Second Generation DNA Methylation Age Predicts Cognitive Change in Midlife: The Moderating Role of Childhood Socioeconomic Status

Second Generation DNA Methylation Age Predicts Cognitive Change in Midlife: The Moderating Role of Childhood Socioeconomic Status | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Second Generation DNA Methylation Age Predicts Cognitive Change in Midlife: The Moderating Role of Childhood Socioeconomic Status Sophie A. Bell, Christopher R. Beam, Ebrahim Zandi, Alyssa Kam, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5551592/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 22 Jul, 2025 Read the published version in Aging → Version 1 posted You are reading this latest preprint version Abstract DNA methylation age (DNAmAge) surpasses chronological age in its ability to predict age-related morbidities and mortality. This study analyzed data from 287 middle-aged twins in the Louisville Twin Study (mean age 51.9 years ± 7.03) to investigate the effect of DNAmAge acceleration on change in IQ (ΔIQ) between childhood and midlife, while testing childhood socioeconomic status (SES) as a moderator of the relationship. DNAmAge was estimated with five commonly used algorithms (Horvath, Horvath Skin and Blood, GrimAge, and PhenoAge). A factor analysis of these measures produced a two-factor structure which we identified as first generation and second generation measures. Results of genetically informed, quasi-causal regression models indicated that accelerated second generation DNAmAge predicted more negative ΔIQ from childhood to midlife, after accounting for genetic and environmental confounds shared by twins. The relationship between DNAmAge and ΔIQ was moderated by childhood SES, with a stronger effect observed among twins from low SES backgrounds. Second generation DNAmAge measures trained to estimate phenotypic biological age show promise in their predictive value for cognitive decline in midlife. Our genetically informed twin design suggested that DNAmAge may represent a biological pathway through which early-life socioeconomic disadvantage impacts midlife cognitive health. Biological sciences/Psychology Biological sciences/Genetics/Behavioural genetics Biological sciences/Genetics/Epigenetics Figures Figure 1 Figure 2 Introduction Aging is highly heterogeneous, and chronological age is not a sensitive measure of the unique physiological and developmental processes that occur across the human lifespan. Patterns of DNA methylation (DNAm) across the genome can be used to estimate DNA methylation age (DNAmAge), a family of measures strongly correlated with chronological age but better able to predict morbidities and mortality. Adjusting for chronological age, DNAmAge measures can be residualized to generate measures of acceleration 1 , such that positive residuals reflect faster biological aging, while negative residuals reflect slower, and presumably healthier, aging. Since the advent of the first measures a decade ago, DNAmAge measures reliably correlate with demographic characteristics like education level and socioeconomic status (SES) as well as clinically relevant outcomes including cognitive performance in adulthood 2 , 3 . Predicting cognitive outcomes from DNAmAge Individuals with accelerated DNAmAge have an increased risk of cancer, stroke, heart disease, and neurodegenerative disease 4 – 6 . While cognitive functioning is a crucial facet of both healthy and pathological aging, the link between DNAmAge and cognitive functioning remains unclear, partly due to inconsistencies in the predictive accuracy of various DNAmAge algorithms 7 . First generation DNAmAge measures, such as Horvath, Horvath Skin and Blood, and Hannum DNAmAge, were designed primarily to estimate chronological age with methylation data from cytosine-phosphate-guanine (CpG) sites sensitive to time-dependent changes 1 , 8 , 9 . While two studies found that Horvath DNAmAge was associated with cognitive decline 6 , 10 , several others have found no statistically significant associations between first generation DNAmAge measures and cognitive performance or decline 11 – 14 . In contrast, second generation DNAmAge measures, including PhenoAge and GrimAge, are trained to predict age-related outcomes by integrating DNAm signatures tied to clinical biomarkers, health behaviors (e.g., smoking), and demographic variables 15 , 16 . These algorithms aim to capture phenotypic aging processes beyond chronological age, such as morbidity and mortality risk. Accelerated PhenoAge and GrimAge are consistently associated with lower cognitive performance in middle and older adulthood 11 , 12 , 17 , 18 . However, findings on cognitive decline are mixed. Neither Hillary et al. 18 nor Maddock et al. 12 reported statistically significant results, yet Reed and colleagues 13 found that those with cognitive decline in midlife showed more accelerated GrimAge scores than those without decline. These findings suggest that second generation clocks, developed to more accurately estimate phenotypic biological age, may better index age-related changes in cognitive ability. Early-life exposures and pace of DNAmAge Disadvantaged socioeconomic status (SES) in childhood is an established risk factor for lower cognitive ability across the lifespan and late-life cognitive decline 19 . Recent research has also begun exploring links between childhood adversity and later-life DNA methylation age (DNAmAge). In retrospective accounts of adversity (e.g., poverty, death of parent, alcohol/drug use in family, abuse), only poverty was related to GrimAge acceleration 20 . Childhood SES but not adult SES predicted accelerated DNAmAge in midlife using the Horvath and Hannum clocks, suggesting the methylome, or DNA methylation modifications across the genome, may be particularly vulnerable to external stressors in childhood 21 , 22 . However, findings remain inconsistent: McCrory et al. 23 found no association between childhood SES and Horvath, Hannum, or PhenoAge DNAmAge in adults over 50, and unexpectedly, Faul et al. 11 found that low childhood SES was associated with slower aging on the Horvath and Hannum clocks later in life. It is not yet known whether SES moderates the effects of DNAmAge on cognitive outcomes. Genetics, Epigenetics, and Classical Twin Designs Associations between early life exposures, mid- to late-life outcomes, and DNAmAge have primarily been correlational. One study decomposed the variance of DNAmAge into additive genetic, shared environmental, and nonshared environmental components using a traditional twin model, while two others used pedigree designs to estimate heritability 16 , 24 , 25 . Our group has shown how to use a longitudinal twin design to go beyond estimating heritability while sharpening causal inference between a predictor and outcome 26 . Only two studies have examined the relationship between DNAmAge and cognitive ability while controlling for between-family confounds 10 , 14 . In the Middle-Aged Danish Twin Study, no relationship was observed between the Horvath and Hannum clocks and cognitive change over 10 years in midlife 14 . Vaccarino and colleagues 10 found no associations between Horvath, Hannum, GrimAge, or PhenoAge clocks and baseline cognition in middle-aged male twins. However, there was a statistically significant within-twin pair effect of Horvath DNAmAge on cognitive decline 11 years later. The reason why only Horvath’s original measure predicted cognitive ability is unclear, as initial validation studies of PhenoAge and GrimAge reported associations with cognitive outcomes 15 , 16 . Using the Louisville Twin Study (LTS), we expanded on this limited amount of research examining DNAmAge and cognitive change and sought to clarify how early life environmental characteristics may influence midlife cognitive outcomes. The LTS, initiated in 1957, includes intensive, prospective data on twins’ development from infancy through adolescence and recent re-evaluations in midlife on their cognitive, physical, and psychosocial functioning as well as epigenetic aging 27 . Parental data collected during the twins' childhood further enabled us to examine how early life exposures, here rearing SES, relate to twins’ midlife epigenetic and cognitive aging. The current study used a twin design to investigate causal effects of DNAmAge on change in IQ (ΔIQ) between childhood and midlife, while testing childhood SES as a moderator of the relationship. While observational designs cannot reproduce the causal inference possible with random assignment, twins offer a quasi-experimental method for studying causal processes while controlling for family-level genetic and environmental confounds. If a monozygotic (MZ) twin has a higher DNAmAge and also a greater ΔIQ than their co-twin, the association cannot be due to genetic differences or early environmental exposures, because identical twins reared together are matched for these factors. For this reason, we refer to associations that survive twin-based controls as “quasi-causal” 26 . To address variability in DNAmAge algorithms, we conducted an exploratory factor analysis of five widely used measures. We expected second generation DNAmAge measures to be sensitive to ΔIQ in midlife, as they were developed to capture phenotypic aging processes beyond chronological age. Finally, given recent findings linking early life stressors with midlife DNAmAge, we hypothesized that the relationship between midlife DNAmAge and ΔIQ would depend on differences in childhood SES. Results Descriptive statistics The sample includes 287 individual twins, comprising 60 monozygotic and 41 dizygotic complete pairs and 85 single twins. All participated in the childhood and midlife phases of the LTS (Supplementary Fig. S1 ). Twins were assessed in person using three versions of the Wechsler Intelligence Scale for Children (WISC) and the Wechsler Adult Intelligence Scale (WAIS-IV). We used Wechsler IQ scores as an overall estimate of an individual’s level of global cognitive ability. The average age for WISC was 14.42 (1.93) years, and the average age for WAIS-IV was 51.87 (7.03). Mean childhood IQ was 102.7 (13.4), and mean midlife IQ was 105.4 (13.8) (Table 1). After regressing DNAmAge values on chronological age and adjusting for cell type, Horvath, Horvath Skin and Blood, Hannum, PhenoAge, and GrimAge 1 , 8 , 9 , 15 , 16 had means close to zero (Table 1). MZ twin correlations (0.60–0.74) were approximately double those for DZ twins (0.28–0.37) across most measures, except PhenoAge, where MZ and DZ correlations were more similar (0.56 vs. 0.47). Pairwise clock correlations are in Supplementary Table S1 . Results from standardized univariate twin models decomposing the variance of individual DNAmAge measures show little to no shared environmental variance, C, in Horvath, Horvath Skin and Blood, Hannum, and GrimAge clocks. Because C was estimated at a negative value in these cases initially, C was set to zero and the constrained AE models are presented (Supplementary Table S2). These four DNAmAge measures were substantially heritable, with the additive genetic effects, A, ranging from 57–73%, and non-shared environment, E, accounting for 27–43% of the variance. The ACE decomposition of PhenoAge differed in that we estimated the A at 0.23 (95% CI: [-0.35, 0.80]), C at 0.33 (95% CI: [-0.16, 0.82]), and E at 0.45 (95% CI: [0.28, 0.62]) (Supplementary Table S2). Table 1. Sample Characteristics of Louisville Twins Characteristic Total n (%) Mean (SD) MZ Correlation 1 r (SE) DZ Correlation 2 r (SE) Sociodemographic factors 287 Age, years 51.87 (7.03) Age at last WISC, years 14.42 (1.93) Female Sex 169 (58.89) White 262 (91.29) Non-Hispanic 279 (99.29) Education, 16 or more years 161 (56.09) Childhood SES 3 46.09 (26.02) Smoker 66 (23.00) Cognitive function WISC Full Scale IQ Score 282 102.67 (13.44) 0.88 (0.06) 0.58 (0.14) WAIS Full Scale IQ Score 255 105.45 (13.84) 0.87 (0.07) 0.50 (0.14) DNAmAge 4 281 Horvath 0.00 (3.37) 0.62 (0.11) 0.37 (0.15 Horvath Skin and Blood -0.03 (3.59) 0.62 (0.10) 0.32 (0.15) Hannum -0.01 (3.25) 0.60 (0.11) 0.28 (0.16) PhenoAge -0.09 (4.20) 0.56 (0.11) 0.47 (0.14) GrimAge -0.12 (3.42) 0.74 (0.09) 0.37 (0.15) Gen 1 Composite 0.00 (2.85) 0.62 (0.10) 0.32 (0.15) Gen 2 Composite 0.00 (1.78) 0.71 (0.09) 0.42 (0.15) 1 Monozygotic Twin Pair Correlation 2 Dizygotic Twin Pair Correlation 3 Childhood socioeconomic status (SES) measured using the Duncan Socioeconomic Index 4 DNAmAge is controlled for chronological age and cell composition. Residuals are in years. Exploratory Factor Analysis of DNAm Clocks We conducted an exploratory factor analysis (EFA) using the five DNAmAge measures residualized for age and cell count. Based on Eigenvalues and fit indices, we selected a two-factor model (RMSEA = 0.08; 95% CI: [0.00, 0.20]) (Supplementary Fig. S2). Results from a likelihood ratio test comparing a two-factor model to a one-factor model suggested that the two-factor model fit substantially better (χ 2 = 70.07, df = 4, p < .001). First generation DNAmAge measures (Horvath, Horvath Skin and Blood, and Hannum) loaded on one factor, while second generation measures (PhenoAge and GrimAge) loaded on a second factor. We refer to these factors as “Gen 1” and “Gen 2”, respectively. Factor loadings (p < .05) are displayed in Supplementary Table S3. The factors were correlated r = 0.57. Composite scores were generated for each participant based on the EFA. Gen 1 is the sum of standardized values for Horvath, Horvath Skin and Blood, and Hannum DNAmAge (M = 0.00, SD = 2.85). Gen 2 is the sum of standardized values for GrimAge and PhenoAge DNAmAge (M = 0.00, SD = 1.78). Twin pair correlations for Gen 1 and Gen 2 are in Table 1. Univariate Twin Models of DNAmAge Factors We decomposed the variances of Gen 1 and Gen 2 into A, C, and E components using the classical twin model. Both factors were moderately heritable, with A accounting for 50% of the variance in Gen 1 and 60% of the variance in Gen 2. The variance attributable to the non-shared environment was 42% in Gen 1 and 31% in Gen 2, while shared environment accounted for less than 1% of the variance in both factors. Standardized estimates are in Supplementary Table S2. Phenotypic and “Quasi-Causal” Association Models Following the twin modeling framework outlined by Turkheimer and Harden (2014), we first estimated a “phenotypic association” model with no control for differential regressions between and within twin pairs. We then fit a “quasi-causal” regression model, using the ACE decomposition of the predictor variable to control for family-level genetic and environmental confounds in the individual-level regressions. This sequence (phenotypic association model and quasi-causal regression model) was estimated for Gen 1 then repeated for Gen 2. Figure 1 illustrates the model: adult IQ was regressed on childhood IQ, DNAmAge, and the A and C components of each; DNAmAge was regressed on childhood IQ and its A and C components; SES (a family-level variable) was included as a linear covariate for DNAmAge and IQ variables and as an interaction term with