The impact of continuous positive airway pressure therapy on the recurrence of atrial fibrillation in patients with obstructive sleep apnea after pulmonary vein isolation

Abstract

Background

While obstructive sleep apnea (OSA) is recognized as a risk factor for atrial fibrillation (AF) recurrence following pulmonary vein isolation (PVI), the preventive efficacy of continuous positive airway pressure (CPAP) remains conflicting across studies, necessitating further validation.

Methods

We conducted a meta-analysis integrating data from Web of Science, PubMed, and OVID databases to evaluate the association between OSA and AF recurrence post-PVI, as well as the effect of continuous positive airway pressure (CPAP).

Results

OSA patients demonstrated a significantly elevated risk of AF recurrence compared to controls (RR = 1.67, 95% CI:1.52–1.83). Stratified analyses revealed differential effects based on left atrial (LA) size: OSA patients without significant LA enlargement exhibited an risk ration (RR) of 2.13 (95% CI:1.63-2.79), and those with enlarged LA size showed an RR of 1.78 (95% CI:1.46-2.17). Subgroup stratification by AF type revealed no significant impact of OSA on recurrence in studies with paroxysmal AF only (RR = 1.15, 95% CI:0.74-1.77), whereas an elevated RR of 1.7 (95% CI:1.54-2.11) emerged in studies combining paroxysmal and persistent AF populations. CPAP therapy reduced AF recurrence risk overall (RR = 0.6, 95% CI:0.52-0.70), but this benefit varied by LA size comparability (no LA difference: RR = 0.58, 95% CI:0.33-1.01; significant LA difference: RR = 0.61, 95% CI:0.49-0.76). CPAP efficacy was evident in mixed AF populations (RR = 0.56, 95% CI:0.48-0.67) but absent in paroxysmal AF cohorts (RR = 1.3, 95% CI:0.71-1.50).

Conclusion

This meta-analysis confirms that OSA is an independent predictor of AF recurrence following PVI. While CPAP therapy confers a general protective benefit, it does not have an effect in patients with paroxysmal AF or in those without left atrial enlargement. These results highlight the necessity of phenotypic stratification when designing and implementing CPAP-based preventive strategies.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12872-025-04846-x.

Introduction

Pulmonary vein isolation (PVI) is widely recognized as a cornerstone treatment for atrial fibrillation (AF) [1]. Nonetheless, managing AF recurrence following PVI continues to present pressing and unresolved challenges in the clinical practice. The likelihood of such recurrence is known to vary broadly across different patient groups, ranging between 20%−50% [2]. Several risk factors such as age, sex, AF type, structural heart disease, hypertension, and obstructive sleep apnea (OSA) have been recognized for their possible contribution to an elevated risk of AF recurrence following PVI [3, 4]. Therefore, effective management of these risk factors is instrumental in preventing AF recurrence. In patients with AF, OSA is of particular concern as a modifiable and treatable risk factor [5]. However, the confluence of studies investigating the effectiveness of continuous positive airway pressure (CPAP) in preventing AF recurrence post-PVI has yielded disparate outcomes [6, 7]. Although earlier meta-analyses proposed a potential advantageous role for CPAP, more recent randomized controlled trials suggest a lack of significant difference in the effect of CPAP on AF recurrence [8]. Consequently, the precise impact of CPAP on AF recurrence post-PVI warrants more detailed exploration.

In light of the context provided earlier, we initiated this meta-analysis. The primary aims of our study were as follows: (1) to examine the consequences of obstructive OSA on the recurrence of AF, and (2) to determine the effectiveness of CPAP in reducing the recurrence rate of AF among patients with OSA who underwent PVI.

Methods

Consent to participate declaration

For our meta-analysis, there is no need to provide a Consent to Participate declaration as we do not involve individual patient data. Our study is based on the aggregation and synthesis of existing research findings, ensuring anonymity and not requiring individual consent from participants.

Clinical trial number

Not applicable.

