The effect of tobacco exposure on the fetal hypothalamic–pituitary–adrenal axis


Dr SD McDonald, McMaster University, Division of Maternal–Fetal Medicine, Department of Obstetrics and Gynecology, 1200 Main St West, HSC 3N52B, Hamilton, Canada ON L8N 3Z5. Email


Objective  Our objective was to determine if maternal smoking is associated with programming of the fetal hypothalamic–pituitary-adrenal (HPA) axis. Cigarette smoking, which induces a state of hypoxia in the fetus, may promote in utero‘programming’ of the HPA axis. In utero, adaptations to the HPA axis, which become maladaptive later in life, have been hypothesised to contribute to the development of adult cardiovascular disease and metabolic disorders.

Design  This was a prospective cohort study of term infants.

Population and setting  The study involved 104 infants born by elective caesarean section, 21 of whom were exposed to in utero tobacco and 83 were nonexposed.

Methods  Healthy women with healthy pregnancies were recruited if they were undergoing elective caesarean section. Maternal blood was drawn for cortisol and cotinine in the morning, and the umbilical blood was drawn immediately after delivery of the baby.

Main outcome measures  Umbilical arterial cortisol and adrenocorticotropin hormone (ACTH) levels.

Results  ACTH levels were significantly elevated in smoke-exposed infants [17 (4–22) pmol/l versus 4 (2–11) pmol/l, respectively, P= 0.005], while cortisol levels were similar [182 (130–240) nmol/l versus 192 (127–265) nmol/l, respectively, P= 0.541].

Conclusions  For the first time, it was shown that infants exposed to in utero tobacco smoke have significantly elevated ACTH levels compared with nonexposed infants. The results of this study warrant further exploration of the effect of smoking on the neonatal HPA axis as a potential set up for ‘programming’.


There is increasing evidence in both animals and humans associating low birthweight with later cardiovascular and metabolic disorders including insulin resistance, hypertension, and coronary artery disease.1–5 This has become known as the Barker hypothesis or fetal origins of adult disease or ‘programming’.5 It is hypothesised that the hypothalamic–pituitary–adrenal (HPA) axis is intimately involved in ‘programming,’6 adaptive changes which are made in response to in utero exposures, which subsequently become maladaptive in extrauterine life. Cigarette smoking may promote in utero‘programming’ of the HPA axis by inducing a hypoxic state.7–9 The hypoxic effect of cigarette smoking on the fetus is due to: (a) an increase in carboxyhaemoglobin, 7–8% compared with 1% in nonsmokers,7 (b) reduction of the unloading pressure both as oxygen moves across the placenta from maternal to fetal blood and again when it crosses from fetal blood into fetal tissues,8 and (c) uterine vessel vasoconstriction.9

Human studies that have previously attempted to evaluate cortisol concentrations in growth-restricted fetuses have been confounded by mode of delivery (labour increases cortisol levels,10–13 and limited by very small samples size.12–14 A small study involving only 28 growth-restricted fetuses and 28 normally grown fetuses found that growth-restricted human fetuses have elevated levels of corticotropin-releasing hormone (CRH), adrenocorticotropin hormone (ACTH), and cortisol,13 however, the specimens analysed were mixed umbilical arterial and venous blood (umbilical arterial concentrations of these hormones are higher than umbilical vein concentrations).15 Economides et al.16 found higher umbilical venous cortisol concentrations (102 nmol/l versus 74 nmol/l P < 0.001) obtained at cordocentesis in 41 severely growth-restricted babies than in 61 normally grown babies despite similar maternal cortisol levels, however, they obtained their samples over a wide span of gestation, between 18 and 38 weeks gestation, and concentrations vary with gestation. In studies in sheep, reductions in uterine artery blood flow resulting in fetal hypoxaemia caused elevations in fetal cortisol concentrations.17,18

If cigarette smoking induces a state of chronic stress in the fetus, the HPA axis may undergo in utero‘programming’. The objective of this study was to determine if smoking is associated with altered fetal HPA-axis hormones at birth.



