY EEG system; Advanced Neuro Technologies), with 32 actively shielded Ag-AgCl electrodes mounted in an elastic electrode cap and arranged according to the International 10-20 System (EasyCap 32, EasyCap). During the entire EEG recording, participants were instructed to keep their gaze fixed on a black cross displayed in front of them and to sit as still as possible. Eye movements were recorded with two adhesive surface electrodes placed at the upper right and lower left sides of the left eye. Impedances were kept below 5 kΩ for all leads. The continuous EEG recordings were processed off-line with Letswave6 (http://www.nocions.org/letswave/).

ERP Waveforms

Nonnociceptive ERPs elicited by transcutaneous electrical stimulation of median nerve.

Within the continuous EEG recordings, the electrical stimulation artifact was suppressed with a linear interpolation of the signals recorded from −1 to +7 ms relative to stimulation onset. The recordings were then high-pass filtered with a 0.3-Hz Butterworth zero-phase filter and segmented into 0.2-s epochs ranging from −0.05 to +0.15 s. Artifacts due to eyeblinks or eye movements were removed with a validated method based on an independent component analysis (FastICA algorithm) (Hyvärinen and Oja 2000). After application of a baseline correction (subtraction of the average amplitude of the signal within the reference interval 7–11 ms), the signals were rereferenced to Fz. Epochs containing signals exceeding ±75 µV were rejected to reduce the contribution of artifacts such as head movements, eyeblinks, or muscular activity. Finally, average waveforms were computed for each participant, session, and stimulation side. Within these waveforms, the N20 wave was identified as the most negative deflection occurring 17–23 ms after stimulus presentation (Fig. 2), at the parietal electrode contralateral to the stimulated hand (left hand: P4; right hand: P3) (Cruccu et al. 2008).

Fig. 2.

Nonnociceptive and nociceptive somatosensory ERPs recorded before and after real HD-tDCS (HD-tDCS experiment, left) and sham HD-tDCS (sham experiment, right) of the left or right sensorimotor cortex (group-level average waveforms). The N120 and P250 waves elicited by vibrotactile stimulation and the N240 and P350 waves elicited by laser stimulation of the ipsilateral and contralateral hand are shown at Cz vs. M1M2. The N20 waves elicited by transcutaneous electrical stimulation of the median nerve are shown at the contralateral parietal electrode (Pc: P3 or P4) vs. Fz. The N160 wave elicited by laser stimulation is shown at the contralateral central electrode (Cc: C3 or C4) vs. Fz. Head plots show the scalp topographies of the different components of nonnociceptive and nociceptive ERPs recorded before (blue frames) and after (red frames) HD-tDCS or sham stimulation. Note the marked reduction of the N120 wave elicited by tactile stimulation of the contralateral hand in the HD-tDCS experiment, the reduction of amplitude and increase of latency of the N20 wave elicited by electrical stimulation of the contralateral median nerve, and the absence of such changes in the sham experiment. Also note the symmetric reduction of the N240 wave in the HD-tDCS experiment and the lack of such a reduction in the sham experiment.

Nonnociceptive vibrotactile ERPs and nociceptive laser ERPs.

After application of a 0.3- to 40-Hz Butterworth zero-phase band-pass filter, the continuous recordings were segmented into 3-s epochs ranging from −0.5 to +2.5 s relative to stimulus onset. Artifacts due to eyeblinks or eye movements were removed with the FastICA algorithm (Hyvärinen and Oja 2000). After baseline correction (reference interval −0.5 to 0 s), epochs containing signals exceeding ±75 µV were rejected before computation of separate average waveforms for each participant, session, stimulation type, and stimulation side. Within the nonnociceptive vibrotactile ERP waveforms, two distinct peaks (N120 and P250) were identified at electrode Cz, referenced to M1M2 (García-Larrea et al. 1995; Kenntner-Mabiala et al. 2008; Miltner et al. 1989). The N120 was defined as the most negative deflection peaking 90–160 ms after stimulus onset. The P250 was defined as the most positive deflection following the N120 (Fig. 2). Within the nociceptive laser ERP waveforms, three distinct peaks were identified (N160, N240, and P350) (Bromm and Treede 1984; Cruccu et al. 2008; Hu et al. 2010; Treede et al. 1988). At electrode Cz referenced to M1M2, the N240 was identified as the most negative deflection peaking 140–260 ms after stimulus presentation and the P350 as the most positive deflection following the N240 (Fig. 2). The N160 was defined as the most negative deflection peaking 140–220 ms at the central electrode C3 or C4 contralateral electrode to the stimulated hand and referenced to Fz (Bromm and Treede 1984; Treede et al. 1988).

High-Frequency Oscillations

Previous studies have shown that, in addition to the N20 waveform, transcutaneous electrical stimulation of the median nerve also elicits an early-latency burst of high-frequency oscillations (HFOs: 400–900 Hz) (Ozaki and Hashimoto 2011). These HFOs are commonly separated into an early component thought to be generated by thalamocortical and pyramidal neurons and a late component reflecting inhibitory interneuronal S1 activity (Ozaki and Hashimoto 2011; Restuccia et al. 2011). Such HFOs have not been reported with mechanical stimulation of skin receptors, probably because identifying this high-frequency activity requires a very phasic stimulus repeated a large number of times (Katayama et al. 2010).

