A 16-gene expression signature to distinguish stage I from stage II lung squamous carcinoma

The present study aimed to perform screening of a gene signature for the discrimination and prognostic prediction of stage I and II lung squamous carcinoma. A microarray meta-analysis was performed to identify differentially expressed genes (DEGs) between stage I and II lung squamous carcinoma samples in seven microarray datasets collected from the Gene Expression Omnibus database via the MetaQC and MetaDE package in R. The important DEGs were selected according to the betweenness centrality value of the protein-protein interaction (PPI) network. Support vector machine (SVM) analysis was performed to screen the feature genes for discrimination and prognosis. One independent dataset downloaded from The Cancer Genome Atlas was used to validate the reliability. Pathway enrichment analysis was also performed for the feature genes. A total of 924 DEGs were identified to construct a PPI network consisting of 392 nodes and 686 edges. The top 100 of the 392 nodes were selected as crucial genes to construct an SVM classifier, and a 16-gene signature (caveolin 1, eukaryotic translation elongation factor 1γ, casein kinase 2α1, tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation η, tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation θ, pleiotrophin, insulin receptor, insulin receptor substrate 1, 3-phosphoinositide-dependent protein kinase-1, specificity protein 1, COP9 signalosome subunit 6, N-myc downstream regulated gene 1, retinoid X receptor α, heat shock protein 90α A1, karyopherin subunit β1 and erythrocyte membrane protein band 4.1) with high discrimination accuracy was identified. This 16-gene signature had significant prognostic value, and patients with stage II lung squamous carcinoma exhibited shorter survival rates, compared with those with stage I disease. Seven DEGs of the 16-gene signature were significantly involved in the phosphoinositide 3-kinase-Akt signaling pathway. The 16-gene signature identified in the present study may be useful for stratifying the patients with stage I or II lung squamous carcinoma and predicting prognosis.