Linking Early Life Adversity to Epigenetic Change: A Meta-Analysis of DNA Methylation as a Molecular Memory of Childhood Trauma.

ABSTRACT

Exposure to adversity in early life can have lasting impacts on later-life health. Childhood adversity may not always be consciously remembered nor leave visible physical scars. However, it marks one’s molecular memory “under the skin” through an epigenetic process, DNA methylation, the addition of a methyl group on a cytosine base. Contemporary research emphasizes specific adversities or methylation in isolated genes, but there has yet to be a systematic compilation of these datasets to understand methylation patterns across diverse sources of adversity. Results were compiled from 49 studies on early-life adversity to create a dataframe of 87,316 genomic locations associated with changes in methylation. We found that the number of CpG sites affected by adversity was not associated with adversity type or timing (p>0.05). Results showed that the number of early-life associated methylated sites decreased as adversity-exposed participants aged (p<0.05) and early life adversity (ELA)-associated sites are more likely to occur in gene bodies than expected by chance (p<0.05). A gene ontology enrichment analysis predicted that adversity affects immune-related pathways (p<0.05). Uncovering patterns and making predictions of how methylation affects participants in broader contexts allows us to more effectively understand how different early life experiences affect one’s epigenetic profile, and thus one’s long-term health.

INTRODUCTION.

A child’s experiences during development shape their risk for adverse health outcomes later in life. For example, consistent exposure to physical abuse early in life has been correlated with increased risk for depression [1] and other cardiometabolic diseases [2]. Epigenetic mechanisms, such as DNA methylation, are thought to mediate this relationship, due to their large role in physiological development, sensitivity to environmental exposures, and influence on long-term health outcomes. Understanding how our early experiences alter biological pathways provides insight into the importance of an early life environment and the measurable changes it may induce.

One of the types of epigenetic modifications that can be measured in children in response to ELA is DNA methylation (DNAm). DNAm acts as a molecular memory of an individual’s early life environment and is one of the most stable and representative changes of many epigenetic mechanisms [3]. During an adverse event, DNA methyltransferases (DNMTs) trigger the addition or removal of a methyl group on a cytosine nucleotide base, creating a methylated area that is referred to as a CpG site [4]. Experiencing ELA causes either an increase or a decrease in methylation. Methylation occurs throughout the body, contributing to distinct tissue-specific functions, meaning that methylation itself is expected in healthy individuals; it is the change in methylation level that raises concern as significant deviations from the norm cause gene expression to be disrupted. Childhood trauma is heavily associated with chronic stress long-term and thus the presence of the stress hormone, cortisol. There is evidence that cortisol inherently reduces DNAm and changes how stress genes involved with inflammation and metabolism may behave. This shows cortisol as a possible mediator in explaining how early life adversity regulates gene expression [5].

Gene expression is essential for carrying out biological functions such as immune response, metabolism, and hormone regulation. Genes in the body are tightly regulated in an extremely specific manner, with precise control mechanisms to ensure each gene is expressed in the right amount and at the right time. There is little space for error in over-expressing or under-expressing genes as even minor deviations of gene expression can significantly damage physiological homeostasis [6]. The addition of a methyl group in genomic regions such as promoters or enhancers interrupts and physically blocks the binding of transcription factors and other important proteins that carry out transcription. This interference with transcription causes many genes to become “silenced”. Dysregulated gene expression as a result of DNAm impacts one’s biological functions and overall health.

Research in this field is extensive in looking at specific types of early life adversities and their corresponding effect on DNA methylation within narrow population cohorts. For example, researchers analyzed methylation levels in blood samples of individuals in utero during the Dutch Hunger Winter of World War 2, identifying CpG sites where their degree of methylation was associated with prenatal nutrition deficiency and maternal war exposure [6]. While such studies provide valuable insight into how famine-related adversity becomes biologically embedded, findings from contexts such as the Dutch Hunger Winter are not generalizable to much of the population. This highlights the need for research that integrates data across multiple types of adversities and contexts. To our knowledge, no study in this field has systematically compiled the known genomic locations of early life adversity-associated methylation sites of twenty-nine adversity types into a single database for large-scale statistical analysis.

This study examined the hypothesis that experiencing prenatal adversities (i.e., maternal stress, war exposure, or tobacco use during pregnancy) would produce significantly stronger marks of methylation as opposed to postnatal adversities (i.e., childhood physical abuse). We predicted that participants’ age across studies would be inversely related to the difference in methylation and that the immune or inflammatory system would particularly be impacted by early-life adversity.

