Journal of Neurology

, Volume 260, Issue 1, pp 260–267

Management of right-to-left shunt in cryptogenic cerebrovascular disease: results from the observational Austrian paradoxical cerebral embolism trial (TACET) registry


    • Department of NeurologyMedical University of Graz
  • Kurt Niederkorn
    • Department of NeurologyMedical University of Graz
  • Thomas Gattringer
    • Department of NeurologyMedical University of Graz
  • Martin Furtner
    • Department of NeurologyMedical University of Innsbruck
  • Raffi Topakian
    • Department of NeurologyAcademic Teaching Hospital Wagner-Jauregg
  • Wilfried Lang
    • Department of NeurologyHospital Barmherzige Brueder
  • Robert Maier
    • Division of Cardiology, Department of Internal MedicineMedical University of Graz
  • Andreas Gamillscheg
    • Division of Paediatric Cardiology, Department of PaediatricsMedical University of Graz
  • Franz Fazekas
    • Department of NeurologyMedical University of Graz
Original Communication

DOI: 10.1007/s00415-012-6629-9

Cite this article as:
Horner, S., Niederkorn, K., Gattringer, T. et al. J Neurol (2013) 260: 260. doi:10.1007/s00415-012-6629-9


Paradoxical embolism due to a patent foramen ovale (PFO) is a possible cause of ischemic stroke, particularly in young cryptogenic stroke patients. In most cases, however, it is difficult to establish a firm etiological association and the debate about management is ongoing. The Austrian Paradoxical Cerebral Embolism Trial was designed as a prospective, national, multi-center, non-randomized registry to add further data on this topic before the completion of randomized controlled trials. Over 27 months 188 cryptogenic stroke/TIA patients ≤55 years were entered by 15 Austrian stroke units. Contrast transesophageal echocardiography demonstrated a cardiac right-to-left shunt (RLS) in 176 patients; a pulmonary RLS was assumed in 10, and 2 showed both. Ninety-seven (55 %) patients with cardiac RLS underwent interventional treatment, and this was more likely for patients with stroke as index event, a symptomatic infarction on MRI and a large size of PFO. Over 2 years, recurrences occurred at a rate of approximately 1.3 % for stroke and 4.3 % for TIA, and were especially frequent in patients with pulmonary RLS. When comparing outcomes in patients with cardiac RLS there was a trend for fewer recurrences with interventional management (closure: four TIA in four patients vs. medical: three strokes and seven TIA in nine patients; p = 0.066 for events, p = 0.085 for patients). The complication rate was 13.4, and 5.7 % had residual shunting. The possible causes for paradoxical embolism in young patients with cryptogenic stroke appear more variable than usually considered, and other causes than PFO should not be neglected. Interventional treatment of a cardiac RLS may offer a small benefit, but has to be weighed against possible complications and the problem of establishing causality.


Paradoxical cerebral embolismCryptogenic strokePatent foramen ovaleRight-to-left shuntPulmonary arteriovenous malformationTranscatheter closure


Cardiac disorders are a major source of cerebral transient ischemic attacks (TIA) and stroke. Recent advances in technology and the availability of more sensitive echocardiographic methods such as contrast transesophageal echocardiography (cTEE) have greatly enhanced the frequency of detected abnormalities of the heart. Their clinical significance and ensuing therapeutic consequences are not always so clear, however. The detection of a patent foramen ovale (PFO) during routine evaluation of patients with cerebral ischemia belongs to this category. While some believe that a PFO is more often an incidental finding, especially when associated with traditional risk factors for stroke [1], others believe it to be a major culprit especially in patients with cryptogenic stroke aged <55 years [2]. Factors that may strengthen the association between PFO and cryptogenic stroke are large-size defects and coincidence with an atrial septal aneurysm (ASA) [1, 3]. Uncertainties also exist regarding the optimal treatment of paradoxical cerebral embolism, and therapeutic considerations have focussed primarily on the management of PFO. Although international guidelines [4, 5] recommend antiplatelet therapy as the first line strategy for treating stroke patients with PFO, transcatheter closure has become common practice in many centers and is one of the most frequent interventional procedures performed in adult congenital heart disease [6]. A large meta-analysis of observational evidence of secondary prevention in cryptogenic stroke/TIA patients with PFO is in favor of mechanical closure [7]. On the contrary recently published data from the first completed randomized clinical PFO management trial (closure study) suggest that PFO closure is not superior to medical management [8].

