External stimulation of the trigeminal nerve causes pupil dilation in healthy volunteers, suggesting locus coeruleus modulation

Abstract

Background:

External trigeminal nerve stimulation (eTNS) is a non-invasive neuromodulation method being investigated as a treatment for epilepsy, depression, and stress-related disorders. eTNS has already received approval for treating migraines and ADHD in children. While there is evidence that eTNS is clinically beneficial, the underlying mechanisms are poorly understood. The trigeminal nerve gives input to several nuclei in the brainstem, which then activate the locus coeruleus (LC), regulating norepinephrine (NE) levels. LC activation has been shown to cause pupil dilation, while NE is involved in regulating processes such as arousal and memory. Thus, the LC-NE system may be important in eTNS mechanisms.

Objective:

We hypothesize that eTNS will cause pupil dilation by activating the LC-NE system. Therefore, we investigated the effect of eTNS on pupil responses in healthy volunteers.

Method:

Participants (n=20) underwent three stimulation conditions (eTNS, sham and median nerve stimulation – MNS) while pupil diameter was recorded. MNS was applied as an additional control.

Results:

eTNS and MNS elicited larger pupil responses compared to sham, with eTNS causing the largest pupil dilation. Additionally, the number of pupil dilations was significantly higher after eTNS than after MNS and sham. Latency of pupil responses did not differ between eTNS and MNS, but was significantly later after sham. Response size was associated with baseline pupil diameter.

Conclusion:

eTNS causes significantly larger pupil dilation than MNS-control stimulation. This suggests that eTNS modulates the release of NE leading to pupil dilation. Accordingly, the LC-NE system could be a key component in mediating eTNS effects.

Introduction

The trigeminal nerve is the 5^th and largest cranial nerve with both motor and sensory components. It divides into three different branches, with the ophthalmic branch innervating the upper part of the face, the maxillary branch innervating the middle part and the mandibular branch the lower part [1-4]. External trigeminal nerve stimulation (eTNS) is a non-invasive neuromodulation method that delivers electrical current via patch electrodes attached to the forehead, cheek or jaw to target one of these branches. eTNS has been successfully used as a treatment for acute and chronic migraines, reducing the number of migraine attacks and reported pain [5,6]. It is also applied to treat patients with trigeminal neuralgia [7] and the U.S. Food and Drug Administration (FDA) has recently approved the first eTNS medical device to treat ADHD in children [8,9]. In addition, eTNS is under investigation for the treatment of drug-resistant epilepsy [10-12], and several neuropsychiatric disorders such as major depressive disorder [13,14] and posttraumatic stress disorder [15]. Despite all these well-documented therapeutic benefits of eTNS, the precise working mechanism is poorly understood.

The three branches of the trigeminal nerve meet in the trigeminal ganglion where they then give input to three sensory trigeminal nuclei (the mesencephalic, principal sensory, spinal nuclei) and the trigeminal motor nucleus [1-4]. These trigeminal nuclei transmit pain and tactile information to the thalamus, while also extending projections to multiple brainstem nuclei including the nucleus of the solitary tract and the locus coeruleus (LC) [16,17]. The LC has widespread projections to the neocortex, as well as strong connections to the orbitofrontal cortex, the anterior cingulate cortex [18-20] and the hippocampus [21,22]. Importantly, the LC is the main nucleus that regulates norepinephrine (NE) levels throughout the entire brain [16,23,24]. NE is a neurotransmitter that is part of the sympathetic nervous system [25-27] and plays an important role in a series of processes. More specifically, it regulates our bodily response to threat, the flight-or-fight response [26] as well as arousal [28,29], wakefulness [30,31] and alertness [27]. NE is also known to enhance synaptic plasticity, modulate cortical excitability [32] and regulate learning [33] and cognitive processes [34]. Furthermore, imaging studies have shown that the LC-NE system is activated during memory retrieval tasks, perceptual rivalry and attention tasks [34], and contributes to the optimization of behavioral performance [18,35].

