Tumor-associated endothelial cell prognostic risk model and tumor immune environment modulation in liver cancer based on single-cell and bulk RNA sequencing: Experimental verification

Background: To build a prognostic and immunotherapeutic response prediction model for liver cancer based on marker genes of tumor-associated endothelial cell (TEC).Method: Single cell sequencing data from Gene Expression Omnibus (GEO) liver cancer patients were utilized to identify TEC subpopulations. Models were built from transcriptomic and clinical data of TCGA liver cancer patients. The GSE76427 and ICGC databases were used as independent validation sets. Time-dependent receiver operating characteristic (ROC) curves and Kaplan-Meier curves were used to verify the ability of the model to predict survival. XCELL, TIMER, QUANTISEQ, CIBERSORT, CIBERSORT-ABS, and ssGSEA were applied to evaluate tumor immune cell infiltration. The TIDE score was used to predict the effect of immunotherapy. Immune blockade checkpoint gene, tumor mutational load and GSVA enrichment analyses were further explored. The expression levels of candidate genes were measured and validated by real-time PCR between liver cancer tissues and adjacent nontumor liver tissues.Results: Eighty-seven genes were identified as marker genes for TECs. IGFBP3, RHOC, S100A16, FSCN1, and CLEC3B were included in the constructed prognostic model. Time-dependent ROC curve values were higher than 0.700 in both the model and validation groups. The low risk group exhibited high immune cell infiltration and function than the higher risk group. The TIDE score indicated that the low-risk group benefited more from immunotherapy than the high-risk group. The risk score and multiple immune blockade checkpoint genes and immune-related pathways were strongly correlated.Conclusion: Novel signatures of TEC marker genes showed a powerful ability to predict prognosis and immunotherapy response in patients with liver cancer.