Comparison of different quantitative evaluation protocols for peri-device leak detection using cardiac computed...

This study seeks to propose and compare different quantitative evaluation methods for identifying patients with peri-device leak (PDL) using cardiac computed tomography angiography (CCTA). Patients who had undergone left atrial appendage (LAA) closure and both transesophageal echocardiography (TEE) and CCTA were enrolled. Hounsfield units (HU) were measured in the proximal and distal regions of the left atrial appendage (p-LAA, d-LAA) on the CCTA, and the average of the two was determined (a-LAA). The relative HU ratios of the LAA to the center of the left atrium (LA) were calculated (p-LAA/c-LA, d-LAA/c-LA, a-LAA/c-LA). The area under the curve (AUC) for the LAA HU and the LAA/LA HU ratio were analyzed and compared. Fifty-one patients were included in this study. Pairwise comparisons showed a statistically significant difference (p = 0.029) in diagnostic performance between the d-LAA (AUC = 0.868) and a-LAA (AUC = 0.972). There were no significant differences between the a-LAA and p-LAA (p = 0.549) or between the d-LAA and p-LAA (p = 0.053). At the optimal cutoff for a-LAA of 115.5 HU, the sensitivity was 100%, the specificity was 88%. At the optimal cutoff for p-LAA of 109 HU, the sensitivity was 100%, the specificity was 84%. The LAA/LA HU ratio did not exhibit better diagnostic performance than HU attenuation in the LAA (p > 0.05). The a-LAA > 115.5 is useful in identifying PDL. Due to its convenience and intuitiveness, p-LAA > 109.0 can also be used as an alternative protocol for a-LAA.