Article Text

Download PDFPDF

Original article
Efficacy of percutaneous closure of patent foramen ovale: comparison among three commonly used occluders
  1. R Thaman1,
  2. G Faganello1,
  3. J R Gimeno2,
  4. G V Szantho1,
  5. M Nelson1,
  6. S Curtis1,
  7. R P Martin1,
  8. M S Turner1
  1. 1Bristol Congenital Heart Centre, Bristol Royal Infirmary, Bristol, UK
  2. 2Dept of Cardiology, University Hospital Virgen Arrixaca, Murcia, Spain
  1. Correspondence to Dr R Thaman, Adult Congenital Heart Unit, Bristol Heart Institute, Level 7, Bristol Royal Infirmary, Bristol BS2 8HW, UK; rajesh.thaman{at}


Background Percutaneous closure of patent foramen ovale (PFO) is standard treatment for patients with paradoxical embolism but studies examining the efficacy of the various occluders are lacking.

Objective To evaluate short- and medium-term closure rates of three common occluders.

Methods One hundred and sixty-six adults (47±12 (18–81 years)) were evaluated with transthoracic bubble echocardiography before and after PFO closure. Only patients with large PFOs were included (>30 bubbles in the left heart after Valsalva).

Results Three occluders were used: Amplatzer (AGA Medical Corporation) (n=80, 48%), Gore Helex (n=48, 29%) and Premere TM (St Jude Medical) (n=38, 23%). One (0.6%) neurological event occurred during follow-up. At 6 months significant residual shunting after Valsalva was highest in the group that received the Helex (58.3%), and lower for Premere (39.5%) and Amplatzer (32.5%). At final follow-up residual shunting remained higher in patients with the Helex (33.3%) than in Premere (18.5%) and Amplatzer (11%). Amplatzer had a significantly lower residual shunt rate than Helex (p<0.05 at 6 months and final follow-up). The Premere had an intermediate residual shunt rate. Septal aneurysm also predicted residual shunting (RR=24.7, 95% CI: 8.2 to 74.4, p<0.0001).

Conclusions Percutaneous PFO closure is an efficacious progressive treatment but closure rates also depend on the presence of aneurysm and differ between occluders.

  • Patent foramen ovale
  • septal occluder
  • stroke
  • transient ischaemic attack (TIA)
  • echocardiography-contrast
  • device closure

Statistics from

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.


Percutaneous closure of patent foramen ovale (PFO) is increasingly considered to be the preferred treatment for young patients who have had a cryptogenic stroke owing to paradoxical embolism. Since Bridges first description of the procedure in 1992 an increasing number of occluders have been introduced in order to facilitate PFO closure.1 Studies examining the comparative efficacy of occluders, however, are limited. Our objective was to evaluate short- and medium-term PFO closure rates of different occluders used in a single referral centre.


One hundred and sixty-six consecutive patients aged 47±12 (18–81 years) underwent PFO closure at the Bristol Royal Infirmary, UK between January 2006 and March 2008. The indications for PFO closure were one or more transient ischaemic events or stroke, peripheral embolism, an episode of decompression illness or orthodeoxia. All patients with either a transient ischaemic event or stroke had been evaluated by a neurologist or stroke physician and had undergone either CT or MRI at the discretion of the physician. All patients had a negative investigation with 24 h Holter, carotid Doppler and haematological tests, also at the discretion of the physician. The presence of a PFO was confirmed using transthoracic bubble contrast echo with multiple Valsalva and only patients with a large right to left shunt were included.

Agitated saline bubble contrast echocardiography

Echocardiographic images were obtained using Vivid 7 (GE Medical Systems, Vingmed Ultrasound, Horten, Norway). The contrast mix consisted of a mixture of 8 ml saline, 1 ml aspirated blood and 1 ml air. This was forced quickly and repeatedly between the two Leuer lock syringes and then injected intravenously. The first injection was performed during normal breathing and then up to five Valsalva manoeuvres were performed.

