Neurobiological mechanisms of TENS-induced analgesia

Abstract

Pain inhibition by additional somatosensory input is the rationale for the widespread use of Transcutaneous Electrical Nerve Stimulation (TENS) to relieve pain. Two main types of TENS produce analgesia in animal models: high-frequency (∼50-100 Hz) and low-intensity 'conventional' TENS, and low-frequency (∼2-4 Hz) and high-intensity 'acupuncture-like' TENS. However, TENS efficacy in human participants is debated, raising the question of whether the analgesic mechanisms identified in animal models are valid in humans. Here, we used a sham-controlled experimental design to clarify the efficacy and the neurobiological effects of 'conventional' and 'acupuncture-like' TENS in 80 human volunteers. To test the analgesic effect of TENS we recorded the perceptual and brain responses elicited by radiant heat laser pulses that activate selectively Aδ and C cutaneous nociceptors. To test whether TENS has a long-lasting effect on brain state we recorded spontaneous electrocortical oscillations. The analgesic effect of 'conventional' TENS was maximal when nociceptive stimuli were delivered homotopically, to the same hand that received the TENS. In contrast, 'acupuncture-like' TENS produced a spatially-diffuse analgesic effect, coupled with long-lasting changes both in the state of the primary sensorimotor cortex (S1/M1) and in the functional connectivity between S1/M1 and the medial prefrontal cortex, a core region in the descending pain inhibitory system. These results demonstrate that 'conventional' and 'acupuncture-like' TENS have different analgesic effects, which are mediated by different neurobiological mechanisms.

Keywords: Analgesia; Electroencephalography (EEG); Human; Pain; Resting state; Transcutaneous electrical nerve stimulation (TENS).

Fig. 1

Experimental design. 80 human participants were randomly assigned to four experimental groups (20 subjects per group), as follows. Group 1: high-frequency active TENS; Group 2: low-frequency active TENS; Group 3: high-frequency sham TENS; Group 4: low-frequency sham TENS. High-frequency (100 Hz) and low-frequency (4 Hz) TENS consisted of constant-current square-wave pulses (duration 200 μs) delivered transcutaneously to the radial nerve at the wrist, either on the left or on the right side. Active and sham TENS lasted for 30 min and 45 s respectively. Five minutes before (“Pre-TENS”) and after (“Post-TENS”) TENS, ongoing brain activity was measured using 64-channel EEG. In addition, 20 nociceptive laser stimuli were delivered to participants' hand dorsum, on both sides (10 stimuli per side). After each stimulus, subjects were instructed to rate the intensity and unpleasantness of the perceived pain using a 0–10 numerical rating scale.

Fig. 2

Effects of different TENS type on laser-elicited pain perception and brain responses. Effects of TENS on laser-evoked pain intensity and unpleasantness (top plots) and brain responses (middle and bottom plots) are evaluated as difference between Pre-TENS and Post-TENS sessions (Post-TENS minus Pre-TENS, normalized by subtracting the respective sham data for displaying purpose; statistical results from non-normalized data are reported in the main text). For both TENS types, the decrease of laser-elicited pain perception and brain responses was significantly larger in the active condition than that in the sham condition. High-frequency TENS induced a larger decrease of both pain perception and brain responses when laser stimuli were delivered to the hand ipsilateral to the TENS side than that contralateral to the TENS side (*: p ​< ​0.05; **: p ​< ​0.01; ***: p ​< ​0.001). In contrast, the decrease in pain perception and brain responses induced by low-frequency TENS was similar when laser stimuli were delivered to the hand ipsilateral and contralateral to the TENS side (ns: not significant). Data are mean ​± ​SEM.

Fig. 3

Group-level laser-evoked responses in the time domain. Group-level waveforms and scalp topographies of N2 and P2 waves (Cz-nose, top panel), as well as N1 wave (Cc-Fz, bottom panel), are displayed for each experimental group. In each experimental group, LEPs elicited by stimulation of the hand ipsilateral and contralateral to the TENS side in the Pre-TENS and Post-TENS sessions are superimposed. Scalp topographies are plotted at the peak latency of the N1,N2, and P2 waves.

Fig. 4

Group-level laser-elicited time-frequency responses. Left panel: Group-level time-frequency distributions and scalp topographies of ERP and α-ERD responses, averaged across experimental groups and conditions. The color scale represents the increase or decrease of the oscillatory magnitude, relative to a prestimulus interval (−400 to −100 ms). The displayed time-frequency distributions contain both phase-locked (ERP: 100–500 ms, 1–10 Hz) and non-phase-locked brain responses (α-ERD: 500–1000 ms, 8–12 Hz), highlighted by the dashed lines. ERP and α-ERD magnitudes were measured at central (top left) and parietal-occipital (bottom left) electrodes respectively. Electrodes showing the maximal response for each time-frequency feature are highlighted in white in the scalp topographies. Right panel: The effect of active TENS on the magnitude of the ERP (top right) and α-ERD response (bottom right) was expressed as difference between Pre-TENS and Post-TENS sessions (Post-TENS minus Pre-TENS, normalized by subtracting the respective sham data for displaying purpose; statistical results from non-normalized data are reported in the main text). Data are mean ± SEM.

Fig. 5

Effects of different TENS type on ongoing brain state.Top left panel: Broadband ongoing EEG oscillations in the four experimental groups. Green waveforms show the difference between the Pre-TENS (blue) and Post-TENS (purple) conditions. Only in the low-frequency active TENS group there was a significant difference in the amplitude of alpha oscillations (8–12 Hz, gray area). Scalp maps show the topographical distribution of alpha oscillation amplitude in the four groups. Top right panel: Changes of ongoing alpha oscillations (Post-TENS minus Pre-TENS) were compared across-groups using two-way ANOVA with two between-subject factors (‘TENS frequency’ and ‘condition’). There was a significant ‘TENS frequency’ ​× ​‘condition’ interaction at bilateral central electrodes (electrodes with FDR-corrected p ​< ​0.05 are shown in white), maximal on the electrodes overlying the primary sensorimotor cortex (S1/M1) contralateral to the TENS side. Ongoing alpha oscillations were increased in the post-TENS period only in the low-frequency active TENS group (ns: not significant; ***: p ​< ​0.001). Data are mean ​± ​SEM. Middle panel: Source-level percentage changes of alpha oscillations in the four experimental groups. Only low-frequency active TENS significantly enhanced alpha power in the bilateral S1/M1. Two-way ANOVA revealed a significant ‘TENS frequency’ ​× ​‘condition’ interaction in the S1/M1 contralateral to the TENS side (p ​= ​0.002, middle right; ns: not significant; *: p<0.05; ***: p ​< ​0.001). Note that the source-level plots displayed in the left part are descriptive: they show the voxels whose absolute percentage change of alpha oscillations (alpha power in the post-TENS session relative to the pre-TENS session) was >10%. Bottom panel: Changes of functional connectivity between S1/M1 and mPFC, in the four experimental groups (Post-TENS minus Pre-TENS). Two-way ANOVA revealed that low-frequency TENS caused a significant enhancement of functional connectivity (indexed by both coherence and DTF measures) between the contralateral S1/M1 and mPFC (ns: not significant; ***: p ​< ​0.001).