Prenatal Psychosocial Factor ands the Neuroendocrine Axi

Prenatal Psychosocial Factors and the Neuroendocrine Axis in HumanPregnancy

PATHK D. WADHWA, MD, PHD, CHRISTINE DUNKEL-SCHETTER, PHD, ALEKSANDRA CHJCZ-DEMET, PHD,

MANUEL PORTO, MD, AND CURT A. SANDMAN, PHD

Objective: Physiological processes including neuroendocrine function have been proposed as mediators of therelationship between prenatal psychological state and pregnancy outcome; however, there are virtually nohuman studies that have systematically assessed such mechanisms. Neuroendocrine processes are signifi-cantly altered during pregnancy, and are characterized by the evolution of a transient neuroendocrine system,the placenta, and modifications in endocrine control mechanisms. Because these alterations have implicationsfor neuroendocrine responsivity to exogenous conditions, the aim of the present study was to examine thecross-sectional association between prenatal psychosocial factors and stress-related neuroendocrine param-eters during human pregnancy.

Method: Fifty-four adult women with a singleton, intrauterine pregnancy were recruited before 28 weeks ofgestation. Maternal antecubital venous blood samples were withdrawn at 28 weeks of gestation for bioassaysof adrenocorticotropin hormone (ACTH), /3-endorphin (jSE), and cortisol. Measures of prenatal stress, socialsupport, and personality were collected using a two-part, self-report questionnaire administered at 28 and 30weeks of gestation. Biomedical data were obtained from the medical record. Factors known to influenceneuropeptide and hormone levels during pregnancy were controlled, including gestational age, circadianvariation, and obstetric risk.

Results: In the present sample, prenatal psychosocial stress, social support, and personality variables wereassociated with neuroendocrine parameters in two primary ways. First, certain psychosocial factors weresignificantly associated with plasma levels of ACTH, j3E, and cortisol, and second, psychosocial factors wereassociated with a measure of disregulation of the normal relationship between two pro-opiomelanocortin(POMC) derivatives, ACTH and /3E. Furthermore, a combination of the maternal psychosocial and sociode-mographic factors during pregnancy accounted for 36% of the variance in ACTH, 22% of the variance in theACTH-/3E disregulation index, 13% of the variance in cortisol, and 3% of the variance in /3E.

Conclusions: The present findings are consistent with the premise that maternal-placental-fetal neuroen-docrine parameters are significantly associated, both in magnitude and specificity, with features of maternalpsychosocial functioning in pregnancy despite the systemic alterations associated with the endocrinology ofpregnancy. These findings provide a basis for further investigations of the role of the neuroendocrine systemas a putative mediating pathway between prenatal psychosocial factors and birth outcome, and possibly alsoas a mechanism linking features of the maternal psychosocial environment to fetal/infant brain development.Key words: pregnancy, psychosocial, stress, social support, neuroendocrine, ACTH, /3E, cortisol, POMC.

INTRODUCTION leading causes of perinatal mortality and morbidityin nonanamalous newborns, and are also associated

Adverse pregnancy and birth outcomes are one of w i t h significantly higher rates of long-term neurode-the most significant problems in maternal-child velopmental impairments and disabilities (1, 3). Bio-health in the United States (1, 2). Outcomes such as medical, or obstetric, risk factors predict only a smallpreterm birth and low birth weight are presently the proportion of the variance in these outcomes (4-6],

and several authors have proposed that maternalpsychological and social factors during pregnancymay influence reproductive and pregnancy out-comes. Results from earlier human studies of prena-

M ^°m the Departments of Obstetrics and Gynecology (P.D.W ^ psychological Stress and social support wereM.P.) and Psychiatry and Human Behavior (A.C-DeM., C.A.S.), ^ J ° ^rUniversity of California, Irvine; and Department of Psychology mixed, but most ol those Studies were limited by(C.D-S.), University of California, Los Angeles. conceptual and methodological weaknesses, includ-

Address reprint requests to: Pathik D. Wadhwa, MD, PhD, ing issues related to sampling, definition and mea-Depaitment of Obstetrics and Gynecology, Division of Maternal- gurement of predictor and outcome variables, re-Fetal Medicine, UCI Medical Center, 101 City Drive South, Build- v. J • J • J * t I £ •*.or rv OA r,oôo r. i ,, „ . . search design, and inadequate control or covariatesing 25, Orange, CA 92668. E-mail: [email protected]. r

Received for publication March 7, 1995; revision received in prediction of adverse outcomes (see 7-9 for recentNovember 20,1995. reviews). Recent prospective human studies ad-

432 Psychosomatic Medicine 58:432-446 (1996)

0033-3174/9 6/5805-0432$03.0O/0Copyright © 1996 by the American Psychosomatic Society

PRENATAL PSYCHOSOCIAL FACTORS AND NEUROENDOCRINE AXIS

dressing several of these limitations (eg, 10-19),however, have more consistently found that psycho-social factors are significantly related to the inci-dence of adverse birth outcomes, and that this asso-ciation is independent of sociodemographic andbiomedical, or obstetric, risk. A large body of animalresearch supports and extends the conclusions of theabove human studies. Experimental studies in rats,sheep, and primates suggest that prenatal stress iscausally associated not only with adverse birth out-comes such as preterm birth and low birth weight,but also with adverse long-term neurodevelopmentaloutcomes related to brain morphology, physiology,and behavior (eg, 20-30).

Various physiological processes have been pro-posed as possible mediators of the relationship be-tween prenatal psychological state and pregnancyoutcome, however there are virtually no humanstudies that have systematically assessed such mech-anisms (18). The participation of the hypothalamic-pituitary-adrenal (HPA) and immune axis in re-sponse to psychological stress has been wellestablished (see 31-33 for reviews) and has beenproposed as a central mechanism linking psychoso-cial factors to health outcomes (34, 35). Severalauthors have suggested that stress-related responsesof the neuroendocrine axis and the autonomic ner-vous system during pregnancy may contribute toadverse outcomes. For instance, elevated levels ofhypothalamic, pituitary, and placental hormoneshave been implicated in the initiation of pretermlabor (36). Vasoconstriction and hypoxia in responseto sympathetic-adrenal-pituitary activation decreaseuteroplacental perfusion, and may thereby contrib-ute to fetal growth restriction (37-39). Endorphiner-gic responses alter pain sensitivity, and may therebyinfluence labor and delivery parameters (40-42).Finally, the immunosuppressive effects of stress andHPA activation may increase susceptibility to infec-tion (43, 44), which, in turn, is a risk factor forpreterm birth (45, 46). Although a few studies haveexamined the association between catecholaminesand anxiety at the onset of labor (eg, 47), andbetween catecholamines and physical activity dur-ing pregnancy (eg, 48), a review of the relevantliterature since 1975 revealed only one study (49)that examined the relation between a psychosocialfactor and neuroendocrine parameter during humanpregnancy. In a sample of 40 pregnant adolescentsassessed during the middle and latter part of gesta-tion and postpartum, subjects with an increase incortisol levels across a 40-minute period measuredbefore 20 weeks of gestation and at 2 to 3 weekspostpartum had fewer symptoms of anxiety and

depression than subjects with no cortisol increase;there was, however, no relation between cortisol andsymptoms of anxiety or depression at the 34 to 36weeks of gestation assessment.

The physiological responses of pregnant women topsychological factors such as stress may be compli-cated by the "background" changes in neuroendo-crine function during pregnancy. Neuroendocrinealterations in pregnancy are characterized by theevolution of a transient endocrine unit, the placenta,and by modification of control mechanisms. Withinthe pregnant uterus, starting at 7 to 9 weeks ofgestation, the fetal-placental-decidual unit producessteroids and peptide hormones, neuropeptides,growth factors, and cytokines, and seems to functionin a manner resembling compressed hypothalamic-pituitary-target systems (50). Although cortisol in-hibits corticotropin-releasing hormone (CRH) ex-pression in the hypothalamus, it stimulates theexpression of the CRH gene in the placenta to causea two- to five-fold increase in CRHmRNA (51, 52).This results in increased synthesis and release ofplacental CRH and other POMC products, includingACTH and /3E. A certain proportion of the CRH andjBE released into the maternal compartment is bio-logically inactive (53, 54), whereas the ACTH re-leased in the maternal compartment is bioactive (55,56). Placental CRH stimulates the maternal pituitary-adrenal axis and causes an increase in the secretionof cortisol from the adrenal cortex (57-59), therebyestablishing a positive placental-adrenal feedbackloop that allows for the simultaneous increase inlevels of CRH, ACTH, and cortisol (60-62). Thesealterations have implications for changes in neu-roendocrine responsivity, especially in the thirdtrimester of gestation. As pregnancy advances, thereis a progressive and significant increase in plasmaconcentrations of stress hormones including CRH,ACTH, /3E, and cortisol, with a peak at labor anddelivery and a rapid return to nonpregnant levelsafter delivery (53, 60, 63-65). Increases in CRHand/or hypercortisolemia are thought to desensitizethe corticotrophe and exert inhibitory influences onthe hypothalamus and pituitary. These inhibitoryinfluences, in turn, may compete with the excitatoryactions of exogenous stimuli to ultimately result inattenuation of neuroendocrine responsivity (60, 66,67).

