Circulation Journal
Online ISSN : 1347-4820
Print ISSN : 1346-9843
ISSN-L : 1346-9843
Arrhythmia/Electrophysiology
Current Status and Clinical Outcomes of Oral Anticoagulant Discontinuation After Ablation for Atrial Fibrillation in Japan ― Findings From the AF Frontier Ablation Registry ―
Yasuo OkumuraKoichi NagashimaMasaru AraiRyuta WatanabeKatsuaki YokoyamaNaoya MatsumotoTakayuki OtsukaShinya SuzukiAkio HirataMasato MurakamiMitsuru TakamiMasaomi KimuraHidehira FukayaShiro NakaharaTakeshi KatoWataru ShimizuYu-ki IwasakiHiroshi HayashiTomoo HaradaIkutaro NakajimaKen OkumuraJunjiroh KoyamaMichifumi TokudaTeiichi YamaneYukihiko MomiyamaKojiro TanimotoKyoko SoejimaNoriko NonoguchiKoichiro EjimaNobuhisa HagiwaraMasahide HaradaKazumasa SonodaMasaru InoueKoji KumagaiHidemori HayashiKazuhiro SatomiYoshinao YazakiYuji Watarion behalf of the AF Ablation Frontier Registry
Author information
JOURNALS FREE ACCESS FULL-TEXT HTML

2019 Volume 83 Issue 12 Pages 2418-2427

Details
Abstract

Background: The safety of discontinuing oral anticoagulant (OAC) after ablation for atrial fibrillation (AF) in Japanese patients has not been clarified.

Methods and Results: A study based on the Atrial Fibrillation registry to Follow the long-teRm Outcomes and use of aNTIcoagulants aftER Ablation (AF Frontier Ablation Registry) was conducted. Data were collected from 3,451 consecutive patients (74.1% men; age, 63.3±10.3 years) who had undergone AF ablation at any of 24 cardiovascular centers in Japan between August 2011 and July 2017. During a 20.7-month follow-up period, OAC therapy was discontinued in 1,836 (53.2%) patients; 51 patients (1.5%) suffered a stroke/transient ischemic attack (TIA), 71 (2.1%) suffered major bleeding, and 36 (1.0%) died. Patients in whom OAC therapy was discontinued were significantly younger than those in whom OACs were continued, and their CHA2DS2-VASc scores were significantly lower. The incidences of stroke/TIA, major bleeding, and death were significantly lower among these patients. Upon multivariate adjustment, stroke events were independently associated with relatively high baseline CHA2DS2-VASc scores but not with OAC status.

Conclusions: Although the incidences of stroke/TIA, major bleeding, and death were relatively low among patients for whom OAC therapy was discontinued, stroke/TIA occurrence was strongly associated with a high baseline stroke risk rather than with OAC status. Thus, discontinuation of OAC therapy requires careful consideration, especially in patients with a high baseline stroke risk.

Atrial fibrillation (AF) affects quality of life and increases the risks of stroke, heart failure, and death.1 Pulmonary vein isolation (PVI) is reported to be more effective than antiarrhythmic drug (AAD) therapy for relief of symptoms and freedom from AF recurrence,2,3 and thus might yield improved clinical outcomes in terms of ischemic stroke, heart failure, and death. For patients in whom PVI is truly curative, it should be possible to discontinue oral anticoagulants (OACs) without fear that an AF-related clinical event will occur. Despite advancements in ablation technologies and strategies, however, the reported 1-year paroxysmal and non-paroxysmal AF recurrence rates are approximately 20% and >30%, respectively, and the rates increase gradually over the long term.4,5 Therefore, it remains unclear whether anticoagulant therapy can be safely terminated after PVI. Several current guidelines68 suggest that OAC discontinuation is limited to patients with a low CHADS2 score, and this has been confirmed by several clinical reports.912 Although most of the reported clinical data has been collected in Western countries, the findings may be applicable to Japanese patients, because the incidence of adverse clinical events among Japanese patients with AF appears to be lower than that among Western patients.1317 Therefore, we conducted a registry-based observational study to investigate the rate at which anticoagulation therapy is terminated after AF ablation among patients in Japan and the relationship between discontinuation and the long-term incidences of stroke, major bleeding, cardiovascular events, and death.

Methods

Study Population

The study began with a review of the records of 3,530 consecutive patients who had undergone catheter ablation for AF at any of 24 cardiovascular centers in Japan between August 2011 and July 2017. All patients were listed in the Atrial Fibrillation registry to Follow the long-teRm Outcomes and use of aNTIcoagulants aftER Ablation (AF Frontier Ablation Registry) (UMIN Clinical Trials Registry: UMIN000026849). The intended follow-up for each patient was a minimum of 12 months. Of the total patients, 79 were lost to follow-up, leaving 3,451 patients for inclusion in the study. Patients consented to the use of their anonymized clinical data for research purposes by the opt-out method, and the study protocol was approved by the institutional review boards of the 24 hospitals where patients were treated.

