A factor XIa inhibitor added nothing after heart attack — and, notably, no extra bleeding

The factor XIa inhibitors were built on a specific bet: that you can uncouple pathological clotting from normal haemostasis, and therefore prevent ischaemic events without buying the bleeding that comes with every anticoagulant ever marketed. LIBREXIA-ACS, published in the New England Journal of Medicine on 29 August, is the largest test of that bet in acute coronary syndrome to date. It settles half the question and leaves the other half open [s1].

What the trial did

LIBREXIA-ACS was a phase 3, randomised, placebo-controlled trial evaluating milvexian, an oral factor XIa inhibitor, added to standard antiplatelet therapy within seven days of an acute coronary syndrome event [s1]. Patients were assigned 1:1 to oral milvexian 25 mg twice daily or matched placebo [s1].

The primary efficacy outcome was a composite of cardiovascular death, myocardial infarction or ischaemic stroke, analysed as time to event [s1]. The principal safety outcome was Bleeding Academic Research Consortium type 3c or 5 bleeding — intracranial or intraocular bleeding compromising vision, or fatal bleeding [s1].

What happened

After a planned interim analysis based on 556 adjudicated efficacy endpoints, the trial was terminated for futility [s1]. By that point 14,194 patients had been enrolled: 7,094 to milvexian and 7,100 to placebo [s1].

Over a median 12.2 months of follow-up, a primary efficacy event occurred in 384 patients (5.4%) on milvexian and 365 (5.1%) on placebo — a hazard ratio of 1.05 (95% CI 0.91–1.21, P = 0.50) [s1]. The point estimate is on the wrong side of one.

BARC type 3c or 5 bleeding occurred in 23 patients (0.3%) on milvexian and 22 (0.3%) on placebo (P = 0.88) [s1].

The authors' conclusion states both halves: among patients with a recent acute coronary syndrome event, milvexian did not decrease the risk of cardiovascular death, myocardial infarction or ischaemic stroke, but did not increase the risk of intracranial or fatal bleeding compared with placebo [s1]. The trial was funded by Janssen Research and Development and Bristol Myers Squibb, and registered as NCT05754957 [s1].

Why the safety result is not a consolation prize

It is tempting to read "no benefit, no harm" as a null result twice over. That misreads what was being tested.

The premise the trial's own framing sets out is that milvexian "may reduce the risk of major adverse clinical events with minimal bleeding risk" [s1] — that is, that factor XI can be inhibited without the bleeding cost that constrains anticoagulation added on top of antiplatelet therapy. If that premise were wrong, adding a factor XIa inhibitor on top of antiplatelet therapy in 7,094 patients should have produced excess intracranial and fatal bleeding. It produced 23 events against 22 [s1].

So LIBREXIA-ACS is evidence about the mechanism and evidence about the indication, and the two point different ways. The safety hypothesis survived a large test. The efficacy hypothesis, in this population and at this dose, did not.

The dose question

The obvious objection to a futility result is that the dose was too low — that a drug engineered not to cause bleeding was given at an exposure that also did not prevent clotting.

The LIBREXIA programme's dose selection was published separately in March, setting out the rationale for the milvexian doses taken into phase 3 [s2]. That the reasoning was documented in advance does not make it correct, but it does mean the 25 mg twice-daily regimen was a stated position rather than a retrospective one. A futility stop at an interim analysis is also, by construction, a decision that the observed trajectory made benefit implausible for the remaining follow-up — not a finding that the drug does nothing at any dose.

What this does not tell you

The trial addresses one indication. Milvexian's phase 3 programme also includes a study in patients with atrial fibrillation, for which dose selection was published separately [s3] — a setting where the comparator is an existing anticoagulant rather than placebo, and where the bleeding trade-off is the whole clinical problem. A futility result against placebo added to antiplatelet therapy after acute coronary syndrome does not transfer to a trial where the drug stands in for an anticoagulant rather than being layered on antiplatelets.

Nor does it establish anything about patients outside the enrolled population, about longer follow-up than a median of 12.2 months, or about the drug in combination with regimens other than the standard antiplatelet therapy used here [s1].

What to watch

The measurable next question is whether the class's remaining phase 3 programmes report the same bleeding profile. If they do, factor XI emerges from this as a validated safety target in search of an indication. If a benefit appears in a setting where the comparator is an anticoagulant rather than placebo, the acute coronary syndrome result will look like a question of context rather than of mechanism.

This article describes trial results. It is not advice about any medicine, and nothing here should be used to start, stop or change treatment.

Sources

• Milvexian with Antiplatelet Therapy after Acute Coronary Syndrome Event, New England Journal of Medicine, 29 August 2026
• Rationale for the milvexian dosing in the phase 3 LIBREXIA program, Journal of Thrombosis and Haemostasis, 4 March 2026
• Quantitative Model-Informed Dose Selection for a Milvexian Phase III Study in Patients With Atrial Fibrillation, Clinical Pharmacology & Therapeutics, 22 August 2025