The in utero environment

Sex hormones: Sex is defined by an individual’s karyotype, which can be either XY (male) or XX (female). This chromosomal difference leads to differentiation of gonads into testes or ovaries, and associated changes in the hormonal milieu of the in utero environment. Human studies have reported a surge of testosterone from the fetal testes, at around the 6th week of pregnancy (Hughes, 2001), peaking between 12 and 18 weeks gestation (Finegan et al., 1989). This results in levels of testosterone at term that are 10 times higher than in females (de Zegher et al., 1992). In males, a second testosterone surge also occurs in the first 3 months after birth (Quigley, 2002). Recent studies suggest that the early testosterone peak(s) plays a significant role in subsequent sex differences in behavioral phenotype and may predispose to sex differences in vulnerability to specific disorders, such as autism (Phoenix et al., 1959).

The in utero effects of testosterone exposure on neurotypical human brain development have been investigated in several ways. Studies in females with congenital adrenal hyperplasia (CAH), for example, have reported increased male-typical play behavior in childhood (Hines et al., 2004) and more aggression in adolescence and adulthood (Berenbaum and Resnick, 1997). Increased aggression has also been reported in girls of opposite, compared to same, sex twins (Cohen-Bendahan et al., 2005). In both cases, this has been attributed to increased testosterone exposure prenatally. A further approach has been to examine directly the relationship between fetal testosterone levels measured at amniocentesis and later behavior. Using this technique, studies have reported a relationship between fetal testosterone and sexually dimorphic behaviors. This has included a positive correlation between fetal testosterone and restricted interests at 48 months old (Knickmeyer et al., 2005) and mental rotation at age 7 (Grimshaw et al., 1995); and a negative correlation with vocabulary level at 24 months old (Lutchmaya et al., 2002a), eye contact at 12 months old (Lutchmaya et al., 2002b), quality of social relationships at 48 months old (Knickmeyer et al., 2005), and empathy at 6–8 years old (Chapman et al., 2006). Fetal testosterone levels have also been found to be positively associated with number of autistic traits (Auyeung et al., 2009, 2012). However, one of the main criticisms when interpreting the results of these studies is the potential confounding effect of the post-natal environment. Future studies are needed to analyze these putative effects closer to delivery. Also, brain-imaging studies are needed to clarify the biological basis of these effects. These suggestions are also pertinent when considering studies into the effects of stress hormones on infant brain development.

Stress hormones: There is increasing evidence that prenatal maternal stress can have significant effects on specific aspects of the development of offspring (for review see Glover, 2011). Studies in rats, for example, have reported that maternal prenatal stress is associated with impaired cognitive and behavioral functioning in their progeny (Weinstock, 2001). Also, in nonhuman primates there is an association between prenatal stress and shorter attention span and irritability in their young (Schneider and Coe, 1993; Schneider et al., 1992; Schneider et al., 2002).

More recently, several prospective studies in humans have also reported that acute and chronic prenatal stress, including anxiety, depression and stressful life events, are associated with a significant increase in childhood emotional, behavioral and cognitive problems (Van den Bergh et al., 2005; Talge et al., 2007a; Austin et al., 2005; Mohler et al., 2006; Davis et al., 2004; Dieter et al., 2001). However, the longer-term effects of severe maternal antenatal stress on human psychopathology such as serious neurodevelopmental disorders (Talge et al., 2007b), including autism (Kinney et al., 2008b; Kinney et al., 2008a), attention deficit hyperactivity disorder (ADHD) (Van den Bergh and Marcoen, 2004; Wadhwa, 1998; O’Connor et al., 2003; Huizink et al., 2007) and conduct disorder (O’Connor et al., 2003; Huizink et al., 2007) is mixed (for review see Glover et al., 2009b). In the Avon Longitudinal Study of Parents and Children (n=7,448), the level of ADHD symptoms doubled from 5% to 10% in children born to mothers who had been most anxious during pregnancy (the top 15%) (O’Connor et al., 2002); however, the exposure variable was a cross-sectional self-report measure on one occasion antenatally at 18 weeks. Similarly, in the South London Child Development study, the rates of any psychopathology in adolescent offspring born from mothers who were depressed during pregnancy doubled from around 24% to 50% (Pawlby et al., 2011). In contrast, Class et al. (2011) examined a prospective population sample of over 3 million births and measured individual-level severe antenatal maternal stress. They found no increase in rates of any psychosis or affective disorder in offspring, but did find a moderately raised risk of offspring ADHD and autism spectrum disorder. Abel et al. (2014) also reported an increase in any psychopathology postnatally following severe antenatal maternal stress. However, when they looked at postnatal events, severe maternal stress or suicide was associated with an increase in risk of later affective psychosis. Clearly prenatal stress does not affect all children, most of whom appear resilient to antenatal maternal psychological effects; there may be significant gene-environment interactions alongside such resilience mechanisms (Kim-Cohen et al., 2006).^, Nevertheless, evidence is more robust in respect of maternal prenatal stress directly affecting children’s subsequent cognitive development, although the biological basis to this is poorly understood.