Because alterations in the neuroendocrine systemduring pregnancy may have implications for sys-temic responsivity to exogenous conditions, it be-comes necessary and crucial to first examinewhether neuroendocrine parameters are related toprenatal psychosocial conditions before proposing a

Psychosomatic Medicine 58:432-446 (1996) 433

P. D. WADHWA et al.

mediating role for the neuroendocrine system in therelation between prenatal psychological state andpregnancy outcome. Hence, the aim of the presentstudy was to examine the cross-sectional associationbetween prenatal psychosocial factors (includingstress, social support, personality, and sociodemo-graphic variables) and stress-related neuroendocrineparameters during human pregnancy.1 Prenatalstress, social support, and sociodemographic factorswere selected as variables of interest because theyhave been implicated in adverse pregnancy out-comes in previous human studies. Personality vari-ables were selected because they have been impli-cated in other adverse health outcomes in humanstudies. ACTH, j8E, and cortisol were selected as theneuroendocrine variables of interest because thesehormones have been shown in previous nonpreg-nant human research to reflect pituitary-adrenal re-sponses to psychological state, and because theyhave been implicated in the physiology of pregnancyand pregnancy outcomes. Because pregnancy-in-duced changes in maternal neuroendocrine functionare more pronounced as pregnancy advances, theearly third trimester was selected as the time periodto conduct the study. Because existing medical con-ditions may relate to both neuroendocrine and psy-chosocial factors, measures of biomedical, or obstet-ric, risk were obtained from the medical record andincluded in the analyses. Finally, because diurnalvariation of neuroendocrine parameters has beenreported during pregnancy, the time of day of eachmaternal blood draw was controlled in the analyses.

METHODS

SubjectsThe sample was comprised of 54 adult (>18 years), English-

speaking women with a singleton, intrauterine pregnancy attend-ing prenatal care at the faculty practice or the residents' clinic ofa large, metropolitan, teaching hospital affiliated with the Univer-

TABLE 1. Sample characteristics

1 It is important to note that the present study was not designedto examine the hypothesized mediating role of the neuroendo-crine system in the relationship between psychosocial factors andbirth outcomes. Given the above-described neuroendocrinechanges over the course of gestation, a test of the neuroendocrine-birth outcome relation would require a measure of rate of neu-roendocrine change, and the study design would thereby requirethe serial assessment of neuroendocrine parameters over thecourse of pregnancy, as opposed to a single assessment at onepoint in time.

N

Age

Parity

Education

Marital

Ethnicity

Occupation

Annual family income

54

30.5 ± 5.2 years

PrimiparousMultiparousHigh school graduatesCollege graduatesOtherMarriedSeparated/divorcedSingleAngloHispanicAfrican-AmericanAsian/otherEmployed for payHousewives

<$ 20,000$20,000-$39,999$40,000-$49,999

>$50,000

Range: 18-42years

29.2%70.8%44.4%37.0%18.6%85.2%9.2%5.6%

79.6%11.1%

1.9%7.4%

79.6%20.4%22.2%16.7%14.8%46.3%

sity of California, Irvine.2 All subjects had enrolled for prenatalcare by the late first or early second trimester of pregnancy, and allsubjects received comparable obstetric care. The sample charac-teristics are described in Table 1.

ProceduresAfter obtaining approval from the Institutional Review Board,

eligible subjects (adult, English-speaking, with a singleton intra-uterine pregnancy, at or before 28 weeks of gestation) wereapproached consecutively by the research staff to participate inthe study. The participation rate was 78%. Neuroendocrine datawere obtained from plasma bioassays of a maternal blood sampleobtained at the subject's 28th week prenatal clinic appointment.Psychosocial data were collected using a two-part, self-reportquestionnaire administered at the prenatal clinic on two occa-sions which coincided with the subject's 28th and 30th weekprenatal clinic appointments. Biomedical data were obtainedfrom the subjects' medical records.

MeasuresNeuroendocrine Measures. Blood samples (20 ml/draw) were

withdrawn by antecubital venipuncture into siliconized EDTA(purple top) vacutainers and placed on ice immediately.3 Samples



were centrifuged at 2000 X g (10 minutes) and the plasma wasdecanted into polypropylene tubes containing 500 KlU/ml apro-tinin (Sigma Chemical Company; St. Louis, MO). The sampleswere stored at -70°C until assayed.

Adrenocorticotropin hormone assay. Plasma levels of ACTHwere measured by a commercially available radioimmunoassay(Nichols Institute Diagnostics; San Juan Capistrano, CA). Theantiserum employed has <0.001% cross-reactivity with /3-endor-phin and ACTH fragments. Samples were assayed in duplicate(200 Î/assay tube). ACTH 12SI-antibody solution (100 /xl) wasadded to the samples, vortexed and incubated at room tempera-ture for 20 ± 2 hours after the addition of an avidin-coated bead.The solid matrix was washed with buffered surfactant in phos-phate-buffered saline to remove unbound components, and thebound radiolabeled antibody complex was quantified using aMicromedic Isoflex Gamma Counter. The ACTH assay has aminimum detectable dose level (MDD) = 1.0 pg/ml (95% confi-dence) with coefficient of variation (CV) = 3.0% (intra-assay) at 35pg/ml and CV = 7.8% (inter-assay) at 36 pg/ml.

Beta-endorphin assay. Plasma levels of /3E were determined bya commercially available solid phase two-site immunoradiomet-ric assay (IRMA; Nichols Institute Diagnostics; San Juan Capist-rano, CA). The antisorum has 1.6% cross-reactivity with beta-lipotropin at 500 pg/ml and has <0.01% cross-reactivity withrelated opiates at 5 ^g/ml. Samples were assayed in duplicate(200 Î/assay tube). 125I-Anti-BE (rabbit) solution (100 /xl) wasadded to each tube and vortexed. The reaction was initiated byadding one anti-BE (rabbit) coated polystyrene bead to the assaytube followed by a stationary incubation at room temperature for20 ± 4 hours. The beads were then washed twice with phosphate-buffered saline and aspirated to dryness. The labeled antibodycomplex bound to the solid phase was measured using a Micro-medic Isoflex Gamma Counter. The Allegro beta-Endorphin Im-munoassay system has a MDD = 10 pg/ml (95% confidence limit)

maternal neuroendocrino parameters over a 20-minute time inter-val and the possible effects of venipuncture on plasma levels ofACTH, /3E, and cortisol. An intravenous catheter was inserted intothe median antecubital vein of the nondominant arm, and twosamples of blood were collected from each subject. The firstsample was collected within 15 seconds of insertion of theintravenous catheter (to simulate the collection procedure used inthe present study), and the second sample was collected exactly20 minutes later. Results indicated there were no significantdifferences in the mean plasma concentrations for all threehormones across the two time points; mean levels of ACTH, j3E,and cortisol were 37.96 ± 11.40 vs 40.97 ± 12.31 pg/ml; 51.51 ±16.35 vs 52.68 ± 14.98 pg/ml; and 31.77 ± 7.20 vs 29.45 ± 6.27/j,g/ml from the first and second samples, respectively. Thesefindings suggest that the three stress hormones are relativelystable across a 20-minute interval of time, that a time interval ofless than 15 seconds between venipuncture and the collection ofthe blood sample does not provide a sufficient period of time forthe neuroendocrine system to evoke a systemic response to thevenipuncture procedure, and that a period of 20 minutes afterprovocation by venipuncture is sufficient for the system to returnto baseline. These data are presented here to validate the plasmasample collection procedure used in the present study and todemonstrate the temporal stability of the three neuroendocrineparameters being assessed.

with a CV = 4.1% (intra-assay) and CV = 9.0% (inter-assay) at thehighest concentrations expected in the present study.

Cortisol assay. Plasma cortisol levels were determined byimmunofluorescence using an automated procedure on an AbbottTDx Analyzer (Abbott Laboratories; Abbot Park, 1L). The assay has<5% cross-reactivity with 11-deoxycortisol, corticosterone, and<1% cross-reactivity with 10 other naturally occurring steroids.The inter-assay and intra-assay CVs are <9% with a MDD (95%confidence) of 0.45 jxg/dL.

Data reduction for the RIA and IRMA assays were done by acomputer-assisted four-parameter logistics program (68).

Psychosocial Measures. The two self-administered question-naires contained measures of prenatal psychosocial stress, per-ceived social support, personality, and sociodemographic factors.Most of the instruments used in the questionnaires consisted ofpreviously validated and published scales. Subjects completedthe questionnaires in a private research suite at their prenatalclinic (study site) during their 28th and 30th week prenatalappointments. On average, subjects required between 25 and 40minutes to complete each questionnaire. A research assistant wasavailable to answer any questions during this time and to checkfor completeness.

Prenatal stress. Prenatal stress was conceptualized in threeways: life event stress, perceived stress, and pregnancy-relatedanxiety. These three constructs were assessed with five instru-ments measuring life events, daily hassles, chronic stress, emo-tional distress, and pregnancy-related anxiety. These instrumentswere derived from past stress research in general (69-71) andavailable research in pregnancy (13).