This was a prospective cohort study designed to examine the relationship between in utero smoke exposure and cortisol and ACTH concentrations in the neonate. Women were eligible if they were healthy and undergoing elective caesarean section (caesarean section without labour) at one of two hospitals in Ontario, Canada (The Ottawa Hospital and Kingston General Hospital). The other inclusion criteria were: spinal or epidural anaesthesia, term pregnancy (38 weeks and 0 days to 41 weeks and 3 days), intact membranes, a healthy singleton fetus, and maternal blood sampling between 08:30 and 11:30 hours. Exclusion criteria included (factors which could potentially affect cortisol): (a) maternal complications [adrenal disease, diabetes mellitus, gestational diabetes mellitus, steroid use during the pregnancy, significant cardiovascular (including hypertension) or respiratory problems or renal disease, thyroid disease, infections including chorioamnionitis, surgery during the pregnancy (excluding caesarean section), alcohol abuse, illicit drug use, depression, anxiety disorder, and post-traumatic stress disorder], (b) fetal complications [known intrauterine growth restriction at any point during the gestation (<10th percentile for gestation), suspected fetal asphyxia, known congenital anomalies], and (c) uteroplacental complications (antepartum haemorrhage, abruption).

Study protocol

Information on demographical variables, pregnancy history, and delivery variables and infant characteristics were collected. The exposure, self-reported cigarette smoking (dichotomous, ‘yes’ or ‘no’), has been validated during pregnancy (correlation coefficient 0.92 between the reported number of cigarettes smoked and cotinine level in pregnant patients).19 Smoking was recorded as the average number of cigarettes smoked per day.

Maternal blood specimens for cortisol and cotinine (the major metabolite of nicotine) were collected between 08:30 and 11:30 hours from the antecubital vein using standard phlebotomy technique by trained research nurses. After the birth of the baby, umbilical cord arterial and venous blood was drawn. Cortisol and ACTH assays were performed on umbilical arterial blood, and cortisol was also assayed on umbilical venous blood.

The study protocol was approved by the institutional ethics review boards, and a written, informed consent was obtained from each participant. Recruitment occurred from May 2002 to December 2004.

Laboratory measurements

Serum cortisol was measured using electrochemiluminescence on the Roche Elecsys 2010 analyzer using the manufacturer’s reagents (Roche Diagnostics, Indianapolis, IN, USA). Plasma was stored at –70°C until batch analysis for all assays. The cortisol assay has an analytical range of 1–1750 nmol/l and recommended adult reference intervals of 171–536 nmol/l (AM) and 64–340 nmol/l (PM). Assay precision (CV) is approximately 5% across the analytical range. ACTH was measured using immunoradiometric assay using the DiaSorin ACTH IRMA (DiaSorin, Stillwater, MN, USA). Samples were transported on ice to the laboratory, centrifuged, and plasma separated within 1 hour of collection. The analytical range of this assay is 0–292 pmol/l. The assay reproducibility was +/− 5% across the analytical range.

Statistical analysis

Normalcy of distribution was checked using the Shapiro–Wilk test. Continuous normally distributed variables were compared using a t test and categorical variables using linear-by-linear association test for trend or Fisher’s exact test as appropriate. A Mann–Whitney U test was used to compare median values of variables that were not normally distributed. Correlations were examined using Spearman correlation coefficients. Backward linear regression was used to explore the relationship of dependent and independent variables.


Approximately 30% of women who presented for caesarean section were eligible, and approximately 90% of women who were eligible agreed to participate.

The cohort was composed of 21 infants of smoking mothers and 83 infants of nonsmoking mothers. Patient demographics are presented in Table 1.

Table 1.  Participant characteristics
Participant characteristicMean (SD)P value
Smokers (%)Nonsmokers (%)
  • *

    Linear-by-linear association test for trend.

  • **

    Fisher’s exact test.

  • ***

    Other: Asian, Black, Aboriginal, Jewish, Arabic, and other ethnicities.

