Diagnosis of Left Atrial Appendage Thrombi by Multiplane Transesophageal Echocardiography Interlaboratory Comparative Study

Abstract

Transesophageal echocardiography (TEE) is regarded as the method of choice for imaging left atrial appendage thrombi (LAAT). However, the interobserver variability among 2 independent echocardiographic laboratories in diagnosing LAAT by multiplane TEE has not yet been assessed. The videorecordings of 50 patients in atrial fibrillation (25 from each laboratory) were blindly reviewed by 1 experienced observer from each institution. LAAT were assessed as present, absent or questionable. Indications for TEE were: cardioversion (n=17), valve disease (n=13), endocarditis (n=12), or embolism (n=8). The prevalence of LAAT was 10% (observer 1) vs 12% (observer 2). A questionable LAAT was assessed in 6% vs 12% and a LAAT was excluded in 84% vs 76%, respectively. By head-to-head comparison, disagreement occurred in 11 cases (22%, kappa=0.5). Discrepant results were not related to the echocardiographic equipment. Problems occurred because of reverberation artifacts of the ridge between the left atrial appendage and left upper pulmonary vein (n=5), and in differentiating LAAT from spontaneous echocardiographic contrast (n=4) or an echogenic atrioventricular groove (n=1). The differentiation of pectinate muscles from LAAT was the reason for disagreement in only 1 case. Eliminating the category of questionable thrombi increased the kappa value to 0.65. In 5 patients undergoing cardiac surgery, both observers had agreed on the presence (n=1) or absence (n=4) of LAAT, and intraoperatively the results of TEE were confirmed. Even with multiplane TEE, interobserver variability among 2 independent echocardiographic laboratories for diagnosing LAAT remains high because of problems in differentiating LAAT from spontaneous echocardiographic contrast and reverberation artifacts.