An NMR metabolomics approach reveals a combined-biomarkers model in a wine interventional trial with validation in free-living individuals of the PREDIMED study

• Published in: Metabolomics Vol. 11; no. 4; pp. 797 - 806
• Main Authors: Vázquez-Fresno, Rosa, Llorach, Rafael, Urpi-Sarda, Mireia, Khymenets, Olha, Bulló, Mònica, Corella, Dolores, Fitó, Montserrat, Martínez-González, Miguel Angel, Estruch, Ramon, Andres-Lacueva, Cristina
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.08.2015; Springer Nature B.V
• Subjects: Biochemistry; Biomedical and Life Sciences; Biomedicine; Cell Biology; Developmental Biology; Life Sciences; Molecular Medicine; Original Article; Vitaceae; Wine; Biomarkers; NMR; Nutrimetabolomics; Cohort study; Interventional study
• ISSN: 1573-3882; 1573-3890
• DOI: 10.1007/s11306-014-0735-x

The development of robust biomarkers of consumption would improve the classification of participants with regard to their dietary exposure. In addition, validation of them in free-living individuals remains an important challenge. The aim of this study is to assess wine intake biomarkers using an NMR metabolomic approach to measure the utility of these biomarkers in a wine interventional study (WIS, n  = 56) and also to evaluate them in a free-living individuals (PREDIMED study, n  = 91). Nine metabolites showed a significantly higher presence in urinary excretion in WIS after wine intake: five food metabolome metabolites (tartrate, ethyl glucuronide [EtG], 2,3-butanediol, mannitol, and ethanol); one related to the endogenous response to wine exposure (3-methyl-2-oxovalerate) and three unidentified compounds. Receiver operating characteristic (ROC) curve for each single metabolite were evaluated and exhibited areas under the curves (AUC) between 67.4 and 86.3 % when they were evaluated individually. Then, a logistic regression model was fitted to generate a combined-biomarkers model using these metabolites. The model generated which included tartrate-EtG, showed an AUC of 90.7 % in WIS. Similarly, the AUC in the PREDIMED study was 92.4 %. Results showed that a model combining tartrate-EtG is more useful for evaluating exposure to wine than single biomarkers, both in interventional studies and epidemiological data. To our knowledge, this is the first time that a combined-biomarker model using an NMR platform in wine biomarkers’ research has been generated and reproduced in a free-living population.