Abstract
Background
Low-level transcutaneous electrical stimulation of the auricular branch of the vagus nerve at the tragus (LLTS) acutely suppresses atrial fibrillation (AF) in humans, but the chronic effect remains unknown.
Objectives
We conducted a sham-controlled, double-blind, randomized clinical trial to examine the effect of chronic LLTS in patients with paroxysmal AF.
Methods
LLTS (20Hz, 1mA below the discomfort threshold) was delivered using an ear clip attached to the tragus (active; n=26) or the ear lobe (sham control; n=27) for 1 hour daily over 6 months. AF burden over 2-week periods was assessed by noninvasive continuous ECG monitoring at baseline, 3 and 6 months. Five-minute ECG and serum were obtained at each visit to measure heart rate variability (HRV) and inflammatory cytokines, respectively.
Results
Baseline characteristics were balanced between the 2 groups. Adherence to the stimulation protocol (≤4 sessions lost per month), was 75% in the active and 83% in the control arm (p>0.05). At 6 months, the median AF burden was 85% lower in the active compared to the control arm (ratio of medians: 0.15, 95% CI: 0.03 to 0.65, p=0.011). Tumor necrosis factor-α was significantly decreased by 23% in the active relative to the control group (ratio of medians: 0.77, 95% CI: 0.63 to 0.94, p=0.0093). Frequency domain indices of HRV were significantly altered with active vs. control stimulation (p<0.01). No device-related side effects were observed.
Conclusion
Chronic, intermittent LLTS resulted in lower AF burden than sham control stimulation, supporting its use to treat paroxysmal AF in selected patients.
Keywords: atrial fibrillation, randomized clinical trial, Neuromodulation
Condensed Abstract
We investigated the chronic effects of low-level transcutaneous vagus nerve stimulation, a noninvasive therapeutic approach, on atrial fibrillation burden in patients with paroxysmal atrial fibrillation. In this sham-controlled, double-blind, randomized clinical trial, low-level transcutaneous vagus nerve stimulation in patients with paroxysmal atrial fibrillation resulted in a significant reduction in the median atrial fibrillation burden in the active compared to the sham group at 6 months.
Our results support the emerging paradigm of noninvasive neuromodulation to treat AF.
Vagus nerve stimulation (VNS) is a neuromodulation method approved for the treatment of drug-refractory epilepsy (1). Transcutaneous VNS, by stimulating the auricular branch of the vagus nerve at the tragus of the external ear, is an emerging noninvasive alternative to VNS (2). Previous functional magnetic resonance imaging studies have shown that central projections of the vagus nerve in the brain stem and other higher centers in the brain are being activated by this approach (3,4). Notably, tragus stimulation has been shown to be effective in treating human diseases including tinnitus (4,5) and epilepsy (6). Moreover, tragus stimulation decreases sympathetic tone in humans (7) and suppresses inflammatory cytokines in both animals (8) and humans (9,10).
Recent evidence suggests that the autonomic nervous system plays a central role in the initiation and maintenance of atrial fibrillation (AF), especially in the early stages (11,12). Importantly, several studies from our group (13–17) and others (18) have shown that autonomic neuromodulation with low-level VNS can suppress AF in experimental models. More recently, in a proof-of-concept human study, we showed that in patients with drug-refractory paroxysmal AF, low-level transcutaneous electrical stimulation of the auricular branch of the vagus nerve at the tragus of the ear (LLTS) for just one hour significantly shortened AF duration and decreased inflammatory cytokines (9). The importance of the latter finding is highlighted by accumulating evidence for a significant role of inflammatory pathways in the pathogenesis of AF (19,20) and is consistent with the well-characterized anti-inflammatory effects of VNS (21,22). Nonetheless, the chronic effects of LLTS in humans remain unknown. In the current study, we evaluated the effect of chronic LLTS on AF burden in patients with paroxysmal AF over a 6-month period relative to sham stimulation.
Methods
This was a prospective double-blind, sham-controlled, randomized clinical trial. Patients with paroxysmal AF, documented by ECG, implantable device or Holter monitor, within 3 months of randomization on 2 separate occasions, at least 1 day apart, of at least 30 seconds duration, were eligible for inclusion in the study. Patients were excluded if they had any of the following: left ventricular ejection fraction <40%, significant valvular disease, recent (<6 months) stroke or myocardial infarction, severe heart failure (New York Heart Association class III or IV), recurrent vaso-vagal syncopal episodes, unilateral or bilateral vagotomy and pregnancy or nursing. In addition, we excluded patients with sick sinus syndrome, 2nd or 3rd degree atrioventricular (AV) block, bifascicular block and prolonged 1st degree AV block (PR>300ms), in the absence of a pacemaker. The study was approved by the Institutional Review Board of the University of Oklahoma Health Sciences Center and informed consent was obtained from all patients prior to enrollment in the study.
Patients were randomly assigned (1:1) to active or sham LLTS with the use of concealed envelopes, stratified by sex. Blocked randomization was used to decrease imbalances in the two groups over time. Active LLTS was performed using a transcutaneous VNS device (Parasym device, Parasym Health, Inc, London, UK) with an ear clip attached to the tragus of the right ear (Figure 1A), which is innervated by auricular branch of the vagus nerve (23). In the sham group, stimulation was delivered to the ear lobe (Figure 1B), which is devoid of vagal innervation (23). Patients were not told which site provides active stimulation to achieve blinding of treatment allocation. The investigators determining the clinical outcomes, the investigators performing the cytokine assays, and the investigators analyzing and interpreting the heart rate variability (HRV) data were also blinded to treatment allocation. The research coordinator who instructed the participant regarding placement of the ear clip according to randomized assignment and the study statistician were unblinded.