A range of mechanisms has been invoked to explain such associations including epigenetic, immune and hormonal. Here, we confine ourselves to consideration of a hormonal mechanism. Genetic and epigenetic considerations are covered in Chapter 5. It has been suggested that the effect of maternal stress on the fetus is partly mediated by the transplacental transport of maternal stress hormones, particularly cortisol (Huizink et al., 2004). This suggestion is supported by some animal and human studies (Trautman et al., 1995), which report a significant association between prenatal maternal exposure to synthetic dexamethasone and increased anxiety and reduced memory and socialization of their children (Rayburn et al., 1997). The fetus is usually protected from maternal cortisol by a placental enzyme, 11ß-hydroxy steroid dehydrogenase Type II (11ß-HSD2), which converts maternal cortisol to inactive cortisone (Mairesse et al., 2007). However, recent studies in mice (Holmes et al., 2006), rats (Mairesse et al., 2007) and humans (Glover et al., 2009a) suggest that maternal stress may cause downregulation of placental 11ß-HSD2, potentially allowing greater placental permeability to maternal cortisol, as well as other potentially relevant maternal factors such as cytokines and macrophages.

Animal studies suggest that maternal stress hormones, such as cortisol, may affect subsequent child behavior by modulating development of the limbic (social) brain system. Rodent and monkey offspring of prenatally stressed mothers, for example, have significant differences in the anatomy (Salm et al., 2004) and the function (Coe et al., 2003b) of limbic/paralimbic (including amygdala) and medial prefrontal cortical regions, which are densely populated with glucocorticoid receptors. Further, exogenous administration of the stress hormone corticosterone to pregnant rats and nonhuman primates significantly affects development of their infants’ limbic system (Jutapakdeegul et al., 2010; Coe et al., 2003a). These findings are supported by a study in humans which reported an association between increased maternal prenatal cortisol levels and increased amygdala volume in later childhood (Buss et al., 2012) and reduced volume in prefrontal cortical brain regions, including orbitofrontal cortex (Buss et al., 2010). However, brain regions do not function in isolation – they form networks connected by white matter tracts. Two of the key paralimbic brain regions that are affected by prenatal maternal stress, the amygdala and prefrontal cortex, are connected by the uncinate fasciculus (UF). A recent study examined the effects of prenatal stress on this tract using DT-MRI (diffusion tensor magnetic resonance imaging) tractography; the authors found a significant correlation between self-reported prenatal maternal stress exposure and proxy measures of microstructural integrity of the right UF in children at age 6–9 (Sagari et al., 2014). These findings suggest that, in addition to prenatal stress modifying the neurodevelopment of the amygdala and prefrontal cortex, it also modifies the white matter that connects these regions (i.e., UF).

These studies are of great interest, but do not take into account the postnatal environment into which children exposed to maternal stress have been born. It is increasingly recognized that family-wide influences throughout childhood also play an important role in the cognitive and behavioral development of children. Thus, early prenatal maternal effects must interact with factors such as social class, maternal social supports, wealth and deprivation, family size and structure, as well as the quality of maternal caregiving (see Chapter 4). Following, we consider the particular experience of childhood trauma and its potential relationship to neuroendocrine development in children.