Maternal age in years29.4 (6.0)32.6 (4.3)0.030
Gestational age at delivery, weeks38.9 (0.5)38.9 (0.8)0.736
Gravida (number of pregnancies)3.0 (1.3)2.7 (1.1)0.288
Term deliveries1.2 (0.7)1.0 (0.7)0.204
Preterm birth <37 weeks0.1 (0.3)0.1 (0.3)0.630
Abortion (miscarriages and terminations)0.6 (1.0)0.6 (1.0)0.950
Live birth1.2 (0.6)1.0 (0.6)0.214
Height, cm163.1 (6.6)162.7 (7.5)0.829
Weight, kg84.7 (17.6)82.8 (15.8)0.632
Number of cigarettes per day12.4 (9.5)00.006
Marital status0.001*
Married/common law18 (85.7)83 (100)
Separated/divorced/single3 (14.7)0 (0)
Educational level<0.001*
Partial high school4 (19)0 (0)
Completed high school9 (42.9)9 (11)
Postsecondary education8 (38.1)57 (69.5)
Graduate school0 (0)16 (19.5)
Household income<0.001*
0-$24 9995 (23.8)3 (3.8)
$25 000–49 9998 (38.1)14 (17.5)
$50,000–79 9994 (19)29 (36.3)
Over $80 0004 (19)34 (42.5)
Mother’s ethnicity0.423**
White17 (81.0)59 (71.1)
Other***4 (19.0)24 (28.9)
Father’s ethnicity0.055**
White19 (90.5)59 (68.7)
Other***2 (9.5)26 (31.3)

Patients were required to present approximately 2.5–3 hours before the operation to the antenatal ward where smoking was not permitted. The majority of caesarean sections were performed under spinal anaesthesia. The common indications for caesarean were because of a previous caesarean section or for a baby in a breech presentation (Table 2). The proportions of male and female infants and the Apgar scores were similar in smokers and nonsmokers (Table 3). Infants of smokers weighed significantly less than those of nonsmokers (3213 ± 443 g versus 3528 ± 453 g, respectively, P= 0.005). There was a trend towards lighter placentas in smokers (685 ± 140 g versus 759 ± 173 g, respectively, P= 0.074), although the ratio of birthweight to placental weight was the same in smokers and nonsmokers.

Table 2.  Birth information
 SmokerNonsmokerP value
  • *

    Previous myomectomy.

  • **

    Fisher’s exact test.

Indication for caesarean section0.538
Repeat caesarean17 (81.0)67 (80.7)
section (%)
Breech (%)3 (14.3)15 (18.1)
Other* (%)1 (4.8)1 (1.2)
Epidural (%)0 (0)1 (1.2)
Spinal (%)21 (100)81 (97.6)
Time of maternal blood draw9:05 (0:53)8:57 (1:02)0.620
Time of cord clamping10:18 (1:05)9:48 (1:01)0.056
Table 3.  Infant characteristics
 Number (%) or mean (SD)P value
Fetal sex0.305
Male8 (38.1%)42 (50.6%)
Female13 (61.9%)41 (49.4%)
Apgar score
1 minute, mean (SD)8.1 (2.0)8.4 (1.3)0.388
5 minute, mean (SD)9.1 (0.5)8.9 (0.7)0.219
Birthweight (g), mean (SD)3213 (443)3528 (453)0.005
Placental weight (g), mean (SD)685 (140)759 (173)0.074
Ratio of birthweight:placental weight, mean (SD)4.8 (0.7)4.8 (0.7)0.938

Infants exposed to tobacco smoke in utero had significantly elevated umbilical arterial ACTH levels compared with nonexposed infants [17 (4–22) pmol/l versus 4 (2–11) pmol/l, respectively, P= 0.005]. (Table 4) Umbilical arterial cortisol concentrations were similar between infants of smokers and nonsmokers [182 (130–240) nmol/l versus 192 (127–265) nmol/l, respectively, P= 0.541]. Umbilical venous cortisol concentrations were similar between infants of smokers and nonsmokers [106 (82–182) nmol/l versus 133 (108–177) nmol/l, respectively, P= 0.346).

Table 4.  Laboratory values
 SmokersNonsmokersP value
Umbilical arterial ACTH, pmol/l median (interquartile range)17 (4–22)4 (2–11)0.005
Umbilical arterial cortisol, nmol/l, median (interquartile range)182 (130–240)192 (127–265)0.541
Umbilical venous cortisol, nmol/l, median (interquartile range)106 (82–182)133 (108–177)0.297
Maternal cotinine, ng/ml, median (interquartile range)80 (25–138)0 (0)<0.001
Maternal cortisol, nmol/l, mean (SD)918 (186)857 (201)0.207

Maternal cotinine concentrations were significantly higher in smokers [80 (25–138) ng/ml versus 0 ng/ml, respectively, P < 0.001]. Maternal cortisol levels showed a trend towards higher levels in smokers than nonsmokers (918 ± 186 nmol/l versus 857 ± 201 nmol/l, respectively, P= 0.207).

