PATIENTS AND METHODS

Patients

We identified all patients with ischaemic stroke, transient ischaemic attacks (TIAs), or amaurosis fugax, who were admitted to our university based stroke centre between January 1994 and July 2000.

The diagnosis of stroke was based on a focal neurological deficit and the corresponding findings on CT or MRI.¹⁵ A TIA was defined as a focal neurological deficit resolving completely within 24 hours.¹⁶ Transoesophageal echocardiography (TEE) had been performed in all patients suspected of paradoxical embolism via right to left shunt. After excluding those with a concurrent aetiology for the cerebrovascular event we identified 318 patients who had a PFO and otherwise unexplained ischaemic stroke or TIA.

The patients were offered several treatment options to prevent a recurrent stroke: endovascular or surgical closure of the patent foramen ovale, or long term antithrombotic treatment. One hundred and fifty nine decided to have endovascular or surgical closure of their PFO and 159 patients chose medical treatment (vitamin K antagonists: n=79, platelet inhibitors: n=80). The patients with medical treatment represent the population of our study. Patients after endovascular closure have been published in part previously.¹⁷

Transoesophageal echocardiographic criteria for PFO and ASA

The diagnosis of PFO and ASA was based on contrast transoesophageal echocardiography. The right to left shunt was judged to be minimal if there were less than five bubbles, moderate if there were six to 20 bubbles, and severe if there were more than 20 bubbles detected in the left atrium.¹⁸ Atrial septum aneurysm was diagnosed when the interatrial septum was abnormally redundant with an excursion of more than 10 mm into the right or left atrium.¹⁹

Vascular risk factors

The following factors considered to increase the risk for stroke or TIA were assessed: hypertension, defined as systolic blood pressure greater than 160 mm Hg, diastolic blood pressure greater than 95 mm Hg, or both; diabetes mellitus defined as symptoms of diabetes plus random blood glucose concentration greater than 11 mmol/l, or fasting blood glucose greater than 7 mmol/l; cigarette smoking; hypercholesterolaemia defined as a total venous plasma cholesterol concentration greater than 5 mmol/l; obesity defined as a body mass index (BMI) greater than 25 kg/m²; coronary artery disease, and positive personal and familial history for cerebrovascular events.

Follow up

Follow up data were obtained retrospectively by reviewing the medical records of all patients and the relevant information concerning recurrent events and treatments. Thereafter, the family physicians of all patients and afterwards the patients themselves were contacted. In order not to miss any recurrent event a standardised telephone questionnaire was used. Follow up data could not be obtained from two patients because of address changes. Transient ischaemic attacks (confirmed by a neurologist), systemic embolism, ischaemic stroke (confirmed by CT), and death were considered retrospectively as end points of follow up. Twenty patients, who had been on antithrombotic treatment for a prolonged period and then decided for some reason to have endovascular or surgical closure of the PFO, were included in the analysis for the time they had been treated medically.

Statistics

Continuous variables are expressed as mean (SD). Nominal variables were compared by χ² test for contingency tables. Statistical significance was assumed at a value of p<0.05. Actuarial analysis of freedom from recurrent cerebral ischaemic events was calculated according to the Kaplan-Meier method. Average annual event rates were calculated according to the formula 1-(1-P)^1/n, where P equals the cumulative event rate at n years of follow up.²⁰ The log rank test was used for univariate analysis of independent variables (age, sex, obesity, coexistence of an ASA, size of the right to left shunt, positive personal history for cerebrovascular ischaemic events, and type of medical treatment) on the rate of recurrence. All analyses were performed with SPSS 10 for MacIntosh statistical software, copyright © 2001, SPSS Inc.

RESULTS

The study population consisted of 93 men and 66 women. Their mean age was 50.7 (13.5) years (range 15 to 77 years). One hundred and twenty one patients (76.1%) were 60 years old or younger at the time when they experienced the ischaemic event that led to the diagnosis of a PFO (fig 1).

This event was a transient ischaemic attack in 38 patients (23.9%), an ischaemic stroke in 119 (74.8%), and an amaurosis fugax in two patients (1.3%). In 90 patients (56.6%) it occurred in the carotid territory and in 69 patients (43.4%) in the vertebrobasilar territory. Forty four patients (27.7%) had experienced more than one cerebrovascular ischaemic event before the diagnosis of the patent foramen ovale.

The vascular risk factors that were considered to increase the risk for stroke or TIA are summarised in table 1. Hypertension, diabetes mellitus, cigarette smoking, hypercholesterolaemia, and familial history for stroke occurred with similar frequencies in patients with one and in those with multiple previous cerebrovascular events. Obesity was more frequent in the group with more than one prior event (p=0.02). There were five patients with coronary artery disease in the group with one cerebral ischaemic event before diagnosis of the PFO. Fifty of the 159 patients (31.4%) did not show any vascular risk factors.

The right to left shunt was minimal in 19 patients (11.9%), moderate in 44 (27.7%), and severe in 96 (60.4%) patients. Thirty three patients (20.7%) had an ASA in addition to the PFO.

Antithrombotic treatment such as anticoagulation, aspirin, or clopidogrel was given according to the judgement of the attending physician after the event leading to the diagnosis of PFO. Seventy nine patients received oral anticoagulant drugs. The target international normalised ratio (INR) was 2.0–3.0. In 26 of them the oral anticoagulation was replaced by antiplatelet drugs after 12 (SD 11) months. Platelet inhibitors were given to 80 patients (77 patients were treated with aspirin at a mean dosage of 233 (SD 83) mg/day, three patients received clopidogrel at a dosage of 75 mg/day). Nine patients were non-compliant and discontinued antithrombotic therapy. No bleeding complications were seen either in patients treated with oral anticoagulants or in those with platelet inhibitors.