Life events: A 99-item modified version (72) of the Schedule ofRecent Life Events (73) was used to assess disruptive changes inpersonal (eg, involved in a lawsuit or court case), family (eg, adeath in the family), interpersonal (eg, a separation/divorce fromone's spouse), social (eg, broke up with a friend), financial (eg,took a cut in wage or salary), and work-related (eg, unemploy-ment) areas that are not usually everyday occurrences. Subjectswere asked to indicate whether they had experienced any of thelisted events since the beginning of their current pregnancy, andif so, whether once or more than once. For each of the events thathad occurred, subjects were asked to make an appraisal of theaverage severity of distress experienced using a 5-point scaleranging from "not stressful at all" to "extremely stressful." Toinclude both occurrence and subjective severity of prenatal lifeevents, Life Event Stress scores (LES) were computed for eachsubject by calculating the standardized score of the product ofnumber of events (frequency) and average severity rating.

Daily hassles: The 117-item Daily Hassles Questionnaire (70)was used to assess frequently occurring daily stressors in thecontent areas of work, family, social activities, the environment,practical considerations, finances, and health. Subjects rated thedegree of severity of each event that had occurred since thebeginning of their current pregnancy with a 4-point scale whichranged from "did not occur" to "extremely severe." These severityratings were summed to compute a daily hassles score for eachsubject.

Chronic stress: The 14-item Perceived Stress Scale (PSS) (69)was used to assess the degree to which any experiences in the lastmonth were appraised as stressful. Specifically, items assess thedegree to which respondents perceive their lives as unpredictable,uncontrollable, and burdensome. It included items such as "In thelast month, how often have you been upset because of somethingthat happened unexpectedly?" and "In the last month, how oftenhave you felt confident about your ability to handle your personalproblems?" which were rated on a 5-point rating scale ranging


P. D. WADHWA et al.

from "never" to "very often." After reversing scores on negativelyworded items, the ratings were summed to compute a chronicstress score.

Emotional distress: The 45-item Hopkins Symptom Checklist(71) was used to measure psychological symptomatology (eg,nervousness, lower back pain, a lump in your throat, feeling blue).Subjects rated the frequency of occurrence of the listed symptomssince the beginning of their pregnancy with a 4-point scale whichranged from "not at all" to "continuously." A total emotionaldistress score, as well as subscores for anxiety and depressionwere computed.

Pregnancy-related anxiety: This was measured with a five-itemscale extracted by factor analysis from a larger set of itemsdesigned for this study by the authors (18). The items consisted ofmodified items from previous work (74, 75) and from a part of thepsychosocial assessment protocol of the Comprehensive PerinatalServices Program (CPSP) of the State of California, Department ofHealth Services. This instrument assesses maternal fears andanxiety specifically related to the health of the baby and the laborand delivery process. Respondents were asked to check either a"true" or a "false" response to each item. After reversing theresponses on the positively worded items, scores were summed toyield a pregnancy-anxiety score for each subject.

Social support. Perceived social support: A commonly used,standard measure of general social support, the InterpersonalSupport Evaluation List (ISEL) (76) was used to assess perceivedsocial support. This 40-item questionnaire is designed for adultpopulations, and assesses the perceived availability of four typesof social support—tangible, appraisal, self-esteem, and belonging.Items include statements such as "When I need suggestions forhow to deal with a personal problem, I know someone I can turnto," and "I am closer to my friends than most other people are totheirs." Based on whether the statement was generally true or notfor them, respondents were asked to check either a "true" or a"false" response to each item. After reversing the responses on thenegatively worded items, scores were summed to yield a totalperceived social support score for each subject. Higher scoresindicate a greater amount of perceived social support.

Pregnancy-specific social support: This was measured with 17items derived from those previously used in pregnancy-relatedstudies (10). This measure had been found to be psychometricallysound for assessing social support in pregnant women of diversesocioeconomic status and ethnic background. Items were in-cluded for different possible sources of support in the subjects'lives, namely, baby's father, family, and friends. For baby's father,four types of support were assessed (eg, emotional, task, material,information) as well as negative behaviors (eg, criticism). Forfamily and friends, only emotional support was assessed. Exam-ples of items were "My husband/partner reacted to this pregnancyby being excited and happy," "I receive help with things I have todo such as errands, household tasks, or children," and "Myemployer (if applicable) is very supportive of my pregnancy andthe special needs that I have during this time (time off forappointments, etc)." Subjects used a 5-point scale, which rangedfrom "never" to "almost always," to respond to each item. Afterreversing scores of negatively worded items, responses weresummed to create a total score of pregnancy-related support foreach subject.

Personality. Three personality variables that have been associ-ated with various physical or mental health outcomes in previousresearch, and for which validated measures were available, wereassessed.

Type A behavior: The 52-item Jenkins Activity Survey (77) wasused to assess Type A behavior pattern. The instrument includes

items such as "How often do you find yourself doing more thanone thing at a time?", "How often do you find yourself hurrying toget places even when there is plenty of time?," and "Wouldpeople you know well agree that you tend to get irritated easily?"After reversing scores of positively worded items, responses weresummed to create a total score of Type A behavior pattern.

Affect intensity: The 40-item Affect Intensity Measure (78) wasused to assess affect intensity. This instrument assesses thetypical strength of an individual's affective responsiveness acrossvarious emotional categories. Items include statements such as"My negative moods are mild in intensity," and "I get overlyenthusiastic." Subjects rated the accuracy of the listed conditionson a 6-point scale which ranged from "never" to "always." Afterreversing scores of positively worded items, responses weresummed and averaged to create a total score of Affect Intensity.

Hardiness: The 50-item Personal Views Survey (79) was used toassess personality hardiness. This instrument assesses positiveand negative expressions of beliefs about self and world expres-sive of commitment, control, or challenge. Examples of itemsinclude, for commitment, "Most of my life gets spent doing thingsthat are worthwhile," for control, "Planning ahead can help solvemost future problems," and for challenge, "It is exciting to learnsomething about myself." Subjects rated each item on a 4-pointscale that ranged from "not at all true" to "completely true." Afterreversing scores of negatively worded items, responses weresummed to create a total score of personality hardiness.

Demographic information. Demographic information, includ-ing age, education, ethnicity, marital status, occupation, andannual family income were also obtained by questionnaire.

Biomedical (Obstetric) Risk. Biomedical, or obstetric, risk ofadverse pregnancy outcome (eg, low birth weight, preterm birth)was determined by the presence of antepartum complicationsduring pregnancy. In the present sample, these included diabetes,eclampsia, heart disease, hypertension, induction of labor, intra-uterine growth retardation, pregnancy-induced hypertension, pre-term labor, placenta previa, and premature rupture of membranes.Because there are currently no standard criteria for the computa-tion of degree or weight of specific antepartum risk conditions, weadopted a conservative strategy and created a dichotomous riskvariable (high-low) (18). A subject was categorized as being at lowbiomedical risk for poor pregnancy outcome if she did notexperience any of the above conditions during pregnancy, and asbeing at high biomedical risk if she experienced one or more of theabove conditions during pregnancy. Based on the presence of oneor more antepartum risk conditions, approximately one third ofthe sample (29.2%) was categorized as high risk, whereas theother two thirds was categorized as low risk.

RESULTS

The data were subjected to three types of analy-ses—descriptive, bivariate, and multivariate. De-scriptive analyses were performed to determine fre-quency distributions of each variable; bivariateanalyses were performed to examine intercorrela-tions among psychosocial, biomedical, and neuroen-docrine factors; and multivariate analyses were per-formed to examine the simultaneous contributionsof sets of psychosocial factors to the variance of eachneuroendocrine parameter.

Table 2 depicts the descriptive statistics for all



TABLE 2. Descriptive Statistics of Psychosocial Variables

Construct

Stress

Life event stress(z (life event frequency)*

(severity)]Daily hasslesChronic stressEmotional distress

Perceived stress(z(hassles)+z(chronic

stress)+z(distress)]Pregnancy anxietyGeneral social supportPregnancy-specific supportPersonality

Type A behaviorAffect intensityHardiness

Measure

Life event frequencyLife event severity

Mean

12.821.250.09

51.8423.2039.34

0 06

0.9425.7430.98

70.193.75

74.22

S.D.

8.980.581.37

39.158.5

21.122.85

1.164.725.07

11.070.388.36

Range

2-480-2.5

-2.73-4.76

5-1587-447-108

-5.63-8.78

0-59-30

18-36

38-882.8-4.3547-92

Cronbach a

NANANA

0.960.860.940.92

0.730.870.75

0.890.820.85

psychosocial variables, including means, standarddeviations, ranges, and internal reliability coeffi-cients. In the present sample, the scores for each ofthe measures were approximately normally distrib-uted and the reliability of each the measures wasdetermined to be adequate (the internal consistencycoefficients, or Cronbach alphas, were moderate tohigh, and ranged between 0.73 and 0.96). As de-scribed in an earlier report (18), the individual stressmeasures were combined to form three indices ofLife Event Stress, Perceived Stress, and Pregnancy-Specific Anxiety on the basis of intercorrelations ofthese measures. Daily hassles, perceived stress, andemotional distress variables were highly intercorre-lated (rs ranged between 0.68 and 0.75; ps<.001),and their scores were, therefore, standardized andsummed to create a composite index labeled Per-ceived Stress. Scores of pregnancy-related anxietywere not significantly correlated with Life EventStress scores, and were only moderately correlatedwith scores of daily hassles, chronic stress, andemotional distress (rs ranged between 0.41 and 0.45;ps<.01), and were therefore, used independently asan index of Pregnancy Anxiety.

