Is auto-servoventilation unnecessary in patients with heart failure and apnoea?
- Michael Arzt⇑,
- Michael Pfeifer and
- Stephan Schroll
- Michael Arzt, Dept of Internal Medicine II, University Hospital Regensburg, Franz-Josef-Strauβ-Allee 11, 93053 Regensburg, Germany. E-mail: michael.arzt{at}ukr.de
From the authors:
D. Jaffuel and colleagues raised several important aspects with respect to our previously published randomised controlled trial of auto-servoventilation (ASV) in heart failure patients with sleep disordered breathing (SDB) [1]. We appreciate the opportunity to comment.
Their first point addresses the type of SDB we studied. Our aim was to include heart failure patients with an at least moderate degree of SDB. Therefore, the inclusion criterion was as clear and simple as an apnoea–hypopnoea index ≥20 events per hour of sleep, assessed by in-laboratory polysomnography. As a consequence, our sample encompassed the full spectrum of SDB types, from “pure” central sleep apnoea (CSA) to “pure” obstructive sleep apnoea (OSA), with some patients with coexisting CSA and OSA in the middle. Randomisation was stratified according to the predominant type of SDB, there being 36 patients with predominant OSA and 32 patients with predominant CSA in the trial. This design aspect allowed us to analyse the outcomes separately for CSA and OSA: both the changes in left ventricular ejection fraction (LVEF) and N-terminal pro-brain natriuretic peptide (NT-proBNP) from baseline to 12 weeks were similar in the ASV and control groups in both the OSA and the CSA patients (LVEF: OSA, 3.6±3.5% versus 4.6±6.9%; CSA, 3.7±6.3 versus 3.4±7.0%; NT-proBNP: OSA, −243±483 ng·mL−1 versus 47±219 ng·mL−1; CSA, −488±659 ng·mL−1 versus 246±912 ng·mL−1). In summary, the subanalyses of CSA and OSA patients confirmed the analysis of the entire sample showing a similar increase of LVEF in the ASV and control groups as well as a greater reduction of NT-proBNP in the ASV compared with the control group. Our data did not support that ASV would be more effective in OSA than CSA or vice versa.
Secondly, D. Jaffuel and colleagues stress the important point that we observed cardiac worsening in 8% of the patients from the ASV group and 14% in the control group. As a consequence, the medical interventions (e.g. dose of diuretics) appeared to be higher in the control arm. One may speculate whether such mechanisms may contribute to a false-negative cardiac outcome in such studies. In the analysis of the present trial, we considered this mechanism and made an effort to alleviate this potential bias by providing the per protocol analyses without the patients with changes in cardiac medication within the trial period. For future trials, the suggested design aspects of a longer run-in period including the assurance of stable cardiac medication and the inclusion of cardiac worsening as an outcome in larger trials may contribute to rule out this potential bias.
The third point raises the question that, despite randomisation, important differences could influence the results as levels of NT-proBNP were higher in the control group compared with the ASV group. We statistically accounted for the higher NT-proBNP level in the control group and found a robust reduction of NT-proBNP in the ASV compared with the control group, respectively (−372±581 ng·mL−1 versus 142±640 ng·mL−1, p=0.010) [1].
Fourthly, in the pre-specified subanalysis according to the time of ASV use, we did not find an increasing effect on LVEF associated with longer daily use of ASV. Similar to the majority of randomised controlled trials of ASV in heart failure patients with SDB, we did not find an effect of ASV on LVEF and confirmed a significant reduction of NT-proBNP by ASV [2–6]. The subanalyses of positive airway pressure (PAP) users (e.g. >4 h) are generally valuable, however, such analyses are only explanatory and can never provide the highest level of evidence. In our opinion, the better alternative would be adequately powered studies or a run-in period identifying potential participants who do not tolerate PAP therapy.
D. Jaffuel and colleagues’ final point raises the question of ventilator mode and continuous PAP therapy as an alternative control group. So far, there is no clear evidence that PAP therapy improves cardiac outcome in heart failure patients on contemporary medical and device therapy. Therefore, the first step would be to provide such evidence before comparing different modes of PAP therapy.
In summary, it is important to note that, to date, PAP therapy is indicated in a large proportion of heart failure patients having SDB with SDB-related symptoms. In addition to symptom relief, in the clinical setting, the use of continuous PAP and ASV in heart failure patients with severe SDB is associated with improved survival [7, 8]. Therefore, at present, ASV is indicated and needed in a large number of heart failure patients with SDB and should be applied in specialised centres.
The present study [1] and the ongoing large long-term trials ADVENT-HF (www.clinicaltrials.gov identifier NCT01128816) and SERVE-HF (www.controlled-trials.com identifier ISRCTN19572887) aim to complement our knowledge by evaluating the question of whether the indication of ASV treatment can be extended to heart failure patients with moderate-to-severe SDB without related symptoms or to other subgroups of heart failure patients with SDB.
Footnotes
Conflict of interest: Disclosures can be found alongside the online version of this article at www.erj.ersjournals.com
- Received March 25, 2014.
- Accepted May 14, 2014.
- ©ERS 2014