Adverse Childhood Experiences, Epigenetic Measures, and Obesity in Youth (original) (raw)

. Author manuscript; available in PMC: 2019 May 13.

Abstract

Objective

To determine if measures of adverse childhood experiences and DNA methylation relate to indices of obesity in youth.

Study design

Participants were derived from a cohort of 321 8 to 15-year-old children recruited for an investigation examining risk and resilience and psychiatric outcomes in maltreated children. Assessments of obesity were collected as an add-on for a subset of 234 participants (56% female; 52% maltreated). Illumina arrays were used to examine whole genome epigenetic predictors of obesity in saliva DNA. For analytic purposes, the cohort analyzed in the first batch comprised the discovery sample (n = 160), and the cohort analyzed in the second batch the replication sample (n = 74).

Results

After controlling for race, sex, age, cell heterogeneity, 3 principal components, and whole genome testing, 10 methylation sites were found to interact with adverse childhood experiences to predict cross-sectional measures of body mass index, and an additional 6 sites were found to exert a main effect in predicting body mass index (P < 5.0 × 10−7, all comparisons). Eight of the methylation sites were in genes previously associated with obesity risk (eg, PCK2, CxCl10, BCAT1, HID1, PRDM16, MADD, PXDN, GALE), with several of the findings from the discovery data set replicated in the second cohort.

Conclusions

This study lays the groundwork for future longitudinal studies to elucidate these mechanisms further and identify novel interventions to alleviate the health burdens associated with early adversity.

Over the past 2 decades, there has been a growing appreciation of the role of adverse childhood experiences (ACEs) on the development of a range of negative health outcomes.1–4 Child maltreatment and other ACE are nonspecific risk factors for multiple psychiatric disorders,5–7 and several health risk behaviors, including smoking, overeating, and excessive alcohol and drug use.4,8,9 Above and beyond the effect of these health risk behaviors, ACE predict a multitude of medical health problems later in life, including ischemic heart disease,9–11 stroke,9 respiratory problems,12,13 diabetes,9,11 and even cancer.9,14

There is preliminary evidence that childhood adversity may confer risk for this broad range of outcomes through epigenetic mechanisms.15–17 Epigenetics refers to functionally relevant chemical modifications to the genome that do not involve a change in DNA nucleotide sequence.18 These chemical modifications can alter gene activity and influence regulation of genes in response to changes in the environment, with epigenetic modifications induced by ACE reported frequently to persist into adulthood.19

DNA methylation is one of the most studied epigenetic mechanisms.20 In a prior investigation, maltreated and control children were found to have significantly different methylation values at 2868 methylation sites.15 The gene sets showing differential methylation between the maltreated and comparison children contained genes involved in biological processes relevant to psychiatric and substance use disorders (eg, neurogenesis, axonal guidance), heart disease (eg, cardiac development), stroke (development of blood vessel morphogenesis), respiratory disease (eg, interleukin regulation), diabetes (eg, leptin signaling), and cancer (eg, Wnt signaling, Notch signaling), as noted above, all medical illnesses that have been linked to adversity in youth.9 That study, however, did not include any health outcomes data, and it is not known if these epigenetic differences were associated with health outcome measures.

The goal of the current investigation was to determine if measures of ACE and DNA methylation predict cross-sectional indices of obesity in youth. Obesity was selected as the health outcome to examine as it is the medical problem associated with childhood adversity most apt to manifest in youth, and there is strong meta-analytic support for the association between child maltreatment and obesity.21

Methods

Participants were derived from a cohort of 321 children recruited between 2013 and 2016 as part of an investigation examining risk and resilience in maltreated children. Data were analyzed in 2017. The 321 children were from 235 families with several siblings and half siblings (range: 0–4 per family) included in the cohort. In the year prior to study enrollment, 37% of the children had an out-of-home placement because of reports of abuse or neglect. Another 15% of the children had prior allegations of maltreatment, and 48% of the sample was never referred to protective services. The children were from diverse socioeconomic status (SES) backgrounds (Hollingshead SES mean 3.34, SD 1.3, range 1–5), and as discussed further below, had experienced a broad range of childhood adversities.

The primary study was designed to assess predictors of mental health outcomes, with the assessment of obesity collected on a subset of the children as an add-on to the primary investigation. Child trauma, DNA methylation, and obesity measures were available on a total of 234 participants. The DNA data were analyzed in 2 batches; once at the end of the second year of recruitment, and once at the completion of the study. For analytic purposes, the cohort analyzed in the first batch was the discovery sample (n = 160), and the cohort analyzed in the second batch was the replication sample (n = 74). Table I depicts the demographic and clinical characteristics of the discovery and replication cohorts.

Table I.

