Identification of cuproptosis-related subtypes and development of a prognostic signature in colorectal cancer

Cuproptosis, a novel form of copper-mediated regulated cell death, participates in tumor progression. However, the role of cuproptosis-related genes (CRGs) in colorectal cancer (CRC) remains unclear. We aimed to investigate the cuproptosis subtypes and build a predictive model to improve the prognosis of patients with CRC. Gene expression data were downloaded from the TCGA database to identify distinct molecular subtypes using a non-negative matrix factorization algorithm. A robust and efficient prognostic signature was constructed by performing multivariate Cox regression analysis and further validated using the Gene Expression Omnibus cohort. Based on the gene expression matrix of CRC, the abundance of infiltrating immune cells and tumour microenvironment scores were calculated using the CIBERSORT and ESTIMATE algorithms, respectively. The pRRophetic algorithm was used to predict the sensitivity of the patients to different chemotherapy drugs. Two distinct molecular subtypes were identified based on 41 CRGs, with subtype C1 being characterized by an advanced clinical stage and worse overall survival. A prognostic signature was constructed based on the DEGs between the two cuproptosis subtypes, and its predictive ability was validated in an external database. Patients with CRC who belonged to the low-risk group had significantly higher survival rates than those who belonged to the high-risk group. Additionally, it remained a valid prognostic indicator in strata of age, sex, tumor location, and TNM stage, and its significance persisted after the multivariate Cox regression analysis. By further analyzing the prognostic signature, a higher immune score was observed in the low-risk group, which presented a better prognosis. AKT.inhibitor.VIII, doxorubicin, lenalidomide, and tipiparnib were more sensitive in the high-risk score group. A highly accurate nomogram was constructed to improve clinical application of the risk score. Compared with an ideal nomogram, our model, consisting of clinicopathological features, performed well in predicting patient survival. In conclusion, our study provides new ways and perspectives for the prediction of the prognosis of patients with CRC and guide more effective treatment regimens.