Lifetime Trauma Exposure and Accelerated Epigenetic Aging among Midlife Women

Abstract

Objective.

Trauma exposure may be linked to accelerated biological aging. However, studies have largely considered childhood abuse, with limited consideration of lifetime trauma exposure, particularly for women. Further, few studies have considered newer epigenetic clocks, which have enhanced links with health outcomes. Among midlife women, we investigated whether lifetime trauma exposure is associated with older epigenetic age with several generations of clocks. We explored associations between childhood maltreatment and epigenetic age and racial differences in associations between trauma and epigenetic age.

Methods.

Two hundred sixteen women (mean age=59 years; 83% non-Hispanic White, 13% Black, 4% other race/ethnicities) underwent physical measures, questionnaires to assess lifetime trauma exposure, and a blood draw. A subset of 123 women completed childhood maltreatment measures. Extrinsic epigenetic age, GrimAge, Principal-Components based PhenoAge, and DunedinPACE were calculated. Clocks were residualized for age and Z-scored for analysis. Associations between trauma and epigenetic age were estimated in linear regression (covariates race, education, body mass index, estimated cell counts). Interactions by race were tested.

Results.

Relative to women without trauma exposure, those with ≥two lifetime traumas had older epigenetic age [GrimAge: 1: B(SE)=0.15(0.15) p=0.31, 2+: B(SE)=0.39(0.13), p=0.004; DunedinPACE: 1: B(SE)=0.23(0.12), p=0.07, 2+: B(SE)=0.33(0.11), p=0.003]. Childhood sexual abuse was also associated with older epigenetic age [GrimAge: B(SE)=0.56(0.24), p=0.021]. Exploratory models suggested that trauma was related to epigenetic age primarily among Black women.

Conclusions.

Among midlife women, greater lifetime trauma and possibly childhood sexual abuse were associated with older epigenetic age, independent of chronologic age. Black women may be particularly affected.

Introduction

Traumatic events, such as sexual and physical violence, accidents, and disasters, are common globally, with an epidemiological survey across 24 countries (n=68,894) indicating that up to 70% of participants had experienced a traumatic exposure in their lifetime (Kessler et al., 2017). Trauma exposure is related to poor physical health (Sowder et al., 2018; Holman et al., 2008; Scott et al., 2013) and mental health in adults (Liu et al., 2021; Prokopez et al., 2020; Bianchini et al., 2022), including increased risk for mental health disorders, chronic disease, disability, and early mortality (Wolf & Morrison, 2017). This association may be particularly pronounced by middle age among those belonging to a racial or ethnic minority group, particularly for Black women (Geronimus et al., 2006). However, the biological processes linking traumatic events to negative health outcomes remains a central question.

Epigenetic Age Acceleration

One putative mechanism is accelerated biological aging. Biological aging is operationalized as an older biological age, based on specific biological markers, relative to one’s chronological age. An indicator of biological age is epigenetic age, derived from DNA methylation (DNAm) patterns. DNAm is one form of epigenetic regulation of gene expression. It is modulated in part by lifestyle and environmental factors, and is thereby a plausible mechanism through which traumatic stress may be related to poorer health (McCrory et al., 2022). In fact, epigenetic markers are among the most robust predictors of biological age developed to date, particularly second generation clocks (Horvath & Raj, 2018; Erema et al., 2022). They are considered a better proxy of biological aging than markers such as telomere length or other -omic-based predictors (Lim et al., 2022); these clocks continue to undergo ongoing development and optimization (Rutledge et al., 2022).

