Vasculogenic mimicry as a potential indicator of drug resistance and prognosis in renal cell carcinoma.

Abstract

4551 Background: Patients with advanced metastatic renal cell carcinoma (RCC) often develop resistance to tyrosine kinase inhibitors (TKIs). Vasculogenic mimicry (VM) refers to the formation of tubular structures by tumor cells mimicking endothelial cells. VM formation is independent of VEGF and endothelial cells, making it inherently resistant to TKIs. Furthermore, hypoxic conditions induced by TKI treatment can promote VM formation, creating a vicious cycle. This study investigates the molecular mechanisms of VM formation and its prognostic significance in RCC. Methods: VM incidence in RCC was assessed using PAS/CD31 staining on tissue microarrays. Single-cell sequencing data were used to identify tumor cells undergoing VM differentiation. Cluster analysis was conducted to characterize these cells, and their prognostic value was validated using TCGA data. Pseudotime trajectory analysis and SCENIC algorithms were used to infer their differentiation pathways and identify transcription factors (TFs) regulating VM formation. Tube formation assays were performed for validation. Results: PAS/CD31 double staining revealed a VM incidence of 15.87% (10/63) among RCC patients. Notably, VM formation was more frequent in recurrent and TKI-resistant patients, suggesting that VM may serve as a mechanism for TKI resistance. Single-cell data from 11 patients with stages T1a-T3 RCC were analyzed, identifying VM-differentiating tumor cells, termed RCC-VM. GSVA revealed that RCC-VM cells were highly enriched in angiogenesis and EMT-related pathways. GO and KEGG analyses also showed enrichment in angiogenesis pathways. Trajectory analysis of tumor cell subpopulations placed RCC-VM at the terminal differentiation state, suggesting it represents a uniquely differentiated tumor cell type. Using SCENIC, we identified FOSL2 as a key TF regulating RCC-VM differentiation. Knockdown of FOSL2 significantly impaired tube formation in 786-O cells in vitro. Additionally, we identified RCC-VM-specific signature genes (VMDEG) and used Lasso-Cox regression to select four key risk factors (PIM1, MT1G, MT-ND4, DDIT3) to construct a survival risk model. Kaplan-Meier survival analysis demonstrated that patients in the high-risk group had significantly shorter survival times compared to the low-risk group (p = 0.0013). The time-dependent ROC curve showed that the model had robust predictive ability, providing potential guidance for personalized treatment of RCC patients. Conclusions: Our study highlights VM as a critical mechanism of TKI resistance in RCC, regulated by the transcription factor FOSL2. VMDEG serves as a valuable prognostic marker for RCC patients. Incorporating VM into staging systems such as pT staging and Fuhrman grading may improve risk stratification and treatment planning for RCC patients.