ABSTRACT
Background Childhood adversity influences long-term health, particularly if experienced during sensitive periods in development when physiological systems are more responsive to environmental influences. Although the underlying mechanisms remain unclear, prior studies suggest that DNA methylation (DNAm) may capture these time-dependent effects of childhood adversity. However, it remains unknown whether DNAm alterations persist into adolescence and how the timing of adversity might influence DNAm trajectories across development.
Methods We examined the relationship between time-dependent adversity and genome-wide DNAm measured at three waves from birth to adolescence using prospective data from the Avon Longitudinal Study of Parents and Children. We first assessed the relationship between the timing of exposure to seven types of adversity (measured 5-8 times between ages 0-11) and DNAm at age 15 using a structured life course modeling approach. We also characterized the persistence into adolescence of associations identified from age 7 DNAm, as well as the influence of adversity on DNAm trajectories from ages 0-15.
Results Adversity was associated with differences in age 15 DNAm at 24 loci (FDR<0.05). Most loci (19 of 24) were associated with adversity (i.e., physical/sexual abuse, one-adult households, caregiver abuse) that occurred between ages 3-5. Although no DNAm differences present at age 7 persisted into adolescence, we identified seven unique types of DNAm trajectories across development, which highlighted diverse effects of childhood adversity on DNAm.
Conclusions Our results suggest that childhood adversity, particularly between ages 3-5, can influence the trajectories of DNAm across development, exerting both immediate and latent effects on the epigenome.
INTRODUCTION
Childhood adversity, such as abuse or maltreatment (1, 2), family disruption or dysfunction (3, 4), or poverty (5, 6), is one of the most potent determinants of poor physical and mental health in both children and adults (7–9). While the mechanisms underlying the biological embedding of childhood adversity are not yet well understood, epigenetic processes, such as DNA methylation (DNAm), have emerged as one potential pathway to bridge genetic factors and life experiences (10). Large-scale population-based studies, systematic reviews, and meta-analyses have shown that DNAm signatures in humans are responsive to life experiences, including exposure to childhood adversity across the life course (11–16). However, prior studies on childhood adversity and DNAm have not fully explored two key dimensions of this relationship, which are critical to assess the biological risk posed by childhood adversity and to better target interventions for health promotion and disease prevention.
First, it remains unclear whether the timing of childhood adversity plays a role in shaping DNAm. Emerging evidence from human populations and animal models suggests there may be sensitive periods in epigenetic programming, when physiological and neurobiological systems may be primed for external influences, which can, in turn, impart more enduring effects on health (17–21). Few studies have investigated whether there are specific periods when childhood adversity may have greater effects on DNAm (15, 22), with no studies investigating epigenetic patterns in adolescence. Thus, it remains unknown whether there are sensitive periods during which adversity may exert greater influences on DNAm, and in turn on adolescent health.
Second, few studies have assessed the role of childhood adversity in relation to longitudinal measures of DNAm across development (i.e., DNAm trajectories). A recent review article argues that analyses examining DNAm at a single moment in development have limited capacity to identify robust associations with health outcomes due to the dynamic nature of epigenetic mechanisms (23). Rather, chrono-epigenetic patterns (i.e., the longitudinal dynamics of epigenetic processes) may more adequately predict the immediate and long-term effects of life experiences. To our knowledge, only six studies have assessed the influence of early-life experiences on the trajectories of DNAm across development. These have mainly focused on biological markers of prenatal environments, such as gestational age and birthweight (24), or have examined early-life stressors individually, including maternal pre-pregnancy BMI and gestational weight gain (25), prenatal maternal smoking (26), socio-economic disadvantage during childhood (27, 28), and adolescent victimization (29). However, no study has examined how the timing of multiple types of childhood adversity may influence DNAm trajectories across development. Such knowledge would provide deeper insight into the molecular underpinnings of human health, while also help identify those at greater risk for the negative effects of adversity on health outcomes.