DNAmAge. By including childhood IQ in the model, the conditional effects of SES and DNAmAge can be interpreted as predicting change in IQ from childhood to midlife (ΔIQ). The phenotypic regression coefficient ( b P3 ) is an estimate of the relationship between the DNAmAge and ΔIQ with between-family genetic and environmental confounds controlled, i.e., within a pair of identical twins reared together. Predictors of DNAmAge In the phenotypic association model, there were no statistically significant associations with Gen 1 DNAmAge. All Gen 1 model results are presented in Supplementary Table S4. Accelerated Gen 2 DNAmAge was associated with lower childhood SES ( b= -0.41, SE = 0.13, p < .05) in the phenotypic association model (Table 2 ). Next, we used the quasi-causal regression model to estimate the effects of childhood IQ and SES on DNAmAge, using the ACE decomposition of childhood IQ to control for genetic and environmental family-level confounds. In the quasi-causal analyses, there were similarly no statistically significant predictors of Gen 1 DNAmAge. Childhood SES again predicted Gen 2 DNAmAge ( b= -0.67, SE = 0.19, p < .001) in this model, while childhood IQ did not predict Gen 2 DNAmAge. Genetic and environmental confounds were not significantly different from zero. Gen 2 DNAmAge results from both models are presented in Table 2 . Predictors of Midlife Cognitive Functioning In the phenotypic association models for both Gen 1 and Gen 2, childhood IQ statistically predicted adult IQ. The strong relationship between childhood IQ and adult IQ ( r = 0.77) is also illustrated in Supplementary Figure S3, where adult IQ is regressed on childhood IQ without controlling for the differential regressions between and within twin pairs. Similarly, SES was positively associated with childhood IQ ( b = 0.56, SE = 0.09, p < .001) in both phenotypic association models. Higher childhood SES predicted higher adult IQ in the Gen 1 model ( b = 0.16, SE = 0.07, p < .05), though this association did not meet statistical significance in the Gen 2 model. We found a statistically significant phenotypic association between Gen 2 and ΔIQ ( b P3 = -0.10, SE = 0 .03, p < .05). In both Gen 1 and Gen 2 phenotypic association models, we then tested the interaction between DNAmAge and SES ( b int ) in the prediction of ΔIQ but found that the interaction was not statistically significant. In quasi-causal models, we then regressed midlife IQ on childhood IQ ( b P1 ), DNAmAge ( b P3 ), and the A and C components of both variables. There was a statistically significant phenotypic association between childhood IQ and adult IQ after accounting for genetic and environmental confounds shared by identical twins in both Gen 1 ( b P1 = 0.41, p < .01) and Gen 2 models ( b P1 = 0.48, p < .001). This association explains that over 40% of the variance in adult IQ can be attributed to childhood IQ. Consistent with the phenotypic association results above, accelerated Gen 2 DNAmAge predicted more negative ΔIQ from childhood to midlife ( b = − 0.18, SE = 0.07, p < .05). Genetic and environmental confounds ( b A3 and b C3 ) were not significantly different from zero. The quasi-causal twin model implies that within a pair of MZ twins raised together, the member of the pair with accelerated DNAmAge also declines more in cognitive ability between childhood and midlife. These findings were only apparent when DNAmAge was measured with second generation (Gen 2) DNAmAge measures. The interaction revealed that twins raised in lower SES families showed a more negative relationship between Gen 2 DNAmAge and ΔIQ ( b = 0.07, SE = 0.03, p < .05) (Table 2 ). Figure 2 illustrates the moderating effect of childhood SES by using a median split of our sample for SES and displaying the regression of adult IQ on Gen 2 DNAmAge in both high- and low-SES groups. The effect of DNAmAge on ΔIQ is greater in twins raised in lower SES households, indicating that socioeconomic disadvantage may be an early exposure that amplifies the negative effects of epigenetic aging on cognitive trajectories in midlife. Table 2 Phenotypic and Quasi-Causal Regression Model Results for Gen 2 DNAmAge Gen 2 DNAmAge Childhood IQ Estimate (SE) DNAmAge Estimate (SE) Adult IQ Estimate (SE) Phenotypic Association Model SES b 0.56 (0.09)** b -0.67 (0.20)** b 0.20 (0.11) Childhood IQ b A2 b A1 b C2 b C1 b P2 -0.08 (0.09) b P1 0.48 (0.14)** DNAmAge b A3 b C3 b P3 -0.10 (0.03)* DNAmAge X SES b int 0.06 (0.03)* Variances A 1.15 (0.34) 1.96 (0.31) 0.50 (0.08) C 0.15 (0.32) 0.00 (0.00) † 0.00 (0.00) † E 0.22 (0.04) 0.85 (0.16) 0.24 (0.05) Quasi-Causal Regression Model SES b 0.56 (0.09)** b -0.67 (0.19)* b 0.18 (0.12) Childhood IQ b A2 -0.78 (0.41) b A1 0.16 (0.23) b C2 0.30 (1.42) b C1 0.80 (0.67) b P2 0.38 (0.26) b P1 0.48 (0.14)** DNAmAge b A3 0.13 (0.15) b C3 0.00 (0.00) b P3 -0.18 (0.07)* DNAmAge X SES b int 0.07 (0.03)* Variances A 1.06 (0.29) 1.73 (0.56) 0.43 (0.12) C 0.21 (0.28) 0.00 (0.00) † 0.00 (0.00)† E 0.23 (0.04) 0.80 (0.15) 0.23 (0.05) b P , phenotypic association between predictor and outcome; b A , amount of variance attributable to additive genetic influences; b C , amount of variance attributable to shared environmental experiences; b int , estimate of the interaction effect. Data are presented as unstandardized regression coefficients with standard errors in parentheses. †C variances have been set to zero *p < .05 **p ≤ .001 Discussion DNAmAge is an effective predictor of specific aspects of biological aging, but challenges remain due to the lack of a gold standard DNAmAge measure and variations in the predictive accuracies of various measures. Most studies have focused on associations between DNAmAge and phenotypic aging outcomes, but few have tested whether such associations occur because DNAm causes biological aging as opposed to merely correlating with it. Our study first identified two factors across five DNAmAge measures and then used a longitudinal twin design to improve causal inference between early-life exposures, DNAmAge, and ΔIQ in midlife. As hypothesized, Gen 2 DNAmAge, but not Gen 1, predicted ΔIQ in midlife. This result was consistent with several previous studies showing associations between second generation measures and cognition cross-sectionally 11 , 17 and longitudinally 13 . PhenoAge and GrimAge algorithms both incorporate methylation signatures of clinical markers—PhenoAge from blood-based biomarkers predicting physical function, multimorbidity, and mortality 15 , and GrimAge from plasma proteins and smoking pack years selected for their prediction of time-to-death 16 . These second generation DNAmAge measures have already shown improvements over first generation DNAmAge in predicting physical functioning, morbidity, and lifespan 16 , 28 . Our finding that accelerated Gen 2 DNAmAge predicted more negative ΔIQ suggests that decline in cognitive ability is interwoven with other age-related physiological processes that associate with the methylome. The within-twin pair association between Gen 2 DNAmAge and ΔIQ that withstood controls for between family level confounds suggests that differences in environmental exposures partially explain their association. Two prior twin studies of DNAmAge and cognition had mixed findings 10 , 14 . Our Gen 1 findings were consistent with Starnawska and colleagues 14 who did not find within-twin pair associations between Horvath and Hannum DNAmAge and cognition cross-sectionally or longitudinally. Our results differed from Vaccarino et al. 10 who found no within-pair association between PhenoAge and GrimAge and cognitive decline. Differences in study design may explain this discrepancy: while our sample included both male and female twins, Vaccarino et al. examined only male veterans. Our study also spanned a longer age range, predicting change from childhood and adolescent IQ to midlife, whereas Vaccarino et al. used two testing points within participants’ 50s and 60s. Importantly, the LTS lifespan longitudinal design allowed us to capture change from premorbid assessments of cognitive ability to midlife, a period when subtle, age-related declines may emerge. While some prior studies found that lower childhood SES was associated with accelerated DNAmAge on the Horvath and Hannum clocks 21 , 22 , others detected no significant effect 23 , 29 . Our results supported the hypothesis that early-life SES would differentially affect Gen 1 and Gen 2 DNAmAge. Childhood SES predicted Gen 2 DNAmAge but not Gen 1 DNAmAge in midlife, consistent with findings from McCrory et al. 20 , who found childhood poverty was associated with accelerated GrimAge in 50- to 87-year-olds. Childhood poverty has been found to influence health across the lifespan by limiting access to resources like education and healthcare 30 . However, the interaction effect between DNAmAge and SES in our study offers a novel explanation for one pathway through which early-life stressors become biologically embedded. The relationship between accelerated epigenetic age and cognitive decline was amplified within twin pairs from lower-income households, suggesting that those raised in poorer families are more vulnerable to the cognitive effects associated with the broader physiological aging measured by DNAmAge. To our knowledge, only one previous study has conducted a factor analysis on multiple DNAmAge measures. In a model predicting multimorbidity and activities of daily living, Faul and colleagues 11 found Horvath and Hannum measures loaded on a first factor and PhenoAge, GrimAge, and DunedinPACE loaded on a second factor. Our EFA results aligned with this, identifying a “Gen 1” factor with Horvath, Horvath Skin and Blood, and Hannum clocks, and a “Gen 2” factor with PhenoAge and GrimAge. Although we did not calculate DunedinPACE, a “third generation” measure that estimates rate of decline over nearly twenty years, future studies should compare its sensitivity with Gen 2 in the LTS sample 31 . These results contribute to an understanding of the shared underlying patterns of measurement in “Gen 1” and “Gen 2” DNAmAge. Distinguishing between these two domains is a step towards refining the application of DNAmAge: our findings suggest second generation DNAmAge measures may have greater utility in studies of cognitive aging than first generation DNAmAge. Strengths and Limitations Unlike prior studies described here that relied on retrospective accounts of childhood financial hardship or parental occupation, ours is the first to use a prospective rating of childhood SES (Duncan Index) which may have helped mitigate recall bias and improve reliability. However, the Duncan Index is a crude measure of parental occupation, and adding other socioeconomic indicators could improve future research. We did not test adult SES in this study, given that previous work had suggested childhood SES, but not adult SES, was predictive of DNAmAge. While our sample covered a broad socioeconomic range, it was predominantly white and recruited from one region of the United States, limiting generalizability. Future studies should examine the relationship between second generation DNAmAge and patterns of cognitive ability in larger and more diverse samples. Finally, we used IQ score as a proxy for global cognitive ability and future work should examine relationships between DNAmAge and specific cognitive domains including memory and executive functioning. Materials and Methods Participants Participants were 287 individual twins from the midlife phase of the Louisville Twin Study (LTS). The LTS began as a study of childhood development in multiple birth pairs. Twins born between 1950–1997 in the Louisville metropolitan area were followed and assessed on cognitive ability, physical development, and temperament between 3 months and 15 years of age. Parents of twins provided data on demographics, parenting factors, and home environment. Zygosity was determined through blood typing during childhood. The sample contains same-sex monozygotic (MZ) and dizygotic (DZ) twin pairs, and opposite-sex DZ pairs. The LTS closed down in 2000 after 1,770 individual twins had been recruited 32 . In 2018, follow-up of LTS twins aged 30–65 was initiated 27 , and the LTS reopened in 2019 to begin the first midlife data collection of the 1,770 twins. The study was approved by the University of Louisville Institutional Review Board (19.0989). Participants provided a 50cc whole blood sample for genotyping and DNA methylation assay. Genotyping confirmed zygosity for any unknown twin pairs. The present sample includes 60 MZ pairs, 41 DZ pairs, and 85 individual twins who completed a midlife study visit between 2019–2023. Supplementary Figure S1 illustrates the participant flow, attrition, and retention across study phases. Socioeconomic Status Childhood socioeconomic status (SES) was measured using the Duncan socioeconomic index which assigns scores of 0-100 based on the occupation of the head of the household 33 . Occupational prestige was based on 1950 US Census income data and education level associated with the occupation. The SES of our sample (M = 47.3, SD = 25.7) was consistent with the overall LTS distribution (M = 46.9, SD = 26.9), evenly distributed, and representative of the Louisville, Kentucky area (Davis et al., 2019). Cognitive Ability Measures LTS participants were assessed using validated Wechsler batteries of cognitive functioning. Children and adolescents were administered one of three forms of the Wechsler Intelligence Scale for Children (WISC, WISC-R, WISC-III; 34 – 36 . Adults in the midlife phase were assessed using the Wechsler Adult Intelligence Scale IV (WAIS-IV; 37 . Test scores are age-standardized based on a standardization sample that is roughly representative of the United States. The Wechsler batteries generate a Full-Scale Intelligence Quotient (IQ), which is an overall estimate of an individual’s current level of global cognitive ability, with a mean of 100 and a standard deviation of 15. Children were typically administered cognitive assessments at ages 7, 8, 9, 12, and 15, though not all twins were assessed at each timepoint; this study used the most recently available WISC assessment for each participant, with over 80% collected at age 15 (see Table 1 for more detailed descriptive statistics). DNAmAge Measures Genomic DNA was extracted from whole-blood samples that were collected at the University of Louisville and Norton Healthcare medical campuses. Venous blood was collected in ethylenediaminetetraacetic acid (EDTA) tubes and shipped to the Norris Comprehensive Cancer Institute at the University of Southern California Keck School of Medicine. DNA was extracted with promega Maxwell 16 LEV blood DNA kits, and then treated with bisulfite reagents at the USC Molecular Genomics Core following manufacturer protocol. Methylation was assayed with the Illumina Infinium Human MethylationEPIC BeadChip (Illumina, San Diego, CA, USA) at the USC Molecular Genomics Core at the Norris Cancer Institute at the Keck School of Medicine. Quality control was conducted in R 4.3.0 38 using the minfi package 39 to identify aberrant