A novel algorithm based on left atrial strain parameters in patients with non-valvular atrial fibrillation could identify the thrombogenic milieu of left atrial appendage

Abstract

Background

The role of the new derived parameters of left atrial (LA) strain in cardioembolic risk stratification is unknown, besides, new algorithm is needed to provide incremental value of cardioembolic risk stratification in patients with atrial fibrillation (AF), especially in those with CHA2DS2-VASc score of 0–2.

Methods

We enrolled 566 consecutive subjects with non-valvular AF, who underwent transesophageal echocardiography. Left atrial appendage (LAA) thrombogenic milieu, as a surrogate for cardioembolic risk, was defined as the presence of a thrombus, severe spontaneous echo contrast, or sludge in the LAA. The impaired LAA emptying velocity was defined as LAA emptying velocity ≤ 30 cm/s. To classify LAA thrombogenic milieu, a decision tree analysis was performed to explore the way of the combination characteristic echocardiographic variables. LA strain parameters includes left atrial reservoir strain (LASr), left atrial systolic stain rate (LASRr), LA filling index, LA filling rate, LA stiffness index, and LA stiffness rate.

Results

Among the 566 subjects, 176 (31.1 %) identified with LAA thrombogenic milieu. Compared those without, LA filling index, LA filling rate, LA stiffness index, and LA stiffness rate was significantly increased in the patients with LAA thrombogenic milieu. The multiple logistic regression analysis suggested that left atrial strain parameters (respectively) were independently correlated with LAA thrombogenic milieu. Left atrial strain parameters (AUC: 0.846–0.916) exhibited good accuracy for identifying LAA thrombogenic milieu, and non-inferior to conventional parameters including CHA2DS2-VASc score (0.645). The decision tree analysis identified LASr, LASRr, LAEF, CHA2DS2VASc score, and LA stiffness index as the most relevant variables to correctly discriminate LAA thrombogenic milieu from patients with AF. The decision tree as a novel algorithm could accurately identify subjects with LAA thrombogenic milieu (AUC 0.930, accuracy 89.31 %) or impaired LAA emptying velocity (AUC 0.868, accuracy 87.98 %). In the subgroup analysis, among 354 the subjects with lower CHA2DS2VASc score (0–2), 85 patients with LAA thrombogenic milieu. The novel algorithm (AUC: 0.938, accuracy 90.39 %) also performed well to discriminate subjects with LAA thrombogenic milieu.

Conclusion

The LA strain parameters were non-inferior to conventional parameters for risk stratification of LAA thrombogenic milieu in AF patients. Furthermore, a novel algorithm, based on left atrial strain parameters and CHA2DS2-VASc score, could accurately identify patients with LAA thrombogenic milieu in non-valvular AF, even in those with lower CHA2DS2VASc score (0–2).