Extending the developmental origins of disease model: Impact of preconception stress exposure on offspring neurodevelopment

Abstract

The concept of the developmental origins of health and disease via prenatal programming has informed many etiologic models of health and development. Extensive experimental research in non-human animal models has revealed the impact of in utero exposure to stress on fetal development and neurodevelopment later in life. Stress exposure, however, is unlikely to occur de novo following conception, and pregnancy health is not independent of the health of the system prior to conception. For these reasons, the preconception period is emerging as an important new focus for research on adverse birth outcomes and offspring neurodevelopment. In this review we summarize the existing evidence for the role of preconception stress exposure on pregnancy health and offspring neurodevelopment across species and discuss the implications of this model for addressing health disparities in obstetrics and offspring outcomes.

Developmental origins of health and disease

Fetal programming is a model for understanding the development of health and disease focused on prenatal conditions that impact the vulnerability of individuals to multiple pathologies (Barker 1995; 2003). Maternal exposure to environmental stressors during the prenatal period is one such condition that has been consistently linked with suboptimal developmental outcomes in the offspring. For example, evidence from animal studies demonstrates that maternal stress during pregnancy can permanently compromise offspring neurodevelopment (Chapillon et al., 2002; Weinstock, 2005) as evidenced by disturbances in executive function (Keenan & Hipwell, 2015; Schneider et al., 2002), impaired learning and disruption in neurogenesis (Chapillon et al., 2002; Coe, Lulbach & Schneider, 2002) and heightened anxiety-like behaviors (Schneider et al., 2002). The strength of the causal claim that maternal stress has a direct impact on offspring development is based on rigorous controlled experiments in animals, including distinguishing prenatal from postnatal effects using methods such as cross-fostering or nursery rearing.

In contrast to models tested in animals, research on the effect of prenatal stress on human offspring is more heterogenous both in terms of operational definitions of stress and in the consistency of the results. In humans, prenatal stress has been conceptualized as bereavement, financial hardship, maternal anxiety and depression, infection, poor nutrition, and exposure to natural disasters or terrorist attacks, each of which may co-occur and may be relatively chronic across development. Despite the heterogeneity in methodologies and operational definitions of prenatal stress, the pattern of findings in this body of research largely mirrors the findings from studies of animals, with some exceptions (see reviews by Kim, Bale & Epperson, 2017 and Talge et al., 2007). Several non-experimental, but prospective, studies of humans have also shown that maternal stress during pregnancy is associated with adverse birth outcomes (Ruth et al., 2012; Kent et al., 2013; Giscombe & Lobel, 2005), shorter gestational length, low birth weight, and preterm birth (Dole et al., 2003; Dominguez et al., 2005; Holzman et al., 2001; Class et al., 2011; Dayan, et al., 2006; Glover, 2015; Wadhwa et al., 2011; Rice et al., 2010). Prenatal stress also is associated with impairments in multiple systems related to offspring neurodevelopment, including alterations in the reactivity of the offspring’s hypothalamic-pituitary-adrenal (HPA) axis (Keenan, Gunthorpe & Grace, 2007; O’Connor et al., 2005; Davis et al., 2011; Entringer et al., 2009), executive functioning (e.g., attention, learning, language problems) (O’Connor, Heron, & Glover, 2002; Buss et al., 2012; Laplante et al., 2008; Bergman et al., 2010), and neural structure and development (Korosi et al., 2012). Furthermore, maternal stress during pregnancy has been associated with an increased risk for neurodevelopmental disorders, including attention deficit/hyperactivity disorder (Linnet et al., 2003; Rodriguez & Bohlin, 2005). These neurodevelopmental deficits and disorders are known to increase risk for later health problems and general impairments (Kelly et al., 2013; Colman, et al., 2014; Sibley et al., 2014; Hankin et al., 2010), underscoring the critical importance of optimizing the health of the fetal environment.

From a developmental and life-course perspective, stress exposure is unlikely to occur de novo following conception. Similarly, pregnancy health is not independent of the health of the system prior to conception. For these reasons, the preconception period is emerging as an important new focus for research on adverse birth outcomes and offspring neurodevelopment (Grandjean et al., 2015), as well as a model for understanding and ultimately preventing health disparities during pregnancy and in child health outcomes (Vaivada et al., 2017). We review the existing evidence for the role of preconception stress exposure on pregnancy health and offspring neurodevelopment across species and outline a program of research to further probe the role of preconception stress exposure within the developmental origins of disease model.

