Abstract
Background
Promoting cognitive health and preventing deficits is crucial for improving the population level health and reducing economic burdens. Biological aging, influenced by DNA methylation, plays a key role in predicting cognitive performance and brain aging. A more recent body of literature shows that social engagement, and volunteering in particular, may play an important role in modifying the relationship between epigenetic age acceleration and cognitive performance.
Methods
Using the Harmonized Cognitive Assessment Protocol in the Health and Retirement Study, the current project tests the association between 5 epigenetic clocks (Horvath, Hannum, PhenoAge, GrimAge, DunedinPoAm) and cognitive performance. It also examines whether the relationship between epigenetic clocks and cognitive performance differs by volunteering frequency (ie, effect modification).
Results
Any level of volunteering was associated with better cognitive performance when compared to no volunteering. All DNA methylation clocks were associated with cognitive performance, except for PhenoAge. Evidence of effect modification was present for Horvath, PhenoAge, GrimAge, and DunedinPoAm. For PhenoAge and GrimAge, 1–100 h of volunteering per year mitigated the influenced of accelerated biological age on cognitive performance. For Horvath and DunedinPoAm, the links between epigenetic age acceleration and cognitive performance were less steep for highly engaged volunteers (101+ h per year).
Conclusions
The findings underscore the cognitive benefits of engagement in volunteer activities. They further elucidate the interplay between volunteering frequency and epigenetic aging on cognitive performance. The relationship between epigenetic age and cognitive performance also varies based on the level of volunteering engagement.
Keywords: Cognitive ability, DNA methylation, Epigenetic clock, Volunteerism
It is estimated that 1 in every 5 individuals has cognitive deficits without the dementia diagnosis (1,2). Furthermore, there is a growing concern that the recent pandemic increases the incidence of neuropathologic issues among severely affected individuals (3). The impact of individuals with cognitive impairment extends beyond health, affecting their social life, family relationships, labor force participation, and financial resources (4). Such cognitive deficits and subsequent dementia also pose a significant economic and public health concern as the population aging trend continues. According to the 2020 Center for Disease Control and Prevention data, the estimated cost of caring for and treating people with Alzheimer’s disease, the most common cause of dementia, was $305 billion dollars and this figure will be projected to $1.1 trillion by 2050 (5). In response to these challenges, promoting cognitive health and preventing deficits within communities can establish a social model of cognitive well-being at a population level. This approach not only addresses immediate health needs but also aims to mitigate long-term economic and societal impacts.
Among predictors of cognitive well-being, chronological age has been a commonly known factor associated with cognitive performance (6). However, chronological age is neither a practical nor modifiable measure of cognitive performance because not all individuals age at the same rate (7). The intraindividual variability in the cognitive aging trajectory is determined by the accumulation of cellular damages and susceptibility or resilience to stressors, defined as biological aging (8). Biological aging may encompass the rate of cellular aging, telomere lengths, and changes in epigenome. Among multiple aspects of biological aging, DNA methylation (DNAm) is an epigenetic mechanism by which specific points of the genome are chemically modified via involving the addition of methyl groups at the 5′ position of cytosine rings in CpG sites to produce 5-methyl-cytosine (9). This methylation process has a pivotal role in gene expression and regulation without changing the genetic sequence (7). DNA methylation is the most well-described type of epigenetic regulation and acts as the “epigenetic aging clock” shown by a growing body of research that it plays an important role in cognitive performance and brain aging in general (10,11).
The potential underlying epigenetic mechanism for the linkage between DMAm and cognitive function is that DNAm regulates gene expression that controls synaptic plasticity and protein synthesis, which was found to play a significant role during normal brain development, learning, and cognitive process in general (eg, memory formation and information processing and consolidating) from early development to late life (12,13). DNA methylation is also essential for optimal development of multiple brain regions (13). It is found to interact with other epigenetic mechanisms such as histone modification and nonprotein-coding RNAs and create reciprocal regulatory circuits that involve in the formation of neuronal and glial cells, which affect critical periods of neurological (ie, sensory and motor) plasticity in aging (14).
Using machine learning, algorithms have been developed and improved to capture DNAm levels of multiple CpGs across the genome (15). There are 3 generations of methylation algorithms or “clocks.” These clocks identified methylation patterns that vary by chronological age (15). The first-generation algorithms include the Hannum clock and the Horvath clock (15,16). The second generation of methylation algorithms includes PhenoAge and GrimAge, which added measures of health risk factors, biological process, and health outcomes (16). The DunedinPoAm is the third generation of methylation algorithms and consists of measures of longevity and health (17).
