A left atrial versus a biatrial lesion set for persistent atrial fibrillation ablation during open heart surgery†

Abstract

OBJECTIVES

To compare the outcomes of left atrial ablation and biatrial (BA) ablation in patients with persistent and long-standing persistent atrial fibrillation undergoing open heart surgery.

METHODS

Between January 2007 and December 2016, 588 consecutive patients with either persistent (20.9%) or long-standing persistent atrial fibrillation (79.1%) were enrolled in this study. To reduce between-group differences, propensity score-matched groups (156 patients/group) were obtained, with similar preoperative and perioperative characteristics.

RESULTS

The propensity score-matched left atrial and BA groups did not differ regarding 30-day mortality (1.9% vs 0.6%; P = 0.617), 5-year overall survival (93.5 ± 2.2% vs 92.8 ± 2.9%, P = 0.998) or survival free from thromboembolic events (97.1 ± 2.8% vs 96.2 ± 2.7%, P = 0.309). The BA lesion set was more beneficial in terms of freedom from atrial arrhythmia recurrence (85.3 ± 4.5% vs 91.9 ± 3.1%, P = 0.049; hazard ratio 3.26; 95% confidence interval 1.33–7.99), but it was associated with higher pacemaker implantation rate (3.8% vs 17.3%; P < 0.001) due to sinus node dysfunction. There was no significant between-group difference regarding the incidence of atrioventricular conduction disturbances (3.2% vs 7.0%, respectively; P = 0.211).

CONCLUSIONS

The 2 lesion sets (left atrial and BA) are associated with similar 30-day mortality, survival rates, incidences of embolic events and atrioventricular conduction disturbances. In patients with persistent atrial fibrillation, concomitant surgical ablation with BA lesion set provided better freedom from atrial arrhythmia recurrence, but BA ablation was associated with a higher incidence of sinus node dysfunction and a higher rate of pacemaker implantation.

INTRODUCTION

Atrial fibrillation (AF) is known to be independently associated with an increased risk of all-cause mortality [1]. Concomitant surgical ablation of AF effectively restores sinus rhythm and improves quality of life, having gained widespread use in patients undergoing open cardiac surgery [2–5]. Nevertheless, there is no consensus regarding the optimal lesion set for treating AF in patients undergoing cardiac surgery. Recent studies, including 2 randomized trials [2, 6], have reported comparable effectiveness for left atrial (LA) ablation and biatrial (BA) ablation [7, 8], whereas other studies, including 2 meta-analyses, reported that BA ablation provides benefit in AF freedom [5, 9, 10]. The main limitation of the BA ablation set is the increased risk of pacemaker implantation [7, 9, 11]. Some authors suggest that BA ablation may represent a reasonable strategy for patients with persistent or long-standing persistent AF [5, 12, 13]. According to current guidelines [14] for the management of AF, maze surgery (preferably BA) should be considered in patients undergoing cardiac surgery. In our clinical practice, we routinely use both lesion sets, giving us ample experience and the opportunity to assess the outcomes in terms of the balance between efficacy and risk of additional morbidity. The aim of this study was to compare the clinical outcomes of LA ablation and BA ablation lesions in patients with persistent and long-standing persistent AF scheduled to undergo open heart surgery.

MATERIALS AND METHODS

Patients and study design

This was a retrospective database study involving propensity score analysis. On obtaining approval from the institutional review board (approval no. NST000231206), data for all patients were extracted from the medical records maintained in the Institutional Research Database.

Between January 2007 and December 2016, 898 consecutive patients with valvular heart disease and documented AF underwent open heart surgery and concomitant AF ablation. The inclusion criteria were age ≥18 years and indication for valve surgery in accordance with the European Society of Cardiology and the European Association for Cardio-Thoracic Surgery guidelines [15]. The exclusion criteria were paroxysmal AF and minimally invasive mitral valve surgery (Fig. 1).

