Paradoxical embolism due to patent foramen ovale (PFO)-mediated right to left shunting is considered a contributing factor to cryptogenic stroke (CS) (1–4). In patients with high-grade PFO shunts, the recurrence of ischemic stroke or transient ischemic attack (TIA) is significantly reduced in those who undergo PFO closure compared to those receiving medical-only treatment (5–8). Consequently, current guidelines recommend PFO closure for young patients with CS and high-grade PFO shunts (9).

For PFO closure, a soft-tipped 0.035-inch guidewire or a multipurpose (MP) catheter is introduced through the PFO tunnel from the right atrium (RA) to the left atrium (LA) (10–13). However, in some cases, the guidewire or catheter fails to traverse the PFO tunnel, even in patients with high-grade shunts (Figure 1). In these patients, transseptal puncture around the PFO tunnel can be performed to facilitate wire or catheter passage (14–17). Nonetheless, our previous data indicate a higher incidence of recurrent stroke or TIA and residual shunt with this technique compared to the standard PFO closure technique (18). Consequently, transseptal puncture is an invasive procedure with a potential for poor outcomes, and its use in the treatment of uncrossable PFOs may be carefully considered.

Currently, there is a scarcity of data concerning the characteristics and long-term clinical outcomes of patients who experience unsuccessful PFO closure due to the inability to pass through the tunnel. In this study, we aimed to determine the long-term outcomes of patients with failed wire or catheter passage through the PFO tunnel, even when PFO closure was indicated.

This study addresses the fundamental question of whether it is safe to leave an uncrossable PFO untreated in patients with high-grade PFO shunts. To our knowledge, this is the first study to compare the prognosis of uncrossable PFO to the PFO closure. We found that: (1) the long-term clinical outcomes of the uncrossable PFO group were similar to those of the PFO closure group, (2) these outcomes were consistent regardless of the baseline PFO grade, and (3) recurrent stroke/TIA before the procedure is a strong prognostic factor for the stroke/TIA after procedure.

The proportion of uncrossable PFO population may have been overlooked. The incidence of uncrossable patent foramen ovale (PFO) varies considerably, ranging from 10%–40% in various studies (15, 21–23). According to existing literature on echocardiographic analysis in patients undergoing PFO closure, 54.5% exhibited complex PFOs, encompassing characteristics such as long-tunnel PFOs, a substantial Eustachian ridge, or an overly redundant Chiari network, which could potentially result in uncrossable PFO (24). Given that the transseptal puncture technique is more commonly used in these complex PFOs, it is reasonable to anticipate an occurrence of uncrossable PFOs within this category. Some studies that reported higher procedure success rates utilized the transseptal puncture technique, a method we intentionally did not employ in our study. Many papers neither discuss how they addressed uncrossable PFO nor report its incidence. Most PFO-related studies only compare PFO closure groups with medication-only groups, potentially neglecting uncrossable PFO. Thus, the true incidence of uncrossable PFOs in real-world clinical practice may be higher than our estimates, particularly if transseptal puncture techniques are not utilized and the incidence of complex PFOs is taken into account. The identified incidence of uncrossable PFO in our study is 33%, and although its exact alignment with actual incidence remains unclear, this figure falls within the range reported in existing literature. It should be emphasized that candidates for PFO closure were carefully selected after consultation with neurologists and cardiac imaging specialists, and all procedures were performed by highly experienced operators. Given our meticulous selection of PFO closure candidates and the skilled execution of procedures, we believe this to be acceptable. Additionally, demographic differences, such as the proportion of grade 3 PFOs, between our study and those reporting higher procedure success rates must also be considered.

Our study suggests that among patients undergoing a PFO closure procedure with high-grade PFO shunts, the prognostic implications of grade 2 vs. grade 3 shunts are not significant. Neither the difference between PFO grades 2 and 3 nor the presence or absence of a resting shunt emerged as independent predictors in a multivariate Cox regression analysis for recurrent stroke or TIA after the procedure. Furthermore, a subgroup analysis comparing the uncrossable PFO group to the PFO closure group, while considering PFO shunt grades 2 and 3, showed no difference in the primary endpoint. High-grade PFO shunts are reportedly associated with a higher risk of stroke recurrence, and PFO closure reduces this risk (8, 25–30). The leading randomized controlled trials, the Gore REDUCE and the RESPECT (Randomized Evaluation of Recurrent Stroke Comparing PFO Closure to Established Current Standard of Care Treatment) trials, did not demonstrate a significant difference in prognosis between the PFO closure group and the medication-only group when including PFO shunt grade 1 (5, 7). No existing studies have focused solely on comparing PFO shunt grades 2 and 3 while excluding grade 1. For this reason, our study chose to exclude patients with PFO shunt grade 1 as candidates for the procedure and included only high-grade PFO shunts.

