Several gene expression signatures have been reported to predict patient survival of gastric cancer after surgical resection. However, the prognostic gene lists have overlapped poorly until now. This study conducted an analysis to characterize gene expression profile and developed a survival prediction model.

The gene expression profile was evaluated in fresh frozen tumor tissue obtained from 48 patients with primary gastric cancer. We measured 84 representative genes involved in transformation and tumorigenesis using quantitative reverse transcription polymerase chain reaction (qRT-PCR) and related the results to overall survival.

In a univariate analysis, 84 genes were ranked on their ability to predict survival, of which nine genes were the strongest predictor (P<0.05). They were PLAU, MAP2K1, THBS1, TWIST1, ITGB5, NME4, ANGPT2, platelet-derived growth factor B (PDGFB), ITGB1. Then, we did a multivariate analysis to further select four genes (ITGB1, PDGFB, THBS1, TWIST1) from the above nine genes for the construction of biomathematics model, which was independent of age, gender, TNM stage and other variables. This model could correctly clarify gastric patients into the high-risk group, median-risk group and low-risk group, as well as predict their survival.

Measurement of the expression of four genes is probable to predict surgery-related survival. This model may be test further for its potential to improve the selection of the resected gastric cancer patients in adjuvant chemotherapy.