The association between adverse childhood experiences and epigenetic age acceleration in the Canadian longitudinal study on aging (CLSA)

Abstract Research examining the association between exposure to a wide range of adverse childhood experiences (ACEs) and accelerated biological aging in older adults is limited. The purpose of this study was to examine the association of ACEs, both as a cumulative score and individual forms of adversity, with epigenetic age acceleration assessed using the DNA methylation (DNAm) GrimAge and DNAm PhenoAge epigenetic clocks in middle and older‐aged adults. This cross‐sectional study analyzed baseline and first follow‐up data on 1445 participants aged 45–85 years from the Canadian Longitudinal Study on Aging (CLSA) who provided blood samples for DNAm analysis. ACEs were assessed using a validated self‐reported questionnaire. Epigenetic age acceleration was estimated by regressing each epigenetic clock estimate on chronological age. Cumulative ACEs score was associated with higher DNAm GrimAge acceleration (β: 0.07; 95% CI: 0.02, 0.11) after adjusting for covariates. Childhood exposure to parental separation or divorce (β: 0.06; 95% CI: 0.00, 0.11) and emotional abuse (β: 0.06; 95% CI: 0.00, 0.12) were associated with higher DNAm GrimAge acceleration after adjusting for other adversities and covariates. There was no statistical association between ACEs and DNAm PhenoAge acceleration. Early life adversity may become biologically embedded and lead to premature biological aging, in relation to DNAm GrimAge, which estimates risk of mortality. Strategies that increase awareness of ACEs and promote healthy child development are needed to prevent ACEs.

between accelerated biological aging, where the epigenetic age is higher than the chronological age, and morbidity and mortality (Fransquet et al., 2019;Horvath, 2013;Horvath & Levine, 2015;Levine et al., 2018;. DNA methylation-based estimators, referred to as 'epigenetic clocks', are composite measures that have been developed to measure aspects of biological aging. Epigenetic clocks use specific CpG sites whose DNA methylation levels produce age estimations. Of the first-generation clocks, the original Horvath DNA methylation (DNAm) clock included 353 CpGs across multiple cell and tissue types from children and adults and is strongly correlated with chronological age, while the Hannum DNAm clock is based on 71 CpGs and was trained solely in whole blood samples from adults (Hannum et al., 2013;Horvath, 2013).
More recently, second-generation epigenetic clocks such as the PhenoAge and GrimAge use selected CpGs associated with risk factors for disease and thus incorporated clinical biomarkers of physiological dysregulation Lu et al., 2019). These newer clocks have shown improved accuracy in predicting physical functioning, time-to-cardiovascular disease, time-to-cancer, and time-to-mortality compared to first-generation clocks Lu et al., 2019).
Studies in children and younger adults have shown associations between exposure to ACEs including childhood exposure to sexual abuse, intimate partner violence, and poor parental mental health, and epigenetic age acceleration (Brody et al., 2016;Davis et al., 2017;Lawn et al., 2018;Nelles-McGee et al., 2022;Simons et al., 2016;Zannas et al., 2015). Further, results from a meta-analysis showed that childhood exposure to traumatic stress, broadly defined as the number of different types of traumatic events experienced in childhood including witnessing family violence, sexual abuse, physical abuse, or neglect occurring prior to age 18, was associated with epigenetic age acceleration; with each additional exposure to a new type of trauma associated with a 6month age acceleration. However, this association was only present for the Hannum DNAm Age and not for the Horvath DNAm Age (Wolf et al., 2018).
Although evidence from the literature suggests an association between exposure to childhood adversity and accelerated biological aging, most of this work has mainly examined one or two types of adversity and included socioeconomic factors as measures of ACEs as opposed to a wide range of ACEs including exposure to emotional abuse and neglect (Klopack et al., 2022). Moreover, studies have focused on the Horvath or Hannum DNAm Age algorithms with discrepant findings (Wolf et al., 2018;Zannas et al., 2015). Studies examining the impact of ACEs on the later generation epigenetic clocks in older individuals are limited. The purpose of this study was to examine the association of ACEs, both as a cumulative score as well as individual forms of adversity, with epigenetic age acceleration assessed using the DNAm GrimAge and DNAm PhenoAge in middle and older-aged adults in Canada. We hypothesized that exposure to ACEs would be associated with epigenetic age acceleration in adults.

