← Back

CHAMPION-AF

Left Atrial Appendage Closure or Anticoagulation for Atrial Fibrillation

Share Share Copied! Compare

Year of Publication: 2026

Authors: Doshi SK, Kar S, Nair DG, ..., Ellenbogen KA; for the CHAMPION-AF Investigators

Journal: The New England Journal of Medicine

Citation: Doshi SK, et al. N Engl J Med. 2026. DOI: 10.1056/NEJMoa2517213.

Link: https://doi.org/10.1056/NEJMoa2517213

PDF: https://www.nejm.org/doi/pdf/10.1056/NEJMoa2517213

Clinical Question

Can left atrial appendage closure replace NOACs for stroke prevention in anticoagulation-eligible AF patients?

Bottom Line

In 3,000 patients with atrial fibrillation eligible for anticoagulation, left atrial appendage closure with the Watchman FLX device was noninferior to NOACs for preventing cardiovascular death, stroke, or systemic embolism at 3 years (5.7% vs 4.8%; HR 1.20; P<0.001 for noninferiority). The device arm had significantly less non-procedure-related bleeding (10.9% vs 19.0%; HR 0.55; P<0.001), establishing LAAC as a viable alternative even in anticoagulation-eligible patients.

        Major Points
        First trial to demonstrate noninferiority of LAAC vs NOACs in anticoagulation-eligible AF patients — prior trials (PROTECT AF, PREVAIL) compared against warfarin only
Primary efficacy (CV death, stroke, SE) at 3 years: 81 events (5.7%) device vs 65 events (4.8%) NOAC; difference 0.9 pp (95% CI −0.8 to 2.6); HR 1.20 (0.87–1.66); P<0.001 for noninferiority
Primary safety (non-procedure-related bleeding): 154 events (10.9%) device vs 260 events (19.0%) NOAC; HR 0.55 (0.45–0.67); P<0.001 for superiority
Net clinical benefit (CV death, stroke, SE, or non-procedure-related bleeding): 15.1% device vs 21.8% NOAC; HR 0.66 (0.56–0.79); P<0.001 for noninferiority
Device successfully deployed in 1,386 of 1,499 patients (92.5%); effective closure (residual leak ≤3 mm) in 98.6% at 4 months
Device-related thrombus in 63/1,320 patients (4.8%) assessed by imaging; 24 (1.8%) were clinically relevant, 2 had a stroke
206 patients (13.7%) in the anticoagulation group crossed over to device before a primary endpoint event
Annualized ischemic stroke/SE rate ~1.1%/year with device vs ~0.8%/year with NOAC — a 0.3% annual excess that needs 5-year assessment

      

Design

Study Type: Randomized, open-label, noninferiority trial

Randomization: 1

Blinding: Open-label with blinded endpoint adjudication

Enrollment Period: 2019-2023

Follow-up Duration: 3 years (5-year ongoing)

Centers: 141

Countries: 16

Sample Size: 3000

Power Calculation: 90% power to show noninferiority with 4.8 percentage point margin, assuming 12% event rate at 3 years, one-sided alpha 0.025, estimated attrition 12.5%

Analysis: Intention-to-treat

Registration: NCT04394546

Inclusion Criteria

Age ≥18 years
Non-valvular atrial fibrillation (paroxysmal, persistent, or permanent)
CHA2DS2-VASc score ≥2 for men or ≥3 for women
Suitable candidates for long-term oral anticoagulation

Exclusion Criteria

Myocardial infarction, stroke, or TIA within 30 days before enrollment
Major bleeding event (ISTH criteria) within prior 30 days
Contraindication to oral anticoagulation
Mechanical heart valve or moderate-to-severe mitral stenosis
Left atrial appendage thrombus
Prior left atrial appendage closure or surgical ligation

Baseline Characteristics

Characteristic	Control	Active
N	1501	1499
Age (mean±SD)	71.8±7.5	71.6±7.5
Female	31.5%	32.4%
White	85.0%	85.1%
CHA2DS2-VASc (mean±SD)	3.5±1.3	3.5±1.2
HAS-BLED (mean±SD)	1.3±0.8	1.3±0.8
Previous ischemic stroke	7.6%	7.9%
Previous hemorrhagic stroke	0.7%	0.3%
Paroxysmal AF	68.5%	69.3%
Persistent AF	25.4%	23.9%
Permanent AF	6.1%	6.8%
Previous AF ablation	46.9%	48.7%

