Search

Little is known about esophageal high-grade intraepithelial neoplasia dominated by cytological atypia (HGINc). We aimed to elucidate the endoscopic features of HGINc compared with esophageal high-grade intraepithelial neoplasia dominated by architectural atypia (HGINa). All patients pathologically diagnosed as esophageal high-grade intraepithelial neoplasia after endoscopic submucosal dissection at our center between January 2018 and December 2019 were included in this study. According to the pathological diagnosis, the patients were divided into two groups: HGINa group and HGINc group. Basic characteristics and endoscopic information were collected in detail. Data were analyzed statistically. Binary logistic regression was performed and a predictive model for HGINc was established. Then we evaluated its predictive value and built a nomogram for clinical application. A total of 175 patients were included in this study (126 with HGINa and 49 with HGINc). Among 228 lesions found in all patients, there were 148 HGINa and 80 HGINc. The independent relevant factors for HGINc were tobacco and alcohol usage, color, and gross type. To predict risk of HGINc, a three-factor model (TFM) was established with a highest area under curve (AUC) as 0.869 (95% CI, 0.852, 0.939). When the cut-off value was set as 0.3569184, the diagnostic accuracy, sensitivity, specificity, positive predictive value, and negative predictive value for HGINc was 81.14%, 88.75%, 77.03%, 67.62%, and 92.68%, respectively. HGINc differs greatly in endoscopic features from HGINa in our study. It's important to reduce misdiagnosis that our model was established with good predictive value for clinical application.