Anticoagulation After Catheter Ablation of Atrial Fibrillation: An Unnecessary Evil?

A Systematic Review and Meta-analysis

Riccardo Proietti MD, PhD; Ahmed AlTurki MD; Luigi Di Biase MD, PhD; Paolo China MD; Giovanni Forleo MD; Andrea Corrado MD; Elena Marras MD; Andrea Natale MD; Sakis Themistoclakis MD

Disclosures

J Cardiovasc Electrophysiol. 2019;30(4):468-478. 

In This Article

Results

Our literature search identified 6918 records, after the exclusion of duplicates. After review of titles and abstracts, 38 of these references were considered potentially eligible for inclusion on the basis of the above-mentioned inclusion criteria, and their full text was analyzed in greater detail. In total, 6 were excluded because they were editorials and 16 were excluded because they did not contain the required data. Finally, a total of 16 studies were included in the analysis[11–15,19–29] (Figure 1). Of the 16 studies included, 7[11,13,15,22,23,25,28] were rated as being of good quality, 8[12,14,19,20,24,26,27,29] were rated as fair, and 1[21] was rated as poor, as assessed by the National Institute of Health quality assessment tool for observational cohort and cross-sectional studies. These ratings and the Newcastle-Ottawa score are summarized in Table 1.

Figure 1.

Forest plot reporting the RR (A) and RD (B) for CVE in off-OAC vs on-OAC patients after ablation. CVE, cerebrovascular events; OAC, oral anticoagulants; RR, risk ratio; RD, risk difference

Study Characteristics

Of the studies included, 10 were prospective cohort studies[12,14,15,19,21,22,25,27–29] and 6 were retrospective cohort studies.[11,13,20,23,25,26] All studies enrolled patients after catheter ablation and had a mean follow-up of at least 12 months. All but three studies[13,20,27] had a blanking period of at least 3 months after catheter ablation before OAC was discontinued. The anticoagulation used was predominantly warfarin in all but three studies,[14,24,26] which included patients on direct-acting oral anticoagulants. In 12 studies,[11,12,14,15,19–21,23–25,28,29] OAC was replaced by aspirin in the discontinuation group. The main indication to stop OAC was the persistence of sinus rhythm and when reported OAC was generally restarted if AF recurred. All 16 studies reported the stroke event rate in both groups of patients, and 14 of the 16 studies reported rates of bleeding in both groups. The studies included enrolled a total of 25 177 patients, divided into 13 166 patients off OAC and 12 011 patients on OAC. The average age ranged from 51 to 66 years in patients off OAC and 58 to 66 years in those on OAC. The proportion of males ranged from 65% to 88% in patients off OAC and 55% to 79% in those on OAC. The average CHADS2 score ranged from 0.5 to 3.1 in patients off OAC and 0.9 to 3.0 in patients on OAC. Three studies[12,13,27] stratified patients according to CHA2DS2-VASc score (<2 or ≥2) and two studies[11,27] according to CHADS2 (<2 or ≥2) and reported CVE and major bleeding in the two groups (on and off OAC) for every subgroup of CHA2DS2-VASc score (<2 or ≥2) and CHADS2 (<2 or ≥2). Baseline characteristics of the studies are summarized in Table 1.

Incidence of CVE in Patients on OAC vs off OAC After AF Ablation

A total of 16 studies reported this outcome.[11–15,19–29] In one study,[28] the number of events was zero in both groups and the risk ratio could not be calculated. The analysis included 25 177 patients, among which 252 CVE occurred during the follow-up: 117 (0.9%) in the group off OAC and 135 (1.07%) a in group on OAC. The pooled RR was 0.66 (CI, 0.38, 1.15, I-squared 69.1%). Repeating the analysis on the RD, the pooled result was RD −0.002 (−0.007, 0.002) (Figure 1). In both analyses, there was no statistically significant difference between the two groups.

Risk of Major Bleeding Events in Patients off OAC and on OAC After AF Ablation

A total of 14 studies[11–15,19–21,23,25–29] considered the incidence of major bleeding in patients off OAC vs on OAC. Overall 23 948 patients were included in the analysis. At the follow-up, 70 (0.5%) and 276 (2.2%) episodes of major bleeding were recorded in the group off OAC and on OAC, respectively.

A significant increase in episodes of major bleeding was observed in patients on OAC compared with those off OAC, the cumulative RR for episodes of major bleeding being 0.17 (CI, 0.09, 0.34 and the pooled RD −0.012 (CI −0.019, −0.006) (Figure 2).

