Depression during pregnancy is relatively common, with prevalence estimates ranging from 7% to 18% in the United States and similar prevalence in many low- and middle-income countries (1,2). Beyond affecting the well-being of the mother herself, maternal depression or depressive symptoms during pregnancy have been associated with adverse birth outcomes (e.g., preterm delivery, low birth weight) (3) and with health-related outcomes in early childhood (3–5). Mechanisms by which maternal depression may lead to adverse outcomes in offspring have not yet been clearly identified, but one hypothesized pathway is through alterations in biological processes related to immune regulation. Healthy immune regulation occurs in part by a counterbalancing of pro- and anti-inflammatory effects of cytokines. Previous work has suggested that these processes are dysregulated in individuals with depressive symptoms and disorders (6–8). This dysregulation is hypothesized to result in part from overactivation of the hypothalamic-pituitary-adrenal (HPA) axis, leading to persistent secretion of glucocorticoids and other hormones that contribute to regulating inflammatory responses (6). As one result, immune cells become less sensitive to glucocorticoids' anti-inflammatory effects; this glucocorticoid insensitivity leads to chronic inflammation, which can cause physiological damage (7) and seems to play a role in the development of diseases such as atherosclerosis, diabetes, and obesity (8).
More specifically, compared with nondepressed individuals, depressed individuals may exhibit higher circulating levels of interleukin-6 (IL-6) and tumor necrosis factor α (TNF-α), two proinflammatory cytokines (8). Although these cytokines play an important and necessary role in responding to acute infection, in a dysregulated system, they may fail to return to baseline after activation, contributing to chronic low-grade inflammation (6). Moreover, counterbalancing anti-inflammatory cytokines, such as interleukin 10 (IL-10), may be lower in depressed compared with nondepressed individuals (8). These relationships seem to hold in pregnancy; although only one small case-control study has looked at maternal depression per se and maternal cytokines (9), studies in the related literature on maternal stress find that in pregnant women, higher maternal stress is associated with higher levels of hormones released by HPA axis activation and higher circulating levels of IL-6 and TNF-α and lower IL-10 (10,11).
Whether maternal depression may exert an intergenerational effect on her offspring's immune response is still unclear. Studies in depressed pregnant mothers and their newborns have demonstrated changes in the infant's hormonal and neurotransmitter patterns that indicate potential dysregulation of the infant's HPA axis and related hormones and neurotransmitters (12). Few studies, however, have assessed the associations of prenatal maternal depression with offspring cytokine responses. The South London Child Development Study, a prospective, observational study of the long-term effects of offspring exposure to prenatal maternal depression followed 103 children of mothers recruited during pregnancy from birth until the age of 16 years. This study found that levels of C-reactive protein, a proinflammatory protein whose production is stimulated by IL-6 and TNF-α, were elevated in young adults whose mothers were depressed during pregnancy (13). Three studies have examined whether maternal prenatal depression may lead to dysregulation of offspring cytokines earlier in life, and they have reported conflicting results: one retrospective study of 136 women recruited at birth (9) found no correlation between major depressive disorder and cord blood immune biomarkers; a small case-control study of preterm and term births (14) found positive correlations between depressive symptoms in pregnancy and cord blood levels of IL-18 (but not IL-10, IL-13, IFN-γ) only in preterm births; however, a prospective study of 98 pregnant women with allergic disease (15) found a global positive association between depressive symptom in pregnancy and the cytokines IL-6, IL-10, IL-13, and IFN-γ in cord blood.
Given the limitations of the existing literature, the current study sought to evaluate, using data from a longitudinal pregnancy cohort, whether maternal depression was associated with alterations of offspring immune responses evident at birth. We hypothesized that maternal depression would lead to higher levels of lymphocyte proliferation (LP) proinflammatory cytokines (IL-6, TNF-α) and IL-13, and lower levels of anti-inflammatory cytokines (IL-10) produced by cord blood mononuclear cell (CBMC) collected at delivery. Because depression is a recurring disease (16,17), and because, as in stress, the effects on health are considered to be the result of cumulative burden rather than acute exposure (18), we considered as our primary predictor mothers who either reported a history of depression or who reported symptoms consistent with depression during pregnancy. In addition, because increased cytokine production in response to immune stimulant exposure can be interpreted as a measure of the immune system's competence, we assessed cytokine production upon stimulation of CBMCs with three immune stimulants (19). We considered as covariates factors that might alter the risk of maternal depression and influence immune responses, including mother's demographics, prepregnancy body mass index (BMI, kilogram per square meter), blood pressure, and parental history of atopy or asthma (20). We also included sex of the child (21), season of birth, and maternal smoking, because they have been shown to influence cord blood immune parameters.
