Background: The global population is rapidly aging, and the demographic shifts exert a substantial burden on the healthcare system and socio-economic growth, underscoring the importance of healthy aging and longevity. Healthy longevity does not simply...
Background: The global population is rapidly aging, and the demographic shifts exert a substantial burden on the healthcare system and socio-economic growth, underscoring the importance of healthy aging and longevity. Healthy longevity does not simply refer to living longer, but rather to extending life free from disease and disability. However, recent increases in the population's life expectancy do not necessarily reflect healthier lives, and evidence on the determinants of healthy longevity remains scarce. Age-related health status denotes inter-individual heterogeneity in aging that cannot be explained by chronological age alone and can be conceptualized as biological age. As an indicator of biological age, epigenetic age, which is derived from DNA methylation in leukocytes or multiple tissues, has received increasing attention as it better reflects age-related diseases and mortality. Multiple lifetime exposures across physical, psychosocial, and physiological domains can induce epigenetic alterations, including DNA methylation, thereby shaping distinct patterns of the complex aging process. However, key determinants across these domains remain understudied, limiting our understanding of pathways toward slower and healthier aging. To date, it remains unclear which health behaviors and clinical indicators are most strongly associated with decelerated epigenetic aging and how these associations may differ by sex. In addition, quality of life (QoL) is a key indicator of subjective well-being, reflecting individual's perceptions of their life circumstances and emotional health within broader social context. However, few studies have examined the association between QoL and epigenetic age acceleration (EAA), particularly in the socioeconomic context. Additionally, while relatively distal metabolic indicators, including hyperglycemia and dyslipidemia, have been widely examined in relation to EAA, more proximal stress factors—such as insulin resistance and hyperinsulinemia—remain understudied. EAA has been related to metabolic syndrome, one of the major age-related diseases; however, its role in the mechanistic pathways linking metabolic stress and incident metabolic syndrome has not been fully elucidated.
To address this research gap, the objectives of our study are as follows. First, we aimed to investigate the collective associations of health behaviors and clinical indicators with EAA and to examine the relative contribution of each component. Second, we aimed to explore the associations of QoL across physical, psychological, social, and environmental domains with EAA, and to assess the potential modifying role of socioeconomic status (SES). Third, we aimed to examine the association between metabolic stress and EAA, as well as the mediating role of EAA in the relationship between metabolic stress traits and the incident metabolic syndrome. Given that these factors—health behaviors, QoL, and metabolic stress indicators—are modifiable at both individual and population levels, elucidating their associations with EAA may offer important insights for public health strategies aimed at slowing biological aging.
Methods: This study draws on epidemiological and epigenomic data from the Korean Genome and Epidemiology Study (KoGES), incorporating two large-scale longitudinal cohorts in KoGES: the Health Examinee (HEXA) study and the Ansan and Ansung study. Participants aged 40 years and older at baseline were enrolled through the national health examinee registry. The HEXA study comprised two survey waves completed by 2016, whereas the Ansan and Ansung study conducted ten survey waves through 2020. Blood-based DNA methylation data were available only from wave 1 of the HEXA study and wave 5 of the Ansan and Ansung study; therefore, these waves were prioritized for analysis. For each study, we additionally selected the wave in which the phenotypic independent variables of interest were available. In the first study, we included the 1,940 participants from the wave 1 of the HEXA study (2004–2013) and the wave 5 of the Ansan and Ansung study (2009–2010), for whom DNA methylation data were available. In the second study, we used two waves of the Ansan and Ansung study: wave 2 (2003–2004) and wave 5 (2009–2010). A total of 1,402 participants who participated in both waves, with available DNA methylation and complete covariate information, were included in the analysis. In the final study, we followed participants over the period from wave 4 (2007–2008) to wave 10 (2019–2020) of the Ansan and Ansung study. We included 1,162 participants in the analytic dataset examining the association between metabolic stress and EAA with complete data. For the mediation analysis, the dataset comprised 518 participants after excluding those with metabolic syndrome at wave 4 or 5 or those who lacked follow-up data for metabolic syndrome.
The independent variables were health behaviors and clinical indicators, QoL, and metabolic stress indicators, respectively, across the three studies. In the first study, health behaviors and clinical indicators were assessed using the American Heart Association (AHA)’s Life’s Essential 8 (LE8), which consists of four health behaviors (healthy diet, healthy sleep, avoidance of nicotine, and engagement in physical activity), and four clinical indicators (body mass index [BMI], blood lipids, blood glucose, and blood pressure). In the second study, the QoL was assessed using the World Health Organization Quality of Life assessment (WHOQOL-BREF), which comprises four domains: physical, psychological, social, and environmental. Lastly, metabolic stress indices included insulin resistance (triglyceride and glucose index [TyG], homeostasis model assessment of insulin resistance [HOMA-IR]), adipose tissue hormone (leptin-to-adiponectin ratio [leptin/adiponectin]), and pancreatic β-cells hormone (C-peptide).
