Heart failure (HF) is recognized as a worldwide epidemic, with high morbidity and mortality. Its prevalence is around 1-2% and can exceed 10% in those aged 70 years or over.1 In Portugal, it ranges from 1.36% in those aged over 25 years to 16% in those over 80; the prevalence of HF with reduced ejection fraction (HFrEF) and HF with preserved ejection fraction (HFpEF) is 1.3% and 1.7%, respectively.2 Despite therapeutic advances, the short- and long-term prognosis of HF remains poor, which has prompted an ongoing search for risk modifiers and treatments that can demonstrably improve prognosis. In this context, in recent years there has been considerable interest in sleep-disordered breathing (SDB), a condition that is frequently associated with worse prognosis in HF patients. The main form of treatment for SDB in HF is non-invasive ventilation, either by continuous positive airway pressure (CPAP) or bilevel positive airway pressure (BPAP) for patients who mainly suffer from obstructive sleep apnea (OSA), or by adaptive servo-ventilation (ASV) for those with predominantly central sleep apnea (CSA). It has been hypothesized that in addition to optimal HF therapy, such treatment could bring added benefits for HF patients.

Preliminary observational studies suggested that treatment of SDB could modify the prognosis of HF, reducing hospitalizations and long-term mortality.3 However, these expectations were frustrated by the recently published results of the SERVE-HF trial,4 in which treatment of SDB by ASV in fact increased mortality in patients with symptomatic HFrEF. The discrepancy between the results of observational studies and those of SERVE-HF call into question the pathogenesis of SDB in HF, challenging clinicians and researchers to improve their knowledge in this area. To this end, we present a review of the subject.

Sleep apnea syndrome is characterized by repeated pauses in breathing resulting in sleep fragmentation, daytime sleepiness and blood oxygen desaturation. According to the American Academy of Sleep Medicine, apnea is defined as a drop of ≥90% in respiratory flow from baseline lasting for ≥10 s, and hypopnea as a drop of ≥30% in respiratory flow from baseline lasting for ≥10 s in association with either ≥3% arterial oxygen desaturation or an arousal.5

Patients with HF can suffer from CSA, OSA, or both. The prevalence of CSA increases with increasing severity of HF. OSA is caused by upper airway collapse with increased reflex respiratory effort. Predisposing factors include obesity and unfavorable craniofacial structure.

In CSA, there is no strengthening of the respiratory reflex during episodes of apnea or hypopnea. The underlying mechanism is increased sensitivity of chemoreceptors in the brainstem and the carotid bodies to small variations in partial pressure of carbon dioxide (PaCO2), leading to inappropriate hyperventilation.6,7 The resulting reduction in PaCO2 weakens the respiratory stimulus and apnea or hypopnea occurs; when PaCO2 rises, the cycle repeats. In addition to this excessive respiratory response to CO2 levels, another mechanism at play in hyperventilation is stimulation of vagal mechanoreceptors (J receptors) due to pulmonary congestion via vagal afferents.8–11 CSA with a cyclic pattern of hyperventilation-apnea-hyperventilation is known as Cheyne-Stokes respiration.

The apnea-hypopnea index (AHI) is the number of events (apnea and/or hypopnea) per hour of sleep. To be considered as CSA, the number of central events must be more than 50% of the total number of events recorded by polysomnography. Sleep apnea can be classified as mild (AHI 6-14/h), moderate (15-29/h) or severe (>30/h). Although this is not universally accepted, there is general agreement that the threshold requiring treatment in symptomatic patients is an AHI >15/h.12 This cutoff is further complicated in HF patients by the fact that the clinical features of SDB in HF are atypical, and that the symptoms of HF can mimic those of SDB and vice versa.

OSA is considered an independent risk factor for HF.13 The main pathophysiological mechanisms by which it worsens HF are increased intrathoracic pressure, resulting in raised pre- and afterload leading to left ventricular compromise, excessive activation of the sympathetic nervous system, and hypoxemia.14

CSA is thought to be a consequence rather than a cause of HF.15 Besides hypoxemia, sympathetic activation is predominant, resulting in increased peripheral vascular resistance, tachycardia, and stimulation of the renin-angiotensin-aldosterone system.

In both forms, variations in blood pressure and heart rate contribute to endothelial dysfunction and elevated inflammatory markers, particularly C-reactive protein.16,17

Patients with SDB are at increased risk of ventricular and other arrhythmias, especially atrial fibrillation.18–20 In an observational study, appropriate therapies from an implantable cardioverter-defibrillator were more frequent at night in patients with OSA and during the day in those with CSA.21 In addition to arrhythmic risk, SDB also is associated with increased risk of atherothrombotic phenomena22 and metabolic alterations, particularly type 2 diabetes.23

Recently, and particularly after the SERVE-HF trial,4 the idea has gained ground that CSA in HF patients is a compensatory mechanism that has some beneficial effects, including the provision of periodic rest to fatigue-prone respiratory muscles, attenuation of excessive sympathetic nervous activity, avoidance of hypercapnic acidosis and increases in end-expiratory lung volume and intrinsic positive airway pressure.24

