A Weighted Gene Co-Expression Network Analysis-Derived Prognostic Model for Predicting Prognosis and Immune Infiltration in Gastric Cancer

• Published in: Frontiers in oncology Vol. 11; p. 554779
• Main Authors: Chen, Qingchuan, Tan, Yuen, Zhang, Chao, Zhang, Zhe, Pan, Siwei, An, Wen, Xu, Huimian
• Format: Journal Article
• Language: English
• Published: Switzerland Frontiers Media S.A 25.02.2021
• Subjects: Cox proportional hazards regression model; gastric cancer; immune infiltration; Oncology; prognostic model; weighted gene co-expression network analysis; Cox proportional hazards regression model; weighted gene co-expression network analysis; gastric cancer; immune infiltration; prognostic model
• ISSN: 2234-943X; 2234-943X
• DOI: 10.3389/fonc.2021.554779

Gastric cancer (GC) is a major public health problem worldwide. In recent decades, the treatment of gastric cancer has improved greatly, but basic research and clinical application of gastric cancer remain challenges due to the high heterogeneity. Here, we provide new insights for identifying prognostic models of GC. We obtained the gene expression profiles of GSE62254 containing 300 samples for training. GSE15459 and TCGA-STAD for validation, which contain 200 and 375 samples, respectively. Weighted gene co-expression network analysis (WGCNA) was used to identify gene modules. We performed Lasso regression and Cox regression analyses to identify the most significant five genes to develop a novel prognostic model. And we selected two representative genes within the model for immunohistochemistry staining with 105 GC specimens from our hospital to verify the prediction efficiency. Moreover, we estimated the correlation coefficient between our model and immune infiltration using the CIBERSORT algorithm. The data from GSE15459 and TCGA cohort validated the robustness and predictive accuracy of this prognostic model. Of the 12 gene modules identified, 1,198 green-yellow module genes were selected for further analysis. Multivariate Cox analysis was performed on genes from univariate Cox regression and Lasso regression analysis using the Cox proportional hazards regression model. Finally, we constructed a five gene prognostic model: Risk Score = [(-0.7547) * Expression ( )] + [(-0.8272) * Expression ( )] + [1.09 * Expression ( )] + [0.5174 * Expression ( )] + [1.66 * Expression ( )]. The prognosis of samples in the high-risk group was significantly poorer than that of samples in the low-risk group (p = 6.503e-11). The risk model was also regarded as an independent predictor of prognosis (HR, 1.678, p < 0.001). The observed correlation with immune cells suggested that this risk model could potentially predict immune infiltration. This study identified a potential risk model for prognosis and immune infiltration prediction in GC using WGCNA and Cox regression analysis.