Article Text

Download PDFPDF

Interventional cardiology
Long-term results after fluoroscopy-guided closure of patent foramen ovale for secondary prevention of paradoxical embolism
  1. A Wahl1,
  2. M Kunz1,
  3. A Moschovitis1,
  4. T Nageh1,
  5. M Schwerzmann1,
  6. C Seiler1,
  7. H P Mattle2,
  8. S Windecker1,
  9. B Meier1
  1. 1
    Cardiology Department, University Hospital Bern, Switzerland
  2. 2
    Neurology Department, University Hospital Bern, Switzerland
  1. Professor B Meier, Cardiology Department, Swiss Cardiovascular Centre Bern, University Hospital, 3010 Bern, Switzerland; bernhard.meier{at}


Objectives: To carry out long-term follow-up after percutaneous closure of patent foramen ovale (PFO) in patients with cryptogenic stroke.

Design: Prospective cohort study.

Setting: Single tertiary care centre.

Participants: 525 consecutive patients (mean (SD) age 51 (12) years; 56% male).

Interventions: Percutaneous PFO closure without intraprocedural echocardiography.

Main outcome measures: Freedom from recurrent embolic events.

Results: A mean (SD) of 1.7 (1.0) clinically apparent embolic events occurred for each patient, and 186 patients (35%) had >1 event. An atrial septal aneurysm was associated with the PFO in 161 patients (31%). All patients were followed up prospectively for up to 11 years. The implantation procedure failed in two patients (0.4%). There were 13 procedural complications (2.5%) without any long-term sequelae. Contrast transoesophageal echocardiography at 6 months showed complete closure in 86% of patients, and a minimal, moderate or large residual shunt in 9%, 3% and 2%, respectively. Patients with small occluders (<30 mm; n = 429) had fewer residual shunts (small 11% vs large 27%; p<0.001). During a mean (SD) follow-up of 2.9 (2.2) years (median 2.3 years; total 1534 patient-years), six ischaemic strokes, nine transient ischaemic attacks (TIAs) and two peripheral emboli occurred. Freedom from recurrent stroke, TIA, or peripheral embolism was 98% at 1 year, 97% at 2 years and 96% at 5 and 10 years, respectively. A residual shunt (hazard ratio = 3.4; 95% CI 1.3 to 9.2) was a risk factor for recurrence.

Conclusions: This study attests to the long-term safety and efficacy of percutaneous PFO closure guided by fluoroscopy only for secondary prevention of paradoxical embolism in a large cohort of consecutive patients.

View Full Text

Statistics from

Since the initial description of a fatal stroke in a young woman with a patent foramen ovale (PFO) by Cohnheim in 1877, the PFO, which remains probe-patent throughout adulthood in about a quarter of the general population,1 has been increasingly recognised as potential mediator of several disease manifestations. These include paradoxical embolism, refractory hypoxaemia due to right-to-left shunt in patients with right ventricular infarction or severe pulmonary disease, orthostatic desaturation in the setting of the rare platypnoea-orthodeoxia syndrome, neurological decompression illness in divers and, more recently, migraine with aura and even high-altitude pulmonary oedema. Percutaneous closure of the PFO, first described in 1992 for secondary prevention of paradoxical embolism,2 has been shown safe and feasible in several studies,310 and its clinical efficacy appeared favourable compared with medical treatment.11 12 Long-term follow-up after percutaneous PFO closure in patients with cryptogenic stroke is not well established.


Patient group

Between April 1994 and September 2004, 525 consecutive patients with PFO and one or more ischaemic stroke, transient ischaemic attack (TIA), or peripheral embolic event presumably related to paradoxical embolism underwent percutaneous PFO closure guided by fluoroscopy only, without intra-procedural echocardiography. Patients undergoing PFO closure for other indications than paradoxical embolism (diving accidents, migraine, etc) were excluded from the analysis. The study protocol was approved by the local ethics committee, and patients gave written informed consent. An embolic event was considered due to paradoxical embolism when the following criteria were fulfilled: (a) presence of PFO with or without atrial septal aneurysm (ASA) with spontaneous or inducible interatrial right-to-left shunt during contrast transoesophageal echocardiography (TOE); (b) clinically and/or radiologically confirmed ischaemic stroke, TIA or peripheral embolism; and (c) exclusion of any other obvious cardiac, aortic or cerebrovascular cause.