These processes regulated by the LC-NE system, i.e. arousal, fight-or-flight, alertness, etc. can be characterized by pupil dilation. Pupil dilation is controlled by the iris dilator muscle which in turn is regulated by the sympathetic nervous system. The iris dilator muscle connects to a subcortical pathway that starts in the hypothalamus and the LC [36-38], explaining why the pupil enlarges when people are aroused. In addition, recent studies have shown that pupil diameter can serve as a measure of LC activity. In animals, direct evidence showed that LC activation closely matches moment-to-moment changes in pupil size [20,39]. In humans, Murphy, O’Connell, O’Sullivan, Robertson and Balsters [40] found that pupil diameter correlated positively with the Blood Oxygen-Level Dependent (BOLD) activity in the LC. Furthermore, changes in pupil diameter are also linked with shifts in neural activity in the cortex, which is believed to result from modulation by the LC-NE system [19,41-44].

Thus, eTNS may exert its therapeutic effect by activating the LC and regulating NE levels in the brain. Therefore, in the current study, we set out to investigate the effect of eTNS on pupil dilation, a marker for LC activity and arousal. Stimulation was applied during a task-free experiment while pupil diameter was recorded in healthy volunteers. We hypothesized that if eTNS would activate the LC-NE system, it should lead to pupil dilation since there is a positive relation between LC activity and pupil diameter [18,40,45-47].

Methods

This study was approved by the Ethics Committee Research UZ/KU Leuven (S63709) and registered on ClinicalTrials.gov (NCT04577677). All subjects provided signed informed consent before participation and received a monetary reward at the end of the experiment. The study was performed in accordance with the 1964 Helsinki Declaration and in agreement with Good Clinical Practice guidelines.

Participants

Twenty healthy volunteers between 19 – 34 years old (12 male, M = 25.80, SD = 4.42) participated in the experiment. None of the participants reported a history of psychiatric or neurological diseases and did not suffer from trigeminal neuralgia or facial pains.

External Trigeminal Nerve Stimulation

Two skin patch electrodes were placed bilaterally over the mental foramen to stimulate the mandibular branch of the trigeminal nerve. Stimulation pulse trains consisted of symmetric biphasic pulses (200 μs per phase) with a frequency of 1500 Hz, 15 ms duration and were repeated every 4000 ms. The stimulation intensity was determined for each participant individually. First, the detectable threshold was defined by asking participants when they felt an unusual sensation on their chin, starting at an intensity of 2 mA and increasing 1 mA at a time. The intensity was increased until participants reported a clear stimulation sensation, without experiencing any pain. In addition to sham stimulation (no stimulation), Median Nerve Stimulation (MNS) was administered as an extra control condition. The median nerve is the main nerve in the front of the forearm and provides sensory and motor function to the forearm, wrist and hand. Two skin patch electrodes were placed over the right forearm and stimulation intensity was determined similarly to eTNS. All other stimulation parameters were identical to eTNS. The three stimulation conditions (eTNS, sham and MNS) were applied alternatingly with 4000 ms between each stimulation condition. The order of the stimulation conditions was pseudorandomized, so the same stimulation type could not be administered more than two times in a row. All participants received stimulation for 20 minutes while pupillometry was performed (see below). To check whether any side effects of the stimulation were experienced, participants rated on a 5-point Likert scale, ranging from 0 = ‘not at all’ to 4 = ‘very much’, how much they experienced headache, anxiety, phosphenes, skin irritation and redness and burning sensation (See Table 1).

Table 1.

Subjective ratings of eTNS/MNS stimulation

	eTNS M ± SD	MNS M ± SD
I feel a burning sensation	0	0
I feel skin irritation	0	0
My skin is red underneath the electrodes	0	0.75 ± 0.85
My skin feels itchy underneath the electrodes	0	0.35 ± 0.75
I see light flashes (phosphenes)	0	0
I have a headache	0.1 ± 0.31
I feel anxiety	0.05 ± 0.75

Table 1. Reported side effects of eTNS and MNS. All items were scored ranging from 0 = “not at all” to 4 = “very much”. eTNS = external trigeminal nerve stimulation; MNS = median nerve stimulation.