Images (harmonic assisted) were obtained in the apical four-chamber view and stored as video tape and/or digital loops. The images were then analysed, the frame in which the greatest number of bubbles was visible in the left heart was frozen and the number of bubbles counted. The PFO was considered small if ≤15 bubbles, moderate if 16–29 bubbles and large if ≥30 bubbles were counted.2 3

Implantation technique

All closures were performed under general anaesthesia with transoesophageal echocardiography (TOE) and fluoroscopic guidance. Patients were given aspirin 300 mg and clopidogrel 300 mg after the procedure. Patients who were taking warfarin, stopped it 3 days before the procedure and treatment was started with aspirin 300 mg/day. Access was gained via the femoral vein using a modified Seldinger technique, patients then received intravenous heparin (100 IU/kg). The position and morphology of the PFO and atrial septum were determined using TOE and fluoroscopy coupled with balloon sizing. TOE was performed using a 5–7 MHz multiplane transducer (Acuson; Siemens Medical Systems, Erlangen, Germany) and a Vivid 7 (GE) or IE 33 (Philips Ultrasound, Bothell Washington, USA) using a standard technique. On TOE the septum was said to be aneurysmal if there was a >10 mm excursion of the atrial septum, a tunnel PFO was reported if the distance from the flap of the valve to the septal wall of the PFO was ≥10 mm.3–6 The PFO was crossed using a MPB3 catheter directly or with an 0.035” guidewire. The catheter and then an Amplatz superstiff wire were advanced into the left upper pulmonary vein and the catheter exchanged for a sizing balloon (AGA Medical Corporation, 24 mm). The maximal recorded balloon waist diameter on TOE or fluoroscopy was taken as the diameter of the PFO. The occluder was then chosen and delivered according to the instructions for use.

Patients were discharged the following day after clinical examination, chest x-ray examination and echocardiography. Patients were given clopidogrel 75 mg/day for 3 months and aspirin 300 mg/day for at least 6 months.

Choice of device

The operators were all experienced at PFO closure. The most appropriate occluder for the morphological variant of the PFO was selected taking into account the balloon waist size and anatomy of the PFO. The atrial size and relation to the mitral valve and aortic root and valve as well as pulmonary and systemic inflows were also considered. In general, attempts were made to avoid a large device impinging on surrounding structures—in particular, the aortic root and valve, leading to potential future complications. If some impingement on the aortic root was likely then a less rigid occluder was preferred. In the cases of septal aneurysm we chose to close the defect rather than the entire aneurysm where possible. The Premere occluder was not used when the balloon waist size was >11 mm.


Patients were followed up initially 6 weeks after the procedure and then at 6-monthly intervals thereafter. At each review patients were evaluated with history, clinical examination, 12-lead ECG, transthoracic echocardiography bubble contrast study (not at 6 weeks) until the defect was considered to have closed (≤5 bubbles crossing right to left). At this stage the aspirin dose was reduced to 75–150 mg/day in patients with cryptogenic transient ischaemic attack (TIA)/stroke or peripheral embolus. In those patients in whom the indication for PFO closure was decompression illness or orthodeoxia the aspirin was stopped.

Statistical methods

Continuous variables were expressed as mean±SD and were compared by a two-sided, unpaired t-test. Categorical variables are reported as counts and percentages (%). Groups were compared using Fisher's exact test. Statistical significance was defined as a p value ≤0.05. Logistic regression analysis was used for multivariate analysis of independent variables (age, sex, device type, PFO size, presence of septal aneurysm) on the rate of closure. Estimates of relative risk (RR) and 95% CIs were obtained.


Patient demographics

Patient demographics are summarised in table 1. The indications for PFO closure were cryptogenic stroke/TIA (n=144, 87%), decompression illness (n=18, 11%), orthodeoxia (n=2, 1%) and peripheral embolism (n=2, 1%). Of those patients with a cryptogenic stroke/TIA, 106 (74%) patients had more than one event; the total number of embolic events was 264.

Table 1

Patient demographics and pre-procedural data

Before PFO closure no patient complained of chest pain, two patients were in New York Heart Association class ≥III, six (4%) patients complained of palpitation but no arrhythmias were recorded. Fifty-three (32%) patients reported migraine headache before procedure and in 36 (22%) migraine was associated with aura.

The septum was aneurysmal in 45 (27%) patients and a tunnel was reported in 29 (17%) patients.

Procedure results

Three different occluders were used: Amplatzer (AGA Medical Corporation, Golden Valley, Minnesota, USA) (n=80, 48%), Gore Helex, Gore Medical, Arizona, USA) septal occluder (n=48, 29%) and Premere (St Jude Medical, St. Paul, Minnesota, USA) septal occluder (n=38, 23%). Of the 80 patients who received an Amplatzer occluder, 14 (17.5%) with a large PFO received an atrial septal defect occluder (9–16 mm). An occluder was deployed successfully in all patients and no patient received more than one. Immediate procedure-related complications were seen in four (2.4%) patients and consisted of femoral pseudoaneurysm (n=1, 0.6%), femoral arteriovenous fistula (n=1, 0.6%) and transient atrial tachycardia (n=2, 1.2%). One additional patient developed a haematoma 2 days after the procedure but this was not associated with either a pseudoaneurysm or fistula. All patients completed 6 Months' echocardiographic follow-up and 159 (96%) patients underwent echocardiography at 12 months or more.