Other possible causes of right-to-left shunting (RLS) leading to cerebral complications like pulmonary arteriovenous malformations (pAVM) have also been noted, but are rarely discussed [9], and the impact of such findings and the risk of subsequent cerebrovascular events are still incompletely understood. Therefore, such scenarios have often been neglected in cryptogenic stroke patients so far.

The goals of this study were to (1) describe the causes of RLS and their frequency in a larger cohort, (2) assess the baseline factors of patients with a RLS and their impact on current treatment decisions, and (3) study the impact of medical therapy or interventional closure of a cardiac RLS (cRLS), mostly a PFO, on the recurrence rate over 2 years.

Subjects and methods

Study design

The Austrian Paradoxical Cerebral Embolism Trial (TACET) was designed as a prospective multicenter observational study and conducted in 15 neurologic departments throughout Austria. Participating centers, investigators and steering committee members are listed in the Appendix. The study protocol was approved by the local ethics committees, and patients gave written informed consent. The primary aim was to assess the baseline characteristics and follow-up results of a cohort of stroke/TIA patients with a cRLS or pRLS by prospective and standardized documentation and data collection irrespective of individually selected treatment strategies, and to compare medical with interventional therapy in regard to stroke/TIA recurrence and complications in patients with a cRLS/PFO. TIA was defined according to common guidelines, i.e., a symptom duration <24 h.

All patients admitted to participating centers with a TIA or stroke between May 2003 and July 2005 were invited to participate in this study if they were aged between 18 and 55 years and if the diagnostic workup had provided evidence of a cRLS or pRLS. In case of consent, patients were managed according to a predefined study protocol with standardized documentation and data collection. Treatment decisions were at the discretion of the individual physician and executed according to local practice. According to the protocol, participating neurologists were asked to exclude patients with other obvious stroke mechanisms. The study had been approved by the ethics committee of the Medical University of Graz and the other responsible local ethics committees, and informed consent had been obtained from all patients.

During the 27 months of recruitment, we entered a total of 201 patients (105 males, 52.2 %; 96 females, 47.8 %; mean age 42.1 ± 10.1 years) who were aged ≤55 years and in whom cTEE had demonstrated a cRLS or pRLS. We subsequently excluded 13 patients from the analysis because of the coexistence of another possible cause for their cerebrovascular event such as a symptomatic extracranial stenosis of ≥70 % (n = 4), a symptomatic intracranial stenosis (n = 4) or electrocardiography (ECG) documented atrial fibrillation (n = 3). Two patients underwent surgical PFO closure.

All patients at least underwent a brain CT or MRI at admission, cervical Doppler and duplex sonography, transcranial Doppler sonography (TCD), ECG and laboratory blood testing. Vascular risk factors were defined according to common recommendations [10].

For diagnosis of RLS, contrast TCD (cTCD) and cTEE were performed. The study protocol included screening for deep-vein thrombosis (DVT) by ultrasound scanning or phlebography of the leg veins. In patients with a cTEE-suspected pRLS, chest CT or pulmonary angiography was performed to search for a pulmonary fistula. Neurological deficits and disability at admission and during all follow-ups were documented using the National Institute of Health Stroke Scale (NIHSS) [11] and modified Rankin Scale (mRS) [12].

After hospital discharge, follow-up occurred at 6 months (questionnaire by telephone), at 1 year (in-hospital visit) and at 2 years (questionnaire by telephone). While in hospital and during all follow-up visits, specific attention was given to the documentation of the medical or interventional treatment strategies in regard to the RLS, associated complications and recurrent cerebral ischemic events.

