Stress, glucocorticoids and liquorice in human pregnancy: Programmers of the offspring brain

Abstract

A suboptimal prenatal environment may induce permanent changes in cells, organs and physiology that alter social, emotional and cognitive functioning, and increase the risk of cardiometabolic and mental disorders in subsequent life (“developmental programming”). Although animal studies have provided a wealth of data on programming and its mechanisms, including on the role of stress and its glucocorticoid mediators, empirical evidence of these mechanisms in humans is still scanty. We review the existing human evidence on the effects of prenatal maternal stress, anxiety and depression, glucocorticoids and intake of liquorice (which inhibits the placental barrier to maternal glucocorticoids) on offspring developmental outcomes including, for instance, alterations in psychophysiological and neurocognitive functioning and mental health. This work lays the foundations for biomarker discovery and affords opportunities for prevention and interventions to ameliorate adverse outcomes in humans.

Keywords::

• Child
• glucocorticoids
• human
• liquorice
• mood
• offspring

Introduction

Mounting epidemiological evidence suggests that smaller body size at birth and/or a shorter length of gestation increase the risk of poorer physical health later in life (Barker et al. 1989; Eriksson et al. 2006, 2007). The brain is particularly affected, manifesting as altered cognitive and affective functioning and as an increased risk of mental health disorders. These associations may not be linear, as children born at the upper end of the birth weight distribution may also be at a higher risk for poorer health later in life (Gunnell et al. 2003). These findings can be best understood within the framework of prenatal “programming” or developmental origins of health and disease (DOHaD; Barker 1998). According to the DOHaD paradigm, adverse environmental experiences during critical periods of early development can permanently alter or programme the structure and function of cells, organs and physiological systems. Recent research has shown that the influences of programming may not be limited to the prenatal period, but may extend across the entire human growth period from preconception to adulthood (Barker et al. 2005). The effects of programming are not limited to development in childhood, but are influential across the life span (Barker et al. 1989; Eriksson et al. 2006, 2007; Räikkönen et al. 2007, 2008a, 2009a).

Recently, much interest has been focused on whether developmental programming—as reflected in alterations in body size at birth and length of gestation—plays a role in determining individual differences in cognitive, social and emotional functioning, including characteristic traits of temperament and personality, in neuroendocrine and autonomic nervous system (ANS) functioning and in mental health. A series of studies have shown that smaller body size at birth and/or shorter length of gestation associate, for instance, with alterations in hypothalamic–pituitary–adrenocortical (HPA) axis and ANS functioning during daily living and under psychosocial stress (Reynolds et al. 2001; IJzerman et al. 2003; Wust et al. 2005; Jones et al. 2006, 2007; Phillips and Jones 2006; Feldt et al. 2007; Kajantie et al. 2007) and with increased risks of depression (Räikkönen et al. 2007, 2008b), suicide (Barker et al. 1995), schizophrenia (Wahlbeck et al. 2001), personality disorders (Lahti et al. 2010b), schizotypal personality traits (Lahti et al. 2009), hostility (Räikkönen et al. 2008a), behavioural symptoms of attention deficit/hyperactivity disorder (Lahti et al. 2006; Strang-Karlsson et al. 2008b; Heinonen et al. 2010), poorer cognitive functioning (Gale et al. 2004; Heinonen et al. 2008; Broekman et al. 2009; Räikkönen et al. 2009a) internalizing and externalizing behaviour problems (Schlotz et al. 2008), poorer sleep (Strang-Karlsson et al. 2008a; Pesonen et al. 2009) and negative and positive affectivity characteristics of temperament and personality (Pesonen et al. 2006, 2008a, Schlotz et al. 2008, Schmidt et al. 2008; Lahti et al. 2010a). These associations may not be linear as children born with higher birth weights may display higher levels of anxiety (Lahti et al. 2010a) and distress (Cheung 2002), and be at an increased risk for psychosis later in life (Gunnell et al. 2003). The extent to which the associations with higher birth weight reflect maternal obesity, poorer health and pregnancy disorders, such as maternal diabetes/glucose intolerance, is not clear. In addition to the original empirical studies, several papers exist that have reviewed the evidence linking smaller body size at birth and/or shorter length of gestation with psychological and psychophysiological functioning and mental health later in life (Shenkin et al. 2004; Gluckman et al. 2005, 2008, 2009; Seckl and Meaney 2006; Seckl and Holmes 2007; Hanson and Gluckman 2008; Kajantie 2008; Seckl 2008; Räikkönen and Pesonen 2009; Kajantie and Räikkönen 2010).

