exposure to postnatal depression predicts elevated cortisol in adolescent offspring

6
Exposure to Postnatal Depression Predicts Elevated Cortisol in Adolescent Offspring Sarah L. Halligan, Joe Herbert, Ian M. Goodyer, and Lynne Murray Background: Animal research shows that early adverse experience results in altered glucocorticoid levels in adulthood, either raised basal levels or accentuated responses to stress. If a similar phenomenon operates in humans, this suggests a biological mechanism whereby early adversity might transmit risk for major depression, glucocorticoid elevations being associated with the development of this disorder. Methods: We measured salivary cortisol at 8:00 AM and 8:00 PM over 10 days in 13-year-old adolescents who had (n 48) or had not (n 39) been exposed to postnatal maternal depression. Results: Maternal postnatal depression was associated with higher, more variable morning cortisol in offspring, a pattern previously found to predict major depression. Conclusions: Early adverse experiences might alter later steroid levels in humans. Because maternal depression confers added risk for depression to children, these alterations might provide a link between early events and later psychopathology. Key Words: Longitudinal, adolescents, cortisol, maternal depres- sion, postnatal depression M aternal depression has been associated with a variety of adverse outcomes in offspring, manifest as early as the neonatal period and through subsequent stages of development to adulthood (Goodman and Gotlib 2001; Murray and Cooper 2003). Ultimately, the offspring of depressed moth- ers are themselves at higher risk for major depression (Beardslee et al 1993; Nomura et al 2002; Weissman et al 1986); this observation has stimulated efforts to understand the biological mechanisms involved in the transmission of intergenerational risk. Individual differences in the activity of the hypothalamic– pituitary–adrenal axis (HPA) have recently been shown to be associated with differences in the risk for depression. Elevations in morning cortisol levels, particularly the occurrence of one or more very high morning cortisol values, predict the development of depressive disorder in those exposed to psychosocial risk factors for depression (Goodyer et al 2000; Harris et al 2000). The question, therefore, is whether early adversity can result in long-standing alterations in the HPA axis in humans. If this were the case, it might represent at least one process mediating the acquisition of risk during early life for later mental disorder. There is considerable experimental evidence that early expe- rience can alter later HPA activity. In general, early adversity increases either the basal secretion of glucocorticoids in adult life or the reactivity of the HPA axis to stress. In adult rats, this includes increased corticotrophin-releasing factor and glucocor- ticoid receptor gene expression in the hypothalamus and hip- pocampus, as well as enhanced behavioral response to stress in those separated from their mothers for varying periods during the postnatal period (Champagne and Meaney 2001; Francis et al 2002; Lehmann et al 2002; Weaver et al 2001). There are no comparable results for adult primates, though early maternal stress, resulting from the imposition of unpredictable foraging demands, has been associated with lasting corticotrophin-releas- ing factor elevations in the offspring of macaque monkeys (Coplan et al 2001); however, other paradigms examining early rearing conditions have produced divergent findings. For exam- ple, peer- versus mother-rearing has variously been found to have no effect on basal cortisol in infant rhesus monkeys (Winslow et al 2003) or to reduce both basal adrenocorticotro- phic hormone and cortisol levels (Clarke et al 1998). Marmoset monkeys exposed to maternal separations on postnatal days 2–28 showed reduced basal cortisol at 18 –20 weeks (Dettling et al 2002). In humans, some evidence for the role of early experience is emerging. Three-year-old children exposed to maternal depres- sion in the postnatal period had higher salivary cortisol levels than nonexposed infants (Hessl et al 1998), although this effect was not observed when the sample was reassessed at 7 years (Ashman et al 2002). Salivary cortisol was elevated in 4.5-year- old children whose mothers were concurrently depressed only if maternal depression had also been present in the first 12 months postpartum (Essex et al 2002). The question of whether there are longer-term associations between early exposure to maternal depression and cortisol levels in offspring has not previously been addressed. We measured salivary cortisol in adolescents who were or were not exposed to maternal postnatal depression (PND). Our aims were 1) to test the prediction that maternal PND is associated with elevated morning cortisol and more extreme morning values in adolescents; and 2) to examine whether any association between maternal PND and adolescent cortisol levels is mediated by other factors (e.g., life stresses) known to contribute to the association between maternal depression and child development. Methods and Materials Participants provided written informed consent before taking part in this Cambridgeshire Local Ethics Committee–approved study. Assessments were carried out by mental-health profes- sionals who were both trained and experienced in conducting clinical interviews. Assessors were supported by a clinical team, and diagnostic decisions were reviewed and confirmed through consensus conference. From the Winnicott Research Unit (SLH, LM), School of Psychology, Univer- sity of Reading; and the Department of Anatomy (JH) and Developmen- tal Psychiatry Section, Department of Psychiatry (IMG), University of Cambridge, United Kingdom. Address reprint requests to Dr. Sarah L. Halligan, University of Reading, School of Psychology, 3 Earley Gate, Whiteknights, Reading RG6 6AL, United Kingdom. Received July 9, 2003; revised September 25, 2003; accepted September 26, 2003. BIOL PSYCHIATRY 2004;55:376 –381 0006-3223/04/$30.00 doi:10.1016/j.biopsych.2003.09.013 © 2004 Society of Biological Psychiatry

