Warfarin and direct oral anticoagulants (DOACs) are commonly used to prevent thromboembolism in patients with atrial fibrillation (AF) patients who are at elevated risk. Warfarin requires frequent monitoring and has multiple food and drug-drug interactions, while DOACs, though easier to manage, are costly. Both increase bleeding risk. All of these factors lead to underuse of anticoagulation in many patients.
The cause of thromboembolism in atrial fibrillation is partly due to thrombus formation in the left atrial appendage (LAA); therefore, percutaneous left atrial appendage occlusion (pLAAO) devices were developed as an alternative strategy to oral anticoagulation to reduce thromboembolism risk. In March 2015, the U.S. Food and Drug Administration (FDA) approved the first pLAAO device in the United States. Since its approval, the use of pLAAO has grown substantially. In the United States alone, over 100,000 pLAAO procedures were performed between October 2020 and September 2022 [1], reflecting a major shift in the management of patients with atrial fibrillation.
In this paper, we ask: is pLAAO ready for routine use in patients with atrial fibrillation? We provide some limitations in the current evidence for pLAAO.
-
1.
Relatively small sample size of pLAAO trials. The five pLAAO trials, PROTET-AF, PREVAIL, PRAGUE-17, WATCH-TAVR and OPTION, collectively randomized 3465 patients. We highlight this to emphasize that these studies were relatively small in size — the largest, OPTION, included 1600 patients. In contrast, the initial four trials comparing DOACs to warfarin enrolled combined over 71,000 patients. While device trials will never be able to enroll as many patients as drug trial, the modest sample size of the pLAAO trials limits both their statistical power and the generalizability of their findings.
-
2.
No significant reduction in major bleeding. Some of the pLAAO studies excluded “nonprocedural” bleeding. Yet when all bleeding events are included, there was no significant reduction in major bleeding with pLAAO in PROTECT-AF and PREVAIL when combined at 5-year follow up (3.1 % with pLAAO vs 3.5 % with control) [2], PRAGUE-17 (9.0 % with pLAAO vs 10.9 % with DOACs “this endpoint included major bleeding and non-major bleeding”), WATCH-TAVR (19.8 % with pLAAO vs 17.4 % with control) or OPTION (3.9 % with pLAAO vs 5.0 % with control).
-
3.
Concerns about non-inferiority margin and power. The pLAAO and the DOACs trial were designed as non-inferiority trials for the primary efficacy endpoint. The non-inferiority margin is chosen ahead of time to define how much worse the new treatment can be, but still be considered “good enough”. In the case of DOACs trials, the non-inferiority margin varied slightly but was under 1.46 in all of them. This means the upper bound of the one-sided 97.5 % confidence interval for the relative risk of DOACs vs warfarin had to be less than 1.46. This non-inferiority margin was chosen to preserve at least 50 % of the warfarin effect based on a meta-analysis of warfarin vs placebo [3] in which the lower bound of the 95 % confidence interval for warfarin vs placebo was 1.92. In PROTECT-AF, the non-inferiority margin was set at 2.0 and was successfully met. PREVAIL used a slightly narrower non-inferiority margin of 1.75, but non-inferiority was not achieved. PROTECT-AF and PREVAIL served as the basis for FDA approval of pLAAO. PRAGUE-17, WATCH-TAVR, and OPTION used non-inferiority margins of less than 1.5. In OPTION, the largest of these studies, the estimated event rate for the efficacy endpoint was approximately 10 %. However, the observed event rate was only 5.3 %, which reduced the study's statistical power. What's more, in OPTION, the primary endpoint included death, an endpoint shown not to be affected by LAAO [4]. In non-inferiority trials, the inclusion of outcomes in composite endpoints that are unlikely to be affected by either treatment strategy makes non-inferiority easier to reach.
-
4.
Open-label design and potential bias in event adjudication. The pLAAO trials were, understandably, open-label. Conducting a double-blind, sham-controlled trial for pLAAO would be logistically challenging. However, the open-label design may introduce unintentional bias in event reporting or adjudication. For example, in PROTECT-AF and as noted by the FDA reviewers [5], of the 10 events classified as hemorrhagic strokes in the warfarin group, 5 occurred after falls and 4 were associated with subdural hematomas while one had subarachnoid hemorrhage. In contrast, 3 patients in the pLAAO group who had falls resulting in subdural hematomas were not adjudicated as having hemorrhagic strokes.
-
5.
Aspirin is not always safer than DOACs. Patients receive long term treatment with aspirin after pLAAO. If a patient cannot take DOACs due to prior bleeding or a high bleeding risk, aspirin may not necessarily offer a safer alternative, and dedicated trials are needed to evaluate this hypothesis. In the ARTESIA trial (4012 patients with subclinical AF randomized to apixaban vs aspirin), major bleeding occurred in 5.3 % patients in the apixaban group and 3.9 % in the aspirin group; p = 0.04. In the AVERROES trial (5599 patients with AF randomized to apixaban vs aspirin), major bleeding occurred in 1.6 % patients in the apixaban group and 1.4 % in the aspirin group; p = 0.57). In the EINSTEIN CHOICE trial (3365 patients with venous thromboembolism randomized to rivaroxaban vs aspirin), major bleeding was 0.5 % with rivaroxaban 20 mg and 0.3 % with aspirin; p = 0.32. In the NAVIGATE ESUS trial (7213 patients with stroke of undetermined source randomized to rivaroxaban 15 mg daily or aspirin), major bleeding was more frequent with rivaroxaban (1.8 % vs 0.7 %; p < 0.001). We highlight these trials with divergent results to show that aspirin is not consistently significantly safer than DOACs, and when observed, the absolute differences were not large.
