The Rationale for PFO Closure: A Series of Arguments For or Against

Abstract

During the last decade, patent foramen ovale (PFO) and intermittent or permanent right-to-left shunting of venous blood has been proposed to play a pathogenetic role in a number of disorders. This review covers the most common and sometimes controversial indications for closure of the PFO. It considers the safety of the procedure, current evidence that supports closure, and counter arguments. It then offers a considered position for each indication.

Introduction

In recent years, there has been a great deal of attention paid to the patent foramen ovale (PFO). In particular, many have accused it of being responsible for a number of complex conditions, such as cryptogenic stroke and migraine.^1,2  For the majority, PFO will remain asymptomatic with no major risk of complications. It is for that reason that PFOs are not systematically closed when detected. However, it is the most common defect diagnosed in young patients that present with an unexplained cerebrovascular event; it is present in about 50% of migraine-with-aura (MA) patients and in patients suffering from cryptogenic stroke; divers with a PFO are 4.5 times more likely to experience decompression sickness than those without a PFO;⁴ it is four times more prevalent in individuals who are susceptible to high-altitude pulmonary edema than in those who are not; and it may be the cause of right-to-left shunting that is responsible for various other forms of hypoxia/hypoxemia. PFO would, therefore, seem to be implicated in a number of medical conditions. In this article, we seek to elucidate the role that the PFO may play in these conditions by referencing literature. We will also consider any circumstances that could justify closure of the PFO.

PFO Closure Complications

Before we decide if PFO closure is an appropriate therapy for our patient, we should be aware of the risks. In a review of literature, Khairy⁵ found an incidence of 7.9% and 1.5% of minor and major complications, respectively, in relation to device closure. We have found that in some of the earlier reports of PFO closure, complication rates varied significantly and were often associated with device designs that may have been superseded. In a more recent report⁶ of a randomized trial comparing Amplatzer (AGA Medical Corporation, Plymouth, Minnesota), CardioSEAL-STARFlex (NMT Medical, Inc., Boston, Massachusetts), and Helex (W.L. Gore and Associates, Flagstaff, Arizona) devices, complication rates were not insignificant. Each of the 3 devices gave rise to episodes of atrial fibrillation (1.4%, 0.9%, and 0.9%, respectively). Just over half of these resolved spontaneously; the remainder required the use of antiarrhythmics. Device embolization, hemopericardium, and transient ischemic attack (TIA) occurred with the Helex device (1.4%, 0.5% and 0.5%, respectively); pericardial tamponade necessitating surgical removal, fever, and peripheral vein thrombosis occurred with the Amplatzer device (0.5%, 0.9% and 0.5%, respectively); and thrombus on the device, paroxysmal supraventricular tachycardia, and fever occurred with the CardioSEAL-STARFlex device (3.6%, 0.5% and 0.5%, respectively). Furthermore, two CardioSEAL-STARFlex devices needed to be removed through auxiliary vascular access because of unsatisfactory positioning. Less serious complications relating to the anticoagulant medication were also reported for the Amplatzer and Helex devices — both at 4.5%. Moderate to severe residual shunt rates at 30 days were 4.1%, 5.5%, and 11.4% for the Amplatzer, CardioSEAL-STARFlex, and Helex devices, respectively. This report does show that the PFO closure procedure is not without risk and that it may not wholly resolve shunting. It is necessary to consider the risk-to-benefit ratio when deciding on the appropriate therapy for the patient. Anatomical features such as Chiari network or prominent eustachian valve may also need to be considered before deciding if closure is appropriate. Such features, in combination with PFO, seem to increase the risk of paradoxical embolism.^7–9 They may contribute to MA⁹ and affect optimal closure device positioning. As technology evolves, devices hopefully become safer relative to the risk of complications.

