Intended for healthcare professionals

Clinical Review State of the Art Review

Advances in the management of cardioembolic stroke associated with patent foramen ovale

BMJ 2022; 376 doi: (Published 09 February 2022) Cite this as: BMJ 2022;376:e063161
  1. Brian Mac Grory, assistant professor1 2,
  2. E Magnus Ohman, professor of medicine2 3,
  3. Wuwei Feng, professor of neurology1,
  4. Ying Xian, professor of neurology and medicine4,
  5. Shadi Yaghi, associate professor of neurology5,
  6. Hooman Kamel, associate professor of neurology6,
  7. Michael E Reznik, assistant professor of neurology5
  1. 1Department of Neurology, Duke University School of Medicine, Durham, NC, USA
  2. 2Duke Clinical Research Institute, Durham, NC, USA
  3. 3Division of Cardiology, Department of Medicine, Duke University School of Medicine, Durham, NC, USA
  4. 4Department of Neurology, UT Southwestern Medical Center, Dallas, TX, USA
  5. 5Department of Neurology, Warren Alpert Medical School of Brown University, Providence, RI, USA
  6. 6Department of Neurology, Weill Cornell Medicine, New York, NY, USA
  1. Correspondence to: B Mac Grory brian.macgrory{at}


Patent foramen ovale (PFO) describes a valve in the interatrial septum that permits shunting of blood or thrombotic material between the atria. PFOs are present in approximately 25% of the healthy population and are not associated with any pathology in the vast majority of cases. However, comparisons between patients with stroke and healthy controls suggest that PFOs may be causative of stroke in certain patients whose stroke is otherwise cryptogenic. Options for the diagnosis of PFO include transthoracic echocardiography, transesophageal echocardiography, and transcranial Doppler ultrasonography. PFOs associated with an interatrial septal aneurysm seem to be more strongly linked to risk of recurrent stroke. Therapeutic options for secondary stroke prevention in the setting of a PFO include antiplatelet therapy, anticoagulation, and percutaneous device closure. Recent randomized clinical trials suggest that percutaneous closure reduces the subsequent risk of stroke in appropriately selected patients, with a large relative benefit but small absolute benefit. Referral for percutaneous PFO closure should therefore be considered in certain patients after a multidisciplinary, patient centered discussion. Areas for future study include structural biomarkers to aid in determining the role of PFO closure in older people with possible PFO associated stroke, the role of direct oral anticoagulants, and very long term outcomes after device closure.


Arterial ischemic stroke associated with patent foramen ovale (PFO) is an important cause of disability and death and represents a challenging clinical scenario for the practicing stroke physician, internist, or cardiologist. The foramen ovale is a conduit between the right and left sides of the heart that is present in all humans during fetal development (supplementary figure). This conduit persists beyond infancy in approximately 25% of people,12 which translates to approximately 1.9 billion people globally. Given its high prevalence and that it is clinically silent in the vast majority of cases, PFO is specifically not classified as a form of congenital heart disease.3 Thus, determining whether a PFO is pathologic (that is, implicated in an ischemic stroke) represents a major challenge, and a potential to impose harm through indiscriminate treatment exists.

Six recent randomized controlled trials (RCTs) reported across eight publications showed that percutaneous closure of a PFO is associated with a reduced risk of subsequent cerebrovascular events in appropriately selected younger patients with cryptogenic stroke (a subclass of ischemic stroke for which no cause can be identified).4567891011 These trials identified “PFO-associated stroke”12 as a therapeutically relevant entity and have prompted further studies into optimal strategies for secondary prevention of stroke. These include investigations on expanding percutaneous PFO closure beyond narrowly defined eligibility criteria from existing clinical trials and the development of biomarkers that more accurately predict future risk of stroke in patients with PFO. In this review, we critically evaluate the literature on the management of stroke associated with PFO, with a particular emphasis on recent literature. We also outline persistent areas of uncertainty in this field and opportunities for future research.

Sources and selection criteria

We searched PubMed, Embase, Medline, Web of Science, and the Cochrane Database of Systematic Reviews for articles published between 1 January 2000 and March 2021. We used the keyword terms “patent foramen ovale”, “paradoxical embolization”, “atrial septal aneurysm”, and “right-to-left shunt”. We reviewed medRxiv ( for any RCTs or prospective observational studies that were pending publication and examined ( to assess the state of ongoing prospective studies on this topic. We then searched the reference lists of clinical guidelines, meta-analyses, major clinical trials, and relevant review articles in the area. We included articles published outside of the above specified date range when they were especially informative and relevant to the subject matter. When discussing treatment options, we mostly restricted our consideration to adequately powered and appropriately controlled clinical trials. We included cohort studies and case-control studies where they provided important detail on the epidemiology or pathophysiology of PFO. We did not include case reports. We included articles from peer reviewed journals only, but we placed no restrictions on language.


An autopsy study of 965 human hearts from people without a history of cardiovascular disease, balanced between the sexes and across the lifespan, described a PFO in 27.3% of all participants, 34.3% of those less than 30 years old, and 20.2% of those above 80.1 Although some studies did not find an association between PFO and stroke,131415 most suggest that PFO is more common in patients with cryptogenic stroke than in matched patients without cryptogenic stroke.1617 A meta-analysis of 23 case-control studies of patients younger than 55 years old found a strong association between the presence of a PFO and cryptogenic stroke subtype compared with stroke arising from a known etiology (pooled odds ratio 5.1, 95% confidence interval 3.3 to 7.8).18

Emerging evidence supports a link between PFO and stroke in older adults. The magnitude of the association between PFO and stroke should be lower in older populations as the incidence of stroke due to other causes increases. However, the substrates for venous thromboembolism, including immobility and malignancy, become more common with age. A prospective study of 503 consecutive patients with ischemic stroke compared 227 patients with cryptogenic stroke and 276 patients with a stroke of known cause on the basis of findings from transesophageal echocardiography.19 Among patients over 55 years old with cryptogenic stroke, the prevalence of PFO was 28% compared with 12% in patients with stroke of known cause, and the association between PFO and stroke in older patients remained significant after differences in age and vascular risk factors were accounted for (adjusted odds ratio 3.0, 1.7 to 5.2; P<0.001). In a population based study from the Oxford Vascular Study, transcranial Doppler ultrasonography (TCD-US) with bubble studies was used to establish the presence of a PFO in patients over 60. Among older patients with cryptogenic stroke, 35.8% had a right-to-left shunt (indicative of a PFO) compared with 21.3% of those with a stroke of known cause.20 This association is not universally observed. A prospective, cross sectional study of consecutive ischemic patients in Italy found no difference in the proportion of patients with PFO among those with cryptogenic stroke compared with stroke of determined subtype.15 Furthermore, no significant interaction was seen between stroke subtype and age on PFO status.

