Asundexian versus Apixaban in Patients with Atrial Fibrillation (OCEANIC-AF)

Authors, Journal, Affiliations, Type, DOI

• Authors: Jonathan P. Piccini, Manesh R. Patel, Jan Steffel, Keith Ferdinand, Isabelle C. Van Gelder, Andrea M. Russo, Chang-Sheng Ma, Shaun G. Goodman, Jonas Oldgren, Christopher Hammett, Renato D. Lopes, Masaharu Akao, Raffaele De Caterina, Paulus Kirchhof, Diana A. Gorog, Martin Hemels, Michiel Rienstra, W. Schuyler Jones, Josephine Harrington, Gregory Y.H. Lip, Stephen J. Ellis, Frank W. Rockhold, Christoph Neumann, John H. Alexander, Thomas Viethen, James Hung, Rosa Coppolecchia, Hardi Mundl, Valeria Caso, for the OCEANIC-AF Steering Committee and Investigators
• Journal: N Engl J Med 2025;392(1):23–32. Published online 1 September 2024; print issue 2 January 2025
• Affiliations: Duke Clinical Research Institute and Duke University Medical Center, Durham, NC (lead, with the executive committee); national and academic sites across 38 countries; Bayer employees among the authors (Wuppertal, São Paulo, Whippany)
• Type: Phase 3, international, double-blind, double-dummy, parallel-group, randomised, active-comparator-controlled non-inferiority trial with a superiority safety objective. Funded by Bayer; oversight by an international executive committee and the Duke Clinical Research Institute in collaboration with the sponsor. Registered NCT05643573 / EudraCT 2022-000758-28
• DOI: 10.1056/NEJMoa2407105

Overview

OCEANIC-AF is the trial that ended the first wave of factor XIa inhibition in atrial fibrillation. In 14,810 patients with AF and a mean CHA₂DS₂-VASc of 4.3, asundexian 50 mg once daily versus apixaban produced more than three times as many strokes or systemic emboli — 98 (1.3%) versus 26 (0.4%), hazard ratio 3.79 (95% CI 2.46–5.83) — with roughly two-thirds fewer major bleeding events (17 [0.2%] versus 53 [0.7%], HR 0.32). The trial was stopped by its data monitoring committee on 19 November 2023, after a median follow-up of only 155 days, and the net clinical benefit composite favoured apixaban (HR 1.61). The design was sound and the population was exactly the one anticoagulation is meant to protect, which is what makes the result so informative: it separates the factor XI hypothesis into two claims — that these drugs bleed less, which survived, and that they prevent stroke equally, which at this dose did not.

Keywords

Not provided — NEJM research articles do not carry author keywords. Indexing terms: atrial fibrillation; factor XIa; asundexian; apixaban; stroke prevention; systemic embolism; major bleeding; non-inferiority; net clinical benefit.

Key Takeaways

Background — the factor XI hypothesis

• AF leads to thromboembolic events with high risk of permanent disability and death; guidelines recommend oral anticoagulation, preferably with DOACs.
• Despite Class I recommendations, many patients are not anticoagulated because of anticipated or actual bleeding — the unmet need this class was built for.
• Factor XI is activated downstream, after the contact pathway is initiated by factor XIIa, and during amplification by thrombin. It therefore contributes to clot progression and pathological thrombosis but has limited effect on haemostasis, because it plays little part in initiation via the extrinsic pathway.
• Most people with factor XI deficiency have no spontaneous bleeding, haemarthroses or haematomas, and have a lower incidence of cardiovascular events, especially cardioembolic stroke.
• Several factor XIa inhibitors had already shown lower bleeding than DOACs in phase 2 work; none had been shown to be as effective at preventing stroke, which is the actual indication.

Asundexian — agent

• Direct, selective, oral inhibitor of factor XIa, once daily, mean terminal half-life 16–18 hours with <15% renal elimination.
• In the phase 2 PACIFIC-AF trial (753 patients), asundexian pooled across 20 mg and 50 mg was associated with a lower incidence of major or CRNM bleeding than apixaban — on a total of 10 events, with no major bleeding, over 12 weeks of follow-up.
• At 50 mg once daily in PACIFIC-AF, trough factor XIa activity fell by 92% and peak activity by 94%.

Design and population

• 1:1 randomisation to asundexian 50 mg daily or apixaban 5 mg twice daily (2.5 mg if ≥2 of: age ≥80, weight ≤60 kg, creatinine ≥1.5 mg/dL), double-dummy with matching placebos.
• 16,436 patients screened at 1,035 sites in 38 countries between 5 December 2022 and 19 November 2023; 14,830 randomised; 20 excluded from one site for Good Clinical Practice violations before unblinding, leaving 14,810 in the intention-to-treat population (7,415 asundexian, 7,395 apixaban) and 14,737 in the safety population.
• Entry: age ≥18, AF documented by ECG at baseline or within 12 months, indication for indefinite oral anticoagulation, and CHA₂DS₂-VASc ≥3 (men) or ≥4 (women); or score 2 (men) / 3 (women) with one additional feature (age ≥70, prior stroke/TIA/systemic embolism, eGFR <50, prior non-traumatic major bleeding, planned single antiplatelet therapy, or ≤6 consecutive weeks of prior oral anticoagulation).
• Mean age 73.9; 35.2% women; mean CHA₂DS₂-VASc 4.3; 18.6% chronic kidney disease; 46.8% heart failure; 88.6% hypertension; 36.9% diabetes; 18.2% prior stroke or TIA; 16.8% had received oral anticoagulants for ≤6 weeks.
• Median follow-up 155 days.
• Primary efficacy: stroke or systemic embolism, intention-to-treat. Primary safety: ISTH major bleeding, on-treatment. Primary net clinical benefit: stroke, systemic embolism, or ISTH major bleeding.
• Non-inferiority margin for the cause-specific hazard ratio was 1.5, chosen to preserve at least 50% of apixaban's relative benefit. No prespecified stopping rules; an interim analysis was planned at 85 primary efficacy events.

