Heart failure (HF) causes a progressive decline in health and physical capacity, and is the leading cause of hospitalization in the elderly (Dharmarajan and Rich, 2017; Kolte et al., 2017). When symptoms of HF persist despite guideline directed medical therapy (GDMT), patients experience progressive fatigue, dyspnea, inability to perform activities of daily living, successive hospitalizations for symptom and co-morbidity management, and ultimately death.

The autonomic nervous system reacts reflexively to HF by stimulating the cardiovascular system in attempt to maintain cardiac output and blood pressure. This reflexive stimulation is maladaptive to a failing heart, however, and ultimately causes HF to progress. While drugs, devices, and cardiac rehabilitation have improved symptoms and survival rates by the progression of HF, HF hospitalization rates and mortality remain high. Although survival from ischemic heart disease has improved, HF prevalence has continued to rise, mainly due to an increase in the prevalence of cardiovascular (CV) risk factors and an increasingly aging population.

In a recently published population-based cohort study of over 300,000 patients, HF was listed on 42% of death certificates and considered to be the cause of 7% of all deaths (Coleman et al., 2011; Brenner et al., 2012; Siegel et al., 2012). The life expectancy of heart failure patients has been less than the survival rates for bowel, breast, and prostate cancer (Coleman et al., 2011; Brenner et al., 2012; Siegel et al., 2012). Survival rates for patients with HF have increased only gradually over the past two decades to 76% at 1 year, 46% at 5 years, and 25% at 10 years (Taylor et al., 2019). About a quarter of patients who are admitted to hospital for HF are readmitted within a month and up to two-thirds within a year, usually for recurrence of HF (Cowie et al., 2014). Individuals who are readmitted with worsening or recurrent symptoms of HF are at a high risk of progressive decline (Neumann et al., 2009).

Existing medical treatment for heart failure and reduced LVEF (HFrEF) using guideline directed medical therapy (GDMT) is directed toward counteracting harmful neurohormonal activation using a combination of beta-blockade, angiotensin-converting enzyme inhibition, angiotensin-receptor blockade in combination with neprilysin inhibition, and mineralocorticoid receptor antagonism.

HF markedly affects patients’ quality of life. Fear, anxiety, and depression are common, and work, travel, social, and leisure activities are affected. Despite the establishment of new prognosis-improving strategies for treatment, patients who continue to have reduced physical capacity, deteriorated quality of life, and hospitalizations for worsening HF desire additional remedies (Jeon et al., 2010; Ågren et al., 2011). The emotional costs are also high for caregivers looking after a member of the family with heart failure (Strömberg, 2013).

In patients with persistent symptoms of advanced HF despite GDMT, electrical stimulation to modulate biological targets within the autonomic nervous system to achieve more physiologic balance between sympathetic and parasympathetic activity has the potential to delay HF progression. Neuromodulation can be achieved with the delivery of autonomic regulation therapy (ART) by stimulating the cervical vagus nerve, which is readily accessible in the neck and provides access to the appropriate central and peripheral nervous system targets for neuromodulation (Salavatian et al., 2017; Hanna et al., 2018).

We will describe the pathophysiology of the autonomic nervous system in heart failure, the progress that has been made in our approach to optimizing VNS delivery for maximal benefit, and a pathway toward clarifying the role of vagus nerve stimulation (VNS) in managing patients with heart failure and reduced left ventricular ejection fraction (LVEF (HFrEF).

In health, homeostasis is maintained through afferent and efferent sympathetic and parasympathetic interactions that occur throughout the neurocardiac reflex hierarchy comprising the brain, peripheral neural circuits, and the intrinsic cardiac network in the cardiac fat pads (Figure 1). Nervous system activities are integrated within and between the peripheral ganglia and the central nervous system to provide networked control of the heart. At the nerve-myocyte interface, normal parasympathetic activity limits norepinephrine release from sympathetic nerve terminals, engendering cellular resistance to stress (Salavatian et al., 2017; Hanna et al., 2018).

