Maternal stress and placental function, a study using questionnaires and biomarkers at birth.

The prevalence of psychosocial stress and anxiety during pregnancy is a cause of concern worldwide, and maternal psychosocial stress has been conceptualized as a teratogen [ 1 ]. Maternal and child mental health is addressed in the WHO Mental Health Action Plan (2013–2020), in which it is estimated that about 10% of pregnant women have experienced mental health problems, primarily depression, and this percentage rises to 15.6% when only considering developing countries [ 2 ]. A Swedish study found that 20–35% of pregnant women evaluated their own emotional and physical health as poor [ 3 ], while 29% of the participating mothers in a survey of a pregnant population from the Copenhagen University Hospital answered “yes” when asked whether they were feeling anxious, stressed or depressed and/or had experienced a traumatizing life event during their pregnancy (unpublished data from LM, n = 48). Several studies have linked maternal psychosocial stress during pregnancy to a range of effects, both in pregnancy, showing effects on incidence of pre-eclampsia [ 4 ] and placental weight [ 5 ], in newborn offspring in the infant cortisol response [ 6 ] and also later in childhood when studying stress reactivity [ 7 ], offspring pediatric disease [ 8 ] and asthma and atopic dermatitis [ 9 ]. Epidemiological studies of the effects of stress during pregnancy have focused on neurological effects and changes in behavior [ 6 ; 10 ; 11 ], and immune system effects such as asthma and allergies [ 9 ; 12 ; 13 ]. Elevated cortisol levels in plasma have been used as a biomarker in diagnosis of chronic stress [ 14 ]. In pregnant women the plasma concentrations of cortisol are elevated by the feed forward mechanism of the corticotropin-releasing hormone produced by the placenta, and elevated plasma cortisol is therefore a poor biomarker during pregnancy [ 15 ; 16 ]. During the first two trimesters of pregnancy, the serum cortisol levels in the fetus are low, except for a peak around gestation week 10 to counteract the effects of hormones from newly formed fetal adrenal tissue. In the third trimester, the fetal serum cortisol levels rise, and is at a maximum at term due to a decline in the placental cortisol metabolizing activity, and the production of cortisol by the fetal adrenal glands [ 17 ]. High levels of cortisol during pregnancy have anti-inflammatory and catabolic properties in both the fetus and the pregnant woman. The mechanisms of maternal psychosocial stress affecting the fetus during pregnancy are assumed to be regulated by placental transfer of hormones, through changes in the expression of placental receptors and enzymes (for reviews, see [ 18 ; 19 ]). The placental barrier consists of different cell layers in the human placenta and the placental cell layer most representative of the placental transport and metabolism of hormones is the syncytiotrophoblast. This cell layer expresses the enzyme 11β-HSD2, expressed in tissues that require protection from cortisol, which transforms 80–90% of the maternal cortisol to cortisone passed on to the umbilical and fetal blood [ 17 ; 20 ]. The activity of this enzyme has been linked to the effect of maternal psychosocial stress on the offspring, as the metabolic activity of this enzyme protects the fetus from the high maternal cortisol plasma levels (for a review see [ 21 ]). The placental gene and mRNA expression of 11β-HSD2, as well as the epigenetic methylation deactivation of 11β-HSD2, has been studied in relation to maternal stress [ 10 ; 22 ; 23 ]. The activity of 11β-HSD2 can be studied by comparing cortisol and cortisone concentrations in maternal and fetal blood. Previously, the ratio of maternal cortisol to fetal cortisol and the fetal or maternal cortisol-cortisone ratio have been used as a measure of activity of this enzyme [ 24 – 26 ]. Most studies have been performed in populations of pregnant women who have been diagnosed with depression or have other mental health problems. In contrast, the stress variables in our study are self-reported experiences of depression, anxiety and stress during pregnancy, measured using questionnaires, including selected personality traits of respondents, as these traits can affect the experience of and emotional reactions to stress exposure [ 27 ; 28 ]. The aim of the Maternal Stress and Placental Function project is to investigate the effect of prenatal maternal psychosocial stress on the adjusted fetal cortisol exposure (AFCE) in a normal pregnant population. AFCE represents the relative amount of cortisone produced by the placenta measured in fetal blood in relation to how much cortisol the placenta has let pass un-metabolized from maternal blood, corrected for the interindividual differences of these hormone levels in the population–i.e. the placental exposure. The AFCE gives us a measure of the activity of the enzyme 11β-HSD2 until the time of birth. An increase in the AFCE represents a relative increase in fetal cortisol exposure.

