Noninvasive vagal nerve stimulation reduces withdrawal-related affective distress and improves functional outcomes in alcohol use disorder – A feasibility and acceptability pilot study

Abstract

Objectives:

Alcohol use disorder (AUD) is characterized by autonomic dysregulation and overactive brain stress systems, which manifests as pronounced negative emotional states in the absence of alcohol and can lead to functional impairment. The vagus nerve presents a promising novel treatment target for these symptoms. The aim of this pilot study was to evaluate feasibility and acceptability of cervical noninvasive vagal nerve stimulation (nVNS) as a new treatment for AUD, and to evaluate if nVNS can alleviate withdrawal-related affective distress and improve functional outcomes.

Materials and methods:

In this double-blind randomized controlled trial, male Veterans with AUD were randomly assigned to receive active (N=9) or sham (N=10) nVNS at baseline, self-administered stimulation for 7 days, and returned for a follow-up visit. Treatment adherence and acceptability ratings were assessed. Behavioral self-report measures of affective distress (PHQ-8, STAI-S, BAI) and functional limitations (DrInC-2R, BIPF) were collected at baseline and follow-up. Linear mixed effects (LME) models were used to assess the impact of treatment group on outcome measures.

Results:

Treatment adherence was high in both groups and did not statistically differ (94% nVNS group, 80% sham group; p= 0.14). Both groups assigned treatment acceptability ratings in the acceptable to highly acceptable range. LME results show significant group by time interactions for the PHQ-8 (p = .04), DrInC-2R (p = .02), and BIPF (p = .01)), indicating greater reductions in depressive symptoms, adverse alcohol-related consequences, and functional limitations following nVNS compared to sham treatment. No statistically significant group by time interactions were observed for anxiety symptoms.

Conclusions:

This is the first study to explore cervical nVNS as a potential new treatment for AUD-related symptoms. Findings provide support for feasibility of treatment delivery and patient acceptance, and initial evidence that nVNS can achieve significant reductions in depressive symptoms as well as adverse alcohol-related consequences and psychosocial functional limitations in AUD.

BACKGROUND

Alcohol use disorder (AUD) is widespread and causes significant functional disability and reduced quality of life ^1,2. Military Veterans present a more clinically complex population that is particularly vulnerable to AUD with a lifetime prevalence of 42.2% ³, which is considerably higher than in civilians who have a lifetime prevalence of approximately 30% ⁴. Despite the immense cost to Veterans and society at large ^1,3, over half of Veterans who complete treatment, will relapse within 6 months, signaling a great need for improved treatments or different treatment targets ⁵.

Long-term excessive alcohol use results in homeostatic dysregulation, or disturbance of the body's natural equilibrium, due to changes in the central and autonomic nervous system ^6-8. In the absence of alcohol, dysregulation of the autonomic nervous system manifests as physiological hyperarousal ⁹, while neuroadaptations in the brain lead to overactive brain stress systems and heightened emotional distress ^7,8,10,11. Alcohol use disorder (AUD) is therefore characterized by pronounced negative affective states during abstinence, such as irritability, anxiety, dysphoria, anhedonia, and negative mood (i.e., hyperkatifeia ^7,12,13). These symptoms, recognized as part of the alcohol withdrawal process, can persist well beyond acute withdrawal periods, and motivate continued harmful drinking and relapse to relieve these symptoms via mechanisms of negative reinforcement ^7,14. Moreover, affective distress in AUD is associated with significant disability and functional impairment ^15-18, which is reflected in a recent shift in the field to focus on functional outcomes rather than abstinence alone when evaluating treatment response and recovery from AUD ^16,19,20.

Current AUD treatments include pharmacotherapy and psychotherapy, however, high rates of drop-out and relapse ^21-23 and low treatment engagement ²⁴ signal a need for alternative treatment options. Recent research provides promising initial evidence that neuromodulation methods, such as noninvasive vagal nerve stimulation (nVNS), may offer innovative treatment approaches to substance use disorders by focusing on novel treatment targets within the addiction circuitry ^25-28. Both autonomic dysregulation and brain regions involved in the perception and regulation of emotional and physical distress can be directly targeted with nVNS ^29-32. As the main nerve of the parasympathetic division of the autonomic nervous system, the vagus nerve regulates the autonomic response to distress to restore homeostasis ^33,34. Since AUD is characterized by sympathetic overdrive and reduced parasympathetic activity, indicating a heightened state of alertness and stress ^9,35,36, the vagus nerve is a promising new treatment target.

Surgically implanted VNS was approved by the Food and Drug Administration (FDA) for epilepsy in 1997. Although the exact mechanism by which VNS exerts its antiepileptic effects is still unclear³⁷, VNS has been shown to increase gamma-aminobutyric acid (GABA) and decrease glutamate neurotransmission in these patients ³⁸. If heavy drinkers reduce or discontinue alcohol consumption, they are at an increased risk for withdrawal-related seizures due to a decrease in GABAergic and an increase in glutamatergic neurotransmission. Since anticonvulsant medications that target the stabilization of these neurotransmitters have established therapeutic benefits in patients with AUD ³⁹, nVNS may have similar effects in a population at risk for seizures when they reduce or stop alcohol use.

