Introduction

Renal sensory afferent and efferent nerves allow bidirectional communication between the central nervous system and the kidney. Peripheral and central inputs alter efferent renal sympathetic nerve activity, influencing the innervated structural and functional components of the kidney, including vessels, glomeruli, and tubuli [1]. The physiological efferent renal sympathetic nerve activity regulates renal blood flow, glomerular filtration rate, tubular sodium and water handling, and stimulates renin release from the juxtaglomerular apparatus which regulates blood pressure and renal perfusion [1, 2]. Under pathophysiological conditions, abnormal efferent renal sympathetic nerve activity can contribute to the associated abnormalities of renal function which, in turn, are of importance in the pathogenesis of several disease conditions like hypertension, renal disease and heart failure [2]. This efferent signaling is further regulated by afferent input from sensory renal chemo- and mechano-receptors in the kidney. Afferent fibers from the kidney travel along with the sympathetic nerves at the level of the kidney and then enter the dorsal roots and project to regions of the brainstem involved in cardiovascular control [2,3,4,5]. Therefore, the kidney represents a source of increased sympathetic activation under certain pathophysiological conditions, such as renal ischaemia, hypoxia, and intrinsic renal disease [6,7,8]. The density of peri-arterial renal sympathetic nerve fibers is lower in distal segments and dorsal locations when compared with proximal superior locations. There is a clear predominance of efferent nerve fibers, with decreasing density of afferent nerves from proximal to distal peri-arterial and renal parenchyma [3].

These observations support the rationale to modulate autonomic innervation of the renal artery, arterioles and tubules in order to reduce norepinephrine spillover from the kidney and to influence sympathetically dependent pathophysiologies [9,10,11]. Renal denervation (RDN) substantially prevents progressive blood pressure rise and associated renal injury and cardiac remodeling in animal models of hypertension [9, 10] (Fig. 1). Factors such as withdrawal of sympathetic nerve activity to the kidney and subsequent changes in fluid mobilization and reduction in circulating angiotensin II, as well as removal of elevated afferent renal nerve activity as a consequence of a pathophysiological alteration in the kidney, may contribute to the antihypertensive effects of RDN. Interestingly, selective destruction of afferent renal nerves had limited antihypertensive effects in hypertensive Dahl salt-sensitive rats [12, 13].

Fig. 1
figure 1

Effects of renal denervation (RDN) on blood pressure, renal catecholamine levels and renal and cardiac interstitial fibrosis in spontaneously hypertensive obese rats (SHR-ob). a Mean arterial blood pressure measured by telemetry in control rats (Ctr), SHRs-ob, and SHRs-ob+renal sympathetic denervation (SHRs-ob+RDN) during 100 days after RDN performed at the age of 34 weeks. b Immunohistochemical renal tyrosine hydroxylase staining and c renal norepinephrine tissue content at the age of 48 weeks. d Interstitial left ventricular (LV) and e kidney fibrosis with representative histological images. (Adopted from [9])

Full size image

In non-randomized studies [14, 15] and registries [16], RDN reduced blood pressure and partially reversed cardiac hypertensive end-organ damage in patients with drug-resistant hypertension [17,18,19]. A recent randomized sham controlled trial (SYMPLICITY HTN-3, ClinicalTrials.gov number, NCT01418261) could not confirm blood pressure reduction by RDN in drug-resistant hypertensive patients [20]. However, in another published, randomized, sham controlled trial, RDN by a new ablation catheter reduced blood pressure in medication naïve hypertensive patients [21]. Several factors such as problems with assessment of renal denervation efficacy, variable experience of the proceduralists and the complex issues of patient drug compliance, including improvement of drug adherence in initially poorly adherent patients (the Hawthorne effect), placebo effect and regression to the mean, makes the interpretation of the results difficult.