Demographic and clinical characteristics of the sample

Assessments from parents were collected during home visits and the majority of the child data for this study were collected at a day camp devised specifically for our research purposes, replicating a methodology used in previous investigations.22–24 The institutional review boards at Yale University, University of Vermont, and Johns Hopkins Schools of Medicine approved this study. Prior to recruitment, an independent child advocate reviewed each case referred through protective services to determine that research participation was in the child’s best interest. The child’s parent or legal guardian provided informed consent and each child provided assent for study participation. Birth parent assent for child participation for children in state custody was obtained when clinically appropriate (eg, ongoing parent-child contact).

Multiple informants and data sources (eg, parents, children, protective services records) were used to obtain a best estimate of each child’s adverse experiences. Permissions were secured to access state child protective services records from all study participants, and records were obtained for all children with prior protective services referrals (n = 122; 87 children recruited through protective services because of recent out-of-home placements and 35 children recruited through the community who were living with their biological families but had a history of prior protective services involvement). The data from these various sources were integrated and rated using the Yale-Vermont Adversity in Childhood Scale (Y-VACS) scoring procedures.25 The Y-VACS assesses a range of intrafamilial (eg, physical abuse, witnessing domestic violence) and extrafamilial (eg, community violence, bullying, natural disasters) adversities, and generates scores that take into account severity and frequency of exposure. The Y-VACS generates a total adversity score, intrafamilial adversity score, extrafamilial adversity score, and individual item scores, with high inter-rater reliability reported in generating these scores.25 As the children with and without prior protective services involvement did not differ on the extrafamial adversity scale of the Y-VACS (Wald χ2 = .237, df = 1, ns), children’s scores on the intrafamilial adversity scale of the Y-VACS were used in subsequent analyses examining obesity risk.

The nurse at the research summer day camp program collected the measurements of height and weight which were used to calculate body mass index (BMI). The same scale was used for all children participating in the study.

Saliva specimens for DNA extraction were refrigerated within 2 hours of collection and DNA extracted using Puregene (Gentra, Minneapolis, Minnesota) kits. To prepare the specimens for the methylation study, 500 ng of genomic DNA were treated with bisulfite reagents included in the EZ-96 DNA methylation kit (Zymo Research, Orange, California) according to the manufacturer’s protocol. Bisulfite-converted DNA samples were then used in the array-based DNA methylation assays.

The Illumina 450K Methylation BeadChip was used to analyze the DNA data of the discovery cohort (Illumina Inc, San Diego, California). Array-based epigenome-wide methylation analyses were completed at the Yale Center for Genome Analysis at Yale University using standard procedures. Quality control was conducted based on published methods.26 Methylation sites with detection P value of >.001 were removed to ensure that only high-confidence probes were included. Probes were also excluded if they had a single nucleotide polymorphism (SNP) within the methylation site map to multiple places in the genome or to sex chromosomes. Methylation data were normalized using the functional normalization method in the “minfi” R package. This function uses internal control probes present on the array to control for between-array technical variation and outperforms other approaches.27 A total of 456 513, (94% of sites) were left for subsequent analysis. The genomic inflation factor for the discovery cohort was 1.004, indicating that the epigenome wide analyses results are negligibly inflated and the results are unlikely due to spurious findings (Figure 1; available at www.jpeds.com). The script and methods used to generate the genomic inflation factor parallel the methods utilized by Gelernter et al.28

Figure 1.

Quantile-quantile (Q-Q) plots. The figure above depicts the Q-Q plots for the P values of the association. The genomic inflation factor for the discovery cohort was 1.004, indicating that the epigenome wide analyses results are negligibly inflated and the results are unlikely due to spurious findings.

The Illumina Infinium Methylation EPIC (850K) BeadChip (Illumina, Inc) was used to assess DNA methylation in the replication cohort. Only the methylation sites that significantly predicted BMI in the children in the discovery cohort after controlling for whole genome testing (P < 5.0 × 10 −7) were examined in the replication cohort. The 850K BeadChip was likewise processed at the Yale Center for Genome Analysis.

Because methylation values at methylation sites can be cell-type specific,29 we conducted a cell composition estimation analysis using a modified version of the method by Housemann et al,30,31 and estimated the relative proportion of each cell type (eg, CD34, CD14, and buccal cells) in the heterogenous peripheral saliva samples.

