Consecutive adult patients with symptomatic, drug-refractory persistent AF referred for catheter ablation who underwent preprocedural CCT were screened between September 2015 and June 2018 in 5 European Cardiology Department (the University Hospital of Poitiers, University Hospital of Dijon and European Hospital Georges Pompidou in France, University Hospital of Innsbruck in Austria and Heart and Vascular Center of Semmelweis University in Hungary). Patients with previously known CAD or revascularization during follow-up were excluded. The study was conducted in accordance with the Declaration of Helsinki. All the patients included in this study received an information. According to institutional policy, approval from the Institutional Review Board was not required.

Cardiac computed tomography was performed routinely 1–4 weeks prior to AF ablation using standard-of-care, site-specific protocols. The complete heart volume was acquired on a high definition CT scanner within one gantry rotation, allowing a good image quality of the coronary arteries (14, 15). CCT analysis was performed by a single reader blinded to the procedural electrophysiology result. According to the previous study published, the severity of CAD was classified in two categories: obstructive (luminal stenosis ≥ 50%) or non-obstructive (absence of CAD or luminal stenosis < 50%) (16). Localization (right coronary, circumflex, and left anterior descending artery) of the lesion was also specified.

Atrial fibrillation ablation procedures were performed using a cryoballoon ablation catheter (Arctic Front Advance; Medtronic, Inc, Minneapolis, MN, USA) or an open-irrigated radiofrequency ablation catheter (NaviStar, ThermoCool, or ThermoCool SmartTouch; Biosense Webster, Inc, Diamond Bar, CA, USA; Flexability, Tacticath, Tacticath SE Abbott, Mineapolis, MN, USA) using an electroanatomical mapping system (Carto 3 or EnSite Precision). Briefly, all the patients underwent standard pulmonary vein isolation and electrical isolation was confirmed by a circular multipolar electrode mapping catheter. Additional ablation lesions, such as ablation of complex fractionated atrial electrograms, or mitral, roof, and cavotricuspid lines, were performed at the discretion of the operator.

The primary objective was to assess the association between obstructive CAD and AF recurrence after persistent AF ablation. Data were collected using an anonymized spread-sheet-based template. Socio-demographic and clinical characteristics were collected at the time of AF ablation. CHA₂DS₂–VASc score was calculated at admission (17). Before AF ablation, all the patients underwent standard two-dimensional transthoracic echocardiography to assess left ventricular ejection fraction, calculated according to the recommendations of the American Society of Echocardiography (18).

Patients were followed up from the time of AF ablation for at least 1 year. All the hospitalization and consultation reports were examined. If necessary, general practitioners and referring cardiologists were contacted to provide previously missing information. Atrial fibrillation recurrences were assessed after a 3-month blanking period (19, 20), and defined as ≥ 1 AF episode recorded during a 12 lead ECG or ≥ 1 AF episode lasting ≥ 30 s documented by Holter monitoring (21). Most of patients had an ECG at 3 months, and a 24-h Holter monitoring at 6, 12, and 24 months. Furthermore, in case of symptoms patients were asked to record an ECG.

Continuous variables were expressed as mean and SD or median and interquartile range as appropriate, and categorical variables were reported as numbers and percentages. Comparisons between groups were performed using the Student t-tests or the Mann–Whitney U tests for continuous variables as appropriate, and χ² test for categorical variables. Cumulative incidence curves were built according to the presence or absence of circumflex artery obstruction using the Kaplan–Meier method and compared using a log-rank test. For analysis of the predictive value of circumflex artery obstruction, recurrences during the first 3 months after the ablation procedure (blanking period) were not taken into account (20). Multivariate Cox analysis using an entry procedure was performed on an initial model including factors known to be associated with AF recurrence (age, body mass index, hypertension, diabetes, indexed left atrial volume, left ventricular ejection fraction, ß-blockers, amiodarone, pulmonary vein isolation alone or with additional lesion set, circumflex artery obstruction). Left anterior descending artery and right coronary artery occlusion were forced into the model in order to see if the association between circumflex artery obstruction and AF recurrence changed. Analyses were performed using SPSS 22 (SPSS, Inc., Chicago, IL, USA) and SAS 9.3 (SAS Institute Inc., Cary, NC, USA) statistical software. Two-sided p-values of less than 0.05 were considered statistically significant.

Among 652 patients who had preprocedural CCT for persistent AF ablation, 156 (23.9%) were not included because of CAD or coronary revascularization, or impossible coronary assessment due to the quality of the exam. Finally, 496 patients were analyzed. Table 1 illustrates the baseline characteristics of the population. Mean age was 61.8 ± 10.0 years, and 378 (76.2%) patients were men. Mean body mass index was 29.0 ± 5.2 kg/m² and 323 (65.1%) had previous hypertension. CHA₂DS₂–VASc score was of 0 or 1 point in 225 (36.6%) patients and 2 or 3 points in 210 (42.3%) patients. Mean left ventricular ejection fraction was 56.4 ± 11.1% and mean left atrial indexed volume was 65.4 ± 23.7 ml/m². In the terms of medications, 419 (84.4%) patients were prescribed anticoagulant therapy, 327 (65.9%) ß-blockers, and 181 (36.5%) amiodarone.

Ablation was performed with radiofrequency ablation catheter in 431 (86.9%) patients and with cryoballoon catheter in 65 (13.1%). Among them, 294 (59.3%) had pulmonary vein isolation alone and 202 (40.7%) underwent additional lesion sets.

Obstructive CAD was present in 86 (17.3%) patients (Table 2). Obstructive CAD was observed in the left anterior descending artery, right coronary artery, and circumflex artery with 76 (15.3%), 24 (4.8%), and 25 (5.0%) patients, respectively.

