Integrating population-based metabolomics with computational microbiome modelling identifies methanol as a urinary biomarker for protective diet–microbiome–host interactions

Background: Diet–microbiome interactions are core to human health, in particular through bacterial fibre degradation pathways. However, biomarkers reflective of these interactions are not well described. Methods: Using the population-based SHIP-START-0 cohort (n = 4017), we combined metabolome-wide screenings with elastic net machine learning models on 33 food items captured using a food frequency questionnaire (FFQ) and 43 targeted urine nuclear magnetic resonance (NMR) metabolites, identifying methanol as a marker of plant-derived food items. We utilised the independent SHIP-START-0 cohort for the replication of food–metabolite associations. Moreover, constraint-based microbiome community modelling using the Human Microbiome data (n = 149) was performed to predict and analyse the contribution of the microbiome to the human methanol pools through bacterial fibre degradation. Finally, we employed prospective survival analysis in the SHIP-START-0 cohort, testing urinary methanol on its predictive value for mortality. Results: Among 21 metabolites associated with 17 dietary FFQ variables after correction for multiple testing, urinary methanol emerged as the top hit for a range of plant-derived food items. In line with this, constraint-based community modelling demonstrated that gut microbiomes can produce methanol via pectin degradation with the genera Bacteroides (68.9%) and Faecalibacterium (20.6%) being primarily responsible. Moreover, microbial methanol production capacity was a marker of high microbiome diversity. Finally, prospective survival analysis in SHIP-START-0 revealed that higher urinary methanol is associated with lower all-cause mortality in fully adjusted Cox regressions. Conclusion: Integrating population-based metabolomics and computational microbiome modelling identified urinary methanol as a promising biomarker for protective diet–microbiome interactions linked to microbial pectin degradation.