To address these gaps, we examined the longitudinal relationship between early-life adversity and genome-wide DNAm across childhood and adolescence, using data collected over two decades from a subsample of the Avon Longitudinal Study of Parents and Children (ALSPAC) cohort. Here, we examined the associations between exposure to seven types of childhood adversity, assessed repeatedly between birth and age 11, and DNAm at age 15. Given the unique availability of three waves of DNAm in this cohort (at birth, age 7, and age 15), we also examined DNAm trajectories from birth to adolescence. Our aims were to: 1) determine whether adolescent DNAm captured the time-dependent effects of childhood adversity; 2) characterize the developmental trajectories of DNAm linked to adversity; and 3) evaluate the persistence of previously-identified associations between adversity and DNAm in childhood (22). To our knowledge, this is the first study to investigate the time-dependent influences of childhood adversity on adolescent DNAm and trajectories of DNAm across development.
MATERIALS AND METHODS
Sample
Data came from the ALSPAC, a large population-based birth cohort from Avon, UK of 14,451 children followed from before birth through early adulthood (30, 31) (see Supplemental Materials for details). The ALSPAC study website contains details of all the data available through a fully searchable data dictionary and variable search tool (http://www.bristol.ac.uk/alspac/researchers/our-data/). Ethical approval for the study was obtained from the ALSPAC Law and Ethics Committee and the Local Research Ethics Committees. Consent for biological samples was collected in accordance with the Human Tissue Act (2004). Informed consent for the use of data collected via questionnaires and clinics was obtained from participants following the recommendations of the ALSPAC Ethics and Law Committee. All data are available by request from the ALSPAC Executive Committee for researchers who meet the criteria for access to confidential data (http://www.bristol.ac.uk/alspac/researchers/access/). Secondary analyses of ALSPAC data were approved with oversight by the Mass General Brigham Institutional Review Boards (IRB) (Protocol 2017P001110).
Measures of childhood adversity
We examined the effect of seven types of childhood adversity previously associated with DNAm (32–35): 1) caregiver physical or emotional abuse; 2) sexual or physical abuse (by anyone)(36–39); 3) maternal psychopathology; 4) one adult in the household; 5) family instability; 6) financial hardship; and 7) neighborhood disadvantage. These adversities were generated from maternal reports via mailed questionnaires, collected 5-8 times between birth and age 11 (Figure 1; see Table S2 for a complete description of childhood adversities).
DNAm data generation
Blood-based DNAm profiles were generated at birth, age 7, and age 15 as part of the Accessible Resource for Integrated Epigenomics Studies (ARIES), a subsample of 1,018 mother-child pairs randomly selected from those with complete data across five or more waves of ALSPAC data collection (Supplemental Materials) (40). DNAm was measured at 485,577 CpG sites using the Illumina Infinium HumanMethylation450 BeadChip microarray (Illumina, San Diego, CA). DNA for this assay was extracted from cord blood at birth, whole blood at age 7, and peripheral blood leukocytes collected at age 15. Laboratory procedures, preprocessing analyses, and quality control steps performed have been described previously (40).
DNA methylation data pre-processing and normalization
DNAm data were processed using the meffil package in R, which performs background correction and functional normalization of DNAm data (41). Twins and samples with >10% of CpG sites with a detection p-value >0.01 or a bead count <3 were removed, as were cross-hybridizing probes and polymorphic probes. To remove possible outliers, we winsorized the beta values (i.e., values that represent the percent of methylation at each CpG site), setting the bottom 5% and top 5% of values to the 5th and 95th quantile, respectively (42). Finally, we removed probes showing little variability across individuals, defined as CpGs with <5% difference in DNAm between the 10th and 90th percentile of values. The final analytic sample after pre-processing consisted of 966 youths and 302,581 CpGs with DNAm data measured at age 15. DNAm measured at age 0 and 7 were similarly pre-processed and normalized (Supplemental materials).