PFO and embolic stroke of undetermined source

Cryptogenic stroke refers to a subclass of ischemic stroke for which no cause can be identified and comprises 25-33% of all ischemic strokes.21222324 The TOAST criteria formalized the classification of cryptogenic stroke,25 with other causes of stroke including large artery atherosclerosis, cardioembolism, small vessel stroke, and other determined causes. By this classification scheme, the “cryptogenic” group is heterogeneous and comprises any stroke without a defined cause. This includes strokes with no candidate mechanism identified despite thorough investigation, strokes caused by more than one potential mechanism, and cases in which the diagnostic investigation remains incomplete. The heterogeneity of this scheme was considered a disadvantage and led to the more refined classification of embolic stroke of undetermined source (ESUS).2627 This classification describes an embolic appearing stroke for which thorough diagnostic investigation (cervical and intracranial vessel imaging, transthoracic echocardiography, electrocardiography, and ≥24 hours of cardiac monitoring with automatic rhythm detection) has yielded no definitive cardioembolic source or high grade arterial stenosis corresponding to the affected brain region. The more sensitive testing modalities such as transesophageal echocardiography and TCD-US are not mandatory as part of the diagnostic investigation of ESUS, leading to likely under-identification of PFO within this framework. Additional consideration should be given to emerging non-cardiac causes of ESUS including occult malignancy,28 the carotid web,29 aortic arch atherosclerosis, and substenotic carotid atherosclerosis within the early phases of a patient’s diagnostic investigation.

One challenge posed by the discovery of a PFO in the context of stroke is to determine its likelihood of causation. A related but separate question is whether an intervention to close the PFO will benefit the patient. Although the likelihood that a PFO is pathogenic rather than incidental varies depending on a patient’s age,30 in clinical practice PFOs are usually detected in younger patients without vascular risk factors.31 Thus, the presence of a PFO paradoxically signifies a group of patients at particularly low risk of recurrence.32 In general, the risk of recurrent stroke in patients with a PFO is approximately 2% per year,33 but most patients with a PFO have more than one mild to moderate risk embolic source,34 further complicating the assignment of causality.

Several strategies have been devised to assign a likelihood that a PFO is causally related to an ischemic stroke. The Risk of Paradoxical Embolism (RoPE) score is a 10 item scale that is calculated on the basis of a person’s age (18-29, 30-39, 40-49, 50-59, 60-69, or ≥70 years), cortical infarct location, and absence of vascular risk factors (hypertension, diabetes, previous stroke, or smoking).3536 In a patient with a maximum RoPE score of 10, the tool estimates an 88% probability of the stroke being attributable to a PFO. By comparison, a RoPE score of 0 would suggest a 0% probability of stroke being attributable to a PFO. Although clinically useful, this score does not account for characteristics of the PFO itself (discussed below) or for the presence of venous thromboembolism, which may increase the probability that a PFO is a causative lesion. Nevertheless, a RoPE score of 7-10 has been proposed to be more likely to reflect a pathogenic PFO,37 and the tool has been suggested for use in predicting the presence of PFO in patients with stroke.38

The success of recent RCTs has led to calls to codify PFO as a causative mechanism of stroke in patients with ESUS.39 A recent consensus statement from the PFO International Workup Group proposed to update the nomenclature related to stroke risk and PFO, coining the term “PFO-associated stroke.”12 This diagnosis requires a superficial, large deep, or retinal infarct in the presence of a medium to high risk PFO and no other likely cause identified. Such cases are then classified according to the probability of the PFO being the causative mechanism (highly probable, probable, possible, or unlikely), with this probability being based on factors such as evidence of straddling thrombus, atrial septal aneurysm (ASA), presence of large shunt, presence of pulmonary embolism, or deep venous thrombosis (DVT).

Pathophysiology of PFO associated stroke

Mechanisms of stroke in patients with PFO

Despite a consistent signal in the observational literature linking PFO with cryptogenic stroke,1617 the presence and risk profile of PFO depends on the presence of other proposed mechanisms. For instance, the presence of a pathogenic PFO is inversely related to the presence of non-stenotic ipsilateral carotid atherosclerosis in patients with ESUS.40 Additionally, the prevalence of a likely pathogenic PFO was found to be markedly lower in ESUS patients with atrial cardiopathy compared with those without atrial cardiopathy (3.3% v 23.7%; adjusted odds ratio 0.2, 0.02 to 0.6).41

Paradoxical embolization refers to any passage of material between the right and left sides of the heart and can occur by way of a PFO, atrial septal defect, ventricular septal defect, or pulmonary arteriovenous fistula. This presumed mechanism of stroke associated with PFO is supported by several observations. Firstly, the frequency of concurrent ischemic stroke among patients with pulmonary embolism is higher among patients with than without PFO.42 Secondly, strokes associated with PFO have a predilection for the cerebral or cerebellar cortices, a pattern consistent with other cardioembolic disorders (such as atrial fibrillation).434445 Thirdly, material other than thrombus may cross a PFO, as in the case of air embolization in divers with PFO.46 Finally, many cases of emboli being visualized straddling the right and left atria through a PFO have been reported.47 Under normal physiology, microthrombi are forming nearly continuously in the venous system and enter the right side of the heart via the inferior or superior vena cava before being filtered out in the pulmonary circulation. In the presence of a right-to-left shunt, these thrombi may pass directly from the right to the left side of the heart and enter the arterial system.4849 A large residual Eustachian valve may potentiate paradoxical embolism by directing a jet of blood toward the PFO, thereby increasing the likelihood of right-to-left passage as opposed to filtration through the pulmonary circulation.5051

A potential alternative explanation to paradoxical embolization as the causative mechanism of PFO associated stroke is that intrinsic differences exist in the bloodstream of people with right-to-left shunting. In addition to oxygenating blood, the pulmonary arterial circulation plays a crucial role in filtering out waste products including prothrombotic and vasoactive metabolites such as serotonin. This property of the pulmonary arterial circulation is partially neutralized in patients with PFO and may also help to explain the association between PFO and migraine.5253 Exploratory work has identified differing patterns of plasma protein expression in patients before and after PFO closure,5455 but whether this mechanism is a contributor to risk of stroke in patients with PFO remains to be seen.

For paradoxical embolization to arise, thrombus formation must be present within the venous system. Many aspects of normal life increase the propensity to formation of a DVT in the extremities. For instance, the risk of DVT in people not wearing compression stockings after a long haul flight is 10.3%.56 Once a lower extremity DVT has developed, ongoing embolization to the right side of the heart ensues. In a prospective study of 60 patients with a lower extremity DVT, 43% showed evidence of active emboli detected at a rate of 5-800 per minute when screened via ultrasonography at the proximal femoral vein or above.57 Stroke in patients with PFO may also be provoked by the Valsalva maneuver or by pulmonary hypertension, which may increase the tendency for right-to-left flow across the PFO (for example, as occurs nightly in patients with sleep apnea).58

Characteristics associated with increased risk of stroke recurrence

Three overlapping, although distinct, reasons exist to examine patient related or PFO related characteristics in a person with a PFO associated stroke: to determine the likelihood of a recurrent event after an index stroke related to PFO; to determine the likelihood that a stroke was caused by a PFO compared with an alternative mechanism; and to determine whether a person meets enrollment criteria for any given clinical trial.

