The effect of polyphenols on DNA methylation-assessed biological age attenuation: the DIRECT PLUS randomized controlled trial

Abstract

Background: Epigenetic age is an estimator of biological age based on DNA methylation; its discrepancy from chronologic age warrants further investigation. We recently reported that greater polyphenol intake benefitted ectopic fats, brain function, and gut microbiota profile, corresponding with elevated urine polyphenols. The effect of polyphenol-rich dietary interventions on biological aging is yet to be determined.

Methods: We calculated different biological aging epigenetic clocks of different generations (Horvath2013, Hannum2013, Li2018, Horvath skin and blood2018, PhenoAge2018, PCGrimAge2022), their corresponding age and intrinsic age accelerations, and DunedinPACE, all based on DNA methylation (Illumina EPIC array; pre-specified secondary outcome) for 256 participants with abdominal obesity or dyslipidemia, before and after the 18-month DIRECT PLUS randomized controlled trial. Three interventions were assigned: healthy dietary guidelines, a Mediterranean (MED) diet, and a polyphenol-rich, low-red/processed meat Green-MED diet. Both MED groups consumed 28 g walnuts/day (+ 440 mg/day polyphenols). The Green-MED group consumed green tea (3-4 cups/day) and Mankai (Wolffia globosa strain) 500-ml green shake (+ 800 mg/day polyphenols). Adherence to the Green-MED diet was assessed by questionnaire and urine polyphenols metabolomics (high-performance liquid chromatography quadrupole time of flight).

Results: Baseline chronological age (51.3 ± 10.6 years) was significantly correlated with all methylation age (mAge) clocks with correlations ranging from 0.83 to 0.95; p < 2.2e - 16 for all. While all interventions did not differ in terms of changes between mAge clocks, greater Green-Med diet adherence was associated with a lower 18-month relative change (i.e., greater mAge attenuation) in Li and Hannum mAge (beta = - 0.41, p = 0.004 and beta = - 0.38, p = 0.03, respectively; multivariate models). Greater Li mAge attenuation (multivariate models adjusted for age, sex, baseline mAge, and weight loss) was mostly affected by higher intake of Mankai (beta = - 1.8; p = 0.061) and green tea (beta = - 1.57; p = 0.0016) and corresponded with elevated urine polyphenols: hydroxytyrosol, tyrosol, and urolithin C (p < 0.05 for all) and urolithin A (p = 0.08), highly common in green plants. Overall, participants undergoing either MED-style diet had ~ 8.9 months favorable difference between the observed and expected Li mAge at the end of the intervention (p = 0.02).

Conclusions: This study showed that MED and green-MED diets with increased polyphenols intake, such as green tea and Mankai, are inversely associated with biological aging. To the best of our knowledge, this is the first clinical trial to indicate a potential link between polyphenol intake, urine polyphenols, and biological aging.

Trial registration: ClinicalTrials.gov, NCT03020186.

Keywords: Epigenetics; Green-MED diet; Methylation age; Tyrosol; Urine metabolomics; Urolithins; Weight loss.

Fig. 1

A flow diagram of the DIRECT PLUS epigenetic study. The first and last participants were enrolled in January and May 2017, respectively. HDG, healthy dietary guidelines; MED, Mediterranean

Fig. 2

a–c Aging clocks pre- and post-intervention. a Correlation between age and different mAge clocks at baseline. b Correlation between age and different mAge clocks post-intervention. c Overlap between available CpGs for each clock (5 selected clocks)

Fig. 3

Green-MED adherence, components, and the association with mAge change. Left: forest plot of GMD adherence score (lower left) and its components (upper left) following 18 months of dietary intervention across groups with between-group differences indicated. Data presented as mean ± SE. Adherence was assessed using a 9-item score, ranging from 0 (non-adherence) to 9 (full adherence). Upper middle: association changes in the corresponding dietary component on the left with relative changes in Li mAge. Lower middle: the association of GMD score and relative change in Li mAge. Upper right: association changes in the corresponding dietary component on the left with relative changes in Hannum mAge. Lower middle: the association of GMD score and relative change in Hannum mAge. Data presented as beta coefficients; multi-variate models adjusted for age, sex, baseline mAge, and weight loss. Red dots = HDG; blue dots MED; green dots = green = MED. *p < 0.05; ^#p < 0.1

Fig. 4

a, b Specific CpGs associated with the GMD score. a Correlations of the change in 217 CpGs from Li et al.’s mAge with GMD score. b Change in cg16290275 across GMD score. GMD score was assessed using a 9-item score, ranging from 0 (non-adherence) to 9 (full adherence). Boxplots represent the median, interquartile range, minimum, and maximum for the GMD score

Fig. 5

Biological aging across the intervention groups in subgroups of health status, men only. Forest plot showing the mean and SE of 18-month change in Li mAge. Data presented as mean and SE across health status and diet subgroups. The presence of DM was defined for participants with baseline fasting plasma glucose levels ≥ 126 mg/dL or hemoglobin-A1c levels ≥ 6.5% or if regularly treated with oral antihyperglycemic medications or exogenous insulin. Liver status was based on MRI-measured live fat, as published before [26], with a cutoff > 5% defining fatty liver. Interactions presented are between the health status and intervention. BMI, body mass index; DM, diabetes mellitus; MS, metabolic syndrome