Abstract
BACKGROUND: Specific metabolites detected in biofluids after food intake have been proposed as objective markers to address limitations with traditional dietary assessment methods. Although hundreds of metabolites have been associated with foods or food groups, they require validation. OBJECTIVES: The aim of this study was to develop a panel from the proposed biomarkers of dietary intake reflecting major food groups and compare them against known and self-reported dietary intakes. METHODS: A randomized crossover trial of 3 interventions, including a day of foods consumed under observation, was performed. Each feeding day was of 3 food groups (e.g., whole grains, dairy, and fish) with optional snacks (e.g., fruit, chicken, and legumes) but absent of 3 other food groups (e.g., meat, vegetables, nuts, and seeds). Fasted whole blood and 24-h urine samples were analyzed by liquid chromatography mass spectrometry to detect previously proposed biomarkers of food intake. Urinary sodium was measured. Pairwise correlation coefficients and generalized linear modeling (GLM) considered relationships between biomarkers and food groups. Comparisons were drawn between self-reported and known dietary intakes. RESULTS: Twenty-one participants [mean age 24.8, standard deviation (SD) 6.0 y, body mass index: 24.1; SD: 4.0] completed the trial. GLM coefficients indicated fish intake was associated with 3-carboxy-4-methyl-5-propyl-2-furanpropanoic acid [62.15 (95% confidence interval: 35.00, 89.29)], wholegrain intake with 3,5-dihydroxybenzoic acid [87.32 (24.28, 150.35)] and fruit intake with fructose [5.39 (2.53, 8.25)], and s-methylcysteine [5.91 (1.24, 10.58)]. GLM rate ratios indicated chicken intake was associated with 3-methylhistidine [0.19 (0.07, 0.31)], anserine [0.21 (0.05, 0.37)], and carnosine [0.11 (0.03, 0.19)], legume intake with glycine betaine [0.21 (0.02, 0.40)] and vegetable intake with sulforaphane [0.30 (0.20, 0.47)], S-methylcysteine [0.23 (0.14, 0.45)], methoxytyramine [0.21 (0.08, 0.35)], and β-carotene [0.05 (0.02, 0.08)]. There was no association between 24-h urinary sodium and known sodium intake [0.11 (-0.06, 0.28)]. Self-reported dietary intake was associated above acceptable level (r > 0.40) with known intake. CONCLUSIONS: We identified some previously reported associations between foods and proposed biomarkers, but not all, outlining the need for assessing dietary biomarkers across a range of study designs including food intakes within realistic ranges. This trial was registered at ACTRN as 12622001036707 (https://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=384292&isReview=true).