Non-Invasive Cervical Vagal Nerve Stimulation Alters Brain Activity During Traumatic Stress in Individuals with Posttraumatic Stress Disorder

Abstract

Objective:

Posttraumatic stress disorder (PTSD) is a disabling condition affecting a large segment of the population, however current treatment options have limitations. New interventions that target the neurobiological alterations underlying symptoms of PTSD could be highly beneficial. Transcutaneous cervical (neck) vagal nerve stimulation (tcVNS) has the potential to represent such an intervention. The goal of this study is to determine the effects of tcVNS on neural responses to reminders of traumatic stress in PTSD.

Methods:

Twenty-two participants were randomized to receive either sham (n = 11) or active (n = 11) tcVNS stimulation in conjunction with exposure to neutral and personalized traumatic stress scripts with High-Resolution Positron Emission Tomography scanning with radiolabeled water for brain blood flow measurements.

Results:

Compared to sham, tcVNS increased brain activations during trauma scripts (p < 0.005) within the bilateral frontal and temporal lobes, left hippocampus, posterior cingulate, and anterior cingulate (dorsal and pregenual), and right postcentral gyrus. Greater deactivations (p < 0.005) with tcVNS were observed within the bilateral frontal and parietal lobes and left thalamus. Compared to tcVNS, sham elicited greater activations (p < 0.005) in the bilateral frontal lobe, left precentral gyrus, precuneus, and thalamus, and right temporal and parietal lobes, hippocampus, insula, and posterior cingulate. Greater (p < 0.005) deactivations were observed with sham in the right temporal lobe, posterior cingulate, hippocampus, left anterior cingulate, and bilateral cerebellum.

Conclusion:

tcVNS increased anterior cingulate and hippocampus activation during trauma scripts, potentially indicating a reversal of neurobiological changes with PTSD consistent with improved autonomic control.

Trial Registration:

#NCT02992899

Introduction

Traumatic stressors such as early childhood trauma, combat exposure, or motor vehicle accidents, can lead to Posttraumatic Stress Disorder (PTSD), which affects ~8% of Americans (1). Symptoms of PTSD represent the behavioral manifestation of stress-induced changes in the brain including intrusive thoughts, avoidance behaviors, sleep disturbances, and hyperarousal. Previous studies also indicate PTSD may be associated with altered neural functional connectivity, elevated cardiovascular reactivity, and increased peripheral inflammatory biomarkers (2-6). Selective Serotonin Reuptake Inhibitor anti-depressants are the only approved medication treatment for PTSD, however patient response (60%) and full remission rate (20-30%) are poor (7, 8). A recent Institute of Medicine report did not find conclusive evidence supporting the efficacy of medication treatments for PTSD (9).

Neuromodulation represents a promising new treatment paradigm for stress-related psychiatric disorders (10, 11). In particular, vagal nerve stimulation (VNS) has shown the potential to mitigate exacerbated sympathetic nerve activity that may lead to intrusions and hyperarousal symptoms in PTSD (12, 13). Implantable VNS is approved for the treatment of depression and epilepsy and appears to partially mitigate peripheral cardiovascular and neural stress responses (14, 15), including during traumatic stress (16). Recently, studies have identified that transcutaneous stimulation of the auricular vagal afferents (taVNS) using surface electrodes can reliably produce vagus sensory evoked potentials (17, 18). Subsequent neuroimaging studies, employing taVNS and transcutaneous cervical vagal afferent stimulation (tcVNS), have replicated deactivations within the limbic brain areas commonly observed with invasive VNS (19, 20). Therefore, it appears that non-invasive VNS has promise as an effective alternative to invasive VNS without the side-effects of invasive surgery and high costs (20, 21).

