Bioinformatics-based identification of differentially expressed genes in endometrial carcinoma: implications for early diagnosis and prognostic stratification.

Introduction: This study aims to identify differentially expressed genes (DEGs) in endometrial carcinoma (EC) through bioinformatics analysis and investigate their roles in early diagnosis and prognosis. Methods: EC-related gene datasets were retrieved from the NCBI and analyzed using R packages to screen for DEGs. Primers were designed for selected DEGs, and their expression levels were validated via qPCR. Logistic regression, survival analysis, Cox proportional hazards models, and random forest models were employed to evaluate associations between DEGs and clinical outcomes. Results: Bioinformatics analysis identified significantly upregulated genes (Erb-B2, PIK3CA, CCND1, VEGF, KIT) and downregulated genes (PTEN, E-cadherin, p53). Logistic regression revealed Erb-B2 as a protective factor against poor prognosis, whereas E-cadherin and P53 were risk genes. Clinical markers CA125, CA199, and IL-9 also emerged as prognostic risk factors. Survival analysis demonstrated significant divergence between good and poor prognosis groups (P < 0.05), with HR < 1 for Erb-B2 and p53 (protective effects) and HR > 1 for E-cadherin, CA125, CA199, and IL-9 (risk effects). The random forest model highlighted CA199 as a pivotal prognostic biomarker, while decision tree analysis enabled effective patient stratification based on CA125 and CA199 thresholds. Conclusion: The identified DEGs and clinical indicators hold significant potential for improving early diagnosis and prognostic evaluation in EC. These findings provide novel biomarkers and theoretical foundations for precision medicine, guiding risk stratification and personalized therapeutic strategies.