Project Viva, a longitudinal pregnancy cohort, has been described in detail elsewhere (22). Briefly, pregnant mothers were enrolled during 1999–2002 at mothers' initial prenatal obstetric visits (median = 9.9 weeks of gestation) at eight locations of Atrius Harvard Vanguard Medical Association, a multispecialty group practice in urban and suburban Eastern Massachusetts. All participants delivered at one of two area hospitals, but only one hospital was willing to facilitate cord blood collection. Women were eligible for enrollment if they were less than 22 weeks pregnant, had a singleton pregnancy, were able to answer questions in English, and had no plans to move out of the study area before delivery. Mothers provided written informed consent upon enrollment, including for access to medical records and cord blood collection at delivery. Delivering clinicians collected venous umbilical cord blood from nonemergent deliveries at the participating hospital (n = 1029 of the 2128 live singleton births), among which we measured cytokines or LP for 463 (see flow chart of sample sizes, Supplemental Figure, Supplemental Digital Content 1, http://links.lww.com/PSYMED/A546).
Study staff visited women after routine appointments during the first and second trimesters and 1 to 3 days after delivery on the postpartum maternity floor. The study was approved by the Institutional Review Board of Harvard Pilgrim Health Care.
We characterized women as having experienced depression if they met criteria for depression before or during pregnancy, as detailed in the following sections. The reference group was women who experienced depression neither before nor during pregnancy.
Depression During Pregnancy
Mothers reported depressive symptoms during the past 7 days at the midpregnancy visit using the Edinburgh Postnatal Depression Scale (EPDS), a 10-item questionnaire that has been validated for use in pregnancy (23). Response options are on a Likert scale ranging from 1 (most of the time) to 4 (not at all), scored so that a higher score indicates more depressive symptoms. Although various EPDS thresholds for probable prenatal depression have been identified (24), a cutoff score of greater than 12 is commonly used (25,26), with previous validation of this cutoff against diagnostic clinical interviews indicating a specificity of 78% and a sensitivity of 86% (23). We categorized women who had EPDS score of higher than 12 as well as women who were prescribed antidepressants during pregnancy regardless of their EPDS score as having depression during pregnancy.
Antidepressants during pregnancy (yes/no) were determined via information drawn from each woman's electronic medical record. We compiled a list of all medications prescribed during pregnancy, and a study physician identified antidepressants.
Prepregnancy depression history (yes/no) was determined using three questions included on the midpregnancy questionnaire. Participants were asked an initial screening question: “Before this pregnancy, was there ever a period of time when you were feeling depressed or down or when you lost interest in pleasurable activities most of the day, nearly every day, for at least 2 weeks?” Women who responded affirmatively were asked to complete two follow-up questions (1): “Before this pregnancy, did you ever see a health care professional who said that you were depressed?” (2) and “Before this pregnancy, did a health care professional ever prescribe a medication for you for depression?” We considered women who responded affirmatively to the screener question and to one or both of the follow-up questions to have a history of depression.
Women were characterized as ever depressed (yes/no) if they scored as having depression during pregnancy or a history of depression or both.
Immune Outcome Assessment
The cytokine panel obtained in Project Viva includes major pro- and anti-inflammatory cytokines (TNF-α and IL-6; IL-10, respectively). In addition, molecules indicating differential T-cell stimulation were measured: interferon-γ (IFN-γ), which promotes differentiation of the T-helper cell 1 (Th1) lineage of the adaptive immune system; and interleukin 13 (IL-13), which is produced mainly by cells in the T-helper cell 2 (Th2) lineage of the adaptive immune system. LP serves as a global measure of immune activation.
Cord blood samples were collected by needle/syringe from the umbilical vein after delivery and were processed within 24 hours without freezing, and CBMC were isolated from umbilical cord blood by density gradient centrifugation with Ficoll-Hypaque Plus (Pharmacia, Uppsala, Sweden). Cells were washed in RPMI 1640 and diluted in 10% human serum (Biowhittaker, Walkersville, MD) to a concentration of 5 × 106 cells/ml. Culture conditions, described hereinafter, were ascertained as appropriate from dose-response and time-kinetic experiments, as previously reported (27).
Four replicates of 0.5 × 106 CBMC/well were incubated in 96-well round-bottom tissue culture plates (Corning, New York, NY) at 37°C in 5% CO2, either unstimulated (incubated in media alone) or stimulated with one of three different immune system stimulants (see hereinafter), and cultured for 72 hours. CBMC were then pulsed with 1 μCi [3H]thymidine for an additional 8 hours. Cultures were maintained at 37°C in a humidified 5% CO2 incubation chamber. Cells were harvested with a Tomcat Mach II harvester (Wallac, Turku, Finland) onto filter plates, which were read in a β-counter. Proliferation was quantified by stimulation index, calculated as the ratio of mean counts per minute of stimulated to unstimulated replicates (28).
