Catheter ablation for AF is associated with a risk for major bleeding due to multiple vascular accesses, transseptal puncture, and catheter manipulation inside left atrium.1,19,20 An international survey of AF ablation procedures found a 4.5% major complication rate.21 Therefore, the key pursuit is to find an optimal balance between thromboembolism and bleeding. To our knowledge, the current meta-analysis is the first to compare procedural characteristics and embolic and bleeding events between uninterrupted and interrupted DOAC regimens for catheter ablation of AF.
Review of Literature
The VENTURE-AF (Study Exploring Two Treatment Strategies in Patients With Atrial Fibrillation Who Undergo Catheter Ablation Therapy) study22 randomized 248 patients to either uninterrupted rivaroxaban or uninterrupted warfarin. In the AXAFA-AFNET 4 (Apixaban During Atrial Fibrillation Catheter Ablation: Comparison to Vitamin K Antagonist Therapy) study,23 633 patients were randomized to uninterrupted apixaban or uninterrupted vitamin K antagonists. Neither of these studies found between-group differences in bleeding or ischemic complication rates.22,23 The RE-CIRCUIT (Uninterrupted Dabigatran Etexilate in Comparison to Uninterrupted Warfarin in Pulmonary Vein Ablation) trial randomized 678 patients to either uninterrupted dabigatran or uninterrupted warfarin; those in the dabigatran arm showed a reduction in bleeding risk, with no symptomatic cerebral events.24 Most recently, the ELIMINATE-AF (Edoxaban Treatment Versus Vitamin K Antagonist in Patients With Atrial Fibrillation Undergoing Catheter Ablation) trial revealed similar bleeding and ischemic complication rates for both uninterrupted edoxaban and uninterrupted warfarin.25
Heterogeneity in Anticoagulation Protocols
The trials described above used direct anticoagulants that have important differences in pharmacodynamics and dosing, and they also used different protocols, resulting in heterogeneity. The 2 studies using a once-daily DOAC shifted the last anticoagulant dose to the night before the procedure. In VENTURE-AF, the last dose of rivaroxaban was administered predominantly on the evening before the procedure. Patients randomized to uninterrupted edoxaban in the ELIMINATE-AF trial also took their scheduled doses in the evening.25 In contrast, more than 80% of the patients treated with dabigatran in the RE-CIRCUIT trial received the last dose <8 hours before the ablation.22,24 In the AXAFA-AFNER study, apixaban treatment was continued without any dose being held back, including on the morning of the ablation.23
Guidelines
Multiple guidelines, international consensus statements, and, most recently, the European Heart Rhythm Association’s Practical Guide on the Use of Non-Vitamin K Antagonist Oral Anticoagulants in Patients with Atrial Fibrillation recommend continuation of oral anticoagulation with vitamin K antagonists or DOACs among patients undergoing AF ablation procedures.5,26,27 The 2017 international expert consensus statement on AF ablation supports the performing of AF ablation procedures without interruption of warfarin or DOACs (Class I), or the holding of 1 to 2 doses of the DOAC before the ablation (Class IIa).5 Furthermore, the European Heart Rhythm Association’s Practical Guide considers it reasonable to administer a last DOAC dose 12 hours before the start of the intervention, especially when transseptal puncture will be performed without periprocedural imaging.27 According to the First Snapshot European Survey, truly uninterrupted antithrombotic regimens (ie, last DOAC dose shortly before the procedure) were used for approximately 30% of DOAC-treated patients undergoing AF ablation.6
Findings from the Current Meta-Analysis
Baseline Characteristics
Reynolds et al12 studied only apixaban, in 2 different doses, Nagao et al13 included apixaban, rivaroxaban, and edoxaban, and the randomized Nakamura et al14 and observational Nakamura et al16 studies included all 4 DOACs; Müller et al15 did not indicate the regimen used. In all studies, the IA group received the last DOAC dose on the day before the ablation. Bridging was done in the IA group in the studies by Müller et al and Nakamura et al.14,15 The observational study by Nakamura et al16 included only patients with paroxysmal AF. The randomized study by Nagao et al13 had a high proportion of patients with chronic kidney disease. Structural heart disease was more prevalent in the randomized studies by Reynolds et al12 and Nakamura et al.14 Coronary artery disease was more prevalent in the study by Reynolds et al.12 LVEF was relatively lower in the studies by Reynolds et al12 and Müller et al.15 Protamine was used to reduce the risk for periprocedural bleeding in the Reynolds et al12 and Nakamura et al14 randomized studies, at the operator’s discretion.