Pupillometry

Continuous pupil diameter was recorded using a Gazepoint GP3 HD eye tracker. This screen-based eye tracker recorded eye movements and pupil dilations while participants focused on a fixation cross in the middle of the screen. Pupil diameter of the left and right eye was sampled at a frequency of 150 Hz. Positions of the eye in relation to the eye tracker were computed using a 9-point calibration, conducted before the start of the experiment. The experiment was carried out in a room with constant ambient lighting conditions.

Experimental procedure

Participants were seated in a soundproof room without any windows to keep luminance constant. First, stimulation electrodes were applied over the chin and wrist and stimulation intensity was determined (See Figure 1A). The experiment consisted of two blocks of 10 min each, with a break of at least 3 min between the blocks. Before the start of the experiment, the subject was taken through a calibration procedure. If necessary, a new calibration took place after the break. In order to keep gaze constant, participants were given verbal instructions to focus on a white cross that was presented in the middle of a black screen. Afterwards, the experimenter left the room and was able to follow the live pupil recording on a separate screen. At the end of the experiment, participants reported whether they experienced any adverse effects of the stimulation. The total duration of the experiment was approximately 45 min.

Figure 1.

A) Procedure of the experiment. Pupil responses were recorded throughout the entire experiment. B) Representative recording of the pupil diameter during eye tracking. A pupil dilation is shown with the pupil metrics (i.e., size and timing) annotated. eTNS = external trigeminal nerve stimulation; MNS = median nerve stimulation.

Data processing

Preprocessing of the pupil data was done in MATLAB (version R2020b [48]), and was based on a pipeline created by Kret and Sjak-Shie [49]. Data points that were invalid due to eye blinks, missing values or other invalidities were linearly interpolated. More specifically, clearly invalid pupil size, such as non-positive values or pupil diameters smaller than 1.5 mm and larger than 9 mm [50], and samples that were marked as ‘invalid’ by the eye tracker were removed. To identify speed dilation outliers, samples with a significantly larger absolute change in pupil size compared to their neighboring samples, the median absolute deviation was calculated [51]. Samples with dilation speed outliers larger than 4 were removed, as well as sample groups within 50 ms of gaps. Remaining outliers were removed and the gaps were linearly interpolated. However, sections with gaps that were larger than 25 ms were considered as ‘missing data’. Of the remaining valid samples, the temporal resolution was increased by smoothing the data using a moving average filter. Afterwards, the filtered signal was grouped according to stimulation condition into eTNS epochs, MNS epochs and SHAM epochs and averaged for each participant to give the mean pupil response signal. Finally, the mean pupil response signal from both eyes was averaged together.

Data analysis

The mean pupil response signal was then analyzed to extract different metric to characterize pupil responses (See Figure 1B). As a primary measure, the area under the curve (AuC) was calculated. It is a robust feature that is often used in pupil studies [52,53] since it considers the time course to maximum pupil response to vary and is mostly constant depending on how pulse trains influence LC firing [53]. First, every trial of every participant was categorized as a pupil dilation or pupil constriction. For AuC, only the trials that were defined as pupil dilations, were included in the analysis. For response time, the latency of the pupil response was evaluated for pupil dilations. The time course of the maxima and minima was detected until 2.5 s after stimulation onset, and then averaged over all trials for the different stimulation conditions (See Figure 1B). Thus, the actual pupil time course were the mean values across trials calculated separately for each participant and condition.

Additionally, the incidence of pupil dilations in the different stimulation conditions was estimated by calculating the percentage of trials where the pupil response was positive. The effect of baseline pupil diameter on the size of pupil responses was examined by investigating the relation between pupil diameter at stimulation onset and pupil response size. For both the incidence and the relation between baseline pupil diameter on pupil response size, all trials were entered into the analysis to avoid biasing the model.

Statistical analysis

To investigate the effect of eTNS on both pupil metrics (AuC and latency of the pupil response), linear mixed effects models were used. As the AuC data was not normally distributed and neither were the residuals of the fitted model, therefore a square root transformation of the AuC data was performed. Stimulation condition (eTNS, MNS and SHAM) was set as a fixed effect and subject as a random factor. For latency, the random effect did not add anything to the model, therefore a Friedman test was performed. The possible modifying effect of stimulation intensity on both pupil metrics was examined by adding it as variable to the model. The backward method was performed to compare the models with and without stimulation intensity. The statistical model with the lowest Bayesian Information Criterion was chosen as the final model.

