Bioinformatics Analysis of Microarray Datasets to Identify Prognostic Factors in Lung Adenocarcinoma

Most patients with lung adenocarcinoma (LUAD) present high recurrence rate and poor prognosis after therapy. Therefore, the purpose of this study was to identify prognostic factors involved in LUAD. Five microarray datasets (including GSE75037, GSE63459, GSE43458, GSE32863, and GSE10072) were downloaded. After data preprocessing and quality control, meta-analysis was performed to screen differentially expressed genes (DEGs) using the MetaDE.ES method in MetaDE package. Subsequently, network construction and module identification were conducted by the Weighted Gene Co-expression Network Analysis method. Moreover, survival-associated genes were identified using the univariate and multivariate Cox regression method in survival package. The risk score model was constructed by prognosis associated genes, followed by the Kaplan-Meier survival analysis. Oncomine expressions analysis of several prognosis associated genes was conducted. The expression levels of key genes were detected using quantitative real-time PCR experiments. A total of 1434 DEGs between LUAD and normal samples were identified. Nine disease-associated modules were identified, in which M8 module was most correlated with LAUD phenotype. A total of 89 indicators (including T stage, M stage, and ADIPOR2) were significantly associated with LAUD prognosis, while only T stage and 9 DEGs (e.g., ARHGEF3, GTSE1, RBM15 and CD52) were retained as the potential prognostic factors following multivariate COX regression analysis. The upregulated adiponectin receptor 2 (ADIPOR2), rho guanine nucleotide exchange factor 3 (ARHGEF3), and CD52 molecule (CD52), and downregulated GTSE1 were validated in LAUD samples of Oncomine database. Importantly, ADIPOR2 and ARHGEF3 were confirmed to be down-regulated in LUAD tissues. ADIPOR2, ARHGEF3, G2 and S-phase expressed 1 (GTSE1) and CD52 might be promising prognostic factors in LUAD.