Translational Oncology (Mar 2025)
Machine learning identification of a novel vasculogenic mimicry-related signature and FOXM1’s role in promoting vasculogenic mimicry in clear cell renal cell carcinoma
Abstract
Background: Clear Cell Renal Cell Carcinoma (ccRCC), the predominant subtype of renal cell carcinoma (RCC), ranks among the most common malignancies worldwide. Vasculogenic mimicry (VM) plays a pivotal role in tumor progression, being closely linked with heightened chemoresistance and adverse prognosis in cancer patients. Nonetheless, the broader impact of vasculogenic mimicry-related genes (VRGs) on ccRCC patient prognosis, tumor microenvironment characteristics, and treatment response remains incompletely understood. Methods: Consensus clustering identified VRG-associated subtypes. We developed a machine learning framework integrating 12 algorithms to establish a consistent VM-related signature (VRG_score). The predictive value of VRG_score for ccRCC prognosis and treatment response was assessed. FOXM1′s clinical relevance was explored using the UCLCAN database. FOXM1 expression in tumor and adjacent tissues was assessed using Western Blotting, IHC, RNA-seq, and Chip-qPCR methods, and its regulatory mechanism was confirmed. Results: We examined VRG mutation and expression patterns in ccRCC at the gene level, identifying two distinct molecular clusters. A consensus VRG_score was formulated using a machine learning computational framework and Cox regression, displaying strong predictive power for prognosis and clinical translation. Additionally, FOXM1 was found to be upregulated in ccRCC, correlating with clinical pathological features and positively regulating PYCR1, thereby activating the PI3K/AKT/mTOR signaling pathway and promoting VM formation. Conclusion: This study constructed a VM-related signature and revealed that FOXM1 promotes VM formation in renal cell carcinoma through the PYCR1-PI3K/AKT/mTOR signaling axis, serving as a prognostic indicator and potential therapeutic target.