While studies investigating the use of nVNS for AUD-related symptoms are still limited ^40,41, evidence supporting the potential therapeutic benefit of vagus nerve stimulation as well as targeting the vagus nerve to treat addiction is growing ^28,42-46. Initial support comes from preclinical work, wherein nVNS has been shown to reduce drug induced anxiety and drug relapse ^44,45. Specific to alcohol, vagotomy, or the resection of the vagus nerve, eliminated relapse-like alcohol behavior in alcohol-dependent rats during abstinence ⁴⁴. This work suggests that absence or modification of physiological feedback from the body to the brain during alcohol withdrawal may decrease alcohol craving and consequently relapse behavior ⁴⁴. Clinical studies additionally demonstrate that cervical nVNS reduces withdrawal-related symptoms associated with opioid use, including anxiety, subjective distress, and pain, which are comparable to withdrawal symptoms associated with chronic, heavy alcohol use ^28,47. In individuals undergoing inpatient medically-assisted detoxification for acute alcohol-withdrawal syndrome, auricular nVNS treatment was associated with increased parasympathetic tone (as measured with pupillometry), and reduced subjective craving, suggesting restored autonomic balance in these individuals ⁴¹. Furthermore, auricular nVNS treatment was also associated with an improvement in depression, but not a change in anxiety symptoms, in men with protracted withdrawal symptoms following one week of medication assisted abstinence. This work provides compelling evidence for the potential effectiveness of nVNS as a treatment for AUD-related affective distress ⁴⁰. Despite this promising early work, there is a clear need for more research to adequately evaluate the potential therapeutic effects of nVNS for withdrawal-related symptoms in AUD.

Finally, nVNS shows significant promise for improving functional disability related to AUD. A recent clinical report on 233 patients with multimorbidity disorders (i.e., diagnosed with two or more conditions, including headache, anxiety, depression, chronic pain, or insomnia), who were prescribed nVNS (self-administration 3x/day for up to one year), showed improved health-related quality of life, including symptoms such as anxiety, depression, and pain ⁴⁸. In addition, other studies reported improvements in quality of life measures following two weeks of nVNS treatment in healthy older adults ⁴⁹ as well as improvements in psychosocial functioning and quality of life in patients with surgically implanted VNS for treatment-resistant depression ^50,51. More research is needed to explore whether these results extend to an AUD population, and via what mechanisms.

While there is a strong conceptual argument for and promising evidence that nVNS targets negative emotional states following discontinuation of chronic, heavy alcohol use, there are currently no published studies in the complex population of Veterans, no studies evaluating cervical nVNS for AUD, and no studies exploring functional outcome improvements. The present research is therefore the first study to examine cervical nVNS as a potential novel treatment option for AUD-related symptoms and is the first in a Veteran population. As such, the aim of this pilot study was to evaluate feasibility of treatment delivery and patient acceptance, and to evaluate if cervical nVNS can alleviate withdrawal-related affective distress (i.e., symptoms of anxiety and depression) and improve functional outcomes in individuals with AUD following 7 days of nVNS treatment.

MATERIALS AND METHODS

Subjects

This double-blind randomized controlled trial (NCT05226130) included 19 male Veterans with a current DSM-5 diagnosis of AUD (average age 39 years, ± 8 years). Participants underwent a detailed phone screening prior to enrollment as well as a structured clinical interview following enrollment to ensure they met study eligibility criteria (i.e., AUD diagnosis with at least one functional disability due to alcohol use, current alcohol craving, and current heavy drinking (>4 drinks on any day or >14 drinks per week). The inclusion criterion of functional disability due to alcohol use was determined with three questions based on DSM-5 criteria for AUD. These questions, which were administered during the phone screening prior to participation, assessed whether drinking or recovering from its effects caused potential participants to spend less time meeting responsibilities at work, school, or at home, whether it caused them to reduce or give up social or recreational activities, or whether it caused problems with family or other people. A total of 123 subjects were screened and excluded based on the phone screening. Exclusion criteria were currently or recently (within the last 90 days) enrolled in an abstinence-based treatment program, evidence of a maladaptive pattern of substance use or abuse other than alcohol one month prior to the screening, and severe mental illness. In addition, participants were screened for MRI-related (e.g., cardiac pacemaker, metal fragments in eyes/skin/body, aortic/aneurysm clips, heart-valve replacement, copper intrauterine device, shunt (ventricular or spinal), or neuro/bio-stimulators) and nVNS-related (e.g., history of carotid endarterectomy, severe carotid artery disease, congestive heart failure, cardiac arrhythmia, known severe coronary artery disease, history of seizure or syncope (within past year), or prior neck surgery) safety exclusion criteria.

The study was approved by the VA San Diego Healthcare System (VASDHS) Institutional Review Board and the Research and Development Committee, and all subjects provided written informed consent prior to participating in study activities.