Despite the increasing evidence pointing to the importance of early life programming in inducing alterations in psychological and psychophysiological functioning and mental health, the mechanisms through which these prenatal influences operate remain largely unknown. Two major hypotheses have been advanced to explain the link between events in utero and the later risk of neuropsychiatric and cardiometabolic disorders: maternal malnutrition and foetal overexposure to glucocorticoid stress hormones (Barker 1991; Edwards et al. 1993). These notions have been extensively explored in preclinical studies. In a range of experimental species, maternal malnutrition (global undernutrition or selective protein deficiency) reduces birth weight and reliably leads to higher blood pressure, glucose levels, altered behaviour and HPA axis function in adult offspring (Warner and Ozanne 2010). Similarly, maternal stress or administration of glucocorticoids, such as dexamethasone or betamethasone that freely crosses the placenta, also reduces birth weight and produces a similar or identical phenotype in the adult offspring (Seckl and Holmes 2007). Similar effects occur in non-human primates exposed to glucocorticoids in the last half of gestation (de Vries et al. 2007). The two hypotheses may be linked. Thus, circulating levels of physiological glucocorticoids (cortisol and corticosterone) are much higher in the maternal than in the foetal blood. This gradient is ensured by a placental enzyme, 11β-hydroxysteroid dehydrogenase type 2 (11β-HSD2), that catalyzes the rapid inactivation of glucocorticoids to their inert 11-keto forms (cortisone, 11-dehydrocorticosterone), thus forming a physiological “barrier” to maternal glucocorticoids. Liquorice contains compounds, such as glycyrrhizic acid and glycyrrhetinic acid, that potently inhibit 11β-HSDs. Maternal administration of these agents or the derived drug carbenoxolone (Lindsay et al. 1996a,b) or genetic knock out of the enzyme (Holmes et al. 2006) reduces birth weight and generates similar programmed CNS and peripheral outcomes in the adult offspring. Moreover, maternal protein malnutrition selectively lowers placental 11β-HSD2 (Langley-Evans et al. 1996), affording a possible link between the proposed mechanisms. Maternal stress in rodents also downregulates placental 11β-HSD2 (Mairesse et al. 2007), perhaps delimited by genotype (Lucassen et al. 2009), suggesting a double hit of elevated maternal glucocorticoids and reduced placental barrier function.

Despite this increasingly coherent body of evidence in model organisms, understanding the key mechanisms of programming in humans has been much less explored. Figure 1 outlines the mechanisms that are potentially involved. A series of human studies have demonstrated that prenatal environmental adversities, such as exposure to malnutrition (Roseboom et al. 2001), maternal pre-pregnancy overweight (Rodriguez 2010), pregnancy disorders, such as hypertension (Tuovinen et al. 2010, 2011), tobacco (Cornelius and Day 2009; Espy et al. 2011; Heinonen et al. 2011), cannabis (El Marroun et al. 2009) and alcohol exposure (Hellemans et al. 2010), have the potential to induce alterations in the offspring growth, birth anthropometry, development and adult functions. In this paper, we concentrate on reviewing the existing human evidence of programming induced by prenatal maternal stress, depression and anxiety, maternal glucocorticoid therapy and intake of liquorice during pregnancy.

Figure 1.  Mechanisms involved in the offspring programming. (Modified from Räikkönen et al. 2008b).

Exposure to prenatal maternal stress, depression and anxiety: Overview of results of studies, examining offspring behaviour with questionnaires, standardized observations, neurocognitive tasks, EEG, MRI and fMRI

Prenatal maternal stress, depression and anxiety may exert effects on offspring development through multiple pathways. These pathways, which include alterations, for instance, in maternal lifestyle, physiological stress-regulatory mechanisms and placental function, are outlined in Figure 1 and described in more detail in previous reviews (O'Regan et al. 2001; Seckl 2001, 2008; Seckl and Meaney 2004b; Van den Bergh et al. 2005b; Seckl and Holmes 2007).