Methylation of CpG (cytosine-phosphate-Guanine) loci varies as a function of age and several epigenetic “clocks” have been developed to track a person’s biological age as compared to their chronological age. First-generation clocks include those developed by Hannum et al. (2013) and Horvath (2013), which correlate strongly with chronological age. Studies with first-generation clocks show these estimates are associated with all-cause mortality, though weakly with disease outcomes (McCrory et al., 2021). Second-generation clocks overcome the weak correlation of first-generation clocks with morbidity and mortality and include clocks such as GrimAge (Lu et al., 2019) and PhenoAge (Levine et al., 2018). These clocks are more predictive of a variety of adverse health outcomes (e.g., cancer, physical function, Alzheimer’s disease, all cause-mortality) than their first-generation counterparts (McCrory et al., 2022). Further evaluation has yielded principal components (PC)-based clocks. PC-based clocks such as PC-based PhenoAge (PCPhenoAge), were developed to enhance the reliability of the PhenoAge clock by using principal components (i.e., shared systematic variation across many CpG sites) rather than individual CpGs to estimate clock age (Higgins-Chen et al., 2022). This clock has been shown to be superior for estimating aging phenotypes and mortality compared to the original PhenoAge clock (Higgins-Chen et al., 2022). Finally, tracking the “pace of aging” using methylation data, sometimes referred to as a “third generation” clock, is Pace of Aging (DunedinPACE) (Belsky et al., 2022). DunedinPACE was trained to predict longitudinal changes in multi-system biomarkers in a cohort and provides a determination of pace of aging. It predicts mortality, chronic disease risk, and time to cardiovascular disease (CVD) better than first generation clocks and adds predictive power over second-generation clocks (Belsky et al., 2022). Less is known about how trauma exposure relates to these newer epigenetic clocks.

Trauma Exposure and Epigenetic Aging

Emerging evidence suggests that the continuum of stress exposures, including traumatic experiences, may accelerate cellular aging (Dee et al., 2023; Protsenko et al., 2023). This stress spectrum ranges from “cumulative stress” (Harvanek et al., 2021; Zannas et al., 2023) and “stressful life events” (Bourassa et al., 2023) to indicators of more severe stress, such as childhood trauma exposure (Parade et al., 2021; Wolf et al., 2018; Lim et al., 2022), and adult/lifetime trauma exposure (Lim et al., 2022). Prior studies have examined the relationship between childhood maltreatment and accelerated epigenetic aging as measured by first- and second-generation clocks (Harvanek et al., 2024; Colich et al., 2020; Belsky et al., 2017). Fewer studies have assessed adult/lifetime trauma exposures in relation to epigenetic aging (Palma-Gudiel et al., 2020).

Childhood Abuse and Neglect (Maltreatment).

A growing body of literature demonstrates a relationship between childhood trauma and accelerated epigenetic aging (Parade et al., 2021; Wolf et al., 2018; Lim et al., 2022). A meta-analysis by Wolf et al. (2018) found that childhood trauma abuse and neglect as measured by the Childhood Trauma Questionnaire (CTQ) was associated with accelerated DNAm via Hannum but not Horvath’s clock. Most of this work is with first-generation (i.e., Hannum and Horvath) epigenetic clocks and research with newer epigenetic clocks is nascent, with most associations observed with GrimAge (Hamlat et al., 2021; McCrory et al., 2022; Klopack et al., 2022) and mixed results for PhenoAge (Graf et al., 2022; Harvanek et al., 2024). The studies that have considered newer epigenetic clocks have typically examined “early life adversity” (e.g., adverse childhood events, ACEs) (Bourassa et al., 2023; Hamlat et al., 2021; Kim et al., 2023; Klopack et al., 2022) which encompasses a range of exposures such as socioeconomic deprivation and parental dysfunction, rather than childhood maltreatment specifically. Notably, newer epigenetic clocks have the advantage of being more closely related to adverse health outcomes and may shed light on the “biological embedding” of childhood traumatic experiences. Thus, this evidence indicates a link between childhood trauma and epigenetic aging, but more research is needed to establish the robustness of this relationship across newer epigenetic clocks.

Lifetime Trauma Exposure.