Cervical nVNS stimulation

Subjects were randomly assigned to receive active (N=9) or sham (N=10) nVNS during their baseline visit, were instructed to self-administer the device at home bilaterally twice a day (morning and night) for 7 days, and returned for a follow-up visit. In addition, subjects were instructed to refrain from drinking alcohol for 24 hours prior to each study visit as symptoms of affective distress due to the cessation of heavy drinking typically occur during this time frame ⁵². One subject consumed alcohol the night before the post-treatment follow-up visit and had an abstinence window of approximately 15 hours. Subjects were allowed to drink as much and as often as desired between the baseline and follow-up visits. Alcohol consumption during the 90 days preceding the baseline interview and during the treatment window was measured with the Timeline follow-back (TLFB) (105).

Stimulation was delivered cervically through the skin with gammaCore nVNS devices (electroCore; Fig 1). During the baseline visit, subjects were trained in localizing the vagus nerve and self-administering the nVNS device during the 7-day treatment period. Active and sham devices were identical in appearance, and sham devices produced slight tingling sensations on the skin similar to active devices. Subjects in both groups were given the same instructions for use of the nVNS/sham devices. To ensure both examiner and subjects remained blinded to the treatment condition, wording was intentionally vague and stated that some individuals experience tingling sensations or a downward pull on the lip while others do not, and that either is normal.

Figure 1. Handheld nVNS device used in the study.

Subjects self-administered the device to stimulate the vagus nerve through the skin via the neck (transcutaneous cervical stimulation). Sham devices look identical to active devices. Picture reproduced with permission from electroCore.

Active devices produced an alternating current (AC) low-voltage electrical signal at 5 kHz sinusoid wave series occurring for 1 ms with a repetition rate of 25 Hz for a duration of 2 minutes (30 seconds of ramp-up followed by 90 seconds of actual stimulation). Sham devices produced a slowly varying direct current (DC) signal of 0.1Hz. Stimulation intensity ranging from 0 to 40 in arbitrary units (corresponding to 0 to 24V) was adjustable by the subject. Subjects were instructed to self-administer stimulation at maximum intensity (i.e., 24V; 14V for sham devices), but were allowed to adjust the intensity to a tolerable level. All subjects in the active treatment condition, including those who adjusted the stimulation intensity, stimulated at intensity levels shown to stimulate the vagus nerve ⁵³. Over the 7-day treatment period, subjects stimulated at an average stimulation strength of 38.45 a.u./~ 22.8V (sd = 2.16/~ 1.2V), ranging from 29-40 a.u. (corresponding to ~17.4V – 24V). In the sham group, the average stimulation strength was 39.9 a.u. (sd = 0.27), corresponding to approximately 14V.

Measures

Feasibility and acceptability

Treatment adherence and adverse events were tracked via a daily treatment completion log. Treatment adherence was calculated by dividing the total number of times subjects were instructed to self-administer the stimulation (2x/day for 7 days = 14 times) by the number of times subjects reported self-administration. Treatment acceptability was measured with item 1 from the Treatment Acceptability Questionnaire (TAQ; “Overall, how acceptable do you find the proposed treatment to be”, which ranges from “1/ very unacceptable” to “7/very acceptable” ⁵⁴).

Behavioral measures

Behavioral self-report measures were collected at baseline (prior to nVNS/sham stimulation) and at follow-up (post 7 days of nVNS/sham self-administration). Affective distress was measured with the Patient Health Questionnaire (PHQ-8 ^55,56), Beck Anxiety Inventory (BAI ⁵⁷), and the State-Trait Anxiety Inventory-State (STAI-S ⁵⁸). One subject missed one item on the PHQ-8, and one subject missed one item on the BAI. Values for these items were replaced with the mean of the completed values for the remaining items, which has been shown to be a reliable method of handling missing questionnaire items ^59,60.

Functional outcomes were measured with the Drinker Inventory of Consequences (DrInC-2R ⁶¹) and the Brief Inventory of Psychosocial Functioning (BIPF ⁶²). The DrInC-2R is a measure of adverse alcohol-related consequences that are typically negatively affected by hazardous drinking or recovering from its effects (i.e., physical, intrapersonal, social responsibility, interpersonal functioning, and impulse control). The BIPF assesses functional limitations in interpersonal relationships, family relationships, work, parenting, education, and self-care. Although this questionnaire was originally developed to assess functional impairment related to posttraumatic stress disorder (PTSD) symptoms, questionnaire items are not specified as PTSD-related.

At the baseline visit, questionnaires covered symptoms experienced during the past week, with the exception of the DrInc-2R (past 90 days) and the STAI-S (rating symptoms “right now”). At the follow-up visit, subjects were instructed to complete questionnaires with regards to symptoms experienced during the past 7 days and post treatment.

Alcohol consumption

Alcohol consumption was measured at baseline (past 90 days) and at the follow-up visit (past 7 days) using the timeline follow-back (TLFB) method ⁶³. The TLFB is a standard tool used to document the number and types of alcoholic drinks consumed per day during a specified time frame. For the present study, percent days drinking and percent days heavy drinking (i.e., 5 or more drinks on any day) was calculated for each interval.