DNA methylation can vary by race or ethnicity.32,33 To adjust for possible population stratification within the predominantly European American sample, we used a methylation-based principal component (PC) approach,34 based on sets of methylation sites within 50 kb of SNPs using the 1000 Genomes Project variants with minor allele frequency of >0.1 following the Barfield et al method.34 This method outperforms other methods that adjust for population stratification in methylation.34

Statistical Analyses

To control for familial correlations due to the inclusion of siblings in the cohort, generalized estimating equations (GEE) were used to examine the impact of intrafamilial adversity on BMI. Age, sex, and race were used as covariates in the GEE analysis (n = 234), and a separate GEE analysis was conducted with SES also included in the model for the subset of subjects with SES data (n = 175). In analyzing the methylation data for the discovery sample (https://cran.r-project.org/web/packages/lme4/index.html), linear mixed-effects models were used to examine the main effect of methylation M-values and intrafamilial childhood adversity scores and the interaction of these 2 variables using the “lme4” function in R software environment (R Foundation for Statistical Computing, Vienna, Austria). Familial correlations were modeled by assigning a random effect to each family. Potential confounding factors including age, sex, race, cell type (eg, CD14, CD34, buccal), and the first 3 PCs were also included in the model. To correct for multiple comparison testing, significant threshold for analyses was set to5.0 × 10−7, consistent with prior recommendations.35 Only the methylation sites that significantly predicted BMI in the children in the discovery cohort after controlling for whole genome testing (P < 5.0 × 10 −7) were examined in the replication cohort, with GEE used in these analyses with the same set of covariates listed above included in the models. The discovery cohort and replication samples were not analyzed together, as at present, methods do not exist to combine 450 K and 850 K data controlling for batch and BeadChip effects.

Results

The children in the cohort experienced a mean of 2.3 (SD = 2.3) intrafamilial adversities (range: 0–8), with 29% of the cohort reported to have experienced 4 or more intrafamilial adversities. Table II (available at www.jpeds.com) provides detailed information about the proportion of children to have experienced each ACE. Among the children in the maltreated cohort, 66% experienced physical abuse, 34% sexual abuse, 57% neglect, and 54% witnessed domestic violence. The Y-VACS scores of the discovery and replication cohorts are included in Table I. The higher intrafamilial adversity scores in the discovery cohort is likely attributable to the majority (82%) of the maltreated children with a recent out-of-home placement being recruited during the first 2 years of the study and included in the discovery cohort.

Table II.

Proportion of children who experienced each adversity

After controlling for age, sex, and race, the Y-VACS intrafamilial childhood adversity measure was a significant predictor of BMI (n = 234; P < .0001). Increased age (P < .0001) and African American ancestry (P < .04) also predicted greater BMI, but after accounting for the other measures in the model, sex (P < .11) was not a significant predictor of BMI. The Y-VACS intrafamilial childhood adversity score was still a significant predictor of BMI (P < .004) when SES was also included in the model, although these analyses were conducted on a smaller subset of subjects as SES data were missing for 25% of the sample, with missing SES data greatest for the most traumatized children— youth in out-of-home care who did not currently have any contact with their biological parents. Children who had high Y-VACS scores (n = 49), defined as 1 SD above the mean, were 16 times more likely to be obese than children who had low Y-VACS scores (n = 49), defined as 1 SD below the mean (percent obese: high Y-VACS: 16.3%; low Y-VACS: 1%, χ2 = 17.6, df = 3, P < .001).

After controlling for age, sex, race, cell heterogeneity, the first 3 PCs, and whole-genome testing, 6 methylation sites were found to exert a main effect in predicting BMI, and an additional 10 methylation sites were found to interact with the adverse childhood experiences measure to predict BMI (Table III). The direction of the interaction effects varied by site, with some methylation sites being stronger predictors of BMI in youth with low intrafamilial adversity scores (eg, HID1, GALE), and other methylation sites being stronger predictors of BMI in youth with high intrafamilial adversity scores (eg, PRDM16, PXDN). Eight of the 16 methylation sites identified in these analyses are in genes previously associated with obesity risk, and 3 were in intergenic regions.

Table III.

Discovery cohort (n = 160): methylation sites that independently and in interaction with intrafamilial childhood adversity predicted BMI

Metacore software was used to identify biological processes associated with the top epigenome wide significant hits. This package accounts for the varying number of methylation sites per gene by assigning a prior probability for each gene based on gene length, followed by a modified hypergeometric test for over-representation of a gene set.36 Methylation sites that had significance values of P < 1.0 × 10−5 were used in this analysis. To correct for multiple comparison testing, false discovery rate (FDR) was set at .05. Enrichment was seen in multiple muscle development and muscle cell proliferation processes, processes involved in responses to several different nutrients and glucocorticoids, as well as multiple processes relevant to translation and transcription. The top 50 significant gene ontology (GO) processes are included online only in Table IV (available at www.jpeds.com).

Table IV.

Enrichment analysis report

Of the 16 sites associated with obesity risk in the discovery cohort, 2 failed to pass quality control tests for analyses in the replication cohort; KCNS3 (cg0155585) and BCAT1 (cg15990629). As depicted in Table V, after controlling for age, sex, race, cell heterogeneity, and the first 3 PCs, the adversity measure and 4 of the methylation sites predicted BMI at nominal levels of significance, and the significance value associated with methylation in PCK2 withstood Bonferroni correction for multiple comparisons (P < .0031). The interaction term of trauma with the methylation value at the methylation site in GALE also approached statistical significance in the replication sample.