Ethical compliance

The Ethics Committee of the Second Affiliated Hospital of South China University of Technology has granted approval for the conduct of this study. The privacy and confidentiality of the participants were respected and protected throughout the research process, and all aspects of the study complied with the principles of the Declaration of Helsinki. The present study was registered on the Prospero website (Registration number: CRD42023452039, https://www.crd.york.ac.uk) in accordance with the guidelines specified by the Meta-analysis of observational studies in epidemiology (MOOSE) group.

Search strategy

The databases searched in this study included PubMed, Web of Science, and Ovid. For the investigation of the impact of obstructive sleep apnea (OSA) on AF recurrence after PVI, the search terms encompassed “atrial fibrillation,” “pulmonary vein isolation,” and “obstructive sleep apnea.” When examining the effect of CPAP therapy on AF recurrence post-PVI, additional keywords “continuous positive airway pressure” were incorporated into the search strategy. Two independent researchers evaluated the eligibility of identified studies for inclusion. In case of disagreements between assessors, a third investigator was consulted to arbitrate and determine whether the studies met the inclusion criteria for the meta-analysis.

Study selection

Only English-language literature was included in this study. “Conference proceedings,” “editorial viewpoints,” and “case reports” were excluded. The inclusion criteria for this study were as follows: 1) participants had to be above 18 years of age; 2) prospective studies, retrospective studies, and case–control studies were eligible for inclusion; 3) For the aim of contrasting the influence of OSA on AF recurrence post-PVI intervention, comprehensive details relating to patients with OSA experiencing PVI therapy were necessitated; 4) When the study goal was to probe the repercussion of CPAP on post-PVI AF recurrence, treatment data focusing on CPAP in patients suffering from sleep apnea syndrome was investigated; and 5) a minimum follow-up period of 6 months was considered.

Data extraction and quality assessment

The data extracted from the literature encompassed the following elements: article titles, publication dates, country of origin, study design, patient demographics (including sample size, gender distribution, age, and body mass index), LA size, comorbidities (such as hypertension and diabetes mellitus), classification of AF subtypes (paroxysmal vs. persistent), AF ablation techniques, diagnostic methodologies for OSA (e.g., polysomnography and Berlin Questionnaire), and post-ablation AF recurrence rates stratified by OSA status (non-OSA group, OSA patients without CPAP therapy, and CPAP-treated OSA cohort). Methodological quality assessment was performed using the Newcastle–Ottawa Scale (NOS), with two independent investigators conducting the literature screening and evaluation process. Any discrepancies in study inclusion decisions were resolved through consultation with a third investigator to determine eligibility for meta-analysis.

The included studies in this meta-analysis employed three primary methodologies for assessing left atrial (LA) size, which were collectively grouped under the term “LA size” in our analysis despite methodological variations across studies. The first method involved measuring LA internal dimension (LAD) through M-mode echocardiography in the standard parasternal long-axis view. The second approach quantified LA volume using the biplane area-length method, which combined apical 4-chamber and 2-chamber views to manually trace the endocardial borders (planimetry), followed by application of the ellipsoid formula: V = (8 × A1 × A2)/(3π × L), where A1 and A2 represent the orthogonal cross-sectional areas, and L denotes the longest atrial dimension. The third methodology focused on two-dimensional LA area measurement through direct endocardial tracing in the 4-chamber view. These distinct measurement techniques, while differing in anatomical planes and calculation parameters, were all considered valid assessments of LA size for the purpose of our comparative analysis.

Statistical analysis

Statistical analysis was performed using R software version 4.0.2 (R Foundation for Statistical Computing, Vienna, Austria), with continuous variables expressed as mean ± standard deviation and categorical variables presented as numbers or percentages. Risk ratios (RR) with 95% confidence intervals were calculated to assess AF recurrence risk. Publication bias was quantitatively evaluated through funnel plots for visual assessment and Egger’s test to determine statistical significance (p < 0.05). Heterogeneity was analyzed using Q-test and I² statistics, with significant heterogeneity defined as a chi-square p-value < 0.1. A fixed-effect model (Mantel–Haenszel method) was selected for meta-analysis when I² values ranged 0–50%, while a random-effects (DerSimonian-Laird method) model was employed when I² exceeded 50%. For studies demonstrating I² > 50%, meta-regression analysis was conducted to explore heterogeneity sources, including: intergroup differences in LA size, patient cohort composition (paroxysmal AF only vs mixed paroxysmal/non-paroxysmal cases), OSA diagnostic methodology, male proportion, follow-up duration, publication year, and mean age.

