Effects of obstructive sleep apnoea on heart rhythm

Intermittent hypoxia and the autonomic nervous system

Under physiological conditions hypoxaemia acts on the carotid body by promoting both hyperventilation and sympathetic activation [14, 15]. In contrast, upper airway obstruction, by preventing lung expansion and stretching of vagolytic fibres in the lung, elicits the diving reflex (increased sympathetic vasoconstriction to muscles and viscera in order to maintain oxygen delivery to the vital organs [9]). This leads to a rise in blood pressure and vagally induced reflex bradycardia [14, 15]. The resultant hyperpnoea on arousal stretches the peripheral afferent fibres in the lung, which is associated with a vagolytic response (Herning–Breuer reflex) and, along with arousal-related increases in sympathetic tone, greatly increase heart rate [14, 15]. Therefore, the responses to obstructive apnoea include both an increase in the sympathetic and the parasympathetic tone to the heart and peripheral vasculature.

In rodents, exposure to intermittent hypoxia leads to an increased density of noradrenergic terminals in the trigeminal sensory and motor nuclei [17]. However, this mechanism also results in maladaptive changes, such as an increase in the noradrenergic pathways within the brainstem, thereby activating the medullary region [17]. Both the activation of the medullary region, and the direct stimulation of the carotid bodies by intermittent hypoxia, lead to increased sympathetic nervous system activity and enhance catecholamine secretion from adrenal medullary chromaffin cells [18, 19].

Hypoxaemia has been suggested to induce both bradycardic and tachycardic arrhythmias. In dogs, hypoxaemic stimulation of the carotid bodies causes bradycardic arrhythmias via vagal activation and this reflex can be prevented by atropine [15]; while tachycardic arrhythmias are due to sympathetic activation and can be prevented by ganglionated plexis ablation or autonomic blockade [14, 15, 20].

Patients with OSA have a greater sympathetic nerve activity during sleep and wakefulness when compared with controls, and as a consequence an elevated heart rate and diurnal hypertension (as well as non-dipping status) are commonly observed in these patients [21, 22]. Continuous sympathetic stimuli repeated during each apnoea may promote carotid chemoreflex and increase the sympathetic drive resulting in increased heart rate and rises in systolic blood pressure at the end of each apnoeic event [22]. Evidence from cross-sectional studies in humans suggests an increase in arrhythmias in relation to the number and degree of oxygen desaturations registered during sleep [23–26]. There is ample evidence from randomised controlled trials proving that treatment of OSA with CPAP reduces sympathetic nervous system activity [27, 28].

Intermittent hypoxia and oxidative stress

In vitro studies have demonstrated that intermittent cycles of hypoxia and re-oxygenation are associated with increased oxidative stress, due to generation of ROS resulting from the inhibition of complex I of the mitochondrial electron transport chain, increased activity of reduced NADPH and xanthine oxidases, and decreased levels of antioxidants [16, 29, 30]. Findings from animal studies have shown that oxidative stress induced by intermittent hypoxia leads to protein oxidation and myocardial lipid peroxidation [31, 32]. Rodents exposed to intermittent hypoxia develop significant myocardial cell injury, such as myocyte hypertrophy, increased myocyte cell length and cell apoptosis as evidenced by rises in serum cardiac troponin I [33, 34]. Moreover, evidence of left ventricular dysfunction as assessed by increased left ventricular end-diastolic pressure has been found in rats exposed to intermittent hypoxia, in which the decrease in left ventricular function was related to the increase in lipid peroxides [31]. Such structural alterations of the myocardium may lead to micro-ischaemia, promoting cardiac repolarisation abnormalities and thus increased susceptibility to develop ventricular dysrhythmias [35].

In humans, augmented oxidative stress has been suggested to increase the risk of cardiac arrhythmia by: depletion of ATP levels, oscillations in the mitochondrial membrane potential, and changes in matrix concentrations of Ca²⁺, K⁺, NADH, ADP and tricarboxylic acid cycle intermediates [36]. These alterations create myocardial regions of inhomogeneous myocardial excitability favouring the development of re-entry arrhythmias [36, 37].

In addition, ROS may not only have a direct effect on intracellular structures, but also promote the sympathetic response in carotid bodies and the adrenal medulla's production of catecholamines in response to hypoxia [19, 38].

In patients with OSA, CPAP therapy has been proven to reduce markers of oxidative stress in a randomised-controlled trial [39].