| RE SULTS
Descriptive characteristics of participants in the study sample are presented in Table 1. The average age of the participants was 63.0 years (SD: 10.3 years), 49.3% were males, majority of the participants had a post-secondary education, and 34.9% reported a total annual income of $100,000 or more. Approximately, twothirds of the participants had experienced at least one ACE, 15.7% had experienced two ACEs, and 12.0% had experienced four or more ACEs. Physical abuse was the most prevalent form of ACE (28.5%), followed by living with a family member with poor mental health (23.5%), emotional abuse (23.2%), and childhood exposure to intimate partner violence (23.0%). More than half of the participants were never smokers, 29.6% participated in the recommended level of physical activity, 84.6% reported consuming high nutrition diet, and 12.7% did not drink alcohol. Overall, 40.2% of participants reported engaging in any two, and almost one-third of participants reported engaging in at least three of four poor health behaviors. The average DNAm GrimAge was 59.1 years (SD: 9.2) and was closer to the chronological age compared to the DNAm PhenoAge (Mean: 46.2, SD: 11.7). Nevertheless, as seen in Figure 1, both epigenetic clocks correlated significantly with chronological age, with a stronger association found for DNAm GrimAge (r = 0.90).
The associations between ACEs and epigenetic clocks in the unadjusted and fully adjusted models are shown in Table 2. For DNAm GrimAge, cumulative ACEs score was significantly associated with faster epigenetic age acceleration (β: 0.07; 95% CI: 0.02, 0.11) after adjusting for covariates. Further, childhood exposure to emotional abuse (β: 0.06; 95% CI: 0.00, 0.12) and parental separation or divorce was positively (β: 0.06; 95% CI: 0.00, 0.11) associated with DNAm GrimAge after adjusting for other adversities and covariates. Regression estimates for many other forms of adversity were elevated but did not reach statistical significance. Apart from emotional abuse, the associations between cumulative ACEs score and individual adversity domains, and epigenetic age acceleration assessed using the DNAm PhenoAge were not statistically significant ( Table 2). Number of poor health behaviors was positively associated with acceleration of epigenetic age measured using both, DNAm GrimAge (β: 0.27; 95% CI: 0.22, 0.32) and DNAm PhenoAge (β: 0.07; 95% CI: 0.02, 0.13) clocks. Further, being a male and having lower annual income was associated with a higher epigenetic age acceleration ( Table 2).
The associations between cumulative ACEs score and individual components (age-adjusted DNAm-based surrogate markers) of the DNAm GrimAge were examined. The results showed a positive association between cumulative ACEs score and DNAm plasminogen activation inhibitor 1 (DNAm PAI-1), DNAm beta-2 microglobulin (DNAm B2M), and DNAm pack-years (DNAm PACKYRS) (Table S1). Further, we explored the associations between cumulative ACEs score and individual ACEs and other epigenetic clocks including Horvath, Hannum, DunedinPoAm, and DunedinPACE. Cumulative ACEs score was positively associated with DunedinPoAm clock, but no statistically significant associations were found for the other three clocks after adjusting for covariates (Tables S2 and S3). We also explored two-way interactions between cumulative ACEs score and number of poor health behaviors, sex, and total income and results were not statistically significant (results not reported).