Beyond regulation of blood pressure, increased sympathetic activity is suggested to contribute to the progression of cardiac arrhythmias [22]. At the cellular level, norepinephrine, the neurotransmitter of the sympathetic nervous system, is released from postganglionic neurons in response to sympathetic stimuli activating cardiac beta-receptors. Subsequently, altered cardiac calcium handling and electrophysiology contribute to arrhythmogenic mechanisms including delayed afterdepolarization-related ectopic firing and re-entry [23, 24]. Sympathetic hyperinnervation is reported in the atria of dogs following rapid atrial pacing [25], and increased sympathetic and vagal nerve discharges can be documented before the onset of atrial fibrillation (AF) in dogs with pacing-induced congestive heart failure [26]. Additionally, recent data have shown that sympathetic activation promotes the development of the arrhythmogenic substrate via neurohumoral mechanisms, such as the renin–angiotensin–aldosterone system, leading to the upregulation of profibrotic pathways in the atria [23, 24]. Therefore, elevated sympathetic activity may contribute to both cardiac structural and electrical remodeling, thus predisposing to arrhythmia. Activation of the autonomic nervous system plays an important role in the initiation and the maintenance of AF, but may also be targeted for the maintenance of sinus rhythm. Beta-receptor blockade by metoprolol was effective in preventing recurrence of AF after successful cardioversion [27], and autonomic modulation by, for example, low-level baroreceptor stimulation can help to maintain sinus rhythm [22, 28].

These observations raise the question of whether modulation of the autonomic nervous system by targeting the renal sympathetic nerves might be an effective strategy to improve rhythm control in patients with cardiac arrhythmias. Here, we will review the current literature on the anti-arrhythmic effects of autonomic nervous system modulation through RDN, with a particular focus on atrial fibrillation and ventricular arrhythmias. We will discuss new insights from preclinical and clinical mechanistic studies and possible clinical implications.

Modulation of renal sympathetic innervation by renal denervation

In small animal models, RDN can be performed by combined surgical and chemical approaches, which modulate both efferent and afferent renal nerve activity [9, 11]. Both kidneys are approached through medial laparotomy, the visible nerves in the area of the renal hilus are removed and approximately 2–4 mm of the adventitia from the renal artery is stripped. The area is then moistened with a phenol/ethanol (10–20%) solution for 10–15 min. Combined surgical and chemical RDN reduces kidney tissue norepinephrine levels by > 90–95% in several animal models. The release of norepinephrine into the urine and renal venous blood, which is inducible by efferent sympathetic nerve stimulation, can be reduced by 80% after RDN in animal studies [5, 29]. Previous studies on RDN in normotensive rats by a 10% phenol/ethanol solution suggest renal reinnervation within 9 weeks [30]. In obese spontaneously hypertensive rats, RDN by a 20% phenol/ethanol solution resulted in a more sustained reduction in blood pressure, renal sympathetic nerve density and renal norepinephrine tissue content (Fig. 1) [9].

Additionally, RDN can prevent cardiac neural remodeling and nerve sprouting under pathophysiological conditions, suggesting reduction of afferent nerve activity. In spontaneously hypertensive rats, RDN by a 20% phenol/ethanol solution resulted in a reduced expression of the norepinephrine transporter in the kidney and in the heart (Fig. 2). In rats with ischemic heart failure, RDN preserves sympathetic nerve innervation in the ventricles, thus improving cardiac function [31], while in goats and dogs with atrial fibrillation induced by atrial tachypacing, RDN achieved a significant reduction in atrial sympathetic nerve sprouting [32,33,34].

Fig. 2
figure 2

a Representative image during a chemical renal denervation (RDN) procedure in a spontaneously hypertensive rat. Both kidneys were surgically denervated by cutting all visible nerves in the area of the renal hilus and by strippingapproximately 2–4 mm of the adventitia from the renal artery. The area was then moistened with a phenol/ethanol solution for 10–15 min using a brush. b Representative western blot to quantify protein expression of norepinephrine transporter in kidney and heart tissue of sham-denervated spontaneously hypertensive rat and of RDN spontaneously hypertensive rats

