The foramen ovale is a natural interatrial channel, delimited inferiorly by the thinner flap-like septum primum and superiorly by the thicker septum secundum (Jeanrenaud & Kappenberger 1991, Movsowitz et al. 1992). During fetal life this anatomic configuration permits a functional RLS of placental oxygenated blood. As long as the pressure in the right atrium exceeds that in the left atrium, blood flows from right to left. At birth, with the development of the pulmonary circulation, the pressures are reversed and the increased pressure in the left atrium forces the septum primum to close the foramen ovale. Fusion of this membrane with the septum secundum usually follows by means of fragile fibrous adhesions during the first years of childhood. When fusion does not occur, an oblique tunnel-like opening persists between the septa, known as PFO. This differs from an atrial septum defect (ASD), which is not a valve-like structure but rather a true defect in the septum (Movsowitz et al. 1992). According to an autopsy study of 965 normal hearts, the size of the PFO varies between 1 and 19 mm, with a mean diameter of 4.9 mm (Hagen et al. 1984).
PFO usually allows RLS, but it may also allow left-to-right shunting, despite its valve-like morphology, especially when associated with left-sided cardiac lesions (Wu et al. 1993), although only rarely without such a pathology (Schneider et al. 1996). RLS through a PFO may occur under several sets of physiological conditions, such as sporting effort, blowing the nose, decompressing the ears, straining at stool, sexual intercourse, coughing and even trumpet playing (Gautier et al. 1991, Evers et al. 1998). It may occur also during positive end expiratory pressure ventilation and diving. Interestingly, RLS across a large PFO may occur during heavy exercise, producing clinically important arterial desaturation in some healthy adults (Wilmhurst et al. 1994). The VM causes a transient rise in right atrial pressure above the left atrial pressure, reversing the interatrial pressure gradient, with resultant RLS across the PFO during the straining phase (Movsowitz et al. 1992). More frequently, RLS occurs immediately on the release of the VM. It has been postulated that, on the release of the VM, systemic venous blood rushes into the right atrium, reversing the interatrial pressure gradient or exacerbating it if it already exists in the straining phase (Lynch et al. 1984). Furthermore, it has been shown that transient reversal of the interatrial pressure gradient may occur during each cardiac cycle (Langholz et al. 1991). This probably explains the echocardiographic findings of alternate bulging of an atrial septal aneurysm into the left and right atria with each cardiac cycle, and the fact that RLS has been shown to exist in the presence of PFO without any provocative manoeuvre or pulmonary hypertension (Lechat et al. 1988, Webster et al. 1984).
Pathological conditions that predispose a subject to RLS in the presence of a PFO include those in which there is a reversal of the interatrial pressure gradient or in which abnormal right atrial flow dynamics exist. Right atrial pressure may exceed left atrial pressure under conditions that raise the intrathoracic pressure, right ventricular failure (right ventricular infarction, cardiomyopathy), pathological conditions of the pulmonary artery causing right ventricular hypertension (pulmonic stenosis, pulmonary embolism, chronic obstructive lung disease, high altitude pulmonary oedema), pulmonary valve stenosis, diseases of the tricuspid valves and during cardiac tamponade (Movsowitz et al. 1992). RLS with a normal right atrial pressure may occur if an abnormal right atrial blood flow exists (cases of right atrial mass, presence of Chiari`s network or after pneumonectomy) or with multifactorial causes, as in platypnoea-orthodeoxia syndrome (Langholz et al. 1991, Movsowitz et al. 1992, Schneider et al. 1995).
In a substantial number of persons, the foramen ovale stays patent throughout life. Autopsy studies performed on adolescents and adults have found an overall incidence of 27%, but this varies with age, from a mean of 35% before 30 years to 22-27% at 30-90 years (Thompson & Evans 1930, Patten 1931, Hagen et al. 1984). Neither the incidence of PFO nor its size varies between the sexes. One study has suggested familial aggregation of PFO (Arquizan et al. 2001).
Contrast TTE studies have shown the incidence of PFO in normal subjects to vary between 10 and 18% (Lynch et al. 1984, Lechat et al. 1988, Webster et al. 1988), but a higher incidence of 26% was observed in a large epidemiological study of healthy adults assessed by contrast TEE (Meissner et al. 1999).
PFO cannot be identified by history, physical examination or chest x-ray. One study has suggested that “crochetage”, a (notch) pattern in the inferior limb leads, could be an electrocardiographic sign associated with its presence (Ay et al. 1998), but others have not been able to reproduce these findings (Tembl et al. 1998). Routine right and left cardiac catheterisations do not usually allow a definite diagnosis (Movsowitz et al. 1992).
