One of the characteristics of the autonomic nervous system is the waxing and waning of its activity, both afferent and efferent. Similarly, the last 50 years have witnessed the waxing and waning of its interest for clinical cardiologists dealing with cardiac arrhythmias, sudden death, and heart failure.

Especially in the 1970s, but also later, much effort was devoted to the study of neural activity through the recording of single fibers in the sympathetic and vagal nerves (Malliani et al., 1969, 1973; Kunze, 1972; Cerati and Schwartz, 1991), which allowed the description of important autonomic reflexes (Schwartz et al., 1973). In the 1970s, 1980s, and 1990s, interest for the autonomic nervous system peaked. Some investigators focused on the stimulation of nerves directed to the heart trying to derive information for potential clinical translation (Schwartz, 1985), others focused on various aspects of the analysis of heart rate, either at rest (Kleiger et al., 1987) or in response to stimuli (Billman, 2009); these analyses of tonic or reflex autonomic activity had post-myocardial infarction (post-MI) and heart failure risk stratification as one significant objective (Schwartz et al., 1992a; Mortara et al., 1997; La Rovere et al., 1998, 2003). During the last 20 years, there was a surge of interest for the possibility of modulating autonomic activity, especially the vagal one, also in chronic conditions such as heart failure (Schwartz et al., 2008a, 2015; De Ferrari et al., 2011); however, the combination of unsatisfactory results and superficial analyses (Hauptman et al., 2012; Gold et al., 2016) has somewhat cooled down these hopes. One area where clinical success has fueled interest is the one related to the prevention of life-threatening events by cardiac sympathetic denervation (Schwartz, 2014).

Here, the three of us, who have collaborated in a number of studies, present our unabashed views on some of these topics.

The neural control of the heart is accomplished throughout a multilevel neural network within the central nervous system and peripheral extracardiac and cardiac ganglia that exert their influence via the sympathetic and parasympathetic nervous systems (Levy and Schwartz, 1994; Shivkumar et al., 2016). Cardiac diseases may profoundly affect central and peripheral mechanisms of neural control of cardiac function, thus resulting in maladaptive responses that may be critically involved in the progression of the disease or in the development of arrhythmias. Neural sensory information from the heart (Paintal, 1963; Schwartz et al., 1973), blood vessels, and other organs is processed at different levels within the neuraxis with a first level of integration represented by the intrinsic cardiac nervous system located in the cardiac ganglia within the heart. The intrinsic cardiac nervous system processes sensory information and provides efferent input to the myocardium under the tonic modulation of the extrinsic sympathetic and parasympathetic input. Arterial baroreceptors play a paramount role in the neural control of the cardiovascular system (Eckberg and Sleight, 1992). Arterial baroreceptors are stretch receptors embedded in the adventitia of the carotid sinus and aortic wall. Increases in arterial blood pressure will result in an increased rate of impulse firing to the nucleus tractus solitarius, which modulates sympathetic and parasympathetic output to the cardiovascular system. The baroreflex control of circulatory homeostasis occurs on a negative feedback basis. Thus, the attending reflex decrease in sympathetic activity and increase in vagal activity will reduce heart rate, cardiac contractility, and peripheral resistance. Opposite changes are associated with an arterial pressure decrease.

By complex interactions between the main neurotransmitters [namely, noradrenaline, acetylcholine, and neuropeptide Y; (Dusi et al., 2020)] and their effects on specific receptors of cardiac cells in the sinoatrial node, atrioventricular node, and left ventricle, the autonomic nervous system affects several aspects of cardiac electrophysiology. At the sinus node level, efferent vagal activity decreases while sympathetic activity increases the spontaneous depolarization rate of sinus node cells.

Furthermore, it has been known for many years that sympathetic nervous system stimulation may be pro-arrhythmic, particularly in conditions of acute myocardial ischemia (Harris et al., 1971; Schwartz and Vanoli, 1981; Janse et al., 1985; Schwartz et al., 1985), while vagal nerve stimulation may reduce the potential for lethal arrhythmias (Kent et al., 1973; Vanoli et al., 1991). Thus, by controlling the autonomic traffic to the heart, the baroreceptors are involved in the susceptibility to ventricular and supraventricular arrhythmias. Moreover, by controlling the hemodynamic adjustments to blood pressure changes, they also play a role in the clinical response to sustained rhythm disorders (De Ferrari et al., 1995; Landolina et al., 1997).