Upload: sarah-l-halligan

Post on 24-Nov-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Exposure to Postnatal Depression Predicts ElevatedCortisol in Adolescent OffspringSarah L. Halligan, Joe Herbert, Ian M. Goodyer, and Lynne Murray

Background: Animal research shows that early adverse experience results in altered glucocorticoid levels in adulthood, either raisedbasal levels or accentuated responses to stress. If a similar phenomenon operates in humans, this suggests a biological mechanismwhereby early adversity might transmit risk for major depression, glucocorticoid elevations being associated with the development ofthis disorder.Methods: We measured salivary cortisol at 8:00 AM and 8:00 PM over 10 days in 13-year-old adolescents who had (n � 48) or hadnot (n � 39) been exposed to postnatal maternal depression.Results: Maternal postnatal depression was associated with higher, more variable morning cortisol in offspring, a pattern previouslyfound to predict major depression.Conclusions: Early adverse experiences might alter later steroid levels in humans. Because maternal depression confers added riskfor depression to children, these alterations might provide a link between early events and later psychopathology.

Key Words: Longitudinal, adolescents, cortisol, maternal depres-sion, postnatal depression

Maternal depression has been associated with a variety ofadverse outcomes in offspring, manifest as early as theneonatal period and through subsequent stages of

development to adulthood (Goodman and Gotlib 2001; Murrayand Cooper 2003). Ultimately, the offspring of depressed moth-ers are themselves at higher risk for major depression (Beardsleeet al 1993; Nomura et al 2002; Weissman et al 1986); thisobservation has stimulated efforts to understand the biologicalmechanisms involved in the transmission of intergenerationalrisk.

Individual differences in the activity of the hypothalamic–pituitary–adrenal axis (HPA) have recently been shown to beassociated with differences in the risk for depression. Elevationsin morning cortisol levels, particularly the occurrence of one ormore very high morning cortisol values, predict the developmentof depressive disorder in those exposed to psychosocial riskfactors for depression (Goodyer et al 2000; Harris et al 2000). Thequestion, therefore, is whether early adversity can result inlong-standing alterations in the HPA axis in humans. If this werethe case, it might represent at least one process mediating theacquisition of risk during early life for later mental disorder.

There is considerable experimental evidence that early expe-rience can alter later HPA activity. In general, early adversityincreases either the basal secretion of glucocorticoids in adult lifeor the reactivity of the HPA axis to stress. In adult rats, thisincludes increased corticotrophin-releasing factor and glucocor-ticoid receptor gene expression in the hypothalamus and hip-pocampus, as well as enhanced behavioral response to stress inthose separated from their mothers for varying periods duringthe postnatal period (Champagne and Meaney 2001; Francis et al

2002; Lehmann et al 2002; Weaver et al 2001). There are nocomparable results for adult primates, though early maternalstress, resulting from the imposition of unpredictable foragingdemands, has been associated with lasting corticotrophin-releas-ing factor elevations in the offspring of macaque monkeys(Coplan et al 2001); however, other paradigms examining earlyrearing conditions have produced divergent findings. For exam-ple, peer- versus mother-rearing has variously been found tohave no effect on basal cortisol in infant rhesus monkeys(Winslow et al 2003) or to reduce both basal adrenocorticotro-phic hormone and cortisol levels (Clarke et al 1998). Marmosetmonkeys exposed to maternal separations on postnatal days2–28 showed reduced basal cortisol at 18–20 weeks (Dettling etal 2002).

In humans, some evidence for the role of early experience isemerging. Three-year-old children exposed to maternal depres-sion in the postnatal period had higher salivary cortisol levelsthan nonexposed infants (Hessl et al 1998), although this effectwas not observed when the sample was reassessed at 7 years(Ashman et al 2002). Salivary cortisol was elevated in 4.5-year-old children whose mothers were concurrently depressed only ifmaternal depression had also been present in the first 12 monthspostpartum (Essex et al 2002). The question of whether there arelonger-term associations between early exposure to maternaldepression and cortisol levels in offspring has not previouslybeen addressed. We measured salivary cortisol in adolescentswho were or were not exposed to maternal postnatal depression(PND). Our aims were 1) to test the prediction that maternal PNDis associated with elevated morning cortisol and more extrememorning values in adolescents; and 2) to examine whether anyassociation between maternal PND and adolescent cortisol levelsis mediated by other factors (e.g., life stresses) known tocontribute to the association between maternal depression andchild development.

Methods and Materials

Participants provided written informed consent before takingpart in this Cambridgeshire Local Ethics Committee–approvedstudy. Assessments were carried out by mental-health profes-sionals who were both trained and experienced in conductingclinical interviews. Assessors were supported by a clinical team,and diagnostic decisions were reviewed and confirmed throughconsensus conference.

From the Winnicott Research Unit (SLH, LM), School of Psychology, Univer-sity of Reading; and the Department of Anatomy (JH) and Developmen-tal Psychiatry Section, Department of Psychiatry (IMG), University ofCambridge, United Kingdom.

Address reprint requests to Dr. Sarah L. Halligan, University of Reading,School of Psychology, 3 Earley Gate, Whiteknights, Reading RG6 6AL,United Kingdom.

Received July 9, 2003; revised September 25, 2003; accepted September 26,2003.