The diagnosis of PFO and ASA was based on contrast TOE, with aerated colloid solution injected into an antecubital vein at the end of a sustained Valsalva manoeuvre. PFO was defined as a flap-like opening in the atrial septum secundum, with the septum primum serving as a one-way valve allowing for permanent or transient right-to-left shunt. ASA was diagnosed as abnormally redundant interatrial septum with an excursion of ⩾10 mm into the right or left atrium and a diameter of the base of the aneurysm of at least 15 mm.13 Spontaneous or provoked right-to-left shunt was semiquantitatively graded according to the amount of bubbles detected in the left atrium after crossing the interatrial septum on a still frame: grade 0  =  none, grade 1  =  minimal (1–5 bubbles), grade 2  =  moderate (6–20 bubbles) and grade 3  =  severe (>20 bubbles).14 Care was taken to document the actual passage of contrast bubbles through the rent but this was not possible in all cases. In three patients, not included in this report, the PFO suspected by contrast TOE was subsequently ruled out by angiography and mechanical probing. In these patients, TOE had failed to depict a true bubble passage through the gap of the PFO, and the mere presence of bubbles in the left atrium close to the foramen ovale had been the basis for assuming a PFO.

Percutaneous PFO closure

The procedure was performed under local anaesthesia and fluoroscopic guidance only as described previously.3 Intra-procedural guidance by TOE or intracardiac echocardiography (ICE), as recommended by others,4 69 was not used. However, all patients underwent TOE before the intervention for initial diagnosis of PFO, as described above. Balloon sizing was not performed. Indeed, the maximal opening of the flap-like PFO is not instrumental for the success of closure. Moreover, there is a finite risk of laceration of a thin septum primum.15 When amplatzer PFO occluders were used, a 25 mm device was selected for all cases except those with particularly large ASA. In 5% of cases, this device was exchanged for a larger one during the procedure because of suboptimal anchoring (negative Pacman sign16). Patients were released to full physical activity a few hours after the procedure, and treated with acetylsalicylic acid 100 mg once daily for 6 months for antithrombotic protection until full device endothelialisation. Transthoracic contrast echocardiography was performed within 24 hours of percutaneous PFO closure to assess for device mal-position. A contrast TOE was repeated 6 months after percutaneous PFO closure to assess for a residual shunt following endothelial overgrowth.

Follow-up evaluation

The outcome after the intervention was prospectively assessed for up to 11 years. All patients were scheduled for a 6-month TOE. When there was a significant residual shunt, a repeat TOE at 1 year was recommended. If the shunt persisted at that time, implantation of a second device was recommended. Thereafter, patients underwent structured telephone interviews, dealing with recurrent embolic events, device-related problems and health status, at regular intervals. Follow-up information was available for all patients at some point in time, but 12 patients (2%) were subsequently lost to follow-up owing to address changes. Death, recurrent ischaemic stroke, TIA or peripheral embolism were considered end points. Patients with suspected recurrent cerebrovascular events were re-examined by a neurologist, and a new imaging study of the brain (computed tomography or magnetic resonance imaging) was performed.

Statistical analysis

Continuous variables are expressed as mean (SD), and were compared by a two-sided, unpaired t-test. Categorical variables are reported as counts and percentages, and were compared by χ2 analysis. Estimates for freedom from recurrent TIA, stroke, peripheral embolism and the composite of TIA, stroke and peripheral embolism were obtained using the Kaplan–Meier method. The log rank test was used for univariate analysis of independent predictors of recurrence. Estimates of the hazard ratio (HR) and 95% confidence intervals (CIs) for each independent variable were obtained by proportional hazard regression analysis. Significance was assumed with a p value <0.05. All data were analysed with the use of SPSS software (version 12.0.1, SPSS Inc.).


In-hospital outcome

Table 1 summarises the demographic data.

Table 1 Baseline clinical characteristics

A total of eight different atrial septal occlusion devices were used depending on historical device availability (table 2).

Table 2 Implanted devices

Percutaneous PFO closure failed in two patients (0.4%) during our early experience. In one patient, the procedure (a Sideris device was planned) was aborted before device delivery because of laceration of the femoral artery during initial placement of an 11F venous sheath with an ensuing retroperitoneal haematoma. This required surgical revision, at which time the PFO was closed surgically. In another patient with PFO and a large ASA, an Amplatzer ASD occluder embolised into the pulmonary artery 12 hours after the procedure. The device was retracted percutaneously into the femoral vein with an Amplatzer retrieval basket, and removed from there by local incision. Repeat PFO closure was not attempted.

