Guidelines for ablation in the management of AF emphasise its role in controlling symptoms.1,2 Previous studies, such as CABANA, have addressed its potential to improve prognosis, focusing on patients at greatest risk of stroke.3 CABANA identified an important benefit in reducing mortality in AF patients with heart failure, but failed to demonstrate a significant reduction in stroke risk. A recent meta-analysis of CABANA and other studies confirms that AF ablation does reduce the risk of stroke, raising the question of greatest clinical importance: after a successful AF ablation, is long-term anticoagulation necessary?4
The OCEAN trial
The OCEAN trial addresses this question.5 This prospective, multicentre, randomised, open-label trial with blinded outcome assessment enrolled 1,284 patients across 56 centres who had undergone catheter ablation for nonvalvular AF and maintained sinus rhythm for at least 1 year. The cohort was mostly male (71.5%), with a mean age of 66 years and a mean CHA2DS2-VASc score of 2.2; approximately one-third had a score of 3 or higher. Participants were randomised to receive either rivaroxaban 15 mg daily or aspirin 75–120 mg daily.
The composite primary endpoint included clinically overt stroke, systemic embolism or new cerebral infarct ≥15 mm detected by MRI at baseline and at 3 years. This endpoint occurred at a low rate (0.8% with rivaroxaban versus 1.4% with aspirin), with no statistically significant difference between the groups (RR 0.56; 95% CI [0.19–1.65]; p=0.28). Major or fatal bleeding was more common with rivaroxaban 1.6% versus 0.6% than with aspirin (HR 2.51, 95% CI [0.79–7.95]), although this difference was not statistically significant. Clinically relevant non-major bleeding occurred in 5.5% of the rivaroxaban group versus 1.6% of the aspirin group (HR 3.51, 95% CI [1.75–7.03]). Together, these results suggest that in low- to intermediate-risk patients, cessation of anticoagulation may be feasible after durable ablation success.
The OCEAN trial was terminated prematurely due to low event rates and the absence of statistically significant differences between treatment arms. We are left with a signal of potential clinical relevance rather than a definitive conclusion. As the implications run counter to current guidelines and as the endpoint includes life-changing strokes, a high level of certainty is required.1,2
Strengths
OCEAN was well-designed and appropriately powered to help fill this gap in our knowledge. A key strength is that it used baseline and 3-year brain MRI to detect silent ischaemic events in addition to the hard end point of stroke, which significantly augmented its statistical power.
The trial protocol was meticulous with clear pathways regarding patient drug switching, medication intolerance and peri-procedural regimens. Rigorous adherence to the trial protocol was reflected in the low drop-out rates, with 97.4% of patients completing the full trial protocol.
Assessment of ablation success relied on discrete periods of ambulatory monitoring. While continuous rhythm monitoring might have been more sensitive, this pragmatic approach improves real-world applicability. Most participants had paroxysmal AF, although outcomes did not differ significantly in those with persistent AF. Despite the open-label design, outcome assessors including MRI reporters were blinded, mitigating bias. Significantly more patients in the aspirin group had a history of previous stroke (17 versus seven), but this discrepancy did not appear to influence the results and overall CHA2DS2-VASc scores across both groups were similar.
The cohort had a low to moderate risk for stroke, but the observed event rate was lower than predicted, and total stroke risks across the trial groups were similar to that observed in cohorts with a CHA2DS2-VASc score of 2 without known AF. The overall event rates of 0.3% in the aspirin group and 0.66% in the anticoagulant group are approximately the rates observed in patients who have device-detected AF episodes between 6 minutes and 24 hours, a cohort considered a grey area for initiating anticoagulation with current data.
Weaknesses
Most studies that identify silent brain infarctions use a cutoff at >3 mm or >5 mm, but in OCEAN >15 mm was specified, due to smaller infarct sizes generally being lacunar in origin with non-embolic aetiology. However, the neuroimaging consensus statement upon which this is based recognises the relatively arbitrary nature of this size cutoff, given the paucity of objective guiding evidence.6 There is a possibility that smaller embolic silent infarcts may have been underrepresented; however, given the low overall event rate, the likelihood of this having a significant impact on the findings is low. When reassessing the data and including the <15 mm lesions, roughly similar numbers occurred in both groups.