Covariates
To account for potential confounding and be consistent with previous work on childhood adversities (15), we included the following covariates: age of blood collection, sex, race/ethnicity, maternal age at birth, maternal education at birth, birthweight, number of previous pregnancies, maternal smoking during pregnancy, and cell type proportions estimated using the Houseman method (43). See Supplemental Materials for variable coding.
Analyses
Structured Life Course Modeling Approach (SLCMA)
Our primary analyses focused on identifying time-dependent associations between each type of childhood adversity and DNAm measured in adolescence (age 15). To identify these associations, we used the structured life course modeling approach (SLCMA; pronounced “slick-mah”), which is a two-stage method that simultaneously compares different a priori-specified hypotheses that explain exposure-outcome relationships (44–46). The SLCMA uses variable selection to identify the life course hypothesis explaining the greatest proportion of outcome variation. Estimates confidence intervals and p-values are calculated for the selected life course hypothesis, using post-selective inference to remove bias associated with multiple testing and variable selection. The SLCMA has been successfully applied to high-dimensional DNAm data to identify sensitive periods in development that can influence genome-wide DNAm patterns (15, 22, 47).
We tested time-dependent exposure to adversity for the timepoint shown in Figure 1. We interpreted exposure to each adversity type through six separate life course hypotheses, including four sensitive periods hypotheses that encoded exposure to each childhood adversity during: 1) very early childhood (ages 0-2), 2) early childhood (ages 3-5), 3) middle childhood (ages 6-7), 4) late childhood (ages 8-11); and two additive hypotheses: 5) total number exposures across childhood (accumulation), and 6) total number of exposures weighted by age (recency), which allowed us to assess if more recent exposures had a stronger impact than distal exposures.
We used selective inference to perform post-selection inference (48) and adjusted for covariates using the Frisch-Waugh-Lovell theorem (49), which has been shown to improve statistical power in penalized regression analyses (47, 50). Only complete cases (i.e., individuals with non-missing data on covariates and exposures from ages 0-11) were analyzed for each adversity (Figure 1). We accounted for multiple-testing using the Benjamini-Hochberg method to control the false-discovery rate at 5% (FDR<0.05) (51), or a Bonferroni-corrected threshold of p<1.65×10−7.
Functional analyses of SLCMA results
We investigated the biological implications of our findings from the SLCMA in four different ways. First, we assessed the enrichment of regulatory elements in top loci compared to the all analyzed loci using chi-squared tests. Second, we examined the correlation of DNAm at the top loci in blood and four different brain regions using the Blood Brain DNA Methylation Comparison Tool (52). Third, we analyzed the enrichment of biological processes in top loci using gene ontology (GO) terms from the DAVID tool (53, 54). Fourth, we assessed the evolutionary constraint of genes linked to top loci using data from the Exome Aggregation Consortium (55).
Trajectories of DNAm response to childhood adversity
The three waves of longitudinal DNAm data available for the ALSPAC cohort provide a unique opportunity to investigate the developmental patterns of DNAm across development. Building from the SLCMA findings, we pursued three additional sets of analyses.
Pre-existence of age 15 associations
First, we determined whether DNAm differences present at age 15 emerged earlier in development. Using linear regression, we tested whether exposure to the adversity selected in the SLCMA at age 15 was associated with DNAm at the same top loci at birth or age 7, while adjusting for covariates.
Types of DNAm trajectories across development
Second, we investigated patterns of DNAm change beyond differences between single time points, focusing on patterns of change and stability among the top loci identified from the SLCMA of age 15 DNAm. Specifically, we examined the longitudinal DNAm patterns of individuals in three distinct exposure groups: 1) adversity exposure during the period identified from the SLCMA (labeled as exposed-SP); 2) adversity exposure outside the period identified from the SLCMA (exposed-other); and 3) never exposed. We then performed an ANOVA of the statistical interaction between age at DNAm collection and exposure group, controlling for DNAm repeated measures in the ANOVA as fixed effects, which allowed us to determine whether any exposure group differed from another in its longitudinal DNAm trajectory. Loci showing significant differences for group-by-age interactions (FDR<0.05) were carried forward into the next stage, as they showed more granular differences between exposure groups across development.