BIOL PSYCHIATRY 2004;55:376–3810006-3223/04/$30.00doi:10.1016/j.biopsych.2003.09.013 © 2004 Society of Biological Psychiatry

ParticipantsParticipants were part of a prospective, longitudinal study of

the development of children of postnatally depressed and wellwomen (Murray 1992). The sample was originally recruited at 2months postpartum, with further assessments when the childrenwere 18 months, 5 years, and 8 1/2 years old, in addition to thecurrent, 13-year follow-up. At initial recruitment, a communitysample of primiparous mothers of healthy, full-term infants wasscreened for PND by administration of the Edinburgh PostnatalDepression Scale (EPDS; Cox et al 1987) at 6 weeks postpartum.Women scoring greater than 12 on the EPDS were interviewed;61 women who met research diagnostic criteria (Spitzer et al1978) for depressive disorder were identified, 58 of whom wererecruited for the study. Forty-two nondepressed mothers ran-domly selected from the same postnatal population were alsorecruited.

Fifty-three PND-group families (91.4%) and 41 comparison-group families (97.6%) were retained at 13 years. Of these, fivePND-group and two control-group adolescents did not completecortisol collections, owing to refusal (n � 3), medical ineligibility(diabetes, n � 1), and scheduling problems (n � 3). The finalsample consisted of 48 PND and 39 comparison families.

Maternal MeasuresMental State. The presence and timing of maternal depres-

sion were assessed with the Structured Clinical Interview forDSM-IV (Spitzer et al 1995). Because maternal clinical interviewswere similarly conducted at 18-month, 5-year, and 8 1/2-yearassessments, detailed longitudinal information relating to mater-nal mental state was available and was used to establish 1) thepresence or absence of current maternal depression; and 2) theoverall duration of maternal depression, calculated as the totalnumber of months of depression over the offspring’s lifetime.

Current maternal psychological symptoms were assessed withthe 28-item version of the General Health Questionnaire (GHQ)(Goldberg and Williams 1988; Goldberg et al 1997). Few mothersmet diagnostic criteria for current depression at 13 years. There-fore, we used the GHQ depression subscale (GHQ-D) scores asour primary measure of current maternal depression to increasethe power of our analyses. Because scores on the GHQ-D werehighly skewed by a large number of zero responses, we dichot-omized scores based on the presence or absence of any depres-sive symptom (van Hemert et al 1995).

Parental Conflict. Mothers completed 10-point rating scaleson perceived and felt criticism as a measure of parental conflict(Hooley and Teasdale 1989).

Adolescent MeasuresMental State. Diagnostic interviews with adolescents were

conducted with the Kiddie Schedule for Affective Disorders andSchizophrenia, Present and Lifetime version (Kaufman et al1997). Participants completed the Mood and Feelings Question-naire, a validated self-report measure of current depressivesymptomatology (Angold et al 1995). Only a small number ofindividuals met diagnostic criteria for past or current depressivedisorder; therefore, we adopted Mood and Feelings Question-naire scores as the primary index of current depression.

Life Events. Adolescents completed an interview assessmentof the occurrence or undesirable life events and difficulties in thepreceding 12 months (Goodyer et al 2000). Few individualsreported more than one undesirable event; therefore, we codedthe presence or absence of any life event.

Puberty. In addition to height and weight, Tanner stage(Tanner 1966) was obtained by adolescent self-report guided bystandard line drawings (Netherton et al 2004).

Cortisol. Adolescents collected saliva samples at 8:00 AM and8:00 PM for 10 consecutive school days, following instructionssupplied for home completion. We measured cortisol withenzyme-linked immunosorbent assay on 20-�L samples of salivawithout extraction (antibody, Cambio, Cambridge, United King-dom). Intra-assay variation was 4.1%, and interassay variationwas 7.6%. There is a good correlation between salivary andplasma cortisol concentrations (Goodyer et al 1996), and salivarylevels represent approximately 5% of those in the blood.

Salivary cortisol is subject to significant day-to-day fluctua-tions. The use of 10 sampling days allowed us 1) to derive arelatively accurate index of mean cortisol for each individual; and2) to examine intraindividual variability over the 10-day samplingperiod, for both morning and evening saliva collections. We usedthe coefficient of variation ([variance/mean]/n) to index variabil-ity, to control for the fact that variance increased with meanlevels. Arcsine transformation of the resulting scores created anunbounded variable suitable for statistical tests. In addition, wecoded individuals according to whether they were “peak posi-tive” for 8:00 AM or 8:00 PM cortisol, that is, one or more of theircortisol concentrations fell above the 90th percentile for thesample mean (percentiles calculated separately by gender andcollection time) (Goodyer et al 2000; Harris et al 2000).

Data Analysis StrategyWe used simple group comparisons to examine the effects of

PND on cortisol measures. When group differences were ob-served, we investigated the role of potential mediators (parentalconflict, adolescent depressive symptoms, and adolescent lifeevents), using regression analyses with blocked entry proceduresto examine whether these factors explained associations be-tween adolescent salivary cortisol and maternal PND. We alsoexamined possible mediating effects of current maternal depres-sive symptoms and the overall duration of maternal depression ina separate set of within-group (PND vs. no PND) analyses, boththese factors being highly confounded with PND. We checkedcontinuous dependent variables for normality before carryingout parametric tests, and for each analysis we verified that all testassumptions were met. No transformation of cortisol variableswas required. Given the modest effect sizes anticipated inassociation with PND (a very distal predictor of offspring corti-sol), the relatively small sample size, and the presence of clearhypotheses, we adopted a p value of .05 as significant for thecurrent study.

