China Oncology ›› 2021, Vol. 31 ›› Issue (8): 746-753.doi: 10.19401/j.cnki.1007-3639.2021.08.008

• Article • Previous Articles     Next Articles

Construction of a prediction model for metastasis in gastric cancer based on the weighted gene co-expression network analysis

GONG Chao 1 , CHEN Kui 1 , ZHANG Dekun 1 , XIE Jingfeng 2 , WU Fanghua 1 , HUANG Yudian 3 , XUE Yuqin 3 , WANG Liqun 1    

  1. 1. Department of General Surgery, The Affiliated Fuzhou First Hospital of Fujian Medical University, Fuzhou 350009, Fujian Province, China; 2. Department of Laboratory, The Affiliated Fuzhou First Hospital of Fujian Medical University, Fuzhou 350009, Fujian Province, China; 3. Department of Pathology, The Affiliated Fuzhou First Hospital of Fujian Medical University, Fuzhou 350009, Fujian Province, China
  • Online:2021-08-30 Published:2021-09-03
  • Contact: WANG Liqun E-mail: fzsypwk@139.com

Abstract: Background and purpose: This study aimed to screen out the bio-markers related to the occurrence and metastasis
of gastric cancer. The weighted gene co-expression network analysis (WGCNA) was performed to analyze the gene chips containing
metastatic and non-metastatic gastric cancer and adjacent tissues in Gene Expression Omnibus (GEO) data set. Methods: The gene
expression differences of 19 patients with gastric cancer were analyzed by WGCNA. In combination with clinical data, gene modules
highly relevant to clinical information were selected to construct the network. Results: Through WGCNA, we screened out the
Lightsteelblue module that was mostly related to gastric cancer metastasis. Then we further analyzed the genes in the module, and
screened out 4 genes: C5AR1, AP3M2, TYMP, ANXA2P1 as “real” hub genes. Through expression analysis and receiver operating
characteristic (ROC) curve analysis, the 4 genes identified were related to the occurrence and metastasis of gastric cancer. Meanwhile,
external ONCOMINE and Kaplan-Meier plot databases were used to verify that the above genes were highly expressed in gastric
cancer, and patients with high expression of the above genes had a worse prognosis. And the GSE14210 dataset was used to build the risk score model to predict the prognosis and progression of disease. These results suggested that the four genes we screened
were potential bio-markers for gastric cancer metastasis and treatment. Conclusion: In the present study, we screened and identified
4 genes related to the occurrence and metastasis of gastric cancer, which could provide evidence for the research on the occurrence,
metastasis and treatment of gastric cancer.

Key words: Gastric cancer, Gene Expression Omnibus, Metastasis, Weighted gene co-expression network analysis