European Stroke Organisation (ESO) Guidelines on the Diagnosis and Management of Patent Foramen Ovale (PFO) after Stroke

Authors, Journal, Affiliations, Type, DOI

• Authors: Valeria Caso (Perugia) and Christian Pristipino (Institut Coeur Paris Centre) as co-chairs, with Guillaume Turc (GHU Paris Psychiatrie et Neurosciences), Azmil H Abdul-Rahim (Liverpool), Pedro Castro (Porto), Heinrich Mattle (Bern), Eleni Korompoki (Athens), Lars Søndergaard (Abbott Structural Heart), Danilo Toni (Sapienza, Rome), Silke Walter (Saarland), and ESO methodologists Salman Hussain and Avtar Lal
• Affiliations: ESO Module Working Group — 7 stroke specialists, 2 cardiologists, 2 methodologists; one author (Søndergaard) is employed by a device manufacturer
• Journal: European Stroke Journal. 2024;9(4):800–834; received 23 Jan 2024, accepted 2 Apr 2024; correction issued June 2024 to fix table placement
• Type: Clinical practice guideline using GRADE, following the ESO standard operating procedure; three overarching themes (diagnosis, treatment of PFO-associated stroke, post-procedural management) across six PICO questions
• DOI: https://doi.org/10.1177/23969873241247978

Overview

The neurology-side guideline, and the more sceptical of the two current guideline documents. Its structural contribution is to insist on a distinction that every earlier document blurred: a PFO found in a patient with a cryptogenic stroke is not the same thing as a PFO that caused the stroke. Everything that follows — the PASCAL classification, the recommendation against closure in "unlikely" patients, the downgrading of anticoagulation — depends on applying that distinction before intervening.

Its second contribution is an admission. There is no gold standard for PFO detection. Contrast TOE, the conventional reference against which every other test has been validated, has a weighted sensitivity of only 89% when checked against autopsy, cardiac surgery or catheterisation. So the guideline issues no evidence-based recommendation on diagnosis at all, only expert consensus, and states plainly that the evidence for closure in patients over 60 and under 18 is insufficient to support a recommendation in either direction.

The recommendations themselves are conventional in direction and unusually honest about certainty: high-quality evidence for closure plus antiplatelet therapy in selected patients aged 18–60; moderate quality for closure specifically in PASCAL "possible" or "probable" patients; low-quality evidence against closure in "unlikely" patients.

Keywords

Patent foramen ovale · stroke · atrial fibrillation · PFO closure · anticoagulation · antiplatelets

Key Takeaways

Epidemiology and the scale of the problem

• PFO is causally involved in 10% of all ischaemic strokes in patients aged 18–60 despite a 25% population prevalence — most carriers never have an event.
• Recurrence on medical therapy is low: approximately 1 event per 100 person-years, corresponding to a cumulative incidence of 4.6% at 3.8 years across the randomised trials. Any intervention is therefore operating on a small absolute risk.
• One-third of ischaemic strokes are cryptogenic, and the proportion is higher in younger patients.
• Three proposed mechanisms: paradoxical embolism of a venous clot across the PFO; in situ thrombus formation within the PFO tunnel; and atrial arrhythmia from disruption of interatrial electrical signalling.

Diagnosis (PICO 1.1–1.3) — no evidence-based recommendation

• Definitions of a large shunt differ by modality: >20 microbubbles in the left atrium within 3 cardiac cycles at c-TTE or c-TOE; >10 high-intensity transient signals at c-TCD. An atrial septal aneurysm is ≥10 mm excursion from the septal plane.
• No study compared c-TCD with c-TTE directly using c-TOE as reference — the third PICO question could not be answered at all.
• Alternatives are worse: cardiac MRI detected a shunt in 29.3% where TOE found 53.7% (P<0.001); single-phase cardiac CT had 85.3% sensitivity but 33.3% specificity and a negative predictive value of 33%; full-cycle CT was better (89.4%/92.3%).
• Expert consensus (not evidence-based): locally agreed algorithms using whatever is available; c-TCD when c-TTE and/or c-TOE is inconclusive (8/9); TCD has superior sensitivity to TTE for screening (6/9); TOE after a positive RLS screen to define anatomy (9/9).
• Adding blood to agitated saline increases c-TCD sensitivity, up to 100% in one comparison; the optimal blood volume is unsettled.