Table V.

Replication sample: methylation and trauma measures together predict BMI (n = 74)

There are multiple methylation sites contained on the 450K BeadChip for each of the genes identified as significant predictors of obesity. As depicted online in Table VI (available at www.jpeds.com), nominal levels of significance in predicting BMI were attained at 8 additional sites in these genes.

Table VI.

Additional methylation sites in replicated genes that predicted BMI

Given the replicated finding with PCK2 across the 2 cohorts, follow-up mediation analyses were conducted using regression analyses and maintaining the demographic, cell heterogeneity, and PC covariates in the model. As depicted in Figure 2, after controlling for the relevant covariates, support for a mediation model was attained, with the effects of intrafamilial childhood adversities on indices of obesity found to be mediated by epigenetic changes in the PCK2, a gene implicated in obesity risk in a prior epigenome wide association study.37

Figure 2.

Indirect effects of intrafamilial childhood adversities on BMI through methylation of PCK2. After controlling for the relevant demographic factors (eg, age, sex, race), cell heterogeneity measures (eg, CD34, CD14, and buccal cells), and the first 3 PCs, support for a mediation model was attained, with the effects of intrafamilial childhood adversities on indices of obesity found to be mediated by epigenetic changes in the PCK2, a gene implicated in obesity risk in a prior epigenome wide study.

** = P < .01; *** = P < .0001.

Discussion

In the current investigation, ACEs predicted cross-sectional assessments of BMI in the children. Children who had high intrafamilial adversity scores, defined as 1 SD above the mean, were 16 times more likely to be obese than children who had low intrafamilial adversity scores, defined as 1 SD below the mean. The findings of this investigation are consistent with the results of a recent meta-analysis of 41 studies examining the association of child maltreatment and obesity.21 The authors of that meta-analysis concluded that child maltreatment is a potentially modifiable risk factor for obesity, and future research should clarify the mechanisms through which child maltreatment affects obesity risk.

The present study suggests child maltreatment may confer risk for obesity through epigenetic mechanisms. Among the subjects in the discovery data set, after controlling for age, sex, race, cell heterogeneity, 3 PCs, and whole-genome testing, 10 methylation sites were found to interact with the ACE measure to predict cross-sectional assessments of BMI, and an additional 6 sites were also found to exert a main effect in predicting BMI.

Eight of the methylation sites identified in the discovery data set are in genes previously associated with obesity risk or functionally related to relevant biological indices. Specifically, PCK2 encodes an enzyme in mitochondria involved in glucose metabolism. It is mainly expressed in liver, pancreas, intestine fibroblasts and is involved in the insulin signaling pathway, with genetic variation in PCK2 previously found to predict individual differences in response to dietary interventions.38 CXCL10 encodes a chemokine, with values of CXCL10 found to correlate significantly with measures of visceral fat area in obese children.39 The protein product of BCAT1 catalyzes the first reaction in the catabolism of the essential branched chain amino acids leucine, isoleucine, and valine and has been identified as a candidate risk gene for obesity.40 HIDI is associated with body fat mass regulation, preadipocyte number and adipocyte size in rats.41 The PRDM16 protein product is involved in the differentiation of brown adipose tissue, and the PRDM16 transcriptional pathway has been identified as promising for the development of novel therapeutic interventions for the treatment of obesity and obesity-related diseases.42 OSBPL9 encodes a member of the oxysterol-binding protein family, a group of intracellular lipid receptors.43 MADD variants are implicated in type 2 diabetes and have been associated with fasting proinsulin levels in genome wide association studies.44 PXDN deletions have been associated with early onset obesity,45,46 and GALE encodes UDP-galactose-4-epimerase which catalyzes 2 distinct but analogous reactions with important metabolic consequences.47 Although the exact methylation sites identified in the current investigation have not been reported in prior epigenome-wide studies of obesity risk,37,48–53 methylation of other methylation sites in 4 of the genes identified in this study (eg, PCK2, MADD, PRDM16, BCAT1) were reported in the prior investigations.37,48–50,53

Three of the significant methylation sites identified in the discovery data set were in intergenic regions. Many methylation islands in intergenic regions are enriched for factor binding sites and are involved in the 3-dimensional organization of the genome and gene regulation.54,55 Transcription factor binding sites and chromatin insulators within intergenic regions are believed to mediate intra- and interchromosomal interactions, affecting gene expression at both proximal and distal locations,55 and there are numerous instances where intergenic genetic variation is associated with disease risk.56

Results of the gene ontology analysis conducted in the discovery cohort with the top epigenome wide significant hits (P < 1.0 × 10-5) found enrichment in multiple biologically relevant processes. Muscle development and muscle cell proliferation processes were enriched among the methylation sites with the highest association with BMI. Processes involved in responses to several different nutrients (eg, vitamin A, vitamin D, iron) and glucocorticoids, as well as multiple processes relevant to translation and transcription were also enriched in the top epigenome significant findings.

