Biatrial versus Isolated Left Atrial Ablation in Atrial Fibrillation: A Systematic Review and Meta-Analysis

Objective The outcomes of biatrial ablation (BA) and isolated left atrial ablation (LA) in atrial fibrillation remain inconclusive. In this meta-analysis, we assess the currently available evidence to compare outcomes between BA and LA. Methods Electronic searches were performed from database inception to December 2016, and relevant studies were accessed. Odds ratios and weight mean differences with 95% confidence intervals are reported. Twenty-one studies comprising 3609 patients were included in the present meta-analysis. Results The prevalence of sinus rhythm in the BA cohort was similar to that in the LA cohort at discharge, at 12 months, and after more than 1 year of follow-up. However, at 6 months, the prevalence of sinus rhythm was higher in the BA cohort than in the LA cohort. The rate of permanent pacemaker implantation was higher in the BA cohort than in the LA cohort. However, 30-day and late mortality and neurological events were similar between the BA and LA groups. Conclusion There was no significant difference in the rate of restored sinus rhythm, the risk of death, and cerebrovascular events between BA and LA, but BA had a higher rate of permanent pacemaker implantation.


Introduction
Atrial fibrillation (AF) is a major healthcare problem worldwide which has enormous economic and public health implications. AF is associated with an increased risk of stroke, heart failure, and all-cause mortality [1][2][3].
Surgical ablation was introduced as a treatment option by Cox et al. [4] in 1991, and it is currently an effective curative strategy for AF. Haïssaguerre et al. [5] suggested that ectopic beats from pulmonary veins may cause AF, and the field of pulmonary vein isolation (PVI) was consequently established and is now performed via catheter or surgical ablation [6]. Left ablation (LA) has historically been the main method used to treat AF, and it has a fairly good clinical effect [7,8]. However, some studies have suggested that LA is less efficacious than biatrial ablation (BA), especially when a right-side AF trigger is present [9]. Hence, because outcomes have been inconclusive, Phan et al. [10] and Zheng et al. [11] reported relevant meta-analyses in 2014, but they arrived at a different conclusion. The authors showed that BA was more effective than LA and that the rate of permanent pacemaker implantation was also higher in the BA cohort than in the LA cohort. However, Zheng et al. [11] suggested that the effects of BA and LA are the same.
In the past two years, several other studies [8,[12][13][14][15] have compared BA and LA in AF, with controversial outcomes. Hence, in this meta-analysis, we sought to assess the current evidence available on this issue.

Literature Search Strategy.
Electronic searches were performed in August 2016 without search restrictions. The primary sources were the electronic Medline, PubMed, Cochrane Library, and EMBASE databases, which were searched from their date of inception to August 2016. The following search terms were used: "maze," "biatrial," "biatrial," "uniatrial," "left atrial," and "ablation." When duplicate published trials with accumulating numbers of patients or increased lengths of follow-up were encountered, the most recent or most complete report was considered. All titles and abstracts identified in the electronic search were uploaded into an EndNote (version X7; Thomson Corporation, Stanford, USA) database ( Figure 1).

Inclusion and Exclusion
Criteria. All available randomized, controlled trials (RCTs) and retrospective comparative studies that compared BA with LA in all age groups were included. Abstracts, case reports, conference presentations, editorials, reviews, and expert opinions were excluded. When institutions published duplicate studies with accumulating numbers of patients or increased lengths of follow-up, only the most complete report was included. Reference lists were also manually searched for further relevant studies.

Data Extraction and Critical
Appraisal. Two reviewers (Hongmu Li and Xifeng Lin) conducted data extraction independent of the included studies. Data on authorship, year of publication, study design, study population, baseline characteristics, characteristics related to outcomes, and duration of follow-up were extracted from each study. Reported percentages were approximated to numbers. The risk of bias was assessed using the Downs and Black checklist [16] for randomized and observational studies. Discrepancies between the reviewers were resolved by discussion until consensus was reached. The final results were reviewed by the senior investigator (Ping Hua).

Quality Assessment and Statistical
Analysis. The included studies were rated to determine the level of quality of the provided evidence according to the criteria of the Centre for Evidence-Based Medicine in Oxford, UK [17]. The methodological quality of the RCTs was assessed with the  Cochrane risk of bias tool [18]. The methodological quality of retrospective studies was assessed with the modified Newcastle-Ottawa scale [19,20]. This meta-analysis was performed using Review Manager Version 5.3 (Cochrane Collaboration, Oxford, UK). Dichotomous variables from individual studies were analyzed using odds ratios (ORs) with 95% confidence intervals (CIs).statistics ( < 0.10) or I 2 statistics were performed to test for heterogeneity between included studies, and values of 50% or higher were considered to be indicative of substantial heterogeneity. If there was substantial heterogeneity, the possible clinical and methodological reasons for this were explored qualitatively. Publication bias was examined through a visual inspection of funnel plots and assessed by applying Egger's weighted regression statistic and considering a value less than 0.05 as indicating significant publication bias. A value of less than 0.05 was considered statistically significant for all analyses.