Echocardiographic examinations

cTEE for diagnosing RLS was performed according to local standards, and the investigators were free in their choice of the contrast agent (agitated saline/air mixture, Echovist®, Oxypolygelatine) used for RLS detection. The diagnosis of a cRLS was made if contrast was observed to cross the interatrial septum from the right to the left atrium within three to four cardiac cycles after right-atrial opacification during Valsalva maneuver (VM). The anatomic size of the PFO was measured in multiple planes by assessing the maximum distance of the septum primum to the septum secundum. A pRLS was assumed if contrast material appeared in the left atrium later than three cardiac cycles after right heart opacification despite apparently intact atrial/ventricular septa and was seen to enter from the pulmonary veins [13]. In case of transcatheter closure of a PFO, a follow-up cTEE was recommended 6 months later.

cTCD monitoring examinations

The recommended TCD monitoring protocol was drawn from generally accepted and published guidelines [14, 15], and investigators were trained in the technique of RLS detection. The cTCD monitoring was performed either by using the unilateral hand-held method or by simultaneous bilateral monitoring of the intracranial arteries. The contrast agent to be used (agitated saline/air mixture, Echovist®, Oxypolygelatine) was not specified. RLS was defined by the presence of at least one contrast-induced microembolic signal on the TCD trace following contrast injection. Testing was done twice with and without a VM.

Statistical analyses

Statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS version 18.0) for Windows. Results were expressed as mean ± standard deviation (SD) and as absolute frequencies and percentages where appropriate. We used two-tailed t tests and chi-square tests to compare proportions. A p < 0.05 was considered to indicate statistical significance. Associations between categorical variables were measured using Pearson's chi-square statistic (Mantel-Haenszel test for ordinal data).


Patient characteristics

The final study cohort consisted of 188 cryptogenic stroke/TIA patients (mean age 42.37 ± 9.69, 51.6 % men). Of these, 108 (57.4 %) patients had suffered from a stroke and 78 (41.5 %) from a TIA; 2 had suffered from an attack of amaurosis fugax. A total of 130 (69.1 %) events were attributed to the carotid territory and 58 (30.9 %) to the posterior circulation. The baseline characteristics of these individuals are shown in Table 1.
Table 1

Patients with right-to-left shunting (RLS)


N (%)




42.37  ± 9.69 (19–55)



97 (51.6)


91 (48.4)



108 (57.4)


78 (41.5)

 Amaurosis fugax

2 (1.1)

 Previous stroke

12 (6.4)

 Previous TIA

22 (11.7)

 Carotid territory

130 (69.1)

 Vertebrobasilar territory

58 (30.9)

Type of RLS

 Cardiac RLS

176 (93.6)

 Pulmonary RLS

10 (5.3)

 Combined RLS

2 (1.1)

Using cTCD and cTEE, a cRLS was diagnosed in 176 (93.6 %) patients, a pRLS was assumed in 10 (5.3 %), and 2 (1.1 %) had a PFO in combination with presumed pulmonary shunting. Of the 176 patients with a cRLS, 168 (95.5 %) showed a PFO. Ten (5.7 %) patients had an atrial septal defect (ASD). Twenty-two (12.5 %) patients with PFO also had an atrial septal aneurysm (ASA). Paradoxical septal motion was seen in 14 (8 %) patients, and in 1 patient an intracardiac thrombus was detected. The maximum diameter of the PFO ranged from 1 to 25 mm (mean 9.3 ± 7.47 mm).

CT of the chest for suspected pAVM was performed in 14 patients, including all 12 patients with cTEE-diagnosed pRLS, and was negative for pAVM in all. However, invasive pulmonary angiography, which was performed in two patients because of recurrent cerebrovascular events, showed multiple arteriovenous fistulas in one of them (Fig. 1).
Fig. 1

Right pulmonary angiogram shows the fine reticular network of pulmonary arteriovenous fistulas in a 54-year-old patient with recurrent strokes and TIAs

DVT screening by ultrasound was done in 138 patients and was suggestive of thrombosis in three (2 %) of them. Phlebography was performed in 15 patients with positive results in two (13 %).