Clinical follow-up

The mean duration of clinical follow-up was 17.2±8.1(6.3–38.5 months). After PFO closure nine (5.4%) patients complained of transient chest pain, 12 (7.2%) patients complained of transient palpitation but only one patient had atrial fibrillation confirmed on 24 h Holter. No patient reported a change in New York Heart Association functional class. Only one (0.6%) neurological event was recorded: this patient reported transient right-arm weakness 12 months after PFO closure with a Helex septal occluder; this patient had normal echocardiography with negative bubble study and a CT scan was also normal.

Device-related outcome

Device-related outcomes are shown in table 2. The indication for PFO closure was similar in the three groups. Mean PFO balloon waist size was significantly smaller in patients undergoing PFO closure with the Premere occluder compared with the Amplatzer or Helex occluders (p<0.01 and <0.01, respectively). The Premere occluder was used significantly more than the Amplatzer (p<0.01) or Helex (p<0.01) in patients where tunnel-like anatomy was seen. There was a trend towards using the Amplatzer and Helex occluders compared with the Premere in patients with septal aneurysm (p=0.1 and 0.06, respectively). There was no significant difference in screening or fluoroscopy times between the three occluders.

Table 2

Procedural data

At 6 months significant residual right-to-left shunting was highest in the group that received the Helex (58.3% (large 45.8%, moderate 12.5%)), and lower for Premere (39.5% (large 23.7%, moderate 15.8%)) and Amplatzer (32.5% (large 17.5%, moderate 15%)). This was also the case at 12 months and final follow-up: residual right-to-left shunt rates at final follow-up were: Helex (33.3% (large 25%, moderate 8.3%)), Premere (18.5% (large 13.2%, moderate 5.3%)) and Amplatzer (11% (large 7.5%, moderate 3.75%)). The Helex was associated with a significantly higher residual shunt rate than the Amplatzer (p<0.05 for 6 months and final follow-up). The Premere had an intermediate residual shunt rate but this did not reach statistical significance in our sample.

Prognostic variables

An analysis of clinical predictors for residual right-to-left shunting at 6 and 12 months showed that the risk of residual shunting was not influenced by age, sex or maximal stretched PFO balloon waist size. Univariate and multivariate predictors of residual right-to-left shunting were septal aneurysm and the type of occluder used. The presence of septal aneurysm was associated with a higher likelihood of residual shunting at 6 and 12 months (RR=24.7, 95% CI 8.2 to 74.4, p<0.001). The Amplatzer was associated with a lower residual shunt rate than the Helex (RR=0.19, 95% CI 0.06 to 0.60, p<0.01). There was no significant difference between Amplatzer and Premere or between the Premere and Helex at 12 months.


Of the 36 patients with migraine and aura, 20 (55.6%) reported a significant improvement in symptoms—that is, resolution or significant reduction in frequency, two of whom continued to have an aura but without any headache. Fifteen (41.7%) patients reported no change and one (2.8%) reported a transient worsening of symptoms. Residual shunts were seen in five (25%) of the patients who reported an improvement in symptoms and in seven (47%) patients with no reported change in symptoms after PFO closure.

Of the 17 patients who reported migraine without aura, six (35.3%) reported a significant improvement in symptoms and 11 (64.7%) reported no change. Residual shunting was seen in three (50%) patients despite improvement in symptoms and in five (45%) patients with no change


This study has three main important findings. First, for patients with appropriate indication, percutaneous PFO closure is efficacious and associated with low recurrent embolic event rate. Second, PFO closure is safe with a low morbidity. Finally, residual right-to-left shunting after PFO closure is dependent on the presence of septal aneurysm and different occluders also have different closure rates; this may have implications for the choice of occluder in certain situations.