Data Collection

Patients’ characteristics and pre-ablation (baseline) and follow-up data were obtained through a review of their hospital charts. The anonymized patient data were collected in the Excel form by physicians or the clinical research coordinator at each institution, and included patients’ characteristics such as sex, age, body mass index (BMI), type of AF; number of ablation sessions (1, 2, 3 or >3 sessions); any comorbidity (hypertension, diabetes, history of a stroke/transient ischemic attack [TIA]); use of an OAC, including the type (warfarin vs. direct OAC [DOAC]), use of an antiplatelet drug; and laboratory test results (hemoglobin concentration and creatinine clearance [CrCl]); transthoracic echocardiography-derived left atrial diameter (LAd) and left ventricular ejection fraction (LVEF), and the ablation method (radiofrequency [RF], cryoballoon or hot balloon). Follow-up variables included OAC discontinuation, AF recurrence after a blanking period of 3 months, occurrence of any adverse clinical event during the follow-up period, and whether AAD use had been discontinued by the end of the follow-up period.

Ablation Protocol

All AADs were discontinued 1 week before ablation. Generally, OACs were continued, but this varied from institution to institution, and the procedure itself was performed according to each institution’s particular protocol but generally as previously described.18,19 Extensive encircling PVI was guided by a circular mapping catheter or multiple-electrode catheter and a 3D mapping system. The ablation catheter was an irrigated-tip contact force (CF) sensing catheter or irrigated-tip standard non-CF sensing catheter, depending on the hospital’s preference or type of catheter available at the time of the procedure. In some patients, adenosine triphosphate was injected intravenously after PVI to provoke dormant PV conduction (performed at the physician’s discretion). If acute PV reconnection or dormant PV conduction was evident, touch-up ablation was performed.

Cryoballoon ablation was performed with an Arctic Front Advance cryoballoon (Medtronic, Minneapolis, MN, USA), as described previously.19 Each PV was encircled once (but twice if necessary). Any touch-up ablation required for dormant conduction or residual PV potentials was performed with a standard irrigated-tip catheter.

Regardless of the ablation method, if sinus rhythm was not restored after PVI, additional linear LA ablation (e.g., mitral isthmus line or LA roof line ablation) was performed, and the residual potentials, including complex fractionated atrial ECGs in the LA, were ablated, as appropriate. Electrical cardioversion was performed when sustained AF/atrial tachycardia was recorded after additional LA ablation. Tricuspid valve isthmus ablation and superior vena cava isolation were also performed when necessary.

Post-Ablation Follow-up

All patients underwent routine follow-up examinations at their hospital’s outpatient clinic. AAD and OAC therapies were generally continued during the 3-month post-ablation blanking period. Thereafter, for patients in whom the arrhythmia did not recur, OAC and AAD therapies were discontinued. The post-ablation medications depended on the physician’s discretion, generally based on the patient’s characteristics or preferences. 24-hour Holter recordings were obtained 3–6 months after ablation. An ECG event recorder was used for any patient who reported cardiac symptoms. Any documented AF episode >30 s in duration on the standard ECG recording, event-activated ECG recording, or 24-hour Holter recording was considered a recurrence.

Study Endpoints

The primary study endpoint was discontinuation of OAC therapy after ablation. There were 2 secondary endpoints: occurrence of a stroke (ischemic stroke or hemorrhagic stroke) or a TIA (efficacy endpoint) and occurrence of major bleeding (defined as a reduction in the hemoglobin concentration of at least 2 g/dL, transfusion of at least 2 units of blood or symptomatic bleeding in a critical area or organ) (safety endpoint). Cardiovascular events, including hospitalization for heart failure, myocardial infarction/unstable angina or other cardiovascular events, and death from any cause (cardiovascular death, stroke-related death, and non-cardiovascular death) were assessed.

Statistical Analysis

Categorical variables are presented as the number and percentage of patients, and differences in these variables between patients in whom OAC therapy was discontinued (OAC discontinuation group) and in whom OAC therapy was not discontinued (OAC continuation group) were analyzed by chi-square or Fisher’s exact test. Continuous variables are presented as mean±SD or median (and interquartile range) values, and between-group differences in these variables were analyzed by Student’s t-test or Wilcoxon rank sum test. Cumulative OAC discontinuation and clinical adverse event rates were estimated by the Kaplan-Meier method, and differences were analyzed by log-rank test. Cox proportional hazards modeling was performed to identify the significant determinants for the incidence of stroke/TIA and major bleeding. Actors shown to be significant by univariate analysis were entered into the multivariate Cox model. Receiver-operating characteristic curves were drawn to identify the best cutoff value for each variable and the prognostic performances variables were compared on the basis of the area under the curve (AUC) by the DeLong test. Statistical analyses were performed with JMP 12 (SAS Institute Inc., Cary, NC, USA) or MedCalc Software Version 16.4.1 (Mariakerke, Belgium), and P<0.05 was considered statistically significant for all analyses.