Studies on cognitive performance have identified links between second-generation DNAm epigenetic clocks (eg, PhenoAge and GrimAge) and slower processing speed for older adults, even after controlling for chronological age (18,19). A study using the Health and Retirement Study (HRS) sample of older adults aged 65–98 shows that slower age acceleration in GrimAge is significantly related to faster processing speed in women (20). Another study using urban-dwelling middle-aged sample shows that the first-generation DNAm epigenetic clock Hannum age acceleration, adjusted for immune functioning, was associated with faster decline in visual memory and processing speed in men (21). In sum, the studies show that second-generation clocks generally outperform the first-generation clocks in predicting the cognitive performance outcomes, but this relationship may be modified by other relevant social factors such as demographics, socioeconomic status, and healthy lifestyle factors. Thus, epigenetic clock can be a relevant metric for investigating how social and biological pathways dynamically interact in influencing cognitive performance outcomes.
Research is still at its nascent stages on whether the link between biological aging and cognitive performance is modified by social and behavioral factors. However, this line of inquiry gains credence guided by the literature on how environmental factors including supportive social relationship, healthy lifestyle factors, and nutritional diet can regulate epigenetic aging that might determine the biological aging of cells, tissues, or organs (22,23). Among modifiable factors, prosocial engagement in later life, and specifically volunteer activity, is found to potentially affect both slower rates of biological aging acceleration (16) and better cognitive performance (24–26). A recent study demonstrates that any level of volunteering (eg, low, medium, high) is associated with better cognitive performance and self-rated memory compared to non-volunteering, after adjusting for selection into volunteering (27). Another study shows that volunteering more than 100 h a year was associated with better cognitive performance (measured by Telephone Interview for Cognitive Status, TICS) and via an increase in cognitive activities (eg, writing, reading, training activities) (26). Importantly, they also investigated a possible reverse causality and concluded that changes in cognitive performance were not predictive of subsequent volunteering frequency. These findings suggest that volunteering entails cognitively stimulating activities, intrinsic rewards, and social engagement, all of which promote cognitive performance and well-being (26). Volunteering may hold promise as a population-level lifestyle intervention to preserve cognitive health in older adults.
Volunteering engagement may play an important role in modulating the intricate relationship between epigenetic age acceleration and cognitive performance. This line of inquiry is guided by 3 separate lines of inquiries mentioned above. First, there is a significant amount of intraindividual variations in cognitive performance in later life that cannot be explained by chronological age, and genetic influence on cognitive performance weakens as epigenetic influence (ie, gene–environment interaction) becomes stronger in later life (7,8). Second, volunteering often involves cognitively and socially stimulating activities that promote neural plasticity. Cognitive stimulation has been linked to better maintenance of cognitive function in older adults (28). Tasks associated with frequent volunteering also involve sustained mental effort, such as planning, communication, and problem-solving, which builds and maintains cognitive reserve—the brain’s ability to compensate for aging-related neural damage. This reserve may help buffer the cognitive decline typically associated with accelerated epigenetic aging (29). Third, volunteering fosters a sense of purpose, accomplishment (30), and social belonging that can influence epigenetic regulation by reducing stress-related epigenetic modifications on cognitive performance.
Guided by the theoretical and empirical evidence that underscores the importance of recognizing the adaptive nature of aging and the potential for continued cognitive performance, we hypothesize that: (a) there is an association between the epigenetic clocks and cognitive performance, and (b) the strengths of the relationship between epigenetic clocks and cognitive performance differ by volunteering frequency (ie, effect modification). In particular, we speculate that second- and third-generation clocks, including PhenoAge, GrimAge, and DunedinPACE, which are associated with mortality, morbidity, age-related biomarkers, immune function, and organ integration, would show stronger negative associations with cognitive outcomes compared to others like Horvath and Hannum.
Method
Sample of the study was selected from the Harmonized Cognitive Assessment Protocol (HCAP) in the HRS. The HRS is funded by the National Institute on Aging (U01 AG009740), with supplemental funding from the Social Security Administration, and is conducted by the University of Michigan. The HCAP study is funded by the National Institute on Aging (U01 AG058499). HRS-HCAP sample was randomly selected from the HRS panel respondents aged 65 and above, who completed their 2016 core interview and the venous blood collection prior to eligibility for the HCAP. Of the 4 426 eligible cases, 3 496 completed the HCAP interview between June 2016 and October 2017 with a final response rate of 79% (31). Compared to the biennial HRS, HRS-HCAP has used an expanded battery of neuropsychological tests that had broad credibility and acceptance within the scientific community (31), which comprised of cognitive measures on respondents’ episodic memory, orientation, language, attention/executive functioning, working memory, processing speed, and fluid and crystallized intelligence; and informant (nominated by the respondent) reported symptom perception and functional capacity measures for older adults who were not able to conduct the HCAP interview themselves (31). The informant HCAP measures were not used in the current study due to the focus of the study on cognitive function of older adults before they have Alzheimer’s disease and related dementias. The current study includes the respondents with complete data on cognitive performance and epigenetic clocks, resulting in the final analytic sample of 2 293.
Core Variables
Cognitive function
General cognitive function was used in the current study and was measured by the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) Word List and Praxis, Brave man story from the East Boston Memory Test, Logical Memory from the Wechsler Memory Scale Fourth Edition (WMS-IV), Animal Fluency, and TICS (31). The total score of all these tests were calculated and the total score of the battery ranged from 7 to 175. Higher scores indicated better global cognition.