Mean follow up was 29 (23) months. During this period 21 patients (13.4%) had a recurrent cerebrovascular event, seven strokes and 14 TIAs. Six strokes occurred in patients who were taking aspirin and one stroke in a patient with anticoagulant drugs. Five TIAs occurred in the aspirin group, six in the oral anticoagulant group, and three in patients who had discontinued the antithrombotic therapy. χ² Tests did not show an association between the treatment regimen and the risk for recurrence (p=0.99). However, six of the seven strokes occurred in patients, who were treated with aspirin. Despite the absence of clear statistical evidence, there was a trend indicating an increased risk for stroke recurrence in this group (p=0.053, two tailed Fisher's exact test).

Neither the severity of the right to left shunt, nor the coexistence of an ASA in addition to the PFO were associated with an increased risk for stroke recurrence.

The overall cumulative estimate of risk of stroke or TIA recurrence, at 2 years of follow up, in the whole cohort of patients with PFO was 9.2% (95% CI 6.5% to 11.9%). When only strokes were considered, the cumulative risk of recurrence was 4.5% (95% CI 2.5% to 6.5%), at 2 years of follow up.

The average annual rate of stroke recurrence was 1.8% and that of both recurrent strokes or TIAs was 5.5% (table 2). When patients with multiple ischaemic events before diagnosis of PFO were analyzed separately, the average annual rate of recurrent strokes was 3.6% and the rate of recurrent strokes or TIAs was 9.9%. The log rank test for survival data showed a significantly higher risk for recurrent events in this group than in patients without ischaemic events before the qualifying event (p=0.02). However, obesity was more common in the group with multiple ischaemic events.

When patients without obesity were analyzed separately, the risk for recurrent events in the subgroup with multiple ischaemic events before diagnosis of PFO was still significantly higher than in the subgroup with first ever stroke or TIA (p=0.02).

Figure 2 shows the Kaplan-Meier survival curves of patients with one event and those with multiple cerebral ischaemic events before diagnosis of PFO.

DISCUSSION

Previous studies have shown an association between PFO and cryptogenic stroke. However, optimal secondary stroke prevention for such patients is not known because randomised trials have not been performed to date.

There are three major studies that assessed the rate of recurrent cerebrovascular events during 2 to 3 years of antithrombotic treatment (table 3). Bogousslavsky et al reported 3.8% of recurrent, non-fatal cerebral ischaemias (stroke and/or TIA)/year during 3 years of prospective follow up in 140 patients.²¹ Mas et al detected 3.4% annual recurrent events (strokes and/or TIAs) during 2 years of retrospective follow up in 132 patients. A subgroup of 38 patients with an ASA in addition to the PFO showed an increased rate of 9.0% strokes and 22% of TIAs and/or strokes after 2 years.¹³ De Castro et al reported 7.2% strokes or TIAs at 3 years of prospective follow up in 74 patients.¹⁰ Our 157 symptomatic patients with PFO presented 9.2% recurrent strokes or TIAs and 4.5% strokes at 2 years of retrospective follow up. The average annual rate was 5.5% for strokes and TIAs and 1.8% for strokes only. The rates we found tended to be slightly higher than in the other studies. One reason might be the age of the patients, which exerts an independent effect on the relation between interatrial septal abnormalities and strokes.⁸ Bogousslavsky et al and Mas et al excluded patients older than 60 years whereas 24% of our patients were older than that. However, there was no association between age and stroke recurrence in our series.

A relevant finding of the present study is that patients with a positive personal history of stroke or TIA before the event leading to the detection of the PFO had a higher rate of recurrent cerebral ischaemia (table 2 and fig 2). Obesity was more frequent in the group with multiple ischaemic events before diagnosis of PFO. However, the higher recurrence rate was unlikely to be associated with body weight. When patients with BMI less than 25 kg/m² were analyzed separately, the risk of recurrence remained significantly higher in patients with multiple ischaemic events than in those with one stroke or TIA before diagnosis of PFO. Mas et al and De Castro et al excluded patients with multiple ischaemic events from their studies and Bogousslavsky et al tended to treat such patients surgically.

Several authors have found that atrial septal hypermobility and ASA combined with PFO and also the size of the PFO are independent risk factors for stroke.^9,22,23 Our data do not support this finding. Neither the coexistence of ASA nor the size of the right to left shunt was associated with an increased recurrence rate in our series. A possible explanation of this disagreement could be the relatively small number of only 33 patients with ASA (21%) in our series. Methodological factors might also play a major part in the proper assessment of the size of a right to left atrial shunt, with difficulties comparing the data of different studies. Grading is usually performed during provocative manoeuvres (Valsalva's manoeuvre or cough) and therefore may depend on the strain exerted by the examined patient, the delay after the beginning of contrast injection, etc. Attempts to standardise the provocative manoeuvres have been shown to improve the sensitivity of the methods used for identification of the PFO.²⁴ In addition, microbubbles are counted from a single two dimensional imaging plane that does not necessarily reflect the exact amount of all the microbubbles shunted into the left atrium. Therefore, results from grading of the right to left shunts should be interpreted with caution.

Our results do not favour any medical treatment regimen. Six strokes occurred in patients with aspirin and one in a patient with anticoagulation; however, this difference was not significant. The PFO In cryptogenic stroke study (PICSS), an ongoing substudy of the warfarin/aspirin recurrent stroke study (WARSS), is about to address this question. It compares aspirin and anticoagulation in patients with PFO and cryptogenic strokes.²⁵

To summarise, our study confirms a risk of recurrent strokes and TIAs that is similar to the rates of previously published series of patients with PFO and cryptogenic strokes. Patients who had experienced multiple cerebrovascular events before diagnosis of PFO were at increased risk of recurrent cerebral ischaemia.