Factors that predict a higher likelihood of recurrence (first point above) may also be associated with a reduced likelihood that a PFO was causative (second point) and vice versa. A PFO with no competing cause of stroke is more likely to be pathogenic but, paradoxically, denotes a population at particularly low risk of recurrence. Furthermore, key morphologic criteria such as a PFO size are important to consider because they were enrollment criteria in two major clinical trials (third point above) but are not clearly associated with an increased likelihood of recurrent stroke (first point). The risk of recurrence after stroke associated with PFO is low (discussed in subsequent sections). Thus, most analyses describing the relation between patient or PFO related factors and recurrent events are exploratory in nature. Box 1 outlines patient and PFO related factors that have been associated with an increased risk of stroke recurrence.Figure 1 and figure 2 depict the evaluation of key morphometric criteria associated with PFO on transthoracic echocardiography and transesophageal echocardiography respectively.

Box 1

Patient related and patent foramen ovale (PFO) related factors with reported association with subsequent stroke

Patient related factors

  • Age5960

  • Hypertension59

  • Diabetes mellitus61

  • Active migraine62

  • Prothrombotic mutations (eg, protein C deficiency, protein S deficiency, factor V Leiden, and prothrombin G20210A)6364

  • Risk of Paradoxical Embolism (RoPE) score <561

  • Stroke in posterior cerebral artery territory62

  • Post-procedure atrial fibrillation61

PFO related factors

  • Interatrial septal aneurysm33

  • Shunt size376566

  • Residual shunting after closure67

Fig 1
Fig 1

Key morphometric features of patent foramen ovale (PFO) on transthoracic echocardiography. A: Orientation of atria and ventricles on apical four chamber view. B: Sonicated air bubbles visible during bubble study with Valsalva maneuver confirming presence of right-to-left shunt. C: Doppler jet associated with PFO. D: Interatrial septal aneurysm without PFO. Note sonicated bubbles present in right atrium and right ventricle but not in left atrium. LA=left atrium; LV=left ventricle; RA=right atrium; RV=right ventricle.

Fig 2
Fig 2

Key morphometric features of patent foramen ovale (PFO) on transesophageal echocardiography. A: Direct visualization of PFO on mid-esophageal short axis view. B: Doppler jet associated with PFO. C: Bubbles (marked with *) traversing interatrial septum during Valsalva maneuver. D: Well seated PFO closure device (marked with **). AA=ascending aorta; AV=aortic valve; LA=left atrium; RA=right atrium.

Two major structural features of PFO are hypothesized to portend a higher risk of recurrent stroke after an index PFO associated stroke. The first is atrial septal aneurysm.5968 An ASA describes the impingement of the atrial septum into the left and right atria according to the cardiac cycle. ASAs have been associated with an increased risk of recurrent stroke in patients with PFO (adjusted hazard ratio 3.3, 95% confidence interval 1.8 to 5.9, in a pooled analysis of four prospective studies of patients with PFO and cryptogenic stroke43) and represent an independent predictor for cryptogenic stroke.686970 However, the reason for this is not yet known. ASAs may cause turbulent blood flow in the region of the PFO, leading to direct thrombus formation, a mechanism separate from paradoxical embolization. Of note, a synergistic effect has been described between PFO and ASA, wherein the risk of subsequent stroke associated with both in combination is greater than the sum of their individual risks.68Figure 1 (D) shows an interatrial septal aneurysm in the absence of a PFO.

The second feature hypothesized to portend a higher risk of recurrent stroke after an index PFO associated stroke is shunt size. Data on the association between shunt size and recurrent stroke are conflicting. Shunt size during life can be inferred on the basis of the number of bubbles that cross the interatrial septum with each cardiac cycle, the degree of maximum separation of the septum primum and septum secundum during a Valsalva maneuver, the maximum number of bubbles observed in the left atrium in a single frame, or whether shunting occurs at rest or only in response to provocative maneuvers. The optimal method of measurement is three dimensional transesophageal echocardiography.71 A large shunt may be expected to increase the propensity for paradoxical embolization, as a great volume of blood is crossing the interatrial septum. Several small, observational studies have associated large shunt size with recurrence,656672 although in each study the number of events was low. For instance, in a single center, retrospective, observational cohort study of consecutive patients with cryptogenic ischemic stroke referred for transesophageal echocardiography, 181 patients were followed for a median of 3.5 years. Fourteen (8%) patients had a recurrent stroke, and PFO size ≥3 mm was independently associated with recurrence of stroke (hazard ratio 3.0, 1.96 to 4.69; P=0.003).65 In randomized trials, shunt size was associated with higher magnitude of benefit of closure in RESPECT (significant interaction between shunt size and treatment effect; P=0.07 with threshold for significance set at 0.10) but not in CLOSE or DEFENSE-PFO.7910 The caveat is that shunt size may alter technical aspects of percutaneous closure, which may mediate this difference, instead of shunt size being itself a risk factor for recurrence. By contrast, two large, observational studies and two meta-analyses found no association between shunt size and risk of recurrent stroke.33596273 Paradoxically, a study of 1324 patients from the RoPE database found that in patients with a RoPE score of >6 (that is, who had a higher likelihood that the PFO was the causative mechanism), small shunt size (maximum number of bubbles in the left atrium observed in a single frame of ≤10) was associated with an increased risk of subsequent stroke (hazard ratio 3.26, 1.59 to 6.67).37 In patients with a RoPE score of ≤6, no association was observed between shunt size and recurrent stroke (hazard ratio 1.29, 0.82 to 2.03).

The relation between other morphologic features (large residual Eustachian valve,51 biomarkers of left atrial dysfunction7475) and PFO pathogenicity or risk of recurrence after PFO associated stroke is unclear.

Diagnostic approaches

A PFO is usually identified during structured diagnostic investigation for potential causes of ischemic stroke.21 Diagnostic modalities to specifically test for the presence of a PFO include transthoracic echocardiography, transesophageal echocardiography, and TCD-US. Cardiac computed tomography does not have adequate sensitivity to exclude a PFO and will not be discussed further.76

Transthoracic echocardiography

Transthoracic echocardiography with an agitated saline bubble study is generally the first test used in clinical practice to examine for the presence of a PFO (fig 1, A-C). In a bubble study, agitated saline is injected into a tributary of the superior vena cava—typically the brachiocephalic vein—which then flows to the right atrium. In the presence of an interatrial septal defect, passage between the right and left sides of the heart may be visualized on echocardiography (a so-called “positive bubble study”). A Valsalva maneuver is often used during the test to increase intrathoracic pressure and thus increase the likelihood that right-to-left passage of bubbles will be visualized. However, detection of a PFO via transthoracic echocardiography is user and technique dependent. For example, the detection of a PFO increases with an increasing number of injections of agitated saline.7778 Additionally, the superior vena cava may not always orient toward the PFO, in which case injection of agitated saline into the femoral vein and inferior vena cava may be a superior technique.79 In older studies, when compared with transesophageal echocardiography, the overall sensitivity of transthoracic echocardiography was 50% and specificity was 92%.80 However, with the advent of more modern approaches including second harmonic imaging (an echocardiographic technique that reduces artifact from the chest wall and improves the quality of the signal from deeper structures), the sensitivity of transthoracic echocardiography approaches 90%.81