Periprocedural complications, including the ones described above, were seen in 13 patients (2.5%), and included embolisation of the device or parts of it with successful percutaneous removal in five patients (two counter-occluder of Sideris devices, two PFO-STAR devices, one Amplatzer ASD occluder), air embolism with transient symptoms in three patients (one Angel-Wings and two PFO-STAR devices), pericardial tamponade requiring pericardiocentesis in one patient (PFO-STAR), and vascular access site problems in four patients (two Sideris, one PFO-STAR and one Amplatzer PFO occluder, all of them with simultaneous coronary angiography). There were no procedural deaths, and none of the procedural complications resulted in long-term sequelae. In the last 100 cases with the Amplatzer PFO occluder, mean (SD) procedure time was 26 (11) minutes (median 25 minutes) and mean (SD) fluoroscopy time was 4.1 (2.8) minutes (median 3.2 minutes).

Patients with occluder devices categorised as small (<30 mm; n = 429 patients) had fewer procedural complications (small 1.6% vs large 6.3%; p = 0.008) than patients with larger devices (⩾30 mm; n = 96). Patients with PFO and an associated ASA (n = 161; 31%) had similar device success (99.4% vs 99.7%; p = 0.55) and complication rates (1.9% vs 2.7%; p = 0.55) than patients with an isolated PFO (n = 364; 69%). Older patients (⩾55 years; n = 224) also had similar device success (100% vs 99.3%; p = 0.22) and complication rates (1.8% vs 3%; p = 0.38) as younger patients (<55 years; n = 301).

Transthoracic contrast echocardiography within 24 hours of percutaneous PFO closure detected a residual shunt in 18% of patients.

Late outcome

Complete PFO closure as assessed by contrast TOE at ⩾6 months was achieved in 86% of patients, whereas a minimal, moderate, or large residual shunt persisted in 9%, 3%, or 2% of patients, respectively (fig 1). Patients with small occluder devices (<30 mm; n = 429 patients) had fewer residual shunts (small 11% vs large 27%; p<0.001) than patients with larger devices (⩾30 mm; n = 96). Patients with PFO and an associated ASA (n = 161; 31%) had similar residual shunt rates (16% vs 13%; p = 0.40) as compared with patients with an isolated PFO (n = 364; 69%). Older patients (⩾55 years; n = 224) also had similar residual shunt rates (14% vs 14%; p = 0.97) as younger patients (<55 years; n = 301).

Figure 1 Patent foramen ovale (PFO)-mediated interatrial shunt at baseline and 6 months after percutaneous closure of the PFO, as assessed by contrast transoesophageal echocardiography.

At 6 months’ follow-up, TOE examination showed a thrombus on the device in four asymptomatic patients. Two patients (one PFO-STAR, one Amplatzer PFO occluder 35 mm) had a small thrombus on the left atrial disc, which resolved after 3 months of oral anticoagulation. One patient (Amplatzer PFO occluder 25 mm) had a probable tiny thrombus on the left atrial disc, which remained unchanged after 4 months of oral anticoagulation. One patient had a 20×7 mm thrombus adherent to the right atrial disc (Amplatzer PFO Occluder 35 mm) which resolved after 6 months of oral anticoagulation. Ten months after cessation of oral anticoagulants, TOE showed a recurrent right atrial thrombus, which resolved once again after oral anticoagulant therapy during 6 months, without further recurrences. The last echocardiography at 7 years’ follow-up was normal.

A total of 14 patients (with two Sideris, one Angel-Wings, two Amplatzer ASD, three Amplatzer PFO and six PFO-STAR devices in place) underwent implantation of a second device (two Sideris, one CardioSEAL, two Amplatzer ASD and nine Amplatzer PFO occluder) owing to a significant residual shunt through the former PFO. In all of these patients, TOE showed the initial closure device in the correct position, but a residual shunt in the region of the former PFO. At fluoroscopy, the guidewire crossed the interatrial septum beside the initial device, probably through a residual gap in the PFO tunnel. This suggests that the usually slit-like opening was incompletely covered by the first device, as opposed to an unrecognised additional atrial septal defect. No periprocedural complications occurred during the second intervention. After implantation of the second device, complete closure was finally achieved in 12/14 patients (86%). One patient treated with two Sideris devices had a minor residual shunt 6 months after the second intervention, which was no longer apparent at 4 years. One patient (Amplatzer ASD occluder followed by Amplatzer PFO occluder ) had a minor residual shunt at 6 months, which persisted at 2 years. Another patient (two Amplatzer PFO occluder 25 mm) still had a moderate residual shunt 9 months after the second intervention.