Results

We initially retrieved a total of 970 articles based on our search criteria, as shown in Fig. 1. After detecting and excluding duplicate articles (n = 604), we conducted a rigorous review of the titles and abstracts of the remaining articles, leading to the further exclusion of 157 articles. The remaining articles were then subjected to a comprehensive full-text review. Among these, 3 articles were excluded based on specific exclusion criteria, as detailed in Fig. 1. At the conclusion of this meticulous screening process, 13 articles were considered eligible and included in our study.

Fig. 1.

Flowchart of the literature screening process for the meta-analysis

Our selection of 13 articles included data from 5,542 patients who underwent PVI for AF. Among these, 3,910 individuals were not diagnosed with OSA, while 1,632 individuals were identified as OSA patients. The duration of follow-up across these studies varied, ranging from a minimum of 7 months to a maximum of 32 months. The characteristics and quality assessment of the included studies was showed in Table 1, while Table 2 presents the basic demographic details of the patients.

Table 1.

Summary of the characteristics of included studies

Investigator	Year Published	Geographical Aera	Study Method	Enrollment of Patients (n)	AF Type at Enrollment	AF treatment	OSA diagnostic method	OSA treatment	Interval of follow-up (months)	Quality score
Zhou [9]	2022	China	Prospective	182	P = 18 Ps = 144	PVI	PSG	CPAP	12	8
Tang [10]	2008	China	Prospective	178	P = 178	PVI	BQ	-	12	9
Szymanski [11]	2015	Poland	Prospective	290	P = 176 Non-P = 114	PVI	PSG	-	12	8
Patel [7]	2010	America	Prospective	3000	P = 1603 Non-P = 1397	PVI plus left atrial linear ablation	PSG	CPAP	32	9
Neilan [12]	2015	Netherland	Prospective	720	P = 249 Ps = 471	PVI plus CAFE ablation	PSG	CPAP	42	9
Jongnarangsin [13]	2008	America	Prospective	324	P = 243 Non-P = 81	PVI plus CAFE ablation	PSG	CPAP	7	7
Fein [14]	2012	America	Prospective	114	P = 53 Ps = 61	PVI	PSG	CPAP	12	9
Chilukuri [15]	2009	America	Prospective	210	P = 119 Non-P = 91	PVI	BQ	-	25	9
Chilukuri [16]	2010	America	Prospective	109	P = 84 Non-P = 25	PVI	BQ	-	11	9
Hunt [6]	2022	Norway	RCT	83	P = 83	PVI	PSG	CPAP	12	9
Naruse [4]	2012	Japan	Prospective	153	P = 82 Non-P = 71	PVI plus left atrial linear ablation or superior vena cava ablation	PSG	CPAP	18	9
Hojo [17]	2018	Japan	Prospective	100	P = 89 Non-P = 11	PVI	PSG	CPAP	24	8
Matiello [18]	2010	Spain	Prospective	174	P = 98 Non-P = 76	PVI plus left atrial linear ablation	BQ	CPAP	17	8

BQ Berlin questionnaire, CAFÉ Complex fractionated atrial electrogram, PSG Polysomnography, PVI Pulmonary vein isolation, AF Atrial fibrillation, OSA Obstructive sleep apnea, (-) Data not available, RCT Randomized controlled trial, P Paroxysmal, Ps Persistent

Table 2.