In terms of the replication analyses, even though the replication sample was small (n = 74) and had a lower prevalence of obesity and history of childhood adversities, several of the findings from the discovery data set were also significant predictors of BMI in the second cohort. The most robust of the findings involved PCK2, with this result withstanding Bonferroni correction for multiple comparisons. In addition, follow-up analyses with PCK2 provided support for a mediation model in which the effects of intrafamilial childhood adversities on indices of obesity were found to be mediated by epigenetic changes in PCK2.

Although prior literature indicates that DNA methylation levels in saliva are similar to those in peripheral blood, skin fibroblasts, and buccal swab DNA, it may not reflect the epigenome of adipose tissue, muscle, pancreas, gastrointestinal system, or the pituitary—tissues most relevant in obesity.57,58 Additional limitations of the current investigation include small size of the replication cohort, use of the 450K BeadChip, which does not include methylation sites at the most replicated previously identified obesity genetic risk SNPs, limiting the test of these genes in the current investigation, use of the 850K platform in the replication cohort, the lack of genetic variant and gene expression data, the absence of food intake and exercise data,59 failure to assess prenatal factors known to affect methylation and obesity risk,60–62 and the cross-sectional nature of the study which limits inferences that can be made regarding the causal consequences of the methylation findings.

The data reported in this article suggest adverse childhood experiences may confer risk for health problems through epigenetic mechanisms. Although epigenetic modifications are frequently long lasting, they can sometimes be reversed. This study lays the groundwork for future longitudinal clinical and translational studies to further elucidate these mechanisms and identify novel prevention and treatment interventions to alleviate the health burden associated with early adversity.

Acknowledgments

Supported by the Zanvyl and Isabelle Krieger Fund (to J.K.), the National Institutes of Health (R01MH098073 [to J.K. and J.H.], R01 MD011746–01 [to J.K.]); the National Center for Posttraumatic Stress Disorder—Veterans Affairs Connecticut (to J.G. and J.K.); the VA Cooperative Study #575B, Genomics of Posttraumatic Stress Disorder in Veterans (to J.G. and J.K.); and the Biological Sciences Training Program through 5T32 MH14276 (to J.M-O.).

We thank the children and families who participated in this research, and the administration of the Vermont Department of Children and Families for their collaboration on this effort.

Glossary

ACEs

Adverse childhood experiences

BMI

Body mass index

GEEs

Generalized estimating equations

PC

Principal component

SES

Socioeconomic status

SNP

Single nucleotide polymorphism

Y-VACS

Yale-Vermont Adversity in Childhood Scale

Footnotes

The authors declare no conflicts of interest.