Figure 1.
Representative examples of active (A) and sham control (B) stimulation. For active stimulation, the ear clip was attached to the tragus, which is innervated by the auricular branch of the vagus nerve (A). For sham control stimulation, the ear clip was attached to the ear lobe, which is devoid of vagal innervation (B). A schematic representation of the study design and timeline of events is shown in (C).
In both groups, stimulation was delivered at a frequency of 20 Hz and pulse width of 200 μs, while the stimulation amplitude was individually titrated to 1 mA below the level that caused mild discomfort. Stimulation parameters were chosen according to previously published acute clinical studies that demonstrated beneficial effects (9,10). Patients were instructed to apply stimulation continuously for 1 hour daily for 6 months, typically at the same time of the day. Patients were requested to keep a daily log with the time and duration of LLTS application, amplitude settings and any comments related to each daily session. The cumulative number of hours used for stimulation, as provided by the device, was correlated with the daily log, as an indirect objective measure of adherence to the stimulation protocol. In patients with pacemakers, interaction between stimulation and the pacemaker was tested before enrollment in the study. None of the patients with pacemakers showed any interaction with the stimulation (no stimulation artifact seen on the pacemaker).
At baseline, 3 months and 6 months, patients underwent noninvasive continuous ECG monitoring for 2 weeks to evaluate their AF burden (defined as the percent of time spent in AF during the monitoring period) using an adhesive continuous ECG patch (Zio® Patch, iRhythm Technologies, Inc, San Francisco, CA). At the same time points, a 5-minute ECG was performed to assess HRV. Following the ECG, 10ml of blood was drawn for cytokine analysis. A schematic representation of the study design and timeline of events is shown in Figure 1C.
Serum cytokine analysis
Blood samples (10ml) were collected at baseline and at 3 and 6 months for cytokine measurement. Serum was stored in aliquots at −80°C until assayed in batches of 10 to 12. Inflammatory cytokines, including tumor necrosis factor (TNF)-α, interleukin (IL)-6, IL-1β, IL-10 and IL-17 were measured using commercially available multiplex assays (MilliporeSigma, Burlington, MA). All immunoassays were run in duplicate and read according to manufacturer’s instructions. The investigators performing the cytokine assays were blinded to group assignment.
Heart rate variability analysis
A 5-min ECG was obtained in the supine position after resting for 15 minutes. Patients were asked to avoid caffeine for 4 hours and alcohol, smoking and exercise for 12 hours prior to the visit to avoid any interference with the results. Analysis and interpretation of the HRV data was performed in a blinded fashion, using the Kubios HRV software (Kubios Inc, Kuopio, Finland), as previously described (24,25).
Statistical analysis
The primary outcome of the study was AF burden over the 2-week monitoring period at the 6-month time point. Secondary outcomes included total duration of AF over the 2-week monitoring period, longest daily duration of AF, premature atrial contraction burden, measures of HRV, as surrogates of the autonomic tone (26) and serum cytokine levels. Continuous variables are presented as mean ± standard deviation or median (interquartile range), as applicable. Categorical variables are presented as percentages. Baseline characteristics were compared between groups using t-test for continuous variables and chi-square for categorical variables. A Generalized Estimating Equations modeling approach was used to compare the mean outcome measure at the 3-month and 6-month visits between the 2 groups after adjusting for the baseline measure. An interaction between time and treatment was considered and dropped from the model if not significant. After dropping the interaction term, the between-group comparisons were made by combining the 3-month and 6-month data. Given the skewed nature of the data, a natural log transformation was used to better satisfy assumptions of normal residuals and constant variance. Because some values were 0, a value of 0.01 was added to each measure before applying the natural log. The estimates reflect the exponentiated regression model terms and are interpreted as the ratio of the median responses in the active arm relative to the median response in the sham arm. Linear regression was used to compare the association between the change in AF burden at follow up and measures of HRV. Analyses were based on the intention-to-treat principle. Statistical significance was declared at p<0.05. All statistical analyses were performed using SAS 9.3 software (SAS Institute, Inc., Cary, North Carolina).
Sample size and power calculations
Assuming a mean±standard deviation AF burden in the control group of 28.4±31.2% (27) and a 87% reduction in the LLTS group to 3.7% (assuming a common standard deviation of 31.2%), a sample size of 53 patients would provide at least 80% power to detect the specified effect sizes at a two-sided significance α level of 0.05.
Results
Study population
From May 2016 through May 2018, 78 patients were screened for eligibility and 53 patients were enrolled in the study (Figure 2). Among the 53 patients, 26 (49%) were randomized to the active stimulation group and 27 (51%) were randomized to the sham control group. The baseline characteristics of the patients were balanced between the 2 groups (Table 1). Notably, 54% and 59% of the patients in the active and control groups, respectively, were on antiarrhythmic medications (class IC or III agents) and 19% in each group did not report any symptoms of AF at baseline. Medical therapy, including beta blockers, angiotensin converting enzyme inhibitors and angiotensin receptor blockers, was balanced between the 2 groups at baseline and remained unchanged during the course of the trial. Among the 26 patients assigned to the active stimulation group, one patient died of septic shock and one patient was lost to follow up. Among the 27 patients assigned to the sham stimulation group, 2 patients died of non-cardiac causes (respiratory failure and colon cancer, respectively) and 2 patients withdrew consent. All the other patients completed the 6-month study protocol and no patient crossed over to the other group. Two patients (one in each group) progressed to persistent AF and underwent electrical cardioversion during the course of the study (between the 3- and 6-month time point). None of the patients underwent catheter ablation during the study. All patients were included in the final analysis (Figure 2). The outcomes of the patients who did not complete the study were included to the point of death, consent withdrawal, or final contact.
Figure 2.
