TY - JOUR

T1 - Dairy consumption, plasma metabolites, and risk of type 2 diabetes

AU - Drouin-Chartier, Jean-Philippe

AU - Hernández-Alonso, Pablo

AU - Guasch-Ferré, Marta

AU - Ruiz-Canela, Miguel

AU - Li, Jun

AU - Wittenbecher, Clemens

AU - Razquin, Cristina

AU - Toledo, Estefanía

AU - Dennis, Courtney

AU - Corella, Dolores

AU - Estruch, Ramon

AU - Fitó, Montserrat

AU - Eliassen, A Heather

AU - Tobias, Deirdre K

AU - Ascherio, Alberto

AU - Mucci, Lorelei A

AU - Rexrode, Kathryn M

AU - Karlson, Elizabeth W

AU - Costenbader, Karen H

AU - Fuchs, Charles S

AU - Liang, Liming

AU - Clish, Clary B

AU - Martínez-González, Miguel A

AU - Salas-Salvadó, Jordi

AU - Hu, Frank B

N1 - © The Author(s) 2021. Published by Oxford University Press on behalf of the American Society for Nutrition.

PY - 2021

Y1 - 2021

N2 - BACKGROUND: Epidemiologic studies have reported a modest inverse association between dairy consumption and the risk of type 2 diabetes (T2D). Whether plasma metabolite profiles associated with dairy consumption reflect this relationship remains unknown.OBJECTIVES: We aimed to identify the plasma metabolites associated with total and specific dairy consumption, and to evaluate the association between the identified multi-metabolite profiles and T2D.METHODS: The discovery population included 1833 participants from the Prevención con Dieta Mediterránea (PREDIMED) trial. The confirmatory cohorts included 1522 PREDIMED participants at year 1 of the trial and 4932 participants from the Nurses' Health Studies (NHS), Nurses' Health Study II (NHSII), and Health Professionals Follow-Up Study US-based cohorts. Dairy consumption was assessed using validated FFQs. Plasma metabolites (n = 385) were profiled using LC-MS. We identified the dairy-related metabolite profiles using elastic net regularized regressions with a 10-fold cross-validation procedure. We evaluated the associations between the metabolite profiles and incident T2D in the discovery and the confirmatory cohorts.RESULTS: Total dairy intake was associated with 38 metabolites. C14:0 sphingomyelin (positive coefficient), C34:0 phosphatidylethanolamine (positive coefficient), and γ-butyrobetaine (negative coefficient) were associated in a directionally similar fashion with total and specific (milk, yogurt, cheese) dairy consumption. The Pearson correlation coefficients between self-reported total dairy intake and predicted total dairy intake based on the corresponding multi-metabolite profile were 0.37 (95% CI, 0.33-0.40) in the discovery cohort and 0.16 (95% CI, 0.13-0.19) in the US confirmatory cohort. After adjusting for T2D risk factors, a higher total dairy intake-related metabolite profile score was associated with a lower T2D risk [HR per 1 SD; discovery cohort: 0.76 (95% CI, 0.63-0.90); US confirmatory cohort: 0.88 (95% CI, 0.78-0.99)].CONCLUSIONS: Total dairy intake was associated with 38 metabolites, including 3 consistently associated with dairy subtypes (C14:0 sphingomyelin, C34:0 phosphatidylethanolamine, γ-butyrobetaine). A score based on the 38 identified metabolites showed an inverse association with T2D risk in Spanish and US populations.

AB - BACKGROUND: Epidemiologic studies have reported a modest inverse association between dairy consumption and the risk of type 2 diabetes (T2D). Whether plasma metabolite profiles associated with dairy consumption reflect this relationship remains unknown.OBJECTIVES: We aimed to identify the plasma metabolites associated with total and specific dairy consumption, and to evaluate the association between the identified multi-metabolite profiles and T2D.METHODS: The discovery population included 1833 participants from the Prevención con Dieta Mediterránea (PREDIMED) trial. The confirmatory cohorts included 1522 PREDIMED participants at year 1 of the trial and 4932 participants from the Nurses' Health Studies (NHS), Nurses' Health Study II (NHSII), and Health Professionals Follow-Up Study US-based cohorts. Dairy consumption was assessed using validated FFQs. Plasma metabolites (n = 385) were profiled using LC-MS. We identified the dairy-related metabolite profiles using elastic net regularized regressions with a 10-fold cross-validation procedure. We evaluated the associations between the metabolite profiles and incident T2D in the discovery and the confirmatory cohorts.RESULTS: Total dairy intake was associated with 38 metabolites. C14:0 sphingomyelin (positive coefficient), C34:0 phosphatidylethanolamine (positive coefficient), and γ-butyrobetaine (negative coefficient) were associated in a directionally similar fashion with total and specific (milk, yogurt, cheese) dairy consumption. The Pearson correlation coefficients between self-reported total dairy intake and predicted total dairy intake based on the corresponding multi-metabolite profile were 0.37 (95% CI, 0.33-0.40) in the discovery cohort and 0.16 (95% CI, 0.13-0.19) in the US confirmatory cohort. After adjusting for T2D risk factors, a higher total dairy intake-related metabolite profile score was associated with a lower T2D risk [HR per 1 SD; discovery cohort: 0.76 (95% CI, 0.63-0.90); US confirmatory cohort: 0.88 (95% CI, 0.78-0.99)].CONCLUSIONS: Total dairy intake was associated with 38 metabolites, including 3 consistently associated with dairy subtypes (C14:0 sphingomyelin, C34:0 phosphatidylethanolamine, γ-butyrobetaine). A score based on the 38 identified metabolites showed an inverse association with T2D risk in Spanish and US populations.

KW - Aged

KW - Animals

KW - Cohort Studies

KW - Dairy Products

KW - Diabetes Mellitus, Type 2/blood

KW - Feeding Behavior

KW - Female

KW - Humans

KW - Male

KW - Middle Aged

KW - Milk

KW - Prospective Studies

KW - Risk Factors

U2 - 10.1093/ajcn/nqab047

DO - 10.1093/ajcn/nqab047

M3 - Journal article

C2 - 33742198

VL - 114

SP - 163

EP - 174

JO - American Journal of Clinical Nutrition

JF - American Journal of Clinical Nutrition

SN - 0002-9165

IS - 1

ER -