Identification of COL3A1, PLAU, and SPP1 as Key Biomarkers for Early Detection of Esophageal Cancer

Esophageal cancer (EC) is a highly lethal malignancy often diagnosed at advanced stages due to the lack of effective early diagnostic markers. This study aimed to identify molecular markers and construct a diagnostic model for early-stage esophageal cancer using bioinformatics approaches. Using bioinformatics, we screened three GEO datasets, locating 506 differentially expressed genes crucial to cancer progression. Our results connect ECM-receptor interaction and cytoskeleton reorganization pathways to EC. Two core gene modules came up during the protein-protein interaction analysis. From the 22 hub genes singled out, COL3A1, PLAU, and SPP1 significantly impacted patient survival, showing considerable overexpression in cancer subjects. These genes' expression patterns changed across cancer stages. The main novelty of our study lies in integrating these three well-known ECM-associated genes into a machine learning-based diagnostic model with an AUC of 0.98, rather than focusing on individual genes. This combined model demonstrates high diagnostic accuracy, suggesting that COL3A1, PLAU, and SPP1 may serve as effective early-stage EC biomarkers. The diagnostic model based on these genes shows high accuracy, making it a promising tool for early-stage cancer screening.