MATERIALS AND METHODS.

Study Selection.

A database-wide literature search of PubMed was run on May 28th, 2025 with search keywords “early life stress”, “childhood adversity”, “early adversity”, “DNA methylation”, and “humans”. 298 papers were found, with an additional 27 papers included from these papers’ reference lists for a total of 49 papers (full paper list in S1 and selection criteria in Chart 1).

Preliminary Data Collection.

An Excel sheet was created for each unique paper containing the genomic coordinates of differentially methylated CpG sites associated with the respective ELA. The genomic coordinates included Chromosome, start position, end position, and CpG ID (unique identifier attached to a CPG site generated by the methylation microarrays, eg. cg00769470).

Creating a Dataset of Genomic Coordinates in RStudio.

There were three main components emphasized when creating the full standardized dataset. First, we ensured that the coordinates of all adversity-associated sites were recorded based on their position in the hg19 human reference genome by converting any sites reported using the hg18 or hg37 genome builds using the Broad Institute Liftover tool. [RS2] (Protocol adapted from [8]). This was a necessary step in combining datasets across studies because different genome assemblies have slightly different coordinates for the same genomic position due to refinement of the human genome mapping over time. The input format of hg19 intervals followed the notation: chromosome number: start position: end position (eg. chr8:141310715-141310720). Second, we converted adversity-associated differentially methylated regions (DMRs) into individual differentially methylated sites utilizing BEDtools intersect [9] and a Bash shell script executed in RStudio (R Version 4.4.3). This created unique CpG sites that only spanned across two base pairs, rather than several thousand. Lastly, genomic coordinates were assigned to sites where only CpG IDs were provided. Information was cross-referenced from Illumina BeadChip Manifest files of different BeadChip models (450K,850K,27K, EPICarray) to the CpG IDs to retrieve a list of matched genomic locations. CpG IDs are transferable across all Illumina BeadChip technologies, meaning they can be matched uniformly with their correct genomic information independent of the model studies used. This step allowed us to fill all blank fields in the initial Excel dataset, resulting in a final dataset of 87,316 unique genomic locations of ELA-associated CpG sites.

Table 1. Organization of Postnatal Adversities into 5 Broad Categories
Abuse	Neglect	Household Challenges	Poor Health	Cumulative
Physical, Emotional, Mental	Physical, Emotional, Maternal sensitivity, Social deprivation, Institutional care	Parental separation/ mental illness/death, Financial stress, Poor Neighborhood	Childhood Illness, Low birth weight, Malnutrition	Experiencing more than 2 of previous adversities, overall early-life stress

Statistical Analyses.

Three statistical tests were performed in RStudio to interpret the dataset of ELA-associated CpG sites. For these models, only utilizing sites from the 450K BeadChip allowed us to simplify results to be more generalizable as all studies using this model tested for the same 400,000 CpG sites. First, we completed a Fisher’s Exact Test and created a contingency table testing the association between significantly differentially methylated sites to find if sites fell significantly within promoter, gene body, or CpG island regions of the genome. Utilizing BedTools Intersect, we defined promoter regions as any site within 10,000 base pairs upstream of a gene’s transcription start site. Information on CpG islands and gene body regions was downloaded from Ensemble through the USCS Table Browser. Second, to answer the hypothesis regarding adversity factors, we conducted a linear model analysis with a linear regression test in RStudio to analyze how the number of significantly differentially methylated sites varied across ELA types, timing, and participant age. Based on the categorization of adversities by the National Adverse Childhood Experiences Survey [9], twenty-nine specific adversities were organized into five broader categories (Table 1). This is how the five categories of adversity were objectively defined in my results. Lastly, we performed a Gene Ontology Enrichment Analysis to predict the biological systems most impacted by ELA.  The rGREAT package was downloaded onto R (Genomic Regions Enrichment of Annotations Tool) along with the Bioconductor package [11]. Version used: rGREAT 2.10.0. To minimize redundancy and overlap in the Gene Ontology (GO) terms, we grouped similarly functioning GO terms through R and took the most significantly enriched term to report. This provided a clearer interpretation of which biological functions are most likely impacted by ELA, giving insight into how ELA truly affects long-term health.

RESULTS.

Linear Regression Model.

The timing of when an adversity was experienced, whether in utero or in childhood/teenage development years, does not have a significant effect on the number of differentially methylated sites found by a study (Fig. 1).