Trauma during childhood and neuroendocrine development

Cumulative prevalence rates for physical abuse in children have been estimated to range from 5–35%, with rates of 4–9% for severe emotional abuse, 15–30% for sexual abuse in girls (5–15% in boys), and 6–12% for neglect (Bick et al., 2012). Apart from the social relevance, this phenomenon has important biological implications: early childhood trauma has been reported to cast a “long shadow” effect on subsequent hormonal changes and risk of psychopathology, particularly for mood and personality disorders (Bick et al., 2012; Heim et al., 2008; Duman, 2009). Cardiovascular, autoimmune, mental and neurodegenerative disorders are all reported to be more common among adults who report childhood abuse (Bick et al., 2012).

Biological mechanisms purported to explain these findings include hyperactivity of the hypothalamic pituitary adrenal (HPA) axis (Bick et al., 2012; Taylor et al., 2006; Cicchetti et al., 2011; De Bellis et al., 1994), and elevated immune system activity (Bick et al., 2012; Shirtcliff et al., 2009; Danese et al., 2007; Danese et al., 2008). These effects have consistently been reported in individuals reporting early trauma, abuse and neglect. Dysregulation of both the stress and immune axes are considered to be associated with risk of psychopathology and ill physical health in adults. Early life trauma may be particularly important as some suggest it may be more likely to have long-term neuroendocrine consequences because of the rapid neuroendocrine development that occurs in this period of life (Shirtcliff et al., 2009). Other psychological factors, such as excessive threat vigilance, poor social relationships, impaired self-regulation and unhealthy lifestyle (Miller et al., 2011) may also be important associations between childhood trauma and later psychopathology.

The reproductive years

Depression: It is important to emphasize that clinical and psychosocial risk factors probably play a more significant role than hormonal fluctuations in most women who suffer depression during the reproductive years. In particular, poor social supports, problematic or violent partner relationships, past poor maternal care (see Chapter 4) and childhood trauma, as well as other family history of psychiatric illness are important risk factors. However, increasing evidence suggests that fluctuations in the concentration of sex hormones across the reproductive years can also significantly influence risk of depression in some women. It has been proposed that this vulnerability represents a discrete nosological entity, sometimes referred to as “reproductive depression” (Payne et al., 2009).

Support for the concept of a reproductive depression includes the observation that, after controlling for relevant risk factors, and comparing with appropriate control groups, the incidence of unipolar depressive episodes doubles (Eberhard-Gran et al., 2002) and the rate of inpatient admissions triples in the first 3 months postpartum (Munk-Olsen et al., 2006; see Chapter 10). Furthermore, women with vulnerability to depressed mood associated with one period of hormonal fluctuation (e.g., premenstrual dysphoric disorder [PMDD]) are particularly vulnerable to depression at other times of hormonal fluctuation (e.g., postnatal and perimenopausal depression) (Bloch et al., 2000; Rubinow et al., 1998; Stewart and Boydell, 1993; Payne et al., 2007).

Additional support for the concept of reproductive depression is derived from family (Widholm and Kantero, 1971) and twin studies (Dalton et al., 1987; Kendler et al., 1992; Condon, 1993), which have reported that the heritability of PMDD is greater than 50% (Kendler et al., 1998) and that PMDD is associated with significant differences in single nucleotide polymorphisms modulating the alpha estrogen receptor (Huo et al., 2007). Similarly, unipolar, nonpsychotic, postnatal depression has been reported to have a heritability of 40% (Treloar et al., 1999) and a significant familial aggregation (particularly when it is narrowly defined as developing within 6 to 8 weeks post delivery) (Forty et al., 2006).

Cooper et al. (1995) describe a significantly greater risk of subsequent postnatal mood disorder in women whose incident mood disorder was also postnatal than in women who developed their first episode of depression outside of the peripartum. These data converge to suggest that some are exquisitely sensitive to “normal” reproductive hormonal fluctuation and may explain why early literature failed to find a link between hormonal levels in women and risk of illness.

In vulnerable women, iatrogenic induction of acute hormonal fluctuation has also been reported to trigger a recurrence of symptoms. For example, using a hormonal regime that involved acute suppression of ovarian function, then administration of high doses of estrogen and progesterone, followed by acute sex hormone withdrawal (i.e., modeling postnatal hormonal changes), a recurrence of depressive symptoms was triggered in women with a history of postnatal depression (Bloch et al., 2000). Importantly, these mood changes were not found in women without a prior history of postnatal depression, further supporting the notion that acute hormonal fluctuation may trigger depressive symptoms in only a specific subgroup of women.