Results

Clinical Characteristics

Clinical characteristics ae shown for the total study patients and per study group in Table 1. The total male:female ratio was 2,558:893 and mean age 63.3±10.3 years. The AF was paroxysmal (lasting <7 days) in 2,157 (62.5%) patients, persistent (lasting ≥7 days to 12 months) in 1,036 (30.0%), and long-lasting persistent (lasting ≥12 months) in 258 (7.5%). Mean CHADS2 and CHA2DS2-VASc scores were 1.2±1.1 and 2.1±1.5, respectively. Standard RF ablation was performed in 1,523 (44.1%) patients, CF-based RF ablation in 1,678 (48.6%), cryoballoon ablation in 248 (7.1%), and hot balloon ablation in 2 (0.1%). Patients’ mean LAd was 40.0±6.6 mm and LVEF was 63.6±9.6%. Of the total 3,451 patients, 3,038 (88.0%) underwent a single AF ablation procedure, and the remaining 413 (12.0%) underwent ≥1 additional ablation procedures. During the post-ablation blanking period, warfarin was used by 711 (20.6%) patients, and DOACs were used by 2,698 (78.0%) patients, but the OAC type in 42 patients (1.2%) was not reported.

Table 1. Clinical Characteristics of the Total Patients and per Study Group
  Total
(n=3,451)
OAC discontinuation
group (n=1,836)
OAC continuation
group (n=1,615)
P value*
Age (years) 63.3±10.3 61.3±10.3 65.4±9.9 <0.001
 <65 1,869 (52.9) 1,130 (61.5) 696 (43.1)  
 ≥65, <75 1,200 (34.8) 567 (30.9) 633 (39.2) <0.001
 ≥75 425 (12.3) 139 (7.6) 286 (17.7)  
Female sex 893 (25.9) 429 (23.4) 464 (28.7) <0.001
Body mass index (kg/m2) 24.0±3.6 23.8±3.6 24.2±3.7 <0.001
Type of AF
 Paroxysmal 2,157 (62.5) 1,231 (67.1) 1,926 (57.3)  
 Persistent 1,036 (30.0) 504 (27.5) 535 (32.9) <0.001
 Long-standing persistent 258 (7.5) 101 (5.5) 157 (9.7)  
Ablation sessions ≥2 413 (12.0) 176 (9.6) 237 (14.7) <0.001
Comorbidities
 Hypertension 1,891 (54.8) 878 (47.8) 1,013 (60.7) <0.001
 Diabetes 561 (16.3) 227 (12.4) 334 (20.7) <0.001
 History of stroke/TIA 275 (8.0) 70 (3.8) 205 (12.7) <0.001
 Heart failure 601 (17.4) 294 (16.0) 307 (19.0)  0.021
 Vascular disease 302 (8.8) 106 (5.8) 196 (12.1) <0.001
CHADS2 score 1.2±1.1 0.9±0.9 1.5±1.2 <0.001
 0 1,022 (29.6) 684 (37.3) [66.9] 338 (20.9) [33.1] <0.001
 1 1,370 (39.7) 761 (41.5) [55.5] 609 (37.7) [44.5]  
 2 683 (19.8) 292 (15.9) [42.8] 391 (24.2) [57.2]  
 ≥3 376 (10.9) 99 (5.4) [26.3] 277 (17.2) [73.7]  
CHA2DS2-VASc score 2.1±1.5 1.6±1.3 2.5±1.6 <0.001
 0 537 (15.6) 372 (20.3) [69.3] 165 (10.2) [30.7]  
 1 871 (25.2) 563 (30.7) [64.6] 308 (19.1) [35.4]  
 2 878 (25.4) 485 (26.4) [55.2] 393 (24.3) [44.8] <0.001
 3 611 (17.7) 253 (13.8) [41.4] 358 (22.2) [58.6]  
 ≥4 554 (16.1) 163 (8.9) [29.4] 391 (24.2) [70.6]  
Echocardiographic variables
 LVEF (%) 63.6±9.6 64.3±8.7 62.9±10.5 <0.001
 Left atrial diameter (mm) 40.0±6.6 38.9±6.3 41.6±6.6 <0.001
Post-ablation therapy
 Oral anticoagulation
  Warfarin 711 (20.6) 338 (18.4) 373 (23.1)  
  Dabigatran 654 (19.0) 353 (19.2) 301 (18.6)  
  Rivaroxaban 828 (24.0) 438 (23.9) 390 (24.2) <0.001
  Apixaban 884 (25.6) 503 (27.4) 381 (23.6)  
  Edoxaban 332 (9.6) 171 (9.3) 161 (10.0)  
  Not reported 42 (1.2) 33 (1.8) 9 (0.6)  
 Antiplatelet therapy 240 (7.0) 83 (4.5) 157 (9.7) <0.001
 Antiarrhythmic drug use at final follow-up 1,113 (32.3) 330 (18.0) 783 (48.5) <0.001
Hb (mg/dL) 14.2±1.5 14.3±1.5 14.0±1.6 <0.001
CrCl (mL/min) 73.4±25.7 76.4±25.0 70.1±26.0 <0.001
AF recurrence 1,046 (30.3) 333 (18.1) 713 (44.2) <0.001