The type of postnatal adversity experienced in childhood does not affect the amount of significantly methylated sites found later in an individual. Whether it is being exposed to overall abuse or a household challenge of parental separation, these differences in adversities do not change the effect childhood trauma has on epigenetic changes (Fig. 2).

As more time passes between early life and when methylation is measured, the number of differentially methylated sites decreases (Fig. 3).

Fishers Exact Test.

ELA-associated sites are approximately three times more likely to be located in gene bodies than would be expected by chance (Fig. 4).

Gene Ontology Enrichment Analysis.

Four out of the nine miscellaneous GO terms were found to be impacted are functions supporting the immune system. Notable biological functions are T-cell selection, Monocyte Differentiation, and B-Cell Activation (Fig. 5).

DISCUSSION.

This meta-analysis synthesized data across studies to link early life adversity to epigenetic changes and identify broad-scale methylation patterns that transcend various contexts. We compiled a comprehensive list of 87,316 unique CpG sites associated with over 20 distinct types of adversities, which allowed us to predict epigenetic trends in relation to participant and adversity factors and evaluate which genomic pathways are most impacted by childhood trauma.

The different timings of adversity have similar effects on the number of differentially methylated sites as shown in Figure 1; a fetus experiencing prenatal trauma has just as significant a biological impact as a child experiencing physical abuse. This supports the concept that prenatal development is also a sensitive time period in shaping one’s methylation landscape, as a fetus’s environment is tied to the mother’s physical and emotional environment. A recent study discovered the lack of concrete methylation-sensitive periods in a child’s life [12], reinforcing the idea that postnatal adversities aren’t the only period that significantly impacts methylation. The overarching idea is that the methylation amount is not affected by adversity timing, as no specific time has a disproportionate effect on changing one’s epigenetic profile.

Figure 2 demonstrates that we cannot discredit one single type of adversity as opposed to another. While abuse and neglect are often perceived as harsher adversities, the data indicates that all types of adversities significantly impact methylation. The day-to-day experiences and situations that children grow up in are significant types of ELA that shouldn’t be discounted in their biological impact, challenging the idea that only overt forms of trauma produce biological change. In all, this finding emphasizes the importance of recognizing all types of adversity as meaningful hardships with biological consequences, which is the first step in reducing childhood trauma stigma and recognizing diverse childhood experiences.

We discovered that the number of differentially methylated sites decreases in strength as participants who experienced a postnatal adversity increase in age in Figure 3. We can correlate the number of methylated CpG sites to the strength of methylation as a greater number of sites within a genome can be associated with how present methylation may be. The results show that the amount of methylation decreases as more time passes from an adverse experience, suggesting the possibility of a biological recovery mechanism of DNA demethylation where the marks of early childhood trauma gradually fade. Because DNA methylation is a dynamic and reversible process, it makes sense that with greater time from facing adversity, the body can heal and “demethylate”. These findings also bring up an interesting perspective that there may be value in younger participants in epigenetic studies as their methylation signatures are more prominent to detect as compared to their older counterparts. This is consistent with a study in this field that found methylation marks fading in adulthood, further suggesting that DNAm isn’t a permanent modification that persists across time without change [13]. Further research is needed to explain why methylation may not be a permanent epigenetic change and determine whether this healing reflects possible active biological interventions within the body.

Figure 4 shows that methylated sites are significantly more likely to be in gene bodies, suggesting that adversity-associated methylation extends beyond traditional methylation regions. Island and promoter region methylation is widely researched and understood. Methylation at those locations inhibits transcription by blocking transcription proteins and thus silencing genes. Gene body methylation, however, is not as emphasized in epigenetic research and remains less represented. Emerging studies suggest preliminary evidence that gene body methylation can have an opposite effect to methylation near the transcription start site and may actually enhance over-transcription to promote gene over-expression. One group of researchers emphasized the possible role of gene body methylation in the development and suppression of cancerous tumors [14]. With mixed results about whether methylation at the gene body encourages tumor replication, the need for research on the molecular-level implications of gene body methylation is absolute, especially in the context of childhood adversity.

Thymic T cell selection self-destructs reactive T cells and selects ones that can recognize pathogens, monocyte differentiation enables monocytes migration through the bloodstream, and B cell activation triggers production of antibodies. My findings of methylation impacting the immune system (Figure 5) align with current literature, as a study that found that childhood abuse is associated with abnormal white blood cell counts and chronic inflammation [15]. This shows that childhood adversity is disrupting important developmental processes related to inflammatory/immune pathways, demonstrating that the biological consequences of adversity extend beyond psychological trauma. Being able to link childhood adversity to a tangible impact traced to a specific pathway significantly broadens the impact of studying genomic changes and methylation patterns.

