Factor XI Inhibition
Definition
Anticoagulation directed at factor XI or its activated form, factor XIa, rather than at thrombin or factor Xa. The premise is mechanistic: factor XI acts mainly in the amplification phase of coagulation, after clotting has been initiated, so blocking it should attenuate pathological thrombus growth while leaving haemostatic plug formation largely intact — "uncoupling thrombosis from hemostasis." Two compounds have randomised data in atrial fibrillation, and they produced opposite verdicts: abelacimab, a subcutaneous monoclonal antibody, cut bleeding by 62–69% against rivaroxaban without an efficacy-powered stroke comparison (sources/abelacimab-azaleatimi71-nejm-2025, rating: high), while asundexian, an oral factor XIa inhibitor, cut bleeding by 68% against apixaban but caused 3.8 times as many strokes or systemic emboli (sources/asundexian-oceanicaf-nejm-2025, rating: very high). No agent in this class is approved for any indication.
Key Concepts
Why factor XI, and why the hypothesis was reasonable
- Factor XI is activated in two ways — by factor XIIa after contact activation, and by thrombin during amplification. It therefore contributes to clot progression and pathological thrombosis but has little role in the initiation of haemostasis through the extrinsic pathway (sources/asundexian-oceanicaf-nejm-2025, rating: very high).
- The human genetics are unusually clean: most people with inherited factor XI deficiency have no spontaneous bleeding, haemarthroses or haematomas, and have a lower incidence of cardiovascular events, especially cardioembolic stroke, than people without the deficiency (sources/asundexian-oceanicaf-nejm-2025). Abelacimab's investigators make the same argument from the same epidemiology (sources/abelacimab-azaleatimi71-nejm-2025).
- The clinical motivation is not theoretical novelty but undertreatment: a substantial share of patients with a Class I indication for anticoagulation do not receive it because of anticipated or actual bleeding, most often gastrointestinal (sources/asundexian-oceanicaf-nejm-2025).
The two agents are not the same drug
| Feature | Abelacimab | Asundexian |
|---|---|---|
| Construct | Fully human monoclonal antibody | Oral small molecule |
| Target | Binds the catalytic domain of factor XI and locks it inactive; also inhibits factor XIa — dual mechanism | Direct, selective factor XIa inhibitor only |
| Route and interval | Subcutaneous, once monthly | Oral, once daily |
| Pharmacokinetics | Near-complete factor XI suppression within 1 h IV, sustained up to 30 days | Terminal half-life 16–18 h; <15% renal elimination |
| Achieved suppression | 99% reduction in free factor XI at 150 mg (IQR 98–99) | 92% trough / 94% peak reduction in factor XIa activity at 50 mg |
| Pivotal AF comparison | AZALEA–TIMI 71 versus rivaroxaban, n=1,287, phase 2b | OCEANIC-AF versus apixaban, n=14,810, phase 3 |
| Result | Bleeding ↓ 62–69%; stroke numerically ↑ | Bleeding ↓ 68%; stroke ↑ 3.8-fold |
(sources/abelacimab-azaleatimi71-nejm-2025, sources/asundexian-oceanicaf-nejm-2025)
Bleeding: the part of the hypothesis that held
Both trials reduced bleeding substantially and consistently across severity grades, which is the strongest evidence in the class.
