Sleep-disordered breathing encompasses a spectrum of respiratory abnormalities occurring during sleep. Among these, central sleep apnoea (CSA) accounts for approximately 5–10% of all cases.1 CSA has been defined as the loss of respiratory drive and subsequent airflow cessation resulting from a delayed brainstem response to arterial blood gases.2
CSA originates from neurological mechanisms, often involving chemoreflex dysregulation or cardiovascular pathology. CSA is commonly associated with heart failure (HF), stroke, renal failure, AF and chronic opioid use, and the presence of CSA has been associated with worse prognosis for patients with these conditions.3,4 Furthermore, emerging data seem to suggest a bidirectional relationship, with CSA independently contributing to adverse cardiovascular outcomes, including increased arrhythmic burden, worsening cardiac function and increased mortality.5 Thus, effective management of CSA represents a critical juncture in both sleep and cardiovascular medicine.
Unlike in obstructive sleep apnoea, positive pressure ventilation has not demonstrated a substantial clinical benefit for CSA.6 This has led to the development of implantable devices targeted to improve CSA, which have been associated with improved outcomes. One of these devices is the implantable transvenous phrenic nerve stimulator (TPNS), which was approved in 2017 by the Food and Drug Administration (FDA) for moderate to severe CSA. In this review we discuss the mechanism of action of this device, its indications with supporting evidence, and its potential cardiovascular benefits.
Pathophysiology of Central Sleep Apnoea
CSA is primarily driven by instability in the feedback loop between the central chemoreceptors and the brainstem respiratory centres. The central chemoreceptors, located in the medulla, respond to changes in arterial CO₂ (PaCO₂), which is mainly reliant on ventilation. While ventilation during wakefulness is controlled by metabolic and behavioural factors, ventilation during sleep becomes solely controlled by the metabolic pathways involving PaCO2.7
In CSA this feedback system involving the respiratory system and metabolic pathway is disrupted, often due to heightened chemosensitivity, prolonged circulation time (e.g. reduced cardiac output in HF), or delayed CO₂ detection, leading to cyclic oscillations in ventilation. This pattern results in alternating phases of hyperventilation, leading to hypocapnia and apnoea due to loss of respiratory drive. Notably, Cheyne–Stokes respiration, a subtype of CSA characterised by crescendo–decrescendo breathing patterns with cyclic apnoeas, is often observed in patients with HF.8
Traditional Treatment Modalities
Treatment of CSA includes a multidisciplinary approach to normalise ventilation, improve oxygenation and reduce sleep fragmentation. Given that CSA typically occurs with other comorbidities, optimisation of conditions such as improving cardiac output in HF, reducing opioid use, or stabilising renal function remains an important initial step. Next, supplementary oxygen stabilises hypoxemia and reduces respiratory instability and has been shown to reduce apnoeas and improve HF symptoms.9 Positive airway pressure (PAP) therapies with continuous positive airway pressure (CPAP) or adaptive servo-ventilation (ASV) are frequently used. However, although CPAP can be helpful for some patients with CSA, it does not provide ventilatory support.10 ASV provides dynamic pressure support in response to ventilatory fluctuations and has shown greater efficacy in reducing central events. However, its use remains controversial given that ASV has been shown to increase mortality in patients with HF with reduced ejection fraction (HFrEF) in the SERVE-HF trial.11 Furthermore, adherence remains an issue with these devices, due to intolerance of these devices overnight, highlighting the need for alternative therapies.12 Last, pharmacological ventilatory support with medications such as acetazolamide or theophylline has been previously used to increase respiratory drive, but this remains an off-label use of these medications.13
Phrenic Nerve Stimulation: Mechanism and Rationale
Exiting from the ventral roots of the mid-cervical spinal cord, the phrenic trunk is formed at the upper lateral border of the scalene anterior muscle and descends along its anterior surface behind the prevertebral fascia.14 Once in the thorax, this nerve passes in front of the lung hilum between the fibrous pericardium and mediastinal pleura and reaches the diaphragm for innervation.15 Notably, there are two phrenic nerves, and their composition is not identical, given that the right phrenic nerve has been shown to have ~20% more axons, potentially due to a larger overall area of diaphragmatic innervation.16 In the mediastinum the right and left phrenic nerves have different anatomical neighbours. The right phrenic nerve descends along the brachiocephalic vein down to the anterolateral superior vena cava, curving posteriorly around the cavoatrial junction, often coursing near the right upper pulmonary vein.17 Oppositely, the left phrenic nerve continues over the aortic arch and pulmonary arch over the left atrial appendage and is accompanied by the left pericardiophrenic vein to the diaphragm.18 The corresponding venous anatomy is the conduit for phrenic nerve stimulation (PNS). The left pericardiophrenic vein (LPPV) typically arises on the lateral border of the heart from the left thoracic vein before the emergence of the left brachiocephalic vein and then forms venovenous collaterals with the left gastric vein and inferior vena cava (IVC) below the diaphragm.19 However, the right pericardiophrenic vein (RPPV) is often very small (<0.5 mm in diameter) or even absent in most patients, making the right brachiocephalic vein the closest venous structure to stimulate the right phrenic nerve.20 Stimulation of either phrenic nerve elicits physiologic diaphragmatic contractions, replicating a natural respiratory effort. Thus, this direct activation of the respiratory system through a peripheral mechanism can bypass any dysfunctional central respiratory control and effectively eliminate the cycles of hyperventilation and apnoea.