TTE is widely available, but is less sensitive than TEE, which is currently regarded as the gold standard for the detection of PFO (Hausmann et al. 1992, Belkin et al. 1994). As a PFO is not a discontinuity in the septum but a valve-like structure, intravenous administration of echo contrast material is usually required to detect RLS. Commonly used contrast agents include air-saline microbubbles, or galactose or oxypolygelatin suspensions. As stated earlier, shunting is present under resting conditions in some persons. A provocation test is usually necessary to reverse the pressure gradient across the interatrial septum, however, and this can be either coughing or a VM. The colour Doppler technique has been used. Contrast and colour Doppler TEE techniques for the detection of PFO have been validated in an autopsy study (Schneider et al. 1996).
The contrast transcranial Doppler (TCD) ultrasound technique is based on the detection of microbubble-induced embolic signals within the intracranial arteries, usually the middle cerebral artery. Intravenous contrast medium injections are performed at rest and after VM in a similar manner to that adopted in TEE. The sensitivity of TCD varies from 68% to 89% relative to contrast TEE, and its specificity from 92% to 100% when studying stroke populations (Karnik et al. 1992, Di Tullio et al. 1993, Job et al. 1994). Only one comparison has been made between TCD and contrast TEE for the detection of PFO in healthy volunteers, giving a sensitivity of 85% and a specificity of 83% (Job et al. 1994).
The dye dilution method, first described as early as 1938 (Boehrer et al. 1993), is based on the detection of intravenously injected dye such as indocyanine green. In the case of RLS, part of the dye enters the left atrium, and thus the systemic arterial circulation, before the lung passage, producing an early deflection from the baseline before the main part of the indicator curve (Swan et al. 1954, Daly 1968, Banas et al. 1971). Earlier investigators used a number of invasive techniques to detect congenital heart defects: injecting of the dye into the inferior vena cava and sampling from the femoral artery (Scott et al. 1976), brachial artery (Gazzaniga & Dalen 1970, Meister et al. 1972) or radial artery (Nicholson et al. 1951), intracardiac injections and sensing from the femoral artery (Cheng 1975), or injections of dye at various sites in the heart and great vessels during cardiac catheterisation and non-invasive detection in the ear (Swan & Wood 1953).
The dye dilution method was used for the first time to detect a PFO by Swan et al. in 1954, there are only a few subsequent reports of its use (Daly 1968, Banas et al. 1971, Meister et al. 1972, Scott et al. 1976, Niggemann et al. 1987).
The oximetric detection of PFO is based on measuring changes in arterial oxygen saturation. In the presence of a PFO, RLS of desaturated blood occurs across the atrial septum during the first few seconds after the end of a VM, creating a transient fall in systemic arterial oxygen saturation, which can be detected in the peripheral circulation (Lee & Gimlette 1957). There appear to be no reports of the use of oximetry after the 1950s.
Several case-control studies have compared the frequency of PFO in patients with ischaemic stroke with that in controls (for reviews, see Overell et al. 2000, Chant & McCollum 2001).
In their recent meta-analysis, Overell et al. (2000) found eight positive and seven neutral or negative results of such comparisons, further analysis of which revealed that when the ages of the patients were considered, a significant difference in mean age emerged between the positive (44.8 years) and neutral or negative series (61.1 years). In nine studies which included young stroke patients aged 55 years or less (Webster et al. 1988, Lechat et al. 1988, Chen et al. 1991, de Belder et al. 1992, Cabanes et al. 1993, Jones et al. 1994, Job et al. 1994, Zahn et al. 1995, Del Sette et al. 1998), PFO was present in 40.3% of the subjects overall as compared with 17.8% of the controls, giving a significant odds ratio (OR) of 3.1 (Overell et al. 2000). Only three studies have been performed on older patients, aged above 55 years of age (de Belder et al. 1992, Jones et al. 1994, Zahn et al. 1995), and these gave an overall prevalence of PFO in the patients (16.3%) that did not differ from the control figure (13.6%) (Overell et al. 2000).
There are nine well performed case-control studies that compare the prevalence of PFO in cryptogenic stroke to that in stroke of known causes among young patients aged 55 years or less (Lechat et al. 1988, Webster et al. 1988, Jeanrenaud et al. 1990, Di Tullio et al. 1992, Cabanes et al. 1993, Ranoux et al. 1993, Job et al. 1994, Jones et al. 1994, Yeung et al. 1996). Here the meta-analysis showed a significantly higher prevalence of PFO in cryptogenic stroke (55.7%) as compared with stroke of known aetiology (17.1%), giving a significant OR of 6.0 (Overell et al. 2000). In older patients, the three studies that exist (Di Tullio et al. 1992, Jones et al. 1994, Yeung et al. 1996) point to a non-significant difference in PFO between cryptogenic stroke (27.1%) and stroke of known origin (14.0%) (Overell et al. 2000).
There have been five comparisons of the prevalence of PFO among young cryptogenic stroke patients (aged 55 years or less) and non-stroke controls (Lechat et al. 1988, Webster et al. 1988, Cabanes et al. 1993, Job et al. 1994, Jones et al. 1994), the overall prevalence of PFO being significantly higher in the cryptogenic stroke cases (54.6% vs. 19.9%), with an OR of 5.0 (Overell et al. 2000). In contrast to this, the two available comparisons of the same kind among persons aged over 55 years (Vella et al. 1991, Jones et al. 1994) show no difference in the prevalence of PFO between the cryptogenic stroke cases (11.6%) and controls (13.4%).