Therapeutic decision-making

Table 2 summarizes patients’ characteristics according to medical treatment (n = 79, 45 %) or percutaneous closure (n = 97, 55 %) in those patients with pure cRLS. As can be seen, stroke as the index event was significantly more frequent in the closure group and a TIA in the medical group. Vascular risk factors were equally distributed, including a history of stroke or TIA, which was present in 20 % of patients with interventional treatment and 15 % with medical treatment. There was however, a significantly greater proportion of patients with symptomatic infarction on MRI among patients undergoing interventional treatment. A larger diameter of the PFO was another factor favouring an endovascular procedure.
Table 2

Demographics and clinical characteristics of patients with cardiac right-to-left shunting


Percutaneous closure, N (%)

Medical treatment, N (%)

P value


97 (55.1)

79 (44.9)



41.52 ± 9.52

43.65 ± 9.88




45 (46.4)

47 (59.5)



52 (53.6)

32 (40.5)




65 (67)

39 (49.4)



30 (30.9)

40 (50.6)


 Amaurosis fugax

2 (2.1)



 Carotid territory

63 (64.9)

59 (74.7)



Median 1 (0–12)

Median 1 (0–15)


(Vascular) risk factors


21 (21.6)

23 (29.1)



31 (31.9)

24 (30.4)



30 (30.9)

35 (44.3)



2 (2.1)

4 (5.1)



23 (23.7)

11 (13.9)


 Valsalva maneuver

7 (7.2)

7 (8.9)


 Previous stroke

4 (4.1)

7 (8.9)


 Previous TIA

15 (15.5)

5 (6.3)




8 (8.2)

4 (5.1)


 Oral anticoagulation

1 (1)

2 (2.5)


 Pill/hormone therapy

21 (21.6)

13 (16.4)




96 (99)

75 (94.9)


 Symptomatic infarcts on MRI

74 (77.1)

37 (49.3)


 Silent infarcts on MRI

22 (22.9)

12 (15.2)




91 (93.8)

77 (97.5)


 PFO size

11.01 ± 7.46

7.2 ± 6.96



14 (14.4)

8 (10.1)



7 (7.2)

3 (3.8)


 Paradoxal septal movement

9 (9.3)

5 (6.3)


 Aortic plaques

4 (4.1)

4 (5.1)


Medication at discharge


76 (78.4)

64 (81)


 Oral anticoagulation

19 (19.6)

14 (17.7)


Medication at follow-up 1 (6 months)


84 (86.6)

53 (67.1)


 Oral anticoagulation

9 (9.3)

12 (15.2)


 Low molecular weight heparin

2 (2.1)

1 (1.3)


The implantation technique for transcatheter closure has been described previously [16]. Most patients were treated under local anesthesia with or without mild sedation; general anesthesia with endotracheal intubation was required in four (3.8 %) patients. The most commonly used devices were the Amplatzer® PFO Occluder and Amplatzer® Septal Occluder (69 %), less frequently the CardiaStar™ (18 %) and CardioSEAL® Occluder (13 %). Some 13 catheter-related complications were noted, i.e., the complication rate was 13.4 %. Complications (Table 3) were considered as mild in 12 instances and included a hematoma at the puncture site (n = 5), transient dysrhythmia (n = 5) and small, asymptomatic device-adherent thrombus (n = 2). One adverse event was categorized as moderate to severe and consisted of device embolization, which had been noticed during regular echocardiographic control. It required surgical removal of the device with defect closure.
Table 3

Catheter-related complications of 97 patients who underwent transcutaneous PFO closure


N (%)


13 (13.4)


12 (12.4)

 Local hematoma at puncture site

5 (5.2)

 Transient dysrhythmia

5 (5.2)

 Small asymptomatic device-adherent thrombus

2 (2.1)

Moderate to severe

 Device embolization

1 (1)

A 6-month post-interventional cTEE control was done in 84 (80.7 %) patients and showed residual shunting in 6 (5.7 %) patients, in 4 treated with the CardiaStar™ Occluder and in 2 who had received an Amplatzer® PFO Occluder.