Aliquots of 5.0 × 106 CBMC were incubated in triplicate as described previously. Cell supernatants were harvested after 72 hours and analyzed for production of the cytokines IFN-γ, IL-13, IL-6, IL-10, and TNF-α using enzyme-linked immunosorbent assay (Endogen, Rockford, IL) according to the manufacturer's protocol. Sensitivities of the assays were less than 2 pg/ml for IFN-γ, less than 7 pg/ml for IL-13, less than 3 pg/ml for IL-10, and less than 2 pg/ml for TNF-α. All IL-6 samples were detectable. The percentage of samples below the detection limit varied from 0% to 72.3% (see Supplemental Table, Supplemental Digital Content 2, http://links.lww.com/PSYMED/A547).
Cell aliquots were stimulated with one of the following: 5 μg/ml of phytohemagglutinin (PHA), 30 μg/ml of cockroach extract (Bla g 2), or 30 μg/ml of house dust mite extract (Der f 1). PHA is a mitogen (29,30) with response not dependent on antigen-presenting cells. Bla g 2 and Der f 1 are two common environmental allergens: previous work has demonstrated that CBMC respond to these extracts by proliferating and secreting cytokines (31,32). Such activation is commonly interpreted as a measure of the competence of the adaptive immune system to react to antigen (19). Current understanding of the dynamics of the neonate's immune system is too limited to identify a specific level of response as “more” or “less” healthy; instead, we seek to determine whether responses differ in offspring of ever depressed mothers compared with mothers who have never experienced depression.
At enrollment, women reported their age, race/ethnicity, marital status, household income, smoking history, history of hypertension, and maternal and paternal history of asthma and atopy. Mothers self-reported prepregnancy weight and height, from which we calculated BMI (kilogram per square meter). We obtained child sex and date of birth from the hospital delivery record. Information on cigarette use during pregnancy was collected by questionnaire (early pregnancy and midpregnancy) and interview (delivery), and categorized as never/former/smoked during pregnancy. To ascertain the presence of hypertensive disorders of pregnancy, we evaluated prenatal records for blood pressure and urine protein results. We created a four-level variable with the categories normotensive, gestational hypertension, pre-eclampsia, and chronic hypertension in line with guidelines from the 2000 National High Blood Pressure Working Group on High Blood Pressure in Pregnancy (33).
To address nonnormality of the cytokine and lymphocyte distributions, we log transformed them. We used χ2, independent t, or analysis of variance tests to compare covariate distribution across women who were ever versus never depressed.
We used separate multivariate regression models to assess relationships of maternal depression with each outcome, namely, LP, IL-10, IFN-γ, TNF-α, IL-6, and IL-13, each upon stimulation with PHA, Bla g 2, or Der f 1 extract and upon incubation without stimulant.
Our primary exposure was lifetime depression (ever depressed, yes/no). We also ran separate models for each dichotomous depression measure (depression during pregnancy, depression history, antidepressant prescription).
For most immune outcomes, we used linear regression models. However, for cytokines with 9% or more of samples falling below the detection level of the assays, we used tobit regression (using SAS PROC QLIM; see Supplemental Table, Supplemental Digital Content 2, http://links.lww.com/PSYMED/A547). Tobit regression adjusts for censored data by combining a probit model to account for the fraction of samples censored (i.e., below limit of detection) with a truncated regression model for those outcome values that score above the censored cutoff.
We tested two separate models for each combination of depression and immune outcome. Model 1 adjusted for likely confounders (mother's age, race/ethnicity, education, and household income, each categorized as reported in Table 1, as well as season of birth and child sex). Model 2 added a set of covariates that might either confound or lie on the pathway between lifetime depression and cord blood immune outcomes (20), including mother's prepregnancy BMI (continuous), history of hypertension and pregnancy disorders of hypertension (categorical), and smoking (categorical) (20,34). Maternal and paternal history of atopy was not associated with any of the immune outcomes, so we did not include these variables in our models.
Because immune outcomes in our models were log transformed, to increase interpretability of the regression coefficients, we exponentiated the coefficients to obtain differences in the ratio of the expected geometric means of the outcome variable and express these in the figures as the percentage difference in the geometric mean of the outcome variable in depressed compared with nondepressed women.
We additionally conducted two sensitivity analyses. Because previous studies have found associations between perinatal complications and cord blood cytokine concentrations (35,36), we reran all models (n = 357) excluding women whose pregnancies resulted in a cesarean delivery, a preterm birth (defined as <34 weeks of gestation), or small for gestational age (defined as <10th percentile of birth weight for gestational age and sex). Antidepressants may serve as an indicator of severe depression but could also have biological effects because of medication per se. To ensure that we were assessing the effect of depression rather than of depression medications, we reran all models (n = 445) excluding women who were taking antidepressants during pregnancy.
