In animal models, VM-AF was defined as AF caused by stimulation of the vagus nerve with observed atrioventricular block, asystolic periods, sinus bradycardia and an increase in heart rate variability ^14,15. In humans, VM-AF was defined as paroxysmal AF that occurs with predominant vagal activation, such as, during sleep, after eating a big meal, or in relationship to other recognized vagal triggers, usually preceded by bradycardia ^8,15,16. Patients may have no underlying heart disease or other systemic disorders that could explain the AF ¹⁷.

The current study was conducted in accordance with the recent Preferred Reporting Items for Systematic Reviews and Meta-Analyses Statement (18, 19). Relevant articles were obtained from a search of EMBASE and MEDLINE databases from inception through 2020. We performed an initial search on February 8, 2020 including the terms “vagal”, “vagal nerve”, “vagus”, “vagus nerve”, and “atrial fibrillation”. We performed a second search on April 26, 2020 using the term “parasympathetic”, “parasympathetic stimulation”, and "atrial fibrillation” (full search strategy and search terms in Figure 2) then the first and second search databases were combined. Only full articles in English were included. The search strategy was done according to published recommendations. To find additional eligible studies, review articles were also screened. A manual search for additional pertinent studies using references from retrieved articles was also completed.

Figure 2. Search methodology and selection process.

---

---

Study eligibility of drug therapy for VM-AF was independently determined by two investigators (PR and JK) and differences were resolved by mutual consensus. The eligibility criteria included the following:

(1) Experimental studies, cohort studies (prospective or retrospective), case-control studies, cross-sectional studies, and randomized control trials reporting the use of antiarrhythmic drugs (AADs) on VM-AF were included. Studies were excluded if intervention, such as ablation, was performed on subjects.

(2) Studies describing the definition of VM-AF or induction protocol of the vagus nerve stimulation, medication name, dose of medication, and outcome were also included in the review.

A standardized data collection form was used to obtain the following information from each study: medication name; name of first author; year of publication; subject model; induction of VM-AF; diagnostic criteria of VM-AF; medication dose; proposed drug mechanisms; and outcome.

Two investigators (PR and JK) independently performed data extraction to ensure accuracy. Any data discrepancy was resolved by reviewing the primary data from the original articles.

Details of the selection flowchart are summarized in Figure 2. Our initial search on February 8, 2020 yielded 2,133 potentially relevant articles (1,498 articles from EMBASE and 635 articles from MEDLINE). The second search on April 26, 2020 further identified 343 potentially relevant articles (224 articles from EMBASE and 119 articles from MEDLINE). After the exclusion of 996 duplicated articles, 1,480 articles underwent title and abstract review. Of the total, 1,273 articles were excluded at this stage as it was conducted in non-VM-AF. This left 207 abstracts for full-length article review; 126 studies were further excluded as these studies did not evaluate pharmacological management or they did not report any AAD use. An additional 36 articles were excluded as they were review articles that referred to the original articles; 28 abstracts were excluded as full articles were not available.

A total of 14 studies ^17,20-32 and 3 practice guidelines ^16,33,34 from 1986-2017 were included Table 1. Only two clinical investigations evaluated the effectiveness of AADs in human subjects with VM-AF ^17,30. Clinical characteristics of patients with vagal-mediated and adrenergic-mediated AF in the included studies are summarized in Table 2.

The main antiarrhythmic mechanism of Class III AADs is to prolong action potential duration by blocking the rapid component of the delayed rectifier current (IKr) which is not specific for VM-AF ²⁰. The first class III AAD evaluated in an animal model was bretylium tosylate. In 1986, Goldberger et al ²³ successfully developed a VM-AF canine model; 100% of AF was terminated by bretylium tosylate. Bretylium has an autonomic effect as it prevents sympathetic neurotransmitter release from nerve terminals ^35,36.

In 1998, Hayashi et al ²¹ reported MS-551, a class III AAD structurally similar to pyrimidinedione, as an effective AAD for VM-AF in 6/8 canine models. MS-551 inhibited IK and decreased the transient outward current (Ito) and the IK1 current ³⁷. However, MS-551 has not been developed as a clinical AAD.

