In this study we examined a previously uncharacterized effect of maternal depression and antidepressive medication use during pregnancy on two IGF2 differentially methylated regions in the offspring. Maternal depression was not found to independently affect the offspring's methylation status at any of the imprint regulatory regions evaluated. Our data suggest a race-specific influence of the use of antidepressive drugs on the methylation outcomes at the DMR upstream of H19, but not significantly at the IGF2 DMR.
Over the years, a rapid increase in antidepressant use in pregnancy has been reported [39, 8], whereas SSRIs are known to transfer across the placenta and the consequences to the offspring's health are still unclear . One study conducted in the United States showed an increase of 5.6% in the use of antidepressants during pregnancy between 1998 and 2005, from 2.5% to 8.1%; with SSRIs being the most commonly used group of drugs (ranging between 71%  and 84% ). Taking into account the increased tendency in antidepressant use, our frequency of reported maternal use is consistent with the literature. Further, the 15% prevalence of depression in our study is slightly higher than the previously reported 9.1% to 14%; this variation depends on the study design and the trimester of pregnancy [41, 42, 7]. Our slightly higher prevalence may be due, in part, to the fact that depression frequencies vary by smoking during pregnancy [43, 44]. Thirty-three percent of smoking mothers were depressed, while 10.7% of the non-smoking mothers reported depression. Besides the harmful effects of smoking, it is still unclear how the fetus reacts to maternal mood changes or intake of antidepressants during pregnancy. As a consequence, research in the field of maternal depression and associated drug use is gaining in importance.
We analyzed the possible association between maternal depression and changes in DNA methylation level at the imprinted control domains of IGF2 (IGF2 DMR and H19 DMR) and did not observe any differences between newborns exposed to maternal depression and individuals from mothers without any reported depression during pregnancy. However, earlier studies conducted in predominantly Caucasian populations suggested that the methylation status of key regulatory regions of certain genes may be sensitive to prenatal maternal mood, stress, or undernutrition [28, 1]. A study in the senior offspring of mothers who were exposed to the Dutch famine of 1944 during periconception, as well as to the related emotional stress, showed a 5% lower mean methylation at the IGF2 DMR compared to the non-exposed same-sex sibling ; and Oberlander et al. reported that prenatal exposure to maternal depression or anxious mood is associated with increased methylation at a CpG-rich region of the NR3C1 gene . Although exposure to SSRIs did not influence the methylation status of NR3C1 CpG sites in this latter study, animal models have shown that SSRIs may significant affect the expression level of this glucocorticoid receptor . In our analysis on the methylation status at the IGF2 and H19 DMRs we did not detect an overall effect of antidepressant use, but when we looked at the interaction term between race and intake of antidepressants at any time during pregnancy (Table 3), and consecutively at the stratification by race (Table 4), our results suggest an association between African American mothers taking antidepressants and hypermethylation at the differentially methylated region of IGF2 upstream of H19 in the offspring. Hypermethylation at H19 DMR has been associated with a remarkable down-regulation of H19 expression, loss of imprinting of IGF2, and several disorders [46–48], including Wilm's tumors in children; a cancer type that is more prevalent in African Americans [49, 50]. Race specific hypermethylation has been reported in a prostate cancer study where black men show a higher percentage of methylation on the CD44 promoter-region, compared to white men [51, 52]; black men are almost twice as likely to exhibit CD44 hypermethylation compared to white men independent of tumor grade or disease stage . A study on the potential epigenetic influences on racial disparity in the progression of endometrial cancer has also shown that cancers from black women demonstrate a significant lower ribosomal DNA methylation than tumors from white women . The racial differences in methylation levels in these studies, as well as in our study, could reflect inherent genetic differences between Caucasians and African Americans, although we do not exclude influences of unmeasured race-related environmental factors. To our knowledge, a race-dependent effect of prenatal exposure to psychotropic medications on gene methylation has not previously been described. Race or ethnicity is still poorly documented in genetic cancer risk studies, especially in the field of epigenetics; and race is often not included in drug-related studies . Considering the fact that pre-pregnancy BMI and education shared some effects on our DNA methylation outcomes, other factors not examined in this study, such as weight gain during pregnancy, specific dietary patterns, or other life-style factors, should be considered in the future. These, or other yet unidentified epigenetic determinants, may possibly explain the racial discrepancy we observe.
