Gastric cancer (GC) patients have high mortality due to late-stage diagnosis, which is closely associated with lymph node metastasis. Exploring the molecular mechanisms of lymphatic metastasis may inform the research into early diagnostics of GC. In the present study, we obtained RNA-Seq data from The Cancer Genome Altas and used Limma package to identify differentially expressed genes (DEGs) between lymphatic metastases and non-lymphatic metastases in GC tissues. Then, we used an elastic net-regularized COX proportional hazard model for gene selection from the DEGs and constructed a regression model composed of 28-gene signatures. Furthermore, we assessed the prognostic performance of the 28-gene signature by analyzing the receive operating characteristic curves. In addition, we selected the gene PELI2 amongst 28 genes and assessed the roles of this gene in GC cells. The good prognostic performance of the 28-gene signature was confirmed in the testing set, which was also validated by GSE66229 dataset. In addition, the biological experiments showed that PELI2 could promote the growth and metastasis of GC cells by regulating vascular endothelial growth factor C. Our study indicates that the identified 28-gene signature could be considered as a sensitive predictive tool for lymphatic metastasis in GC.
Keywords: COX; Elastic net; VEGF-C; gastric cancer; lymph node metastasis; prognosis.
© 2019 The Author(s).