-
6.
pLAAO does not save all patients from anti-thrombotic drugs. Elderly patients with AF may need anti-thrombotic drugs for different reasons such as percutaneous coronary interventions and venous thromboembolism. These patients may still have to take a potent anti-thrombotic drug even after pLAAO. Indeed, we emulated a target trial where patients with AF were randomly assigned to receive pLAAO or DOACs in a 1:3 ratio using a nationally representative 15 % sample of Medicare beneficiaries (more detail of pseudo-randomization) [6]. Among 3692 patients who underwent pLAAO, 7.4 % and 55.0 % of them were still taking DOACs and P2Y12 inhibitors at 6–12 months post pLAAO, respectively.
-
7.
Competing risks for stroke may diminish benefit from pLAAO. Most patients who receive pLAAO are elderly with multiple comorbidities [7]. These patients have multiple potential causes of stroke, and the extent to which LAA thrombus contributes to that is not well defined. As competing risks increase, such as carotid disease, valvular heart disease, kidney failure, cancer etc., the likelihood of benefiting from a localized intervention like pLAAO may diminish.
-
8.
Other risks and costs: There are other limitations of pLAAO that should be carefully weighed against long-term oral anticoagulation. Peri-device leak was presented in 20 % of the patients in the OPTION trial but most leaks were small (3 mm or less). While these leaks raise concerns about possible residual thromboembolic risk, it remains unclear whether they have on long-term clinical outcomes. Device-related thrombus occurs in approximately 3–4 % of cases and is associated with increased risk of stroke and systemic embolism. Moreover, the costs of pLAAO should be taken into consideration, not only from the device and implantation procedure itself but also from the need for pre and post-procedural imaging.
Over time, newer generation pLAAO devices and greater operators experience have improved safety and reduced complications rates. This current evidence, however, does not support the enthusiasm for this non-pharmacologic stroke preventive approach. We believe further studies in diverse populations are needed before pLAAO replaces DOACs. This includes a trial enrolling patients of advanced age with multiple comorbidities and contraindications to oral anticoagulants, with one arm receiving neither anticoagulation nor a device.
CRediT authorship contribution statement
Mohammed Ruzieh: Writing – original draft, Conceptualization. Tianze Jiao: Writing – review & editing, Conceptualization. John Mandrola: Writing – review & editing, Conceptualization. Andrew J. Foy: Writing – review & editing, Conceptualization.
Disclosure
None.
Funding
None.
Declaration of competing interest
None.
References
- 1.Vora A.N., Pereira L.A., Du C., et al. Association of Physician Certification and Outcomes among Patients Undergoing Left Atrial Appendage Occlusion. J. Am. Coll. Cardiol. Intv. 2025;18(5):591–602. doi: 10.1016/j.jcin.2024.10.020. [DOI] [PubMed] [Google Scholar]
- 2.Reddy V.Y., Doshi S.K., Kar S., et al. 5-year outcomes after left atrial appendage closure: from the PREVAIL and PROTECT AF trials. J. Am. Coll. Cardiol. 2017;70(24):2964–2975. doi: 10.1016/j.jacc.2017.10.021. [DOI] [PubMed] [Google Scholar]
- 3.Hart R.G., Pearce L.A., Aguilar M.I. Meta-analysis: antithrombotic therapy to prevent stroke in patients who have nonvalvular atrial fibrillation. Ann. Intern. Med. 2007;146(12):857–867. doi: 10.7326/0003-4819-146-12-200706190-00007. [DOI] [PubMed] [Google Scholar]
- 4.Whitlock R.P., Belley-Cote E.P., Paparella D., et al. Left atrial appendage occlusion during cardiac surgery to prevent stroke. N. Engl. J. Med. 2021;384(22):2081–2091. doi: 10.1056/NEJMoa2101897. [DOI] [PubMed] [Google Scholar]
- 5.United States Food and Drug Administration . 2025. FDA executive summary prepared for the October 8, 2014 meeting of the Circulatory System Devices Panel. P130013: Boston Scientific WATCHMAN® Left Atrial Appendage Closure Therapy. Accessed May 19. chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://wayback.archive-it.org/7993/20161024183019/http:/www.fda.gov/downloads/AdvisoryCommittees/CommitteesMeetingMaterials/MedicalDevices/MedicalDevicesAdvisoryCommittee/CirculatorySystemDevicesPanel/UCM417199.pdf. [Google Scholar]
- 6.Zhao J.Z., Ruzieh M., Du F., et al. Association between use of WATCHMAN device and 1-year mortality using high-dimensional propensity scores to reduce confounding. Circ. Cardiovasc. Qual. Outcomes. 2025;18(4) doi: 10.1161/CIRCOUTCOMES.124.011188. [DOI] [PubMed] [Google Scholar]
- 7.Freeman J.V., Varosy P., Price M.J., et al. The NCDR left atrial appendage occlusion registry. J. Am. Coll. Cardiol. 2020;75(13):1503–1518. doi: 10.1016/j.jacc.2019.12.040. [DOI] [PMC free article] [PubMed] [Google Scholar]