Gene Expression Profile Changes In Early Gastric Cancer Progression And Their Diagnostic And Prognostic Value

Background&Objects:The molecular mechanism of early gastric cancer(EGC)formation is still unclear.This study will use the Weighted Gene Co-expression Network Analysis(WGCNA)method to explore the trajectory of gene expression profile in the carcinogenesis of EGC,and combine the clinical information to discuss the value of differentially expressed genes(DEGs)in diagnosis and prognosis.Materials&Methods:A total of two independent biopsy tissue samples were included in this study,which were numbered GSE6480 and GSE17077:GSE6480 consisted of 19 cases of Non-Atrophic Gastritis(NAG)and 19 cases of low grade intraepithelial neoplasia LGIN),20 cases of High Grade Intraepithelial Neoplasia(HGIN)and 19 cases of EGC;GSE17077 is composed of 15 cases of LGIN,16 cases of HGIN,16 cases of EGC and the same amount of corresponding background mucosa.The transcriptome of lesion mucosa was compared with NAG or corresponding background mucosa to obtain DEGs.DEGs were divided into different co-expression modules by WGCNA to describe their expression profile trajectories and make gene function analysis.Besides,module hub genes were filtered by network connectivity.According to the results of the functional annotation,the Cibersort tool was used to further assess the infiltration of immune cells in the sample tissue.For diagnosis,the random forest model was used to screen the genes with the most differential diagnostic efficacy among DEGs,and a three-class molecular diagnostic model was established to identify non-atrophic gastritis(NAG)or background mucosa,low-grade intraepithelial neoplasia(LGIN),High Grade Intraepithelial Neoplasia(HGIN)and EGC.For prognosis,transcriptome data and prognostic data of the advanced cancers from Asian Cancer Research Group(ACRG)were downloaded,the samples were categorized into two groups by unsupervised clustering according to the module hub genes,and Kaplan Meier method was used to compare overall survival(OS)and disease-free survival(DFS)between the two groups.Results:A total of 3739 DEGs were found,which can be divided into 8 co-expression modules,distinguished by different colors.Among them,the brown module genes are mainly involved in intestinal metaplasia,and their expressions peak at the LGIN stage;the blue module genes are mainly involved in nucleic acid processing and the cell cycle,and their expressions peak at the HGIN stage;the black module genes are mainly-related to innate immunity,and their expressions increase continuously in the carcinogenesis of EGC;the pink module genes are mainly involved in the T cell-related immune response,the expression of which is presented as a biphasic form with decreasing at the precancerous stages.The tissue infiltration degree of CD8 T cells in the carcinogenesis of EGC shows a similar biphasic form as pink module;and the proportion of CD8 T cells in immune infiltrating cells is smaller in lesion mucosa than the corresponding background mucosa.Based on DEGs,the random forest three-class diagnostic model is established to the distinguish of NAG or background mucosa,LGIN,HGIN and EGC.The overall misclassification probability varies from 21.28%to 27.66%,and the corresponding mcAUC(Multi-class area under the curve)is between 0.8714 and 0.8824.In advanced gastric cancer,the group with higher expression level of black module hub genes has a better prognosis than the lower expression group(OS,P=0.046;DFS,P=0.028).Conclusions:Among DEGs related to EGC carcinogenesis,the expressions of most genes,including genes related to intestinal metaplasia,nucleic acid processing and cell cycle,peak at the precancerous stage,suggesting that they mainly play a role in the early stage of carcinogenesis.The random forest three-class diagnostic model based on DEGs can effectively distinguish NAG or background mucosa,LGIN,HGIN and EGC.The tissue infiltration of innate immune cells,especially monocytes and mast cells,continues to increase during the carcinogenesis,suggesting that these two cells may be an important cause of tumor-promoting inflammation.Down-regulation of CD8 T cell function occurs before tumor invasion,which may be the basis of immune escape.Gastric cancer patients with strong immune response characteristics have a good prognosis,and the hub genes of the innate immune related module can be used as molecular signature to predict the prognosis of gastric cancer.