Most of the research on trauma and epigenetic aging focuses on childhood, while the existing literature on lifetime trauma is smaller with inconsistent results across epigenetic clocks and a focus on clinical and Veteran populations. For example, the meta-analysis by Wolf and colleagues (2018) concluded that lifetime trauma was not associated with accelerated DNAm as measured by both Horvath and Hannum clocks, but this meta-analysis did not include second-generation clocks. Studies using newer clocks have produced mixed results. For example, Katrinli et al. (2020) found that both greater lifetime trauma burden and current or lifetime PTSD were associated with GrimAge acceleration in a sample of mostly low-income Black adults from the Grady Trauma Project compared to trauma-exposed controls without PTSD. In contrast, McKenna et al. (2021) found no significant association with GrimAge in a smaller subset from the same study. Likewise, Katrinli et al. (2023) found no differences in GrimAge acceleration between combat trauma U.S. Veterans and non-combat exposed Veterans. However, others studying combat-exposed male U.S. Veterans (Tamman et al., 2023) have found that lifetime trauma is associated with older epigenetic age (GrimAge).

While the variability in these findings might reflect differences in how lifetime traumas are measured, most of these studies were conducted in samples with high levels of traumatic events, which may limit the generalizability of findings to community samples with more typical levels of trauma exposure. Thus, the evidence suggests that there may be a relationship between lifetime trauma exposure and faster epigenetic aging as measured by second-generation clocks, although studies are few, show mixed results, have focused on clinical and male Veteran populations, and have not examined this relationship with pace of aging clocks. No epigenetic studies of lifetime trauma have used a comprehensive set of clocks, such as extrinsic epigenetic age acceleration (EEAA), GrimAge, DunedinPACE, and PCPhenoAge in a community sample of women.

Current Study

Considering trauma and epigenetic aging among midlife women is particularly important. Women are more likely than men to experience traumatic events, such as sexual violence, that have a high conditional risk of PTSD (Kessler et al., 2017). Studying women at midlife may be particularly important as midlife includes the menopause transition: a time of rapid biologic change (Levine et al., 2016; Shepherd et al., 2020). Finally, considering racial differences is critical given differences in both exposures and the epigenetic aging among racial groups (Crimmins et al., 2021; Horvath et al., 2016; Roberts et al., 2011). However, no studies have focused on lifetime trauma exposure and epigenetic aging in midlife women, nor have considered racial differences in relationships between lifetime trauma exposure and epigenetic aging among midlife women.

Among a well-characterized sample of midlife women who underwent assessments of trauma exposure and epigenetic aging, we investigated whether exposure to trauma would be associated with older epigenetic age, controlling for chronologic age. We primarily considered lifetime trauma exposure in relation to epigenetic age, and in secondary/exploratory models, childhood abuse and neglect (maltreatment). Further, we explored racial differences in associations between trauma exposure and epigenetic aging.

Methods

The study was conducted among 274 women ages 45–67 years recruited in 2017–2020 for a study of menopause and brain health (MsBrain) (Thurston et al., 2022). MsBrain participants were recruited from the surrounding Pittsburgh, PA community via advertisements, fliers, and mailings to local registries and listservs as well as from a study on menopause and cardiovascular health (MsHeart) (Thurston et al., 2016) that was conducted 5 years earlier among nonsmoking midlife women. MsBrain study exclusion criteria included pregnancy; hysterectomy and/or bilateral oophorectomy; history of stroke/cerebrovascular accident; Parkinson’s disease, chemotherapy; a history of dementia; seizure disorder; brain tumor; active substance abuse (established via urine toxicology screen and interview); history of head trauma with loss of consciousness >60 min; and current use of oral or transdermal systemic estrogen and/or progesterone, selective estrogen receptor modulators, aromatase inhibitors, gabapentin, selective serotonin reuptake inhibitors, or serotonin norepinephrine reuptake inhibitors.

A total of 274 women were enrolled. Of the 274 women, 50 women did not undergo epigenetic testing, yielding a sample of N=224 that entered into any model (lifetime trauma or child maltreatment models) considered here. Women with epigenetic data were less likely to be Black (relative to White) and more likely to have used illicit drugs or have a history of emotional neglect than women without epigenetic data (Supplemental Table 1). An additional eight women did not have lifetime trauma exposure data, yielding a sample of N=216 for these models. In addition, analyses of childhood maltreatment were conducted among the subset of participants (N=123) who were recruited from the MsHeart study and who had epigenetic data, yielding a sample size for these models of N=123. Women with child maltreatment data were more likely to be Black (relative to White) and had a higher BMI than women without these data (Supplemental Table 2).

