Atrial fibrillation (AF) confers a substantial risk of stroke and thromboembolism, which is effectively reduced by oral anticoagulation (OAC), whether as a vitamin K antagonist (VKA, eg, warfarin) or one of non-VKA oral anticoagulants (NOACs). With the use of OACs, the risk of bleeding remains a problem associated with these agents. Thus, a clinical dilemma remains in the optimal management of patients with AF who are at high risk of stroke but have a history of bleeding.

Among patients with non-valvular AF, over 90% of the thrombus is formed in the left atrial appendage (LAA).1 Thus, the exclusion of LAA could greatly decrease the risk of stroke in patients with AF, given that blood stasis in a fibrillating LAA is likely to lead to thrombus formation. Interventional devices to occlude the LAA are not ‘new’ and various LAA closure devices have been available for years. More recently, the Watchman (Atritech) and Amplatzer Cardiac Plug (ACP) device (St Jude Medical) are the more widely used ones in clinical practice.

The PROTECT AF (Watchman Left Atrial Appendage System for Embolic Protection in Patients with AF) trial was the first prospective randomised controlled trial (RCT) on LAA closure with 707 patients with non-valvular atrial fibrillation.2 This showed that the Watchman device was non-inferior to warfarin in patients with a high risk of thromboembolism for the primary end point, which was a composite of stroke, systemic embolism (SE) and cardiovascular (CV) death with a risk ratio of 0.62 (95% CI 0.35 to 1.25). However, the rate of intraoperative complications (composite of major bleeding, pericardial effusion, procedure related stroke and device embolisation) was relatively high at 8.7%. Also, the mean CHADS₂ score (congestive heart failure, hypertension, age, diabetes, history of stroke) was 2.2 in the Watchman device group and only <15% of patients who were screened entered the trial (707/4998), raising issues about generalisability. Latest data from 4 years follow-up in PROTECT AF trial showed that the Watchman device was superior to warfarin in terms of 40% relative reduction of stroke/SE/CV death.3 With the improvement of the device and the procedure technique, the PREVAIL (Prospective Randomised Evaluation of the WATCHMAN LAA Closure Device in Patients With Atrial Fibrillation vs Long-Term Warfarin Therapy) trial4 showed a lower rate of intraoperative complications (4.2%), but the PREVAIL trial did not achieve the prespecified criteria for non-inferiority.

In their Heart paper, Mínguez et al report the 2-year clinical outcomes from the Iberian registry are consistent with the PREVAIL trial.5 In the Iberian registry, LAA closure with ACP device was performed in 167 patients with non-valvular AF from 12 hospitals. During the follow-up period, 5.4% of subjects had intraoperative complications, and major bleeding and stroke/transient ischaemic attack events within 2 years were less frequent than the expected outcomes based on conventional risk scores. No actual ‘control group’ of patients undergoing medical therapy was studied, and reliance on event rates from an entirely different cohort, which was a large ‘all comers’ Swedish nationwide registry6—rather different from the selected patients undergoing left atrial appendage occlusion (LAAO) from 12 Spanish hospitals.

Although the Iberian registry results are promising, there are several pragmatic issues. First, after LAA closure device implantation, it is generally recommended that patients should be maintained on a short period of OAC and/or long-term antiplatelet therapy. Given that the risk of major bleeding with antiplatelet therapy is similar to OAC,6 it remains uncertain how ‘safe’ antiplatelet therapy would be when patients are at a high risk of bleeding. Second, the pathophysiology of stroke in AF may be related to a hypercoagulable state and the presence of various stroke risk factors, such as diabetes and hypertension. Indeed, one of the strongest predictors of ischaemic stroke is complex aortic plaque on the descending aorta,7 and patients with AF also have a systemic prothrombotic state that increases the risk of thromboembolism with the LAA. Third, the efficacy of LAA closure will depend on the proportion of strokes caused by LAA thrombus. Therefore, the concept of LAA closure for stroke prevention should be well established in RCTs before it can be accepted in clinical practice; indeed, the results from non-randomised trial observational data do not provide decisive evidence that LAA closure reduces the risk of stroke in patients with AF, especially if they have no control group and depend on comparisons against event rates from a totally separate population from a different study setting. Finally, it is uncertain how the LAA closure devices would perform when compared with NOACs.

A recent European Heart Rhythm Association (EHRA)/European Association of Percutaneous Cardiovascular Interventions (EAPCI) expert consensus statement on catheter-based LAA occlusion recommends that the device procedure requires training and knowledge based on a structured mentoring process. Furthermore, institutions carrying out the procedure should have proper echocardiography, anaesthesia and preferably surgical support.8

In conclusion, the current indications of LAA closure should be limited to those patients with AF who are at a very high risk of stroke and have clear contra-indications for any form of anticoagulation or a high risk of bleeding on anticoagulation. Overenthusiasm should be tempered.