With progressive HF, there is a generalized increase in sympathetic activity and a concurrent decrease in parasympathetic tone throughout the neurocardiac network (Hauptman and Mann, 2011; Kishi, 2012; Florea and Cohn, 2014). The biological consequences of these changes are summarized in Table 1.

There have been more than three decades of clinical experience with VNS to modulate afferent and efferent neurological pathways as an approved therapy for drug-refractory epilepsy and treatment-resistant depression. In-vitro and in vivo investigations of HF during this time have included explorations of right versus left versus bilateral stimulation, electrode and waveform configuration, efferent versus afferent versus bidirectional stimulation, continuous versus cyclic stimulation, duration of stimulation on/off cycles and number of pulses delivered per cycle, stimulation frequency, open-loop stimulation independent of any trigger, closed-loop stimulation with a physiologic trigger such as ventricular depolarization, current amplitude, effects of VNS intensity on heart rate (HR), and safety and tolerance of VNS titration (De Ferrari and Schwartz, 2011; Ardell et al., 2017). These have provided major contributions to our understanding of the potential use of VNS for the treatment of patients with HFrEF (Supplementary Appendix S1).

Preclinical studies have demonstrated that monotherapy with VNS improves LV ejection fraction, decreases LV end-systolic and end-diastolic volumes, improves indices of LV diastolic function, significantly reduces LV end-diastolic circumferential wall stress, a determinant of myocardial oxygen consumption, and decreases plasma levels of n-terminal pro-brain natriuretic peptide, and reduces of minimum, average and maximum heart rate. These measures suggest that VNS can reduce preload, improve LV relaxation and improve LV function without increasing myocardial oxygen consumption (Premchand et al., 2014; Hanna et al., 2018). Vagal stimulation, performed shortly after the onset of an acute ischemic episode in conscious animals with a healed myocardial infarction, has been effective in preventing ventricular fibrillation (Vanoli et al., 1991). The addition of VNS to beta-blockade improves LV systolic function, improves indices of LV diastolic function, lowers LV end-diastolic pressure and LV end-diastolic wall stress, and increases deceleration time of rapid mitral inflow velocity beyond that seen with beta-blockade alone (Sabbah, 2011).

VNS can be delivered clinically using open-loop stimulation with a programmable pulse generator and an electrode lead that surrounds the cervical vagus nerve and requires no intraoperative mapping for placement (Figure 2) (Inder S et al., 2019).

An external programmer, similar to the type used to adjust sensing and stimulation of a pacemaker, can be used to adjust variables of VNS intensity. Intensity comprises the combination of the pulse amplitude (current), pulse frequency, pulse duration (width), and duty cycle. The duty cycle consists of a period of vagus nerve stimulation (“on-time”) alternating with a period of no stimulation (“off-time”), is quantified as the ratio of duration of the on-time to duty cycle duration (i.e., on-time plus off-time), and is repetitive over time (Figure 3).

The electrical lead of the VNS system that is in current investigational use for HF has a bipolar silicone helical design with platinum-iridium conductors, is self-sizing and atraumatic, and is identical in design to the electrical lead that has been utilized in more than 31,000 VNS system implantations for epilepsy or depression and has had 72,100 device-years of patient exposure since 2009 (LivaNova Inc, Houston TX, United States). Cumulative lead survival has exceeded design requirements and there have been a low rate of complications, with common causes being infection (0.87%) and vocal cord dysfunction (0.68%), and a low rate of lead failure. Feasibility studies of chronic utilization of this VNS system for HF have had no device or therapy-related serious adverse events, device malfunctions, or therapy discontinuations (Anand et al., 2020a).

The axons that comprise the cervical vagus nerve include approximately 80% afferent and 20% parasympathetic preganglionic efferent projections (Libbus et al., 2016). The efferent vagal fibers that are directed to the heart usually operate with discharge frequencies in the range of 5–10 Hz (Olshansky et al., 2008). VNS for epilepsy and VNS for heart failure both manifest a dose (intensity)-dependency in achieving the desired therapeutic effect, but differs in the neurologic pathways targeted, the technology platforms utilized, VNS characteristics, and the anatomic portal for VNS administration. Left cervical VNS is used for epilepsy and cathodal stimulation is directed caudally. Right cervical VNS is used for HF and cathodal stimulation is directed peripherally. The therapies also differ in the frequency, pulse width, duty cycle, and current outputs that are utilized. Titration of VNS for epilepsy is empiric while a biomarker of autonomic nervous system engagement and cardiac control is used to guide titration of VNS for HF (Inder S et al., 2019).