From these loci, we identified groupings of loci with similar longitudinal patterns (i.e., types of DNAm trajectories). Similarities were defined in relation to three main distinguishing features: 1) mean exposure group differences across ages 2) mean age differences across exposure groups, and 3) exposure group differences within each age. These features were determined using Tukey post-hoc analyses of the ANOVA described above. We then performed hierarchical clustering of these features to identify homogeneous subsets of trajectories (Supplemental materials).
Persistence of childhood DNAm differences to adolescence
Third, we assessed whether DNAm alterations linked to childhood adversity, which we previously identified at age 7 (22), persisted until adolescence. For these analyses, we performed linear models between adversity and age 15 DNAm data for these 48 childhood loci only (Supplemental materials).
RESULTS
Sample characteristics and prevalence of exposure to adversity
Demographic characteristics of the ARIES sample and subset of children with any exposure to adversity between ages 0-11 can be found in Table S1. The prevalence of exposure to a given adversity between the ages of 0-11 ranged from 15.1% (sexual/physical abuse) to 34.8% (maternal psychopathology) (Figure S1; Table S3). The correlation of exposure within each adversity across different developmental ages ranged from 0.36 (family instability) to 0.786 (one adult in the household). The different types of adversity were weakly correlated with each other, ranging from an average correlation of −0.04 (family instability) to 0.16 (maternal psychopathology), suggesting that these measures of childhood adversity captured distinct exposures.
Childhood adversity showed time-dependent associations with adolescent DNAm profiles
Across all types of adversity, 24 loci showed significant associations between exposure to adversity and DNAm levels at age 15 at FDR<0.05; 7 of these loci were also significant at a Bonferroni-corrected p-value <1.65×10−7 (Table 1).
Among FDR-significant loci, sensitive period models were most often selected, with 21 loci showing associations with childhood adversity that occurred during very early childhood (2 of 24), early childhood (18 of 24), or late childhood (1 of 24) (Figure 2A). By contrast, only 3 loci showed associations with the accumulation of adversity across development.
A large proportion of effects were for one-adult households in the FDR-significant loci (18 of 24 loci). We also identified associations with caregiver physical/emotional abuse (2 loci), sexual or physical abuse by anyone (3 loci), and maternal psychopathology (1 locus). No associations were observed with family instability or neighborhood disadvantage.
Exposure to adversity was most often associated with a decrease in DNAm levels (21 out of 24). On average, childhood adversity exposure was linked to a 3.6% absolute difference in DNA methylation levels (range 1.5-10.4%). For the three loci associated with accumulated time with one adult in the household, each additional timepoint with one adult was associated with a 1% difference in DNAm (range 0.3-1.4%).
Biological relevance of adolescent loci associated with time-dependent childhood adversity
To further understand the biological implications of loci associated with childhood adversity, we assessed their genomic context, relationship with brain DNA methylation levels, and functional relevance at the biological and evolutionary level (Table S4).
From a genomic enrichment standpoint, the 24 FDR-significant loci were overrepresented in enhancers (χ2=5.1; p=0.025) but not in gene promoters (χ2=2.1; p=0.15; Figure S2A) away from CpG islands (‘Open Sea’) but not in CpG Islands, shores, or shelves (χ2=13.7; p=0.018; Figure S2B).
Overall, the top loci identified in adolescence tended to show higher representation in regions of lower CpG density, suggesting these genomic regions may be more responsive to childhood adversity.
Most loci (19/24) had weak, but positive correlations between brain and blood (prefrontal cortex ravg=0.06, range=−0.19-0.65; entorhinal cortex ravg=0.07, range=−0.23-0.60; superior temporal gyrus ravg =0.06, range=−0.18-0.61; cerebellum ravg=0.07, range=−0.13-0.54) (Table S5; Figure S3). Thus, adversity-induced alterations to blood DNAm levels may reflect similar changes in the central nervous system.