Umbilical arterial cortisol was not correlated with umbilical arterial ACTH (r=−0.066, P= 0.523), or maternal cortisol levels (r= 0.047, P= 0.633), or time of the delivery of the placenta (r= 0.137, P= 0.164).

Umbilical arterial ACTH was not significantly correlated with maternal cotinine concentrations in smokers (r= 0.193, P= 0.413). Maternal cortisol in smokers was significantly negatively correlated with education level, r=−0.470, P= 0.031, although not with income, r=−0.213, P= 0.355.

Using backward linear regression, a full model with ACTH as the dependent variable and all potential confounders and explanatory variables were included (maternal self-reported smoking status, maternal cortisol, income, fetal sex, gestational age at delivery and birthweight) and explained only 15.4% of the variance in umbilical ACTH. A simplified model with self-reported smoking status and birthweight explained 13.8% and hence these two variables were focused on in this simplified model. Using regression, it was noted that with birthweight, self-reported smoking status and an interaction term in the model, and a statistically significant interaction between birthweight and ACTH was noted (P= 0.020). In tobacco-exposed fetuses, there was a statistically significant negative correlation between ACTH concentrations and birthweight (r=−0.573, P= 0.010). In nonexposed fetuses, there was no significant correlation between ACTH concentrations and birthweight (r=−0.149, P= 0.195).


This study showed for the first time to our knowledge that infants exposed to tobacco smoke in utero have significantly elevated ACTH levels compared with nonexposed infants [17 (4–22) pmol/l versus 4 (2–11) pmol/l, respectively, P= 0.005], although cortisol concentrations were not significantly different.

Although ACTH levels were increased in umbilical arterial blood, cortisol levels were not, which we speculate may reflect reduced negative feedback of cortisol on the hypothalamus/pituitary, down regulation of the ACTH receptor or a blunted response to ACTH in the fetus. Additionally, the adrenal gland may be less responsive, hence, permitting elevations in ACTH without increases in cortisol. Animal models of fetal hypoxaemia are not helpful in trying to explain this phenomenon as they show increases in cortisol in response to elevations in ACTH.4 The increased umbilical arterial ACTH seen in infants of smokers may, in part, be due to secondary chronic stress from tobacco exposure leading to hypoxia through increases in carboxyhaemoglobin,7 decreased unloading pressure as oxygen moves across the placenta from maternal to fetal blood and again when it crosses from fetal blood into fetal tissues,8 and from uterine vessel vasoconstriction.9 Alternatively, ACTH may be elevated due to direct stimulation of the fetal HPA axis by nicotine, which crossed the placenta, although this is less likely given that patients had to present to hospital approximately 2.5–3 hours prior to their operation. In hypertensive nonpregnant adults, ACTH levels have been reported to rise immediately after smoking, while the peak in plasma cortisol was seen 20–30 minutes later.20 In our study, we did not see a correlation between cotinine and ACTH, which would be expected if direct stimulation by nicotine was the only mechanism of action, however, our sample size may not have been sufficient to demonstrate this. Animal models have shown that ACTH does not cross the placenta and hence this is not believed to represent elevations in maternal ACTH being transferred to the fetus. Fetal ACTH does not cross to the mother either.21 In our study, the umbilical cord blood was drawn slightly later in the smoke-exposed group, however, most studies suggest that a circadian rhythm has not been established in human fetuses.22,23

There was an effect modification between birthweight and ACTH in smoke-exposed fetuses. It is believed that the lower birthweight is not likely to be the cause of the increased ACTH in smoke-exposed fetuses, given both the size of the increase, and that lower birthweight in nonsmoke-exposed fetuses was not associated with increased ACTH. It is rather believed that smoke-exposed fetuses with lower birthweights may be particularly prone to elevated ACTH levels.