Table 3 depicts the mean concentrations, range,and standard deviations of plasma levels of ACTH,j3E, and cortisol in third trimester maternal venousblood. Values for each of these hormones wereapproximately normally distributed, and theirplasma concentrations were comparable to thoseearly third trimester levels reported in other studies(64, 80-81). Plasma concentrations of ACTH and /3Ewere strongly intercorrelated (r=0.51, p<.001),whereas those of ACTH and cortisol were slightly

intercorrelated (r=0.24, p<.01). Scatterplots ofplasma concentrations of ACTH and /3E depicted thehigh degree of correlation between the levels ofACTH and j3E, as well as the cases when thisrelationship was uncoupled or disregulated (Fig. 1).ACTH and j3E are products of the same parentmolecule—POMC, and are thought to be released ina pulsatile manner in approximately equimolaramounts into peripheral circulation. Hence, a dis-regulation index (DI) was created to examine thedegree to which the ACTH-/3E correlation patternwas disrupted (40). It was computed as shown be-low:

DI = absolute value [(/3E-ACTH)/j3E) X 100]

The DI score measures the absolute value of thepercentage difference between BE and ACTH levels.A larger value on the DI indicates a greater relativedifference between BE and ACTH levels (ie, moredisregulation), whereas a lower value indicates asmaller relative difference between BE and ACTHlevels (ie, less disregulation). In the present sample,the mean disregulation index score was 31.56, with astandard deviation of 20.12 and a range between 0.51and 84.4

Bivariate analyses were performed to examine theassociations of the psychosocial and demographicfactors with maternal plasma concentrations ofACTH, j3E, cortisol, and the magnitude of disregula-

1 Three outliers were coded in.


P. D. WADHWA et al.

TABLE 3.

MeanRangeSD

Maternal Plasma Levels of ACTH, 0-Endorphins,and Cortisol at 28 Weeks of Gestation

ACTH(pg/ml)

33.226-9514.48

(pg/ml)

47.623-10718.67

Cortisol(ptg/dl)

27 667-5510.19

tion score. Pearson product-moment correlation co-efficients were computed to examine associationsbetween continuous variables, and Spearman rhocorrelation coefficients were computed to examineassociations between dichotomous variables. Alltests for statistical significance were two-tailed.Given the relatively modest sample size and thelarge number of variables, the bivariate analyseswere considered exploratory in nature, and wereconducted mainly to guide subsequent multivariateanalyses (82, 83).

As shown in Table 4, married subjects had lowerACTH levels and had larger ACTH-/3E disregulationcompared to subjects not married. Subjects reportinggreater perceived stress (a combination of chronicstress, daily hassles, and emotional distress) hadsignificantly higher concentrations of plasma ACTH(Fig. 2) and had significantly lower ACTH-/3E dis-regulation scores. Perceived social support was as-sociated with significantly lower levels of ACTH(Fig. 3), J3E, and cortisol, and significantly greaterACTH-/3E disregulation scores. Furthermore, preg-nancy-specific social support was associated withsignificantly lower levels of ACTH, and cortisol, andsignificantly greater ACTH-/3E disregulation. HigherType A behavior scores were associated with signif-icantly higher concentrations of /3E and significantlygreater ACTH-/3E disregulation, and higher hardi-ness scores were significantly associated with lowercortisol levels. Affect Intensity was not significantlyassociated with any of the neuroendocrine parame-ters.

Spearman rho correlation coefficients were com-puted to examine the association between the bio-medical risk index and levels of ACTH (r=-.23,p=NS), j3E (r=.O4, p=NS), and cortisol (r=.O2,p=NS). Because biomedical or obstetric risk was notassociated with plasma concentrations of eitherACTH, /3E, or cortisol in the present sample, it wasnot included in subsequent multivariate analyses.

Multiple regression analyses were performed toexamine the joint contribution of demographic andpsychosocial factors to each of the four third trimes-ter neuroendocrine parameters—ACTH, /3E, cortisol,and the ACTH-/3E disregulation index. A step-wise,

B-Endorphin (pg/ml)

Fig. 1. Scatterplot of the correlations between maternal ACTHand pE at 28 weeks of gestation (r=. 51, p<.00l).

hierarchial multiple regression procedure was usedfor these analyses. Variables were entered in threesteps. Demographic variables (maternal age, andmarital status) were entered as a set in Step 1,personality variables (type A behavior, affect inten-sity, and hardiness) were entered as a set in step 2,and prenatal stress and social support variables (lifeevent stress, perceived stress, pregnancy anxiety,general social support, and pregnancy support) wereentered as a set in step 35. Analyses thus permittedtests of the contributions of each set of variables aftercontrolling for the effects of the prior set.

Results of each step of the regression analyses areshown in Table 5. The sociodemographic variables(age and marital status) significantly predicted levelsof ACTH (p<.001) and ACTH-/3E disregulation(p<.05), and there was a marginally significant asso-ciation with cortisol (p<.10). After sociodemo-graphic variables were controlled, personality vari-ables (type A, affect intensity, hardiness) weresignificantly associated with ACTH (p<.0l), andwere marginally significantly associated with /3E andcortisol (p<.10). In the third step, after sociodemo-graphics and personality were controlled, the three

5 Multiple regression analyses were also performed by revers-ing the order of steps 2 (personality variables) and 3 (stress andsocial support variables). There were no significant changes intheoverall multivariate or univariate results.



TABLE 4. Correlations Between Psychosocial and Neuroendocrine Parameters

DemographicAgeMarital status

StressLife event stress (LES)Perceived stress (PS)Pregnancy anxiety

Social SupportGeneral support (ISEL)Pregnancy support

PersonalityType A behaviorAffect intensity (AIM)Hardiness

Two-tailedN = 54, *** p < .001, ** p < .01,

ACTH

0.00- 0 . 4 7 " *

0.100 44***0.23

-0.48***-0.48***

0.01-0.04-0.19

* p < .05.

0E

0.01-0.20

0.190.170.08

-0.27*-0.17

-0.30*-0.06-0.18

Cortisol

-0.24-0.23

-0.080.150.07

- 0 . 3 1 *-0.29*

-0.22-0.10-0.28*

Dissregulation index

0.090.31*

-0.07-0.38**-0.19

0.27*0.37"

0.29*0.080.13

stress and two social support variables togethersignificantly predicted ACTH (p<.001) andACTH-/3E disregulation (p<.05), and there was amarginally significant association with cortisol(p<.10). In total, adjusted R2 indicated that the threesets of predictors accounted for 36% of the variancein ACTH, 22% of the variance in ACTH-/3E disregu-lation, 13% of the variance in cortisol and 3% of thevariance in /3E.

Plasma samples had been collected between thehours of 8 AM and 5 PM. To test the effects ofdiurnal variations of neuroendocrine parameters, thetime of day that blood samples were obtained wasconverted to nautical time (eg, 2:50 PM was coded as1450), and Pearson product-moment correlation co-efficients were computed to examine the associationbetween time of draw and levels of /3E, ACTH andcortisol. Time of blood draw was not associated withlevels of ACTH (r=-.16, p=.30) or /3E (r=-.21,p=.2O), but was negatively associated with cortisol(r=-.37,p<.05), indicating the presence of a diurnalrhythm for plasma cortisol with higher plasma con-centrations in the morning and lower plasma con-centrations in the evening. To control for the effectsof diurnal variation of cortisol, an additional set ofregression analysis was performed to predict levelsof cortisol by adding a step. Time of day when bloodwas drawn was the predictor variable in step 1, andsociodemographics, personality variables, and stressand social support variables were added in steps 2through 4, respectively, as in earlier analyses. Aftertime was controlled, the contributions of sociodemo-graphic variables in step 2 and personality variablesin step 3 were each significant (p<.05), as before. Inthe fourth step, with time, sociodemographics, andpersonality controlled, the three stress and two so-

cial support variables together had a marginallysignificantly association with cortisol (p<.10). Intotal, adjusted R2 indicated that the four sets ofpredictors accounted for 22% of the variance incortisol, a 9% increase due to the addition of thetime of day variable.

Regression analyses provided results of t tests ofassociations between the individual variables (con-tained within steps) and the four outcomes. Theseare summarized below because they may provideinsight as to which variables were most likely toaccount for significant steps in regression analyses.However, these results must be interpreted cau-tiously given the sample size. The only significantpredictor of J3E was type A (t=2.42, p<.05). ACTHwas significantly associated with marital status(t=-3.92, p<.001), perceived stress (t=2.21, p<.05),pregnancy anxiety (t=2.00, p<.05), and supportavailability (t=—2.22, p<.05). Cortisol was associ-ated significantly with perceived social support(t=-2.23, p<.05) and marginally with hardiness(t=-1.95, p=.O6). ACTH-jSE disregulation was asso-ciated significantly with marital status (t=2.47,p<.05) and with perceived stress (t=-2.07, p<.05).

DISCUSSION

The present study is among the first to examinethe relationship between psychosocial factors andneuroendocrine parameters in human pregnancy. Inthe present sample, the measurement of psychoso-cial factors was reliable, and the shared variance ofrelated individual stress measures was combined,thereby enhancing the measurement of prenatalstress conceptually and empirically (18). The plasma


P. D. WADHWA et al.

IO<

-4 -2 0 2 4 6

Perceived Stress8 10

Fig. 2. Scatterplot of the correlations between perceived stressand maternal ACTH at 28 weeks of gestation (r=.44,p<.001).