Fedorov et al ²⁰ reported that nibentan is highly effective in reducing the number of reentrant wavelets in VM-AF in a canine model. This study showed significant dose-dependent increases in AERP (55±9%, 82±12%, and 90±6% at the dose of 0.063, 0.125, and 0.250 mg/kg; p < 0.01) and wavelength (47±7%, 68±12%, and 72±4%; at the dose of 0.063, 0.125, and 0.250 mg/kg; p< 0.01). Nibentan has not been developed as an AAD for clinical use.

Tedisamil was investigated to treat VM-AF ²⁶. Tedisamil was effective in terminating 11/11 episodes of VM-AF in a canine model. It also prevented re-induction of AF within 30 minutes of drug administration. The fibrillation cycle length increased before conversion (112±25 ms to 232±15 ms; p<0.01) suggesting lengthening of the AERP ²⁶. Tedisamil has not been developed as an AAD for clinical use.

Sotalol has class III antiarrhythmic and beta-blocker effects. Yesil et al ³⁰ compared sotalol in patients with presumed VM-AF to those with purported adrenergic AF. Only 36% of the patients with VM-AF returned to sinus rhythm; however, 71% of adrenergic AF patients were successfully terminated with sotalol. The study concluded that sotalol is more effective in treating adrenergic AF due to its combined class III and beta blocking effect ^30,33,38.

IKAch inhibition is a specific therapeutic target and strategy for VM-AF ^22,27,28. Tertiapin was the first IKAch selective inhibitor tested in a canine model. Hashimoto et al ²² reported that tertiapin prolongs the AERP (134±9 to 162±7 ms; p<0.05) without affecting ventricular repolarization (171±8 to 172±3 ms) during vagal nerve stimulation and terminates AF with 100% efficacy. Moreover, tertiapin did not affect PR, QRS and corrected QT intervals (raw data was not presented by the authors) ²².

NIP-151 is an IKAch selective inhibitor in an in vitro model (single whole cell voltage clamp experiment) with minimal effect on IKr ²⁸. In a canine model, NIP-151 dose-dependently prolonged AERP (28±5, 32±7, and 33±7 ms at the dose of 15,30, and 75 μg/kg/min, respectively; p<0.05) but did not affect the ventricular effective refractory period (raw data were not shown). Given its minimal effect on IKr, there were no significant electrocardiographic changes as opposed to dofetilide (IKr blocker), which causes QT prolongation (23±4 and 32±2 ms at the dose of 0.3 and 3μg/kg/min, respectively) ²⁸.

NTC-801, another IKAch selective inhibitor, was studied in guinea pig atrial cells and a VM-AF rapid pacing canine model ²⁷. NTC-801 selectively and dose dependently prolonged AERP (21.3±3.3, and 30.0±2.6 ms at the dose of 1 and 3 μg/kg/min; p=0.0204 and p=0.0006, respectively) but was frequency-independent (38.7±7.9, 38.0±7.4, and 35.7±7.1 ms at basic cycle length of 300, 250, and 200 ms; p=0.0038, p=0.0038, and p=0.0037, respectively) without affecting ventricular effective refractory period (raw data was not presented by the authors) ²⁷.

Plasmid DNA vectors expressing Gαi2 C-terminal peptide (Gαictp) in combination with Gαo1 C-terminal peptide (Gαoctp) was injected in the posterior left atrium to selectively disrupt M2R/Gαi2 stimulation and impede the Gαi2/o1βγ uncoupling signal in a canine model ²⁵. Vagal-induced AERP shortening and VM-AF inducibility index (raw data were not shown) were nearly eliminated in atria receiving a combination of Gαi2ctp and Gαo1ctp plasmid DNA vectors compared to Gαi2ctp alone (2.8±1.5 versus 19.5±5.0 msec, respectively; P<0.01) ²⁵. Lou et al ²⁴ reported similar results after Gαictp injection to the anterior atrial wall of beagle dogs.