Flow diagram of participant recruitment and follow-up
Table 1.
Baseline characteristics of the patients
| Characteristic | Active (n=26) | Control (n=27) | P value |
|---|---|---|---|
| Age (years) | 65.2±14.5 | 68.0±10.6 | 0.42 |
| Male sex, n (%) | 12 (46) | 12 (44) | 0.90 |
| Race, n (%) | 0.61 | ||
| Non white | 1 (4) | 3 (11) | |
| White | 25 (96) | 24 (89) | |
| Body mass index (kg/m2) | 30.3±6.2 | 31.6±9.1 | 0.55 |
| Years with AF, n (%) | 0.79 | ||
| ≤1 year | 5 (19) | 6 (22) | |
| >1 year | 21 (81) | 21 (78) | |
| AF severity score | 0.68 | ||
| Class 0 to 2 | 14 (52) | 13 (44) | |
| Class 3 or 4 | 12 (48) | 14 (56) | |
| Heart failure, n (%) | 6 (23) | 5 (19) | 0.85 |
| Coronary artery disease, n (%) | 5 (19) | 7 (26) | 0.56 |
| Diabetes, n (%) | 4 (15) | 7 (26) | 0.50 |
| Sleep Apnea, n (%) | 5 (19) | 9 (33) | 0.24 |
| Treatment for Sleep Apnea, n (%) | 4 (80)* | 7 (78)* | 1.0 |
| Hypertension, n (%) | 17 (65) | 23 (85) | 0.12 |
| Sick sinus syndrome, (%) | 8 (31) | 11 (41) | 0.45 |
| Antiarrhythmic medications, n (%) | 14 (54) | 16 (59) | 0.69 |
| Flecainide, n (%) | 4 (15) | 3 (11) | 0.70 |
| Propafenone, n (%) | 1 (4) | 1 (4) | 1.0 |
| Dronedarone, n (%) | 1 (4) | 2 (7) | 1.0 |
| Amiodarone, n (%) | 2 (8) | 1 (4) | 0.61 |
| Sotalol, n (%) | 2 (8) | 1 (4) | 0.61 |
| Dofetilide, n (%) | 4 (15) | 8 (29) | 0.33 |
| Beta blockers, n (%) | 18 (69) | 20 (74) | 0.77 |
| Metoprolol, n (%) | 15 (58) | 16 (59) | 1.0 |
| Carvedilol, n (%) | 3 (11) | 4 (15) | 1.0 |
| ACE inhibitors/ARBs, n (%) | 11 (42) | 12 (44) | 0.88 |
| Calcium channel blockers, n (%) | 3 (12) | 5 (19) | 0.70 |
| Statins, n (%) | 6 (23) | 8 (30) | 0.59 |
| CHA2DS2-VASc Score | 2.9±2.1 | 3.0±1.4 | 0.84 |
| Left ventricular hypertrophy, n (%) | 19 (74) | 20 (73) | 0.93 |
| Left ventricular ejection fraction (%) | 61.2±8.5 | 58.1±8.7 | 0.20 |
| Left atrial diameter (cm) | 4.4±0.6 | 4.6±0.8 | 0.31 |
| Heart rate (bpm) | 66.9±10.7 | 67.4±10.9 | 0.87 |
| PR interval (ms) | 186.1±44.9 | 192.2±45.7 | 0.63 |
| Correct QT interval (ms) | 441.2±21.4 | 440.3±22.9 | 0.99 |
percent of those with sleep apnea
ACE = angiotensin converting enzyme; ARB = angiotensin receptor blocker
Adherence to the protocol of daily stimulation, defined as ≤4 sessions missed on average per month, was 96%, 80% and 74% at 1, 3 and 6 months, respectively, in the active group and 85%, 88% and 82%, respectively, in the control group (p=0.49 for comparison between groups). Among patients randomized to the active group, the mean monitoring period was 12.4, 12.2 and 13.1 days, respectively at the baseline, 3-month and 6-month time point. Among patients randomized to the control group, the respective monitoring periods were 12.8, 11.8 and 10.8 days, consistent with those previously reported (27). The average stimulation amplitude was 16.8±12.6 mA and 19.9±12.1 mA in the active and control group, respectively (p=0.36).
AF burden
The outcomes related to AF burden are summarized in Table 2. AF burden was non-normally distributed, with a baseline median (interquartile range) of 4.5% (0.2% to 31.0%) and 1.0% (0% to 15.0%) in the active and control groups, respectively (p=0.43). At 6 months, the median AF burden was 85% lower in the active compared to the sham control group (ratio of medians: 0.15, 95% CI: 0.03 to 0.65, p=0.011; Figure 3). In the active group, AF burden at 6 months was significantly lower than at baseline (ratio of medians: 0.24, 95% CI 0.07 to 0.78; p=0.018). In the control group, AF burden at 6 months, although numerically higher, was not statistically significantly higher compared to baseline (ratio of medians: 1.88, 95% CI 0.68 to 5.16; p=0.22). After combining across the 3- and 6-month time points, the median AF burden was significantly reduced by 75% in the active compared to control group (ratio of medians: 0.25, 95% CI: 0.08 to 0.77, p=0.016). Similarly, the total duration of AF at 6 months was 83% lower in the active compared to control arm (ratio of medians: 0.17, 95% CI: 0.03 to 0.85, p=0.032). The longest daily duration of AF did not differ significantly between the 2 groups (ratio of medians: 0.38, 95% CI: 0.07 to 2.03, p=0.26). Patient-level data for AF burden are shown in Figure 4. Seven patients in each group had AF burden of 0 during their baseline and follow-up 2-week monitoring periods. Considering the patients with non-zero AF burden at baseline, 9 patients with in the active group (47%) vs. 1 patient in the control group (5%) experienced >75% reduction in AF burden during follow up (p=0.003).
