Integrated and predictive approach for identifying determinants of health changes: role of nutrition - The DIAPASON project

The objective of the project is to develop accurate and robust markers of the evolution of health status toward metabolic syndrome (MetS), and to determine to what extent nutrition is a determinant, by using a multidisciplinary approach, putting together sociology, epidemiology, nutrition, statistics, and computer science. Within the GAZEL French cohort set up in 1989 (20,000 EDF-GDF employees), the project integrate demographic, socioeconomic, clinical, and biological data (annual questionnaires including food frequency questionnaires (FFQs)) to analyze food trajectories between 1998 & 2009. Metabolic signatures will be characterized using untargeted metabolomics within a Case Control sub-cohort in order to identify early discriminant factors predictive of MetS development. 52-64 y.o. males with high BMI (25≤BMI<30) who developed MetS after the follow up (‘Case’) were selected and compared for some parameters (socio-demographic, clinical, parameters, food habits) with at risk subjects who did not (‘Control’ matched for BMI, age, sex). Correlations between these data and metabolic signatures will be done to build predictive models and determine whether integration of multidimensional parameters improves prediction. Among the 9,000 individuals, 3 dietary patterns were identified from FFQs in 1998 and confirmed from 2004 & 2009 FFQs: Western Diet (WD), Hea -lthy, and Traditional French after Multiple Correspondence Analyses based on 22 food items. Individuals’ food trajectories were different ac-cording to their dietary patterns: decrease of ‘WD’ with time, in opposite to an increase in ‘Healthy’ and ‘Traditional’ behaviours. The sub-cohort appears representative of the whole cohort, but the behaviors’ evolution of Cases and Controls is different in amplitude: a deeper decrease in ‘WD’ in the Cases vs an increase in ‘Traditional’ behavior. This approach should provide new tools to better stratify at risk populations