We estimated five EAA (intrinsic EAA [Horvath DNAmAge acceleration], extrinsic EAA [Hannum DNAmAge acceleration], PhenoAge acceleration [PhenoAA], GrimAge2 acceleration [Grim2AA], and Dunedin Pace of Aging Calculated from the Epigenome [DunedinPACE]). EAA was utilized as a dependent variable in all studies, and as a potential mediator between metabolic stress and metabolic syndrome, additionally in the third study. Incident metabolic syndrome was defined as the presence of at least three of five metabolic abnormalities based on Adult Treatment Panel (ATP) III criteria.
Regarding methodologies, in the first study, quantile-based g-computation (QGC) was used to assess the relative contribution of health behaviors and clinical indicators to decelerated epigenetic age, as well as their collective associations. In the second study, we employed multiple linear regression to examine associations between QoL and EAA, and conducted SES-stratified analyses and interaction analyses between QoL and SES to investigate the modifying role of SES. In the third study, we applied multiple linear regression to examine associations between metabolic stress and EAA and mediation analysis with a counterfactual framework to explore the metabolic stress–EAA–incident metabolic syndrome pathway.
Results: In the first study, better LE8 components were collectively associated with decelerated epigenetic aging (ψ ranged from -4.29 to -0.79, depending on the EAA measure). The LE8 components contributing most to decelerated epigenetic aging varied by sex. Among males, nicotine avoidance and better blood glucose showed the greatest contributions to lower EAA. The contribution of nicotine avoidance accounted for 91% and 77% of the overall associations of four health behaviors with lower Grim2AA and DunedinPACE, respectively, whereas better blood glucose accounted for 94% and 86% of the overall associations of four clinical indicators. Among females, physical activity and better blood glucose or BMI were the greatest contributors to lower EAA. Physical activity accounted for 44% of the LE8–Grim2AA association. Better blood glucose explained 54% and 50% of the association with lower Grim2AA and DunedinPACE, respectively. Better BMI contributed 46% to lower PhenoAA.
In the second study, higher physical and psychological QoL were associated with lower EAA across SES groups, while higher social and environmental QoL were associated with elevated EAA, among individuals with low SES (lower income, educational attainment), manual workers, and married individuals. In the social domain, higher QoL was associated with an average of 0.679 years higher Grim2AA among individuals with low income (β = 0.679 for Q2 vs. Q1, P < 0.05) and with an average of 1.001 years higher Grim2AA among individuals with low education (β = 1.001 for Q2 vs. Q1, P < 0.05). Higher environmental QoL was associated with elevated DunedinPACE among individuals with low income (β = 0.036, Q4 vs Q1, P < 0.05) and with higher Grim2AA and DunedinPACE among individuals with low educational attainment (β = 0.826, Q2 vs Q1 for Grim2AA; β = 0.040, Q4 vs Q1 for DunedinPACE; all P < 0.05). The associations between higher social and environmental QoL and elevated EAA were stronger among manual workers than among non-manual workers. Higher environmental QoL was associated with increased EAA among married individuals, but with lower EAA among unmarried individuals.
For the third study, high levels of TyG, HOMA-IR, C-peptide, and leptin/adiponectin were associated with elevated EAA. The associations of TyG and the leptin/adiponectin ratio with incident metabolic syndrome were mediated by DunedinPACE. DunedinPACE mediated 4.78% of the association between higher TyG and metabolic syndrome (P = 0.021). DunedinPACE also mediated 9.71% of the total effect of the high leptin/adiponectin group (P = 0.020) and 6.46% per 1-unit increase in the leptin/adiponectin ratio (P = 0.021) on incident metabolic syndrome.
Conclusion: This study investigated multidimensional factors associated with EAA. We propose that nicotine avoidance, physical activity, and better blood glucose levels or BMI may contribute to slow aging. In addition, our findings suggest that while higher physical and psychological QoL may support slower aging, higher social and environmental QoL that conflicts with socioeconomic constraints may instead accompany accelerated aging. Finally, our results on potential epigenetic aging pathways linking insulin resistance and adipose tissue-related inflammatory response to metabolic syndrome suggest that management of metabolic stress—such as weight and glycemic control—may help mitigate both metabolic syndrome risk and epigenetic aging.
Taken together, we suggest that healthy behaviors, subjective well-being supported by socio economic resource, and improved metabolic function may aid in promoting slower biological aging. These findings underscore the need for multidimensional public health strategies to support healthy longevity by targeting lifestyle modification, psychosocial support, and metabolic health. In the context of global population aging, such approaches may help reduce the growing healthcare and socioeconomic burden.