samples and CpG sites, remove cross-reactive probes, conduct background correction, and adjust for batch effects prior to estimating DNAmAge variables. Background correction was performed using the normal-exponential out-of-band (noob) method 40 . The ComBat method was used to adjust for laboratory batches 41 . We estimated blood cell composition using the Houseman method 42 and obtained percentages of CD8 + T cells, CD4 + T cells, natural killer cells (NK), B cells, monocytes, and granulocytes. Cell proportions are highly correlated, so in the current analysis we included the first and second principal components from a principal components (PC) analysis (collectively, they explained 86.12% of the variance) in all analyses. As replicate blood samples often have different methylation values due to noise in the CpG sites, batch effects, and sample preparation, Higgins-Chen et al. 43 developed a principal components-based method to bolster reliability in DNAmAge algorithms by training them based on PC analysis, extracting the covariance between multicollinear CpGs, including age-related covariance. We estimated DNAmAge using five PC-trained algorithms, Horvath, Horvath Skin and Blood, Hannum, PhenoAge, and GrimAge 1 , 8 , 9 , 15 , 16 . Data Analysis Descriptive statistics for the sample’s demographic characteristics, survey data, and cognitive ability data were computed in R version 4.3.2 38 . Methylation data were cleaned and prepared in R. DNAmAge was calculated as the residual difference in PC-trained values predicted from chronological age in a bivariate linear regression model. DNAmAge values were adjusted for cell composition. Age and cell-adjusted variables were then winsorized such that values more than two standard deviations from the mean DNAmAge score were replaced with a value two standard deviations above or below the mean. Pairwise correlations for the five DNAmAge measures were then estimated using maximum likelihood in M plus Version 8 44 . We used the classical twin model to decompose the phenotypic variance of DNAmAge into additive genetic (A), shared environmental (C), and nonshared environmental (E) components. Given that MZ twins share 100% of their genetic makeup and DZ twin pairs share half, on average, the A variance is correlated 1.0 within MZ twin pairs and 0.5 within DZ twin pairs. In both MZ and DZ twin pairs, the C component includes experiences that make siblings raised together more similar, and is correlated 1.0. Nonshared environmental components are unique to individuals and by definition uncorrelated within twin pairs. We estimated a univariate ACE model for all five DNAmAge measures individually using M plus . Exploratory factor analysis was conducted in M plus to identify underlying dimensions across the five DNAm clock variables (Horvath, Horvath Skin and Blood, Hannum, PhenoAge, and GrimAge). The number of factors to be rotated was determined by the scree plot of Eigenvalues (Supplementary Fig. S2) and a likelihood ratio test comparing a two-factor model to a one-factor model. EFA models were estimated using full-information maximum likelihood with an oblique (geomin) rotation. Factor loadings were evaluated for statistical significance, with a criterion of p < .05 (Supplementary Table S3). We estimated composite scores by standardizing the five DNAmAge variables and using the sum of the standardized values for the variables in each factor. Once composite scores were computed, univariate ACE models were estimated for each factor, decomposing the variance of DNAmAge composite scores into genetic, shared environmental, and nonshared environmental components. To explore the primary aims of the study, the relationship between each DNAmAge factor and ΔIQ was modeled in two ways. Both the phenotypic association model and “quasi-causal” regression model were fit in M plus . The phenotypic association model is equivalent to a simple phenotypic regression where the relationship between predictor and outcome is examined without controlling for genetic and shared environmental effects. This was estimated prior to all genetically informed models. Using a dataset wide by twin pair and estimated parameters set to be equal across the members of a pair, we regressed midlife IQ on childhood IQ, childhood SES, DNAmAge, and the interaction between SES and DNAmAge. Next, we fit the quasi-causal regression model to estimate the phenotypic regression coefficients conditional on b A and b C paths. The classical twin method was used to partition childhood IQ, adult IQ, and DNAmAge into A, C, and E components. In this case, midlife IQ was regressed on childhood IQ, DNAmAge, and the A and C components of DNAmAge and childhood IQ, as illustrated in Fig. 1 . Childhood SES was used as a linear covariate for DNAmAge and the IQ variables. We also estimated an interaction between childhood SES and DNAmAge to determine whether the strength of the relationship between DNAmAge and ΔIQ depended on the SES of the family in which twins were raised. A statistically significant b P3 coefficient affirmatively answers the question: Within a pair of identical twins, does the twin with the higher DNAmAge value also have the larger positive or negative ΔIQ, statistically adjusting for the effects of the A and C components of DNAmAge ( b A3 and b C3 ). Data Availability The datasets analyzed during the current study are not publicly available but deidentified data and/or code are available from the corresponding author on reasonable request. Declarations Funding Funding was provided by the National Institute on Aging Grant No. R01AG063949. Competing Interest Statement No competing interests. Author Contribution S.A.B., E.T., and C.R.B. conceptualized and designed the research and methodology. C.R.B., E.Z., D.W.D. helped collect, generate and/ or process the data. S.A.B. and E.T. executed the statistical analysis and data visualization and wrote the original draft. S.A.B., C.R.B., E.Z., A.K., E.A., J.B., D.F., D.W.D., and E.T. contributed to the interpretation of the results and revisions of the final manuscript. Data Availability The datasets analyzed during the current study are not publicly available but de-identified data and/or code are available from the corresponding author on reasonable request. References Horvath, S. DNA methylation age of human tissues and cell types. Genome Biol. 14 , R115 (2013). Oblak, L., van der Zaag, J., Higgins-Chen, A. T., Levine, M. E. & Boks, M. P. A systematic review of biological, social and environmental factors associated with epigenetic clock acceleration. Ageing Res. Rev. 69 , 101348 (2021). Ryan, J., Wrigglesworth, J., Loong, J., Fransquet, P. D. & Woods, R. L. A Systematic Review and Meta-analysis of Environmental, Lifestyle, and Health Factors Associated With DNA Methylation Age. Journals Gerontology: Ser. A . 75 , 481–494 (2020). Fransquet, P. D., Wrigglesworth, J., Woods, R. L., Ernst, M. E. & Ryan, J. The epigenetic clock as a predictor of disease and mortality risk: a systematic review and meta-analysis. Clin. Epigenetics . 11 , 62 (2019). Horvath, S. & Ritz, B. R. Increased epigenetic age and granulocyte counts in the blood of Parkinson’s disease patients. Aging (Albany NY) . 7 , 1130–1142 (2015). Levine, M. E., Lu, A. T., Bennett, D. A. & Horvath, S. Epigenetic age of the pre-frontal cortex is associated with neuritic plaques, amyloid load, and Alzheimer’s disease related cognitive functioning. Aging (Albany NY) . 7 , 1198–1211 (2015). Bell, C. G. et al. DNA methylation aging clocks: challenges and recommendations. Genome Biol. 20 , 249 (2019). Hannum, G. et al. Genome-wide Methylation Profiles Reveal Quantitative Views of Human Aging Rates. Mol. Cell . 49 , 359–367 (2013). Horvath, S. et al. Epigenetic clock for skin and blood cells applied to Hutchinson Gilford Progeria Syndrome and ex vivo studies. Aging (Albany NY) . 10 , 1758–1775 (2018). Vaccarino, V. et al. Epigenetic Age Acceleration and Cognitive Decline: A Twin Study. J. Gerontol. Biol. Sci. Med. Sci. 76 , 1854–1863 (2021). Faul, J. D. et al. Epigenetic-based age acceleration in a representative sample of older Americans: Associations with aging-related morbidity and mortality. Proceedings of the National Academy of Sciences 120, e2215840120 (2023). Maddock, J. et al. Childhood growth and development and DNA methylation age in mid-life. Clin. Epigenet . 13 , 155 (2021). Reed, R. G., Carroll, J. E., Marsland, A. L. & Manuck, S. B. DNA methylation-based measures of biological aging and cognitive decline over 16-years: preliminary longitudinal findings in midlife. Aging (Albany NY) . 14 , 9423–9444 (2022). Starnawska, A. et al. Blood DNA methylation age is not associated with cognitive functioning in middle-aged monozygotic twins. Neurobiol. Aging . 50 , 60–63 (2017). Levine, M. E. et al. An epigenetic biomarker of aging for lifespan and healthspan. Aging 10 , 573–591 (2018). Lu, A. T. et al. DNA methylation GrimAge strongly predicts lifespan and healthspan. Aging 11 , 303–327 (2019). Graves, A. J. et al. Accelerated epigenetic age is associated with whole-brain functional connectivity and impaired cognitive performance in older adults. Sci. Rep. 14 , 9646 (2024). Hillary, R. F. et al. An epigenetic predictor of death captures multi-modal measures of brain health. Mol. Psychiatry . 26 , 3806 (2019). Aartsen, M. J. et al. Advantaged socioeconomic conditions in childhood are associated with higher cognitive functioning but stronger cognitive decline in older age. Proc. Natl. Acad. Sci. U S A . 116 , 5478–5486 (2019). McCrory, C. et al. Early life adversity and age acceleration at mid-life and older ages indexed using the next-generation GrimAge and Pace of Aging epigenetic clocks. Psychoneuroendocrinology 137 , 105643 (2022). Austin, M. K. et al. Early-life socioeconomic disadvantage, not current, predicts accelerated epigenetic aging of monocytes. Psychoneuroendocrinology 97 , 131–134 (2018). Hughes, A. et al. Socioeconomic Position and DNA Methylation Age Acceleration Across the Life Course. Am. J. Epidemiol. 187 , 2346–2354 (2018). McCrory, C. et al. How does socio-economic position (SEP) get biologically embedded? A comparison of allostatic load and the epigenetic clock(s). Psychoneuroendocrinology 104 , 64–73 (2019). Jylhävä, J. et al. Longitudinal changes in the genetic and environmental influences on the epigenetic clocks across old age: Evidence from two twin cohorts. eBioMedicine 40 , 710–716 (2019). Marioni, R. E. et al. DNA methylation age of blood predicts all-cause mortality in later life. Genome Biol. 16 , 25 (2015). Turkheimer, E. & Harden, K. P. Behavior genetic research methods: Testing quasi-causal hypotheses using multivariate twin data. in Handbook of research methods in social and personality psychology, 2nd ed 159–187 (Cambridge University Press, New York, NY, US, (2014). Beam, C. R. et al. Midlife Study of the Louisville Twins: Connecting Cognitive Development to Biological and Cognitive Aging. Behav. Genet. 50 , 73–83 (2020). Levine, M. E. Assessment of Epigenetic Clocks as Biomarkers of Aging in Basic and Population Research. Journals Gerontology: Ser. A . 75 , 463–465 (2020). Fiorito, G. et al. Social adversity and epigenetic aging: a multi-cohort study on socioeconomic differences in peripheral blood DNA methylation. Sci. Rep. 7 , 16266 (2017). Barboza Solís, C. et al. Mediating pathways between parental socio-economic position and allostatic load in mid-life: Findings from the 1958 British birth cohort. Soc. Sci. Med. 165 , 19–27 (2016). Belsky, D. W. et al. DunedinPACE, a DNA methylation biomarker of the pace of aging. eLife 11, (2022). Davis, D. W., Turkheimer, E., Finkel, D., Beam, C. & Ryan, L. The Louisville Twin Study: Past, Present and Future. Twin Res. Hum. Genet. 22 , 735–740 (2019). Duncan, O. A socioeconomic index for all occupations. Occupations social status (1961). Wechsler, D. Wechsler Intelligence Scale for Children (Psychological Corporation, 1949). Wechsler, D. Wechsler Intelligence Scale for Children-Revised (The Psychological Corporation, 1974). Wechsler, D. Wechsler Intelligence Scale for Children (The Psychological Corporation, 1991). Wechsler, D. Wechsler Adult Intelligence Scale - Fourth Edition Administration and Scoring Manual. (2008). https://psycnet.apa.org/doiLanding?doi=10.1037%2Ft15169-000 R Core Team. R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, 2023). Aryee, M. J. et al. Minfi: a flexible and comprehensive Bioconductor package for the analysis of Infinium DNA methylation microarrays. Bioinformatics 30 , 1363–1369 (2014). Triche, T. J., Weisenberger, D. J., Van Den Berg, D., Laird, P. W. & Siegmund, K. D. Low-level processing of Illumina Infinium DNA Methylation BeadArrays. Nucleic Acids Res. 41 , e90 (2013). Johnson, W. E., Li, C. & Rabinovic, A. Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics 8 , 118–127 (2007). Houseman, E. A. et al. DNA methylation arrays as surrogate measures of cell mixture distribution. BMC Bioinform. 13 , 86 (2012). Higgins-Chen, A. T. et al. A computational solution for bolstering reliability of epigenetic clocks: implications for clinical trials and longitudinal tracking. Nat. Aging . 