Arousals and the autonomic nervous system

In animal models of OSA, experimentally simulated obstructive apnoea led to arousals which were associated with an increase in sympathetic activity. In dogs, spontaneous arousals from sleep were associated with acute rises in heart rate due to both sympathetic activation and parasympathetic withdrawal, and could be prevented by sympathetic block [40]. In pigs, arousals induced by tracheal obstruction during sleep were associated with increases in blood pressure, heart rate and coronary vascular resistance [41]. α-Adrenergic receptor blockade eliminated increases in blood pressure and decreased coronary vascular resistance [41].

In healthy subjects, bursts of sympathetic nerve activity and increases in blood pressure were recorded in response to arousal stimuli during sleep [42]. In patients with OSA, arousals occurring at the end of apnoeas and hypopnoeas are associated with marked transient increases in sympathetic nervous activity which led to considerable blood pressure rises [43]. Moreover, by acute hyperactivation of the sympathetic nervous system, arousals provoked coronary vasoconstriction [44] resulting in micro-ischaemia, which may increase dispersion of, and prolong, myocardial repolarisation [45]. In a case–control study, including patients with OSA and snorers and healthy subjects as a control, ST-segment depression in the electrocardiogram (ECG) as a surrogate of ventricular micro-ischaemia was observed in patients with OSA but not in controls [46]. In the latter study, daytime and nocturnal ST-segment depression episodes were related to the arousal index and increased daytime urinary epinephrine excretion [46].

Negative intrathoracic pressure and increased transmural gradients

Obstructive apnoea and hypopnoea are associated with repeated inspiratory efforts against the collapsed upper airways producing considerable negative intrathoracic pressure, which may be as low as -80 mmHg. This mechanism, repeated during each apnoeic phase, may stretch the cardiac wall and intrathoracic vessels possibly leading to both short-term electrical and long-term mechanical remodelling of both atria and the left ventricle, thereby increasing the risk for the onset of atrial and ventricular dysrhythmias [47, 48].

In vitro and in vivo studies on myocardial tissue have provided evidence for acute stretch-activated Ca²⁺ channels which produce functionally significant repolarisation gradients and promote both early and delayed after-depolarisations, thereby predisposing to ventricular arrhythmias [49, 50]. Similar results were found in an animal model, in which negative intrathoracic pressure during obstructive respiratory events induced shortening of the right atrial refractory period and consequently increased the susceptibility to premature beats and atrial fibrillation [51].

In humans, the Mueller manoeuvre, an inspiratory effort made against a closed mouthpiece and nose, has been proven to be an adequate way to induce intrathoracic pressure changes resembling those observed during an obstructive apnoea; but without the confounding effects of hypoxaemia, arousals from sleep, and comorbidities often present in patients with OSA. During the Mueller manoeuvre, the negative intrathoracic pressure transmitted to the pericardial cavity leads to an increase in the left ventricular afterload, in the pressure developed during the left ventricular isovolumetric phase, and in left ventricular volume [52, 53]. In healthy subjects, simulated obstructive apnoea or hypopnoeas were associated with an acute increase in proximal aortic diameter and in left ventricular volume, as assessed by echocardiography, while left atrial volume and left ventricular ejection fraction were reduced [53–55]. These findings suggest that repeated intrathoracic pressure changes in OSA might play a role in the onset of cardiac arrhythmias by promoting both mechanical and electrical remodelling of the heart, as evidenced by recent findings [48].

Negative intrathoracic pressure and the autonomic nervous system

In an animal model of OSA, negative tracheal pressure leads to a pronounced shortening of the atrial effective refractory period and increased atrial fibrillation inducibility, which are mediated by sympathetic α- and β-pathways, vagal activation and ganglionated plexus [56].

In humans, the negative intrathoracic pressure provoked by the Mueller manoeuvre is associated with a substantial rise, of more than 200%, in postganglionic sympathetic nerve activity leading to a significant increase in mean blood pressure at the end of the apnoea [57].

In summary, the increased susceptibility to arrhythmia in patients with OSA may primarily be due to both structural and electrical remodelling of the heart over time, which are promoted by excessive negative intrathoracic pressure and increased sympathetic activation, and thus provide an arrhythmogenic substrate for the acute triggers resulting from apnoea and hypopnoea.

Observational studies regarding atrial fibrillation

The prevalence of nocturnal atrial fibrillation in patients with OSA has been estimated to be between 3% and 5%, compared with a prevalence of between 0.4% and 1% in the general population or control subjects without OSA [10, 23, 60].

In an observational study, including obese patients, the 5-year incidence of atrial fibrillation in patients with OSA has been estimated to be 4.3% compared to a 2.1% incidence in obese patients without OSA, thus resulting in a hazard ratio for atrial fibrillation of 2.2 in patients with OSA [59].