Damage to cardiac sensory nerve endings caused by acute myocardial infarction and left ventricular remodeling directly affects the baroreceptor system. The attending reflex autonomic dysfunction, characterized by reduced parasympathetic and increased sympathetic activity coupled with neural remodeling and nerve sprouting (Cao et al., 2000), promotes arrhythmogenesis.

The prominent neurohumoral mechanism at play in heart failure is the sympathetic nervous system whose increased activity coupled with vagal withdrawal is initiated by the arterial baroreflex (Hartupee and Mann, 2017). Although other mediators, including sympatho-excitatory reflexes, humoral factors, and central mechanisms (Floras and Ponikowski, 2015) contribute to the development of sympathetic-parasympathetic imbalance in heart failure, an impairment of baroreflex control of heart rate is a prominent characteristic of the heart failure syndrome and a reliable marker of the severity of the disease (Mortara et al., 1997).

Relevant insights into the pathophysiological implications of heart disease-related baroreflex impairment date back to the early 1970s when it was recognized that baroreceptor reflexes can be modulated by cardiac afferent sympathetic activity activated by mechanical and chemical stimuli (Malliani et al., 1973; Schwartz et al., 1973). An animal model provided the first evidence that reduction in cardiac parasympathetic control is associated with an increased risk for sudden death. In this canine model (Billman et al., 1982; Schwartz et al., 1984), baroreflex sensitivity (BRS) was impaired by myocardial infarction, with the greatest impairment noted in animals particularly susceptible to sudden death (Schwartz et al., 1988). Similarly in humans, a tight relationship between reduced baroreceptor activity and heart disease state was first reported by Eckberg et al. (1971) and was subsequently found to be associated with an increased risk of cardiac mortality and sudden cardiac death in post-MI and heart failure patients (Mortara et al., 1997; La Rovere et al., 1998, 2001).

The initial results with BRS led some investigators to consider the possibility that powerful baroreceptive reflex would imply that the attendant increase in vagal activity to the sinus node would extend to the ventricles as well. The high specificity of the cardiac innervation (Pagani et al., 1974; Randall, 1984), the central organization of cardiovascular reflexes (Wurster, 1984), and the important report by Inoue and Zipes (1987) indicate clearly that such an extrapolation would be both naïve and unwarranted. Indeed, a heart-rate response indicative of increased vagal activity does not exclude the possibility of a dominant sympathetic activity at ventricular level (e.g., the diving reflex, hypoxia, and inferior myocardial ischemia). Nonetheless, the most frequent reflex response is synergistic, i.e., one limb of the autonomic nervous system is excited with simultaneous inhibition of the other (Wurster, 1984), and the reduction in heart rate produced by the baroreflex is accompanied by a reflex withdrawal of sympathetic activity that is generalized and extends to the ventricles. In the sudden death animal model that played such an important role in the development of the clinical interest for BRS (Schwartz et al., 1984), the dogs with higher BRS were also those with larger heart rate reductions during acute myocardial ischemia despite continuation of exercise (Schwartz et al., 1984). One logical implication is that the animals responding with strong vagal reflexes to blood pressure increases are likely to respond similarly to acute myocardial ischemia. The animals with the greatest sinus node response to the baroreflex test are less prone to sudden death during myocardial ischemia, and conversely, those with the most reduced BRS are more vulnerable to ventricular fibrillation. This does not mean that the baroreflex test predicts the autonomic changes at the ventricular level during myocardial ischemia but indicates that it can often predict the outcome during an ischemic episode, which is what really matters. Although, as correctly stated (Inoue and Zipes, 1987), the use of spontaneous or reflex changes in heart rate as an indicator of what might happen at the ventricular level would certainly be naive, their use to identify individuals at varying risk of life-threatening events is a rational exploitation of the current understanding of cardiovascular pathophysiology.

