← Back
NeuroTrials.ai
Neurology Clinical Trial Database

COMPETE

Antithrombotic treatment for migraine in patients with patent foramen ovale: multicentre, randomised, active controlled, open label trial (COMPETE)

Year of Publication: 2026

Authors: Li Z, Wang C, Tang Y, et al.; COMPETE Investigators

Journal: BMJ

Citation: BMJ 2026;394:e100103

Link: https://doi.org/10.1136/bmj-2026-100103

Bottom Line

In patients with migraine and PFO, rivaroxaban was superior to metoprolol for migraine responder rate (78.4% vs 61.8%; absolute difference 16.2%, P<0.001), while aspirin and clopidogrel were non-inferior; no major bleeding occurred.

Major Points

  • First large randomized head-to-head comparison of anticoagulant, antiplatelets, and beta blocker for migraine prevention in PFO (1000 randomized, 39 Chinese centers).
  • Rivaroxaban 20 mg daily produced the highest ≥50% responder rate (78.4%) and was superior to metoprolol; aspirin (61.7%) and clopidogrel (66.8%) were non-inferior but not superior.
  • Benefit of rivaroxaban was consistent across subgroups and especially pronounced in migraine with aura and grade 3 right-to-left shunt — groups thought to have the strongest microembolic mechanism.
  • No major bleeding; 2 SAEs total. GI symptoms clustered with aspirin; bradycardia/hypotension with metoprolol; clopidogrel had fewest drug-related AEs.
  • Open-label PROBE design and a lower-than-Western metoprolol dose (25 mg BID) are important caveats when generalising the rivaroxaban benefit.

Design

Study Type: Randomized Controlled Trial (investigator-initiated, active-controlled, open-label with blinded endpoint adjudication; hypothesis-generating)

Randomization: 1

Blinding: Open-label with blinded outcome assessors (PROBE design)

Enrollment Period: Registered NCT05546320

Follow-up Duration: 12 weeks of treatment (post 12-week screening)

Centers: 39

Countries: China

Sample Size: 1000

Analysis: Full analysis set (n=984); hierarchical hypothesis testing (non-inferiority → superiority → pairwise); LOCF and multiple imputation sensitivity analyses


Inclusion Criteria

  • Age 18-64 years
  • Migraine (with or without aura) diagnosed by ICHD-3 criteria for >1 year
  • ≥4 migraine days per month on average
  • PFO confirmed by transthoracic or transesophageal echocardiography
  • Right-to-left shunt verified by contrast echocardiography or transcranial Doppler
  • Completed 12-week prospective screening/diary period

Exclusion Criteria

  • Prior transient ischaemic attack, stroke, or intracranial haemorrhage
  • Right-to-left shunt from a source other than PFO
  • Contraindications to antiplatelet, anticoagulant, or beta blocker treatment
  • Previous use of any of the four investigational medicinal products
  • Any condition rendering the patient unsuitable per investigator

Baseline Characteristics

Overall:

  • Median Age (years): 38.5
  • Sex - Female: 75.1% (739/984)
  • Migraine with aura: 23.8% (234)
  • Right-to-left shunt grade 3 (>30 microbubbles): 59.6% (586)
  • Right-to-left shunt grade 2 (10-30 microbubbles): 23.7% (233)
  • Right-to-left shunt grade 1 (1-9 microbubbles): 16.8% (165)
  • Snoring: 7.9% (78)
  • Hypertension: 6.8% (67)
  • Smoking: 6.3% (62)
  • Prior preventive migraine treatment (non-IMP): 42.0% (413)
  • Transesophageal echocardiography performed: 13.2% (130)

Arms

FieldAspirinClopidogrelRivaroxabanControl
InterventionAspirin 300 mg once dailyClopidogrel 75 mg once dailyRivaroxaban 20 mg once dailyMetoprolol 25 mg twice daily
Duration12 weeks12 weeks12 weeks12 weeks