Biological clocks (DNA methylation measures)
DNA methylation measures were derived from the 2016 HRS Venous Blood Study. We included 2 first-generation, 2 second-generation, and 1 third-generation epigenetic clocks. The first gen clocks, trained on chronological age, include Horvath’s clock from DNAm measured at 353 CpG sites and Hannum’s clock from 71 CpG sites. The second-generation clocks include PhenoAge, developed using 9 phenotypic aging measures and GrimAge, trained on 13 DNAm-based surrogates to predict all-cause mortality. Finally, a third-generation clock, Dunedin Pace of Aging (DunedinPACE) was created based on changes in 18 biomarkers of organ system integrity and health outcomes to approximate pace of aging based on data from the Dunedin cohort over 12 years and is expressed in years of epigenetic aging compared to 1 chronological year. Epigenetic aging measures are available as restricted health data at https://hrsdata.isr.umich.edu/data-products/epigenetic-clocks. More detailed information about epigenetic aging measures is available in Crimmins, Thyagarajan, Levine, Weir, and Faul (32)
Volunteering activity
The frequency of volunteering was assessed by asking HRS participants: “Have you spent any time in the past 12 months doing volunteer work for religious, educational, health-related, or other charitable organizations?” For those who answered yes, a cascading scale was provided on how many hours they volunteered a year: 1–49, about 50, 51–99, about 100, 101–200, about 200, and more than 200 h. For the current analyses, the adjacent categories were combined which resulted in 5 categories: non-volunteers (0); 1–50 h (1), 51–100 h (2), 101–200 h (3), and more than 200 h (4). Guided by the extant literature, we also tested several specifications of volunteering measurement including binary (no volunteering vs any volunteer activity), 3 (0, 1–99, 100+), and the original 8 categories (0, 1–49, 50, 51–99, 100, 101–199, 200, 201+). Though there is not a unified and widely accepted volunteering categories, the final categories were chosen based on 2 grounds. First, the supplementary analyses provided similar patterns regarding the effect modification. Second, literature shows that there might be a potential nonlinear or threshold effects of volunteering and health outcomes (31).
Covariates
As potential confounder, we adjusted for relevant covariates in the analyses, as suggested in the literature (16,33). Demographic factors included age, gender (female, male), race/ethnicity (non-Hispanic White, Black, Hispanic, other races), and marital status (partnered, not partnered). Socioeconomic status was assessed by educational attainment (less than high school, GED or high school diploma, college degree or higher), wealth (household wealth [assets minus debts, inverse hyperbolic sine transformed]), and income (individual earning, inverse hyperbolic sine transformed). Health behaviors included frequency of physical activity (3 times a month or less, once a week, more than once a week, every day), smoking status (currently smoking, currently not smoking), binge drinking (more than 3 drinks per day for men, more than 2 drinks per day for women), and obesity (body mass index lower than 30 kg/m2, body mass index 30 kg/m2 or higher). Health-related factors were assessed by self-rated health (excellent, very good, good, fair, poor), activities or daily living (having any activities of daily living [ADL] limitations, not having ADL limitation), and chronic conditions (an index variable for heart disease, diabetes, cancer, stroke, hypertension). Finally, APoE-4 gene carrier status, which is strongly predictive of Alzheimer’s disease, was categorized as ε2/ε2, ε2/ε3, ε2/ε4, ε3/ε3, ε3/ε4, and ε4/ε4. In the analysis, the presence of any copy of APOE-ε4 allele (yes/no) was used.
Analytic Plans
To estimate age acceleration rate using the DNAm measures, we regressed each epigenetic clock on chronological age and cell types (monocyte, natural killer cell, B cell, T cell) to calculate the residuals, except for DunedinePoAm, which is already indicative of the rate of epigenetic aging. Adjusting for cell types ensures that the biological aging measures are not confounded by individual variation in cell type proportions, which can significantly influence epigenetic patterns (34,35). We also used the principal components version of the clocks for improved reliability (36). The calculated measures represent the average pace of biological aging per year. For the ease of comparison between the clocks, we standardized all the clocks. For the effect modification by volunteering status, we followed the recommendations by Knol and VanderWeele (37). Generalized linear models were used to estimate the relationship between the 5 epigenetic clock measures and cognitive performance within each stratum of volunteering frequency. Non-volunteers served as a reference group. Although the current dataset includes less than 10% missing data with the largest missingness coming from ApoE allele status, we handled missing data on volunteering and covariates using multiple imputation by chained equations (m = 5) using R package, mice.
Several additional analyses were performed to test the robustness of the analytical model. First, sensitivity analyses were conducted using binary cognitive impairment status available in the HCAP dataset. Several additional covariates were considered, such as being underweight, frequency of informal helping, retirement status, and network size. They were then excluded from the final analyses for a more parsimonious model since they did not change the substantive conclusions for either volunteering or epigenetic clocks. All analyses were performed in early 2024 using R, version 4.2.1. The data for the current analyses are available from the Health and Retirement Study website with restricted data user agreement.
