PFO Indications Beyond Stroke

Definition

This page covers the clinical scenarios in which PFO closure is considered in patients who have not had a PFO-associated stroke: migraine, decompression illness in divers, platypnea-orthodeoxia syndrome, thrombophilia, isolated atrial septal aneurysm, systemic (non-cerebral) embolism, transient ischaemic attack, and deep vein thrombosis. None of these populations was included in the six randomised closure trials, so the evidence is observational, indirect, or absent — and every recommendation below is conditional, with all but one resting on very low certainty (sources/pfo-jscai-2022, rating: very high).

Key Concepts

The full recommendation set — and what is behind each one

Migraine — the paradox

• Three RCTs, three failed primary endpoints. MIST (STARFlex), PRIMA (migraine with aura) and PREMIUM (with and without aura) showed a mean difference of 0.59 attacks/month (0.15–1.03) and 1.33 migraine days/month (0.33–2.32) — statistically detectable, clinically marginal, and below each trial's prespecified threshold.
• And yet: complete cessation of migraine in RR 3.46 (0.65–18.40) — a large point estimate with a confidence interval crossing 1. A post-hoc pooled analysis of PRIMA and PREMIUM using the Amplatzer occluder found −3.1 versus −1.9 migraine days (P=0.02), −2.0 versus −1.4 attacks (P=0.01), and complete cessation in 9% versus 0.7% (P=0.001).
• ~48% of patients with migraine with aura have a PFO (Anzola), against 25% of the population, and aura patients may respond better than those without.
• SCAI's resolution: recommend against routine closure, and add a remark that patients "particularly those with debilitating migraines who have failed to benefit from conventional medical therapy, who place a high value on the uncertain benefits … may reasonably choose PFO closure" (sources/pfo-jscai-2022).

Decompression illness in divers

• The mechanism is the cleanest in this list: a right-to-left shunt lets nitrogen bubbles bypass the pulmonary filter and enter the arterial circulation. DCI risk is approximately 5× higher in divers with a PFO.
• But the treatment recommendation is against routine closure — recurrent DCI RR 0.31 (0.08–1.13), from three observational studies comparing closure against conservative diving practices. Non-randomised, small, and non-significant.
• The alternative offered is behavioural: limit high-risk diving, or stop diving. SCAI notes the panel judged this a "reasonable conservative option for most patients," which is a value judgement rather than a trial result.
• Acute DCI requires immediate hyperbaric oxygen therapy — that part is uncontroversial and unrelated to the PFO question.
• For recreational divers without DCI, screening for PFO is not indicated (AHA 2025 sports statement, sources/competitive-sports-aha-2025).

Platypnea-orthodeoxia syndrome

• The one non-stroke indication with a compelling objective signal. Positional dyspnoea and hypoxaemia from right-to-left shunting that increases on sitting or standing; closure raised oxygen saturation by 14.21% (12.18–16.25) and 76% of patients reported improvement in quality of life and symptoms.
• But the cohort experienced AF 9.6%, mortality 8.7% and other procedural complications 8.7% — this is a sick population, often with an enlarged aortic root or a pneumonectomy distorting atrial geometry, and the mortality figure is not attributable to the procedure but is not separable from it either.
• SCAI recommends closure (conditional, very low certainty) — the only "for" recommendation among the eight non-stroke scenarios that rests on a measurable physiological outcome rather than analogy.

Systemic embolism without prior stroke

• Recommended for closure, supported by three patients. No study directly addresses the question; the recommendation rests on a subgroup analysis in which three patients with prior systemic embolism underwent closure and none had a recurrent stroke or TIA.
• The physiological argument is real but incomplete: emboli reach the cerebral circulation before the renal, splenic or peripheral arteries, so systemic events are rarer, the brain is more sensitive to brief ischaemia, and randomised trials in systemic embolism are infeasible for that reason. SCAI therefore extrapolates from the stroke evidence.
• The reasoning is explicit and defensible — the mechanism is identical — but it is a recommendation built on mechanistic similarity and three outcomes.

TIA — why it was excluded

• Both guidelines decline to treat TIA as an index event. CLOSE I included TIA, but ESO excluded it because only one RCT enrolled TIA patients, and the subgroup showed no difference.
• The reason is definitional, not evidential: by definition TIA patients have normal imaging and no persistent deficit, so a vascular cause cannot be established and the event cannot be distinguished from complex migraine, seizure, or other transient neurological symptoms.
• SCAI suggests against closure for TIA without prior stroke, while noting that patients with recurrent, high-probability TIAs may reasonably choose it.
• In practice, many patients labelled TIA with a diffusion-negative MRI and a PFO are treated as stroke — a gap between guideline definitions and clinical behaviour that neither document resolves.