Closure versus antiplatelet therapy in patients aged 18–60 (PICO 2)

• The pivotal evidence: individual-patient-data meta-analysis of 6 RCTs (SCOPE; 1,882 device vs 1,846 medical) — annualised stroke 0.47% vs 1.09%, adjusted HR 0.41 (0.28–0.60). All-cause mortality did not differ (RR 0.98).
• Closure versus antiplatelet therapy alone, restricted to the two trials that used antiplatelet-only comparators (Gore REDUCE, CLOSE): HR 0.18 (0.05–0.63) — but most of those patients had a large shunt, an aneurysm, or both.
• Closure versus any antithrombotic therapy: HR 0.38 (0.19–0.77).
• The benefit is concentrated, not average. RoPE ≥7: HR 0.21 (0.11–0.42); RoPE <7: HR 0.61 (0.37–1.00); P for interaction 0.02. PASCAL "probable" HR 0.10 (0.03–0.35), "possible" 0.38 (0.22–0.65), "unlikely" 1.14 (0.53–2.46).
• Absolute benefit is small in the aggregate and largest in a narrow group. Two-year ARR 2.1% in both "possible" and "probable" (NNT 47). For large shunt plus atrial septal aneurysm: 2-year ARR 5.5% (2.7–8.3), 5-year 7.1% (3.7–10.6), NNT 14 (9–27).
• Harm is concentrated in the patients who do not benefit. Post-45-day atrial fibrillation absolute risk increase: 4.41% (1.02–7.80) in PASCAL "unlikely", 1.53% in "possible", 0.65% in "probable".
• Major bleeding: access-site or retroperitoneal haemorrhage 1%, pericardial tamponade 0.17%, cardiac perforation 0.06%.
• Harm and benefit therefore point the same way: the patients with most to gain have least to lose.

Patient selection — RoPE and PASCAL

• PASCAL was not an inclusion criterion or stratification variable in any of the six RCTs and has not been prospectively validated — which is why the working group rated its evidence down by one level.

Closure versus anticoagulation (PICO 3) and anticoagulation versus antiplatelet therapy (PICO 4)

• Closure plus antiplatelet therapy is suggested over long-term anticoagulation in PASCAL possible/probable, aged 18–60 — weak for, low quality — on the basis of the pivotal trials plus the cumulative major bleeding risk of anticoagulation.
• The head-to-head evidence is one post-hoc analysis of the 3-arm CLOSE trial: 0 strokes in 238 closed patients over 963 patient-years versus 3 in 187 anticoagulated over 967 patient-years (HR 0.14, 0.00–1.45, P=0.08) — underpowered.
• Anticoagulation versus antiplatelet therapy: no recommendation. Pooled anticoagulants had lower stroke recurrence (RR 0.76, 0.43–1.32) but more major bleeding (RR 1.63, 0.77–3.43), with no difference in the composite of stroke/TIA/death (RR 0.93, 0.41–2.09).
• Expert consensus: an individualised choice for patients who decline closure, balancing lower stroke recurrence against bleeding risk and the patient's preference (9/9).
• Long-term anticoagulation is not recommended in PFO-associated stroke unless indicated for another reason.

Post-procedural management (PICO 5 and 6)

• Antiplatelet therapy after closure is mechanistic: the coagulation system is part of normal device sealing, and complete endothelialisation takes 6 months in most patients but may take up to 5 years. Minor cerebrovascular events occur in 1–2% within the first 4 weeks.
• Dual antiplatelet therapy then single antiplatelet therapy is suggested — but based on trial protocol rather than comparative evidence. The only two available observational studies found no significant difference between DAPT and SAPT for either major bleeding or myocardial infarction.
• Expert consensus: DAPT for 1–6 months, then single antiplatelet therapy long term, for at least 5 years after closure (9/9).
• No recommendation on duration of single antiplatelet therapy — no evidence exists.
• Prolonged cardiac monitoring (PICO 6): no recommendation. New-onset AF is 4.3 times more frequent after closure; 46% of AF events are transient and occur within 45 days; beyond day 45 there was no excess AF risk in the IPD meta-analysis.
• Pooled AF detection with long-term monitoring after closure: 10% (5–15) across five observational studies; no recurrent stroke recorded in the monitored cohort.
• Determinants of post-closure AF: age (aOR 1.67 per 10 years), left disc diameter 25 mm versus smaller (aOR 2.67), and male sex (aOR 4.78).
• Expert consensus on monitoring: 24-hour telemetry or Holter before closure in patients under 55 (8/9); ILR for patients over 60 (7/9); continue monitoring to end-of-life in anyone already implanted (8/9); against systematic device implantation after closure (8/9); ILR for recurrent stroke after closure with negative short-term monitoring (8/9).

