Great attention has been paid to the SERVE-HF trial (“Treatment of Sleep-disordered Breathing with Predominant Central Sleep Apnea by Adaptive Servo Ventilation in Patients with Heart Failure”), which showed increased all-cause mortality and cardiovascular mortality in the Adaptive Servo-ventilation (ASV) group compared with the control group of conventional heart failure management alone. The results of this trial led to the recommendation by multiple ASV manufacturers and medical societies to withdraw clinical use of ASV from patients with heart failure and a reduced ejection fraction (HFrEF) less than 45%.
Sleep-disordered breathing is common in patients with HFrEF with prevalence rates of 50% to 75%. Central sleep apnea (CSA) is associated with increased mortality in heart failure (HF) and is found in 25% to 40% of this subpopulation. It is estimated that the severity of CSA increases in parallel with the severity of the HF. For several years, treatment of CSA with positive pressure ventilation was believed to favor outcomes in HFrEF with a protective effect.
In the Canadian Positive Airway Pressure for Patients with CSA and HF (CANPAP) trial, subjects were randomized to treatment with CPAP or no CPAP. This trial was terminated early; it did not show an advantage of CPAP in morbidity or mortality. A post-hoc analysis suggested that mortality might be reduced if the frequency of respiratory events per hour or apnea hypopnea index (AHI) is reduced to 15/hour or less while using CPAP.
Hoping to improve the outcomes of HF, SERVE-HF was the first randomized, large scale, multinational trial directed to treat CSA in patients with HFrEF < 45% and concomitant clinically significant sleep apnea with AHI > 15/hour of central predominance (CSA index >10/hour). Treatment arms compared the addition the ASV, one of the most effective noninvasive positive pressure ventilation technologies for central apneas that offers automated inspiratory pressure support in addition to expiratory positive pressure vs conventional medical treatment alone in the control group.
The study published in the New England Journal of Medicine in September 2015 was designed in an intention-to-treat basis with the primary end point of time to first event, a composite of death from any cause, lifesaving cardiovascular intervention (heart transplant, implantation of LVAD, resuscitation after sudden cardiac arrest, or defibrillation for ventricular arrhythmia), or unplanned hospitalization for heart failure. The study did not show significant differences in the primary end point between the ASV and control group (54.1% and 50.8%, respectively; hazard ratio, 1.13; 95% confidence interval, 0.97 to 1.31; P=.10).1
The most interesting and unexpected outcome was an increase in the all cause mortality and cardiovascular mortality in the ASV group (hazard ratio for death from any cause, 1.28; 95% CI, 1.06 to 1.55; P=.01; and hazard ratio from cardiovascular death, 1.34; 95% CI, 1.09 to 1.65; P=.006).1 These findings led to the above recommendations from manufacturers, as well as a position statement from the American Academy of Sleep Medicine. These findings cannot be extrapolated to the obstructive sleep apnea population with concomitant HFrEF or to patients with HF with preserved ejection fraction, where positive pressure ventilation has offered an advantage1 likely by a different physiologic mechanism not fully uncovered at this time, believed to be an overall effect of afterload reduction.
Selection and self-selection bias in this study was addressed in a new analysis by the same SERVE-HF investigator group published August 2017, where a time-dependent model of on-treatment analysis (done to tease out if the original results were related to the treatment assignment or to poor adherence) was conducted to understand potential causes of the initial findings in the original study. It showed patients randomized to ASV who crossed over to the control group were at higher risk of cardiovascular death than control subjects; also the control patients with crossover to ASV had a signal of lower risk of cardiovascular death risk compared with patients assigned to ASV.2 Reduced adherence to ASV treatment during SERVE-HF was a concern, since it resulted in a reduced exposure to the treatment. The on-treatment analysis showed again an increase of cardiovascular death in HFrEF patients with predominant CSA treated with ASV in addition to conventional heart failure treatment compared with the control group.2 There was no increase in cardiovascular death risk associated with ASV use intervals (dose effect). This effect is not related to the amount of hours used per night.
The effect of the recommended withdrawal of treatment in HFrEF patients with EF<45% and moderate to severe central predominant sleep apnea is being addressed in smaller studies. A single center retrospective analysis observed the effects after ASV discontinuation in this population. Thirteen out of 126 patients treated with ASV who met SERVE-HF criteria were followed for at least a year; 93% of the subjects who met criteria had ASV removed; immediate recurrence of the central apnea was observed in most (except two patients), while adverse events were not identified (defined as need for emergency hospitalization). Day and nighttime symptoms were reported by 61% of the group, and they were started on alternative treatments.3 At 1 year after ASV removal, 88% of patients were still alive, overall cardiac function did not change in 1 year (P=0.17), and seven patients required adjustment of medications for heart failure. Symptomatic patients were treated with oxygen supplementation for nocturnal symptoms or CPAP if they had daytime sleepiness. None was treated with bi-level PAP, acetazolamide, or phrenic nerve stimulation. Four patients insisted on continuation of ASV despite understanding physician concerns. 3 This study helps to demonstrate that ASV discontinuation is feasible but requires close follow-up. However, larger, long-term prospective reviews are required to draw statistically meaningful conclusions about the consequences and safety of ASV removal; these studies will be difficult to conduct under the current indications for ASV in the interest group.
At this time, investigators have shifted to further understand the causes of the increase in cardiovascular mortality, overall mortality, and the understanding of the pathophysiologic processes associated with ASV use in HFrEF. It is not known whether the effect in mortality is related to the specific ASV device/algorithm used to suppress CSA or is related to the ASV principle itself. Upcoming studies will assist in clarifying these details. Currently, there is an ongoing trial looking at the effect of ASV on survival and hospital admissions in heart failure (ADVENT–HF) using a different ASV device; this study will hopefully elucidate the impact of class effect vs device effect. It may also provide better insight of the etiology of mortality and the impact of improved ASV compliance, first addressed by the on-treatment analysis of the SERVE-HF.4
Although the reasons for increased mortality related to ASV are unclear, proposed hypotheses include: central apnea is an adaptive mechanism to HFrEF and the reversal of central apneas might adversely affect the underlying disease process,1 low adherence to ASV may impact outcomes, and specific devices may induce hyper-/hypoventilation generated by the algorithm designs of the specific ASV device and this may result in electrolyte abnormalities that generate arrhythmias.
The ADVENT-HF trial, although similar in design, has significant differences from SERVE-HF: different ASV devices may have a different impact on cardiac output and ventilation, recruited patients included those with less daytime sleepiness, and the potential to assess the effect of ASV in patients with OSA and low daytime sleepiness in patients with reduced EF.5,6 This ongoing study may help us to further understand why there is increased mortality and what effect ASV has on the treatment of sleep apnea in patients with HFrEF.