Thrombophilia, DVT and PE

• Thrombophilia without prior stroke: against closure added to antithrombotic therapy, on one fragile cohort with confounding.
• Thrombophilia with prior PFO stroke on antiplatelet therapy: closure is suggested over antiplatelet therapy alone (RR 0.17, 0.07–0.44, from two observational studies). In a 591-patient cohort, thrombophilia raised recurrent event risk (HR 1.85, 1.09–3.16) and closure reduced it (HR 0.16, 0.09–0.30).
• Where lifelong anticoagulation is already required for thrombophilia, DVT or PE, closure is suggested in addition to anticoagulation — the rationale being non-compliance and planned interruptions for procedures rather than an additional protective effect. Every one of these recommendations is very low certainty.
• DVT is found in 7–27% of patients worked up for PFO-associated stroke, is not in any risk score, and early lower-limb Doppler (with pelvic MR venography in selected cases) is recommended by society guidelines despite no outcome data (sources/pfo-jama-2025, rating: high).

AF with ischaemic stroke — the recommendation against closing

• SCAI suggests against routine closure in patients with a history of AF who have had an ischaemic stroke — there is no direct study, and two forces point the same way: occult AF is a likelier explanation for the stroke than the PFO, and closure itself provokes AF (RR 7.33 versus antiplatelet therapy, 86% within 45 days).
• ESO notes that the more likely a patient is to benefit from closure, the less likely they are to develop device-related AF, and vice versa — the same inverse relationship visible across the PASCAL strata.
• The stratified safety data come from the same pooled analysis: the absolute risk increase in atrial fibrillation beyond day 45 with a device was 4.41% (1.02–7.80) in PASCAL "unlikely", 1.53% (0.33–2.72) in "possible" and 0.65% (−0.41 to 1.71) in "probable" (sources/pfo-jama-2021, rating: very high).

Contradictions / Open Questions

• The migraine evidence is internally contradictory and the recommendation reflects a value judgement, not the data. Three randomised trials failed their primary endpoints; a post-hoc pooled analysis of two of them showed a significant benefit on four secondary endpoints including a nine-fold increase in complete cessation. SCAI chose to recommend against routine closure while writing a remark that licenses it — which is honest but leaves the clinician with no discriminating criterion. (sources/pfo-jscai-2022)
• Systemic embolism is recommended for closure on three patients. The mechanism is identical to stroke, which is why the extrapolation is defensible, but the same logic would justify closure for any paradoxical embolic event, and the guideline is explicit that randomised evidence is infeasible. A recommendation that cannot be tested is a recommendation that will never be revised.
• Platypnea-orthodeoxia shows the largest treatment effect and carries the lowest certainty grade. A 14-point rise in oxygen saturation is a physiological outcome, not a clinical one; the cohort had 8.7% mortality and 9.6% AF, and no randomised comparison exists. The recommendation is right and the evidence is still very low — a combination worth noticing.
• Decompression illness is recommended against routine closure while closure is offered as an option — the same structure as migraine. The alternative on offer (dive conservatively or stop) is a behaviour change with no comparative outcome data, so "against closure" here means "for behavioural restriction" without evidence either way.
• TIA is excluded as an index event by definition rather than by evidence, yet TIA patients are routinely treated as stroke patients in practice when imaging is negative and a PFO is present. The gap between the guideline's definitional exclusion and real-world management is unaddressed by either document.
• The thrombophilia recommendations are the weakest in the guideline and the most consequential for anticoagulation decisions — two observational studies, a relative risk of 0.17, and a recommendation to add a device to lifelong anticoagulation. Patients who need lifelong anticoagulation regardless have the least to gain from closure and the same procedural risk.
• Isolated atrial septal aneurysm is recommended against closure on the basis of a single observational study reporting three adverse events, while aneurysm is simultaneously one of the two anatomical features that defines the group with the largest treatment benefit in PASCAL. The apparent inconsistency is resolved by the absence of a prior stroke — but it is not resolved in the text.

Connections

• Part of concepts/Patent-Foramen-Ovale — the parent overview
• Related to concepts/PFO-Closure-Selection — the stroke population, where the evidence actually is
• Related to concepts/PFO-Detection-and-Anatomy — aneurysm and shunt size as the anatomical axis
• Related to concepts/Sports-Cardiology-SDM — diving advice and exercise restriction
• Related to entities/Pulmonary-Embolism — the venous source; PFO as a right-to-left shunt
• Related to entities/Atrial-Fibrillation — the competing mechanism and the procedural complication
• Related to entities/Ischemic-Stroke — where the evidence base comes from

Sources

• sources/pfo-jscai-2022
• sources/pfo-jama-2021
• sources/pfo-eso-2024
• sources/pfo-jama-2025
• sources/competitive-sports-aha-2025