Full size image

Most of the brainstem regions involved in cardiovascular control receive input from the renal afferent fibers. Afferent signals arising from the renal sympathetic nerves are able to influence nerve activity of several ganglia including those innervating the heart (Fig. 3). Studies suggest a direct link between renal sympathetic nerve activity and left stellate ganglion (LSG) activity [35, 36]. Renal sympathetic nerve stimulation is associated with LSG neuronal activity and upregulates the level of LSG nerve growth factor expression [36]. Importantly, the stimulation procedure not only activates the efferent sympathetic nerves but also the afferent fibers going from the kidney to the central nervous system. Additionally, the stimulation frequency used in these renal nerve activation studies are likely to impact renal hemodynamic or excretory renal function, which has not been sufficiently characterized and studied.

Fig. 3
figure 3

Effects of renal denervation (RDN) on efferent and afferent signaling of the kidney. NE Norepinephrine, MSNA muscle sympathetic nerve activity, LSG Left stellate ganglion

Full size image

Besides modulation of circulating catecholamines and whole body sympathetic nerve activity, bilateral RDN caused significant central and peripheral sympathetic nerve remodeling, improved baroreflex sensetivity [37] and reduced stellate ganglion nerve activity in ambulatory dogs [36]. Catheter-based RDN can also affect cardiac sympathetic activity. Selective RDN significantly reduces cardiac sympathetic overdrive assessed by 123I-MIBG scintigraphy [39, 40].

A catheter-based approach for RDN has been developed for human use. RDN can result in an up to 50% reduction of renal norepinephrine spillover measured with a radiochemical tracer methodology using 3H-norepinephrine in humans [12, 13]. Additionally, firing of single sympathetic vasoconstrictor fibers (measured by single muscle sympathetic nerve activity), was reduced by 37% [38]. These findings indicate that sympathetic activation can be reduced by RDN beyond just the kidneys, suggesting a combined modulation of efferent and afferent signaling (Fig. 3).

The first-generation devices used radiofrequency pulses emitted from a monopolar electrode positioned under fluoroscopic guidance in each of the renal arteries. The latest generation devices are multi-electrode catheters allowing a more standardized and more intense ablation procedure [41]. During the development of RDN catheters, the ablation strategy changed from ablation limited to the main vessels to more extended ablation in the main and subsequent branch vessels [41]. This revised approach was associated with higher efficacy and lower variability in treatment effects in preclinical studies. Alternatively, ultrasound or chemical ablation have been investigated in clinical studies [41].

Despite great advantages of catheter techniques, there is no universally accepted measure by which the completeness of RDN can be properly assessed intraprocedurally. Although periprocedural changes in circulating norepinephrine or neuropeptide Y [42], nerve stimulation-induced blood pressure changes [43, 44] and anatomy-guided lesion placement [45] have been introduced to improve ablation outcome, a standardized protocol to directly measure renal sympathetic nerve activity in humans is still lacking. In sedated and unrestrained conscious rodent models, assessment and quantification of renal nerve activity is feasible [46, 47]. The development of endovascular recording of renal nerve action potentials may provide a useful accessory tool to assess successful RDN. Innovation in this area will be crucial to predict and monitor the therapeutic value of RDN [48].

Effects of RDN on cardiac arrhythmias

Beyond hypertension and hypertensive end-organ damage, cardiac rhythm disorders, namely atrial fibrillation and ventricular arrhythmias, have been identified as a promising and emerging target of RDN [49].

Atrial electrophysiological effects of renal denervation

In experimental data, RDN resulted in a reduction in heart rate and AV-conduction velocity in pigs [50]. In chloralose/urethane anasthetized pigs, neither atrial effective refractory period nor P-wave duration was influenced by acute RDN, thus excluding a direct impact of RDN on atrial refractoriness or atrial conduction during sinus rhythm [50]. Additionally, the sensitivity of ganglionated plexi was not modulated by RDN. In an AF pig model with rapid atrial pacing, RDN reduced the duration of pacing-induced AF [51]. However, AF-induced electrical remodeling was not attenuated [50]. Reduced AV-conduction velocity was associated with lower ventricular heart rate during AF [50].