Recently, we have observed that tcVNS attenuates neural activity in traumatized individuals without PTSD during exposure to personalized traumatic stress scripts within the insula, prefrontal cortex, orbitofrontal cortex, premotor cortex, hippocampus, and anterior cingulate (22). Additionally, tcVNS also elicited advantageous sympathetic downregulation during the traumatic scripts in peripherally measured physiological biomarkers (16). However, because PTSD disrupts many neurobiological systems and associated feedback loops (3, 5), it is unclear whether tcVNS is similarly effective in this population. For example, during personalized traumatic reminders, PTSD elicits hypoactivity in brain areas such as the anterior cingulate, hippocampus, and orbito- and medial pre-frontal cortex, that may remove inhibition and lead to a hyperactive amygdala (23). Because of these neurobiological changes and other mechanisms, PTSD patients experience exacerbated stress during traumatic scripts, as identified by increased endocrine response and aberrant neural activity compared to traumatized non-PTSD controls (24, 25). Therefore, a non-invasive device with the potential to decrease the stress reactivity in PTSD may alleviate the typified acute and chronic neurobiological changes present in this population.

The purpose of this study was to examine how tcVNS alters neural responses to personalized traumatic scripts in PTSD patients. We hypothesized tcVNS would significantly improve activation in brain areas altered by PTSD—the anterior cingulate, hippocampus, and medial/orbital frontal cortex—along with blocking stress-related activation within stress-related brain areas such as the insula.

Materials and Methods

Emory University (#IRB00091171), Georgia Institute of Technology (#H17126), SPAWAR Systems Center Pacific, and the Department of Navy Human Research Protection Program institutional review boards provided approval for this study which is posted on ClinicalTrials.gov (ClinicalTrials.gov # NCT02992899). It should be noted that, although the ClinicalTrials.gov study description included the neuroimaging methods presented in this study, brain activity findings were not listed as primary or secondary outcomes. Verbal / written informed consent was provided by all participants before enrollment.

Participants

Healthy individuals between the ages of 18 and 70 with a history of prior trauma were recruited. Supplementary Figure 1 (Supplemental Digital Content) presents the Consolidated Standards of Reporting Trials (CONSORT) for this study. Of the 127 individuals assessed for eligibility, 100 were excluded based on declining to participate (n = 37) or not meeting inclusion criteria (n = 63) including non-PTSD traumatized controls analyzed separately (22). The diagnosis of schizophrenia, schizoaffective disorder, bipolar disorder, bulimia, or anorexia, as defined by The Diagnostic and Statistical Manual of Mental Disorders (DSM-5) (26), excluded individuals from participating. Further exclusion criteria were current pregnancy, traumatic brain injury, meningitis, presence of an active implanted device, evidence or history of serious medical or neurological illness, or positive toxicology screen. Trained staff administered the Structured Interview for the Diagnostic and Statistical Manual of Mental Disorders (SCID) (27) for psychiatric diagnosis. The Clinician-Administered PTSD Scale for DSM-5 (28) was used to detect both the severity and presence of current and lifetime PTSD. The remaining 27 participants with PTSD were randomized into either sham or tcVNS groups using an online tool. Five participants were excluded from data analysis due to a temporary malfunction of the HR-PET scanner (n = 4; all participants in the current manuscript were enrolled after repair) or withdrawing from the study due to adverse reaction to trauma script (n = 1). The remaining 22 participants were included in the analysis, with n = 11 receiving tcVNS and sham (Table 1). Across all participants: one (4.5%) had a history of agoraphobia, one (4.5%) had a history of cocaine abuse, two (9%) had a history of alcohol abuse, twelve (54.5%) had a history of major depression (seven tcVNS, five sham), one (4.5%) had a history of obsessive-compulsive disorder, one (4.5%) had a history of opioid abuse, three (13.6%) had a history of panic disorder, one (4.5%) had a history of sedative-hypnotic-anxiolytic abuse, two (9%) had a history of social phobia, and one (5%) had a history of stimulant abuse.

Table 1: Participant Demographics for the active non-invasive vagal nerve stimulation (tcVNS) and sham groups.