To evaluate the effects of HD-tDCS on the magnitude of the HFOs elicited by stimulation of the ipsilateral and contralateral median nerves, the continuous EEG recordings were band-pass filtered with a 400- to 1,000-Hz band-pass Butterworth zero-phase filter after suppression of the electrical stimulation artifact (−1 to 7 ms) and segmented into 0.2-s epochs ranging from −0.05 to +0.15 s relative to stimulus onset. A baseline correction (reference interval 7–11 ms) was performed, and the signals were averaged across trials after rereferencing to Fz. A Hilbert transform was then used to obtain an estimate of the envelope of HFOs (Restuccia et al. 2011). As in previous studies (Katayama et al. 2010; Restuccia et al. 2007), the early and late subcomponents of HFOs were defined relative to the latency of the N20 wave. The early subcomponent extended between −5 and 0 ms relative to the N20 peak, and the late subcomponent extended between 0 and +8 ms (Fig. 3A). The magnitudes of these two subcomponents were estimated by averaging the result of the Hilbert transform within these two intervals. Averaged across participants and conditions, the amplitudes of HFOs were maximal at the central-parietal electrodes contralateral to the stimulated hand (left hand: CP6; right hand: CP5) (Fig. 3B). These electrodes were thus chosen to estimate the magnitude of early and late HFOs across participants and conditions.

Fig. 3.

A: high-frequency oscillations (HFOs) elicited by nonnociceptive electrical stimulation of the median nerve can be separated into an early component (−5 to 0 ms relative to the latency of the N20 wave) and a late component (0 to +8 ms relative to the latency of the N20 wave). Dashed line represents the EEG signal band-pass filtered with a 400- to 1,000-Hz Butterworth zero-phase filter; solid line represents its Hilbert transform (average waveform from 1 recording performed in 1 subject while stimulating the right hand; contralateral central-parietal electrode CP5 vs. Fz). An estimate of the magnitude of early and late HFOs components was computed by calculating the area under the curve of the Hilbert transform, from −5 to 0 ms (early subcomponent) and from 0 to +8 ms (late subcomponent). B: scalp topography of the maximum peak amplitude of HFOs averaged across all participants and all conditions. The amplitude of HFOs was maximal at the contralateral central-parietal electrode (CP5 or CP6 vs. Fz). C: magnitudes of early and late components of HFOs in the HD-tDCS experiment and the sham experiment. Scatterplots represent for each subject the change in amplitude of the responses elicited by stimulation of the contralateral and ipsilateral hands, after vs. before treatment. Box plots show group-level average ± SD. Note in the HD-tDCS experiment as compared with the sham experiment the increase in magnitude of late-latency HFOs most evident when stimulating the contralateral hand.

Statistical Analyses

A mixed-model ANOVA with the between-subject factor Group (real HD-tDCS vs. sham HD-tDCS) and the within-subject factors Time (before vs. after HD-tDCS) and Side (somatosensory stimuli delivered to the ipsilateral vs. contralateral hand relative to the sensorimotor cortex onto which the neuromodulation was applied) was used to test directly the differential effects of real vs. sham HD-tDCS on the perception and ERPs elicited by nociceptive and nonnociceptive stimuli delivered to the ipsilateral and contralateral hands. Indeed, a significant three-way interaction between the Group, Time, and Side would demonstrate a differential effect of HD-tDCS vs. sham stimulation on the responses to stimuli delivered to the ipsilateral vs. contralateral hand; whereas a two-way Time × Group interaction would indicate a bilateral effect of HD-tDCS vs. sham stimulation and a two-way Time × Side interaction would indicate an asymmetric effect on the responses to stimuli delivered to the ipsilateral vs. contralateral hands present after both real and sham HD-tDCS. Finally, a main effect of Time would indicate a bilateral change in the responses.

In a second step, the effects of real HD-tDCS and sham HD-tDCS were assessed within each experiment separately with a repeated-measures ANOVAs with the within-subject factors Time (before vs. after HD-tDCS) and Side (somatosensory stimuli delivered to the ipsilateral vs. contralateral hand relative to the hemisphere onto which the neuromodulation was applied).

A Greenhouse-Geisser correction was used when necessary. When a significant interaction was found, post hoc pairwise comparison or paired t-tests were performed. Significance threshold was set at P < 0.05.

RESULTS

Intensity of Perception

The intensity of the percept elicited by nonnociceptive tactile stimuli was largely unchanged (Fig. 4). Accordingly, the mixed-model ANOVA showed no main effect of Time [F(1,26) = 0.013; P = 0.908] and no interaction between the factor Time and the factors Group and Side (Table 1).

Fig. 4.

Effect of real HD-tDCS (HD-tDCS experiment) and sham HD-tDCS (sham experiment) on the intensity of the perception elicited by nonnociceptive vibrotactile and nociceptive laser stimuli delivered to the contralateral and ipsilateral hands. Scatterplots represent for each subject the average % change in percept before vs. after HD-tDCS or sham stimulation. Box plots show the group-level average ± SD. Note the bilateral reduction of the perception elicited by nociceptive laser stimulation, which is most pronounced after real HD-tDCS.

Table 1.

