Back to StatUp
Journal Club

AQUATIC: When "More" Is Worse

A trial stopped for harm - understanding why aspirin hurt patients on anticoagulation

Lemesle G, et al. Aspirin in Patients with Chronic Coronary Syndrome Receiving Oral Anticoagulation. NEJM 2025;393:1578-88.

Section 1: Why Did We Need This Trial?

Picture this: A 72-year-old man with atrial fibrillation, two years out from a drug-eluting stent. He's on apixaban. Should he also be on aspirin?

The Clinical Dilemma

About 15% of patients with chronic coronary syndrome also need long-term anticoagulation - usually for atrial fibrillation. These patients are caught between two risks:

  • Atherothrombotic risk: MI, stent thrombosis, stroke from plaque
  • Bleeding risk: Already elevated on anticoagulation

The intuition: "More antithrombotic therapy = better protection." But is that true?

2017: COMPASS Trial

Low-dose rivaroxaban + aspirin reduced CV events in stable CAD. Sparked interest in "dual-pathway" therapy.

2019: AFIRE Trial (Japan)

Rivaroxaban alone vs. rivaroxaban + antiplatelet in AF + stable CAD. Stopped early - combination therapy increased bleeding AND death. But it was open-label.

2019: OAC-ALONE Trial (Japan)

Similar findings in Japanese patients >1 year post-stent. Also open-label, also stopped early.

2024: EPIC-CAD Trial (Korea)

Edoxaban alone vs. edoxaban + antiplatelet. Open-label. Similar conclusions.

The problem: All these trials were open-label. Physicians and patients knew who was on combination therapy. This introduces bias in event reporting, crossovers, and clinical decisions.

The Gap AQUATIC Filled

AQUATIC was the first double-blind, placebo-controlled trial asking this question. And it enrolled a high-risk European population - not the lower-risk Asian cohorts of prior trials.

Why double-blind matters: In an open-label trial, if a patient on combination therapy has chest pain, the physician might be more likely to call it an MI. If they have minor bleeding, it might get classified as "major." Blinding eliminates this bias.

Section 2: The Protocol - A True Placebo-Controlled Design

AQUATIC's key strength was its rigorous double-blind design. Let's understand what made it different.

Aspirin Group

433

Patients randomized to
OAC + Aspirin 100mg daily

Placebo Group

439

Patients randomized to
OAC + Identical placebo

Key Design Elements

  • Enrollment: May 2020 to April 2024
  • Sites: 51 centers in France
  • Design: Double-blind, placebo-controlled
  • Randomization: 1:1, stratified by OAC type and baseline antiplatelet status
  • Follow-up: Median 2.2 years (planned 24-48 months)

The Two Strata

Patients entered based on their baseline antithrombotic regimen:

Stratum A (67.7%): Already on OAC + aspirin

Randomized to continue aspirin vs. switch to placebo

Stratum B (32.3%): On OAC alone

Randomized to add aspirin vs. add placebo

The Primary Endpoint

A composite of:

  • Cardiovascular death
  • Myocardial infarction
  • Stroke
  • Systemic embolism
  • Coronary revascularization
  • Acute limb ischemia

This composite captures both thrombotic events AND the consequences of bleeding (which can trigger revascularization and death).

What they expected: The trial was powered to detect a 25% reduction in events with aspirin (HR 0.75). They expected 16% event rate in placebo at 24 months. Spoiler: that's not what happened.

Section 3: Who Got In? The High-Risk Selection

AQUATIC deliberately enrolled SICK patients. This wasn't a low-risk cohort - it was designed to test aspirin where we might expect it to help most.

Core Eligibility

  • Age 18 or older
  • Chronic coronary syndrome (stable CAD)
  • Prior coronary stent >6 months before enrollment
  • Currently on oral anticoagulation (any indication)
  • Features of high residual atherothrombotic risk

Definition of "High Atherothrombotic Risk"

Patients needed at least ONE of these:

  • Prior ACS with stent implantation, OR
  • Prior elective PCI PLUS one of:
    • Diabetes
    • Diffuse 3-vessel disease
    • Chronic kidney disease (CrCl <50 ml/min)
    • Prior stent thrombosis
    • Peripheral artery disease
    • Complex PCI (left main, 3+ stents, bifurcation, >60mm stent length, CTO)

Population Characteristics

Mean age 71.7 years
Male 85.3%
Prior MI 72.1%
AF indication for OAC 89.0%
Diabetes 37.4%
PAD 14.4%
Heart failure 26.5%
CHA2DS2-VASc 4 (high)
This was a HIGH-RISK population. Compare to prior trials:
  • AFIRE: 35% prior MI
  • EPIC-CAD: 35% prior MI
  • AQUATIC: 72% prior MI

Event rates in AQUATIC were 7-8x higher than prior trials. If aspirin was going to help anyone, it should have been these patients.