The prevalence of atrial fibrillation has also been found to be higher in patients with coronary artery disease (CAD) and OSA, compared to patients with CAD but without OSA (32% and 18%, respectively) [61]. Atrial fibrillation has also been more frequently observed in patients with heart failure and OSA, compared to those with heart failure but without OSA (22% versus 5%, respectively) [62], and in patients with both hypertrophic cardiomyopathy (HCM) and OSA, compared to those with HCM but without OSA (31% versus 6%, respectively) [63].

In patients with atrial fibrillation, the prevalence of OSA has been found to be between 21% and 49% [64, 65], and it is higher in patients with lone atrial fibrillation, compared to matched control subjects with other cardiovascular disease (49% versus 32%) [64, 66].

OSA has been suggested to increase the risk for atrial fibrillation recurrence after cardioversion or catheter ablation. In an uncontrolled study of 3000 patients having pulmonary vein isolation (PVI) therapy for atrial fibrillation, 21% of the study population was diagnosed with OSA [65]. More recently, moderate-to-severe OSA was found to be an independent predictor for atrial fibrillation recurrence in patients undergoing PVI [67].

After an average follow-up period of 32 months, 32% of the patients without CPAP therapy experienced atrial fibrillation recurrence, while only 21% of patients with effective CPAP therapy did [65]. Furthermore, in 174 patients who had been treated with catheter ablation therapy for atrial fibrillation, the risk of atrial fibrillation ablation failure has been shown to be independently associated with the severity of OSA [68]. The probability of relapse was 52% among patients without OSA, and 86% among patients with severe OSA [68].

Studies regarding the prevalence of atrial fibrillation in patients with OSA are summarised in table 1.

Observational studies regarding atrial arrhythmias

The prevalence of supraventricular (atrial and sinus) arrhythmias has been estimated to be approximately 50% among patients with severe OSA, compared with about 25% in patients with mild OSA and 20% in control subjects without OSA [71]. In a study looking at patients with newly diagnosed OSA, a prevalence of supraventricular ectopics (SVE) of up to 98%, and SVT of up to 35%, was found [70]. Moreover, patients with OSA were found to have more sinus tachycardia and SVE compared with simple snorers and control subjects without OSA. The number of both nocturnal SVT and SVE has been shown to be significantly higher in patients experiencing nocturnal hypoxia and was related to minimum oxygen saturation during the night [46, 71]. In one of these studies the incidence of daytime SVT was correlated with increased nocturnal and diurnal urinary catecholamine excretion, and the number of daytime sinus tachycardia were correlated with apnoea/hypopnoea index (AHI) [46].

Interventional studies

In a small prospective study, recurrence of atrial fibrillation one year after cardioversion was found in 82% of patients who were not, or were poorly, treated with CPAP compared with 42% in patients who were effectively treated with CPAP [69]. However, the use of CPAP was not randomised in the latter study, and patients not using CPAP might generally be less compliant with medical therapy, thus potentially introducing a bias. In another uncontrolled study, including 316 patients with newly diagnosed OSA, CPAP therapy significantly reduced the amount of nocturnal paroxysmal atrial fibrillation and SVE from 14% to 4% [60].

A randomised-controlled trial (RCT) including patients with newly diagnosed moderate to severe OSA showed no effect of therapeutic CPAP on the frequency of SVE and SVT, when compared to patients on sub-therapeutic CPAP [70]. However, the latter study was not adequately powered to definitely answer the question of whether CPAP improves the frequency of SVE and SVT. Interventional studies on the effect of CPAP on atrial fibrillation in patients with OSA are summarised in table 1.

Due to the lack of appropriately powered RCTs, specifically looking at the effect of CPAP on atrial fibrillation and supraventricular arrhythmias, there is currently very little robust evidence to support the hypothesis that OSA is causally related to atrial fibrillation and supraventricular arrhythmias. Further RCTs investigating the effects of CPAP on the occurrence of atrial fibrillation and supraventricular arrhythmias, as well as RCTs looking at the recurrence rates of atrial fibrillation after cardioversion, are urgently needed.

Observational studies

Bradycardia, with a reported prevalence of between 8% and 95%, is commonly found in patients with OSA [72, 73], and the occurrence of bradycardia seems to be related to the extent of hypoxaemia [26]. Bradyarrhythmias, such as atrio-ventricular block, sinus pause and asystole, occur in up to 18% of patients with OSA, even in absence of cardiac conduction disease [23, 25, 26, 72, 74]. In comparison, in a healthy elderly population aged between 60 and 85 years, the prevalence of nocturnal bradyarrhythmias was only 3% [75]. In the European multicentre polysomnographic study, 58% of patients implanted with a pacemaker for sinus node dysfunction, and 68% of those treated with a pacemaker for atrio-ventricular block, had minimally symptomatic OSA and 27% fulfilled the criteria for severe OSA [76].

