Immune infiltration of colorectal cancer (CRC) is closely associated with clinical outcome. However, previous work has not accounted for the diversity of functionally distinct cell types that make up the immune response. In this study, based on a deconvolution algorithm (known as CIBERSORT) and clinical annotated expression profiles, we comprehensively analyzed the tumor-infiltrating immune cells present in CRC for the first time. The fraction of 22 immune cells subpopulations was evaluated to determine the associations between each cell type and survival and response to chemotherapy. As a result, profiles of immune infiltration vary significantly between paired cancer and paracancerous tissue and the variation could characterize the individual differences. Of the cell subpopulations investigated, tumors lacking M1 macrophages or with an increased number of M2 macrophages, eosinophils, and neutrophils were associated with the poor prognosis. Unsupervised clustering analysis using immune cell proportions revealed five subgroups of tumors, largely defined by the balance between macrophages M1, M2, and NK resting cells, with distinct survival patterns, and associated with well-established molecular subtype. Collectively, our data suggest that subtle differences in the cellular composition of the immune infiltrate in CRC appear to exist, and these differences are likely to be important determinants of both prognosis and response to treatment.
Keywords: clinicopathological features; colorectal cancer; genomic signature; nomogram.
© 2018 The Authors. Cancer Medicine published by John Wiley & Sons Ltd.