|
Establishment of prognostic risk evaluation model for gastric cancer integrating ferroptosis and immune-related genes
ZHANG Yujun, WANG Yan, ZHU Lin, CAI Yingbin
Chinese Clinical Oncology. 2023, 28 (01):
38-48.
Objective To explore the prognostic value of ferroptosis related genes (FRGs) and immune related genes (IRGs) in gastric cancer and build a prognostic risk model to predict the overall survival (OS). MethodsThe transcriptome data and clinical data of gastric cancer were downloaded from the Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases, and FRGs and IRGs were extracted from FerrDb and Immport databases. The TCGA expression profile genes were intersected with FRGs and IRGs to obtain gastric cancer FRGs and IRGs. The differential FRGs and IRGs with prognostic value were screened by differential analysis and prognostic correlation analysis, and the intersection was obtained by Venn plot. The ferroptosis death related gene risk score (FRG risk-score) and immune related gene risk score (IRG risk-score) models were further screened by LASSO-Cox regression. FRG riskscore and IRG risk-score models were evaluated by Kaplan-Meier survival analysis, receiver operating characteristic curve (ROC), principal component analysis (PCA), t-distributed stochastic neighbor embedding (t-SNE), univariate and multivariate Cox regression analysis, and nomograms and calibration curves combining clinicopathological features, FRG riskscore and IRG riskscore were constructed. Meanwhile, the GSE84437 dataset was used to validate the model. Cibersort algorithm was used to estimate the abundance of 22 kinds of infiltrating immune cells in all samples of TCGA cohort and the correlation between FRGs, IRGs and immune cell infiltration was analyzed using TIMER database. TCIA and TIDE websites were used to analyze the difference of immune drug sensitivity and immune escape score between different risk groups. ResultsA total of 8 overlapping candidate FRGs and 4 overlapping candidate IRGs were obtained through screening. LASSO-Cox regression results showed that five FRGs (ZFP36, NOX4, SLC1A5, MYB, DUSP1) and four IRGs (CGB5, GHR, GLP2R, NPR1) were included in the prognostic model. Kaplan-Meier survival analysis, area under ROC curve (AUC), PCA and tSNE analysis all verified that the model had good prediction ability. Combining the clinicopathological features, nomograms of FRG risk-score and IRG risk-score and their calibration curves, it is shown that the model has a good predictive ability for 1-year and 3-year survival rates of gastric cancer patients. The results of correlation analysis between immune cell infiltration and model genes showed that there was a significant correlation between FRGs and IRGs and the level of immune cell infiltration (P<0.05). Immunopharmacotherapy analysis showed that patients in the low-risk group with FRGs prognostic characteristics had better immunotherapeutic response to CTLA4 blockers. Immune escape results showed that patients in the high-risk group of FRGs and IRGs prognostic models had higher TIDE scores. Conclusion The prognosis model integrating FRGs and IRGs can better predict the prognosis of gastric cancer patients, and can be used as an early detection biomarker and an anti-tumor immunotherapy target, which has clinical therapeutic significance.
Related Articles |
Metrics
|