Results

We assessed 87 mothers and their 13-year-old adolescentoffspring (44 girls, 43 boys). The sample comprised white,low-to-middle-class families, consistent with the Cambridge pop-ulation from which it was drawn. The PND and control groupswere comparable in terms of demographic characteristics (Table1), including age in months, gender distribution, socioeconomicstatus, parental marital status, height, weight, mean Tanner Stage,body mass index, and birth weight. Adolescents in the PNDgroup were less likely to have been breast-fed to 6 weeks thanthose in the control group (Cooper et al 1993). Consistent withprior reports, girls had higher mean 8:00 AM cortisol than boys[girls � 3.20 � 1.19 ng/mL, boys � 2.70 � 1.10 ng/mL; t(85) �2.07, p � .042], and level of pubertal development was positivelyassociated with mean 8:00 AM cortisol (r � .20, p � .035)

S.L. Halligan et al BIOL PSYCHIATRY 2004;55:376–381 377

www.elsevier.com/locate/biopsych

(Netherton et al 2004). We have controlled for gender, Tannerstage, and breast-feeding status in subsequent analyses.

CortisolGroup comparisons of cortisol data indicated significant

differences (Table 2). For 8:00 AM samples, the PND groupshowed higher mean scores and more variability in cortisol levelsacross the 10 collections compared with the control group. Whenwe examined maximum recorded 8:00 AM cortisol value over the10 days of assessment, calculated for each individual, meanmaximum concentration was significantly higher in the PNDversus the control group [PND � 5.71 � 2.12 ng/mL, control �4.48 � 1.71 ng/mL; t(85) � 2.94, p � .004], whereas there was nogroup difference in minimum cortisol concentrations [PND �

1.49 � .94 ng/mL, control � 1.33 � 0.70 ng/mL; t(85) � .92, ns].Thus, group differences in 8:00 AM mean cortisol and cortisolvariability were driven by differences in the occurrence ofcortisol values at the high, rather than the low, end of the range.Accordingly, a higher proportion of individuals in the PND thanin the control group were peak positive (i.e., one or more cortisolconcentrations greater than the 90th percentile) for morningcortisol (Table 2). There were no significant group differences in8:00 PM cortisol measures.

Multivariate AnalysesWe considered maternal marital/partner conflict, current ad-

olescent depressive symptoms, and adolescent life events in the12 months before interview as potential mediators betweenmaternal PND and morning cortisol in adolescents. Adolescentlife events were not associated with maternal PND, and we didnot examine them further; however, PND was associated withmore maternal partner conflict and higher adolescent depressivesymptoms, justifying further investigation of these measures(Table 1). We conducted linear regression analyses to examine8:00 AM cortisol mean and variability. In each instance, weentered child gender, Tanner stage, breast-feeding status, childdepressive symptoms, and maternal partner conflict into theregression equation in the first step, and examined the additionalcontribution of PND in a second step (Table 3). Maternal PNDsignificantly predicted variance in 8:00 AM cortisol measures overand above other factors. Pubertal development was significantly,independently, and positively associated with mean 8:00 AM

cortisol, and child depressive symptoms, gender, and breast-feeding status also showed effects at the trend level of signifi-cance (p � .10).

Intervening Maternal DepressionMothers with PND were significantly more likely to experi-

ence further depression in the following years than non-PND

Table 1. Sample Characteristics, Reported by Group

Control(n � 39)

PND(n � 48) Statistics

Maternal CharacteristicsSocial class I, II, and III nonmanual (%) 66.7 60.4 �2(1) � .36Percent separated from child’s father 12.8 22.9 �2(1) � 1.46Percent reporting current depressive symptoms (GHQ) 5.1 54.2 �2(1) � 23.7a

Total time depressed in course of study (months)b 2.74 (4.49) 22.46 (16.7) t(55.2) � �7.84a

Reported parental conflict (0 –10) 4.33 (2.28) 5.35 (2.11) t(85) � �2.15c

Child CharacteristicsProportion of boys (%) 51.3 47.9 �2(1) � .10Age at assessment (months) 160.4 (3.0) 160.0 (1.5) t(85) � .86Height (M) 1.62 (.08) 1.63 (.07) t(85) � �.49Weight (kg) 53.7 (11.4) 56.9 (10.8) t(85) � �1.33Body mass index 20.3 (3.6) 21.3 (3.7) t(85) � �1.32Tanner stage (1–5)d 3.3 (.9) 3.3 (.6) Z (n � 87) � �.26Birth weight (kg) 3.4 (.4) 3.5 (.4) t(85) � �1.11Proportion breast-fed to 6 weeks (%) 76.9 51.1 �2(1) � 6.10c

Lifetime depressive disorder (%) 7.7 22.9 �2(1) � 3.69e

Current depressive symptoms (MFQ) 13.9 (8.8) 17.9 (11.0) t(85) � �1.84e

Experienced life event in 12 months before interview (%) 59.0 64.6 �2(1) � .29

Data are presented as % or mean (SD).PND, postnatal depression; GHQ, General Health Questionnaire; MFQ, Mood and Feelings Questionnaire.ap � .001.bt test corrected for unequal variances.cp � .05.dMann–Whitney test used for nonparametric statistics.ep � .10.

