Construction and Validation of a T Cell Exhaustion–Related Prognostic Signature in Cholangiocarcinoma

Objective: T cell exhaustion (TEX) is a critical determinant of immune resistance. This study was performed to investigate the key genes linked to TEX in cholangiocarcinoma (CCA) and construct a TEX-associated gene signature to forecast the prognosis of patients with CCA.

Methods: Based on the expression data acquired from the E-MTAB-6389 dataset, the TEX-related modules and module genes were identified using weighted coexpression network analysis (WGCNA). Subsequently, a TEX-related prognostic signature was built by using the univariate and least absolute shrinkage and selection operator (LASSO) Cox regression analysis. The immune cell infiltration in each CCA sample was evaluated using the single-sample gene set enrichment analysis (ssGSEA) package, followed by single-cell RNA sequencing (scRNA-seq) analysis. Furthermore, the expression of TEX-related genes in the gene signature was experimentally validated in CCA cells by quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) and western blot analysis.

Results: A total of 15 TEX-associated modules and 23 module genes were identified. Then, a four-gene signature related to TEX was established, containing Palladin, Cytoskeletal Associated Protein (PALLD), Member RAS Oncogene Family (RAB31), ADAM Metallopeptidase With Thrombospondin Type 1 Motif 2 (ADAMTS2), and WISP1, which could predict prognosis of patients with CCA. Moreover, neutrophils, endothelial cells, B cells, and T cells exhibited significant infiltration in CCA samples, and these four TEX-related genes were both significantly positively correlated with T cells, endothelial cells, and B cells while negatively correlated with neutrophils. Moreover, a total of 13 cell types were annotated after scRNA-seq analysis. Notably, RAB31 was mainly highly expressed in monocytes, macrophages, DC2 (Dendritic Cells 2), and DC3 (Dendritic Cells 3), and PALLD, ADAMTS2, and WISP1 were mainly overexpressed in fibroblasts. Furthermore, experimental validation revealed that the expression levels of PALLD, RAB31, ADAMTS2, and WISP1 were consistent with the trend results of bioinformatics analysis.

Conclusion: A prognostic signature was developed by four TEX-related genes, including PALLD, RAB31, ADAMTS2, and WISP1, which might be a powerful predictor for the prognosis of patients with CCA. These TEX-related genes were related to the infiltration of neutrophils, endothelial cells, B cells, and T cells in CCA.