Table 2.
Effect of LLTS on AF burden and cytokine levels
| Active | Control | Outcome (3 and 6 months combined) | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Outcome | Baseline (N=26) | 3 months (N=25) | 6 months (N=24) | Baseline (N=27) | 3 months (N=26) | 6 months (N=23) | Ratio of medians (Active vs. sham) | 95% CI | p-value |
| AF burden | |||||||||
| AF burden (%) | 4.5 (0.2 – 31) | 2.0 (0 – 21) | 2.0 (0 – 11) | 1.0 (0 – 15) | 3.0 (0.1 – 11) | 8.5 (0 – 42) | 0.25 | 0.08 – 0.77 | 0.016 |
| Total AF duration (hours) | 12.0 (0.1 – 77) | 6.5 (0 – 62) | 7.0 (0 – 37) | 2.0 (0 – 43) | 4.0 (0.2 – 37) | 19.8 (0 – 77) | 0.27 | 0.07 – 1.08 | 0.064 |
| Longest daily AF duration (hours) | 6.0 (0 – 21) | 6.5 (0 – 24) | 6.0 (0 – 19) | 2.0 (0 – 13) | 3.5 (0 – 16) | 10.3 (0 – 23) | 0.51 | 0.13 – 1.97 | 0.33 |
| Cytokines | |||||||||
| TNF-α (pg/ml) | 6.9 (4.6 – 9.3) | 5.2 (3.8 – 9.2) | 5.5 (3.9 – 8.3) | 5.4 (4.3 – 9.1) | 6.5 (4.5 – 9.4) | 5.4 (4.4 – 8.3) | 0.77 | 0.63 – 0.94 | 0.0093 |
| IL-6 (pg/ml) | 1.6 (0.5 – 3.7) | 1.6 (0.5 – 2.7) | 1.5 (0.7 – 3.2) | 2.0 (0.9 – 3.7) | 2.2 (1.2 – 4.3) | 2.3 (1.0 – 4.4) | 0.99 | 0.64 – 1.51 | 0.94 |
| IL-1β (pg/ml) | 0.9 (0.4 – 15) | 0.7 (0.5 – 12) | 0.8 (0.5 – 11) | 1.0 (0.4 – 13) | 0.8 (0.3 – 12) | 0.7 (0.3 – 12) | 1.13 | 0.97 – 1.30 | 0.11 |
| IL-10 (pg/ml) | 6.2 (1.3 – 10.4) | 4.4 (2.1 – 9.6) | 5.0 (2.0 – 7.3) | 11.7 (3.5 – 19.5) | 10.0 (4.1 – 21.2) | 9.0 (3.7 – 15.3) | 1.02 | 0.79 – 1.31 | 0.88 |
| IL-17 (pg/ml) | 5.2 (2.2 – 9.6) | 4.7 (2.8 – 10.0) | 4.6 (2.7 – 9.3) | 5.6 (2.9 – 9.5) | 5.1 (2.3 – 11.6) | 3.8 (2.2 – 6.4) | 1.11 | 0.90 – 1.38 | 0.34 |
Data are presented as median (interquartile range). The p-value is based on a comparison of median levels at the 3-month and 6-month time points combined, after adjusting for baseline measures using Generalized Estimating Equations methodology to fit linear regression models.
Figure 3.
Comparison of atrial fibrillation (AF) burden between the 2 groups. The data are presented as median values and interquartile range. The p-value is based on a comparison of median AF burden levels at the 6-month time point after adjusting for baseline measures.
Figure 4.
Patient-level data on atrial fibrillation (AF) burden change in the 2 groups. Patients whose AF burden decreased by >75% at follow up were categorized as responders. The proportion of responders was significantly larger in the active compared to the sham control group (47% vs. 5%, respectively, p=0.003). B = baseline; 3M = 3 months; 6M = 6 months.
The burden of premature atrial contractions was also decreased in the active compared to control group (ratio on medians: 0.69, 95% CI 0.48 to 0.98, p=0.04). There was no significant effect of active compared to sham stimulation on heart rate, PR interval or QT interval during follow up (70.5±11.2 bpm vs. 69.6±11.9 bpm, p=0.78; 190.1±49.5 ms vs. 178.4±50.4 ms, p=0.87; 438.2±23.5 ms vs. 436.3±24.9 ms, p=0.95, respectively).
Cytokines
Serum TNF-α levels were significantly decreased in the active relative to control group by 23% (ratio of medians: 0.77, 95% CI: 0.63 to 0.94, p=0.0093). Consistent with our previous acute study (9), the levels of the rest of the cytokines examined were low, as expected in the absence of acute infection, inflammatory or autoimmune condition in our patients (28,29) and were not significantly different between the 2 groups (Table 2).
Heart rate variability
Data on HRV were available in ≤13 patients in each group at each time point, due to the presence of either atrial pacing (in 8 and 11 patients in the active and control group, respectively), AF, or multiple premature ventricular complexes at the time of the ECG in the remaining patients. In light of the relatively high percentage of missing values, we advise caution when interpreting these results. In the frequency domain, low frequency (LF) was significantly higher (ratio of medians: 1.59, 95% CI: 1.14 to 2.23, p=0.007; Table 3) and high frequency (HF) was significantly lower in the active compared to the sham group (ratio of medians: 0.74, 95% CI: 0.59 to 0.93, p=0.01). The LF/HF ratio, which reflects sympathovagal balance (26), was significantly higher in the active compared to the sham group (ratio of medians: 2.16, 95% CI: 1.29 to 3.63, p=0.003). There was a trend towards significance in the linear association between the change in LF/HF ratio and the respective change in AF burden at follow up in the active group (r=−0.51, p=0.077), but not in the control group (r=−0.14, p=0.67) (Figure 5). There were no differences between the 2 groups in the time domain (Table 3), consistent with the notion that frequency domain measures perform better than time domain measures when short duration (5 minute) recordings are examined (24).