Statistical analysis

Baseline differences in demographic and behavioral measures were assessed with the Welch two sample t-test (or Wilcoxon test for skewed variables) for continuous variables and a Pearson Chi-Squared or Fisher’s Exact Test for categorical variables. For variables that showed a skewed distribution, a Wilcoxon test was used to account for non-normality. Linear mixed effects (LME) models were used to assess the impact of treatment group on behavioral measures, with treatment group (sham = 0, nVNS = 1) and time (baseline = 0, 7-day follow-up visit = 1) modeled as fixed effects as independent variables, and subject modeled as a random effect.

The treatment group by time interaction was examined to determine whether nVNS had a differential impact on behavioral measures relative to sham. Outcome variables were examined for potential skewness. For variables with significant skew (e.g., some of the DrInC-2R subscales and TLFB heavy drinking %), the same predictive analyses were run with a generalized LME modeling these outcomes as an inverse gaussian or a gamma distribution (as recommended for right-skewed data; see Lo & Andrews, 2015⁶⁴. Results and statistical significance were either unchanged or enhanced, highlighting the robustness of the LME analyses. For interpretability purposes, we present model coefficients and effect sizes based on LME analyses. Post-hoc contrast analyses correcting for multiple comparisons (Tukey method) were conducted to explore group differences at baseline and follow-up, as well as pre- to post-treatment changes within each treatment group. Analyses were conducted using R statistical software ⁶⁵.

RESULTS

Sample Demographics

Groups did not differ significantly in baseline demographic variables, including years of education, race, and age (ps>=.21; Table 1). In addition, both groups had a comparable level of problematic drinking, in the moderate-severe range, as measured with the Alcohol Use Disorders Identification Test (AUDIT⁶⁶) and a clinical interview. In the active group, two subjects met for a clinical diagnosis of moderate AUD per DSM-5 criteria and 7 for severe AUD. In the sham group, all 10 subjects met for a clinical diagnosis of severe AUD. There was no statistically significant group difference in baseline percentage of days during which subjects consumed alcohol or in heavy drinking days. A moderate percentage of subjects met diagnostic criteria for co-occurring depression and anxiety disorders, and approximately half of the sample met for posttraumatic stress disorder, which is common in Veteran populations⁶⁷. Groups did not differ significantly in number of subjects who met for these psychiatric disorders (p=1; Table 1). Groups did not differ in baseline severity of depression (PHQ-8), anxiety (STAI-S, BAI), or functional limitations (BIPF, DrInC-2R) (ps>=.23; Table 2).

Table 1.

Baseline demographics and alcohol use of nVNS/sham groups

Mean (SD)	nVNS (N=9)	Sham (N=10)	Diff. (p)
Age yrs mean (SD), range	37.4 (8.6), 25-49	39.6 (7.2), 28-50	0.56
Yrs of education	15.3 (2.6)	13.8 (2.4)	0.21
Race (N/%)			0.30
African American	2 (66.7%)	1 (33.3%)
Caucasian	6 (60%)	4 (40%)
Asian	1 (25%)	3 (75%)
Mixed race	0 (0%)	2 (100%)
AUDIT Total	21.2 (6.7)	20.8 (5.7)	0.88
TLFB % days drinking	59.9 (20.2)	72.8 (22.2)	0.20
TLFB % days heavy drinking	40.1 (28.0)	51.7 (26.9)	0.37
Major depression (N/%)	4 (44%)	4 (40%)	1
Panic disorder (N/%)	2 (22%)	3 (30%)	1
Social anxiety (N/%)	2 (22%)	2 (20%)	1
GAD (N/%)	3 (33%)	3 (30%)	1
PTSD (N/%)	5 (56%)	5 (50%)	1

Note. Demographics at baseline. AUDIT = Alcohol Use Disorders Identification Test, TLFB = Timeline Follow-Back, GAD = generalized anxiety disorder. TLFB data timeframe: past 90 days. SD = standard deviation. Diff. = differences in mean baseline scores between nVNS and sham group based on Welch two sample t-test for all other variables. Differences for race were calculated based on a Pearson Chi-squared test and for psychiatric disorders based on a Fisher’s Exact test.

Table 2.

Baseline behavioral scores of nVNS/sham groups

Mean (SD)	nVNS (N=9)	Sham (N=10)	Diff. (p)
PHQ-8 Total	13.2 (5.9)	12.5 (7.2)	0.81
STAI-S Total	48.6 (16.9)	40.1 (14.2)	0.23
BAI Total	16.8 (9.2)	17.0 (11.6)	0.96
DrInC-2R Total	47.9 (18.0)	44.4 (28.1)	0.75
BIPF Total	56.9 (26.5)	47.7 (32.9)	0.51

Note. Mean symptom scores at baseline. PHQ-8 = Patient Health Questionnaire, STAI-S = State-Trait Anxiety Inventory-State, BAI = Beck Anxiety Inventory, DrInC-2R = Drinker Inventory of Consequences, BIPF = Brief Inventory of Psychosocial Functioning, SD = standard deviation. Diff. = differences in baseline scores between nVNS and sham group based on Welch two sample t-test.