| DISCUSS ION
The present study investigated the association of ACEs with two measures of epigenetic age acceleration in a population-based sample of middle-aged and older adults. The results showed that exposure to greater number of ACEs was associated with accelerated epigenetic aging, measured using the DNAm GrimAge but not DNAm PhenoAge. When examining individual forms of adversity, childhood exposure to emotional abuse and parental separation or divorce were positively associated with DNAm GrimAge acceleration after adjusting for other adversities and covariates. Further, engaging in greater number of poor health behaviors including physical inactivity, smoking, nutritional risk, and high-risk alcohol consumption was associated with epigenetic age acceleration measured using both, DNAm GrimAge and DNAm PhenoAge clocks.
Our finding that greater number of ACEs is associated with acceleration of biological aging is largely congruent with previous studies. Early life adversity has been associated with biological age acceleration assessed using the Horvath and Hannum DNAm clocks and childhood abuse was associated with acceleration of DNAm GrimAge in younger and middle-aged adults (Hamlat et al., 2021;Wolf et al., 2018;Zannas et al., 2015). Moreover, DNAm GrimAge has been associated with age-related decline in various health outcomes that are also associated with ACEs. In the Irish Longitudinal Study on Ageing, DNAm GrimAge acceleration was associated with walking speed, polypharmacy, frailty, and mortality, factors that are also associated with ACEs, after adjusting for age, sex, socioeconomic, and lifestyle factors (Bellis et al., 2015;McCrory et al., 2021;Mian et al., 2022). In our study, childhood exposure to emotional abuse and parental separation or divorce were associated with a higher DNAm GrimAge acceleration. Estimates for some of the other individual forms of adversity such as childhood exposure to intimate partner violence was elevated, but did not reach statistical significance, likely due to small sample size. Nevertheless, these results provide indication of epigenetic programming from exposure to early life stress. Exposure to ACEs may induce DNAm changes that may be persistent across the life course, especially in the absence of health behavior and lifestyle interventions (Dunn et al., 2019).
Guided by the stress theory, research shows that individuals who have experienced early life adversity have shown increased methylation levels, which are associated with hypothalamic-pituitaryadrenal axis dysregulation, thus suggesting the potential early life origins of adult disease (Liu & Nusslock, 2018).
Further, our results showed that co-occurrence of poor health behaviors was associated with DNAm GrimAge and DNAm PhenoAge acceleration. Studies have shown consumption of vegetables and omega-3 fatty acid intake were negatively associated with epigenetic age, and fat intake, insulin and glucose levels, body mass index, waist-to-hip ratio, and liver fat and visceral adipose tissue volume, smoking, and alcohol consumption were positively associated with epigenetic age (Ecker & Beck, 2019;Zhao et al., 2019). Research has demonstrated epigenetic aging to be associated with increased risk of frailty, time-to-coronary heart disease, time-to-cancer, and time-to-mortality (Li et al., 2020;Lu et al., 2019). In fact, our results showed that cumulative ACEs score was positively associated with individual components of the DNAm GrimAge including DNAm PAI-1, DNAm B2M, and DNAm PACKYRS. PAI-1 is an inhibitor of tissue plasminogen activator and higher PAI-1 levels are associated with inflammation, thrombotic complications, and cardiovascular conditions (Vaughan, 2005). Likewise, B2M is a sensitive biomarker for inflammatory conditions, infections, and cancer, and is positively associated with coronary heart disease, stroke, and mortality (Bethea & Forman, 1990;Shi et al., 2021).

| Study design and participants
The Canadian Longitudinal Study on Aging (CLSA) is a national, longitudinal research platform, which included participants from all 10 Canadian provinces. Using stratified random sampling, 51,338 individuals aged 45-85 years at the time of recruitment (2011)(2012)(2013)(2014)(2015) were recruited from the community. These participants will be followed every 3 years until 2033 or until death or loss to follow-up.
All study participants provided data on the demographic, biological, TA B L E 2 Association between ACEs and epigenetic age acceleration measures

| Adverse childhood experiences (ACEs)
Exposure to ACEs before the age of 16 was assessed using a 14-item self-reported questionnaire that was adapted from the Childhood Experiences of Violence Questionnaire (CEVQ) (Tanaka et al., 2012;Walsh et al., 2008) and the National Longitudinal Study of Adolescent to Adult Health Wave III questionnaire (Harris & Udry, 2022).
Frequency and severity of childhood exposure to physical, emotional, and sexual abuse, neglect, and intimate partner violence were assessed on an ordinal scale with five response options: never, 1-2 times, 3-5 times, 6-10 times, or more than 10 times. Presence or absence of exposure to these five forms of ACEs were based on the CEVQ guidelines (Tanaka et al., 2012). Participants were identified as having experienced physical abuse if they reported being slapped on the face, head or ears, or hit or spanked with something hard 3 or more times; pushed, grabbed, or shoved, or have something thrown with the intention of hurting 3 or more times, or kicked, bit, punched, choked, burned, or physically attacked by an adult 1 or more times (Tanaka et al., 2012). Sexual abuse was identified by asking participants to report if they experienced unwanted touching or if were threatened or forced into unwanted sexual activity (Tanaka et al., 2012). Participants were identified as having experienced emotional abuse if they reported their parents or guardians swearing or saying hurtful or insulting things that made them feel like they were not wanted or loved 3 or more times. Neglect was present if participants reported that their parents, step-parents, or guardians did not take care of their basic needs 1 or more times. Childhood exposure to intimate partner violence was present if participants witnessed verbal abuse in the home between parents or guardians 6 or more times; or if they witnessed physical abuse between parents or guardians 3 or more times (Tanaka et al., 2012). The psychometric properties of the CEVQ have been reported previously (Dube et al., 2004;Tanaka et al., 2012). Additionally, participants were asked to report on 3 other forms of ACEs including whether or not they had experienced parental divorce/separation, parental death/ serious illness, or lived with a family member with mental or psychiatric illness. Participants' responses on each of the 8 individual forms of ACEs were added to calculate a total ACEs score.