Extending the developmental origins of disease model

Evidence suggests that one primary mechanism by which stress exposure confers risk for offspring neurodevelopment is via maternal neuroendocrine functioning, including the HPA axis. Support for this mechanism comes from decades of research linking prenatal maternal stress and adverse offspring outcomes with dysregulation of maternal gluccocorticoids (GCs) and placental stress regulation on the rapidly developing fetal nervous system. Briefly, the normal response to stress has the effect of maintaining physiological homeostasis, which typically results in adrenal activation and the release of cortisol. However, repeated or chronic increases in stress exposure, and hence chronically increased plasma cortisol, can lead to responses that have pathological and illness-inducing consequences for the offspring: maternal GCs are transported across the placenta, entering the fetal circulation leading to alterations in fetal stress architecture (Weinstock, 2005).

Additionally, the maternal HPA axis system is impacted by the growth and development of the placenta, (O’Donnell, O’Connor & Glover, 2009). In contrast to the inhibitory control of cortisol on the expression of CRH in a non-pregnant state, maternal cortisol activates the promoter region in the placenta and stimulates further CRH synthesis during pregnancy (King, Smith, & Nicholson, 2001). Maternal GCs also induce placental CRH production that stimulates the fetal HPA axis, leading to increased levels of fetal cortisol (Beijers, Buitelaar, & de Weerth, 2014). Typically, the placental enzyme 11-β hydroxysteroid dehydrogenase-type 2 (11-β-HSD2) converts maternal cortisol to inactive cortisone, thereby decreasing the level of GCs in the placenta. Prenatal stress exposure, however, results in the down-regulation of the 11-β-HSD2 enzyme, which in turn exposes the fetus to higher levels of maternal GCs (O’Donnell, et al., 2012).

Substantial work has led to the identification of numerous candidate genes associated with individual differences in stress regulation. The glucocorticoid receptor (GR) gene, for example, is a well-researched gene; several single nucleotide polymorphisms (SNPs) have been functionally characterized including. Experimental manipulations of the sensitivity of GR have been associated with impaired functioning in the negative feedback of the HPA-axis and depression in animals (Pariante, 2004). In humans, the BclI GG genotype is associated with the high levels of cortisol in response to a controlled laboratory stressor (Kumsta et al., 2007). Variations in the FK506-binding protein 5 (FKBP5) gene also are associated with the regulation of the HPA axis, particularly in the recovery following termination of the stressor. Ising and colleagues (2008) found that three SNPs of the FKBP5 gene were associated with cortisol response to a social stressor. Thus, genotype likely accounts for some variability in sensitivity to stress exposure prior to conception as well as changes in HPA axis functioning over the course of pregnancy.

There is increasing evidence that epigenetic effects are likely to play a major role in the molecular mechanisms underlying the long-lasting effect of stress exposure on adult health. Indeed, there is a growing body of evidence that the epigenome is responsive to environmental exposures including the social environment both across development in animals and in humans. A potential mechanism by which prenatal stress impacts offspring outcome is via epigenetic changes in placental 11β-HSD2. In rodent models, maternal stress during pregnancy has been shown to decrease placental 11β-HSD2 mRNA levels (Mairesse et al., 2007; Welberg, Seckl & Holems, 2000), thus impairing the conversion of maternal glucocorticoids to less active forms. In humans, maternal depression and anxiety during pregnancy has been associated with increased maternal NR3C1 methylation, and greater methylation of placental NR3C1 and 11β-HSD2, as well as indications of offspring neurodevelopmental problems (Hompes et al., 2013; Jensen Peña, Monk & Champagne, 2012; O’Donnell et al., 2012; Togher, et al., 2017).

A life-course perspective (Elder & Rockwell, 1979) naturally extends the prenatal stress model to the integrity of the system prior to conception. Numerous studies have shown links between stress exposure and compromised health earlier in development. For example, psychosocial stress exposure (e.g., poverty, neighborhood crime) during childhood and adolescence is associated with higher ambulatory blood pressure (Beatty & Matthews, 2009) and higher levels of C-reactive protein (CRP) (Broyles et al., 2012; Fuligni et al., 2009; Lande et al., 2008), both of which signal risk for later disease. Exposure to stress in childhood and adolescence is also associated with cortisol reactivity to a controlled stressor (Jaffee et al., 2015), higher concentrations of hair cortisol (Simmons et al., 2016), greater inflammatory activity including Tumor Necrosis Factor (TNF)-alpha (Hartwell et al., 2013), as well as altered Interleukin-6 (IL-6) immune response to a bacterial stimulus in the laboratory (Miller, Rohleder & Cole, 2009).

