Background: Microsatellite instability (MSI) and microsatellite stability (MSS) define distinct molecular subtypes of colorectal cancer (CRC) with differential responses to immune checkpoint inhibitors (ICIs). While MSI tumours exhibit high immunogenicity and favourable ICI outcomes, MSS tumours remain largely resistant. Metabolic reprogramming may contribute to these differences, yet its role in shaping immune response is poorly understood.

Methods: We analysed transcriptomic and metabolomic profiles from CRC cohorts to identify metabolic pathways associated with MSI and MSS phenotypes. A machine-learning approach was applied to derive a predictive metabolic signature (IMMETPRED) and assess its correlation with immune-related features and clinical outcomes.

Results: The IMMETPRED signature stratified patients by metabolic phenotype and predicted ICI response with high accuracy. Functional analysis revealed strong associations between metabolic divergence and immune infiltration patterns.