It has been proposed that the incidence of bradycardic arrhythmias depends on the severity of OSA, and has been observed in about 8% of patients diagnosed with an AHI <60 events·h⁻¹ and in about 20% of patients with very severe OSA (AHI >60 events·h⁻¹) [73]. Another study showed similar results, with a prevalence of bradyarrhythmias of 8% in OSA patients with an AHI >30 events·h⁻¹ compared with 2% in patients with an AHI <30 events·h⁻¹ [11]. Atrio-ventricular blocks, primarily grade II and III, are generally registered during rapid eye movement (REM) sleep when apnoeas are typically longer and oxygen desaturation is more pronounced, and have been reported to occur in about 10% of patients with OSA, compared with 1% in the healthy elderly population [23, 73, 75]. Studies reporting the prevalence of bradyarrhythmias in patients with OSA are summarised in table 2.

Interventional studies

In an early uncontrolled study including 15 patients with OSA and severe bradycardia (heart rate <30 beats·min⁻¹), or asystoles of 2.5 to 3.6 s or second-degree atrio-ventricular block, atropine was partially effective and treatment of OSA by tracheostomy was highly effective, in preventing these arrhythmias during sleep [72]. Sinus pauses are suspected to be longer in OSA patients compared with age-matched controls, and seem to be prevented by treatment with CPAP in 80–90% of patients [73, 75]. Moreover, CPAP seems to be effective in preventing both cardiac pauses of >3 s duration and bradycardic episodes (<40 beats·min⁻¹), as the frequency of these dysrhythmias was significantly decreased at 8 weeks; and they completely disappeared after 6 months of CPAP treatment in two uncontrolled interventional studies [25, 77]. In contrast, in a randomised and controlled study including patients with newly diagnosed moderate-to-severe OSA, CPAP was not associated with a reduction of bradycardic episodes [70]. However, this discordant finding may be due to differences in the study populations or in treatment times [70]. Studies describing the effect of OSA treatment on bradyarrhythmias are summarised in table 2.

In summary, to date there is preliminary evidence supporting the hypothesis that OSA is causally associated with bradyarrhythmias. However, currently there are no data from RCTs proving that CPAP is an effective therapy to reduce the number of both cardiac pauses and bradycardic episodes. This issue will need to be clarified in interventional RCTs including a carefully selected population of OSA patients with frequent bradycardic episodes.

Observational studies

In patients with OSA, the QT interval (corrected for heart rate; QT_c) has been found to be considerably prolonged at the onset of the apnoea and to shorten during the post-apnoea hyperventilation period [85]. Furthermore, in patients with OSA, QT dynamicity, a marker of augmented myocardial vulnerability to arrhythmias, has been found to be related to the severity of AHI [86].

Interventional studies

In a small uncontrolled study, repolarisation alterations were registered mainly during non-REM sleep and disappeared after CPAP therapy, thus suggesting a causal role played by OSA [85]. Moreover, OSA might lead to abnormal ventricular repolarisation, characterised by a flattened relationship between QT duration and heart rate; this finding reflects alterations in the autonomic function and seems also be reversible by CPAP treatment [87, 88].

In a randomised-controlled trial, analysis of QT_c and TpTe_c (TpTe corrected for heart rate) intervals from a 12-lead ECG found that CPAP withdrawal for 2 weeks was associated with both prolongation, and an increase in dispersion, of repolarisation, providing a possible mechanistic link between OSA, cardiac arrhythmias and SCD [89]. The increase in the length of these intervals was positively correlated with the change in the severity of OSA, suggesting that the risk for malignant dysrhythmias increases with the severity of OSA. In addition, there was a correlation between the change in the QT_c and change in the urinary noradrenaline levels, suggesting that increased sympathetic nervous system activity may be one of the underlying mechanisms between OSA and the increased dispersion of cardiac repolarisation [89]. According to Panikkath et al. [83] an uncorrected TpTe interval >100 ms, and a QT_c interval >430 ms in the resting ECG, are associated with an increased risk of SCD. Using these thresholds in the latter randomised controlled trial, more than 50% of patients with OSA were found to be at increased risk for SCD after 2 weeks of CPAP therapy withdrawal [89].

Thus, there is strong RCT evidence showing that OSA is associated with cardiac repolarisation disturbances, and CPAP seems to be effective in normalising cardiac repolarisation in patients who are highly compliant with CPAP. However, future interventional trials will need to address the question as to whether long-standing OSA is associated with chronic and irreversible alterations of cardiac repolarisation. Due to the ethical issues related to a long-term control group, it will be very difficult to perform RCTs to clarify this point.

