MethodsStudy population

Consecutive patients between 2006 and 2021, who underwent percutaneous PFO closure for cryptogenic stroke or transient ischaemic attack and received echocardiographic follow-up including transthoracic echocardiography with contrast bubble study, were retrospectively enrolled at a single centre (Amsterdam University Medical Centers, AMC, Amsterdam). Baseline patient characteristics were recorded and baseline echocardiographic parameters were collected, including shunt size and presence of atrial septal aneurysm (>10 mm deviation of the septum). Baseline right-to-left shunt grade was assessed according to the number of bubbles passing through the atrial septum within three heartbeats following complete opacification of the right atrium, measured during transthoracic echocardiography with contrast bubble study and Valsalva manoeuvre, categorised in four groups: no shunt, small shunt (1–9 bubbles), moderate shunt (10–20 bubbles) and large shunt (>20 bubbles). All enrolled patients underwent comprehensive evaluation by the multidisciplinary adult congenital heart disease team, where percutaneous closure with a double-disc device was advised. This decision followed extensive neurological and cardiovascular evaluation to rule out other stroke aetiologies. Closure was advised if no alternative causes were identified apart from the presence of a PFO, as ascertained by echocardiography with contrast bubble study, and if the risk of paradoxical embolism (RoPE) score was ≥6.

Closure procedure

Patients underwent PFO closure under general anaesthesia with transoesophageal echocardiography guidance. Using three-dimensional transoesophageal echocardiography, PFO measurements were performed to select the appropriate device size at the discretion of the treating physicians, based on PFO characteristics. Measurements of PFO width and PFO height were conducted subsequent to the passage of a guidewire through the PFO, as described before (online supplemental figure 1).15 The choice of the implanted device type was determined by the treating physicians. The delivery system was introduced through venous femoral access and haemostasis was obtained with a ‘figure-of-eight’ stitch. All patients received treatment with either dual antiplatelet therapy or single antiplatelet therapy combined with direct oral anticoagulant for a minimum duration of 6 months. After evaluation if residual right-to-left shunt was present at 6 months, decisions regarding the continuation or cessation of dual antiplatelet therapy were made accordingly. At the 12-month follow-up, the majority of patients in our study were referred back to their local hospital or general practitioner for further follow-up. We recommended extended treatment with single antiplatelet therapy, primarily clopidogrel, and advocated for follow-up with transthoracic echocardiography and contrast bubble study to further evaluate patients with residual right-to-left shunt after the 12-month period.

Outcomes and follow-up evaluation

The primary outcomes encompassed the incidence of residual right-to-left shunt at 6 and 12 months after percutaneous PFO closure. Follow-up assessments using transthoracic echocardiography with contrast bubble study (agitated saline) and Valsalva manoeuvre were performed at 6 months and, in case a residual shunt was present, repeat transthoracic echocardiography with contrast bubble study was conducted 12 months after closure. If patients had no residual right-to-left shunt at 6-month follow-up, this was registered as no residual right-to-left shunt at 12 months. Successful closure was defined as no or small residual right-to-left shunt (<10 bubbles), whereas unsuccessful closure (relevant residual shunt) was defined as moderate or large residual right-to-left shunt (≥10 bubbles). Residual right-to-left shunt grades were consolidated into two groups for univariable and multivariable analyses: (1) successful closure and (2) unsuccessful closure. All univariables were included in the full model for multivariable regression analyses. The final model was derived using a backward selection approach, with stepwise exclusion of non-significant variables, a significance level of p<0.05 was used. Interaction terms were considered for PFO height, width and length, which were not statistically significant.

Statistical analysis

Data were summarised as number of patients (%) for categorical variables; mean (±SD) for normally distributed continuous variables and median (IQRs) for non-normally distributed continuous variables. Normality of all variables was assessed via histograms. Categorical variables were compared using the χ2 test. Continuous variables were compared using the t-test or Mann-Whitney U test according to the distribution of the variable. Univariable and multivariable logistic regression analyses were performed. Statistical analyses were performed with IBM SPSS Statistics for Windows V.28 (IBM). A p value <0.05 was considered statistically significant.

Patient and public involvement

Patients were not involved in the coproduction of this research.

