A risk model constructed using 14 N-methyladenosine-related lncRNAs as a new prognostic marker that correlates with the immunomodulatory effect and drug sensitivity in colorectal cancer.

Background: Colorectal cancer (CRC) remains the most common gastrointestinal malignancy. Despite multimodal therapy, its mortality is high due to recurrence and metastasis. This study developed and verified a risk model consisting of 14 N-methyladenosine (mA) long noncoding RNAs (lncRNAs) to assess the prognosis of patients with CRC and investigated its relevance to immune regulation and drug sensitivity.

Methods: The gene expression profiles and clinical data of 446 patients with CRC were retrieved from The Cancer Genome Atlas (TCGA). 14 lncRNAs were screened using the Gene Co-expression Network (corFilter =0.5, P<0.001), and univariate and least absolute shrinkage and selection operator (LASSO) Cox regression analysis to construct the optimal risk model. The predictive performance and clinical applicability of the model were next verified. In addition, we performed Gene Ontology (GO) enrichment analysis to identify potential biological functions and detected the difference in tumor mutational burden (TMB), immune function, and sensitivity to immunotherapy and other drugs between the high- and low-risk groups to evaluate the application of the constructed risk model in depth.

Results: The model was found to be an appropriate marker for predicting the prognosis of patients with CRC, independent of other clinical features, and demonstrated good precision and broad clinical applicability. It correlated with pathways in the development of cancer and immune-related functions, and patients in the high-risk group had higher tumor immune dysfunction and escape (TIDE) scores. Furthermore, we found significant differences in the overall survival (OS) between patients in the high- and low-tumor mutation burden (TMB) groups, which may work in conjunction with the constructed model to better predict patients' prognosis. Finally, we identified 12 drugs, including A-443654 and sorafenib, with lower half maximal inhibitory concentration (IC) values in the high-risk group. Conversely, 21 drugs, including gemcitabine and rapamycin, had lower IC values in the low-risk group.

Conclusions: We constructed a risk model based on 14 mA-related lncRNAs that could predict the prognosis of patients with CRC and provided additional therapeutic ideas for their treatment. These findings may additionally serve as a foundation for further studies on regulating CRC via mA-related lncRNAs.