In humans, prospective follow-up studies have provided accumulating evidence for an association between prenatal exposure to high levels of maternal stress, anxiety and depression and altered neurodevelopment in the offspring at stages ranging from birth up to 20 years of age. Importantly, these associations persisted after controlling for postnatal maternal mood and/or other potentially important pre- and postnatal confounders, such as alcohol consumption and smoking, during pregnancy, birth weight and socio-economic status, and in some studies also use of medication, such as antidepressants, have been taken into account. Results of the studies reviewed here are described in Table I. The studies are organized by the age of the offspring at the final follow-up assessment. In newborn babies, maternal prenatal stress was associated with neurodevelopmental alterations manifesting as less optimal scores on the Brazelton Neonatal Assessment Scale (Brouwers et al. 2001; Rieger et al. 2004; Hernandez-Reif et al. 2006) and on quantitative neurological examination (Lou et al. 1994). The newborn babies also displayed higher cardiac vagal tone, lower Apgar score (Ponirakis et al. 1998), less optimal behavioural states (Van den Bergh 1990, 1992) and higher cortisol levels 1 week postpartum (Diego et al. 2004). In the study of Harvison et al. (2009), neonates of mothers scoring lower on perinatal anxiety displayed more negative frontal slow wave amplitudes in response to their mother's voice compared to a female stranger's voice, while neonates of mothers scoring higher on perinatal anxiety showed the opposite pattern. The latter results indicate that maternal anxiety may induce neurophysiologically based differences in attentional allocation, processes known to be crucial to an infant's ability to maintain homoeostasis.

Table I.  Studies testing associations between prenatal maternal stress, anxiety and depression with offspring development.

Age of child	Measure or type of >prenatal exposure	Prenatal timing >weeks gestation	Outcome measure(s)	Reference(s)
2–3 days	Anxiety (state & trait); Stress (PSS)	20; 30	NBAS: habituation, social interaction, motor system, state organization, state regulation, autonomic system	Hernández-Reif et al. (2008)
3–5 days	Total distress	< 20; 30–34	Regulation problems (NBAS)	Rieger et al. (2004)
1 week	Perinatal depression	26	Cortisol level	Diego et al. (2004)
Neonate	Anxiety (general)	40^*	Auditory evoked responses (AER)	Harvison et al. (2009)
12 days	Depression	Not reported	Attentiveness (NBAS)	Hernández-Reif et al. (2006)
4–14 days	Life events stress	Mid gestation	Birth weight, head circumference, Prechtl's neurological observation	Lou et al. (1994)
Birth, 1 day, >3–4 weeks	Anxiety (state and trait)	≤ 16; 32–34	Medical records (e.g. Apgar scale), cardiac vagal tone (EKG Porges’ method)	Ponirakis et al. (1998)
6 weeks	Foetal reactivity to maternal stimulation (viewing of video)	32	Infant negative reactivity during a standardized lab procedure	DiPietro et al. (2008)
3 months	Depression	15–28; 29–40	Infant cortisol level pre and post infant stress test	Oberlander et al. (2008)
4 months	Anxiety (state) and depression	32	Behavioural reactivity (HIBRP)	Davis et al. (2004)
2, 6 months	Depression	Third trimester	Temperament ratings	McGrath et al. (2008)
3–5 months	Stress and depression	28–40	Face-to-face play interactions (CCTI)	Field et al. (1985)
3–6 months	Depressive symptoms, anxiety (pregnancy-specific), parenting stress; job strain	7–40	Excessive crying	van der Wal et al. (2007)
4–6 months	Anxiety (trait); depression, life events stress	Third trimester (32)	Difficult temperament (SITQ)	Austin et al. (2005)
6 months	Depression and anxiety	26	Infant cortisol level pre and post infant stress test	Brennan et al. (2008)
6 months	Stress experience	1st week postpartum	Temperament (IBQ); negative and overall temperamental reactivity	Pesonen et al. (2005)
6 months	Anxiety (pregnancy-specific; general)	12; 20	Temperament (IBQ-R); infant difficulties	Henrichs et al. (2009)
1 week, 12 weeks, >7 months	Anxiety (state and trait)	12–22; 23–31; 32–40	Prechtl's neurological observation, behavioural state observation, feeding score, mother–infant interaction (ITQ, ICQ, BSID)	Van den Bergh, (1990;1992)
7 months	Anxiety	36–39^*	Cortisol level of the child, pre and post still face procedure	Grant et al. (2009)
4–8 months	Anxiety (trait)	21; 26–34, 35	Temperament (ITQ-revised)	Vaughn et al. (1987)
8 months	Daily hassles	15–17, 27–28, 37–38	Mental development (BSID-MDI)	Huizink et al. (2002, 2003)
9 months	9/11	Not reported	Temperament (IBQ)	Brand et al. (2006)
9 months	9/11	28–40	Cortisol level of the child	Yehuda et al. (2005)
1.5 years	Life events stress (i.e. relationship problems)	14 months postpartum^*	Cognitive development (BSID-MDI), fear reactivity (Lab-TAB)	Bergman et al. (2007)
2 years	Anxiety (general: POMS and STAI), stress (DSI, PSS); depression (POMS, CES-D); pregnancy-specific stress (PES)	24, 28, 32	BSID: MDI, PDI; IBR	Dipietro et al. (2006)
3 weeks; 1 year; 2 years	Anxiety (state and trait)	32	Orientation (NBAS), Cognitive development (BSID-MDI), Task orientation and motor co-ordination (IBR)	Brouwers et al. (2001)
2 years	Perceived stress pregnancy-related anxiety	15–38	Behavioural problems (CBCL) temperament (ICQ), attention regulation (BSID)	Gutteling et al. (2005b)
13–15 and 42 months	Depression	Monthly, from >conception onwards	Frontal and parietal EEG asymmetry scores	Dawson and Ashman (2000); Dawson et al. (2001)
14–54 months	Anxiety, depression	Each trimester	Behavioural and emotional problems (CBCL); saliva cortisol after stressors	de Bruijn et al. (2009a,b)
6 months, 5 years	Psychological distress	1–16; 17–28; 29–40	Temperament (ITQ, PTQ)	Martin et al. (1999)
2 years	Ice storm	4–24	Cognitive development (BSID-MDI), language production (MCDI)	Laplante et al. (2004)
5.5 years		–	IQ (WISC)	Laplante et al. (2008)
6 months, 6 years	Pregnancy risks	16–20	Temperament (ITQ), school grades and marks	Niederhofer and Reiter (2004)
4–6 years	Daily hassles, anxiety, perceived stress, life events	16	Cortisol level of the child at first school day; after inoculation memory (TOMAL)	Gutteling et al. (2004, 2005a, 2006)
4 and 7 years	Anxiety	18, 32	Emotional and behavioural problems (SDQ)	O'Connor et al. (2002, 2003)
7–8 years	Perceived stress	10, 12, 20, 28, 32, 36	ADHD (DSM-IV)	Rodriguez and Bohlin (2005)
6–9 years	Anxiety (pregnancy specific)	19,25,31	Grey matter density (MRI study)	Buss et al. (2009)
10 years	Self rated anxiety	18, 32	Cortisol level of the child	O'Connor et al. (2005)
10 years →	Death of a close relative	0–12	Schizophrenia (ICD8/ICD10)	Khashan et al. (2008)
14 years	Chernobyl	14 →	Depression/MDD (C-SSAGA-A: DSM-III-R) ADHD	Huizink et al. (2007)
8–9 years, 14–15,>17 years	Anxiety (state and trait)	12–22	ADHD (CBCL, TRF), computerized encoding task and stop task, vocabulary and block design (WISC-R), sustained attention (CPT), depression (CDI), cued attention, N-back, Go/NoGo, dual task (ERP), response-shifting.	Van den Bergh and Marcoen (2004); >Van den Bergh et al. (2005, 2006, 2008), Mennes et al. (2006, 2009)
25 years	Severe life events	On average 25 years >postpartum^*	Cortisol and ACTH levels	Entringer et al. (2009)
–	Hurricane	24–40	Autism (DSM-II-R/DSM-IV)	Kinney et al. (2008)

