Integrative Transcriptomic Analysis and Mechanistic Modeling Reveal Critical Regulators of Thyroid Carcinoma

Abstract

Background: Thyroid carcinoma (THCA) is a prevalent endocrine cancer with rising incidence, but its molecular drivers remain incompletely understood. Methods: We conducted an integrated bioinformatics analysis of four GEO microarray datasets to identify genes differentially expressed in THCA. A protein–protein interaction (PPI) network constructed from common DEGs yielded seven hub genes (FN1, PTEN, TP53, ERBB3, MET, HRAS, PIK3CB) by centrality. Functional enrichment of these genes highlighted pathways such as extracellular matrix remodeling and MAPK signaling. Validation: We confirmed hub gene expression differences using UALCAN (TCGA data; all p<1×10-6) and examined survival and clinical stage associations via GEPIA2. Protein-level expression was corroborated by immunohistochemical data from the Human Protein Atlas. Independent validation using the GSE35570 cohort showed that FN1, MET, and HRAS were significantly overexpressed in THCA (p<0.0001) while PTEN and PIK3CB were significantly underexpressed (p<0.0001); TP53 and ERBB3 changes were not statistically significant. Receiver operating characteristic (ROC) curves demonstrated excellent diagnostic power for FN1 (AUC=0.9900) and MET (AUC=0.9551). Mutational profiling via cBioPortal highlighted frequent TP53, PTEN, and HRAS alterations. We also developed a mechanistic model linking these hub genes in THCA-related signaling. Conclusions: This multi-step analysis establishes FN1, MET, HRAS, PTEN, and PIK3CB as robust THCA biomarkers with potential diagnostic and therapeutic value.