10 -

10 15LOW

20 25 30 35

Social Support

40 45HIGH

Fig. 3. Scatterplot of the correlations between social supportand maternal ACTH at 28 weeks of gestation (r=-.48,p<.00l).

concentrations of the three HPA-placental hormonesas well as the direction and magnitude of the asso-ciations between them conformed to expected normsand patterns during pregnancy. The small associa-tion between ACTH and cortisol during pregnancy

reflects the expected modification of normal controlexerted by cortisol on ACTH, whereas the highcorrelation between ACTH and |8E is indicative oftheir co-release from the common precursor or par-ent molecule, POMC. Although estimation of neu-roendocrine activity was limited in the present studyby only one cross-sectional assessment of hormoneconcentrations, factors known to influence hormonelevels during pregnancy were controlled includinggestational age (50, 62), diurnal variations (64, 84-88), and biomedical or obstetric risk (46, 50, 89).

There were significant associations between pre-natal psychosocial factors and maternal neuroendo-crine parameters despite the increased level of neu-roendocrine activation and other backgroundchanges related to pregnancy. Prenatal psychosocialstress, social support, and personality variables wereassociated with maternal neuroendocrine parame-ters in at least two ways. First, psychosocial factorswere significantly associated with plasma concentra-tions of ACTH, )3E, and cortisol, and second, psycho-social factors were associated with disregulation ofthe normal relationship between two POMC deriva-tives, ACTH and jSE. Further, a combination of thematernal psychosocial and sociodemographic fac-tors during pregnancy accounted for considerablevariance in the measures of all neuroendocrine pa-rameters but j3E.

Of the three neuroendocrine parameters, plasmaACTH was the most sensitive indicator of maternalpsychological stress. Subjects reporting greater pre-natal perceived stress and greater pregnancy anxietyhad significantly higher plasma concentrations ofACTH. This stress-ACTH relationship during humanpregnancy is similar to that reported in studies ofnonpregnant human subjects (eg, 90-94) and ofpregnant animals (eg, 23, 27, 95-97). This findingraises two issues. First, it appeared that only sometypes of prenatal psychological stress were related toACTH levels in the early third trimester of gestation.In the present sample, perceived stress and preg-nancy anxiety were associated with ACTH but pre-natal life event stress was not. It is possible that thematernal-placental-fetal neuroendocrine axis may bemore responsive to chronic stress than to episodicstress during the early third trimester of gestation. Itis also possible that measures of neuroendocrineparameters at points in time more temporally prox-imal to the occurrence of the stressful life events mayprovide better estimates of the hypothesized lifeevent stress-ACTH relationship. Second, most hu-man studies of neuroendocrine response to psycho-social stress have used adrenal hormones (cortisol orcatecholamines) as indices of stress, and have in-



TABLE 5. Stepwise Regression Analyses Predicting Neuroendocrine Parameters

Step 1 • Sociodemographic variablesF(2,48)R2

Adjusted R2

Step 2: Personality variablesF(5,45)R2

Adjusted R2

Step 3: Stress and social support variablesF(10,40)R2

Adjusted R2

ACTH

7.69*".24.21

3.39***.27.19

3.78*".49.36

BE

1.26.05.01

2.02 +

.18

.09

1.18.23.03

Cortisol

2.72+

.10

.06

2.15+

.19

.10

1.76+

.31

.13

DisregulationIndex

3.30*.12.08

1.87.17.08

2.40*.38.22

p < .10; * p < .05; *• p < .01; *** p < .001.

ferred the involvement of higher centers such as thepituitary (ie, ACTH] in the stress response (eg, 98-105). However, the present finding suggests that itmay be important to use multiple measures of neu-roendocrine activity, rather than relying only oncortisol levels, in studies of stress and pregnancy.

The present study is the first to examine theassociation between social support and neuroendo-crine function in human pregnancy. Although asmall animal literature (eg, 106-111) suggests thatthe presence of familiar co-species (usually thoughtreflective of social support), reduces HPA responseto stress, a review of the relevant literature since1975 found only one recent report (112) of therelationship between social support and neuroendo-crine function in human subjects. In a sample ofaging adults, this study found that for men, emo-tional support had the strongest associations withlower levels of norepinephrine, epinephrine andcortisol, and that for women, married women hadsignificantly lower epinephrine levels than unmar-ried women. In the present study, perceived socialsupport had a significant association with each ofthe three HPA hormones, while pregnancy-specificenacted support was significantly associated withACTH and cortisol. Subjects who reported higherlevels of social support availability had significantlylower plasma concentrations of ACTH, )3E, and cor-tisol. Thus, this finding is consistent with the argu-ment that one of the mechanisms linking socialsupport and health may be physiological (113).

Some of the personality measures were also re-lated to maternal neuroendocrine levels. For exam-ple, subjects reporting more Type A behavior hadhigher plasma concentrations of j3E, and subjectshigh in psychological hardiness had lower levels ofcortisol. Previous research has established a relation

between components of the type A behavior patternand hypertension and cardiac reactivity (114-117).Other research has established the relation betweenj3E and fetal hypoxia in pregnancy (118-119). Thepresent finding may lend tentative support for aputative mechanism linking Type A behavior duringpregnancy with fetal hypoxia and its sequelae, in-cluding fetal growth restriction and low birthweight.

In addition to the association between maternalpsychological state and plasma concentrations ofACTH, /3E, and cortisol, the present study foundprenatal psychosocial factors were associated withdisregulation of the normal relationship betweentwo POMC products, ACTH and J3E. The computa-tion of the disregulation index in the present reportto quantify the magnitude of the relationship ofACTH and )3E offers an important advantage overconventional measures of concentrations, or levels,of individual stress hormones (40, 120). It offers anew metric to facilitate assessment of HPA axis andplacental feedback and control systems. Subjectsreporting greater perceived stress had smallerACTH-/3E disregulation, while subjects reportingmore social support or those who scored higher onthe type A personality dimension had greaterACTH-/3E disregulation. Consistent with our stressand ACTH-/3E disregulation findings, the studiesthat have measured both ACTH and /3E (eg, 92-94,121—122) have concluded that various forms of stressincrease the association (ie, is associated withsmaller disregulation) of these peptides. The precisemechanism of uncoupling between ACTH and j3E isunknown, and may reflect differences in pre- orposttranslational POMC processing or in rate ofdegradation of ACTH and j8E. Recent evidence sug-gests ACTH-/3E uncoupling could result from differ-


P. D. WADHWA et al.

ential proteolytic processing of the POMC molecule(123, 124). In addition, the differential processing ofPOMC in pregnancy may be related to the expressionof POMC in the placenta (61). Further, the directionand magnitude of this differential processing ofPOMC may be determined by qualitative factors inthe environment (33). The present finding may besignificant because the dynamics of the POMC sys-tem have been shown to exert a critical influence onembryonic and fetal development. The POMC sys-tem is one of the earliest peptidergic systems toappear during mammalian CNS development (125),and POMC-derived peptides have been specificallyimplicated in early neurogenesis (126, 127) andorganogenesis (128, 129).

Although the present study was not designed totest the proposed mediating role of the neuroendo-crine system in the relationship between maternalpsychosocial factors and pregnancy outcome, thesefindings have implications for this biopsychosocialmodel of pregnancy. Placental stress hormones mod-ulate the synthesis of placental progesterone, estro-gen, and prostaglandins and control adrenal steroi-dogenesis, and are thereby implicated in themaintenance of pregnancy, fetal growth and matura-tion, and the timing of delivery (36, 62, 130, 131).Increases in maternal ACTH concentrations afterexperimental induction of prenatal stress have beenassociated with prematurity in sheep (132), fetalHPA activation in rats (23), and low birth weight andimpaired infant neurodevelopment and immuno-competence in non-human primates (20, 26). Stress-induced alterations of the normal trajectory of ma-ternal-placental-fetal neuroendocrine hormonesover the course of gestation may cause co-release andfurther elevation of other related HPA and placentalproducts including CRH, oxytocin, and prostaglan-dins, which, in higher concentrations, are known tocontribute directly to preterm labor and prematuredelivery (36, 46, 133, 134). Type A behavior pattern,through its association with plasma /3E, may exacer-bate the effects of vasoconstriction and hypoxia onfeta] development and may thereby contribute tofetal growth restriction and low birth weight. Dis-regulation of the ACTH-/3E co-release pattern mayinfluence early embryogenesis (135) and has beenshown recently to alter pain sensitivity and influ-ence the use of anesthesia during vaginal delivery(40). Finally, prenatal social support may downregu-late the maternal-placental-fetal neuroendocrine sys-tem and may thereby influence fetal developmentand contribute to more optimal pregnancy outcomes.At the present time these remain speculations, but

our results offer a basis for further investigation inhuman pregnancy.

The relationships between prenatal psychosocialfactors, neuroendocrine parameters and birth out-comes are complex. For instance, we reported in anearlier paper (18) that pregnancy anxiety and lifeevent stress over the first two trimesters of preg-nancy were prospectively associated with gesta-tional age at birth and infant birth weight, respec-tively. The significant association betweenpregnancy anxiety and ACTH in the present report isconsistent with an overall mediational model, butwe found no evidence of an association between lifeevent stress over the first two trimesters of gestationand third trimester maternal neuroendocrine levels.It is conceivable that the influences of stress-relatedneuroendocrine dysfunction on birth outcomes maybe modulated by the nature of the stressor (episodicvs. chronic stress) and its time of occurrence duringpregnancy (early, middle, or late gestation). Sucheffects may also be outcome-specific, and may showa dose-response relationship with outcome. Serialneuroendocrine assessments conducted over theearly, middle, and late course of pregnancy mayfacilitate a more precise examination of the magni-tude and duration of the association between prena-tal psychosocial factors and maternal-placental-fetalneuroendocrine activity.