Descriptive data, lifestyle, self-reported health and birth-related outcomes are shown in Table 2 . Caesarean sections were the birth method of 151 (55%) women. Seventy (26%) pregnant women reported having one or more chronic diseases, primarily asthma or allergies and various metabolic, gastrointestinal and dermatological diseases (see Table 2 ). Correlations between maternal and fetal levels of cortisol and cortisone are shown in Fig 1 . The y-axis is log-scale. Both fetal and maternal plasma cortisone (blue circle) and fetal and maternal cortisol (orange square) values correlate. The results from DASS-42, PRA and Major Life Events questionnaires are shown in Fig 2 . The study population divided into groups according to severeness of maternal (A) Depression, (B) Anxiety and (C) Stress based on the Depression Anxiety and Stress Scales, (D) Major Life events and (E) Pregnancy related anxiety (PRA). Results from the regression analysis (10^β (CI) and p-value for each category of DASS-42, PRA and Major Life Events) are presented in Table 3 . Outcomes from the DASS were analyzed as dichotomous “normal” and “increased” due to small group sizes in the “moderate” and “severe” categories (see Fig 2 ). The data are presented as 10^β representing the factor by which the mean AFCE is multiplied for each unit increase in exposure. a Unadjusted ᵇ Adjusted for age, BMI, parity, neuroticism upper quartile and conscientiousness lower quartile ᶜ Adjusted for smoking and alcohol during pregnancy ᵈ Adjusted for asthma medication, chronic disease and gestational complications ᵉ Adjusted for gestational age, gender, delivery mode, birth strain, placenta weight and symmetry, and time from delivery to maternal blood samples. *statistically significant. A test of correlation between the primary outcome and time from delivery to maternal and fetal blood sample collection was performed using the Pearson Correlation and showed positive correlation between time from delivery to maternal blood sample and log AFCE (r = 0.16, p<0.01). The maternal time variable was therefore included in step 5 of the regression model. Significant results were seen for the primary outcome AFCE and PRA with approximately 1.5 times increase in AFCE for a unit increase in PRA in models 1 (10^β = 1.349, p = 0.032), and when adjusting for personal traits in model 2 (10^β = 1.455, p = 0.013), lifestyle in model 3 (10^β = 1.462, p = 0.012) and health in model 4 (10^β = 1.466, p = 0.013), but the increase was no longer statistically significant when adjusting for delivery factors in model 5 (10^β = 1.315, p = 0.080) (see Table 3 ). When stratifying data according to delivery mode, the effect of PRA on AFCE was found primarily in vaginal delivery (model 5: elective Caesarean section: 10^β = 1.102, p = 0.638, vaginal delivery: 10^β = 1.549, p = 0.050) (See S2 Table ). No other exposure variable showed any significant effect on AFCE. For all exposure variables ( state -depression, anxiety and stress, PRA and Major life events) in model 5, the possible confounders showing a significant effect on AFCE were delivery mode (10^β = 0.49, p = 0.012, less AFCE in caesarean section), birth strain (10^β = 1.77, p = 0.034, more AFCE with higher birth strain) and placental weight (10^β = 1.42, p = 0.020, more AFCE in larger placentas) (see S3 Table ). As with AFCE, the only exposure variable showing significant effects in models 1–4 was PRA, which was significantly associated with maternal plasma cortisol (10^β = 0.72, p = 0.022 in model 4) and fetal plasma cortisone (10^β = 0.74, p = 0.024 in model 4), but not in model 5 when adjusting for delivery variables. Like for AFCE, a significant confounder in the exposure effect relationship was delivery mode for all exposure variables, in maternal cortisol (10^β~3.3, p<0.000), and fetal cortisol (10^β~3.1, p<0.000) and cortisone (10^β~4.5, p<0.000). For the two fetal outcomes fetal cortisol and fetal cortisone, there was a significant effect of parity; more fetal cortisol (10^β~1.3, p~0.004) and fetal cortisone (10^β~1.3, p~0.004) in first pregnancies. Cortisone in both maternal and fetal plasma was significantly correlated with maternal age (10^β~0.6, p~0.001) and (10^β~0.8, p~0.03). In maternal cortisol and fetal cortisone there was an effect of time from delivery to maternal blood sample, with a reverse correlation (10^β~0.6, p<0.000) and (10^β~0.8, p~0.03). The only effect of personality was seen in maternal cortisone, were there was a significant effect of conscientiousness , with an increased content of cortisone with increased conscientiousness (10^β~1.4, p~0.03).

Associations between Pregnancy-Related Anxiety (PRA) and adjusted fetal cortisol exposure (AFCE) were seen in our study population of 273 mother-fetus dyads, and these associations were strongest in the vaginal delivery group. The AFCE was significantly associated with placental weight independently of the individual plasma hormone levels, which supports that the AFCE is a measure of placental function. The PRA questionnaires were shown to be more sensitive to the target population than the DASS-42 questionnaire, which demonstrates the importance of validating stress scales to be used in a population of pregnant women.