Values are shown as mean±SD or n (%). [%] indicates the % of the OAC discontinuation rate or continuation rare per each stroke risk score. *Student’s t-test, Wilcoxon rank sum test, or chi-square test, as appropriate, for the OAC discontinuation group vs. OAC continuation group. AF, atrial fibrillation; CHADS2, congestive heart failure=1, hypertension=1, age ≥75 years=1, diabetes=1, and stroke/TIA=2; CHA2DS2-VASc, congestive heart failure=1, hypertension=1, age ≥75 years=2, diabetes=1, stroke/TIA=2, vascular disease=1, age 65–74 years=1, and female sex=1; CrCl, creatinine clearance; Hb, hemoglobin; LVEF, left ventricular ejection fraction; OAC, oral anticoagulant; TIA, transient ischemic attack.

Outcomes and OAC Discontinuation/Continuation During Follow-up

During the final follow-up (median 20.7 [12.7–33.2] months), 2,405 (69.7%) of the 3,451 patients were arrhythmia-free, with 565 (23.5%) of these patients still taking an AAD. Kaplan-Meier freedom rate from AF recurrence among patients who underwent 1 ablation procedure was 78.4% at 1 year and 69.8% at 2 years, whereas that among patients who underwent ≥2 ablation procedures was 76.0% at 1 year and 65.0% at 2 years. OACs were discontinued in 1,836 (53.2%) patients (discontinuation group) at 5.4 (3.1–11.0) months after ablation. The OAC discontinuation rate increased predominantly in 43.3% at 1 year but thereafter increased modestly to 56.2% at 2 years (Figure 1A). The OAC discontinuation rate decreased gradually without a tipping point as the CHADS2 and CHA2DS2-VASc scores increased (Figure 1B,C; Table 1). As shown in Table 1, male sex was more prevalent in the OAC discontinuation group than in the OAC continuation group, patients in the OAC discontinuation group were significantly younger than those in the OAC continuation group, and BMI was significantly lower in this group. Antiplatelet drugs use was less prevalent, and the baseline hemoglobin concentration and CrCl were higher in this group. Also higher in this group were the prevalence of paroxysmal AF, the percentage of patients who required only 1 ablation session, the percentage of patients without a comorbidity, and the percentage of patients in whom AF did not recur during the follow-up period (P<0.05 for all). CHADS2 and CHA2DS2-VASc scores were lower, LVEF was greater, and LAd smaller in this group than in the OAC continuation group. DOAC use was greater in this group, as was non-use of AAD by the time of the final follow-up examination (P<0.05 for all).

Figure 1.

Kaplan-Meier curves for oral anticoagulant (OAC) discontinuation rates in the total patients (A), for each CHAD2 score (B) and each CHA2DS2-VASs score (C).

By the time of the final follow-up examination, stroke/TIA had occurred in 51 (1.5%) of the total patients, major bleeding had occurred in 71 (2.1%), there was a cardiovascular event in 106 (3.1%), and 36 (1.0%) patients had died. Cardiovascular events consisted of 23 (0.7%) cases of myocardial infarction/unstable angina, 44 (1.3%) hospitalizations for heart failure, and 39 (1.1%) other cardiovascular events. Deaths comprised 5 (0.1%) cardiovascular deaths, 4 (0.1%) stroke-related deaths, and 27 (0.8%) non-cardiovascular deaths. The overall incidence of clinical adverse events, including stroke/TIA, major bleeding, cardiovascular events, and death was significantly lower in the OAC discontinuation group than in the OAC continuation group (P<0.05 for all) (Figure 2). In the OAC discontinuation group 18 patients suffered a stroke/TIA, and these patients were significantly older than those in the same group who did not suffer a stroke/TIA (n=1,818) (69.7±10.9 years vs. 61.3±10.3 years, P<0.001). In addition, their CHADS2 (1.8±1.2 vs. 0.9±0.9, P<0.001) and CHA2DS2-VASc scores were higher (3.0±1.5 vs. 1.6±1.3, P<0.001). In the OAC discontinuation group, AF recurrence was more prevalent among these patients also (38.8% [7/18] vs. 17.9% [326/1,818], P=0.031). Similarly, patients in this group who suffered major bleeding (n=29) were significantly older than those who did not (n=1,807) (71.8±7.3 years vs. 61.2±10.3 years, P<0.001), and their CHADS2 (1.7±1.4 vs. 0.9±0.9, P<0.001) and CHA2DS2-VASc scores were higher (2.9±1.8 vs. 1.6±1.3, P<0.001). AF recurrence tended to be more prevalent among these patients than among those who did not suffer major bleeding (31.0% [9/29] vs. 17.9% [324/1,807], P=0.09).