Adolescence may be a point in development during which stress exposure is especially impactful on health. Data from cross-sectional studies suggest that there are significant changes in the functioning of the HPA axis during late adolescence as measured by diurnal cortisol and cortisol reactivity (Gunnar et al., 2009; Oskis et al., 2009; Sumter et al., 2010). There also appear to be changes in functioning of brain regions involved in the regulation of the HPA axis during late adolescence (McCormick et al., 2010). In a study of rodent behavior, (Jankord et al., 2011), the effects of chronic, variable stress were exaggerated in animals exposed in late adolescence compared to those exposed in early adolescence and adulthood, and the late adolescent exposed animals were the only group for which stress was associated with an increase in basal corticosterone. Evidence for the impact of early stress exposure of epigenetic changes that impact HPA-axis functioning is emerging. Childhood adversity, such as parental loss and maltreatment, has been associated with increased methylation and alterations in HPA axis response to a controlled stressor (Tyrka et al., 2012). In one of only a handful of studies testing the impact of preconception stress exposure on perinatal gene expression, preadolescent female mice who were exposed to chronic variable stress differed from controls in their HPA axis response to stress during pregnancy, but not postpartum, and in expression of a number of genes in the paraventricular nucleus of the hypothalamus (Morrison et al., 2016). These findings suggest that adolescence may be a developmental period during which the HPA-axis may be highly sensitive to environmental factors, and that exposures during this period may shape the responsiveness of the system in the future. Stress exposure, therefore, is associated with health indices across development, and dimensions of exposure such as developmental timing, chronicity and acuity are likely important factors. There has been relatively little research, however, linking work on stress exposure over the course of development on later pregnancy health and offspring development. Among the existing studies, the period immediately preceding conception has been the focus. We review that literature in the following section.

Non-human models of preconception stress exposure on offspring neurodevelopment

Animal models provide compelling evidence for preconception stress effects on offspring neurodevelopment. Exposure to highly translatable stressors (e.g., overcrowding, temperature, and pain stress), results in altered behavior, neurobiology and brain morphology; effects which have been shown to persist into adulthood. Results from these studies are summarized in Table 1. Shachar-Dadon and colleagues (2009) exposed female rats to unpredictable and variable stress (e.g., swim test, isolation, water and food deprivation) for either a week or two weeks prior to mating and compared the offspring of those two groups (n = 83 and 25, respectively) to offspring of unexposed females (n= 86). Five probes were used to examine behavior of the offspring in adulthood: navigation of an elevated maze, shock avoidance learning, acoustic startle response, open field test, and social interaction test; with offspring from each litter distributed across each test. Stress exposure experienced within one week of, but not two weeks prior to, conception resulted in increased anxiety and activity and decreased social interaction in the offspring. There were also significant interactions between stress exposure and sex of the offspring on later development: preconception stress exposure resulted in greater shock avoidance among male offspring whereas the opposite effect was observed for females. In contrast, male offspring of dams exposed to preconception stress showed less startle than control males, whereas the female offspring stress showed greater startle than control females. In a follow-up to the study described above, Bock and colleagues (2016) examined dendrite morphology in the medial prefrontal cortex (mPFC) in a subsample of the offspring. These analyses revealed an effect for preconception stress experienced two weeks prior to conception: dendrites in the left mPFC were more complex in male and female offspring and the spines longer in the male offspring of the stress exposed dams. Within the same sample, therefore, effects of preconception stress varied as a function of timing, sex of the offspring, and phenotype, as well as the characterization of altered offspring behavior as heightened or dampened.

Table 1.

Animal models providing evidence for preconception stress effects on offspring neurodevelopment.