ResultsPatient characteristics

A total of 227 consecutive patients were included in this study (figure 1). The mean age was 43±11 years, 50.2% were women and the median RoPE score was 7.0 (IQR 6–8). Atrial septal aneurysm was present in 34.8%. Transthoracic echocardiography with contrast bubble study and Valsalva manoeuvre at baseline showed that 6.2% (14 patients) had small right-to-left shunt, 19.8% (45 patients) had moderate right-to-left shunt and 74.0% (168 patients) had large right-to-left shunt. The mean PFO width measured with three-dimensional transoesophageal echocardiography during the closure procedure was 11.0±3.9 mm, the mean PFO height was 6.3±2.4 mm and the mean tunnel length was 7.0±2.4 mm. Most PFOs were closed using the Amplatzer PFO occluder (55.9%, 127 patients), followed by the Amplatzer Cribriform MF septal occluder (35.7%, 81 patients), the Premere PFO closure device (7.0%, 16 patients) and the Gore Cardioform septal occluder (1.3%, 3 patients). All baseline, echocardiographic and procedural characteristics are presented in tables 1 and 2.

Figure 1

Flow chart of all patients included. 28 patients had a different indication for percutaneous closure than cryptogenic stroke or transient ischaemic attack (TIA). 42 patients did not undergo transthoracic echocardiography (TTE) with contrast bubble study (CBS) for shunt assessment at baseline or during follow-up (FU). PFO, patent foramen ovale.

Table 1

Baseline characteristics of all 227 consecutive patients with PFO and cryptogenic stroke or transient ischaemic attack

Table 2

Procedural characteristics of all 227 patients

Procedural complications

Minor complications occurred in 4.4% (10 patients) and there were no periprocedural complications in 95.6% (217 patients). All complications are specified in table 2.

Residual right-to-left shunt at follow-up

Follow-up transthoracic echocardiography with contrast bubble study and Valsalva manoeuvre at 6 months after the PFO closure procedure showed that 72.7% (165 patients) had no residual right-to-left shunt, 12.3% (28 patients) had small residual right-to-left shunt, 6.6% (15 patients) had moderate residual right-to-left shunt and 8.4% (19 patients) had large residual right-to-left shunt. At 12-month follow-up, 87.7% (199 patients) had no residual right-to-left shunt, 6.6% (15 patients) had small residual right-to-left shunt, 2.6% (6 patients) had moderate residual right-to-left shunt and 3.1% (7 patients) had large residual right-to-left shunt. Thus, successful closure at 6 months after the closure procedure, defined as no or small residual right-to-left shunt, was achieved in 85.0% (95% CI 79.8% to 89.1%), which increased to 94.3% (95% CI 90.5% to 96.6%) at 12 months after the closure procedure. The temporal course of right-to-left shunt grade is illustrated in figure 2.

Figure 2

Successful closure was defined as no or small residual RLS at FU and unsuccessful closure (relevant residual shunt) as moderate or large residual RLS at FU. (A) Transition of RLS grade between baseline and 6-month FU. At baseline, 6.2% (95% CI 3.7% to 10.1%) (14/227 patients) had small RLS. After PFO closure, 85.0% (95% CI 79.8% to 89.1%) (193/227 patients) had successful closure at 6-month FU. (B) Transition of RLS grade between 6 and 12 months of FU in patients with moderate or large residual RLS at 6-month FU. At 12-month FU, successful closure increased to 94.3% (95% CI 90.5% to 96.6%) (214/227 patients). No RLS (0 bubble); small RLS (1–9 bubbles); moderate RLS (10–20 bubbles); large RLS (>20 bubbles). FU, follow-up; PFO, patent foramen ovale; RLS, right-to-left shunt.

The percentage of patients that improved ≥1 right-to-left shunt grade at 6 months after closure compared with baseline was 89.9% (95% CI 85.3% to 93.2%) (204/227 patients). At 12 months after the closure procedure, the percentage of patients that improved ≥1 right-to-left shunt grade compared with baseline was 96.5% (95% CI 93.2% to 98.2%) (219/227 patients). The percentage of patients that improved ≥1 right-to-left shunt grade from 6 to 12 months after the closure procedure was 77.4% (95% CI 65.6% to 86.0%) (48/62 patients). The change in number of patients for each right-to-left shunt grade between baseline, 6-month follow-up and 12-month follow-up is presented in online supplemental figure 2. The likelihood of successful closure at 12-month follow-up for patients with unsuccessful closure at 6-month follow-up was 73.3% (95% CI 48.1% to 89.1%) (11/15 patients) for moderate right-to-left shunt and 52.6% (95% CI 31.7% to 72.7%) (10/19 patients) for large right-to-left shunt. The characteristics and clinical course of the patients with moderate or large residual shunt at 12-month follow-up are presented in table 3.

