Integrating single-cell and bulk RNA sequencing to predict prognosis and immunotherapy response in prostate cancer

Abstract

Abstract Prostate cancer (PCa) is one of the leading causes of death for men worldwide. Cancer-associated fibroblasts (CAFs) are considered to be closely connected to tumour growth, invasion, and metastasis. We explored the role and characteristics of CAFs in PCa through bioinformatics analysis and built a CAFs-based risk model to predict prognostic treatment and treatment response in PCa patients. First, we downloaded the signal-cell RNA sequencing (scRNA-seq) data of PCa from the GEO database. We extracted bulk RNA-seq data and microarray data of PCa from the TCGA and GEO databases respectively, and adopted "ComBat" to remove batch effects. Then, we created a Seurat object for the scRNA-seq data using the package "Seurat" of R and identified CAF clusters based on the CAF-related genes (CAFRGs). Based on CAFRGs, a prognostic model was constructed by univariate Cox, LASSO, and multivariate Cox analyses. And the model was validated internally and externally by Kaplan-Meier analysis, respectively. We further performed GO and KEGG analysis of differentially expressed genes between risk groups. Besides, we investigated differences in somatic mutations between different risk groups. We explored differences in the immune microenvironment landscape and immune checkpoint gene expression levels in the different groups. Final, we predicted the response to immunotherapy and the sensitivity of antitumour drugs between the different groups.We screened 4 CAF clusters and identified 463 CAFRGs in PCa scRNA-seq. We constructed a model containing 10 prognostic CAFRGs by univariate Cox, LASSO, and multivariate Cox analysis. Somatic mutation analysis revealed that TTN and TP53 were significantly more mutated in the high-risk group than in the low-risk group, suggesting that the high-risk group may have a poor prognosis. Finally, we screened 31 chemotherapeutic drugs and targeted therapeutic drugs for PCa.In conclusion, we identified four clusters based on CAFs and constructed a new CAFs-based prognostic signature that could predict PCa patient prognosis and response to immunotherapy and might suggest meaningful clinical options for the treatment of PCa.