Abstract
Purpose: To assess food-specific metabolites as objective markers of dietary intake, and their suitability for real-world application.
Methods: Medline, Scopus, and Embase were searched from inception to 20 August 2024 with no restrictions independently by two reviewers and augmented by searching reference lists and bibliographies. Eligible studies were of dietary interventions where food was provided to participants and known metabolite markers of dietary intake were quantified. One reviewer extracted all data into a pretested excel sheet, a second reviewer then checked every extracted value.
Results: 113 publications reporting on at least one of 23 prespecified metabolites were identified. Pairwise meta-analyses of correlations between food intake and metabolite markers revealed moderate correlations, whereas dose response meta-analyses of post-intervention metabolite concentrations showed linear associations between alkylresorcinols and whole grains, daidzein and genistein with soy foods, and theobromine with chocolate intakes. Meta-regression analyses indicated these associations were largely stable by population group, analytical method of metabolite assessment, and whether food interventions were acute, or under observation. High heterogeneity in analyses indicated there were other factors driving differences between studies. Non-linear associations were observed between α-carotene and vegetables and sulforaphane with broccoli, limiting their use in real-world applications.
Conclusions: Metabolites showing the most potential for real-world application were alkylresorcinols for whole grains, daidzein or genistein for soy foods and theobromine for chocolate. More dietary interventions are needed to establish viable metabolite markers reflecting all food groups, and meta-research to identify potential drivers of differences in food-metabolite associations between studies.