Finally, the study included 588 (65.5%) patients, of whom 123 patients (20.9%) had persistent AF and 465 patients (79.1) had long-standing persistent AF. The patients were stratified according to the choice of lesion set: LA ablation in 419 (71.3%) patients and BA ablation in 169 (28.7%) patients. The primary end point was freedom from atrial arrhythmia recurrence (AAR). Secondary end points included 30-day mortality, long-term survival and rate of permanent pacemaker implantation. Drug management and cardiac rhythm assessment are provided in the related Supplementary Material section.

Surgical ablation technique

In all procedures, exposure was achieved through midline sternotomy, with standard cardiopulmonary bypass and mild hypothermia (33–34°C). For myocardial protection, cold cardioplegic crystalloid solution was used (Custodiol^®; Kohler Pharma, Alsbach-Hahnlein, Germany) with antegrade flow.

Lesion set choice was based on the surgeon’s preference. Bipolar radiofrequency ablation and a cryoablation probe were used. The ligament of Marshall was dissected by electrocautery. LA lesion set procedures were performed according to the Cox maze IV pattern, which included encircling pulmonary vein isolation and endoepicardial connecting lines between the myocardial tissue islands on the roof and inferior part of the posterior wall of both pulmonary veins (box lesion). The ablations around the right and left pulmonary vein orifices and LA appendage were performed epicardially, usually before aorta cross-clamping. Ablation between the left upper pulmonary vein orifice and the LA appendage base connection was performed endocardially. Right atrial lesions included superior–inferior vena cava lines, lesions to the tricuspid valve annulus at the 12 o’clock position and a lateral line. In all cases, the LA appendage was excluded using a double-layer 3/0 polypropylene suture. LA appendage closure was assessed on follow-up transoesophageal echocardiography, with successful (complete) closure defined as no leak across the suture line and stump diameter below 10 mm. Since 2015, we have typically performed right atrial ablation after declamping the aorta on the beating heart during reperfusion time.

Ablation lines were created using either a dry bipolar radiofrequency clamp (Isolator Synergy; AtriCure Inc., Cincinnati, OH, USA) or an irrigated radiofrequency clamp (CardioBlate; Medtronic Inc., Minneapolis, MN, USA). The bipolar radiofrequency clamp was applied 6–10 times for each line, and/or cryolesions were created by applying cryoICE^® cryoablation probes (AtriCure), with an application time of 2 min. Between 2007 and 2012, the completeness of lesions across the left and right atrioventricular grooves was achieved using a bipolar clamp and a monopolar radiofrequency probe (Isolator Transpolar Pen; AtriCure) or using only the bipolar clamp and following the technique described by Benussi et al. [16, 17]. Since 2012, we have been using the nitrous oxide-based CryoIce probe (AtriCure), employing either solely cryothermal energy or a combination of bipolar radiofrequency and cryothermal energy for mitral and tricuspid annulus lesions. When cryothermia was used alone, the pulmonary veins and LA posterior wall were isolated en bloc. To ablate the LA isthmus, an epicardial lesion over the coronary sinus was added endocardially in the corresponding area.

Statistical analysis

Statistical analysis was performed using STATA version 13.0 (StataCorp LP, College Station, TX, USA). Continuous data are presented as mean ± standard deviation for variables with normal distribution and median (25th–75th percentile) for variables with non-normal distribution. Categorical data are described as absolute numbers and relative frequencies. Before propensity score matching, the 2 groups (LA vs BA) were compared using the independent samples t-test (normal distribution) or the Mann–Whitney U-test (non-normal distribution) for continuous variables and the Pearson’s χ² test with an (N−1)/N correction factor for categorical variables. After propensity score matching, between-group differences were compared using the paired t-test or the Wilcoxon signed rank test for continuous variables and using McNemars’s test for categorical variables. Using multiple logistic regression analysis, a propensity score was calculated for each patient, defined as the probability that the patient received the BA lesion set. The baseline variables included in the propensity score analysis are listed in Table 1. The nearest neighbour approach, without replacement and utilizing 5-to-1 digit matching, was used to identify 1:1 matched patients. The balance between variables before and after matching was assessed in terms of the percentage standardized bias (reported as a percentage). The Kaplan–Meier method was used to evaluate overall survival and survival free from AAR, and the results are presented as 95% confidence intervals (CIs). Survival curves were compared using a stratified log-rank test for matched pairs. After matching, uni- and multivariable logistic regression analyses were performed to determine independent predictors of permanent pacemaker implantation. Cox proportional hazards regression models were used to identify predictors of late AAR. The constant relative hazard assumption was confirmed as valid based on the correlation of sets of scaled Schoenfeld residuals for each covariate, with a suitable transformation of time, along with a global test for the model as a whole. The inclusion criterion for the full regression model was P ≤ 0.200 or clinical significance, and the limit for stepwise backward elimination was P < 0.100. The results of logistic regression analyses are expressed as odds ratios (ORs) and those of the Cox regression analysis as hazard ratios, both with their respective 95% CIs. A P-value <0.05 was considered statistically significant.