In summary, the aforementioned studies support the suggestion that the concentration of sex hormones across the reproductive years can significantly influence depression in some women. Studies also suggest that fluctuations in sex hormones can act as significant triggers for schizophrenia and bipolar disorder in some women.

Schizophrenia and bipolar disorder: Women with schizophrenia have an increased vulnerability to relapse postpartum and around the time of menopause (Craig et al., 2005). Those with bipolar disorder have also been estimated to have greater than a 1 in 4 risk of relapse during the postpartum period (Jones and Craddock, 2001), particularly during the first 2 weeks (Heron et al., 2007). This risk doubles in those with a past history of severe postpartum psychosis or in bipolar women with a family history of postpartum psychosis (Jones and Craddock, 2001; Robertson et al., 2005). Familial (probably genetic) factors have been implicated in this postpartum susceptibility (Jones and Craddock, 2001) and linkage studies have reported the location of possible susceptibility genes (Jones and Craddock, 2007).

It has been suggested that this risk is modulated by estrogen and that women are protected from psychotic illness when estrogen levels are relatively high. This hypothesis has received support from a recent RCT of 100 acutely psychotic women. Those randomized to transdermal 17β-estradiol, combined with antipsychotic medication, reported significantly reduced positive and general psychopathological symptoms after 28 days, compared with women prescribed antipsychotic medication alone (Kulkarni et al., 2008). It may be that transdermal estrogen increased the bioavailability (i.e., plasma levels) of antipsychotic medication in treated women. Results of this pilot have yet to be replicated and it is not recommended that women or men with residual psychotic symptoms are treated with adjunctive hormonal therapy unless there are indications of sex hormone deficiency or a lack of ovulation.

In summary, sex hormones, particularly estrogen, may play a role in the expression of schizophrenia and other major mental illnesses during reproductively active years. The reader is also referred to Chapters 22 and 23.

The post-reproductive years

Memory: One of the most significant factors to have driven recent research into the effects of sex hormones on brain were early studies reporting that estrogen replacement therapy (ERT) reduced the risk of cognitive impairment and Alzheimer’s Disease (AD) in aging postmenopausal women. However, the direction of this risk may be critically dependent on a woman’s age on starting ERT and the length of time postmenopause for which ERT is prescribed. It has been reported that ovariectomy in younger women (i.e., prior to menopause) significantly increases the risk for development of cognitive impairment and dementia in later life (Nee and Lippa, 1999; Rocca et al., 2007). This risk disappeared if women were prescribed ERT until at least 50 years of age (Rocca et al., 2007). The suggestion of a “critical period” during which ERT may be protective is consistent with many prospective observational studies that have reported that ERT reduced the risk of AD by up to 50%.

Conversely, a large multi-center randomized controlled trial, the Women’s Health Initiative Memory Study (WHIMS), reported an increased risk of dementia in women over 65 years old randomized to treatment with estrogen (and medroxyprogesterone acetate [MPA]) versus placebo (Shumaker et al., 2003). In other words, in the WHIMS, women were only eligible to start ERT from the age of 65 if they had not received it before. The effects on cognitive risk were only significant in women randomized to conjugated estrogen (CEO) plus MPA rather than to CEO alone, suggesting that MPA (not CEO) may have negative effects on cognition. Also, both arms of the study were underpowered to analyze the risk of AD. In addition, it has been noted that during the period this study was undertaken (1990s and early 2000s), most women were approaching primary care for ERT to treat menopausal symptoms at much younger ages.

Mood: Longitudinal studies have reported that the perimenopause is associated with an increased risk of depression compared to both the premenopausal and postmenopausal periods (Bromberger et al., 2001; Freeman et al., 2004; Schmidt et al., 2004; Maartens et al., 2002; Chim et al., 2003). These findings have been replicated across a variety of cultures, and in both rural and urban settings (Jablensky et al., 1992; Castle and Murray, 1993). The precise role of sex hormones in the etiology of symptoms at this time remains unclear. However, reports into the efficacy of estradiol for the treatment of depressive disorders in perimenopausal women (Soares et al., 2001) indicates that further research is still needed.