Detection of left atrial appendage thrombus by dual-energy computed tomography-derived imaging biomarkers in patients with atrial fibrillation

AimsThis study aimed to assess the diagnostic performances of dual-energy computed tomography (CT)-derived iodine concentration and effective atomic number (Zeff) in early-phase cardiac CT in detecting left atrial appendage (LAA) thrombus and differentiating thrombus from spontaneous echo contrast (SEC) in patients with atrial fibrillation using transesophageal echocardiography (TEE) as the reference standard.Methods and resultsA total of 389 patients with atrial fibrillation were prospectively recruited. All patients underwent a single-phase cardiac dual-energy CT scan using a third-generation dual-source CT. The iodine concentration, Zeff, and conventional Hounsfield units (HU) in the LAA were measured and normalized to the ascending aorta (AA) of the same slice to calculate the LAA/AA ratio. Of the 389 patients, TEE showed thrombus in 15 (3.9%), SEC in 33 (8.5%), and no abnormality in 341 (87.7%) patients. Using TEE findings as the reference standard, the respective sensitivity, specificity, positive predictive value, and negative predictive value of the LAA/AA HU ratio for detecting LAA thrombus were 100.0, 96.8, 55.6, and 100.0%; those of the LAA/AA iodine concentration ratio were 100.0, 99.2, 83.3, and 100.0%; and those of the LAA/AA Zeff ratio were 100.0, 98.9, 79.0, and 100.0%. The areas under the receiver operator characteristic curve (AUC) of the LAA/AA iodine concentration ratio (0.978; 95% CI 0.945–1.000) and Zeff ratio (0.962; 95% CI 0.913–1.000) were significantly larger than that of the LAA/AA HU ratio (0.828; 95% CI 0.714–0.942) in differentiating the thrombus from the SEC (both P < 0.05). Although the AUC of the LAA/AA iodine concentration ratio was larger than that of the LAA/AA Zeff ratio, no significant difference was found between them (P = 0.259).ConclusionThe dual-energy CT-derived iodine concentration and the Zeff showed better diagnostic performance than the conventional HU in early-phase cardiac CT in detecting LAA thrombus and differentiating the thrombus from the circulatory stasis. However, these results need to be validated in large-cohort studies with late-phase images.