Figure 2.

Kaplan-Meier curves for stroke/transient ischemic attack (TIA), major bleeding, cardiovascular events, and all-cause death after ablation for atrial fibrillation in patients in whom oral anticoagulant (OAC) therapy was discontinued (OAC discontinuation) and those in whom it was continued (OAC continuation).

The major determinants of stroke/TIA and major bleeding are shown in Table 2. Age, CHA2DS2-VASc score, and LAd were positively related to the incidence of stroke/TIA (hazard ratio [HR] for per year increase in age: 1.06; HR per 1-point increase in CHA2DS2-VASc: 1.44; and HR per millimeter increase in LAd: 1.06). Long-lasting-AF (vs. paroxysmal AF), hypertension, heart failure, vascular disease, and OAC continuation were shown to be significantly associated with the incidence of stroke/TIA but not with AF recurrence after ablation. On multivariate Cox hazard analysis, long-lasting-AF (HR: 3.07 [95% confidence interval (CI): 1.31–6.63] vs. paroxysmal AF) and CHA2DS2-VASc score (HR: 1.31 [95% CI: 1.06–1.60]) remained significant predictors for the incidence of stroke/TIA. Age and CHA2DS2-VASc score were positively related to the incidence of a major bleeding event (HR per year increase in age: 1.10; HR per 1-point increase in CHA2DS2-VASc: 1.43). BMI, hemoglobin concentration, and CrCl were inversely related to the occurrence of a major bleeding event. Use of antiplatelet drugs and OAC continuation were significantly associated with the occurrence of a major bleeding event. Multivariate Cox hazard model revealed age (HR: 1.06 [95% CI: 1.02–1.11] per year increase) as a significant predictor of a major bleeding event.

Table 2. Univariate and Multivariate Cox Hazard Models for Predicting Stroke/TIA and Major Bleeding Events
  Stroke/TIA Major bleeding
Univariate analysis Multivariate analysis Univariate analysis Multivariate analysis
HR (95% CI) P value HR (95% CI) P value HR (95% CI) P value HR (95% CI) P value
Age (+1 year) 1.06 (1.03–1.10) 0.003 1.03 (0.99–1.07) 0.16 1.10 (1.06–1.13) <0.001 1.06 (1.02–1.11) 0.020
Female 0.94 (0.48–1.72) 0.85     1.59 (0.96–2.58) 0.07    
BMI (+1 kg/m2) 1.00 (0.92–1.00) 0.92     0.91 (0.84–0.97) 0.007 0.95 (0.88–1.03) 0.26
AF type
 Paroxysmal Ref.   Ref.   Ref.      
 Persistent 1.07 (0.54–2.00) 0.64 0.91 (0.45–1.73) 0.76 1.05 (0.62–1.73) 0.85    
 Long-standing persistent 3.38 (1.55–6.76) 0.003 3.07 (1.31–6.63) 0.012 0.82 (0.25–2.03) 0.70    
Ablation sessions ≥2 1.54 (0.63–3.24) 0.32     0.40 (0.10–1.07) 0.07    
Comorbidities
 Hypertension 2.01 (1.11–3.85) 0.020     1.14 (0.71–1.86) 0.56    
 Diabetes 1.73 (0.89–3.16) 0.11     2.08 (1.21–2.82) 0.009    
 Heart failure 2.17 (1.15–3.90) 0.019     1.74 (0.99–2.93) 0.05    
 Stroke/TIA 2.09 (0.91–4.22) 0.08     1.07 (0.42–2.28) 0.87    
 Vascular disease 2.46 (1.20–4.64) 0.016     2.43 (1.30–4.25) 0.007    
 CHA2DS2-VASc 1.44 (1.24–1.68) <0.001 1.31 (1.06–1.60) 0.015 1.43 (1.25–1.63) <0.001 1.03 (0.84–1.25) 0.78
Echocardiography variables
 LVEF (+1%) 0.99 (0.96–1.02) 0.40     1.01 (0.98–1.04) 0.48    
 LAd (+1 mm) 1.06 (1.02–1.10) 0.006 1.04 (0.99–1.08) 0.12 1.01 (0.98–1.05) 0.42    
Post-ablation therapy
 Warfarin (vs. DOAC) 1.40 (0.76–2.51) 0.28     1.47 (0.87–2.42) 0.15    
 Antiplatelet therapy 1.15 (0.40–2.63) 0.78     2.91 (1.55–5.08) 0.002 1.79 (0.91–3.31) 0.09
 AAD use 1.25 (0.69–2.19) 0.45     1.39 (0.98–1.96) 0.07    
Hb (mg/dL) 0.85 (0.71–1.01) 0.07     0.73 (0.63–0.85) <0.001 0.88 (0.75–1.04) 0.15
CrCl (mL/min) 0.99 (0.98–1.00) 0.11     0.97 (0.96–0.98) <0.001 1.00 (0.98–1.01) 0.57
AF recurrence 1.34 (0.76–2.34) 0.31     1.56 (0.96–2.51) 0.07    
OAC discontinuation 0.44 (0.24–0.78) 0.004 0.74 (0.39–1.35) 0.32 0.53 (0.33–0.86) 0.010 0.75 (0.45–1.23) 0.26

AAD use, antiarrhythmic drug use at final follow-up; BMI, body mass index; CI, confidence interval; DOAC, direct oral anticoagulant; HR, hazard ratio; LAd, left atrial diameter; other abbreviations as in Table 1.