Botulinum toxin injection in epicardial fat pads rich in autonomic ganglia can temporarily suppress VM-AF inducibility for approximately 1 week in a canine model compared to control (20±11% versus 58±14%, respectively; p=0.025) before the effects wear off in week 2 (30±21% versus 58±14%, respectively; p=0.11), and week 3 (56±13% versus 67±14%, respectively; p=0.41) ³². The mechanism may have been associated with parasympathetic autonomic modification by ganglionic block that reduced AERP dispersion ³².

According to the 2014 AHA/ACC/HRS guidelines, clinical characteristics of AF are classified into categories by duration of episodes including paroxysmal, persistent, long-standing persistent, and permanent AF. These definitions for AF pattern have been useful for the management of AF but are limited by the lack of correlation to their underlying pathogenic mechanisms ³³.

The original observational study of 18 VM-AF patients by Coumel et al ¹ showed a predominantly male population with a ratio of men to women of 4:1. These patients were younger; the ages of first symptom onset were 25-60 years. The frequency of AF episodes was variable from patient to patient, ranging from sporadic events to recurrent daily events, and lasting from a few minutes to several hours. The AF episodes usually occurred during the night, after the digestive period of a large meal when vagal tone surged, and often ended in the morning. Interestingly, many patients in this study reported that AF episodes were preventable by exercising, but the resting or relaxation period following emotional distress was often followed by onset of AF episodes and symptoms ¹Table 2.

Makrides et al ⁴⁰ reported a case with transient ST elevation in the inferolateral leads in VM-AF that resolved after cardioversion. The mechanism was hypothesized to be from high parasympathetic tone. Increased vagal stimulation may augment early repolarization in the inferolateral leads causing transient ST elevation ^39,40. Other plausible explanation such as coronary vasospasm cannot be excluded. VM-AF was also reported in ischemic stroke due to the enhanced central vagal activity in basal ganglion infarction ⁴¹. Cardioinhibitory neurocardiogenic syncope, which was aggravated by nausea, was also found to co-present with VM-AF ⁴².

In contrast to VM-AF, adrenergic AF is caused by a high adrenergic state (high sympathetic tone), such as, exercise, is associated with emotion, daytime hours, and higher heart rate. It can also occur in older patients with structurally abnormal hearts, and improves with beta-blocker use. The differences in clinical characteristics between VM-AF and adrenergic AF are summarized in Table 2^{1,8,14,33,43-46}.

The identification of triggers for paroxysmal AF is based on clinical autonomic characteristics, is qualitative with expected overlaps, and is not likely due to isolated vagal activation. Nevertheless, the characteristics clinically distinguish a select group of young patients with a clear association between episodes of AF and vagal activation. This distinction warrants consideration of novel therapeutic options for these young patients who are otherwise healthy.

In two small studies of AF patients who were referred for pulmonary vein isolation, VM-AF and adrenergic AF prevalence was approximately 12-27% and 7-16%, respectively ^8,47. In the larger Euro Heart Survey of 1,517 patients with paroxysmal AF, the prevalence of VM-AF was 6%, and the prevalence of adrenergic AF was 15%. ³⁸. This VM-AF prevalence estimated from selected populations are likely an over-estimation from the general population because patients with AF from the general population are more likely to be elderly with comorbidities and a persistent pattern of AF and are less likely to be referred for ablation. The true prevalence of VM-AF is still unknown.

The first large descriptive observational study of VM-AF was reported by the Euro Heart Survey Registry. In contrast to Coumel’s definition, VM-AF was defined as AF that occurred after meals and/or was present during the night only without evidence of any adrenergic triggers regardless of structural heart disease. In the Euro Heart Survey Registry, the prevalence of VM-AF in paroxysmal AF, age, body mass index, and other characteristics were similar between men and women. Of the total, 19% of VM-AF patients experienced progression from paroxysmal AF to persistent AF ³⁸.

VM-AF is also believed to play a role in endurance athletes presenting with AF. The incidence of AF in male athletes was 1.8-8.8 fold higher than non-active men. Endurance athletes are known to have elevated vagal tone as well as dilated atria causing stretch to the atrial wall. Stretching of the atrial wall activates the stretch receptors, shortening the action potential and AERP, thus precipitating AF (48). However, the prevalence of VM-AF has not been well established in the athletic population.