Results
Table 1 shows the descriptive statistics of the study variables. The sample included older adults of 75.2 years of age, more females (57%), non-Hispanic Whites (74%), married individuals (60%), and retirees (88%). Most of the respondents reported at least a high school education (82%), no smoking (68%), no binge drinking (96%), no obesity (68%), no heart conditions (65%), and good health (63%). Total wealth (raw values) ranged from −$1 728 200 to $43 500 000. In 2016, about 1 in every 5 individuals volunteered 1–100 h, 7% volunteered 101–200 h, and 6% volunteered 201 h or more. Across 10 different cognitive assessments, the sample respondents reported the average score of 124.9, on a scale of 7–175. Descriptive statistics of the core study variables stratified by volunteering level are presented in Supplementary Table 1. In terms of the epigenetic age acceleration, Figure 1 shows the heatmap of epigenetic age acceleration measures in the study sample. First- and second-generation clocks were quite highly correlated, and DunedinPACE shows moderate yet significant positive correlations with the rest of the clocks.
Table 1.
Descriptive Statistics of Study Variables (n = 2 293)
| Characteristic | Mean (SD) or % |
|---|---|
| Cognitive performance (7–175) | 124.88 (22.32) |
| Volunteering frequency | |
| None | 1 455 (64%) |
| 1–50 h | 299 (13%) |
| 51–100 h | 216 (9.5%) |
| 101–200 h | 168 (7.4%) |
| 201 + h | 140 (6.1%) |
| Horvath (raw values) | 70.05 (8.48) |
| Hannum (raw values) | 59.11 (7.75) |
| PhenoAge (raw values) | 61.94 (8.76) |
| GrimAge (raw values) | 72.43 (7.06) |
| DunedinPoAm | 1.08 (0.09) |
| Age | 75.3 (7.19) |
| Gender | |
| Male | 981 (43%) |
| Female | 1 303 (57%) |
| Race/ethnicity | |
| Non-Hispanic White | 1 682 (73%) |
| Non-Hispanic Black | 311 (14%) |
| Hispanic | 181 (7.9%) |
| Other races | 108 (4.7%) |
| Married/partnered | 1 377 (60%) |
| Education | |
| Less than high school | 390 (17%) |
| High school | 807 (35%) |
| Some college | 535 (23%) |
| College graduate | 551 (24%) |
| Wealth (raw values) | 540 601.8 (1 064 890) |
| Income (raw values) | 6 835 (30 154.1) |
| Physical activity | |
| 3 times a month or less | 1 338 (59%) |
| Once a week | 185 (8.1%) |
| More than once a week | 231 (10%) |
| Every day | 528 (23%) |
| Currently smoking | 185 (8%) |
| Binge drinking | 87 (4%) |
| Obese (BMI > 30 kg/m2) | 725 (32%) |
| Self-rated health | |
| Poor | 157 (6%) |
| Fair | 574 (22%) |
| Good | 930 (36%) |
| Very good | 752 (29%) |
| Excellent | 190 (7%) |
| ADL limitations | |
| No limitation | 2 447 (94%) |
| At least one limitation | 158 (6%) |
| Chronic conditions | |
| None | 1 475 (65%) |
| At least one | 776 (34%) |
| ApoE-4 carrier status | |
| No copy | 1 909 (89%) |
| At least one copy | 226 (11%) |
Notes: ADL = activities of daily living; BMI = body mass index; SD = standard deviation.
Figure 1.
Heatmap of epigenetic age acceleration measures in the HRS-HCAP study.
Figure 2 presents the findings from generalized linear models for all participants. The results show that any level of volunteering was associated with better cognitive performance when compared to no volunteering (1–50; β = 2.33, p < .001, 51–100; β = 1.61, p = .009, 101–200; β = 4.61, p < .001, 201+; β = 2.97, p < .001, Supplementary Table 2). Volunteering was significantly associated with better cognitive performance, with the strongest effects seen at 101–200 h of annual volunteering. It is also noteworthy that an increasing level of volunteering was generally associated with better cognitive performance, suggesting a potential dose-response relationship. For epigenetic clocks, GrimAge showed a negative association with cognitive performance (β = −0.18, p < .01), suggesting that accelerated aging per GrimAge is linked to lower cognitive performance. Other clocks (Horvath, Hannum, PhenoAge, DunedinPACE) did not show significant associations, though Hannum (β = 0.15, p = .054) and PhenoAge (β = −0.08, p = .065) approached significance.
Figure 2.
Generalized linear model predicting cognition (N = 2 293). The analyses account for all the relevant covariates, survey weights, and the other 4 epigenetic clocks. No volunteering serves as a referent group. CI = confidence interval.
