Intercellular Communication-Related Molecular Subtypes and a Gene Signature Identified by the Single-Cell RNA Sequencing Combined with a Transcriptomic Analysis

BACKGROUND: The tumor microenvironment (TME) of lung adenocarcinoma (LUAD) comprise various cell types that communicate with each other through ligand-receptor interactions. This study focused on the identification of cell types in LUAD by single-cell RNA sequencing (scRNA-seq) data and screening of intercellular communication-related genes. METHODS: The Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo) provided the RNA-seq data of LUAD patients in the GSE149655, GSE31210, and GSE72094 datasets. Quality control of the scRNA-seq data in GSE149655 was performed by the Seurat package (http://seurat.r-forge.r-project.org) for identifying highly variable genes for principal component analysis (PCA) and cell clustering. The CellPhoneDB (http://www.cellphonedb.org) was used for filtering intercellular communication-related ligand-receptor pairs. According to ligand and receptor expressions, LUAD samples were clustered using ConsensusClusterPlus (https://www.bioconductor.org/packages/release/bioc/html/ConsensusClusterPlus). Additionally, the identification of prognosis-related ligand and receptor genes was conducted along with the development of a risk prediction model by the least absolute shrinkage and selection operator (LASSO) Cox regression analysis. RESULTS: This study identified twelve cell types in 8170 cells of LUAD tissues along with 219 ligand and receptor genes. LUAD was classified into three different molecular subtypes, among which cluster 3 (C3) had the longest overall survival (OS) time and cluster (C1) had the shortest OS time. In comparison with the other two molecular subtypes, it was observed that C1 had a higher rate of somatic mutations and lower levels of infiltrating immune cells and immune scores. Ten genes were screened from the total ligand and receptor genes to construct a risk model, which showed a strong prediction power in the prognosis of patients with LUAD. CONCLUSION: The results of this study revealed cell types specific to LUAD, which were classified into different molecular subtypes according to intercellular communication-related genes. A novel prognostic risk model was developed in this study, providing new insights into prognostic assessment models for LUAD.