Mixed-model ANOVAs with between-factor Group and within-subject factors Time and Side

	Time × Side × Group		Time × Group		Time × Side		Time
	F value	P	F value	P	F value	P	F value	P
Nonnociceptive stimulation
Intensity of perception	0.15	0.701	0.48	0.495	5.32	0.029*	0.01	0.908
N20 amplitude	2.07	0.163	0.03	0.858	8.42	0.007*	0.00	0.989
N20 latency	6.93	0.014*	1.89	0.181	0.37	0.549	0.68	0.417
N120 amplitude	7.35	0.012*	0.65	0.429	0.08	0.785	3.11	0.09
P250 amplitude	0.57	0.457	0.01	0.921	0.36	0.556	4.76	0.038*
HFO late subcomponent	0.92	0.347	7.03	0.013*	3.29	0.081	1.36	0.254
Nociceptive stimulation
Intensity of perception	0.14	0.716	1.71	0.203	2.97	0.097	11.4	0.002*
N160 amplitude	0.08	0.787	1.27	0.269	0.40	0.531	8.33	0.008*
N240 amplitude	0.47	0.501	6.06	0.021*	0.36	0.556	12.78	0.001*
N240 latency	0.03	0.874	2.12	0.157	4.61	0.041*	0.24	0.625
P350 amplitude	3.46	0.074	0.28	0.602	2.26	0.145	21.91	0.000*

Values are results of mixed-model ANOVAs with between-factor Group (HD-tDCS experiment vs. sham experiment) and within-subject factors Time (before vs. after HD-tDCS) and Side (stimulation of ipsilateral vs. contralateral hand) A 3-way Time × Side × Group interaction indicates a differential effect of HD-tDCS vs. sham stimulation on the responses to stimuli delivered to the ipsilateral vs. contralateral hand. A 2-way Time × Group interaction indicates a bilateral effect of HD-tDCS vs. sham stimulation on the responses to stimuli delivered to both hands, whereas a 2-way Time × Side interaction indicates an asymmetric effect on the responses to stimuli delivered to the ipsilateral vs. contralateral hand both after real HD-tDCS and after sham HD-tDCS. Finally, a main effect of Time indicates a bilateral change in the responses in both experiments. Significant values are in bold.

^*P < 0.050.

In contrast, the intensity of the percept elicited by nociceptive laser stimuli was reduced both after real HD-tDCS and after sham HD-tDCS. This reduction was symmetric at both hands and, on average, more pronounced in the group who received real HD-tDCS (average reduction at both hands: −15% to −20%) compared with the group who received sham HD-tDCS (average reduction at both hands: −1% to −11%). The mixed-model ANOVA confirmed a main effect of Time [F(1,26) = 11.4; P = 0.002] but showed no significant interaction between the factor Time and the factors Group and Side (Table 1). The within-subject ANOVAs conducted separately for each experiment showed a significant main effect of Time in the HD-tDCS experiment [F(1,13) = 9.4; P = 0.009] but not in the sham experiment [F(1,13) = 2.57; P = 0.133] (Tables 2 and 3, Fig. 4).

Table 2.

Repeated-measures ANOVAs for HD-tDCS experiment with factors Time and Side

	Main Effect of Time		Interaction Time × Side
	F value	P	F value	P
Nonnociceptive stimulation
Intensity of perception	0.40	0.536	2.14	0.167
N20 amplitude	0.01	0.907	9.27	0.009*
N20 latency	2.09	0.172	9.24	0.009*
N120 amplitude	4.05	0.065	11.03	0.006*
P250 amplitude	2.02	0.179	0.62	0.445
HFO late subcomponent	6.45	0.025*	3.82	0.072
Nociceptive stimulation
Intensity of perception	9.4	0.009*	0.89	0.362
N160 amplitude	4.88	0.046*	0.05	0.833
N240 amplitude	13.20	0.003*	0.60	0.453
N240 latency	1.59	0.229	2.53	0.136
P350 amplitude	10.82	0.006*	0.08	0.777

Values are results of repeated-measures ANOVAs for the HD-tDCS experiment with factors Time (before vs. after HD-tDCS) and Side (stimulation of ipsilateral vs. contralateral hand). Significant values are in bold.

^*P < 0.050.

Table 3.

Repeated-measures ANOVAs for sham experiment with factors Time and Side

	Main Effect of Time		Interaction Time × Side
	F value	P	F value	P
Nonnociceptive stimulation
Intensity of perception	0.14	0.714	3.19	0.098
N20 amplitude	0.02	0.894	1.09	0.315
N20 latency	0.18	0.679	1.43	0.253
N120 amplitude	0.39	0.545	0.74	0.406
P250 amplitude	2.81	0.118	0.02	0.880
HFO late subcomponent	1.27	0.280	0.37	0.555
Nociceptive stimulation
Intensity of perception	2.57	0.133	2.24	0.158
N160 amplitude	4.42	0.056	0.70	0.418
N240 amplitude	1.00	0.336	0.01	0.941
N240 latency	0.57	0.462	2.09	0.172
P350 amplitude	11.56	0.005*	4.55	0.052

Values are results of repeated-measures ANOVAs for the sham experiment with factors Time (before vs. after sham HD-tDCS) and Side (stimulation of ipsilateral vs. contralateral hand). Significant values are in bold.

^*P < 0.050.

Early-Latency ERPs Elicited by Electrical Stimulation of Median Nerve

Transcutaneous electrical stimulation of the median nerve elicited a consistent N20 wave in each participant and condition (Fig. 2). Compared with the N20 wave elicited by stimulation of the hand ipsilateral to the sensorimotor cortex onto which HD-tDCS was applied, the magnitude of the N20 wave elicited by stimulation of the contralateral hand was, on average, reduced both after real HD-tDCS and after sham HD-tDCS. This asymmetric reduction in amplitude was more pronounced after real HD-tDCS compared with sham HD-tDCS. This observation was confirmed by the mixed-model ANOVA, which showed a significant interaction between the factors Time and Side [F(1,26) = 8.42; P = 0.007] but no significant interaction between these two factors and the between-subject factor Group [F(1,26) = 2.07; P = 0.163; Table 1]. In the HD-tDCS experiment, the within-subject ANOVA showed a significant Time × Side interaction [F(1,13) = 9.27; P = 0.009; Table 2], and post hoc comparisons confirmed that, after real HD-tDCS, the magnitude of the N20 wave elicited by stimulation of the contralateral hand was significantly reduced (−0.31 ± 0.48 μV; t = 2.436; P = 0.030) whereas the magnitude of the N20 wave elicited by stimulation of the ipsilateral hand tended to increase (+0.28 ± 0.69 μV), but this increase was not significant (t = 1.52; P = 0.152). In the sham experiment, the Time × Side interaction was not significant [F(1,13) = 1.09; P = 0.315].

