← Back

CLOSE

Patent Foramen Ovale Closure or Anticoagulation vs Antiplatelets after Stroke

Share Share Copied! Compare

Year of Publication: 2017

Authors: Mas, Jean-Luc et al.

Journal: New England Journal of Medicine

Citation: N Engl J Med 2017;377:1011–1021

Link: https://www.nejm.org/doi/full/10.1056/NEJMoa1705915

PDF: https://www.nejm.org/doi/pdf/10.1056/NEJMoa1705915

Clinical Question

In patients with recent cryptogenic stroke and PFO, is PFO closure superior to antiplatelet therapy in preventing recurrent stroke?

Bottom Line

PFO closure significantly reduced recurrent stroke compared with antiplatelet therapy in select patients with cryptogenic stroke.

        Major Points
        One of three landmark 2017 PFO closure RCTs (with RESPECT and REDUCE). Unique design: three-arm trial comparing PFO closure vs antiplatelet therapy vs oral anticoagulation — the only PFO trial to include an anticoagulation arm.
Strictest patient selection among PFO trials: required PFO with either atrial septal aneurysm (ASA, excursion ≥10 mm) OR large interatrial shunt (≥30 microbubbles) — enriching for 'high-risk' PFO anatomy.
Zero recurrent strokes in the PFO closure group (0/238) vs 14 in the antiplatelet group (14/235) over median 5.3 years (HR not calculable, p=0.002). This 0-event result was the most dramatic outcome of any PFO trial.
New-onset atrial fibrillation occurred in 4.6% of closure patients vs 0.9% antiplatelet (p=0.02) — typically peri-procedural and transient, but mandated anticoagulation per protocol.
Anticoagulation vs antiplatelet analysis: no strokes in anticoagulation group (0/187) vs 7 in antiplatelet group (7/174), suggesting anticoagulation may also prevent recurrence in PFO-related stroke.
Devices used: multiple PFO closure devices permitted (not limited to one manufacturer). Most patients received Amplatzer PFO Occluder or Amplatzer Cribriform Occluder.
Conducted in France, Germany, and Switzerland (32 centers). Enrollment was slow (2007–2015, 8 years) due to strict inclusion criteria.
Youngest PFO trial population (mean age 43) with very low vascular risk factor prevalence (hypertension 14%, diabetes 2%) — ideal cryptogenic stroke population.

      

Design

Study Type: Open-label randomized controlled trial

Randomization: 1

Blinding: Open-label

Enrollment Period: 2007–2015

Follow-up Duration: Median 5.3 years

Centers: 32

Countries: France, Germany, Switzerland

Sample Size: 663

Analysis: Intention-to-treat

Inclusion Criteria

Age 16–60 years.
Recent cryptogenic ischemic stroke (within 6 months of randomization).
PFO with associated atrial septal aneurysm (ASA, excursion ≥10 mm) OR large interatrial shunt (≥30 microbubbles on contrast echocardiography).
TEE confirmation of PFO anatomy.
Complete etiological workup excluding other stroke causes: brain MRI, cervical/intracranial vascular imaging, ECG + ≥24h cardiac monitoring, standard blood tests.

Exclusion Criteria

Identifiable cause of stroke (large-vessel atherosclerosis ≥50% stenosis, cardioembolic source, lacunar mechanism, vasculitis, dissection).
Small PFO without ASA and without large shunt (PFO with minimal shunt excluded).
Chronic atrial fibrillation or flutter.
Indication for long-term anticoagulation.
Contraindication to antiplatelet therapy or anticoagulation.
Prior PFO closure attempt.
PFO anatomy unsuitable for percutaneous closure.
Severe disability from index stroke (mRS >3).
Pregnancy or planned pregnancy.

