Abstract
BACKGROUND: The CHADS2 scoring system is simple and widely accepted for predicting thromboembolism in patients with nonvalvular atrial fibrillation (NVAF). Although congestive heart failure (CHF) is a component of the CHADS2 score, the definition of CHF remains unclear. We previously reported that the presence of CHF was a strong predictor of left atrial appendage (LAA) thrombus. Therefore, the present study aimed to elucidate the relationship between LAA thrombus and the brain natriuretic peptide (BNP) level in patients with unanticoagulated NVAF. METHODS: The study included 524 consecutive patients with NVAF who had undergone transesophageal echocardiography to detect intracardiac thrombus before cardioversion between January 2006 and December 2008, at Hiroshima City Asa Hospital. The exclusion criteria were as follows: paroxysmal atrial fibrillation, unknown BNP levels, prothrombin time international normalized ratio ≥2.0, and hospitalization for systemic thromboembolism. RESULTS: Receiver operating characteristic analysis yielded optimal plasma BNP cut-off levels of 157.1 pg/mL (area under the curve, 0.91; p<0.01) and 251.2 pg/mL (area under the curve, 0.70; p<0.01) for identifying CHF and detecting LAA thrombus, respectively. Multivariate analyses demonstrated that a BNP level >251.2 pg/mL was an independent predictor of LAA thrombus (odds ratio, 3.51; 95% confidence interval, 1.08-10.7; p=0.046). CONCLUSIONS: In patients with unanticoagulated NVAF, a BNP level >251.2 pg/mL may be helpful for predicting the incidence of LAA thrombus and may be used as a surrogate marker of CHF. The BNP level is clinically useful for the risk stratification of systemic thromboembolism in patients with unanticoagulated NVAF.