Multiple potential causes of interatrial shunting can be discriminated on the basis of echocardiography according to the location, size, and stability of an interatrial septal defect. These include a PFO, an atrial septal defect, a pulmonary arteriovenous fistula, an unroofed coronary sinus, and rarer congenital cardiac defects. A PFO is a potential defect in the interatrial septum and is visible only when atrial hemodynamics promote shunting of blood from one side to another. By contrast, an atrial septal defect is a structural deficiency of the septum that is visible regardless of atrial pressure differentials. Pulmonary arteriovenous fistula (pAVF) describes a connection between a pulmonary artery and pulmonary vein without an intervening capillary bed. The time elapsed between injection and visualization of bubbles in the left atrium may aid in discriminating between a PFO and a pAVF. An “early positive” bubble study is indicative of a PFO, whereas a “late positive” bubble study is suggestive of a pAVF. An unroofed coronary sinus (coronary sinus septal defect) describes a deficiency of the wall of the coronary sinus permitting transmission of blood from the right atrium to the left atrium via this deficiency.

Transesophageal echocardiography

Transesophageal echocardiography permits more direct visualization of relevant cardiac structures than transthoracic echocardiography (fig 2). Transesophageal echocardiography is performed by inserting an ultrasonic probe into the esophagus and bringing it in close proximity to the left side of the heart, thus providing a superior acoustic window. Thus, less artifact is imposed by skin, fat, muscle, bone, and other chest wall structures. A postmortem study of 35 patients who had transesophageal echocardiography performed before death found nine PFOs, of which all were diagnosed by color Doppler transesophageal echocardiography and eight were diagnosed by contrast transesophageal echocardiography.82

When a PFO is already known, transesophageal echocardiography is the gold standard to determine its size and to further characterize the interatrial septum. One measure of right-to-left shunt size considers early microbubble appearance in the left atrium, with the shunt classified as trace (0-5 microbubbles), moderate (6-25 microbubbles), or large (>25 microbubbles). Transesophageal echocardiography is also useful to examine for competing structural causes of stroke that may not be visualized on transthoracic echocardiography, including right atrium appendage thrombus, atrial myxoma, fibroelastoma, aortic arch atheroma, or small aortic or mitral valvular vegetations.8384 Transesophageal echocardiography is well tolerated even in older patients.85

Absolute contraindications to transesophageal echocardiography include active gastrointestinal bleeding, recent gastroesophageal surgery, or esophageal stricture, diverticulum, or neoplasm; relative contraindications include recent upper gastrointestinal bleeding, hiatal hernia, coagulopathy, and cervical osteoarthritis.86 The risk of esophageal perforation is less than 0.1%8788; other more common risks include transient bronchospasm, cardiac arrythmias, minor pharyngeal bleeding, or procedural failure due to intolerance of the probe.89 A judicious approach to the use of transesophageal echocardiography is important because, although a low risk procedure, it is resource intensive and requires a large team to deploy. Furthermore, staff members performing the procedure are exposed to large volume aerosol because of coughing and gagging during probe insertion, which magnifies the risk of transmission of respiratory viruses including SARS-CoV-2.90

Transcranial Doppler ultrasonography

TCD-US in tandem with an agitated saline bubble study is a sensitive modality for detection of a PFO.91 Inference of a right-to-left shunt is possible when bubbles are insonated in the cerebral circulation. TCD-US also permits quantification of microemboli as an indirect measure of the magnitude of right-to-left shunting. The Spencer grading system considers five levels of right-to-left shunt size: grade I (trace, 1-10 microemboli), grade II (small, 11-30 microemboli), grade III (moderate, 30-100 microemboli), grade IV (large, 101-300 microemboli), and grade V (very large, >300 microemboli),92 although such fine gradations between shunt size are unlikely to prove useful in clinical practice

A bubble study is less reliable in transesophageal echocardiography than in TCD-US because the presence of the ultrasonic probe in the throat may render performance of the Valsalva maneuver more difficult for the person under study and because sedation may be needed for the procedure. One head-to-head comparison of transesophageal echocardiography and TCD-US found that the latter had a sensitivity of 95% and specificity of 100% for the detection of PFO relative to transesophageal echocardiography. Provocative maneuvers such as coughing or the Valsalva maneuver further increase its sensitivity.93 In general, the concordance between transesophageal echocardiography and TCD-US is approximately 90%.94 However, TCD-US is more sensitive than transesophageal echocardiography for the presence of right-to-left shunting in regular breathing,94 so it should be strongly considered in patients who are unable to perform a Valsalva maneuver. Although most studies consider transesophageal echocardiography as the reference standard, a proportion of patients have small PFOs that are not visible on transesophageal echocardiography but are identified via TCD-US.819395 Thus, transesophageal echocardiography and TCD-US should be considered complementary studies, with both needed for confident exclusion of a PFO.

TCD-US may also be used to examine for the presence of spontaneous microembolism via high intensity transient signals, screen for extracardiac shunting, and assay for the presence of residual shunting after percutaneous PFO closure.

Testing for venous thromboembolism

The role of testing for venous thromboembolism in patients with a cryptogenic stroke, a PFO, and no signs or symptoms suggestive of venous thromboembolism is unclear. An assay of serum D-dimer is a highly sensitive test and, when negative, is useful for the exclusion of venous thromboembolism in patients with a low pre-test probability of venous thromboembolism.96 Further options for testing including venous Doppler ultrasonography of the upper and lower extremities, magnetic resonance venography of the pelvis, and computed tomography angiography of the chest. A retrospective, single center study of 114 patients with cryptogenic ischemic stroke and a PFO found that 9% also had a DVT and 4% had a pulmonary embolism.97 Another retrospective study found that 29% of patients with cryptogenic stroke and a PFO had a concurrent DVT and 7% had a pulmonary embolism.98 When present, venous thromboembolism strongly implicates the PFO as a mechanism for stroke and warrants treatment with anticoagulation,99 although it does not necessarily bolster the argument for pursuing percutaneous PFO closure.

