Left Atrial Appendage Closure or Medical Therapy in Atrial Fibrillation (CLOSURE-AF)

Authors, Journal, Affiliations, Type, DOI

• Landmesser U, Skurk C, Kirchhof P, Lewalter T, Hartung J, et al.; CLOSURE-AF Trial Investigators
• N Engl J Med 2026;394:1270–1280 (CME article)
• Coordinated by Charité University Medicine Berlin; 42 German centres; sponsored by German Center for Cardiovascular Research (DZHK)
• Pragmatic, prospective, open-label, multicentre, randomised, controlled trial with blinded outcome assessment (PROBE design); parallel two-group
• DOI: 10.1056/NEJMoa2513310

Overview

CLOSURE-AF enrolled 888 patients with atrial fibrillation and simultaneously high stroke risk (CHA₂DS₂-VASc ≥2, mean 5.2) and high bleeding risk (HAS-BLED ≥3 or prior major bleeding, CKD stage 4, or recurrent bleeding — mean HAS-BLED 3.0). Left atrial appendage closure (LAAO) was compared with physician-directed best medical care (85.1% DOAC). After a median 3 years of follow-up, LAAO failed to demonstrate noninferiority for the composite of stroke, systemic embolism, major bleeding, or cardiovascular/unexplained death (difference in restricted mean survival time −0.36 years; P=0.44 for NI; device group numerically worse). The trial delivers a definitive negative verdict on LAAO for the highest-risk AF population — the exact group for whom it has been most widely advocated.

Keywords

Left atrial appendage closure, atrial fibrillation, stroke prevention, anticoagulation, DOAC, Watchman, Amulet, noninferiority, high-risk, bleeding, CHA₂DS₂-VASc, HAS-BLED

Key Takeaways

Background

• Percutaneous LAAO was developed as an alternative to long-term anticoagulation for stroke prevention in AF patients with contraindications to or inability to tolerate anticoagulation.
• ~90% of cardiac thrombi form in the left atrial appendage; occlusion of the LAA should theoretically reduce thromboembolism.
• Previous pivotal trials (PROTECT AF, PREVAIL) compared Watchman vs. warfarin — not vs. DOACs — in lower-to-moderate risk patients, limiting contemporary relevance.
• PRAGUE-17 (Osmancik 2020/2022) showed LAAO non-inferior to NOACs at 4 years in patients at moderate risk; OPTION (2025 NEJM) showed NI in patients with AF undergoing catheter ablation (lower bleeding risk).
• No large RCT had tested LAAO vs. contemporary DOACs in the highest-risk AF population (high stroke AND high bleeding risk simultaneously) — the group most likely to receive the device in clinical practice.

Trial Design

• Population: AF with CHA₂DS₂-VASc ≥2 AND (HAS-BLED ≥3 or prior BARC ≥3a/3c bleeding, CKD stage IV [eGFR 15–29], or recurrent bleeding precluding anticoagulation).
• Device group (n=446): Watchman, Watchman FLX (Boston Scientific), Amplatzer Cardiac Plug, Amulet (St. Jude Medical), or LAmbre (LifeTech) at selected sites as bailout; 94.4% underwent assigned treatment; 79.6% received DAPT post-implant.
• Medical-therapy group (n=442): Physician-directed best medical care — DOAC as default; 85.1% received a DOAC.
• Primary endpoint: Composite of stroke (ischemic or haemorrhagic), systemic embolism, major bleeding (BARC ≥3), or cardiovascular/unexplained death — all independently adjudicated.
• Non-inferiority margin: HR 1.3 (tested with Cox model; restricted mean survival time as primary metric due to proportional-hazards assumption violations).
• Median follow-up: 3.0 years (IQR 1.7–4.7); maximum 6.7 years.

Patient Characteristics

• Mean age 77.9±7.1 years; 38.6% women — an elderly, comorbid population.
• Mean CHA₂DS₂-VASc 5.2±1.5; mean HAS-BLED 3.0±0.9.
• Well-balanced baseline characteristics between groups.
• Implantation success: 98.3% (414/421 attempted); technical success 97.6%; procedural success 96.4%.

Primary Results — Noninferiority NOT Demonstrated

• Primary endpoint events: Device 155/446 (16.8/100 pt-yr) vs Medical 127/442 (13.3/100 pt-yr).
• Difference in restricted mean survival time: −0.36 years (95% CI −0.70 to −0.01).
• P = 0.44 for noninferiority — NI not demonstrated; device group numerically worse.
• Consistent results in per-protocol analysis (RMST difference −0.40 years; 95% CI −0.76 to −0.04).
• All prespecified subgroup analyses consistent with primary result.

