Comprehensive characterization and functional implications of mesenchymal stem cell-related genes in pancancer: A new horizon for immunotherapy.

Abstract

e14577 Background: Accumulated evidence demonstrates that mesenchymal stem cells (MSCs) contribute to tissue regeneration, tumor progression, chemotherapy resistance, and mediate tumor immune escape by modulating the stroma and immunity of the tumor microenvironment. However, little is known about the role of MSCs in cancer patients receiving immunotherapy. Therefore, we investigated the comprehensive characterization of MSC-related genes and their functional implications in pan-cancer immunotherapy. Methods: We performed a pan-cancer analysis of 59 MSC-related genes, which was used to calculate an MSC score with single sample gene set enrichment analysis. Multivariate Cox regression analysis was performed to identify a potential independent prognostic marker for cancer. TIDE algorithm and neural network were utilized to assess the predictive accuracy of MSC-related genes for immunotherapy. Results: MSC-related gene expression showed significant differences inconsistently between normal and tumor samples across the 33 cancer types (p < 0.05). Cox regression analysis suggested that the MSC score is an independent prognostic marker for papillary renal cell carcinoma (TCGA; OS, RFS and DSS, all p < 0.05), mesothelioma (TCGA and GSE29354, OS, all p < 0.05), glioma (TCGA and GSE107850, OS, all p < 0.05), and stomach adenocarcinoma (TCGA, GSE84437, GSE15459, GSE13861 and GSE62254, OS; GSE13861 and GSE62254, RFS; all p < 0.05). The abundance of fibroblasts was also more representative of the MSC score compared to the stromal score. Samples with a high MSC score were predicted to have low responsiveness to immunotherapy, which was validated across seven immune checkpoint therapy datasets (IMvigor210, 27.52% (low) vs 18.12% (high); GSE78220, 64.28% (low) vs 42.86% (high); GSE135222, 30.77% (low) vs 28.57% (high); GSE165252, 36.11% (low) vs 28.57% (high); GSE79671; 38.89% (low) vs 33.33% (high); PRJEB25780, 45.45% (low) vs 8.70% (high); GSE176307, 22.22% (low) vs 13.64% (high)). TIDE algorithm and Neural network demonstrated a good predictive accuracy of MSC-related genes for immunotherapy (CESC, COAD, DLBC, GBM, KIRC, LAML, LIHC, LUAD, LUSC, MESO, OV, PCPG, PRAD, SARC, SKCM, THCA, UCEC, and UCS; all ROC ≥0.9). Conclusions: We characterized the MSC-related genes across multiple cancer types and highlighted their potential as a predictive biomarker for immunotherapy response and prognosis. A prospective clinical trial is warranted to confirm these findings.