Abstract
AIM: To build a ferroptosis-related prognostic model for patients with colon adenocarcinoma (COAD). METHODS: COAD expression profiles from The Cancer Genome Atlas were used as the training set and GSE39582 from Gene Expression Omnibus as the validation set. Differentially expressed ferroptosis-related genes between patients with COAD and normal controls were screened, followed by tumor subtype exploration based on ferroptosis-related gene expression levels. A ferroptosis score (FS) model was constructed using least absolute shrinkage and selection operator penalized Cox analysis. Based on FS, patients were subgrouped into high- and low-risk subgroups and overall survival was predicted. The potential prognostic value of the FS model and the clinical characteristics were investigated using receiver operating characteristic curves. RESULTS: Twenty-four differentially expressed ferroptosis-related genes were identified, four of which (CYBB, PRNP, ACSL4, and ACSL6) were included in the prognostic signature. Moreover, age, pathological T stage, and tumor recurrence were independent prognostic factors for COAD. The FS model combined with three independent prognostic factors showed the best prognostic value (The Cancer Genome Atlas: area under the curve = 0.897; GSE39582: area under the curve = 0.858). CONCLUSION: The novel prognostic model for patients with COAD constructed by pairing the FS model with three important independent prognostic factors showed promising clinical predictive value.