Note: ADHD, attention deficit hyperactivity disorder; AER, Auditory Evoked Responses; BSID, Bayley Scales of Infant Development; CBCL, Child Behaviour Checklist; CCTI, Colorado Child Temperament Inventory; CDI, child depression inventory; CPT, continuous performance task; C-SSAGA-A-, child semi-structured assessment of genetics and alcoholism; DSI, Daily Stress Inventory; DSM-R, Diagnostic and Statistical Manual-revised; ECG, electrocardiogram; ERP, event-related potential; HIBRP, Harvard Infant Behavioural Reactivity Protocol; IBQ, Infant Behaviour Questionnaire; IBR, Infant Behaviour Record; ICD, International Classification of Diseases; ICQ, Infant Characteristics Questionnaire; ITQ, Infant Temperament Questionnaire; IQ, Intelligence Quotient; MCDI, MacArthur Communicative Development Inventory; MDD, Major Depressive Disorder; MDI, Mental Development Index; Lab-TAB, laboratory temperament assessment Battery; NBAS, Brazelton neonatal behavioural assessment scale; PES, Pregnancy Experience Scale; POMS, Profile of Moods Scale; PSS, Perceived Stress Scale; PDI, Psychomotor Development Index; PTQ, Preschool Temperament Questionnaire; SDQ, Strengths and Difficulties Questionnaire; SITQ: Short Infant Temperament Questionnaire; TOMAL, Test of Memory and Learning; TRF, Teacher's Report Form; WISC, Wechsler Intelligence Scale for Children; WPPSI, Wechsler Preschool and Primary Scale of Intelligence

* Retrospective.