2 , 644–661 (2022). Muthén, L. K. & Muthén, B. O. Mplus: Statistical Analysis with Latent Variables: User’s Guide (Version 8) (Muthén & Muthén, 1998). Additional Declarations No competing interests reported. Supplementary Files SciReportsSIDNAmAge.docx Cite Share Download PDF Status: Published Journal Publication published 22 Jul, 2025 Read the published version in Aging → Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-5551592","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":393502786,"identity":"c6d4c2d1-2d7f-4c88-93dd-1c4543d7b75d","order_by":0,"name":"Sophie A. Bell","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA5klEQVRIiWNgGAWjYBACAwbG5z8YGGwY2ICcA0AuECQQ0sJsIMHAkEa6lsPIYgS0mEskMxgX1JxP7JM+fvFwQYEdAz97jgFeLZYzkhmSZxy7ndjGl1NweIZBMoNkzxv8WgzOnD9wmIfttjEbD0/CYR4DZgaDGwRsMThzmLGZ5985mJZ6BnuCWo43MzPzth2QY+NhB1pncJjBQIKQX9qb2Zhn9iUDtfAwALUc55E486wArxZzZmY25oJvdjzyPeyPP/P8qZbjb0/egFcLCDBDKB6we3gIKkfSwv6AKNWjYBSMglEw8gAAOFU+t+7Bz6IAAAAASUVORK5CYII=","orcid":"","institution":"Department of Psychology, University of Virginia","correspondingAuthor":true,"prefix":"","firstName":"Sophie","middleName":"A.","lastName":"Bell","suffix":""},{"id":393502788,"identity":"f0760e3c-2200-45b2-927f-846e602fb504","order_by":1,"name":"Christopher R. Beam","email":"","orcid":"","institution":"Department of Psychology, University of Southern California","correspondingAuthor":false,"prefix":"","firstName":"Christopher","middleName":"R.","lastName":"Beam","suffix":""},{"id":393502789,"identity":"5349f076-1873-4292-acf4-7d78852c4a8f","order_by":2,"name":"Ebrahim Zandi","email":"","orcid":"","institution":"Keck School of Medicine, University of Southern California","correspondingAuthor":false,"prefix":"","firstName":"Ebrahim","middleName":"","lastName":"Zandi","suffix":""},{"id":393502790,"identity":"8b8f914d-e712-482f-aa65-ae739dbb96f9","order_by":3,"name":"Alyssa Kam","email":"","orcid":"","institution":"Department of Psychology, University of Southern California","correspondingAuthor":false,"prefix":"","firstName":"Alyssa","middleName":"","lastName":"Kam","suffix":""},{"id":393502791,"identity":"8fcc2bfa-458e-4da9-b7ad-8664ce2f2ddf","order_by":4,"name":"Emily Andrews","email":"","orcid":"","institution":"Department of Psychology, University of Virginia","correspondingAuthor":false,"prefix":"","firstName":"Emily","middleName":"","lastName":"Andrews","suffix":""},{"id":393502792,"identity":"a0e87216-27d6-40e7-af43-aad790256bb3","order_by":5,"name":"Jonathan Becker","email":"","orcid":"","institution":"Department of Family and Geriatric Medicine, University of Louisville School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Jonathan","middleName":"","lastName":"Becker","suffix":""},{"id":393502793,"identity":"db1c2b16-2db4-49a9-b53c-99174b2baaff","order_by":6,"name":"Deborah Finkel","email":"","orcid":"","institution":"Institute for Gerontology, Jönköping University","correspondingAuthor":false,"prefix":"","firstName":"Deborah","middleName":"","lastName":"Finkel","suffix":""},{"id":393502794,"identity":"f9d6e339-8de6-497c-87e6-36b1af73901d","order_by":7,"name":"Deborah W. Davis","email":"","orcid":"","institution":"Department of Pediatrics, University of Louisville School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Deborah","middleName":"W.","lastName":"Davis","suffix":""},{"id":393502795,"identity":"0673f254-8a46-4ca6-8e7a-02e41f082aeb","order_by":8,"name":"Eric Turkheimer","email":"","orcid":"","institution":"Department of Psychology, University of Virginia","correspondingAuthor":false,"prefix":"","firstName":"Eric","middleName":"","lastName":"Turkheimer","suffix":""}],"badges":[],"createdAt":"2024-11-29 23:08:05","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5551592/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5551592/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.18632/aging.206284","type":"published","date":"2025-07-23T00:00:00+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":72224529,"identity":"70e55ae0-26f7-4b14-b7af-a069e4acd5a8","added_by":"auto","created_at":"2024-12-24 01:27:42","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":35258,"visible":true,"origin":"","legend":"\u003cp\u003ePath diagram depicting the association between DNAmAge and change in IQ, modified by childhood socioeconomic status (SES). Variances of ACE components are estimated, with paths to observed variables set to 1.0.\u0026nbsp;All regression coefficients are unstandardized. \u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003eA1\u003c/em\u003e\u003c/sub\u003e\u003csub\u003e \u003c/sub\u003eand \u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003eC1\u003c/em\u003e\u003c/sub\u003e are regression coefficients from the A and C components of Childhood IQ to Adult IQ. \u003cem\u003eb\u003c/em\u003e\u003csub\u003eP1\u003c/sub\u003e is the phenotypic regression conditional on \u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003eA1\u003c/em\u003e\u003c/sub\u003e\u003csub\u003e \u003c/sub\u003eand \u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003eC1\u003c/em\u003e\u003c/sub\u003e. \u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003eA2 \u003c/em\u003e\u003c/sub\u003eand \u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003eC2\u003c/em\u003e\u003c/sub\u003e\u003cem\u003e \u003c/em\u003eare regression coefficients from the A and C components of Childhood IQ to DNAmAge. \u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003eP2\u003c/em\u003e\u003c/sub\u003e\u003cem\u003e \u003c/em\u003eis the phenotypic regression conditional on \u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003eA2\u003c/em\u003e\u003c/sub\u003e\u003csub\u003e \u003c/sub\u003eand \u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003eC2\u003c/em\u003e\u003c/sub\u003e. \u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003eA3\u003c/em\u003e\u003c/sub\u003e\u003csub\u003e \u003c/sub\u003eand \u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003eC3\u003c/em\u003e\u003c/sub\u003e\u003cem\u003e \u003c/em\u003eare regression coefficients from the A and C components of DNAmAge to Adult IQ. \u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003eP3\u003c/em\u003e\u003c/sub\u003e is the phenotypic regression conditional on \u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003eA3\u003c/em\u003e\u003c/sub\u003e\u003csub\u003e \u003c/sub\u003eand \u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003eC3\u003c/em\u003e\u003c/sub\u003e. SES is a covariate for DNAmAge and the IQ variables, and a moderator of \u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003eP3\u003c/em\u003e\u003c/sub\u003e.\u003c/p\u003e","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-5551592/v1/d3e92be4d70890c15f4d4392.png"},{"id":72225042,"identity":"cdf15b93-7be5-4b42-ad25-21a07afb607b","added_by":"auto","created_at":"2024-12-24 01:43:42","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":87227,"visible":true,"origin":"","legend":"\u003cp\u003eRegression of Adult IQ residuals on Gen 2 DNAmAge for high and low SES groups. IQ residuals are derived from the regression of adult IQ on childhood IQ and childhood socioeconomic status.\u003c/p\u003e","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-5551592/v1/b70e6b373a65f660f5ca68bd.png"},{"id":88566026,"identity":"a6691f46-b00b-4bc7-af50-d131a3a6ed94","added_by":"auto","created_at":"2025-08-07 19:51:13","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1794824,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5551592/v1/c538a72f-782f-45a7-9c64-36fc98caf019.pdf"},{"id":72224664,"identity":"36278a15-a479-410b-8591-9cc35ca593d1","added_by":"auto","created_at":"2024-12-24 01:35:42","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":329780,"visible":true,"origin":"","legend":"","description":"","filename":"SciReportsSIDNAmAge.docx","url":"https://assets-eu.researchsquare.com/files/rs-5551592/v1/7431d933fa4c21dfe86d847c.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Second Generation DNA Methylation Age Predicts Cognitive Change in Midlife: The Moderating Role of Childhood Socioeconomic Status","fulltext":[{"header":"Introduction","content":"\u003cp\u003eAging is highly heterogeneous, and chronological age is not a sensitive measure of the unique physiological and developmental processes that occur across the human lifespan. Patterns of DNA methylation (DNAm) across the genome can be used to estimate DNA methylation age (DNAmAge), a family of measures strongly correlated with chronological age but better able to predict morbidities and mortality. Adjusting for chronological age, DNAmAge measures can be residualized to generate measures of acceleration \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e, such that positive residuals reflect faster biological aging, while negative residuals reflect slower, and presumably healthier, aging. Since the advent of the first measures a decade ago, DNAmAge measures reliably correlate with demographic characteristics like education level and socioeconomic status (SES) as well as clinically relevant outcomes including cognitive performance in adulthood \u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e,\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\n\u003ch3\u003ePredicting cognitive outcomes from DNAmAge\u003c/h3\u003e\n\u003cp\u003eIndividuals with accelerated DNAmAge have an increased risk of cancer, stroke, heart disease, and neurodegenerative disease \u003csup\u003e\u003cspan additionalcitationids=\"CR5\" citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e. While cognitive functioning is a crucial facet of both healthy and pathological aging, the link between DNAmAge and cognitive functioning remains unclear, partly due to inconsistencies in the predictive accuracy of various DNAmAge algorithms \u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e. First generation DNAmAge measures, such as Horvath, Horvath Skin and Blood, and Hannum DNAmAge, were designed primarily to estimate chronological age with methylation data from cytosine-phosphate-guanine (CpG) sites sensitive to time-dependent changes \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e. While two studies found that Horvath DNAmAge was associated with cognitive decline \u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e,\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e, several others have found no statistically significant associations between first generation DNAmAge measures and cognitive performance or decline \u003csup\u003e\u003cspan additionalcitationids=\"CR12 CR13\" citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eIn contrast, second generation DNAmAge measures, including PhenoAge and GrimAge, are trained to predict age-related outcomes by integrating DNAm signatures tied to clinical biomarkers, health behaviors (e.g., smoking), and demographic variables \u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e,\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e. These algorithms aim to capture phenotypic aging processes beyond chronological age, such as morbidity and mortality risk. Accelerated PhenoAge and GrimAge are consistently associated with lower cognitive performance in middle and older adulthood \u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e,\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e,\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e,\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e. However, findings on cognitive decline are mixed. Neither Hillary et al. \u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e nor Maddock et al. \u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e reported statistically significant results, yet Reed and colleagues \u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e found that those with cognitive decline in midlife showed more accelerated GrimAge scores than those without decline. These findings suggest that second generation clocks, developed to more accurately estimate phenotypic biological age, may better index age-related changes in cognitive ability.\u003c/p\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eEarly-life exposures and pace of DNAmAge\u003c/h2\u003e \u003cp\u003eDisadvantaged socioeconomic status (SES) in childhood is an established risk factor for lower cognitive ability across the lifespan and late-life cognitive decline \u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e. Recent research has also begun exploring links between childhood adversity and later-life DNA methylation age (DNAmAge). In retrospective accounts of adversity (e.g., poverty, death of parent, alcohol/drug use in family, abuse), only poverty was related to GrimAge acceleration \u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e. Childhood SES but not adult SES predicted accelerated DNAmAge in midlife using the Horvath and Hannum clocks, suggesting the methylome, or DNA methylation modifications across the genome, may be particularly vulnerable to external stressors in childhood \u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e,\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e. However, findings remain inconsistent: McCrory et al. \u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e found no association between childhood SES and Horvath, Hannum, or PhenoAge DNAmAge in adults over 50, and unexpectedly, Faul et al. \u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e found that low childhood SES was associated with slower aging on the Horvath and Hannum clocks later in life. It is not yet known whether SES moderates the effects of DNAmAge on cognitive outcomes.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eGenetics, Epigenetics, and Classical Twin Designs\u003c/h3\u003e\n\u003cp\u003eAssociations between early life exposures, mid- to late-life outcomes, and DNAmAge have primarily been correlational. One study decomposed the variance of DNAmAge into additive genetic, shared environmental, and nonshared environmental components using a traditional twin model, while two others used pedigree designs to estimate heritability \u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e,\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e,\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e. Our group has shown how to use a longitudinal twin design to go beyond estimating heritability while sharpening causal inference between a predictor and outcome \u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e. Only two studies have examined the relationship between DNAmAge and cognitive ability while controlling for between-family confounds \u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e,\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. In the Middle-Aged Danish Twin Study, no relationship was observed between the Horvath and Hannum clocks and cognitive change over 10 years in midlife \u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. Vaccarino and colleagues \u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e found no associations between Horvath, Hannum, GrimAge, or PhenoAge clocks and baseline cognition in middle-aged male twins. However, there was a statistically significant within-twin pair effect of Horvath DNAmAge on cognitive decline 11 years later. The reason why only Horvath\u0026rsquo;s original measure predicted cognitive ability is unclear, as initial validation studies of PhenoAge and GrimAge reported associations with cognitive outcomes \u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e,\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eUsing the Louisville Twin Study (LTS), we expanded on this limited amount of research examining DNAmAge and cognitive change and sought to clarify how early life environmental characteristics may influence midlife cognitive outcomes. The LTS, initiated in 1957, includes intensive, prospective data on twins\u0026rsquo; development from infancy through adolescence and recent re-evaluations in midlife on their cognitive, physical, and psychosocial functioning as well as epigenetic aging\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e. Parental data collected during the twins' childhood further enabled us to examine how early life exposures, here rearing SES, relate to twins\u0026rsquo; midlife epigenetic and cognitive aging.\u003c/p\u003e \u003cp\u003eThe current study used a twin design to investigate causal effects of DNAmAge on change in IQ (ΔIQ) between childhood and midlife, while testing childhood SES as a moderator of the relationship. While observational designs cannot reproduce the causal inference possible with random assignment, twins offer a quasi-experimental method for studying causal processes while controlling for family-level genetic and environmental confounds. If a monozygotic (MZ) twin has a higher DNAmAge and also a greater ΔIQ than their co-twin, the association cannot be due to genetic differences or early environmental exposures, because identical twins reared together are matched for these factors. For this reason, we refer to associations that survive twin-based controls as \u0026ldquo;quasi-causal\u0026rdquo; \u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e. To address variability in DNAmAge algorithms, we conducted an exploratory factor analysis of five widely used measures. We expected second generation DNAmAge measures to be sensitive to ΔIQ in midlife, as they were developed to capture phenotypic aging processes beyond chronological age. Finally, given recent findings linking early life stressors with midlife DNAmAge, we hypothesized that the relationship between midlife DNAmAge and ΔIQ would depend on differences in childhood SES.