The use of 15 mg rivaroxaban rather than 20 mg may be a departure from usual practice for many practitioners. With a lower dosage of anticoagulant, a higher stroke risk might be expected. The fact that major or fatal haemorrhages outnumbered clinical embolic events in the rivaroxaban group suggests that the choice of 15 mg was appropriate for the population in question.
Though the trial offered low-dose aspirin as an alternative, some patients took neither aspirin nor an anticoagulant, and previous evidence has shown no benefit of aspirin in this setting. The question that OCEAN comes close to answering, but does not directly address, is whether we can leave patients with neither anticoagulant nor antiplatelet after ablation. The argument presented by the authors of OCEAN is that the anti-embolic effects of aspirin are negligible and, therefore, aspirin is equivalent to placebo. This was perhaps a missed opportunity, but other trial results can help close the evidence gap.
OCEAN in Context
Whereas retrospective data on cessation of anticoagulation after AF ablation has been mixed, there was already a trend suggesting a reduced stroke risk post-ablation, a trend that has grown stronger.7–10 The concept of AF burden as a determinant of stroke risk has emerged from data on device-detected atrial arrhythmias. Outcome ascertainment is now more nuanced, quantifying the degree of success rather than reducing it to a binary variable. Methods in AF ablation have evolved quickly, and experience has accumulated at an institutional level and for individual operators. AF ablation is a full-time job for a large proportion of cardiologists, and few would dispute that results have improved accordingly.
When seen in the larger context of literature examining post AF ablation stroke rates, the large, prospective ALONE-AF trial should be considered alongside OCEAN (Table 1 ).11 The statistical strength and strikingly congruent results of these prospective studies demand a re-evaluation of current practices.
ALONE-AF used no anticoagulant rather than aspirin, and the stroke risk post successful AF ablation appears to be low enough to discontinue anticoagulation. This prospective, multicentre, open-label, randomised trial involved 840 arrhythmia-free patients at least 1-year post-ablation with a mean CHA2DS2-VASc of 2.3. Participants were randomised to continue anticoagulation (apixaban 5 mg twice daily or rivaroxaban) or to discontinue anticoagulation without antiplatelet therapy. Over a median follow-up of 2 years, the primary composite endpoint of stroke, systemic embolism or major bleeding occurred in 0.3% (off-oral anticoagulant) versus 2.2% (on-oral anticoagulant continuation). Thromboembolic events were rare in both arms (0.3% versus 0.8%), while major bleeding and clinically relevant non-major bleeding were substantially more common in patients continuing anticoagulation. The overall event rate in this trial was higher than observed in OCEAN, possibly due to marginally less rigorous monitoring to define successful AF ablation.
OPTION and OCEAN
The OPTION trial compared left atrial appendage occlusion to direct oral anticoagulant after AF ablation without a prerequisite for ablation success (though achieved in 88.1% of patients).12 Across a total of 1,600 patients, left atrial appendage occlusion (LAAO) demonstrated non-inferiority for both composite death, stroke or systemic embolism (5.3% versus 5.8%), major bleeding (3.9% versus 5%) and superiority for relevant non-major bleeding (8.5% versus 18.1%). Although not meeting statistical significance, the trends favoured LAAO. The population had a significantly higher CHA2DS2-VASc score than OCEAN, but did include lower-risk patients. As the apparent benefit with LAAO in the OPTION trial was driven by bleeding, the results of OCEAN raise the question as to whether the OPTION patients with a successful ablation might have done better with neither anticoagulation nor LAAO.
Applying the Results
The results of OCEAN and ALONE raise the possibility of stopping anticoagulation after successful ablation and, therefore, of offering ablation as an alternative to anticoagulation in some patients. It is important to remember the context of these individual trials and limit their application to patients with similar characteristics. The results cannot be generalised to higher-risk patients (CHA2DS2-VASc >2) or those whose ablation result is uncertain. In patients who do stop anticoagulation, structured surveillance for arrhythmia recurrence would be required both before cessation of anticoagulation could be considered, and after to ensure durable success; surveillance could include implanted devices or smart devices worn by the patient.13
Ablation can now be considered as part of stroke risk modification strategies as an adjunct or an alternative to LAAO and anticoagulant medications. There are significant implications for the use of AF ablation in asymptomatic patients. The results are likely to increase the number of patients who chose AF ablation, seeking to modify their stroke risk, in turn to eliminate or delay the need for long-term anticoagulation. The OCEAN trial substantiates what many electrophysiologists had suspected; successful AF ablation can reduce the risk of stroke.