Patient characteristics of included studies

Primary investigators	Year Published	OSA					Non-OSA
		Patients	Age(years)	CPAP Patients	HT	DM	Patients	Age(years)	HT	DM
Zhou [9]	2022	182	-	62	119	76	60	61	42	40
Tang [10]	2008	104	58	-	70	17	19	56	17	5
Szymanski [11]	2015	115	60	-	91	12	136	56	92	9
Patel [7]	2010	640	51	315	224	116	2360	57	1062	283
Neilan [12]	2015	142	57	71	90	33	578	56	275	73
Jongnarangsin [13]	2008	32	59	18	23	-	292	57	127	-
Fein [14]	2012	62	-	32	42	12	30	59	21	6
Chilukuri [15]	2009	92	58	-	63	9	118	58	43	10
Chilukuri [16]	2010	48	-	-	-	-	61	-	-	-
Hunt [6]	2022	83	62	37	32	4	21	-	-	-
Hojo [17]	2018	34	-	11	24	10	66	60	25	4
Naruse [4]	2012	116	-	82	-	-	37	-	-	-
Matiello [18]	2010	42	-	-	28	-	132	51	45	-

HT Hypertension, DM Diabetes mellitus, (-) Data not available, OSA Obstructive sleep apnea, CPAP Continuous positive airway pressure

This meta-analysis of 13 studies assessed OSA’s impact on AF recurrence after PVI. Among included studies, 11 identified OSA as a significant risk factor for AF recurrence (RR = 1.73, 95% CI:1.47–2.03), while 3 showed no significant association (Fig. 2A). Moderate heterogeneity (I² = 56.1%) necessitated random-effects modeling. The pooled analysis demonstrates OSA increases AF recurrence risk by 73% following PVI. Meta-regression revealed AF subtype distribution explained heterogeneity in OSA’s impact on post-PVI recurrence (Supplementary result 1). Non-paroxysmal AF served as an effect modifier, with OSA showing stronger recurrence association in this subgroup (Q = 4.26, p = 0.04). No other covariates were found to significantly contribute to heterogeneity. Sensitivity analyses confirmed OSA’s robust AF recurrence association (pooled estimates unaffected by study exclusion, Supplementary Fig. 1A). The meta-analysis results generated using R are presented in Supplementary Result 2.

Fig. 2.

A Forest plot comparing atrial fibrillation (AF) recurrence rates after pulmonary vein isolation (PVI) between patients with obstructive sleep apnea (OSA) and non-OSA controls. B Subgroup analysis of AF recurrence rates following PVI in OSA versus non-OSA patients. (AF: atrial fibrillation; OSA: obstructive sleep apnea; LA: left atrial; RR: risk ration; CI: confidential interval)

We performed subgroup analyses to evaluate how OSA affects AF recurrence following PVI (Fig. 2B). The risk ratio for AF recurrence in OSA patients was 2.13 (95% CI: 1.63—2.79) when their LA size showed no significant difference from controls, and changed to 1.78 (95% CI: 1.46—2.17) when LA size was significantly enlarged. Subgroup stratification by AF type revealed that OSA did not significantly influence recurrence in studies involving only paroxysmal AF patients (RR = 1.15, 95% CI: 0.74—1.77). However, in studies including both paroxysmal and persistent AF patients, the RR increased to 1.70 (95% CI: 1.54—2.11). Sensitivity analyses across all subgroups consistently supported these findings without altering the conclusions (Supplementary Fig. 1).

We evaluated the effect of CPAP therapy on AF recurrence in patients with OSA (Fig. 3). The results demonstrated that CPAP treatment reduced the risk of AF recurrence in OSA patients (RR 0.6, 95% CI 0.52—0.70). Sensitivity analyses yielded consistent findings (Supplementary Fig. 2 A). The results of the meta-analysis conducted using R statistical software are presented in Supplementary Result 3.

Fig. 3.