In addition to their possible influences on birthoutcomes, stress-related neuroendocrine parametersduring pregnancy may directly influence the devel-opment of the fetal central nervous system. A largenumber of experimental studies using animal mod-els (see 22, 136-138 for reviews) have demonstratedthat prenatal stress and maternal neuroendocrineresponses to stress during critical periods of fetaldevelopment are causally associated with perma-nent changes in fetal/infant brain morphology (re-ceptor number, distribution, density), physiology(HPA reactivity, neurochemical levels), and function(emotionality, sexual activity, neuromotor function,learning) (21-30). These influences on the fetal ner-vous system may not be exhibited at birth but atsubsequent stages of infant growth and develop-ment. Moreover, because many of these events tran-spire in utero, exposure of the immature and devel-oping fetal brain to maternal neuroendocrinechanges may have larger and longer-lasting effectsthan similar alterations in neonates or adults.

In conclusion, the present findings are consistentwith the premise that maternal-placental-fetal neu-roendocrine parameters are significantly associated,both in magnitude and specificity, with features ofthe maternal psychological and social environment



despite the systemic alterations associated with theendocrinology of pregnancy. These findings providea basis for further investigations of the role of theneuroendocrine system as a putative mediating path-way between prenatal psychosocial factors and birthoutcome, and possibly also as a mechanism linkingfeatures of the maternal psychosocial environmentto fetal/infant brain development. Future researchefforts to enhance the understanding of biopsycho-social interactions in pregnancy may be well servedby serial assessments of the reactivity of the mater-nal-placental-fetal neuroendocrine axis to variousenvironmental conditions at multiple times over thecourse of human gestation.

This work was supported in part by United StatesPublic Health Service Grants ROl HD-28413 andP30 HD-28202.

REFERENCES

1. Creasy RK: Preterm labor and delivery. In Creasy RK andResnik R (eds), Maternal Fetal Medicine: Principles andPractice. Philadelphia: WB Saunders, 1994

2. Creasy RK, Merkatz IR: Prevention of preterm birth: Clinicalopinion. Obstet Gynecol 76:(1 Suppl):2-4, 1990

3. Knoches AM, Doyle LW: Long-term outcome of infants bornpreterm. Bailliere's Clin Obstet Gynecol 7:633-651, 1993

4. Institute of Medicine: Preventing low birth weight. Washing-ton DC: National Academy Press. 1985

5. Main DM, Gabbe SG: Risk scoring for preterm labor: Wheredo we go from here? Am J Obstet Gynecol 157:789-793,1987

6. Shiono PH, Klebanoff MA: A review of risk scoring forpreterm birth. Clin Perinatol 20:107-125, 1993

7. Lobel M: Conceptualizations, measurement, and effects ofprenatal maternal stress on birth outcomes. J Behav Med17:225-272, 1994

8. McLean DE, Hatfield-Timajchy K, Wingo PA, et al: Psycho-social measurement: Implications for the study of pretermdelivery in black women. Am J Prevent Med 9:38-81, 1993

9. Paarlberg KM, Vingerhoets AJ, Passchier J, et al: Psychoso-cial factors and pregnancy outcome: A review with emphasison methodological issues. J Psychosom Res 39:563-595,1995

10. Collins NL, Dunkel-Schetter C, Lobel M, et al: Social supportin pregnancy: Psychosocial correlates of birth outcomes andpostpartum depression. J Pers Soc Psychol 65:1243-1258,1993

11. Hedegaard M, Henriksen TB, Sabroe S, et al: Psychologicaldistress in pregnancy and preterm delivery. Br Med J 307:234-239, 1993

12. Homer CJ, James SA, Siegel E: Work-related psychosocialstress and risk of preterm, low birth weight delivery. Am JPublic Health 80:173-177, 1990

13. Lobe! M, Dunkel-Schetter C, Scrimshaw SCM: Prenatal ma-ternal stress and prematurity: A prospective study of socio-economically disadvantaged women. Health Psychol 11:32-40, 1992

14. Lou HC, Nordentoft M, Jensen F, et al: Psychosocial stressand severe prematurity. Lancet 340:54, 1992

15. Pagel MD, Smilkstein G, Regen H, et al: Psychosocial influ-ences on new born outcomes: A controlled prospectivestudy. Soc Sci Med 30:597-604, 1990

16. Pascoe JM, Chessare J, Baugh E, et al: Help with prenatalhousehold tasks and newborn birth weight: Is there anassociation? J Dev Behav Pediatr 8:207-212, 1987

17. Rothberg AD, Lits B: Psychosocial support for maternalstress during pregnancy: Effect on birth weight. Am J ObstetGynecol 165:403-407, 1991

18. Wadhwa PD, Sandman CA, Porto M, et al: The associationbetween prenatal stress and infant birth weight and gesta-tional age at birth: A prospective investigation. Am J ObstetGynecol 169:858-865, 1993

19. Williamson HA, LeFevre M, Hector M: Association betweenlife stress and serious perinatal complications. J Fam Prac29:489-496, 1989

20. Coe CL: Psychosocial factors and immunity in nonhumanprimates: A review. Psychosom Med 55:298-308, 1993

21. Insel TR, Kinsley CH, Mann PE, et al: Prenatal stress haslong-term effects on brain opiate receptors. Brain Res 511:93-97, 1990

22. McEwen BS: Steroid hormones: Effect on brain developmentand function. Horm Res 37(Suppl 3):l-10, 1992

23. Ohkawa T, Rohde W, Takeshita S, et al: Effect of an acutematernal stress on the fetal hypothalamo-pituitary-adrenatsystem in late gestational life of the rat. Exp Clin Endocrinol98:123-129, 1991

24. Peters DAV: Maternal stress increases fetal brain and neona-tal cerebral cortex 5-hydroxytryptamine synthesis in rate: Apossible mechanism by which stress influences brain devel-opment. Pharmacol Biochem Behav 35:943-947, 1990

25. Sandman CA, Yessaian N: Persisting subsensitivity of thestriatal dopamine system after fetal exposure to beta-endor-phin. Life Sci 39:1755-1763, 1986

26. Schneider ML, Coe CL, Lubach GR: Endocrine activationmimics the adverse effects of prenatal stress on the neuro-motor development of the infant primate. Dev Psychobiol25:427-439, 1992

27. Takahashi LK, Kalin NH: Early developmental and temporalcharacteristics of stress-induced secretion of pituitary-adre-nal hormones in prenatally stressed rat pups. Brain Res558:75-78, 1991

28. Takahashi LK, Kalin NH, Barksdale CM, et al: Stressorcontrollability during pregnancy influences pituitary-adre-nal hormone concentrations and analgesic responsiveness inoffspring. Physiol Behav 42:323-329, 1988

29. Takahashi LK, Turner JG, Kalin NH: Prenatal stress altersbrain catecholaminergic activity and potentiates stress-in-duced behavior in adult rats. Brain Res 574:131-7, 1992

30. Weinstock M, Matlina E, Maor GI, et al: Prenatal stressselectively alters the reactivity of the hypothalamic-pituitaryadrenal system in the female rat. Brain Res 595:195-200,1992

31. Axelrod J, Reisine TD: Stress hormones: Their interactionand regulation. Science 224:452-459, 1984

32. Herbert TB, Cohen S: Stress and immunity in humans: Ameta-analytic review. Psychosom Med 55:364-379, 1993

33. Ur E: Psychological aspects of hypothalamo-pituitary-adre-nal activity. Bailliere's Clin Endocrinol Metab 5:79-96,1991

34. Chrousos GP, Gold PW: The concepts and stress and stresssystems disorders. J Am Med Assoc 267:1244-1252, 1992

35. McEwen BS, Stellar E: Stress and the individual: Mecha-nisms leading to disease. Arch Int Med 153:2093-2101,1993

36. McLean M, Bisits A, Woods R, et al: A placental clock


P. D. WADHWA et al.

controlling the length of human pregnancy. Nature Med1:460-463, 1995

37. Cosmi EV, Luzi G, Gori F, et al: Response of uteroplacentalfetal blood flow to stress situation and drugs. Eur J ObstetGynecol Reprod Biol 35:139-145, 1990

38. Myers RE: Maternal psychological stress and fetal asphyxia:A study in the monkey. Am J Obstet Gynecol 122:47-59,1975

39. Shepherd RW, Stanczyk FZ, Bethea CL, et al: Fetal andmaternal endocrine responses to reduced uteroplacentalblood flow. J Clin Endocrinol Metabol 75:301-307, 1992

40. Sandman CA, Wadhwa PD, Chicz-DeMet A, et al: Thirdtrimester POMC disregulation predicts use of anesthesia atvaginal delivery. Peptides 16:187-190, 1995

41. Scheinin M, Scheinin H, Ekblad U, et al: Biological corre-lates of mental stress related to anticipated cesarean section.Acta Anaesthesiol Scand 34:640-644, 1990

42. Wuitchik M, Hesson K, Bakal DA: Perinatal predictors ofpain and distress during labor. Birth 17:186-191, 1990

43. DeSouza EB: Corticotropin-releasing factor and interleukin-1receptors in the brain-endocrine-immune axis. Role in stressresponse and infection. Ann NY Acad Sci 697:9-27, 1993