Table 2. Group Differences in 10-Day Cortisol Measures

10-Day Salivary MeasuresControl(n � 39)

PND(n � 48) Statistic

8:00 AM CortisolMean cortisol (ng/mL) 2.68 (1.05) 3.17 (1.22) t(85) � 2.00a

Coefficient of variationb .045 (.028) .069 (.050) t(84) � �3.00c

Percent peak positiveindividuals

46.2 72.9 �2(1) � 6.47a

8:00 PM CortisolMean cortisol (ng/mL) .33 (.17) .37 (.35) t(85) � �.74Coefficient of variationb .017 (.027) .030 (.054) t(84) � �1.05Percent peak positive

individuals51.3 48.8 �2(1) � .26

Data are presented as % or mean (SD).PND, postnatal depression.ap � .05.bAnalyses carried out on arcsine transformed scores. Raw scores are

presented for clarity.cp � .01.

378 BIOL PSYCHIATRY 2004;55:376–381 S.L. Halligan et al

www.elsevier.com/locate/biopsych

mothers. Eighty-three percent of the PND-group mothers expe-rienced an episode of depression subsequent to the postnatalperiod, versus 33% of control-group mothers [�2(1) � 22.60, p �.0005]. Therefore, we considered overall duration of maternaldepression as a possible contributor to the association betweenmaternal PND and 8:00 AM cortisol measures in adolescents. Theextent to which overall duration of maternal depression wasconfounded with the occurrence of PND precluded controllingfor this factor across groups. Therefore, we used correlationalanalyses to examine associations between duration of maternaldepression and adolescent cortisol measures separately for thePND and control groups. Number of months of maternal depres-sion showed no association with offspring 8:00 AM cortisolmeasures, within either the PND group (mean cortisol r � �.10,variability r � .11, peak positive r � �.05; all p � .10) or thecontrol group (for mean cortisol r � �.31, p � .06; variability r� �.10, ns; peak positive r � �.25, ns), although in the latterthere was a trend-level correlation with mean 8:00 AM cortisol, inthe opposite-to-predicted direction.

Mothers who became depressed postnatally also had moresymptoms of depression than control-group mothers at the timeof the current assessment (details in Table 1), which suggests thatconcurrent maternal symptoms could be contributing to highermorning cortisol levels in the PND-group adolescents; however,when we compared 8:00 AM cortisol levels for PND-groupadolescents who either were (n � 26) or were not (n � 22)exposed to concurrent maternal depressive symptoms, we foundno differences in mean cortisol [current maternal depression �3.30 � 1.27 ng/mL, no current depression � 3.03 � 1.17 ng/mL;t(46) � .72, ns], cortisol variability [mean � .071 � .054 ng/mLand .066 � .046 ng/mL, respectively; t(46) � .32, ns], or in theproportion of peak positive individuals [20 (76.9%) and 15(68.2%), respectively; �2(1) � .46, ns]. Only two mothers in thecontrol group reported current depressive symptoms, precludingan analysis of the effects of concurrent maternal depression inthe absence of PND.

Discussion

We show that maternal depression occurring in the postnatalperiod is associated with higher and more variable morningcortisol in adolescents 13 years later. Furthermore, current de-pressive symptoms in the mother or the adolescent, maternal

partner conflict, overall duration of maternal depression, and theexperience of undesirable life events by the adolescent, did notexplain the observed association between maternal PND andadolescent cortisol. To our knowledge, this is the first study todemonstrate a direct association between maternal PND andcortisol levels in adolescent offspring.

The current findings are striking in view of the differences incortisol that have been found to predict the development ofdepressive disorder. Two longitudinal studies examined salivarycortisol in adolescents and adult women with psychosocial riskfactors for the development of depression (Goodyer et al 2000;Harris et al 2000). For both cohorts, individuals who had highermean morning salivary cortisol levels, or one or more very highmorning cortisol values, during initial, repeated cortisol assess-ments, were more likely to develop depression over a 12-monthfollow-up period. Thus, perturbations in morning cortisol mightrepresent an antecedent risk factor for depressive disorder.

Research on cross-fostered rat pups has provided evidencefor nongenetic, intergenerational transmission of stress reactivityand HPA-axis activity via quality of maternal care (Francis et al1999). In humans, the availability of a sensitive caregiver servesto buffer HPA responses to environmental demand in thedeveloping infant (Gunnar 1990; Nachmias et al 1996), andimpairments in maternal responsiveness have been associatedwith accentuated cortisol reactivity in infants (Dawson andAshman 2000; Field 1994). Maternal PND has detrimental effectson care-giving behaviors (Murray et al 1993, 1996). Thus, acompelling interpretation of the current data is that elevatedcortisol in the PND group is an enduring consequence ofadversity in infancy resulting from impairments in mother–infantinteractions.