The current study is a part of the Maternal Stress and Placental Function project, conducted in Copenhagen, Denmark. Participants were pregnant women giving birth at Copenhagen University Hospital, where the Department of Obstetrics has around 6000 births a year, of which approximately 22% are Caesarean Sections. Patients admitted to the department are healthy women and women with medical and obstetric complications, as well as women with psychosocial problems. We aimed to investigate the normal population and the criterion for exclusion was age less than 18. The project was approved by the Regional Scientific Ethical Committee of Copenhagen (H-15006254) and the Danish Data Protection Agency (2015-41-4208). All women were informed about the aim of the study and gave written informed consent. Recruitment was carried out at four locations connected to the hospital: at the information meetings for all pregnant women; at information meetings for women giving birth to their first child; in the waiting room at the midwives’ offices; and at the information meeting specifically for women giving birth by planned cesarean section. The pregnant women were instructed to answer four written questionnaires in order to measure: 1: relevant personal factors such as socioeconomic status, use of medication, smoking, and alcohol consumption during and before pregnancy, 2: pregnancy-related anxiety, 3: personality, and 4: prevalence of prenatal depression, anxiety and stress. Over a period of 11 months from June 2015 to May 2016, out of 2058 invited families, 562 decided to participate in the study. Only participants with returned questionnaires and successful blood sampling from both mother and umbilical cord directly after birth were included in this study, resulting in 273 participants. Sampling was primarily conducted from Monday to Sunday in the timeframe 7 am to 8 pm. Information regarding lifestyle, BMI, parity, chronic illness, pregnancy complications and medication used during pregnancy was obtained via self-reported questionnaires, and variables connected to the birth and the infant were collected via hospital records. A variable representing the strain of birth (birth strain) was constructed taking into account the length of active labor (vaginal birth 1 point, active labor >12 hours: 2 points, pushing contractions>1 hour: 2 points), augmentation of labor using synthetic oxytocin: (1 point), the use of pain alleviation (non-medical: 1 point or medical: 2 points) and the interventions used during delivery (forceps or vacuum assisted: 1 point or acute section: 2 points). Resulting in a birth strain score of 0 points for elective caesarean section, and a possible score of 1 to 10 points for vaginal births. Placental symmetry was calculated as the measure of the widest place on the placental diameter minus the shortest place on the placental diameter. Maternal state stress was defined as the individual degree of depression, anxiety and stress experienced during the pregnancy, pregnancy- and birth-related thoughts and anxiety, and the experience of major life events during pregnancy. These were assessed using the Depression Anxiety Stress Scales (DASS), Pregnancy Related Anxiety (PRA), and a Major Life Events question. DASS (DASS-42 translated to Danish by Dr Mikael Thastum from the University of Aarhus) contains 42 questions, with depression, anxiety and stress represented by 14 items each [ 29 ]. Items were scored 0 to 3 and total scores for each condition were categorized into normal, mild, moderate and severe according to the DASS manual [ 30 ]. PRA were assessed using a Danish translation of the 10-item Pregnancy-Related Thoughts (PRT) questionnaire [ 31 ] and an additional four items (Birth-Related Thoughts, BRT) (see S1 Table ) used by the Copenhagen University Hospital to screen for severely anxious pregnant women. The PRT questions were answered on a four-point Likert scale rating from “not at all (1)” to “very much (4)” and the BRT questionnaire had a five-point scale from 0 to 5. The BRT was transformed and integrated into the PRT with an acceptable internal consistency (α = 0.79), producing a measure of Pregnancy-Related Anxiety (PRA). PRA was categorized into four groups based on total scores in this study: 10–19 defined “no PRA”, 20–25 “some PRA”, 26–31 “moderate PRA” and scores more than 31 “high PRA”. The Major Life Events question, “ have you experienced any major life events during your pregnancy that have led to changes in your state of mind ?”, was answered “yes/no” and defined with examples from the Holmes-Rahe stress inventory of Major Life Events [ 32 ]. Maternal trait stress was assessed using the Danish version of the NEO-FFI inventory, a well-known measure of the “big five” personality traits of Neuroticism, Extraversion, Openness, Conscientiousness and Agreeableness [ 33 ]. The two traits of Neuroticism and Conscientiousness were individually included in the analyses as possible confounders. A high score for Neuroticism was defined as the upper 25% of scores (upper quartile) and a low score on Conscientiousness as the lowest 25% of scores (lower quartile). The participants donated a maternal 20 ml blood-sample and 20–30 ml umbilical