Elevations in ACTH during fetal life, even without elevations in cortisol, could potentially have adverse long-term consequences in terms of programming of the fetal HPA axis. If the adrenal gland became less responsive to ACTH, and potentially more autonomous, the usual regulation of cortisol could be disrupted. Exposure to dexamethasone in the last week of gestation in rats has been shown to decrease the glucocorticoid receptor gene expression in the hippocampus, which reduces sensitivity to feedback, and may permit elevated circulating glucocorticoids without the usual negative feedback.24 In a study of 9-year-old children, a 1-kg decrease in birthweight was associated with a trend towards an 11.9% (95% CI −3.9 to +30.4%) increase in glucocorticoid metabolites adjusted for sex and weight (partial correlation −0.1, P= 0.150). The highest average urinary glucocorticoid metabolites were found in the lightest and heaviest babies at birth, with a U-shaped or quadratic relationship.25 Low birthweight was associated with raised fasting plasma cortisol concentrations in three different adult populations.26 Thinness at birth has been associated with raised blood pressure,27 decreased glucose tolerance in children28 and a higher incidence of metabolic syndrome in adults,2 all of which could be accounted for by changes in cortisol. (‘Thinness’ has been defined differently in each of the above studies: per SD decrease in Ponderal index,5 lowest quartile of the Ponderal index ≤23.36 and ≤6.5 lbs,7 respectively.)

The mothers who smoked in this study had a mean cortisol level of 918 nmol/l compared with 857 nmol/l in nonsmokers. This difference was not statistically significant (P= 0.207), possibly due to the fact that the mothers in our study were moderate smokers, smoking on average 12 cigarettes per day. In nonpregnant adults, it has been shown that smoking causes elevations in cortisol levels, although many of the studies reported higher cigarette consumption levels of 18.829 to 2030 cigarettes per day.

Other factors besides smoking may affect maternal cortisol levels. In smokers, maternal education level correlated negatively with maternal cortisol levels, r=−0.483, P= 0.027, although the correlation in nonsmokers was not significant. Several studies in the literature have shown that adult cortisol levels are not related to social class at birth.26,31 In this study, maternal income level did not correlate significantly with cortisol levels in either smokers or nonsmokers.

The effect of smoking on birthweight has previously been well documented,32–34 however, the approximate 300-g difference in infants in our study (3213 ± 443 g versus 3528 ± 453 g, P= 0.005), reinforces the importance of educating women about the adverse perinatal outcomes associated with maternal smoking, a modifiable risk factor.

Limitations of this study include the fact that umbilical CRH and maternal CRH and ACTH were not measured (initially due to budgetary constraints and they could not be added subsequently due to specimen volume requirements and the fact that the ACTH sample must be placed immediately on ice, which was not done with any of the maternal samples). Our samples, which were stored at −20°C for times ranging between 3 months and 2 years, were run in batches. Cortisol is stable ‘indefinitely’ at −20°C,35 and ACTH has been documented as being stable at this temperature for at least 5 to 6.5 months, the longest previous storage documented in the literature.36 Cotinine stability on urine stored at −20°C has been documented over a 10-year span,37 and we do not anticipate that there would be differences with serum, but will verify this in further studies. Given that the lack of correlation between cotinine and ACTH could be due to low power, further studies with larger sample sizes are required to explore this possible relationship.

Although this study was conducted in two academic centres in Ontario, Canada, the inclusion criteria selected healthy women with healthy pregnancies, and not women requiring tertiary care centre admission. Hence, it is anticipated that the results of this study would be generalisable to most pregnant women and their infants, at least of Caucasian descent, who met the inclusion/exclusion criteria.

The main strengths of this study are its prospective nature, which allowed adequate collection of the important confounders, and the fact that its sample size was approximately twice as large as the only previously well-designed study, which examined cortisol levels in infants of nonsmokers who underwent elective caesarean section.15

Future larger studies will involve CRH measurements, maternal ACTH measurements, placental receptor studies for nicotine and hormones, and long-term follow up of a smoking cohort.


This study established for the first time that ACTH levels were significantly elevated in infants of smokers. Normal values for cortisol in infants of smokers born at term by elective caesarean section were also established. Further research related to in utero smoking exposure and other endocrine disruption is underway in a larger cohort. This research will provide more information on the harm of tobacco exposure to the developing fetus.


This work was supported by a grant from the physicians of Ontario through the Physicians’ Services Incorporated, grant number 2004-673.