Figure 3 shows the associations between the 5 epigenetic clocks and cognitive performance, stratified by volunteering frequency. Supplementary Table 3 shows the findings from this set of analyses. The relationship between epigenetic age acceleration and cognitive performance reveals nuanced patterns that vary by the frequency of volunteering. For first-generation clocks, Horvath and Hannum, both clocks show significant negative associations with cognitive performance at 101–200 h of volunteering (β = −0.32, p < .001 for Horvath, β = −.38, p < .001 for Hannum). This suggests that volunteering at moderate levels might offer less of protective cognitive benefits to individuals experiencing faster biological aging for first-generation clocks, trained on chronological age.
Figure 3.
The associations between epigenetic clocks and cognitive performance, stratified by volunteering status (N = 2 293). The analyses account for all the relevant covariates, survey weights, and the other 4 epigenetic clocks. No volunteering serves as a referent group. The DunedinPACE graph is separately depicted due to the significant changes in scale. CI = confidence interval.
Second-generation clocks, PhenoAge and GrimAge, show interesting insights on the effect modification by volunteering. The link between PhenoAge and cognitive performance is stronger at 101–200 h of volunteering (β = −0.30, p < .001). However, for GrimAge, up to 200 h of volunteering offer a significant protective effect for cognition, in that individuals show better cognitive performances despite their higher epigenetic age acceleration (1–50 h β = 0.41, p < .001; 51–100 β = 0.46, p = .03; 101–200 β = 0.33, p = .04). Similarly, for 1–50 h of volunteering, DunedinPACE shows a significant positive association (β = 8.76, p = .02), indicating faster aging is linked to better cognitive performance in this low-volunteering group.
Discussion
To our best knowledge, this is the first study that examines the interplay between biological age (epigenetic clocks) and social engagement in predicting cognitive performance. The study gains credence since it builds on the past literature documenting the social (16,24,25) and biological (10,11,23) factors associated with cognitive performances in later adulthood. The analyses revealed intriguing associations between volunteering behavior, epigenetic aging, and cognitive performance. The association between epigenetic aging and cognitive performance exhibited nuanced patterns across different epigenetic clocks. Higher GrimAge predicted lower cognitive performance, which is as expected given that GrimAge clock is specifically designed to predict time-to-death (19,35) and is strongly associated with cancer, cardiovascular disease, and Alzheimer’s disease (20). Consistent with previous research, any level of volunteering was associated with better cognitive performance compared to no volunteering, underscoring the cognitive benefits of engagement in volunteer activities (24,25,27,38,39). Volunteering may function as a social intervention that reduces stress and stimulates new learning, social connections, and physical activities (39,40), which assist in maintaining cognitive performance and enhancing cognitive health and older adults. Augmenting social capital and social support, in turn, protects older adults’ cognitive function from decline (27,40).
The effect modification model further elucidates the interplay between volunteering frequency and epigenetic aging on cognitive performance. Evidence of effect modification suggests that the relationship between epigenetic age and cognitive performance varies based on the level of volunteering engagement and the choice of epigenetic clocks. The effect modification by 101–200 h of volunteering was significant for Hannum, Horvath, and PhenoAge, but greater epigenetic age acceleration was associated with worse cognitive performance for this volunteering group. It is important to note that both Horvath and Hannum are designed primarily to predict chronological age, thus offering limited insight into biological and cognitive functioning. PhenoAge predicts morbidity and mortality, and the finding might indicate that while moderate levels of volunteering may provide cognitive and social benefits, the demands of this level of engagement could also exacerbate the effects of biological aging, particularly for individuals who experience greater biological age acceleration.
However, for later-generation clocks such as GrimAge and DunedinPACE, active volunteers exhibit better cognitive performance despite the greater biological age acceleration. This suggests that higher levels of engagement in meaningful activities like volunteering may mitigate some of the cognitive risks associated with biological aging as captured by these clocks. GrimAge and DunedinPACE, which incorporate health-specific markers like inflammation and lifestyle factors, might reflect the benefits of social and cognitive stimulation provided by volunteering. In particular, DunedinPACE predicts the current rate of biological aging and is useful in measuring the effectiveness of interventions aimed at slower aging, such as diet, exercise, and social activities (41,42). These results suggest the potential resilience-building effects of active participation, highlighting that even individuals with accelerated biological aging may preserve or improve cognitive function through structured and socially meaningful activities (40).
Although the study has implications for epigenetic aging and cognitive health in the context of prosocial activities like volunteering, some limitations suggest the need for further research investigations. First, results of the study are constrained to the cross-sectional study design, thus these findings can’t explain the causal pathway of cognitive performance in relation to epigenetic clocks. Moreover, changes in volunteering, cognitive performance, and epigenetic clocks overtime should be taken into consideration. Although research shows that dramatic epigenetic changes are rare in later life (43), changes still occur over time that could have important implications for age-related diseases and cognitive performance. Given that sustained volunteering activity is associated with decelerated epigenetic aging (44), future studies should examine how changes in epigenetic age acceleration affect cognitive performance over time and how volunteering frequency influences this process.
