Development and Validation of an Individualized Nomogram for Predicting Patent Foramen Ovale-Associated Stroke: Patent Foramen Ovale Morphology-Based Analysis

Background

We previously reported four patent foramen ovale (PFO) morphological types that influenced right-to-left shunt (RLS) grades. Herein, we aimed to study the relationship between PFO morphology and cryptogenic stroke (CS). We further developed a nomogram based on four PFO morphological types and functional parameters to guide clinicians in judging the risk of PFO-associated stroke.

Methods

This was a retrospective observational study involving adult patients with PFO between January 2020 and November 2022. Patients were divided into a PFO-associated stroke group (CS group) and a group without cryptogenic stroke (non-CS group). Four types of PFO and RLS grades were analyzed. Nomograms were made to predict PFO-associated stroke using multivariable logistic regression analysis. The discrimination performance of the model was internally validated and assessed using the receiver operating characteristic.

Results

We enrolled 389 patients (male, 182 patients; female, 207 patients) with PFO, the mean age was 43.3 ± 8.1 years. The derivation cohort comprised 293 patients (CS group, 186 patients; non-CS group, 107 patients). The predictive nomogram comprised PFO morphological types, interatrial septum (IAS) mobility distance, septum primum thickness, PFO channel length at rest, and contrast-transthoracic echocardiography (c-TTE) RLS grade during the Valsalva maneuver. A validation cohort was established (CS group, 61 patients; non-CS group, 35 patients). The model area under the curve (AUC) was 0.891 (95% confidence interval = 0.855−0.928) in the derivation cohort and 0.935 (95% confidence interval = 0.885−0.986) in the validation cohort. Calibration curve analysis showed that the nomogram had a C-index of 0.891 in the derivation cohort and 0.935 in the validation cohort. The decision curve analysis (DCA) indicated that the nomogram had clinical applicability.

Conclusions

Adding four PFO morphological types improved the risk stratification capability for PFO-associated stroke. The nomogram can identify high or low-risk PFO individuals and select patients who will likely benefit from interventional device closure.