PFO and stroke

The relation between PFO and stroke has been well demonstrated by a number of studies.7–9 In patients with cryptogenic stroke and PFO, recurrent neurological events have been reported in around 6–8% of patients depending on the age and size of the PFO and 8–15% in the presence of septal aneurysm or septal hypermobility.4 10 11 Although to date there have been no reported randomised trials, observational studies in patients with cryptogenic stroke have demonstrated that percutaneous PFO closure is at least as effective as conventional medical treatment in reducing the incidence of recurrent cerebral ischaemic events, with a recurrence rate of around 3–4% after 2–3 years of medical treatment (although 8% in patients receiving aspirin)1 3 12–19 and between 0 and 3.5% after percutaneous PFO closure according to recent studies.14 20–23 As well as cryptogenic stroke or transient ischaemia, right-to-left shunting through PFO has also been related to decompression illness,24 25 arterial hypoxia via platypnoea orthodeoxia syndrome26 and migraine.27 28

The results of this study provide further evidence for the clinical efficacy of PFO closure. This study examined a high-risk cohort as reflected by the high incidence of recurrent neurological events before PFO closure, atrial septal aneurysm and the fact that only patients with large shunts were included. In this cohort only one patient had a recurrent event after PFO closure and this was transient and not associated with any abnormality on CT scanning. Similarly there were no subsequent decompression illnesses in divers; however, patients with previous decompression episodes were advised to refrain from high decompression stress diving until they had a small (≤15 bubbles) shunt only at follow-up. In the two patients with orthodeoxia, PFO closure was associated with an immediate increase in systemic O2 saturation, implying successful shunt reduction.

Complications associated with PFO closure

Recent studies have reported lower complication rates than earlier studies, with the main morbidity relating to vascular access injury.14 29–31 In this study one patient developed an arteriovenous fistula and one a pseudoaneurysm owing to inadvertent injury to the femoral artery. In order to reduce this complication we now perform ultrasound-guided femoral vein puncture. There were no deaths and we did not observe any device fractures, displacements or erosions. The lower incidence of complications seen in this and other more recent studies may relate to the use of more contemporary occluders with easier implantation techniques, a reduction in size and improvements in delivery systems and newer technology allowing for repositioning of occluders if required. Other reasons might also include increased experience in deployment and the more judicious use of antiplatelet treatment. Nine (5.4%) patients complained of a new onset of chest pain after the procedure but this was short lived. Twelve (7.2%) patients reported palpitation; however, transient atrial fibrillation was seen in one individual only. Finally, complications need to be weighed up against the risks of long-term anticoagulation. In various series fatal haemorrhagic events have been reported in 2–7% of patients with cerebrovascular disease treated with oral anticoagulation.32–34

Residual shunting

In this study persistence of right-to-left shunting was associated with factors related to the type of occluder used and the morphology of the septum. A number of different occluders have been developed and modified in order to facilitate PFO closure. These differ in their design and technical properties. In this study occluders that are in common use within Europe were used. The basic design and principles behind each of these are similar, although the delivery systems and methods used to deploy them differ. In general, each occluder consists of a double disc, plate or anchor (consisting of a fabric or mesh attached to a framework) with a small connecting central waist. Complete PFO closure depends on growth of endothelium to cover the device and the PFO. The factors which influence this endothelialisation process are unknown but the PFO occluder must flatten and stabilise the interatrial septum aiding in the process of endothelialisation.

In this study the Amplatzer occluder had a better short- and mid-term closure rate than the Helex septal occluder. The Premere septal occluder had an intermediate closure rate, though the differences were not statistically significant. The reasons for differences in closure rate are unclear. One possible explanation might have been that the Amplatzer and Premere were used for the less anatomically challenging defects. However, this did not appear to be the case, in fact the Amplatzer was often used for the most challenging defects. There are a number of reasons for this including its ease of delivery and the ability to recapture and reposition the device if required and for the larger PFOs an atrial septal defect occluder was thought to be more appropriate. Similarly, the Premere was often used in patients with tunnel PFO (owing to its design which often suits tunnel morphology), though this device was not used in patients with a balloon diameter >11 mm. Another possible explanation is that the endothelialisation process may be faster in patients with these occluders because of their greater rigidity and ability to oppose the septum more closely whereas the increased separation of the left and right atrial discs sometimes observed with the ‘softer’ Helex might delay this process in some individuals. Notwithstanding this hypothesis, as this study shows PFO closure is progressive and the difference in residual shunt rates between the devices reduces with time. Therefore with continued follow-up this difference will become smaller. Moreover, it could be argued that residual shunt rates are less important than recurrent neurological event rates, which were not different between the three groups, though we would point out that this study was underpowered for such a comparison.