Table 3

Characteristics of patients with significant residual shunt at 12 months

At 6-month follow-up, 7.6% (6/79) of the patients with atrial septal aneurysm had large residual shunt compared with 8.8% (13/148) of the patients without atrial septal aneurysm (p=0.758). At 12-month follow-up, 3.8% (3/79) of the patients with atrial septal aneurysm had large right-to-left shunt compared with 2.7% (4/148) of the patients without atrial septal aneurysm (p=0.659). All residual right-to-left shunt grades for patients with and without atrial septal aneurysm during follow-up are presented in online supplemental table 1 and 2.

Table 4 lists univariable logistic regression analyses comparing patient characteristics with unsuccessful closure at 6 and 12 months of follow-up. For each millimetre increase in PFO size, both in width and height, the ORs for unsuccessful closure were significantly increased at 6-month follow-up (OR=1.13, 95% CI 1.03 to 1.24, p=0.012 and OR=1.25, 95% CI 1.06 to 1.48, p=0.008, respectively) and at 12-month follow-up (OR=1.24, 95% CI 1.08 to 1.42, p=0.002 and OR=1.38, 95% CI 1.08 to 1.76, p=0.010, respectively). In addition, using the Amplatzer Cribriform MF septal occluder was significantly associated with unsuccessful closure at 6-month follow-up (OR=2.86, 95% CI 1.37 to 5.97, p=0.005) and at 12 months not at the conventional significance level; however, the upper limit of the CI and OR itself indicate a potentially meaningful association that may be clinically relevant (OR=3.09, 95% CI 0.98 to 9.79, p=0.055). For each incremental increase in PFO device size the OR to have unsuccessful closure was significantly higher at 6-month follow-up (OR=1.16, 95% CI 1.04 to 1.29, p=0.007) and at 12 months (OR=1.35, 95% CI 1.14 to 1.59, p<0.001).

Table 4

Univariable logistic regression analyses between baseline or procedural characteristics and unsuccessful closure at 6 and 12 months of follow-up

Multivariable logistic regression analyses were performed for the presence of unsuccessful closure at 6 and 12 months of follow-up (table 5). Multivariable analyses at 6 months showed that the OR for each incremental step in PFO width to have unsuccessful closure was significantly higher, OR=1.13, 95% CI 1.03 to 1.24, p=0.012 (table 5). At 12-month follow-up, the OR for Amplatzer Cribriform MF septal occluder and device size to have unsuccessful closure was significantly higher, OR=4.24, 95% CI 1.13 to 15.88, p=0.032 and OR=1.43, 95% CI 1.18 to 1.72, p<0.001, respectively (table 5).

Table 5

Multivariable logistic regression analysis between baseline or procedural characteristics and unsuccessful closure at 6 and 12 months of follow-up

The differences between residual right-to-left shunt grade per device size during follow-up are presented in online supplemental tables 3 and 4.

Discussion

As we have learnt from large randomised controlled trials, the efficacy of percutaneous PFO closure using a double-disc device in preventing recurrent PFO-associated strokes has been established.1–4 Patients undergoing percutaneous PFO closure had a lower risk of stroke recurrence, ranging from 0% to 2% compared with 4.8–10.5% in patients treated with dual antiplatelet therapy alone.1–4

Previous studies identified residual shunting after percutaneous PFO closure in approximately a quarter of the patients, with its presence notably linked to an increased risk of stroke recurrence, specifically in cases of large residual shunt.1–12 Windecker et al reported a relative risk of 4.2 (95% CI 1.1 to 17.8, p=0.03), with an average annual rate of recurrent transient ischaemic attack, cerebrovascular accident or peripheral embolism of 6.8% in patients with residual right-to-left shunt compared with 2.1% in those without.5 Additionally, Deng et al observed an HR of 3.05 (95% CI 1.65 to 5.62, p<0.001) and a cumulative 5-year probability of recurrent stroke or transient ischaemic attack of 9.3% in patients with residual right-to-left shunt versus 2.5% in those without.11 Conversely, von Bardeleben et al found no association between residual right-to-left shunt and recurrent stroke after cumulative follow-up of 1265 patient-years (mean follow-up duration: 3.8 years) with a re-event rate of 0.7% annually.7

In our study, any residual right-to-left shunt decreased from 27.3% (62/227 patients) at 6 months after percutaneous PFO closure to 12.3% (28/227 patients) at 12 months. Notably, the patients with moderate or large residual right-to-left shunt (unsuccessful closure) decreased from 15.0% (34/227 patients) to 5.7% (13/227 patients). Our study demonstrated a significant reduction in residual right-to-left shunt over time after transoesophageal echocardiography-guided PFO closure in a real-world cohort. This observed trend aligns with findings from Deng et al, who similarly reported a decrease in any grade of residual shunt between 6 and 12 months after closure from 29.5% to 19.8%, and from 8.3% to 7.2% in patients with moderate or large shunt.11 Our study, showing a reduction in residual right-to-left shunt, particularly in patients with moderate or large shunt (clinically significant), adds valuable insights to the existing knowledge regarding clinically relevant residual right-to-left shunt after percutaneous PFO closure and its management strategies.

