Accuracy and reliability of left atrial appendage morphology assessment by new 3D transesophageal echocardiographic rendering modalities: a comparative study with computed tomography

Abstract

Abstract Funding Acknowledgements Type of funding sources: None. Background/Introduction. Left atrial appendage (LAA) morphology assessed by contrast-enhanced computed tomography (CT) has been associated to the risk of cardioembolic stroke in non-valvular atrial fibrillation. Subsequent studies with the traditional LAA classification system (CS) into 4 morphologies (Chicken wing, Cauliflower, Cactus and Windsock) yielded mixed results in terms of reliability and stroke risk association. Recently, a simple LAA morphology CS (new-LAAcs) based on the LAA bend angle measurement has been suggested. Three-dimensional transesophageal echocardiography (3D TOE) quality imaging has been improved and new volume rendering modalities developed. Purpose. Aim of this study was to evaluate the accuracy and reliability of 2D and new 3D TOE rendering modalities compared to CT in assessing LAA morphology. We used and validated a new simple LAA morphology classification system (new-LAAcs) based on the LAA bend angle in contrast to the traditional CS. Methods. 50 consecutive patients who underwent both cardiac CT and TOE were enrolled. LAA morphology was assessed by three different TEE modalities: (1) 2D TOE inspective evaluation (2D TOE), (2) 3D TOE multiplanar reconstruction (3D TOE MPR) and (3) 3D TOE Philips TrueVue Glass rendering (3D TOE GLASS). We assessed TOE accuracy compared to CT by sensitivity, specificity, accuracy, and Cohen’s kappa. Two trained readers independently adjudicated LAA morphologies in the new-LAAcs and the inter-rater reliability was obtained by percentage agreement and Cohen’s kappa. The reliability of the new- vs. traditional-LAAcs was assessed by CT in terms of reliability rates and influence on LAA morphology prevalence. Results. CT and TOE imaging analyses were feasible in all patients. 2D TOE was fairly accurate in identifying LAA morphology (κ 0.38, p = 0.022) and had only moderate inter-rater (κ 0.46, p = 0.027) and substantial intra-rater (κ 0.62, p = 0.003) reliability rates. 3D TOE showed high validity: 3D TOE MPR had an almost perfect accuracy (κ 0.84, p < 0.001) and substantial (κ 0.77, p < 0.001) inter-rater reliability; 3D TOE GLASS substantial accuracy (κ 0.67, p < 0.001) and almost perfect (κ 0.82, p < 0.001) inter-rater reliability. Intra-rater agreement was almost perfect for both 3D TOE modalities (κ 0.84, p < 0.001). In the comparison among CS the traditional-LAAcs inter-rater reliability was moderate (κ 0.47, p < 0.001) and the intra-rater reliability substantial (κ 0.68, p < 0.001) while the new-LAAcs yielded an almost perfect reliability level (inter-rater κ 0.84, p < 0.001 and intra-rater κ 0.93, p < 0.001). With the traditional-LAAcs, the prevalence of CW LAA was 30 (60%), while with the new-LAAcs the prevalence of low-risk-LAA was 13 (26%), leading to classify 17 (57%) CW morphologies as high-risk-LAA. Conclusions. 3D TOE is an accurate, reliable, and feasible alternative to CT in assessing LAA morphology with the new-LAAcs. The new-LAAcs shows higher reliability rates than the traditional one. Abstract Figure. Abstract Figure.