Treatment adherence, acceptability ratings, and adverse events

Treatment adherence in the study sample was high in both groups and did not statistically differ (i.e., 94% in the nVNS group and 80% in the sham group; p= 0.14, Table 3). Two subjects (both in the sham group) were not treatment compliant as defined a priori by >75% treatment adherence. One of these subjects additionally forgot to stimulate on both sides of the neck for 5 (out of 14) administrations; these were included in the total count as 0.5 administrations. Follow-up behavioral data and treatment acceptability ratings for these two subjects were consequently excluded from the analyses (although both subjects gave ratings of 7 on the TAQ).

Table 3.

Treatment adherence and acceptability ratings of nVNS/sham groups

Mean (SD)	nVNS (N=9)	Sham (N=10)	Diff. (p)
Treatment adherence	94.4 (7.8)	79.76 (26.4)	0.19
Treatment acceptability	5.7 (1.0)	6.4 (0.8)	0.11

Note. Mean treatment adherence and acceptability ratings. SD = standard deviation. Diff. = differences in baseline scores between nVNS and sham group based on a Welch two sample t-test for treatment acceptability and on a Wilcoxon for treatment adherence (as this variable was significantly skewed).

On the treatment acceptability questionnaire, the nVNS group assigned an average score of 5.7 (SD=1.0) relative to 6.4 (SD=0.8) in the sham group; Table 3). A rating above the midpoint of the TAQ (i.e., between 5 and 7) is the established criterion for “acceptable to highly acceptable” across previous studies ^68,69.

No serious adverse events were reported. Other adverse outcomes were temporary and consistent with widely reported expected outcomes such as lip pulling, tingling sensations, muscle twitching, or headache.

Treatment effects on affective measures

A significant group by time interaction was observed for PHQ-8 scores (F(1, 16.53) = 4.76, p = .04; Table 4), with post-hoc analyses indicating greater reductions in depressive symptoms following nVNS (Cohen d = 1.40; t = 2.78, p = .05) compared to sham (Cohen d = −0.14; t = −0.25, p = .99; Fig 2). Following treatment, symptoms in the nVNS group decreased from moderate to mild levels, while in the sham group symptoms stayed in the moderate range (Table 5).

Table 4.

Results from linear mixed effects (LME) models of functional outcomes and affective distress

	Estimate	Std. Error	t value (df)	p value
PHQ8 Total
Time	0.40	1.466	0.27 (16.9)	0.79
Group	0.72	2.940	0.25 (22.3)	0.81
Time x group	−4.29	1.966	−2.18 (16.5)	0.04 *
STAI-S Total
Time	0.37	2.816	0.13 (11.7)	0.90
Group	6.25	6.931	0.90 (20.0)	0.38
Time x group	−3.57	3.550	−1.01 (11.4)	0.34
BAI Total
Time	−2.07	2.662	−0.78 (17.1)	0.45
Group	−0.22	5.227	−0.043 (22.6)	0.97
Time x group	−2.22	3.573	−0.62 (16.7)	0.54
DrInC-2R Total
Time	−15.79	5.432	−2.91 (16.1)	0.01*
Group	3.49	9.473	0.37 (22.5)	0.72
Time x group	−18.21	7.304	−2.49 (15.6)	0.02 *
BIPF Total
Time	−8.05	6.232	−1.29 (17.1)	0.21
Group	9.19	12.707	0.72 (22.9)	0.47
Time x group	−23.47	8.360	−2.81 (16.7)	0.01 *

Note. LME including time (baseline, 7-day follow up), group (sham, nVNS), and treatment group by time interaction as fixed effects. PHQ-8 = Patient Health Questionnaire, STAI-S = State-Trait Anxiety Inventory-State, BAI = Beck Anxiety Inventory, DrInC-2R = Drinker Inventory of Consequences, BIPF = Brief Inventory of Psychosocial Functioning, Significance code: p<.05*.

Figure 2.

Estimated marginal means (EMM) of measures of affective distress (i.e, BAI, STAI-S, and PHQ-8) for both groups at baseline and follow-up visits. Error bars represent standard errors (SE). Significant group by time interaction for PHQ-8 only (F(1, 16.53) = 4.76, p = .04), with post-hoc analyses indicating greater reductions in depressive symptoms following nVNS (Cohen d = 1.40; t = 2.78, p = .05).

Table 5.

Estimated marginal means of behavioral scores of nVNS and sham groups at baseline and post treatment

EMM (SE)	nVNS (N=9) Baseline / Follow-up	Sham (N=10) Baseline / Follow-up
PHQ-8 Total	13.2 (2.2) / 9.3 (2.3)	12.5 (2.1) / 12.9 (2.3)
STAI-S Total	48.6 (5.3) / 45.4 (5.4)	42.3 (5.08) / 42.7 (5.45)
BAI Total	16.8 (4.02) / 12.5 (4.02)	17.0 (3.81) / 14.9 (4.11)
DrInC-2R Total	47.9 (7.28) / 13.9 (7.28)*	44.4 (6.9) / 28.6 (7.57)
BIPF Total	56.9 (9.8) / 25.3 (9.8)*	47.7 (9.3) / 39.6 (9.9)

Note. Estimated marginal means (EMM) and standard errors (SE) for symptom scores at baseline and post treatment visits. PHQ-8 = Patient Health Questionnaire, STAI-S = State-Trait Anxiety Inventory-State, BAI = Beck Anxiety Inventory, DrInC-2R = Drinker Inventory of Consequences, BIPF = Brief Inventory of Psychosocial Functioning. Significance codes: p<.05*.