Arms

Field	Device: Watchman FLX	Control
Intervention	Percutaneous left atrial appendage closure with the Watchman FLX device (Boston Scientific). Post-implantation: NOAC plus aspirin, NOAC monotherapy, or dual antiplatelet therapy for 3 months, then aspirin or P2Y12 inhibitor monotherapy recommended.	Physician-choice non-vitamin K antagonist oral anticoagulant (apixaban, rivaroxaban, edoxaban, or dabigatran) at guideline-recommended doses for stroke prevention in AF.
Duration	3 years follow-up	3 years follow-up

Outcomes

Outcome	Type	Control	Intervention	HR / OR / RR	P-value
Composite of death from cardiovascular causes, stroke, or systemic embolism at 3 years	Primary	4.8% (65 events)	5.7% (81 events)	1.2	<0.001 for noninferiority
Non-procedure-related bleeding (primary safety endpoint — superiority)	Secondary	19.0% (260 events)	10.9% (154 events)	0.55	<0.001
Procedure-related and non-procedure-related ISTH major bleeding (secondary safety)	Secondary	6.4% (87 events)	5.9% (83 events)	0.92	<0.001 for noninferiority
Net clinical benefit (CV death, stroke, SE, or non-procedure-related bleeding)	Secondary	21.8% (300 events)	15.1% (215 events)	0.66	<0.001 for noninferiority
Death from cardiovascular causes	Secondary	2.7% (36 events)	2.7% (38 events)	1.01
Stroke (ischemic or hemorrhagic)	Secondary	2.5% (33 events)	3.6% (50 events)	1.46
Systemic embolism	Secondary	0.1% (2 events)	0% (0 events)
ISTH major bleeding (non-procedure-related)	Secondary	6.4% (87 events)	5.1% (71 events)	0.78
Clinically relevant nonmajor bleeding (non-procedure-related)	Secondary	14.2% (193 events)	7.0% (99 events)	0.48
All-cause death	Secondary	4.9% (67 events)	5.0% (71 events)
Ischemic stroke or systemic embolism	Secondary	2.2% (29 events)	3.2% (45 events)
Hemorrhagic stroke	Secondary	0.4% (5 events)	0.4% (5 events)
All bleeding events (3-year cumulative)	Secondary	19.0%	12.8%
Device-related thrombus	Adverse	N/A	4.8% (63/1,320 assessed)
Clinically relevant device thrombus	Adverse	N/A	1.8% (24 patients; 2 had stroke)
Pericardial effusion requiring intervention (≤30 days)	Adverse	N/A	0.7% (10 patients)

Subgroup Analysis

Results for the primary end points appeared to be consistent across most subgroups including age, sex, CHA2DS2-VASc score, type of AF, and prior stroke/TIA.

Criticisms

Open-label design — despite blinded adjudication, knowledge of treatment assignment may influence patient behavior, medication adherence, and clinical decision-making
Numerically higher stroke rate with device (3.6% vs 2.5%; HR 1.46) raises concern about a small efficacy trade-off; 5-year data needed to determine if this gap persists or widens
Noninferiority margin of 4.8 percentage points on absolute scale is wide given low event rates — allows a potentially higher relative risk of events
13.7% crossover from anticoagulation to device group dilutes the between-group difference in an ITT analysis
Only Watchman FLX tested — results cannot be extrapolated to other LAAC devices or techniques
CHA2DS2-VASc ≥5 subgroup underrepresented (18.4%) — highest-risk patients who may have most to lose from any efficacy trade-off are less well studied
Post-implantation antithrombotic protocol varied (NOAC+aspirin, NOAC alone, or DAPT for 3 months) — no standardized regimen
Device-related thrombus in 4.8% of assessed patients, with CT imaging detecting more thrombus than TEE — clinical significance of CT-detected thrombus remains debated

Funding

Boston Scientific

Based on: CHAMPION-AF (The New England Journal of Medicine, 2026)

Authors: Doshi SK, Kar S, Nair DG, ..., Ellenbogen KA; for the CHAMPION-AF Investigators

Citation: Doshi SK, et al. N Engl J Med. 2026. DOI: 10.1056/NEJMoa2517213.

Content summarized and formatted by NeuroTrials.ai.

CHAMPION-AF

(2026)

Objective

To determine whether percutaneous left atrial appendage closure with Watchman FLX is noninferior to NOAC therapy for stroke prevention in AF patients who are eligible for long-term anticoagulation.

Study Summary

• In 3,000 patients with atrial fibrillation eligible for anticoagulation, left atrial appendage closure with the Watchman FLX device was noninferior to NOACs for preventing cardiovascular death, stroke, or systemic embolism at 3 years (5.7% vs 4.8%; HR 1.20; P<0.001 for noninferiority).
• The device arm had significantly less non-procedure-related bleeding (10.9% vs 19.0%; HR 0.55; P<0.001), establishing LAAC as a viable alternative even in anticoagulation-eligible patients.