Hypercoagulability/thrombophilia testing

Testing for both arterial and venous hypercoagulability disorders may be considered as part of the investigation of diagnostic stroke in younger patients. However, no evidence associates hypercoagulablity with reduction of risk for recurrent stroke, so appropriate patients must be carefully selected. Most hypercoagulability panels contain at least 10 individual tests, so the cost of testing and the probability of obtaining at least one value outside the reference range are both high.100 Nevertheless, disorders of hypercoagulability become more prevalent with age,101 and they should still be considered even in older patients when clinical suspicion is high and traditional vascular risk factors are absent. For definitive results, testing must be repeated at an interval of 12 weeks,102 as many individual tests may be spuriously elevated in the presence of acute thrombosis. Of note, both arterial and venous hypercoagulability disorders may predispose to stroke in the presence of a PFO.103 A retrospective, observational cohort study conducted at a tertiary referral center found that laboratory testing for hypercoagulability in patients below the age of 65 resulted in a change in management (defined for the purposes of the study as starting anticoagulation or PFO closure) in one in 12 patients.104

Common disorders tested for across most hypercoagulable panels include those causing venous hypercoagulability (for example, protein C and S deficiency, factor V Leiden, prothrombin gene mutation, and anti-thrombin III deficiency105), arterial hypercoagulability (for example, hyperhomocysteinemia106), and mixed arterial and venous hypercoagulability (for example, antiphospholipid antibody syndrome107). Cancer increases the tendency to both arterial and venous hypercoagulability and should be suspected in people with indicative physical examination findings (such as lymphadenopathy), unexplained weight loss, other B symptoms (fever or night sweats), toxin exposure, or a family history of cancer at an early age and in those who have not had age appropriate cancer screening. Other hypercoagulable states such as exposure to hormonal contraception, hormone replacement therapy, or tobacco should also be considered.

Importantly, all clinical trials of percutaneous PFO closure have excluded patients with hypercoagulable disorders. However, several observational studies have shown that concurrent hypercoagulability in patients with stroke and a PFO is associated with a higher risk of recurrent stroke. These findings suggest that the benefit of closure may therefore be even further magnified in this population,63108 but additional prospective studies are needed to definitively answer this question.

Long term cardiac monitoring

Two recent studies explored the diagnostic yield of extended cardiac monitoring after a stroke or transient ischemic attack. EMBRACE was an RCT in which 572 patients were randomized to 30 days of ambulatory cardiac monitoring or standard medical therapy.109 The primary endpoint (newly detected atrial fibrillation of ≥30 seconds within 90 days of randomization) was attained by 16.1% of patients assigned to cardiac monitoring and 3.2% of control patients (an absolute difference of 12.9%, 95% confidence interval 8.0% to 17.6%). CRYSTAL-AF was an RCT in which 441 patients were randomized to insertion of an insertable cardiac monitor (ICM) or standard medical therapy.110 The primary endpoint was time to detection of new atrial fibrillation of ≥30 seconds. By six months, this endpoint was reached by 8.9% of patients in the ICM group compared with 1.4% of controls (hazard ratio 6.4, 1.9 to 21.7), with a sustained, incremental improvement in detection rates of paroxysmal atrial fibrillation up to three years post-implantation. Such monitoring has also been shown to increase anticoagulant use and is associated with the risk of subsequent stroke.111 However, patients enrolled in these trials were markedly older and had a high burden of vascular risk factors relative to patients enrolled in the major trials of percutaneous PFO closure, which mostly consisted of young patients without vascular risk factors. Age is the strongest predictor for the development of atrial fibrillation,112 with fivefold higher rates of detection of atrial fibrillation described in patients older than 65 compared with those younger than 65.110

Importantly, none of the PFO closure trials mandated cardiac monitoring before enrollment. Two small studies have subsequently examined the yield of cardiac monitoring in patients eligible for both PFO closure and enrollment in CRYSTAL-AF or EMBRACE. One such study documented a yield of 27% for the detection of new atrial fibrillation after three years of recording via an ICM,113 and another used external cardiac monitoring and reported a diagnostic yield of 9.5% for new atrial fibrillation at three weeks.114 As a result, a reasonable approach may be to pursue short term cardiac monitoring before PFO closure, reserving longer term monitoring for patients with risk factors for atrial fibrillation or biomarkers of left atrial dysfunction,115 which increase the likelihood of detection of atrial fibrillation in patients with ESUS.116117118

Using long term cardiac monitoring in all patients under consideration for PFO closure may not be reasonable, as the trials of PFO closure all mandated enrollment within six months of the index event (with the exception of RESPECT8). Whether their findings may be similarly applied beyond this time frame is therefore uncertain. Long term cardiac monitoring is also costly and associated with a risk of detecting brief, transient, incidental runs of atrial fibrillation (which can itself be provoked by ischemic stroke119120).

Medical treatment after PFO associated stroke

Pharmacologic secondary prevention is generally recommended after stroke associated with PFO, with options including antiplatelet therapy and anticoagulation. Of the two, anticoagulation has been posited to be more effective, as stroke attributable to PFO is hypothesized to arise from thrombi originating in the venous blood supply. However, this position has not been conclusively tested in head-to-head comparisons with antiplatelet therapy, and the increased bleeding risk associated with anticoagulation may eclipse any benefit with respect to reduction of stroke risk.

The PFO Cryptogenic Stroke Study (PICSS)121 enrolled 630 patients with cryptogenic stroke nested within the Warfarin-Aspirin Recurrent Stroke Study (WARSS),122 of whom 33.8% had a PFO. This PFO substudy was underpowered to detect a difference in rates of recurrent stroke between patients treated with warfarin and those treated with aspirin (hazard ratio 0.52, 0.16 to 1.67; P=0.28). The PFO Closure or Anticoagulants versus Antiplatelet Therapy to Prevent Stroke Recurrence (CLOSE) study was also underpowered to find a significant difference in rates of subsequent stroke between its anticoagulation and antiplatelet therapy medical treatment arms (hazard ratio 0.44, 0.11 to 1.48; P=0.18).7 Even after an individual participant meta-analysis of prospective observational cohort studies and the medical arms of three RCTs,123 conclusive evidence to compare anticoagulation treated patients and antiplatelet therapy treated patients with respect to stroke recurrence rates is still not available (adjusted hazard ratio 0.75, 0.44 to 1.27).

No dedicated trials have specifically compared the direct oral anticoagulants (DOACs) apixaban, rivaroxaban, edoxaban, or dabigatran with antiplatelet therapy in patients with PFO and stroke, although subgroup analyses have been done using data from recent trials in patients with ESUS. A pre-specified post hoc analysis of the NAVIGATE-ESUS trial was not powered to examine recurrent stroke rates in patients treated with rivaroxaban compared with those treated with aspirin (hazard ratio 0.54, 0.22 to 1.36).124125 However, the addition of data from the PICCS and CLOSE trials in a subsequent meta-analysis resulted in a net benefit of anticoagulation over antiplatelet therapy (odds ratio 0.48, 0.24 to 0.96; P=0.04). A similar subgroup analysis of the RESPECT ESUS trial included 680 patients with a PFO, for whom the risk of recurrent stroke in those treated with dabigatran was 2.9% compared with 3.2% in patients treated with aspirin.126127 A meta-analysis combining participants with PFO from all four of these trials did not find a significantly lower likelihood of recurrent stroke in DOAC treated patients (odds ratio 0.70, 0.43 to 1.14).127 Two major limitations are inherent in this analysis: morphometric properties of the PFO were not considered in this analysis as transesophageal echocardiography was not an enrollment criterion for NAVIGATE-ESUS or RESPECT, and, even in combination, these data lack statistical power to conclusively establish which treatment, if either, is better.

