Central sleep apnoea (CSA) is common in patients with chronic systolic heart failure (HFrEF). Various trials have shown a prevalence of 21 - 37% in this group of people. Up to 66% of patients with CSA and HFrEF present with periodic breathing (PB), which is considered being a marker of HF severity and poor prognosis. Brack et al. summarized data from cohorts, longitudinal studies and retrospective analyses showing an independently increased risk of death in HF patients with PB (HR 2.1-5.7 in five of seven studies). Furthermore, PB in HF patients is known to reduce quality of life and exercise performance and to increase sympathetic nerve activity as well as the probability of malignant cardiac arrhythmias.

The pathogenesis of PB is characterized by an instability of ventilatory drive. The level of carbon dioxide (CO2) in blood and cerebrospinal fluid correlates linearly with minute ventilation. A high level of CO2 increases ventilation while hypocapnia dampens it. This control theory is based on the loop gain (LG), which represents the sensitivity and reactivity of the ventilatory system and comprises three components: The plant gain defines the capacity of the system to change PaCO2 in response to a change in ventilation (metabolic response). It is influenced by the lung volume as well as the anatomy of the thorax and the upper airways. The feedback gain is defined by the chemoreceptor responsiveness in reaction to blood gas changes. The controller gain is represented by the respiratory control center in the brain stem and defines the capacity of the system to change ventilation in response to a change in PaCO2 (ventilatory response).

Sands et al. proposed and validated a mathematical model based on the ventilatory cycle pattern that quantifies the feedback loop. The ratio of ventilatory and cycle duration within the PB pattern is defined as the duty ratio (DR), which is the basis to calculate the LG. Any temporary breathing disturbance causing a PB pattern with a LG < 1 stabilizes within a few breathing cycles. A LG > 1 represents an unstable ventilatory response and slight changes of CO2 are accompanied by overshooting and undershooting of the ventilation. In that case, the polysomnography shows the typical pattern of waxing and waning of the tidal volume and effort.

HF patients typically present with an increased LG due to an impaired left ventricular function and a hyperstimulation of pulmonary vagal receptors. Furthermore, Khoo showed an increased chemosensitivity (controller gain) as well as a decreased ventilatory capacity (plant gain) in this group of people.

Sands and colleagues characterized PB considering the mean LG derived from several ventilatory cycles during non-REM sleep. This retrospective study of PB in HFrEF patients addresses the following questions:

1. Is a single LG value appropriate to characterize the individual PB?
2. Does the LG depend on sleep stage and body position?
3. Does the intraindividual LG variability allow for the discrimination of different PB phenotypes and, if so, do these phenotypes differ in further characteristics?

Inclusion Criteria:

• Systolic heart failure with left-ventricular ejection fraction <45%
• Apnea-Hypopnea index >15 per hour as determined by diagnostic polysomnography
• Predominant central sleep apnea as defined by >50% central respiratory events

Exclusion Criteria:

• <50 evaluable respiratory events for loop gain analysis during diagnostic polysomnography

Patients admitted to the sleep lab for diagnostic polysomnography.