HD-tDCS exerted a significant effect on the latency of the N20 wave. On average, the latency of the N20 wave elicited by stimulation of the contralateral hand was significantly increased after real HD-tDCS but not after sham HD-tDCS (Fig. 2). The mixed-model ANOVA showed a significant three-way interaction between the factors Group, Time, and Side [F(1,26) = 6.93; P = 0.014; Table 1]. In the HD-tDCS experiment, the within-subject ANOVA showed a significant Time × Side interaction [F(1,13) = 9.24; P = 0.009], and the post-hoc comparisons confirmed that, after HD-tDCS, the latency of the N20 elicited by stimulation of the contralateral hand was significantly increased (+0.3 ± 0.4 ms; t = 2.51; P = 0.026) whereas the latency of the N20 elicited by stimulation of the ipsilateral hand was unchanged (+0.0 ± 0.4 ms; t = 0.000; P = 1.0). In the sham experiment, the within-subject ANOVA showed no significant Time × Side interaction [F(1,13) = 1.43; P = 0.253; Table 3].

HFOs Elicited by Electrical Stimulation of Median Nerve

In all conditions of both experiments, transcutaneous electrical stimulation of the median nerve elicited a significant burst of HFOs, centered around the latency of the N20 wave.

On average, the magnitude of the early subcomponent of HFOs was not changed after real HD-tDCS and after sham HD-tDCS (Fig. 3C). The mixed-model ANOVA showed no main effect of Time and no interaction between the factor Time and the factors Group and Side.

In contrast, the magnitude of the late subcomponent of HFOs was, on average, increased after HD-tDCS but not after sham HD-tDCS. This was confirmed by the results of the mixed-model ANOVA, which showed a significant interaction between the factors Time and Group [F(1,26) = 7.03; P = 0.013; Table 1]. In the HD-tDCS group, although the increase in HFOs magnitude was, on average, more pronounced for stimuli delivered to the contralateral hand compared with the ipsilateral hand, the within-subject ANOVA showed a main effect of Time [F(1,13) = 6.45; P = 0.025] but no significant Time × Side interaction [F(1,13) = 3.82; P = 0.072; Table 2]. In the sham group, the within-subject ANOVA showed no significant changes in HFO magnitude (Table 3).

ERPs Elicited by Nonnociceptive Vibrotactile Stimulation of Hand Dorsum

Nonnociceptive vibrotactile stimuli delivered to the index fingertip elicited a consistent negative-positive potential (N120, P250) maximal at the scalp vertex in each participant and condition (Fig. 2).

After real HD-tDCS, the magnitude of the N120 wave was, on average, reduced after stimulation of the hand contralateral to the sensorimotor cortex onto which HD-tDCS was applied but not after stimulation of the ipsilateral hand. After sham HD-tDCS, the magnitude of the N120 wave was virtually unchanged. This differential effect of real HD-tDCS on the magnitude of the N120 waves elicited by vibrotactile stimulation of the contralateral and ipsilateral hands was confirmed by the results of the mixed-model ANOVA, which revealed a significant three-way interaction between the factors Group, Time, and Side [F(1,26) = 7.35; P = 0.012]. In the HD-tDCS group, the within-subject ANOVA showed a significant Time × Side interaction [F(1,13) = 11.03; P = 0.006], and post hoc comparison showed that the magnitude of the N120 elicited by stimulation of the contralateral hand was significantly reduced after HD-tDCS (−3.1 ± 3.3 µV; t = 3.46; P = 0.004) whereas the magnitude of the N120 elicited by stimulation of the ipsilateral hand was not (−0.1 ± 3.5 µV; t = 0.113; P = 0.912). In the sham group, the within-subject ANOVA showed no significant changes in N120 magnitude (Table 3).

Contrasting with the selective effect of real HD-tDCS on the magnitude of the N120 elicited by stimulation of the contralateral hand, the magnitude of the later P250 was, on average, slightly reduced at both hands, after both real HD-tDCS and sham HD-tDCS. This was corroborated by the results of the mixed-model ANOVA, which showed a main effect of Time [F(1,26) = 4.76; P = 0.038] and no interaction between the factor Time and the factors Group and Side (Table 1; Fig. 5).

Fig. 5.

Single-subject and group-level average change in the magnitude of nonnociceptive (N20, N120, P250) and nociceptive (N160, N240, P350) ERPs before vs. after real HD-tDCS (HD-tDCS experiment) and sham HD-tDCS (sham experiment). Black connected lines show the single-subject differences in amplitude (after − before HD-tDCS or sham stimulation) of the responses elicited by stimulation of the ipsilateral and contralateral hands. Box plots show the group-level average ± SD. Note in the HD-tDCS experiment the asymmetric reduction of the N20 and N120 waves elicited by nonnociceptive stimulation of the contralateral hand and the symmetric reduction of the N160 and N240 waves elicited by nociceptive stimulation of the contralateral and ipsilateral hands.