RESULTS

Before propensity score matching

Baseline demographic and echocardiographic data and perioperative data are presented in the Supplementary Material, Table S1. Before propensity score matching, patients in the LA group were more likely to be female and to undergo cryoablation (35.3% vs 13.6%), whereas those in the BA group were more likely to undergo mitral valve repair (29.6% vs 21.2%) and radiofrequency ablation (86.4% vs 64.7%). Although the non-matched groups did not differ regarding the length of hospital stay (Supplementary Material, Table S2), 30-day mortality rate (2.4% vs 0.6%, P = 0.146) or main indication for pacemaker implantation (sinus node dysfunction), the incidence of permanent pacemaker implantation in the early postoperative period was higher in the BA group (16.6% vs 6.4%, P < 0.001).

Additionally, the non-matched groups did not differ regarding 5-year all-cause mortality (the log-rank test P = 0.629; Supplementary Material, Fig. S1), survival free from cardiac-related death (the LA group 96.5 ± 1.0% with 95% CI 93.8–98.0%; BA group 98.5 ± 1.0% with 95% CI 94.3–99.6%; log-rank P = 0.690) or survival free from thromboembolic events (the LA group 96.9 ± 1.2% with 95% CI 93.1–98.6%; the BA group 95.2 ± 2.8% with 95% CI 85.6–98.5%; log-rank P = 0.764). Nevertheless, at 5-year follow-up, survival free from AAR off antiarrhythmic drugs was significantly better for BA (90.7 ± 3.1% with 95% CI 82.5–95.2%) than for LA lesions (81.6 ± 3.1% with 95% CI 74.5–86.8%) (Supplementary Material, Fig. S2; hazard ratio 2.2 with 95% CI 1.1–4.3 and P = 0.017).

After propensity score matching

To reduce between-group differences and account for the contribution of confounders, a propensity score analysis was performed, and 2 groups (LA vs BA) were obtained, each with 156 patients.

Early outcomes

Even after propensity score matching, the LA and BA groups differed significantly regarding cardiopulmonary bypass time (102.4 ± 68.8 vs 126.6 ± 83.7 min, mean difference 24.2 min; P = 0.006) and aortic cross-clamp time (77.2 ± 51.6 vs 92.3 ± 56.4 min, mean difference 15.1 min; P = 0.014). On the other hand, the matched groups did not differ regarding surgeon identity, ablation device, concomitant procedures (coronary surgery 12.2% vs 12.8%, P = 0.865; tricuspid valve repair 64.1% vs 62.2%, P = 0.726; Table 1), energy sources (cryothermal energy alone 15.4% vs 19.2%, P = 0.371; in all other cases, radiofrequency ablation with or without ‘cryolines’), length of hospital stay, mean clinical follow-up or 30-day mortality (1.9% vs 0.6%, P = 0.617) (Table 2). However, a significantly higher rate of permanent pacemaker implantation was observed in the BA group (17.3% vs 3.8%; P < 0.001).