The prevalence of SDB in HF is high (over 50%). In a study of 700 patients with HFrEF in New York Heart Association (NYHA) class ≥II, Oldenburg et al. found a prevalence of 76% (40% CSA and 36% OSA) using a cutoff AHI value of ≥5/h.25 Schulz et al., studying 203 patients with HFrEF in NYHA class II-III, observed a prevalence of 71% (43% CSA and 28% OSA) with a cutoff of AHI ≥10/h,26 while in a study by Vazir et al. of 55 men with mildly symptomatic HFrEF, 53% presented SDB defined as AHI ≥15/h.27

In patients with HFpEF, the prevalence of SDB is also high. Using a cutoff of AHI >10/h, Chan et al. found a prevalence of 56%, mainly OSA,28 while Bitter et al., using a definition of AHI ≥15/h, observed a prevalence of 69%, mainly OSA.29

A prospective observational study in Germany including 1500 patients with HFrEF revealed a prevalence of SDB of 80% despite optimal HF therapy.30 Snoring was more prevalent in patients with OSA, while atrial fibrillation and ischemic cardiomyopathy were more common in the CSA group. The most frequent symptoms were nocturia, witnessed apnea and fatigue, which illustrates the difficulty of distinguishing these symptoms from those of HF itself.

In daily clinical practice, HF patients with SDB rarely report daytime sleepiness, possibly because of sympathetic hyperactivation, and thus many of them remain undiagnosed.31 Questionnaires such as the Epworth Sleepiness Scale,32 the Berlin Questionnaire33 and the Maislin Questionnaire34 are used to screen for SDB, although their usefulness has been questioned.

Polysomnography remains the gold standard for diagnosis of SDB. However, despite the comprehensive and detailed information it provides, polysomnography requires specialized technical and human resources, and is slow and expensive, with long waiting lists. There is thus a need for simpler and more accessible protocols, and a large number of portable devices have been developed for this purpose using techniques that include simplified cardiorespiratory polygraphy with single or multiple channels, oximetry and/or nasal air flow,35,36 which could be alternatives to polysomnography for screening purposes.

One such device is the ApneaLink (ResMed Corp, San Diego, CA). A study by Erman et al. validating this device in screening for SDB in 59 diabetic patients showed that the ApneaLink provided reliable information and was easy to use, with high sensitivity and specificity in calculating AHI compared to polysomnography, especially for AHI ≥15/h (sensitivity 91% and specificity 95%).37 Ragette et al. also found an excellent correlation between the ApneaLink and polysomnography, with better results for AHI ≥10/h.38 In addition, the device showed good diagnostic ability compared to polysomnography (sensitivity 87.1% and specificity 94.9%) for detecting CSA events in a study of 70 individuals (41 with HF) with Cheyne-Stokes respiration.39

Given its pathophysiology, it is no surprise that SDB is associated with worse prognosis in HF patients. OSA is described as a cardiovascular risk marker that is more often found in mild forms of HF (NYHA I-II), while CSA tends to be an indicator of more severe cases (NYHA III-IV).

OSA was associated with increased mortality in patients with HFrEF over a mean follow-up of three years (hazard ratio [HR] 2.81, p=0.029)40 and in patients with ischemic HF after a mean follow-up of 32 months (HR 3.03, p=0.043).41 Other authors report similar results in patients in whom CSA was identified during hospital stay, with lower survival in those with more severe disease.42

In a study including 784 patients with systolic HF, CSA was an independent predictor of six-month cardiac readmission,43 and in another study, the same authors reported that both OSA and CSA are independent risk factors for long-term post-discharge mortality in patients hospitalized for acute HF.3

In patients with HF and CSA, one of the most important prognostic indicators is the nocturnal hypoxemic burden. In a study of 963 patients with HFrEF, time with peripheral oxygen saturation <90% (T90) was associated with a 16% increase in all-cause mortality per hour of T90 in multivariate analysis, while AHI was no longer significant after adjustment for confounding factors.44 The best cutoff for mortality in this population was a T90 of at least 22 min. According to these authors, nocturnal hypoxemic burden could be an important therapeutic target and reducing hypoxemia could be more useful than reducing AHI.

SDB is a common comorbidity in HF that worsens morbidity and mortality, and thus HF specialists, especially cardiologists and internists, must be able to suspect and diagnose the condition and refer patients to sleep specialists, as well as to monitor the effects of treatments administered to their patients.

Results of studies on various treatments have been inconsistent. Following SERVE-HF, ASV is contraindicated for patients with HFrEF and CSA. However, this finding cannot necessarily be extrapolated to other populations, such as those with HFpEF or CSA. SERVE-HF has opened new horizons for research, and studies currently under way may be crucial in defining the most appropriate approach to SDB in HF.

The authors have no conflicts of interest to declare.