| End point | Abelacimab 150 mg vs rivaroxaban | Asundexian 50 mg vs apixaban |
|---|---|---|
| Major or CRNM bleeding | 3.22 vs 8.38/100 py; HR 0.38 (0.24–0.60) | 1.1% vs 2.6%; HR 0.44 (0.34–0.57) |
| Major bleeding | 1.22 vs 3.73/100 py; HR 0.33 (0.16–0.66) | 0.2% vs 0.7%; HR 0.32 (0.18–0.55) |
| Intracranial haemorrhage | 2 vs 4 patients; HR 0.51 (0.09–2.78) | 3 (<0.1%) vs 18 (0.2%); HR 0.16 (0.05–0.55) |
| Gastrointestinal bleeding | 2 vs 18 events; HR 0.11 (0.03–0.48) | Not a prespecified primary |
| Fatal bleeding | Not reported separately | 0 vs 4 patients |
- The gastrointestinal result in AZALEA–TIMI 71 is the most mechanistically interesting bleeding finding: the gut is the commonest bleeding site on DOACs and the one site where DOACs do not beat warfarin, and a parenteral agent nearly abolished it (two events per abelacimab arm versus 18). The authors suggest local gastrointestinal exposure to an oral direct-acting drug may itself contribute to bleeding (sources/abelacimab-azaleatimi71-nejm-2025).
- A dose–response was present in AZALEA–TIMI 71 — 90 mg bled less than 150 mg — which argues against the result being an artefact of the open-label design.
Stroke prevention: the part that did not
- OCEANIC-AF is decisive. Stroke or systemic embolism occurred in 98 patients (1.3%) on asundexian and 26 (0.4%) on apixaban, cause-specific HR 3.79 (95% CI 2.46–5.83); the excess was present from the start and persisted. The divergence was driven by ischaemic stroke (HR 4.06), and it overwhelmed the bleeding advantage: the composite of stroke, systemic embolism or major bleeding favoured apixaban, HR 1.61 (1.21–2.15) (sources/asundexian-oceanicaf-nejm-2025, rating: very high).
- AZALEA–TIMI 71 had the same directional signal but no power to interpret it. Ischaemic stroke occurred at 1.21 and 1.24 per 100 person-years on abelacimab 150 and 90 mg versus 0.59 on rivaroxaban — hazard ratios above 2 with confidence intervals crossing 1 (2.06, 0.70–6.02; 2.10, 0.72–6.14). Haemorrhagic stroke was numerically lower on abelacimab (0 and 1 versus 2 patients). The trial had 80% power to detect a 40% bleeding reduction and none to exclude a stroke difference, and the authors say so plainly (sources/abelacimab-azaleatimi71-nejm-2025, rating: high).
- The net clinical outcome flipped between trials — HR 0.55 favouring abelacimab against rivaroxaban, HR 1.61 favouring apixaban against asundexian. In both cases the composite is dominated by whichever component the trial generated most of.
The three candidate explanations for the divergence
| Explanation | Supporting detail | What would test it |
|---|---|---|
| Mechanism — factor XI and XIa inhibition is required; XIa inhibition alone is insufficient | Abelacimab prevents generation of XIa as well as inhibiting it; asundexian inhibits XIa activity only | A factor XI-directed agent in a properly powered AF efficacy trial |
| Potency — 99% suppression is necessary, 92–94% is not enough | Effective thrombus prevention may require complete or near-complete (>99%) factor XIa suppression; abelacimab 150 mg achieved 99%, asundexian 50 mg 92–94% | Dose-ranging with measured target engagement and outcome, which the authors note is not feasible in AF because event rates are too low |
| Comparator and benchmark — apixaban in 2024 outperformed its own historical event rate | Ischaemic stroke on apixaban was lower than CHA₂DS₂-VASc predicted and lower than in ARISTOTLE a decade earlier, attributed to better risk-factor treatment and more rhythm control; a non-inferiority margin of 1.5 is harder to meet when the control arm beats its history | Contemporary event rates rather than historical margins |
(sources/asundexian-oceanicaf-nejm-2025, sources/abelacimab-azaleatimi71-nejm-2025)
- The mechanistic case cannot be settled from two separate trials. The OCEANIC-AF investigators state the question directly: whether upstream or combined inhibition of factor XI and factor XIa would lead to effective stroke prevention is unknown.
