Clinical risk prediction for left atrial appendage thrombus among patients with atrial fibrillation

Abstract

Abstract Background Exclusion of left atrial appendage thrombus (LAAT) by transoesophageal echocardiography (TOE) is recommended in patients with inadequate anticoagulation prior to direct cardioversion (DCR) or catheter ablation for atrial fibrillation/flutter (AF/AFL). LAAT risk factors in this population remain poorly defined. Purpose Determine LAAT predictors in AF/AFL patients undergoing pre-procedural TOE. Methods We evaluated available clinical and transthoracic echocardiographic (TTE) parameters in AF/AFL patients undergoing TOE between 1999–2022 in our institution in Melbourne, Australia. Regression analysis identified predictors of LAAT, which were applied to a weighted score developed in the derivation cohort (70%) and validated in the remaining 30%. Results Of 627 patients (age 62±12 years, 27% female, AF 84%,AFL 16%, left ventricular ejection fraction (LVEF) 44±20%), 24% had LAAT and 13.8% dense spontaneous echo contrast precluding DCR. Anticoagulation was NOAC 56.5%, warfarin 32.1% and none in 11.4%. In the LAAT cohort, thrombus resolution occurred in 39% on serial transoesophageal imaging with a median time to resolution of 131 days (IQR 54–398). Diabetes (p=0.004), prior stroke (p=0.009), coronary disease (p=0.015), renal impairment (p<0.001) and CHADS2VASc >2 (73% vs. 55%, p<0.001) were higher in the LAAT cohort. Age (p=0.093), gender (p=0.689), BMI (p=0.828), anticoagulant type (p=0.316) and diabetes (p=0.107) were not univariate predictors, whereas anticoagulation duration (<30 days), creatinine and TTE markers of remodeling (LVEF, LAVI, RVSP and TAPSE) were independent predictors on univariate and multivariate regression; CHADS2VASc was not significant after adjustment (p=0.090). The weighted risk model included continuous (age, creatinine, LVEF, LAVI, TAPSE and RVSP) and categorical (anticoagulation duration) variables with excellent predictive performance: AUC 0.872 (95% CI 0.798–0.946), PPV 91%, NPV 70% and accuracy 80%. Conclusion A novel LAAT risk model comprising clinical and echocardiographic parameters enhances risk prediction over CHADS2VASc in AF/AFL and may guide the need for pre-procedural TOE imaging. Funding Acknowledgement Type of funding sources: None.