Development and internal validation of a risk prediction model for gastric precancerous lesions integrating...

This study aimed to develop and internally validate a clinical risk prediction model for precancerous lesions of gastric cancer (PLGC) in Helicobacter pylori (H. pylori)-positive individuals by integrating H. pylori strain typing, serological gastric function markers, lifestyle factors, and host susceptibility polymorphisms. A total of 82 H. pylori-positive outpatients who underwent gastroscopy with histopathology between July 2020 and December 2021 were enrolled and categorized into a PLGC group (n = 38; chronic atrophic gastritis, intestinal metaplasia, and/or dysplasia) and a non-PLGC control group (n = 44). Demographic, clinical, dietary, and lifestyle information was collected using standardized questionnaires. Fasting blood samples were obtained for pepsinogen I (PG I), pepsinogen II (PG II), pepsinogen I/II ratio (PGR), gastrin-17 (G-17), and serologic H. pylori typing (Type I vs Type II), and 4 gastric cancer susceptibility loci (PSCA rs2976392, PLCE1 rs2274223, PRKAA1 rs59133000, and MUC1 rs4072037) were genotyped using TaqMan assays. Multivariable logistic regression identified older age and Type I H. pylori infection as independent risk factors, whereas higher PG I, higher PGR, and frequent whole-grain intake were protective. Susceptibility loci were not retained in the final model. The derived model showed adequate calibration (Hosmer-Lemeshow P = .426) and strong discrimination (AUC = 0.969), with an optimal cutoff of 0.403 yielding 94.7% sensitivity and 93.2% specificity; bootstrap validation indicated stable performance. This model may support risk stratification and targeted surveillance among H. pylori-positive patients.