Identification of differentially expressed genes at the single-cell level and prognosis prediction through bulk RNA sequencing data in breast cancer

Abstract

Background: The invention and development of single-cell technologies have contributed a lot to the understanding of tumor heterogeneity. The objective of this research was to investigate the differentially expressed genes (DEGs) between normal and tumor cells at the single-cell level and explore the clinical application of these genes with bulk RNA-sequencing data in breast cancer.Methods: We collected single-cell, bulk RNA sequencing (RNA-seq) and microarray data from two public databases. Through single-cell analysis of 23,909 mammary gland cells from seven healthy donors and 33,138 tumor cells from seven breast cancer patients, cell type-specific DEGs between normal and tumor cells were identified. With these genes and the bulk RNA-seq data, we developed a prognostic signature and validated the efficacy in two independent cohorts. We also explored the differences of immune infiltration and tumor mutational burden (TMB) between the different risk groups.Results: A total of 6,175 cell-type-specific DEGs were obtained through the single-cell analysis between normal and tumor cells in breast cancer, of which 1,768 genes intersected with the bulk RNA-seq data. An 18-gene signature was constructed to assess the outcomes in breast cancer patients. The efficacy of the signature was notably prominent in two independent cohorts. The low-risk group showed higher immune infiltration and lower TMB. Among the 18 genes in the signature, 16 were also differentially expressed in the bulk RNA-seq dataset.Conclusion: Cell-type-specific DEGs between normal and tumor cells were identified through single-cell transcriptome data. The signature constructed with these DEGs could stratify patients efficiently. The signature was also closely correlated with immune infiltration and TMB. Nearly all the genes in the signature were also differentially expressed at the bulk RNA-seq level.