Diet Quality and Epigenetic Aging in the Women's Health Initiative

Abstract

Background: Higher diet quality scores are associated with a lower risk for many chronic diseases and all-cause mortality; however, it is unclear if diet quality is associated with aging biology.

Objective: This study aimed to examine the association between diet quality and a measure of biological aging known as epigenetic aging.

Design: A cross-sectional data analysis was used to examine the association between three diet quality scores based on self-reported food frequency questionnaire data and five measures of epigenetic aging based on DNA methylation (DNAm) data from peripheral blood.

Participants/setting: This study included 4,500 postmenopausal women recruited from multiple sites across the United States (1993-98), aged 50 to 79 years, with food frequency questionnaire and DNAm data available from the Women's Health Initiative baseline visit.

Main outcome measures: Five established epigenetic aging measures were generated from HumanMethylation450 Beadchip DNAm data, including AgeAccelHannum, AgeAccelHorvath, AgeAccelPheno, AgeAccelGrim, and DunedinPACE.

Statistical analyses performed: Linear mixed models were used to test for associations between three diet quality scores (Healthy Eating Index, Dietary Approaches to Stop Hypertension, and alternate Mediterranean diet scores) and epigenetic aging measures, adjusted for age, race and ethnicity, education, tobacco smoking, physical activity, Women's Health Initiative substudy from which DNAm data were obtained, and DNAm-based estimates of leukocyte proportions.

Results: Healthy Eating Index, Dietary Approaches to Stop Hypertension, and alternate Mediterranean diet scores were all inversely associated with AgeAccelPheno, AgeAccelGrim, and DunedinPACE (P < 0.05), with the largest effects with DunedinPACE. A one standard deviation increment in diet quality scores was associated with a decrement (β ± SE) in DunedinPACE z score of -0.097 ± 0.014 (P = 9.70 x 10^-13) for Healthy Eating Index, -0.107 ± 0.014 (P = 1.53 x 10^-14) for Dietary Approaches to Stop Hypertension, and -0.068 ± 0.013 (P = 2.31 x 10^-07) for the alternate Mediterranean diet.

Conclusions: In postmenopausal women, diet quality scores were inversely associated with DNAm-based measures of biological aging, particularly DunedinPACE.

Keywords: Aging; DNA methylation; Diet quality; Women; epigenetic Aging.

Figure 1.

Overview of selection of participants from Women’s Health Initiative (WHI) participants for a study of the association between diet quality and epigenetic aging. Existing DNA methylation data from blood samples collected at baseline were generated by three WHI sub-studies: AS311 (Bladder Cancer and Leukocyte Methylation Study), BAA23 (The Integrative Genomics for Risk of Coronary Heart Disease and Related Phenotypes in WHI cohort study), and EMPC (Epigenetic Mechanisms of Particulate Matter-Mediated Cardiovascular Disease Study).

Figure 2.

Associations between three diet quality scores and five epigenetic aging measures in 4,500 Women’s Health Initiative (WHI) participants. ^a Healthy Eating Index 2015 ^b Dietary Approaches to Stop Hypertension ^c alternate Mediterranean Diet ^d Effect size (Beta) and 95% confidence intervals for decrement epigenetic aging measure (z-score) per one-standard deviation increment in diet quality score, calculated using linear mixed model adjusting for age, race and ethnicity, education, smoking status and pack-years of smoking, physical activity, WHI ancillary study from which the methylation data were obtained (random and fixed effect), and DNA methylation-based estimates of leukocyte proportions

Figure 4.

Associations between diet quality scores and DNA methylation-based surrogate measure components of GrimAge in 4,500 Women’s Health Initiative (WHI) participants. ^a Healthy Eating Index 2015 ^b Dietary Approaches to Stop Hypertension ^c alternate Mediterranean Diet ^d DNAm-based surrogate measure of plasma adrenomedullin ^e DNAm-based surrogate measure of plasma beta-2-microglobulim ^f DNAm-based surrogate measure of plasma cystatin C ^g DNAm-based surrogate measure of plasma growth differentiation factor 15 ^h DNAm-based surrogate measure of plasma leptin ⁱ DNAm-based surrogate measure of cumulative cigarette smoking exposure ^j DNAm-based surrogate measure of plasma plasminogen activator inhibitor 1 ^k DNAm-based surrogate measure of plasma tissue inhibitor metalloproteinases 1 ^l Effect size (beta) and 95% confidence intervals (CI) of DNAm-based surrogate measure components of GrimAge (z-score) per one-standard deviation increment in diet quality score calculated using linear mixed models, adjusting for age, race and ethnicity, education, smoking status and pack-years of smoking, physical activity, WHI ancillary study from which the methylation data were obtained (random and fixed effect), and DNA methylation-based estimates of leukocyte proportions

Figure 5.

Associations between components of diet quality scores and epigenetic aging measures in 4,500 Women’s Health Initiative (WHI) participants, including components from the A) Healthy Eating Index 2015, B) Dietary Approaches to Stop Hypertension, and C) alternate Mediterranean Diet. ^a moderation component (reverse scored) ^b reverse scored component ^c ratio of monounsaturated fatty acids to saturated fatty acids ^d moderate alcohol consumption: ~1 12-oz can of beer, 5oz of wine, or 1.5oz of liquor ^e effect size (beta) and 95% confidence intervals (CI) of epigenetic aging measure (z-score) per one-standard deviation increment in diet quality score component, calculated using linear mixed models, adjusting for age, race and ethnicity, education, smoking status and pack-years of smoking, physical activity, WHI ancillary study from which the methylation data were obtained (random and fixed effect), and DNA methylation-based estimates of leukocyte proportions