Anticoagulation with Edoxaban in Patients with Atrial High-Rate Episodes (NOAH-AFNET 6)

Authors, Journal, Affiliations, Type, DOI

• Authors: Kirchhof P, Toennis T, Goette A, Camm AJ, Diener HC, Becher N, Bertaglia E, Blomstrom Lundqvist C, Borlich M, Brandes A, Cabanelas N, Calvert M, Chlouverakis G, Dan G-A, de Groot JR, Dichtl W, Kravchuk B, Lubiński A, Marijon E, Merkely B, Mont L, Ozga A-K, Rajappan K, Sarkozy A, Scherr D, Sznajder R, Velchev V, Wichterle D, Sehner S, Simantirakis E, Lip GYH, Vardas P, Schotten U, Zapf A; for the NOAH-AFNET 6 Investigators
• Journal: New England Journal of Medicine (NEJM)
• Affiliations: 206 sites, 18 European countries; coordinated by University Heart and Vascular Center Hamburg and Atrial Fibrillation Network (AFNET), Germany
• Type: Investigator-initiated, double-blind, double-dummy, randomized controlled trial (RCT)
• DOI: https://doi.org/10.1056/NEJMoa2303062
• Funding: German Center for Cardiovascular Research (DZHK); drug supply and additional funding from Daiichi Sankyo Europe (manufacturer of edoxaban; no role in design, analysis, or manuscript)

Overview

NOAH-AFNET 6 was a landmark European RCT testing whether oral anticoagulation with edoxaban reduces cardiovascular events in patients aged ≥65 with device-detected atrial high-rate episodes (AHREs) and at least one stroke risk factor, but without ECG-documented atrial fibrillation. Among 2,536 patients (median AHRE duration 2.8 hours; median CHA₂DS₂-VASc 4), edoxaban did not significantly reduce the primary composite of cardiovascular death, stroke, or systemic embolism (HR 0.81; P=0.15), but significantly increased the safety composite of death or major bleeding (HR 1.31; P=0.03). The trial was stopped early at a median 21 months for futility and safety concerns. The unexpectedly low placebo-group stroke rate (~1%/yr — far below ECG-confirmed AF comparators) is the trial's central mechanistic finding, suggesting AHREs carry fundamentally lower embolic risk due to low arrhythmia burden.

Keywords

Atrial high-rate episodes, AHRE, subclinical atrial fibrillation, edoxaban, anticoagulation, stroke prevention, implanted cardiac devices, non-vitamin K antagonist oral anticoagulants, NOAH-AFNET 6

Key Takeaways

Background and Trial Rationale

• Device-detected AHREs are detected by pacemakers, ICDs, CRT devices, and implanted loop recorders; they resemble AF in electrogram morphology but are rare (median 2.8 episodes) and brief (median duration 2.8 hours).
• An estimated 10–30% of patients with implanted devices have AHREs lasting >5–6 minutes.
• AHREs are often undiagnosed in patients without long-term rhythm monitoring; their occurrence does not constitute ECG-documented AF.
• Oral anticoagulation (VKA or NOAC) reduces ischemic stroke risk in ECG-documented AF; NOACs are preferred due to lower bleeding risk. However, OAC had not been proven effective without ECG-confirmed AF (e.g., ESUS trials, HF-in-sinus-rhythm trials were all neutral).
• Some clinicians initiate OAC in AHRE patients — especially those with stroke risk factors or AHREs lasting >24 hours — based on electrogram similarity to AF.

Trial Design and Population

• Design: Investigator-initiated, event-driven, double-blind, double-dummy, 1:1 randomized trial; 206 sites; 18 European countries; June 2016–September 2022.
• Eligibility: Age ≥65; CIED-detected AHREs (atrial rate ≥170 bpm; duration ≥6 min; ≥2 months after device implantation); ≥1 stroke risk factor (HF, hypertension, DM, prior stroke/TIA, vascular disease, or age ≥75); no ECG-documented AF.
• Exclusion: ECG-documented AF; ACS/PCI/CABG within 30 days; life expectancy <12 months; contraindication to OAC or edoxaban; dual antiplatelet therapy; other indication for OAC.
• Randomization: 1:1, variable block sizes, stratified by indication for acetylsalicylic acid.
• Intervention: Edoxaban 60 mg OD (standard AF dose) vs. matching placebo; dose reduced to 30 mg OD for body weight ≤60 kg, CrCl 15–50 mL/min, or strong P-gp inhibitor use. ASA 100 mg was given in the placebo arm only for established vascular indications.
• Follow-up: Every 6 months; AHRE data retrieved from devices; ECG at each visit; planned enrollment: 2,538 patients with ~220 primary outcome events.

Patient Characteristics

• Modified intention-to-treat population: 2,536 patients (1,270 edoxaban; 1,266 placebo).
• Mean age 78 years; 37.4% women; predominantly White European.
• Median AHREs: 2.8 hours duration; >200 bpm atrial rate; 2.8 episodes per group.
• Median CHA₂DS₂-VASc score: 4 (all patients CHA₂DS₂-VASc ≥2 by design).
• 28.7% of edoxaban patients received reduced dose (30 mg) at baseline.
• 53.9% of placebo patients received ASA 100 mg.
• Discontinuation occurred after median ~17 months in both groups.
• 134 patients withdrew consent in each group.
• ECG-diagnosed AF developed in 462/2,536 patients (18.2% total; 8.7%/yr).