We also identified 6 distinct clusters of biological processes overrepresented in FDR-significant loci (n=24 genes). Although none reached statistical significance, these clusters were broadly implicated in muscle tissue development, cell adhesion, MAPK cascades, and cellular regulation of biosynthetic processes (Figure S4). These results suggest that different types of childhood adversity may act through diverse biological process to influence a multitude of downstream biological processes, rather than through a concerted network of biological pathways.
Genes linked to FDR-significant loci showed little evidence of strong evolutionary conservation overall, as measured by the intolerance to loss-of-function estimates (Table S4; Figure S5). However, 3 genes showed high probability of intolerance to genetic variation resulting in their loss-of-function (pLI>0.9; DSP, CUX2, and STK38L), which were all linked to the accumulation of exposure to one adult in the household. These findings highlight a potential role for genes influenced by parental and social environment in human survival and evolution.
DNAm differences at age 15 were not present earlier in childhood
Among the 24 FDR-significant loci observed in DNAm at age 15, three associations were observed with DNAm at birth (Table S6) and two with DNAm at age 7 (p<0.05) (Table S7). However, none of these association survived adjustment for multiple tests (24 tests at FDR < 5%), and effect estimates were considerably smaller at age 7 than at age 15 with consistent direction for less than half (11 of 24) (Figure 3A).
Childhood adversity was linked to distinct trajectories of DNAm across development
Moving beyond single time points of DNAm, we found that 21 of the 24 FDR-significant loci had significant adversity exposure group-by-age interactions (FDR<0.05), suggestive of more complex patterns of change and stability across development. From these loci, we further identified six types of longitudinal DNAm trajectories, which showed distinct differences in DNAm patterns across ages and adversity exposure groups, both across and within specific ages (Figure 3; Table 1; Figure S6; Table S8). For the three loci that did not show exposure group-by-age interactions, we identified slight differences between exposed-SP and unexposed youths at age 7, which fully emerged by age 15 (i.e., stable). Table 2 provides a full description of the patterns that distinguish these different types of DNAm trajectories, as well as examples.
Associations between adversity and childhood DNAm did not persist into adolescence
We previously identified associations between time-varying adversity exposures before age 7 and DNAm at 48 CpG sites measured at age 7, which we assessed for persistence into adolescence at age 15. Of these 48 CpG sites, only one showed an association between childhood adversity and DNAm at age 15 (p<0.05; Table S9). However, this association did not survive adjustment for multiple tests (48 tests at an FDR<0.05), with just over half showing consistent direction of effect (25/48) (Figure 3D).
DISCUSSION
The main finding from this study is that childhood adversity has unique and time-dependent associations with DNAm, which manifest through varying patterns of persistence and latency across development. This work highlights the role of sensitive periods in development and their effects on chrono-epigenetic patterns. To our knowledge, this is the first study to incorporate time-dependent measures of adversity in the study of longitudinal epigenetic patterns.
Our findings point to early childhood (i.e., between the ages of 3 and 5) as a sensitive period for the biological embedding of childhood adversity, as reflected by DNAm differences present in adolescence. These findings are consistent with multiple prior studies in humans (13, 15) and animal models (56, 57), which have shown that exposures earlier in life may have greater influence on epigenetic patterns. This emphasis on sensitive periods, over other alternative life course hypotheses, is also in line with previous analyses of childhood adversity and DNAm in the ALSPAC cohort, which found that most sensitive period effects arose from exposures during early childhood (22). Early childhood (between ages 3-5) is a critical time period when children begin rapidly developing cognitive, social, emotional, linguistic, and regulatory skills (58). As developmental processes during preschool ages provide an important foundation for future executive functioning (59), this period is ripe for interventions that may limit or prevent the long-term effects of childhood adversity. Indeed, prior studies on children exposed to domestic violence have shown that child-parent psychotherapy between age 3-5 can improve PTSD and depressive symptoms, as well as child behavior problems (60, 61).