Reference	Model/design	Dependent Measure	Results
Bock et al (2016)	Compared offspring of female rats assigned to an unstressed control group (n = 11) to two experimental groups that were stressed one week before mating (n= 15) or 2 weeks before mating (n= 9).	Spine number/density and dendritic length/complexity in offspring at postnatal day 65.	There was a significant increase in spine number in the left hemisphere of the medial prefrontal cortex (mPFC) of male and female offspring of dams who were stressed 2 weeks before mating versus the control group (p = 0.011). Male offspring of dams who were stressed 2 weeks before mating also had longer dendrites (p = 0.02) compared to the control group. There were no other significant group differences.
Shachar et al. (2009)	Compared adult offspring (n = 146) of dams assigned to an unstressed control group (n = 86) to two experimental groups: one that mated immediately following 7 days of stress exposure (n = 83) and one that mated 2 weeks after the stress exposure (n = 25).	Activity (lines crossed on runway), tone and shock response (avoidance, escape or escape failure), startle response, prosocial behavior, asocial behavior.	Offspring of preconception stress dams were more active, in terms of lines crossed in the elevated maze plus, versus those in the control group (p < .001). There were also group effects in prosocial behavior (p < 0.05), and asocial behavior (p < 0.001). There were no other group differences.
Huang et al (2010)	Compared 2-month old offspring of adult female rats assigned to an unstressed control group(n = 8) versus a chronic, unpredictably stressed (CUS) group (n = 12).	Escape latency in the Morris water maze task, serum corticosterone (COR) levels, expression of brain-derived neurotrophic factor (BDNF), and N-methyl-D-aspartate (NMDA) 2A (NR2A) and 2B (NR2B) receptors in the hippocampus.	Escape latency of the offspring of dams who were stressed preconception was significantly longer than that of the control group (p < 0.01). There was a significant increase in the COR levels of the offspring of preconception stress dams versus the control group (p < 0.01). Decreased expression of BDNF and NR2B in the hippocampus in offspring of preconception stress dams versus offspring of the control group (p < 0.05).
Li et al., (2010	Compared adult offspring of female rats assigned to an unstressed control group (n = 8) versus a 21-day chronic, unpredictable stress (CUS) group (n=8)	Dam behavior pre- and post-administration of CUS; offspring behavior in the Morris Water Maze test; 5-HT in the hypothalamus, NE levels in the hippocampus, P-CREB in the hippocampus	Exposed offspring had decreased spatial memory, lower sucrose consumption, higher levels of norepinephrine (NE) in the hippocampus, lower levels of serotonin (5-HT) in the hypothalamus, and less protein expression of phosphorylated Cyclic AMP responsive element binding protein (P-CREB).
Zaidan et al (2013; 2015)	Compared brains of male (n= 6) and female (n=5) F1 offspring of rats exposed to 1 week of chronic, unpredictable stress two weeks prior to conception to male (n=5) and female (n= 7) controls; and corticosterone in F2 (7 males and 5 females) and controls (7 males and 7 females)	expression of CRF1 receptors in the brain in first generation (F1), and corticosterone levels in F1 and second generation (F2) offspring compared to offspring of controls	Expression of corticotropin releasing factor (CRF) 1 and 2 receptors were increased in neonatal and adult brains, comparedto brains of control offspring. Corticosterone levels were increased in the F1 females, and F2 males anddecreased in F2 females compare to offspring of controls

Li and colleagues (2010a) reported statistically significant differences in behavior and neurochemistry between adult male offspring of maternal rats exposed to 21 days of chronic unpredictable stress prior to conception (n = 8) to male offspring of control dams (n=8). Regarding impact on behavior, exposed offspring had decreased spatial memory, and lower sucrose consumption, which may indicate altered reward sensitivity. In terms of neurochemistry, exposed offspring had higher levels of norepinephrine (NE) in the hippocampus, lower levels of serotonin (5-HT) in the hypothalamus, and less protein expression of phosphorylated Cyclic AMP responsive element binding protein (P-CREB), which also may indicate dysregulated serotonergic and noradrenergic neurotransmitter systems.

Neurochemistry and behavior also were altered as a function of chronic, unpredictable, and variable preconception stress administered for a period of three weeks immediately prior to conception (Huang et al., 2010). In this study, behavior of maternal rats was observed prior to and after stress exposure. Weight loss and decreases in sucrose intake and motor activity during the open field test, all of which are animal analogues of depression phenotypes, were observed in the stress exposed rats. Adult male and female offspring of stress exposed rats had longer escape latencies during the Morris water maze task and higher serum corticosterone levels following the task. Expression of brain-derived neurotrophic factor and N-methyl-D-aspartate receptor in the hippocampus, both of which are critical to synaptic plasticity, was decreased in the stress exposed offspring compared to the controls. In this study, preconception stress impacted several systems – maternal behavior and offspring learning, and offspring HPA axis activity, and neural integrity.

