A patent foramen ovale (PFO) is a remnant of the normal fetal circulation that can persist into adulthood. It was first described in 1564 by Leonardi Botali. The clinical significance of PFOs remained poorly understood for a long period of time. The advent of echocardiography allowed for the identification of PFO as a risk factor for several clinical syndromes, including ischemic stroke, myocardial infarction, decompression sickness in divers due to paradoxical embolism,1-3) migraines, and platypnea-orthodeoxia syndrome.4) PFOs are now amenable to interventional percutaneous therapy, and multiple novel technologies are either available or under development for lesion closure. This paper reviews the current knowledge of the relationship between PFOs and cryptogenic strokes and summarizes current therapeutic options.

PFOs are present in about a quarter of the adult population. The prevalence may decline with age. However, PFOs increase in size with age. This trend may reflect size selection as larger defects remain patent while smaller defects close spontaneously.5)

PFOs are residual, oblique, slit-shaped defects resembling tunnels, which normally close secondary to the formation of fibrous adhesions between the septum primum and secundum during the first months of life. These adhesions help to form a valve-like structure, with the "door-jam" located in the left atrium (LA) side of the atrial septum.5)6) After birth, the increase in pulmonary blood flow and rising LA pressure push the septum primum rightward against the septum secundum, shutting the flap of the PFO. Flap fusion is complete by age two years in 70 to 75 percent of children, with the remaining 25 to 30 percent having a PFO. When right atrium (RA) pressure rises intermittently with Valsalva or other isometric strain, the leaflets of the PFO may separate, resulting in leftward excursion of the septum primum and permitting flow from the RA to the LA (Fig. 1).

PFOs are associated with other cardiac anomalies, including atrial septal aneurysms (ASAs) and Chiari networks.7) ASAs consist of redundant congenital atrial septal tissue in the region of the fossa ovalis, bulging into the RA or LA during respiration. The prevalence of ASAs is 1% in autopsybased studies, a number differing from echocardiographic studies.7)8) One adult study showed that 33% of patients with ASAs also had PFOs, although 32% had isolated ASAs.9) The odds of having a PFO are 4.6 times greater in patients with ASAs than in those without ASAs. ASAs are more frequent in stroke patients, but they are also more common in patients with PFOs, making cause and effect uncertain.

Cryptogenic stroke refers to an ischemic cerebrovascular accident (CVA) that occurs in the absence of significant risk factors or clear cause. The majority of ischemic strokes are due to cardioembolism, large vessel atherothromboembolism, small vessel occlusive disease, or other mechanisms. Cryptogenic strokes account for 30 to 40 percent of all ischemic strokes.10) The risk factors for cryptogenic stroke are not clearly different from the risk factors for other types of ischemic stroke, although hypertension may be less common in patients with cryptogenic stroke compared to those with other types of ischemic strokes. The pathophysiology of cryptogenic stroke is likely heterogeneous, and proposed mechanisms include paradoxical embolism from atrial septal abnormalities such as PFOs, occult cardiac embolism secondary to aortic atheromatous disease or other cardiac sources, hypercoagulable states, preclinical or subclinical cerebrovascular disease, and inflammatory processes.

The relationship between paradoxical (right-to-left) embolism through a PFO and stroke remains controversial because of variability in the reported stroke risk in patients with PFOs.2)11-14) PFO itself may be an innocent bystander or etiologic mechanism involved in paradoxical embolism, especially in patients younger than 55 years of age.15) The clinical diagnosis of paradoxical embolism is presumptive and based on the absence of a left-sided thromboembolic source, a right-to-left shunt, and optionally, the detection of thrombus in the venous system or right heart chambers. The proposed CVA mechanism entails paradoxical embolism, with passage of thrombi from the peripheral venous system to the left cardiac cavities through a septal defect after Valsalva maneuver. Thrombus formation in the atria as a consequence of possible atrial arrhythmias is often associated with PFOs.16) Thrombus formation within the PFO tunnel secondary to blood stasis inside the tunnel and hypercoagulability is associated with PFO.17)18)

