Aims
Gastric low-grade intraepithelial neoplasia (LGIN) carries a risk of progression to gastric cancer (GC) in a short time. Identifying predictors of pathological upgrade is critical for precision treatment.
Methods
This retrospective study analyzed 260 LGIN patients undergoing ESD within 90 days after biopsy. Clinical data, lesion characteristics (size, location, morphology, ulceration), and immunohistochemical markers (Ki67, c-MYC, MUC5AC,MUC6, CDX-2, P53, H.pylori, MLH1) were assessed. Bioinformatic analysis identified DEGs in GIN. A nomogram was developed using multivariable logistic regression and validated via ROC, calibration, and decision curve analysis.
Results
In our study, Ki67 positivity>30% (OR=2.47), depressed-type lesions (OR=2.21), and larger lesion size (OR=1.33) are the independent risk factors for pathological upgrade. Bioinformatic analysis furtherly confirmed Ki67 is one of the key gene in high-grade lesions. The ROC, calibration, and decision curve analysis showed the nomogram has good discrimination (AUC=0.73) and calibration for predicting pathological upgrade of LGIN patients.
Conclusions
The nomogram incorporating Ki67, lesion morphology and size effectively predicts pathological upgrade of LGIN patients in a short period of time,which helps precision treatment decision-making for gastric LGIN.