Forest plot evaluating the effect of continuous positive airway pressure (CPAP) therapy on AF recurrence after PVI in OSA patients. (CPAP: continuous positive airway pressure; RR: risk raito; CI: confidential interval)

Subgroup analyses were conducted to examine the impact of CPAP therapy on AF recurrence following PVI. In studies where no significant difference in LA size was observed between CPAP and non-CPAP groups, CPAP exhibited no significant effect on AF recurrence risk (RR = 0.58, 95% CI: 0.33—1.01) (Fig. 4A). Notably, sensitivity analysis showed a different conclusion after excluding Hunt’s study (Supplementary Fig. 2D). In contrast, when significant differences in LA size were present, CPAP was associated with a decreased risk of AF recurrence (RR = 0.61, 95% CI: 0.49—0.76) (Fig. 4B). Examining populations with both paroxysmal and non-paroxysmal AF, CPAP use correlated with reduced AF recurrence (RR = 0.56, 95% CI: 0.48—0.67) (Fig. 4C). However, in a study focusing solely on paroxysmal AF patients, CPAP therapy did not significantly influence the risk of AF recurrence (RR = 1.3, 95% CI: 0.71—1.50) (Fig. 4D).

Fig. 4.

Subgroup analysis of atrial fibrillation (AF) recurrence after pulmonary vein isolation (PVI) in OSA patients based on continuous positive airway pressure (CPAP) utilization. A Studies reporting no significant difference in left atrial (LA) size between CPAP and non-CPAP groups. B Studies demonstrating significant differences in LA size between CPAP and non-CPAP groups. C Both CPAP and non-CPAP groups included patients with paroxysmal and non-paroxysmal AF. D CPAP and non-CPAP groups exclusively enrolled patients with paroxysmal AF. (CPAP: continuous positive airway pressure; RR: risk raito; CI: confidential interval)

Discussion

In our meta-analysis, we identified OSA as a risk factor for AF recurrence. Treatment with CPAP in OSA patients was found to significantly reduce the risk of AF recurrence. However, subgroup analysis revealed that CPAP treatment did not significantly affect AF recurrence after PVI when there was no noticeable difference in LA size between patients who received CPAP and those who did not. Significant effects were observed only when there was a notable difference in LA size. For populations with paroxysmal AF, CPAP had no significant impact on AF recurrence post-ablation. In contrast, for populations including both paroxysmal and non-paroxysmal AF, CPAP treatment significantly reduced AF recurrence after PVI.

The association between OSA and the onset of AF has been confirmed by several studies [19], a finding that our research further supports. OSA can promote the occurrence of AF through a variety of mechanisms, one of the most important of which is alterations in cardiac structure [20]. OSA-induced obstruction of the upper respiratory tract leads to significant variations in thoracic pressure, which substantially increase the difference in pressure inside and outside the heart chambers, eventually leading to atrial dilation [21]. Enlarged heart chambers can cause changes in the overall structure and fibrous makeup of the heart, thereby promoting the occurrence and progression of AF [22]. Compared to the control group, patients with OSA typically exhibit larger LA size and more severe left ventricular diastolic dysfunction [23].

An enlarged LA size has been identified as a risk factor for the onset of AF, a view supported by numerous studies [24]. Our research also underscores this conclusion, as we found that the frequency of AF recurrence after PVI significantly increased when the LA size parameters in the OSA group were substantially larger than those in the non-OSA group. However, an important point underscored by our study is that the probability of AF recurrence in the OSA group is still higher than that in the non-OSA group even when there is no significant difference in LA size parameters between the two. This suggests that in addition to the mechanism of altered cardiac structure promoting AF, OSA also involves other mechanisms that promote the occurrence and progression of AF, such as endothelial system dysfunction, inflammatory response, and oxidative stress [25].

While polysomnography remains the primary diagnostic method for OSA, the user-friendly Berlin questionnaire has emerged as a significant adjunct in clinical practice. In an earlier meta-analysis, Chee Yuan Ng et al. discovered the Berlin questionnaire effectively identifies the risk of recurrence of AF (AF) in patients with OSA following PVI [26]. This finding was subsequently corroborated by my research. These discoveries further attest to the practicality and efficacy of the Berlin questionnaire in anticipating the risk of postoperative AF recurrence in OSA patients.

