Individual and additive effects of vitamin D, omega-3 and exercise on DNA methylation clocks of biological aging in older adults from the DO-HEALTH trial

Abstract

While observational studies and small pilot trials suggest that vitamin D, omega-3 and exercise may slow biological aging, larger clinical trials testing these treatments individually or in combination are lacking. Here, we report the results of a post hoc analysis among 777 participants of the DO-HEALTH trial on the effect of vitamin D (2,000 IU per day) and/or omega-3 (1 g per day) and/or a home exercise program on four next-generation DNA methylation (DNAm) measures of biological aging (PhenoAge, GrimAge, GrimAge2 and DunedinPACE) over 3 years. Omega-3 alone slowed the DNAm clocks PhenoAge, GrimAge2 and DunedinPACE, and all three treatments had additive benefits on PhenoAge. Overall, from baseline to year 3, standardized effects ranged from 0.16 to 0.32 units (2.9-3.8 months). In summary, our trial indicates a small protective effect of omega-3 treatment on slowing biological aging over 3 years across several clocks, with an additive protective effect of omega-3, vitamin D and exercise based on PhenoAge.

Fig. 1. Flowchart of the DO-HEALTH Bio-Age trial in the Swiss subset of DO-HEALTH.

The flowchart shows the allocation of participants across the eight treatment arms.

Fig. 2. Treatment effects of vitamin D, omega-3 and SHEP individually and in combination on changes in DNAm measures from baseline to year 3.

a–d, Treatment effects are expressed as standardized estimates of the change in DNAm measures from baseline to year 3 at the respective 95% CI. For the PhenoAge (a) and GrimAge (b) clocks, we analyzed versions constructed from DNAm PCs, which have superior technical reliability compared to the original versions of these measures. For GrimAge2 (c) and DunedinPACE (d), we used the original versions as they already demonstrate high technical reliability^,,. All analyses were done in samples of n = 777 participants sampled at baseline and the 3-year follow-up, without technical replicates. In a–d, estimates and 95% CIs from analyses of covariance adjusted for chronological age (continuous + spline at 85 years), sex, falls before the study, BMI, study site and the respective baseline biological age measures are shown. The graphs show the main effects under the assumption of additive effects between treatments in the 2 × 2 × 2 factorial trial design; that is, comparing all individuals treated with vitamin D across the eight treatment arms of the trial to those who did not receive vitamin D (same for omega-3 and SHEP compared to control SHEP). Only for GrimAge2 (c) is each treatment arm compared to placebo because of treatment interactions. Consistent effects of omega-3 were observed in decelerating biological aging as measured by PhenoAge (a), GrimAge2 (c) and DunedinPACE (d). PhenoAge additionally showed an additive benefit between treatments, as observed for the clinical outcomes incident prefrailty and incident invasive cancer in the same trial. Detailed findings are shown in Extended Data Table 3. For second-generation clocks, standardized units can be converted to months of age retardation using the following equation: (standardized estimate × s.d. + mean) × 12. Overall, the standardized treatment effects from baseline to year 3 were small, ranging from 0.16 to 0.32 units (2.9–3.8 months).

Fig. 3. Treatment effects of vitamin D, omega-3 and SHEP individually and in combination on changes in DNAm-based surrogate biomarkers of plasma proteins based on GrimAge.

a–g, For the seven DNAm-based surrogate markers of plasma proteins underlying GrimAge, we analyzed versions constructed from DNAm PCs. The seven DNAm-based biomarkers estimate the abundance of GDF-15 (a), PAI-1 (b), TIMP-1 (c), B2M (d), adrenomedullin (ADM; e), leptin (f) and cystatin C (g). All analyses were done in samples of n = 777 participants sampled at baseline and the 3-year follow-up, without technical replicates. In a–g, estimates and 95% CIs from analyses of covariance adjusted for chronological age (continuous + spline at 85 years), sex, falls before the study, BMI, study site and the respective baseline DNAm-based biomarkers are shown. Omega-3 stands out as an individual treatment with a decline in three of the seven DNAm-based surrogate biomarkers of plasma proteins (PAI-1, leptin and TIMP-1). However, the data also show a consistent additive benefit of combining two or all three treatments for several plasma proteins (PAI-1, B2M, TIMP-1 and GDF-15). Detailed findings are shown in Extended Data Table 4.

Extended Data Fig. 1. Second generation clocks, Pace of Aging and GrimAge-based plasma proteins explained.

Overview of life-style factors associated with second generation DNAm biology of aging measures^,, and DNA methylation based surrogate markers of plasma proteins. While DNAm clocks measure biological aging, surrogate proteins are ordinal/quantitative measures, where high values correspond to a higher hazard of mortality^,. For more details and references on the DNAm clocks see Extended Data Table 5. Icons from Noun Project (https://thenounproject.com/).

Extended Data Fig. 2. Correlation of chronological age and biological age for DNAm clocks and Pace of Aging.

All DNAm measures of age correlated with chronological age at baseline in DO-HEALTH including the pan tissue Horvath clock assessed with Spearman’s rank correlation coefficient. (Spearman’s rank correlation coefficient r = 0.56,), Hannum (r = 0.6), PhenoAge (r = 0.6), GrimAge (r = 0.92), and GrimAge2 (r = 0.71). The correlation between DunedinPACE and chronological age was substantially weaker (r = 0.19).

Extended Data Fig. 3. Stratified treatment effects of vitamin D, omega-3 and SHEP (simple home exercise program) individually and in combination on change of DNAm measures from baseline to year 3.

Panels A and B show regression coefficients, and their 95% CI derived from analysis of covariance adjusted for chronological age (continuous + spline at 85 years), sex, falls before study, body mass index, study site, respective baseline biological age measure. All analyses were done in n = 777 participants sampled at baseline and at 3-year follow-up, without technical replicates. The treatment effects in Panel A are stratified by biological sex (male vs. female) and Panel B by chronological age (70-74 vs 75+ years).

Extended Data Fig. 4. Stratified treatment effects of vitamin D, omega-3 and SHEP (simple home exercise program) individually and in combination on change of DNAm measures from baseline to year 3.

Panels A and B show regression coefficients, and their 95% CI derived from analysis of covariance adjusted for chronological age (continuous + spline at 85 years), sex, falls before study, body mass index, study site, respective baseline biological age measure. All analyses were done in n = 777 participants sampled at baseline and at 3-year follow-up, without technical replicates. The treatment effects in Panel A are stratified by BMI (≤25 kg/m2 vs >25 kg/m2) and in Panel B by baseline serum Vitamin D concentrations (<20 ng/mL vs ≥20 ng/mL).

Extended Data Fig. 5. Stratified treatment effects of vitamin D, omega-3 and SHEP (simple home exercise program) individually and in combination on change of DNAm measures from baseline to year 3.

The forest plot shows regression coefficients stratified by serum omega-3 concentration (≤100 ng/mL and >100 ng/mL), and their 95% CI derived from analysis of covariance adjusted for chronological age (continuous + spline at 85 years), sex, falls before study, body mass index, study site, respective baseline biological age measure. All analyses were done in n = 777 participants sampled at baseline and at 3-year follow-up, without technical replicates.