The intrinsic ganglia and nerves (ganglionated plexi) are mostly found in the atria especially at the pulmonary vein to left atrial junction, superior vena cava, right atrial junction, and atrioventricular node Figure 1A. These ganglionated plexi are innervated by nerve fibers derived from the parasympathetic nervous system and sympathetic nervous system Figure 1A. The complex sympathetic and vagal interactions at the cardiac ganglion, cellular and subcellular levels are shown schematically Figure 1B and Figure 1C.

Figure 1. A) Extrinsic and intrinsic cardiac autonomic innervation; 2B) Autonomic innervation at the cardiac myocyte; C) The cellular/subcellular channel and signal transduction pathways of substrate components of vagally-mediated AF (VM-AF) and mechanism of drug therapy for VM-AF. (IK1: Inward rectifier potassium channel; Iks : Slow delayed rectifier channel; LTCC: L-type calcium channel; FCC: Funny current channel; IKAch: Acetylcholine-activated potassium channel; RyR2: Ryanodine receptor 2 receptor; M2R : Type-2 muscarinic cholinergic receptor; SR: Sarcroplasmic reticulum).

---

---

Sympathetic activation leads to focal ectopic firing through enhanced automaticity, early after-depolarization (EAD), or delayed after-depolarization (DAD)¹⁰. Enhanced automaticity is mediated by β-adrenergic activation of the funny current (If) ⁴⁹ and an increased slope of phase-4 depolarization ⁵⁰. Phase-3 EAD is associated with lengthening of the action potential duration, which is augmented by β-adrenergic activation through the enhanced L-type calcium channel current (ICaL) ⁵¹. Beta-adrenergic activation also increases the opening of intracellular calcium and ryanodine receptor 2 (RyR2) through protein kinase A/calcium calmodulin-dependent protein kinase II (PKA/CaMKII) activation, followed by diastolic RyR2 calcium leakage mediating DAD ¹⁰. Parasympathetic stimulation activates the acetylcholine-dependent potassium current (IKAch) causing shortening of action potential duration thereby increasing the propensity for inhomogeneity of refractoriness and reentry mediated AF substrate ⁵².

In summary, vagal nerve stimulation enhances the spatially heterogeneous action potential duration and refractory period, which creates a reentry substrate for AF. Together with adrenergic activity, increased automaticity, EAD and DAD provide a plausible triggering mechanism for AF. The dynamics of sympatho-vagal interaction are possible underlying mechanisms of initiation and maintenance of VM-AF ⁵³.

Key pharmacological targets for VM-AF are shown in Figure 1C. Disopyramide may be useful for VM-AF although clinical data are limited. It is the only drug clinically available in the U.S. with a combined Class IA antiarrhythmic and anticholinergic effect. Cibenzoline is another Class IA AAD with anticholinergic effect but not available in the U.S. The mechanisms of anticholinergic activities between cibenzoline and disopyramide are different. Both cibenzoline and disopyramide have an antimuscarinic effect on atrial M2 muscarinic receptors; however, cibenzoline also has a direct blocking effect on IKAch channel ⁵⁴. Theoretically, cibenzoline could diminish vagal tone more than disopyramide with less side effects given weaker anticholinergic actions on type-1 muscarinic cholinergic receptors in the brain and type-3 muscarinic cholinergic receptors in skeletal muscle and glands than type-2 muscarinic cholinergic receptors in the heart ¹⁷.

The use of Class IC and Class III AADs for VM-AF is based on the same principles of antiarrhythmic effects in patients with AF as in the general population. Flecainide, a Class IC AAD, can be considered for use in patients with VM-AF. Sotalol, a Class III AAD with beta-blocking effect, could be considered in selected patients with VM-AF without baseline bradycardia. Observations from three IKAch selective inhibitor studies suggest that IKAch selective inhibitors could be future therapeutic options for VM-AF without the associated risk of ventricular arrhythmia ^22,27,28.