Effects of preconception stress exposure on neuroendocrine functioning have been observed in both the first (F1) and second (F2) generation offspring. Zaidan and colleagues (2013; 2015) exposed female rats to one week of chronic, unpredictable stress two weeks prior to conception. First, expression of corticotropin releasing factor (CRF) type 1 and 2 receptors in the brain, which mediate the initial activation of the HPA axis response to stress, were increased in neonatal and adult brains, compared to brains of control offspring. In addition, corticosterone levels were increased in the F1 females. Corticosterone levels were also altered as a function of preconception stress exposure in the second generation: males showed higher levels, whereas females showed lower levels, than offspring of controls.

Human models of preconception stress and offspring health

A small but growing evidence base in human studies provides preliminary support for the impact of preconception stress exposure on birth outcomes. Many of these studies are based on maternal retrospective report of stress exposure (see Table 2 for a summary). For example, in a Swedish population-based cohort, preconception stress, operationalized as maternal bereavement of a first-degree relative within 6 months of conception, was associated with elevated risk of adverse birth outcomes (e.g., preterm birth, low birth weight, small for gestational age), as well as increased risk for infant mortality (Class et al., 2013). These findings have been replicated and extended to predict childhood mortality in a Danish population-based sample (Class et al., 2015), in which bereavement experienced prior to conception, but not post-conception, increased the risk of neonatal and infant mortality.

Table 2.

Human models providing evidence for preconception stress effects on offspring health and neurodevelopment.

Reference	Sample	Model/design	Dependent Measure	Results
Class et al. (2013)	3,055,361 Swedish population based sample of offspring born between 1973 and 2008	Tested the association between preconception (6 months before conception) and prenatal (across pregnancy) bereavement stress, on birth outcomes and infant mortality.	Preterm birth (PTB), defined as < 37 weeks gestation; small for gestational age (SGA) defined by birth weight less than 2 standard deviations below the mean for GA; infant mortality (8,398 cases) from medical records.	Preconception stress increased risk for PTB (AOR=1.19; 95% CI=1.12–1.26) and SGA (AOR=1.14; 95% CI=1.05–1.23); Preconception stress increased risk for infant morality controlling for PTB and SGA (AOR=1.37; 95% CI=1.11–1.70. Prenatal stress was not associated with birth outcomes or infant mortality
Class et al. (2015)	1,865,454 Danish population-based sample of offspring born 1979 to 2009	Tested the association between preconception (6 months before conception), prenatal (across pregnancy) bereavement stress, on infant mortality.	Neonatal (0–28 days), infant (29–364 days), and early childhood (1–5 years) mortality from medical records	Preconception stress was associated with neonatal (AOR = 1.87, 95% confidence interval = 1.53–2.30) and infant mortality (AOR = 1.52, 95% confidence interval = 1.15–2.02). Bereavement during the prenatal period was not associated with offspring mortality
Baird et al. (2009)	874,874 mother and offspring pairs from the Southampton Women’s Survey (SWS) followed through pregnancy and 12 months of age.	Tested the association between maternal psychological distress before conception and offspring sleep disturbance during infancy.	Number of times babies woke on average between the hours of midnight and 6:00am each night during a 2-week period at 6 and 12 months of age	At 6 and 12 months, offspring whose mothers experienced preconception psychological distress had an increased risk of waking compared to offspring of women not reporting preconception stress.
Harville et al. (2010)	Sample of 4,865 women who were enrolled as children in the prospective National Child Development Study in Great Britain in 1958, and who later had at least 1 singleton birth	Tested the association between prospectively collected data on childhood stress exposure (e.g., including financial hardship, family dysfunction, violence/mental health issues, and family structure) and low birth weight and preterm birth.	Low birth weight (LBW) defined as below 2500 grams; preterm birth (PTB), defined as more than 3 weeks prior to due date. Participants were also asked if they smoked before or during the pregnancy.	Childhood stressors were associated with risk of LBW (OR = 1.51, 95% CI 1.10–2.06) and PTB (OR = 1.44, 95% CI 1.08–1.92)., controlling for smoking status and adult social class.
Li et al. (2009; 2010)	Registry study of 1,015,912 singletons born in Denmark from 1987 to 2001.	Tested the association between preconception maternal bereavement and offspring risk for attention-deficit/hyperactivity disorder (ADHD), and cerebral palsy.	ADHD hospitalization or administration of ADHD medication after 3 years of age; Diagnosis of cerebral palsy.	Maternal bereavement due to death of a child or spouse during preconception was associated with increased risk of ADHD (HR 1.44, 95% CI 1.00 – 2.16) in male offspring; maternal bereavement during the periods of 7 to 12 months before pregnancy 2.29 (95% CI, 1.54 – 3.43) and 6 to 0 months before pregnancy 1.93 (95% CI, 1.34 −2.78) was associated with risk of cerebral palsy.