What is still running, and what it will and will not answer
- LILAC–TIMI 76 (NCT05712200) is testing abelacimab against placebo in patients with AF at high stroke risk in whom current anticoagulation is deemed unsuitable. That design sidesteps the need to match a DOAC, but it answers a narrower question — whether the drug beats nothing in a population no other drug is being given to — and it will not tell us whether factor XI inhibition is a DOAC substitute (sources/abelacimab-azaleatimi71-nejm-2025).
- Population definitions differ in a way that matters: AZALEA–TIMI 71 enrolled patients with median CHA₂DS₂-VASc 5 of whom 92% had already taken an anticoagulant for ≥60 days, whereas OCEANIC-AF deliberately included 16.8% who had received oral anticoagulation for ≤6 weeks — a transition population in which escape mechanisms after stopping a DOAC are plausible (sources/abelacimab-azaleatimi71-nejm-2025, sources/asundexian-oceanicaf-nejm-2025).
- Consequence for practice: factor XI inhibition is not an option to offer a patient who cannot take a DOAC because of bleeding, unless that patient is enrolled in a trial. The only evidence-backed alternatives in that situation remain dose-adjusted DOACs, warfarin, and left atrial appendage occlusion — see entities/LAAO and concepts/NOAC-AF-Management.
Contradictions / Open Questions
- The class claim is split, and the trials are not interchangeable. One trial shows large bleeding reductions with a favourable net outcome against rivaroxaban; a much larger trial shows large bleeding reductions with a stroke excess large enough to reverse net benefit against apixaban. "Factor XI inhibitors bleed less" is well supported. "Factor XI inhibitors are safer anticoagulants" is not, because safety is not the same as net benefit (sources/asundexian-oceanicaf-nejm-2025, rating: very high; sources/abelacimab-azaleatimi71-nejm-2025, rating: high).
- Whether the abelacimab ischemic stroke signal is noise is unknowable from 28 events. The point estimates (HR >2 in both dose arms) are uncomfortable, the confidence intervals are wide, and the trial was never designed to answer the question. Dismissing it as chance and acting on it as a class signal are both defensible readings of the same table.
- Mechanism versus dose cannot be separated. Abelacimab and asundexian differ in target, potency, route, comparator, population and trial design simultaneously. Attributing the outcome difference to "factor XI versus factor XIa" is a hypothesis the data cannot confirm.
- The dose question may be unanswerable prospectively. The OCEANIC-AF authors note that phase 2 efficacy-based dose-finding in AF is not possible because stroke event rates are too low — so the field may have to choose doses on pharmacodynamics alone, which is exactly what produced the 92–94% versus 99% problem.
- Two trials, two early terminations, two different truncations. AZALEA–TIMI 71 stopped early for unexpected bleeding benefit; OCEANIC-AF stopped early for unexpected stroke harm. In both cases the stopping rule was appropriate and in both cases the resulting data are less informative than a completed trial would have been.
- No reversal agent is described for either drug, and no data exist on managing major bleeding on factor XI inhibition. Both trials report adverse-event profiles similar to the DOAC comparator, but periprocedural and trauma management are unaddressed.
- Long-term safety is unknown — the longest follow-up in the class is a median 2.1 years in a 1,287-patient trial. A drug intended for indefinite use in an elderly population has no data beyond that horizon.
- Cost, access and the actual clinical niche are unexamined. If the class eventually proves useful, the plausible role is in patients who cannot be anticoagulated, which is a small population unlikely to drive the pricing assumed by the trials that tested it against DOACs.
Connections
- concepts/NOAC-AF-Management — the DOAC standard this class was built to beat, and the dose-reduction rules used in both comparator arms
- entities/Atrial-Fibrillation — the indication, and the stroke-prevention thresholds that define who would ever be offered a novel anticoagulant
- concepts/CHA2DS2-VA — risk stratification; both trials enrolled on CHA₂DS₂-VASc ≥4 or equivalent
- entities/LAAO — the evidence-backed alternative when anticoagulation is unsuitable and a device is being considered
- concepts/Subclinical-AF — the adjacent question of whether lower-risk patients benefit from anticoagulation at all