Atrial fibrillation (AF) is the most common sustained cardiac arrhythmia and is associated with significantly increased morbidity and mortality (1- 2). Radiofrequency catheter ablation has brought a paradigm shift to the management of AF and has currently evolved to become the standard of care for symptomatic patients in whom antiarrhythmic drugs have failed (1,3). AF ablation is technically challenging and is associated with a small but definite risk of periprocedural thromboembolic and bleeding complications (4- 6). The increased risk of thromboembolic complications is likely related to the exacerbation of the baseline pro-thrombotic state by catheters in the left atrium (LA), endothelial denudation, char formation, and tissue inflammation from ablation in the LA (4- 5,7). The thromboembolic risk can be minimized by adequate periprocedural anticoagulation, which could potentially increase the risk of bleeding complications (8).

Optimal periprocedural anticoagulation protocols to minimize these complications are still largely debated and are nonuniform (4- 5). AF ablation on therapeutic anticoagulation with warfarin, with an international normalized ratio between 2 and 3, has been shown to be safe and is becoming increasingly adopted by several institutions (9- 13). Recently, dabigatran, an oral direct thrombin inhibitor, was approved for the prevention of stroke in patients with nonvalvular AF (3,14). Management of periprocedural anticoagulation in patients on dabigatran undergoing AF ablation is not known. We aimed to evaluate the feasibility and safety of continuing dabigatran during the periprocedural period of AF ablation in a multicenter, observational study.

We performed a multicenter, observational study from a prospectively collected registry of patients undergoing AF ablation for drug-refractory, symptomatic AF at 8 high-volume electrophysiology laboratories between January 2010 and July 2011. The study protocol was approved by local institutional review board. The dabigatran group consisted of all consecutive patients receiving anticoagulation therapy with 150 mg dabigatran etexilate (Pradaxa) twice daily for at least 30 days before the AF ablation procedure. An equal number of patients who were matched for age, sex, and type of AF undergoing AF ablation at respective institutions during the same period receiving therapeutic anticoagulation with warfarin for at least 30 days before the procedure comprised the warfarin group. Patients were excluded if the international normalized ratio was not between 2.0 and 3.5 at the time of the procedure and if the ablation was performed with the help of a remote navigation system.

Patients taking dabigatran were instructed to hold the dose on the morning of the procedure. Dabigatran was resumed within 3 h after hemostasis and when the patient was ready to have oral intake after the ablation procedure. Patients in the warfarin group underwent catheter ablation on uninterrupted warfarin therapy throughout the periprocedural period, including the evening of the procedure.

Transesophageal echocardiography was performed on all dabigatran patients to rule out a left atrial appendage thrombus, whereas it was not performed on patients in the warfarin group. During the ablation procedure, a bolus of 10,000 U of unfractionated heparin (UH) was given before the transseptal puncture. The activated clotting time (ACT) was checked 15 min after the bolus and every 20 min thereafter. Further weight-adjusted UH boluses were given to maintain an ACT of 300 to 400 s while catheters remained in the LA. Pulmonary vein antral isolation (PVAI) using a double transseptal approach was described in detail elsewhere (15). Briefly, under intracardiac echocardiography guidance, 2 transseptal accesses were obtained using standard needles and sheaths. The location, number, and size of venous sheaths and recording catheters were at the operator's discretion. A circular mapping catheter (Lasso, Biosense Webster Inc., Diamond Bar, California or Spiral, St. Jude Medical, Minneapolis, Minnesota) was used to map the LA. A 3-dimensional geometry of the LA was reconstructed with the CARTO (Biosense Webster Inc.) or the NavX mapping system (St. Jude Medical). A 3.5-mm open irrigated tip catheter (ThermoCool, Biosense Webster Inc.) was used to ablate the antrum of the pulmonary veins to achieve electrical isolation. Radiofrequency energy output was titrated to a maximum of 40 to 45 W along the anterior segments and 30 to 35 W while ablating the posterior segments. The exact number and extent of lesions were according to the operator's preference and were tailored to the individual patient.