Note: AOR = adjusted odds ratio; CI = confidence interval.

Data from representative U.S. samples also support the association between retrospective recall of preconception stress exposure and offspring outcome, as well as the unique effects of preconception exposure, controlling for post-conception exposure. In a study of nearly 1,000 participants, women who reported stressful life events (e.g., death of a family member) prior to conception were 40% more likely to deliver a child of very low birth weight, controlling for pregnancy complications; stressful life events experienced during pregnancy were not associated with birth weight (Witt et al., 2014a). The magnitude of this association was greatest among offspring of women residing in low-income neighborhoods (Witt et al., 2014a).

One important study to this emerging field in humans is The National Child Development Study, a cohort study of children born in Britain during one week of March 1958. Over 18,000 participants were enrolled and then followed up at ages 7, 11, 23, 33, and 41 (Atherton et al., 2008). Stressors were assessed by parental report and interviewer observation in the home and included financial, parenting, family, and community stressors. At ages 33 and 41 years, female participants (n≈5,000) were asked about pregnancy outcomes, including gestational age and birth weight. The results revealed that exposure to stressors during childhood was associated with higher rates of preterm birth and low offspring birth weight. These results held even after controlling for stress exposure and smoking during pregnancy (Harville et al., 2010).

Investigation of associations between preconception stress and indices of later neurodevelopment are scarce. First, in a cohort of women from the Southampton Women’s Survey (SWS), who were recruited between 20–34 years of age and followed through their subsequent pregnancies, associations between maternal preconception psychological stress (e.g., symptoms of depression and anxiety) and infant sleep were tested. Based on a sample of 874 mother-infant pairs, preconception psychological distress was prospectively associated with compromised sleep in offspring, including middle of the night awakening, at 6 and 12 months even when controlling for postnatal symptoms of distress (Baird et al., 2009). In a nationwide population-based cohort study, including all 1,015,912 singletons born in Denmark from 1987 to 2001, associations between bereavement stress and ADHD were tested: 29,094 children were born to women who lost a close relative during pregnancy or up to 1 year before pregnancy. Maternal bereavement stress in the 6 months prior to pregnancy was shown to increase the risk for attention-deficit hyperactivity disorder (ADHD) in male offspring by almost fifty percent (Li et al., 2010b).

Although relatively few studies have been conducted in human samples, there is growing evidence of an association between preconception stress exposure and birth outcomes, using both retrospective and prospective data. Extant data from animal studies converge to highlight the potential critical influence of preconception stress exposure on offspring neurodevelopment in ways that may affect learning, memory, and stress reactivity.

Developmental timing and chronicity of preconception stress exposure

The timing of preconception stress exposure may differentially influence offspring outcomes. Although the period immediately prior to pregnancy may present the most vulnerable period for uterine priming, successful implantation of the zygote and healthy early placental development, more distal environmental exposures leading up to that point may be as important in shaping maternal biological, physiological, and psychological responses to stress. Additionally, a life-course maternal “health capital” perspective that considers all gains and losses in biological, psychological, and physical health over the lifetime is likely to provide the fullest picture of the impact of environmental exposures.