The present study has several limitations. The NEST population is hospital based and limited to pregnant women visiting Hospitals affiliated to Duke University. Although, home birth and home consults are very uncommon in the U.S.. Eighteen percent of the mothers refused to participate, and for practical reasons our analysis was focused on the first 436 participants only. We verified the distributions of the characteristics in all groups (with and without refusal/exclusions) and did not see any significant changes in distributions. Our study population is representative for pregnant mothers from Caucasian or African American origin in Durham County, NC. We verified the population from NEST with the US Census in Durham County , and the distribution of maternal age in our study is similar to the American Community Survey Estimates (ACSE) (2005-2009), reporting the numbers of women with birth in the last 12 months. Although the ACSE data suggest equal proportions of Caucasian and African American women giving birth (38.8% and 37.4%, respectively), we have slightly more African American mothers (49.3%) participating the study than Caucasians (44.3%). The Census reports 23.8% of "other races or origin", while only 5.0% of the NEST participants report this. A possible explanation can be the restriction to English speaking women in our study design. When comparing education, the Census data suggest that 56.3% of the mothers have lower education, compared to 65.1% in NEST. However, adjusting for education and race in our models did not alter the results.
In our study, twenty percent of the mothers smoke during pregnancy. This is six percent higher than what is expected in the US . As mentioned in our earlier published study , NEST originally focused on smokers, and was later expanded to all pregnant mothers (see Materials and Methods). As a consequence our high percentage of pregnant smokers is not representative of the general population and smoking is oversampled. In order to verify if possible bias from this study design would affect our results we did a sensitivity analysis and rerun our data without the first part of the cohort (where smoking mothers were focused) and found the same associations. We also verified the distributions of all variables in smoking mothers in both phases of recruitments, and found no significant differences frequencies of depression, antidepressant intake, race distribution, education, age, birth weight or BMI.
Further, while low birth weight has been associated with depression  or intake of antidepressants , we do not detect an association between depression and birth weight. In mothers taking antidepressants we see a slightly higher percentage (+6%) of newborns with a birth weight less than 2.5 kg, although the difference is not significant (p = 0.11). Another limitation in our study is the reliance on self-report to identify individuals with depression, despite our verification with medical records. We do not exclude the possibility that depression may be slightly underreported: people with depression may not always report their mood changes or be diagnosed with depression. Moreover, no data on duration, history, dosage or rating scales were collected to evaluate any dose-response patterns. In addition, the use of antidepressants was based on medical records without verification of dose-compliance from the pregnant women. Further, only 45.5% of the mothers taking antidepressants report depression (Table 1). The reason for this could be attributed to the following: psychotropic drugs are widely prescribed for reasons other than depression , and as earlier mentioned the prevalence of depression may be underestimated. Nevertheless, our data suggest that methylation was not affected by the reported depression, while the effect of psychotropic drugs remained significant. As a consequence, the intake of antidepressants in African American pregnant women may be an independent predictor of hypermethylation at H19 DMR. We do not exclude a similar effect at the IGF2 DMR, given the small sample size of mothers taking antidepressants at this region studied. Further, we did not stratify by the class of antidepressants women used, given the low numbers, but the proportions of mothers taking SSRIs in both, African American and Caucasian, did not differ significantly and was relatively high: 71.4% of the African American mothers and 75.9% of the Caucasian mothers took SSRIs (p = 0.75). A larger study is warranted to understand more about the possible effects of antidepressant use in pregnancy. In conclusion, we report a race-dependent association between maternal use of antidepressants and hypermethylation at at least one of the imprint regulatory regions of IGF2 in the offspring. Although we do not know the exact underlying cause, infants born to African American mothers using antidepressants in pregnancy suggest an adverse effect on the methylation status of the H19 DMR, indicating a higher risk for loss of imprinting of IGF2 and potentially pernicious consequences for their health status in later life.