Synthesis of Evidence.
Our electronic literature search resulted in the retrieval of 398 citations. Of these, 372 were excluded after duplicate and irrelevant references were excluded, and 49 potentially relevant articles were retrieved. Finally, following a manual search of reference lists and a critical appraisal, 21 studies comprising 3609 patients were included in this meta-analysis. Two articles had redundant publications but covered different characteristics [21,22].

Discussion
Two recent meta-analyses [10,11] that compared BA, LA, and surgical ablation in AF arrived at conflicting conclusions. However, these meta-analyses excluded several studies that compared BA with LA [8,[12][13][14][15]25]. Therefore, we performed a new meta-analysis to compare BA with LA. This metaanalysis included three RCTs and 18 retrospective studies that collectively contained 3609 patients and compared the efficacy and safety of BA and LA. There was no significant difference between BA and LA in the rate of restored SR, but BA groups had a higher probability of SR after 6 months of follow-up. We also found that while BA and LA had similar rates of death and cerebrovascular events, the BA groups had a higher rate of permanent pacemaker implantation.
A pooled analysis of restored postoperative SR showed that there was no difference between the BA and LA groups. However, several recent studies [21,23,36] have shown that BA is inferior to the more complete LA when used alone. Patients in the LA group had shorter aortic cross-clamping times and cardiopulmonary bypass times than were observed in the BA group. Furthermore, the techniques used in AF ablation vary widely, even within the same procedure group, and if the different lesion sets used for ablation were included, the results may have indicated that this procedure has greater efficacy.
In contrast, some studies have reported that BA is superior to LA for restoring SR [8,12,15,[25][26][27][28][30][31][32][33]. This finding rests mainly on the finding that BA groups have much more tissue damage and a higher rate of cardiac conduction system injury. However, all of these studies have common limitations. First, some patients took antiarrhythmic drugs (including amiodarone) perioperatively and continued the use of these drugs until the operation, and few researchers sufficiently addressed this variable. Second, the sample size in most of the articles was small (less than 150 individuals), weakening the power of the studies. Third, long rhythm registration during follow-up was not available in all of the patients. Furthermore, only a few of the studies were RCTs. Unlike previous reviews, we included the largest studies in our meta-analysis, and our inclusion criteria did not limit our search to articles published in English. We also conducted Breda       The findings of the present meta-analysis confirm that BA increases the risk of permanent pacemaker implantation. This finding may be attributed to the fact that LA has shorter aortic cross-clamping and cardiopulmonary bypass times and promotes more extensive lesions. There was no significant increase in the risk of cerebrovascular events or early and late mortality between the two groups. We hypothesize that report selection resulted in fewer such events, and these results remain to be discussed.
The most important findings of our meta-analysis include the following: (1) LA and BA were equally effective in restoring SR, (2) BA resulted in higher prevalence of SR at the 6-month follow-up, and (3)   meta-analysis included the largest studies, and its inclusion criteria did not limit the search to articles published in the English language.
The results of our study show that there was no difference in the rate of restored SR between LA and BA. While some previous studies have proposed that BA alone is inferior to a more complete LA, this significance disappeared in a multivariate analysis. The difference in these results may have been caused by differences in inclusion criteria between previous studies and our study. The other studies limited inclusion to articles reported in the English language. Additionally, the techniques used for AF ablation varied widely, including, for example, the use of different lesion sets, even within same procedure group.
One of the most important reasons that researchers have suggested for why BA is better than LA at restoring SR is that there is a significant difference in electrical activity between patients with chronic and paroxysmal AF [37,38]. Lazar et al. [37] demonstrated that a left-to-right atrial frequency gradient exists in paroxysmal but not persistent AF. This prompted them to propose that the maintenance of persistent or chronic AF may be less dependent on the posterior left atrium. Additionally, Sanders et al. [39] proposed that, in patients with paroxysmal AF, the dominant sources of activity are often localized in the pulmonary veins. In contrast, in patients with permanent AF, the dominant sites are more often localized in the atria, including right atrial sites. Unsurprisingly, patients with persistent or long-standing persistent AF are more likely to receive BA, and this may affect clinical outcomes [25].
The present meta-analysis has the following limitations. Its main limitation is that only three small-sample RCTs were included. Inadequate random sequence generation and blinding tend to increase the risk of bias. Hence, larger RCTs are needed to determine the best treatment. Another limitation is that the original meta-analysis was based on the assumption that the surgical subgroups (BA and LA) were sufficiently similar to be assessed together, but the operation methods and ablation technologies used in these procedures are continually developing. Additionally, there was extreme heterogeneity among the studies in study design, data, and energy source, and a subgroup analysis yielded results that differed from those obtained in the original analysis. Future systematic reviews should, when sufficient literature is available, evaluate different indications separately. Finally, follow-up periods were generally short. Therefore, the long-term outcomes of BA and LA remain to be explored.

Conclusion
In this comparative meta-analysis, we show that BA is not more efficacious than LA in restoring SR. Additionally, the risks of death and cerebrovascular events are significantly different between BA and LA, but BA results in a higher rate of permanent pacemaker implantation.