Folic acid plays a crucial role in cellular regulation and metabolism, commonly included in dietary supplements. Despite this, its involvement in colorectal cancer (CRC), particularly in metabolic pathways and immune evasion, remains poorly understood. We developed the FMRG_score system using machine learning algorithms on TCGA and GEO data to assess modification patterns influencing clinical outcomes and immune characteristics in CRC. The system's reliability was validated across multiple external immunotherapy cohorts. We examined associations between FMRG-related features and clinical traits, mutation profiles, biological functions, immune infiltration, therapy responses, and drug sensitivities. By integrating in vitro and in vivo experiments with bioinformatics, we built a nine-gene risk model linked to folate metabolism for CRC prognosis. Notably, CYP26A1, a key gene in the model, was upregulated in CRC tissues, promoting cell proliferation, migration, invasion, and contributing to an immunosuppressive tumor microenvironment. Significant differences in clinical traits, immune infiltration, checkpoint expression, therapy response, and drug sensitivity were observed between risk groups. This folate-based scoring system provides a novel tool for evaluating CRC prognosis, tumor microenvironment, and response to immunotherapy. We also propose CYP26A1 as a potential oncogene in CRC, offering new therapeutic insights.
© 2025. The Author(s).