In a goat model of persistent AF induced by atrial tachypacing [32], RDN reduced renal norepinephrine concentrations, blunted atrial neural remodeling and reduced AF complexity along with structural remodeling (Fig. 4). Moreover, increased AF-inducibility, shortening and dispersion of atrial refractoriness as well as elevated plasma norepinephrine levels were almost completely prevented by RDN in animals subjected to stimulation of left stellate ganglion and rapid atrial pacing for three hours [52, 53].

Fig. 4
figure 4

Effects of renal denervation (RDN) compared to sham intervention (SHAM) in goats with atrial fibrillation induced by an implanted pace maker. a Representative X-ray images of the kidney. Locations of ablation delivery are indicated by superimposed small black points in the vessels. b Renal tissue norepinephrine concentrations in the left (white bars) and right (black bars) kidneys determined after the sacrifice experiments. c Representative perivascular tyrosine hydroxylase (TH) staining (brown twigs) of cardiac sympathetic nerves in SHAM and RDN goats with 6 weeks AF (magnification, ×1000). d Quantification of TH-positive (indicating sympathetic nerve structures) fraction of the perivascular area in the anterior left atrium (LAant), posterior left atrium (LApost) and right atrium (RA). e Quantification of AF complexity: representative spatial and temporal distribution of AF activation pattern during one AF cycle length of the right and left atrium in SHAM (top) and RDN (bottom) goats with 6 weeks. Isochronal maps: time between isochrones 10 ms, red earliest, blue latest activation. (Adopted from [32])

Full size image

In a pig model of obstructive sleep apnea (OSA), shortened atrial refractoriness [54, 55] acutely induced by applied negative thoracic pressure was primarily mediated by combined sympathovagal activation, since it could be influenced by atropine, bilateral vagotomy or beta-receptor blockade [54, 56]. Compared to beta-blocker treatment, RDN resulted in more pronounced attenuation of the shortening in atrial refractoriness during OSA maneuvers, which might explain the superior anti-arrhythmic effect of RDN compared with beta-blocker therapy in this animal model [55]. Importantly, anti-arrhythmic drugs such as amiodarone or sotalol displayed a considerably less pronounced anti-arrhythmic effect compared to RDN in the same model [55]. In pigs undergoing repetitive OSA maneuvers over 4 h, RDN inhibited spontaneous atrial premature beats, and reduced the number AF episodes as well as AF duration [56]. The observed reduction in spontaneous atrial extra beats by RDN may reduce the trigger for AF in OSA [56]. Additionally, RDN has been shown to reduce susceptibility to AF in a canine models of renal impairment, induced by embolization of small branches of the renal artery in the right kidney using gelatin sponge granules [57], and tachycardiomyopathy, induced by ventricular tachypacing [58]. Taken together, these studies provide strong preclinical evidence towards an anti-arrhythmogenic effect of RDN.

Clinical data

RDN can reduce heart rate in patients with resistant hypertension [59]. In persistent AF, RDN can improve rate control [60], which might improve clinical symptoms and outcomes in patients with AF. RDN has been shown to prevent or even reverse atrial remodeling determined by echocardiography [61] or by electroanatomical mapping [62]. In a small study of patients with symptomatic AF and resistant hypertension, the atrial anti-arrhythmic effects of circumferential pulmonary vein isolation (PVI) combined with RDN were investigated [63,64,65]. Patients who received the combined procedures showed significant reductions in average systolic and diastolic blood pressure, whereas those in the PVI-only group did not show any significant improvement in blood pressure. Other electrophysiological parameters were not obviously changed by RDN. At 1-year follow-up, 69% of patients who received both procedures maintained sinus rhythm, compared to 29% of those in the PVI-only group [63]. In chronic kidney disease patients, the addition of RDN to pulmonary vein isolation reduced recurrence of paroxysmal atrial fibrillation [66]. In a case report, even RDN without PVI reduced blood pressure and attenuated paroxysmal AF episodes, which were symptomatic and drug-resistant before RDN [67]. It remains unknown whether sympathetic modulation of the autonomic nervous system by RDN can display anti-arrhythmic effects in hypertensive AF patients independent from its blood pressure-lowering effects. This important clinical and pathophysiological question should be addressed in a randomized sham controlled trial, in which RDN is compared to aggressive and adopted up-titration of antihypertensive drugs in the control group.