Design and Procedures

As part of the MsBrain study, participants underwent screening procedures, physical measurements, a medical history interview, questionnaires, and phlebotomy. Five years earlier (at the MsHeart assessment), a subset of MsBrain participants also underwent assessments of childhood maltreatment, described below. Study procedures were reviewed and approved by the University of Pittsburgh Human Research Protection Office, and all participants provided written, informed consent.

Measures

Lifetime trauma exposure

Trauma was assessed via a modified Brief Trauma Questionnaire (BTQ) which was developed for the Nurses Health Study II (Koenen et al., 2009) and adapted from the Brief Trauma Interview (Schnurr, Lunney, Sengupta, & Spiro, 2005). The BTQ is a self-report questionnaire that assesses traumatic events, including car accidents, natural disasters, life threatening illness, being beaten or mugged, sexual assault, death of a child, sexual harassment, threat of injury or violence, or witnessing a severe injury or death. Interrater reliability for the presence of Criterion A1 trauma exposure according to the DSM-IV was high [average kappa=0.70 (range 0.74–1.00) for all events except illness (0.60)] (Koenen et al., 2009) and has demonstrated good convergent validity with participant interviews (Schnurr et al., 2002) and criterion validity (i.e., strongly associated with diagnostic criteria for PTSD and symptom severity) (Lancaster, Melka, and Rodriguez, 2009).

Childhood maltreatment

Childhood abuse and neglect (i.e., childhood maltreatment) were assessed at the MsHeart visit via the 28-item short form of the Child Trauma Questionnaire (CTQ), a validated multi-dimensional scale of childhood abuse and neglect experienced at or before age 18 (Bernstein et al., 1994; Scher, Stein, Asmundson, McCreary, & Forde, 2001). The CTQ has strong test-retest reliability (0.79–0.86), internal consistency (Cronbach alpha=0.74–0.95 across subscales in this investigation), and convergent validity with clinical interview (Bernstein et al., 1994). Respondents rate each item on a 5-point scale ranging from 1 (never) to 5 (very often true). Items are summed to yield scores on 5 subscales (emotional abuse, physical abuse, emotional neglect, physical neglect, and sexual abuse) with scores ranging from 5 to 25. CTQ short form clinical cut points validated by Bernstein et al. (2003) have sensitivity and specificity at 0.85 or above relative to clinical interview (Bernstein et al., 2003). If scoring above the clinical threshold on any one subscale, women are considered to have been exposed to childhood maltreatment.

Epigenetic clocks

Participants underwent phlebotomy after an overnight fast. Whole blood leukocytes were collected in PAXgene tubes and stored at −70˚C until extraction. Samples were extracted using an automated QiaSymphony DNA extractor following standard protocol. The Illumina EPIC array (Illumina, San Diego, CA) with the standard Illumina methylation assay protocol was used to measure the methylation levels. Samples were distributed across the array by individuals blinded to any phenotypic data including trauma exposure. Using standard quality control with the minfi package, no samples or probes were identified as low quality. The beta (β) values were calculated from the intensity of the methylated and unmethylated alleles as the ratio of fluorescent signals. We then performed Noob background correction. Finally, we estimated the various epigenetic clocks through the various packages (DunedinPACE, DunedinPoAm38, PC-clocks-main), and the online age calculator (https://dnamage.genetics.ucla.edu). For our main analysis, we focused on Dunedin PACE, GrimAge, Principal-Components (PC) based PhenoAge (PCPhenoAge), and extrinsic epigenetic age (EEA). EEA used the enhanced version of the Hannum age clock and 3 estimated measures of age-associated blood cells, including naïve cytotoxic T cells, exhausted cytotoxic T cells and plasma B cells (Chen et al., 2016; Soerensen, et al., 2022) and estimates age acceleration by using the difference in biological age from chronological age. One extreme outlier (>5 SD from the mean) for EEA was excluded. We used the PC version of PhenoAge, as for the PhenoAge clock, the PC version shows a marked improvement in reliability compared to PhenoAge, thereby reducing technical noise and improving performance (Higgins-Chen et al., 2022); an incremental improvement in reliability not as apparent for other clocks (e.g., GrimAge) (Higgins-Chen et al., 2022; Borrus et al., 2024).