In our study, several PFO tunnels were not crossable with a wire or catheter, even at baseline PFO grade 3 (28/82, 34.1%). These findings contradict the assumption that a larger shunt, defined as a large number of agitated saline microbubbles observed within three cycles, may indicate a larger PFO defect. The agitated saline test indirectly estimates the PFO size based on the number of microbubbles. It does not directly measure the PFO size (16). Several studies have reported that the PFO shunt grade reportedly does not correlate with the actual size PFO size (29, 31, 32). Thus, the estimated PFO shunt size may not be accurate, which may explain the uncrossable tunnels encountered in patients with grade 3 PFO. Anatomical complexities of the PFO, such as the long tunnel or the angle between the guiding catheter and PFO tunnel, have been previously elucidated (8, 14, 17, 22, 23, 33). We believe that regardless of the initial shunt severity, stroke incidence may be comparable to the crossable group wherein patients are inserted with the closure device. This might be because anatomical barriers, such as long-tunnel PFOs, a prominent Eustachian ridge, or a redundant Chiari network (24), may prevent emboli, thrombi, or linear structures like guidewires from passing through the PFO while still allowing smaller microbubbles to traverse. The relationship between these anatomical complexities and passage of wire or catheter requires further investigation.

We found that a previous history of recurrent stroke/TIA was the only indicator for future episodes of recurrent stroke/TIA. Patients with a previous history of recurrent stroke/TIA continued to be at a high risk for recurrent stroke/TIA even after successful PFO closure (HR_adj 11.08, 95% CI 2.05–59.98, P = 0.005). This indicates that the cause of stroke/TIA in patients with PFO who underwent closure has not been thoroughly evaluated. Although we tried to rule out other causes of stroke before the procedure, there was a possibility of missing causal factors such as atrial fibrillation (AF). In our study, three new-onset AF cases were noted after successful device implantation; one patient had persistent AF and two had paroxysmal AF. It is difficult to accurately determine whether AF in these patients was a device-induced complication or had gone undetected before the procedure. Regardless of the cause, AF is a strong risk factor for stroke or TIA.

There are some important limitations to this study. One of the most significant is the relatively high uncrossable rate of 33%. While this rate is higher compared to established trials such as Gore-REDUCE, it is crucial to take into account the patient exclusion criteria in those trials (7). Unlike these trials, our study did not exclude patients with anatomically complex PFO that have a higher likelihood of procedural failure, including those requiring trans-septal puncture. Therefore, the uncrossable rate in this study may reflect a broader, more diverse patient population with varying anatomical complexities of PFOs. However, the high rate of uncrossable PFO remains a limitation of our study, and the findings should be corroborated by further studies focusing on this specific issue. As a single-center observational study with a relatively small number of patients analyzed, the results may not be generalizable to all populations. Additionally, approximately 10% of patients in the study were not evaluated using cardiac computed tomography to rule out pulmonary arteriovenous fistulas or anomalous pulmonary venous return, which may have affected the results. An additional limitation of the study is the lack of detailed investigation of the prevalence of liver disease in the patients, which is important because a right to left shunt identified on TEE in patients with undiagnosed hepatopulmonary syndrome may be mistaken for a shunt due to PFO. However, because we judged a right to left shunt with PFO only if the agitated saline contrast was detected in the left atrium within 3 cardiac cycles, we expect that the proportion of patients with hepatopulmonary syndrome in whom microbubbles are typically observed after 4 cardiac cycles is not as high as we fear. Additionally, we note a significant disparity in the duration of post-procedure follow-up between the two groups. While we strived for extended follow-up periods due to the limited number of clinical events, we acknowledge that this difference in follow-up duration poses a limitation in interpreting the study results. Future studies with larger sample sizes and more rigorous evaluation methods are needed to further investigate the long-term clinical outcomes of patients with uncrossable PFOs and to better understand the factors that contribute to recurrent stroke/TIA events in this population. Also, despite our efforts to thoroughly rule out other causes of CS, we acknowledge the possibility of excessive measures of PFO shunt grade or other biases, since this was a single center study.

Our study suggests that the clinical outcomes of patients with uncrossable PFOs are comparable to those of patients who undergo successful PFO closure, regardless of the baseline PFO grade. This suggests that medical therapy alone might be sufficient for preventing recurrent stroke or TIA in patients with uncrossable PFOs, even when a high-grade PFO is present. These findings may have implications for the management of patients with uncrossable PFOs, potentially reducing the need for invasive closure procedures in this population. Further research is needed to validate these findings and to optimize treatment strategies for patients with uncrossable PFOs.