References

• 1.CDC. Adverse childhood experiences reported by adults—five states, 2009. MMWR Morb Mortal Wkly Rep 2010;59:1609–13. Reprinted in JAMA, 2011; 305(7):666–669. [PubMed] [Google Scholar]
• 2.Kuehn BM. AAP: toxic stress threatens kids’ long-term health. JAMA 2014;312:585–6. doi: 10.1001/jama.2014.8737. [DOI] [PubMed] [Google Scholar]
• 3.Shonkoff JP, Boyce WT, McEwen BS. Neuroscience, molecular biology, and the childhood roots of health disparities: building a new framework for health promotion and disease prevention. JAMA 2009;301:2252–9. [DOI] [PubMed] [Google Scholar]
• 4.Anda RF, Croft JB, Felitti VJ, Nordenberg D, Giles WH, Williamson DF, et al. Adverse childhood experiences and smoking during adolescence and adulthood. JAMA 1999;282:1652–8. [DOI] [PubMed] [Google Scholar]
• 5.Kendler KS, Bulik CM, Silberg J, Hettema JM, Myers J, Prescott CA. Childhood sexual abuse and adult psychiatric and substance use disorders in women: an epidemiological and cotwin control analysis. Arch Gen Psychiatry 2000;57:953–9. [DOI] [PubMed] [Google Scholar]
• 6.Molnar BE, Buka SL, Kessler RC. Child sexual abuse and subsequent psychopathology: results from the National Comorbidity Survey. Am J Public Health 2001;91:753–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 7.Kaufman J Child abuse. Lewis’ child and adolescent psychiatry: a comprehensive textbook 4th ed. Baltimore (MD): Lippincott Williams & Wilkins; 2007. [Google Scholar]
• 8.Burke NJ, Hellman JL, Brandon SG, Weems CF, Carrion VG. The impact of adverse childhood experiences on an urban pediatric population. Child Abuse Negl 2011;35:408–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 9.Felitti VJ, Anda RF, Nordenberg D, Williamson DF, Spitz AM, Edwards V, et al. Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults. The Adverse Childhood Experiences (ACE) Study. Am J Prev Med 1998;14:245–58. [DOI] [PubMed] [Google Scholar]
• 10.Dong M, Giles WH, Felitti VJ, Dube SR, Williams JE, Chapman DP, et al. Insights into causal pathways for ischemic heart disease: adverse childhood experiences study. Circulation 2004;110:1761–6. [DOI] [PubMed] [Google Scholar]
• 11.Romans S, Belaise C, Martin J, Morris E, Raffi A. Childhood abuse and later medical disorders in women. An epidemiological study. Psychother Psychosom 2002;71:141–50. [DOI] [PubMed] [Google Scholar]
• 12.Anda RF, Brown DW, Dube SR, Bremner JD, Felitti VJ, Giles WH. Adverse childhood experiences and chronic obstructive pulmonary disease in adults. Am J Prev Med 2008;34:396–403. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 13.Dube SR, Fairweather D, Pearson WS, Felitti VJ, Anda RF, Croft JB. Cumulative childhood stress and autoimmune diseases in adults. Psychosom Med 2009;71:243–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 14.Brown DW, Anda RF, Felitti VJ, Edwards VJ, Malarcher AM, Croft JB, et al. Adverse childhood experiences are associated with the risk of lung cancer: a prospective cohort study. BMC Public Health 2010;10:20. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 15.Yang B-Z, Zhang H, Ge W, Weder N, Douglas-Palumberi H, Perepletchikova F, et al. Child abuse and epigenetic mechanisms of disease risk. Am J Prev Med 2013;44:101–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 16.Zhang TY, Meaney MJ. Epigenetics and the environmental regulation of the genome and its function. Annu Rev Psychol 2010;61:439–66, C1–3. [DOI] [PubMed] [Google Scholar]
• 17.Bredy TW, Sun YE, Kobor MS. How the epigenome contributes to the development of psychiatric disorders. Dev Psychobiol 2010;52:331–42. doi: 10.1002/dev.20424. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 18.Zhang H, Kranzler HR, Poling J, Gelernter J. Variation in the nicotinic acetylcholine receptor gene cluster CHRNA5-CHRNA3-CHRNB4 and its interaction with recent tobacco use influence cognitive flexibility. Neuropsychopharmacology 2010;35:2211–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 19.Nestler EJ. Epigenetic mechanisms in psychiatry. Biol Psychiatry 2009;65:189–90. [DOI] [PubMed] [Google Scholar]
• 20.Nestler E, Hyman SE, Malenka RC. Molecular neuropharmacology: a foundation for clinical neuroscience 2nd ed. New York (NY): McGraw-Hill Professional; 2008. [Google Scholar]
• 21.Danese A, Tan M. Childhood maltreatment and obesity: systematic review and meta-analysis. Mol Psychiatry 2014;19:544–54. [DOI] [PubMed] [Google Scholar]
• 22.Kaufman J Depressive disorders in maltreated children. J Am Acad Child Adolesc Psychiatry 1991;30:257–65. [DOI] [PubMed] [Google Scholar]
• 23.Kaufman J, Cicchetti D. The effects of maltreatment on school-aged children’s socio-emotional development: assessments in a day camp setting. Dev Psychol 1989;25:516–24. [Google Scholar]