Additional measures

Demographics, medical history, and health behaviors were assessed by questionnaires/ interview. Race, ethnicity, gender, and sex were self-reported at the study visit. Education was reported years of completed education. Height and weight were measured via standard methods and body mass index (BMI) calculated. Women underwent a urine toxicology screen. Current substance use was exclusionary; lifetime use of amphetamines, opiates, hallucinogens, benzodiazepines, marijuana, and other substances were assessed via questionnaire. PTSD symptoms were assessed via the PTSD Checklist Civilian Version (PCL-C) (Weathers et al., 2001). Depressive symptoms were assessed via the Center for Epidemiologic Studies Depression Scale and categorized according to its clinical cut point of 16 (Weissman et al., 1977).

Statistical analysis

Relationships of trauma to epigenetic clocks were tested in linear regression models. Epigenetic clocks were residualized by age and z-scored for analysis. Covariates were selected in an a priori manner and included race, education, and BMI (log transformed) and estimated cell counts (CD8+ T cells, CD4+ T cells, B cells, natural killer cells, monocytes) based on the Horvath (2013) and Houseman (2012) methods. We also considered granulocytes in these models but their high correlation with other cell types did not allow for their inclusion with adequate model fit. BMI and select cell types (CD8+ T cells, natural killer cells) were log transformed due to skew. To account for multiple testing and reduce the risk of Type I errors, we additionally applied the Bonferroni correction to control the Family-Wise Error Rate at a threshold of 0.05, adjusting for the number of tests conducted in primary models testing associations between trauma exposure and epigenetic age. Secondarily, we considered child maltreatment scores in relation to epigenetic clocks in linear regression models. In additional models, effect modification of relationships by race were tested in multivariable models. The lifetime trauma measure was dichotomized for analyses examining racial differences (i.e., any trauma versus none) due to small cell sizes for these analyses. Stratified models were presented where relevant. These models were limited to Black and White participants given the small number of women of other races or ethnicities. Covariates considered in secondary models were PTSD symptoms, depressive symptoms, smoking, alcohol use, and current or past illicit drug use. All tests were two tailed with α=0.05. Analyses were conducted in SAS v9.4 (Cary, NC).

Results

Participants were on average 59 years old, overweight, and college educated (Table 1). The sample was largely non-Hispanic White (82%) or Black (14%).

Table 1.

Participant characteristics (N=224)

Age, M (SD)	58.98 (4.38)
Race/ethnicity, N (%)
White	183 (81.70)
Black	32 (14.29)
Asian or mixed race/ethnicities	9 (4.02)
Years of education, Median (IQR)	16.00 (14.00, 18.00)
BMI, M (SD)	28.77 (6.88)
Smoking, current, N (%)	3 (1.34)
Alcohol use, current, N (%)
<1/month	81 (36.16)
≤1/week	67 (29.91)
>1/week	76 (33.93)
Illicit drug use history, N (%)	44 (19.64)
PTSD symptoms, M (SD)†	24.14 (7.92)
Trauma exposure, N (%)
0	76 (35.19)
1	55 (25.46)
2+	85 (39.35)
Childhood maltreatment, N (%)*
Any abuse or neglect	57 (47.11)
Physical abuse	28 (22.76)
Emotional abuse	35 (28.46)
Sexual abuse	15 (12.50)
Physical neglect	21 (17.07)
Emotional neglect	26 (21.14)
GrimAge, M (SD)	57.04 (4.48)
DunedinPACE, M (SD)	0.93 (0.13)
PCPhenoAge, M (SD)	55.01 (6.72)
EEA, M (SD)	41.95 (5.91)

^*Note: child maltreatment measures obtained on N=123 participants, and participants could be represented in more than one category

^†PCL-C scores

We considered lifetime trauma exposure in relation to epigenetic age. Women who had experienced two or more traumas had significantly greater epigenetic age (GrimAge) and a faster pace of aging (DunedinPACE), controlling for chronologic age and other covariates, compared to women with no lifetime traumas (Table 2). Statistically significant associations between trauma exposure and epigenetic age were apparent after correction for multiple testing [e.g., 2+ lifetime traumas relative to 0, GrimAge: B(SE)=0.34(0.14), p=0.03; DunedinPACE: B(SE)=0.27(0.12), p=0.02].

