Establishment and validation of a clinical diagnostic model for gastric low-grade intraepithelial neoplasia

Objective:A clinical diagnostic model of gastric low-grade intraepithelial neoplasia (LGIN) was developed and validated to improve the identification of precancerous lesions in gastric cancer.Methods:A retrospective analysis of 1211 patients with chronic atrophic gastritis (CAG) and 1089 patients with LGIN admitted to the Endoscopy Center of the First Affiliated Hospital of Bengbu Medical College from January 2016 to December 2021 was performed to record basic clinical and pathological information.A total of 1756 patients were included after screening and were divided unequally and randomly into 2 groups, one for establishing an LGIN predictive nomogram (70% of patients) and the other for external validation of the model (30% of patients). R software was used for statistical analysis.Methods:A retrospective analysis of 1211 patients with chronic atrophic gastritis (CAG) and 1089 patients with LGIN admitted to the Endoscopy Center of the First Affiliated Hospital of Bengbu Medical College from January 2016 to December 2021 was performed to record basic clinical and pathological information.A total of 1756 patients were included after screening and were divided unequally and randomly into 2 groups, one for establishing an LGIN predictive nomogram (70% of patients) and the other for external validation of the model (30% of patients). R software was used for statistical analysis.Results:The nomogram was built with 10 predictors: age, sex, lesion location, intestinal metaplasia, multiple location, lesion size, erosion, edema, surface white fur, and form. The calibration curves showed good agreement between the predicted and actual diagnoses. The C-indexes were 0.841 (95% CI: 0.820-0.863) in the training dataset, 0.833 in the internal validation dataset, and 0.842 in the external validation dataset (Hosmer-Lemeshow test, P = .612), showing satisfactory stableness.Conclusions:This study provides a visual mathematical model that can be used to diagnose high-risk LGIN, improve follow-up or endoscopic treatment and the detection rate of precancerous gastric cancer lesions, reduce the incidence of gastric cancer, and provide a reliable basis for the treatment of LGIN.