The anticoagulants used: 62% apixaban, 25% rivaroxaban, 3% dabigatran, 10% warfarin. All approved DOACs were allowed - making results more generalizable than AFIRE (rivaroxaban only) or EPIC-CAD (edoxaban only).

Section 4: A Trial Stopped for Harm

The Data Safety Monitoring Board met five times during the trial. On April 16, 2024, they made a dramatic recommendation.

TRIAL STOPPED EARLY

Due to excess deaths from any cause in the aspirin group

What Happened

The trial enrolled 872 of a planned 2000 patients. The DSMB observed:

  • More deaths in the aspirin group
  • More bleeding in the aspirin group
  • No signal of ischemic benefit

Within a month, all patients were contacted to stop aspirin or placebo. Final database lock: May 2025.

Median Follow-up and Treatment Duration

Median Follow-up

2.2

years

Median Treatment

1.7

years

Early stopping implications: When trials stop early, effect sizes can be exaggerated. But the direction of harm - more events with aspirin - is reliable. And this is consistent with prior open-label trials.
Remember: This trial expected aspirin to HELP (HR 0.75). Instead, it found aspirin caused HARM. The signal was so strong the DSMB stopped enrollment.

Section 5: The Results - Aspirin Made Things Worse

Here are the numbers. Pay attention to the direction of every hazard ratio.

Primary Efficacy Outcome

CV death, MI, stroke, systemic embolism, coronary revascularization, or acute limb ischemia

OAC + Aspirin

16.9%

73 of 433 patients

OAC + Placebo

12.1%

53 of 439 patients

Hazard Ratio: 1.53 (95% CI: 1.07 to 2.18)

P = 0.02 - Aspirin INCREASED events by 40%

Death from Any Cause

Aspirin

13.4%

58 patients died

Placebo

8.4%

37 patients died

HR 1.72 (95% CI 1.14-2.58), P = 0.01

Cardiovascular Death

Aspirin

7.6%

33 CV deaths

Placebo

4.3%

19 CV deaths

HR 1.90 (95% CI 1.07-3.35)

Major Bleeding (ISTH Criteria)

Aspirin

10.2%

44 major bleeds

Placebo

3.4%

15 major bleeds

HR 3.35 (95% CI 1.87-6.00), P < 0.001

Every endpoint favored placebo. Aspirin increased the primary composite, increased death, increased CV death, and tripled major bleeding. There was no silver lining.

Section 6: Parsing the Paradox - Where Did the Harm Come From?

If aspirin is supposed to prevent clots, why did it increase cardiovascular events? Let's look at the components.

Atherothrombotic Events (the things aspirin should prevent)

Event Aspirin Placebo HR
All atherothrombotic 10.6% 9.1% 1.27
MI 3.0% 2.1% 1.56
Stroke 1.4% 2.5% 0.57
Stent thrombosis 1 pt 1 pt 0.74

None of these differences were statistically significant

The Critical Finding

Aspirin did NOT reduce atherothrombotic events. The rates were virtually identical (10.6% vs 9.1%, not significant). And stent thrombosis was extremely rare - just 1 event per group.

The theoretical benefit of aspirin - preventing clots - simply did not materialize.

So Where Did the Extra Events Come From?

Bleeding. Aspirin tripled major bleeding (10.2% vs 3.4%).

And bleeding kills. Major bleeding events are associated with:

  • Hypotension and shock
  • Withholding of antithrombotic therapy (increases subsequent clots)
  • Blood transfusions (prothrombotic)
  • Direct mortality from hemorrhagic stroke, GI bleeds, etc.
The mechanism of harm: Aspirin didn't cause heart attacks. It caused bleeding. And bleeding caused death and CV events. The composite endpoint captured this cascade.
Why no stent thrombosis benefit? All patients were >6 months post-stent. By this point, the stent has endothelialized. OAC alone provides adequate protection against late/very late stent thrombosis. Aspirin adds bleeding risk without adding protection.

