Systematic Screening via WGCNA, Machine Learning, and Validation Identifies C3 and CFD as Prognostically Significant Regulators in GIST Progression

Gastrointestinal stromal tumor (GIST), the most common gastrointestinal mesenchymal neoplasm, remains poorly understood at the molecular level, limiting precise diagnosis and targeted therapy. This study aimed to systematically identify key GIST-associated genes through multiomic integration and experimental validation. We analyzed three GIST transcriptomic datasets from GEO, corrected batch effects via surrogate variable analysis (SVA), and identified 61 differentially expressed genes (DEGs) using limma. Weighted gene co-expression network analysis (WGCNA) highlighted progression-related modules, which were refined using random forests and LASSO regression to prioritize C3 and complement factor D (CFD), both of which showed robust diagnostic performance (AUC: 0.928 for C3; 0.955 for CFD). Experimental validation confirmed C3/CFD downregulation in GIST tissues, correlating with advanced stage and poor survival. Functional assays demonstrated their tumor-suppressive roles, inhibiting GIST cell proliferation, colony formation, and migration. CIBERSORT analysis linked C3/CFD to altered immune infiltration, while ssGSEA/GSEA implicated their involvement in lipid metabolism and oxidative phosphorylation. These findings establish C3 and CFD as critical tumor-suppressive biomarkers that modulate the immune response and reprogram metabolism, offering new avenues for GIST diagnosis and therapy.