\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003eDescriptive statistics\u003c/h2\u003e \u003cp\u003eThe sample includes 287 individual twins, comprising 60 monozygotic and 41 dizygotic complete pairs and 85 single twins. All participated in the childhood and midlife phases of the LTS (Supplementary Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). Twins were assessed in person using three versions of the Wechsler Intelligence Scale for Children (WISC) and the Wechsler Adult Intelligence Scale (WAIS-IV). We used Wechsler IQ scores as an overall estimate of an individual\u0026rsquo;s level of global cognitive ability. The average age for WISC was 14.42 (1.93) years, and the average age for WAIS-IV was 51.87 (7.03). Mean childhood IQ was 102.7 (13.4), and mean midlife IQ was 105.4 (13.8) (Table\u0026nbsp;1). After regressing DNAmAge values on chronological age and adjusting for cell type, Horvath, Horvath Skin and Blood, Hannum, PhenoAge, and GrimAge \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e,\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e had means close to zero (Table\u0026nbsp;1). MZ twin correlations (0.60\u0026ndash;0.74) were approximately double those for DZ twins (0.28\u0026ndash;0.37) across most measures, except PhenoAge, where MZ and DZ correlations were more similar (0.56 vs. 0.47). Pairwise clock correlations are in Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003eResults from standardized univariate twin models decomposing the variance of individual DNAmAge measures show little to no shared environmental variance, C, in Horvath, Horvath Skin and Blood, Hannum, and GrimAge clocks. Because C was estimated at a negative value in these cases initially, C was set to zero and the constrained AE models are presented (Supplementary Table S2). These four DNAmAge measures were substantially heritable, with the additive genetic effects, A, ranging from 57\u0026ndash;73%, and non-shared environment, E, accounting for 27\u0026ndash;43% of the variance. The ACE decomposition of PhenoAge differed in that we estimated the A at 0.23 (95% CI: [-0.35, 0.80]), C at 0.33 (95% CI: [-0.16, 0.82]), and E at 0.45 (95% CI: [0.28, 0.62]) (Supplementary Table S2).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Taba\" border=\"1\"\u003e \u003ccolgroup cols=\"1\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"char\" char=\".\" colname=\"c1\"\u003e \u003cp\u003eTable\u0026nbsp;1. Sample Characteristics of Louisville Twins\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabb\" border=\"1\"\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCharacteristic\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTotal n (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMean (SD)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMZ Correlation\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003er (SE)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eDZ Correlation\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003er (SE)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eSociodemographic factors\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e287\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge, years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e51.87 (7.03)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge at last WISC, years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e14.42 (1.93)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale Sex\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e169 (58.89)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWhite\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e262 (91.29)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNon-Hispanic\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e279 (99.29)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEducation, 16 or more years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e161 (56.09)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eChildhood SES\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e46.09 (26.02)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSmoker\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e66 (23.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eCognitive function\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWISC Full Scale IQ Score\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e282\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e102.67 (13.44)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.88 (0.06)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.58 (0.14)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWAIS Full Scale IQ Score\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e255\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e105.45 (13.84)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.87 (0.07)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.50 (0.14)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eDNAmAge\u003c/em\u003e\u003csup\u003e\u003cem\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/em\u003e\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e281\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHorvath\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.00 (3.37)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.62 (0.11)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.37 (0.15\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHorvath Skin and Blood\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-0.03 (3.59)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.62 (0.10)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.32 (0.15)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHannum\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-0.01 (3.25)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.60 (0.11)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.28 (0.16)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePhenoAge\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-0.09 (4.20)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.56 (0.11)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.47 (0.14)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGrimAge\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-0.12 (3.42)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.74 (0.09)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.37 (0.15)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGen 1 Composite\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.00 (2.85)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.62 (0.10)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.32 (0.15)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGen 2 Composite\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.00 (1.78)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.71 (0.09)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.42 (0.15)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003eMonozygotic Twin Pair Correlation\u003c/p\u003e \u003cp\u003e\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003eDizygotic Twin Pair Correlation\u003c/p\u003e \u003cp\u003e\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003eChildhood socioeconomic status (SES) measured using the Duncan Socioeconomic Index\u003c/p\u003e \u003cp\u003e\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003eDNAmAge is controlled for chronological age and cell composition. Residuals are in years.\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eExploratory Factor Analysis of DNAm Clocks\u003c/h3\u003e\n\u003cp\u003eWe conducted an exploratory factor analysis (EFA) using the five DNAmAge measures residualized for age and cell count. Based on Eigenvalues and fit indices, we selected a two-factor model (RMSEA\u0026thinsp;=\u0026thinsp;0.08; 95% CI: [0.00, 0.20]) (Supplementary Fig. S2). Results from a likelihood ratio test comparing a two-factor model to a one-factor model suggested that the two-factor model fit substantially better (χ\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;70.07, df\u0026thinsp;=\u0026thinsp;4, p\u0026thinsp;\u0026lt;\u0026thinsp;.001). First generation DNAmAge measures (Horvath, Horvath Skin and Blood, and Hannum) loaded on one factor, while second generation measures (PhenoAge and GrimAge) loaded on a second factor. We refer to these factors as \u0026ldquo;Gen 1\u0026rdquo; and \u0026ldquo;Gen 2\u0026rdquo;, respectively. Factor loadings (p\u0026thinsp;\u0026lt;\u0026thinsp;.05) are displayed in Supplementary Table S3. The factors were correlated \u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.57. Composite scores were generated for each participant based on the EFA. Gen 1 is the sum of standardized values for Horvath, Horvath Skin and Blood, and Hannum DNAmAge (M\u0026thinsp;=\u0026thinsp;0.00, SD\u0026thinsp;=\u0026thinsp;2.85). Gen 2 is the sum of standardized values for GrimAge and PhenoAge DNAmAge (M\u0026thinsp;=\u0026thinsp;0.00, SD\u0026thinsp;=\u0026thinsp;1.78). Twin pair correlations for Gen 1 and Gen 2 are in Table\u0026nbsp;1.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eUnivariate Twin Models of DNAmAge Factors\u003c/h2\u003e \u003cp\u003eWe decomposed the variances of Gen 1 and Gen 2 into A, C, and E components using the classical twin model. Both factors were moderately heritable, with A accounting for 50% of the variance in Gen 1 and 60% of the variance in Gen 2. The variance attributable to the non-shared environment was 42% in Gen 1 and 31% in Gen 2, while shared environment accounted for less than 1% of the variance in both factors. Standardized estimates are in Supplementary Table S2.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003ePhenotypic and “Quasi-Causal” Association Models\u003c/h3\u003e\n\u003cp\u003eFollowing the twin modeling framework outlined by Turkheimer and Harden (2014), we first estimated a \u0026ldquo;phenotypic association\u0026rdquo; model with no control for differential regressions between and within twin pairs. We then fit a \u0026ldquo;quasi-causal\u0026rdquo; regression model, using the ACE decomposition of the predictor variable to control for family-level genetic and environmental confounds in the individual-level regressions. This sequence (phenotypic association model and quasi-causal regression model) was estimated for Gen 1 then repeated for Gen 2. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e illustrates the model: adult IQ was regressed on childhood IQ, DNAmAge, and the A and C components of each; DNAmAge was regressed on childhood IQ and its A and C components; SES (a family-level variable) was included as a linear covariate for DNAmAge and IQ variables and as an interaction term with DNAmAge. By including childhood IQ in the model, the conditional effects of SES and DNAmAge can be interpreted as predicting change in IQ from childhood to midlife (ΔIQ). The phenotypic regression coefficient (\u003cem\u003eb\u003c/em\u003e\u003csub\u003eP3\u003c/sub\u003e) is an estimate of the relationship between the DNAmAge and ΔIQ with between-family genetic and environmental confounds controlled, i.e., within a pair of identical twins reared together.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003ePredictors of DNAmAge\u003c/h3\u003e\n\u003cp\u003eIn the phenotypic association model, there were no statistically significant associations with Gen 1 DNAmAge. All Gen 1 model results are presented in Supplementary Table S4. Accelerated Gen 2 DNAmAge was associated with lower childhood SES (\u003cem\u003eb=\u003c/em\u003e-0.41, SE\u0026thinsp;=\u0026thinsp;0.13, p\u0026thinsp;\u0026lt;\u0026thinsp;.05) in the phenotypic association model (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Next, we used the quasi-causal regression model to estimate the effects of childhood IQ and SES on DNAmAge, using the ACE decomposition of childhood IQ to control for genetic and environmental family-level confounds. In the quasi-causal analyses, there were similarly no statistically significant predictors of Gen 1 DNAmAge. Childhood SES again predicted Gen 2 DNAmAge (\u003cem\u003eb=\u003c/em\u003e-0.67, SE\u0026thinsp;=\u0026thinsp;0.19, p\u0026thinsp;\u0026lt;\u0026thinsp;.001) in this model, while childhood IQ did not predict Gen 2 DNAmAge. Genetic and environmental confounds were not significantly different from zero. Gen 2 DNAmAge results from both models are presented in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003ePredictors of Midlife Cognitive Functioning\u003c/h2\u003e \u003cp\u003eIn the phenotypic association models for both Gen 1 and Gen 2, childhood IQ statistically predicted adult IQ. The strong relationship between childhood IQ and adult IQ (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.77) is also illustrated in Supplementary Figure S3, where adult IQ is regressed on childhood IQ without controlling for the differential regressions between and within twin pairs. Similarly, SES was positively associated with childhood IQ (\u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.56, SE\u0026thinsp;=\u0026thinsp;0.09, p\u0026thinsp;\u0026lt;\u0026thinsp;.001) in both phenotypic association models. Higher childhood SES predicted higher adult IQ in the Gen 1 model (\u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.16, SE\u0026thinsp;=\u0026thinsp;0.07, p\u0026thinsp;\u0026lt;\u0026thinsp;.05), though this association did not meet statistical significance in the Gen 2 model. We found a statistically significant phenotypic association between Gen 2 and ΔIQ (\u003cem\u003eb\u003c/em\u003e\u003csub\u003eP3\u003c/sub\u003e= -0.10, SE\u0026thinsp;=\u0026thinsp;0 .03, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.05). In both Gen 1 and Gen 2 phenotypic association models, we then tested the interaction between DNAmAge and SES (\u003cem\u003eb\u003c/em\u003e\u003csub\u003eint\u003c/sub\u003e) in the prediction of ΔIQ but found that the interaction was not statistically significant.