cord blood directly after birth. Blood was sampled by venipuncture into 10 ml vacuum tubes containing EDTA, centrifuged for 10 minutes at 4000g without brake and the plasma was collected and stored at -80°C until analysis. All sampling and handling times and times of freezing the samples were noted. The quantification method of cortisol and cortisone is described in detail in the supporting material ( S1 File ). In brief: for each analyte, specific isotopically labelled internal standards (IS) (cortisol-d 4 for cortisol and cortisone-d 8 for cortisone) were used. The working solution of IS was prepared fresh and consisted of 25 mL 50 mM sodium diphosphate dibasic pentahydrate, pH unadjusted and 1.5 mL of IS mix from Chromsystems. The simplified liquid extraction method was adapted. Sample analysis was performed using a Waters (Milford, MA, USA) Acquity UPLC system with a Kinetex 2.6 μm EVO C18 column (100Å 100x2.1 mm; Phenomenex, Torrance, CA, USA). Column temperature was 50°C, flow rate was 500 μl/min, and injection volume was 5 μl. The total analysis time was 9 minutes per sample. The mobile phase was a gradient of a mixture of an aqueous mobile phase 0.1% NH 4 OH (v/v) in water (mobile phase A) and an organic phase containing 0.1% NH 4 OH (v/v) in MeOH (mobile phase B). The passing criteria were defined by measuring the concentration of cortisol and cortisone in QC samples: In each batch, a set of matrix match QC samples were run in order ensure the validity of the batches (low and high). The low control contained cortisol and cortisone at concentrations of 73 nmol/L and 5,5 nmol/L respectively. The high control contained cortisol and cortisone at concentrations of 494 nmol/L and 81,5 nmol/L respectively. If the high controls were within ±15% of the target values and low controls were within ±20% of the target value the batch was considered acceptable. The QC values for the batches that were deemed acceptable the QC values were within ±10% of the target value. The primary outcome was cortisol-cortisone ratio between umbilical cord (fetal) blood and maternal blood, referred to as AFCE (adjusted fetal cortisol exposure): A F C E = F e t a l c o r t i s o l c o r t i s o n e M a t e r n a l c o r t i s o l c o r t i s o n e Maternal and fetal plasma cortisol and cortisone concentrations were included as secondary outcomes. In order to test for normality in the distribution of cortisol, cortisone and AFCE levels, the Shapiro-Wilk test was performed and yielded a non-normal distribution. Log-transformation was performed and only log fetal cortisol did not obtain normal distribution. All analyses were performed using linear regression, with stepwise adjustment for confounders and covariates. The associations were assessed in five models. Model 1 was unadjusted, and the other models were adjusted with an increasing selection of potential confounders and intermediate variables: Model 2: personal traits (maternal age, BMI before pregnancy (categorized 1: underweight 30kg/m 2 ), parity (primipartum) including maternal trait stress (neuroticism (upper quartile) and conscientiousness (lower quartile)), Model 3: lifestyle (smoking during pregnancy, alcohol during pregnancy), Model 4: health (chronic disease, gestational complications (normal birth), asthma medication) and Model 5: delivery (gestational age, gender, mode of delivery (caesarean section or vaginal birth), birth strain, placental weight (log), placenta symmetry (log), time from delivery to maternal blood sample) (see Table 1 ). The measure of effect is the exponentiated regression coefficient of the exposure– 10^ β–which, because of the log-transform of the outcome, is the factor by which the outcome is multiplied for a unit increase of the exposure. a DASS-42 is the results from the depression anxiety stress scales. b PRA is the results from the questionnaire on pregnancy related anxiety. c AFCE is the adjusted fetal cortisol response. ANOVA was used to test the differences in hormone levels and AFCE between vaginal birth and elective caesarean groups, when stratifying the data by birth mode. All statistical modeling and testing was performed in IBM SPSS Statistics 24. Descriptive results are presented categorically with frequency tables and numerically as mean ±SD. A p-value less than 0.05 is deemed statistically significant.

Questions about Birth-Related Thoughts, used by Copenhagen University Hospital to screen for anxious pregnant women. Tick a number next to each statement to show how much concern you feel at the moment (only one number for each line). (DOCX) Click here for additional data file. Regression coefficients (10^β) and levels of significance (p-value) in the regression analysis of AFCE and state stress exposures stratified by birth mode. Regression coefficients marked *statistically significant, # borderline statistically significant. (DOCX) Click here for additional data file. Regression coefficients (10^β) and levels of significance (p-value) from step five in the regression analysis of AFCE and state stress exposures including all co-variables. (DOCX) Click here for additional data file. (DOCX) Click here for additional data file. (XLSX) Click here for additional data file.