There is a well-established link between early-life adversity and poorer health outcomes. Individuals exposed to childhood adverse events are at higher risk of negative physical, mental, and behavioural-related health outcomes, in comparison to those without adversity (1–3). In addition, evidence suggests certain types of early-life adverse events may have different impacts. For example, while childhood abuse was found to be strongly associated with depression, the experience of family violence in childhood was found to be an important risk factor for the development of anxiety disorder (4). Furthermore, the associations may vary at different ages. Financial difficulties in childhood for example, may be associated with poorer mental health in adulthood, but not in older age (5). Childhood adversities can lead to acute and/or chronic lasting stress, which has been shown capable of influencing the programming of early-life stress-response system (6, 7) and brain development (8), with the potential for biological embedding and thus longer-term effects (9).

One promising mechanism underlying the long-term effects of early-life stress is epigenetic mechanisms. It refers to changes in DNA function that affect gene expression and protein synthesis, but without involving mutations in DNA sequence (10, 11). Three types of epigenetic mechanisms, including DNA methylation, histone modification, and microRNA-mediated processes, are of primary interest in research. DNA methylation is one of the most extensively studied epigenetic modifications, in which a methyl group is covalently bound to cytosine-phosphate-guanine (CpG) and may lead to the reduction of gene expression (12). Moreover, DNA methylation has been shown to be highly sensitive to environmental stimuli, especially in the critical early-life period, as this special time is essential for the development of neural plasticity and potential epigenetic marks may contribute to the change of vital physiological and neuroendocrine systems (13, 14). Studies have shown the adverse exposures early in life can result in changes to DNA methylation patterns which then remain stable over time (15, 16).

The brain-derived neurotrophic factor (BDNF), is a protein that is essential in neuronal development and function, as well as synaptic plasticity (17). BDNF levels have been shown to be highly dynamic in response to stress (18), and interacts directly with the hypothalamic-pituitary-adrenal (HPA) axis in regulating the stress response (19). Numerous studies have identified an association between psychiatric disorders, in particular depression, and altered BDNF promoter methylation (20–22). There is also some evidence to indicate differential BDNF promoter methylation following previous stress-related events. For example, maternal stress during pregnancy has been associated with differential BDNF methylation in the offspring (23, 24), as has neighbourhood disadvantage in early-life (25, 26). Furthermore, rodent models suggest early-life adversity alters BDNF methylation and persists into adulthood (27–29), and this was supported by a study of 85 human adults where differential blood BDNF methylation was observed in response to low levels of maternal care in childhood (6). To date, however, no studies have looked at the long-term effects of childhood adverse events on differential BDNF methylation in older adult populations.

Therefore, this study aimed to investigate whether major adverse events in childhood are associated with altered BDNF methylation in later life, investigating DNA methylation in both blood and buccal tissue.

Overall, we found no strong evidence for an association between three major childhood adverse experiences and blood and buccal BDNF methylation in later life. Childhood abuse/maltreatment was not associated with differential BDNF methylation, in either blood or buccal tissue, and financial difficulties/poverty in childhood was only associated with decreased blood methylation at one CpG. The strongest finding was with regards to childhood experience of war/natural disasters, where significant associations were found with BDNF methylation, although this was not consistent across gene regions, or tissues, nor was the direction of association consistent (at 3 sites lower methylation was observed with war/natural disaster, but increased methylation at a 4th site). Our significant findings could merely be chance associations, resulting from multiple testing and thus increasing the risk of type 1 error (false positives). Indeed, none of the associations would remain after Bonferroni adjustment based on the number of statistical tests.

One possibility for these largely null findings, is that childhood adverse events have no significant long-term effects on blood or buccal BDNF promoter methylation in later life, particularly for those individuals aged 65 years and over. Although findings from some prior rodent and human studies lead to the hypothesis that early-life stress may have a long-lasting effect in altering BDNF methylation, which is detectable in peripheral tissue (6, 27, 41–43), none of these studies have measured DNA methylation in older age. Methylation alterations are complex and not fully understood (41). Knowledge about the extent to which peripheral tissues can represent brain functions, and how methylation patterns differ between specific tissues over the life course (29), is still evolving. Further, a growing body of literature suggested the epigenetic patterns are potentially reversible through interventions. For example, mouse models have shown that epigenome-modifying drugs not only have the ability to prevent maltreatment-induced increase of BDNF promoter methylation (44), but are also capable of reversing DNA methylation changes resulting from early-life adversity (27, 45–47). Similar findings were reported from human studies, with several studies of depression observing an attenuated BDNF methylation change after receiving antidepressant treatment (35, 48–50). Other interventions, such as psychotherapeutic approaches (42) and aerobic exercises (51) also showed potential therapeutic ability in the context of childhood adverse experiences. Taken together, these findings suggest that it might be quite unlikely that DNA methylation changes resulting from early adversity can persist over several decades of life.