Furthermore, in a patient with a PFO grade III associated with a large ASA, TOE 2 years after implantation of an Amplatzer PFO occluder 35 mm (performed owing to persistence of a residual shunt after 6 months) disclosed a new tiny atrial septal defect at the lower rim of the device, probably corresponding to an erosion of the interatrial septum17 owing to wear and tear of the ASA undulating incessantly between the right and left disc of the device. In another patient, routine TOE 6 months after implantation of an Amplatzer PFO occluder 25 mm showed a completely occluded PFO, but a new small atrial septal defect was seen at the lower rim of the device. In both cases, these iatrogenic small atrial septal defects were successfully closed using an Amplatzer PFO occluder 25 mm. There were no further device related complications, in particular no erosions of the atrial walls.

During a mean (SD) 2.9 (2.2) years of follow-up (median 2.3 years, total 1534 patient- years), three deaths (one road traffic accident, one cancer, one cirrhosis), six ischaemic strokes, nine TIAs, and two peripheral emboli were reported. Freedom from the composite end point of recurrent ischaemic stroke, TIA, or peripheral embolism was 98% at 1 year, 97% at 2 years, and 96% at 5 and at 10 years, respectively (fig 2). The presence of a residual right-to-left shunt after transcatheter treatment of PFO (HR = 3.4; 95% CI 1.3 to 9.2; fig 3), and procedural complications (HR = 10.4; 95% CI 3.0 to 36.3) were predictors for recurrent embolic events. Gender, older age (⩾55 years), multiple events before PFO closure, the presence of an ASA associated with the PFO, arterial hypertension, diabetes mellitus, smoking status, family history, hypercholesterolaemia, and device size did not adversely affect the outcome.

Figure 2 Freedom from the composite end point of recurrent transient ischaemic attack, ischaemic stroke, or peripheral embolism after percutaneous closure of patent foramen ovale.
Figure 3 Freedom from the composite end point of recurrent transient ischaemic attack, ischaemic stroke or peripheral embolism after percutaneous closure of the patent foramen ovale. Patients are stratified according to the presence or absence of a residual shunt.


As far as we know, this study is the largest series of consecutive patients treated at a single centre with the longest follow-up reported to date to investigate the safety, feasibility and long-term clinical efficacy of percutaneous PFO closure, and to analyse the risk factors for recurrence. In addition, the most economical approach to PFO closure was used, affording a lean and short intervention, both for the patient and for the operator. The principal findings are (a) safety and feasibility of percutaneous PFO closure with the simple technique described (no echocardiographic guidance) were confirmed in a large series; (b) using contemporary devices, such as the Amplatzer PFO occluder, complication rates were <1%, and high rates of complete PFO closure could be achieved; (c) larger devices seemed to be associated with higher complication and residual shunt rates; (d) excellent long-term results in the secondary prevention of paradoxical embolism were documented; (e) important differences were found between the different PFO closure devices used, some of them being clinically relevant; (f) a residual shunt after transcatheter treatment of PFO was a risk factor for recurrence; (g) an ASA associated with the PFO had no influence on device success, or on the risk of periprocedural complications, the residual shunt rate or the risk of recurrent events; (h) concern has been raised that the current focus on cryptogenic stroke as an indication for PFO closure may deprive the elderly who have the highest risk18 of paradoxical embolism of a simple preventive treatment.19 20 In this series, the procedure proved as feasible, safe and effective in preventing recurrent embolic events in selected older patients (⩾55 years) as in younger patients; (i) device-related late events were rare and without clinical sequelae.

Transcatheter treatment of patients with cryptogenic stroke and PFO has been shown to be safe and feasible using a variety of occlusion devices, both with4 69 and without echocardiographic guidance.3 5 10 Routine TOE guidance provides little additional information to that which can be gleaned from a hand injection of contrast medium16 in a profile-adjusted view (fig 4). TOE is poorly tolerated by the supine patients, and comes therefore at the cost of sedation or general anaesthesia, which considerably lengthens the procedure. ICE is a costly alternative that is more comfortable for the patient, but it may increase vascular complications because of additional venous access. The procedure can be performed on outpatients, with total procedure times <30 minutes, and fluoroscopy times <5 minutes. In this large series of PFO closure without intra-procedural TOE or ICE guidance, device success was close to 100%, and the periprocedural complication rate was 2.5%. Importantly, most complications occurred when we were first using the method with older devices, which reflects in part a learning curve.3 None of these complications could have been avoided by additional echocardiographic guidance.

