Maternal prenatal depressive symptoms and toddler behavior: an umbilical cord blood epigenome-wide association study

Abstract

Children of mothers with prenatal depressive symptoms (PND) have a higher risk of behavioral problems; fetal programming through DNA methylation is a possible underlying mechanism. This study investigated DNA methylation in cord blood to identify possible "at birth" signatures that may indicate susceptibility to behavioral problems at 18 months of age. Cord blood was collected from 256 children of mothers who had self-reported on symptoms of depression during pregnancy and the behavior of their child at 18 months of age. Whole genome DNA methylation was assessed using Illumina MethylationEPIC assay. The mother and child pairs were categorized into four groups, based on both self-reported depressive symptoms, PND or Healthy control (HC), and scores from the Child Behavior checklist (high or low for internalizing, externalizing, and total scores). Adjustments were made for batch effects, cell-type, and clinical covariates. Differentially methylated sites were identified using Kruskal-Wallis test, and Benjamini-Hochberg adjusted p values < 0.05 were considered significant. The analysis was also stratified by sex of the child. Among boys, we observed higher and correlated DNA methylation of one CpG-site in the promoter region of TPP1 in the HC group, with high externalizing scores compared to HC with low externalizing scores. Boys in the PND group showed lower DNA methylation in NUDT15 among those with high, compared to low, internalizing scores; the DNA methylation levels of CpGs in this gene were positively correlated with the CBCL scores. Hence, the differentially methylated CpG sites could be of interest for resilience, regardless of maternal mental health during pregnancy. The findings are in a relatively healthy study cohort, thus limiting the possibility of detecting strong effects associated with behavioral difficulties. This is the first investigation of cord blood DNA methylation signs of fetal programming of PND on child behavior at 18 months of age and thus calls for independent replications.

Fig. 1. Groups for comparison of differentially methylated (DM) CpGs in association with Child Behavior Checklist (CBCL) scores and maternal mood assessed by use of the Edinburgh Postnatal Depression Scale.

Results with DM CpGs with significant adjusted p value are marked by bold lines for the whole study population, by dotted lines when considering only female children, and by double lines when considering only male children. PND perinatal depression, HC healthy control, High CBCL scores median and above, Low CBCL scores 5–40 percentile. *Excluded due to below percentile and between 41 percentile to median.

Fig. 2. Box plots displaying group-wise distribution (median and percentiles) of beta values of differentially methylated CpGs in the whole study cohort and stratified by sex of the child.

PND Perinatal depression, HC healthy control, High CBCL scores median and above, Low CBCL scores 5–40 percentile, int internalizing scores, and ext externalizing scores.