\u003c/p\u003e \u003cp\u003eIn quasi-causal models, we then regressed midlife IQ on childhood IQ (\u003cem\u003eb\u003c/em\u003e\u003csub\u003eP1\u003c/sub\u003e), DNAmAge (\u003cem\u003eb\u003c/em\u003e\u003csub\u003eP3\u003c/sub\u003e), and the A and C components of both variables. There was a statistically significant phenotypic association between childhood IQ and adult IQ after accounting for genetic and environmental confounds shared by identical twins in both Gen 1 (\u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003eP1\u003c/em\u003e\u003c/sub\u003e = 0.41, p\u0026thinsp;\u0026lt;\u0026thinsp;.01) and Gen 2 models (\u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cem\u003eP1\u003c/em\u003e\u003c/sub\u003e = 0.48, p\u0026thinsp;\u0026lt;\u0026thinsp;.001). This association explains that over 40% of the variance in adult IQ can be attributed to childhood IQ. Consistent with the phenotypic association results above, accelerated Gen 2 DNAmAge predicted more negative ΔIQ from childhood to midlife (\u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;0.18, SE\u0026thinsp;=\u0026thinsp;0.07, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.05). Genetic and environmental confounds (\u003cem\u003eb\u003c/em\u003e\u003csub\u003eA3\u003c/sub\u003e and \u003cem\u003eb\u003c/em\u003e\u003csub\u003eC3\u003c/sub\u003e) were not significantly different from zero. The quasi-causal twin model implies that within a pair of MZ twins raised together, the member of the pair with accelerated DNAmAge also declines more in cognitive ability between childhood and midlife. These findings were only apparent when DNAmAge was measured with second generation (Gen 2) DNAmAge measures.\u003c/p\u003e \u003cp\u003eThe interaction revealed that twins raised in lower SES families showed a more negative relationship between Gen 2 DNAmAge and ΔIQ (\u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.07, SE\u0026thinsp;=\u0026thinsp;0.03, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.05) (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Figure\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e illustrates the moderating effect of childhood SES by using a median split of our sample for SES and displaying the regression of adult IQ on Gen 2 DNAmAge in both high- and low-SES groups. The effect of DNAmAge on ΔIQ is greater in twins raised in lower SES households, indicating that socioeconomic disadvantage may be an early exposure that amplifies the negative effects of epigenetic aging on cognitive trajectories in midlife.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePhenotypic and Quasi-Causal Regression Model Results for Gen 2 DNAmAge\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"13\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c11\" colnum=\"11\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c12\" colnum=\"12\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c13\" colnum=\"13\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colspan=\"11\" nameend=\"c11\" namest=\"c1\"\u003e \u003cp\u003eGen 2 DNAmAge\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c13\" namest=\"c12\"\u003e\u0026nbsp;\u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003eChildhood IQ\u003c/p\u003e \u003cp\u003eEstimate (SE)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003eDNAmAge\u003c/p\u003e \u003cp\u003eEstimate (SE)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c10\" namest=\"c9\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c13\" namest=\"c11\"\u003e \u003cp\u003eAdult IQ\u003c/p\u003e \u003cp\u003eEstimate (SE)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"15\" rowspan=\"16\"\u003e \u003cp\u003ePhenotypic Association Model\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"10\" nameend=\"c11\" namest=\"c2\"\u003e \u003cp\u003eSES\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c13\" namest=\"c12\"\u003e\u0026nbsp;\u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.56 (0.09)**\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e \u003cp\u003e-0.67 (0.20)**\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e \u003cp\u003e0.20 (0.11)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"10\" nameend=\"c11\" namest=\"c2\"\u003e \u003cp\u003e\u003cb\u003eChildhood IQ\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c13\" namest=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eA2\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eA1\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eC2\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eC1\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eP2\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e \u003cp\u003e-0.08 (0.09)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eP1\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e \u003cp\u003e0.48 (0.14)**\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"10\" nameend=\"c11\" namest=\"c2\"\u003e \u003cp\u003e\u003cb\u003eDNAmAge\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c13\" namest=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eA3\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eC3\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eP3\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e \u003cp\u003e-0.10 (0.03)*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"10\" nameend=\"c11\" namest=\"c2\"\u003e \u003cp\u003e\u003cb\u003eDNAmAge X SES\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c13\" namest=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eint\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e \u003cp\u003e0.06 (0.03)*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"10\" nameend=\"c11\" namest=\"c2\"\u003e \u003cp\u003e\u003cb\u003eVariances\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c13\" namest=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eA\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.15 (0.34)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e \u003cp\u003e1.96 (0.31)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e \u003cp\u003e0.50 (0.08)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eC\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.15 (0.32)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e \u003cp\u003e0.00 (0.00)\u003csup\u003e\u0026dagger;\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e \u003cp\u003e0.00 (0.00)\u003csup\u003e\u0026dagger;\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eE\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.22 (0.04)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e \u003cp\u003e0.85 (0.16)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e \u003cp\u003e0.24 (0.05)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"15\" rowspan=\"16\"\u003e \u003cp\u003e\u003cb\u003eQuasi-Causal Regression Model\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"10\" nameend=\"c11\" namest=\"c2\"\u003e \u003cp\u003e\u003cb\u003eSES\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c13\" namest=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.56 (0.09)**\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e \u003cp\u003e-0.67 (0.19)*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e \u003cp\u003e0.18 (0.12)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"10\" nameend=\"c11\" namest=\"c2\"\u003e \u003cp\u003e\u003cb\u003eChildhood IQ\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c13\" namest=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eA2\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e \u003cp\u003e-0.78 (0.41)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eA1\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e \u003cp\u003e0.16 (0.23)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eC2\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e \u003cp\u003e0.30 (1.42)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eC1\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e \u003cp\u003e0.80 (0.67)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eP2\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e \u003cp\u003e0.38 (0.26)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eP1\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e \u003cp\u003e0.48 (0.14)**\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"10\" nameend=\"c11\" namest=\"c2\"\u003e \u003cp\u003e\u003cb\u003eDNAmAge\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c13\" namest=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eA3\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e \u003cp\u003e0.13 (0.15)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eC3\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e \u003cp\u003e0.00 (0.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eP3\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e \u003cp\u003e-0.18 (0.07)*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"10\" nameend=\"c11\" namest=\"c2\"\u003e \u003cp\u003e\u003cb\u003eDNAmAge X SES\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c13\" namest=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e\u003csub\u003e\u003cb\u003eint\u003c/b\u003e\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e \u003cp\u003e0.07 (0.03)*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"10\" nameend=\"c11\" namest=\"c2\"\u003e \u003cp\u003e\u003cb\u003eVariances\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c13\" namest=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eA\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.06 (0.29)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e \u003cp\u003e1.73 (0.56)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e \u003cp\u003e0.43 (0.12)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eC\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.21 (0.28)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e \u003cp\u003e0.00 (0.00)\u003csup\u003e\u0026dagger;\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e \u003cp\u003e0.00 (0.00)\u0026dagger;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eE\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.23 (0.04)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e \u003cp\u003e0.80 (0.15)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c12\" namest=\"c10\"\u003e \u003cp\u003e0.23 (0.05)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"1\" nameend=\"c13\" namest=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"11\" nameend=\"c11\" namest=\"c1\"\u003e \u003cp\u003eb\u003csub\u003eP\u003c/sub\u003e, phenotypic association between predictor and outcome; b\u003csub\u003eA\u003c/sub\u003e, amount of variance attributable to additive genetic influences; b\u003csub\u003eC\u003c/sub\u003e, amount of variance attributable to shared environmental experiences; b\u003csub\u003eint\u003c/sub\u003e, estimate of the interaction effect. Data are presented as unstandardized regression coefficients with standard errors in parentheses.\u003c/p\u003e \u003cp\u003e\u0026dagger;C variances have been set to zero\u003c/p\u003e \u003cp\u003e*p\u0026thinsp;\u0026lt;\u0026thinsp;.05 **p\u0026thinsp;\u0026le;\u0026thinsp;.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c13\" namest=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eDNAmAge is an effective predictor of specific aspects of biological aging, but challenges remain due to the lack of a gold standard DNAmAge measure and variations in the predictive accuracies of various measures. Most studies have focused on associations between DNAmAge and phenotypic aging outcomes, but few have tested whether such associations occur because DNAm causes biological aging as opposed to merely correlating with it. Our study first identified two factors across five DNAmAge measures and then used a longitudinal twin design to improve causal inference between early-life exposures, DNAmAge, and ΔIQ in midlife.\u003c/p\u003e \u003cp\u003eAs hypothesized, Gen 2 DNAmAge, but not Gen 1, predicted ΔIQ in midlife. This result was consistent with several previous studies showing associations between second generation measures and cognition cross-sectionally \u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e,\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e and longitudinally \u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. PhenoAge and GrimAge algorithms both incorporate methylation signatures of clinical markers\u0026mdash;PhenoAge from blood-based biomarkers predicting physical function, multimorbidity, and mortality \u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e, and GrimAge from plasma proteins and smoking pack years selected for their prediction of time-to-death \u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e. These second generation DNAmAge measures have already shown improvements over first generation DNAmAge in predicting physical functioning, morbidity, and lifespan \u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e,\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e. Our finding that accelerated Gen 2 DNAmAge predicted more negative ΔIQ suggests that decline in cognitive ability is interwoven with other age-related physiological processes that associate with the methylome.\u003c/p\u003e \u003cp\u003eThe within-twin pair association between Gen 2 DNAmAge and ΔIQ that withstood controls for between family level confounds suggests that differences in environmental exposures partially explain their association. Two prior twin studies of DNAmAge and cognition had mixed findings \u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e,\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. Our Gen 1 findings were consistent with Starnawska and colleagues \u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e who did not find within-twin pair associations between Horvath and Hannum DNAmAge and cognition cross-sectionally or longitudinally. Our results differed from Vaccarino et al. \u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e who found no within-pair association between PhenoAge and GrimAge and cognitive decline. Differences in study design may explain this discrepancy: while our sample included both male and female twins, Vaccarino et al. examined only male veterans. Our study also spanned a longer age range, predicting change from childhood and adolescent IQ to midlife, whereas Vaccarino et al. used two testing points within participants\u0026rsquo; 50s and 60s. Importantly, the LTS lifespan longitudinal design allowed us to capture change from premorbid assessments of cognitive ability to midlife, a period when subtle, age-related declines may emerge.\u003c/p\u003e \u003cp\u003eWhile some prior studies found that lower childhood SES was associated with accelerated DNAmAge on the Horvath and Hannum clocks \u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e,\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e, others detected no significant effect \u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e,\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e. Our results supported the hypothesis that early-life SES would differentially affect Gen 1 and Gen 2 DNAmAge. Childhood SES predicted Gen 2 DNAmAge but not Gen 1 DNAmAge in midlife, consistent with findings from McCrory et al. \u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e, who found childhood poverty was associated with accelerated GrimAge in 50- to 87-year-olds. Childhood poverty has been found to influence health across the lifespan by limiting access to resources like education and healthcare \u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e. However, the interaction effect between DNAmAge and SES in our study offers a novel explanation for one pathway through which early-life stressors become biologically embedded. The relationship between accelerated epigenetic age and cognitive decline was amplified within twin pairs from lower-income households, suggesting that those raised in poorer families are more vulnerable to the cognitive effects associated with the broader physiological aging measured by DNAmAge.