Table 3.
Heart rate variability analysis
| Active | Control | Outcome (3 and 6 months combined) | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Outcome | Baseline (N=12) | 3 months (N=10) | 6 months (N=13) | Baseline (N=12) | 3 months (N=11) | 6 months (N=10) | Ratio of medians (Active vs. sham) | 95% CI | p-value |
| RMSSD (ms) | 20.2 (14.6 – 31.7) | 17.0 (10.0 – 26.5) | 12.3 (8.5 – 30.7) | 18.2 (12.5 – 28.5) | 17.1 (15.6 – 50.9) | 27.0 (13.4 – 41.7) | 0.71 | 0.45 – 1.12 | 0.15 |
| SDNN (ms) | 17.7 (12.0 – 23.8) | 15.9 (14.4 – 23.2) | 20.1 (7.4 – 33.3) | 15.3 (11.5 – 28.5) | 19.1 (13.0 – 37.6) | 20.8 (12.4 – 25.9) | 0.82 | 0.54 – 1.25 | 0.36 |
| pNN50 (%) | 1.2 (0 – 14.0) | 1.3 (0 – 5.7) | 0 (0 – 10.8) | 0.5 (0 – 9.0) | 1.4 (0 – 14.9) | 8.5 (0 – 20.1) | 0.38 | 0.06 – 2.36 | 0.23 |
| LF power (n.u.) | 55.5 (34.4 – 59.9) | 52.2 (34.6 – 81.8) | 50.2 (39.1 – 64.1) | 39.0 (16.4 – 73.5) | 20.5 (13.3 – 48.7) | 42.0 (20.9 – 56.6) | 1.59 | 1.14 – 2.23 | 0.007 |
| HF power (n.u.) | 44.4 (40.0 – 65.5) | 47.6 (18.0 – 65.4) | 49.7 (35.8 – 60.3) | 60.8 (26.3 – 83.4) | 79.5 (51.1 – 86.5) | 57.7 (43.4 – 78.9) | 0.74 | 0.59 – 0.93 | 0.010 |
| LF/HF ratio | 1.3 (0.5 – 1.5) | 1.1 (0.5 – 4.5) | 1.0 (0.6 – 1.8) | 0.6 (0.2 – 2.8) | 0.3 (0.2 – 10) | 0.7 (0.3 – 13) | 2.16 | 1.29 – 3.63 | 0.003 |
Data are presented as median (interquartile range). N.u. = normalized units
Figure 5.
Linear regression of the change in AF burden at follow up (Δ AF burden) as a function of the respective change in low frequency to high frequency ratio (Δ LF/HF) in the active (n=13; r=−0.51, p=0.077) and sham control group (n=11; r=−0.14, p=0.67).
Adverse events
There were no adverse events related to the use of the device.
Discussion
In this study, we showed for the first time that chronic, intermittent, transcutaneous electrical stimulation of the auricular branch of the vagus nerve at the tragus suppressed AF in patients with paroxysmal AF over a 6-month period. These data suggest that LLTS, a low-cost, low-risk intervention, applied for a short period of time in selected patients with paroxysmal AF, may result in a significant decrease in their AF burden (Central illustration). Notably, a growing body of evidence suggests that AF burden is a clinically meaningful endpoint associated with cardiovascular and neurological outcomes (30). Moreover, the importance of this finding is highlighted by recent evidence that lower AF burden over a 2-week monitoring period is associated with a lower risk of ischemic stroke independent of known stroke risk factors in patients with paroxysmal AF (31). Our results corroborate previous experimental evidence from our group and others, that stimulation of either the cervical vagus nerve, or the tragus is able to suppress AF and reverse atrial electrical and autonomic remodeling (13–18). These promising results provide the basis for the design of further randomized clinical trials to evaluate the long-term efficacy of autonomic neuromodulation for the treatment of patients with paroxysmal AF.
Central illustration.
Noninvasive neuromodulation using low level tragus stimulation (LLTS) significantly decreased atrial fibrillation (AF) burden and decreased tumor necrosis factor (TNF)-α levels. The potential mechanisms of this effect are shown. Also shown are the neural pathways involved in this effect. LLTS targets the auricular branch of the vagus nerve, an afferent nerve (blue arrows) which relays information to central vagal projections in the brain stem. The signal undergoes processing in the brain stem and in higher centers (green arrows), which in turn provide the efferent neural signal to the heart (red arrows), which reaches the target organ through the vagus nerve.
The use of low-level VNS to prevent AF may appear a paradox, given that VNS has been used for almost a century to induce AF in experimental models (32,33). However, recent evidence suggests that the ability of VNS to induce AF is proportional to the degree of heart rate slowing, with no increase in AF inducibility until VNS slows the heart rate by at least 40% (34). On the other extreme, we and others have consistently shown that VNS at levels significantly below the bradycardia threshold induces a strong antiarrhythmic effect (13–18). Taken together, these observations suggest that the final effect of VNS depends on the level of stimulation, with anti-arrhythmic effects prevailing at low levels which do not slow the sinus rate (35) and proarrythmic effects predominating at high levels, associated with significant sinus rate slowing.
