MALMPS: A Machine Learning-Based Metabolic Gene Prognostic Signature for Stratifying Clinical Outcomes and Molecular Heterogeneity in Stage II/III Colorectal Cancer.

Abstract

Colorectal cancer (CRC) is a frequently lethal disease, with stage II/III CRC accounting for ≈70%. Metabolic reprogramming plays a pivotal role in deciphering cancer heterogeneity and progression. Here, 9 datasets and 83 machine learning algorithm combinations are leveraged‌ to develop the Machine Learning-based Metabolic gene Prognostic Signature (MALMPS) model. The MALMPS model outperformed traditional clinical traits and molecular features in predicting prognosis for stage II/III CRC patients across training and validation datasets. COX7B, a key gene in MALMPS, is shown to promote CRC malignancy through multi-omics analysis and in vitro assays. CRC patients are stratified into high- and low-risk groups based on the median cutoff of MALMPS. Notably, the high-risk subgroup exhibited poor prognosis, activated inflammation, and enriched carbohydrate, glycosaminoglycan, and lipid metabolism, with therapeutic potential for IGF-1R and Wnt/β-catenin inhibitor. In contrast, the low-risk group displayed a TGF-β pathway inactivating mutation and enriched in nucleotides, cofactors, and amino acids metabolism. Metabolite profiling in the in-house SDCRC dataset validated the distinct metabolic alterations between the two groups. These findings indicate that MALMPS is a valuable instrument for predicting the recurrence risk of stage II/III colorectal cancer, particularly for identifying individuals at high risk.