Spectral Computed Tomography for the Quantitative Assessment of Patients With Carcinoma of the Gastroesophageal Junction: Initial Differentiation Between a Diagnosis of Squamous Cell Carcinoma and Adenocarcinoma.

Abstract

This study aimed to distinguish between esophageal squamous cell carcinoma (ESCC) and esophageal adenocarcinoma (EAC) using spectral computed tomography (CT) and to discuss the accuracy according to an optimal threshold of single and combined parameters. In this monoinstitution study, 61 patients, 35 of whom had ESCC and 26 had EAC confirmed by surgery or esophagoscopy, were recruited from August 2016 to March 2017. Enrolled patients underwent dual-phase chest CT enhancement. The spectral CT parameters (NIC, NICD, NICratio, Zeff, Zeff-C, K40-70 keV, K80-100 keV, and K110-140 keV) were measured during arterial phase (AP) and venous phase (VP). Binary logistic regression was used to calculate combined predictive probability. Thresholds of quantitative parameters and diagnostic accuracy were calculated using receiver operating characteristic curve. Compared with ESCC, higher NICAP, NICVP, NICD, Zeff AP, Zeff VP, Zeff-C AP, and Zeff-C VP were observed for EAC, whereas NICratio was lower for EAC. Higher K40-70 keV, K80-100 keV, and K110-140 keV were exhibited in EAC than in ESCC. Area under the curve (AUC) of NICAP, K40-70 keV AP, and Zeff AP were 0.720, 0.730, and 0.706, respectively. The area under the curve of new combined predictive value of NICAP and λ40-0 keV AP was 0.804. The sensitivity and specificity were 77.80% and 80.60%, respectively, when the threshold of new predictive value was 0.60. The diagnostic accuracy obtained by using NICAP and K40-70 keV AP combined is better than that obtained using a single parameter in differentiation between a diagnosis of squamous cell carcinoma and adenocarcinoma.