Experimental studies have emphasized the primacy of expe-riences occurring during early development in determining HPAaxis parameters (Anisman et al 1998; Coplan et al 2001). Innonhuman primates there seems to be a developmental windowbeyond which the effects of early social adversity can no longerbe reversed by a change in social environment (Harlow andSuomi 1971). This period corresponds to 24–30 weeks postpar-tum in humans, suggesting that early infancy might be a timeduring which developing neurobiological systems are uniquelysensitive to environmental influences. The current findings are inkeeping with this possibility. Maternal depression occurring inthe postnatal period was related to cortisol in adolescent off-

Table 3. Regression Analyses Examining the Prediction of 8:00 AM Cortisol Concentrations in Adolescents

Variables Entered

8:00 AM Cortisol Mean 8:00 AM Cortisol Variability

Step 2B SE (B)

Step 2�

Step 2B SE (B)

Step 2�

Step 1 R2 � .18: F(5,80) � 3.50, p � .01 R2 � .07: F(5,79) � 1.20, nsGender .43 .25 .18a .036 .068 .06Tanner stage .34 .16 .22b .039 .043 .08Adolescent depressive symptoms .02 .01 .19a .002 .003 .08Maternal dyadic conflict �.06 .05 �.11 .013 .015 .09Breast-fed to 6 weeks (yes/no) .49 .25 .20� .11 .068 .18a

Step 2 R2 � .04: F(1,79) � 4.27, p � .05 R2 � .09: F(1,78) � 8.04, p � .01Postnatal depression .52 .25 .22b .195 .069 .32c

Overall Model R2 � .22: F(6,79) � 3.75, p � .01 R2 � .16: F(6,78) � 2.43, p � .05

B, unstandardized coefficient; �, standardized coefficient.ap � .10.bp � .05.cp � .01.

S.L. Halligan et al BIOL PSYCHIATRY 2004;55:376–381 379

www.elsevier.com/locate/biopsych

spring, but measures of maternal depression occurring subse-quent to that time were not.

Although our interpretation of the current data emphasizesthe primacy of the early environment, there are genetic factorsregulating individual differences in levels of cortisol (Bartels et al2003; Meikle et al 1988). Nothing is known of potential geneticcontributions to the features of cortisol secretion or their re-sponses to environmental demand during adolescence, or ofwhether associations between adrenal steroids and the onset ofdepression are genetically mediated. At this stage, gene–envi-ronment interactions, to explain both variations in cortisol andthe onset of major depression, have yet to be specified. Similarly,intrauterine environment has been shown to have implicationsfor offspring HPA axis development (Maccari et al 2003). To theextent that postnatally depressed mothers also differ from non-depressed mothers in the prenatal period, intrauterine effectsmight also be confounded with those of early environment.

It could be argued that the reliance on salivary cortisol collec-tions limits the conclusions that can be drawn from the currentresearch. First, prior research with adults has indicated poor com-pliance with collection protocols (Kudielka et al 2003; Yehuda etal 2003). Our restriction of sampling to school days, the involve-ment of both parents and children in salivary collections, and thelong-term nature of the current study might have enhancedprotocol adherence. Nonetheless, deviations in collection timesmight have contributed to variability in morning and eveningcortisol levels. Second, morning cortisol was reported withoutreference to the interval between awakening and collection time(Pruessner et al 2003), and reported levels might not reflect theapex of early waking secretion. Third, although salivary cortisolprovides a good index of plasma cortisol (Goodyer et al 1996), itsrelationship to the central measures of HPA-axis activity thathave been studied in animal research is complex. Furtherresearch that uses biological challenge procedures or assess-ments of glucocorticoid receptor activity might elucidate thebiological processes underlying the observed cortisol alterations.

Whatever the exact nature of the factors determining themarked differences in morning cortisol levels observed in ado-lescents exposed to maternal depression early in life, it remainsto be seen how such differences might contribute to anysubsequent psychopathology. The role of the HPA axis in theetiology of depression is still being defined (Cowen 2002; Gold etal 2002), and emergent observations that elevations in morningcortisol secretion are related to the development of depressionrequire further study (Angold 2003). There are also broaderimplications to be considered. Elevated glucocorticoids increasethe susceptibility of the brain to adverse events (Gubba et al2000; Sapolsky 1985) and can have deleterious effects on hip-pocampal structure and function (Newcomer et al 1999; Sapolsky2000). It will be of considerable interest to examine whether theobserved alterations in cortisol secretion have implications foremotional and cognitive development, as well as for affectivedisorders.

This study was supported by the Tedworth Charitable Trustand a Medical Research Council Program Grant.

We thank Sheelah Seeley for her assistance with data collectionand Peter J. Cooper for helpful comments on the manuscript.

Angold A (2003): Adolescent depression, cortisol and DHEA. Psychol Med33:573–581.

Angold A, Costello EJ, Messer SC, Pickles A, Winder F, Silver D (1995): Thedevelopment of a short questionnaire for use in epidemiological studies

of depression in children and adolescents. Int J Methods Psychiatr Res5:237–249.

Anisman H, Zaharia MD, Meaney MJ, Merali Z (1998): Do early-life eventspermanently alter behavioral and hormonal responses to stressors? Int JDev Neurosci 16:149 –164.

Ashman SB, Dawson G, Panagiotides H, Yamada E, Wilkins CW (2002): Stresshormone levels of children of depressed mothers. Dev Psychopathol14:333–349.

Bartels M, Van den Berg M, Sluyter F, Boomsma DI, de Geus EJ (2003):Heritability of cortisol levels: Review and simultaneous analysis of twinstudies. Psychoneuroendocrinology 28:121–137.