Thrombosis Risk
The incidence of periprocedural thromboembolism in patients with AF undergoing ablation ranges from 0.9% to 5% and depends on the diagnostic modality.13 Possible mechanisms include blood coming in contact with foreign surfaces, endothelial injury and inflammation in the left atrium, cellular damage and release of components, and blood flow alteration after sinus rhythm is established.28 Unfractionated heparin prevents common extrinsic and intrinsic coagulation pathway activation when administered before septal puncture.14 Artificial surface–induced thrombosis is not prevented effectively by DOACs.14,16,17 Thus, even with UA, intraprocedural unfractionated heparin is required to prevent thromboembolic events. Moreover, there is a hypothesis that dabigatran downregulates expression of antithrombin, with a compensatory prothrombin upregulation leading to diminution of unfractionated heparin effect.18
Müller et al15 reported greater incidence of asymptomatic, magnetic resonance imaging (MRI)-detected, so-called SCE in the IA group. At 1 to 2 days after radiofrequency catheter ablation, MRI was done by using a 1.5 Tesla MRI scanner. Acute lesions showed focal hyperintensities in diffusion-weighted imaging. Apparent diffusion coefficient mapping was used to differentiate true lesions from a shine-through artifact. In the study by Nagao et al,13 SCE was independently predicted by CHA2DS2-VASc score in the UA group and by intraprocedural cardioversion and procedure time in the IA group. Overall, SCE was significantly more frequent in the IA group (P<.005).13 The Nakamura et al16 observational study found that uninterrupted dabigatran was an independent predictor of SCE. The SCE rate did not differ significantly between the UA and IA groups in the randomized study by Nakamura et al.14 In our meta-analysis, the incidence of stroke or TIA did not differ significantly between the 2 groups, but SCE was significantly more frequent with IA, further emphasizing the prothrombotic milieu during AF ablation and need for bridging with unfractionated heparin. This is supported by the need for a higher total heparin dose in the IA group. Moreover, ACT before first heparin bolus was significantly longer in the UA cohort, supporting lesser thrombotic risk in this group.
Bleeding Risk
Reynolds et al12 stated that patients taking DOACs may have lower risk for periprocedural bleeding than patients taking warfarin. The Nakamura et al14 randomized trial found similar rebleeding rates at venous puncture sites in both the UA and IA groups. Although the presence of chronic kidney disease increased periprocedural bleeding risk in a study by Yanagisawa et al,29 similar findings were not reported in the studies incorporated in this meta-analysis. The same study found antiplatelet use to be an independent predictor of adverse events in AF ablation; conversely, Reynolds et al12 reported that aspirin was not significantly associated with bleeding in multivariate model results.29 Several studies found low rates of major bleeding in both UA and IA groups and similar incidences of minor bleeding, which was attributed to postprocedural protamine use and postprocedural unfractionated heparin use.12-14 In keeping with the above findings, total bleeding, major bleeding, and minor bleeding were similar in the 2 groups in our meta-analysis. Similarly, total pericardial effusion, cardiac tamponade, and total puncture complications did not differ significantly between the IA and UA groups, nor did protamine use. A recently published meta-analysis found that the rate of vascular complications in electrophysiology procedures—and thus, major and minor bleeding—can be reduced by using ultrasound-guided femoral access.30
MACCVE
MACCVE is a novel composite endpoint, we looked into, which comprised of major bleeding events as well as thrombotic events. In our meta-analysis, MACCVE did not differ significantly between the UA and IA groups. Although SCE were noted more in relation to interrupted DOACs, the overall outcomes were comparable between the two groups which suggests that even with uniterrupted periprocedural anticoagulation, patients can be discharged safely from hospital following AF ablation on the same day.29
Predictors of Silent Cerebral Events
To date, the clinical relevance of SCE remains unclear. Some data suggest that SCE is associated with cognitive impairment occurring after an AF ablation procedure.31 This represents a real cause for concern for some authors,23,32 whereas the relationship between SCE and cognitive impairment is disputed by others.2,7,33 Increased incidence of SCE has been reported with reinsertion and application of a previously withdrawn cryoballoon, multielectrode catheter use for additional left atrial mapping, and transient coronary air embolism.34 Additional radiofrequency ablation within the left atrium in patients undergoing nonpulmonary vein isolation ablation was an independent risk factor for cerebral ischemic events in a study by Nakamura et al.35 In a very recent meta-analysis published, uninterrupted DOAC was found to of similar bleeding events with comparison to minimally interrupted DOAC and also mirrored our findings of lesser SCE.36 However, this study did not explore the procedural aspects, specially relation to use of Heparin and ACT. Also our results are statistically more relevant as we accounted the necessary modifications to address sparse binary events.
Limitations
First, we were able to include only 5 studies, 2 of which were observational trials. Second, the overall follow-up duration was less. Third, there was considerable difference in the periprocedural anticoagulation regime across the studies. Fourth, subgroup analyses (e.g. paroxysmal vs persistent AF, mapping vs balloon strategy) could not be done due to lack of data. Finally, given the infrequent outcomes, the overall sample size (despite pooling the number of patients) across the studies maybe inadequate.