The latencies of the N120 and P250 were not significantly affected after real or sham HD-tDCS (Fig. 5).

ERPs Elicited by Nociceptive Laser Stimulation

Nociceptive laser stimuli delivered to the hand dorsum elicited a consistent negative-positive complex maximal at the scalp vertex (N240-P350) in each participant and each condition. This complex was preceded by an earlier N160 wave, maximal at central-temporal regions contralateral to the stimulated hand (Fig. 2).

After real HD-tDCS, the magnitude of the N240 was, on average, markedly reduced both for stimuli delivered to the ipsilateral hand and for stimuli delivered to the contralateral hand. In contrast, the magnitude of the N240 was virtually unchanged after sham HD-tDCS (Fig. 2; Table 4). This symmetric reduction of the N240 in the HD-tDCS group was confirmed by the results of the mixed-model ANOVA, showing a significant interaction between the factors Group and Time [F(1,26) = 6.06; P = 0.021]. The within-subject ANOVAs confirmed a main effect of Time in the HD-tDCS experiment [F(1,13) = 13.2; P = 0.003] and the lack of effect of Time in the sham experiment [F(1,13) = 1.0; P = 0.336].

Table 4.

ERP magnitude, ERP latency, HFO amplitude, and intensity of perception before and after real or sham HD-tDCS following stimulation of ipsilateral or contralateral hand

	HD-tDCS Experiment				Sham Experiment
	Contralateral hand		Ipsilateral hand		Contralateral hand		Ipsilateral hand
	Before	After	Before	After	Before	After	Before	After
Nonnociceptive stimulation
N20
Amplitude, µV	−2.47 ± 1.76	−2.15 ± 1.52	−2.30 ± 1.42	−2.58 ± 1.12	−2.14 ± 0.95	−2.06 ± 1.13	−2.36 ± 0.92	−2.48 ± 1.26
Latency, ms	19.3 ± 1.5	19.6 ± 1.3	19.5 ± 1.3	19.5 ± 1.0	19.2 ± 0.5	19.1 ± 0.6	19.1 ± 0.7	19.2 ± 0.7
N120
Amplitude, µV	−9.92 ± 4.94	−6.85 ± 4.51	−8.70 ± 5.12	−8.60 ± 3.62	−6.17 ± 4.48	−6.08 ± 4.64	−5.98 ± 3.89	−4.90 ± 5.35
Latency, ms	130 ± 7	131 ± 9	130 ± 11	135 ± 9	124 ± 15	126 ± 12	125 ± 14	127 ± 14
P250
Amplitude, µV	16.7 ± 6.56	15.2 ± 5.23	16.2 ± 5.55	15.5 ± 5.14	17.1 ± 5.78	15.9 ± 5.68	17.5 ± 5.26	16.3 ± 5.26
Latency, ms	233 ± 44	248 ± 45	247 ± 43	247 ± 40	258 ± 56	265 ± 49	244 ± 47	276 ± 43
HFO
Early component, µV·ms	0.130 ± 0.064	0.137 ± 0.062	0.100 ± 0.040	0.109 ± 0.039	0.121 ± 0.045	0.117 ± 0.047	0.118 ± 0.049	0.124 ± 0.023
Late component, µV·ms	0.106 ± 0.055	0.135 ± 0.068	0.115 ± 0.064	0.118 ± 0.044	0.092 ± 0.027	0.090 ± 0.021	0.109 ± 0.032	0.099 ± 0.033
Intensity of perception, NRS	2.8 ± 1.7	2.6 ± 1.6	2.8 ± 1.6	2.6 ± 1.5	2.6 ± 1.1	2.8 ± 1.3	2.8 ± 1.3	2.8 ± 1.2
Nociceptive stimulation
N160
Amplitude, µV	−9.29 ± 6.27	−6.73 ± 4.42	−9.46 ± 8.83	−6.67 ± 5.42	−5.96 ± 4.34	−5.07 ± 3.47	−6.68 ± 5.75	−5.23 ± 5.78
Latency, ms	175 ± 20	178 ± 16	176 ± 20	182 ± 21	179 ± 28	188 ± 25	182 ± 29	184 ± 25
N240
Amplitude, µV	−19.7 ± 12.7	−14.0 ± 10.1	−20.6 ± 15.6	−13.3 ± 10.3	−13.2 ± 10.3	−12.0 ± 9.67	−12.5 ± 8.60	−11.3 ± 9.81
Latency, ms	219 ± 21	219 ± 19	208 ± 24	222 ± 26	224 ± 21	215 ± 28	223 ± 21	226 ± 27
P350
Amplitude, µV	21.7 ± 11.02	16.0 ± 11.92	21.3 ± 11.66	15.9 ± 14.00	15.6 ± 9.67	13.0 ± 7.04	17.8 ± 8.65	11.6 ± 6.99
Latency, ms	326 ± 31	326 ± 34	325 ± 33	339 ± 41	344 ± 44	352 ± 53	341 ± 43	337 ± 48
Intensity of perception, NRS	4.4 ± 1.8	3.9 ± 2.1	4.5 ± 1.7	3.8 ± 2.2	3.8 ± 1.3	3.7 ± 1.4	3.9 ± 1.1	3.6 ± 1.3

Values are group-level average (±SD) ERP magnitude, ERP latency, HFO amplitude, and intensity of perception [numerical rating scale (NRS) extending between 0 and 10] obtained before and after real or sham HD-tDCS following stimulation of ipsilateral or contralateral hand.

