Colorectal cancer (CRC) is one of the most common malignant tumors worldwide. Patients with different immunophenotypes of CRC could achieve different effect of immunotherapy and yield different prognosis. With the advancement of bioinformatics, multi-omics analysis of the variations at both genomics and epigenomics levels helps a lot to provide a molecular basis for immunophenotype. Gene expression and clinical data of CRC patients were obtained from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO). We calculated Spearman correlation of CD274 (programmed cell death ligand-1, PD-L1) and IFNG (interferon gamma, IFN-γ) expressions with immune cell fraction, and screened different immune cell types with CD274 and IFNG by Lasso regression analysis. Multi-omics analysis was exploited to screen out candidate genes with differential in genetic and epigenetic landscapes between two CRC subtypes with the greatest difference in immune infiltration. Finally, a risk scoring model was established and the role of candidate genes in prognosis and oncoimmunology was evaluated at the pan-cancer level. Two CRC types (cluster A and cluster B) including five subtypes (subclusters A1, A2, B1, B2A, and B2B) were identified by unsupervised clustering analysis. Somatic mutations, CNVs, and DNA methylation differed between subcluster A2 and B2, and analysis of DEGs correlated with CRC immune phenotypes identified FUT9 and MS4A3 as key genes related to CRC immune-phenotypes and prognosis. Furthermore, FUT9 was validated to act as a key gene related to CRC immune escape in vitro. The present study established a risk model for CRC immunophenotyping and prognosis, and highlighted the significance of FUT9 and MS4A3 in oncoimmunology of CRC.
Zhu et al. (Mon,) studied this question.