No group by time interactions were observed for state anxiety (STAI-S; F(1,11.39) = 1.01, p = .34), or overall anxiety (BAI; F(1,16.65) = 0.39, p = .54).

Treatment effects on alcohol consumption

No statistically significant group by time interactions were found in percent days drinking (F(1,16.32) = 3.06, p=.09), or percent days heavy drinking (F(1,14.63) = 0.62, p = .44).

Treatment effects on functional outcomes

Analyses revealed statistically significant group by time interactions for the DrInC-2R total (F(1, 15.59) = 6.22, p = .02; Table 4). In addition, group by time interactions were statistically significant for the DrInC-2R physical (F(1, 16.09) = 10.63, p = .005) and intrapersonal (F(1, 16.12) = 9.99, p = .006; Table 6) subscales. Specifically, post hoc analyses showed greater pre- to post-treatment decreases on the DrInC-2R total (Cohen d = 3.28, t = 6.52, p < .0001; Fig 3), the DrInC-2R physical (Cohen d = 3.07, t = 6.12, p < .0001), and the DrInC-2R intrapersonal (Cohen d = 3.12, t = 6.21, p < .0001) subscales in the nVNS relative to the sham group (range of Cohen d = 0.78 to 1.52, range of p values = .06 to .52). These results for the DrInC-2R physical and intrapersonal subscales are consistent with greater reductions in perceived adverse physical states (e.g., hangovers, sleeping problems, harm to health) and negative feelings (e.g., feeling bad or guilty) due to excessive drinking in the nVNS group. Following the 7-day treatment window, both groups reported problems in the low to very low range, with the nVNS group showing a larger decrease. Group by time interactions for the other DrInC-2R subscales were not statistically significant. Scores for the DrInc-2R should be interpreted with caution. Although the time frame of the DrInc-2R can be modified to assess baseline to post treatment changes, normative data is only available for the past 30-day time frame version of the questionnaire ^61,70.

Table 6.

Results from linear mixed effects (LME) models of DrInC-2R subscales

	Estimate	Std. Error	t value (df)	p value
DrInC-2R Physical
Time	−1.95	1.230	−1.501 (16.8)	0.15
Group	0.86	1.978	0.433 (24.5)	0.67
Time x group	−5.72	1.753	−3.26 (16.1)	0.005 **
DrInC-2R Interpersonal
Time	−3.951	1.803	−2.19 (16.7)	0.04*
Group	2.900	2.681	1.08 (24.8)	0.29
Time x group	−4.60	2.433	−1.89 (16.0)	0.08
DrInC-2R Intrapersonal
Time	−2.636	1.551	−1.70 (16.6)	0.11
Group	2.811	2.836	0.99 (22.6)	0.33
Time x group	−6.59	2.084	−3.16 (16.1)	0.006 **
DrInC-2R Impulse
Time	−5.26	1.410	−3.73 (14.0)	0.002**
Group	−1.66	1.825	−0.91 (24.9)	0.37
Time x group	0.04	1.911	0.02 (13.1)	0.99
DrInC-2R Social Responsibility
Time	−2.17	1.184	−1.83 (16.9)	0.09
Group	−1.42	1.512	−0.94 (27.5)	0.36
Time x group	−1.16	1.606	−0.72 (15.9)	0.48

Note. LME including time (baseline, 7-day follow up), group (sham, nVNS), and treatment group by time interaction as fixed effects. DrInC-2R = Drinker Inventory of Consequences. Significance codes: p<.05* p<.01**.

Figure 3.

Estimated marginal means (EMM) of measures of functional outcomes (i.e, DrInC-2R and BIPF) for both groups at baseline and follow-up visits. Error bars represent standard errors (SE). Significant group by time interaction for BIPF total (F(1, 16.72) = 7.88, p = .01) and DrInC-2R total (F(1, 15.59) = 6.22, p = .02).

For the BIPF total, analyses revealed a statistically significant group by time interaction (F(1, 16.72) = 7.88, p = .01). Specifically, significantly greater pre- to post-treatment decreases in BIPF scores (Cohen d = 2.67; t = 5.30, p = .0003) were observed in the nVNS relative to the sham group (Cohen d = 0.68; t = 1.20, p = .63). Following treatment, impairments decreased from severe to mild levels in the nVNS group, while in the sham group, impairments only decreased to moderate levels.

DISCUSSION

The present pilot study is the first to explore feasibility and acceptability of cervical nVNS as a potential novel treatment for AUD in Veterans, and whether nVNS can achieve a reduction in withdrawal-related affective distress and improve functional outcomes in this population. Findings provide support for feasibility of treatment delivery and patient acceptance, and initial evidence that 7-days of nVNS treatment relative to sham can achieve significant reductions in depressive symptoms as well as adverse alcohol-related consequences and psychosocial functional limitations.