PTSD = Posttraumatic stress disorder, CADSS = Clinician-Administered Dissociative States Scale, ETI-SR = Early Trauma Inventory-Self Report, ATI = Adulthood Trauma Inventory, SUDS = subjective units of distress, CAPS = Clinician-Administered PTSD Scale

Measure	tcVNS (n = 11)	Sham (n = 11)	p value
Age (y)	35.3 ± 12.7	37.6 ± 13.7	0.69
Sex	9 F, 2 M	6 F, 5 M	0.36
BMI (kg·m−2),	29.6 ± 7.3	28.6 ± 4.0	0.71
Race/Ethnicity			0.66
White/Caucasian	5 (45.4%)	7 (63.6%)
Black/African American	6 (54.5%)	3 (27.3%)
Asian	-	1 (9.1%)
Education			0.20
High School - Graduate	1 (9.1%)	1 (9.1%)
College – Not Complete	5 (45.5%)	1 (9.1%)
Associate Degree	2 (18.2%)	2 (18.2%)
Bachelor’s degree	3 (27.3%)	7 (63.6%)
Marital Status			0.92
Never Married	5 (50.0%)	5 (41.7%)
Married / Civil Partnership	2 (20.0%)	2 (16.7%)
Divorced / Separated	2 (20.0%)	4 (33.3%)
Widowed	1 (10.0%)	1 (8.3%)
PTSD Measures
PTSD Checklist Score	56.1 ± 15.3	57.1 ± 13.7	1.0
CAPS-5 Total Severity	16.3 ± 2.8	14.5 ± 4.3	0.55
Psychometric Measures
Beck Depression Inventory	28.7 ± 16.2	27.5 ± 11.5	0.84
CADSS	8.1 ± 9.0	8.3 ± 9.7	0.84
Hamilton Anxiety	22.3 ± 8.8	17.1 ± 7.6	0.16
ETI-SR	24.3 ± 10.9	17.9 ± 13.4	0.24
ATI	7.0 ± 1.9	6.8 ± 3.2	0.87
Anger
Total	127.4 ± 19.4	120.0 ± 11.0	0.31
State	22.0 ± 11.1	16.8 ± 2.4	0.21
Trait	22.3 ± 7.2	21.0 ± 4.0	0.62
SUDS (Baseline)	34.5 ± 29.4	33.9 ± 28.5	0.87

Supplementary Table 1 details the individual traumatic experiences of the study sample. The most common traumatic experiences were molestation, rape, and/or sexual abuse (41%). Among the other traumatic memories, the remaining were related to physical abuse by family or partner (32%), death of a family member or child (27%), emotional abuse (18%), military-related stress, violence, or death (23%), observed violence or a suicide attempt or threatened with violence (18%), car accident (4.5%), a victim of gun violence (4.5%), and death of a friend (4.5%).

Study Design

Data collection occurred between May 2017 and September 2019 at Emory University School of Medicine. The tcVNS manufacturer (ElectroCore, Basking Ridge, NJ) provided pre-numbered active and sham devices to a member of the research staff not involved in any other aspect of the protocol. Furthermore, research staff conducting enrollment and data collection were blinded to stimulus type. Research staff conducting analyses were unblinded to device type but were not involved in data collection.

At the initial screening, a trained member of the research staff conducted a psychiatric interview and facilitated a written traumatic history of the participant. From there, the written traumatic experiences were converted into a 60-second script which was recorded by a member of the research team (29). Participants also completed the PTSD Checklist (30), Beck Depression Inventory (31), Clinician-Administered Dissociative States Scale (32), Hamilton Anxiety Scale (33), Early Childhood Trauma-Self Report (34), Adulthood Trauma Inventory (35), and State Trait Anger Scale (36).

During the second visit, participants underwent a high-resolution positron emission tomography (HR-PET) scan session (Figure 1) consisting of auditory delivery of a series of both neutral and traumatic scripts. In total, there were fourteen HR-PET scans (neutral scans = 1-2, 7-8, 11-12; active scans = 3-4, 9-10, 13-14; tcVNS/sham only scans = 5-6), with a 90 min break in between scans 10 and 11 employed to assess the prolonged effects of tcVNS. The neutral scripts were descriptions of nature designed to induce neutral-to-positive affective responses. All scripts were delivered using headphones as the participant lay supine in the HR-PET scanner. Following traumatic scripts, stimulation (tcVNS or sham) was delivered to the left side of the neck during an automated two-minute interval. For scans five and six, the stimulation was delivered while the participant rested with eyes open. Following all scans, participants were asked to complete the Subjective Units of Distress Scale, rating their level of distress from 0 (no distress) to 100 (extreme distress).