As the baroreflex affects the balance between parasympathetic inhibition and sympathetic excitation of the sinoatrial node of the heart, sinus node activity (either spontaneous or in response to a provocation) can provide information on the underlying regulatory system.

Effective risk stratification for patients who might develop life-threatening ventricular arrhythmia and sudden cardiac death is one of the main unsolved areas in clinical cardiology. Arrhythmic risk represents the sum of several different risk-augmenting processes and factors. Understanding the relation between changes in autonomic activity and cardiac electrophysiological properties has led to the view that the autonomic nervous system modulates interactions between triggering factors and the underlying electrophysiologic substrate. This points to a significant potential prognostic value of markers of autonomic activity.

Since the 1990s, the analysis of BRS has been considered as a tool that might help identifying “high-risk” patients. A multicenter study on more than 1,200 post-infarction patients demonstrated the incremental prognostic value provided by an impaired BRS when combined to left ventricular ejection function and to the potential trigger of non-sustained ventricular tachycardia (La Rovere et al., 1998, 2001). Specifically, a depressed BRS, a reduced left ventricular ejection fraction, and the presence of non-sustained ventricular tachycardia were all independent predictors of mortality, but depressed BRS almost doubled the risk of death provided by the other two markers. Moreover, among patients with either reduced or preserved left ventricular function but without signs of electrical instability, mortality differed significantly according to the presence or absence of preserved autonomic function (La Rovere et al., 2001; De Ferrari et al., 2007).

The role of baroreflex-mediated responses in the control of hemodynamic stability is particularly relevant during the course of a sustained ventricular rhythm. Inadequate baroreflex-mediated sympatho-excitation during a sustained ventricular tachycardia in post-infarction patients was the leading cause of an unfavorable hemodynamic profile leading to circulatory collapse (Landolina et al., 1997).

Randomized trials, demonstrating that among post-infarction patients mortality can be effectively reduced by prophylactic implantation of a cardioverter defibrillator, established a paradigm shift in risk stratification through the assessment of left ventricular ejection fraction as the gold standard risk predictor. However, this does not deprive autonomic markers of their clinical value (Wellens et al., 2014). It is now clear that left ventricular ejection fraction measurement has both limited sensitivity and specificity as a tool for arrhythmic risk stratification and that the field of risk stratification should move from the “high-risk ejection fraction” to the broader concept of the “high-risk patients” (Chugh, 2017). This transition implies a novel opportunity for autonomic markers to be re-evaluated in their involvement in the pathogenesis of arrhythmic risk and incorporated in novel prediction models. Moreover, novel ECG-based risk markers that quantify sympathetic activity-associated repolarization instabilities are promising in their ability to guide decisions about the prophylactic implantation of a cardioverter defibrillator (Bauer et al., 2019). The markers tested by Bauer et al. (2019) are framed in an emerging area of biomedical signal processing aiming at monitoring relevant electrocardiographic fiducial points and time intervals under the hypothesis such that their evolution over time might provide information about cardiac control.

In a founder population of long QT syndrome type 1 (LQT1), which avoids the confounding factors due to different mutations and segregates the malignant KCNQ1-A341V mutation (Brink et al., 2005; Crotti et al., 2007; Brink and Schwartz, 2009), the characterization of cardiac autonomic control and baroreflex function was found to be useful to improve the risk stratification of arrhythmic events (Schwartz et al., 2008b; Crotti et al., 2012; Bari et al., 2014a,b, 2015; Porta et al., 2015). In this population which is at the highest risk of fatal events in situations of high sympathetic drive (Schwartz et al., 2001), it was found that subjects who did not experience arrhythmic events, namely the asymptomatic mutation-carriers, have a completely different autonomic profile compared to those experiencing syncope or cardiac arrest requiring resuscitation. Indeed, asymptomatic individuals exhibited longer RR (Schwartz et al., 2008b), lower BRS (Schwartz et al., 2008b), higher QT variability in the LF band during daytime (Porta et al., 2015), lower respiratory sinus arrhythmia during nighttime and under β-blockers (Porta et al., 2015), slower heart rate recovery after exercise test (Crotti et al., 2012), and lower QT variability complexity (Bari et al., 2014a,b, 2015). These findings suggested that, besides RR lengthening, the combination of a more reactive sympathetic drive to the ventricles (i.e., adapting more rapidly QT to RR changes and limiting irregularity of QT changes during a sympathetic stressor) and of a sluggish vagal responsiveness after exercise represents a protective mechanism. Remarkably, non-mutation carriers belonging to the same family line (Brink et al., 2005; Brink and Schwartz, 2009) have an autonomic profile more similar to symptomatic patients than asymptomatic ones, thus suggesting that there are peculiar traits of the autonomic control that might be key for survival because they reduce the severity of the disease (Schwartz et al., 2008b; Porta et al., 2015).