Second, volunteering frequency is based on a single item and does not address the heterogeneity in the types of volunteering, levels of involvement, and longevity of the engagement. These characteristics of volunteering should be considered by longitudinal studies in the context of epigenetic aging and cognitive performance. Similarly, although composite measures provide an overall cognitive performance score, multiple cognitive domains (eg, language, executive function, orientation, and memory) (45) might be differently affected by epigenetic age acceleration and volunteering. Our supplementary analyses show that the 4 cognitive domains are highly correlated with general cognitive performance score, with correlation coefficients ranging from 0.55 to 0.76, future studies should examine the types of volunteering associated with specific cognitive dimensions.
Third, there may be potential selection bias in effect estimates and the generalizability of our finding needs to be considered with caution. Fourth, we cannot rule out biases due to unmeasured confounders. We minimized this concern owing to very extensive data on potential confounding variables in HRS and the robustness of our findings across the moderation models.
Despite the limitations, our study reveals that epigenetic age acceleration affects cognitive performance in older adults, while the relationship might be mitigated by volunteering frequency. These findings underscore the complex interplay between social engagement, biological aging, and cognitive health, emphasizing the importance of tailored interventions to support cognitive well-being in aging populations. Further research is needed to understand the underlying mechanisms and develop effective strategies for promoting cognitive health in older adults.
Supplementary Material
Contributor Information
Seoyoun Kim, Department of Sociology, Texas State University, San Marcos, Texas, USA; Institute for Social Research, University of Michigan, Ann Arbor, Michigan, USA.
Xi Pan, Department of Sociology, Texas State University, San Marcos, Texas, USA.
Roger A Fielding,, (Medical Sciences Section).
Funding
This research was made possible using the data collected by the Harmonized Cognitive Assessment Protocol (HCAP), funded by the National Institute on Aging (U01 AG058499). The parent study, the Health and Retirement Study (HRS) is funded by the National Institute on Aging (U01 AG009740). This study was partly supported by the National Heart, Lung, and Blood Institute (R01HL171806).
Conflict of Interest
None.
Author Contributions
Both S.K. and X.P. were involved in the conceptualization of the study. X.P. curated data for the HCAP and S.K. conducted the formal analyses. S.K. drafted the manuscript. All authors contributed to the interpretation of the data, provided critical revisions of the manuscript, and approved the final version of the manuscript.
References
- 1. Rajan KB, Weuve J, Barnes LL, McAninch EA, Wilson RS, Evans DA.. Population estimate of people with clinical Alzheimer’s disease and mild cognitive impairment in the United States (2020–2060). Alzheimers Dement. 2021;17(12):1966–1975. https://doi.org/ 10.1002/alz.12362 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Manly JJ, Jones RN, Langa KM, et al. Estimating the prevalence of dementia and mild cognitive impairment in the US: the 2016 Health and Retirement Study Harmonized Cognitive Assessment Protocol Project. JAMA Neurol. 2022;79(12):1242–1249. https://doi.org/ 10.1001/jamaneurol.2022.3543 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Hampshire A, Trender W, Chamberlain SR, et al. Cognitive deficits in people who have recovered from COVID-19. EClinicalMedicine. 2021;39:101044. https://doi.org/ 10.1016/j.eclinm.2021.101044 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Amano T, Halvorsen CJ, Kim S, Reynolds A, Scher C, Jia Y.. An outcome-wide analysis of the effects of diagnostic labeling of Alzheimer’s disease and related dementias on social relationships. Alzheimer’s Dementia. 2024;20(3):1614–1626. https://doi.org/ 10.1002/alz.13574 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Alzheimer’s Association. 2020 Alzheimer’s Disease Facts and Figures. Alzheimer’s Impact Movement and Alzheimer’s Association. 2022;16(3):391–460. https://doi.org/ 10.1002/alz.12068. [DOI] [Google Scholar]
- 6. Salthouse TA. When does age-related cognitive decline begin? Neurobiol Aging. 2009;30(4):507–514. https://doi.org/ 10.1016/j.neurobiolaging.2008.09.023 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Horvath S, Raj K.. DNA methylation-based biomarkers and the epigenetic clock theory of ageing. Nat Rev Genet. 2018;19(6):371–384. https://doi.org/ 10.1038/s41576-018-0004-3 [DOI] [PubMed] [Google Scholar]
- 8. Barter JD, Foster TC.. Aging in the Brain: new roles of epigenetics in cognitive decline. Neuroscientist. 2018;24(5):516–525. https://doi.org/ 10.1177/1073858418780971 [DOI] [PubMed] [Google Scholar]
