Introduction

High prevalences of sleep-disordered breathing (SDB), including obstructive sleep apnea (OSA), central sleep apnea (CSA), and Cheyne–Stokes respiration (CSR) in patients with heart failure (HF) were observed [1–3]. SDB, particularly OSA, has been confirmed to be correlated with increased incidences of left ventricular (LV) dysfunction and hypertrophy [4, 5].

Natriuretic peptide, either B-type natriuretic peptide (BNP) or N-terminal part of the propeptide of BNP (NT-pro BNP) has been recommended as a novel biomarker for diagnosis and management of HF and can reflect left ventricular systolic/diastolic dysfunction [6]. Positive airway pressure (PAP) ventilation, including continuous PAP (CPAP), bilevel PAP (BiPAP), adaptive servo-ventilation (ASV), is a widely acceptable approach for treatment of patients with HF and SDB. PAP not only ameliorates SDB, improves LV function, but also alleviates the symptoms of HF [7, 8]. However, it has yielded conflicting results that whether PAP can ameliorate BNP or NT-pro BNP [9, 10].

The purpose of the present meta-analysis was to quantitatively evaluate the efficacy of PAP ventilation on BNP or NT-pro BNP in patients with HF and SDB.

Methods

This meta-analysis was conducted in accordance with the ‘preferred reporting items for systematic reviews and meta-analyses’ [11].

Literature Search

A systematic computerized search of PubMed, Embase, Web of Science and Cochrane library was undertaken from inception to July 20, 2014 by two independent reviewers. All searches included the keywords and corresponding Mesh term: (sleep apnea or sleep-disordered breathing) and (positive airway pressure or non invasive ventilation) and (brain natriuretic peptide) and (heart failure) and (randomized controlled trial). Additionally, references in published studies were manual searched.

Study Selection Criterion

Studies were eligible if they met the following inclusion criteria: (1) study population was adult (age ≥ 18 years) with SDB (OSA, CSA or CSR); (2) HF was diagnosed according to the HF symptoms (New York Heart Association Class I-IV) and mean left ventricular ejection fraction (LVEF) less than 45 % in echocardiography. (3) study was RCT, with reasonable control group; (4) the mean duration of PAP was at least 1 week; (5) BNP levels were reported for both the experimental and control groups. Abstracts, reviews, case reports, editorials, conference articles, and non-English studies were excluded. If important data were ambiguous or lacked, the corresponding author was contacted by email, after twice non-response, the study was ruled out.

Data Extraction

The following variables were extracted from each included study by two reviewers: first author, year of publication, study population characteristics, study design, type of PAP, control type, and BNP levels in each group (experimental and control groups). If several therapeutic durations were reported in one study, the different duration was considered as a separate cohort and pooled into overall meta-analysis.

Quality Assessment

Study quality was evaluated using the Cochrane’s tool for assessing RCTs risk of bias [12]. Six items were assessed: random sequence generation (selection bias), concealed allocation (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias), and other bias. Two reviewers independently analyzed and assessed the quality of individual study, when discrepancy appeared, a third reviewer was consulted.

Statistical Analysis

All statistical analyses were performed with Review Manager Version 5.2 (The Cochrane Collaboration, Oxford, UK) and Stata Version 12.0 (Stata Corporation, College Station, Texas, USA). Standardized mean difference (SMD) with 95 % confidence interval (95 % CI) was calculated using post-intervention BNP value in each group. Random effect model was applied if high heterogeneity (I 2 > 50 %) presented, otherwise, fixed effect model was used. When data showed as median and range, mean and standard deviation was appropriately estimated [13]. Further subgroup analysis was performed after stratifying by PAP type and therapeutic duration. Analysis of publication bias was performed by Begg’s test and Egger’s test [14, 15]. Statistical significance was set at p value less than 0.05.

Results

Search Results

Figure 1 shows the flow diagram. A total of twenty-eight studies were initially found from the databases. Six studies were duplicate articles; 11 studies were excluded after browsing the titles and abstracts. The remaining 11 studies were enrolled for further full text scrutiny. Five studies were subsequently ruled out after reviewing of the full text: 3 were not RCTs [16–18], one lacked exact data of BNP [19], and one conducted PAP treatment less than 1 week [20]. Finally, 6 RCTs were pooled into the present meta-analysis [9, 10, 21–24].

Fig. 1
figure 1

Flow diagram

Full size image

The Quality of Included Studies

Assessment of the risk of bias of each study is outlined in Table 1. In general, the quality of included studies was moderate to high.