\u003c/p\u003e \u003cp\u003eTo our knowledge, only one previous study has conducted a factor analysis on multiple DNAmAge measures. In a model predicting multimorbidity and activities of daily living, Faul and colleagues \u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e found Horvath and Hannum measures loaded on a first factor and PhenoAge, GrimAge, and DunedinPACE loaded on a second factor. Our EFA results aligned with this, identifying a \u0026ldquo;Gen 1\u0026rdquo; factor with Horvath, Horvath Skin and Blood, and Hannum clocks, and a \u0026ldquo;Gen 2\u0026rdquo; factor with PhenoAge and GrimAge. Although we did not calculate DunedinPACE, a \u0026ldquo;third generation\u0026rdquo; measure that estimates rate of decline over nearly twenty years, future studies should compare its sensitivity with Gen 2 in the LTS sample \u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e. These results contribute to an understanding of the shared underlying patterns of measurement in \u0026ldquo;Gen 1\u0026rdquo; and \u0026ldquo;Gen 2\u0026rdquo; DNAmAge. Distinguishing between these two domains is a step towards refining the application of DNAmAge: our findings suggest second generation DNAmAge measures may have greater utility in studies of cognitive aging than first generation DNAmAge.\u003c/p\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eStrengths and Limitations\u003c/h2\u003e \u003cp\u003eUnlike prior studies described here that relied on retrospective accounts of childhood financial hardship or parental occupation, ours is the first to use a prospective rating of childhood SES (Duncan Index) which may have helped mitigate recall bias and improve reliability. However, the Duncan Index is a crude measure of parental occupation, and adding other socioeconomic indicators could improve future research. We did not test adult SES in this study, given that previous work had suggested childhood SES, but not adult SES, was predictive of DNAmAge. While our sample covered a broad socioeconomic range, it was predominantly white and recruited from one region of the United States, limiting generalizability. Future studies should examine the relationship between second generation DNAmAge and patterns of cognitive ability in larger and more diverse samples. Finally, we used IQ score as a proxy for global cognitive ability and future work should examine relationships between DNAmAge and specific cognitive domains including memory and executive functioning.\u003c/p\u003e \u003c/div\u003e"},{"header":"Materials and Methods","content":"\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eParticipants\u003c/h2\u003e \u003cp\u003eParticipants were 287 individual twins from the midlife phase of the Louisville Twin Study (LTS). The LTS began as a study of childhood development in multiple birth pairs. Twins born between 1950\u0026ndash;1997 in the Louisville metropolitan area were followed and assessed on cognitive ability, physical development, and temperament between 3 months and 15 years of age. Parents of twins provided data on demographics, parenting factors, and home environment. Zygosity was determined through blood typing during childhood. The sample contains same-sex monozygotic (MZ) and dizygotic (DZ) twin pairs, and opposite-sex DZ pairs. The LTS closed down in 2000 after 1,770 individual twins had been recruited \u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e. In 2018, follow-up of LTS twins aged 30\u0026ndash;65 was initiated \u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e, and the LTS reopened in 2019 to begin the first midlife data collection of the 1,770 twins. The study was approved by the University of Louisville Institutional Review Board (19.0989). Participants provided a 50cc whole blood sample for genotyping and DNA methylation assay. Genotyping confirmed zygosity for any unknown twin pairs. The present sample includes 60 MZ pairs, 41 DZ pairs, and 85 individual twins who completed a midlife study visit between 2019\u0026ndash;2023. Supplementary Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e illustrates the participant flow, attrition, and retention across study phases.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eSocioeconomic Status\u003c/h2\u003e \u003cp\u003eChildhood socioeconomic status (SES) was measured using the Duncan socioeconomic index which assigns scores of 0-100 based on the occupation of the head of the household \u003csup\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e. Occupational prestige was based on 1950 US Census income data and education level associated with the occupation. The SES of our sample (M\u0026thinsp;=\u0026thinsp;47.3, SD\u0026thinsp;=\u0026thinsp;25.7) was consistent with the overall LTS distribution (M\u0026thinsp;=\u0026thinsp;46.9, SD\u0026thinsp;=\u0026thinsp;26.9), evenly distributed, and representative of the Louisville, Kentucky area (Davis et al., 2019).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eCognitive Ability Measures\u003c/h2\u003e \u003cp\u003eLTS participants were assessed using validated Wechsler batteries of cognitive functioning. Children and adolescents were administered one of three forms of the Wechsler Intelligence Scale for Children (WISC, WISC-R, WISC-III; \u003csup\u003e\u003cspan additionalcitationids=\"CR35\" citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e. Adults in the midlife phase were assessed using the Wechsler Adult Intelligence Scale IV (WAIS-IV; \u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e. Test scores are age-standardized based on a standardization sample that is roughly representative of the United States. The Wechsler batteries generate a Full-Scale Intelligence Quotient (IQ), which is an overall estimate of an individual\u0026rsquo;s current level of global cognitive ability, with a mean of 100 and a standard deviation of 15. Children were typically administered cognitive assessments at ages 7, 8, 9, 12, and 15, though not all twins were assessed at each timepoint; this study used the most recently available WISC assessment for each participant, with over 80% collected at age 15 (see Table\u0026nbsp;1 for more detailed descriptive statistics).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eDNAmAge Measures\u003c/h2\u003e \u003cp\u003eGenomic DNA was extracted from whole-blood samples that were collected at the University of Louisville and Norton Healthcare medical campuses. Venous blood was collected in ethylenediaminetetraacetic acid (EDTA) tubes and shipped to the Norris Comprehensive Cancer Institute at the University of Southern California Keck School of Medicine. DNA was extracted with promega Maxwell 16 LEV blood DNA kits, and then treated with bisulfite reagents at the USC Molecular Genomics Core following manufacturer protocol. Methylation was assayed with the Illumina Infinium Human MethylationEPIC BeadChip (Illumina, San Diego, CA, USA) at the USC Molecular Genomics Core at the Norris Cancer Institute at the Keck School of Medicine.\u003c/p\u003e \u003cp\u003eQuality control was conducted in R 4.3.0 \u003csup\u003e38\u003c/sup\u003e using the \u003cem\u003eminfi\u003c/em\u003e package \u003csup\u003e\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e to identify aberrant samples and CpG sites, remove cross-reactive probes, conduct background correction, and adjust for batch effects prior to estimating DNAmAge variables. Background correction was performed using the normal-exponential out-of-band (noob) method \u003csup\u003e\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e. The ComBat method was used to adjust for laboratory batches \u003csup\u003e\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eWe estimated blood cell composition using the Houseman method \u003csup\u003e\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e and obtained percentages of CD8\u0026thinsp;+\u0026thinsp;T cells, CD4\u0026thinsp;+\u0026thinsp;T cells, natural killer cells (NK), B cells, monocytes, and granulocytes. Cell proportions are highly correlated, so in the current analysis we included the first and second principal components from a principal components (PC) analysis (collectively, they explained 86.12% of the variance) in all analyses.\u003c/p\u003e \u003cp\u003eAs replicate blood samples often have different methylation values due to noise in the CpG sites, batch effects, and sample preparation, Higgins-Chen et al. \u003csup\u003e\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e developed a principal components-based method to bolster reliability in DNAmAge algorithms by training them based on PC analysis, extracting the covariance between multicollinear CpGs, including age-related covariance. We estimated DNAmAge using five PC-trained algorithms, Horvath, Horvath Skin and Blood, Hannum, PhenoAge, and GrimAge \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e,\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eData Analysis\u003c/h2\u003e \u003cp\u003eDescriptive statistics for the sample\u0026rsquo;s demographic characteristics, survey data, and cognitive ability data were computed in R version 4.3.2 \u003csup\u003e38\u003c/sup\u003e. Methylation data were cleaned and prepared in R. DNAmAge was calculated as the residual difference in PC-trained values predicted from chronological age in a bivariate linear regression model. DNAmAge values were adjusted for cell composition. Age and cell-adjusted variables were then winsorized such that values more than two standard deviations from the mean DNAmAge score were replaced with a value two standard deviations above or below the mean. Pairwise correlations for the five DNAmAge measures were then estimated using maximum likelihood in M\u003cem\u003eplus\u003c/em\u003e Version 8 \u003csup\u003e44\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eWe used the classical twin model to decompose the phenotypic variance of DNAmAge into additive genetic (A), shared environmental (C), and nonshared environmental (E) components. Given that MZ twins share 100% of their genetic makeup and DZ twin pairs share half, on average, the A variance is correlated 1.0 within MZ twin pairs and 0.5 within DZ twin pairs. In both MZ and DZ twin pairs, the C component includes experiences that make siblings raised together more similar, and is correlated 1.0. Nonshared environmental components are unique to individuals and by definition uncorrelated within twin pairs. We estimated a univariate ACE model for all five DNAmAge measures individually using M\u003cem\u003eplus\u003c/em\u003e.\u003c/p\u003e \u003cp\u003eExploratory factor analysis was conducted in M\u003cem\u003eplus\u003c/em\u003e to identify underlying dimensions across the five DNAm clock variables (Horvath, Horvath Skin and Blood, Hannum, PhenoAge, and GrimAge). The number of factors to be rotated was determined by the scree plot of Eigenvalues (Supplementary Fig. S2) and a likelihood ratio test comparing a two-factor model to a one-factor model. EFA models were estimated using full-information maximum likelihood with an oblique (geomin) rotation. Factor loadings were evaluated for statistical significance, with a criterion of p\u0026thinsp;\u0026lt;\u0026thinsp;.05 (Supplementary Table S3). We estimated composite scores by standardizing the five DNAmAge variables and using the sum of the standardized values for the variables in each factor. Once composite scores were computed, univariate ACE models were estimated for each factor, decomposing the variance of DNAmAge composite scores into genetic, shared environmental, and nonshared environmental components.\u003c/p\u003e \u003cp\u003eTo explore the primary aims of the study, the relationship between each DNAmAge factor and ΔIQ was modeled in two ways. Both the phenotypic association model and \u0026ldquo;quasi-causal\u0026rdquo; regression model were fit in M\u003cem\u003eplus\u003c/em\u003e. The phenotypic association model is equivalent to a simple phenotypic regression where the relationship between predictor and outcome is examined without controlling for genetic and shared environmental effects. This was estimated prior to all genetically informed models. Using a dataset wide by twin pair and estimated parameters set to be equal across the members of a pair, we regressed midlife IQ on childhood IQ, childhood SES, DNAmAge, and the interaction between SES and DNAmAge.\u003c/p\u003e \u003cp\u003eNext, we fit the quasi-causal regression model to estimate the phenotypic regression coefficients conditional on \u003cem\u003eb\u003c/em\u003e\u003csub\u003eA\u003c/sub\u003e and \u003cem\u003eb\u003c/em\u003e\u003csub\u003eC\u003c/sub\u003e paths. The classical twin method was used to partition childhood IQ, adult IQ, and DNAmAge into A, C, and E components. In this case, midlife IQ was regressed on childhood IQ, DNAmAge, and the A and C components of DNAmAge and childhood IQ, as illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Childhood SES was used as a linear covariate for DNAmAge and the IQ variables. We also estimated an interaction between childhood SES and DNAmAge to determine whether the strength of the relationship between DNAmAge and ΔIQ depended on the SES of the family in which twins were raised. A statistically significant \u003cem\u003eb\u003c/em\u003e\u003csub\u003e\u003cb\u003eP3\u003c/b\u003e\u003c/sub\u003e coefficient affirmatively answers the question: Within a pair of identical twins, does the twin with the higher DNAmAge value also have the larger positive or negative ΔIQ, statistically adjusting for the effects of the A and C components of DNAmAge (\u003cem\u003eb\u003c/em\u003e\u003csub\u003eA3\u003c/sub\u003e and \u003cem\u003eb\u003c/em\u003e\u003csub\u003eC3\u003c/sub\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eData Availability\u003c/h2\u003e \u003cp\u003eThe datasets analyzed during the current study are not publicly available but deidentified data and/or code are available from the corresponding author on reasonable request.\u003c/p\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003ch2\u003eFunding\u003c/h2\u003e \u003cp\u003eFunding was provided by the National Institute on Aging Grant No. R01AG063949.\u003c/p\u003e \u003c/p\u003e\u003cp\u003e \u003ch2\u003eCompeting Interest Statement\u003c/h2\u003e \u003cp\u003eNo competing interests.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eS.A.B., E.T., and C.R.B. conceptualized and designed the research and methodology. C.R.B., E.Z., D.W.D. helped collect, generate and/ or process the data. S.A.B. and E.T. executed the statistical analysis and data visualization and wrote the original draft. S.A.B., C.R.B., E.Z., A.K., E.A., J.B., D.F., D.W.D., and E.T. contributed to the interpretation of the results and revisions of the final manuscript.