One hundred two women were missing data on the primary exposure variable. To maintain sample size, reduce bias, and improve model precision, we imputed depression scores and covariate values for all individuals missing these measures using a chained equation multiple imputation model (PROC MI in SAS). Imputed values were derived using the full Project Viva cohort (n = 2128), by including exposure and outcome variables, all covariates, and other potential predictors of the outcome in the imputation model (37). We generated 50 imputed data sets and combined them using PROC MIANALYZE (38). Final models included participants with imputed depression and covariate data. Participants missing measurements of outcome for a given exposure-outcome analysis were excluded from that particular analysis.
We performed all analyses in SAS Version 9.4 (SAS Institute Inc, Cary, NC).
Women in versus excluded from the study sample showed few differences (see Supplemental Table, Supplemental Digital Content 3, http://links.lww.com/PSYMED/A548). Compared with those excluded, women in the study sample were more likely white (71% versus 65%) and had fewer cesarean deliveries (19% versus 25%). In the study sample, 80 women (22%) of 361 with data on depression were ever depressed. Of these 80, 34 women had pregnancy depression without a history of depression, 27 had a history of depression without pregnancy depression, and 19 had both pregnancy depression and a history of depression.
Comparing women who were ever versus never depressed highlights several differences (Table 1). Women who were ever depressed were slightly younger at enrollment (mean age = 31.9 versus 32.5 years) and were more likely to have a household income of less than US $40,000/year (18% versus 9%) and smoke during pregnancy (21% versus 8%). Women who were ever depressed also had more babies born in winter and fewer in spring or fall.
Associations Between Depression and Immune Outcomes
Unadjusted Spearman correlation coefficients for log-transformed immune outcomes and EPDS score ranged from −006 to 0.08 (see Supplemental Table, Supplemental Digital Content 4, http://links.lww.com/PSYMED/A549). In our models, we first looked at the anti-inflammatory cytokine, IL-10. In models adjusted for likely confounders, levels of IL-10 from unstimulated cells did not differ by ever versus never maternal depression (Figure 1). However, after stimulation with cockroach or dust mite allergen, IL-10 levels were substantially lower in ever depressed women (percentage difference [95% confidence interval] = Bla g 2–41.4 [−63.4 to −6.1], p = .027; Der f 1–36.0 [−60.6 to 4.0], p = .071), and a similar trend was evident in IL-10 stimulated with PHA (−24.2 [−56.8 to 33.0], p = .333).
Adding mother’s prepregnancy BMI and smoking history to the model did not change the results nor did adding pregnancy variables (See Supplemental Table, Supplemental Digital Content 5, http://links.lww.com/PSYMED/A550).
We next looked at TNF-α, IL-6, IFN-γ, IL-13, and LP. Although there was a general trend for most stimulated markers to be lower in women who were ever versus never depressed, none of these trends reached statistical significance (Figure 2). Adding additional covariates to the base model did not change any of the findings. IL-13 had a high proportion of samples (>50%) below the assay’s level of detection (see Supplemental Table, Supplemental Digital Content 2, http://links.lww.com/PSYMED/A547). There was no evidence that the proportion of samples above the level of detection was associated with maternal depression (data not shown).
We separately considered each depression indicator’s association with stimulated or unstimulated levels of IL-10 (but not with cytokines for which no primary associations were found). Findings were largely similar to those with the combined measure of depression (Figure 3), although confidence intervals were generally wide and crossed the null given relatively small numbers in each group. Excluding women with perinatal complications did not change results, and excluding women who had prescriptions for antidepressants during pregnancy caused a slight attenuation in effect estimates but the basic pattern of associations was unchanged (data not shown).
In a longitudinal cohort of 463 pregnant women, mothers who reported a history of depression or depression during pregnancy had lower CBMC production of the anti-inflammatory cytokine IL-10 in response to stimulation. Neither LP nor levels of the proinflammatory cytokines TNF-α and IL-6 differed in children of mothers who ever (versus never) experienced depression. Although this finding was unexpected, intergenerational studies in both animals (39) and humans (40) have suggested that exposure to depression or stress in utero may lead to a general downregulation of cytokines and other immune processes in the offspring.
As a predominantly anti-inflammatory cytokine, IL-10 plays a central role in control of both innate and cell-mediated immunity. Thus, a decrease in levels of this cytokine may dampen the potential of the newborn to effectively downregulate proinflammatory stimuli (41). If effects of prenatal depression on reduction in IL-10 production persisted, they might adversely affect a child’s response to immune challenges through the life course.