During the mean follow-up of 24 ± 19 months, AF recurrence occurred in 207 (41.7%) patients. AF recurrence rate was not different when considering the number of obstructed coronary arteries: 42.9% in the non-lesion group, 41.1% in the single-vessel lesion group, 41.1% in the 2-vessel lesion group, and 40.8% in the 3-vessel lesion group (P = 0.98).

In addition, AF recurrence rate did not differ between patients according to the presence of obstructive CAD [absence of obstructive CAD: 170 (41.5%) patients; obstructive CAD: 37 (43.0%) patients; P = 0.79]. When considering the location of the obstruction there was no difference in relation to outcome seen between those with vs. without obstruction in the left anterior descending and right coronary arteries. On the other hand, the recurrence rate was significantly higher in case of circumflex artery obstruction (Figure 1). The Kaplan–Meier analysis confirmed a higher risk of AF recurrence in the group in which circumflex artery was stenosed. At 1 year, documented recurrence of AF had occurred in 48.0% (12 out of 25) of patients with circumflex artery obstruction and in 26.1% (123 of the 471) of patients without circumflex artery obstruction. At 2 years, documented recurrence of AF had occurred in 56.0% (14 of the 25) of patients with circumflex artery obstruction and in 32.1% (151 of the 471) of patients without circumflex artery obstruction (Log-rank hazard ratio [HR], 2.06; 95% confidence interval [CI], 1.02 to 4.16; P < 0.01) (Figure 2). Baseline characteristics of the patients according to circumflex artery obstruction are given in Supplementary file 1.

Multivariable Cox regression analysis showed that body mass index (HR 1.04; 95% CI 1.01–1.07 per 1 kg/m² increase; P = 0.02), left atrial indexed volume (HR 1.01; 95% CI 1.01–1.02 per 1 ml/m² increase; P < 0.001), and circumflex artery obstruction (HR 2.32; 95% CI 1.36–3.98; P < 0.01) were independently associated with AF recurrences at follow-up after persistent AF ablation (Table 3). When left anterior descending artery and right coronary artery occlusion were forced in the model, those two variables were not associated with AF recurrence and circumflex artery obstruction was still associated with AF recurrences (HR 2.45; 95% CI 1.37–4.36; P < 0.01).

Previous studies found prevalence of asymptomatic obstructive CAD ranging from 24 to 41% (16, 22), which was substantially higher than the proportion in our study (17.3%). This difference might be explained by the lower prevalence of cardiovascular risk factors and above all by the higher prevalence of women in our study compared to the one authored by Nucifora et al. Mito’s work had also an older population compared to ours.

This is the first multicenter study assessing the impact of coronary artery stenosis on AF recurrence after persistent AF ablation. Coronary artery disease might be responsible for AF with several mechanisms: chronic inflammation, heart failure causing atrial stretch, and finally ischemia, all of which are responsible for fibrosis (4, 23–25). In our study, AF recurrence was not significantly different between patients with obstructive CAD and patients without obstructive CAD. These results are in line with a previous study carried out on a limited number of patients with shorter follow-up (12). In order to explore a possible effect of the CAD localization, Kornej et al. tried to determine whether a right coronary stenosis was associated with outcomes after AF ablation but the recurrence rate did not depend on obstruction (26). The present work is the first to evaluate the role of each coronary artery.

The anatomy of coronary arteries supplying the left atrium has been described as originating from the first segment of the circumflex artery (27, 28), and Alasady et al. demonstrated that the involvement of the left atrial coronary branch during myocardial infarction was associated with AF onset (29). Moreover, a study on experimental myocardial infarction on swine demonstrated that proximal circumflex artery occlusion involving the left atrial branch was associated with atrial infarction and atrial structural remodeling, characterized by early left atrial dilation, dysfunction, and fibrosis. When proximal circumflex artery occlusion was performed in swine without involving the left atrial branch, only interstitial atrial fibrosis was found with a lesser degree of left atrial dilation or dysfunction, and in cases of left anterior descending occlusion, no atrial fibrosis was found, and there was no left atrial remodeling (24). Taken together, these studies highlight the involvement of the circumflex artery in left atrial vascularization and suggest atrial ischemia and infarction as potential mechanisms of atrial fibrosis and AF. In the present work, only circumflex stenosis, without known clinical infarction, was associated with a higher rate of AF recurrence. These results argue for a regional and targeted, not just global (30), pathophysiological mechanism of atherosclerosis on the perpetuation of atrial fibrillation (31). The other consequence of these findings is that not only symptomatic myocardial infarction, but also silent CAD, is associated with AF recurrence. The mechanism leading to AF recurrence could consequently be silent chronic atrial ischemia related to circumflex coronary stenosis. To shed further light on these results, atrial substrate evaluation through electroanatomical mapping and cardiac magnetic resonance imaging combined with detailed coronary angiography could add new data on the role of the circumflex artery and ischemia in atrial fibrillation.

The absence of uniformity in CCT protocols is a methodological limitation. Nevertheless, we were able to highlight the role of circumflex artery stenosis in AF recurrence, which strengthens this result on account of the assessment heterogeneity. Second, the latest classification of coronary artery disease (CAD RAD) was not used in our study because it was not available on all CCT reports (32). However, we have chosen a cutoff (50%) that has already been used in the literature and which reflects a significant rate of CAD (16). In addition, the small number of patients with circumflex artery stenosis (25 patients with LCX obstruction vs. 471 patients without LCX) may have a significant impact on the statistics. Minor change in number recurrence might come to statistic difference. Finally, given the limited CT scan resolution for small artery assessment, we were not able to assess lesions located on the atrial branch of the circumflex artery.