Dropping aspirin during a heart attack traded fewer bleeds for more events

The direction of travel in antiplatelet therapy has been to use less of it. Dual therapy — aspirin plus a P2Y12 inhibitor — prevents clots and causes bleeding, and a decade of trials has been spent finding places where one drug can be dropped without paying for it in ischaemic events.

PREMIUM, published in the New England Journal of Medicine on 29 August and presented at ESC Congress 2026, tried to drop aspirin at the earliest and highest-risk moment: before primary percutaneous coronary intervention for ST-segment elevation myocardial infarction [s1]. It did not work.

What the trial did

PREMIUM was a multicentre, open-label, randomised trial in Japan involving patients with STEMI undergoing primary PCI [s1]. Patients were randomly assigned 1:1, before the procedure, to low-dose prasugrel monotherapy or to dual antiplatelet therapy with aspirin plus low-dose prasugrel for 12 months [s1].

The primary outcome was a composite of death from any cause, stroke, or myocardial infarction at 12 months, assessed for noninferiority with a prespecified margin of 1.50 on the upper bound of the 95% confidence interval of the hazard ratio [s1]. The major secondary outcome was major bleeding — Bleeding Academic Research Consortium type 3 or type 5 — at 12 months, to be tested for superiority only if noninferiority on the primary outcome was established [s1].

That gating matters. The bleeding comparison was designed to be interpreted only after the ischaemic question had been answered safely.

A total of 2,216 patients were included in the full analysis population: 1,109 assigned to monotherapy and 1,107 to dual therapy [s1].

What happened

At 12 months, death from any cause, stroke, or myocardial infarction had occurred in 124 patients receiving monotherapy (Kaplan-Meier estimate 11.0%) and 94 receiving dual therapy (Kaplan-Meier estimate 8.5%) [s1]. The hazard ratio was 1.34 (95% CI 1.02 to 1.75), with P = 0.40 for noninferiority [s1].

The whole interval sits above one. Noninferiority was not established, and the point estimate is in the direction of harm.

Major bleeding occurred in 61 patients receiving monotherapy (Kaplan-Meier estimate 5.6%) and 92 receiving dual therapy (Kaplan-Meier estimate 8.4%), a hazard ratio of 0.66 (95% CI 0.47 to 0.91) [s1]. Definite or probable stent thrombosis and serious adverse events appeared similar between the two groups [s1].

So the trade the trial was testing did occur — less bleeding without aspirin — but it was bought with ischaemic events, and by the trial's own prespecified logic the bleeding result does not stand as a finding because the gate it sat behind did not open.

The authors' conclusion is unhedged: low-dose prasugrel monotherapy initiated before PCI was not noninferior to dual antiplatelet therapy for 12 months with respect to the composite outcome [s1].

Timing is the variable, not aspirin

PREMIUM should not be read as evidence that aspirin can never be dropped after a stent. A trial published the same day illustrates why.

A-CLOSE, an open-label noninferiority trial in South Korea, enrolled 3,203 patients with high-risk clinical or lesion characteristics 12 months after drug-eluting stent implantation and randomly assigned them to clopidogrel monotherapy or extended dual antiplatelet therapy [s2]. Over 24 months, the primary net adverse clinical events end point occurred in 80 patients (5.0%) on monotherapy and 81 (5.1%) on dual therapy, a risk difference of -0.1 percentage points (90% CI -1.3 to 1.2, P = 0.001 for noninferiority) [s2].

But A-CLOSE's own secondary endpoints are not reassuring in one direction: death, myocardial infarction, stent thrombosis or stroke occurred in 60 patients (3.7%) on monotherapy and 26 (1.6%) on dual therapy (hazard ratio 2.33, 95% CI 1.47 to 3.69, P<0.001), while BARC type 2, 3 or 5 bleeding occurred in 28 patients (1.8%) on monotherapy and 65 (4.1%) on dual therapy (hazard ratio 0.43, 95% CI 0.27 to 0.67, P<0.001) [s2].

Read together, the two trials describe the same trade at different points on the timeline: dropping one antiplatelet reduces bleeding and increases ischaemic events, and whether the net comes out acceptable depends on when you do it and to whom. At the moment of a STEMI, PREMIUM found it did not [s1].

Limits

PREMIUM was open-label, conducted in Japan, and used low-dose prasugrel — a dosing convention specific to Japanese practice [s1]. Neither the drug exposure nor the background bleeding risk transfers automatically to other populations.

A-CLOSE was likewise conducted in South Korea and in a different clinical situation entirely, 12 months after stenting rather than at the index event [s2]. The comparison drawn above is thematic, not statistical; the two trials share no population, no comparator and no end point definition.

PREMIUM was funded by Boston Scientific Japan and registered as NCT05709626 [s1]. A-CLOSE was funded by Chong Kun Dang and Samjin and registered as NCT03947229 [s2].

What to watch

Whether the deescalation literature now starts distinguishing sharply by timing rather than by drug. PREMIUM is the clearest evidence so far that the acute phase of a STEMI is not where aspirin can be removed, and the size of its ischaemic excess — a hazard ratio of 1.34 with a lower bound of 1.02 — makes that a difficult finding to explain away [s1].

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

• Aspirin Omission at the Time of Primary Percutaneous Coronary Intervention in STEMI, New England Journal of Medicine, 29 August 2026
• Clopidogrel or Dual Antiplatelet Therapy in High-Ischemic-Risk Patients, New England Journal of Medicine, 29 August 2026