Figure 1:

Protocol timeline for the current study depicting timing of script delivery and non-invasive cervical vagal nerve (tcVNS)/sham stimulation. All scripts were delivered via headphones with stimulation initiated immediately following. Total session duration was five hours including a prolonged (90 min) rest period.

Non-Invasive Transcutaneous Cervical Vagus Nerve Stimulation

All stimulation (tcVNS or sham) was administered with hand-held devices, identical in appearance and operation (GammaCore, ElectroCore, Basking Ridge, NJ), using collar electrodes applied to the left side of the neck for 120 seconds using previously described placement protocols (18). The tcVNS devices produced an AC voltage signal of five 5kHz sine pulses repeating at 25 Hz. The sham devices produced an AC biphasic voltage of 0.2 Hz square pulses which create a mild sensation similar to tcVNS. Sham stimulation is commonly used in tcVNS protocols (37); the sham signal in this study, being low frequency, is mainly attenuated at the skin-electrode interface due to the high impedance and therefore is unlikely to activate the afferent nerve fibers. However, the sham device does deliver relatively high voltage and current at the skin, and thus the nociceptors are activated, feeling like a pinch, and giving a similar sensation as active tcVNS. Stimulation intensity was adjustable from 0-5 arbitrary units with outputs of 0-30 V (~0-60 mA) and 0-14 V (~0-60 mA) for tcVNS and sham, respectively. Stimulation intensity was recorded as the maximum tolerable without pain following an acclimation period where the level was gradually increased. The stimulation intensity was 3.9 ± 1.1 (range: 2 – 5) and 4.9 ± 0.3 au (range: 3.5 – 5) for active tcVNS and sham, respectively. All participants reported feeling the stimulation for both active and sham.

Neuroimaging and Analysis

Regional cerebral blood flow was measured using HR-PET (CTI, Knoxville, TN) (38). Before the scans, participants were instructed to minimize all movement. Five seconds before the script (or stimulation period in scans five and six), a 20 mCi of radio-labeled water (H₂[O¹⁵]), produced in an onsite cyclotron, was intravenously administered. Following the radio-labeled water injection, the HR-PET scan was initiated along with the script recording to measure brain blood perfusion.

HR-PET image analysis was completed similar to previous research (39) within the statistical parametrical mapping (SPM12; www.fil.ion.ucl.ac.uk/spm) suite. Scans were pre-processed by spatially normalizing to a mean intensity image across the fourteen individual scans, transformed into a common anatomical space (SPM PET Template), smoothed using a three-dimensional Gaussian filter at 5-mm full width half maximum, and then normalized to whole-brain activity. First level (individual) models were computed using the neutral script and traumatic script conditions with the factor of scan pairs. Because the first trauma script occurs without prior stimulation, this scan was omitted from the analyses. The first level model was grand mean scaled, estimated, and contrasts computed for activation (trauma scripts – neutral scripts) and deactivation (neutral scripts – trauma scripts) for each trauma script block (n = 3). Second-level analysis (between-participant) was completed for both activation and deactivation in separate models using the first-level contrast images in a flexible factorial model with stimulation type (independent) and trauma script block (dependent) as factors. The second-level analysis also included the covariates of age (numeric) and sex (binary).

Statistical Analysis

Data normality was assessed using the Shapiro-Wilk test in R (v3.4.0; www.r-project.org), with comparisons between tcVNS and sham groups completed using a two-sample t-test or Mann-Whitney-Wilcoxon test for continuous and Fisher’s exact test for discrete variables, respectively. For assessments during the scripts, linear mixed-effects models were fit to the data (lme4; cran.r-project.org/web/packages/lme4) using between-participant fixed effects of stimulation type (sham, tcVNS) and within-participant fixed effects of script type (neutral, traumatic) and duration (before, after break), random effect of participant, and baseline value as a covariate. The before- and after-break time points were chosen as more finite time decompositions (i.e., trauma script blocks, scan number) produced singularity or lack of convergence model fit errors. Model selection was based on the lowest Akaike Information Criterion with residual and qq plots checked to ensure appropriate fit.