Outcomes

OutcomeTypeControlInterventionHR / OR / RRP-value
Proportion of participants achieving ≥50% reduction in monthly migraine days or attacks from baseline to weeks 9-12PrimaryMetoprolol: 61.8% (144/233) · Aspirin: 61.7% (148/240) — non-inferior vs metoprolol · Clopidogrel: 66.8% (157/235) — non-inferior vs metoprolol · Rivaroxaban: 78.4% (185/236) — non-inferior AND superior vs metoprolol (difference 16.2%, 98.33% CI 6.0-26.4; P<0.001)
Reduction in monthly migraine days (rivaroxaban vs metoprolol)SecondaryEffect: Difference 1.0 (95% CI 0-2.0)
Reduction in monthly migraine attacks (rivaroxaban vs metoprolol)SecondaryEffect: Difference 1.0 (95% CI 0-2.0)
Reduction rate in migraine days (rivaroxaban vs metoprolol)SecondaryEffect: Δ 12.5% (95% CI 3.3-20.0)
Complete migraine cessation at 12 weeks (rivaroxaban vs metoprolol)SecondaryEffect: Δ 12.5% (95% CI 4.1-20.8)
MSQ role-restrictive change (rivaroxaban vs metoprolol)SecondaryEffect: Difference 6.4 (95% CI 2.9-11.4)
MSQ total score change (rivaroxaban vs metoprolol)SecondaryEffect: Difference 18.7 (95% CI 7.9-30.0)
Rivaroxaban vs aspirin (responder rate)SecondaryEffect: Δ 16.6% (6.7-26.4)
Rivaroxaban vs clopidogrel (responder rate)SecondaryEffect: Δ 11.6% (1.6-21.6)
Clopidogrel vs aspirin (responder rate)SecondaryEffect: Δ 4.9% (-5.7 to 15.5), NS
Any adverse event (participants)AdverseMetoprolol 43 (17.3%) | Aspirin 60 (24.3%) | Clopidogrel 34 (13.9%) | Rivaroxaban 53 (21.6%)
IMP-related adverse eventsAdverseMetoprolol — | Aspirin 48 (19.4%) | Clopidogrel 23 (9.4%) | Rivaroxaban —
Serious adverse eventsAdverse2 total — 1 corpus luteum rupture/pelvic haemorrhage (rivaroxaban, IMP-related, resolved); 1 acute cholecystitis (metoprolol, unrelated, recovered)
Major bleedingAdverse0 in all four groups
Discontinuation due to AEAdverseMetoprolol 7 | Aspirin 0 | Clopidogrel 1 | Rivaroxaban 2
Gastrointestinal symptomsAdverseAspirin 29 (11.7%) — highest
Bradycardia (metoprolol)Adverse9 (3.6%)
Hypotension (metoprolol)Adverse17 (6.9%)

Subgroup Analysis

Rivaroxaban vs metoprolol responder-rate benefit largest in participants with migraine aura (Δ 27.3%, 98.33% CI 5.3-49.4) and grade 3 right-to-left shunt (Δ 18.8%, 5.8-31.8); direction of benefit consistent across other subgroups.


Criticisms

  • Open-label design (PROBE) and self-reported headache diaries create risk of reporting bias despite blinded endpoint adjudication.
  • Metoprolol dose (25 mg twice daily) at the lower end of the effective range — may have underestimated the active-control effect and magnified rivaroxaban's apparent superiority.
  • 12-week treatment period is too short to assess long-term efficacy or cumulative bleeding risk of chronic anticoagulation in a young migraine population.
  • Enrolment restricted to Chinese centres, predominantly East Asian participants, limiting generalisability; East Asian pharmacogenomics (CYP2D6) may affect metoprolol exposure.
  • Only ~24% of participants had migraine with aura — the subgroup most mechanistically linked to PFO; subgroup analyses underpowered.
  • Framed as 'hypothesis generating' with hierarchical testing rather than a confirmatory pivotal trial; requires replication before rivaroxaban can be recommended for migraine prevention.

Funding

CAMS Innovation Fund for Medical Sciences; National Natural Science Foundation of China; Noncommunicable Chronic Diseases-National Science and Technology Major Project; National High Level Hospital Clinical Research Funding; National Key R&D Programme of China; CAMS Clinical and Translational Medicine Research Fund; Yunnan Pan Xiangbin Expert Workstation. No industry funding declared.

Based on: COMPETE (BMJ, 2026)

Authors: Li Z, Wang C, Tang Y, et al.; COMPETE Investigators

Citation: BMJ 2026;394:e100103

Content summarized and formatted by NeuroTrials.ai.