Skip to main content
NeuroWiki
Back to Trials

Clinical Question

PFO closure for migraine?

Synthesises 3 trialsThree randomized trials, three missed primary endpoints; every society document that has addressed it recommends against

Clinical Synthesis

Every randomized trial of PFO closure for migraine has missed its primary endpoint, and every society document that has addressed it recommends against the procedure; the encouraging numbers that circulate are all secondary or post-hoc.

Bottom line

Do not offer PFO closure to prevent migraine. Three randomized trials missed their primary endpoint, the pooled analysis that appears to support it re-defined its endpoints and still could not move the conventional responder rate, and every society document that has addressed it recommends against. Two of those are graded guidelines, SCAI 2022 and VA/DoD 2023; the third is a 2025 Asian-Pacific expert consensus whose summit was funded by the device manufacturer, which makes its recommendation against harder to dismiss rather than easier. For a patient with disabling refractory migraine who raises it themselves, SCAI allows that closure may reasonably be chosen after shared decision-making that is explicit about the uncertain benefit.

Sources

Migraine Intervention With STARFlex Technology (MIST) trial· 2008Percutaneous closure of patent foramen ovale in migraine with aura, a randomized controlled trial· 2016Percutaneous Closure of Patent Foramen Ovale in Patients With Migraine· 2017Pooled Analysis of PFO Occluder Device Trials in Patients With PFO and Migraine· 2021PFO closure for migraine (conditional recommendation against)· 2022Recommendation 40 (Strength: Weak against)· 2023Expert Opinion Table 3 (expert consensus, not a graded recommendation)· 2025Pooled efficacy outcomes (5 RCTs + 6 observational, search to 2024-03-12)· 2024

Trials in this question· 3

MIST2008
Sham-controlled STARFlex trial: primary endpoint of migraine cessation was not met.
Primary not met3/74 vs 3/73 (P=0.51)
PRIMA2016
Unblinded Amplatzer trial: primary endpoint of monthly migraine days was not met.
Primary not met2.9 vs 1.7 days (P=0.17)
PREMIUM2017
Sham-controlled Amplatzer trial: efficacy co-primary not met, safety co-primary met.
Efficacy not met38.5% vs 32.0% (P=0.32)

Related questions

PFO closure for cryptogenic stroke?

From the early trials that missed their primary endpoint (CLOSURE-I, PC, RESPECT 2013) to the positive 2017 cluster and DEFENSE-PFO; AF excess is the trade-off

Was this useful?

NeuroWiki is a clinical reference. It does not substitute for your clinical judgment, current guidelines, or your institution's protocol. Verify before acting. Do not enter patient names, MRNs, or dates of birth. Privacy Policy