Prediction of poor exposure in endoscopic mitral valve surgery using computed tomography.

Abstract

In endoscopic mitral valve surgery, optimal exposure is crucial. This study aims to develop a predictive model for poor mitral valve exposure in endoscopic surgery, utilizing preoperative body profiles and computed tomography images. We enrolled patients undergoing endoscopic mitral valve surgery with available operative video and preoperative computed tomography. The degree of valve exposure was graded into 0 (excellent), 1 (fair), 2 (poor) and 3 (very poor). Intrathoracic dimensions-anteroposterior width (chest anteroposterior) and left-to-right width (chest width) of the thorax, height of right hemi-thorax (chest height), angle between the left ventricular axis and the horizontal plane (left ventricle apex angle), heart width, level of diaphragm in midline, and vertical distance between the midline diaphragm level and the highest top of the right diaphragm (Δdiaphragm) were measured. Among 263 patients, mitral valve exposure was graded as 0 in 131 (49.8%), 1 in 72 (27.4%), 2 in 46 (17.5%) and 3 in 14 (5.3%). Body mass index, chest width, left ventricle apex angle, heart width and Δdiaphragm were identified as independent predictors of grades 2 and 3 exposure by stepwise logistic regression analysis, with an area under the receiver operating characteristic curve of 0.822 (P < 0.001). Univariate logistic regression for grade 3 exposure prediction revealed that Δdiaphragm had the largest area under the curve (0.826, P < 0.001). Poor mitral valve exposure occurred in approximately one-fourth of the endoscopic surgery series and might be predicted preoperatively using body mass index and computed tomography measurements to help determine the surgical approach.