44. Jemmot J, Locke S: Psychosocial factors, immunologic me-diation, and human susceptibility to infectious diseases:How much do we know? Psychol Bull 95:78-108, 1984

45. Garite TJ: Premature rupture of the membranes. In CreasyRK, Resnick R (eds), Maternal-Fetal Medicine: Principlesand Practice. Philadelphia, PA: WB Saunders, 1994

46. Lockwood CJ: Recent advances in elucidating the pathogen-esis of preterm delivery, the detection of patients at risk, andpreventative therapies. Curr Opin Obstet Gynecol 6:7-18,1994

47. Lederman R, Lederman E, Work B, et al: Anxiety andepinephrine in multiparous women in labor: Relationship toduration of labor and fetal heart rate pattern. Am J ObstetGynecol 153:820-827, 1985

48. Katz VL, Jenkins T, Haley L, et al: Catecholamine levels inpregnant physicians and nurses: A pilot study of stress andpregnancy. Obstet Gynecol 77:338-342, 1991

49. Dorn LD, Susman EJ, Petersen AC: Cortisol reactivity andanxiety and depression in pregnant adolescents: A longitu-dinal perspective. Psychoneuroendocrinology 18:219-239,1993

50. Yen SSC: Endocrinology of pregnancy. In Creasy RK,Resnick R (eds), Maternal-Fetal Medicine: Principles andPractice. Philadelphia, PA: WB Saunders, 1994

51. Frim DM, Emanule RL, Robinson BG, et al: Characterizationand gestational regulation of corticotropin-releasing hor-mone messenger RNA in human placenta. J Clin Invest82:287-292, 1988

52. Robinson BG, Emanuel EL, Frim DM, et al: Glucocorticoidstimulation expression of corticotropin-releasing hormonegene in human placenta. Proc Natl Acad Sci USA 85:5244-5248,1988

53. Chan EC, Smith R: /3-endorphin immunoreactivity duringhuman pregnancy. J Clin Endocrinol Metab 75:1453-1458,1992

54. Linton EA, Behan DP, Saphier PW, et al: Corticotropin-releasing hormone (CRH)-binding protein: Reduction in theadrenocorticotropin-releasing activity of placental but nothypothalamic CRH. J Clin Endocrinol Metab 70:1574-1580,1990

55. Castro MI, Valego NK, Zehnder TJ, et al: Bioactive-to-immu-

noreactive ACTH activity changes with the severity of stressin late-gestation ovine fetus. Am J Physiol 265:68-73, 1993

56. Waddell BJ, Burton PJ: Releasing of bioactive ACTH byperifused human placenta at early and late gestation. JEndocrinol 136:345-353, 1993

57. Goland RS, Jozak S, Conwell I: Placental corticotropin-releasing hormone and the hypercortisolism of pregnancy.Am J Obstet Gynecol 171:1287-91, 1994

58. Magiakou M, Mastorakos G, Rabin D, et al: The maternalhypothalamic-pituitary-adrenal axis in third trimester hu-man pregnancy. Clin Endocrinol

59. Wadhwa PD, Sandman CA, Chicz-DeMet AC, et al: PlacentalCRH modulates, maternal pituitary-adrenal function in hu-man pregnancy. Ann NY Acad Sci

60. Goland RS, Conwell IM, Warren WB, et al: Placental corti-cotropin-releasing hormone and pituitary-adrenal functionduring pregnancy. Neuroendocrinology 57:742-749, 1992

61. Margioris AN, Grino M, Rabin D, et al: Human placenta andthe hypothalamic-pituitary-adrenal axis. Adv Exp Med Biol245:389-398, 1988

62. Petraglia F, Volpe A, Genazzani AR, et al: Neuroendocrinol-ogy of the human placenta. Frontiers Neuroendocrinol 11:6-37, 1990

63. Chan EC, Smith R, Lewin T, et al: Plasma corticotropin-releasing hormone, j3-endorphin, and cortisol interrelation-ships during human pregnancy. Acta Endocrinol 128:339-344, 1993

64. Raisanen I: Plasma levels and diurnal variation of /3-endor-phin and fi-lipotropin and corticotropin during pregnancyand early puerperium. Eur J Obstet Gynecol Reprod Biol27:13-20, 1988

65. Wolfe CDA, Patel SP, Campbell EA, et al: Plasma corticotro-phin-releasing factor (CRF) in normal pregnancy. Br J ObstetGynaecol 95997-1002, 1988

66. Goland RS, Wardlaw SL, Blum M, et al: Biologically activecorticotropin-releasing hormone in maternal and fetalplasma during pregnancy. Am J Obstet Gynecol 159:884-890, 1988

67. Schulte HM, Weisner D, Allolio B: The corticotropin-releas-ing hormone test in late pregnancy: Lack of adrenocortico-tropin and cortisol response. Clin Endocrinol 33:99-106,1990

68. Rodbard D, Munson P, De Lean A: Improved curve fitting,parallelism testing, characterization of sensitivity and spec-ificity, validation, and optimization for radioligand assays.Radioimmunoassay Related Proced Med 1:469-504, 1978

69. Cohen S, Kamarck T, Mermelstein R: A global measure ofperceived stress. J Health Soc Behav 24:385-396, 1983

70. DeLongis A, Coyne JC, Dakof G, et al: Relationships of dailyhassles, uplifts, and major life events to health status. HealthPsychol 1:119-136, 1982

71. Derogatis LR, Lipman RS, Rickels K, et al: The HopkinsSymptom Checklist (HSCL): A measure of primary symptomdimensions. In Pichot P (ed), Psychological Measurementsin Psychopharmacology: Modern Problems in Pharmacopsy-chiatry, vol 7. Basel: Karger, 1974

72. Kobasa SC, Maddi SR, Kahn S: Hardiness and health: Aprospective study. J Pers Soc Psychol 42:168-177, 1982

73. Holmes TH, Rahe RH: The social readjustment rating scale.J Psychosom Res 11:213-218, 1967

74. Lederman RP: Psychosocial adaptations in pregnancy.Englewood Cliffs, NJ: Prentice-Hall, 1984

75. Engle PL, Scrimshaw SC, Zambrana RE, et al: Prenatal and



postnatal anxiety in Mexican women giving birth in Los 95.Angeles. Health Psychol 9:285-299, 1990

76. Cohen S, Mermelstein R, Kamarck T, et al: Measuring thefunctional components of social support. In Sarason IG, 96.Sarason BR (eds), Social Support: Theory, Research andApplications. The Hague, Netherlands: Martines Niijhoff,1985

77. Jenkins CD, Zyzanski SJ, Rosenman RH: Jenkins Activity 97.Survey. New York: The Psychological Corporation, 1979

78. Larsen RJ, Diener E, Emmons RA: Affect intensity andreactions to daily life events. J Pers Soc Psychol 51:803-814,1986 98.

79. Maddi SR, Kobasa SC: The development of hardiness. InMonat A, Lazarus RS (eds), Stress and Coping: An anthology.New York: Columbia University Press, 1991

80. Hoffman D, Abboud T, Haase H, et al: Plasma /3-endorphin 99.concentrations prior to and during pregnancy, in labor, andafter delivery. Am J Obstet Gynecol 150:492-496, 1984

81. Stalla GK, Bost H, Kaliebe T, et al: Human corticotropin- 100.releasing hormone during pregnancy. Gynecol Endocrinol3:1-10, 1989

82. Cohen J, Cohen C: Applied Multiple Regression/Correlation 101.Analysis for the Behavioral Sciences, 2nd Ed. Hillsdale, NJ:Lawrence Erlbaum Associates, 1983 102.

83. Stevens J: Applied Multivariate Statistics for the SocialSciences, 2nd Ed. Hillsdale, NJ: Lawrence Erlbaum Associ-ates, 1992 103.

84. Carr BR, Parker CR, Madden JD, et al: Maternal plasmaadrenocorticotropin and cortisol relationships throughouthuman pregnancy. Am J Obstet Gynecol 139:416-422, 1981

85. ChallisJR, Patrick JE: Changes in diurnal rhythms of plasma 104.cortisol in women during the third trimester of pregnancy.Gynecol Obstet Invest 16:27-32, 1983 105.

86. Cousins L, Rigg L, Hollingsworth D, et al: Qualitative andquantitative assessment of the circadian rhythm of cortisolin pregnancy. Am J Obstet Gynecol 145:411-416, 1983 106.

87. Jawad MJ, Wilson EA, Lagrew DC: Diurnal variation ofsteroid hormones in maternal serum during late pregnancy.Am J Obstet Gynecol 148:339-340, 1984 107.

88. Nolten WE, Lindheimer MD, Rueckert PA, et al: Diurnalpatterns in regulation of cortisol secretion in pregnancy.J Clin Endocrinol Metab 51:466-472, 1980

89. Raisanen I, Salminen K, Laatikainen T: Response of plasma 108.immunoreactive beta-endorphin and corticotropin to iso-metric exercise in uncomplicated pregnancy and in pregnan-cy-induced hypertension. Eur J Obstet Gynecol Reprod Biol35:119-24, 1990 109.