Prognostic Performance of Age, CHAD2 and CHA2DS2-VASc Scores for Stroke/TIA and Major Bleeding Events

The prognostic performance of age and traditional risk scores for stroke/TIA and major bleeding events was investigated (Figure 3). Performance of the CHA2DS2-VASc score for predicting stroke/TIA was best (AUC, 0.70; best cutoff value, ≥3), followed by the CHADS2 score (AUC, 0.69, P=0.45 by DeLong test vs. CHA2DS2-VASc score; best cutoff value, ≥2) and age (AUC, 0.65, P=0.07 vs. CHA2DS2-VASc score; best cutoff value, ≥63 years), and the performance of age for the prediction of major bleeding was best (AUC, 0.73; best cutoff value ≥70 years), followed by CHA2DS2-VASc score (AUC, 0.67, P<0.001 vs. age; best cutoff value, ≥3) and CHADS2 score (AUC, 0.60, P<0.001 vs. age; best cutoff value, ≥2). Kaplan-Meier curves for the occurrence of adverse clinical events among patients with a CHA2DS2-VASc score ≥3 are shown in Figure 4. Among patients with a CHA2DS2-VASc score ≥3 the cumulative incidence of stroke/TIA, major bleeding, cardiovascular events, and all-cause death was greater than that among patients with a CHA2DS2-VASc score <3.

Figure 3.

Prognostic performance of traditional predictors of (A) stroke risk and (B) major bleeding after ablation for atrial fibrillation. AUC, area under the curve; CI, confidence interval.

Figure 4.

Kaplan-Meier curves for stroke/transient ischemic attack (TIA), major bleeding, cardiovascular events, and all-cause death after ablation for atrial fibrillation in patients with CHA2DS2-VASc scores of <3 and ≥3.

Discussion

Our main findings were as follows. First, OAC use was discontinued at 5.1 (3.1–11.0) months after ablation in 53.2% of the 3,451 AF Frontier Ablation Registry patients included in our study. OAC discontinuation was associated with a younger age, male sex, paroxysmal AF, a single AF session, absence of comorbidities, a smaller LAd, AAD discontinuation by the time of the final follow-up examination, and absence of AF recurrence after ablation. Second, stroke/TIA and major bleeding events occurred in patients with high stroke risk scores, and such events were associated with patients’ baseline clinical characteristics (i.e., CHA2DS2-VASc scores and age), rather than with discontinuation vs. continuation of OAC therapy. Third, the CHA2DS2-VASc score tended to be superior to age for prediction of post-ablation stroke/TIA, and age was found to be superior to the CHADS2 and CHA2DS2-VASc scores for prediction of a major bleeding event.

Post-Ablation OAC Discontinuation and Clinical Outcomes

OACs were discontinued within 21 months after ablation in slightly more than 50% of patients included in our study. Discontinuation of OACs beyond 3 months after ablation is reported in 20–80% of patients,913,20,21 and observational studies have reported that discontinuation of OACs is dependent on multiple factors, such as ablation success and initial stroke risk, but mainly on the treating physician’s inclination or patient’s preference. In our study, age, male sex, and smaller LAd were associated with OAC discontinuation. A single AF session, absence of comorbidities, and absence of AF recurrence were also associated with OAC discontinuation. Despite the widely varying OAC discontinuation rates documented in previous observational studies, most authors have concluded that OACs can be discontinued in patients judged to be at low risk for a stroke after successful AF ablation,913,20,21 and the increased incidence of major bleeding events associated with continuous OAC use outweighs its benefit in terms of stroke prevention. According to current guidelines,68 OAC therapy should be continued when, even after AF, the CHA2DS2-VASc score is ≥2. Interestingly, we found a score of 2 to be the best cutoff CHA2DS2-VASc score for OAC discontinuation but the AUC of 0.65 was low (data not shown in the Results). The OAC discontinuation rate gradually decreased as this score increased and there was no tipping point (Figure 1). Moreover, more than one-quarter of high-risk patients with CHADS2 score ≥3 or CHA2DS2-VASc score ≥4 discontinued OAC during the follow-up (Table 1). These findings suggested that in real-world settings, physicians might not always consider the guidelines. Therefore, clarifying whether the physicians’ inclinations regarding OAC discontinuation are appropriate is clinically important, and no pertinent data are available in Japan.