In cases of paroxysmal AF, only PVAI was performed. The procedural endpoint of this ablation strategy was achieving entry and exit blocks. In patients with persistent AF, additional substrate modification involving complex fractionated atrial electrograms, identified by either the mapping catheter or the 3-dimensional map was also performed. Complex fractionated atrial electrograms along the posterior wall, left atrial septum, left atrial roof, coronary sinus, left atrial appendage base, and crista terminalis were mapped and ablated. If spontaneous intra-atrial tachycardia occurred during ablation, it was mapped and ablated. Roof lines and mitral isthmus lines to ablate residual left atrial tachycardia after PVAI and complex fractionated atrial electrogram ablation together with coronary sinus and left atrial appendage isolation was also performed in some patients according to the operator's preference. Further ablation at the superior vena cava and right atrial junction was also performed at operator's discretion if mapping revealed double potentials around this region and when high-output (30 mA) pacing did not capture the phrenic nerve. In addition, 20 μg/min isoproterenol was administered for 15 min to disclose non-pulmonary vein triggers that were ablated. Finally, if AF did not terminate after PVAI and substrate modification, direct current cardioversion was performed to achieve normal sinus rhythm.

Demographic, procedural, and complication data were obtained from prospectively collected registries in each of the participating centers. Events occurring within the first 30 days after the ablation procedure were included in this current analysis.

Hematomas and pericardial effusions were considered as bleeding complications. Cerebrovascular accidents and transient ischemic attacks were considered thromboembolic complications after ruling out intracranial hemorrhage. Any bleeding requiring blood transfusion, hematomas requiring surgical intervention, and pericardial effusions requiring drainage (tamponade) were considered as major bleeding complications. Minor bleeding complications included small hematomas and pericardial effusions not requiring an intervention (nontamponade). Late pericardial tamponades were those occurring >48 h after the procedure. The primary safety outcome measured was a composite of bleeding and thromboembolic complications. Miscellaneous nonanticoagulation-related events were also recorded.

In the event of a bleeding complication that required reversal of anticoagulation, fresh frozen plasma with possible hemodialysis was considered in the dabigatran group. In the warfarin group, vitamin K was considered in addition to fresh frozen plasma at operator's discretion.

Patients in both groups were matched by age (±2 years), sex, type of AF, and the institution where the procedure was performed. Tests for independent samples were selected throughout because other unmatched variables could remain that influence treatment efficacy. Both groups were compared using the chi-square test or Fisher exact test where appropriate for categorical variables and the independent Student t test for continuous variables. Bleeding complications (major, minor, and total), thromboembolic complications, and composite of bleeding and thromboembolic complications were compared between both groups.

A multivariable logistic model was used for identifying significant predictors of complications. All potential confounders were entered into the model based on clinical significance or observed univariable association. The controlling variables forced in the model were age (dichotomized at 75 years), sex, and AF type. The odds ratio (OR) and 95% confidence interval (CI) of composite bleeding and thromboembolic complications were computed. A p value <0.05 (2-sided) was considered statistically significant. All analyses were performed with SPSS version 19.0 for Windows (SPSS, Inc., Chicago, Illinois).

The patient population comprised 290 patients with 145 patients in each group. The baseline characteristics of dabigatran and warfarin groups are shown in (Table 1). The mean age of the study population was 60 years with 79% being male and 57% having paroxysmal AF. There were no differences in the individual components of the CHADS₂ score, mean CHA₂DS₂-VASc score, HAS-BLED score, left atrial size, left ventricular ejection fraction, and the presenting rhythm on arrival at the electrophysiology laboratory between both groups. In addition, there were no differences in the procedure time (208 min vs. 203 min; p = 0.78), radiofrequency ablation time (52 min vs. 55 min; p = 0.49), fluoroscopy time (58 min vs. 53 min; p = 0.19) between both groups (Table 2). Acute procedural success and successful PVAI did not differ between both groups.

The comparison of complications between both groups is shown in (Table 3). A total of 32 patients (11%) had either bleeding and/or embolic complications, of whom 29 (10%) had bleeding complications and 3 (1%) had thromboembolic complications. No intracranial hemorrhage or deaths occurred in the study population Of note, all 3 thromboembolic complications occurred in nonparoxysmal patients in the dabigatran group, whereas no thromboembolic complications occurred in the warfarin group (p = 0.25 for comparison). Neurologic symptoms improved in all 3 patients with no residual deficits noted at 30-day follow-up. Compared with patients taking warfarin, patients taking dabigatran had a significantly higher major bleeding rate (6% vs. 1%; p = 0.019), total bleeding rate (14% vs. 6%; p = 0.031), and composite of bleeding and thromboembolic complications (16% vs. 6%; p = 0.009). All major bleedings were pericardial effusions with tamponade requiring drainage (9 patients in the dabigatran group and 1 patient in the warfarin group), and none of the patients with a groin hematoma needed an intervention. One patient in the dabigatran group had both a pericardial effusion with tamponade and a groin hematoma. One patient in the warfarin group had both a pericardial effusion without tamponade and a groin hematoma. None of the patients needed hemodialysis for elimination of dabigatran from the systemic circulation.