Ventricular arrhythmias, heart failure and renal denervation

In dogs, 3 h of renal sympathetic nerve stimulation increased LSG neuronal activity, and facilitated the incidence of ventricular arrhythmias during acute myocardial ischemia. Interestingly, the increase in ventricular arrhythmias could be attenuated by LSG ablation [34]. In different pig models with acute ischemia/reperfusion [66] (Fig. 5) or with myocardial infarction induced by a permanent coronary occlusion [69,70,71], RDN has been shown to reduce ventricular ectopic activity and ventricular fibrillation. Also in a dog model of heart failure induced by atrial tachycardiomyopathy induced by ventricular high-rate pacing, RDN attenuated the ventricular remodeling process [72, 73]. The occurrence of spontaneous premature ventricular contractions and the subsequent ventricular dysfunction could be prevented by sympathetic modulation by the procedure [74]. Similarly, arrhythmogenic prolongation of ventricular repolarization (QT-interval) induced by simulated sleep apnea [75] or by cesium [76] could be attenuated by RDN.

Fig. 5
figure 5

Effects of renal denervation (RDN) on ventricular fibrillation (VF) in a pig model for ventricular ischemia and reperfusion. a Representative view of the left ventricular during ischemia reperfusion experiments. Atrial electrophysiology was recorded by an epicardial catheter. b Incidence of VF during ischemia and the reperfusion phase in RDN-treated compared to SHAM-treated pigs. c Representative hemodynamics and electrocardiographic (ECG) tracings during 20 min of left anterior descending coronary artery ligation followed by reperfusion in a SHAM-treated and a RDN-treated animal. (Adopted from [68])

Full size image

Clinical data

In patients with heart failure with reduced ejection fraction, circulating norepinephrine concentrations predict mortality [77]. Interestingly, cardiac and renal norepinephrine spillover is increased in mild to moderate and severe chronic heart failure compared to healthy subjects, with the absolute renal norepinephrine spillover higher than the cardiac spillover suggesting that the kidney contributes more to total norepinephrine spillover than the heart in heart failure [78, 79].

A small case series has provided evidence that, in patients with dilated cardiomyopathy and an electrical storm, RDN was able to reduce ICD shocks and ventricular ectopic activity [80]. These anti-arrhythmic effects could be confirmed in several other case series [81, 82] and in an international multicenter registry [83]. RDN may be most beneficial for patients with heart failure with recurrent, refractory arrhythmias that cannot tolerate maximal beta-blockade and are not eligible for anti-arrhythmic catheter-based ablation of ventricular tachycardia. Alternatively, RDN may also be used as an adjunct strategy in patients undergoing catheter ablation. In patients with heart failure, RDN reduced NT-proBNP and was safe without any adverse event deterioration of other indices of cardiac and renal function [84]. The role of autonomic modulation by RDN in patients with heart failure and ventricular arrhythmias needs to be further investigated in randomized, sham-controlled trials.

Conclusions

RDN is a promising and safe strategy to modulate the autonomic nervous system activity. Beyond its blood pressure-lowering effects, animal models of atrial fibrillation have demonstrated favorable electrophysiological changes, reverse remodeling and potential anti-arrhythmic effects of sympathetic modulation by RDN. Interestingly, adjunct RDN improved outcome of catheter ablation in atrial fibrillation patients in small non-randomized clinical trials; these findings warrant further clinical trials. Additionally, animal experiments and early registry data also suggest promising ventricular anti-arrhythmic effects by RDN in the setting of heart failure. Further studies are needed to investigate the anti-arrhythmic effects of RDN.