ABSTRACT
The main mechanism of the CHADS2 and CHA2DS2-VASc scores to predict stroke in nonvalvular atrial fibrillation (NVAF) is still controversial. We evaluated the association of the CHADS2 and CHA2DS2-VASc scores with left atrial thrombus (LAT) as detected by transesophageal echocardiographic (TEE) and compared the predictive ability of these risk stratification schemes with nonvalvular atrial fibrillation (NVAF). Data from 2,695 consecutive NVAF patients in whom TEE was performed for screening LAT from July 2007 to February 2014 were analyzed. Only 3% of the subjects had LAT. Presence of LAT was not significantly associated with either CHADS2 (P = 0.07) or CHA2DS2-VASc score (P = 0.12). The area under the curve (AUC) concerning LAT prediction using CHADS2 and CHA2DS2-VASc was 0.574 and 0.569, respectively. A composition model includes previous stroke or transient ischemic attack, nonparoxysmal AF, moderate to severe left ventricular systolic dysfunction, left atrial enlargement, and cardiomyopathy which improved the discrimination significantly (AUC = 0.743). In our cohort, both CHADS2 and CHA2DS2-VASc scores were of limited value for predicting LAT in patients with NVAF. This questions the CHADS2/CHA2DS2-VASc score predicting stroke mainly through the mechanism of cardiogenic embolism. A scoring scheme combining clinical and echocardiographic parameters may better predict LAT as a surrogate for cardioembolic risk in NVAF patients.