Individual Endpoint Components (Secondary Endpoints)

• Stroke: Equal — 27 vs 27 patients (2.6 vs 2.7/100 pt-yr); RMST difference 0.01 years (95% CI −0.17 to 0.20). LAAO provides stroke reduction consistent with its mechanism, but periprocedural risk offsets it.
• Systemic embolism: 3 vs 1 patients — numerically higher in device group.
• Major bleeding (BARC ≥3): 70 vs 61 patients (7.4 vs 6.2/100 pt-yr); RMST difference −0.12 years (95% CI −0.40 to 0.15). No bleeding reduction with LAAO; 18 of device-group bleeds were procedure-related; 52 were not.
• Cardiovascular/unexplained death: 99 vs 81 patients (9.5 vs 7.7/100 pt-yr); RMST difference −0.19 years (95% CI −0.50 to 0.11) — numerically worse with LAAO.
• All-cause death: 155 vs 141 patients; RMST difference −0.13 years (95% CI −0.48 to 0.21) — similar.
• Serious adverse events: 82.5% (device) vs 77.4% (medical) — consistently higher in device group.

Procedural Data

• Periprocedural complications in first 7 days: 5.7% of device implantations.
• Major bleeding within 7 days (procedural): 18 patients (the dominant periprocedural complication).
• Two periprocedural deaths; one device embolisation requiring surgical retrieval.
• Device leaks: 20 patients; >5 mm leak in 3 patients.
• Highest major bleeding incidence was in the first 6 months after LAAO — driven by DAPT and non-procedural bleeding.

Key Mechanistic Insight

• The early periprocedural phase + DAPT-driven bleeding post-LAAO offsets any stroke prevention benefit in high-risk elderly patients. The expected "delayed benefit" of LAAO (when DAPT is completed and antiplatelet therapy discontinued) does not compensate for early harm in this population.
• The 3-year median follow-up may be insufficient to detect late benefits, but the observed trajectory is unfavourable throughout.
• The medical-therapy event rate was higher than anticipated (13.3/100 pt-yr), reflecting genuine high-risk population; LAAO performed worse.

Contextual Comparison with Other LAAO Trials

• PROTECT AF / PREVAIL: Watchman vs. warfarin in lower-risk patients → noninferiority demonstrated, but DOAC comparator was not used.
• PRAGUE-17 (Osmancik 2022): LAAO vs. NOACs in moderate-risk AF (4-year follow-up) → NI demonstrated; lower-risk population (mean CHA₂DS₂-VASc ~4.7, HAS-BLED ~3.0 but fewer prior bleeds).
• OPTION (NEJM 2025): LAAO after AF ablation vs. OAC → NI demonstrated; critically, lower-risk population (patients undergoing ablation, not those who have failed/cannot take anticoagulation).
• LAAOS III (NEJM 2021): Surgical LAA closure during open-heart surgery → reduced ischaemic stroke (HR 0.67) — confirms the anatomical concept, but this is surgical occlusion in a completely different procedural context.
• CLOSURE-AF is the only trial to test LAAO specifically against contemporary DOACs in the highest-risk group (high stroke + high bleeding simultaneously) — and it failed.

Limitations of the Document

• Sample size reduced twice (from 1,586 to 912) during interim analyses based on published data from PRAGUE-17; trial may have been underpowered for individual endpoint components.
• Open-label design; complete blinding of procedural/device complications to clinical events committee was not possible.
• Conducted exclusively in Germany — predominantly White population; generalisability to other ethnicities and healthcare systems limited.
• Medical-therapy group: physician decided best medical care; DOAC use was high (85.1%) but not universal — reflects real-world practice.
• More patients lost to follow-up in medical-therapy group (5.2% vs 2.2%); more withdrawals in device group (12.6% vs 10.9%).
• Post-implant antithrombotic therapy was individualised — DAPT protocol variation may influence bleeding outcomes; ongoing trials (ADALA, LAAO +anticoagulation combinations) are addressing periprocedural antithrombotic strategy.
• Results apply to Watchman/FLX and Amulet devices — may evolve with device improvements and coatings reducing antithrombotic requirements.

Key Concepts Mentioned

• concepts/Left-Atrial-Appendage-Closure — percutaneous LAAO strategy; NI failure in highest-risk AF population
• concepts/Anticoagulation-in-AF — DOAC vs LAAO comparison

Key Entities Mentioned

• entities/Atrial-Fibrillation — stroke prevention strategy; pLAAO section updated; major contradiction with Class IIa recommendation

Wiki Pages Updated

• wiki/sources/laao-closureaf-nejm-2026.md — created
• wiki/entities/Atrial-Fibrillation.md — LAAO section in stroke prevention updated with CLOSURE-AF; new contradiction added
• wiki/sourceindex.md — new entry added
• wiki/wikiindex.md — Atrial-Fibrillation description updated