PFO and Cryptogenic Stroke

An ischemic stroke is classified as cryptogenic if no cause can be defined following an adequate diagnostic evaluation. The basis of such an evaluation was used to classify stroke sub-types in a study of low molecular weight heparin and has become known as the Trial of ORG 10172 in Acute Stroke Treatment (TOAST) classification.¹⁰ Prevalence of cryptogenic stroke. Findings published from the National Institute of Neurological and Communicative Disorders databank showed that the prevalence of cryptogenic stroke in a population 45 years of age, followed for a median of 5.1 years, Meissner¹⁸ found that those with a PFO were at no greater risk of stroke or ischemic attack than those without a PFO. It was acknowledged that those patients with an ASA could be at greater risk. However, in the editorial comment, Meier¹⁹ suggested that the study is limited by a lack of sufficient patient numbers to draw meaningful conclusions. Di Tullio et al,²⁰ in a population study of subjects > 39 years of age and mean follow-up of 6.6 years, reached conclusions similar to those of Meissner. The number of patients with PFO was also similar. This higher prevalence does not seem to translate into a greater risk in these relatively small PFO populations. However, there have been several reports on the effect of PFO closure in stroke patients (secondary prevention).

What is the effect of closure of the PFO? There have been no published data from a prospective randomized clinical trial in stroke patients comparing the effect of PFO closure to medical therapy. To our knowledge, there are currently five such studies that are actively recruiting patients. Four are sponsored by PFO closure companies and are being conducted in the United States (2), Europe (1), and in both (1). The results from these studies are not expected to be available before 2010, at best. The fifth study is sponsored by the Assistance Publique — Hôpital de Paris, and is being conducted in France. This study is not expected to be completed until the end of 2012. Justification for PFO closure in cryptogenic stroke patients could be provided by the results of these studies. However, these results are unlikely to be available in the near term.

Is there any justification for PFO closure in these patients? Various authors have published their findings on the recurrence of stroke or TIA. In a meta-analysis, Khairy⁴ compared recurrent events in patients that had transcatheter closure of their PFO with patients receiving medical therapy. Homma²¹ provided some additional data. We have updated this comparative information, but we only included prospective studies of a reasonable number of patients and only one report if multiple reports existed from the same center, with the exception of Schuchlenz et al.⁷⁷ In this case, two different sets of patients are analyzed. Overall event rates for both sets of patients were calculated as a percentage of the total number studied. We have also calculated overall annual recurrence rates for a more relevant comparison. Note that anti-platelet treatment has also been usual in those who had their PFO closed. Although the studies may differ with respect to inclusion and exclusion criteria, etc., there was a difference in event rates between PFO closure and medical therapy. It could be argued that the number of events in a relatively short follow-up period is likely to be low, biasing the PFO closure group. We, therefore, took respective data from these two tables based only on those studies with a reported follow-up period > 24 months. When having a follow-up period of > 24 months, there also appears to be a difference between PFO closure and medical therapy, in favor of PFO closure.

Should we close the PFO? We believe that in spite of the absence of a multicenter-controlled study, there is a strong case for the closure of a PFO in a young patient following a cryptogenic stroke or TIA. The strongest case exists in patients with a co-existing ASA.

PFO and Migraine

Prevalence of PFO in migraine patients. In the past decade, there has been growing evidence that patients with migraine, particularly those with aura, are more likely to have a PFO.  While there is currently no proof of a cause-effect relationship, these reports show a higher incidence of PFO in MA patients.^81–83

Migraine and stroke. Migraine, especially MA is an independent risk factor for stroke,²² and the higher incidence of PFO in these individuals may suggest a causative relationship. Pierangeli et al²³ show the strong association between PFO with ASA and stroke, and the strong association between PFO with atrial septal aneurysm and MA. As the authors point out, it is tempting to suggest that the migraine may be caused by paradoxical embolism; which, in turn, may be responsible for the increased stroke risk. Medication to minimize migraine will not necessarily be effective in reducing the risk of paradoxical embolism. In a review of the literature, Bousser et al²⁴ conclude that while migrainous infarcts are known to occur and a bidirectional relationship between MA and cerebral ischemia may be suggested, migraine itself is too complex to allow simple interpretations to be made. Kruit etl al²² reported that in the cerebellar region of the posterior circulation, migraine patients were more likely to demonstrate infarcts, and particularly at higher migraine frequencies and in MA. Female patients were more likely to have deep white matter lesions (WML) irrespective of migraine type. Vermeer²⁵ has shown that in an elderly population, the presence of WML and silent brain infarcts are associated with an increased risk of stroke. However, there has been no study showing if there is any inter-relationship between the presence of WML and silent brain infarcts and the risk of stroke in migraine patients. In a recent study performed in individuals affected by MA (n = 87), WML were present in 61% and right-to-left shunts in 45% of the patients.²⁶ The presence of WML did not correlate with any migraine clinical feature, whereas the presence, number, and volume of WML increased with subjects’ age. There was no significant difference in the total volume and number of WML in the group with and without right-to-left shunts. The authors conclude that the presence of a right-to-left shunt does not increase the likelihood of finding WML in migraineurs.²⁶