Contrasting with this specific but symmetric effect of HD-tDCS on the magnitude of the N240 wave, the magnitudes of the N160 and P350 waves were, on average, reduced both after real HD-tDCS and after sham HD-tDCS, in a symmetric fashion. The mixed-model ANOVAs revealed a main effect of Time [N160: F(1,26) = 8.33; P = 0.008; P350: F(1,26) = 21.9; P < 0.001] and no interaction between the factor Time and the factors Group and Side (Table 1).

The mixed-model ANOVA showed a marginal interaction between the factors Time and Side on the latency of the N240 wave [F(1,26) = 4.61; P = 0.041]. However, the within-subject ANOVAs showed no significant differences in N240 latencies, both in the HD-tDCS experiment and in the sham experiment (Tables 2 and 3, respectively). There was no significant effect of real or sham HD-tDCS on the latencies of the N160 and P350 (Fig. 5).

DISCUSSION

Our results show that cathodal HD-tDCS applied over the hand area of S1 exerts a different effect on the cortical processing of nonnociceptive and nociceptive somatosensory input in humans. Specifically, cathodal HD-tDCS significantly affected the responses to nonnociceptive stimuli delivered to the hand contralateral to the sensorimotor cortex onto which HD-tDCS was applied, as demonstrated by the reduced magnitude and increased latency of the N20 wave elicited by electrical stimulation of the contralateral median nerve and the reduced magnitude of the later-latency N120 wave elicited by vibrotactile stimulation of the contralateral hand dorsum. In contrast, cathodal HD-tDCS of the sensorimotor cortex induced a symmetric effect on the responses to nociceptive stimuli. Rather than reducing the responses elicited by stimulation of the contralateral hand, HD-tDCS led to a symmetric reduction of the N240 wave that was, at least in part, due to a true neuromodulatory effect of HD-tDCS, as it was not observed after sham HD-tDCS.

HD-tDCS of Sensorimotor Cortex Decreases Responsiveness of S1

After cathodal HD-tDCS, the magnitude of the N20 wave elicited by stimuli delivered to the contralateral hand was significantly reduced compared with the N20 wave elicited by stimuli delivered to the ipsilateral hand. This finding is consistent with the results of a previous study showing that cathodal tDCS over the sensorimotor cortex results in a reduction of magnitude of the N20 wave (Dieckhöfer et al. 2006). Considering that the N20 wave originates from Brodmann area 3b of S1 and that it constitutes the earliest measurable cortical response to nonnociceptive somatosensory input (Allison et al. 1989a; Hari et al. 1984; Hari and Forss 1999; Valeriani et al. 2000, 2004; Wood et al. 1985), our finding demonstrates that cathodal HD-tDCS delivered over the sensorimotor cortex significantly reduces the responsiveness of S1 to thalamocortical input ascending within the lemniscal pathways.

Further supporting the fact that HD-tDCS exerted an inhibitory effect on S1 was the finding that the latency of the N20 wave elicited by stimulation of the contralateral hand was significantly increased after cathodal HD-tDCS. Because it seems unlikely that HD-tDCS exerts an effect on the time required for somatosensory afferent volleys to reach the cortex, a possible explanation for the increased latency of the N20 peak is that because of the reduced responsiveness of S1 neurons generation of the postsynaptic cortical activity leading to the N20 wave required accumulation of more afferent input over time. This postsynaptic interpretation is also supported by our finding that HD-tDCS did not modulate early-latency HFOs thought to predominantly reflect synchronized action potentials ascending the thalamocortical projections to S1 (Curio et al. 1997; Hashimoto et al. 1996; Ozaki and Hashimoto 2011; Restuccia et al. 2011).

Cathodal HD-tDCS tended to exert an opposite, excitatory effect on the responsiveness of the contralateral S1. Indeed, whereas the magnitude of the N20 wave elicited by stimulation of the contralateral hand was significantly decreased after HD-tDCS, the magnitude of the N20 wave elicited by stimulation of the ipsilateral hand tended to increase (Fig. 2). Because the four return anode electrodes were located immediately adjacent to the cathode electrode, this opposite effect of HD-tDCS on the responsiveness of the contralateral S1 cannot be explained by an anodal stimulation of the contralateral hemisphere. One possibility could be that it resulted from interhemispheric inhibitory interactions between the left and right sensorimotor cortices (Brodie et al. 2014; Mochizuki et al. 2007; Ragert et al. 2011): applying cathodal HD-tDCS on the sensorimotor cortex could lead to a reduced interhemispheric inhibitory drive toward the contralateral homotopic sensorimotor cortex.

Finally, cathodal HD-tDCS led to a significant increase of the late-latency HFOs immediately following the N20 wave, and this enhancement was more pronounced for the responses elicited by stimulation of the contralateral median nerve (Fig. 3C). Although this constitutes further evidence that HD-tDCS modulated the state of S1, further studies are needed to understand why HD-tDCS reduced the magnitude of the N20 wave but tended to increase the magnitude of late-latency HFOs. Nevertheless, the genuineness of our results is supported by several previous studies showing that various experimental manipulations can lead to dissociated effects on the N20 wave and HFOs (Gobbelé et al. 2003; Katayama et al. 2010; Ogawa et al. 2004). Although it is generally assumed that cathodal tDCS decreases cortical responsiveness because it hyperpolarizes the stimulated neurons (Datta et al. 2009; Nitsche et al. 2008; Nitsche and Paulus 2000), Rahman et al. (2013) recently suggested that during tDCS different cellular elements can become hyperpolarized or depolarized in any given brain region. Such variable effects could be an explanation for the differential effect of HD-tDCS on the magnitude of the N20 wave and that of late-latency HFOs.