We also analyzed the predictors of complications. The use of dabigatran in patients with bleeding and/or thromboembolic complications was much higher than in patients without these complications (Table 4). In univariable analysis, the only predictors of bleeding were the use of dabigatran (69% vs. 48%; p = 0.031) and age older than 75 years (17% vs. 4%; p = 0.004). Dabigatran use (72% vs. 47%; p = 0.009) and age older than 75 years (16% vs. 4%; p = 0.008) were also the only univariable predictors of the composite of bleeding and thromboembolic complications. In multivariable logistic regression analysis including age older than 75, sex, AF type, and the use of dabigatran, the only independent predictors of bleeding or thromboembolic complications were dabigatran use (OR: 2.76, 95% CI: 1.22 to 6.25; p = 0.01) and age older than 75 (OR: 3.82, 95% CI: 1.09 to 13.35; p = 0.04). Dabigatran use was found to be an independent predictor of both bleeding complications (OR: 2.34, 95% CI: 1.02 to 5.39; p = 0.046) and composite of bleeding and thromboembolic complications.

In our multicenter, observational study, we found that the use of dabigatran periprocedurally for AF ablation was associated with an increased risk of bleeding and composite of bleeding and thromboembolic complications compared with uninterrupted warfarin anticoagulation.

Dabigatran at a dose of 150 mg twice daily has been shown to be better than warfarin in preventing stroke in AF patients with an equivalent bleeding risk and has recently been approved for use in the United States for nonvalvular AF (3,14). Although a large number of patients are either switched from warfarin or started de novo on dabigatran for systemic anticoagulation, managing its use periprocedurally for catheter ablation is currently unclear. Our study is the first such study to systematically evaluate the feasibility and safety of continuing dabigatran periprocedurally for AF ablation compared with the more accepted warfarin anticoagulation.

The risk of thromboembolism during and after an AF ablation procedure remains a significant concern due to the inherent nature of the procedural components (16- 19). Some of the common underlying causes of periprocedural thromboembolism during AF ablation are thrombus formation on the catheters and guide sheaths, endothelial denudation, local tissue inflammation, dislodgment of an unrecognized left atrial thrombus, char formation on the catheter tip, and de novo clot formation due to atrial stunning after restoration of sinus rhythm with inadequate anticoagulation (5,7,17). These findings have led to the adoption of several safety practices such as using an open irrigated ablation catheter, transesophageal echocardiography to screen for left atrial thrombus before ablation, maintaining adequate periprocedural anticoagulation to prevent thrombus formation on the atrial surface, and pulling the sheaths back into the right atrium during ablation. As a result of these aggressive strategies, the risk of systemic thromboembolism decreased significantly from 5% to 6% to <1% in the more recent literature (9,20- 21). It is a routine practice to anticoagulate patients both with UH during the procedure and alternate anticoagulants for at least 4 weeks after the procedure (21- 23). Aggressive anticoagulation in the background of multiple venous accesses with moderate to large sheaths, transseptal accesses, and extensive ablation increases the risk of bleeding complications (8,16).

Currently, the periprocedural anticoagulation protocols for AF ablation are primarily driven by operator comfort and experience. Several different anticoagulation protocols consisting of varied combinations of antiplatelet agents, warfarin, UH, and low molecular weight heparin combined with transesophageal echocardiography screening stratified by baseline risk of systemic thromboembolism and type of AF have been put to work (1,4- 5). Although there are no established guidelines supporting the superiority of one periprocedural anticoagulation protocol over the other, it has become increasingly evident that performing AF ablation with uninterrupted anticoagulation with warfarin is safe and not associated with increased bleeding complications compared with other anticoagulants (9- 13,24). The recent consensus document from the European Heart Rhythm Association does mention that uninterrupted warfarin anticoagulation is a potential alternative to bridging with UH or low molecular weight heparin (25). Moreover, using an open irrigated ablation catheter on a background of therapeutic anticoagulation with warfarin has been shown to be associated with a dramatic reduction in the risk of thromboembolic events (9).

