MiRNA-based model for predicting the TMB level in colon adenocarcinoma based on a LASSO logistic regression method

Abstract

AbstractSome patients with advanced colon adenocarcinoma (COAD) are not sensitive to radiotherapy and chemotherapy, and as such, immunotherapy has become the most popular option for these patients. However, different patients respond differently to immunotherapy. Tumor mutational burden (TMB) has been used as a predictor of the response of advanced COAD patients to immunotherapy. A high TMB typically indicates that the patient's immune system will respond well to immunotherapy. In addition, while microRNAs (miRNA) have been shown to play an important role in treatment responses associated with the immune system, the relationship between miRNA expression levels and TMB has not been clarified in COAD.We downloaded miRNA data and mutational files of COAD from the Cancer Genome Atlas database. Differentially expressed miRNAs were screened in the training group, and miRNAs used to construct the model were further identified using the LASSO logistic regression method. After building the miRNA-based model, we explored the correlation between the model and TMB. The model was verified by a receiver operating characteristic curve, and the correlation between it and 3 widely used immune checkpoints (programmed death receptor-1, programmed death-ligand 1, and cytotoxic T-lymphocyte associated protein-4) was explored. Functional enrichment analysis of the selected miRNAs was performed, and these respective miRNA target genes were predicted using online tools.Our results showed that a total of 32 differentially expressed miRNAs were used in the construction of the model. The accuracies of the models of the 2 datasets (training and test sets) were 0.987 and 0.934, respectively. Correlation analysis showed that the correlation of the model with programmed death-ligand 1 and cytotoxic T-lymphocyte associated protein-4, as well as TMB, was high, but there was no correlation with programmed death receptor-1. The results of functional enrichment analysis indicated that these 32 miRNAs were involved in many immune-related biological processes and tumor-related pathways.Therefore, this study demonstrated that differentially expressed miRNAs can be used to predict the TMB level, which can help identify advanced COAD patients who will respond well to immunotherapy. The miRNA-based model may be used as a tool to predict the TMB level in patients with advanced COAD.