Table 1 Risk of bias
Full size table

Characteristics of Eligible Studies

A total of 6 studies (7 cohorts) with 222 patients were included into meta-analysis. The mean age was 62.3 ± 7.2 years; mean BMI was 27.6 ± 3.0 kg/m2. One study was crossover in design [10]; the remaining 5 were parallel. SDB of included studies patients varied from OSA, CSA to CSR. Patients in one study received BiPAP, and patients in the remaining 5 studies received ASV. The characteristics of the included studies are listed in Table 2.

Table 2 Characteristics of included studies
Full size table

Impact of PAP on BNP

Moderate heterogeneity (I 2 = 58.1 %) was noted, and random effect model was performed. When compared with control group, a significant reduction in BNP was observed in patients treated with PAP (SMD −0.517, 95 % CI −0.764 to −0.270, z = 4.11, p = 0.000), Fig. 2.

Fig. 2
figure 2

Efficacy of PAP on BNP in patients with HF and SDB

Full size image

Further subgroup analysis showed that the significant reduction was not yet changed after stratifying by PAP type and therapeutic duration (Table 3).

Table 3 Subgroup analysis of the efficacy of PAP on BNP by different variables in patients with HF and SDB
Full size table

Publication Bias

There was no statistical significance of publication bias in the present meta-analysis (Begg’s funnel plots in Fig. 3, p = 0.072; Egger’s test, p = 0.157).

Fig. 3
figure 3

Publication bias

Full size image

Discussion

The present meta-analysis evaluated the efficacy of PAP on BNP in patients with HF and SDB. The results including 6 RCTs demonstrated that PAP can decrease BNP levels in patients with HF and SDB.

The prevalence of SDB, including OSA, CSA, and CSR, is up to 40 %, and SDB contributes to poor prognosis in HF patients [1, 3]. Accumulating evidence elucidates that PAP can improve LVEF and left ventricular hypertrophy in patients with HF and SDB [7, 8, 19]. The potential mechanism is multi-factors: reducing sympathetic nerve activity, increasing cardiac ejection, decreasing ventricular afterload, ameliorating pulmonary congestion, improving oxygen saturation, improving dyspneic symptom. BNP is released from ventricular in response to volume expansion and pressure overload and is associated with left ventricular function and prognosis in HF [25]. BNP has been studied as primary or secondary outcome in many interventional studies. PAP types vary from CPAP, BiPAP to ASV etc., whose efficacy on BNP is inconsistent. The PAP types enrolled in our meta-analysis were ASV and BiPAP, and the results proved that both ASV and BiPAP do ameliorate the BNP concentrations in subjects with HF and SDB.

Previous studies indicated that ASV is superior to CPAP in ameliorating CSA or CSR in patients with HF [26, 27]. One previous RCT showed that CPAP did not decrease BNP levels in SDB patients without HF [28], several studies, however, demonstrated an significant improvement of BNP in OSA patients without HF as a result of CPAP treatment [18, 29]. CPAP can only alleviate 50 % of CSA [7]. Study found that ASV can not only suppress all types of SDB, but also improve sleep quality [30]. Subjects in most studies of our meta-analysis suffered HF and CSA or CSR, and the BNP levels were improved in patients received ASV therapy. The reasons of why ASV is more effective than CPAP may be as follow: ASV modality generates fixed or automatic expiratory PAP to eliminate the obstruction of upper airway and provides flexible inspiratory PAP to relieve CSA or CSR [23, 31]. Therefore, ASV may stabilize respiration, eliminate hypoxia, alleviate ventricular afterload, and even decrease ventricular hypertrophy.

There are some strengths of the present meta-analysis. Firstly, all enrolled studies were RCTs, and all of them had a high quality. Secondly, there was a large sample size (Totally 222 patients) to strengthen our conclusion. Thirdly, no evidence showed any publication bias in the present meta-analysis. Lastly, all subjects had a good PAP compliance, the PAP usage time per night in all studies was more than 4 h.

Several limitations have to be mentioned in our analysis. Firstly, control type varied in accordance with different studies design, including medical therapy, CPAP, oral appliance, and subtherapeutic ASV. Secondly, the PAP therapeutic duration was various across each study, ranging from 7 days to 12 months; however, its efficacy did not change.

In conclusion, the present meta-analysis indicated that PAP can significantly lower BNP levels in patients with HF and SDB.