Percutaneous PFO closure

Percutaneous PFO closure was first proposed almost 30 years ago.128 The underlying rationale is that preventing transmission of a thrombus across the interatrial septum would reduce the risk of subsequent stroke and that the magnitude of treatment benefit would overcome any procedural complications or long term negative effects of device insertion. However, percutaneous closure has been a controversial therapy for most of the past three decades.129

A low risk of recurrent events exists in patients with PFO associated stroke whether or not closure is pursued. This, along with the relatively small absolute benefit expected with treatment, has rendered conducting a definitive and adequately powered clinical trial difficult. Nevertheless, six recent RCTs have provided compelling evidence that percutaneous PFO closure is associated with a reduced risk of recurrent events and is cost effective.4567891011130 This conclusion was supported by a study level meta-analysis,131 with an individual participant level meta-analysis under way.132 All six recent RCTs were open label with no sham procedure but used blinded endpoint adjudication. Five trials enrolled patients within six months of their qualifying event, and one trial enrolled patients up to nine months (RESPECT). Of these trials, five were restricted to patients younger than 60, and one enrolled patients up to age 80 (DEFENSE-PFO). This trial was noteworthy as the only one to include patients aged between 60 and 80; however, the mean age was 49 (standard deviation 15) in the percutaneous closure arm and 52 (12) in the medical treatment arm, meaning that no conclusions about patients between 60 and 80 can be definitively drawn. Table 1 and table 2 outline key aspects of each trial, including patients’ characteristics, key enrollment criteria, device type, study endpoints, and adverse outcomes.

Table 1

Overview of study design for randomized controlled trials comparing patent foramen ovale (PFO) closure with medical therapy

View this table:
Table 2

Overview of outcomes for randomized controlled trials comparing patent foramen ovale (PFO) closure with medical therapy

View this table:

Technical aspects of percutaneous PFO closure

Morphological PFO features such as size and presence of an ASA influence the complexity of percutaneous closure.133 The six key clinical trials used two major classes of closure device: an older, umbrella-clamshell design (CLOSURE) and a more modern “double disc” device (PC, RESPECT, CLOSE, REDUCE, DEFENSE-PFO). Figure 3 depicts placement of a double disc device across the interatrial septum. With both device classes, the objective is to permanently close the PFO and prevent transmission of thrombus from the right atrium to the left atrium while minimizing the risk of complications. Technically successful closure occurred in 93-96% of patients in the double disc trials and 87% in the umbrella-clamshell trials.30 One study level meta-analysis of these RCTs suggested a significant reduction in recurrent stroke with double disc devices but not with umbrella-clamshell devices.30 The primary goal of these interventions is the attainment of satisfactory closure measured via degree of residual shunting, as the presence of residual shunting is associated with increased rates of recurrent stroke (hazard ratio 3.1 (1.7 to 5.6) for any residual shunt and 4.5 (2.2 to 9.2) for a moderate to large residual shunt).67134

Fig 3
Fig 3

Deployment of double disc closure device. A. Orientation of interatrial septum compared with rest of adult heart. After cannulation of femoral vein, catheter is introduced into right atrium (RA). B. Patent foramen ovale (PFO) is traversed with catheter and distal-most disc is expanded in left atrium (LA). Catheter is retracted to generate snug seal between interatrial septum and disc. As catheter is further retracted, right atrial disc self-expands. C. Oppositional forces created by two discs seal PFO

Procedural complications of percutaneous closure include procedural failure (that is, substantial residual right-to-left shunting after deployment), post-procedural atrial fibrillation, cardiac tamponade, pneumothorax, hemothorax, and aortic dissection; device complications include device migration or thrombosis. Given these risks, the performance of individual centers should be assessed by key benchmarks, including the proportion of PFO closure procedures that result in a technically successful result and the rate of procedural complications. A retrospective study based on administrative claims data suggested that, in real world practice, the rate of any adverse event was 10.9% in patients over the age of 60 and 4.9% in patients younger than 60.135 These complications included atrial fibrillation or atrial flutter (3.7%), vascular complications requiring surgical repair (3.0%), surgical site hematomas (2.7%), cardiac tamponade (0.5%), pneumothorax or hemothorax (0.1%), and death (0.3%). Post-procedural atrial fibrillation is associated with an increased risk of recurrent events61; it may be a complication of the procedure itself or an independent event and an indicator of left atrial dysfunction that may have predated the stroke. A meta-analysis of RCTs suggested that the weighted mean incidence of atrial fibrillation after PFO closure was 3.2% (95% confidence interval 1.8% to 5.0%) over a mean follow-up of 2.8 years, compared with 0.47% (0.15% to 0.91%) in patients managed medically.136 In a retrospective, observational cohort study including 1349 patients who underwent PFO closure, 53 (3.9%) patients developed new atrial fibrillation or atrial flutter.137 Of those 53 patients, 33 (62%) developed atrial fibrillation within four weeks of the procedure and 23 (43%) had only one documented episode of <48 hours throughout study follow-up. Data on the management of atrial fibrillation after PFO closure are lacking. One reasonable approach may include temporary anticoagulation with a vitamin K antagonist or DOAC in tandem with placement of an implantable loop recorder, which would permit calculation of the overall atrial fibrillation burden to determine whether lifelong anticoagulation is warranted.

Options for medical treatment after percutaneous closure include single antiplatelet therapy with either aspirin or clopidogrel, dual antiplatelet therapy, or anticoagulation. Most trials comparing percutaneous closure with medical treatment mandated some duration of post-procedure dual antiplatelet therapy, typically for at least one month. Table 1 outlines antithrombotic therapy after percutaneous closure for each of the major percutaneous closure trials. Single antiplatelet therapy should be continued indefinitely in all patients for secondary prevention of stroke (even with a technically successful PFO closure), given that one can never be absolutely certain that the PFO was the causative lesion.

Consideration for percutaneous PFO closure

A discussion between a neurologist and a cardiologist is necessary for all patients referred for percutaneous closure according to US Food and Drug Administration mandates for the Amplatzer and Gore Septal Occluder devices.138139 Input from other specialties (for example, internal medicine/primary care, hematology, rheumatology) should also be sought when some ambiguity surrounds the results of diagnostic testing or a patient’s unique medical history. A multidisciplinary clinic or case conference is a practical framework allowing a forum for such discussion, with simultaneous opinions from a neurologist and structural cardiologist. This dialog should include the results of all relevant diagnostic testing as discussed above, including an evaluation of PFO specific and patient specific factors (fig 4). An age cutoff of 60 is proposed on the basis of the enrollment criteria for most of the clinical trials (detailed in table 1); however, this should not be considered absolute. For instance, a patient of 55-60 with more than one vascular risk factor and an enlarged left atrium might meet enrollment criteria for the recent randomized trials but may benefit from antiplatelet therapy, long term cardiac monitoring, and risk factor modification as opposed to percutaneous closure.