Table 2.

Association of lifetime trauma exposure to epigenetic age

	GrimAge		DunedinPACE		PCPhenoAge		EEA
	B	(SE)	B	(SE)	B	(SE)	B	(SE)
Trauma exposure¶
0	ref		ref		ref		ref
1	0.13	(0.16)	0.21	(0.14)	0.03	(0.16)	0.07	(0.18)
2+	0.34	(0.14)**	0.27	(0.12)**	0.03	(0.14)	0.04	(0.15)
Race (White)	−0.42	(0.17)*	−0.69	(0.14)***	−0.74	(0.15)***	0.20	(0.18)
Education (years)	−0.05	(0.02)*	−0.07	(0.02)***	−0.03	(0.02)	0.00	(0.03)
BMI	1.30	(0.28)***	1.92	(0.23)***	1.18	(0.25)***	0.91	(0.30)**
CD8T	−0.08	(0.08)	−0.14	(0.06)*	−0.07	(0.07)	−0.09	(0.08)
CD4T	−5.28	(0.98)***	−4.96	(0.82)***	−6.75	(0.89)***	−6.28	(1.04)***
NK	−0.20	(0.08)*	−0.20	(0;07)**	−0.05	(0.07)	0.08	(0.09)
B cell	−0.38	(2.54)	3.22	(2.12)	−8.07	(2.32)***	−2.46	(2.71)
Mono	0.59	(3.06)	0.04	(2.25)	−3.25	(2.79)	−1.16	(3.26)

^*p<.05

^**p<.01

^***p<.001

^¶p-values for associations between trauma exposure and epigenetic age with Bonferroni correction, GrimAge: 1 vs. 0 lifetime traumas, p=1.00; 2+ vs. 0 traumas, p=0.03; DunedinPACE: 1 vs. 0 lifetime traumas, p=0.52; 2+ vs. 0 traumas: p=0.02 (DunedinPACE).

Covariates: Race, years of education, body mass index, estimated cell counts (CD8+ cells, CD4+ cells, B-cells, natural killer cells (NK), and monocytes (mono).

Note: Epigenetic age variables residualized for age and z-scored

In secondary models, we tested interactions by race. Notably, Black women were somewhat less likely to endorse any lifetime trauma than were White women (Black women: 50%; White women: 67.6%, X²(1, 207)=3.31, p=0.07). We observed interactions between lifetime trauma exposure and race in relation to epigenetic age (for GrimAge: p=0.04; DunedinPACE: p=0.19; EEA: p=.03), with associations between any trauma exposure and epigenetic age only observed among Black women (Figure 1, Supplemental Table 3). We considered childhood maltreatment, which was assessed on a subset of participants (N=123). Women with a history of childhood sexual abuse had significantly greater epigenetic age via GrimAge [B(SE)=0.56 (0.24), p=0.02, covariates race, education, BMI, and estimated cell counts]. We did not observe associations of other forms of maltreatment with epigenetic age (Supplemental Table 4).

Figure 1.

Associations between trauma exposure and GrimAge (A) and DunedinPACE (B) by race

Note: Black women with no trauma, n=14; one or more traumatic events, n=16. White women with no trauma, n=58; one or more traumatic events, n=125; *p<0.05 for trauma/no trauma comparisons stratified by race

In additional secondary models, we considered the further covariate of PTSD symptoms in models of lifetime trauma exposure and epigenetic age in multivariable models; findings were broadly consistent with primary models (Supplemental Table 5, model 1). We also considered covariates of depressive symptoms; drug use history, smoking, and alcohol use; and blood cell counts in associations of lifetime trauma to epigenetic aging; conclusions are consistent with primary models (Supplemental Table 5, models 2–4).