S1 Is Obligatory Relay for Higher-Order Cortical Processing of Tactile Input

Cathodal HD-tDCS delivered over the sensorimotor cortex did not only reduce the early-latency responses to tactile stimuli originating from within S1. Indeed, cathodal HD-tDCS also affected later brain responses to vibrotactile stimulation of the contralateral hand, specifically the N120 wave, which is thought to predominantly reflect later stages of cortical processing within the left and right operculo-insular cortex and the cingulate cortex (Allison et al. 1989b, 1992; García-Larrea et al. 1995; Hu et al. 2015; Kunde and Treede 1993; Mouraux et al. 2011). Because there was no reduction of the N120 wave elicited by vibrotactile stimuli delivered to the ipsilateral hand, and no reduction of the N120 wave after sham HD-tDCS in the sham experiment, this finding suggests that late responses to tactile stimuli originating from outside S1 are dependent on the state of S1. In other words, this finding provides support for a serial processing of tactile input from the thalamus to S1 and from S1 to other brain areas such as the operculo-insular cortex and the cingulate cortex. This interpretation is also supported by the observation of Pons et al. (1992), showing that the responses in S2 to tactile stimuli delivered to the hand of Rhesus monkeys are reduced after lesions of the S1 hand area. However, we cannot exclude that the modulation of the N120 wave observed after HD-tDCS resulted from a neuromodulatory effect of HD-tDCS on other brain regions located close to S1, such as S2 or M1 (see Bilateral Effect of HD-tDCS on Nociceptive Processing vs. Response Habituation).

Bilateral Effect of HD-tDCS on Nociceptive Processing vs. Response Habituation

Both the nociceptive ERPs elicited by stimulation of the contralateral hand and the nociceptive ERPs elicited by stimulation of the ipsilateral hand were reduced after cathodal HD-tDCS of the left or right sensorimotor cortex. To examine whether this symmetric reduction of amplitude was due to a neuromodulatory effect of HD-tDCS or to unrelated time-dependent effects such as response habituation (Greffrath et al. 2007) or decreased vigilance (García-Larrea et al. 1997; Legrain et al. 2002; Miltner et al. 1989), we conducted a second experiment in which participants received sham HD-tDCS over the left or right sensorimotor cortex. The reduction of the N240 was present only after real HD-tDCS, indicating that this effect was not merely the consequence of habituation or decreased vigilance. In contrast, the magnitudes of the earlier N160 wave and the later P350 wave were similarly reduced after real HD-tDCS and after sham HD-tDCS, suggesting that they could, at least in part, be due to habituation or decreased vigilance.

It seems unlikely that the bilateral effect of HD-tDCS on the N240 waves of nociceptive ERPs could be explained by a change in the responsiveness of S1 to ascending nociceptive input, as such a change would be expected to preferentially affect the responses to nociceptive input originating from the contralateral hemibody. Considering the size of the electric field generated by the HD-tDCS montage, it is likely that the effects of HD-tDCS were not restricted to S1 but also extended to nearby areas such as M1 and S2. Furthermore, HD-tDCS can be expected to affect not only the targeted area but also remote areas having strong connections with the targeted area (Rahman et al. 2013). The bilateral effect of HD-tDCS on the N240 wave of nociceptive ERPs could thus be due, at least in part, to an indirect modulation of other brain areas (Antal and Paulus 2010; Lefaucheur et al. 2006; Mylius et al. 2012; Tamura et al. 2004). One possibility could be that the bilateral reduction of the N240 resulted from an effect of HD-tDCS on S2 or the highly connected insular and cingulate cortices, as these areas are thought to be the main sources of the N240 and are known to respond to nociceptive stimuli delivered to both the ipsilateral and contralateral hemibodies (Chen et al. 1998; Frot and Mauguière 2003; Garcia-Larrea et al. 2003; Kakigi et al. 1995; Kanda et al. 2000; Tarkka and Treede 1993; Valeriani et al. 1996, 2000; Vogel et al. 2003).

García-Larrea et al. showed with PET that direct electrical epidural stimulation of M1 (a procedure sometimes used for the treatment of intractable chronic pain) induces a significant increase in cerebral blood flow in the ipsilateral thalamus, the anterior cingulate and orbitofrontal cortex, the insula, and the upper brain stem (García-Larrea et al. 1999). This has led some authors to propose that the rTMS or tDCS delivered over the sensorimotor cortex may activate descending inhibitory control mechanisms acting on the spinal transmission of ascending nociceptive inputs (Garcia-Larrea and Peyron 2007). This hypothesis, which is also supported by the results of Onesti et al. (2013) showing that rTMS delivered over the lower limb representation of M1 in patients suffering from diabetic neuropathic pain leads to a reduction of the spinal nociceptive withdrawal reflex (RIII), could also explain our finding that cathodal HD-tDCS leads to a symmetric reduction of the N240 waves of laser-evoked potentials.

In addition to reducing the magnitude of the N240 of both hands, HD-tDCS also appeared to reduce the intensity of the percept elicited by laser stimulation of both hands, and this decrease was, on average, more pronounced after real HD-tDCS compared with sham HD-tDCS. This symmetric effect on pain perception contrasts with the results of some previous studies suggesting that tDCS exerts a stronger effect on the responses elicited by nociceptive stimulation of the contralateral hand (Antal et al. 2008; Csifcsak et al. 2009). However, these studies did not compare directly the responses elicited by stimulation of the ipsilateral and contralateral hands. Furthermore, several previous studies have shown that rTMS or tDCS delivered over the sensorimotor cortex induces a bilateral reduction of pain perception in healthy volunteers (Nahmias et al. 2009; Poreisz et al. 2008b; Terney et al. 2008).

