New type of blood thinner fails to prevent repeat heart attacks

A new type of blood thinner designed to prevent clots without causing dangerous bleeding has failed to cut repeat heart attacks, strokes and heart deaths in people who had just survived a heart attack or unstable angina.

In LIBREXIA ACS, a trial of more than 14,000 people, milvexian added to standard antiplatelet medicines made no difference to major cardiovascular events compared with placebo. The trial had already been stopped early, in November 2025, after an interim analysis showed it was unlikely to succeed. The full results show the drug didn't raise the risk of bleeding into the brain or fatal bleeding either, which keeps hopes alive for trials in other conditions.

Professor P. Gabriel Steg of Hôpital Bichat in Paris presented the results in the Hot Line 3 session at ESC Congress 2026 in Munich, at 11:00 on 29 August 2026 (11:00 in Madrid). They were published simultaneously in the New England Journal of Medicine.

The problem after a heart attack

Acute coronary syndrome (ACS) is the umbrella term for a sudden drop in blood supply to the heart muscle. It includes heart attacks and unstable angina. Usually, a fatty plaque in a coronary artery ruptures and a blood clot forms on top of it.

Treatment typically involves opening the artery with a stent and then giving two antiplatelet medicines, such as aspirin plus clopidogrel, ticagrelor or prasugrel, along with a statin. Even so, the risk of another heart attack, stroke or cardiovascular death remains high in the following year.

Platelets are only one part of clotting. The other is a chain of proteins called the coagulation cascade. Adding a traditional anticoagulant to antiplatelet treatment can cut repeat events further, but it also increases bleeding. In the ATLAS ACS 2 trial, a low dose of rivaroxaban reduced cardiovascular events after ACS, but raised the risk of major bleeding. An earlier trial of apixaban in the same setting, APPRAISE-2, was stopped because of bleeding.

"Adding current anticoagulants to standard therapy has been tested to reduce recurrent thrombotic events, but bleeding risk was increased," Steg said.

The factor XI idea

Milvexian works on a protein called factor XIa. It sits in part of the clotting cascade that appears to contribute more to harmful clots inside blood vessels than to the clotting that stops bleeding after an injury.

Evidence for this comes partly from people born with low levels of factor XI. They have fewer strokes and clots in the veins, but they rarely have serious spontaneous bleeding. That raised the possibility that blocking factor XIa could prevent clots with less bleeding than existing anticoagulants.

"There has been increasing interest in developing factor XIa inhibitors to prevent harmful thrombosis, while preserving normal clotting processes to minimise bleeding," Steg said. "We conducted the LIBREXIA ACS trial to assess the efficacy and safety of the selective factor XIa inhibitor, milvexian, after ACS when added to standard antiplatelet therapy."

Milvexian, a tablet, is being developed by Bristol Myers Squibb and Johnson & Johnson. LIBREXIA is a programme of three large phase 3 trials, in ACS, atrial fibrillation and stroke prevention.

The LIBREXIA ACS trial

LIBREXIA ACS (NCT05754957) was a randomised, double-blind, placebo-controlled trial run at 893 sites in 44 countries and funded by Bristol Myers Squibb and Janssen Research and Development, part of Johnson & Johnson.

To take part, people needed to have had an ACS within the previous seven days. They could have had a stent or been treated with medicines alone, and they needed at least two factors linked to a higher risk of further events. Everyone received standard antiplatelet medicines chosen by their doctor.

In total, 14,194 people were randomly assigned to:

• Milvexian 25 mg twice a day (7,094 people)
• A matching placebo (7,100 people)

The main goal was a combination of cardiovascular death, heart attack and ischaemic stroke (a stroke caused by a blocked artery). The main safety measure was severe bleeding defined by the Bleeding Academic Research Consortium as type 3c (bleeding inside the skull) or type 5 (fatal bleeding).

Why the trial stopped early

An independent data monitoring committee reviewed the results at a planned interim analysis based on 556 adjudicated events. It recommended stopping because the trial was unlikely to meet its main goal. The sponsors ended the trial in November 2025. The futility finding was confirmed when the full data were analysed.

What the trial found

The main outcome occurred in 5.4% of people on milvexian and 5.1% on placebo (hazard ratio 1.05, 95% confidence interval 0.91 to 1.21, p=0.50). There was no difference in any of the individual parts of that outcome, or in any of the major secondary measures, including death from any cause.

The most serious types of bleeding were rare and identical in both groups, at 0.3% (hazard ratio 1.04, 95% confidence interval 0.58 to 1.87, p=0.88).

The ESC press release gave the median follow-up as 10 months, while the trial summary on Synapse, drawing on the journal paper, gave 12.2 months.

Blood tests showed that people taking milvexian had a longer activated partial thromboplastin time, a measure of clotting. That indicates the drug was doing what was expected biologically at this dose. It simply didn't translate into fewer events.

Why it might have failed

There are several possible explanations, and the trial can't distinguish between them.

• The dose. LIBREXIA ACS used 25 mg twice a day, a relatively low dose chosen to limit bleeding when combined with two antiplatelet drugs. The atrial fibrillation and stroke trials use 100 mg twice a day.
• The disease. Heart attacks are driven mainly by platelets clumping on ruptured plaques in fast-flowing arteries. Factor XI may play a smaller role there than in clots that form in slow-flowing blood, such as in the heart during atrial fibrillation.
• Strong background treatment. Modern antiplatelet medicines, stents and statins may leave little room for another drug to make a difference.
• Follow-up. The trial was stopped early, so follow-up was around a year at most.

How experts reacted

Clinicians quoted after the presentation broadly agreed that milvexian was safe but ineffective after ACS at this dose. Robert Harrington, dean of Weill Cornell Medicine, said the result "confirms the complexity of treating ACS and the need to further advance knowledge of the disease," adding that, with no new safety concerns, factor XIa inhibition "continues to be a promising mechanism."

What it means for other trials

The absence of extra bleeding is the main positive. Steg said: "While no effect on efficacy was shown, the absence of an observed increase in intracranial or fatal bleeding is reassuring given that two further trials are ongoing: LIBREXIA AF for patients with atrial fibrillation and LIBREXIA Stroke for secondary stroke prevention. These studies are distinct from LIBREXIA ACS in several aspects, including patient populations, endpoints, type and duration of background therapy and disease pathology."

Those trials will test whether factor XIa inhibitors can match or beat existing anticoagulants in settings where clots in slow-moving blood are a bigger problem. Other factor XI drugs are also in development. Results from these trials will decide whether this class of medicines finds a place in practice.

What it means for patients

For people who have had a heart attack or unstable angina, nothing changes. Standard treatment with antiplatelet medicines, statins and other proven therapies remains the approach. Milvexian is not approved for any use. People taking part in the ongoing LIBREXIA trials should follow the advice of their trial team.

This article explains published research and isn't medical advice. Talk to your own doctor or specialist team before changing any treatment.