Closing the Appendage in 2026: From Rescue Device to Default Option

The Clinical Question

For a patient with atrial fibrillation, when does a one-time mechanical closure of the left atrial appendage genuinely rival the pill they'd otherwise take for life?

In March 2026, two New England Journal of Medicine trials gave opposite answers within weeks of each other. CHAMPION-AF enrolled 3,000 anticoagulation-eligible patients with standard-risk AF and showed that left atrial appendage closure (LAAC) was non-inferior for efficacy and superior for bleeding prevention compared to direct oral anticoagulants. CLOSURE-AF enrolled 888 elderly high-risk patients and failed to show non-inferiority — the device arm saw more cardiovascular death and no bleeding benefit.

Same question. Same device category. Different verdicts.

This isn't a story about conflicting evidence — it's a story about asking the right question in the right population. The trials don't actually disagree; they answer different questions. Let's walk through this step by step.

Why the Appendage Matters

Approximately 90% of thrombi in non-valvular atrial fibrillation form in the left atrial appendage. Oral anticoagulation works — we have decades of trial data proving it — but it costs bleeding risk, adherence burden, polypharmacy complexity, and lifestyle restriction. For some patients, those costs are untenable. For others, they're just undesirable.

Mechanical exclusion of the LAA is the competing hypothesis: if you can't form thrombus in the appendage, you can't embolize it. The hypothesis has now been tested across multiple populations, in several device generations, with several different control strategies and post-implant antithrombotic regimens.

The question has evolved. A decade ago, we asked: Can LAAC substitute for OAC in people who can't tolerate anticoagulation? In 2026, the question is: Should LAAC be offered as a routine alternative to DOACs for AF stroke prevention in anticoagulation-eligible patients?

The answer depends on which trial you read — and more importantly, which patient is in front of you.

State of Play in 2026: Devices, Regimens, and Complications

Current Devices

Watchman FLX (Boston Scientific) is the contemporary standard: five sizes, wider landing-zone tolerance, improved deliverability. It anchored CHAMPION-AF (100% of implants), OPTION-LAAC (100%), and was the majority device in CLOSURE-AF (54.3%).

Amplatzer Amulet (Abbott) uses a dual-seal design with lobe and disc. In the Amulet IDE trial, it demonstrated superior LAA occlusion versus the older Watchman 2.5 (98.9% vs 96.8%, P=0.003) but higher procedural complications (4.5% vs 2.5%, HR 1.86, P=0.02). Amulet was used in PRAGUE-17 (61.3%) and CLOSURE-AF (41.8%).

Earlier-generation Watchman 2.5 and the discontinued WaveCrest have largely exited clinical practice. In CLOSURE-AF, LAmbre accounted for 3.9% of implants — the only non-Watchman-FLX, non-Amulet device documented in the trial.

Post-Implant Antithrombotic Regimens

This is where the evidence gets interesting — because different trials used different strategies, and those differences matter for interpretation.

CHAMPION-AF (Watchman FLX): 85% of implanted patients received DOAC for 45 days, then transitioned to dual antiplatelet therapy (DAPT) at imaging confirmation of adequate seal, then to single antiplatelet therapy (SAPT) at 6 months, then aspirin indefinitely. 12% of patients were started on dual antiplatelet therapy from the beginning, rather than DOAC.

OPTION-LAAC (Watchman FLX): Similar strategy — 45-day NOAC bridge followed by DAPT, then SAPT.

Amulet protocols (Amulet IDE, PRAGUE-17, CLOSURE-AF Amulet patients): Aspirin plus clopidogrel from day one, with no mandatory OAC bridge. DAPT continued for 1–6 months, then transitioned to aspirin monotherapy.

EWOLUTION real-world registry: 83% of patients were managed with antiplatelets only — reflecting the OAC-ineligible population that dominated early device adoption.

The real question isn't whether these regimens work — it's whether the differences between them explain trial outcome heterogeneity. We'll return to this.