90. Gerra G, Volpi R, Delsignore R, et al: ACTH and beta-endorphin responses to physical exercise in adolescent 110.women tested for anxiety and frustration. Psychiatry Res41:179-86, 1992

91. Gerra G, Caccavari R, Delsignore R, et al: Parental divorce 111.and neuroendocrine changes in adolescents. Acta PsychiatrScand 87:350-354, 1993

92. Meyerhoff JL, Oleshansky MA, Mougey EH: Psychologic 112.stress increases plasma levels of prolactin, cortisol, andPOMC-derived peptides in man. Psychosom Med 50:295-303, 1988 113.

93. Mutti A, Ferroni C, Vescovi PP, et al: Endocrine effects ofpsychological stress associated with neurobehavioral perfor-mance testing. Life Sci 44:1831-1836, 1989 114.

94. Oltras CM, Mora F, Vives F: Beta-endorphin and ACTH inplasma: Effects of physical and psychological stress. Life Sci40:1683-1686, 1987

Cadet R, Pradier P, Dalle M, et al: Effects of prenatal maternalstress on the pituitary adrenocortical reactivity in guinea pigpups. J Dev Physiol 8:467-475, 1986Erisman S, Carnes M, Takahashi LK, et al: The effects ofstress on plasma ACTH and corticosterone in young andaging pregnant rats and their fetuses. Life Sci 47:1527-1533,1990Ohkawa T, Takeshita S, Murase T, et al: Ontogeny of theresponse of the hypothalamo-pituitary-adrenal axis to ma-ternal immobilization stress in rats. Endocr Jpn 38:187-194,1991Cummins SE, Gevirtz RN: The relationship between dailystress and urinary cortisol in a normal population: Anemphasis on individual differences. Behav Med 19:129-134,1993Fredrikson M, Sundin O, Frankenhaeuser M: Cortisol excre-tion during the defense reaction in humans. Psychosom Med47:313-319, 1985Lundberg U, Frankenhaeuser M: Pituitary-adrenal and sym-pathetic-adrenal correlates of distress and effort. J Psycho-som Res 24:125-130, 1980Mason JW: A historical view of the stress field. J Hum Stress1:22-36, 1975McCann BS, Carter J, Vaughan M, et al: Cardiovascular andneuroendocrine responses to extended laboratory challenge.Psychosom Med 55:497-504, 1993Rose RM, Fogg LF: Definition of a responder: Analysis ofbehavioral, cardiovascular, and endocrine responses to var-ied workload in air traffic controllers. Psychosom Med55:325-338, 1993Schaeffer MA, Baum A: Adrenal cortical response to stress atThree Mile Island. Psychosom Med 46:227-37, 1984Weizman R, Laor N, Barber Y, et al: Impact of the Gulf Waron the anxiety, cortisol, and growth hormone levels of Israelicivilians. Am J Psychiatry 151:71-75, 1994Gust DA, Gordon TP, Brodie AR, et al: Effect of a preferredcompanion in modulating stress in adult female rhesusmonkeys. Physiol Behav 55:681-684, 1994Gust DA, Gordon TP, Wilson ME, et al: Removal from natalsocial group to peer housing affects cortisol levels andabsolute numbers of T cell subsets in juvenile rhesus mon-keys. Brain Behav Immun 6:189-199, 1992Keverne EB, Martensz ND, Tuite B: Beta-endorphin concen-trations in cerebrospinal fluid of monkeys are influenced bygrooming relationships. Psychoneuroendocrinology 14:155—161, 1989Levine S: The influence of social factors on the response tostress. Psychother Psychosom 60:33-38, 1993Levine S, Wiener SG: Psychoendocrine aspects of mother-infant relationships in nonhuman primates. Psychoneuroen-docrinology 13:143-154, 1988Stanton ME, Patterson JM, Levine S: Social influences onconditioned cortisol secretion in the squirrel monkey. Psy-choneuroendocrinology 10:125-134, 1985Seeman TE, Berkman LF, Blazer D, et al: Social ties andsupport and neuroendocrine function: The MacArthur stud-ies of successful aging. Ann Behav Med 16:95-106, 1994Cohen S: Psychosocial models of the role of social support inthe etiology of physical disease. Health Psychol 7:269-297,1988Burg MM, Jain D, Soufer R, et al: Role of behavioral andpsychological factors in mental stress-induced silent leftventricular dysfunction in coronary artery disease. J Am CollCardiol 22:440-448, 1993


P. D. WADHWA et al.

115. Friedman HS, Booth-Kewley S: Personality, type A behavior,and coronary heart disease: The role of emotional expres-sion. J Pers Soc Psychol 53:783-792, 1987

116. Lambert WVV, MacEvoy B, Klackenberg-Larsson I, et al: Therelation of stress hormone excretion to type A behavior andto health. J Hum Stress 13:128-135, 1987

117. Niaura R, Goldstein MG: Psychological factors affectingphysical condition. Cardiovascular disease literature review.Part II: Coronary artery disease and sudden death andhypertension. Psychosomatics 33:146-155, 1992

118. Stark RI, Wardlaw SL, Daniel SS: Characterization of plasmabeta-endorphin immunoactivity in the fetal lamb: Effects ofgestational age and hypoxia. Endocrinology 119:755-761,1986

119. Wardlaw SL, Stark RI, Daniel S, et al: Effects of hypoxia onbeta-endorphin and beta-lipotropin release in fetal, new-born, and maternal sheep. Endocrinology 108:1710-1715,1981

120. Sandman CA, Wadhwa PD, Dunkel-Schetter C, et al: Psycho-biological influences of stress and HPA regulation on thehuman fetus and infant birth outcomes. Ann NY Acad Sci739:198-210, 1994

121. Mauri A, Serri F, Caminiti F, et al: Correlation betweenamniotic levels of alpha-MSH, ACTH and beta-endorphin inlate gestation and labour in normal and complicated preg-nancies. Acta Endocrinol 123:637-642, 1990

122. Odagiri E, Adachi T, Shirota M, et al: The changes of humanplasma ACTH, immunoreactive beta-endorphin and cortisolduring pregnancy, labor, and delivery. Nippon Sanka Fu-jinka Gakkai Zasshi. Acta Obstet Gynecol Jpn 34:920-928,1982

123. Benjannet S, Rondeau N, Day R, et al: PCI and PC2 areproprotein convertases capable of cleaving pro-opiomelano-cortin at distinct pairs of basic residues. Proc Natl Acad SciUSA 88:356-368, 1991

124. Friedman TC, Peng Loh Y. Birch NP: In vitro processing ofpro-opiomeilanocortin by recombinant PCI, SPC3. Endocri-nology 135:854-862, 1994

125. Khachaturian H, Lewis ME, Alessi NE, et al: Time of originof opioid peptide-containing neurons in the rat hypothala-mus. J Comp Neurol 236:538-546, 1985

126. Bartolome JV, Bartolome MB, Daltner LA, et al: Effects of(3-endorphin on ornithine decarboxylase in tissues of devel-

oping rats: A potential role for this endogenous neuropep-tide in the modulation of tissue growth. Life Sci 38:2355-2362, 1986

127. Ilyinsky OB, Kozlova MV, Kondrikova ES, et al: Effect ofopioid peptides and naloxone on nervous tissue in culture.Neuroscience 22:719-735, 1987

128. Cossu G, Cusella-DeAngelis MG, Senni MI, et al: Adrenocor-ticotropin is a specific mitogen for mammalian myogeniccells. Dev Biol 131:331-336, 1989

129. Estivariz FE, Carino M, Lowrey PJ, et al: Further evidencethat N-terminal pro-opiomelanocortin peptides are involvedin adrenal mitogenesis. J Endocrinol 116:201-206, 1988

130. Jones SA, Challis JRG: Steroid, corticotropin-releasing hor-mone, ACTH, and prostaglandin interactions in the amnionand placenta of early pregnancy in man. J Endocrinol 125:153-159, 1990

131. Petraglia F, Angioni S, Coukos G, et al: Neuroendocrinemechanisms regulating placental hormone production.Contr Gynecol Obstet 18:147-56, 1991

132. Maclsaac RJ, Carson RS, Horvath AP, et al: Premature birthof live lambs induced by pulsatile infusion of ACTH. JEndocrinol 124:99-107, 1990

133. Goland RS, Jozak S. Warren WB, et al: Elevated levels ofumbilical cord plasma corticotropin-releasing hormone ingrowth-retarded fetuses. J Clin Endocrinol Metab 77:1174-1179, 1993

134. Warren WB, Patrick SL, Goland RS: Elevated maternal andplasma corticotropin-releasing hormone levels in pregnan-cies complicated by preterm labor. Am J Obstet Gynecol166:1198-1207, 1992

135. Rius RA, Chikuma T, Loh YP: Prenatal processing of pro-opiomelanocortin in the brain and pituitary of mouse em-bryos. Dev Brain Res 60:179-185, 1991

136. Dorner G: Hormone-dependent brain development and neu-roendocrine prophylaxis. Exp Clin Endocrinol 94:4-22,1989

137. Lauder JM: Hormonal and humoral influences on braindevelopment. Psychoneuroendocrinology 8:121-155, 1983

138. Silverman A, Hou-Yu A, Kelly DD: Modification of hypotha-lamic neurons by behavioral stress. In Tache Y, Morley JE,Brown MR, eds, Neuropeptides and Stress. New York: Sprin-gier-Verlag, 1989


Documents

Prenatal Psychosocial Factor ands the Neuroendocrine Axi