Our Japanese registry-based study showed significantly lower incidences of stroke/TIA, major bleeding, cardiovascular events, and death among patients for whom OAC use was discontinued after AF ablation than among those for whom OAC use was continued. The incidence of stroke/TIA was 0.4%/year among patients for whom OAC use was discontinued, but 1.1%/year among patients for whom OAC use was continued. We assume this is because OAC use was discontinued in patients with low stroke risk scores following successful ablation. Our data suggested that a decision to discontinue OAC therapy based mainly on a patient’s characteristics is clinically reasonable. Despite the low incidence of stroke/TIA events among the present patients in whom OAC therapy was discontinued, physicians should be cautious when considering OAC discontinuation, because 35% (18/51) of patients who suffered a stroke/TIA and 40.8% (29/71) of those who suffered major bleeding were part of the OAC discontinuation group. Importantly, those patients were significantly older and their stroke risk scores were higher than those of patients in the same group who did not suffer a stroke/TIA or major bleeding. Furthermore, a CHA2DS2-VASc score ≥3, rather than OAC continuation vs. discontinuation, was identified as an independent predictor for each of these clinical events.

There is little data supporting post-AF ablation OAC discontinuation, especially in patients at high risk of stroke/TIA. Numerous investigators have also reported a strong association between advanced age, prior stroke/TIA, high CHADS2, CHA2DS2-VASc, and HAS-BLED scores, and stroke/TIA and major bleeding.912,20,21 Although some recent studies have shown a significant association between AF recurrence and stroke/TIA events,22 our study did not. We cannot say with certainty that AF did not recur asymptomatically in some of the patients who suffered a stroke/TIA. The incidence of asymptomatic AF recurrence increases after ablation, especially among patients at high risk for a stroke/TIA.23 Therefore, OAC discontinuation in high-stroke risk patients should be carefully considered (or may be unsafe) even if sinus rhythm is thought to be maintained after ablation, because the stroke risk may be high once AF recurs asymptomatically.

Comparison of several Japanese AF registries (majority of patients did not undergo ablation) with a similar age and stroke risk score as this study population (62.3±10.3 years, CHADS2 1.2±1.1 and CHA2DS2-VASc 2.1±1.5) provided further insight into understanding the effect of OAC discontinuation after ablation on the clinical outcomes. Among 3,588 patients (68.1±13.5 years) without OAC therapy in the pooled analysis from 3 large Japanese registries,24 the incidence of strokes was 0.69 and 0.98 events per 100 person-years for those aged <65 years and 65–75 years, respectively. Those events rates seemed higher than that of our OAC discontinuation group (0.4%/year), and therefore, ablation might have beneficial effects on stroke/TIA events in this group even if OACs are discontinued after ablation. In contrast, the incidences of stroke/systemic embolism (SE) and major bleeding events in a warfarin therapy cohort (2,822 patients aged <70 years in the J-Rhythm Registry25) were lower (∼0.4%/year and 0.7%/year, respectively) than that in our OAC continuation group (1.1%/year and 1.6%/year, respectively) despite similar ages and stroke risk (61.0±7.0 years and CHADS2 score 1.2±1.0) as our population. In another OAC therapy cohort (1,895 patients aged <75 years in the SAKURA AF registry:26 65.9±6.9 years, CHAD2 score 1.3±1.0, CHA2DS2-VASc score 2.3±1.3, and 52.0% DOAC users and 48.0% warfarin users), the incidences of stroke/SE and major bleeding events were 1.2 and 1.1 events person-years, respectively, which were similar to those in our OAC continuation group. Those data suggested that in such high-risk patients, the stroke risk might persist even if OACs are continued after ablation. Therefore, our results and those of the other studies are a strong indication that the decision to continue long-term post-ablation OAC therapy in Japanese patients should depend on the patient’s age and stroke risk profile derived from the CHA2DS2-VASc score and not on the perceived success or failure of ablation, as indicated in the recent guidelines.27

Clinical Implications

Our most important finding was that the CHA2DS2-VASc score or CHADS2 score was a better predictor of stroke/TIA than age, but age was the best predictor of major bleeding. The best cutoff CHA2DS2-VASc score for stroke/TIA events was 3, and the best cutoff age for major bleeding events was 70 years. The incidences not only of stroke/TIA and major bleeding but also of cardiovascular events and death were relatively high among patients with a CHA2DS2-VASc score ≥3. Our data implied that, for patients aged ≥71 years and for those with a CHA2DS2-VASc score ≥3, OAC continuation after ablation may be safe for stroke prevention. Improving lifestyle factors and, if needed, taking medications appropriately, including antithrombotic agents, antihypertensive drugs or statins, are important for reducing the bleeding risk and adverse events, regardless of ablation success. Nevertheless, the AUCs of each factor for predicting stroke/TIA and bleeding events were relatively low, so the clinical judgement for OAC discontinuation after ablation should be done according to the CHA2DS2-VASc score and age while considering other factors including the AF type, history of major bleeding, comorbidities, and severity of atrial remodeling, etc.