The indications for pacemaker implantation included symptomatic sick sinus node syndrome (junctional rhythm below 60 bpm or rigid sinus rhythm with inappropriate response to physical activity) and severe atrioventricular conduction disturbance (see related section in Supplementary Material). In this population, the main indication was sinus node dysfunction (0.6% vs 10.3% for the LA and BA groups, respectively; P < 0.001), although there was no significant between-group difference regarding the incidence of atrioventricular conduction disturbances (3.2% vs 7.0%, respectively; P = 0.211). Average timing for pacemaker implantation was 10–14 days following the procedure. Uni- and multivariable logistic regression analyses indicated that BA lesion set (OR 5.75, 95% CI 2.16–15.29; P < 0.001) and AF duration (OR 1.01, 95% CI 1.00–1.02; P = 0.039) were associated with permanent pacemaker implantation (Supplementary Material, Table S3).

Follow-up data

The mean follow-up duration was 37.3 ± 34.1 and 37.1 ± 34.2 months in the LA and BA groups, respectively. Anticoagulation therapy was continued due to mechanical valve replacement, AAR or other indications in 107 and 101 patients in the LA and BA groups, respectively (P = 0.361). Among these, 1.5% patients (3 and 2 patients, respectively) had to continue anticoagulation therapy due to incomplete closure of the LA appendage. Patients who were discontinued from anticoagulation therapy underwent transoesophageal echocardiography annually thereafter. There was no new-onset leakage through the LA appendage suture line. In all cases, the LA appendage stump diameter was smaller than 5 mm. During follow-up, bleeding was noted in 8 (5.1%) and 11 (7.1%) patients from the LA and BA groups, respectively (P = 0.646).

The propensity score-matched groups did not differ regarding 5-year overall survival (Fig. 2; 93.5 ± 2.2% with 95% CI 87.3–96.7% vs 92.8 ± 2.9% with 95% CI 84.5–96.8%; the log-rank test, P = 0.998), 5-year survival free from cardiac-related death (95.1 ± 2.0% with 95% CI 89.3–97.8% vs 98.4 ± 1.1% with 95% CI 93.9–99.6%; the log-rank test, P = 0.101) or 5-year survival free from thromboembolic events (97.1 ± 2.8% with 95% CI 81.4–99.6% vs 96.2 ± 2.7% with 95% CI 85.2–99.1%; log-rank test, P = 0.309). In the LA group, survival free from AAR off antiarrhythmic drugs (Fig. 3) was 97.5 ± 1.4% (95% CI 92.4–99.2%), 91.8 ± 2.9% (95% CI 84.0–95.9%) and 85.3 ± 4.5% (95% CI 73.7–92.0%) at the 1-, 3- and 5-year follow-up, respectively; in the BA group, these values were 98.2 ± 1.2% (95% CI 93.2–99.6%), 93.6 ± 2.6% (95% CI 86.2–97.1%) and 91.9 ± 3.1% (95% CI 83.2–96.1%), respectively. Data regarding freedom from AAR at various time points, stratified according to the choice of lesion set, are shown in Supplementary Material, Fig. S3. The between-group differences in freedom from AAR were statistically significant (the log-rank test, P = 0.049), with the BA lesion set providing lower recurrence rates. Multivariable Cox proportional hazards regression analysis (Supplementary Material, Table S4) confirmed that not performing the additional right atrium lesion set was associated with the primary end point (hazard ratio 3.26; 95% CI 1.33–7.99).

At the last follow-up, 18 patients from the LA group and 8 patients from the BA group had late AAR. Among them, 3 patients from the LA group had typical atrial flutter and were successfully restored to sinus rhythm using catheter ablation. Typical atrial flutter was not noted in the BA group. Atypical LA flutters were observed in 3 and 4 patients from the LA and BA groups, respectively. Among these, 2 patients from each group had incompleteness of the roof line (cryothermia was used as the sole energy source) and were successfully treated with catheter ablation. Three patients (1 from the LA group and 2 from the BA group) had incomplete ablation of the mitral line; although 1 was successfully treated, sinus rhythm could not be restored in the other 2. The remaining 12 patients from the LA group and 4 patients from the BA group had long-standing persistent AF. Pacemaker interrogation reports reviewed that all 16 patients with atrioventricular conduction disturbances were pacemaker dependent during follow up; however, 4 (from the BA group) of 17 patients who received pacemaker due to sinus node dysfunction at discharge had returned to normal sinus rhythm.

