Diagnostic Performance of Transesophageal Echocardiography and Cardiac Computed Tomography in Infective Endocarditis

Background: Multimodality imaging is essential for infective endocarditis (IE) diagnosis. The aim of this work was to evaluate the agreement between transesophageal echocardiography (TEE) and cardiac computed tomography (CT) findings in patients with surgically confirmed IE.

Methods: Sixty-eight patients (mean age 63 ± 2 years) with a definite diagnosis of left-side IE according to the modified European Society of Cardiology Duke criteria, on both native and prosthetic valves, underwent TEE and cardiac CT before surgery. The presence of valvular (vegetations, erosion) and paravalvular (abscess, pseudoaneurysm) IE-related lesions were compared between both modalities. Perioperative inspection was used as reference.

Results: TEE performed better than CT in detecting valvular IE-related lesions (TEE area under the curve [AUC_TEE] = 0.881 vs AUC_CT = 0.720, P = .02) and was similar to CT with respect to paravalvular IE-related lesions (AUC_TEE = 0.830 vs AUC_CT = 0.816, P = .835). The ability of TEE to detect vegetation was significantly better than that of CT (AUC_TEE = 0.863 vs AUC_CT = 0.693, P = .02). The maximum size of vegetations was moderately correlated between modalities (Spearman's rho = 0.575, P < .001). Computed tomography exhibited higher sensitivity than TEE for pseudoaneurysm detection (100% vs 66.7%, respectively) but was similar with respect to diagnostic accuracy (AUC_TEE = 0.833 vs AUC_CT = 0.984, P = .156).

Conclusions: In patients with a definite diagnosis of left-side IE according to the modified European Society of Cardiology Duke criteria, TEE performed better than CT for the detection of valvular IE-related lesions and similar to CT for the detection of paravalvular IE-related lesions.