Implantation of the Phrenic Nerve Stimulator
The remedē System (developed by Respicardia, now a part of ZOLL Medical Corporation) is the first FDA-approved implantable device specifically designed for CSA. This system bears many similarities to that of a permanent pacemaker system, consisting of a pulse generator with a transvenous pacing wire that can be implanted in the right brachiocephalic vein (Figure 1A) or LPPV (Figure 1B).21 As previously mentioned, the LPPV remains the target of choice because it is much larger and more often available than the RPPV, and it is closer in proximity to its corresponding nerve than the right brachiocephalic vein.22
While implantation techniques vary among operators, the typical method involves delivering a 7 Fr guide catheter into the left brachiocephalic vein for non-selective venography to identify the take-off of the LPPV. Subsequently, a 5 Fr inner catheter is telescoped into the guide catheter and slowly dragged medially on the floor of the left brachiocephalic vein to cannulate the LPPV and advance a coronary wire over which the stimulation lead can be advanced.22 If the LPPV is not suitable for lead implantation, the stimulation lead can be implanted in the right brachiocephalic vein to capture the right phrenic nerve with a right phrenic lead that has a helical bias to facilitate stability.21 A sensing lead can optionally be implanted in the azygos vein to detect respiration via thoracic impedance, although it has not shown any impact on outcomes or effectiveness.23
Before generator implant and pocket closure, the stimulation lead is assessed for threshold testing. The lead is a quadripolar lead with two IS-1 terminal pins and four ring electrodes, and different electrode combinations are assessed for optimal capture of the phrenic nerve to induce full contraction of the diaphragm.21 This step requires active communication with the patient to ensure stimulation and to assess any associated discomfort. Thus, general anaesthesia is typically not recommended for implantation. Once thresholds and stability are satisfactory, the lead(s) is/are sutured into place and connected to the generator with closure of the pocket in similar fashion to that of a permanent pacemaker (PPM) or ICD.24 In the case of coexistence with a PPM or ICD, it is recommended to test at maximum output to check for device-to-device interaction. The device is then activated at 6–8 weeks after implantation to allow for healing and lead stability during an in-office visit for testing.22 Once activated, the device delivers low-energy electrical pulses during sleep, which are synchronised with the respiratory cycle, to stimulate diaphragmatic contraction and restore a stable breathing pattern (Figure 2).