Beardslee WR, Keller MB, Lavori PW, Staley J, Sacks N (1993): The impact ofparental affective disorder on depression in offspring: A longitudinalfollow-up in a nonreferred sample. J Am Acad Child Adolesc Psychiatry32:723–730.

Champagne F, Meaney MJ (2001): Like mother, like daughter: Evidence fornon-genomic transmission of parental behavior and stress responsivity.Prog Brain Res 133:287–302.

Clarke AS, Kraemer GW, Kupfer DJ (1998): Effects of rearing condition on HPAaxis response to fluoxetine and desipramine treatment over repeatedsocial separations in young rhesus monkeys. Psychiatry Res 79:91–104.

Cooper PJ, Murray L, Stein A (1993): Psychosocial factors associated with theearly termination of breast-feeding. J Psychosom Res 37:171–176.

Coplan JD, Smith ELP, Altemus M, Scharf BA, Owens MJ, Nemeroff CB, et al(2001): Variable foraging demand rearing: Sustained elevations in cister-nal cerebrospinal fluid corticotropin-releasing factor concentrations inadult primates. Biol Psychiatry 50:200 –204.

Cowen PJ (2002): Cortisol, serotonin and depression: All stressed out? Br JPsychiatry 180:99 –100.

Cox JL, Holden JM, Sagovsky R (1987): Detection of postnatal depression.Development of the 10-item Edinburgh Postnatal Depression Scale. Br JPsychiatry 150:782–786.

Dawson G, Ashman SB (2000): On the origins of a vulnerability to depression:The influence of the early social environment on the development ofpsychobiological systems related to risk for affective disorder. In NelsonCA, ed. The Effects of Adversity On Neurobehavioral Development. Mah-wah, NJ: Lawrence Erlbaum Associates, 245–280.

Dettling AC, Feldon J, Pryce CR (2002): Early deprivation and behavioral andphysiological responses to social separation/novelty in the marmoset.Pharmacol Biochem Behav 73:259 –269.

Essex MJ, Klein MH, Cho E, Kalin NH (2002): Maternal stress beginning ininfancy may sensitize children to later stress exposure: Effects on cortisoland behavior. Biol Psychiatry 52:776 –784.

Field T (1994): The effects of mother’s physical and emotional unavailabilityon emotion regulation. Monogr Soc Res Child Dev 59:208 –227.

Francis DD, Diorio J, Liu D, Meaney MJ (1999): Nongenomic transmissionacross generations of maternal behavior and stress responses in the rat.Science 286:1155–1158.

Francis DD, Diorio J, Plotsky PM, Meaney MJ (2002): Environmental enrich-ment reverses the effects of maternal separation on stress reactivity.J Neurosci 22:7840 –7843.

Gold PW, Drevets WC, Charney DS (2002): New insights into the role ofcortisol and the glucocorticoid receptor in severe depression. Biol Psy-chiatry 52:381–385.

Goldberg DP, Gater R, Sartorius N, Ustun TB, Piccinelli M, Gureje O, et al(1997): The validity of two versions of the GHQ in the WHO study ofmental illness in general health care. Psychol Med 27:191–197.

Goldberg DP, Williams P (1988): A User’s Guide to the GHQ. London: NFER-Nelson.

Goodman SH, Gotlib IH (2001): Children of Depressed Parents: Mechanisms ofRisk and Implications for Treatment. Washington, DC: American Psycho-logical Association.

Goodyer IM, Herbert J, Altham PM, Pearson J, Secher SM, Shiers HM (1996):Adrenal secretion during major depression in 8- to 16-year-olds, I. Al-tered diurnal rhythms in salivary cortisol and dehydroepiandrosterone(DHEA) at presentation. Psychol Med 26:245–256.

Goodyer IM, Tamplin A, Herbert J, Altham PM (2000): Recent life events,cortisol, dehydroepiandrosterone and the onset of major depression inhigh-risk adolescents. Br J Psychiatry 177:499 –504.

Gubba EM, Netherton CM, Herbert J (2000): Endangerment of the brain byglucocorticoids: Experimental and clinical evidence. J Neurocytol29:439 –449.

380 BIOL PSYCHIATRY 2004;55:376–381 S.L. Halligan et al

www.elsevier.com/locate/biopsych

Gunnar MR (1990): The Psychobiology of Infant Temperament. Hillsade, NJ:Erlbaum.

Harlow HF, Suomi SJ (1971): Social recovery by isolation-reared monkeys.Proc Natl Acad Sci U S A 68:1534 –1538.

Harris TO, Borsanyi S, Messari S, Stanford K, Brown GW, Cleary SE, et al (2000):Morning cortisol as a risk factor for subsequent major depressive disor-der in adult women. Br J Psychiatry 177:505–510.

Hessl D, Dawson G, Frey K, Panagiotides H, Self H, Yamada E, et al (1998): Alongitudinal study of children of depressed mothers: Psychobiologicalfindings related to stress. In: Hann DM, Huffman LC, Lederhendler KK,Minecke D, eds. Advancing Research on Developmental Plasticity: Integrat-ing the Behavioral Sciences and the Neurosciences of Mental Health. Beth-seda, MD: National Institutes of Mental Health, 256.

Hooley JM, Teasdale JD (1989): Predictors of relapse in unipolar depressives:Expressed emotion, marital distress, and perceived criticism. J AbnormPsychol 98:229 –235.

