Abstract
Melanoma is a malignancy of aggressive behaviour and related with poor prognosis. Immunotherapy for metastatic melanoma shows great promise. However, the development of rapid resistance remains a largely insurmountable challenge. In this study, we aim to identify macrophage marker genes and construct an immune risk model, hoping to provide clinical significance in prognosis and immunotherapy response prediction. A total of 16,291 cell samples from 48 melanoma tissues in GSE120575 were enrolled. Firstly, 1,662 macrophages were identified with marker gene annotation. Subsequently, we acquired 6 macrophage subtypes in TCGA-SKCM dataset based on the expression characteristics of 724 gene that differentially expressed between macrophages and other immune cells. Since macrophage subtypes A and B had most distinguishing differences of immune- and tumour-related pathway enrichment, prognosis and immune microenvironment features that 102 immunity- and prognosis-related genes were further identified from them. Ultimately, we developed a risk signature of 21 immune genes through multivariate Cox regression, dividing patients into high- and low-risk groups. Explicitly, low-risk patients had a longer survival than high-risk patients, and similar results were also found in GSE65904 and GSE59455. Moreover, low-risk patients were found to have more favourable anti-tumour immune environment including more immune, stromal components, less tumour components, and higher infiltration of immune effector cells like activated memory CD4 + T cells, CD8 + T cells, M1 macrophages, plasma cells. Also, low-risk groups with higher gene expression of PD-1, PD-L1, CTLA4 were associated with better responses to immune checkpoint inhibitors (ICIs). Additionally, the differential expression of gene and protein levers of the 21 genes in normal and melanoma tissues demonstrated their importance in cellular physiology and disease mechanisms. To conclude, we developed an immune risk signature which could distinguish patients with different clinical prognoses and contribute to predicting the response to immunotherapy in melanoma.