Contemporary Complication Rates

Based on CHAMPION-AF and OPTION-LAAC data with Watchman FLX:

• Procedural/device success: 92.5% (1,386 of 1,499 attempted implants in CHAMPION-AF)
• Effective closure (residual leak ≤3 mm) at 4 months: 98.6%
• Pericardial effusion requiring intervention within 30 days: 0.7%
• Device-related thrombus on imaging: 4.8% (63 of 1,320 patients)
• Clinically relevant device-related thrombus: 1.8% (24 patients, including 2 ischemic strokes)

For context, CLOSURE-AF's high-risk cohort saw periprocedural complications in 5.7% within 7 days: 2 deaths, 5 tamponades requiring drainage, 18 bleeds requiring transfusion, and 1 device embolization requiring surgical retrieval.

The procedure has matured substantially since 2009, but it is not risk-free — particularly in elderly patients with advanced comorbidity.

How the Evidence Evolved

Act I: LAAC as a Bleeding-Risk Bailout (2009–2017)

PROTECT-AF (JAMA 2014, N=707, 3.8-year follow-up) was the foundational trial. The primary composite outcome occurred at 2.3 versus 3.8 per 100 patient-years in device versus warfarin arms — the device arm was superior with 96% posterior probability of superiority. Hemorrhagic stroke was reduced by 85% (RR 0.15). Cardiovascular mortality was reduced by 60% (HR 0.40, P=0.005).

But early procedural safety was concerning: pericardial effusion occurred in 4.8%, and procedure-related ischemic stroke in 1.3%.

PREVAIL (JACC 2014, N=407) followed with a more conservative design. The first efficacy endpoint — stroke, systemic embolism, and cardiovascular/unexplained death — narrowly missed the prespecified non-inferiority margin due to wide credible intervals (RR 1.07; 95% CrI 0.57–1.89) rather than a signal of device inferiority. The second efficacy endpoint — ischemic stroke or systemic embolism more than 7 days post-randomization — did meet non-inferiority.

Crucially, 7-day procedural complications fell to 4.2% versus PROTECT-AF's higher rate (P=0.004). The investigators concluded that "safety has matured."

EWOLUTION (Heart Rhythm 2017, real-world registry, N=1,025) extended the evidence to unselected populations: 73% were OAC-ineligible, mean CHA₂DS₂-VASc was 4.5. Procedural success was 98.5%. Annual ischemic stroke rate was 1.1% — an 84% relative risk reduction versus predicted risk. Device-related thrombus was detected in 3.7%.

Takeaway from Act I: If anticoagulation is dangerous or impossible, the device is a defensible second choice — and by 2017, it was safe enough for real-world use outside highly selected trial populations.

Act II: LAAC in the AF Ablation Pathway

OPTION-LAAC (NEJM 2025, N=1,600, 36-month follow-up, Watchman FLX only) asked a different question: What if we implant the device during catheter ablation for AF?

Primary efficacy (all-cause death, stroke, or systemic embolism): 5.3% versus 5.8% — non-inferior (P<0.001).

Primary safety (non-procedural major bleeding or clinically relevant non-major bleeding): 8.5% versus 18.1% — superior (P<0.001).

We already knew from STROKE-AF and CRYSTAL-AF that restoring sinus rhythm doesn't abolish stroke risk — something must keep protecting the brain post-ablation. OPTION-LAAC says that "something" can be mechanical, implanted during the ablation the patient was getting anyway, sparing them lifelong anticoagulation without sacrificing stroke protection.

Takeaway from Act II: If you're already in the left atrium manipulating catheters, closing the door on the way out eliminates bleeding risk without compromising efficacy.

Act III: CHAMPION-AF and CLOSURE-AF — One Question, Two Verdicts

This is where the narrative accelerates.

CHAMPION-AF: The Paradigm Shift

CHAMPION-AF (NEJM March 2026, N=3,000, Watchman FLX versus NOAC, anticoagulation-eligible AF patients) was explicitly designed to test whether LAAC could compete with contemporary anticoagulation in a standard-risk population.