For females, adolescence is a likely sensitive period with regard to preconception stress exposure. In fact, a number of prevention scientists have argued that preconception care should begin in puberty (Witt et al., 2014; Dean et al., 2013). This is based, in part, on evidence from neurodevelopmental studies that have revealed significant changes in brain morphology and function during adolescence (Casey, Getz & Galvan, 2008; Zehr et al., 2006). The pruning of brain regions involved in emotion regulation and higher cognitive functioning is hypothesized to render these regions particularly vulnerable to the experience of stressors (McCormick et al., 2010). Changes in adrenal hormones also occur during adolescence. Among girls, these changes result in slower recovery to baseline of cortisol in response to a stressor with more advanced maturation (Stroud et al., 2004). The effects of age on stress response are also evident in animal models: pubertal female rats have prolonged corticosterone release in response to restraint stress compared to adult females (Romeo, Lee & McEwen, 2005). This potential adolescent programming of the HPA axis may have implications for later functioning. For example, female, but not male, rats who were exposed to stress during adolescence showed altered behavioral response to stress in adulthood (McCormick et al., 2005). In humans, individual differences in HPA-axis functioning in adolescence are associated with self-reported exposure to stressors, and these individual differences appear to be relatively stable (Doane et al., 2015). Thus, late adolescence may be a particularly vulnerable period for sensitization of the HPA axis to environmental stress resulting in stable individual differences in reactivity and regulation that continue into adulthood.

We are aware of only one study in which the developmental timing of preconception stress on offspring development has been tested. In the aforementioned prospective cohort study of nearly 5,000 women (Harville et al., 2010), exposures most strongly influenced the birth outcomes if they occurred in adolescence: the highest risk for both low birth weight and preterm birth was among females who experienced multiple hardships in adolescence only, although exposure at any time during childhood/adolescence raised the risk of poor birth outcomes. These results are consistent with the conceptualization of adolescence as a critical period with regard to the development and sensitivity of biological systems involved in later reproductive health. However, a much clearer understanding of the developmental effects of stress exposure is needed given that life stressors are known to cluster and interact, leading to chronic and accumulating patterns across development (Evans, Li & Whipple, 2013). Furthermore, a wealth of data has shown that exposure to sustained, repeated or multiple stressors is associated with more severe and chronic impairment and distress in individuals (Clancy et al., 2006; Cloitre et al., 2001; Dohrenwend et al., 2006; Gerard & Buehler, 2004; Larson et al., 2008; Margolin et al., 2010; Mistry et al., 2010) than is exposure to single, discrete forms of stress. Gaining an understanding of the developmental effects of stress exposure on health, including prenatal health, is fundamental to informing the timing and type of prevention efforts.

Relevance of preconception stress exposure to understanding health disparities

As stated earlier, a developmental and life-course perspective would posit that because stress exposure is unlikely to occur de novo following conception, pregnancy health is not independent of the health of the system prior to conception. In the U.S., higher levels of acute and chronic stress are found among families living in low-income environments than among families living in other income environments. Neighborhood disorder, lack of safety and exposure to violence are all significantly higher in areas with lower per capita income (Evans, 2003; Ewart & Shuchdat, 2002). African Americans live in poverty at a disproportionately high rate; more than a quarter of African Americans live in poverty (DeNavas-Walt & Proctor, 2014).

Both minority race and living in a low resourced environment appear to impact both the diurnal rhythm and feedback loop of the stress response system, and the interface between HPA-axis and other systems critical for maintaining health such as immune functioning beginning in adolescence and continuing into adulthood. Discrimination and unfair treatment as a result of minority status (e.g., race, poverty) are associated with health risks in adolescence, including higher ambulatory blood pressure (Beatty & Matthews, 2009) and higher levels of CRP (Lande et al., 2008). Higher levels of cortisol in the afternoon and evening have been reported among African Americans and individuals living in lower SES environments than among European Americans and individuals living in higher SES environments (Chen & Paterson, 2006). African American women also are more likely to demonstrate a significant increase in cortisol in response to a psychosocial stressor than are European American women (Fowles & Gabrielson, 2005). Furthermore, data from studies using exposure to a controlled stressor provide evidence for racial differences in inflammatory response (i.e., interleukin-6) to stress, with African American pregnant and non-pregnant women show higher responses than European American women (Christian et al., 2013). In a study in which both cortisol and pro-inflammatory cytokines were measured during pregnancy, minority race and low-income status was characterized by high levels of cortisol without a compensatory decrease in cytokines, suggesting impaired feedback between the neuroendocrine and immune systems (Corwin et al., 2013).

