Abstract
Background:Colorectal cancer (CRC) remains one of the most common malignancies across the world. Thus far, a biomarker, which can comprehensively predict the survival outcomes, clinical characteristics, and therapeutic sensitivity, is still lacking. Methods: We leveraged retrieved transcriptomic data of CRC from the public database, and constructed irlncRNA pairs. After integrating with clinical survival data, we performed differential analysis and constructed 11 irlncRNA pairs signature using Lasso regression analysis. We next drew the 1-, 5-, 10-year curve line of receiver operating characteristics, calculated the areas under the curve, and recognized the optimal cutoff point. Then, we validated the pair-risk model in terms of the survival outcomes of the patients. Moreover, we tested the reliability of the pair-risk model for predicting tumor aggressiveness and therapeutic responsiveness of CRC. Additionally, we reemployed the 11 of irlncRNAs involved in the pair-risk model to construct an expression risk model that was also highly predictive of prognostic outcomes of CRC patients. Results:We recognized a total of 377 DEirlcRNAs, including 28 low-expressed and 349 high-expressed irlncRNAs in CRC patients. After performing a univariant Cox analysis, we identified 115 risk irlncRNAs that were significantly correlated with survival outcomes of patients with CRC. By taking the intersection of the DEirlcRNAs and the risk irlncRNAs, we ultimately recognized 55 irlncRNA as core irlncRNAs. Then, we established a Cox HR model (pair-risk model) as well as an expression HR model (exp-risk model). We found that both of the two models significantly outperformed the common-used clinical characteristics, including age, T, N, and M stages, in terms of predicting survival outcomes. Moreover, we validated the pair-risk model can serve as a potential tool for studying the tumor microenvironment of CRC and drug response. Additionally, we noticed that combining the pair-risk model and exp-risk model yielded a more robust approach for predicting the survival outcomes of patients with CRC.Conclusions:We suggest that the irlncRNA-based risk models can be utilized as prognostic tools to predict survival outcomes and clinical characteristics and guide treatment regimens of CRC.