Integrated Prognostic Model for Young Breast Cancer Patients: Insights from SEER, METABRIC, and TCGA Databases.

Abstract

This study aims to develop a prognostic model for breast cancer in young women (BCY) by integrating clinical and genomic data. We analyzed clinical data from SEER and identified key prognostic genes using METABRIC and TCGA datasets. Machine learning (LASSO and XGBoost) was used for gene selection, and a survival prediction model was built. The model was validated in independent datasets, and an interactive online nomogram was developed. The study included 139,994 breast cancer patients, confirming that age under 40 is an independent adverse factor for overall survival (OS). Using gene expression data from METABRIC and TCGA, PLA2G2A and C8orf76 were identified as critical prognostic biomarkers. These genes were incorporated into a predictive model that stratified patients into high- and low-risk groups with significant survival differences. The model achieved strong predictive performance, with AUC values of 0.862, 0.813, and 0.756 for 3-, 5-, and 8-year overall survival (OS), respectively. Additionally, a web-based tool (https://lyx00.shinyapps.io/BCY-Model-CG/) was developed to facilitate clinical implementation, allowing individualized survival predictions based on patient-specific clinical and genetic data. This integrated prognostic model, combining clinical and genomic data, improves survival prediction for breast cancer in young women (BCY) patients. It offers a more precise tool for risk assessment and personalized treatment planning.