Infants of mothers scoring higher on prenatal stress were rated by an observer as having poorer interactions with their mother (Field et al. 1985), being more irritable (DiPietro et al. 2008) and reactive (Davis et al. 2004, 2007), having problems with regulation of attention (Huizink et al. 2002, 2003) and having poorer language abilities and a lower IQ (Laplante et al. 2004, 2008). Their mothers rated them as having more sleeping, feeding and activity problems (Van den Bergh 1990, 1992), and being more irritable, difficult and showing more negative affectivity characteristics of temperament (Vaughn et al. 1987; Van den Bergh 1990, 1992; Huizink et al. 2002, 2003; Austin et al. 2005; Pesonen et al. 2005; McGrath et al. 2008; Henrichs et al. 2009) and crying more excessively (van der Wal et al. 2007). Scores on the Bayley Scales of Infant Development were worse at 8 and 24 months (Brouwers et al. 2001; Huizink et al. 2002, 2003; Laplante et al. 2004), but not in another study of 7-month-old infants (Van den Bergh 1990, 1992). Methodological differences, such as differences in sample size and variables that were used as covariates, may explain the conflicting findings between the studies using the Bayley Scales. Dawson and colleagues found that during mother–infant interaction, children of depressed mothers showed higher than normal heart rates and higher levels of cortisol, and reduced activity in brain regions that mediate positive approach behaviour. The authors indicate that there is suggestive evidence from their follow-up study that the postnatal experience with the mother may have had more effect on infant frontal EEG than on prenatal factors (Dawson and Ashman 2000; Dawson et al. 2001).

Pre-school children and children were rated by their mothers (Martin et al. 1999; O'Connor et al. 2002, 2003; Niederhofer and Reiter 2004; Van den Bergh and Marcoen 2004), teachers (Niederhofer and Reiter 2004; Rodriguez and Bohlin 2005), an external observer (Van den Bergh and Marcoen 2004) or themselves (Van den Bergh and Marcoen 2004) as showing poorer attention, hyperactivity, behavioural and emotional problems, and they were rated by their teacher as having lower school grades and problem behaviour (Niederhofer and Reiter 2004). O'Connor et al. (2002) found that the effects of antenatal anxiety were stronger than the effects of antenatal depression. Finally, the results of Obel et al. (2003)indicated that stressful life events in the mother during pregnancy increased the risk for ADHD problems in pre-adolescence between ages, 9–11 years.

Adolescents showed problems with cognitive control when performing computerized cognitive tasks measuring prefrontal cortex functioning and scored lower on intelligence subtests at the age of 14–15 and 17 years (Mennes et al. 2006; Van den Bergh et al. 2005a, 2006). Girls displayed higher levels of depressive symptoms (Van den Bergh et al. 2008). Mennes et al. (2009) showed that 17-year-old children of mothers who were more anxious between 12 and 22 weeks of gestation displayed changes in event-related potentials (ERP) measured with EEG, when performing a gambling task, requiring endogenous control. Functional magnetic resonance (fMRI) measures revealed that in the 20-year-old offspring, differences related to the level of antenatal maternal anxiety were found in the activation patterns of several important prefrontal regions (Mennes 2008). Buss et al. (2009) showed in a MRI study that high pregnancy-specific anxiety in mid gestation, but not later, is associated with decreased grey matter density in specific brain areas, in 8-year-old children.

Some evidence also suggests that the major life events inducing severe prenatal stress, such as death of a close relative during pregnancy, may be associated with stillbirth (Wisborg et al. 2008), low birth weight and smallness for gestational age status (Khashan et al. 2008b), cerebral palsy (Li et al. 2009b) and schizophrenia (Khashan et al. 2008a) in the offspring. One study has linked such events also with the risk of autism in the offspring (Kinney et al. 2008), but another study failed to replicate this association (Li et al. 2009a).

In humans, birth anthropometry associates with subsequent HPA axis function. Higher plasma and urinary glucocorticoid levels are found in children and adults who were of lower birth weight (Clark et al. 1996; Phillips et al. 1998). Cortisol responses to ACTH stimulation are exaggerated in those of low birth weight (Levitt et al. 2000; Reynolds et al. 2001, 2007). This occurs in myriad populations (Phillips et al. 2000; Kajantie and Räikkönen 2010) and precedes overt adult disease (Levitt et al. 2000). It is well known that psychological stress causes activation of the HPA axis with a consequent elevation in the levels of cortisol. Elevated levels of cortisol are also among the most consistently demonstrated biological abnormalities found in depression (Knorr et al. 2010), and in some contexts in anxiety (Roelofs et al. 2009). As maternal and foetal levels of cortisol are correlated (Gitau et al. 1998; O'Keane et al. 2011), the experience of stress and higher levels of depression and anxiety during pregnancy may directly result in higher foetal exposure to maternal cortisol. In a recent study, Sarkar et al. (2008) demonstrated that the correlation between amniotic fluid cortisol (reflecting foetal exposure) and maternal plasma cortisol becomes stronger the higher with levels of maternal prenatal anxiety, while maternal prenatal anxiety correlated with higher plasma cortisol levels of the mother, it was not directly related to amniotic fluid cortisol (Glover et al. 2009).

