Abstract:

Objective ·To construct a prediction model of immune-related long non-coding RNA (lncRNA) in gastric cancer patients by bioinformatics method, and explore its application value. Methods ·Transcriptome sequencing (RNA sequencing, RNA-seq) data of 413 gastric cancer samples were downloaded from the cancer genome atlas (TCGA) database, including 32 normal samples and 381 tumor samples. Immune-related genes were obtained from ImmPort website. The immune-related lncRNAs (irlncRNAs) were obtained by correlation analysis. The differentially expressed irlncRNAs (DEirlncRNAs) were obtained by the limma R package, and heat maps and volcano maps were drawn. Batch effects of sample were corrected by constructing DEirlncRNA pairs. The clinicopathological characteristic data of TCGA gastric cancer patients were downloaded, and the DEirlncRNA pairs related to prognosis were obtained by univariate analysis, and screened by LASSO regression analysis. Finally, a risk prediction model was constructed by COX proportional hazards regression analysis. The predictive performance of the model and traditional clinicopathological features were analyzed and compared by calculating the area under the curve (AUC). The patient risk value was calculated according to the formula, and the patients were divided into high and low risk groups according to the optimal cutoff value. Survival maps were drawn by using Kaplan-Meier curves, and differences in survival rates between the two groups were compared by Log-rank test. The relationship between risk scores and clinicopathological characteristics was analyzed according to the Wilcoxon signed-rank test. The independent prognostic factors of gastric cancer patients were verified by univariate analysis and multivariate analysis. The relationship between risk scores and immune infiltrating cells and immune-related genes was validated according to Spearman correlation analysis. The half-maximal inhibitory concentration (IC₅₀) values of drugs in high and low risk groups were compared by using the pRRophetic R package. Results ·Compared with normal tissues, 106 irlncRNAs were differentially expressed in gastric cancer tissues, of which 11 were low-expressed and 95 were high-expressed. A total of 32 DEirlncRNAs pairs were included in the COX proportional hazards model, 20 of which were independent prognostic factors for gastric cancer. The 1-, 2- and 3-year AUC values of the risk prediction model were 0.889, 0.966 and 0.935, respectively, which were significantly higher than those of traditional clinicopathological features. The survival rate of patients in the high risk group was significantly lower than that in the low risk group (P=0.000). High risk scores were more closely related to high tumor stage, distant metastasis and patient death. Univariate analysis showed that age, TNM stage, T stage, N stage, M stage and risk score were closely related to the prognosis of gastric cancer patients (all P<0.05). Multivariate analysis showed that age, TNM stage and risk score were independent prognostic factors for gastric cancer patients (all P<0.05). Risk scores were negatively correlated with various T cells and mast cells, and positively correlated with tumor-associated fibroblasts, macrophages and endothelial cells. In the high risk group, the expression levels of immune-related genes IFNG and MSH2 were lower than those in the low risk group. In the high risk group, the sensitivity of the drugs doxorubicin, cisplatin, tipifarnib and mitomycin were lower than those in the low risk group (all P<0.05). Conclusion ·The COX proportional hazards model constructed with irlncRNAs pairs can accurately predict the survival status and survival rate of gastric cancer patients and its sensitivity to chemotherapy drugs.

Key words: gastric cancer, immune-related long non-coding RNA (irlncRNA), COX proportional hazards model, LASSO regression, area under the curve (AUC)