BDNF is one of the most well-studied genes in response to early life stress. It is highly expressed in the central nervous system and plays a vital role in regulating brain development and facilitating neuron functioning, including proliferation, differentiation, and survival (52–54). Human BDNF gene contains eleven functional exons, nine of which are regulated by specific promoter regions (55, 56). Evidence suggests increased methylation at any promoter regions typically leads to decreases in BDNF gene expression, as the addition of methyl group inhibits transcription factors to bind properly to regulatory elements (57). Also, stress can suppress BDNF expression (58), and decreased BDNF expression has been shown to be associated with epigenetic modification of the BDNF gene (59), as well as depressive disorder and other mental disorders in late life (22, 60, 61), which eventually leads to our primary interest whether BDNF methylation play a role in mediating early life adversities and mental health issues in adults over 65 years of age. However, another possibility for largely insignificant findings is that other genes are involved. For example, NR3C1, a stress related gene that codes for glucocorticoid receptors that bind with cortisol or other glucocorticoids and are directly involved in HPA responsivity to stress (62), and some prior studies have reported significant associations between a history of childhood adverse events and NR3C1 promoter methylation in adulthood (63–66). There is also evidence that other candidate genes, such as interleukin-6 (67) and serotonin transporter gene promoter (68), may be responsive to early-life adverse experiences.

A reason for our largely null findings could also be because individuals with the most severe early-life exposures were not included in our study. The ESPRIT cohort recruited participants who were living in the community with at least 65 years of age, thus those hospitalized or who had died before that age would not have been included. This so-called survival bias may have influenced the findings. Of note, however, prior studies in ESPRIT have demonstrated associations between early-life adversity and later health outcomes, including an increased risk of depression (31) and poorer cognitive performance (33). It has also been shown that some specific events in early life were associated with an increased risk of mortality in older ESPRIT participants, and in particular, war/disaster increased the risk of death for women (32). Post-hoc sex-stratified analysis in the current study also found a significant association between war/disaster and decreased BDNF methylation in women at two CpG sites of promoter IV in buccal tissue (CpG 8: β = –0.17, 95% CI: –0.31 to –0.04, p = 0.01 and CpG 9.10: β = –0.12, 95% CI: –0.22 to –0.02, p = 0.02), but not men (CpG 8: β = –0.04, 95% CI: –0.20 to 0.12, p = 0.61 and CpG 9.10: β = –0.04, 95% CI: –0.16 to 0.08, p = 0.54). The concordance of these associations may be a chance finding, but could also indicate true sex-specific associations. Sex-specific epigenetic changes are a relatively common phenomena in the field, and could be explained by biological differences between the sexes, as well as boys and girls differing in their experience of events. Together these findings would support epigenetic processes as potentially contributing to the long-term effects of early-life adverse events on later-life health, but further work is needed.

There are several limitations to this study, one of the most important to consider is the measure of adversity. Our study only examined the occurrence of these major childhood adverse events, but did not further quantify the severity or exact nature of the adversity. These adverse events would all have differed in intensity and duration, so as the presence of family support, thus the associated stress and impact on the individual would have varied. The likely substantial heterogeneity in exposures may partially account for the overall null associations found here. We didn’t have information either on the level of family support available in childhood, which could help buffer the negative effects of an adverse environment (69–71). Additionally, three major childhood adverse events were specifically selected in this study based on our prior hypothesis, however, other types of childhood adversities, such as parental separation or the loss of a family member, should also be carefully considered in future research. Another limitation is the retrospective reporting of early-life adverse experiences. This is likely to lead to some misclassification, considering the long time since its occurrence. However, given that we are measuring an objective outcome in this study, this should not have strongly bias the results in a particular way. Lastly, with the exception of the war/natural disaster exposure in childhood, the two other adverse childhood events were less frequent, especially childhood abuse/maltreatment (reported by < 15%). This meant insufficient power to detect smaller methylation differences, in particular after adjusting for potentially important confounders.

This study also has multiple strengths. Firstly, to authors’ knowledge, this is the first study, focusing on older adults with an age of 65 and over, to examine the relationship between childhood adverse events and BDNF promoter methylation. The data collection was embedded in the large longitudinal ESPRIT study, with participants unaware of the objectives of this analysis and childhood adverse experiences obtained in a rigorous manner, which provided high-quality data for this study. Notably, we concentrated on the occurrence of adversities themselves, rather than individuals’ perceptions of their childhood adverse experiences, so that adverse events data was measured objectively. DNA methylation was measured in two tissue types, and for one of these tissues, across two promoter regions of the gene. In addition, a large variety of covariates was controlled in adjustment analysis, including not only the common sociodemographic and lifestyle factors, but also measures of physical and psychological health that were assessed through medical records, trained professionals, standardized interviews, or validated questionnaires.

There is insufficient evidence of an association between major childhood adverse events, namely abuse/maltreatment, war/natural disaster, and financial difficulties/poverty, and differential BDNF promoter methylation in older adults. We suggest future studies to investigate methylation across the genome, rather than focusing on a specific gene, and large, longitudinal studies, with sufficient power to overcome the added risk of false-positive from testing such a number of genes, are preferable. Other epigenetic mechanisms, including histone acetylation and microRNAs, could also be investigated.

The original contributions presented in this study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

The studies involving human participants were reviewed and approved by Ethical Committee of the University Hospital of Kremlin-Bicêtre, France. The participants provided their written informed consent to participate in this study.

M-LA and KR led the ESPRIT study and the collection of primary data. JR designed the study, undertook the measurement of DNA methylation, and performed the analyses. AZ interpreted the data and wrote the first draft of the manuscript, with supervision from JR. All authors revised the manuscript and approved the final version.