Clearly, factors related to PFO morphology should have some bearing on PFO closure rate and it would seem entirely plausible that patients with a larger PFO, a septal aneurysm or possibly a tunnel would take longer to obtain full closure.4 5 11 Any of these defects may limit the ability of the occluder to oppose the interatrial septum or prevent good disc apposition and may predispose it to residual shunting. In this study PFO closure rate was not associated with shunt size, balloon waist size or the presence of a tunnel. The only factor that correlated with PFO closure rate was the presence of septal aneurysm. In cases of septal aneurysm we chose to cover the PFO rather than the entire aneurysm. Some centres prefer to use larger occluders to cover the entire aneurysm and splint the septum, whether this reduces residual shunting is unknown.

In this study we acknowledge that residual shunt rates are higher than reported in some other studies.21 35 36 However, some of these studies used TOE follow-up where the quality of Valsalva may be less and therefore may have underestimated residual shunts. We used stringent criteria with vigorous Valsalva as evidenced by a reduction in left ventricular cavity size on transthoracic echocardiography. More importantly, the majority of our patients studied had extremely large shunts before PFO closure, many with complete opacification of the left ventricle after Valsalva release. Therefore after PFO closure the presence of a moderate shunt is still indicative of improvement and in some cases where a large shunt remains but has been significantly reduced this might also represent success as the physical presence of the device alone should help to prevent emboli from crossing the defect.

Clinical implications

As more is understood about the morphology of PFO and the number of occluders increases then the complexity of PFO closure as an intervention also increases. One of the main challenges remains which occluder to use. We believe that no single type of occluder is sufficient, different occluders are required in order to meet the different sizes and morphologies of PFOs. The most important consideration is ‘which occluder best suits the PFO’. In this study we attempted to select the most appropriate occluder for the defect—that is, the choice was driven by anatomy. It might be argued that the type of occluder does not matter as the clinical significance of residual shunts in our study was negligible. However, in certain circumstances a faster ‘seal’ may be desirable—for example, in patients with multiple thromboembolic events, divers who wish to return to diving as quickly as possible, patients who can only tolerate short periods of antiplatelet treatment or patients contemplating pregnancy. In these circumstances it may be appropriate to select an occluder with a lower rate of residual shunts—that is, faster ‘closure rate’, if the anatomy is suitable. The presence of residual shunting also poses a significant clinical dilemma. In this situation many cardiologists prefer to re-evaluate patients and use an additional occluder if necessary. Some of these patients will, of course, have an additional, separate previously unrecognised defect whereas the majority will have a residual shunt in the region of the former PFO. Using a further occluder may pose a technical challenge both in its delivery and the choice of device; furthermore, the long-term consequences of multiple occluders within the atrial septum are unknown.


The association between migraine and PFO is well recognised.27 28 Several studies have shown a higher incidence of PFO in patients with migraine and aura.28 37 38 Similarly, a higher rate than expected of migraine has been reported in divers with decompression illness, and in one study migraine was associated with larger shunts.39 The link between PFO and migraine without aura is more contentious. Improvement in migraine following PFO closure has also been reported in observational studies.27 29 40 41 In this study we observed a high proportion of patients with migraine. In those patients with migraine with aura there was a perceived improvement in migraine frequency in around a third of patients after PFO closure. In patients without aura there was no significant change in headache after closure. In patients who had migraine with aura there was a non-significant tendency towards migraine improvement in those patients whose shunt was abolished; however, migraine did persist in some individuals. Clearly, the inability to predict those patients who may benefit from PFO closure remains a major impediment to performing PFO closure in this group. Further work is required to determine the effect of PFO closure on migraine.


There are a number of limitations to this study. First, this is not a randomised comparison but a ‘real-life’ clinical series, but this is a deficiency of many PFO studies to date. Second, although the choice of occluder was based on PFO morphology and the operators were experienced with all of the devices, the final choice of occluder was at the discretion of the operator. Finally, as in all PFO studies our data would benefit from a longer follow-up. However, most studies have indicated that the neurological recurrences tend to occur within the first 2 years after treatment, suggesting that our low annual recurrent event risk is unlikely to increase dramatically.14 22 We accept, however, and indeed point out that residual right-to-left shunt rates decrease with time and that a longer follow-up is likely to show a smaller difference between occluders. Within these limitations our data examine a large high-risk consecutive cohort and provide useful additional information in a rapidly increasing field of interest.


Percutaneous PFO closure in patients is a safe and efficacious strategy. Closure rate is reduced in patients with atrial septal aneurysm and PFO closure rates also vary depending on the occluder used. This may have important implications in certain individuals in whom more rapid PFO closure is required.



  • Competing interests MST is a proctor for St Jude Medical and AGA Medical, RPM is also a proctor for AGA Medical. There are no other conflicts of interest to declare.

  • Provenance and peer review Not commissioned; externally peer reviewed.