As yet, data are sparse concerning the relation between maternal prenatal depression, offspring cytokines, and offspring health outcomes. In contrast to our findings, a small prospective study found higher levels of CBMC IL-10 (either unstimulated or stimulated with dust mite antigen) from women classed as “mildly depressive” (Beck Depression Inventory ≥10) in pregnancy. Levels of proinflammatory cytokines and LP were higher as well. All of these women had allergic disease (15), so it is unclear how generalizable this finding might be to nonallergic women. Another study reported that children of women with prenatal depression had higher levels of the proinflammatory biomarker C-reactive protein (after accounting for their own depression status) when they were adults at the age of 25 years (13). Here, cytokine outcomes were not evaluated early in life, and although the authors adjusted for mother’s and child’s depression, other unmeasured factors after birth and throughout childhood may have contributed to this inflammatory response. Similar to our findings, a prospective birth cohort study recently found significant inverse associations between prenatal maternal depression and Th2 cytokines (IL-4, IL-5, and IL-13) from peripheral blood mononuclear cells collected from children at the age of 3 years, upon treatment in vitro with a variety of immune stimuli including PHA and dust mite antigen (40).
Broadly speaking, evidence is emerging for links among prenatal maternal stress or depression, alterations in maternal HPA axis and immune system functioning, and offspring HPA axis function or immune response at birth and beyond. Glucocorticoids are known to cross the placenta, and it has been suggested that fetal exposure to higher levels of maternal glucorticoids, which might be due to either prenatal maternal stress (42) or prenatal maternal depression (43), could adversely affect fetal immune maturation (42,44). Although details differ, maternal psychosocial stress and prenatal depression have both been linked to epigenetic changes in placental genes involved in maternal-fetal glucocorticoid transfer (45). Specific to depression, studies have consistently shown that women with versus without untreated depression or depressive symptoms in pregnancy are more likely to have newborns with higher levels of cortisol (as well as norepinephrine, a hormone released when the sympathetic nervous system is activated) (3). Recent epigenetic studies have also found evidence of differences in newborns exposed to mother’s prenatal depression (12). For example, a candidate gene study found that increased maternal depressed mood in the third trimester was associated in the infant with increased methylation of a binding site on the gene NR3C1, which codes for the glucocorticoid receptor (46). Thus, the present study adds to a small but growing body of literature supporting the hypothesis that exposure to maternal prenatal depression affects offspring immune parameters at birth.
Because we conceived of depression as a recurring disease, we examined the effects on the newborn of the mother’s ever having experienced depression before the birth. We also separately examined the effects of probable depression during pregnancy and of reporting a history of depression before the pregnancy and found that they were similar, supporting the conclusion that timing of overt depressive symptoms is not the critical factor in the association between maternal depression and neonate immune markers.
Another debate in the literature revolves around whether effects of maternal depression on offspring outcomes are primarily due to the impact of antidepressant medication itself, rather than the underlying depression (47). Given the similarity of our findings including or excluding women on antidepressants, it seems less likely that results are primarily due to a drug effect.
Our study has several limitations. Not all members of the cohort had cord blood drawn at delivery and assayed (see Supplemental Table, Supplemental Digital Content 6, http://links.lww.com/PSYMED/A551). Although in most respects those in the analytic sample were similar to the full cohort, they were less likely to have had a cesarean delivery (see Supplemental Digital Content 3, http://links.lww.com/PSYMED/A548) because cord blood samples were only collected during nonemergency deliveries. Previous work has found higher concentrations of IFN-γ, TNF-α, and IL-6 in cord blood of babies born vaginally versus by cesarean section (35), which might result in finding a stronger positive association between maternal depression and cord blood levels of these cytokines in the analytic sample. However, we did not find any evidence of positive associations. Alternatively, because those with emergency cesareans were more likely excluded from the analytic sample, and perinatal complications may be on the pathway from maternal depression to offspring immune responses, we may be underestimating effects. However, after excluding any pregnancies that resulted in cesarean delivery, our results were substantially unchanged.
Depressive symptoms were measured only once during pregnancy. Exposure misclassification due to experience of depression earlier or later in pregnancy would bias results toward the null if the timing of prenatal insults on fetal development is important. There is some support for such timing effects. For example, a study of methylation status of brain-derived neurotrophic factors in neonatal cord blood found an association with maternal depressed mood in the third, but not the second trimester (48).
Although all of the cytokine levels were measurable, those that fell within the linear portion of the standard curve may have been measured with more precision than the lowest and highest values, which may make it harder to see effects. We examined patterns of association across correlated outcomes and interpreted the results according to a priori hypotheses, without correcting for multiple outcomes, because doing so increases the possibility of a type II error and may overlook the possibility of an informative finding (49–51). However, when we controlled for the false discovery rate for IL-10 outcomes (52), we found that the p value for IL-10 stimulated with Bla g2 remained significant (for rationale, see Supplemental Digital Content 7, http://links.lww.com/PSYMED/A552). In addition, although our sample size of 463 was large enough to detect a somewhat small or medium effect (53), the study may not be sufficiently powered to make a null result fully informative regarding a small effect.
The Project Viva cohort is composed of predominantly white, well-educated women and their children residing in the Boston area. Whether these results will generalize to populations with a higher proportion of lower SES or minority groups remains to be seen. However, of note is that the prevalence of depression among poor and minority groups is higher than in white/higher SES groups. Thus, if our finding that prenatal depression affects neonate immune responses is replicated in more disadvantaged populations, it may provide insight into the origins of social disparities in health.
