ABSTRACT
BACKGROUND: Persistent foramen ovale (PFO) contributes to cryptogenic stroke and is associated with stroke recurrence, although the exact mechanism of ischemic events is not fully understood. Several biomarkers have been developed for the prediction of atrial fibrillation after stroke, but there are currently only limited data on their potential value for the diagnosis of PFO-related stroke. METHODS: This study was a prospective single-center study that included all patients hospitalized between March 31, 2018, and January 18, 2020, in the stroke department of the Dijon University Hospital for ischemic stroke without obvious cause and without a history of atrial fibrillation. PFO was systematically screened by transthoracic echocardiography and images were reviewed by an independent cardiologist blinded from clinical data. PFO was defined according to the CLOSE trial criteria: PFO associated with interatrial septal aneurysm or significant interatrial shunt (> 30 microbubbles in the left atrium within three cardiac cycles after right atrial opacification). The potential association of PFO-related stroke with biomarkers of cardiac fibrosis and inflammation such as galectin-3, GDF-15, ST-2, osteoprotegerin and NT-proBNP was tested using multivariate backward stepwise logistic regression. RESULTS: Of the 240 patients included in the SAFAS study, 229 had complete echocardiographic data, and 23 (10%) had PFO-related stroke. Patients with PFO-related stroke were significantly younger (58±14 vs. 69±14, P<0.001), had less frequent previous arterial hypertension (30 vs. 60%, P=0.008), and more frequent cerebellar territory involvement (26 vs. 9%, P=0.014) compared to the other patients. In addition, they had less frequently left atrial dilatation (left atrial index volume>34mL/m2 [9 vs. 35%, P=0.009]). After ROC curve analysis for definition of thresholds, PFO-related stroke patients more often had galectin-3<9.5ng/mL (59 vs. 27%, P=0.002), ST2<13380pg/ml (23 vs. 50%, P=0.007), GDF-15<1200ng/mL (59 vs. 27%, P=0.002), osteoprotegerin<1133pg/mL (82 vs. 58%, P=0.033) and NT-proBNP<300pg/mL (88 vs. 55%, P=0.009). After multivariate analysis, only galectin-3<9.5ng/mL (OR [95% CI] 3.4 [1.18; 9.8], P=0.024) and osteoprotegerin<1133pg/L (OR [95% CI] 5.0 [1.1; 22.9], P=0.038) were independently associated with PFO-related stroke. CONCLUSION: Patients in whom cryptogenic stroke is attributed to a significant PFO have a specific clinical and biological phenotype. Low levels of galectin-3 and osteoprotegerin may help identify patients with PFO-related strokes.