Colorectal cancer (CRC) is a highly prevalent malignancy marked by significant inter-patient heterogeneity in immune and metabolic traits, which critically influences therapeutic response and clinical outcomes. This complexity reflects the dynamic tumor ecosystem, where diverse cellular and molecular interactions drive disease progression and resistance. Despite therapeutic advances, metastatic CRC (mCRC), particularly microsatellite stable (MSS) tumors, remains largely refractory to immunotherapy and prone to relapse. A deeper understanding of tumor-specific metabolic reprogramming is essential to identify actionable vulnerabilities and guide personalized therapies.

In this study, we present a multi-omics framework that integrates transcriptomic, metabolomic, and functional data into Genome-Scale Metabolic Models (GSMMs) to computationally model the tumor ecosystem and redefine CRC stratification based on metabolic phenotypes. This approach enabled the identification and preclinical validation of specific metabolic vulnerabilities, advancing precision oncology by translating multi-omics data into actionable targets. Although not spatially resolved, the framework is compatible with spatially annotated datasets, offering future opportunities to dissect intratumoral metabolic heterogeneity.

In parallel, to further refine stratification, we developed an immune–metabolic gene signature capturing key immune and metabolic traits in CRC. This signature supported a predictive model that classified tumors into two main clusters: one predominantly glycolytic and another characterized by enhanced metabolic flexibility and oxidative phosphorylation (OXPHOS). These profiles support more precise patient selection and inform tailored therapeutic strategies targeting immune–metabolic vulnerabilities.

Altogether, our results demonstrate that multi-omics data modeling enables refined stratification of CRC tumors, uncovering metabolic vulnerabilities that can be exploited for precision therapy design. This approach provides a robust framework for the development of personalized treatment strategies based on tumor-specific metabolic rewiring.

Acknowledgments
Authors acknowledge support from MICIU/AEI/10.13039/501100011033–European Commission FEDER funds (PID2023-150539OB-I00); CIBER-EHD (EHD20PI03, CB17/04/00023); AGAUR (2021-SGR-00350); ICREA Foundation (ICREA Academia award to M.C.); and the Spanish Structures of María de Maeztu program (CEX2021-001202-M).