The influence of the different stimulation conditions on the number of pupil dilations was investigated using linear mixed models. Pearson’s correlation coefficients were used to examine the relation between baseline pupil diameter and the pupil response size.

Estimated marginal means were computed for post-hoc analyses and Tukey’s honest significant difference was used to adjust for multiple comparisons. All analyses were done in RStudio [54], and a significance level of α < 0.05 was adopted.

Results

Figure 2A shows an example of the average pupil response to eTNS, MNS and SHAM for one subject. The shaded area is the 95% confidence interval. Every stimulation condition consisted of 100 repetitions. In Figure 2B, the group average pupil dilation of all participants for every stimulation condition is depicted. eTNS elicited a pupil response ± 1.5 s after stimulation onset. MNS elicited a smaller pupil response compared to eTNS, while SHAM elicited a small to no pupil response.

Figure 2.

A) Average pupil response to the three stimulation conditions for one subject. B) Group average pupil dilation of all participants for every stimulation condition. eTNS = external trigeminal nerve stimulation; MNS = median nerve stimulation. The shaded area is the 95% confidence interval

Response size - AuC

Figure 3A shows the effect of stimulation condition on AuC. The linear mixed effect model revealed a main effect of stimulation condition (F(2,4846.1) = 61.99, p < 0.001). Post-hoc analyses showed that eTNS caused pupil responses with larger AuC than MNS (t(inf) = 3.69, p < 0.001) and SHAM stimulation (t(inf) = 10.95, p < 0.001). MNS also generated pupil responses with larger AuC than SHAM (t(inf) = 7.28, p < 0.001). These results were quite consistent across participants individually (Figure 3B) and were observed in the majority of the participants (16 of 20).

Figure 3.

A) The bar plot shows the average area under the curve for every stimulation condition. B) The plot shows the consistency across individual participants. C) The latency of the pupil response for every stimulation condition. D) The percentage of pupil dilations in every stimulation condition. eTNS = external trigeminal nerve stimulation; MNS = median nerve stimulation. Error bars represent standard errors and colored dots represent different subjects.

Response time - Latency

For the latency of the pupil response, a significant main effect of stimulation condition was found (X²(2,20) = 12.03, p = 0.002). The pupil response happened later after SHAM stimulation compared to eTNS and MNS stimulation (p = 0.046 and p = 0.028, respectively). No difference regarding response time was found between eTNS and MNS (p = 1) (See Figure 3C).

Incidence

Figure 3D shows the percentage of pupil dilations in every stimulation condition. The analysis revealed a significant main effect of stimulation (F(2,36) = 15.19, p < 0.0001). More specifically, the percentage of dilations was significantly higher in the eTNS condition compared to MNS (t(2,36) = −3.07, p = 0.0111) and SHAM (t(2,36) = −5.50, p < 0.0001). The number of pupil dilations did not differ between MNS and SHAM (t(2,36) = −2.43, p = 0.0512).

Effect of baseline pupil diameter on pupil response size

Baseline pupil diameter did not differ between the three different stimulation conditions (X² (2,4982) = 0.004, p = 0.9978). The correlation analysis of the pupil dilations after eTNS showed a negative relation between baseline pupil diameter and pupil response size (Pearson’s R = −0.273, p < 0.001) (see Figure 4A). For pupil dilations after MNS stimulation, pupil response size decreased with increasing baseline pupil diameter (R = −0.222, p < 0.001) (see Figure 4B).

Figure 4.

A) The relation between baseline pupil diameter and pupil response size for pupil dilations after eTNS. B) The correlation between baseline pupil diameter and pupil response size for pupil dilations after MNS. eTNS = external trigeminal nerve stimulation; MNS = median nerve stimulation.