Kaufman J, Birmaher B, Brent D, Rao U, Flynn C, Moreci P, et al (1997):Schedule for Affective Disorders and Schizophrenia for School-Age Chil-dren-Present and Lifetime Version (K-SADS-PL): Initial reliability and va-lidity data. J Am Acad Child Adolesc Psychiatry 36:980 –988.

Kudielka BM, Broderick JE, Kirschbaum C (2003): Compliance with salivasampling protocols: Electronic monitoring reveals invalid cortisol day-time profiles in noncompliant subjects. Psychosom Med 65:313–319.

Lehmann J, Russig H, Feldon J, Pryce CR (2002): Effect of a single maternalseparation at different pup ages on the corticosterone stress response inadult and aged rats. Pharmacol Biochem Behav 73:141–145.

Maccari S, Darnaudery M, Morley-Fletcher S, Zuena AR, Cinque C, Van ReethO (2003): Prenatal stress and long-term consequences: Implications ofglucocorticoid hormones. Neurosci Biobehav Rev 27:119 –127.

Meikle AW, Stringham JD, Woodward MG, Bishop DT (1988): Heritability ofvariation of plasma cortisol levels. Metabolism 37:514 –517.

Murray L (1992): The impact of postnatal depression on infant development.J Child Psychol Psychiatry 33:543–561.

Murray L, Cooper PJ (2003): Intergenerational transmission of affective andcognitive processes associated with depression: Infancy and the pre-school years. In Goodyer IM, ed. Unipolar Depression: A Lifespan Perspec-tive. Oxford: Oxford University Press, 17–46.

Murray L, Fiori-Cowley A, Hooper R, Cooper P (1996): The impact of postnataldepression and associated adversity on early mother-infant interactionsand later infant outcome. Child Dev 67:2512–2526.

Murray L, Kempton C, Woolgar M, Hooper R (1993): Depressed mothers’speech to their infants and its relation to infant gender and cognitivedevelopment. J Child Psychol Psychiatry 34:1083–1101.

Nachmias M, Gunnar M, Mangelsdorf S, Parritz RH, Buss K (1996): Behavioralinhibition and stress reactivity: The moderating role of attachment secu-rity. Child Dev 67:508 –522.

Netherton C, Goodyer I, Tamplin A, Herbert J (2004): Salivary cortisol anddehydroepiandrosterone in relation to puberty and gender. Psychoneu-roendocrinology 29:125–140.

Newcomer JW, Selke G, Melson AK, Hershey T, Craft S, Richards K, et al (1999):Decreased memory performance in healthy humans induced by stress-level cortisol treatment. Arch Gen Psychiatry 56:527–533.

Nomura Y, Wickramaratne PJ, Warner V, Mufson L, Weissman MM (2002):Family discord, parental depression, and psychopathology in offspring:Ten-year follow-up. J Am Acad Child Adolesc Psychiatry 41:402–409.

Pruessner M, Hellhammer DH, Pruessner JC, Lupien SJ (2003): Self-reporteddepressive symptoms and stress levels in healthy young men: Associa-tions with the cortisol response to awakening. Psychosom Med 65:92–99.

Sapolsky RM (1985): A mechanism for glucocorticoid toxicity in the hip-pocampus: Increased neuronal vulnerability to metabolic insults. J Neu-rosci 5:1228 –1232.

Sapolsky RM (2000): Glucocorticoids and hippocampal atrophy in neuropsychi-atric disorders. Arch Gen Psychiatry 57:925–935.

Spitzer RL, Endicott J, Robins E (1978): Research diagnostic criteria: Rationaleand reliability. Arch Gen Psychiatry 35:773–782.

Spitzer RL, Williams JBW, Gibbon M (1995): Structured Clinical Interview forDSM-IV (SCID). New York: New York State Psychiatric Institute, BiometricsResearch.

Tanner JM (1966): Growth at Adolescence. New York: Appleton.van Hemert AM, den Heijer M, Vorstenbosch M, Bolk JH (1995): Detecting

psychiatric disorders in medical practice using the General Health Ques-tionnaire. Why do cut-off scores vary? Psychol Med 25:165–170.

Weaver IC, La Plante P, Weaver S, Parent A, Sharma S, Diorio J, et al (2001):Early environmental regulation of hippocampal glucocorticoid receptorgene expression: Characterization of intracellular mediators and poten-tial genomic target sites. Mol Cell Endocrinol 185:205–218.

Weissman MM, John K, Merikangas KR, Prusoff BA, Wickramaratne P, Gam-mon GD, et al (1986): Depressed parents and their children. Generalhealth, social, and psychiatric problems. Am J Dis Child 140:801–805.

Winslow JT, Noble PL, Lyons CK, Sterk SM, Insel TR (2003): Rearing effects oncerebrospinal fluid oxytocin concentration and social buffering in rhe-sus monkeys. Neuropsychopharmacology 28:910 –918.

Yehuda R, Halligan SL, Yang RK, Guo LS, Makotkine I, Singh B, et al (2003):Relationship between 24-hour urinary-free cortisol excretion and sali-vary cortisol levels sampled from awakening to bedtime in healthy sub-jects. Life Sci 73:349 –358.

S.L. Halligan et al BIOL PSYCHIATRY 2004;55:376–381 381

www.elsevier.com/locate/biopsych