Cardiac computed tomography for the localization of mitral valve prolapse: scallop-by-scallop comparisons with echocardiography and intraoperative findings.

We compared the diagnostic accuracy of cardiac computed tomography (CT) with that of echocardiography for the detection of mitral valve prolapse (MVP) on a scallop-by-scallop basis, using surgical inspection as a reference standard. This retrospective study included 145 patients (mean age 53 years; 94 men) who underwent surgical MVP repair or replacement and preoperative cardiac CT between May 2011 and October 2013. The prolapsed scallop was localized using cardiac CT and echocardiography according to the Carpentier method (anterior leaflet: from lateral to medial A1, A2, A3; posterior leaflet: P1, P2, P3). The per-scallop sensitivity and specificity of each method were compared, using surgical inspection as a reference standard. Interobserver agreement for the CT analysis was tested between three independent readers. Surgically, MVP was confirmed in 26% (226/870) scallops, with 56% (81/145) of the patients showing a single-scallop prolapse. The per-scallop sensitivity of cardiac CT was lower than that of echocardiography (80% vs. 87%, P = 0.004), with similar specificity (both 95%). For single-scallop lesions, cardiac CT showed good sensitivity (94%) and specificity (95%), with no significant difference to echocardiography. For the 64 patients with multiple scallop prolapse, CT underestimated the extent of MVP in 31 (49%) patients and echocardiography in 22 (34%) patients. Interobserver agreement was good, with kappa = 0.72-0.74. Cardiac CT provides a feasible method for localizing MVP on a per-scallop basis, but it may underestimate the extent of prolapsed scallop compared with echocardiography, particularly in patients with multiple-scallop lesions.