Regional brain blood flow changes (activation, deactivation) between device types were encoded similar to previous recommendations (40) resulting in t-statistic brain maps. For both activation and deactivation analyses, a threshold of p < 0.005 (uncorrected) and a minimum voxel size of eleven was employed to minimize Type I and Type II errors in neuroimaging research (41). Significant cluster peaks were identified using the distance from the anterior commissure with x, y, and z coordinates transformed from Montreal Neurological Institute (MNI) space to those of the Talairach stereotaxic atlas (42). Cluster peaks were identified using Brodmann Areas (BA) from the Talairach daemon (www.talairach.org). The a priori α level for non-brain imaging data was chosen at 0.05. All data are presented as mean ± SD.

Results

Demographic

Demographic variables and psychosomatic scales were not significantly different between participants receiving tcVNS or sham (Table 1). tcVNS and sham also demonstrated similar PTSD checklist and CAPS-5 scores which were indicative of current PTSD (Table 1).

Psychometric Measures During Traumatic Scripts

Across all participants, the experimental environment elicited an increased state of distress (t(21) = 5.7, p < 0.0001; range: 0 - 80); baseline distress was also a significant predictor of distress across stimulation and script type (p < 0.0001). Furthermore, subjective distress decreased following the break (by 13.5 ± 16.8, p = 0.001). Given the greater initial baseline, trauma scripts (mean difference: 1.7 ± 8.4, p = 0.36) and tcVNS (mean difference: −3.4 ± 26.8, p = 0.56) did not alter subjective distress compared to neutral scripts and sham, respectively (script by stimulation interaction p = 0.22). A similar pattern of results was observed for visual analog ratings (anxiousness, fear, nervousness, high, and anger) and CADSS (moving slow, unreal, separation).

Neuroimaging

Figure 2 and Supplementary Tables 2-3 present the overall patterns of activation and deactivation while listening to the traumatic scripts in individuals with PTSD. When the study sample was examined as a whole (collapsed across stimulation type), trauma scripts resulted in significant (p < 0.005) activity within the bilateral ventromedial prefrontal cortex including the subgenual anterior cingulate (BA 10-11, 25), right temporoparietal areas (BA 7, 39), superior frontal lobe and dorsal anterior cingulate (BA 6, 32), bilateral insula, bilateral post- and pre-central gyri, and bilateral parietal lobe (BA 39, 40). Across the whole sample of individuals with PTSD, trauma scripts elicited significant areas of hypoactivation (p < 0.005) within the bilateral parietal precuneus, bilateral superior frontal gyrus (BA 11), right temporal lobe (BA 39), bilateral occipital lobe (BA 18), bilateral parahippocampal gyrus, bilateral posterior cingulate (BA 29,30).

Figure 2:

Sagittal slices presenting significant (p < 0.005) activation (red) and deactivation (blue) while listening to personalized trauma scripts in participants receiving cervical non-invasive vagal nerve and sham stimulation with posttraumatic stress disorder. Talairach x coordinates below slices indicate the location with negative and positive coordinates located in the left and right hemispheres, respectively. Color bars indicate z-values of the cluster.

Figure 3 and Supplementary Table 4 present brain areas with greater activations and deactivations during the traumatic scripts when sham stimulation was applied compared to active tcVNS. Compared to active tcVNS, sham elicited greater (p < 0.005) activations within the left medial, middle, and inferior gyri (BA 6, 11, 47), left precentral gyrus (BA 4, 44), left precuneus (BA 7), right postcentral gyrus and inferior and superior parietal lobules (BA 3, 5, 7, 40), left thalamus, left lentiform nucleus, right inferior frontal lobe (BA 46), right insula (BA 13), right superior temporal gyrus (BA 22), and right posterior cingulate (BA 31). Compared to active tcVNS, sham elicited greater (p < 0.005) deactivations within the right inferior (BA 20) and superior (BA 39) temporal lobe, left middle occipital gyrus (BA 19), right parahippocampal gyrus (BA 36) and hippocampus, right posterior cingulate (BA 29, 30), left middle temporal gyrus (BA 21), and left anterior cingulate (BA 24).