The antiarrhythmic effects of low-level cervical VNS have been attributed to its antiadrenergic effects (36). Importantly, transcutaneous VNS has been shown to activate central vagal projections in the brain in humans, leading to decreased sympathetic output (3). Furthermore, it has been argued that transcutaneous VNS, which preferentially activates afferent rather than efferent vagal fibers, may offer a therapeutic advantage (2). Unlike cervical VNS, which may inadvertently result in stimulation of sympathetic fibers, that are co-localized with vagal fibers in the vagus nerve (37), transcutaneous VNS may induce minimal or no concomitant sympathetic stimulation (2). The fact that AF burden and total AF duration, but not the longest AF duration were decreased by LLTS in our study, suggests that LLTS suppressed the initiation of AF, but did not terminate the AF episode once it had started. This is consistent with the notion that different mechanisms operate for the initiation compared to the maintenance of an AF episode (11) and is supported by the finding that LLTS significantly suppressed atrial ectopic beats, which trigger AF.
The reduction of TNF-α levels by LLTS in the present study is similar in magnitude to our previous human studies (9,38) and is consistent with the well-established anti-inflammatory effects of VNS (21,22). The significance of this result is highlighted by accumulating evidence for an important role of inflammatory pathways in the pathogenesis of AF (19,20). Although historically the immune system was considered to be self-regulating through autonomous humoral and cellular pathways, recent evidence indicates that neural circuits regulate immunity, through the inflammatory reflex, a prototypical reflex circuit that maintains immunological homeostasis (39). Specifically, the vagus nerve provides the efferent and possibly the afferent limb of this inflammatory reflex, also known as the cholinergic anti-inflammatory pathway, by which the brain modulates inflammation (21,39). Transcutaneous VNS acts through the same pathway to suppress inflammatory cytokines (8). A recent preliminary clinical trial in patients with rheumatoid arthritis showed that VNS for 4 minutes daily inhibited whole blood lipopolysaccharide-induced TNF-α production and decreased serum IL-6 levels at 6 weeks, raising the intriguing possibility that VNS for only short periods of time may be sufficient to induce a long-lasting anti-inflammatory response (22). Consistent with this notion, in our study, one hour of stimulation daily resulted in a decrease in TNF-α levels at follow up. The degree to which this anti-inflammatory effect contributed to decreased AF burden cannot be determined with the current study design.
Although we did not specifically examine the mechanism of LLTS’s effect in this study, we hypothesize that it may result, at least partially, from neural remodeling (Central illustration). It is well known that the stimulatory or inhibitory effects of neuro-stimulation can greatly outlast the duration of stimulation (long-term potentiation or long-term depression) (40). Moreover, there is evidence suggesting that synaptic plasticity, the ability of neurons to alter their strength of communication at synapses, which in turn results in long-term potentiation or depression, occurs not only in the brain, but also in the cardiac autonomic ganglia (41). Importantly, it was previously shown in mice that percutaneous VNS for 30 minutes decreased lipopolysaccharide-induced brain neuro-inflammation and improved cognitive function 24 hours later (42). In addition, transcutaneous VNS in humans resulted in activation of central vagal projections in the brain, as assessed by functional magnetic resonance imaging, which persisted well after cessation of stimulation (3). Collectively, these observations support the notion that LLTS is characterized by a “memory” effect, which allows for brief periods of stimulation to result in long-lasting effects. Whether this effect occurs at the level of the cardiac autonomic ganglia, or the brain, or both, remains to be determined.
The HRV data are limited by a significant number of missing values, which may introduce bias. Limitations notwithstanding, our results support the notion that LLTS restored sympathovagal balance towards a more physiologic range (43).
In our study, the response to LLTS was variable among individual patients (Figure 4), suggesting that patient selection is critical to optimize results and highlighting the need for an acute biomarker of response to therapy, which in turn would maximize the efficacy of this novel therapeutic modality. An ideal biomarker should change acutely with LLTS and this change should be able to predict chronic response to therapy. Unfortunately, such a biomarker is lacking at present. Several biomarkers have been shown to change acutely with LLTS in humans, including muscle sympathetic nerve activity (7), heart rate variability (7,25), TNF-α (9) and global longitudinal strain (25). Further studies are necessary to investigate whether acute change in these parameters with LLTS may be a marker for predicting the chronic effect of LLTS on AF burden.
Limitations
AF burden was very low in some patients during the 2-week monitoring, reflecting the intermittent nature of AF. Nonetheless, the median AF burden observed in our study was consistent with that reported in a large cohort of paroxysmal AF patients with similar age and stroke risk undergoing 2-week monitoring (31). Moreover, in light of the less prominent role of the autonomic nervous system in later stages of AF (11,12), we intentionally included patients with early stages of the disease, as reflected by the relatively low AF burden. Continuous monitoring for AF burden with an implantable loop recorder may have increased the assessment of the true treatment effect. The optimal dosing and stimulation parameters for LLTS have not been systematically determined. Notably, the disappointing results of recent VNS trials in HF, despite the clear rationale for decreasing sympathovagal imbalance in this disease, highlight the notion that optimizing stimulation parameters is crucial to achieve a favorable effect with neuromodulation therapies (35,44). In addition, the shortest, effective duration of LLTS necessary to produce a long-lasting effect has not been determined. It is possible that the long-term effect of LLTS would have been even more pronounced had we chosen to use 2 or even 3 hours of stimulation daily. However, this approach would likely have had a negative impact on patient adherence. Therefore, we chose to only test the effect of LLTS applied for 1 hour daily. Although all patients reported that they followed the stimulation protocol as instructed, we cannot exclude the possibility that a patient may have searched for and become aware of the site of active stimulation, affecting the double blind nature of the study.
Conclusion
Our results support the emerging paradigm of noninvasive neuromodulation to treat AF. Further studies to optimize patient selection in order to maximize the efficacy of this novel noninvasive therapy are warranted.
Perspectives.