This study has several strengths as well. Exposures were assessed before outcomes occurred, a variety of measures of prenatal depression were available, and measurements were available for a rich array of potential confounders and predictors of cord blood immune parameters.
Prenatal depression is relatively common, and recent randomized controlled trials have demonstrated that it is treatable (54,55). Understanding the mechanisms by which a mother’s experience of depression before the birth of her children can affect offspring health will provide insights into how children get started on a more or less healthy developmental trajectory and potentially make more clear when and how one might buffer such effects to ensure that all children have a healthy start in life. In concert with other studies implicating the mother’s HPA activity and level of inflammation in dysregulation of the fetus’s own HPA and immune responses, our study found that neonates who were exposed to maternal depression had lower levels of an important anti-inflammatory cytokine at birth. The implications for future immune dysregulation or disease as the child grows are still not clear. Further longitudinal studies in this cohort (56) and others can shed light on whether immune differences observed at birth persist into childhood and beyond and whether they have an impact on an individual’s health. In addition, such studies can determine whether other alterations not yet seen at birth, such as reduced lymphocyte effector proliferation, globally diminished immune responses, or elevated inflammatory responses (13,39,57) become apparent later in life.
Conflicts of Interest and Source of Funding: Project Viva is supported by the National Institutes of Health (R01 HD 034568, R01AI102960). The authors report no conflicts of interests.
1. Howard LM, Molyneaux E, Dennis CL, Rochat T, Stein A, Milgrom J. Non-psychotic mental disorders in the perinatal period. Lancet. 2014;384:1775–88.
2. O’Hara MW, Wisner KL. Perinatal mental illness: definition, description and aetiology. Best Pract Res Clin Obstet Gynaecol 2014;28:3–12.
3. Gentile S. Untreated depression during pregnancy: short- and long-term effects in offspring. A systematic review. Neuroscience 2017;342:154–66.
4. Andersson NW, Hansen MV, Larsen AD, Hougaard KS, Kolstad HA, Schlünssen V. Prenatal maternal stress and atopic diseases in the child: a systematic review of observational human studies. Allergy 2016;71:15–26.
5. Entringer S, Buss C, Wadhwa PD. Prenatal stress, development, health and disease risk: a psychobiological perspective-2015 Curt Richter Award Paper. Psychoneuroendocrinology 2015;62:366–75.
6. Dhabhar FS, Burke HM, Epel ES, Mellon SH, Rosser R, Reus VI, Wolkowitz OM. Low serum IL-10 concentrations and loss of regulatory association between IL-6 and IL-10 in adults with major depression. J Psychiatr Res 2009;43:962–9.
7. Nathan C. Points of control in inflammation. Nature 2002;420:846–52.
8. Slavich GM, Irwin MR. From stress to inflammation and major depressive disorder: a social signal transduction theory of depression. Psychol Bull 2014;140:774.
9. Akbaba N, Annagür BB, Annagür A, Akbulut H, Akyürek F, Çelik Ç. Neurotrophins and neuroinflammation in fetuses exposed to maternal depression and anxiety disorders during pregnancy: a comparative study on cord blood. Arch Womens Ment Health 2017;1–7.
10. Coussons-Read ME, Okun ML, Nettles CD. Psychosocial stress increases inflammatory markers and alters cytokine production across pregnancy. Brain Behav Immun 2007;21:343–50.
11. Coussons-Read ME, Okun ML, Schmitt MP, Giese S. Prenatal stress alters cytokine levels in a manner that may endanger human pregnancy. Psychosom Med 2005;67:625–31.
12. Cao-Lei L, Laplante DP, King S. Prenatal maternal stress and epigenetics: review of the human research. Curr Mol Biol Rep 2016;2:16–25.
13. Plant D. Psychological and Biological Sequelae of Exposure to Prenatal Maternal Depression: Findings From the 25-Year Prospective South London Child Development Study [dissertation]. London: King’s College London (University of London); 2014.
14. Fransson E, Dubicke A, Byström B, Ekman-Ordeberg G, Hjelmstedt A, Lekander M. Negative emotions and cytokines
in maternal and cord serum at preterm birth. Am J Reprod Immunol 2012;67:506–14.
15. Mattes E, McCarthy S, Gong G, van Eekelen JA, Dunstan J, Foster J, Prescott SL. Maternal mood scores in mid-pregnancy are related to aspects of neonatal immune function. Brain Behav Immun 2009;23:380–8.
16. Eaton WW, Anthony JC, Gallo J, Cai G, Tien A, Romanoski A, Lyketsos C, Chen L-S. Natural history of Diagnostic Interview Schedule/DSM-IV major depression: the Baltimore epidemiologic catchment area follow-up. Arch Gen Psychiatry 1997;54:993–9.
