Background: Liquid chromatography coupled with mass spectrometry (LC-MS)-based metabolomics has uncovered numerous metabolic changes associated with colorectal cancer (CRC), yet reproducibility across studies and study designs remains a major challenge. Integrating multiple CRC metabolomics studies conducted at IARC’s Metabolomics Laboratory using harmonized analytical workflows enables the identification of robust and mechanistically informative metabolic signatures.
Objectives: To evaluate reproducibility and overlap of CRC-associated metabolites across multiple case-control and prospective LC-MS studies, to characterize their involvement in major metabolic pathways, and to assess cross-validation with obesity and weight-loss metabolomics studies conducted at IARC’s Metabolomics Laboratory.
Methods: We performed a harmonized integrative analysis of multiple independent LC-MS metabolomics studies of CRC conducted in a single laboratory, including four case-control studies with diagnostic blood samples and one prospective study with pre-diagnostic blood samples (EPIC). All studies used standardized sample handling protocols, identical analytical platforms and methods, and common quality control procedures. Metabolites were evaluated for analytical reliability. Study-specific associations with CRC were estimated using multivariable-adjusted models. Cross-study overlap and directional consistency were evaluated via vote-counting and rank-based methods, with pathway-level analyses to identify consistently dysregulated metabolic processes. Unannotated but reproducible features were tracked for consistency across studies. CRC-associated metabolites were further compared with findings from obesity and weight-loss intervention metabolomics studies conducted in the same laboratory using identical LC-MS instrumentation and analytical protocols, to explore shared metabolic signatures.
Results: Across studies, we identified a reproducible subset of circulating metabolites associated with CRC. Consistent alterations were observed in pathways related to amino acid metabolism, fatty acid oxidation, polyamine metabolism, tryptophan-microbiota interactions, heme metabolism, and nitrogen metabolism/uremic toxins. Some of the metabolites were also altered years prior to CRC diagnosis in prospective analyses, indicating early metabolic perturbations. Notably, unannotated LC-MS features also showed reproducible associations across studies, highlighting candidates for further annotation. A subset of these metabolites overlapped with features associated with BMI, waist circumference, or significantly altered by weight-loss interventions in independent obesity studies, suggesting shared metabolic mechanisms.
Conclusions/Implications: This cross-study integrative analysis identifies early and reproducible metabolic alterations in CRC with strong directional concordance, despite differences in study design and populations, implicating key pathways including amino acid, lipid, and microbiota-linked metabolism. Furthermore, several circulating metabolites also overlap with those associated with obesity-related metabolic traits and weight-loss interventions. Harmonized multi-study analyses provide insights into CRC biology, prioritize both annotated and unannotated robust features for further study, and support their potential utility in early detection, biomarker development, and future mechanistic investigations.