Identification of Novel Lung Adenocarcinoma Subtypes and Prognostic Gene Sets Based on GSVA Analysis

Background Lung cancer is one of the most common malignant tumors of the respiratory system in the world. According to the traditional histological classification, it can be divided into many subtypes. In recent years, the incidence rate of lung adenocarcinoma (LUAD) has been rising rapidly. In this study, we identified the biomarkers related to the prognosis of LUAD through the protein-protein interaction (PPI) network analysis, gene set variation analysis (GSVA) and the "CancerSubtypes" software package in R, with a view to having a positive impact on the future treatment and new drug development. Method We obtained the relevant data needed for the study from The Cancer Genome Atlas (TCGA) database and the GEO database. Through GSVA analysis, the gene expression matrix was further transformed into the gene set expression matrix for subsequent research. Then, we applied the package "CancerSubtypes" in R to transform the samples into different subtypes, and established a LUAD-related prognosis model based on the differential expression gene sets (DEGSs) among the subtypes. Finally, we performed functional and pathway enrichment analysis together with PPI network analysis of genes from prognosis related DEGSs. Results A total of 600 LUAD samples were obtained from TCGA database, including 541 tumor samples and 59 normal samples. We screened 507 tumor samples for further classification, including 166 cases of subtype 1, 138 cases of subtype 2, and 203 cases of subtype 3. Subsequently, we identified 63 DEGSs and constructed the prognostic characteristics of LUAD with 4 of them (“T_GSE24634_TREG_VS_TCONV_POST_DAY5_IL4_CONVERSION_UP” “T_GSE25088_WT_VS_STAT6_KO_MACROPHAGE_DN” “T_GSE45365_HEALTHY_VS_MCMV_INFECTION_CD11B_DC_DN” and “T_HALLMARK_MYC_TARGETS_V2”). Finally, we established the corresponding PPI network with 6 subnets, and identified 15 core proteins including CCNB2, KIF2C, TPX2, PES1, BRIX1, NIP7, PSMB4, PSMD12, PSMC3, MPHOSPH10, WDR43, POLA1, MCM4, PAICS and GART. Conclusions In this study, we identified four gene sets related to the prognosis of LUAD and obtained 15 core proteins. This study could provide relevant theoretical basis and guidance for the update of treatment methods and the development of new drugs, related to LUAD and other cancers.