Abstract
Most studies on dietary polyphenol intake and type 2 diabetes (T2D) risk have focused on total or specific subclasses of polyphenols. Since polyphenols are often consumed simultaneously, the joint effect of an intake of multiple subclasses should be explored. We aimed to identify profiles of the dietary polyphenol subclasses intake associated with T2D. A total of 60,586 women from the Etude Epidémiologique auprès de femmes de l'Education Nationale (E3N) cohort study were followed for 20 years between 1993 and 2014. T2D cases were identified and validated. The individual energy-adjusted daily intakes of 15 subclasses of polyphenols were estimated at baseline using a food frequency questionnaire and the PhenolExplorer database. We used Bayesian profile regression to perform the clustering of the covariates by identifying exposure profiles of polyphenol intakes and, simultaneously, link these to T2D risk by using multivariable Cox regression models. We validated 2740 incident T2D cases during follow-up, and identified 15 distinct clusters with different intake profiles and T2D risk. When compared to the largest cluster (n = 6298 women), higher risks of T2D were observed in three of those clusters, which were composed of women with low or medium intakes of anthocyanins, dihydroflavonols, catechins, flavonols, hydroxybenzoic acids, lignans, and stilbenes. One cluster (n = 4243), characterized by higher intakes of these polyphenol subclasses, exhibited lower T2D risk when compared to the reference cluster. These results highlight the importance of a varied diet of polyphenol-rich foods such as nuts, fruits, and vegetables to prevent T2D risk.