Left atrial appendage filling defects on 64-slice multidetector computed tomography in patients undergoing pulmonary vein isolation: predictors and comparison to transesophageal echocardiography.

OBJECTIVES

To evaluate predictor variables and accuracy of left atrial appendage (LAA) filling defects seen on multidetector computed tomographic (MDCT) scan in predicting LAA thrombus in patients undergoing pulmonary vein (PV) isolation procedure.

METHODS

Electrocardiogram-gated 64-slice MDCT and transesophageal echocardiography (TEE) were undertaken in 51 consecutive patients with nonvalvular atrial fibrillation who were referred for circumferential antral pulmonary vein isolation.

RESULTS

In 51 patients (37 men; mean age, 64 years), left atrium (LA) diameter emerged as the predictor of LAA filling defects (odds ratio, 4.9; 95% confidence interval, 1.19-20.25). Left atrial appendage filling defects had sensitivity of 100%, specificity of 95.92%, positive predictive value of 0.5, and negative predictive value of 1, for thrombi seen on TEE image. A mean LAA/ascending aorta Hounsfield unit ratio of 0.78 or less was identified as a sensitive predictor of thrombus on TEE (sensitivity, 100%; specificity, 87.8%; positive predictive value, 0.25; and negative predictive value, 1).

CONCLUSIONS

A larger LA predisposes to LAA filling defects on MDCT scan. Pending prospective validation, absence of LAA filling defects on 64-slice MDCT may reliably exclude LAA thrombi in patients with nonvalvular atrial fibrillation obviating the need for TEE.