17. Otte C, Gold SM, Penninx BW, Pariante CM, Etkin A, Fava M, Mohr DC, Schatzberg AF. Major depressive disorder. Nat Rev Dis Primers 2016;2:16065.
18. Danese A, Moffitt TE, Pariante CM, Ambler A, Poulton R, Caspi A. Elevated inflammation levels in depressed adults with a history of childhood maltreatment. Arch Gen Psychiatry 2008;65:409–15.
19. Friberg D, Bryant J, Shannon W, Whiteside TL. In vitro cytokine production by normal human peripheral blood mononuclear cells as a measure of immunocompetence or the state of activation. Clin Diagn Lab Immunol 1994;1:261–8.
20. Willwerth BM, Schaub B, Tantisira KG, Gold DR, Palmer LJ, Litonjua AA, Perkins DL, Schroeter C, Gibbons FK, Gillman MW. Prenatal, perinatal, and heritable influences on cord blood immune responses. Ann Allergy Asthma Immunol 2006;96:445–53.
21. Beijers R, Jansen J, Riksen-Walraven M, de Weerth C. Maternal prenatal anxiety and stress predict infant illnesses and health complaints. Pediatrics 2010;126:e401–e9.
22. Oken E, Baccarelli AA, Gold DR, Kleinman KP, Litonjua AA, De Meo D, Rich-Edwards JW, Rifas-Shiman SL, Sagiv S, Taveras EM. Cohort profile: Project Viva. Int J Epidemiol 2014;44:37–48.
23. Cox JL, Holden JM, Sagovsky R. Detection of postnatal depression. Development of the 10-item Edinburgh Postnatal Depression Scale. Br J Psychiatry 1987;150:782–6.
24. Matthey S, Ross-Hamid C. Repeat testing on the Edinburgh Depression Scale and the HADS-A in pregnancy: differentiating between transient and enduring distress. J Affect Disord 2012;141:213–21.
25. Milgrom J, Skouteris H, Worotniuk T, Henwood A, Bruce L. The association between ante- and postnatal depressive symptoms and obesity in both mother and child: a systematic review of the literature. Womens Health Issues 2012;22:e319–e28.
26. Rich-Edwards JW, Kleinman K, Abrams A, Harlow BL, McLaughlin TJ, Joffe H, Gillman MW. Sociodemographic predictors of antenatal and postpartum depressive symptoms among women in a medical group practice. J Epidemiol Community Health 2006;60:221–7.
27. Gold DR, Willwerth BM, Tantisira KG, Finn PW, Schaub B, Perkins DL, Tzianabos A, Ly NP, Schroeter C, Gibbons F, Campos H, Oken E, Gillman MW, Palmer LJ, Ryan LM, Weiss ST. Associations of cord blood fatty acids with lymphocyte proliferation, IL-13, and IFN-gamma. J Allergy Clin Immunol 2006;117:931–8.
28. Schaub B, Tantisira KG, Gibbons FK, He H, Litonjua AA, Gillman MW, Weiss S, Perkins DL, Gold DR, Finn PW. Fetal cord blood: aspects of heightened immune responses. J Clin Immunol 2005;25:329–37.
29. Baran J, Kowalczyk D, Ozóg M, Zembala M. Three-color flow cytometry detection of intracellular cytokines
in peripheral blood mononuclear cells: comparative analysis of phorbol myristate acetate-ionomycin and phytohemagglutinin stimulation. Clin Diagn Lab Immunol 2001;8:303–13.
30. Stites DP, Carr MC, Fudenberg HH. Development of cellular immunity in the human fetus: dichotomy of proliferative and cytotoxic responses of lymphoid cells to phytohemagglutinin. Proc Natl Acad Sci 1972;69:1440–4.
31. Chan-Yeung M, Ferguson A, Chan H, Dimich-Ward H, Watson W, Manfreda J, Becker A. Umbilical cord blood mononuclear cell proliferative response to house dust mite does not predict the development of allergic rhinitis and asthma. J Allergy Clin Immunol 1999;104:317–21.
32. Clerici M, DePalma L, Roilides E, Baker R, Shearer GM. Analysis of T helper and antigen-presenting cell functions in cord blood and peripheral blood leukocytes from healthy children of different ages. J Clin Invest 1993;91:2829–36.
33. Chapman DP, Perry GS, Strine TW. The vital link between chronic disease and depressive disorders. Prev Chronic Dis 2005;2:A14.
34. Wood RA, Bloomberg GR, Kattan M, Conroy K, Sandel MT, Dresen A, Gergen PJ, Gold DR, Schwarz JC, Visness CM, Gern JE. Relationships among environmental exposures, cord blood cytokine responses, allergy, and wheeze at 1 year of age in an inner-city birth cohort (Urban Environment and Childhood Asthma study). J Allergy Clin Immunol 2011;127:913–9.