Figure 3:

Sagittal slices presenting greater (p < 0.005) activation (red) and deactivation (blue) while listening to personalized trauma scripts in participants with posttraumatic stress disorder receiving sham cervical non-invasive vagal nerve stimulation compared to active. Talairach x coordinates above slices indicate the location with negative and positive coordinates located in the left and right hemispheres, respectively. Color bars indicate z-values of the cluster.

Figure 4 and Supplementary Table 5 present brain areas with greater activations and deactivations during the traumatic scripts when active tcVNS was applied compared to sham. Compared to sham tcVNS, active elicited greater (p < 0.005) activations within the left hippocampus and parahippocampal gyrus (BA 36), right parahippocampal gyrus (BA 35), left inferior, medial, middle, and superior temporal gyri (BA 10, 20, 21, 38), right middle and superior temporal gyri (BA 22, 38, 39), right postcentral gyrus (BA 2), left medial frontal gyrus (BA 10), left fusiform gyrus (BA 18), left occipital lobe (BA 18, 19), left posterior cingulate (BA 23, 29-31), left anterior dorsal anterior cingulate (BA 24, 33), and left pregenual anterior cingulate (BA 24), and left cerebellum. Compared to sham tcVNS, active elicited greater (p < 0.005) deactivations within the right inferior, medial, and middle frontal lobe (BA 6, 8-10, 45), left medial, middle, and superior frontal lobe (BA 6, 8, 10), bilateral pre- and post-central gyri (BA 1-4, 6, 7, 40), bilateral inferior and superior parietal lobules (BA 7, 40), bilateral cerebellum, left occipital lobe (BA 19), left thalamus, left caudate, and left lentiform nucleus.

Figure 4:

Sagittal slices presenting greater (p < 0.0025) activation (red) and deactivation (blue) while listening to personalized trauma scripts in participants with posttraumatic stress disorder receiving active cervical non-invasive vagal nerve stimulation compared to sham. Talairach x coordinates above slices indicate the location with negative and positive coordinates located in the left and right hemispheres, respectively. Color bars indicate z-values of the cluster.

Figure 5 and Supplementary Table 6 present brain areas with greater activity during only stimulation. Compared to active tcVNS, sham stimulation had greater (p < 0.005) activity during the application of the device within the right inferior and superior frontal gyrus (BA 9, 45), right pre- and post-central gyrus (BA 2, 6), right inferior lobule (BA 40), right temporal lobe (BA 38), left middle and superior frontal gyrus (BA 6, 8, 9), and right insula (BA 13). Compared to sham stimulation, active tcVNS had greater (p < 0.005) activity within the bilateral fusiform gyrus (BA 20, 37), right posterior cingulate (BA 31), right middle temporal gyrus (BA 21, 22), right precuneus (BA 7, 39), left superior temporal gyrus (BA 38), left uncus, and bilateral occipital lobe (BA 18, 19).

Figure 5:

Sagittal slices presenting greater (p < 0.005) activity during either sham (red) and active transcutaneous vagal nerve stimulation (tcVNS; blue) during periods of only stimulation in participants with posttraumatic stress disorder. Talairach x coordinates above slices indicate the location with negative and positive coordinates located in the left and right hemispheres, respectively. Color bars indicate z-values of the cluster.

Discussion

Application of tcVNS in PTSD patients exposed to traumatic scripts resulted in increased brain activity within the anterior cingulate, including the subgenual (sgACC), anterior dorsal (adACC), and posterior dorsal (pdACC) subdivisions. Prior studies have observed a relative decrease in sgACC activity, located primarily within left BA 24, in PTSD patients exposed to traumatic reminders, including traumatic scripts (23, 43, 44), and cognitive challenges such as oddball paradigms (45, 46). Downregulated activity of the sgACC in PTSD is thought to reflect deficient emotional regulation associated with traumatic remembrance (23) possibly via reduced inhibition of the amygdala by the anterior cingulate/medial prefrontal cortex as these areas have shown an inverse relationship during an emotional conflict task (47). Furthermore, greater sgACC activity was also associated with decreased arousal levels as measured by skin conductance (47). Therefore, increased sgACC activity would facilitate inhibition of the amygdala leading to improved emotional regulation. Since skin conductance is driven by the sympathetic nervous system, increased activity of which underlies hyperarousal symptoms in PTSD (48), the findings in the current study indicate that tcVNS may have the capacity to improve hyperarousal symptoms through a well-defined neural mechanism (23) associated with PTSD during a simulated intrusive memory which may have direct clinical application.