Dabigatran has a rapid onset of action (0.5 to 2 h), but the elimination half-life ranges from 12 to 14 h when used for long-term therapy with a normal creatinine clearance (26). However, in our study, dabigatran was discontinued on the morning of the procedure, and almost all procedures were performed within 1.5 times of the elimination half-life, when there was a significant residual pharmacodynamic effect. The safety of overlapping UH and dabigatran has not been established, in contrast to that of overlapping UH and warfarin. Dabigatran is generally recommended to be discontinued at least 24 to 48 h before an invasive procedure with a standard risk of bleeding and normal renal function (27). The overlapping pharmacodynamic effects of UH and dabigatran during the procedure probably explain the increased risk of bleeding in the dabigatran group of our study population. Our study finding is consistent with that of previous studies that elderly patients have a higher risk of complications including bleeding complications during AF ablation (6,28).

When sorted by AF type, all the thromboembolic complications were seen in nonparoxysmal AF patients. The incidence of thromboembolic complications in nonparoxysmal AF patients in the dabigatran group, although not statistically different from that of the warfarin group, was very high (5%) and probably has important clinical relevance. This may be due to the greater extent of ablation in persistent AF, although our study was not powered to identify the correlation. The role of atrial substrate or type of AF in periprocedural thromboembolic events in patients taking dabigatran needs to be explored further.

It is well recognized that recent (<3 to 4 weeks) initiation of anticoagulation is associated with an increased risk of bleeding in AF patients and this might have potentially confounded a part of our results (29). However, the impact of recent anticoagulation initiation on periprocedural bleeding is not clear. Moreover, all patients in our study were receiving therapeutic anticoagulation for at least 30 days. This likely minimized the impact, if any, of recent anticoagulation initiation on our study results.

Very recently, Winkle et al. (30) described their experience of using dabigatran post-procedurally after AF ablation in 123 patients. Less than 30% of them were taking pre-procedural dabigatran, and it was stopped 36 to 60 h before the procedure based on patients' glomerular filtration rate. Approximately 50% of the patients were taking warfarin pre-procedurally, which was stopped 5 days prior to the procedure, and a subcutaneous low molecular weight heparin bridge was used in them until the procedure. The intraprocedural ACT was targeted at 225 s. All patients were started on low molecular weight heparin immediately after the procedure, and dabigatran was started/restarted 22 h after the procedure. None of the patients experienced bleeding or thromboembolic complications until 30 days after the procedure. Their study is significantly different from our current study in that we evaluated the safety of periprocedural dabigatran compared with warfarin. Compared with our protocol, Winkle et al. stopped dabigatran earlier pre-procedurally and restarted it later post-procedurally. Also, they targeted a much lower ACT (225 s) compared with our study (300 to 400 s). These differences in the anticoagulation protocols probably explain the differences in the bleeding complications between both the studies.

Although our study provides initial data on slightly increased bleeding with dabigatran compared with warfarin in patients undergoing AF ablation, large randomized, controlled studies are required to confirm our results and identify an optimal periprocedural anticoagulation protocol. It is possible that holding dabigatran for >24 h without delaying the first dose after the procedure may decrease these bleeding complications and needs to be evaluated in future studies. The lack of any recommended acute reversal agents for dabigatran, at present, makes the risk of excessive major bleeding complications all the more important (27).

This was a relatively small observational study with a matched-control design. Even though, we conducted a multivariate analysis after adjusting for known predictors of bleeding complications, it is possible that other confounding variables affecting the results were unaccounted for in the study. Another limitation of the study is that the procedural techniques were operator dependent, which could potentially confound the results, even though all our centers predominantly practice a similar protocol. Nevertheless, a scarcity of data on the safety of continuing dabigatran during AF ablation procedure makes our study very important to the current electrophysiology practice, especially with the increasing use of dabigatran in clinical practice.

In patients undergoing AF ablation, continuation of dabigatran during the periprocedural period is associated with an increased risk of bleeding and composite of bleeding or embolic complications compared with uninterrupted warfarin therapy. Further studies are needed to identify the optimal periprocedural anticoagulation strategies in patients on dabigatran undergoing AF ablation.