Construction and Validation of A MiRNAs Prediction Model for Tumor Mutational Burden in Gastric Cancer / 胃肠病学

ABSTRACT

Background: Tumor mutational burden (TMB) is a promising predictor, which can evaluate the efficacy of immune checkpoint inhibitors (ICIs) for tumor patients. MicroRNAs (miRNAs) act as the key regulators of anti - cancer immune response. However, the relationship between TMB and miRNAs expression profiles in gastric cancer is not elucidated. Aims: To construct and validate the TMB prediction model of gastric cancer by analyzing the sequencing data and clinical data in TCGA database. Methods: Gastric cancer patients in TCGA database were divided into TMB high group and TMB low group, and differentially expressed miRNAs were screened. Gastric cancer patients were randomly divided into training group and validation group. A miRNAs prediction model for predicting TMB level was developed by lasso regression analysis method in training group, and was validated in validation group. KEGG functional enrichment analysis was used to analyze the screening related miRNAs. The correlation between miRNAs and three ICIs was analyzed. Results: Fifty-four differentially expressed miRNAs were screened, and 20 of them were identified as TMB-related miRNAs. The sensitivity, specificity, AUC in the miRNAs prediction model were relatively high. Functional enrichment results revealed that TMB-related miRNAs were mainly involved in biological process associated with immune response and signaling pathways related with cancer. The miRNAs prediction model showed a median positive correlation with PD-L1 (r=0.36, P<0.01), a median negative correlation with PD-1 (r=-0.31, P<0.01) and no significant correlation with CTLA4 (r=0.14, P<0.01). Conclusions: This study presents a miRNAs prediction model which can stratify gastric cancer patients with different TMB levels.