- 9. Cui D, Xu X.. DNA methyltransferases, DNA methylation, and age-associated cognitive function. Int J Mol Sci. 2018;19(5):391–460. https://doi.org/ 10.3390/ijms19051315 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Maity S, Farrell K, Navabpour S, Narayanan SN, Jarome TJ.. Epigenetic mechanisms in memory and cognitive decline associated with aging and Alzheimer’s disease. Int J Mol Sci. 2021;22(22):12280. https://doi.org/ 10.3390/ijms222212280 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Vaccarino V, Huang M, Wang Z, et al. Epigenetic age acceleration and cognitive decline: a twin study. J Gerontol A Biol Sci Med Sci. 2021;76(10):1854–1863. https://doi.org/ 10.1093/gerona/glab047 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Grigorenko EL, Kornilov SA, Naumova OY.. Epigenetic regulation of cognition: a circumscribed review of the field. Dev Psychopathol. 2016;28(4pt2):1285–1304. https://doi.org/ 10.1017/S0954579416000857 [DOI] [PubMed] [Google Scholar]
- 13. Dauncey MJ. Nutrition, the brain and cognitive decline: insights from epigenetics. Eur J Clin Nutr. 2014;68(11):1179–1185. https://doi.org/ 10.1038/ejcn.2014.173 [DOI] [PubMed] [Google Scholar]
- 14. Huffman K. The developing, aging neocortex: how genetics and epigenetics influence early developmental patterning and age-related change. Front Genet. 2012;3:212. https://doi.org/ 10.3389/fgene.2012.00212 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Sugden K, Caspi A, Elliott ML, et al. ; Alzheimer's Disease Neuroimaging Initiative. Association of Pace of Aging measured by blood-based DNA methylation with age-related cognitive impairment and dementia. Neurology. 2022;99(13):e1402–e1413. https://doi.org/ 10.1212/WNL.0000000000200898 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Nakamura JS, Kwok C, Huang A, Strecher VJ, Kim ES, Cole SW.. Reduced epigenetic age in older adults who volunteer. Psychoneuroendocrinology. 2023;148:106000. https://doi.org/ 10.1016/j.psyneuen.2022.106000 [DOI] [PubMed] [Google Scholar]
- 17. Belsky DW, Caspi A, Arseneault L, et al. Quantification of the pace of biological aging in humans through a blood test, the DunedinPoAm DNA methylation algorithm. Elife. 2020;9:e54870. https://doi.org/ 10.7554/eLife.54870 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Maddock J, Castillo-Fernandez J, Wong A, et al. DNA methylation age and physical and cognitive aging. J Gerontol Ser A. 2019;75(3):504–511. https://doi.org/ 10.1093/gerona/glz246 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. McCrory C, Fiorito G, Hernandez B, et al. GrimAge outperforms other epigenetic clocks in the prediction of age-related clinical phenotypes and all-cause mortality. J Gerontol Ser A. 2020;76(5):741–749. https://doi.org/ 10.1093/gerona/glaa286 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. O’Shea DM, Maynard T, Tremont G.. DNA methylation “GrimAge” Acceleration mediates sex/gender differences in verbal memory and processing speed: findings from the Health and Retirement Study. J Gerontol Ser A. 2022;77(12):2402–2412. https://doi.org/ 10.1093/gerona/glac133 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Beydoun MA, Shaked D, Tajuddin SM, Weiss J, Evans MK, Zonderman AB.. Accelerated epigenetic age and cognitive decline among urban-dwelling adults. Neurology. 2020;94(6):e613–e625. https://doi.org/ 10.1212/WNL.0000000000008756 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Rentscher KE, Klopack ET, Crimmins EM, Seeman TE, Cole SW, Carroll JE.. Social relationships and epigenetic aging in older adulthood: results from the Health and Retirement Study. Brain Behav Immun. 2023;114:349–359. https://doi.org/ 10.1016/j.bbi.2023.09.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Penner M, Roth T, Barnes C, Sweatt D.. An epigenetic hypothesis of aging-related cognitive dysfunction. Hypothesis and Theory. Front Aging Neurosci. 2010;2:9. https://doi.org/ 10.3389/fnagi.2010.00009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Infurna FJ, Okun MA, Grimm KJ.. Volunteering is associated with lower risk of cognitive impairment. J Am Geriatr Soc. 2016;64(11):2263–2269. https://doi.org/ 10.1111/jgs.14398 [DOI] [PubMed] [Google Scholar]
- 25. Han SH, Roberts JS, Mutchler JE, Burr JA.. Volunteering, polygenic risk for Alzheimer’s disease, and cognitive functioning among older adults. Soc Sci Med (1982). 2020;253:112970. https://doi.org/ 10.1016/j.socscimed.2020.112970 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Villalonga-Olives E, Majercak KR, Almansa J, Khambaty T.. Longitudinal impact of volunteering on the cognitive functioning of older adults: a secondary analysis from the US Health and Retirement Study. Int J Nurs Sci. 2023;10(3):373–382. https://doi.org/ 10.1016/j.ijnss.2023.06.018 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Kail BL, Carr DC.. More than selection effects: volunteering is associated with benefits in cognitive functioning. J Gerontol B Psychol Sci Soc Sci. 2020;75(8):1741–1746. https://doi.org/ 10.1093/geronb/gbaa101 [DOI] [PubMed] [Google Scholar]