Primary efficacy (cardiovascular death, stroke, or systemic embolism at 3 years): 5.7% in the device arm versus 4.8% in the NOAC arm — HR 1.20 (95% CI 0.87–1.66), P<0.001 for non-inferiority. ✓

Primary safety (non-procedure-related major bleeding): 10.9% in the device arm versus 19.0% in the NOAC arm — HR 0.55 (95% CI 0.45–0.67), P<0.001 for superiority. ✓

Net clinical benefit (composite of efficacy and safety endpoints): 15.1% versus 21.8% — HR 0.66, P<0.001 for non-inferiority. ✓

Stroke rates were numerically higher in the device arm (3.6% vs 2.5%) but did not reach statistical significance — this will bear watching at 5-year follow-up. Notably, 13.7% of patients randomized to the NOAC arm crossed over to receive a device before experiencing a primary endpoint.

This was the first trial to show net clinical benefit from LAAC in a non-high-bleeding-risk AF population.

CLOSURE-AF: The Guardrail

CLOSURE-AF (NEJM 2026, N=888, catheter LAAC versus best-medical-care, high-risk AF) enrolled a much older and sicker cohort: mean age 78, mean CHA₂DS₂-VASc 5.2, mean HAS-BLED 3.0, 24% with stage IV chronic kidney disease, 30% with prior major bleeding.

Primary composite (stroke, systemic embolism, major bleeding, or cardiovascular/unexplained death): 16.8 versus 13.3 per 100 patient-years — failed non-inferiority (P=0.44).

Stroke rates were nearly identical: 2.6 versus 2.7 per 100 patient-years.

Major bleeding was not reduced: 7.4 versus 6.2 per 100 patient-years.

Cardiovascular or unexplained death was numerically higher in the device arm: 9.5 versus 7.7 per 100 patient-years.

The trial was terminated early after enrolling 912 of 1,586 planned patients.

Why the Difference?

Both trials asked fundamentally the same question: Is LAAC non-inferior to DOAC for AF stroke prevention? But they differed in five specific dimensions that together explain the opposite verdicts.

Clinical read: CHAMPION-AF was a fair contest that the device won on its strongest card — bleeding prevention in otherwise-healthy anticoagulation-eligible patients. CLOSURE-AF stacked the deck against the device: older, sicker patients; a weaker post-implant antithrombotic regimen; unfavorable baseline arm imbalance. The device lost on its weakest card — tolerating an invasive procedure in frail elderly patients with advanced CKD.

LAAC has crossed the threshold from "alternative when OAC is contraindicated" to "alternative for standard-risk AF patients who prefer mechanical closure." But it doesn't rescue the octogenarian with stage IV CKD whose fundamental problem is frailty, not the pill.

The Trial Between the Trials

It's worth mentioning PRAGUE-17 (N=402, 2020–2022 extension), the first device-versus-DOAC head-to-head trial in high-risk AF. It met non-inferiority (10.99% vs 13.42% per year, subdistribution HR 0.84, P=0.004 for non-inferiority) and sits between CHAMPION-AF and CLOSURE-AF in both design rigor and sample size. The Abbott Amulet-versus-DOAC pivotal trial (sometimes referenced as CATALYST-AF) has not yet reported.

Where Experts Disagree

There is no meaningful disagreement about what the 2026 data show. The disagreement is about what they mean for patient selection and shared decision-making.

Conservative interpreters argue that CLOSURE-AF's failure in high-risk patients is a warning: elderly patients with CKD and prior bleeding are not good device candidates, and expanding LAAC to standard-risk populations (as CHAMPION-AF suggests) risks procedural harm in patients who would have done fine on a DOAC.

Interventionalists counter that CLOSURE-AF was underpowered, used mixed devices with heterogeneous post-implant regimens, enrolled an unbalanced cohort, and was terminated early. They argue CHAMPION-AF is the cleaner trial — uniform device, uniform protocol, well-balanced arms — and its positive result should guide contemporary practice.

Stroke neurologists flag the 3.6% versus 2.5% stroke signal in CHAMPION-AF and ask whether a 44% relative increase — which corresponds to a 1.1 percentage-point absolute difference over 3 years, meaningful over a lifetime but below statistical significance at current follow-up — is acceptable in exchange for bleeding reduction. They want 5-year and 10-year data before accepting the device as equivalent for stroke prevention.

