Introduction: The purpose of the present academic work was to identify genes with worse prognoses of gastric cancer (GC). Methods: GSE54129 & GSE64951 were extracted from GEO database. Differentially expressed genes (DEGs) between GC and normal specimen were identified via GEO2R online tools with logFC > 1 and adjust P value < 0.05. Raw data were checked in Venn software online to detect the common DEGs. The DEGs with logFC > 1 were considered as up-regulated genes. Gene ontology (GO) analysis and KEGG pathway analysis were conducted and visualized via DAVID and Bioinformatics online tool. PPI network was evaluated by STRING and hub genes were selected via CytoHubba app in Cytoscape. Kaplan Meier-plotter was used to conduct survival analysis. GEPIA website was used to analyze the expression. The best discriminate cut-off points between the high and low expression groups were assessed with the Receiver operating characteristic (ROC) curve and area under the curve (AUC) for overexpressed DEGs by using R software with pROC and ggplot2 packages to analyze data from TCGA-STAD with Log2 transformed FPKM. Results: There are 174 GC & 52 normal tissue samples in the two profile datasets, and 605 and 1,860 up-regulated DEGs from GSE64951 and GSE54129, respectively. Venn diagram detected 62 commonly up-regulated DEGs. GO analysis indicated that they were enriched in integrin-mediated signaling pathway, extracellular matrix organization, and collagen fibril organization for biological processes (BP), and enriched in integrin complex, integrin alpha1-beta1 complex, and endoplasmic reticulum lumen for GO cell component (CC), and enriched in cell adhesion molecule binding, collagen-binding involved in cell-matrix adhesion, and transcription corepressor activity for molecular function (MF). KEGG analysis found that they were enriched in Axon guidance, Cell adhesion molecules (CAMs), and Leishmaniasis. Analysis of PPI network complex found top 10 genes as shown in the PPI network figure. Kaplan Meier plotter identified 3core genes (COL5A1, COL12A1, ITGBL1) associated with a worse prognosis (P < 0.05), and GEPIA found a higher expression level of them in GC tissues than normal tissue. ROC curve showed that AUC of COL5A1, COL12A1, ITGBL1 were 0.813, 0.831, and 0.646, respectively, and their cutoff values were 4.494, 2.789, and 0.900. Conclusion: In GC, COL5A1, COL12A1, and ITGBL1 are associated with poor prognoses that could be potential therapeutic targets., COL5A1 and COL12A1 could be potential diagnostic markers.Figure 1.: A) PRISMA flow chart; B) Forest plot of the overall results; C) Funnel Plot.
Read full abstract