Discussion

This study tested whether a greater number of lifetime traumatic exposures was associated with accelerated biological aging as measured by first- and second- generation epigenetic clocks among a well-characterized cohort of midlife women. We found that women who experienced two or more lifetime traumas had significantly greater epigenetic age (~1.8 years of aging for GrimAge), controlling for chronological age and other covariates. In exploratory models, racial differences were observed, with greater lifetime trauma exposure associated with accelerated epigenetic age largely among Black women, among whom lifetime trauma exposure was associated with 3.6 greater years of epigenetic aging (GrimAge) relative to their non-exposed Black counterparts. Additional secondary models indicated that a history of childhood sexual abuse associated with 2.2 greater years of epigenetic aging (GrimAge). These findings support the relationship between lifetime trauma, childhood trauma, and accelerated epigenetic aging using newer epigenetic clocks.

Our findings are notable given their use of updated epigenetic clocks. Findings are broadly concordant with other studies on trauma exposure and accelerated biological aging (Boks et al., 2015; Carvalho et al., 2022; Roberts et al., 2017) and with earlier findings that lifetime trauma may be related more to second-generation clocks (Katrinli et al., 2020; Tamman et al., 2023; Yang et al., 2021), particularly GrimAge, than with first-generation clocks (Yang et al., 2021). Despite associations with GrimAge, we found no evidence that lifetime trauma was associated with PCPhenoAge, which is an enhanced version of the second-generation clock PhenoAge. Although GrimAge and PhenoAge are both second generation clocks, they differ in how they were trained. For example, the original DNAm PhenoAge was trained on measures of blood-based phenotypic biomarkers associated with mortality (e.g., albumin, glucose, white and red blood cell counts, and chronological age), and GrimAge was trained to predict death based on DNAm surrogates of morbidity- and mortality-related plasma proteins and smoking after adjusting for age in the training model. Future studies of trauma exposure and DNAm should continue to consider multiple generations epigenetic clocks to facilitate our understanding of the specificity of associations of trauma to certain clocks.

We also considered the relationship between childhood maltreatment and accelerated epigenetic aging. We found that women with a history of childhood sexual abuse, but not other forms of traumatic exposure during childhood, had significantly greater epigenetic age as measured by GrimAge. Our findings correspond with other studies of childhood sexual abuse and accelerated epigenetic aging in adulthood (Lawn et al., 2018; Shenk et al., 2022), particularly those using the CTQ to measure childhood maltreatment (Hamlat et al., 2021). The research to date on childhood abuse and epigenetic aging is mixed, likely due in part to variations in childhood trauma measurement method and demographic factors. For example, a meta-analysis by Wolf et al. (2018) found that childhood trauma exposure was associated with Hannum’s clock, but only for studies measuring childhood maltreatment via the CTQ. Demographic factors may also contribute to the heterogeneity in the literature. For example, McCrory et al. (2022) found that childhood sexual abuse was not associated with accelerated epigenetic aging in a nationally representative sample of 8,175 older (mean age=62.3 years) community-dwelling Irish men and women, as measured by GrimAge and Pace of Aging methylation (PoAm). However, Lawn et al. (2018) found that childhood sexual abuse was associated with epigenetic age acceleration in the Avon Longitudinal Study of Parents and Children (ALSPAC) (n=989) as measured by Horvath’s clock (UK sample of women and men). Given our sample was a well-characterized group of midlife American women and used a diverse array of epigenetic clocks, our findings provide unique context to the existing literature on childhood sexual abuse and epigenetic aging regarding demographic factors and allows for comparisons among multiple generations of epigenetic clocks.