Efficacy — the trial's defining result

• The excess was present from the start of the trial and persisted through follow-up — not a late divergence.
• Stroke or systemic embolism occurred in more than three times as many asundexian patients; the diagnosis rests on 124 events in total, 98 of them in the asundexian arm.

Safety — bleeding was genuinely lower

• The last row is the verdict. The safety advantage did not compensate for the stroke excess: the composite of stroke, systemic embolism or major bleeding favoured apixaban by a clear margin.
• No formal testing for safety or other end points was conducted because of early termination, so all of these are point estimates and confidence intervals, not hypothesis tests.
• Adverse events were similar: any event 34.9% in both groups; leading to discontinuation 2.0% versus 1.6%.

Why the stroke rate was higher — three proposed explanations

1. Factor XIa inhibition may not prevent stroke effectively in AF. Despite the genetic epidemiology, escape mechanisms may operate, especially in patients transitioned from ongoing oral anticoagulation — a group deliberately included here (16.8% with ≤6 weeks of prior OAC).
2. 50 mg may be an inadequate dose. PACIFIC-AF showed 92–94% reduction in factor XIa activity at trough and peak, but effective thrombus prevention may require complete or near-complete (>99%) suppression — which is the figure abelacimab achieved at 150 mg in AZALEA–TIMI 71.
3. The benchmark was unusually strong. Ischemic stroke in the apixaban arm was lower than predicted by CHA₂DS₂-VASc and lower than in ARISTOTLE more than a decade earlier, attributed to better treatment of coexisting conditions and more rhythm control in contemporary populations. A fixed non-inferiority margin built on historical event rates is harder to meet when the control arm outperforms its own history.

What the field took from it

• The trial's own conclusion: there is a need to determine "whether inhibition of factor XI or XIa can reduce the occurrence of stroke or systemic embolism in patients with atrial fibrillation, and if so, in which population and what degree of factor XIa inhibition is necessary."
• The companion AZALEA–TIMI 71 trial (abelacimab, subcutaneous monoclonal antibody, 99% free factor XI reduction) had already shown a large bleeding reduction with numerically more ischemic strokes in a smaller, open-label, non-efficacy-powered comparison. The two trials together read as a class signal unless the potency difference is real — a question only LILAC–TIMI 76, testing abelacimab against placebo in patients unsuitable for anticoagulation, may settle.

Limitations of the document

Authors' own stated limitations

• Stopped prematurely, so follow-up was abbreviated despite substantial total patient-years of exposure.
• The number of patients approached is unknown, so generalisability to the overall AF population cannot be assessed.
• The trial addressed only patients at risk of both stroke and bleeding with an indication for indefinite anticoagulation; efficacy or safety in other populations is undetermined.
• Only one dose of asundexian was tested; whether higher doses with greater factor XIa inhibition would perform differently is unknown.

Additional limitations evident in the design

• The efficacy conclusion rests on a trial that did not reach its planned size (18,000 planned versus 14,810 randomised) and was stopped at a median of five months — which is enough to see a large harm signal but not to characterise absolute annual risks. The reported hazard ratio of 3.79 is plausible in a transition population and possibly inflated by early events.
• The dose question is untested, not answered. The authors are careful to say 50 mg "may not be a high enough dose", but the trial cannot distinguish a dose failure from a target failure: it does not report achieved factor XIa suppression in OCEANIC-AF itself, only the PACIFIC-AF figure for the same dose.
• The 20-patient exclusion for Good Clinical Practice violations is disclosed but the effect of removing a whole site's patients before unblinding is not analysed in sensitivity form.
• The apixaban dose-reduction criteria were applied as standard; whether the comparator's performance is representative of real-world apixaban use is not addressed.
• No head-to-head comparison with a factor XI (as opposed to XIa) inhibitor exists; the mechanistic explanation offered for the asundexian/abelacimab divergence is a hypothesis from two different trials, not an experimental finding.
• Industry funding and authorship — Bayer funded, designed and employs several authors. The Duke Clinical Research Institute performed the analysis, but the trial hypothesis belonged to the sponsor.

Key Concepts Mentioned

• concepts/Factor-XI-Inhibition — the class-defining failure; OCEANIC-AF is the efficacy anchor for the whole field
• concepts/NOAC-AF-Management — apixaban dosing and dose-reduction criteria as applied in the comparator arm
• entities/Atrial-Fibrillation — the indication
• Factor XIa; contact activation; amplification phase; uncoupling thrombosis from haemostasis
• Non-inferiority margins and the preservation-of-effect principle
• Net clinical benefit composites
• ISTH major bleeding definition

Key Entities Mentioned

• entities/Atrial-Fibrillation
• Asundexian (BAY 2433334) — Bayer
• Apixaban — the active comparator, and the benchmark every subsequent AF anticoagulant must beat
• Abelacimab — the factor XI/XIa antibody that took the opposite result in AZALEA–TIMI 71
• ARISTOTLE — historical apixaban benchmark, cited for the lower-than-expected contemporary stroke rate
• PACIFIC-AF — the phase 2 predecessor
• Duke Clinical Research Institute — independent analysis

Wiki Pages Updated

• Created: wiki/sources/asundexian-oceanicaf-nejm-2025.md
• Created: wiki/concepts/Factor-XI-Inhibition.md
• Updated: wiki/concepts/NOAC-AF-Management.md
• Updated: wiki/entities/Atrial-Fibrillation.md
• Updated: wiki/sourceindex.md
• Updated: wiki/wikiindex.md