CONCLUSION.

Our major findings are that all types of postnatal adversities can leave significant epigenetic impacts, and that the immune system is a highly compromised pathway in response to early life adversities. Our comprehensive CPG-site list provides a valuable resource for future studies in understanding how childhood experiences shape long-term health.

Future steps include pulling papers from additional databases. The list of papers used in this study was obtained from PubMed; however, pulling from other databases, such as Google Scholar, could broaden the scope of our work, as more papers would improve sample size, diversity, and applicability. It would also be useful to extend our statistical work in performing a Principal Component Analysis (PCA) or Independent Component Analyses on redundant CpG sites across papers. These tests could separate individual sites based on which sites are often methylated together, which would be incredibly useful in isolating groups of sites that respond specifically to certain types of adversity. A Tukey’s test could be performed in addition to the Fisher’s Exact Test we conducted to support the idea that each location is statistically significantly different than another in methylation levels. Future research referencing this dataset can help identify targets for early intervention, models of disease risk based on epigenetic profiles, and mediate methylation patterns across diverse populations. The promotion of healthier psychological and biological outcomes from childhood trauma for future generations starts with understanding the patterns and relationships among the epigenome in relation to DNA methylation.

ACKNOWLEDGMENTS.

I am extremely grateful to my mentor, Rachel Petersen, for supporting and leading me in this project and my lab PI, Amanda Lea. I also thank the SSMV for giving me this internship opportunity, and more specifically to my project advisor, Rebekah Stanton.

SUPPORTING INFORMATION.

Supporting information available online includes:

Table S1. List of 49 Papers Used in Data Collection

REFERENCES

1. H.Peng, et al. Childhood Trauma, DNA Methylation of Stress-Related Genes, and Depression: Findings From Two Monozygotic Twin Studies. Psychosoma Med. 80, 599-608 (2018).
2. C.Ling, L.Groop, Epigenetics: A Molecular Link Between Environmental Factors and Type 2 Diabetes. Diabetes. 58, 2718-2725 (2009).
3. M.Essex, et al. Epigenetic vestiges of early developmental adversity: childhood stress exposure and DNA methylation in adolescence. Child Development. 84, 58-75 (2013).
4. S.Moore, et al. Epigenetic correlates of neonatal contact in humans. Development and Psychopathology. 29, 1517-1538 (2017).
5. M. Feingold, The Genetics of Stress: How Cortisol Impacts Your DNA and Methylation. 3X4 Genetics. (2025).
6. S.Romens, J.McDonald, J.Svaren, S.Pollak, Associations between early life stress and gene methylation in children. Child Development. 86, 303-309 (2015).
7. E.Tobi, et al. DNA methylation signatures link prenatal famine exposure to growth and metabolism. Nature Communications. 5, 5592 (2014).
8. G.Genovese, et al. BCFtools/liftover: an accurate and comprehensive tool to convert genetic variants. Bioinformatics. 40, btae038 (2024).
9. A.Quinlan, I.Hall, BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics. 26, 841-842 (2010).
10. “Adverse Childhood Experiences International Questionnaire (ACE-IQ)” (World Health Organization, 2020).
11. A.Gu, D.Hübschmann, rGREAT:an R/bioconductor package for functional enrichment on genomic regions. Bioinformatics. 39, btac745 (2023).
12. S.Marzi, et al. Analysis of DNA Methylation in Young People: Limited Evidence for an Association Between Victimization Stress and Epigenetic Variation in Blood. AJP. 175, 517-529 (2018).
13. A.Lussier, et al. Association between the timing of childhood adversity and epigenetic patterns across childhood and adolescence: findings from the Avon Longitudinal Study of Parents and Children (ALSPAC) prospective cohort. The Lancet Child & Adolescent Health. 7, 532-543 (2023).
14. Q.Wang, et al. Gene body methylation in cancer: molecular mechanisms and clinical applications. Clinical Epigenetics. 14, 154 (2022).
15. A.Danese, C.Pariante, A.Caspi, A.Taylor, R.Poulton, Childhood maltreatment predicts adult inflammation in a life-course study. Proc Natl Acad Sci USA. 104, 1319-1324 (2007).

---

---

Posted by buchanle on Thursday, May 14, 2026 in May 2026.

Tags: Early life adversity, Immune System, meta-analysis, methylation