In the U.S., African American women living in urban low-income environments are more likely to experience pregnancy complications than other women (Giscombé & Lobel, 2005; Kent et al., 2013). A primary cause of maternal and child health disparities for African Americans in the U.S. is likely due to compromised health from earlier stress exposure. If this is the case, then successfully improving health of the offspring via prenatal interventions will be challenging, as the maternal systems for maintaining health have already been impacted for the current generation. The application of the developmental origins of health and disease model to health disparities may serve to identify targets that will disrupt intergenerational cycles of poverty and impairment.

Conclusion and future directions

There is strong evidence for negative effects of maternal prenatal stress on the developing fetus; effects that continue to impact development throughout childhood. There is developing evidence that the observed association between prenatal stress and offspring neurodevelopment may be largely due to stress exposures that occur prior to conception. Consistent with a kindling or stress sensitization model, we posit that the development of the stress architecture during childhood and adolescence is in part based on environmental inputs, and that observed differences in stress regulation during pregnancy are largely due to variability in those earlier inputs. This hypothesis is testable via several interrelated approaches. First, more research is needed to understand the developmental timing of stress exposure on pregnancy health and offspring neurodevelopment, including models that compare chronicity and type of stressor (e.g., social, safety, health). Second, animal models will be critical for characterizing the unique effects of preconception stress exposure on pregnancy health and postnatal caregiving and offspring development. For example, cross-fostering is needed to determine the relative contribution of preconception stress on fetal development as opposed to postnatal caregiving effects on offspring neurodevelopment. Third, examining prevention effects will provide further information on the relative impact of preconception stress on offspring development. An example is using an enriched environment to attenuate earlier stress exposure. Cutuli and colleagues (2017) demonstrated that exposure to an enriched environment prior to reproduction had neuroprotective effects on the offspring even after exposure to a stressor; adolescent offspring of enriched dams had a more modulated immune response and more climbing behavior in response to the forced swim test, and greater expression of gluccocorticoid receptors in the amygdala, which was comparable to adolescent offspring of dams who were not stress exposed. Testing the timing and dose of an enriched environment on the attenuation of the effects of preconception stress exposure on offspring neurodevelopment is necessary for pursuing a preventive intervention program in humans.

In parallel to hypothesis-testing in animal models, prospective studies of humans from childhood through pregnancy are required to provide further evidence and to characterize the stress phenotype in terms of type, timing, and chronicity. Much of the current research in humans is limited by retrospective, self-reports of exposures and or emotional distress. Bias in recall and individual differences in perceptions and definitions of stress exposure may obscure true effects between preconception stress exposure and pregnancy health. Alternatively, observed associations based on recall and individual differences in perceptions may be in fact due to third variables such as genetic factors that account for variance in both maternal recall of stress exposure and infant outcomes. The likelihood of the co-occurrence stressors in humans (e.g., parental bereavement, financial hardship, poor nutrition) provides an additional challenge. In addition, although most models of social stress exposure (e.g., economic, housing, community stress) in humans assume chronicity of exposure or equivalence of effects across development, there is often substantial variability in timing and chronicity within and across domains of stressors (e.g., Keenan et al., unpublished manuscript). Capturing these dimensions will be important for testing the specificity of timing of exposure on offspring development, especially for stress exposure that occurs both pre- and post-conception. Key elements to future programs of research in humans, therefore, include prospective, repeated assessments of a range of stress exposure in childhood and adolescence and objective biomarkers of stress sensitivity and stress related health conditions in controlled settings. Characterizing stress phenotypes at multiple levels (e.g., behavioral, immune, endocrine), as well as severity, timing, and chronicity, will be important for articulating measurable targets of preventive interventions.

The results from the proposed research agenda will fill a critical gap in knowledge about the developmental origins of disease. To date, the model has been limited by right-hand censoring, with the possibility that effects attributed to exposures during pregnancy are in fact causally linked to environmental exposures occurring prior to conception and the resulting alterations in biological systems critical in supporting healthy fetal development. Filling these gaps could lead to the design and deployment of public health initiatives that communicate the importance of reduction of stress and improved stress regulation during childhood, adolescence and early adulthood to support later maternal and offspring health. Consequently, the proposed research is uniquely poised to elucidate the type and timing of biobehavioral targets for early preventive interventions.