Fig 4
Fig 4

Clinical decision making algorithm for management of patent foramen ovale (PFO) associated stroke. *In certain cases, other mechanisms causing interatrial shunting may require targeted medical or surgical therapy—for example, an atrial septal defect or very large pulmonary arteriovenous fistula. ARCADIA=AtRial Cardiopathy and Antithrombotic Drugs in Prevention After Cryptogenic Stroke; MRI=magnetic resonance imaging; RCT=randomized clinical trial

Patients should be counseled that the decision to pursue PFO closure is rarely a straightforward one and that other options including medical management and surgical closure may be reasonable depending on an individual’s preferences and tolerance of risk. Patients should be counseled that up to a third of PFOs are incidental, even in cases of cryptogenic stroke, meaning that PFO closure might be futile or potentially harmful,18 aside from the small, theoretical benefit of primary prevention of PFO associated stroke.

Although the number needed to treat to prevent one ischemic stroke over five years is 24 across all patients receiving PFO closure (and 13 in patients with PFO and ASA),30 consideration should be given to the risk of subsequent stroke even if the procedure is performed. Closure reduces the risk of recurrent cryptogenic stroke but not other stroke subtypes.140 Stroke recurrence may result from technical failure of the procedure, device complications including atrial fibrillation, the emergence of new stroke substrates over time (carotid stenosis, atrial fibrillation, small vessel disease), or covert mechanisms that were overlooked at the time of the initial evaluation. Long term follow-up from the RESPECT trial found that 28.3% of recurrences arose owing to a mechanism unrelated to the PFO or its closure.10 A thoughtful decision to refer a patient for percutaneous closure depends on the following steps:

  1. Making the diagnosis of PFO definitively (including ruling out other potential causes of interatrial shunting)

  2. Careful history taking to elicit any provoking maneuvers at onset

  3. Assessing a patient’s overall vascular health (including calculating their RoPE score);

  4. Consideration of a potential concurrent venous thromboembolism

  5. Examining the characteristics of the PFO, including its size and the presence of an ASA

  6. Examining characteristics of the stroke itself (for example, small, deep strokes are unlikely to be associated with a PFO)

  7. Completing a thorough investigation for competing mechanisms including paroxysmal atrial fibrillation, structural cardiac lesions, and non-atherosclerotic vasculopathies.

Surgical closure

Open surgical closure of a PFO is sometimes done at the same time as open heart surgery for other indications and involves direct oversewing of the PFO. A small study of 11 patients who underwent surgical closure of a PFO (six of whom had concomitant open valve replacement) examined technical success rates via transesophageal echocardiography.141 Residual shunting remained present in eight patients, of whom six had incomplete sealing of the septum primum and septum secundum and two had complete closure of the original PFO but a new iatrogenic septal defect in the middle or inferior part of the fossa ovalis. This high rate of technical failure may explain why open surgical PFO closure is associated with a higher risk of subsequent stroke,142 and its use should therefore be avoided.


Although some existing guidelines published before 2018 are agnostic to the benefits of percutaneous closure,143144145 several major statements have come from major professional organizations,146147148149 as well as position papers and conference proceedings,150151152 that are now endorsing its use in appropriately selected patients. The American Heart Association/American Stroke Association (AHA/ASA) Guidelines for the Prevention of Stroke in Patients With Stroke and Transient Ischemic Attack advocate consideration of targeted investigation for PFO in patients for whom a cause of stroke is not identified after electrocardiography, basic laboratory testing, echocardiography (either transthoracic echocardiography or transesophageal echocardiography) and non-invasive cervical vessel imaging based on individual patient related factors.146 The same guidelines state that choosing percutaneous closure over antiplatelet therapy in patients aged 18-60 with a non-lacunar ischemic stroke of undetermined cause (after thorough investigation) and when the PFO has high risk features is reasonable. The American Academy of Neurology (AAN) states that percutaneous closure may be recommended in patients <60 years old with an embolic stroke for which no other cause can be found.147153 Whereas the AHA/ASA guidelines recommend considering “prolonged cardiac monitoring to screen for intermittent atrial fibrillation,” the AAN recommends monitoring for four weeks in patients older than 40 and one to two weeks in those under 40 before referring for percutaneous closure.153 All statements stress the importance of a multidisciplinary approach to management decisions and detailed diagnostic investigation before intervention. Box 2 summarizes guidelines and position papers from major professional organizations.

Box 2

Recommendations from professional societies on patent foramen ovale (PFO) closure for secondary stroke prevention

American Heart Association/American Stroke Association secondary stroke prevention guideline 2021146

  • “In patients with a nonlacunar ischemic stroke of undetermined cause and a PFO, recommendations for PFO closure versus medical management should be made jointly by the patient, a cardiologist, and a neurologist, taking into account the probability of a causal role for the PFO (level C-EO)”

  • “In patients 18 to 60 years of age with a nonlacunar ischemic stroke of undetermined cause despite a thorough evaluation and a PFO with high-risk anatomic features, it is reasonable to choose closure with a transcatheter device and long-term antiplatelet therapy over antiplatelet therapy alone for preventing recurrent stroke (level B-R)”

American Academy of Neurology Practice Advisory Update 2020147

  • “If a higher risk alternative mechanism of stroke is identified, clinicians should not routinely recommend PFO closure (level B)”

  • “In patients younger than 60 years with a PFO and an embolic appearing infarct and no other mechanism of stroke identified, clinicians may recommend closure following a discussion of potential benefits (reduction of stroke recurrence) and risks (procedural complication and atrial fibrillation) (level C)”

  • “PFO closure may be offered to younger patients (e.g., <30 years) with a single, small, deep stroke (<1.5 cm), a large shunt, and absence of any vascular risk factors that would lead to intrinsic small-vessel disease such as hypertension, diabetes, or hyperlipidemia (level C)”

Society for Cardiovascular Angiography and Interventions Expert Consensus Statement 2019148

  • “A successful PFO program must have a rigorous process for selection in order to offer the procedure to only the patients with unexplained stroke who will benefit the most in order to mitigate risk and avoid unnecessary procedures”

  • “Patient selection should involve close collaboration between the PFO proceduralist and a neurologist (preferably a stroke neurologist)”

  • “Prior to considering PFO closure, a careful evaluation should be done to rule out other causes of stroke”

Canadian Best Stroke Practice Recommendation 2018149

  • “For patients requiring long-term anticoagulation, the decision regarding PFO closure remains unclear, and decisions should be based on individual patient characteristics and risk versus benefit profile [Evidence C]”

  • “There is insufficient evidence to make a recommendation regarding the comparative effectiveness of PFO closure vs. anticoagulant therapy”