A number of case-control studies have reported an increased prevalence of PFO and ASA in patients with a history of cryptogenic stroke.8)9)19-24) A meta-analysis of these case-control studies found that the presence of a PFO, ASA, or both was significantly associated with ischemic stroke in patients <55 years of age {odds ratios (OR) 3.1, 6.1, and 15.6, respectively}.15) By comparison, the association in patients over the age of 55 was less certain (OR 1.3, 3.4, and 5.1, respectively). However, a recent prospective analysis showed that concurrence of PFO and ASA is a high-risk characteristic in older patients, as well.25)

The Patent Foramen Ovale and Cryptogenic Stroke Study (PICSS) prospectively followed 630 patients.26) The patients were randomly assigned to receive either aspirin (325 mg/day) or warfarin {mean international normalized ratio (INR) 2.04}. In this cohort, 265 patients (42%) had experienced a cryptogenic stroke in the past; on transesophageal echocardiography (TEE), the patients with a history of cryptogenic stroke had a higher rate of PFO than those with a known cause of stroke (39 versus 29%) and a significantly higher rate of large PFO (20 versus 9.7%). The overall incidence of ASA was 11.5%, but the incidence in patients with PFOs was not given. Among all patients in PICSS, no association was found between the presence of PFO alone and recurrent stroke or death. There was also no association between concurrent PFO and ASA and recurrent stroke or death. Among those patients with cryptogenic stroke, the two-year risk of recurrent stroke or death was not significantly different between those with PFOs (14.3%) and those without PFOs (12.5%). Moreover, there was no evidence of reduction in risk among those cryptogenic stroke patients with PFOs assigned to the warfarin group (hazard ratio 0.52, 95% CI 0.16-1.67).

The Northern Manhattan Stroke Study (NOMAS) prospectively evaluated 1,100 stroke-free subjects 40 years of age or older (mean age 69 years) from Manhattan using transthoracic echocardiography to look for the presence of PFOs and ASAs.27) PFOs were detected in 14.9 percent of subjects, and ASAs were detected in 2.5 percent. During a mean follow-up period of 80 months, ischemic strokes occurred in 68 subjects. PFO was associated with a statistically insignificant increase in stroke risk {hazard ratio (HR) 1.64, 95% CI 0.87-3.09}, as was the coexistence of PFO and ASA (HR 1.25, 95% CI 0.17-9.24).

The Stroke Prevention: Assessment of Risk in a Community (SPARC) study prospectively evaluated 588 randomly sampled subjects 45 years of age or older from Olmsted County, Minnesota, using TEE.11) PFOs were present in 24.3 percent of subjects, and ASAs were present in 1.9 percent. During a mean follow-up period of 5.1 years, cerebrovascular events {transient ischemic attack (TIA), cerebral infarction, or death related to cerebral infarction} occurred in 41 patients. PFO was not a significant independent predictor of cerebrovascular events after adjustment for age and comorbid conditions (HR 1.46; 95% CI 0.74-2.88). Furthermore, there was no association between PFO size and risk of cerebrovascular events.

To summarize all the data, multiple case-control trials have found an association between cryptogenic stroke and the presence of PFO. However, population-based cohort studies have found no statistically significant association between the risk of first stroke and presence of PFO.

Treatment options vary and include medical options (antiplatelet agents or anticoagulants) and invasive methods (surgical treatment in the past and transcatheter device closure in recent years). The clinical course of a patient with a PFO who has a history of stroke or TIA and who does not receive treatment is unknown.

The goals of medical therapy are to prevent recurrence of an event once a sentinel event has already occurred or to prevent the initial occurrence of an event. Evaluation of the results of medical therapy is complicated for several reasons. First, the strength of the relationship between PFO occurrence and the central nervous system event must be established. If the central nervous system event is from another cause such as ascending aortic atheromatous debris, then medical therapy or surgical or percutaneous closure of the PFO will obviously be ineffective. Second, the specific anatomic pathology (e.g., the presence of PFO versus presence of ASA versus a combination) and the size of the PFO and shunt must be established at rest. The effectiveness of specific medical therapy (e.g., aspirin versus warfarin) must also be investigated.