With regard to affective distress, the present study found a reduction in depressive symptoms but no significant decrease in anxiety symptoms following nVNS treatment. Improvement in depressive symptoms following (implanted) VNS is well established in the literature ^71-73 and VNS is FDA-approved for treatment-resistant depression. In addition, there is increasing evidence that noninvasive VNS can achieve similar results ^74-76. Although there is currently only one other study that evaluated the effects of nVNS on withdrawal-related affective distress in AUD, our results are directly in line with this work. Specifically, Wang et al. (2021) found that nVNS treatment was associated with improvements in depression, but not a change in anxiety symptoms, in men with protracted withdrawal symptoms after 4-weeks of auricular nVNS treatment ⁴⁰.

There are a number of potential mechanisms that may explain the anti-depressant effect of nVNS in general, and more specifically, in AUD. Long-term excessive alcohol consumption causes a state of chronic physiological distress during abstinence due to changes in the autonomic nervous system (i.e., increased sympathetic/decreased parasympathetic tone) and in the central nervous system (i.e., overactive brain stress systems) ^7-9. This imbalance in homeostasis during abstinence results in affective distress (e.g., depression, anxiety, irritability) and the urge to drink to relieve these symptoms. The vagus nerve plays a critical role in maintaining and restoring homeostasis ³¹ and noninvasive stimulation of the vagus nerve has been shown to reduce sympathetic and increase parasympathetic activation, which corresponds to a reduced physiological stress response ^77-80. Neuroimaging studies have demonstrated that nVNS can modulate neural activation in brain regions involved in the perception and regulation of emotion and distress, including the insula, amygdala, prefrontal cortex, and anterior cingulate ^30-32. nVNS may therefore alter a stress response by modulating physiological symptoms of distress (sympathetic activation), and thus reduce cognitive manifestations of affective distress ⁸¹. While negative affective states associated with chronic, heavy alcohol use include a variety of symptoms in addition to anxiety, such as irritability, dysphoria, anhedonia, and negative mood ^7,12,13, our proposed mechanism of nVNS for alcohol-related withdrawal maps most directly onto combatting anxiety rather than depression with treatment. Nevertheless, elevated sympathetic and less active parasympathetic nervous system activity has also been shown in depression ^82-84, therefore, a rebalance of the ANS system could explain our current finding of improved depressive symptoms.

Alternatively, or complimentary, increased serotonin and noradrenaline levels following nVNS (due to its projections to the solitary nucleus, dorsal raphe nucleus, and locus coeruleus) have been proposed as a potential mechanism for reduced depressive symptoms ⁷². Reductions in both neurotransmitters have been implicated in the pathophysiology of AUD, including alcohol withdrawal and relapse, and the noradrenaline system has been proposed as a treatment target for AUD ^85-88. It is also plausible that activation of anti-inflammatory pathways with nVNS^89,90 may reduce chronic inflammation associated with AUD ^91-93. Long-term, heavy alcohol use may lead to increased cytokine levels and a chronic inflammatory state ⁹⁴, and (neuro)inflammation has been linked to overactive sympathetic nervous system and symptoms of negative affect, including anxiety and depression ^95-99. Reduction in inflammatory responses in AUD with nVNS may therefore present an alternative potential mechanism of action for the symptom improvement observed in the present study. In addition, Wang et al. (2021) hypothesize that improved depression in their cohort of AUD patients, may be due to the effects of nVNS on plasma levels of brain derived neurotrophic factor (BDNF), which were significantly higher in AUD patients who received nVNS relative to a control group ⁴⁰. Increased BDNF has been previously implicated as a potential mechanism of improved depressive symptoms in a range of studies by promoting neuroplasticity ^82,100. Moreover, the authors of this work argue that the lack of reductions in anxiety in their study may be due to the lack of reductions in alcohol craving as they have been proposed to be causally related ⁴⁰. In the present study, the lack of significant changes in anxiety symptoms may also be due to the small sample size or the questionnaires used to measure anxiety. As such, it is possible that our measures did not capture the physiological or somatic, but rather the cognitive aspects of anxiety. However, nVNS has been found to reduce both anxiety and depressive symptoms in individuals with major depressive disorder ^74,101-103, pointing out the need for further studies to better understand whether anxiolytic effects of nVNS may be specific to certain psychiatric disorders and what the underlying mechanism of action may be. In sum, while this early work suggests that depressive symptoms may be more sensitive to nVNS treatment than anxiety, both remain important targets in those with AUD. Future research with larger cohorts and additional or more comprehensive, mechanistic measures may provide further insight into this question.

The present study also found improvements in functional outcomes, as measured with the DrInC-2R and the BIPF, following nVNS but not sham treatment. The DrInC-2R and the BIPF assess functional outcomes in several domains such as interpersonal relationships, family relationships, work, parenting, education, and self-care (BIPF) as well as physical, intrapersonal, social responsibility, interpersonal functioning, and impulse control (DrInC-2R). While the composite scores of these measures show improvements in functional outcomes in general, the subscales of the DrInC-2R indicate that these changes may be specific to perceived adverse physical states (e.g., harm to physical health due to drinking) and negative feelings (e.g., shame or guilt, loss of interest, harm to moral life due to drinking). Mechanistically, it is possible that these improvements in physical health and negative feelings may be driven by reductions in physiological hyperarousal and depressive symptoms due to restored balance of the autonomic nervous system. Although we were not powered to assess a mediation mechanism in this study, a reduction in affective (and physiological) distress following nVNS treatment is consequently likely to improve functional outcomes and decrease adverse alcohol-related consequences in those with AUD. In addition, it is important to point out that the BIPF is a brief 7-item measure that does not provide domain-specific information. Future studies using the expanded 80-item version of this measure may shed more light onto which areas in particular can be improved with nVNS treatment.

