Anatomical and functional predictors of uncrossable PFO in patients with cryptogenic stroke

Background

Transcatheter patent foramen ovale (PFO) closure is widely recommended to prevent recurrent cryptogenic stroke in patients with moderate-to-large shunts. However, a substantial proportion of patients have “uncrossable” PFOs in which guidewire passage fails, potentially leading to futile procedures or the need for transseptal puncture. Echocardiographic markers that reliably predict the technical feasibility of PFO closure have not been clearly defined.

Methods

We retrospectively evaluated patients with cryptogenic stroke or transient ischemic attack who underwent TEE and were referred for catheter-based PFO closure at a tertiary center between April 2010 and March 2022. Patients with concomitant atrial septal defect or transseptal puncture–assisted closure were excluded. Uncrossable PFO was defined as persistent failure of guidewire advancement through the PFO tunnel despite escalation to support sheaths. TEE parameters included shunt grade, PFO height, tunnel length, presence of bubbles at the PFO outlet during the Valsalva maneuver, resting shunt, atrial septal aneurysm or septal hypermobility, Eustachian valve or Chiari network, and the inferior vena cava–PFO (IVC–PFO) angle. Predictors of successful guidewire passage were identified using multivariable logistic regression and receiver operating characteristic (ROC) analyses.

Results

Among 240 patients undergoing attempted PFO closure, 90 (37.5%) had uncrossable PFOs. Baseline demographic and clinical characteristics, including shunt grade, were similar between groups. In contrast, uncrossable PFOs were characterized by smaller PFO height, longer tunnel length, narrower IVC–PFO angle, and markedly lower frequencies of resting shunt and presence of bubbles at the PFO outlet.

In multivariable analysis, the presence of bubbles at the PFO outlet emerged as the strongest independent predictor of successful guidewire passage (adjusted OR 7.47), followed by PFO height ≥2 mm (adjusted OR 5.67), larger IVC–PFO angle (per 1° increase; adjusted OR 1.05), and resting shunt (adjusted OR 10.7). Shunt grade alone was not predictive. Sequential incorporation of these anatomical and functional markers resulted in progressive improvement in discrimination, with the highest predictive accuracy achieved when outlet-level bubble presence was included (AUC 0.845).

Conclusions

Approximately one-third of patients with cryptogenic stroke and moderate-to-large shunts had uncrossable PFOs despite similar shunt grades. TEE-derived functional and anatomical markers—particularly the presence of bubbles at the PFO outlet during the Valsalva maneuver, along with PFO height and IVC–PFO angle—were strongly associated with guidewire passage and substantially enhanced prediction beyond shunt grade alone. Incorporation of these parameters into pre-procedural assessment may refine patient selection, reduce futile closure attempts, and guide consideration of alternative medical therapy.