Groups outside the evidence

• Over 60: only DEFENSE-PFO randomised any — 13 to closure, 21 to medical therapy. No recurrent stroke or TIA after closure, 4 events in controls (P=0.07); HR 0.14 (0.005–3.57). No evidence-based recommendation. Expert consensus: enrol in trials or a registry; otherwise use PASCAL plus clinical judgement (8/9).
• Under 18: CLOSE allowed 16–17, and enrolled exactly one patient under 18. No evidence-based recommendation. Expert consensus: closure in selected patients aged 13–17 according to PFO anatomy (9/9).
• Timing: closure within 6 months of the index stroke, as soon as possible given lesion size and risk profile (9/9). No trial was powered to test very early closure.
• TIA was not treated as an index event in this guideline, because only one RCT included TIA patients; that subgroup showed no difference.

Limitations of the document

• Enormous reliance on expert consensus. Almost every recommendation outside the 18–60 closure question is either conditional with very low certainty or replaced by a consensus statement. The diagnosis section has no evidence-based recommendation at all.
• The conventional reference standard is itself unvalidated. Every diagnostic estimate in the document is measured against TOE, whose own sensitivity is 89% against invasive or post-mortem truth. The pooled sensitivities for c-TCD and c-TTE are therefore relative to a flawed comparator, which the guideline acknowledges but cannot correct.
• One author is employed by a device manufacturer (Søndergaard, Abbott Structural Heart, Medical Affairs) — disclosed, and the guideline declares no conflicts of interest, but it is a conflict of the kind GIN standards are designed to surface.
• PASCAL is used to drive recommendations without having been validated prospectively or used as a stratification variable in any trial — the guideline says so explicitly and rates the evidence down, but the recommendations still turn on it.
• TIA is excluded as an index event, and pregnancy is excluded entirely. The Discussion concedes that PFO-related stroke in pregnancy "necessitates dedicated pathways" and is not addressed, leaving a common clinical scenario with no guidance.
• The age boundaries are arbitrary in practice. 18–60 is the trial envelope, not a biological threshold; a 61-year-old with a large shunt and an aneurysm has better odds than a 55-year-old without either, and only one of them is covered by the recommendation.
• The guideline does not adjudicate which antiplatelet agent to use long term, because only one trial compared devices against investigator-chosen therapy.
• Post-procedural antithrombotic strategy rests on trial protocol rather than comparison — no RCT has tested DAPT versus SAPT after closure, and the two observational studies available found nothing.
• No cost-effectiveness analysis, despite closure being recommended for a population in which the absolute risk reduction is 2% at two years and the NNT is 47 for the average PASCAL patient.

Key Concepts Mentioned

concepts/Patent-Foramen-Ovale · concepts/PFO-Closure-Selection · concepts/PFO-Detection-and-Anatomy · concepts/PFO-Non-Stroke-Indications · concepts/Poststroke-AF-Monitoring · concepts/mWHO-Classification

Key Entities Mentioned

entities/Ischemic-Stroke · entities/Atrial-Fibrillation · entities/Pulmonary-Embolism · concepts/Atrial-Septal-Defect

Named but unlinked: RoPE · PASCAL · SCOPE · CLOSE · RESPECT · REDUCE · DEFENSE-PFO · CLOSURE I · PC Trial · Amplatzer PFO Occluder · Gore Cardioform · ILR · c-TCD · c-TTE · c-TOE

Wiki Pages Updated

• sources/pfo-eso-2024 — created
• concepts/Patent-Foramen-Ovale — created
• concepts/PFO-Closure-Selection — created
• concepts/PFO-Detection-and-Anatomy — created
• concepts/PFO-Non-Stroke-Indications — created
• entities/Ischemic-Stroke — cryptogenic stroke and ESUS
• concepts/Poststroke-AF-Monitoring — AF before and after closure
• concepts/Intracardiac-Shunts — PFO as the commonest intracardiac shunt
• entities/Pulmonary-Embolism — PFO as a right-to-left shunt
• concepts/Atrial-Septal-Defect — PFO is not an ASD
• wiki/wikiindex · wiki/sourceindex