Central sleep apnoea in heart failure: one size does not fit all

Abstract

Over the last decade, results from cluster analyses have offered new insights into pathophysiological endotypes of obstructive sleep apnoea (OSA), opening the possibility to developing a more personalised approach to OSA treatment. In contrast, therapeutic approach to patients with heart failure (HF) and central sleep apnoea (CSA) remains quite undifferentiated. If optimisation of the underlying HF fails to improve CSA, continuous positive airway pressure might be offered in the majority of symptomatic patients. While adaptive servoventilation (ASV) has been developed suppressing CSA and CheyneStokes respiration/periodic breathing, its use is limited to patients without HF or to those with preserved left ventricular ejection fraction of >45%. The data of the French Cohort Study of Chronic Heart Failure Patients With Central Sleep Apnoea Eligible for Adaptive ServoVentilation (FACE) trial add important puzzle pieces to our understanding of the outcome of patients with HF with CSA, thus fostering a more differentiated perspective to treatment and supporting the concepts of tailoring treatment to individual patients: 1. FACE includes patients who are typically excluded from randomised controlled trials due to their characteristics. 2. It raises the question of specific phenotypes with unfavourable outcome. The Treatment of Predominant Central Sleep Apnoea by Adaptive Servo Ventilation in Patients With Heart Failure (SERVEHF) trial studied a specific HF population with predominant CSA and left ventricular ejection fraction ≤45%. The FACE trial, which enrolled patients from 2009 to 2018, included patients who were prescribed ASV before and after publication of the SERVEHF trial in 2015. Based on realworld data in a less selective HF patient population with wide ranges of left ventricular ejection fraction and sleep breathing disturbances, the authors describe six phenotypes using cluster analysis according to anthropometric parameters, comorbidities, heart function, predominance of CSA or OSA, and ASV adherence and report the 3month cardiovascular outcome of such patient clusters. Other important prespecified outcome data of the FACE trial such as 12month and 24month cardiovascular outcome and change of diseasespecific quality of life were not included in this report. Cluster analysis and descriptive data did not include race/ethnicity of the patient population. Cluster 1 (men, low left ventricular ejection fraction, predominant CSA with CheyneStokes respiration pattern and low ASV acceptance) had the worst outcomes in the composite primary outcome at 3 months (allcause death, lifesaving cardiovascular intervention or unplanned hospitalisation for worsening of chronic HF). Secondary outcomes at 3 months (deaths and hospitalisation of any cause) were also significantly increased in cluster 2 (men, low left ventricular ejection fraction and predominant OSA). Cluster 1, representing 16% of the FACE trial population, is very similar to the SERVEHF highrisk population that showed an increased cardiovascular mortality on ASV, which is probably driven by the subgroup with even lower left ventricular ejection fraction (<30%). The FACE trial shows strikingly that in the ‘real world’ ASV is prescribed for and accepted by patients who have not been enrolled in longterm randomised clinical trials investigating cardiovascular outcomes (Clusters 4, 5 and 6; table 1). Clusters 4, 5 and 6 compromising 53% of the FACE trial population had adequate ASV adherence and represented HF patient population with factors that may increase the likelihood for positive cardiovascular effects of positive airway pressure therapy. Those factors include, in different combinations, mildly reduced or preserved left ventricular ejection fraction, more hypertension, higher body mass index, more hypoxia, and coexistence or predominance of OSA. Some of these patient clusters could challenge the principle of therapeutic equipoise, a prerequisite for randomised clinical trials. To overcome the dilemma between ethically problematic trials and lack of evidence, the full repertoire of clinical trial designs (eg, observational, randomisedregistry trials as well as randomised withdrawal and randomised clinical trials) is needed to adequately address the different patient groups and different research questions (quality of life vs longterm cardiovascular outcomes). In addition to these considerations, the FACE data raise the fact that there is a substantial variability of patients with HF with CSA, and the underlying phenotype is most relevant for outcome independent of ASV treatment. Although by design the nonrandomised FACE trial cannot in a definitive manner distinguish whether the cardiovascular outcomes were due to treatment effects of ASV or related to patient characteristics, the