Section 7: Statistical Considerations - Early Stopping

The trial stopped at 872 patients instead of the planned 2000. What does this mean for interpreting the results?

Planned

2000

patients

Enrolled

872

patients (44%)

What Early Stopping Does to Results

Effect size exaggeration: When you stop a trial because you hit a boundary, the point estimate is often more extreme than the "true" effect. Random high variation triggered the stop.

Wider confidence intervals: Fewer patients = less precision.

Underpowered for subgroups: Individual endpoints (MI alone, stroke alone) may not reach significance.

But the Direction Is Reliable

While the HR of 1.53 might be somewhat inflated, the direction of harm is almost certainly real:

  • Consistent across multiple endpoints (primary, death, bleeding)
  • Consistent with prior open-label trials (AFIRE, EPIC-CAD)
  • Biologically plausible (bleeding mechanism)
  • Independent DSMB recommendation (they saw unblinded data)

Power Considerations

The trial was designed to detect HR 0.75 (aspirin benefit) with 80% power.

Instead they found HR 1.53 (aspirin harm) - the opposite direction with a magnitude even larger than expected benefit.

Even if the true HR is closer to 1.3, the conclusion doesn't change: aspirin hurts these patients.

Don't dismiss early-stopped trials. When a trial stops for harm, that's signal, not noise. The DSMB has access to unblinded data and makes this call to protect patients.
For clinical application: You don't need to know the exact hazard ratio. You need to know the direction. Aspirin + OAC is worse than OAC alone. That's the actionable finding.

Section 8: How AQUATIC Differs from Prior Trials

AQUATIC wasn't the first trial in this space. But it was different in important ways.

Trial Comparison

AQUATIC AFIRE EPIC-CAD
Design Double-blind Open-label Open-label
Population European Japanese Korean
All stented 100% 70% 60%
Prior MI 72% ~35% ~35%
Event rate ~14% ~2% ~1.5%
OAC type All approved Rivaroxaban Edoxaban

Why These Differences Matter

  • Double-blind: Eliminates bias in event ascertainment and reporting
  • European population: Different bleeding/clotting profiles than Asian cohorts; different DOAC dosing
  • Higher risk: 7-8x higher event rates validate that enrichment criteria worked
  • All OACs allowed: Results generalizable to apixaban, rivaroxaban, dabigatran, warfarin

The Convergent Evidence

Despite methodological differences, ALL trials reached the same conclusion:

Adding antiplatelet therapy to OAC in stable patients increases bleeding without reducing ischemic events.

AQUATIC's contribution: It confirmed in a double-blind design, in a Western population, in high-risk stented patients, with any OAC, that the findings from open-label Asian trials are real and generalizable.

Test Your Understanding

Apply what you've learned to interpret AQUATIC correctly.

Key Takeaways

  • 1
    First double-blind trial in this space

    AQUATIC eliminated the bias inherent in prior open-label trials. The findings are more robust.

  • 2
    Trial stopped early for HARM

    The DSMB halted enrollment due to excess deaths in the aspirin group. This is a strong safety signal.

  • 3
    Aspirin INCREASED the primary composite by 40%

    16.9% vs 12.1%, HR 1.53, P=0.02. The opposite of what was expected.

  • 4
    72% more deaths with aspirin

    13.4% vs 8.4%, HR 1.72, P=0.01. This drove the early termination.

  • 5
    3x more major bleeding

    10.2% vs 3.4%, HR 3.35, P<0.001. This is where the harm came from.

  • 6
    NO reduction in atherothrombotic events

    MI, stroke, stent thrombosis rates were similar. Aspirin didn't prevent clots - it just caused bleeding.

  • 7
    Stent thrombosis was RARE (1 per group)

    OAC alone provides adequate protection >6 months post-stent. Aspirin adds risk without benefit.

The Bottom Line

For patients with chronic coronary syndrome on long-term oral anticoagulation who are more than 6 months post-stent, stop the aspirin. Even in high-risk patients, aspirin adds bleeding without preventing ischemic events.

Before You Go: A Quick Poll

Your patient has AF, a stent placed 2 years ago, and is currently on apixaban + aspirin 81mg. Based on AQUATIC, what do you do?