As a consequence of this biology, a number of studies have explored the effects of maternal psychopathology on offspring HPA axis function, mainly observing the effects by comparing the offspring of high vs. low anxiety, stressed or depressed pregnant women (Glover et al. 2010). Prenatal maternal anxiety and depression are in general associated with raised cortisol in the offspring, but infants of mothers exposed to the trauma of 9/11 who themselves developed symptoms of posttraumatic stress disorder (PTSD) had lower cortisol levels, though this is a common finding in PTSD (Yehuda et al. 2005). One study has provided evidence that an altered cortisol diurnal profile in the offspring associated with prenatal maternal anxiety was also associated with an altered behavioural phenotype, i.e. with depressed mood (Van den Bergh et al. 2008). Differences have been reported throughout the lifespan from infancy, ages 1 week to 9 months (Diego et al. 2004; Yehuda et al. 2005; Brennan et al. 2008; Oberlander et al. 2008; Grant et al. 2009), 4–15 years of age (Gutteling et al. 2004, 2005; O'Connor et al. 2005; Huizink et al. 2008; Van den Bergh et al. 2008) and in adults (Entringer et al. 2009). These data suggest that, as in animal models, the HPA axis is a prime target for prenatal influences and programming.

The studies on prenatal maternal stress, anxiety and depression, even when controlling for postnatal maternal stress, anxiety and depression, cannot overrule a possibility that a shared genetic basis may underlie the associations with offspring neurodevelopmental outcomes. Most evidence pointing to a hereditary component comes from the studies of depression (Belmaker and Agam 2008). It is, however, unlikely that individual differences in offspring neurodevelopmental outcomes could be attributed to a single genetic, epigenetic or pre- or postnatal environmental factor. Multiple mechanisms do exist (see Figure 1), and these may act in concert. Further studies on the exact underlying mechanisms through which maternal stress, depression and anxiety induce programming of the offspring are clearly warranted.

Maternal glucocorticoid therapy

In humans, therapeutic treatment with synthetic glucocorticoids is common in women at risk of preterm delivery, serving to accelerate foetal lung maturation and thus reduce neonatal morbidity and mortality (Seckl and Meaney 2004a). The long-term effects of this treatment are not fully determined, although there are suggestions that foetal growth and subsequent development may be impaired (for a review see Bolt et al. 2001). Glucocorticoid treatment during pregnancy typically reduces birth weight (for a review see Sloboda et al. 2005), but long-term follow up studies are few. Antenatal glucocorticoid administration has been linked with higher blood pressure in adolescence (Doyle et al. 2000) and higher insulin levels in adulthood (Dalziel et al. 2005). Studies aimed at establishing the long-term neurological and developmental effects of antenatal glucocorticoid therapy are complicated by the frequency of neurological sequelae common in such children anyway. However, in a group of 6-year-old children, antenatal glucocorticoid exposure is associated with subtle effects on neurological function, including reduced visual closure and visual memory (MacArthur et al. 1982). Multiple doses of antenatal glucocorticoids given to women at risk of preterm delivery reduced birth weight and head circumference in the offspring (French et al. 1999). There are also effects on behaviour: three or more courses of glucocorticoids associate with an increased risk of externalizing behaviour problems, distractibility and inattention (Yeh et al. 2004), and children exposed to a longer (>24 h to delivery) relative to a shorter duration ( < 24 h to delivery) of a single, repeat dose of betamethasone were rated by their mothers as more impulsive at the age of 3 years (Pesonen et al. 2009a). Children exposed to dexamethasone in early pregnancy because of risk of congenital adrenal hyperplasia and born at term showed increased emotionality, unsociability, avoidance and behavioural problems (Trautman et al. 1995). Postnatally, dexamethasone in premature babies lowers later IQ and other higher brain functions (Yeh et al. 2004). Overall, whilst far from complete, evidence indicates that exposure of the foetus (or premature neonate) to excess glucocorticoids that readily cross the placental and foetal tissue barriers to physiological cortisol (because they are poor substrates for 11β-HSD2) may induce alterations in the developing CNS.