In conclusion, the current evidence clearly shows a significant association of PFO with ischaemic stroke in general, and especially with cryptogenic stroke, among patients younger than 55 years. No firm conclusions can be made among older patients because of the limited data. There are several reports (Homma et al. 1994, Job et al. 1994, Steiner et al. 1998) demonstrating a more significant association of a large PFO than a small one with stroke, and it has been shown that stroke patients with a large PFO show more brain imaging features of embolic infarcts than those with a small PFO (Steiner et al. 1998). These results suggest a “dose-response” relationship and support a causality link between PFO and ischemic stroke (Overell et al. 2000). No prospective study has been made of the PFO-stroke association, however.
Kasper et al. (1992) observed that the presence of PFO was associated with significant arterial hypoxaemia and a high incidence of cerebral and peripheral ischaemic events in patients with haemodynamically significant pulmonary embolism. In a further study, the group found patients with a PFO and major pulmonary embolism to have a particularly high risk of death or arterial thromboembolic complications (Konstantinides et al. 1998).
Patients with severe chronic obstructive pulmonary disease have an increased prevalence of PFO relative to controls, and their PFO has more pronounced systemic arterial oxygen desaturation associated with it than is observed without PFO (Soliman et al. 1999).
There is an increased prevalence of PFO in patients with obstructive sleep apnoea. RLS across the PFO can contribute to significant systemic arterial hypoxaemia after a VM in up to a third of these patients (Shanoudy et al. 1998).
PFO may act as a pathway for the arterialisation of venous gas bubbles. RLS through a PFO is thought to be associated with neurological, cutaneous and cardiorespiratory decompression sickness as a result of paradoxical gas embolism (Wilmhurst et al. 1989). Furthermore, the presence of a large RLS has been shown to be associated with multiple brain lesions in magnetic resonance imaging (MRI) in the case of sport divers (Knauth et al. 1997). On the other hand, it has recently been found that diving itself is associated with the presence of ischaemic lesions rather than diving in the presence of a PFO (Schwerzermann et al. 2001), although divers with a PFO have an increased risk of decompression sickness events and suffer more ischaemic lesions in the brain than do divers without a PFO. PFO may also be associated with hypobaric decompression sickness in high altitude aviators and astronauts (Kerut et al. 2001). Venous air embolism, a complication of surgery above the level of the heart in which non-collapsible veins are present at the operative site, may serve as an embolic source of paradoxical embolism through the medium of a PFO. Although rare, this may be a serious complication of posterior fossa surgery (Movsowitz et al. 1992).
Paradoxical fat embolism can occur during invasive intramedullary orthopaedic procedures. In a survey of 111 consecutive operations on 110 patients for fractures of the femur and tibia or hemiarthroplasty of the hip, Christie et al. (1995) observed paradoxical embolisation in four cases. All of them involved major embolic phenomena, leading to pulmonary hypertension before the embolic material was observed to be passing through a PFO. Two of the patients died. The exact mechanism of fat embolism is unclear, however, since systemic embolisation was also observed in one patient in the absence of a septal defect.
This rare syndrome refers to dyspnoea induced by assuming an upright position and relieved by returning to a recumbent position (platypnoea), or accentuated arterial hypoxaemia experienced in a standing position that is improved by lying down (orthodeoxia). It has been shown to be associated with PFO and RLS through the lesion despite normal pressure in the right side of the heart. It has been claimed that surgical closure of the PFO leads to clinical improvement. (Robin & McCauley 1997).
There have been several suggestions of an association between PFO and migraine with aura (Del Sette et al. 1998, Anzola et al. 1999, Wilmhurst & Nightingale 2001). Interestingly, it has been suggested that there may be a subgroup of patients who have severe migraine associated with major RLS in whom closure of the PFO may improve or abolish the migraine (Wilmhurst et al. 2000).
One study has suggested that paradoxical embolism in patients with a PFO may play a role in causing transient global amnesia (Klötzsch et al. 1996).
Paradoxical embolism of any systemic artery is probably possible, and cases have been demonstrated in many locations, such as the coronary, renal, splenic, peripheral extremity and spinal cord arteries (Loscalzo 1986, Mori et al. 1993, Carey et al. 1999).
PFO that permits paradoxical embolism of a thrombus will also allow shunting of other emboli such as tumour or hepatic tissue. Even cerebellar tissue embolism in the coronary artery and a case of bullet embolism after a gunshot wound in the leg have been reported (Thompson & Evans 1930, Johnson 1951). Septic embolism is a cause of cerebral abscess in patients with congenital heart defects and RLS, but reports of an association between PFO and cerebral abscesses are scarce (Wilmhurst & de Belder 1994, Kawamata et al. 2001).