Health economists note that CHAMPION-AF's crossover rate (13.7% from NOAC to device) suggests patient preference is driving demand, but the cost-effectiveness of a $15,000–$25,000 device plus procedure versus $200/month lifelong DOAC has not been rigorously modeled in standard-risk populations.

What's not in dispute: the device works. The question is for whom — and that requires nuanced patient selection, not blanket recommendations.

Practical Takeaway

In 2026, the evidence supports offering LAAC in three clinical scenarios:

1. OAC Contraindication — Unchanged

This remains the clearest indication. For patients with prior life-threatening bleeding, recurrent falls with intracranial hemorrhage, or absolute refusal of anticoagulation despite counseling, LAAC is a reasonable mechanical alternative. EWOLUTION and real-world registries confirm safety and efficacy in this population.

2. AF Ablation Candidates — OPTION-LAAC Supports Concurrent Implant

If a patient is undergoing catheter ablation for symptomatic AF, concurrent LAAC eliminates lifelong anticoagulation without compromising stroke protection and significantly reduces bleeding risk. This is now evidence-based practice, not off-label extension.

3. Standard-Risk Anticoagulation-Eligible AF Who Prefer Mechanical Closure — CHAMPION-AF Supports This

For patients with standard-risk AF (CHA₂DS₂-VASc ≥2 in men, ≥3 in women), normal renal function, no advanced frailty, who express informed preference for a one-time procedure over lifelong anticoagulation, LAAC with Watchman FLX is a non-inferior alternative that reduces bleeding risk. This should be offered in experienced centers with robust post-implant imaging and antithrombotic protocols.

Temper Expectations In:

Very elderly patients with advanced CKD and prior major bleeding: CLOSURE-AF suggests the device does not rescue this population. The procedural risk and lack of bleeding benefit make continued anticoagulation — or even no anticoagulation with acceptance of stroke risk — potentially preferable.

Patients where 5-year stroke protection absolutely cannot be compromised: CHAMPION-AF's nominal stroke signal (3.6% vs 2.5%) deserves the 5-year readout before dismissal. If a patient's primary concern is stroke prevention and bleeding risk is tolerable, a DOAC remains the evidence-based choice.

Still Unresolved:

• Device-related thrombus rate of ~4.8% with 1.8% clinically relevant events — what is the optimal surveillance and treatment strategy?
• Optimal post-implant antithrombotic regimen: DOAC bridge versus DAPT-only — CHAMPION-AF and Amulet protocols differ, and we lack head-to-head data
• Long-term durability: Does the device remain effective at 10 years? Does late DRT increase over time?
• Cost-effectiveness in low-risk mainstream AF populations
• Operator and center volume thresholds for maintaining contemporary safety standards

Bottom Line

The question has evolved from "Can LAAC substitute for OAC in people who can't take it?" to "Should LAAC be offered as a routine alternative to DOACs for AF stroke prevention?"

In 2026, the answer is a qualified yes — for standard-risk AF in experienced centers, with Watchman FLX, using a DOAC bridge protocol.

It is not yes for the octogenarian with stage IV CKD and a prior GI bleed.

It is emphatically yes for patients undergoing AF ablation.

CHAMPION-AF's verdict is the paradigm shift. CLOSURE-AF's verdict is the guardrail. Together, they tell us who the device is for — and who it isn't.

The 5-year CHAMPION-AF readout, expected in 2028, will determine whether the stroke signal is noise or a real late-effect — and whether 2026's qualified yes becomes 2028's unqualified one.

Key Clinical Principle: LAAC competes with DOACs on bleeding risk, not stroke prevention. In populations where bleeding drives morbidity and procedural risk is low, the device wins. In populations where frailty drives mortality and procedural risk is high, the device loses. Patient selection is everything.

Cross-referenced trials: PROTECT-AF · PREVAIL · EWOLUTION · Amulet IDE · PRAGUE-17 · OPTION-LAAC · CLOSURE-AF · CHAMPION-AF · STROKE-AF · CRYSTAL-AF