Figure 4 Fluoroscopic picture with hand injection of contrast medium to ascertain correct placement (Pacman sign16) of a 25 mm Amplatzer PFO occluder before release. The appearance of the device and of the septum secundum (SS) is a reminder of the arcade figure Pacman. The left (cranial) halves of the discs are separated by the thick muscular septum SS, while the thin septum primum (SP) is undulating between the right (caudal) halves of the discs. The patent foramen ovale (PFO) channel is indicated by the arrow. LA, left atrium; RA, right atrium.

The success and complication rates depend on the device used.10 21 In 396 patients implanted with an Amplatzer PFO occluder, device success was 100%, and the complication rate 0.25% (one arteriovenous fistula requiring elective surgical closure). While larger devices are easier to implant, and preferred by most operators in the case of larger PFOs or associated ASAs, there are concerns about the risk of impairment or erosion of adjacent structures. On the other hand, smaller devices fit more snugly into the fossa ovalis, and may thus be more likely to close the PFO completely. However, they are more likely to be embolised or to incompletely cover a slit-like PFO or a cribriform septum primum. In this non-randomised comparison most likely biased towards smaller devices (eg, only 26% of patients receiving a smaller device had an associated ASA vs 51% for larger devices; p<0.001), smaller devices (<30 mm) were associated with fewer complications and fewer residual shunts.

During long-term follow-up, the risk of recurrent events after transcatheter treatment of PFO with or without associated ASA was <1% a year. This excellent efficacy compares favourably with medical treatment.2226 Of note, recurrent embolic events occurred up to 27 months after PFO closure, which is considerably longer than the 8 months reported by Braun et al.9 A residual shunt after transcatheter treatment of PFO was a risk factor for recurrence.3 5 10 This lends further support to the theory of paradoxical embolism being the most prevalent mechanism for vascular events in this patient population. Therefore, one should aim for complete PFO occlusion in order to maximise therapeutic efficacy, a goal which can be achieved in >90% of patients using contemporary devices.7 8

Patients with both PFO and ASA constitute a high-risk population with a three- to fivefold increased risk for recurrent embolic events compared with patients with PFO alone.27 Furthermore, secondary prevention with acetylsalicylic acid alone has been found insufficient for protection against recurrent cerebrovascular events in patients with both PFO and ASA.23 25 If paradoxical embolism is assumed to be the most likely stroke mechanism in these patients, surgical or transcatheter treatment of PFO associated with ASA constitute alternative treatment options. Procedural success, periprocedural complication and residual shunt rates of transcatheter treatment of ASA associated with PFO were similar to the treatment of patients with isolated PFO.10 The long-term risk of recurrent embolic events after transcatheter treatment of ASA associated with PFO was also comparable with that of patients with PFO alone.10 This high-risk patient population might thus derive particular benefit from transcatheter treatment.

Most importantly, during the longest follow-up reported to date, which extended beyond 4 years in 147 patients (28%), long-term device-related events were rare, and none had clinical sequelae. In particular, thrombi were seen in four patients (0.8%), and a new tiny atrial septal defect at the lower rim of the device, probably corresponding to an erosion of the interatrial septum, was seen in two patients (0.4%). No further device-related complications were seen, in particular no erosions of the atrial walls, and no need for subsequent surgical removal of a device. This is an important finding for an interventional treatment of a condition that has a low annual event rate in natural history.


The following limitations should be considered in the interpretation of our results: (a) The diagnosis of PFO-mediated paradoxical embolism is presumptive, and therefore PFO and cryptogenic stroke may coexist without causal relation in certain patients. Percutaneous PFO closure in such patients will not influence recurrent cerebrovascular events, a circumstance contributing to the small recurrence rate despite successful PFO closure in our series and those of others. (b) Percutaneous PFO closure was performed using eight different device types during different time periods. Device allocation was non-randomised, and left to the discretion of the operator. Contemporary devices for percutaneous PFO closure, such as the Amplatzer PFO occluder, achieve complete PFO occlusion in about 90% of patients and are associated with complications in <1% of cases. Since a residual shunt and procedural complications were risk factors for recurrence, these improvements in device performance are likely to have a positive impact on clinical outcome in the future. They are honoured in our current device selection. (c) While percutaneous PFO closure is safe and feasible, with remarkable speed and ease of implantation using contemporary devices, and thus continuously growing patient numbers, it has to be emphasised that its true therapeutic efficacy as adjunct or alternative to medical treatment can only be ascertained by randomised studies.


View Abstract


  • Competing interests: BM: research grant and speaker bureau for AGA Medical.

  • Ethics approval: Obtained from local ethics committee.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.