Intervention

Watchman FLX left atrial appendage closure device with post-implantation NOAC+aspirin, NOAC monotherapy, or DAPT for 3 months then aspirin/P2Y12 monotherapy, vs physician-choice NOAC (apixaban, rivaroxaban, edoxaban, or dabigatran).

Inclusion Criteria

Non-valvular AF, CHA2DS2-VASc ≥2 (men) or ≥3 (women), eligible for long-term OAC, no MI/stroke/TIA or major bleed within 30 days

Study Design

Arms: Array

Patients per Arm: Device: 1,499; Anticoagulation: 1,501

Outcome

Primary efficacy (CV death/stroke/SE): 5.7% vs 4.8% (HR 1.20; P<0.001 NI). Primary safety (non-procedural bleeding): 10.9% vs 19.0% (HR 0.55; P<0.001). Net clinical benefit: 15.1% vs 21.8% (HR 0.66; P<0.001 NI).

Bottom Line

Major Points

First trial to demonstrate noninferiority of LAAC vs NOACs in anticoagulation-eligible AF patients — prior trials (PROTECT AF, PREVAIL) compared against warfarin only
Primary efficacy (CV death, stroke, SE) at 3 years: 81 events (5.7%) device vs 65 events (4.8%) NOAC; difference 0.9 pp (95% CI −0.8 to 2.6); HR 1.20 (0.87–1.66); P<0.001 for noninferiority
Primary safety (non-procedure-related bleeding): 154 events (10.9%) device vs 260 events (19.0%) NOAC; HR 0.55 (0.45–0.67); P<0.001 for superiority
Net clinical benefit (CV death, stroke, SE, or non-procedure-related bleeding): 15.1% device vs 21.8% NOAC; HR 0.66 (0.56–0.79); P<0.001 for noninferiority
Device successfully deployed in 1,386 of 1,499 patients (92.5%); effective closure (residual leak ≤3 mm) in 98.6% at 4 months
Device-related thrombus in 63/1,320 patients (4.8%) assessed by imaging; 24 (1.8%) were clinically relevant, 2 had a stroke
206 patients (13.7%) in the anticoagulation group crossed over to device before a primary endpoint event
Annualized ischemic stroke/SE rate ~1.1%/year with device vs ~0.8%/year with NOAC — a 0.3% annual excess that needs 5-year assessment

Study Design

Study Type: Randomized, open-label, noninferiority trial
Randomization: Yes
Blinding: Open-label with blinded endpoint adjudication
Sample Size: 3000
Follow-up: 3 years (5-year ongoing)
Centers: 141

Primary Outcome

Definition: Composite of death from cardiovascular causes, stroke, or systemic embolism at 3 years

Control	Intervention	HR/OR	P-value
4.8% (65 events)	5.7% (81 events)	1.2 (0.87-1.66)	<0.001 for noninferiority

Limitations & Criticisms

Open-label design — despite blinded adjudication, knowledge of treatment assignment may influence patient behavior, medication adherence, and clinical decision-making
Numerically higher stroke rate with device (3.6% vs 2.5%; HR 1.46) raises concern about a small efficacy trade-off; 5-year data needed to determine if this gap persists or widens
Noninferiority margin of 4.8 percentage points on absolute scale is wide given low event rates — allows a potentially higher relative risk of events
13.7% crossover from anticoagulation to device group dilutes the between-group difference in an ITT analysis
Only Watchman FLX tested — results cannot be extrapolated to other LAAC devices or techniques
CHA2DS2-VASc ≥5 subgroup underrepresented (18.4%) — highest-risk patients who may have most to lose from any efficacy trade-off are less well studied
Post-implantation antithrombotic protocol varied (NOAC+aspirin, NOAC alone, or DAPT for 3 months) — no standardized regimen
Device-related thrombus in 4.8% of assessed patients, with CT imaging detecting more thrombus than TEE — clinical significance of CT-detected thrombus remains debated

Citation

Doshi SK, et al. N Engl J Med. 2026. DOI: 10.1056/NEJMoa2517213.

View Original Publication →

CHAMPION-AF

Left Atrial Appendage Closure or Anticoagulation for Atrial Fibrillation

Clinical Question

Bottom Line

Major Points

Design

Inclusion Criteria

Exclusion Criteria

Baseline Characteristics

Arms

Outcomes

Subgroup Analysis

Criticisms

Funding

Ask about CHAMPION-AF