Machine Learning-Driven Multi-Omics Analysis Identifies CHP2 as a Key PANoptosis-Related Dual-Function Biomarker in Colorectal Cancer

The heterogeneity of colorectal cancer (CRC) represents a great challenge in therapy. We integrated multiomics and machine learning, interpreted by SHAP models to provide a clinical rationale, to identify Calcineurin B Homologous Protein 2 (CHP2) as a core candidate, which was further validated via in vitro and zebrafish models. The expression of CHP2 are decreased in CRC, which is associated with a poor prognosis and an immune suppressed "cold" TIME. Functionally, CHP2 overexpression inhibits cell growth and invasion by inducing PANoptosis. Clinically, specific CHP2 expression profiles discriminate patients at high risk that are resistant to standard chemotherapy (e.g., 5-FU) but sensitive to targeted inhibitors. CHP2 is a powerful dual-function biomarker-prognostic for survival and predictive for the response to therapy-that could lead to a personalized approach in treating drug-resistant CRC.