• 24.Kaufman J, Yang BZ, Douglas-Palumberi H, Houshyar S, Lipschitz D, Krystal J, et al. Social supports and serotonin transporter gene moderate depression in maltreated children. Proc Natl Acad Sci USA 2004;101:17316–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 25.Holbrook H, O’Loughlin K, Althoff R, Douglas-Palumberi H, Kaufman J, Hudziak J. The Yale-Vermont adversity in childhood scale: a quantitative approach to adversity assessment. American Academy of Child and Adolescent Psychiatry’s 61st Annual Meeting San Diego, CA: 2015. [Google Scholar]
• 26.Maksimovic J, Phipson B, Oshlack A. A cross-package Bioconductor workflow for analysing methylation array data. F1000Res 2016;5:1281. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 27.Fortin JP, Labbe A, Lemire M, Zanke BW, Hudson TJ, Fertig EJ, et al. Functional normalization of 450k methylation array data improves replication in large cancer studies. Genome Biol 2014;15:503. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 28.Jensen KP, Smith AH, Herman AI, Farrer LA, Kranzler HR, Sofuoglu M, et al. A protocadherin gene cluster regulatory variant is associated with nicotine withdrawal and the urge to smoke. Mol Psychiatry 2017;22:242–9. doi: 10.1038/mp.2016.43. Epub Apr 12. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 29.Jaffe AE, Irizarry RA. Accounting for cellular heterogeneity is critical in epigenome-wide association studies. Genome Biol 2014;15:R31. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 30.Houseman EA, Accomando WP, Koestler DC, Christensen BC, Marsit CJ, Nelson HH, et al. DNA methylation arrays as surrogate measures of cell mixture distribution. BMC Bioinformatics 2012;13:86. doi: 10.1186/471-2105-13-86. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 31.Smith AK, Kilaru V, Klengel T, Mercer KB, Bradley B, Conneely KN, et al. DNA extracted from saliva for methylation studies of psychiatric traits: evidence tissue specificity and relatedness to brain. Am J Med Genet B Neuropsychiatr Genet 2015;168B:36–44. doi: 10.1002/ajmg.b.32278. Epub 2014 Oct 29. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 32.Adkins RM, Krushkal J, Tylavsky FA, Thomas F. Racial differences in genespecific DNA methylation levels are present at birth. Birth Defects Res A Clin Mol Teratol 2011;91:728–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 33.Heyn H, Moran S, Hernando-Herraez I, Sayols S, Gomez A, Sandoval J, et al. DNA methylation contributes to natural human variation. Genome Res 2013;23:1363–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 34.Barfield RT, Almli LM, Kilaru V, Smith AK, Mercer KB, Duncan R, et al. Accounting for population stratification in DNA methylation studies. Genet Epidemiol 2014;38:231–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 35.Rakyan VK, Down TA, Balding DJ, Beck S. Epigenome-wide association studies for common human diseases. Nat Rev Genet 2011;12:529–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 36.Phipson B, Maksimovic J, Oshlack A. missMethyl: an R package for analyzing data from Illumina’s HumanMethylation450 platform. Bioinformatics 2016;32:286–8. [DOI] [PubMed] [Google Scholar]
• 37.Huang RC, Garratt ES, Pan H, Wu Y, Davis EA, Barton SJ, et al. Genome-wide methylation analysis identifies differentially methylated CpG loci associated with severe obesity in childhood. Epigenetics 2015;10:995–1005. doi: 10.80/15592294.2015.1080411. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 38.Larsen LH, Angquist L, Vimaleswaran KS, Hager J, Viguerie N, Loos RJ, et al. Analyses of single nucleotide polymorphisms in selected nutrient-sensitive genes in weight-regain prevention: the DIOGENES study. Am J Clin Nutr 2012;95:1254–60. doi: 10.3945/ajcn.111.016543. Epub 2012 Apr 4. [DOI] [PubMed] [Google Scholar]
• 39.Ishii M, Araki S, Goto M, Yamamoto Y, Kusuhara K. CCL2 level is elevated with metabolic syndrome and CXCL10 level is correlated with visceral fat area in obese children. Endocr J 2016;63:795–804. Epub 2016 Jul 2. [DOI] [PubMed] [Google Scholar]
• 40.Chen J, Meng Y, Zhou J, Zhuo M, Ling F, Zhang Y, et al. Identifying candidate genes for Type 2 Diabetes Mellitus and obesity through gene expression profiling in multiple tissues or cells. J Diabetes Res 2013;2013:970435. doi: 10.1155/2013/970435. Epub 2013 Dec 26. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 41.Weingarten A, Turchetti L, Krohn K, Kloting I, Kern M, Kovacs P, et al. Novel genes on rat chromosome 10 are linked to body fat mass, preadipocyte number and adipocyte size. Int J Obes (Lond) 2016;40:1832–40. doi: 10.038/ijo.2016.127. Epub Jul 27. [DOI] [PubMed] [Google Scholar]
• 42.Kajimura S Promoting brown and beige adipocyte biogenesis through the PRDM16 pathway. Int J Obes Suppl 2015;5:S11–4. doi: 10.1038/ijosup.2015.4. Epub Aug 4. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 43.GeneCards. OSBPL9 Gene: Oxysterol Binding Protein Like 9 GeneCards Human Gene DataBase: Weizmann Institute of Science; 2017. [Google Scholar]