Figure 2.

Forest plot reporting the RR (A) and RD (B) for major bleeding in off-OAC vs on-OAC patients after ablation. CVE, cerebrovascular events; OAC, oral anticoagulants; RD, risk difference; RR, risk ratio

Risk of CVE and Major Bleeding in Patients off OAC and on OAC After AF Ablation Stratified by CHA2DS2-VASc Score

Three studies[12,13,27] reported the number of CVE and episodes of major bleeding in the two groups stratified by CHA2DS2-VASc score (<2 and ≥2). Overall, 2983 patients with CHA2DS2-VASc score less than 2 and 1358 with CHA2DS2-VASc score equal to or higher than 2 were included in the analysis. In the group with CHA2DS2-VASc score ≥2, 514 patients were on-OAC and 844 off-OAC, while among those with CHA2DS2-VASc score <2, 1786 patients were on-OAC and 1197 were off-OAC.

There was no significant difference in the incidence of CVE between the two groups (on-OAC and off-OAC) on considering both CHA2DS2-VASc score <2 and ≥2, the pooled RR being 0.76 (CI 0.26, 2.25) and 1.51 (CI 0.63, 3.30), respectively (Figure 3). However, the RR for bleeding was significantly lower in off-OAC patients with CHA2DS2-VASc score <2 (RR 0.08; CI 0.02, 0.37), while it proved nonsignificant in those with CHA2DS2-VASc score ≥2 (RR 0.27; CI 0.05, 1.53) (Figure 4).

Figure 3.

Forest plot reporting the RR for CVE in off-OAC vs on-OAC patients after ablation, stratified by CHA2DS2-VASc. CVE, cerebrovascular events; OAC, oral anticoagulants; RR, risk ratio

Figure 4.

Forest plot reporting the RR for major bleeding in off-OAC vs on-OAC patients after ablation, stratified by CHA2DS2-VASc. OAC, oral anticoagulants; RR, risk ratio

Risk of CVE and Major Bleeding in Patients off OAC and on OAC after Ablation, Stratified by CHADS2 Score

Two studies[11,27] provided data on CVE and major bleeding in off-OAC and on-OAC patients stratified by CHADS2 score. Overall, 4397 patients with CHADS2 <2 and 948 with CHADS2 ≥2 were included. The pooled analysis showed no statistically significant difference in the risk of CVE between patients on OAC and those off OAC, whether their CHADS2 was <2 or ≥2 (Figure 5).

Figure 5.

Forest plot reporting the RR for CVE in off-OAC vs on-OAC patients after ablation, stratified by CHADS2 score.CVE, cerebrovascular events; OAC, oral anticoagulants; RR, risk ratio

The pooled RR of major bleeding was significantly different between on-OAC and off-OAC patients with CHADS2 score <2 (RR 0.07; CI 0.01, 0.31). However, in those with CHADS2 score ≥2 there was only a trend towards a higher incidence among those on OAC (RR 0.12; CI 0.01, 1,89) (Figure 6).

Figure 6.

Forest plot reporting the RR for major bleeding in off-OAC vs on-OAC patients after ablation, stratified by CHADS2 score. OAC, oral anticoagulants; RR, risk ratio

Meta-regression

A meta-regression was carried out for two variables: percentage of patient with AF recurrence and aspirin. Of note, the percentage of patients with AF recurrence impacts the treatment effect in the two groups (P = 0.001) (Table 2). The use of aspirin also reaches the statistic significance (P = 0.05).

Publication Bias Assessment

Publication bias was assessed visually by means of a funnel plot for the primary outcome of CVE in patients off-OAC vs on-OAC patients. Visually, a trend toward asymmetry was seen mainly among studies with smaller sample sizes, which would suggest publication bias (Figure 7). Accordingly, the Egger's test showed that bias existed in the studies included in the meta-analysis (P = 0.01). The Galbraith plot displayed in Supplementary Information Figure 2 shows the α intercept at −1.93, indicating a trend toward a lower level of accuracy in studies with smaller sample sizes.

Figure 7.

Publication bias: the orange line reflects a regression line of Egger's test

Supplemental Figure 1.

Flow Chart of published studies

Supplemental Figure 2.

Galbraith Plot displaying an explanation of the asymmetry of funnel plot. The intercept value of the regression line is at -1.96, showing a lower accuracy of the studies with smaller sample sizes. SND=standard normal deviation.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....