Comprehensive Analysis of Ligand-receptor Interactions in Colon Adenocarcinoma to Identify of Tumor Microenvironment Oxidative Stress and Prognosis Model

Background: Single-cell technology enables a deep study on the mechanism of cancers. This work delineated the function of ligand-receptor interaction in colon adenocarcinoma (COAD), and developed a LR pairs-based prognostic model. Methods: For identifying important LR pairs, Single-cell RNA sequencing data of COAD was included. Unsupervised consensus clustering constructed molecular subtypes. LASSO established a prognostic model. Infiltration of 22 immune cells was evaluated by Cibersort. Enrichment score of oxidative stress related pathways was determined by SsGSEA in each patient. Results: Forty-seven LR pairs were closely associated with the prognosis of COAD. Three molecular subtypes were differentiated according to 47 LR pairs, which displayed differential clinical features and molecular features. There were significant differences in immune T cell lytic activity among different subtypes. In clust1 with poor prognosis, significantly enriched oncogenic pathways were found, especially epithelial-mesenchymal transition (EMT). Additionally, it has been found that clust3 had significantly higher immune infiltration. A prognostic model containing eight LR pairs (PDGFB-PDGFRA, FLT4-VEGFC, CSF1R-CSF1, DLL1-NOTCH4, PDGFB-LRP1, DLL1- NOTCH3, FLT4-PDGFC, and NRP2-PGF) was established, which could effectively divide samples into low-risk and high-risk groups. Significantly higher oxidative stress was found among high-risk patients. Conclusions: This study integrated expression data and single-cell data for demonstrating the effectiveness of LR pairs in establishing the prognostic model and constructing molecular subtypes. Prognostic LR pairs may contribute to tumorigenesis and progression in COAD. The prognostic model was the potential for predicting prognosis and guiding immunotherapy for COAD patients.