• 44.Strawbridge RJ, Dupuis J, Prokopenko I, Barker A, Ahlqvist E, Rybin D, et al. Genome-wide association identifies nine common variants associated with fasting proinsulin levels and provides new insights into the pathophysiology of type 2 diabetes. Diabetes 2011;60:2624–34. doi: 10.337/db11-0415. Epub 2011 Aug 26. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 45.Bonaglia MC, Giorda R, Zanini S. A new patient with a terminal de novo 2p25.3 deletion of 1.9 Mb associated with early-onset of obesity, intellectual disabilities and hyperkinetic disorder. Mol Cytogenet 2014;7:53. doi: 10.1186/755-8166-7-53. eCollection 2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 46.Doco-Fenzy M, Leroy C, Schneider A, Petit F, Delrue MA, Andrieux J, et al. Early-onset obesity and paternal 2pter deletion encompassing the ACP1, TMEM18, and MYT1L genes. Eur J Hum Genet 2014;22:471–9. doi: 10.1038/ejhg.2013.189. Epub Oct 16. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 47.GeneCards. GALE Gene: UDP-Galactose-4-Epimerase Gene Cards: Human Gene Database: Weizmann Institute of Science; 2017. [Google Scholar]
• 48.Demerath EW, Guan W, Grove ML, Aslibekyan S, Mendelson M, Zhou YH, et al. Epigenome-wide association study (EWAS) of BMI, BMI change and waist circumference in African American adults identifies multiple replicated loci. Hum Mol Genet 2015;24:4464–79. doi: 10.1093/hmg/ddv161. Epub 2015 May 1. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 49.Meeks KAC, Henneman P, Venema A, Burr T, Galbete C, Danquah I, et al. An epigenome-wide association study in whole blood of measures of adiposity among Ghanaians: the RODAM study. Clin Epigenetics 2017;9:103. doi: 10.1186/s13148-017-0403-x. eCollection 2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 50.Samblas M, Milagro FI, Mansego ML, Marti A, Martinez JA. PTPRS and PER3 methylation levels are associated with childhood obesity: results from a genome-wide methylation analysis. Pediatr Obes 2017;14:12224. [DOI] [PubMed] [Google Scholar]
• 51.Sayols-Baixeras S, Subirana I, Fernandez-Sanles A, Senti M, Lluis-Ganella C, Marrugat J, et al. DNA methylation and obesity traits: an epigenome-wide association study. The REGICOR study. Epigenetics 2017;3:1363951. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 52.Wahl S, Drong A, Lehne B, Loh M, Scott WR, Kunze S, et al. Epigenome-wide association study of body mass index, and the adverse outcomes of adiposity. Nature 2017;541:81–6. doi: 10.1038/nature20784. Epub 2016 Dec 21. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 53.Samblas M, Milagro FI, Mansego ML, Marti A, Martinez JA. PTPRS and PER3 methylation levels are associated with childhood obesity: results from a genome-wide methylation analysis. Pediatr Obes 2018;13:149–58. doi: 10.1111/ijpo.12224. Epub 2017 Jun 14. [DOI] [PubMed] [Google Scholar]
• 54.Hodges E, Molaro A, Dos Santos CO, Thekkat P, Song Q, Uren PJ, et al. Directional DNA methylation changes and complex intermediate states accompany lineage specificity in the adult hematopoietic compartment. Mol Cell 2011;44:17–28. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 55.Yang J, Corces VG. Chromatin insulators: a role in nuclear organization and gene expression. Adv Cancer Res 2011;110:43–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 56.Cotnoir-White D, Laperriere D, Mader S. Evolution of the repertoire of nuclear receptor binding sites in genomes. Mol Cell Endocrinol 2011;334:76–82. [DOI] [PubMed] [Google Scholar]
• 57.Oelsner KT, Guo Y, To SB, Non AL, Barkin SL. Maternal BMI as a predictor of methylation of obesity-related genes in saliva samples from preschool-age Hispanic children at-risk for obesity. BMC Genomics 2017;18:57. doi: 10.1186/s12864-016-3473-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 58.Souren NY, Tierling S, Fryns JP, Derom C, Walter J, Zeegers MP. DNA methylation variability at growth-related imprints does not contribute to overweight in monozygotic twins discordant for BMI. Obesity (Silver Spring) 2011;19:1519–22. doi: 10.038/oby.2010.353. Epub 1 Jan 27. [DOI] [PubMed] [Google Scholar]
• 59.Mathers JC, Strathdee G, Relton CL. Induction of epigenetic alterations by dietary and other environmental factors. Adv Genet 2010;71:3–39. doi: 10.1016/B978-0-12-380864-6.00001-8. [DOI] [PubMed] [Google Scholar]
• 60.Cao-Lei L, Dancause KN, Elgbeili G, Massart R, Szyf M, Liu A, et al. DNA methylation mediates the impact of exposure to prenatal maternal stress on BMI and central adiposity in children at age 13(½) years: Project Ice Storm. Epigenetics 2015;10:749–61. doi: 10.1080/15592294.2015.1063771. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 61.Heijmans BT, Tobi EW, Stein AD, Putter H, Blauw GJ, Susser ES, et al. Persistent epigenetic differences associated with prenatal exposure to famine in humans. Proc Natl Acad Sci USA 2008;105:17046–9. doi: 10.1073/pnas.0806560105. Epub 2008 Oct 27. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 62.Veenendaal MV, Costello PM, Lillycrop KA, de Rooij SR, van der Post JA, Bossuyt PM, et al. Prenatal famine exposure, health in later life and promoter methylation of four candidate genes. J Dev Orig Health Dis 2012;3:450–7. doi: 10.1017/S2040174412000396. [DOI] [PubMed] [Google Scholar]