Risk assessment of type I gastric neuroendocrine tumors based on endoscopic and clinical features of autoimmune gastritis

BACKGROUND Autoimmune gastritis (AIG) is frequently associated with one or more comorbid conditions, among which type I gastric neuroendocrine tumors (gNETs) warrant significant clinical concern. However, risk factors for the development of gNETs in AIG populations remain poorly defined. AIM To characterize the clinical and endoscopic profiles of AIG and identify potential risk factors for gNETs development. METHODS In this single-center cross-sectional study carried out at a tertiary hospital, 303 patients with AIG over an 8-year period were retrospectively categorized into gNETs (n = 116) and non-gNETs (n = 187) groups. Endoscopic and clinical parameters were analyzed. Endoscopic features were systematically reevaluated according to the 2023 Japanese diagnostic criteria for AIG. Feature selection was performed using the Boruta algorithm, and the model discriminative ability was evaluated via receiver operating characteristic curve analysis. RESULTS Among the 303 patients with AIG, 116 had gNETs and 187 did not. Compared with the non-gNETs group, patients in the gNETs group were younger (54.3 years vs 60.6 years, P < 0.001), had higher rate of vitamin B12 deficiency (77.2% vs 55.8%, P < 0.001), lower pepsinogen I (4.3 ng/mL vs 7.4 ng/mL, P < 0.001) and pepsinogen I/II ratios (0.7 vs. 1.1, P < 0.001), and lower prior Helicobacter pylori infection rate (3.4% vs 21.4%, P < 0.001). Endoscopically, the gNETs group showed a lower incidence of oxyntic mucosal remnants, hyperplastic polyps, and patchy antral redness. The predictive model incorporating age, prior Helicobacter pylori infection, vitamin B12 level, gastric hyperplastic polyps, and patchy antral redness showed an area under the curve of 0.830. CONCLUSION Patients with AIG or gNETs exhibit specific clinical and endoscopic features. The predictive model demonstrated favorable discriminative ability and may facilitate risk stratification of gNETs in patients with AIG.