The adACC plays a central role in the integration of cognitive and emotional neural processes. Located within BA 24 and 32, the adACC is part of the dorsal cognitive division, which is a distributed attentional network with reciprocal connections with the lateral prefrontal cortex, parietal cortex, and motor areas (premotor cortex, supplementary motor area) (49, 50). Studies show a relative decrease, in function in this area during exposure to traumatic scripts in PTSD (23, 44). In the current study, there were similar results with traumatic script exposure paired with sham stimulation when compared to tcVNS. The adACC has three general functions: monitor, controller, and economic, which, according to a recently published integrative theory (51), allows it to act as a ‘storage buffer’ following the processing of an initial stimulus before guiding the individual to appropriate action. This processing appears advantageous in cognitive processes leading to fear extinction and/or emotional reappraisal (52). Therefore, greater activity within the adACC may represent a similar mechanism to the sgACC—and indeed areas within the anterior cingulate are functionally connected (52)—to improve emotional processing and regulation. PTSD also decreased activity during executive cognitive processing within the adACC, and this relationship was associated with PTSD severity (53). These results additionally present a direct clinical application for tcVNS and suggest the breadth of benefits may span across multiple aspects of emotional regulation and cognitive function.

The second main finding of this study is that tcVNS increased brain activity within the left hippocampus during personalized traumatic stress. The hippocampus is involved in processing explicit memory and fear extinction and is seen during context-specific traumatic reminders (54). In PTSD, a failure of hippocampal activation is commonly observed during trauma reminders (44), potentially as a result of PTSD-mediated crenation (55) leading to declarative memory deficits (56). Additionally, PTSD disrupts a memory network consisting of the hippocampus, anterior cingulate, and orbitofrontal cortex, that is employed in the recall of emotionally valenced words (57).

PTSD also appears to disrupt the resting functional connectivity of the hippocampus. In previous traumatized control participants, the anterior and posterior hippocampus differed in their connection to areas within the default mode and salience network (58). However, this separation was not observed with PTSD, in addition to exhibiting a less inhibitory effect of the hippocampus on the precuneus (58). The left precuneus was more active in the sham condition in the current study, reinforcing this inefficient inhibition and potentially indicating a lessened ability to inhibit the default mode network at-large which is known in PTSD (59).

The hippocampus is also sensitive to stress. For example, injecting yohimbine, an ɑ-2 antagonist which stimulates norepinephrine release, decreased hippocampus metabolism in PTSD patients (60), suggesting a high sensitivity to norepinephrine. One potential mechanism of action for tcVNS is that stimulation will decrease downstream norepinephrine release from the locus coeruleus during traumatic stress (61), which would decrease noradrenergic innervation of the hippocampus and therefore increase metabolic activity. While circulating norepinephrine levels are not available in the current study, previous studies in both the periphery (16) and brain (22, 62) indicate tcVNS can decrease reactivity to both somatosensory and traumatic stressors. Furthermore, whether this potential benefit of tcVNS is limited to acute effects or has the potential to modulate long-term neurobiological changes—such as increasing hippocampal volume through neural regeneration—requires investigation.

The third major finding of this study was that tcVNS decreased brain reactivity to traumatic stress in a similar manner to our previous non-PTSD findings (22). In non-PTSD individuals experiencing previous trauma, tcVNS decreased activity within similar areas as observed in the current study such as the left frontal lobe (middle, inferior, and superior gyri), left precentral gyrus, left thalamus, and insula (although discordant hemispheres) (22). Many of these areas coincide with the conjunction analysis for cohort-specific brain activations (Figure 1) and are commonly activated in response to traumatic stress in both healthy and PTSD patients (43, 63). These findings provide evidence that tcVNS may decrease general emotional stress reactivity in PTSD and provide support for use as a treatment modality. Interestingly, nodes of the default mode network, such as the posterior cingulate and precuneus, were more active with sham stimulation. Therefore, an alternative interpretation of greater activity with sham during trauma scripts is a greater efficacy of tcVNS to silence resting-state networks during cognitive processing.