Effect of device closure. We have reviewed published retrospective data on the effect of closing the PFO or ASD on patients with migraine. These patients have presented for closure because they had experienced a neurological event or decompression sickness.  Patients in these studies presented for closure for indications such as stroke, decompression sickness, and hypoxemia. Rreports are based on migraineurs with a PFO only and a comparison is made with migraineurs whose PFO was not closed. While many of these reports seem to support PFO closure in migraine patients, all reports are single-center experiences. No patients were blinded to their treatment, nor were they randomized to closure or non-closure. A large number are retrospective in respect of migraine assessment, which, in itself, is unreliable because patients do not easily recall the timing and severity of migraine events. There has been only one prospective, multicenter, randomized, double-blind, sham-controlled clinical trial designed to assess the effect of PFO closure on a migraine headache in patients 18 to 60 years of age with a history of frequent, disabling MA having failed at least two classes of preventive medication because of inefficacy or intolerance (Migraine Intervention with STARFlex® Technology [MIST I Trial]).²⁷ This study was conducted in the UK only and was sponsored by a PFO closure device company. Aspirin and clopidogrel were given to all patients as a loading dose in the 24 hours before the procedure (300 mg each) and for 90 days after the procedure (75 mg each daily). The primary efficacy end point was migraine headache cessation. The anticipated cessation rate was 40% in the implant group, compared with 15% of the sham group during the analysis phase (days 91 to 180). Secondary efficacy comparisons were incidence of migraine during the healing phase, change in the severity of migraine attacks based on Migraine Disability Assessment Scale (MIDAS) (over a 3-month retrospective period), and Headache Impact Test (HIT-6) (over a 1-month retrospective period) scores. No significant difference was observed in the primary end point of migraine headache cessation between implant and control (sham) groups (3 of 74 versus 3 of 73, respectively; P = 0.51). Secondary end points did not differ significantly between groups for either the intention-to-treat or per-protocol populations. Most patients in both groups reported PFO and Decompression Sickness Decompression sickness in scuba divers results from the release of nitrogen bubbles from solution in venous blood as the pressure is released. For a diver with a PFO or ASD, it is possible for paradoxical embolization of bubbles to occur. This leads to rapid onset of neurological decompression sickness, even when divers adhere to decompression algorithms. Literature reports focusing exclusively on this type of subject are relatively few. Torti²⁸ studied 230 scuba divers, 63 of whom had a PFO. Although they showed that the risk of suffering decompression sickness is five times higher in divers with a PFO, the absolute risk is still low. However, they did find that the risk increases with increasing rate of shunting during Valsalva. Germonpré et al²⁹ provide a comprehensive review of the subject of decompression sickness. He shows why there are not more frequent reports of decompression sickness from divers with a PFO: diagnosis of decompression sickness is not easy and symptoms may be overlooked by the diver, or it may be assumed that the diver’s ascent did not conform to decompression protocol. These factors limit routine PFO diagnosis and closure. However, he discloses the results of one of his earlier studies, which showed that divers are able to increase rates of shunting through diving over a 6–8 year period. He suggests that this potentially increases the risk of decompression sickness.

Should we close the PFO? We conclude that routine diagnosis of a PFO is warranted only in cases where divers have taken precautions to limit the risk of decompression sickness through adherence to diving algorithms, and yet they experience frequent and/or severe cases of decompression sickness. If diagnosed in these cases, the PFO should be closed. Whether PFO is the reason behind unexplained deaths among divers has not been thoroughly studied.

