By Jeffrey Zimmet, MD, PhD

Associate Professor of Medicine, University of California, San Francisco; Director, Cardiac Catheterization Laboratory, San Francisco VA Medical Center

Dr. Zimmet reports no financial relationships relevant to this field of study.

SYNOPSIS: After years of uncertainty, three large randomized trials have shown a benefit to patent foramen ovale closure in reducing recurrence after cryptogenic stroke in the right patients.

SOURCES: Saver JL, Carroll JD, Thaler DE, et al. Long-term outcomes of patent foramen ovale closure or medical therapy after stroke. N Engl J Med 2017;377:1022-1032.

Mas JL, Derumeaux G, Guillon B, et al. Patent foramen ovale closure or anticoagulation vs. antiplatelets after stroke. N Engl J Med 2017;377:1011-1021.

Søndergaard L, Kasner SE, Rhodes JF, et al. Patent foramen ovale closure or antiplatelet therapy for cryptogenic stroke. N Engl J Med 2017;377:1033-1042.

Patent foramen ovale (PFO) is a common condition in the general population, with autopsy studies estimating a prevalence of more than 25%. Prior studies of contrast echocardiography diagnosing PFO placed it between roughly 10% and 30% of patients. Retrospective studies have long drawn an association between PFO and cryptogenic stroke, particularly in patients with larger shunts or with atrial septal aneurysm (ASA), but until very recently randomized evidence supporting PFO closure in these patients has been lacking. The first of three negative contemporary trials, CLOSURE I in 2012 found a clearly negative result with a device known as the STARFlex. Although it has been noted that this trial included lower-risk patients, as well as patients with lacunar strokes who would not be expected to benefit from PFO closure, this information could not prevent the downfall of its parent company. Two subsequent trials published the following year, the PC trial and the RESPECT trial, both used the Amplatzer PFO Occluder. The PC trial investigators randomly assigned cryptogenic stroke patients to PFO closure or to medical therapy, which likewise failed to show a significant treatment effect of PFO intervention.

The results of the largest of these studies, RESPECT, were more nuanced. Enrollment was much slower than expected, taking eight years to complete. Patients in the medical therapy arm were more likely to withdraw from the study (in many cases to seek intervention). In the intention-to-treat analysis, although there were half as many strokes in the closure arm compared with medical therapy, this difference did not meet statistical significance (P = 0.08). More than four years later, Saver et al conducted an analysis of the extended follow-up from this trial, increasing the median duration of follow-up to 5.9 years (vs. 2.2 years in the original trial). During that period, the number of patients experiencing recurrent stroke increased to 18 in the closure group and 28 in the medical therapy group, yielding the all-important significant P value for the difference between groups (P = 0.046). The authors reported a relative risk reduction of 35% and a number needed to treat of 43 to prevent one recurrent stroke at five years.

The results of the CLOSE trial and the Gore REDUCE trial also showed that PFO closure resulted in reduced rates of recurrent stroke compared to medical therapy. As an investigator-initiated study, the CLOSE trial was about the strategy of PFO closure rather than a particular device. Eleven different devices ultimately were used, which reflects the greater availability of devices outside the United States. More important was the stringent inclusion criteria of this trial, which required either an ASA or a large interatrial shunt to qualify, presumably to increase the odds that the PFO was causative in the original event. Strikingly, CLOSE reported no recurrent strokes at all in the PFO closure group, whereas stroke occurred in 14 of the 235 patients in the antiplatelet therapy group (P < 0.001).

In terms of patient selection, REDUCE represents a middle ground between CLOSE and earlier trials. Although > 80% of patients in each group in REDUCE presented with moderate or large shunts, small shunts also were included, and such enrollees were not required to exhibit other risk factors such as ASA. Nonetheless, closure was associated with a lower likelihood of recurrent stroke, with a relative risk of 0.51 (95% confidence interval, 0.29-0.91; P = 0.04). In each trial, procedural success rates were high, while serious procedural complications were uncommon. As in previous reports, the procedure was associated with a small but significant increase in episodes of atrial fibrillation, which in most cases were self-limited. In these three randomized trials comparing interventional to medical therapy for PFO in the setting of cryptogenic stroke, the primary conclusions were similar. In younger patients with cryptogenic stroke and PFO, closure of the PFO was associated with lower rates of recurrent stroke compared with medical therapy consisting primarily of antiplatelet medication.


After so many years without clear supportive data, there are three positive, independently performed, randomized, controlled trials, which truly represent a tipping point for PFO closure in cryptogenic stroke. Before applying these data to average stroke patients, it is worthwhile to review several key points.

First, the patients in these trials were not all-comers with stroke. The patients included in these trials were younger overall, with average ages in the low-to-mid-40s. None of the trials enrolled patients > 60 years of age. Cryptogenic stroke is defined primarily by what is not found, rather than by what is present.

This means that the event is not the result of small-vessel occlusive disease (a lacunar stroke, which is recognized as unlikely to be of embolic origin), and that imaging has excluded proximal arterial stenosis and alternate cardioembolic sources, including atrial fibrillation. Before being considered for PFO closure, the average patient should undergo not only basic brain imaging for stroke but also MRI or CT angiography of the intracranial arteries, cervical arteries, aortic arch, TEE, and screening for atrial fibrillation.

Most strokes are not caused by PFO, and PFOs are very common. The more risk factors a patient exhibits for run-of-the-mill atherosclerotic or embolic stroke, the less likely it is that closing a PFO will be effective. On the other hand, it should be clear from this discussion that not all PFOs are created equally regarding magnitude of the interatrial shunt, presence of atrial septal aneurysm, and their potential role in the etiology of a stroke.

Some have argued that young stroke patients undergoing PFO with a large shunt and otherwise negative complete screening no longer should be labeled as cryptogenic and should be treated accordingly.

Based on the available evidence, PFO closure is now a viable option for U.S. patients who have suffered a cryptogenic stroke. It is essential that potential candidates are screened appropriately so that patients who are most likely to benefit are targeted for therapy.