Stimulation intensity

Stimulation intensities ranged from 2 to 10 mA across stimulation conditions (Figure 5). Stimulation intensity was slightly higher with eTNS (5.85 ± 1.82 mA) compared to MNS (4.95 ± 2.0 mA). The paired t-test showed a significant difference of stimulation intensity between eTNS and MNS (t(19) = 3.76, p = 0.001). However, the analysis showed that there was no effect of stimulation intensity on AuC (F(1,40.87) = 0.07, p = 0.800), nor on the latency of the pupil response (F(8,46.35) = 1.40, p = 0.223). In addition, the model for AuC and latency both had a lower Bayesian Information Criterion without stimulation intensity included as a variable (AuC: −62.40 vs. −58.43; Latency: 98.79 vs. 118.1; without vs. with stimulation intensity). Both eTNS and MNS were well tolerated and caused minor to no side effects (See Table 1). The tingling sensation caused by eTNS and MNS was perceived as similar (t(19) = −0.52, p = 0.606). MNS caused significantly more redness (t(19) = −3.94, p < 0.001) and skin irritation under the electrodes compared to eTNS ((t(19) = −2.10, p = 0.049)).

Figure 5.

The mean stimulation intensity for eTNS and MNS. eTNS = external trigeminal nerve stimulation; MNS = median nerve stimulation. Error bars represent standard errors.

Discussion

We investigated the effects of eTNS (and MNS-control stimulation) on pupil dilation metrics in healthy volunteers. We found that eTNS induced larger pupil dilations than MNS and SHAM stimulation. The number of pupil dilations was higher after eTNS than after the MNS-control conditions, and changes in pupil size were associated with pupil diameter at stimulation onset. These results support the hypothesis that eTNS activates noradrenergic pathways such as the LC-NE system.

Only few studies have investigated the effect of eTNS and its underlying working mechanism in healthy volunteers. Mercante and colleagues [55] investigated the acute effects of eTNS on cortical and brainstem excitability. Brainstem function was studied by measuring the blink reflex and its recovery cycle [56,57], while transcranial magnetic stimulation (TMS) was used for intercortical facilitation and inhibition [58]. The results showed that cyclic eTNS only affected brainstem interneuron excitability but not the activity of both facilitatory and inhibitory intracortical excitability. Axelson, Isberg, Flink and Amandusson [59] used a similar TMS protocol to examine the effect of continuous eTNS on cortical excitability as a possible working mechanism behind eTNS effects on drug-resistant epilepsy. Although, other anti-epileptic treatments have been shown to alter cortical excitability, their results also showed no effect of eTNS on excitability parameters. It is possible that the absence of effect of eTNS on the cortical excitability of healthy volunteers can be explained in terms of baseline levels [60]. More specifically, if baseline levels of intracortical inhibition are high or low, it may result in a 'floor' or 'ceiling' effect, rendering the physiological demonstration of the potential influence of eTNS unfeasible. In addition, in the majority of clinical studies, eTNS is applied for long periods [11,13,61-63], and stimulation parameters might not be optimal to alter excitability. Moreover, in a pilot study, Gadeyne and colleagues [64] investigated the effect of eTNS on P300 event-related potentials (ERP), a proxy of noradrenergic signaling in the brain [65,66]. The P300 response was induced by using an auditory oddball paradigm, in which subjects have to press a button upon hearing a rare oddball tone and, refrain when a frequent standard tone is presented. The authors found that eTNS resulted in a clinically relevant increase in P300 amplitude in comparison to no stimulation, albeit not statistically significant, potentially reflecting an increase in noradrenaline release in the brain. However, another study on the effect of eTNS on ERPs in an oddball paradigm did not find the same results [67]. Short-term, cyclic eTNS did not affect any of the ERP parameters.

Since Aston-Jones and Cohen [18] first demonstrated a correlation between pupil diameter and LC single-unit activities in a monkey, pupil diameter has been proposed as a reliable indicator of noradrenergic signaling. Numerous studies have since confirmed this relationship in monkeys [20,68], rats [39,69], mice [70,71], and even in human BOLD fMRI [40]. The time course of pupil dilation induced by eTNS, as observed in Figure 2, resemble the pupil responses observed in monkeys following LC electrical stimulation [20] and optogenetic stimulation in rats [69]. This similarity provides support for the hypothesis that eTNS activates the LC. However, although pupil dilation has been widely used as an indicator of LC activity, it is an indirect measure and can be influenced by other factors, such as mental state and fatigue during the experiment [72-75]. Therefore, while pupil dilation provides valuable insights into the influence of eTNS on the LC-NE system, considering additional measures such as neuroimaging might be interesting for future research.

