A new risk-stratified scoring system for predicting left atrial appendage thrombus in patients with nonvalvular atrial fibrillation
摘要
The predictive capacity of the CHA2DS2-VASc score for left atrial appendage thrombus (LAAT) detection in nonvalvular atrial fibrillation (NVAF) patients shows significant limitations.
ObjectiveTo recognize predictors of LAAT and create a more precise risk-assessment model.
MethodsBetween January 2019 and December 2023, consecutive NVAF patients without a previous history of anticoagulation who underwent transesophageal echocardiography were recruited from two centers. Clinical data, biomarker information, and transthoracic echocardiographic parameters were collected in a standardized manner. The derivation cohort comprised patients from one center, while the validation cohort consisted of participants from the other center.
Results1505 patients (mean age 63.4±9.77 years; 895 male) were analyzed. LAAT was detected in 109 (7.24%) of the total 1505 patients. The final parameters selected for the LEFT-AF risk model included left atrial diameter (LAD), left ventricular ejection fraction (LVEF), history of heart failure (HF), history of stroke, and non-paroxysmal AF (NPAF). In the derivation cohort, the novel scoring system demonstrated superior discriminative performance compared to the currently used CHA2DS2-VASc score (0.693, 95% CI 0.634-0.752) (P<0.001), CHA2DS2 score (0.679, 95% CI 0.621-0.731) (P<0.001) and CLOTS-AF score (Creatinine>1.5mg/dL, LVEF<50%, LAVI>34ml/m2, TAPSE<17mm, Stroke, AF rhythm) (0.762, 95% CI 0.704-0.821), with an area under the receiver operating characteristic curve (AUC) of 0.855 (95% CI 0.808-0.902) (P<0.001). This same superior performance was maintained in the validation cohort.
ConclusionAmong patients with NVAF who had not previously received anticoagulation therapy, the prevalence of LAAT was 7.24%. The LEFT-AF score improves LAAT risk stratification, particularly in patients with low CHA2DS2-VASc scores, potentially guiding anticoagulation decisions.
Graphical abstract