Renal sympathetic nerve activity plays a crucial role in the control of cardiovascular homeostasis and is involved not only in the pathogenesis of hypertension but also in other cardiovascular processes such as heart failure, and perhaps cardiac arrhythmias. Renal afferent and efferent nerves function in a reflex loop where afferent input from the kidney to the central nervous system is integrated with inputs from other neural reflexes to determine the level of sympathetic outflow to individual organs. Renal denervation (RDN) as a method of modulating sympathetic activity by interrupting afferent and efferent sympathetic nerve signaling appears to be an attractive therapeutic target in patients with cardiovascular disease triggered by sympathetic overactivity such as hypertension, heart failure, and – according to some – even atrial or ventricular arrhythmias. RDN has been initially introduced to reduce blood pressure in subjects with resistant hypertension (Mahfoud et al., 2013). While initial clinical trial results failed to reach a consensus on the efficacy of RDN in this context (Bhatt et al., 2014), three subsequent sham-controlled studies that were carefully designed and rigorously conducted have shown that RDN significantly reduces blood pressure regardless of the use of antihypertensive drugs (Townsend et al., 2017; Azizi et al., 2018; Kandzari et al., 2018). Notably, the use of multi-electrode catheter and more ablations per artery have definitely improved the RDN procedure in the more recent studies that also took into account the distribution of the sympathetic nerves among the renal arteries.

Clinical implications of RDN are well beyond blood pressure control. Interestingly, a recent meta-analysis including 17 studies revealed that RDN improved a number of cardiovascular markers of organ damage including left ventricular mass index, central augmentation index, and carotid-femoral pulse wave velocity independent of blood pressure (Kordalis et al., 2018). Moreover, several clinical RDN studies report beneficial effects on ventricular and supraventricular arrhythmias (Ukena et al., 2012; Pokushalov et al., 2014). In the recently reported ERADICATE-AF trial that randomized 302 patients with paroxysmal atrial fibrillation and hypertension, RDN added to catheter ablation, compared with catheter ablation alone, significantly increased the likelihood of freedom from atrial fibrillation at 12 months (Steinberg et al., 2020). While we report these studies and their claims, we cannot help expressing a certain degree of skepticism on the rationale by which RD should provide protection against life-threatening cardiac arrhythmias. The underlying concept seems to be that reduction or elimination of renal afferent would decrease the efferent sympathetic activity directed to the heart, which is mediated by the stellate ganglia. We find it difficult to conceive how this “reduction” could be greater than that produced by the direct section of these nerves, as it is produced by the cardiac sympathetic denervation performed by removing the lower half of the stellate ganglion/ganglia with the first four thoracic ganglia.

RDN has been reported to exert beneficial effects on cardiac function and remodeling in animal models of heart failure (Sharp et al., 2018), but the results in patients are largely inconsistent due in part to limited power with small sample sizes. In a meta-analysis including two controlled (80 patients) and two uncontrolled studies (21 patients) (Fukuta et al., 2017), 6 months after RDN, there was a greater increase in left ventricular ejection fraction and a greater decrease in left ventricular end-diastolic diameter in the RDN group than in the control group. No serious adverse events such as acute renal artery stenosis and dissection occurred.

A paper like this one does not really need a traditional conclusion, which would merely be a pale summary of what has been a serious effort to share with the interested reader our experience and our views. Our hope is that more and more young investigators will be attracted by this fascinating field of research, which is endowed with so many areas of clinical relevance.

PS, AP, and MR contributed to the conception and design of research, drafted the manuscript, edited and revised critically the manuscript, and approved the final version of the manuscript.