Observational studies

In patients with OSA, the prevalence of nocturnal ventricular premature complexes (VPCs) has been found to be between 14% and 74%, whereas VPCs have been estimated to occur in about 5% of the general population [1, 23, 24, 72]. The wide range of the reported prevalence may be explained by different OSA severities in the studied populations. Recent findings from the Sleep Heart Health Study, based on one night of monitoring, suggest that subjects with severe OSA have a prevalence of 5% for non-sustained ventricular tachycardia and of 25% for complex ventricular ectopy, such as bigeminy, trigeminy or quadrigeminy, with an odds ratio of 3.4 and 1.8, respectively, when compared with the general population [10]. VPCs were observed mainly during the night at the end of apnoeas, and when oxygen desaturations were most pronounced [90]. Recently, two studies investigated the time of VPCs occurrence during the night and revealed an increased onset during the apnoeic phase and shortly after events of disturbed breathing when compared with normal breathing periods, thus supporting possible direct effects of OSA on the inducibility of arrhythmias [91, 92].

Similar results were observed in a prospective study, where 58% of patients with an AHI >10 events·h⁻¹, and 82% of patients with mean nocturnal oxygenation <90%, had supra- and ventricular ectopics [71]. In a study including 38 patients with reduced left ventricular ejection fraction treated with a cardioverter-defibrillator, 41% had sleep-related breathing disorders (OSA and Cheyne–Stokes breathing) and VPCs occurred significantly more often during sleep disordered breathing than during normal respiration [93]. Koshino et al. [90] found that 60% of all patients with ventricular arrhythmias referred for catheter ablation or implantation of a cardioverter-defibrillator met the criteria for OSA (AHI ≥5 events·h⁻¹), and 34% met the criteria for moderate-severe OSA (AHI ≥15 events·h⁻¹).

In a larger study, including 472 patients with congestive heart failure, 32% were found to have untreated OSA and 34% had CSA. After a 48 month follow-up period, sleep disordered breathing was found to be independently associated with an increased risk for ventricular arrhythmias [94]. Moreover, OSA patients treated with catheter ablation for ventricular dysrhythmias seem to have a higher relapse rate compared to those without OSA (45% versus 6%) [95]. In a case–control study, comparison of the occurrence rate of VPCs between patients with OSA, simple snorers and healthy subjects showed that patients with OSA have a significantly higher rate of VPCs than control subjects, which seems to be related to increased sympathetic activation [46]. However, this is in contrast to Miller [96] who reviewed 24 h Holter ECGs of 23 patients with OSA, and found that the prevalence of VPCs and ventricular arrhythmias during sleep is not different when compared to wakefulness [96]. In a more recent study, ECG data from 257 patients with newly diagnosed OSA were analysed [11]. Only 9% of patients had VPCs and no correlation with OSA severity was found [11]. Because of these conflicting findings it is uncertain whether ventricular arrhythmias are more frequently observed in patients with OSA than in comparable populations. The studies describing the prevalence of ventricular arrhythmias in patients with OSA and vice versa are summarised in table 3.

Interventional studies

In an uncontrolled study including 15 patients with OSA, nocturnal VPCs and episodes of ventricular tachycardia could be partially resolved after atropine administration, and almost completely abolished after tracheostomy, suggesting a possible role of OSA in eliciting these arrhythmias [72].

In a randomised-controlled study including 18 patients with OSA and heart failure, CPAP therapy for 1 month reduced the number of VPCs by 58% when compared to patients without CPAP therapy, and this effect was related to a reduction of urinary noradrenaline excretion [98]. In another randomised-controlled study, which included 83 patients with moderate-severe OSA, 24 h Holter monitoring was performed before and after 1 month of therapeutic or sub-therapeutic CPAP therapy [70]. The authors did not find any statistically significant change for any ventricular arrhythmia during night or day, although a trend towards less daytime ventricular tachycardia was observed in the therapeutic CPAP group. However, the interpretability of this study is limited by the relative small number of patients with arrhythmias included [70]. Interventional studies regarding the association between ventricular arrhythmias and OSA are summarised in table 3.

Due to the very limited data available from RCTs, it is currently not known whether there is a causal relationship between OSA and ventricular arrhythmias. To date, there is also no evidence proving that CPAP is an effective therapy to re-establish a normal heart rhythm in OSA patients with ventricular arrhythmias. Thus, there is a need for well-designed and appropriately powered controlled trials evaluating the impact of CPAP on ventricular arrhythmias in patients with OSA.