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe datasets analyzed during the current study are not publicly available but de-identified data and/or code are available from the corresponding author on reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eHorvath, S. DNA methylation age of human tissues and cell types. \u003cem\u003eGenome Biol.\u003c/em\u003e \u003cb\u003e14\u003c/b\u003e, R115 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOblak, L., van der Zaag, J., Higgins-Chen, A. T., Levine, M. E. \u0026amp; Boks, M. P. A systematic review of biological, social and environmental factors associated with epigenetic clock acceleration. \u003cem\u003eAgeing Res. Rev.\u003c/em\u003e \u003cb\u003e69\u003c/b\u003e, 101348 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRyan, J., Wrigglesworth, J., Loong, J., Fransquet, P. D. \u0026amp; Woods, R. L. A Systematic Review and Meta-analysis of Environmental, Lifestyle, and Health Factors Associated With DNA Methylation Age. \u003cem\u003eJournals Gerontology: Ser. A\u003c/em\u003e. \u003cb\u003e75\u003c/b\u003e, 481\u0026ndash;494 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFransquet, P. D., Wrigglesworth, J., Woods, R. L., Ernst, M. E. \u0026amp; Ryan, J. The epigenetic clock as a predictor of disease and mortality risk: a systematic review and meta-analysis. \u003cem\u003eClin. Epigenetics\u003c/em\u003e. \u003cb\u003e11\u003c/b\u003e, 62 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHorvath, S. \u0026amp; Ritz, B. R. Increased epigenetic age and granulocyte counts in the blood of Parkinson\u0026rsquo;s disease patients. \u003cem\u003eAging (Albany NY)\u003c/em\u003e. \u003cb\u003e7\u003c/b\u003e, 1130\u0026ndash;1142 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLevine, M. E., Lu, A. T., Bennett, D. A. \u0026amp; Horvath, S. Epigenetic age of the pre-frontal cortex is associated with neuritic plaques, amyloid load, and Alzheimer\u0026rsquo;s disease related cognitive functioning. \u003cem\u003eAging (Albany NY)\u003c/em\u003e. \u003cb\u003e7\u003c/b\u003e, 1198\u0026ndash;1211 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBell, C. G. et al. DNA methylation aging clocks: challenges and recommendations. \u003cem\u003eGenome Biol.\u003c/em\u003e \u003cb\u003e20\u003c/b\u003e, 249 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHannum, G. et al. Genome-wide Methylation Profiles Reveal Quantitative Views of Human Aging Rates. \u003cem\u003eMol. Cell\u003c/em\u003e. \u003cb\u003e49\u003c/b\u003e, 359\u0026ndash;367 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHorvath, S. et al. Epigenetic clock for skin and blood cells applied to Hutchinson Gilford Progeria Syndrome and ex vivo studies. \u003cem\u003eAging (Albany NY)\u003c/em\u003e. \u003cb\u003e10\u003c/b\u003e, 1758\u0026ndash;1775 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVaccarino, V. et al. Epigenetic Age Acceleration and Cognitive Decline: A Twin Study. \u003cem\u003eJ. Gerontol. Biol. Sci. Med. Sci.\u003c/em\u003e \u003cb\u003e76\u003c/b\u003e, 1854\u0026ndash;1863 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFaul, J. D. et al. Epigenetic-based age acceleration in a representative sample of older Americans: Associations with aging-related morbidity and mortality. \u003cem\u003eProceedings of the National Academy of Sciences\u003c/em\u003e 120, e2215840120 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMaddock, J. et al. Childhood growth and development and DNA methylation age in mid-life. \u003cem\u003eClin. Epigenet\u003c/em\u003e. \u003cb\u003e13\u003c/b\u003e, 155 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eReed, R. G., Carroll, J. E., Marsland, A. L. \u0026amp; Manuck, S. B. DNA methylation-based measures of biological aging and cognitive decline over 16-years: preliminary longitudinal findings in midlife. \u003cem\u003eAging (Albany NY)\u003c/em\u003e. \u003cb\u003e14\u003c/b\u003e, 9423\u0026ndash;9444 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStarnawska, A. et al. Blood DNA methylation age is not associated with cognitive functioning in middle-aged monozygotic twins. \u003cem\u003eNeurobiol. Aging\u003c/em\u003e. \u003cb\u003e50\u003c/b\u003e, 60\u0026ndash;63 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLevine, M. E. et al. An epigenetic biomarker of aging for lifespan and healthspan. \u003cem\u003eAging\u003c/em\u003e \u003cb\u003e10\u003c/b\u003e, 573\u0026ndash;591 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLu, A. T. et al. DNA methylation GrimAge strongly predicts lifespan and healthspan. \u003cem\u003eAging\u003c/em\u003e \u003cb\u003e11\u003c/b\u003e, 303\u0026ndash;327 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGraves, A. J. et al. Accelerated epigenetic age is associated with whole-brain functional connectivity and impaired cognitive performance in older adults. \u003cem\u003eSci. Rep.\u003c/em\u003e \u003cb\u003e14\u003c/b\u003e, 9646 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHillary, R. F. et al. An epigenetic predictor of death captures multi-modal measures of brain health. \u003cem\u003eMol. Psychiatry\u003c/em\u003e. \u003cb\u003e26\u003c/b\u003e, 3806 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAartsen, M. J. et al. Advantaged socioeconomic conditions in childhood are associated with higher cognitive functioning but stronger cognitive decline in older age. \u003cem\u003eProc. Natl. Acad. Sci. U S A\u003c/em\u003e. \u003cb\u003e116\u003c/b\u003e, 5478\u0026ndash;5486 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcCrory, C. et al. Early life adversity and age acceleration at mid-life and older ages indexed using the next-generation GrimAge and Pace of Aging epigenetic clocks. \u003cem\u003ePsychoneuroendocrinology\u003c/em\u003e \u003cb\u003e137\u003c/b\u003e, 105643 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAustin, M. K. et al. Early-life socioeconomic disadvantage, not current, predicts accelerated epigenetic aging of monocytes. \u003cem\u003ePsychoneuroendocrinology\u003c/em\u003e \u003cb\u003e97\u003c/b\u003e, 131\u0026ndash;134 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHughes, A. et al. Socioeconomic Position and DNA Methylation Age Acceleration Across the Life Course. \u003cem\u003eAm. J. Epidemiol.\u003c/em\u003e \u003cb\u003e187\u003c/b\u003e, 2346\u0026ndash;2354 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcCrory, C. et al. How does socio-economic position (SEP) get biologically embedded? A comparison of allostatic load and the epigenetic clock(s). \u003cem\u003ePsychoneuroendocrinology\u003c/em\u003e \u003cb\u003e104\u003c/b\u003e, 64\u0026ndash;73 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJylh\u0026auml;v\u0026auml;, J. et al. Longitudinal changes in the genetic and environmental influences on the epigenetic clocks across old age: Evidence from two twin cohorts. \u003cem\u003eeBioMedicine\u003c/em\u003e \u003cb\u003e40\u003c/b\u003e, 710\u0026ndash;716 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMarioni, R. E. et al. DNA methylation age of blood predicts all-cause mortality in later life. \u003cem\u003eGenome Biol.\u003c/em\u003e \u003cb\u003e16\u003c/b\u003e, 25 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTurkheimer, E. \u0026amp; Harden, K. P. Behavior genetic research methods: Testing quasi-causal hypotheses using multivariate twin data. in Handbook of research methods in social and personality psychology, \u003cem\u003e2nd ed\u003c/em\u003e 159\u0026ndash;187 (Cambridge University Press, New York, NY, US, (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBeam, C. R. et al. Midlife Study of the Louisville Twins: Connecting Cognitive Development to Biological and Cognitive Aging. \u003cem\u003eBehav. Genet.\u003c/em\u003e \u003cb\u003e50\u003c/b\u003e, 73\u0026ndash;83 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLevine, M. E. Assessment of Epigenetic Clocks as Biomarkers of Aging in Basic and Population Research. \u003cem\u003eJournals Gerontology: Ser. A\u003c/em\u003e. \u003cb\u003e75\u003c/b\u003e, 463\u0026ndash;465 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFiorito, G. et al. Social adversity and epigenetic aging: a multi-cohort study on socioeconomic differences in peripheral blood DNA methylation. \u003cem\u003eSci. Rep.\u003c/em\u003e \u003cb\u003e7\u003c/b\u003e, 16266 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBarboza Sol\u0026iacute;s, C. et al. Mediating pathways between parental socio-economic position and allostatic load in mid-life: Findings from the 1958 British birth cohort. \u003cem\u003eSoc. Sci. Med.\u003c/em\u003e \u003cb\u003e165\u003c/b\u003e, 19\u0026ndash;27 (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBelsky, D. W. et al. DunedinPACE, a DNA methylation biomarker of the pace of aging. \u003cem\u003eeLife\u003c/em\u003e 11, (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDavis, D. W., Turkheimer, E., Finkel, D., Beam, C. \u0026amp; Ryan, L. The Louisville Twin Study: Past, Present and Future. \u003cem\u003eTwin Res. Hum. Genet.\u003c/em\u003e \u003cb\u003e22\u003c/b\u003e, 735\u0026ndash;740 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDuncan, O. A socioeconomic index for all occupations. \u003cem\u003eOccupations social status\u003c/em\u003e (1961).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWechsler, D. \u003cem\u003eWechsler Intelligence Scale for Children\u003c/em\u003e (Psychological Corporation, 1949).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWechsler, D. \u003cem\u003eWechsler Intelligence Scale for Children-Revised\u003c/em\u003e (The Psychological Corporation, 1974).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWechsler, D. \u003cem\u003eWechsler Intelligence Scale for Children\u003c/em\u003e (The Psychological Corporation, 1991).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWechsler, D. Wechsler Adult Intelligence Scale - Fourth Edition Administration and Scoring Manual. (2008). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://psycnet.apa.org/doiLanding?doi=10.1037%2Ft15169-000\u003c/span\u003e\u003cspan address=\"https://psycnet.apa.org/doiLanding?doi=10.1037%2Ft15169-000\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eR Core Team. \u003cem\u003eR: A Language and Environment for Statistical Computing\u003c/em\u003e (R Foundation for Statistical Computing, 2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAryee, M. J. et al. Minfi: a flexible and comprehensive Bioconductor package for the analysis of Infinium DNA methylation microarrays. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cb\u003e30\u003c/b\u003e, 1363\u0026ndash;1369 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTriche, T. J., Weisenberger, D. J., Van Den Berg, D., Laird, P. W. \u0026amp; Siegmund, K. D. Low-level processing of Illumina Infinium DNA Methylation BeadArrays. \u003cem\u003eNucleic Acids Res.\u003c/em\u003e \u003cb\u003e41\u003c/b\u003e, e90 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJohnson, W. E., Li, C. \u0026amp; Rabinovic, A. Adjusting batch effects in microarray expression data using empirical Bayes methods. \u003cem\u003eBiostatistics\u003c/em\u003e \u003cb\u003e8\u003c/b\u003e, 118\u0026ndash;127 (2007).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHouseman, E. A. et al. DNA methylation arrays as surrogate measures of cell mixture distribution. \u003cem\u003eBMC Bioinform.\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e, 86 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHiggins-Chen, A. T. et al. A computational solution for bolstering reliability of epigenetic clocks: implications for clinical trials and longitudinal tracking. \u003cem\u003eNat. Aging\u003c/em\u003e. \u003cb\u003e2\u003c/b\u003e, 644\u0026ndash;661 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMuth\u0026eacute;n, L. K. \u0026amp; Muth\u0026eacute;n, B. O. \u003cem\u003eMplus: Statistical Analysis with Latent Variables: User\u0026rsquo;s Guide (Version 8)\u003c/em\u003e (Muth\u0026eacute;n \u0026amp; Muth\u0026eacute;n, 1998).\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-5551592/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5551592/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eDNA methylation age (DNAmAge) surpasses chronological age in its ability to predict age-related morbidities and mortality. This study analyzed data from 287 middle-aged twins in the Louisville Twin Study (mean age 51.9 years\u0026thinsp;\u0026plusmn;\u0026thinsp;7.03) to investigate the effect of DNAmAge acceleration on change in IQ (ΔIQ) between childhood and midlife, while testing childhood socioeconomic status (SES) as a moderator of the relationship. DNAmAge was estimated with five commonly used algorithms (Horvath, Horvath Skin and Blood, GrimAge, and PhenoAge). A factor analysis of these measures produced a two-factor structure which we identified as first generation and second generation measures. Results of genetically informed, quasi-causal regression models indicated that accelerated second generation DNAmAge predicted more negative ΔIQ from childhood to midlife, after accounting for genetic and environmental confounds shared by twins. The relationship between DNAmAge and ΔIQ was moderated by childhood SES, with a stronger effect observed among twins from low SES backgrounds. Second generation DNAmAge measures trained to estimate phenotypic biological age show promise in their predictive value for cognitive decline in midlife. Our genetically informed twin design suggested that DNAmAge may represent a biological pathway through which early-life socioeconomic disadvantage impacts midlife cognitive health.\u003c/p\u003e","manuscriptTitle":"Second Generation DNA Methylation Age Predicts Cognitive Change in Midlife: The Moderating Role of Childhood Socioeconomic Status","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-12-24 01:27:38","doi":"10.21203/rs.3.rs-5551592/v1","editorialEvents":[{"type":"communityComments","content":1}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"32f3ef6f-ddee-46b9-94fa-01349f6bfa96","owner":[],"postedDate":"December 24th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":41942692,"name":"Biological sciences/Psychology"},{"id":41942693,"name":"Biological sciences/Genetics/Behavioural genetics"},{"id":41942694,"name":"Biological sciences/Genetics/Epigenetics"}],"tags":[],"updatedAt":"2025-08-07T19:51:04+00:00","versionOfRecord":{"articleIdentity":"rs-5551592","link":"https://doi.org/10.18632/aging.206284","journal":{"identity":"aging","isVorOnly":true,"title":"Aging"},"publishedOn":"2025-07-23 00:00:00","publishedOnDateReadable":"July 23rd, 2025"},"versionCreatedAt":"2024-12-24 01:27:38","video":"","vorDoi":"10.18632/aging.206284","vorDoiUrl":"https://doi.org/10.18632/aging.206284","workflowStages":[]},"version":"v1","identity":"rs-5551592","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5551592","identity":"rs-5551592","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}