Recurrent cerebrovascular attacks and treatment

During the 2 years of follow-up, 21 recurrent cerebrovascular events (CVEs) were reported in 16 (8.5 %) patients (5 strokes in 4 patients and 16 TIAs in 13 patients). This reflects an overall annual event rate for stroke of approximately 1.3 % and for TIA of about 4.3 %. The follow-up rates were 71 % for in-hospital visit 1 year after the index event and 98 % for standardized final telephone follow-up at 2 years.

In the group of 176 patients with a cRLS, a total number of 14 cerebrovascular events (3 strokes, 11 TIAs) occurred in 13 patients (7.4 %). In the medically treated group, ten events occurred in nine patients. Six of those patients subsequently underwent interventional closure without further attacks. Among the patients who had undergone closure of the PFO or of the ASD directly after the index attack, a further TIA was noted in four patients for each. When comparing the number of events of all patients while on medical therapy against those treated with endovascular closure of the PFO, there was a trend for a better outcome in the interventional group (closure: four TIAs in four patients vs. medical: three strokes and seven TIAs in nine patients; p = 0.066 for events, p = 0.085 for patients).

In the remaining 12 patients with presumed pulmonary shunting (combined with a PFO in two patients), there were seven events (two strokes and five TIAs) in three patients. The patient with angiographically confirmed pulmonary fistula had two stroke and two TIA recurrences.

The time course of the recurrent cerebrovascular events is shown in Fig. 2. As can be seen, most recurrent CVEs in patients with cRLS occurred soon after the index event. Most recurrent strokes/TIAs could not be attributed to other etiologies than paradoxical embolism. Potential other explanations were present in 4 of 16 patients (25 %). These included: essential thrombocythemia, antiphospholipid antibody syndrome, aortic arch atheromatosis and significant bradyarrhythmia following cardiac pacemaker implantation.
Fig. 2

Distribution and timing of recurrent cerebrovascular events in patients with cardiac right-to-left shunt and medical versus interventional treatment and in those with pulmonary shunting


Interventional closure of a PFO, especially in young patients with a cryptogenic stroke, is a frequently performed strategy for secondary prevention of cerebrovascular attacks despite missing evidence from appropriate trials. The impact of other RLS causes was also mostly neglected in clinical studies so far. We therefore attempted to add insights into the actual management of these patients and their outcomes from clinical practice in several stroke units throughout Austria.

As a first observation we found the possibility of paradoxical embolism to be quite variable. RLS was associated with an isolated cardiac shunt in the majority (93.6 %) of the TACET patients, and this was associated with a PFO in 95.5 % of patients. A pulmonary RLS, however, was also found in ten (5.3 %) patients, and two (1.1 %) patients had a PFO in combination with presumed pulmonary shunting. Importantly, 3 of these 12 patients had recurrent CVEs. Pulmonary angiography was therefore performed in two patients following a negative chest CT and confirmed pulmonary arteriovenous malformations in one of them. Although not often discussed [4], this observation supports pulmonary shunting as a possible cause of stroke. The incidence of ischemic events reported in one series of consecutive patients with pAVM was as high as 18 % for stroke and 37 % for TIA [17]. In a review on pAVMs the incidence of stroke was reported to vary from 2.6 to 25 %, with an average of 8.5 %. However, that may underestimate the actual risk as investigations for stroke were not done systematically in all reviewed studies [18]. This suggests recommending more detailed workup for a pAVM when cTEE localizes a shunt at the pulmonary level, especially following repeated CVEs. Unfortunately, chest CT appears to be of limited sensitivity for a pAVM so that we and others may have underestimated their actual prevalence.