The functional biological effects of VNS for HF are achieved through a combination of efferent and afferent stimulation. Efferent effects occur through cathode stimulation in a caudal direction at the interface of the electrical lead with the various axons that traverse the vagus nerve interface. An acute decrease in heart rate detected at the time of stimulation of the vagus nerve during the duty cycle is indicative of autonomic engagement at the time of stimulation (Ardell et al., 2015).

Afferent activation affects efferent sympathetic and parasympathetic function centrally by modulating continuous sympathetic and parasympathetic activity (tonic effect) and activity at rest (basal effect). The actions result peripherally in vasorelaxation through activation of the nitric oxide pathway (Premchand et al., 2014; Hanna et al., 2018). Vagus nerve efferent activation causes anti-adrenergic effects both within the intrinsic cardiac nervous system and via pre-synaptic and post-synaptic interactions at the end terminus (DiCarlo et al., 2013).

At the myocyte level, increases in acetylcholine signaling through muscarinic receptors reduce oxidative stress, decrease contractile force, improve calcium signaling, and restore gene expression. At the same time, cholinergic trans-differentiation of sympathetic neurons occurs, playing a normalizing role against sympathetically mediated pathogenesis (Premchand et al., 2014; Hanna et al., 2018).

As a consequence of augmenting parasympathetic tone and reducing sympathetic hyperactivity, there is a medium-to long-term reduction of endothelial dysfunction, mitigation of cardiac remodeling, improved left ventricular function with restoration of physiologic sympathetic myocyte responsiveness, improved myocardial oxygen supply-demand ratio, and a reduced risk for arrhythmias that compromise cardiovascular function and/or longevity. These effects are complementary to those of pharmacotherapy for HFrEF (Figure 4), (Ardell et al., 2016) and improvements have occurred in symptoms and funcion when VNS is administered in conjunction with GDMT to patients with HFrEF.

A large-scale, multi-national randomized controlled trial (ANTHEM-HFrEF) is now underway in order to fully test the hypotheses raised by results of ANTHEM-HF, and to further assess the impact of ART on morbidity and mortality. The key inclusion criteria are stable GDMT for at least 4 weeks, NYHA class III or class II if hospitalized for HF within the previous 12 months, LVEF ≤35%, LV end diastolic diameter <8.0 cm, NT-proBNP ≥800 pg/ml, and 6MWD of 150–450 m, limited by HF symptoms.

A Bayesian adaptive sample size selection is being utilized in determining the impact of VNS on morbidity and mortality. The effects of ART on symptoms and function are also being evaluated. Patients are randomized 2:1 to VNS plus GDMT or GDMT alone. Evaluation of morbidity and mortality utilizes a conventional frequentist approach for analysis of the primary outcome end point—reduction in the composite of cardiovascular death or first HF hospitalization.

A second study is embedded within the ANTHEM-HFrEF Pivotal Study to evaluate improvement in symptoms and function in the study population. Successful demonstration of improvement in symptoms and function will require meeting two risk-related conditions—a 60% or greater probability of a reduction in the combination of cardiovascular death/HF hospitalization at the end of the study (i.e., no evidence of harm) and sufficient freedom from device-related serious adverse events—and demonstrating statistically significant improvements in three efficacy endpoints: LVEF, 6MWD, and KCCQ overall score (Figure 6) (Zaeem et al., 2019).

Progress has continued during the past decade in our understanding of how ART may be delivered using VNS in a safe and tolerable manner to alleviate symptoms, increase functional capacity, and improve the quality of life in patients with HFrEF. Findings to date, derived from pilot investigation of VNS delivered at the natural operating frequency of the vagus nerve, have provided hypotheses that are now being tested in a large multicenter randomized investigation of VNS effects on symptoms, functional status, quality of life, morbidity and mortality.