Study Limitations

Because this study was retrospective and observational, only associations could be established; causal relationships could not. Such a study is subject to patient selection bias, and the diagnostic and therapy/intervention methods were not controlled for. For example, the decision to discontinue or maintain AAD/OAC therapy depended on the physicians’ preference based on the patient’s characteristics or preferences. Several patients were lost to follow-up during the follow-up period, and therefore, the clinical event rates might have been underestimated. It is also possible that the AF recurrence rate was underestimated. We might have missed patients with asymptomatic recurrence, even though our method of detecting recurrence was that used routinely in Japan. Finally, the follow-up period was relatively short for clarifying stroke/SE and other clinical events. AF recurrence after ablation increases gradually as time passes, and thus, clarifying the association between late recurrence and stroke events will require further prospective long-term studies. Nevertheless, the study included a relatively large number of patients from multiple centers in Japan, and thus we believe that, despite the study limitations, our findings will help physicians judge whether OAC therapy can be discontinued after ablation in Japanese AF patients.

Conclusions

In approximately half of our registry patients who underwent AF ablation, OACs were discontinued, and some patients experienced AF recurrence, but neither factor was associated with the occurrence of stroke/TIA or major bleeding after AF ablation. Rather, age and initial stroke risk were related to the stroke/TIA and major bleeding events, suggesting that although OACs should be continued in high-risk patients, OACs can be discontinued in low-risk patients, especially those in a low age category.

Acknowledgments

The authors thank all of the centers that participated in this study and all of the patients who gave consent to participate. We thank Ms. Wendy Alexander-Adams and Mr. John Martin for their encouragement and assistance with the reporting of our findings in English.

Funding

This study was supported by a research grant from Bristol-Meyers Squibb.

Disclosures

The following authors have potential conflicts of interest: Y.O. received research funding from Bayer Healthcare, Daiichi-Sankyo, Bristol-Meyers Squibb, Nippon Boehringer Ingelheim, Pfizer Japan, TORAY, and Boston Scientific Japan and has accepted remuneration from Bayer Healthcare, Daiichi-Sankyo, and Bristol-Meyers Squibb. N.M. received research funding from Daiichi-Sankyo. S.S. received research funding from Daiichi-Sankyo and Mitsubishi-Tanabe. A.H. accepted remuneration from Nippon Boehringer Ingelheim, Bayer Healthcare, Bristol-Myers Squibb, Daiichi-Sankyo. M.K. accepted remuneration from Johnson & Johnson K.K., Medtronic Japan, Bayer Healthcare. H.F. received lecture fees from Nippon Boehringer Ingelheim and Daiichi-Sankyo. S.N. received lecture fees from Bayer Healthcare, Daiichi-Sankyo, Bristol-Meyers Squibb, Pfizer Japan, and Nippon Boehringer Ingelheim. T.K. received a research grant from Daiichi-Sankyo, and lecture fees from Bristol-Myers Squibb, Daiichi-Sankyo and Nippon Boehringer Ingelheim, and honoraria for writing promotional material for Bristol-Myers Squibb. W.S. received research funding from Bristol-Myers Squibb, Daiichi-Sankyo, and Nippon Boehringer Ingelheim, and patent royalties/licensing fees from Daiichi-Sankyo, Pfizer Japan, Bristol-Myers Squibb, Bayer Healthcare, and Nippon Boehringer Ingelheim. T.H. serves as a consultant to Medtronic Japan and has received lecture fees from Daiichi-Sankyo. I.N. received a scholarship from the Japanese Heart Rhythm Society and speaking honoraria from Medtronic Japan. K.O. received remuneration from Nippon Boehringer Ingelheim, Daiichi-Sankyo, Johnson & Johnson, and Medtronic Japan. M. Tokuda serves as a consultant to Medtronic Japan. T.Y. received lecture fees from Daiichi-Sankyo and Abbott Medical Japan. K.T. received lecture fees from Daiichi-Sankyo, Nippon Boehringer Ingelheim, Bristol-Myers Squibb, Pfizer Japan, Bayer Healthcare. K. Soejima received research funding from Daiichi-Sankyo and Nippon Boehringer Ingelheim, and accepted remuneration from Medtronic Japan, Jonson & Jonson, and Abbott Medical Japan. N.H. accepted remuneration from Nippon Boehringer Ingelheim, Bristol-Myers Squibb, Bayer Healthcare, and research funding from Bayer Healthcare, Nippon Boehringer Ingelheim, Daiichi-Sankyo. M.H. received lecture fee from Nippon Boehringer Ingelheim and Bristol-Myers Squibb. K. Satomi and Y.A. received research funding from BIOTRONIK Japan. Other authors have no conflicts of interest.

References
 
© 2019 THE JAPANESE CIRCULATION SOCIETY
feedback
Top