  • “For carefully-selected patients with a recent ischemic stroke or TIA attributed to a PFO, PFO device closure plus long-term antiplatelet therapy is recommended over long-term antithrombotic therapy alone provided all the following criteria are met [Evidence Level A]: (i) age 18–60 years; (ii) The diagnosis of the index stroke event is confirmed by imaging as a nonlacunar embolic ischemic stroke or a TIA with positive neuroimaging or cortical symptoms; (iii) The patient has been evaluated by a neurologist or clinician with stroke expertise, and the PFO is felt to be the most likely cause for the index stroke event following a thorough etiological evaluation to exclude alternate etiologies”

European Stroke Organization Consensus Statement 2017145

  • “In patients aged 18–60 years old with cryptogenic stroke/TIA and with high risk PFO features (moderate or severe shunt, atrial septal aneurysm (ASA), atrial septal hypermobility) we recommend percutaneous closure plus medical therapy instead of antiplatelet therapy alone (Grade A)”

  • “In patients between 60 and 65 years, percutaneous closure plus medical therapy instead of antiplatelet therapy alone can be offered (Grade B). Percutaneous closure plus medical therapy can be considered in place of antiplatelet therapy alone also for patients aged <18 and >65 years old on an individual basis (Grade C)”

  • “Based on the few available data, percutaneous closure and oral anticoagulation (OAC) therapy seem to perform equally (Grade C). Therefore, while waiting for further evidence and based on the superiority of percutaneous closure over medical therapy as a whole, patient engagement in the choice is pivotal”

  • TIA=transient ischemic attack


Emerging treatments

As percutaneous PFO closure has been shown to be effective within narrowly defined clinical trial populations, opportunities remain to establish the utility of this therapy in other populations. In particular, the management of PFO in older people remains controversial. Older patients are more likely to harbor occult mechanisms of stroke such as paroxysmal atrial fibrillation, non-stenotic atherosclerotic carotid lesions, and aortic arch atherosclerosis. Concurrently, the incidence of venous thromboembolism also increases with advancing age,154155156157 which represents an increasingly likely substrate for paradoxical embolization. High quality observational evidence now links PFO with ESUS in older people,1920 and studies of stroke recurrence in older patients with ESUS and a PFO suggest that PFOs are associated with a meaningfully increased risk of stroke.158 Further observational work suggests that PFO closure is feasible in patients above the age of 60 in real world clinical practice.159 A trial comparing percutaneous closure with medical treatment is therefore rational in this patient population,60 especially in those without traditional vascular risk factors. Two observational studies are under way to assess the role of PFOs in older patients with a stroke or transient ischemic attack (NCT00859885) and the interrelation of PFOs and atrial fibrillation in older people (DefenseElderly; NCT04285918).

Further studies are also needed to determine which patients are most likely to benefit from PFO closure on the basis of individual patient level biomarkers. For instance, exploratory analyses of existing trial data suggest that PFO closure may not be efficacious in patients with very small shunts,160 and whether features of the eustachian valve can be used to inform clinical decision making and the risk of future stroke remains unknown.51 An individual participant meta-analysis of the recent major RCTs is under way and will examine associations between morphological PFO biomarkers (including shunt size and presence of an ASA) and benefit of treatment.132 Additionally, as PFO closure is a relatively modern therapy, the very long term stability of PFO closure devices remains unknown. Several registries are actively enrolling patients to determine the ongoing stability of these devices, including the REDUCE Post-Approval Study (NCT03821129), the Trevisio Post-Approval Study (NCT04433520), and the Amplatzer PFO Occluder Post-Approval Study (NCT03309332).

From the standpoint of medical treatment of PFO, although previous studies did not find a significant net benefit of anticoagulation over antiplatelet therapy with respect to secondary prevention of stroke in patients with PFO, such studies were collectively underpowered and the use of DOACs has not been tested in a randomized clinical trial. These drugs are associated with fewer drug-drug interactions than warfarin and confer a lower risk of hemorrhagic complications.64161 They may thus be compelling targets for further study in patients with cryptogenic stroke and PFO who do not undergo PFO closure, especially in those over the age of 60. Several observational studies suggest that the use of DOACs is feasible in patients with PFO associated stroke.159161


PFOs are common structural cardiac lesions that have been associated with cardioembolic stroke, particularly in young patients without other vascular risk factors. Options for secondary prevention after PFO associated stroke include medical management (with antiplatelet therapy or anticoagulation) or percutaneous device closure in selected patients. Percutaneous PFO closure is associated with a high relative risk reduction but a low absolute risk reduction for recurrent stroke. Before proceeding with PFO closure, other known causes of ischemic stroke should be definitively excluded, relevant morphological features of the PFO should be fully characterized, and each available treatment option corresponding to patients’ individual preferences should be considered as part of a multidisciplinary, structured decision making process.

Glossary of abbreviations

  • AAN—American Academy of Neurology

  • AHA/ASA—American Heart Association/American Stroke Association

  • ASA—atrial septal aneurysm

  • DOAC—direct oral anticoagulant

  • DVT—deep venous thrombosis

  • ESUS—embolic stroke of undetermined source

  • ICM—insertable cardiac monitor

  • pAVF—pulmonary arteriovenous fistula

  • PFO—patent foramen ovale

  • RCT—randomized controlled trial

  • RoPE—Risk of Paradoxical Embolism

  • TCD-US—transcranial Doppler ultrasonography

Research questions

  • Is percutaneous patent foramen ovale (PFO) closure of benefit in older patients?

  • Can morphological features of PFOs identify patients poised to benefit most from PFO closure?

  • What is the role of novel oral anticoagulants in the prevention of future stroke in patients with PFO?

  • Is there a subset of patients with migraine who may benefit from PFO closure?

Patient involvement

A patient of BMG’s who had had an embolic stroke of undetermined source and subsequently underwent percutaneous closure of a high risk patent foramen ovale (PFO) contributed to the drafting of this manuscript. This man and his wife (who had been extensively involved at all stages of the decision making process) reviewed an early draft of the manuscript and provided extensive comments and suggestions on its content and style. They emphasized in particular the need to promote awareness of the risk of PFO associated stroke in young patients. They also wished to highlight that cryptogenic strokes may present with mild symptoms at their first iteration and therefore can easily be misdiagnosed.


We thank Franklin Schneider (Warren Alpert Medical School of Brown University) for echocardiographic images used in figure 2, Eric Feng for the creation of figure 3, and Vinald Francis for the creation of the supplementary figure.


  • Series explanation: State of the Art Reviews are commissioned on the basis of their relevance to academics and specialists in the US and internationally. For this reason they are written predominantly by US authors

  • Contributors: All listed authors satisfy all four ICMJE authorship criteria. BMG conceived and designed the work, drafted the manuscript, and revised it for critically important intellectual content. EMO, WF, YX, SY, and HK made substantial contributions to the design of the work and revised it for critically important intellectual content. MER conceived and designed the work and revised it for critically important intellectual content. BMG, EMO, WF, YX, SY, HK, and MER provided final approval of the version to be published and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. BMG is the guarantor.

  • Competing interests: We have read and understood the BMJ policy on declaration of interests and declare the following interests: none.

  • Provenance and peer review: Commissioned; externally peer reviewed.


View Abstract