The rationale for aspirin therapy comes from clinical data and from evidence suggesting that the paradoxical particles associated with transient ischemic events are small platelet/fibrin aggregates.28) PFOs are treated with aspirin (300 mg/day) for secondary prevention of stroke or TIA among young patients after a single first event. At four years, aspirin therapy did not improve the frequency of recurrent cerebrovascular events in high-risk patients, such as those with septal abnormalities.29) The efficacy of aspirin therapy is suggested by the French PFO-ASA study, which found that among 216 patients with a history of cryptogenic stroke who had a PFO alone, the incidence of recurrent stroke on aspirin therapy was only 2.3 percent after four years, a value comparable to the 4.2 percent risk seen in patients with neither PFOs nor ASAs.29) Support for the use of aspirin also comes from the PICCS study, which did not demonstrate a statistically significant difference between the effects of aspirin and those of warfarin on the risk of subsequent stroke or death among patients with cryptogenic stroke and PFO.26) However, PICSS was primarily designed as a prognostic study and was underpowered to demonstrate a treatment effect.

The American College of Chest Physicians Conference on Antithrombotic and Thrombolytic Therapy and the American Heart Association (AHA)/American Stroke Association Council on Stroke Practice Guidelines recommend antiplatelet therapy after the occurrence of cryptogenic stroke in the majority of patients. Warfarin therapy is suggested in the setting of known deep venous thrombosis or documented hypercoagulability (Table 1).30)31)

Despite medical therapy, up to 25% of patients with a history of cryptogenic stroke experience recurrent stroke or TIA within 4 years of the initial event.25)28) This has led a number of expert clinicians to conclude that mechanical closure should be the primary treatment modality for PFO patients after cryptogenic stroke. The treatment of symptomatic patients with PFOs who have failed medical therapy is either surgical or percutaneous closure with permanently implanted closure devices.

The reported efficacy of surgical closure in PFO patients with prior cerebrovascular ischemic events is variable,32-35) and randomized trials comparing medical therapy have not been performed. Surgical PFO closure consists of direct suture closure by open thoracotomy. In an age of excellent percutaneous PFO closure methods and results, surgical closure has become rare. The major inconvenience of this method is the invasive character and elevated complication rate.33)

The goals of PFO closure are to prevent neurologic events and to avoid the need for long-term anticoagulant therapy. Percutaneous transcatheter closure of a PFO is a catheter-based technique utilizing an atrial septal occlusion device.

A nonrandomized retrospective study directly compared 150 patients undergoing endovascular closure to 158 patients receiving medical therapy (79 oral anticoagulants and 79 antiplatelet agents). At four years, the incidence of death or recurrent events in patients undergoing defect closure (8.5%) was significantly lower than that seen in patients treated with aspirin (28.3%), but not significantly lower than that seen in patients treated with oral anticoagulation (13.3%).36) Defect closure was more effective than medical treatment in patients with complete closure and more than one cerebrovascular event at baseline. A systematic review based on the 10 studies of percutaneous PFO closure after a first embolic event in 1,355 patients and six studies of medical therapy in 895 patients showed: 1) The rate of recurrent neurologic events at one year was 0 to 4.9 percent with percutaneous closure and 3.8 to 12.0 percent with medical management; 2) Major complications associated with percutaneous closure occurred in 1.5 percent of patients; 3) Minor complications occurred in 7.9 percent.37) Because these studies were not randomized comparisons, definitive conclusions regarding the superiority of medical versus interventional management could not be reached from this analysis.

It has also been suggested that patients at high risk for recurrent events may benefit from defect closure after the first event.38) In prospective studies, only ASAs, increased shunt volume, and shunting at rest have been associated with increased recurrence risk.