DISCUSSION

The key findings of our study are that the BA maze procedure provides better freedom from AAR; however, the additional right atrial ablation was associated with a higher incidence of permanent sinus node dysfunction in the early postoperative period, which was reflected in a significantly higher rate of pacemaker implantation.

Atrial arrhythmia recurrence

Current European guidelines [14] indicate that the BA lesion set may be more effective than the LA lesion set alone in patients with persistent or long-standing persistent AF. Several investigations including propensity-matched studies, randomized trials and meta-analyses have focused on obtaining evidence in support of such recommendations.

In a recent study involving propensity score analysis of LA and BA procedures (n = 93 per group), Ad et al. [8] reported comparable results for 2-year freedom from AF recurrence (75% vs 86%, P = 0.13), as well as for freedom from thromboembolic events (stroke and transient ischaemic attack), which may be related to the limited small sample size and event rate in their cohort. In contrast, our study revealed significant between-group differences regarding 5-year freedom from AAR (85% vs 92%, P = 0.049). Although it is difficult to compare our present findings with the observations of Ad et al. [8], the discrepancy is likely related to differences in study population and clinical presentation. Specifically, we included relatively younger patients and overall fewer men; moreover, there were differences in AF duration and prevalence of various types of AF, as well as in the type of concomitant procedures. Ad et al. [8] concluded that LA ablation represents a reasonable option only in patients with shorter AF duration and smaller LA size (i.e. patients without risk factors for late AF recurrence), which is consistent with our observations. However, our present findings also indicate that BA procedures are preferable in patients with traditional predictors of failure.

In a subgroup analysis of patients with LA and BA lesion sets (86 per group), Soni et al. [7] did not find any significant difference in 1-year freedom from AF recurrence (76.1% vs 80.0%). Although Soni et al. included various lesions in each subgroup (e.g. the pulmonary vein, mitral valve and LA appendage), we used a standard institutional protocol (described above) for each lesion set in all enrolled patients. Moreover, Soni et al. did not perform rigorous rhythm monitoring using the Holter monitoring or loop recording. We believe that adequate follow-up and appropriate monitoring are key to revealing potential differences and reducing the overestimation of success rates.

Churyla et al. [18] reported the outcomes of 147 propensity score-matched pairs who received LA or BA ablation for various types of AF during mitral valve surgery. Unlike in our study, they did not find significant differences between the LA and BA groups regarding freedom from AF (69% vs 79%, P = 0.09) at the last follow-up. Nevertheless, Churyla et al. included patients with paroxysmal AF (>40% in each group), which likely contributed to their observation that both LA and BA procedures provided excellent and comparable 3-year results. The authors hypothesized that BA ablation might be beneficial in high-risk patients but found no between-group differences in a subgroup analysis involving only patients with long-standing persistent AF (66% vs 71%, P = 0.51), which could be explained by insufficient sample size and small event rate (58 and 53 patients, respectively, were analysed in each group).

A prospective randomized study by Wang et al. [6] included patients with permanent AF (LA, n = 149; BA, n = 150) and also reported no benefit of 1 technique over the other; specifically, at final follow-up, sinus rhythm was maintained in 85.2% and 84.1% of cases, respectively (P = 0.87).

A meta-analysis by Phan et al. [9] reported better sinus rhythm maintenance for BA ablation at 6 months (64% vs 74%, P = 0.03) and 12 months (70% vs 77%, P = 0.005). However, there were no significant differences at 1-year follow-up (59% vs 64%, P = 0.87). In contrast, a meta-analysis authored by Barnett and Ad in 2006 had shown that BA ablation provides superior freedom from AF at 1, 2 and 3 years after the procedure: 88.9 ± 8.2% vs 75.9 ± 8.4% (P = 0.001), 85.8 ± 5.0% vs 74.5 ± 1.9% (P = 0.001) and 87.1 ± 4.7 vs 73.4 ± 0.0% (P = 0.001), respectively [5].