Primary Outcomes — Efficacy

• Primary efficacy composite (CV death + stroke + systemic embolism):
  • Edoxaban: 83 patients (3.2%/yr) vs placebo: 101 patients (4.0%/yr)
  • HR 0.81 (95% CI 0.60–1.08; P=0.15) — NOT significant
  • Trial was stopped at 184/220 planned events: underpowered
• The trial was terminated early at median 21-month follow-up based on DSMB recommendations for safety concerns and informal futility assessment.

Primary Outcomes — Safety

• Safety composite (death from any cause + major bleeding):
  • Edoxaban: 149 patients (5.9%/yr) vs placebo: 114 patients (4.5%/yr)
  • HR 1.31 (95% CI 1.02–1.67; P=0.03) — SIGNIFICANTLY WORSE with edoxaban
• Major bleeding (ISTH criteria): 53 edoxaban vs 25 placebo — HR 2.10 (95% CI 1.30–3.38)
• Death from any cause: 111 edoxaban vs 94 placebo — HR 1.16 (95% CI 0.88–1.53; NS)

Secondary Outcomes — Individual Components

• Stroke: 22 edoxaban (0.9%/yr) vs 27 placebo (1.1%/yr) — numerically lower, NS
• Systemic embolism: 14 edoxaban (0.5%/yr) vs 28 placebo (1.1%/yr) — numerically lower, NS
• Stroke + SE composite: 25 edoxaban (1.0%/yr) vs 38 placebo (1.5%/yr); HR 0.65 (95% CI 0.39–1.07) — numerically lower, NS (secondary outcome)
• Cardiovascular death: 52 edoxaban (2.0%/yr) vs 57 placebo (2.2%/yr) — HR NS
• CRITICAL: Stroke incidence in placebo group (~1%/yr) was substantially lower than expected and lower than:
  • Swedish AF registry (CHA₂DS₂-VASc 4, no OAC)
  • AVERROES trial (AF patients on aspirin)
  • ELDERCARE-AF trial (AF patients on no anticoagulation)
  • ENGAGE AF-TIMI 48 trial (AF patients on edoxaban)

Discussion — Key Mechanistic Insights

• The paucity and brevity of AHREs (low arrhythmia burden) is the main distinguishing feature from ECG-documented AF.
• Paroxysmal AF carries lower stroke risk than persistent/permanent AF; reducing arrhythmia burden (e.g., early rhythm control in EAST-AFNET 4) reduces CV events. AHREs represent an even lower arrhythmia burden than paroxysmal AF.
• The unexpectedly low baseline stroke rate means the absolute benefit-harm ratio for OAC is unfavourable — the bleeding risk (HR 2.10 for major bleeding) cannot be offset by a small absolute stroke reduction.
• These findings suggest AHREs without ECG-confirmed AF are better managed without anticoagulation.
• The implication for consumer device-detected brief AF episodes (smartwatches, patches) is also relevant — episodes detected by consumer devices are typically even shorter than CIED-detected AHREs.

Statistical Considerations

• Event-driven design: planned 220 primary efficacy events for 80% power (assumed HR 0.68 with edoxaban); trial stopped with 184 events — underpowered.
• Cause-specific Cox proportional hazards model with frailty term for site; Aalen–Johansen estimator for cumulative incidence curves.
• Data from Ukrainian sites censored February 24, 2022 (start of Russian invasion).
• Sensitivity analyses (per-protocol, safety population, AHRE verification by independent lab, consent withdrawal exclusion) consistent with primary results.
• No imputation for primary/safety outcomes; worst-case imputation for cause of death.

Limitations of the Document

• Underpowered: Trial stopped at 184/220 planned events (21 months median, not full follow-up); cannot definitively exclude a small beneficial effect on stroke.
• Premature termination: Safety concerns and informal futility assessment led to early stopping — formal alpha-spending applied only at termination.
• Single NOAC: Results with edoxaban may not generalize to other NOACs (apixaban, rivaroxaban, dabigatran); the ongoing ARTESIA trial tested apixaban vs aspirin in SCAF ≤24h (different population and comparator).
• Population: Predominantly White European patients; results may differ by race/ethnicity.
• Device eligibility: AHRE had to be ≥2 months post-implantation; applies only to CIED-detected AHREs, not consumer-detected.
• No data on AHRE duration subgroups: Cannot determine if very long AHREs (e.g., >24 h) might benefit; enrollment criteria had no upper duration limit.
• Stroke risk estimation: CHA₂DS₂-VASc overestimates stroke risk in AHRE patients; improved risk stratification tools needed.

Key Concepts Mentioned

• concepts/Subclinical-AF — AHREs defined as CIED-detected atrial tachyarrhythmias; edoxaban neutral for primary composite; low arrhythmia burden explains low baseline stroke rate
• concepts/CHA2DS2-VA — median CHA₂DS₂-VASc 4 in trial population; CHA₂DS₂-VASc overestimates embolic risk in AHRE patients without ECG-confirmed AF

Key Entities Mentioned

• entities/Atrial-Fibrillation — ECG-documented AF was an exclusion criterion; AHREs are distinct from established AF; AF developed in 18.2% during follow-up
• entities/Ischemic-Stroke — primary outcome component; stroke incidence ~1%/yr in both groups — far lower than expected in matched ECG-confirmed AF cohorts

Wiki Pages Updated

• wiki/sources/edoxaban-ahre-noahafnet6-nejm-2023.md — created (this file)
• wiki/concepts/Subclinical-AF.md — expanded NOAH-AFNET 6 trial section with detailed results; updated NOAH source citation; updated source_count 5→6; added NOAH to Sources
• wiki/sourceindex.md — added NOAH-AFNET 6 entry
• wiki/wikiindex.md — updated Subclinical-AF description
• wiki/log.md — appended