In our experimental design, we used individual stimulation intensities based on participant’s sensory threshold. The results revealed that a significantly higher current was necessary for eTNS to achieve a similar sensation with MNS stimulation. However, the analysis showed no effect of stimulation intensity on any of the pupil metrics, and the statistical models without intensity had a lower Bayesian Information Criterion. Moreover, based on our current understanding, there have been no previous studies on eTNS that have included an extra control condition alongside the standard SHAM condition. In our study, we introduced MNS to address the startle response, an automatic and mostly unconscious reaction to abrupt or unfamiliar stimuli, which can also result in pupil dilation. This additional control allows us to gain a more accurate understanding of the actual impact of the different stimulation conditions. Furthermore, by randomly alternating between the three stimulation conditions, we considered the natural fluctuations in pupil size that occur over time.

As one of the pioneering studies, our findings provide valuable insight into the potential mechanisms underlying the effects of eTNS, and its therapeutic applications. This knowledge is crucial for optimizing eTNS protocols, refining treatment approaches, and expanding the potential clinical applications of this promising neuromodulatory technique. Moreover, the use of non-invasive neuromodulation methods has grown exponentially over the past decade. In the clinic, a small number of eTNS therapies have received approval, while some are in the clinical trial stage. Additionally, there is a growing interest in investigating the involvement of the LC-NE system in several facets of human cognition, including learning and memory, decision making, and aging and neurodegeneration. Subsequently, eTNS emerges as a valuable, novel tool.

Pupil dilation could be used as a biomarker to quantify the effectiveness of eTNS to optimize stimulation parameters. In our study, we stimulated between the sensory threshold and pain threshold of every participant for a duration of 15 ms. Future studies could systematically compare different stimulation intensities, durations and frequency per condition. Regarding stimulation duration and frequency, future research can investigate the precise combination of parameters required to elicit substantial effects on arousal measures, brain activity and behavior.

In addition, to gain a more comprehensive understanding of the underlying mechanisms of eTNS, it is valuable to explore additional metrics. More specifically, measures such as electroencephalography (EEG), functional magnetic resonance imaging (fMRI), and salivary α-amylase could provide further insights into the physiological and neurobiological processes associated with eTNS. These complementary metrics have the potential to reveal detailed information regarding neural activity, connectivity, and physiological changes, enhancing our understanding of the mechanism underlying eTNS effects.

We studied the role of the LC-NE system on eTNS effects in healthy volunteers. However, it could be worthwhile to further investigate this method in population with LC degeneration. More specifically, the LC is implicated in several neurodegenerative diseases, such as Alzheimer’s and Parkinson’s disease (PD). Recent studies have reported neuronal loss of LC cells in PD patients [76,77] and that the dysregulation of the LC-NE system is involved in cognitive and motor impairment in these patients [78-80]. eTNS could be used to study LC degeneration in these populations, which might lead to valuable insides regarding the pathophysiology of PD and other neurodegenerative diseases. Furthermore, eTNS might also have therapeutic benefits for these populations and lead to a treatment for certain cognitive and motor symptoms.

In conclusion, 20 minutes of eTNS was able to induce significantly larger pupil dilation than MNS-control stimulation. In addition, the incidence of pupil dilations was higher after eTNS compared to the control conditions. These results suggest that eTNS modulates the release of NE leading to pupil dilation. Accordingly, the LC-NE system could be a key component in mediating eTNS effects. Further experiments using different stimulation parameters and proxies of LC activity, might provide more information on eTNS mechanisms of action.

Highlights:

• eTNS causes larger pupil dilation compared to MNS-control stimulation

• The number of pupil dilations is larger after eTNS compared to control stimulation

• Pupil response size is linked to baseline pupil diameter

• eTNS might activate noradrenergic pathways such as the LC-NE system