Systematic analysis of multigene predictors in gastric cancer exploiting gene expression signature.

Gastric cancer (GC) is the second most common cause of cancer death worldwide but could be more curable if diagnosed at an earlier stage. At present, the capability to predict the efficaciousness of molecular diagnosis for GC for each patient remains elusive. The purpose of this study was to identify tumor biomarkers through systems analysis of multigene predictors exploiting the available data resource. In this study, we investigated the top 10% overexpressed genes in GC from five data sets of the Oncomine platform, with 265 GC samples versus 174 normal gastric mucosa samples. Sixteen candidate genes were identified as predictors of GC, of which 14 genes were verified through the comparison of expression levels in specimens from normal (chronic gastritis, 21 samples) and GC groups (38 samples). In addition, unique molecular portraits of diffuse adenocarcinoma (DA), intestinal adenocarcinoma (IA), and mixed adenocarcinoma (MA) were studied through Gene Ontology and Kyoto Encyclopedia of Genes and Genomes pathway analysis, where DA showed higher extracellular matrix alteration while IA and MA showed higher cell-cycle alteration than other types. We also found that the elevated expressions of genes during GC progression were independent of gene mutations, and high core-binding factor subunit beta expression is correlated with a high overall survival rate in GC patients. Our research may provide an efficient clinical diagnosis of GC at an early stage with high accuracy and thus help improve the overall survival rate through early therapeutic interventions.