Clinical Evidence
Given that the TPNS system has been approved for the treatment of moderate to severe CSA, it is important to understand this definition from the initial multicentre, randomised, controlled study of 151 patients with CSA.25 Patients were included if they had an apnoea–hypopnoea index (AHI) of at least 20 on polysomnography, and at least 50% central events or at least 30 central events, and not more than 20% of these events could be obstructive.21,26 Major exclusion criteria included phrenic nerve palsy, American College of Cardiology stage D HF, stroke in the past 12 months, CSA due to opioid use, severe hypoxemia, severe chronic obstructive pulmonary disease, or renal dysfunction (including creatinine >2.5 mg/dL or calculated creatinine clearance <30 ml/min).21,26
In the initial remedē System pivotal trial, PNS demonstrated a reduction in AHI of at least 50% in 51% of the treatment group compared with 11% in the control group (p<0.001) in the first 6 months while also improving oxygenation, reducing arousals, and improving quality of life as measured with the Epworth Sleepiness Scale and SF-36.26 Further follow-up data at 5 years showed a consistent reduction in AHI from 46 events per hour to 17, and in central apnoea index (CAI) from 23 events per hour to 1.23 At 24 and 36 months, results showed sustained reductions in AHI, CAI, and improvements in sleep quality. In terms of safety, an overall rate of 14% of patients (n=21) reported adverse events. Most of these events were related to device implantation rather than to PNS, with lead component failure, lead displacement, lead dislodgement, and implant site infection being the most common.23 Furthermore, the number of adverse events occurred mostly in the first year of implant (15/21 events), and there were only three occurrences of concomitant device interaction, which were all resolved with system revision.23 In general, device implantation appears be safe with minimal procedure-related safety events, including hematoma, migraine and atypical chest pain with initiation of therapy.21,26
In patients with HF at baseline, PNS was shown to reduce CSA severity and was associated with an improvement of HF outcomes, decreasing the 6-month rate of HF hospitalisations to 4.7% compared with 17.0% in the control arm.27 Unlike ASV, PNS has not demonstrated increased cardiovascular risk or death in patients with HFrEF, making it appealing.11 In fact, win ratio (WR) analysis of the remedē System pivotal trial suggests that TPNS may be superior to untreated CSA in HF patients, using a hierarchical clinical benefit endpoint composed of mortality, HF hospitalisation and health status.28 The post-approval study for TPNS remains ongoing and will help to provide further data for safety and effectiveness, but, thus far, there remains no evidence of significant long-term adverse effects, and most patients have experienced improvement in daytime symptoms.29
Cardiovascular Implications
Heart Failure
CSA is an often underrecognised comorbidity in patients with HF, with a reported prevalence of 30–50%.30 Although the mechanism linking HF and CSA is not completely understood, it seems to be centred on changes in PaCO2 levels above and below the apnoeic threshold.31 In HF, the metabolic milieu is greatly altered, and patients chronically hyperventilate. The physiology of this breathing pattern is not completely understood, but pulmonary congestion has been shown to activate pulmonary stretch receptors and induce ventilation.32 Reductions in cardiac output can alter the response and feedback of peripheral chemoreceptors, inducing apnoea from lower levels of CO2 compared with the normal population.7 The subsequent increase in CO2 caused by apnoea produces hyperventilation, which then leads to the cyclic breathing pattern of Cheyne–Stokes respiration.8 Furthermore, patients with HF also have impaired cerebrovascular reactivity, with an impaired ability to dampen ventilatory undershoots and overshoots, further perpetuating CSA.33
Post-hoc analyses of the initial TPNS trials evaluating patients with HF have demonstrated a significant benefit in a composite of mortality, HF hospitalisations and health status.28 At 6 months these data demonstrated a significant reduction of HF hospitalisations by 12.3% with activation of therapy.27 Recent prospective data have shown improvements in hypoxemic burden and quality of life with TPNS in patients with HF and CSA.34 The WR analysis, including longest survival, lowest HF hospitalisation rate, and >2 category difference in patient global assessment at 6 months in the subgroup analysis of the pivotal study, showed that the TPNS group experienced clinical benefit compared with the control group (WR 4.92; 95% CI [2.27–10.63]; p<0.0001).28 With these benefits, the TPNS device remains the only FDA-approved treatment for CSA in patients with left ventricular ejection fraction less than 45%.25
Atrial Fibrillation
The relationship between sleep-related disorders and AF has mostly focused on obstructive sleep apnoea; however, CPAP has failed to decrease AF burden.35 Now CSA is becoming an increasingly recognised comorbidity: specifically in the Sleep Heart Health Study, CSA was shown to be an independent risk factor for AF with a twofold higher risk.36,37 Metabolic disturbance is common in sleep disorders and can induce oxidative stress and cause endothelial dysfunction, all of which can lead to electrical and structural remodelling to facilitate the development of AF.38
While the complexity of AF is likely to complicate the mechanistic link between the two diseases, these conditions seem to be linked to simultaneous sympathovagal discharges.39 In animal models, recurrent apnoea was associated with an increase in activity of the right anterior ganglia plexus, leading to an increase in vagal tone; and further oxygen desaturation would lead to an increase in the sympathetic activity of the stellate ganglion, which would leave the atria vulnerable to induction of AF.40 When the intrinsic nervous system was chemically ablated, AF was no longer inducible.40 Other studies evaluating autonomic modulation demonstrated similar results.41 Given that there appears to be a link between apnoea and AF, the effect of the treatment of CSA for patients with pre-existing AF remains an area of interest for research, given the number of patients with concomitant devices.
Conclusion
CSA is an increasingly recognised disease entity with substantial comorbidities. The TPNS device provides a safe and effective option for treatment, with demonstrated benefit in patients with HF. Given its overlap with HF and its link to sleep-disordered breathing, AF may have an important relationship to CSA, which should be targeted in future studies.