PFO and Hypoxia

Hypoxemia. There are many reasons for a reduced level of arterial blood oxygen. The partial or total association with an ASD or PFO can be assumed if the defect is diagnosed and right-to-left shunting can be demonstrated.

Mechanical ventilation patients. There have been reports of hypoxemia in mechanically ventilated patients due to the increased level of shunting that results. Cujec³⁰ showed that patients with a PFO experience little or no benefit from positive end-expiratory pressure compared to those without a PFO. Baldwin³¹ failed to diagnose the PFO in one patient, but in a second, the patient survived through prompt diagnosis and closure of the PFO. Ventricular-assist device patients. Right-to-left shunting may be increased during use of a ventricular assist device. Nguyen³² describes a patient on left-ventricular device support whose degree of shunting was not deemed to be the principle cause of his hypoxemia. It was not until the PFO was first occluded then closed that the arterial oxygen saturation values normalized. Loeffelbein et al³³ suggest that the negative pressures necessary for blood aspiration may open a PFO, which would remain closed under normal circumstances.

Right-ventricular infarction patients. Hypoxemia due to right-to-left shunting may be a complication in right ventricular infarction. Although rare, Bansal,³⁴ Sterling,³⁵ Matsuo,³⁶ and Crawford³⁷ are among those that have reported such cases, with resolution following closure.

Adult respiratory distress syndrome patients. Dewan et al³⁸ and Legras et al³⁹ show that the presence of a PFO can contribute significantly to the hypoxemia experienced in a patient with adult respiratory distress syndrome (ARDS).

Chronic obstructive pulmonary disease patients. Hacievliyagil et al⁴⁰ found during Valsalva a PFO in 23 of 52 patients with chronic obstructive pulmonary disease (COPD) (44%) and in 10 of 50 healthy controls (20%).

Platypnea-Orthodeoxia Syndrome Symptoms. Platypnea-orthodeoxia is a rare condition. The term is derived from platypnea (flat breathing) and orthodeoxia (dyspnea on assuming an upright posture). Patients exhibit normal oxygen saturation levels while in a recumbent position, but exhibit a reduction when assuming other postures, and/or during exertion. Platypnea-orthodeoxia is most commonly recognized in patients with major pulmonary disorders such as emphysema, recurrent pulmonary embolism, ARDS, pulmonary atriovenous malformations, post-pneumectomy, hepato-pulmonary syndromes, or cirrhosis of the liver. However, it may also manifest in patients with an atrial septal defect or PFO that permits right-to-left atrial shunting. In patients with a PFO, various mechanisms have been suggested to explain this process, including physiologically transient spontaneous reversal of the left-to-right atrial pressure differential, preferential blood flow from the inferior vena cava toward the atrial septum as a part of the remnant prenatal circulation, or a physiologic change in the relationship of the compliance of right and left ventricles.⁴³ The mere adoption of an upright posture may stretch the intra-atrial communication, facilitating right-to-left shunting, due to a less compliant or stiffer right ventricle. When standing, there is a drop in the right ventricular filling pressures, making the left ventricle relatively more compliant while maintaining right ventricular compliance. In addition, during early diastole, the compliance of the right ventricle is reduced, thus offering greater resistance to blood flow from the right atrium. This leads to enhancement of the right-to-left shunting across the septal defect without elevation of the right heart pressures. Reduction of intravascular volume can change the right ventricular compliance and cardiac output, thereby increasing the right-to-left shunting. Mercho et al⁴⁴ report two cases of platypnea-orthodeoxia following pneumonectomy and discusses reasons for how this can exacerbate right-to-left shunting, principally by anatomical distortion.

Therapy options. Many patients with platypnea-orthodeoxia syndrome have co-morbid conditions that make surgical correction of the ASD/PFO a risky procedure. By contrast, percutaneous device closure offers a safer option in patients without pulmonary hypertension.

