Serum metabolomics modelling reveals stage-associated purine and endocannabinoid metabolic signatures in colorectal cancer progression

Abstract

Background Colorectal cancer (CRC) remains one of the major challenges in contemporary oncology. Despite advances in imaging diagnostics and histopathological assessment, we still lack tools for the early detection of micrometastases and molecular alterations. The purinergic and endocannabinoid systems play an important role in CRC pathogenesis, yet their mutual interactions are not fully understood. Methods The study involved serum samples from 117 patients undergoing CRC surgical treatment. We performed targeted serum metabolomic profiling of purine metabolites and endocannabinoid-related lipids. Advanced statistical modeling (elastic-net regression and stepwise selection) was applied to link metabolite concentrations with TNM stage and pathological features, such as neuroinvasion and vascular infiltration. Results AMP was the strongest marker positively correlated with higher TNM stage. On the other hand, 2-AG level showed consistent negative correlation with neuroinvasion and lymphatic nodes metastasis. Xanthine level and N stage correlation was also observed. Conclusions The results allow us to propose a model in which AMP and 2-AG constitute opposing poles of the metabolic profile in CRC patients An increase in serum AMP was associated with more advanced CRC stage, whereas lower 2-AG levels were associated with invasive features, particularly neuroinvasion. This suggests that altered purine/nucleotide metabolism may be associated with reduced endocannabinoid tone in advanced CRC. However, the proposed interaction between purinergic signalling, COX-2 activity and 2-AG metabolism should be interpreted as a hypothesis-generating model requiring direct tissue-level validation. These findings support further investigation of purine–endocannabinoid crosstalk as a potential biologically relevant pathway in CRC progression, but therapeutic implications require direct tissue-level and functional validation.