Pre-Interventional Transesophageal Echocardiography as a Reliable Predictor of Residual Shunt Following Patent Foramen Ovale Closure

Background
Closure of a patent foramen ovale (PFO) is an effective strategy in the prevention of recurrent stroke after cryptogenic stroke. Residual shunt (RS) is a common issue following PFO closure and may affect safety and efficacy. Transesophageal echocardiography (TEE) is the key diagnostic tool, but standardized assessment of morphological parameters to prevent RS remains challenging.

Objectives
In this study, we investigate the diagnostic value of different anatomical parameters assessed by transesophageal echocardiographic to predict RS after PFO closure.

Methods
We consecutively enrolled five-hundred and twenty-seven (n=527) patients undergoing PFO closure. We performed pre-interventional TEE and after PFO closure we then screened for RS at six-months follow-up.  

Results
Pre-interventional TEE measures of PFO morphology revealed significant differences in patients with RS in comparison to those with closed PFO. Incidence of RS was significantly more frequent in patients with increasing PFO size (p=0.025) (Figure 1A) and atrial septum aneurysm (p=0.022) (Figure 1B).  In patients with RS, we found significantly increased excursion (p=0.005) (Figure 1C) and length (p=0.005) of septum primum (Figure 1D), as well as increased PFO tunnel length (p=0.036) (Figure 1E). Thus, anatomical complexity contributed to a clinically relevant residual shunting following percutaneous PFO closure (Figure 1F).  Further, we observed that patients with a larger device size (right atrial disk diameter >25 mm) had a significantly (p<0.0001) higher incidence of RS compared to those who received a smaller device indicating a device-associated risk for incomplete PFO closure (Figure 2A). However, the ratio between RA disk size and septal excursion was significantly lower (p=0.025) in patients with RS compared to those with closed PFO (Figure 2B). Similarly, the occluder size-to-shunt ratio was lower (p=0.002) in RS patients (Figure 2C) suggesting that the device was undersized. Multivariable logistic regression, adjusted for device size, identified PFO shunt size (OR 1.65; p=0.022) and septum primum excursion (OR 1.99; p=0.048) as independent predictors of RS, with an AUC of 0.68 (p<0.0001) (Figure 2D).

By training machine learning models on TEE parameters, stratification of PFO morphology resulted in high diagnostic accuracy to predict RS after PFO closure (Figure 3A). Correspondingly, transformed risk scores from the predictive model significantly (p<0.0001) distinguished between patients with RS and those with sufficient PFO closure (Figure 3B). Incorporation of echocardiographic and device-related features unveiled specific parameters critical for prediction of RS (Figure 3C&D).

Conclusions
Our study shows that baseline assessment of PFO morphology using TEE improves the identification of patients at risk for RS after PFO closure. A standardized TEE approach may enhance the procedure's safety and effectiveness. Key clinical implications include: (1) Routine pre-interventional TEE enables precise planning for safer PFO closure; (2) Identifying patients with complex anatomy aids in risk stratification for RS; (3) High-risk patients may need closer follow-up with TEE and could benefit from prolonged or intensified antiplatelet or anticoagulant therapy to prevent recurrent embolic events.