Device closure. There have been several reports of percutaneous device closure of PFO in patients with platypnea-orthodeoxia syndrome, many as single-case studies. Those with a greater number of patients include an early report⁴⁵ of successful closure and consequential increase in oxygen saturation in 8 patients. In follow up, all living patients remained free from shunt-related arterial desaturation and symptomatology. Godart et al⁴6 closed 11 patients with small ASDs or PFOs presenting with hypoxemia, 6 having associated platypnea-orthodeoxia. During follow up (up to 30 months), no patient experienced any episode of desaturation due to inter-atrial shunting. Rao et al⁴⁷ studied device closure in 10 patients with ASD or PFO. Oxygen saturation levels were found to be acceptable in all patients at final follow up. Delgado et al⁴⁸ successfully closed 18 patients using devices. Complete resolution of symptoms was seen in all patients, despite a small shunt in 2 patients and a moderate shunt in 1 other on follow up. Guérin et al⁴⁹ reported on the results from 78 patients in a multicenter French registry. Closure was successful in 76 patients (97%). On follow up, 6 patients were found to have small residual shunts without symptoms or cyanosis. Oxygen saturation increased immediately after PFO occlusion in all patients and was normal in 78%. Cyanosis observed in all patients in an upright position before occlusion was only 24% after occlusion (those patients with pulmonary associated pathology.) Dyspnea decreased significantly after PFO closure. Five patients still had marked symptoms with no residual shunt, but these symptoms were probably related to the underlying lung disease.

Should we close the PFO? We conclude that if a PFO is detected in patients with platypnea-orthodeoxia syndrome, percutaneous closure is justified.

Obstructive Sleep Apnea

Intermittent hypoxemia results during obstructive sleep apnea syndrome (OSAS). Because the events occur during a period of sleep, when the metabolic processes are different than those during a period of wakefulness, the consequences are varied and certainly result in an increased risk of cerebrovascular disease (ischemic stroke and TIA)^,50–55 and cardiovascular disease.^56–60 Patients with severe depletion of arterial oxygen require continuous positive airway pressure (CPAP). The prevalence of PFO in OSAS patients was observed to be higher (69%) than in patients without sleep apnea (17%).⁶¹ Furthermore, saturated arterial oxygen levels after a Valsalva maneuver fell significantly more in OSAS patients with a PFO than in those without a PFO. Beelke et al⁶² found a prevalence of PFO of 27% and 15% in OSAS patients and control patients, respectively — still significantly different, but at lower values, in spite of Valsalva maneuver provocation. Johansson et al,⁶³ using the proportion of desaturation to respiratory events calculated as the ratio of oxygen desaturation index (ODI)/apnoea–hypopnoea index (AHI), studied 15 cases with high proportional desaturation (ODI/AHI > 0.66) that were individually matched with 15 controls with low proportional desaturation.

Discussion

It is clear that PFO may allow interatrial right-to-left shunting of venous blood. This may lead to or worsen various disease states. We have made an attempt to summarize the information available with respect to possible indications for PFO closure. In the case of ischemic stroke, indications for closure vary between the US and Europe. In the US, as there are several ongoing trials, closure of PFO is currently recommended only with recurrent stroke while on warfarin, or with stroke and contraindication to chronic warfarin therapy, unless in the setting of a clinical trial. In Europe, closure is more commonly undertaken in many cryptogenic stroke patients. This review shows that resolution of cryptogenic stroke, hypoxemia, platypnea-orthodeoxia and decompression sickness is possible through closure of the PFO. In all cases, the procedural risks must be taken into account before deciding to close. However, the many questions on the conceivable relationship between PFO and diseases like migraine and obstructive sleep apnea remain to be answered. If the correct trials to investigate the effect of PFO closure can be designed and supported, much scientific knowledge can be achieved and eligible patients treated.

Dr. Dahlöf is from Gothenburg Migraine Clinic, Gothenburg, Sweden, and Prof. Søndergaard is from Rigshospitalet, Copenhagen, Denmark. Mr. Hannam is a consultant in private practice.

Manuscript submitted February 18, 2009, provisional acceptance given March 18, 2009, article accepted April 16, 2009.

Disclosure: Mr. Hannam disclosed that he is consultant to St. Jude Medical. Correspondence: Peter H. Hannam, FIMMM, MRSC, 64 The Boulevard, Worthing, West Sussex, BN13 1LA. United Kingdom. E-mail: peter.hannam@btconnect.com.