E880 JACC April 5, 2011 Volume 57, Issue 14
IMAGING AND DIAGNOSTIC TESTING USE OF CARDIAC COMPUTED TOMOGRAPHY IN DETECTION (QUALITATIVE AND QUANTITATIVE) OF LEFT ATRIAL APPENDAGE THROMBUS ACC Poster Contributions Ernest N. Morial Convention Center, Hall F Tuesday, April 05, 2011, 9:30 a.m.-10:45 a.m.
Session Title: Advances in Noncoronary Applications of Cardiac CT Abstract Category: 37. CT Coronary Calcium and Noncoronary CT Applications Session-Poster Board Number: 1170-195 Authors: Adekunle Shittu, Yalcin Hacioglu, Eli Gang, Dong Li, Harpreet Bhatia, Ronald P. Karlsberg, Juan Alvergue, Matthew J. Budoff, Los Angeles Biomedical Research Institute, Torrance, CA, Cardiovascular Research Foundation of Southern California, Beverly Hills, CA Background: Transesophageal Echocardiography (TEE) is the gold standard for detecting Left Atrial Appendage (LAA) thrombus. Previous studies have reported cardiac computed tomography (CCT) to have a high sensitivity and negative predictive value for detecting thrombus in left atrial appendage. Two previous studies determined optimal LAA/Aorta (AA) Hounsfield (HU) density ratio for detection of LAA thrombus to be 0.25 and 0.75 respectively. We thus sought to determine both the optimal cut-off for LAA HU and LAA/Aorta HU in detecting LAA thrombus compared to TEE. Methods: We retrospectively reviewed images/results of 84 patients who had both CCT and TEE done at a mean average interval of 24 days between studies.LAA was evaluated by multidetector computed tomography qualitatively (visual filling defect) and quantitatively (measurement of LAA HU and LAA/Aorta HU densities) using a 1cm2 area of interest in the same axial plane. Results were compared to TEE visualization of thrombus or spontaneous echo contrast. Results: Compared to TEE detection of thrombus in LAA , qualitative identification by multidetector computed tomography had a sensitivity 100%, specificity 77.9%, positive predictive value 51.6%, negative predictive value 100% and total accuracy of 82.1%. The optimal LAA HU density cutoff for thrombus detection was 119 with sensitivity 88% , specificity 86% and area under receiver-operator characteristic curve (AUC) 0.923 (p=0.0004). The optimal LAA/Aorta HU density cut off point was 0.242 with sensitivity 87%,specificity 88%, AUC 0.921 (p=0.0011). There is no significant difference (p=0.72) between both AUC. In comparison to the derived optimal cut-off points for thrombus detection , LAA HU of 77 has sensitivity 76% and specificity 90% and LAA/Aorta HU of 0.728 has sensitivity 100% and specificity 70%. Conclusions: Multidetector computed tomography is an imaging modality to exclude LAA thrombus qualitatively ( high negative predictive value). Quantitative measurement of LAA HU density (cut off 119) or LAA/Aorta HU density ratio (cut-off 0.242) improves accuracy of positively detecting LAA thrombus.