Despite the restrictive recommendations of professional organizations [5, 6], more than half of the physicians/patients opted for an interventional closure of the PFO when identified as the cause of an RLS. Several factors were obviously favoring the decision for intervention over medical therapy. Stroke as the index event and a symptomatic infarction on MRI were significantly more often noted with PFO closure. Another factor favoring intervention was the increasing diameter of the PFO. It is of interest that a history of preceding TIA or stroke was almost equally distributed in both groups. Also there was a very low rate of actual evidence for peripheral venous thrombosis as the possible cause of paradoxical embolism with a somewhat higher rate of detection by phlebography than Doppler sonography. This attests to the well-known problem of ascertaining a causal relationship between RLS and the CVE [19]. Another important finding of this observational study is that medical management of PFO patients was quite diverse. The choice of treatment (e.g., anticoagulants vs. antiplatelets) was not pre-specified and was taken by the individual neurologists together with cardiologists according to presumed “best practice” at this time. Moreover, a relatively high proportion of patients (approximately 15 %) in the medical group did not take any antithrombotics already at the 6-month follow-up. This also points to limited compliance, which may be especially problematic in young stroke patients and together with previously mentioned different treatment strategies might have influenced recurrence rates.

Although proof or disproof of an efficacy of PFO closure would certainly have been the most interesting aspect of this study, it was beyond its scope as this was not a randomized trial. Recently published data from the closure trial [8] indicated a non-significantly different rate of recurrent stroke or TIA of 5.5 (mechanical closure) versus 6.8 % (medical management).

The rate of recurrent cerebrovascular attacks in our series was 5.6 % for all patients with an RLS and 4 % for those with a PFO. This is somewhat higher than suggested by earlier observational studies that reported an average annual risk of recurrent ischemia in patients with PFO between 2.4 and 3.8 % while on antithrombotic treatment [2022]. It needs to be noted, however, that the majority of recurrent events consisted of TIAs and that the close follow-up of our patients may have served to readily pick up every short-lived neurologic phenomenon. When comparing treatment strategies we noted a trend in favor of interventional treatment. Nine patients experienced a further CVE while on medical treatment compared to only four following closure of the cardiac RLS. These results are very similar to and confirm a recent report from an Italian registry [23]. Differences in baseline demographic and clinical factors are unlikely to have accounted for this difference and should rather have been in favor of the medical treatment group.

It is also interesting to note that most of recurrent events in TACET occurred relatively soon after the index event. This could imply a period of increased risk for further embolization around the acute attack where, if considered, intervention might be most beneficial. Such a setting is already well known for patients with high-grade carotid stenosis [24] and might also deserve attention in trials assessing the benefit of PFO closure. Another potential explanation for the recurrent CVE could be found in 25 % of patients. This is in line with a recent report [19] and indicates that a thorough workup is required to rule out other etiologies than paradoxical cerebral embolism. In this context, it needs to be noted that the presence of vascular risk factors like hypertension might have provided an alternative explanation for the ischemic event already at baseline for some, but this consideration was also at the discretion of the individual physician.

Regarding risks of the intervention in TACET, PFO closure was associated with a rate of 12.5 % of mild complications, and one patient experienced a dislocation of the device requiring surgical intervention. Atrial fibrillation has been associated with PFO closure and mainly occurred periprocedurally in the closure study [8]. In our series cardiologists described dysrhythmias in five patients that were all transient and did not lead to any embolic event. In addition, the possibility of residual shunting following intervention, which was around 6 % at 6 months in our series, should also not be neglected when counseling patients regarding management of their RLS.


The study was financially supported by the Austrian Society of Stroke Research.

Conflicts of interest

The authors declare that they have no conflict of interest.

Ethical standard

This study has been approved by the appropriate ethics committee and have therefore been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki.

Copyright information

© Springer-Verlag 2012