- 28. Guiney H, Machado L.. Volunteering in the community: potential benefits for cognitive aging. J Gerontol Ser B. 2017;73(3):399–408. https://doi.org/ 10.1093/geronb/gbx134 [DOI] [PubMed] [Google Scholar]
- 29. Hugo J, Ganguli M.. Dementia and cognitive impairment: epidemiology, diagnosis, and treatment. Clin Geriatr Med. 2014;30(3):421–442. https://doi.org/ 10.1016/j.cger.2014.04.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Kim E, Nakamura JS, Strecher VJ, Cole SW.. Reduced epigenetic age in older adults with high sense of purpose in life. J Gerontol A Biol Sci Med Sci. 2023;78(7):1092–1099. https://doi.org/ 10.1093/gerona/glad092 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Langa KM, Ryan LH, McCammon RJ, et al. The Health and Retirement study Harmonized Cognitive Assessment Protocol Project: study design and methods. Neuroepidemiology. 2020;54(1):64–74. https://doi.org/ 10.1159/000503004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Crimmins EM, Thyagarajan B, Levine ME, Weir DR, Faul J.. Associations of age, sex, race/ethnicity, and education with 13 epigenetic clocks in a nationally representative U.S. Sample: The Health and Retirement Study. J Gerontol A Biol Sci Med Sci. 2021;76(6):1117–1123. https://doi.org/ 10.1093/gerona/glab016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Kim S, Halvorsen CJ.. Volunteering as an equalizer: a quasi-experimental study using propensity score analysis. Am J Prev Med. 2021;61(5):628–635. https://doi.org/ 10.1016/j.amepre.2021.05.004 [DOI] [PubMed] [Google Scholar]
- 34. Levine ME, Lu AT, Quach A, et al. An epigenetic biomarker of aging for lifespan and healthspan. Aging (Milano). 2018;10(4):573–591. https://doi.org/ 10.18632/aging.101414 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Lu AT, Quach A, Wilson JG, et al. DNA methylation GrimAge strongly predicts lifespan and healthspan. Aging (Milano). 2019;11(2):303–327. https://doi.org/ 10.18632/aging.101684 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Higgins-Chen AT, Thrush KL, Wang Y, et al. A computational solution for bolstering reliability of epigenetic clocks: implications for clinical trials and longitudinal tracking. Nat Aging. 2022;2(7):644–661. https://doi.org/ 10.1038/s43587-022-00248-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Knol MJ, VanderWeele TJ.. Recommendations for presenting analyses of effect modification and interaction. Int J Epidemiol. 2012;41(2):514–520. https://doi.org/ 10.1093/ije/dyr218 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Fried LP, Carlson MC, Freedman M, et al. A social model for health promotion for an aging population: initial evidence on the experience corps model. J Urban Health. 2004;81(1):64–78. https://doi.org/ 10.1093/jurban/jth094 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Proulx CM, Curl AL, Ermer AE.. Longitudinal associations between formal volunteering and cognitive functioning. J Gerontol Ser B. 2017;73(3):522–531. https://doi.org/ 10.1093/geronb/gbx110 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Sharifi S, Babaei Khorzoughi K, Rahmati M.. The relationship between volunteering and cognitive performance in older adults: a systematic review. Geriatr Nurs (New York, N.Y.). 2024;55:89–96. https://doi.org/ 10.1016/j.gerinurse.2023.10.020 [DOI] [PubMed] [Google Scholar]
- 41. Belsky DW, Caspi A, Corcoran DL, et al. DunedinPACE, a DNA methylation biomarker of the pace of aging. eLife. 2022;11:e73420. https://doi.org/ 10.7554/eLife.73420 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Farina MP, Klopack ET, Umberson D, Crimmins EM.. The embodiment of parental death in early life through accelerated epigenetic aging: implications for understanding how parental death before 18 shapes age-related health risk among older adults. SSM Popul Health. 2024;26:101648. https://doi.org/ 10.1016/j.ssmph.2024.101648 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Snir S, Farrell C, Pellegrini M.. Human epigenetic ageing is logarithmic with time across the entire lifespan. Epigenetics. 2019;14(9):912–926. https://doi.org/ 10.1080/15592294.2019.1623634 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Kim S, Halvorsen C, Potter C, Faul J.. Does volunteering reduce epigenetic age acceleration among retired and working older adults? Results from the Health and Retirement Study. Soc Sci Med. 2025;364:117501. https://doi.org/ 10.1016/j.socscimed.2024.117501 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Gross AL, Li C, Briceño EM, et al. Harmonisation of later-life cognitive function across national contexts: results from the Harmonized Cognitive Assessment Protocols. lancet Healthy Longevity. 2023;4(10):e573–e583. https://doi.org/ 10.1016/S2666-7568(23)00170-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.