Maternal intake of liquorice

If glucocorticoid exposure underpins some developmental effects on the foetus, what does 11β-HSD2 do? Rare children homozygous for deleterious mutations of the HSD11B2 gene encoding 11β-HSD2 are of substantially lower birth weight than their siblings, many of whom will be heterozygous for the mutations (Dave-Sharma et al. 1998). In humans, as in rodents (Benediktsson et al. 1993), placental 11β-HSD2 activity correlates with birth weight in some (Stewart et al. 1995), but not all studies (Rogerson et al. 1997). However, such correlations are weak and mechanisms can only be inferred.

A handful of recent studies have focused on prenatal maternal intake of liquorice confectionery as a natural experimental platform to address mechanisms of programming. These have been conducted in Finland where consumption of large quantities of liquorice confectionery is common among young women: nearly half of the pregnant women consume at least some liquorice (Räikkönen et al. 2009b; Strandberg et al. 2001). Glycyrrhizin (3β-d-diglucuronyl-18β-glycyrrhetinic or -glycyrrhizic acid) is a natural constituent of liquorice. Its hemisuccinate synthetic analogue, carbenoxolone, was a clinical drug formerly used to treat peptic ulcers. Its water solubility makes it attractive for use in preclinical and human studies (Welberg et al. 2000). These agents are potent (low nanomolar Ki) inhibitors of 11β-HSD2. Though they also potently inhibit 11β-HSD type 1, this is little expressed in the foetus at least in rodents (Speirs et al. 2004). In normal circumstances, this placental enzyme metabolizes up to 80–90% of maternal active cortisol to inactive cortisone, a function inhibited by carbenoxolone, at least in intact placenta ex vivo (Benediktsson et al. 1997).

In 1998, a project was initiated to test whether varying levels of maternal consumption of glycyrrhizin in liquorice during pregnancy was associated with body size at birth and length of gestation. In over 1000 pregnant women, prenatal exposure to high (>500 mg/wk) compared to zero-low (0–249 mg/wk) or moderate (250–499 mg/wk) glycyrrhizin in liquorice was associated with a slightly shorter duration of gestation (Strandberg et al. 2001). Glycyrrhizin intake during pregnancy was not significantly associated with birth anthropometry. The findings on shorter length of gestation were replicated in another Finnish cohort: high intake of glycyrrhizin during pregnancy was associated with over a twofold increased risk in the rate of preterm delivery (Strandberg et al. 2002).

In a follow-up study, the long-term effects of maternal intake of liquorice during gestation on 8-year-old offspring psychological development, mental health and psychophysiological functioning were determined. In comparison to the group whose mothers had zero-low glycyrrhizin exposure, those with high exposure scored significantly lower on verbal and visuo-spatial abilities and in narrative memory. In addition, children in the high-exposure group had significant 2.4–3.0-fold increased risk of externalizing symptoms, attention, rule breaking, aggression problems and DSM IV-based symptoms of obsessive defiant disorder (Räikkönen et al. 2009b). The effects on cognitive performance appeared dose-related. None of the associations were affected by birth weight or duration of gestation and other pre- and perinatal, maternal and child characteristics implicated as risks for pregnancy and/or cognitive and psychiatric outcomes.

The HPA axis of these children was also affected. Children of mothers consuming high levels of glycyrrhizin had a higher salivary cortisol peak and area under the curve upon awakening and higher overall salivary cortisol throughout a Trier Social Stress Test for Children (TSST-C; Räikkönen et al. 2010c). The associations with salivary cortisol peak and salivary cortisol baseline during the TSST-C were dose-related. As far as could be ascertained, the results were not due to confounding factors such as maternal health during pregnancy, including blood pressure levels, obesity and pregnancy disorders, birth anthropometry and length of gestation, maternal social class at birth and in a follow-up at the child's age of 8 years, maternal smoking and alcohol consumption during pregnancy, and child characteristics, such as sex, growth in height, head circumference and difficulties in cognitive functioning that might interfere with cognitive performance, associate with psychiatric symptomatology, and alter performance in the TSST-C. We have previously reported, in this same sample of 8-year-old children, that shorter sleep duration and lower sleep efficiency (percent time spent asleep whilst in bed) are associated with alterations in the diurnal salivary cortisol pattern and in salivary cortisol responses during the TSST-C (Räikkönen et al. 2010b). Poorer sleep may thus introduce another factor that may confound the associations between liquorice consumption and HPA axis activity. Maternal intake of liquorice during pregnancy is, however, not associated with the children's sleep patterns (Räikkönen, unpublished data). Thus, when we further controlled the associations of maternal liquorice intake during pregnancy and their children's HPA axis function for the children's sleep duration and sleep efficiency, the associations remained identical.