The occurrence of residual right-to-left shunt in the initial months following device closure is common, primarily attributed to the time required for complete endothelialisation of the device.16 Therefore, antithrombotic therapy is recommended for at least 1–6 months.16 The gradual process of endothelialisation to cover of the device and thereby, sealing off the PFO defect, partly explains the reduction of residual right-to-left shunt over time. Persistent residual right-to-left shunt can arise from multiple mechanisms, but is mostly due to a residual defect through or adjacent to the PFO device or an additional unidentified pathway facilitating right-to-left shunt.10 These mechanisms include cardiac aetiologies, such as fenestrations, second septal defects or pulmonary arteriovenous malformations, but also pathways from non-cardiac origins, including connections of the brachiocephalic trunk, innominate vein, superior caval vein, inferior caval vein or hepatic vein.10 Stroke recurrence may be linked to another cause which was not initially identified, including the ones mentioned before, and not necessarily due to shunting through the PFO. A comprehensive review of studies addressing the risks of residual shunts after closure and the risk of recurrent stroke is displayed in online supplemental table 5.8 11 17–19

As we look into the different types of closure devices, we noticed divergent rates of residual right-to-left shunt associated with different device types, as previously described by Hammerstingl et al, who reported a closure rate at 12 months of 92.3% for Amplatzer PFO occluder, 87.9% for CardioSeal occluder, 87.5% for Premere occluder and 62.3% for Helex occluder.8 These discrepancies in residual right-to-left shunt between different device types, paralleled by our findings with the Amplatzer Cribriform MF septal occluder, suggest that anatomical patient characteristics (influencing device-type selection) or structural device features may impact the grade of residual right-to-left shunt. This variability might stem from differences in device design, such as stiffness, radial force and equal or different left and right atrial disc sizes. Additionally, variations in device sizes could contribute to the differences in residual right-to-left shunt, as revealed in our multivariable analyses, where device size emerged as a significant predictor for residual right-to-left shunt.

Furthermore, atrial septal aneurysm is recognised as an important anatomical risk factor for PFO-attributed stroke and also for the presence of residual right-to-left shunt after closure, with a successful closure rate of 90.1% at 12 months in patients with both PFO and atrial septal aneurysm compared with 97.4% in patients with PFO alone.7 Although we did not find an association between the presence of atrial septal aneurysm and residual right-to-left shunt, successful closure was numerically lower at 12 months in patients with atrial septal aneurysm (92.4%, 73/79 patients) compared with patients without (95.3%, 141/148 patients).

Moreover, it remains crucial to determine the underlying mechanism of the shunt to accurately assess the severity of residual right-to-left shunt, comprehend its clinical implications and assess the risk of stroke recurrence. This diagnostic process can be facilitated through transoesophageal echocardiography for cardiac origins or additional contrast-enhanced CT. In case of residual right-to-left shunt due to contrast passage adjacent to the PFO device or an additional defect, a second closure device, plug, coil or percutaneous stitch can be implanted to effectively seal off the connection between the right and left atria, as has been reported in several small cohort studies and case reports.13 14 20–23 Hence, in future studies, diagnosing the mechanism and severity of residual right-to-left shunt should be a priority to tailor the appropriate treatment strategy for each individual patient. The optimal timing for evaluation extends to at least 12 months after procedure, considering that most cases of residual right-to-left shunt during the first months tend to diminish and, in some cases, completely resolve over time. Additional treatment options include a second percutaneous closure procedure, surgical closure, prolonged antithrombotic regimen or a decision to accept the residual right-to-left shunt.

Limitations

This study has several limitations. Primarily, this was a retrospective analysis conducted at a single-centre study. Second, the absence of transcranial Doppler assessment implies the possibility of overlooked residual shunts. Third, our study did not investigate residual stroke occurrences due to being underpowered for this analysis, considering a recurrence rate ranging from 1% to 5%. Finally, the majority of PFOs in our study used Amplatzer devices, hence, the incidence of residual right-to-left shunt might differ for other closure devices.