In secondary models, we considered differences by race, finding interactions by race for lifetime trauma exposure in relation to epigenetic aging, with associations primarily observed among the Black women in the sample. Our findings are consistent with studies of trauma exposure and epigenetic aging in cohorts comprised mostly of Black individuals (Katrinli et al., 2020; Zannas et al., 2023). Further, other studies have found associations primarily among Black individuals. For example, in a longitudinal cohort study (CARDIA) examining the association between ACEs and epigenetic aging, Kim and colleagues (2023) found an association for increased epigenetic aging by PhenoAge among the Black but not White participants. Some studies have examined potential psychosocial factors that might help to explain this relationship between trauma and accelerated biological/epigenetic aging among Black Americans, including unspoken anger (McKenna et al., 2021), violence exposure (Jovanovic et al., 2017), chronic financial pressure (Simons et al., 2016), and discrimination (Carter et al., 2019; Chae et al., 2020). For example, one study of predominantly low-income Black adults found that lifetime stress exposure was indirectly associated with age acceleration through increases in internalized anger (McKenna et al., 2021). Our findings suggest the particular importance of trauma exposure for epigenetic aging for Black women, and are broadly consistent with the large body of work on the “weathering hypothesis” that finds Black women, regardless of socioeconomic status, are more likely to experience early health deterioration compared to their White female and male counterparts (Geronimus et al., 2006).

Our study includes limitations. Our sample sizes of N=216 for lifetime trauma analyses and particularly N=123 for the childhood trauma exposure analyses, were small and should be interpreted with caution. Multiple secondary models were conducted; these analyses should be interpreted with caution, particularly given the multiple secondary tests performed. The racial and ethnic makeup of our sample was limited, with few non-White women, and no women who were not White or Black, limiting the generalizability of our findings. This study was conducted among a cohort recruited for studies about menopause; study exclusions (e.g., key health conditions, key medications such as major antidepressants, hormone therapy, certain neurological medications) yielded a study sample that likely is healthier and less distressed than the general population. Approximately 65% of our sample the endorsed one or more of the lifetime traumas assessed, which is similar to some estimates from national samples showing rates of 60% (Elliot et al. 2018) but lower than other estimates showing rates such as 89.5% (Kilpatrick et al., 2013) and 82.2% (Sampson et al., 2022). These features of our sample may limit the generalizability of findings to sicker or more traumatized populations, yet it is also notable that associations were observed even in this relatively low risk sample.

In exploratory models, childhood maltreatment was measured via retrospective self-report. Ideally, studies should use both retrospective and prospective approaches to measuring childhood maltreatment (Cecil et al., 2020) given potential limitations in agreement between retrospective and prospective assessment (Baldwin et al., 2019; Cecil et al., 2020). However, such an approach has feasibility limitations, retrospective reports of childhood maltreatment have been linked to adverse health outcomes even in the absence of objective evidence of maltreatment (Danese and Widon, 2020), and both retrospective and prospective assessment are independently associated with self-reported physical and mental health outcomes at midlife (Reuben et al., 2016). Thus, despite differences between prospective and retrospective report of childhood maltreatment, evidence suggests that retrospective reports are informative and related to important adult outcomes. Finally, the number of women scoring positive for a history of sexual abuse were few. Future work should consider assessing this relationship in larger samples with a more racially/ethnically heterogeneous group of women and potential targeted sampling of women with history of trauma.

Our study had several strengths. This included the novel assessment of midlife women in the relationship between trauma exposure and epigenetic aging. Though there have been some studies that examine the relationship between trauma and epigenetic aging in all-female or predominantly female samples, none of these have considered lifetime trauma. Likewise, we used a comprehensive array of DNAm methods to measure accelerated epigenetic aging. This allowed for comparison across first-generation, second-generation, PC-based, and pace of aging measures, as recommended by Levine (2020). We considered racial differences in associations of trauma to epigenetic age.

This study is the first to examine the association between both lifetime and childhood trauma and epigenetic aging in a well-characterized community sample of midlife women, and the first to do so with a diverse array of epigenetic clocks. This study points to the importance of trauma exposure in accelerating biological aging and the potential importance of trauma prevention in promoting women’s physical health as they age.

Public significance:

This study suggests that lifetime trauma exposure is associated with accelerated biological age in midlife women.