These findings suggest that high maternal liquorice consumption during pregnancy may exert deleterious effects upon cognitive and psychiatric outcomes and HPAA functioning in children paralleling the effects seen in preclinical models. Whilst it remains to be ascertained whether the dose of liquorice consumed affects placental 11β-HSD2 function appreciably in vivo (mice lacking one copy of the gene encoding 11β-HSD2 have intermediate reductions in birth weight), it is plausible to suggest that this may occur and that placental glucocorticoid “leakiness” may have an impact on foetal brain development during periods critical for affective and cognitive development with persisting impacts. If humans resemble rodents, then maternal stress-related disorders may further lower placental glucocorticoid barrier function.

Conclusion

This review shows, first, that accumulating evidence exists suggesting that maternal negative emotions during pregnancy are linked with the long-term behaviour and physiology of her child, even after controlling for relevant covariates. The range of maternal “stressors” that predict the child outcomes is quite wide. Moreover, a wide range of different outcomes have been found to be affected by these maternal stressors (Glover et al. 2010). This should come as no surprise. In terms of mechanisms, we are far from understanding how and when the early hormonal environment may affect the refinement of neural circuits in specific brain layers and areas which will later determine the way in which sensory-cognitive, motor, arousal and emotional structure–function relationships are affected (Van den Bergh et al. 2005a; Fox et al. 2010).

Second, this review shows that there are links between prenatal maternal intake of glucocorticoids and of liquorice confectionery and long-term behaviour and physiology of her child, even after controlling for relevant covariates. These findings may not generalize merely to intake of liquorice confectionery.

Because of its sweetening (50–200 times sweeter than refined sugar) and flavouring capacity, glycyrrhizin is also used as a natural sweetener and is also found in other foodstuffs, including some candies and chewing gum, herbal teas, alcoholic and non-alcoholic drinks, tobacco and some traditional (e.g. cough medicine) as well as some herbal medicine (to treat stomach ulcers, sore throat and viral infections). Hence, not surprisingly, according to some estimates the daily consumption levels of glycyrrhizin range from 1.6 to 215.2 mg (Isbrucker and Burdock 2006). This estimate was derived from the USA where consumption of liquorice confectionery is not popular. Glycyrrhizin is generally recognized as safe for use in foods, though the European Community's Scientific Committee on Food and the FAO/WHO Expert Committee (http://ec.europa.eu/food/fs/sc/scf/out186_en.pdf) have considered that a consumption of 100 mg/day may be a reasonable upper limit for the majority of the population, and the FDA recommends that if glycyrrhizin-containing foods are not consumed in excess or by sensitive individuals, these foods do not pose a health hazard (http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?fr = 184.1408). The findings reviewed here suggest that consumption of large quantities of glycyrrhizin in foodstuffs may alter offspring's brain development and subsequent function, and therefore, it may be prudent to inform pregnant women of the potential hazards of eating large quantities of liquorice.

Future studies should try to increase knowledge about mechanisms underlying the effect of prenatal influences on social-emotional, cognitive and behavioural problems, particularly about whether prenatal stress and gene-environment interactions affecting the HPA-axis play a role in early developmental processes. It has become increasingly clear that epigenetic changes, including DNA methylation and histone modifications, are causally involved in a range of developmental processes by affecting transcriptional activity. For the future, it is important that these epigenetic processes, that take place also in the placenta, are integrated in the studies of prenatal stress. They provide “a physical basis for the influence of the perinatal environmental signals over the life of the individual” (Meaney 2010). Moreover, the eventual moderating influence of other environmental factors, such as the postnatal caregiving environment, should also be studied. There is evidence that severe early life stress may carry consequences on a wide range of developmental outcomes, including physical and mental health, psychophysiological functioning and cognitive abilities and attained social class in adulthood (Pesonen et al. 2007, 2008b, 2010, 2011; Alastalo et al. 2009; Räikkönen et al. 2011). Whether the postnatal environmental influences buffer or add to the prenatal influences is little studied. Results of these research efforts would open the way for the development of targeted pre- and perinatal prevention and intervention strategies that could reduce the risk that prenatal environmental adversities carry for early functioning and later mental health of children, and have significant consequences for the long-term health and well-being of the children.

Declaration of interest: This work was sponsored by the grants from the European Science Foundation, Stress and Mental Health programme (EuroSTRESS), the Finnish Academy, the Medical Research Council (UK), The Netherlands Organisation for Scientific Research (NWO). The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.