Premium
Identifying Cranberry Juice Consumers with Predictive OPLS‐DA Models of Plasma Metabolome and Validation of Cranberry Juice Intake Biomarkers in a Double‐Blinded, Randomized, Placebo‐Controlled, Cross‐Over Study
Author(s) -
Zhao Shaomin,
Liu Haiyan,
Su Zhihua,
Khoo Christina,
Gu Liwei
Publication year - 2020
Publication title -
molecular nutrition and food research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.495
H-Index - 131
eISSN - 1613-4133
pISSN - 1613-4125
DOI - 10.1002/mnfr.201901242
Subject(s) - metabolome , cranberry juice , placebo , opls , metabolomics , urine , randomized controlled trial , linear discriminant analysis , chemistry , food science , medicine , chromatography , urinary system , statistics , mathematics , pathology , hydrogen bond , alternative medicine , organic chemistry , molecule
Scope Methods to verify cranberry juice consumption are lacking. Predictive multivariate models built upon validated biomarkers may help to verify human consumption of a food using a nutrimetabolomics approach. Methods A 21‐day double‐blinded, randomized, placebo‐controlled, cross‐over study was conducted among healthy young women aged 1829. Plasma was collected at baseline and after 3‐day and 21‐day consumption of cranberry or placebo juice. Plasma metabolome was analyzed using UHPLC coupled with high resolution mass spectrometry. Results 18 discriminant metabolites in positive mode and 18 discriminant metabolites in negative mode from a previous 3‐day open‐label study were re‐discovered in the present blinded study. Predictive orthogonal partial least squares discriminant analysis (OPLS‐DA) models were able to identify cranberry juice consumers over a placebo juice group with 96.9% correction rates after 3‐day consumption in both positive and negative mode. This present study revealed 84 and 109 additional discriminant metabolites in positive and negative mode, respectively. Twelve of them were tentatively identified. Conclusion Cranberry juice consumers were classified with high correction rates using predictive OPLS‐DA models built upon validated plasma biomarkers. Additional biomarkers were tentatively identified. These OPLS‐DA models and biomarkers provided an objective approach to verify participant compliance in future clinical trials.