Competency in Medical Knowledge
In selected patients with paroxysmal atrial fibrillation, noninvasive neuromodulation using low level tragus stimulation decreases atrial fibrillation burden and inflammatory cytokines.
Translational outlook
Further research is needed to optimize patient selection in order to maximize the efficacy of this novel noninvasive therapy.
Acknowledgments
Funding source: Funded by NIH/NIGMS #8P20GM103447 and American Heart Association #15MCPRP2579000 to Stavros Stavrakis. Partial funding by NIH/NIGMS #1U54GM10493
Abbreviations
- AF
atrial fibrillation
- VNS
vagus nerve stimulation
- LLTS
low level tragus stimulation
- AV
atrioventricular
- HRV
heart rate variability
- TNF
tumor necrosis factor
- IL
interleukin
- LF
low frequency
- HF
high frequency
Footnotes
Disclosures: None
ClinicalTrials.gov Identifier: NCT02548754
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Uthman BM, Reichl AM, Dean JC et al. Effectiveness of vagus nerve stimulation in epilepsy patients: a 12-year observation. Neurology 2004;63:1124–6. [DOI] [PubMed] [Google Scholar]
- 2.Deuchars SA, Lall VK, Clancy J et al. Mechanisms underpinning sympathetic nervous activity and its modulation using transcutaneous vagus nerve stimulation. Exp Physiol 2018;103:326–331. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Frangos E, Ellrich J, Komisaruk BR. Non-invasive Access to the Vagus Nerve Central Projections via Electrical Stimulation of the External Ear: fMRI Evidence in Humans. Brain Stimul 2015;8:624–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Yakunina N, Kim SS, Nam EC. BOLD fMRI effects of transcutaneous vagus nerve stimulation in patients with chronic tinnitus. PLoS One 2018;13:e0207281. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Lehtimaki J, Hyvarinen P, Ylikoski M et al. Transcutaneous vagus nerve stimulation in tinnitus: a pilot study. Acta Otolaryngol 2013;133:378–82. [DOI] [PubMed] [Google Scholar]
- 6.Hamer HM, Bauer S. Lessons learned from transcutaneous vagus nerve stimulation (tVNS). Epilepsy Res 2019;153:83–84. [DOI] [PubMed] [Google Scholar]
- 7.Clancy JA, Mary DA, Witte KK, Greenwood JP, Deuchars SA, Deuchars J. Non-invasive Vagus Nerve Stimulation in Healthy Humans Reduces Sympathetic Nerve Activity. Brain Stimul 2014;7:871–7. [DOI] [PubMed] [Google Scholar]
- 8.Huston JM, Gallowitsch-Puerta M, Ochani M et al. Transcutaneous vagus nerve stimulation reduces serum high mobility group box 1 levels and improves survival in murine sepsis. Crit Care Med 2007;35:2762–8. [DOI] [PubMed] [Google Scholar]
- 9.Stavrakis S, Humphrey MB, Scherlag BJ et al. Low-level transcutaneous electrical vagus nerve stimulation suppresses atrial fibrillation. J Am Coll Cardiol 2015;65:867–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Yu L, Huang B, Po SS et al. Low-Level Tragus Stimulation for the Treatment of Ischemia and Reperfusion Injury in Patients With ST-Segment Elevation Myocardial Infarction: A Proof-of-Concept Study. JACC Cardiovasc Interv 2017;10:1511–1520. [DOI] [PubMed] [Google Scholar]
- 11.Stavrakis S, Nakagawa H, Po SS, Scherlag BJ, Lazzara R, Jackman WM. The role of the autonomic ganglia in atrial fibrillation. JACC Clin Electrophysiol 2015;1:1–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Chen PS, Chen LS, Fishbein MC, Lin SF, Nattel S. Role of the autonomic nervous system in atrial fibrillation: pathophysiology and therapy. Circ Res 2014;114:1500–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Li S, Scherlag BJ, Yu L et al. Low-level vagosympathetic stimulation: a paradox and potential new modality for the treatment of focal atrial fibrillation. Circ Arrhythm Electrophysiol 2009;2:645–51. [DOI] [PubMed] [Google Scholar]
- 14.Sha Y, Scherlag BJ, Yu L et al. Low-level right vagal stimulation: anticholinergic and antiadrenergic effects. J Cardiovasc Electrophysiol 2011;22:1147–53. [DOI] [PubMed] [Google Scholar]
- 15.Sheng X, Scherlag BJ, Yu L et al. Prevention and reversal of atrial fibrillation inducibility and autonomic remodeling by low-level vagosympathetic nerve stimulation. J Am Coll Cardiol 2011;57:563–71. [DOI] [PubMed] [Google Scholar]
- 16.Yu L, Scherlag BJ, Li S et al. Low-level vagosympathetic nerve stimulation inhibits atrial fibrillation inducibility: direct evidence by neural recordings from intrinsic cardiac ganglia. J Cardiovasc Electrophysiol 2011;22:455–63. [DOI] [PubMed] [Google Scholar]
- 17.Yu L, Scherlag BJ, Li S et al. Low-level transcutaneous electrical stimulation of the auricular branch of the vagus nerve: a noninvasive approach to treat the initial phase of atrial fibrillation. Heart Rhythm 2013;10:428–35. [DOI] [PubMed] [Google Scholar]
- 18.Shen MJ, Shinohara T, Park HW et al. Continuous low-level vagus nerve stimulation reduces stellate ganglion nerve activity and paroxysmal atrial tachyarrhythmias in ambulatory canines. Circulation 2011;123:2204–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Aviles RJ, Martin DO, Apperson-Hansen C et al. Inflammation as a risk factor for atrial fibrillation. Circulation 2003;108:3006–10. [DOI] [PubMed] [Google Scholar]
- 20.Hu YF, Chen YJ, Lin YJ, Chen SA. Inflammation and the pathogenesis of atrial fibrillation. Nat Rev Cardiol 2015;12:230–43. [DOI] [PubMed] [Google Scholar]
