Identification and Validation of Signature Genes in Invasiveness-Associated Modules of Nonfunctioning Pituitary...

Background: Invasive non-functional pituitary adenomas (NFPAs) are associated with high recurrence and unfavorable clinical outcomes, yet their underlying molecular mechanisms remain incompletely understood. This study aimed to identify robust biomarkers of invasiveness by integrating transcriptional networks, machine learning, and epigenetic regulation. Methods: RNA sequencing was performed on 32 NFPA samples (15 invasive, 17 non-invasive). Weighted gene co-expression network analysis (WGCNA) was used to identify invasiveness-associated modules, which were validated in public datasets (GSE169498, GSE51618). Candidate genes were prioritized using machine learning, and their epigenetic regulation was studied using DNA methylation datasets (GSE207937, GSE115783). Results: We identified a five-gene signature associated with invasiveness (KIFC3, PNMA3, ARHGAP18, LRRC10B, and KCNC4). All five genes were consistently downregulated in invasive NFPAs (all p < 0.01) and were enriched in oxidative phosphorylation and neuroactive ligand-receptor interaction pathways. A machine learning validation approach (Random Forest followed by forward stepwise logistic regression) showed strong discriminative performance for this signature (mean AUC = 0.919). DNA methylation analyses indicated no robust differences at the genome-wide level or across promoter regions of the core genes; nevertheless, several locus-specific CpG sites (e.g., near KIFC3) showed suggestive methylation changes. Conclusions: Using an integrative multi-omics framework, we identified a novel five-gene signature associated with NFPA invasiveness. The coordinated downregulation of these genes may reflect alterations in cellular energy metabolism and microenvironmental signaling. Although the signature demonstrated promising diagnostic potential, its transcriptional repression is unlikely to be primarily explained by DNA methylation. These findings provide candidate markers and mechanistic hypotheses for understanding invasive NFPA and developing risk-stratification tools.