Low-Level Transcutaneous Electrical Vagus Nerve Stimulation Suppresses Atrial FibrillationFree Access

Study population

Patients with paroxysmal AF referred to the electrophysiological laboratory for AF ablation were eligible for enrollment. Patients were excluded if they had left ventricular dysfunction (defined as left ventricular ejection fraction <40%), significant valvular disorder (i.e., prosthetic valve or hemodynamically relevant valvular diseases), or recent (<6 months) stroke. All patients underwent baseline transthoracic and transesophageal echocardiography within 2 days prior to ablation. Antiarrhythmic medications were discontinued 5 days before the procedure except amiodarone, which was discontinued for at least 6 weeks. All patients received pre-procedure anticoagulation with either warfarin, to maintain an international normalized ratio between 2.0 and 3.0, or a new oral anticoagulant agent. Warfarin was continued throughout the procedure, whereas new oral anticoagulant agents were discontinued for 24 h pre-procedure. The study was approved by the Institutional Review Board of the University of Oklahoma Health Sciences Center. After providing written informed consent, patients were randomized to either LLTS for 1 h or sham LLTS (control group).

Experimental protocol

Our laboratory routinely performs the ablation procedure under general anesthesia, using desflurane, vecuronium, and propofol. Before induction of general anesthesia, incremental voltages were applied to the right tragus (20 Hz, 1-ms square wave) through a Grass S88 stimulator (Natus Neurology Incorporated, Warwick, Rhode Island) connected to a stimulus isolation unit, until the patient experienced discomfort. The discomfort threshold was defined as the lowest voltage that resulted in any discomfort at the stimulation site. After induction of general anesthesia, sheaths were inserted in both femoral veins and coronary sinus. An octapolar catheter was positioned at the His bundle to monitor the atrial-His (AH) interval and a duodecapolar catheter was positioned in the coronary sinus for atrial pacing. In the baseline state, the atrial effective refractory period (AERP) was measured using programmed stimulation (S1-S1 = 600 ms) at 10× diastolic threshold from the right atrial appendage and the distal coronary sinus electrodes using 10 ms S1-S2 decrements. Subsequently, burst atrial pacing (cycle length decreasing from 250 ms to 200 ms) was delivered through the coronary sinus catheter to induce AF. This process was repeated until AF was induced. AF inducibility, defined as the number of attempts required to induce AF, was recorded, as was the duration of AF induced by burst atrial pacing. The AF atrial cycle length was determined manually by averaging 30 consecutive beats at a paper speed 100 mm/s, using screen calipers, as previously described (16). Additionally, interelectrogram intervals of <100 ms and continuous electrical activity were counted as a single interval (16).

After acquiring the baseline parameters, LLTS or sham treatment was initiated, regardless of AF continuation (Figure 1A). Stimulation of the right ear tragus was accomplished by attaching a flat metal clip onto the tragus, which served as the cathode, with another clip at an adjacent site for the anode (Figure 1B). Incremental voltages were applied to the tragus (20 Hz, 1-ms duration, square wave) using a Grass S88 stimulator until slowing of the sinus rate or AV nodal conduction (AH prolongation) was achieved (Figure 2). As soon as the sinus rate decreased and/or the AH interval prolonged, stimulation was turned off to avoid excessive bradycardia and/or heart block. The lowest voltage to slow the sinus rate or prolong the AH interval was defined as the threshold for setting the LLTS in each patient. LLTS was set at 50% below the voltage required to slow the sinus rate or prolong the AH interval. In the experimental group, LLTS was applied continuously for 1 h; in the control group, the threshold was measured but no LLTS was delivered. To prevent inadvertent supra-threshold vagus nerve stimulation, the sinus rate and AH interval were monitored continuously to ensure that they were not altered by LLTS. The bispectral index was continuously monitored, too, during the procedure. The bispectral index was maintained between 40 and 60 during the maintenance phase of general anesthesia. During the 1-h period of LLTS or sham stimulation, transseptal puncture and electroanatomical mapping, but not ablation, were performed. If the patient remained in AF at the end of the hour, cardioversion was performed to restore sinus rhythm. Within 5 min following the end of the LLTS application, programmed stimulation and burst atrial pacing were repeated using the same protocol as before, to measure the AERP and induce AF, respectively. Figure 1A illustrates the study protocol.