- 21.Chavan SS, Pavlov VA, Tracey KJ. Mechanisms and Therapeutic Relevance of Neuroimmune Communication. Immunity 2017;46:927–942. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Koopman FA, Chavan SS, Miljko S et al. Vagus nerve stimulation inhibits cytokine production and attenuates disease severity in rheumatoid arthritis. Proc Natl Acad Sci U S A 2016;113:8284–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Peuker ET, Filler TJ. The nerve supply of the human auricle. Clin Anat 2002;15:35–7. [DOI] [PubMed] [Google Scholar]
- 24.Heart rate variability: standards of measurement, physiological interpretation and clinical use. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology. Circulation 1996;93:1043–65. [PubMed] [Google Scholar]
- 25.Tran N, Asad Z, Elkholey K, Scherlag BJ, Po SS, Stavrakis S. Autonomic Neuromodulation Acutely Ameliorates Left Ventricular Strain in Humans. J Cardiovasc Transl Res 2019;12:221–230. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Freeman R Assessment of cardiovascular autonomic function. Clin Neurophysiol 2006;117:716–30. [DOI] [PubMed] [Google Scholar]
- 27.Rosenberg MA, Samuel M, Thosani A, Zimetbaum PJ. Use of a noninvasive continuous monitoring device in the management of atrial fibrillation: a pilot study. Pacing Clin Electrophysiol 2013;36:328–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Qian Y, Kang Z, Liu C, Li X. IL-17 signaling in host defense and inflammatory diseases. Cell Mol Immunol 2010;7:328–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Tanaka T, Narazaki M, Kishimoto T. IL-6 in inflammation, immunity, and disease. Cold Spring Harb Perspect Biol 2014;6:a016295. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Chen LY, Chung MK, Allen LA et al. Atrial Fibrillation Burden: Moving Beyond Atrial Fibrillation as a Binary Entity: A Scientific Statement From the American Heart Association. Circulation 2018;137:e623–e644. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Go AS, Reynolds K, Yang J et al. Association of Burden of Atrial Fibrillation With Risk of Ischemic Stroke in Adults With Paroxysmal Atrial Fibrillation: The KP-RHYTHM Study. JAMA Cardiol 2018;3:601–608. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Liu L, Nattel S. Differing sympathetic and vagal effects on atrial fibrillation in dogs: role of refractoriness heterogeneity. Am J Physiol 1997;273:H805–16. [DOI] [PubMed] [Google Scholar]
- 33.Andrus EC, Carter EP, With an Appendix by Harold A. Wheeler PD. The Refractory Period of the Normally-Beating Dog’s Auricle; with a Note on the Occurrence of Auricular Fibrillation Following a Single Stimulus. J Exp Med 1930;51:357–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Zhang Y, Ilsar I, Sabbah HN, Ben David T, Mazgalev TN. Relationship between right cervical vagus nerve stimulation and atrial fibrillation inducibility: therapeutic intensities do not increase arrhythmogenesis. Heart Rhythm 2009;6:244–50. [DOI] [PubMed] [Google Scholar]
- 35.Ardell JL, Nier H, Hammer M et al. Defining the neural fulcrum for chronic vagus nerve stimulation: implications for integrated cardiac control. J Physiol 2017;595:6887–6903. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Chinda K, Tsai WC, Chan YH et al. Intermittent left cervical vagal nerve stimulation damages the stellate ganglia and reduces the ventricular rate during sustained atrial fibrillation in ambulatory dogs. Heart Rhythm 2016;13:771–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Verlinden TJ, Rijkers K, Hoogland G, Herrler A. Morphology of the human cervical vagus nerve: implications for vagus nerve stimulation treatment. Acta Neurol Scand 2016;133:173–82. [DOI] [PubMed] [Google Scholar]
- 38.Stavrakis S, Humphrey MB, Scherlag B et al. Low-Level Vagus Nerve Stimulation Suppresses Post-Operative Atrial Fibrillation and Inflammation: A Randomized Study. JACC Clin Electrophysiol 2017;3:929–938. [DOI] [PubMed] [Google Scholar]
- 39.Pavlov VA, Chavan SS, Tracey KJ. Molecular and Functional Neuroscience in Immunity. Annu Rev Immunol 2018;36:783–812. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Gomes-Osman J, Indahlastari A, Fried PJ et al. Non-invasive Brain Stimulation: Probing Intracortical Circuits and Improving Cognition in the Aging Brain. Front Aging Neurosci 2018;10:177. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Ashton JL, Burton RAB, Bub G, Smaill BH, Montgomery JM. Synaptic Plasticity in Cardiac Innervation and Its Potential Role in Atrial Fibrillation. Front Physiol 2018;9:240. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Huffman WJ, Subramaniyan S, Rodriguiz RM, Wetsel WC, Grill WM, Terrando N. Modulation of neuroinflammation and memory dysfunction using percutaneous vagus nerve stimulation in mice. Brain Stimul 2019;12:19–29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Lombardi F, Tarricone D, Tundo F, Colombo F, Belletti S, Fiorentini C. Autonomic nervous system and paroxysmal atrial fibrillation: a study based on the analysis of RR interval changes before, during and after paroxysmal atrial fibrillation. Eur Heart J 2004;25:1242–8. [DOI] [PubMed] [Google Scholar]
- 44.Byku M, Mann DL. Neuromodulation of the Failing Heart: Lost in Translation? JACC Basic Transl Sci 2016;1:95–106. [DOI] [PMC free article] [PubMed] [Google Scholar]