This study included patients with uniform risk for late failure (those with paroxysmal AF were excluded from the analysis) and moreover treated following a single institutional protocol for lesion sets. Importantly, we showed that both LA and BA lesions have low 30-day mortality, high freedom from thromboembolic events and acceptable mid-term overall survival and freedom from cardiac-related death. A key result of our study is the finding that BA surgical ablation provides better 5-year freedom from AAR. The 2017 guidelines [19, 20] stated that patients with advanced disease (long AF duration, enlarged LA size) should have BA ablation. Another interesting finding is that most late AAR recurrences occurred later than 5 years after surgery, suggesting that the longevity of LA ablation alone is very limited in this complex group of patients. Since we found that both lesion sets are effective in terms of early and mid-term outcomes, we believe that right atrial ablation may play a key role in decreasing long-term AAR, especially in patients with long-standing persistent AF.

Pacemaker implantation requirements

A systematic review of the Cochrane database [21] and analysis of data from the Society of Thoracic Surgeons database [22] showed that concomitant AF surgery resulted in increased pacemaker implantation rate (relative risk 1.69, 95% CI 1.12–2.54; OR 1.26, 95% CI 1.07–1.49). However, according to the experience of the Inova Fairfax Hospital [23], pacemaker implantation rates following the Cox maze procedure have decreased recently and are currently comparable with those noted in patients who do not undergo surgical ablation.

Churyla et al. [18] reported no difference in pacemaker implantation rates (10% vs 12%, P = 0.57), with atrioventricular node dysfunction as the prevalent indication. Gillinov et al. [2] also reported that the rates of permanent pacemaker implantation did not differ between the LA and BA subgroups (10 patients, 14.9%; 16 patients, 24.2%; P = 0.22). Among these patients, the indications for pacemaker implantation were heart block in 53.9% of cases and sinus node dysfunction in 34.6% of cases (11.5% unknown).

Wang et al. [6] observed that the BA lesion set was associated with higher incidence of pacemaker implantation in the early postoperative period (6.0% vs 9.3%, P = 0.03). Unfortunately, this previous study includes several limitations such as underestimation of AF prevalence, continuing antiarrhythmic drug use and adding a right isthmus line in all patients with LA ablation pattern.

Similar to our findings, the observations of Soni et al. [7] also indicated that the BA lesion set was associated with higher incidence of pacemaker implantation (7.5% vs 16.5%, P = 0.02) due to sinus node dysfunction (3.7% vs 12.1%, P < 0.01), which was independent of other operative factors. In a recent meta-analysis [9], the permanent pacemaker implantation rate in the LA and BA subsets was 5.4% and 7.0%, respectively (P = 0.008).

Although the BA lesion set can help improve rhythm outcomes, the expected benefit in freedom from AAR might be diluted by an increased rate of pacemaker implantation for sinus node dysfunction. In our study, we did not investigate the consequences of pacemaker implantation. Ad et al. [23] reported that pacemaker implantation after the Cox maze procedure was not associated with increased short- or long-term morbidity or AAR. Nevertheless, we recommend avoiding the BA lesion set in patients without conventional predictors for late failure, provided that the risk of pacemaker implantation is greater than the benefit provided by rhythm-control strategies. Currently, we do perform the BA lesion set in patients with long-standing persistent AF, atrial flutter or indication for concomitant tricuspid valve surgery.

Limitations

This study has several limitations that should be addressed. First, our findings are limited by the confounding and bias associated with the retrospective study design. Furthermore, a substantial number of patients were lost to follow-up (almost 20%). Moreover, we used 24-h Holter monitoring instead of continuous loop recording or 72-h Holter monitoring. The results may be affected by the fact that 4 surgeons participated and 2 types of energy sources were used. Using propensity score analysis, we sought to minimize the effect of bias, confounding factors and inter-surgeon variability. Finally, it was not possible to collect rhythm information at specific time points so as to provide appropriate statistics for time-based analyses. The findings of our analysis suggest that BA ablation should be used in selected patients with high risk of AF recurrence; however, this conclusion remains to be confirmed in future prospective randomized studies with an adequately powered sample size.

Conflict of interest: none declared.

REFERENCES

Author notes