Identification and characterization of ferroptosis-related genes in therapy-resistant gastric cancer

Abstract

Therapy resistance in gastric cancer poses ongoing challenges, necessitating the identification of ferroptosis-related genes linked to overall survival for potential therapeutic insights. The purpose of the study was to identify ferroptosis-related genes contributing to therapy resistance in gastric cancer and explore their associations with overall survival. Differentially expressed ferroptosis-related genes were identified in therapy-resistant versus therapy-responsive gastric cancer patients. Hub genes were selected from these genes. Enrichment analysis focused on oxidative stress and ROS metabolism. Validation was conducted in a TCGA stomach adenocarcinoma dataset. A hub gene-based risk model (DUSP1/TNF/NOX4/LONP1) was constructed and assessed for overall survival prediction. Associations with the tumor immune microenvironment were examined using the ESTIMATE algorithm and correlation analysis. Ten hub genes were identified, enriched in oxidative stress and ROS metabolism. Validation confirmed their aberrant expressions in the TCGA dataset. The hub gene-based risk model effectively predicted overall survival. High G6PD/TNF expression and low NOX4/SREBF1/MAPK3/DUSP1/KRAS/SIRT3/LONP1 expression correlated with stromal and immune scores. KRAS/TNF/MAPK3 expression positively correlated with immune-related SREBF1/NOX4 expression. DUSP1/NOX4/SREBF1/TNF/KRAS expression was associated with immune cell infiltration. The hub gene-based risk model (DUSP1/TNF/NOX4/LONP1) shows promise as an overall survival predictor in gastric cancer. Ferroptosis-related hub genes represent potential therapeutic targets for overcoming therapy resistance in gastric cancer treatment.

Figure 1.

Data preprocessing of the GSE31811 dataset. The RNA-sequencing data of the GSE31811 dataset were obtained from the NCBI Gene Expression Omnibus and normalized using logarithm transformation. (A, B) Box plots were generated to compare data distributions before (A) and after (B) normalization. (C, D) PCA was performed to compare data separations before (C) and after (D) normalization. PCA = principal component analysis.

Figure 2.

Identification of DEGs between therapy-responsive and therapy-resistant patients with gastric cancer. GSE31811 (A, B) and GSE26253 (C, D) datasets were divided into therapy-responsive and therapy-resistant groups, respectively. Volcano plots (A, C) and heatmaps (B, D) depict the differential RNA expressions in therapy-resistant group versus therapy-responsive group. Genes with |fold change| > 1 and adjusted P value < .05 were defined as DEGs. The red dots represent upregulated DEGs. The blue dots represent downregulated DEGs. DEGs = differentially expressed genes.

Figure 3.

Identification of ferroptosis-related DEGs. A total of 430 ferroptosis-related genes were obtained from GeneCards and FerrDB databases after removing the duplicates. Intersection analysis was conducted to identify ferroptosis-related DEGs in GSE31811 (A) and GSE26253 (B) datasets. DEGs = differentially expressed genes.

Figure 4.

Identification of hub genes. (A) A protein-protein interaction network was constructed to identify the hub genes in 21 ferroptosis-related DEGs using String. (B) The top 10 DEGs with the highest interaction degrees were defined as hub genes. DEGs = differentially expressed genes.

Figure 5.

Construction of regulatory networks of the hub genes. (A) The RNA-microRNA interaction network. (B) The RNA-transcription interaction network. (C) The RNA-RNA binding protein interaction network. (D) The RNA-drug interaction network.

Figure 6.

GO annotation and KEGG enrichment analysis of the hub genes. (A) GO enrichment bar chart. The y-axis represents the significance of the terms. The bars are sorted according to the z-scores. (B–D) Circle diagrams of enriched GO terms in biological process (B), molecular function (C), and cellular component (D) categories. The outer circle shows a scatter plot of the expression level (logFC) of each gene in the assigned GO term. Red dots represent upregulated genes. Blue dots represent downregulated genes. The inner ring is a bar plot indicating the significance of the GO terms, and the color corresponds to the z-score. (E) A bar chart of enriched KEGG signaling pathways. (F) The top 5 enriched KEGG signaling pathways. BP = biological process, CC = cellular component, GO = gene ontology, KEGG = Kyoto Encyclopedia of Genes and Genomes, MF = molecular function.

Figure 7.

GSEA. The enriched gene sets in the GSE31811 dataset. (B) The enriched gene sets in the GSE26253 dataset. (C, D, F) Enrichment analysis of GO terms RNA binding (C) and mRNA metabolic process (D) as well as KEGG pathway focal adhesions (F). (E) Enrichment analysis of KEGG pathways focal adhesion, regulation of actin cytoskeleton, and neuroactive ligand receptor interaction. GO = gene ontology, GSEA = gene set enrichment analysis, KEGG = Kyoto Encyclopedia of Genes and Genomes.

Figure 8.

The predictive value of the hub gene mRNA expressions in the overall survival of TCGA-STAD. (A–D) Comparison of mRNA expression of DUSP1 (A), TNF (B), NOX4 (C), and LONP1 (D) between gastric tumor tissue and normal gastric tissue samples from the TCGA-STAD (n = 350). (E) Total mRNA expression of 10 hub genes. (F) Forest plot of the multivariate Cox regression analysis of potential overall survival predictors. (G) The correlation matrix of the potential predictors. (H) Time-dependent receiver operating characteristic curve analysis. AUC = area under the curve, TCGA-STAD = The Cancer Genome Atlas stomach adenocarcinoma dataset.

Figure 9.

The performance of hub gene-based risk model in predicting the overall survival of patients in TCGA-STAD. (A–D) Wilcoxon rank-sum test was carried out to assess the correlation of hub gene-based risk scores with the age (A), gender (B), clinical staging (C), and tumor grading (D) of the TCGA-STAD. (E) Decision curve analysis. (F) A nomogram of the risk model. TCGA-STAD = The Cancer Genome Atlas stomach adenocarcinoma dataset.

Figure 10.

Correlations of hub gene expression with the immune microenvironment of gastric cancer patients from TCGA-STAD. (A, B) Correlations of hub gene expression with stromal scores (A) and immune scores (B). (C) Correlations of hub gene expression with immune-related gene expression. (D) Correlations of hub gene expression with immune cell infiltration. TCGA-STAD = The Cancer Genome Atlas stomach adenocarcinoma dataset.