Recent evidence also suggests that interventions that promote executive function skills may help build resilience among preschool children exposed to adversity and help build toward future success (62).
Of the seven types of adversity examined, exposure to single parent households had the greatest number of associations in adolescence. By contrast, previous research on DNAm from the same children at age 7 identified no associations with one-adult households, suggesting that these effects may be adolescent-specific (22). Single parent family structures are associated with onset of puberty, as well as other mental health outcomes such as self-esteem, depressive symptoms, and externalizing behaviors, especially in girls (64–66) and when exposure occurs during the first few years of life (67). This relationship may reflect an increased likelihood of children in one-adult household to form maladaptive attachment styles (64), or result from the decreased emotional and material support that can be provided by single parents due to societal and financial pressures. In turn, these gaps in expected versus experienced inputs in early life could manifest during adolescence, a period of rapid maturation for higher order functions. Our findings suggest the biological effects of one-adult households may extend to the DNAm level, which could potentially influence developmental and health outcomes in adolescence. Curiously, fewer associations were observed for other adversities, such as maternal psychopathology and experiences of sexual, physical, or emotional abuse. These adversities may have subtler influences on the adolescent epigenome, which require larger sample sizes or meta-analyses to uncover associations. Of note, none of our top loci overlapped between different types of childhood adversity, nor were they present in a previous study of DNAm trajectories and adolescent victimization (29). These results provide additional insight into ongoing debate on the “lumping or splitting” of childhood adversities in clinical research (68), showing that distinct dimensions of adversity may result in unique epigenetic signatures. However, it remains unknown whether these subtypes of adversity might have differential effects on downstream biological processes and vulnerability to disease, highlighting the need for studies that compare and contrast the effects of adversity on both epigenetic mechanisms and health outcomes.
Arguably the most novel set of findings from our study concerned the relationships of adversity and patterns of stability and change in DNAm over time. Most DNAm trajectories showed primarily latent effects of adversity, meaning they did not emerge until age 15 in youths exposed to adversity.
These findings are in line with previous longitudinal studies of genome-wide DNAm, which have shown that early-life stressors, such as prenatal maternal smoking (26) and socio-economic disadvantage during childhood (27, 28), can have both immediate and latent effects on DNAm during childhood and adolescence. However, some trajectories also showed effects that emerged at earlier ages, which may reflect a desynchronization of epigenetic patterns in response to childhood adversity that became more apparent later in development. Importantly, Oh and Petronis recently suggested that these “sleeper” temporal rhythms in DNAm may explain why complex diseases unfold over years of development, rather than immediately after exposures or risk factors (23). Alternatively, these alterations could reflect specific development windows when the impacts of adversity on biological processes may begin to emerge. These findings suggest the effects of childhood adversity on the epigenome may not instantly take effect, but rather remain latent or alter the developmental trajectories of DNAm in subtle ways that evade immediate detection. As such, future research should investigate whether these latent effects of childhood adversity on the epigenome persist into adulthood and whether they are indeed more likely to influence physical and mental health than alterations that arise earlier in development.
Similarly, the DNAm differences we had previously observed at age 7 did not persist into adolescence (22). Studies on early-life stressors (27, 28) and markers of prenatal environments, such as birthweight and gestational age (24), have revealed parallel insights, showing that DNAm differences linked to early-life environments do not generally persist across developmental time. By contrast, some DNAm signatures of prenatal smoking persist across development, suggesting that exposures with more clearly delineated biological responses may induce more lasting consequences on the epigenome (26). Although these findings suggest that early signatures of childhood adversity may fade from the epigenome, these short-term alterations may alter the developmental trajectories of downstream physiological systems or cellular pathways, which may not be reflected in the epigenome despite their potential influences on health and disease (69). An alternative hypothesis is that the effects of adversity may indeed resolve over time, suggestive of adaptive alterations to the epigenome across development, without the need for focused interventions. Although these findings may have important ramifications for the prevention and treatment of adversity-related health conditions, additional research is needed to identify the true consequences of these differences and determine whether short- and/or long-term DNAm changes mediate the link between childhood adversity and health outcomes.