Baseline Characteristics

Characteristic	Control	Active
Mean age ± SD - yr	43.4 ± 10.2	43.3 ± 10.3
Male sex - %	66.4	66.8
Hypertension - %	13.9	14.7
Diabetes mellitus - %	1.8	2.1
Hyperlipidemia - %	13.3	12.6
Current smoker - %	29.6	31.5
Prior stroke or TIA - %	5.4	5.5
Migraine - %	28.6	30.7
Atrial septal aneurysm - %	67.2	66.0
Large shunt (≥30 microbubbles) - %	86.4	87.0
Both ASA and large shunt - %	53.6	53.4

Arms

Field	PFO Closure + Antiplatelet	Control	Oral Anticoagulation
Intervention	Percutaneous PFO closure using any CE-marked device (Amplatzer PFO Occluder, Amplatzer Cribriform Occluder, STARFlex, or other approved devices). Procedure performed under general anesthesia or conscious sedation via femoral venous access with TEE or ICE guidance. Post-procedure: DAPT (aspirin + clopidogrel) for 3 months, then aspirin monotherapy. TEE at 6 months to confirm closure.	Antiplatelet monotherapy at investigator discretion: aspirin 75–300 mg/day (most common) or clopidogrel 75 mg/day. DAPT (aspirin + clopidogrel) was not routinely used. Therapy continued for the entire follow-up period.	Oral anticoagulation with a vitamin K antagonist (warfarin or fluindione, target INR 2.0–3.0). Patients who refused randomization to closure vs antiplatelet could be randomized to anticoagulation vs antiplatelet. This arm was smaller and not the primary comparison.
Duration	Median 5.3 years; device is permanent	Median 5.3 years	Median 5.3 years

Outcomes

Outcome	Type	Control	Intervention	HR / OR / RR	P-value
Recurrent stroke	Primary	14/235 (4.9% KM at 5yr)	0/238 (0% at 5yr)	0.03	<0.001
New-onset atrial fibrillation	Secondary	0%	4.6%		0.02
Device-related complications	Adverse		5.9%
Serious adverse events	Adverse		6.6% in closure group

Subgroup Analysis

ASA vs large shunt only: both subgroups showed 0 strokes in closure arm, but the ASA subgroup had a higher event rate in the antiplatelet arm (3.4% vs 2.0% per 100 patient-years), suggesting ASA may be a stronger risk marker. The 0-event result in the closure arm precluded formal HR calculation in subgroups. Anticoagulation arm: 0 strokes in 187 patients (0/187) vs 7/174 in antiplatelet arm — suggesting anticoagulation may be equally effective to closure for PFO-related stroke prevention. Atrial fibrillation on closure: 4.6% new-onset AF, mostly within first month, suggesting device-related atrial irritation. Patients with both ASA and large shunt (53%) had the highest risk of recurrence on antiplatelet therapy.

Criticisms

Open-label design — patients and physicians knew treatment assignment, potentially influencing care decisions and outcome reporting.
Extremely selective enrollment: required ASA or large shunt, excluding ~60% of PFO patients seen in clinical practice — limits generalizability.
8-year enrollment period (2007–2015) — prolonged accrual raises concerns about evolving diagnostic standards, cardiac monitoring technology, and background medical therapy.
Zero events in closure group prevents HR calculation — while dramatic, it limits statistical precision and makes it impossible to quantify effect size.
The anticoagulation arm had a non-randomized component (patients who refused closure randomization) — introducing selection bias into the anticoagulation comparison.
No mandatory prolonged cardiac monitoring (only ≥24h required) — modern practice uses 30-day or implantable loop monitors, potentially missing occult AF.
Age cap of 60 years — excludes a significant proportion of cryptogenic stroke patients, particularly those where age-related AF prevalence increases.
New-onset AF rate of 4.6% post-closure is concerning — while mostly transient, long-term AF risk post-device implantation remains unclear.
French healthcare system context — access patterns and follow-up intensity may differ from US/UK settings.

Funding

French Ministry of Health, Fondation CNP Assurances

Based on: CLOSE (New England Journal of Medicine, 2017)