Our research findings suggest that there is no significant difference in the influence of OSA on the recurrence of AF among patients with paroxysmal AF compared to those without OSA. However, when the study population includes both paroxysmal and persistent AF patients, OSA prominently emerges as a notable risk factor for the recurrence of AF. Existing literature does not provide a detailed explanation for this phenomenon. We hypothesize that this may be attributable to minimal differences in atrial structural alterations between OSA and non-OSA patients among those with paroxysmal AF, resulting in no significant divergence in their risk of recurrence.

Consistent with previous meta-analysis results, our study further confirms that CPAP can effectively reduce the risk of AF recurrence after PVI [8]. However, when the LA size of the patients participating in the study showed no significant differences, we found that the effect of CPAP in preventing recurrence of AF was not significant. Similarly, there was not a significant difference found in the impact of CPAP among patients with paroxysmal AF. Regarding this phenomenon, a definitive explanation was not identified from our review of related literature. We speculate that this observation could be related to the reasons we have previously mentioned, i.e., the atrial structural changes are likely minor among OSA patients with paroxysmal AF and those without significant changes in LA size, which might be the primary reason for the lack of significant differences in recurrence rates. Therefore, the understanding of the influence of CPAP on AF recurrence requires reevaluation, and more clinical trials are needed to further substantiate this point of view.

Limitations

In our meta-analysis, all data were not adjusted for cardiovascular related risk factors, including gender, age, baseline diseases (such as hypertension, diabetes, etc.), and baseline medication status. Furthermore, our study did not differentiate the severity of OSA. Additionally, there was substantial heterogeneity among the studies in some of the comparisons, which might have influenced the results. Lastly, most of the results selected for our study were from retrospective studies, potentially leading to bias between the experimental and control groups, thus potentially affecting the study outcomes.

Conclusions

OSA is an independent risk factor for the recurrence of AF following catheter ablation. CPAP therapy has been shown to reduce the risk of AF recurrence in patients with OSA. However, it is important to note that CPAP therapy does not significantly improve recurrence risk in patients without significant changes in LA size. The efficacy of CPAP in preventing AF recurrence differs between patients with paroxysmal and persistent AF.

Authors’ contributions

Contributions Jun Fan, MD, PhD: Conceptualized and designed the study; Contributed significantly to the writing and editing of the manuscript. Shao-Xi Sun, MD, PhD: Contributed to the selection of studies for analysis and interpretation of the findings. Li–Li Cao, MD: Assisted in data collection, extraction, and verification for the meta-analysis. Shao-Ling Luo, MD: Involved in quality assessment of selected studies and manuscript editing. Shao-Hua Wang, MD, PhD: Assisted in formulating search strategies and conducting literature search. Wei-Jie Li, PhD: Participated notably in study selection and data extraction. Yi-Chao Pan, MD: Worked on statistical analyses and interpretation of data. Tian-Yuan Wu, MD: Provided critical revisions and suggestions; Oversaw research progress. Jian Liu, MD: Reviewed the final manuscript thoroughly for intellectual content; Ensured the integrity and accuracy of data. Bing-Bo Yu, MD, PhD: Provided overall supervision and guidance for the project; Had full access to the study data, taking final responsibility to submit for publication. All authors contributed to and have approved the final version of the manuscript and agree to be accountable for all aspects of the work. #J.F., S.X.S., and L.L.C. contributed equally to this work. *Corresponding Authors: Both J.L. and B.B.Y. equally contributed to the supervisory role of this study.

Funding

This study was supported by several grants. JF was funded by the Guangzhou City Science and Technology Program (202102021100) and the Guangzhou Traditional Chinese Medicine and Integrated Traditional Chinese and Western Medicine Science and Technology Project (20222 A010003). S-hW’s efforts were financed bythe Guangzhou City Science and Technology Program (2023 A04 J0611) and Guangdong Basic and Applied Basic Research Foundation (2023 A1515111095). Additionally, B-bY’s work received financial support from the Guangzhou City Health and Wellness Science and Technology Guiding Project (20241 A011017) and the Guangzhou City School-Enterprise Joint Funding Project (SL2023 A03 J01200). Furthermore, Q-sD’s work received financial support from the Guangzhou City Major Health and Wellness Science and Technology Project (CZ0102112201).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.