35. Malamitsi-Puchner A, Protonotariou E, Boutsikou T, Makrakis E, Sarandakou A, Creatsas G. The influence of the mode of delivery on circulating cytokine concentrations in the perinatal period. Early Hum Dev 2005;81:387–92.
36. Macaubas C, De Klerk NH, Holt BJ, Wee C, Kendall G, Firth M, Sly PD, Holt PG. Association between antenatal cytokine production and the development of atopy and asthma at age 6 years. Lancet. 2003;362:1192–7.
37. White IR, Royston P, Wood AM. Multiple imputation using chained equations: issues and guidance for practice. Stat Med 2011;30:377–99.
38. Rubin DB. Multiple Imputation for Nonresponse in Surveys. Hoboken, NJ: John Wiley & Sons; 2004.
39. Kay G, Tarcic N, Poltyrev T, Weinstock M. Prenatal stress depresses immune function in rats. Physiol Behav 1998;63:397–402.
40. Ramratnam SK, Visness CM, Jaffee KF, Bloomberg GR, Kattan M, Sandel MT, Wood RA, Gern JE, Wright RJ. Relationships among maternal stress and depression, type 2 responses, and recurrent wheezing at age 3 years in low-income urban families. Am J Respir Crit Care Med 2017;195:674–81.
41. Iyer SS, Cheng G. Role of interleukin 10 transcriptional regulation in inflammation and autoimmune disease. Crit Rev Immunol 2012;32:23–63.
42. Merlot E, Couret D, Otten W. Prenatal stress, fetal imprinting and immunity. Brain Behav Immun 2008;22:42–51.
43. Stroud LR, Papandonatos GD, Parade SH, Salisbury AL, Phipps MG, Lester BM, Padbury JF, Marsit CJ. Prenatal major depressive disorder, placenta glucocorticoid and serotonergic signaling, and infant cortisol response. Psychosom Med 2016;78:979–90.
44. Coussons-Read ME. Stress and Immunity in Pregnancy. In: Segerstrom S, Segerstrom SC, editors. The Oxford handbook of psychoneuroimmunology. New York, NY: Oxford University Press; 2012. p.3–17.
45. Palma-Gudiel H, Cirera F, Crispi F, Eixarch E, Fañanás L. The impact of prenatal insults on the human placental epigenome: a systematic review. Neurotoxicol Teratol 2018;66:80–93.
46. Oberlander TF, Weinberg J, Papsdorf M, Grunau R, Misri S, Devlin AM. Prenatal exposure to maternal depression, neonatal methylation of human glucocorticoid receptor gene (NR3C1) and infant cortisol stress responses. Epigenetics 2008;3:97–106.
47. Hannestad J, DellaGioia N, Bloch M. The effect of antidepressant medication treatment on serum levels of inflammatory cytokines
: a meta-analysis. Neuropsychopharmacology 2011;36:2452–9.
48. Appleton AA, Buka SL, Loucks EB, Rimm EB, Martin LT, Kubzansky LD. A prospective study of positive early-life psychosocial factors and favorable cardiovascular risk in adulthood. Circulation 2013;127:905–12.
49. Rothman KJ. No adjustments are needed for multiple comparisons. Epidemiology (Cambridge, Mass) 1990;1:43–6.
50. Schulz KF, Grimes DA. Multiplicity in randomised trials I: endpoints and treatments. Lancet 2005;365:1591–5.
51. Streiner DL. Best (but oft-forgotten) practices: the multiple problems of multiplicity—whether and how to correct for many statistical tests. Am J Clin Nutr 2015;102:721–8.
52. McDonald JH. Handbook of Biological Statistics. 3rd ed. Baltimore, MD: Sparky House Publishing: 254-60.
53. Cohen J. A power primer. Psychol Bull 1992;112:155.
54. Tandon SD, Perry DF, Mendelson T, Kemp K, Leis JA. Preventing perinatal depression in low-income home visiting clients: a randomized controlled trial. J Consult Clin Psychol 2011;79:707–12.
55. O’mahen H, Himle JA, Fedock G, Henshaw E, Flynn H. A pilot randomized controlled trial of cognitive behavioral therapy for perinatal depression adapted for women with low incomes. Depress Anxiety 2013;30:679–87.
56. Ly NP, Rifas-Shiman SL, Litonjua AA, Tzianabos AO, Schaub B, Ruiz-Pérez B, Tantisira KG, Finn PW, Gillman MW, Weiss ST, Gold DR. Cord blood cytokines
and acute lower respiratory illnesses in the first year of life. Pediatrics 2007;119:e171–8.
57. O’Connor TG, Moynihan JA, Caserta MT. Annual Research Review: the neuroinflammation hypothesis for stress and psychopathology in children–developmental psychoneuroimmunology. J Child Psychol Psychiatry 2014;55:615–31.