There are no clear PFO therapy guidelines based on randomized trials. There are several ongoing randomized studies {Randomized Evaluation of Recurrent Stroke Comparing PFO Closure to Established Current Standard of Care Treatment (RESPECT) trial, Evaluation of the STARFlex Septal Closure System in Patients with a Stroke or TIA Due to the Possible Passage of Clot of Unknown Origin Through a Patent Foramen Ovale (CLOSURE-1), and Percutaneous closure (PC)-Trial: Patent Foramen Ovale and Cryptogenic Embolism}; their results may clarify the effectiveness of percutaneous closure as compared with medical therapy. Because prospective, randomized, controlled trials are not yet completed, there is no definite answer as to which therapy provides the best long-term prophylaxis against recurrent stroke. No prospective trial of percutaneous PFO closure among patients who have experienced cryptogenic stroke has been completed, and no device has been approved by the Food and Drug Administration (FDA) for PFO closure after cryptogenic stroke. Thus, the safety and effectiveness of devices for this indication are unknown. In 2007, the Food and Drug Administration Circulatory System Devices Panel agreed that randomized trials are absolutely necessary to determine the efficacy of percutaneous septal occluders in preventing recurrent cryptogenic stroke.39)

Although many other devices have been tested and used worldwide (Table 2), the most widely used PFO closure devices in the United States have been the CardioSEAL septal occluder device (NMT Medical, Boston, MA) and the Amplatzer PFO occluder device (AGA Medical, Golden Valley, MN). In the United States, both devices were recently withdrawn from humanitarian device exemption approval status, making them available for use exclusively in clinical research trials. Off-label use of other devices to close PFOs clearly occurs, but it has been impossible to accurately determine either the true extent of off-label use or the indications for which it occurs. Outside of the United States, and particularly in Europe, numerous devices have been approved and are being used for a variety of clinical indications.

The Amplatzer PFO occluder device is engineered as a self-expanding, double-disk device made from Nitinol wire tightly woven into two disks covered by Dacron patches that facilitate device endothelialization (Fig. 2). On the PFO occluder, the right atrial disk is larger (except on the 18-mm device, on which both disks are of the same size) than is the left atrial disk.40) The Amplatzer device has many advocates who recognize its straightforward deliverability, retrievability, and utility in many PFO anatomical variants as some of its many positive attributes. Conversely, the overall rigidity, large amount of Nitinol used in its manufacturing, and occurrence of device erosion (albeit rare) remain noticeable shortcomings.41)

The CardioSEAL occluder device consists of two square Dacron patches mounted among four spring arms made from a cobalt-based alloy, which are designed to enhance adherence to the interatrial septum (Fig. 3). A built-in spring-back mechanism allows the umbrella frame to resume its original shape after deformation during delivery.42) Advocates of the CardioSEAL occluder device cite as its positive attributes its conformability and well-established use in large numbers of patients. By contrast, the device's propensity for surface thrombus formation and inability to redeploy following unsuccessful implantation remain major criticisms of the CardioSEAL device.43)

Experimental devices under investigation include HeartStitch (Sutura Inc., Fountain Valley, CA), which is based on SuperStitch technology (Fig. 4); the PFx™ closure system (Cierra, Redwood City, CA), which employs monopolar radiofrequency energy to effect closure of a PFO by welding the tissues of the septum primum and septum secundum together (Fig. 5); BioTREK™ (NMT Medical, Boston, MA), which is a totally bioabsorbable septal repair implant; and the Coherex FlatStent™ PFO closure system (Coherex Medical, Inc., Salt Lake City, UT), which is very lightweight and leaves minimal surface exposed in the LA (Fig. 6).44)

PFOs are an important risk factor for TIA and stroke, 40% of which are "cryptogenic". The relationship between paradoxical (right-to-left) embolism through a PFO and stroke remains controversial. Conventional therapy for secondary prevention of recurrent stroke in patients with PFO includes medical therapy with antiplatelet agents or anticoagulants. The treatment of symptomatic patients with PFOs who have failed medical therapy is percutaneous closure with permanently implanted closure devices. Although the effect of percutaneous PFO closure has been superior to medical therapy in nonrandomized trials, the optimal therapy for patients with PFOs who have had cerebrovascular events is not well defined, because definitive randomized controlled trials have not been completed.