There are several noteworthy advantages of nVNS: it is a low-risk, low-cost treatment with few strict contraindications and can be safely self-administered using a handheld pocket-sized device - thus its implementation presents a low burden to both patients and healthcare providers. The present study provides support that nVNS as a treatment for AUD is feasible and that non-invasive neuromodulation is well accepted and tolerated without any serious adverse effects. While earlier technology allowed only for invasive vagus nerve simulation via a surgically implanted pulse generator connected via wires to the cervical branch of the vagus nerve, non-invasive forms of nVNS are now available to stimulate the vagus nerve through the skin via either transcutaneous auricular nVNS (auricular branch of the vagus nerve in the ear) or transcutaneous cervical nVNS (cervical branch of the vagus nerve in the neck). Both auricular and cervical nVNS are well tolerated and have been shown to reliably stimulate the vagus nerve in humans similar to implantable VNS devices as evidenced by vagal somatosensory evoked potentials ^32,53,104. Importantly, noninvasive VNS does not carry the risks associated with surgery and having an implanted medical device ³¹, and has been shown to be generally safe and well tolerated ^104-107. Cervical nVNS, as used in this study, may therefore present a promising novel treatment option for AUD. The results of this study strongly support a larger clinical trial to assess device effectiveness.

The present study has several limitations that are important to address. Most notably, the sample size is small, stressing the need for further research with larger cohorts. In addition, since the study included only male Veterans, present findings may not be generalizable to civilians with AUD or to women. Furthermore, ideal dosing of nVNS (i.e., number of administrations per day, treatment duration, unilateral vs. bilateral stimulation, etc.) for the treatment of AUD symptoms are still unclear and need further research. There are also no studies yet that have evaluated whether these therapeutic effects extend beyond active treatment, signaling a need for studies that evaluate sustained intervention effects of nVNS. Furthermore, the device used in this study unfortunately does not have any compliance monitoring capabilities. Treatment adherence was therefore measured by self-report (i.e., completion of administration log). It is noteworthy that treatment compliance was lower in the sham treatment group (although the difference between groups was not statistically significant), which could possibly indicate functional unblinding due to perceptual differences in stimulation. Another limitation is that the time frame of the DrInC-2R assessment covered 90 days prior to baseline but only the past week at follow-up, which likely biased our findings to detect decreases in problems. However, time intervals compared between groups were equivalent. These timeframes were chosen to capture a comprehensive assessment of alcohol-related adverse consequences at baseline and to assess post-intervention effects. When interpreting findings, it is important to consider the short time window of follow-up survey data. Lastly, the BIPF is a measure of PTSD-related functional impairments in Veterans and has not been validated in an AUD population. However, this questionnaire is designed to query general functional outcomes and does not require subjects to attribute impairments to PTSD-specific symptoms.

The present study did not control for potential effects of past or current treatment for AUD symptoms and included subjects who received psychotherapy and/or medication management at the time of study participation (current or recent participation in abstinence-based treatment was an exclusion criterion and subjects did not have a recent change in treatment status). Future study designs could be improved by accounting for treatment status (e.g., no treatment, psychotherapy, medication, psychotherapy and medication). It would also be beneficial to conduct trials in both inpatient and outpatient settings and to evaluate effectiveness of nVNS in supporting different stages of AUD treatment (i.e., acute alcohol detoxification, protracted withdrawal, relapse prevention, etc.). Furthermore, in the present study, subjects were instructed to drink as much and as often as desired during the treatment period (i.e., they were not instructed to reduce or abstain from drinking). It would be important in future study designs to account for patient motivation to reduce or abstain from alcohol use in addition to measuring alcohol consumption, and to evaluate effectiveness of nVNS in supporting reduction of alcohol use versus achieving abstinence, as well as maintaining this change in drinking (i.e., durability of effects).

In conclusion, the present pilot study provides valuable initial support that nVNS is a feasible and well accepted potential new treatment for AUD-related symptoms, and that it may reduce affective distress and improve functional outcomes in individuals with AUD. nVNS may offer an innovative approach to treating AUD symptoms without the limitations associated with medication (side effects, risks) or psychotherapy (stigma, time commitment, access constraints). Further research with larger study cohorts is needed to evaluate the efficacy of nVNS in improving a wide range of AUD-related symptoms, as well as to explore ideal dosing parameters, durability of effects, and (physiological and neural) mechanisms of action. Future work is also important to assess whether nVNS may be beneficial as a stand-alone treatment or as an adjunctive treatment to conventional medication management or psychotherapy (or to non-responders) in order to support sobriety and to reduce relapse risk.

Abbreviations:

nVNS

Noninvasive vagal nerve stimulation

AUD

alcohol use disorder