In addition to decreasing general emotional stress reactivity to trauma scripts, tcVNS may block activation to areas important in regulating the peripheral stress response such as the insula (64). In the current cohort, tcVNS blocked activation within the mid-to-posterior insula, which is the subdivision that processes somato-visceral information from the body and translates to an emotional state (65). Furthermore, there is emerging evidence that the right insula processes sympathetic effects (66), with greater activity observed with more physiological stress (67). These data support earlier findings (22) in a non-PTSD sample and reinforces the potential efficacy for tcVNS to decrease both central and peripheral arousal during traumatic stress.

While the current study presents novel findings of neural activity responses during traumatic stress in PTSD with tcVNS, the study is not without limitations. While multiple studies have observed evidence of tcVNS stimulating the vagal nerve (18, 68), similar data are not available for the current manuscript. However, it should be noted that the presence of sham or active tcVNS could be predicted with an accuracy of 96% using peripheral biomarkers (69), providing evidence of a distinct change in physiology with stimulation of the vagus nerve. Secondly, it should be noted that changes in brain activity were observed without concomitant changes in perceptual measures of distress. This was likely an artifact of participants reporting elevated baseline stress, potentially in response to the upcoming traumatic reminder scripts, which limits the degree into which changes values be calculated. This limits interpretability of the study results, as a direct causal relationship between perceptual distress and changes in neural activity cannot be established. Further studies are required to fully examine the relationship between perceptual responses during traumatic stress, tcVNS, and brain activity in PTSD. Thirdly, while the application of tcVNS to a clinical population is novel, this is a cross-sectional study and therefore longitudinal changes cannot be determined. This is particularly salient in PTSD with known neural changes, possibly related to altered neurotransmitter sensitivity/levels (48), and future studies are required to determine the long-term efficacy of tcVNS. This study would be improved by probing whether participants thought they received either sham or active tcVNS.

In conclusion, this study has provided novel insight into the potential benefits of tcVNS applied following personalized traumatic reminders in PTSD. Stress-related disorders such as PTSD are important public health problems, affecting ~8% of Americans (1). Furthermore, approved pharmacological treatments of PTSD suffer from poor response, elevated remission rates, and insufficient evidence supporting their efficacy (9). The current study uniquely addressed whether a novel, non-invasive neuromodulation device, tcVNS, altered neural activity during personalized traumatic scripts. Personalized trauma scripts allow for an individualized approach to inducing hyperarousal symptoms and provides insight into potential efficacy in real-world scenarios where tcVNS may have an application.

Specific to PTSD, the application of tcVNS following traumatic scripts increased activity in brain areas known to be hypoactive, the anterior cingulate and hippocampus. Reversal of this hypoactivity likely indicates tcVNS-mediated improvements in emotional processing, memory retrieval, and a decreased fear response, providing neural evidence for the clinical efficacy of tcVNS. Like non-PTSD but previously traumatized individuals, tcVNS also was able to decrease trauma script-related brain activity and therefore may be indicative of lessened arousal. These findings also indicate differences between PTSD and non-PTSD, supporting the need for a targeted investigation into PTSD populations. Specifically, whether the potential beneficial effects of tcVNS on the anterior cingulate and hippocampal circuitry persist during prolonged usage merits further investigation.

Abbreviations:

PTSD

Posttraumatic Stress Disorder

HR-PET

High-Resolution Position Emission Tomography

VNS

vagal nerve stimulation

taVNS

transcutaneous auricular vagal nerve stimulation

tcVNS

transcutaneous cervical vagal nerve stimulation

DSM

Diagnostic and Statistical Manual of Mental Disorders

SCID

Structural Clinical Interview for DSM

SPM

Statistical Parametric Mapping