During baseline sinus rhythm, 5 ml of blood were simultaneously obtained from the peripheral femoral venous sheath (systemic sample) and the coronary sinus sheath (cardiac sample). Simultaneous cardiac and systemic samples were drawn again following 1 h of LLTS or sham treatment (Figure 1A). For each blood draw, the first 10 ml were discarded, and the blood was immediately transferred into tubes containing no anticoagulant agent. After incubation in an upright position at room temperature for 45 min to allow clotting, the blood was centrifuged for 15 min at 2,000 g to obtain serum. Patients’ serum was saved frozen at –80° C and processed in batches of 6 to 8. The investigators performing the biomarker assays were blinded to group assignment. Pro-inflammatory cytokines, including tumor necrosis factor (TNF)–alpha, C-reactive protein (CRP), interleukin (IL)-6, and IL-10 were measured using a commercially available assay analyzed on a flow cytometer (multiplex assay; R&D Systems, Minneapolis, Minnesota). All immunoassays were run in duplicate and read according to manufacturer’s instructions.

Statistical analysis

Data are presented as mean ± SD or percentages for continuous and categorical variables, respectively. Comparisons in baseline characteristics between groups were performed using Student t test or chi-square test for continuous and categorical variables, respectively. The differences in burst atrial pacing-induced AF duration, number of attempts, AERP, and inflammatory cytokines before and after 1 h of LLTS or sham treatment were compared between the groups with 2-way analysis of variance. The analysis of variance modeling assumptions were evaluated by plotting the residuals by the predicted values (for the constant variance assumption) and comparing the normal QQ plot with the QQ plot of the residuals (for the normality assumption). The assumptions of constant variance and normality were reasonable for all analysis of variance models based on the residual plots. The difference in AF recurrence at follow-up between the 2 groups was evaluated using the log-rank test. Statistical significance was declared at p < 0.05. All statistical analyses were performed using SAS software (version 9.2, SAS Institute, Inc., Cary, North Carolina).

Based on experimental data (11), the present study was powered to detect a 50% reduction in the duration of burst atrial pacing-induced AF after LLTS versus in control subjects. A sample size of 40 patients (20 in each group) would provide at least 80% power to detect this difference, at a 2-sided significance alpha level of 0.05.

Antiarrhythmic effects of LLTS

CANS is a neural network that controls the heart’s vascular, contractile, and electrophysiological functions. The neural signals in this network are integrated and processed at the ganglionated plexi (GP), which may contain hundreds to thousands of autonomic neurons and are typically located in the epicardial fat pads (18,19). Hyperactivity of the major atrial GP plays a critical role in AF initiation and maintenance, whereas ablating these GP provides additional benefits to the standard circumferential pulmonary vein isolation procedure (20). Previous experimental studies demonstrated that LLVNS and LLTS are antiadrenergic and anticholinergic (10,11,14,15). When LLTS was delivered to the cranial end of the transected vagal trunk, all the antiarrhythmic effects of LLTS disappeared, whereas LLVNS applied at the distal end of the transected vagal trunks retained its antiarrhythmic effects (9,15). Although LLVNS and LLTS probably activate both the afferent and efferent vagal fibers, these observations suggest that the efferent fibers of the vagus nerves are crucial to the antiarrhythmic effects. The seemingly “paradoxical” anticholinergic effects of LLVNS and LLTS, which require efferent vagal fibers for their actions, are mediated by inhibiting the activity of the major atrial GP, as evidenced by direct neural recordings within canine GP (10,12,13). Likewise, LLVNS decreases the activity of the left stellate ganglion, the gateway of sympathetic innervation to the heart (14). In the present study, similar mechanisms are likely underlying the antiarrhythmic effects of LLTS. The anticholinergic effects prolong the AERP and the antiadrenergic effects suppress the myocyte calcium transients, leading to the inhibition of both re-entry and triggered firing (21). Moreover, suppressing CANS activity also breaks the vicious cycle formed by hyperactivity of the CANS and atrial remodeling (13).

Our results can be interpreted in light of the currently accepted mechanism for AF initiation and maintenance, which requires both a trigger and a substrate (3). The trigger for paroxysmal AF often lies in the pulmonary vein myocardium, whereas atrial and autonomic remodeling provides the substrate for AF to sustain once initiated (3). We have provided evidence that LLTS modified the substrate for AF by prolonging the AERP and AF cycle length. Although this study offers no direct evidence that LLTS could inhibit triggered activity, this effect has been demonstrated in animal studies, in which LLVNS inhibited induction of rapid firing and AF from the pulmonary veins (9,10). Moreover, the antiarrhythmic effects of LLVNS and LLTS were very similar in various animal models of AF, including rapid atrial pacing, acetylcholine-induced AF, and high-frequency stimulation-induced triggered firing (9,13,15). In this study, we used widely accepted parameters, such as AF duration, AF cycle length, and AERP. Although AF inducibility lacks accuracy and its clinical significance is debatable (22), the favorable effects of LLTS on other endpoints, such as AF duration and AERP, lend credence to our results.

