Exploratory Screening and Replication of Urinary Biomarkers of Fruit and Vegetable Intake in Free-living European Children and Adolescents Using Untargeted Metabolomics

Background: Accurately measuring fruit and vegetable intake is challenging in epidemiological studies, and this difficulty is even greater in children. Biomarkers of fruit and vegetable intake may enhance objective assessment. Objectives: This exploratory study aimed to screen for potential biomarkers and assess the replicability of previously reported biomarkers associated with fruit and vegetable intake in a free-living population. Methods: Using an untargeted metabolomics approach, we quantified the molecular features in urine from the Identification and Prevention of Dietary- and Lifestyle-induced Health Effects in Children and Infants (IDEFICS) and I.Family cohort. To explore complementary temporal dimensions of diet-metabolite associations, we examined both short-term and habitual fruit and vegetable intake in parallel. Partial least squares and random forest were applied using the MUVR algorithm (Multivariate Methods with Unbiased Variable Selection in R) to identify molecular features related to fruit and vegetable intake. A linear mixed regression model was then fitted to selected features. In addition, the fruit and vegetable intake-metabolites associations were explored in the Dortmund Nutritional and Anthropometric Longitudinal Designed (DONALD) cohort to replicate analyses from the IDEFICS/I.Family cohort. Results: Fruit and vegetable intake were significantly related to 59 features. Ten metabolites could be annotated in the IDEFICS/I.Family cohort. We observed positive associations of total vegetable intake with octenoylcarnitine, total fruit intake with 5-hydroxyindoleacetate and D-pantothenic acid, orange intake with ferulic acid 4-O-glucuronide and ferulic acid 4-O-sulfate, potato intake with gentisic acid, and apple and total fruit as well as total vegetable intake with hippuric acid. The positive associations between hippuric acid and apple/total fruit intake were replicated in the DONALD cohort. Background: Accurately measuring fruit and vegetable intake is challenging in epidemiological studies, and this difficulty is even greater in children. Biomarkers of fruit and vegetable intake may enhance objective assessment. Objectives: This exploratory study aimed to screen for potential biomarkers and assess the replicability of previously reported biomarkers associated with fruit and vegetable intake in a free-living population. Methods: Using an untargeted metabolomics approach, we quantified the molecular features in urine from the Identification and Prevention of Dietary- and Lifestyle-induced Health Effects in Children and Infants (IDEFICS) and I.Family cohort. To explore complementary temporal dimensions of diet-metabolite associations, we examined both short-term and habitual fruit and vegetable intake in parallel. Partial least squares and random forest were applied using the MUVR algorithm (Multivariate Methods with Unbiased Variable Selection in R) to identify molecular features related to fruit and vegetable intake. A linear mixed regression model was then fitted to selected features. In addition, the fruit and vegetable intake-metabolites associations were explored in the Dortmund Nutritional and Anthropometric Longitudinal Designed (DONALD) cohort to replicate analyses from the IDEFICS/I.Family cohort. Results: Fruit and vegetable intake were significantly related to 59 features. Ten metabolites could be annotated in the IDEFICS/I.Family cohort. We observed positive associations of total vegetable intake with octenoylcarnitine, total fruit intake with 5-hydroxyindoleacetate and D-pantothenic acid, orange intake with ferulic acid 4-O-glucuronide and ferulic acid 4-O-sulfate, potato intake with gentisic acid, and apple and total fruit as well as total vegetable intake with hippuric acid. The positive associations between hippuric acid and apple/total fruit intake were replicated in the DONALD cohort. Background: Accurately measuring fruit and vegetable intake is challenging in epidemiological studies, and this difficulty is even greater in children. Biomarkers of fruit and vegetable intake may enhance objective assessment. Objectives: This exploratory study aimed to screen for potential biomarkers and assess the replicability of previously reported biomarkers associated with fruit and vegetable intake in a free-living population. Methods: Using an untargeted metabolomics approach, we quantified the molecular features in urine from the Identification and Prevention of Dietary- and Lifestyle-induced Health Effects in Children and Infants (IDEFICS) and I.Family cohort. To explore complementary temporal dimensions of diet-metabolite associations, we examined both short-term and habitual fruit and vegetable intake in parallel. Partial least squares and random forest were applied using the MUVR algorithm (Multivariate Methods with Unbiased Variable Selection in R) to identify molecular features related to fruit and vegetable intake. A linear mixed regression model was then fitted to selected features. In addition, the fruit and vegetable intake-metabolites associations were explored in the Dortmund Nutritional and Anthropometric Longitudinal Designed (DONALD) cohort to replicate analyses from the IDEFICS/I.Family cohort. Results: Fruit and vegetable intake were significantly related to 59 features. Ten metabolites could be annotated in the IDEFICS/I.Family cohort. We observed positive associations of total vegetable intake with octenoylcarnitine, total fruit intake with 5-hydroxyindoleacetate and D-pantothenic acid, orange intake with ferulic acid 4-O-glucuronide and ferulic acid 4-O-sulfate, potato intake with gentisic acid, and apple and total fruit as well as total vegetable intake with hippuric acid. The positive associations between hippuric acid and apple/total fruit intake were replicated in the DONALD cohort. Conclusions: This study shows several plausible urinary biomarkers of fruit and vegetable intake in children and adolescents, with the association between hippuric acid and apple and total fruit intake replicated in an independent cohort, and consistent with previous literature.