Introduction: Multidetector CT (MDCT) is used prior to atrial fibrillation ablation (AFA) to anatomically guide ablation procedures. Whether 64-slice MDCT also can be used to diagnose left atrial thrombus is not known.
Methods: We sought to determine the accuracy and interobserver variability of MDCT in the evaluation of left atrial thrombus prior to AFA. We enrolled 50 patients scheduled for AFA who underwent 64-slice MDCT scan and transesophageal echocardiography prior to the procedure. Three experienced observers reviewed all the MDCT images for the presence of a left atrial thrombus, and two different readers interpreted the transesophageal echocardiograms (TEE), which were used as the gold standard. All observers were blinded to clinical data and each other.
Results: Interobserver variability between the three MDCT readers was poor (highest kappa statistic 0.43, P = 0.001). Diagnostic accuracy was highly variable, with sensitivities ranging from 100% to 50% and specificities ranging from 85% to 44%. TEE reader agreement was 98%.
Conclusion: MDCT demonstrates high interobserver variability and has only modest diagnostic accuracy for the detection of left atrial thrombus in patients undergoing AFA procedure. Potential factors affecting the accuracy of MDCT include image quality and the difficulty of distinguishing clot from pectinate muscle. MDCT likely is not the optimal method to detect left atrial thrombus using current techniques and standards of interpretation.