A study published in Nature, Faecal metabolites as a readout of habitual diet capture dietary interactions with the gut microbiome offers compelling evidence that the metabolite composition of stool may serve as a reliable functional readout of what people eat and how their gut microbes respond.
In the largest investigation of its kind to date, researchers combined dietary questionnaire data, untargeted faecal metabolomics and metagenomic sequencing from two cohorts: 1,810 individuals from the TwinsUK cohort and 837 participants from ZOE PREDICT1.
Machine-learning models based only on faecal metabolites (with or without basic covariates such as BMI) were able to predict habitual consumption of major food groups, as well as adherence to established dietary patterns (for example, DASH, plant-based, Mediterranean). In particular, intake of meat, nuts and seeds, wholegrains, tea/coffee and alcohol were predicted with high accuracy (area-under-the-curve above 0.80).
Beyond diet prediction, the study found that a relatively small panel of 54 faecal metabolites could recapitulate much of the predictive power of the full dataset. That panel encompassed metabolites across diverse chemical classes including lipids, amino acids, vitamins/cofactors, and xenobiotics.
Importantly, many of these dietary-associated metabolites also correlated with gut microbiome composition and microbial diversity. Consumption of fibre-rich and plant-based foods was linked to metabolites associated with greater microbial diversity, whereas metabolites associated with higher meat consumption correlated with reduced diversity.
The results suggest that metabolite profiling can serve not simply as a passive biomarker of diet but as a mediator linking diet to microbiome structure. Mediation analyses revealed that the faecal metabolome provides a stronger statistical bridge between dietary intake and microbiome species than direct diet–microbe associations.
These findings carry important implications for nutritional science and microbiome research. If stool metabolomics can reliably reflect and summarise dietary habits, it may provide a more objective and functional alternative to self-reported food frequency questionnaires. In addition, because these metabolites mirror the metabolic consequences of diet–microbe interactions, they may be valuable in identifying candidate metabolites for targeted dietary or prebiotic interventions aimed at improving health outcomes.