| DNA methylation analysis
The methodology for profiling genome-wide DNA methylation in peripheral blood mononuclear cells (PBMCs) in the CLSA participants has been described in the CLSA Data Support Document (David et al., 2020). Briefly, the proportion of methylation on cytoside-  (Lu et al., 2019). In addition, since smoking is an important risk factor for morbidity and mortality, a DNA methylation-based estimator of smoking pack-years was also included (Lu et al., 2019). Age and sex were also included as covariates (Lu et al., 2019). The DNAm PhenoAge was based on phenotypic age score developed from chronological age and nine clinically relevant blood biomarkers including albumin, creatinine, serum glucose, C-reactive protein, lymphocyte percent, mean cell volume, red cell distribution width, alkaline phosphatase, and white blood cell count . The DNAm PhenoAge was trained to predict all-cause mortality, and the DNAm GrimAge was trained to predict time-to-death. In addition, the DNAm age acceleration residuals were also estimated for each participant by regressing the biological clock estimate on chronological age.

| Covariates
Regression analyses were adjusted for sex, total annual household income (<$50,000, $50,000-< $100,000, $100,000-$ < 150,000, $150,000 or more), cigarette smoking (never smoker, smoker), physical activity (adequate activity, low activity), nutritional intake (low risk, high risk), and alcohol consumption (never, occasional/regular drinker). Physical activity was assessed using the Physical Activity Scale for the Elderly (PASE) and dichotomized as meeting the World Health Organization's age specific guidelines for physical activity of at least 150 minutes of moderate-intensity or at least 75 min of vigorous-intensity physical activity per week (Washburn et al., 1999;World Health Organization, 2010).
Nutritional intake was assessed using the "Seniors in the Community: Risk Evaluation for Eating and Nutrition (SCREEN-II)" tool, and a previously established and validated cut-point of <32 was used to identify participants as "high risk" (Keller et al., 2005). In the analysis, the number of poor health behaviors were summed for each participant with the score ranging from 0 to 4. These covariates were identified a priori in the literature for their association with the aging process.

| Statistical analysis
Descriptive statistics including mean and standard deviation (SD) for continuous variables and count and percentage for categorical variables were reported. Pearson correlation was used to examine the correlation between epigenetic clocks and chronological age. A multivariable ordinary least squares linear regression analysis was used to examine the association between cumulative ACEs score and epigenetic age acceleration assessed using DNAm GrimAge and DNAm PhenoAge, after adjusting for the covariates mentioned above. Multivariable linear regression models were also used to examine the association between individual forms of adversity and epigenetic age acceleration for each clock. The associations between cumulative ACEs score and individual components (age-adjusted DNAm-based surrogate markers) of DNAm GrimAge were also examined. We also explored associations between cumulative ACEs score and Horvath, Hannum, DunedinPoAm, and DunedinPACE.
Further, we explored two-way interactions between ACEs and number of poor health behaviors, sex, and total income. Standardized regression estimates, 95% confidence intervals, and p-value for the unadjusted and fully adjusted models were reported. All analyses were performed on SAS version 9.4 for a two-tail test and at a significance level of 0.05.

AUTH O R S ' CO NTR I B UTI O N S
D.J., A.G., and P.R. were involved in the conceptualization and design of the study. D.J. and D.L. conducted the data analyses. D.J. drafted the manuscript. All authors contributed to the interpretation of the data, provided critical revisions of the manuscript, and approved the final version to be published. Health and Preventive Interventions.

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data are available from the Canadian Longitudinal Study on Aging (www.clsa-elcv.ca) for researchers who meet the criteria for access to de-identified CLSA data.