S1 Is Differentially Involved in Processing Nonnociceptive and Nociceptive Inputs

Regardless of the mechanism explaining the bilateral and symmetric reduction of nociceptive ERPs after HD-tDCS delivered over the sensorimotor cortex, our finding that cathodal HD-tDCS exerts a clearly lateralized effect on the responses to tactile input originating from the contralateral hand but does not exert any lateralized effect on the responses to nociceptive input indicates that S1 is not similarly involved in the processing of nonnociceptive and nociceptive inputs. Considering that the early-latency N160 wave of nociceptive ERPs is thought to reflect, at least in part, activity originating from the contralateral S1, one may wonder why HD-tDCS did not induce a lateralized reduction of the N160 similar to the lateralized reduction of the N20 and N120 elicited by nonnociceptive stimulation. This lack of a lateralized effect suggests that HD-tDCS over the sensorimotor cortex does not similarly affect the ability of S1 to respond to nociceptive and nonnociceptive somatosensory inputs. It has been suggested that area 3b of S1 constitutes the primary target of vibrotactile input, whereas nociceptive input predominantly elicits responses in areas 1 and 2 (Bushnell et al. 1999; Valeriani et al. 2004; Vierck et al. 2013; Whitsel et al. 2009). Differences in the orientation of the cortical surface of the different subregions of S1, being more radial or tangential to the scalp surface, could lead to differential effects of HD-tDCS. Modeling studies have shown that, even directly under the stimulating electrode, tDCS predominantly produces currents that are tangential to the scalp surface, and studies on the effects of direct current stimulation of cortical slices have suggested that the aftereffects of tDCS mainly result from changes in the synaptic efficacy of pyramidal neurons whose somatodendritic axis is parallel to the current flow (Rahman et al. 2013).

Study Limitations

A first limitation of our study is the lack of behavioral evidence that HD-tDCS over the sensorimotor cortex modulated the perception of vibrotactile stimuli delivered to the contralateral hand. However, this was also the case in previous studies assessing the effect of cathodal tDCS or TBS over S1 (Grundmann et al. 2011; Torta et al. 2013) and could be related to the fact that subjective reports of the intensity of perception elicited by brief variations of constant amplitude are not a sensitive means to assess tactile discrimination performance (Tamè and Holmes 2016). Future studies should examine whether changes in vibrotaction induced by HD-tDCS over S1 can be identified by using more sensitive tasks to assess intensity, frequency, or spatial discrimination abilities (Morley et al. 2007; Rogalewski et al. 2004).

A second limitation of our study is that the mixed-model ANOVA conducted to compare directly the effects of real HD-tDCS vs. sham HD-tDCS on the magnitude of the N20 wave elicited by electrical stimulation of the median nerve revealed a significant interaction between the factors Time (before vs. after HD-tDCS) and Side (somatosensory stimuli delivered to the ipsilateral vs. contralateral hand) but no interaction with the factor Group (real vs. sham HD-tDCS). This suggests that, even though the lateralized reduction in N20 magnitude was clearly more pronounced after real HD-tDCS, a lateralized reduction might also have been present after sham HD-tDCS. This raises the question as to whether HD-tDCS delivered for 110 min (40 s ramp-up from 0 to 1 mA, 30 s plateau at 1 mA, 40 s ramp-down from 1 to 0 mA), which is commonly used as a sham condition (Nitsche et al. 2008; Tanaka et al. 2009), might actually exert a slight neuromodulatory effect.

Finally, because nociceptive laser stimuli were delivered to the hand dorsum and nonnociceptive vibrotactile were delivered to the index fingertip, one should consider whether slight differences in the somatotopic representation of the hand dorsum and index fingertip could have explained the differential effects of HD-tDCS on nociceptive and vibrotactile ERPs. Source analysis studies using MEG (Omori et al. 2013) and high-resolution functional MRI studies (Nelson and Chen 2008) indicate that the distance between the S1 response to nociceptive stimuli delivered to the hand dorsum and vibrotactile stimuli delivered to the index fingertip is below 1 cm, i.e., well below the focus of the HD-tDCS montage used in the present study, which is thought to generate an electric field having a grossly approximate radius of 5 cm. More importantly, considering interindividual variations in anatomy and the fact that the position of the electrodes was defined on the basis of standard scalp locations, slight differences in the location of the cortical patches processing hand dorsum vs. fingertip input cannot be expected to result in a differential effect of HD-tDCS that was consistent across individuals.

Conclusions

We show that cathodal HD-tDCS delivered over the hand area of the sensorimotor cortex clearly affects the responses to tactile input originating from the contralateral hand in a lateralized fashion, whereas it affects the responses to nociceptive input in a symmetric fashion. Taken together, these results demonstrate, in humans, a differential involvement of S1 in vibrotaction and nociception.

GRANTS

C. Lenoir and A. Mouraux are supported by the European Research Council (ERC Starting Grant PROBING PAIN Grant 336130).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

C.L. and A.M. conceived and designed research; C.L. and G.H. performed experiments; C.L. and A.M. analyzed data; C.L. and A.M. interpreted results of experiments; C.L. prepared figures; C.L. and A.M. drafted manuscript; C.L., G.H., Y.V., S.M.H., and A.M. edited and revised manuscript; C.L., G.H., Y.V., S.M.H., and A.M. approved final version of manuscript.