A multicenter retrospective study: risk factors and nomogram construction for pathological escalation of gastric...

Low-grade intraepithelial neoplasia (LGIN) is recognized as a precancerous lesion of gastric carcinoma. However, discrepancies have been observed between initial biopsy findings and post-Endoscopic Submucosal Dissection (ESD) histopathological results. This study aims to develop and clinically validate a multifactorial nomogram for predicting the risk of pathological upgrading in gastric LGIN.

We retrospectively analyzed clinical data from patients who underwent gastric ESD at three hospitals from June 2018 to July 2024. Sixty-eight patients from Zhejiang Provincial Hospital of Chinese Medicine and fifteen patients from Zhejiang Provincial People's Hospital constituted the training cohort; sixty-eight patients from Shaoxing People's Hospital formed the validation cohort. We performed detailed comparisons of 13 baseline patient characteristics and conducted logistic regression analyses. Subsequently, a nomogram was developed and evaluated.

The pathological upgrade rate was 54.1%. Independent risk factors included male sex, mixed IIc lesion morphology, clear demarcation lines, lesion size ≥ 15 mm, and Helicobacter pylori antibody positivity. The area under the receiver operating characteristic curve (AUC) of the training cohort was 0.891, with a sensitivity of 0.882 and specificity of 0.592, indicating strong discriminative ability. The calibration curve closely aligned with the diagonal, suggesting excellent predictive accuracy. DCA indicated significant clinical benefit. The predictive performance of the constructed nomogram in the validation cohort is moderate, with an AUC of approximately 0.67.

This multicenter retrospective study identified several independent risk factors for pathological escalation of LGIN to high-grade intraepithelial neoplasia (HGIN) or gastric cancer, including mixed lesion morphology, clear demarcation lines, and lesion size ≥ 15 mm. Further prospective studies are warranted to validate the nomogram's performance across diverse clinical settings.