Anti-inflammatory effect of LLTS

Our findings’ significance is highlighted by substantial evidence linking AF with inflammation (23–26). In a large population-based study, CRP levels predicted both the presence of AF at baseline and AF development during follow-up, even after adjusting for other known cardiovascular risk factors (23). Risk of AF was also progressively higher with increasing CRP quartiles (23). In another case-control study, TNF-α levels were significantly increased in patients with AF compared with control patients, and there was a graded increase in TNF-α among patients with paroxysmal, persistent, and permanent AF (25). In the present study, it is noteworthy that blood levels of inflammatory cytokines in both the LLTS and control groups were elevated at baseline (27) and similar to those reported in large population-based cohorts and case-control studies (23,25). Moreover, the magnitude of decrease in TNF-α and CRP levels by LLTS is consistent with the difference between control and pre-rheumatoid arthritis patients (27) as well as for patients with active versus inactive inflammatory diseases (17), suggesting a significant biological effect. Extrapolating from the post-myocardial infarction literature, a decrease in CRP and/or TNF-α of a similar magnitude to that observed in our study, if sustained, would predict a lower incidence of recurrent cardiovascular events (28,29). Nonetheless, given that the sustainability of inflammatory cytokine reduction and the long-term effects of LLTS remain unknown, these results should be interpreted cautiously.

More than 80% of the nerve fibers in the vagal trunks are afferent vagal fibers carrying neural input from nearly the entire body, and the tonic activity of the vagus nerves is essential to maintain immune homeostasis (30). Our results are consistent with the recently recognized anti-inflammatory properties of the vagus nerve (30,31). Stimulation of the cholinergic anti-inflammatory pathway, which is composed of efferent vagus nerve signals, leads to acetylcholine-dependent activation of the alpha-7 nicotinic acetylcholine receptor subunit on monocytes and macrophages, resulting in reduced production of the inflammatory cytokines TNF-α, IL-1β, and IL-6, but not IL-10 (31,32). Reduction in inflammatory markers with vagus nerve stimulation was also observed in a canine model of HF (33). The finding that LLTS suppresses systemic but not coronary sinus levels of TNF-α and CRP suggests that the reduction of inflammatory cytokines was the result of activation of the cholinergic anti-inflammatory pathway and not just a local cardiac vagal effect. The cholinergic anti-inflammatory pathway involves efferent vagus nerve fibers, traveling to the spleen through the celiac ganglion and the splenic nerve to inhibit inflammatory responses (30). If the anti-inflammatory effects observed resulted from a local cardiac effect, the coronary sinus cytokine levels would have decreased before the systemic cytokine levels did. Importantly, we have provided evidence that the cholinergic anti-inflammatory pathway can be activated without heart rate slowing, consistent with a previous experimental study, which demonstrated that transcutaneous vagus nerve stimulation inhibited inflammatory cytokine production in mice with endotoxemia (34).

Clinical implications

In this study, we demonstrated that neuromodulation by LLTS is a promising, noninvasive therapy to treat AF and AF-related inflammation. Given the recently demonstrated beneficial effects of vagus nerve stimulation in patients with HF (8,35), neuromodulation may provide a means of treating patients with both HF and AF with a single treatment modality. Moreover, these results may provide the basis to potentially expand the therapeutic targets of this treatment modality into other inflammatory conditions, including rheumatoid arthritis and systemic lupus erythematosus (30). In the present study, the stimulation voltage (50% below the cardiac threshold) was lower than the discomfort threshold in the average patient. Importantly, LLTS with the stimulation strength 80% below the cardiac threshold exerted similar antiarrhythmic effects as 50% below the cardiac threshold (15), indicating that this noninvasive approach can be tolerated by ambulatory patients. Moreover, transcutaneous stimulation of the auricular branch of the vagus nerve in patients with epilepsy was safe and tolerable in a recent pilot study (36).

Study limitations

The optimal stimulation parameters for LLTS have not been determined. In this study we used 20 Hz based on our previous animal experience (9–12). Further studies are necessary to evaluate the shortest time period and the lowest stimulation strength of LLTS that would have a beneficial effect. Intermittent stimulation may be more efficacious than continuous stimulation, based on recent experimental evidence (37). Although any LLTS-caused discomfort may have been masked by general anesthesia, the average stimulation level (<50% lower than the voltage threshold) was lower than the average discomfort threshold. In every patient, the discomfort threshold was higher than the LLTS voltage level. Thus, it is unlikely that discomfort is a major contributing factor to the effect of LLTS. There was a trend toward an increased prevalence of diabetes in the control group, which appears to be clinically significant. However, the 2 groups were otherwise well matched; thus, it is unlikely that these small differences influenced the results. In this study, we examined the acute effects of LLTS; further trials are warranted to evaluate the long-term clinical importance of our findings.