Our study had several limitations. First, DNAm data were generated from slightly different tissue types at each wave, which may have induced additional variability between ages. Although we corrected for cell type composition using well-established bioinformatic methods, differences in DNAm between waves may have been partially driven by tissue-based differences. As such, additional research using consistent tissues for DNAm measurement across development are needed to fully parse the longitudinal effects of adversity on DNAm. Furthermore, we were unable to assess the effects of childhood adversity on the types of DNAm trajectories across development, as the timing of DNAm and adversity overlapped. Future studies with additional waves of DNAm after the measurement of childhood adversity may help untangle the role of adversity in shaping trajectories of DNAm, as would the use of causal inference methods that can handle time-varying exposures and confounders (70).
Finally, our analytic subset was mainly composed of children from European descent and families with socioeconomic privilege, limiting the generalizability of our findings to broader populations, given existing disparities in distributions of childhood adversity and health outcomes (71). Our findings should be replicated in more diverse cohorts to fully assess the impact of childhood adversity on DNAm across development.
CONCLUSIONS
In sum, this study highlights the complex relationship between childhood adversity and longitudinal DNAm trajectories across development, which vary not only based on the timing of adversity, but also the age at which DNAm is measured. Our findings also provide further insight into the sensitive periods that shape the biological embedding of experiences during early-life, while placing further emphasis on the analysis of chrono-epigenetic patterns in the context of human health. In particular, our results suggest that adversity during early childhood may alter the overall trajectory of DNAm across development, which may, in turn, influence health across the life course. As such, future studies should continue to investigate longitudinal measures of DNAm to identify the potential role of latent and occasionally persistent epigenetic alterations in driving short- and long-term health outcomes. Ultimately, this line of research will help guide intervention strategies and identify individual who are at higher risk for physical and mental disorders arising from exposure to childhood adversity.
Data Availability
All data are available by request from the ALSPAC Executive Committee for researchers who meet the criteria for access to confidential data (http://www.bristol.ac.uk/alspac/researchers/access/).
CONFLICTS OF INTEREST
The authors have no conflicts of interest to declare.
ACKNOWLEDGEMENTS
This work was supported by the National Institute of Mental Health of the National Institutes of Health (grant number R01MH113930 awarded to ECD). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Dr. Dunn and Dr. Lussier were also supported by a grant from One Mind. We are extremely grateful to all the families who took part in the ALSPAC study, the midwives for their help in recruiting them, and the whole ALSPAC team, which includes interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists and nurses. The UK Medical Research Council and Wellcome (Grant ref: 217065/Z/19/Z) and the University of Bristol provide core support for ALSPAC. A comprehensive list of grants funding is available on the ALSPAC website (http://www.bristol.ac.uk/alspac/external/documents/grant-acknowledgements.pdf); This research was specifically funded by grants from the BBSRC (BBI025751/1; BB/I025263/1), IEU (MC_UU_12013/1; MC_UU_12013/2; MC_UU_12013/8), National Institute of Child and Human Development (R01HD068437), NIH (5RO1AI121226-02), and CONTAMED EU (212502). This publication is the work of the authors, whom will serve as guarantors for the contents of this paper. Dr. Walton is funded by CLOSER, whose mission is to maximise the use, value and impact of longitudinal studies (www.closer.ac.uk). CLOSER was funded by the Economic and Social Research Council (ESRC) and the Medical Research Council (MRC) between 2012 and 2017. Its initial five-year grant has since been extended to March 2021 by the ESRC (grant reference: ES/K000357/1). The funders took no role in the design, execution, analysis or interpretation of the data or in the writing up of the findings. Dr. Walton is also supported by the European Union’s Horizon 2020 research and innovation programme (grant n° 848158).
Finally, we would also like to thank Dr. Garrett Fitzmaurice for his guidance in the characterization of DNAm trajectories across development.