Morphological assessment of the aortic valve using coronary computed tomography angiography, cardiovascular magnetic resonance, and transthoracic echocardiography: comparison with intraoperative findings

To compare the diagnostic accuracies of coronary computed tomography angiography (CCTA), cardiovascular magnetic resonance (CMR), and transthoracic echocardiography (TTE) in aortic valve (AV) morphological assessments with operative findings. We retrospectively enrolled 262 patients who underwent CCTA, CMR, and TTE before AV surgery. Two independent blinded observers assessed AV morphology as being tricuspid, bicuspid, or quadricuspid using three imaging modalities. Interobserver and intermodality agreements were obtained with kappa statistics. The diagnostic accuracies of CCTA, CMR, and TTE for identifying AV morphology (tricuspid vs. non-tricuspid) were compared with intraoperative findings as the reference standard. At surgery, tricuspid AV, bicuspid AV, and quadricuspid AV were present in 179, 80, and 3 patients, respectively. The CCTA and CMR image qualities were all diagnostic. Thirteen cases of TTE were not evaluable due to severe AV calcification. An excellent correlation between CMR and CCTA was seen for the identification of AV morphology (κ = 0.97). Good correlations existed between CCTA and TTE (κ = 0.72) and between CMR and TTE (κ = 0.74). CCTA, CMR, and TTE had an excellent or good interobserver agreement (κ = 0.90, 0.95, and 0.72, respectively). Sensitivity, specificity, and positive and negative predictive values for AV morphology assessment (tricuspid vs. non-tricuspid) were: 97, 95, 98, and 94 % with CCTA (n = 262); 98, 96, 98, and 95 % with CMR (n = 262); and 98, 88, 95, and 96 % with TTE (n = 249). CCTA and CMR are highly accurate for identifying AV morphology.