Diethylstilbestrol Exposure in Utero and Depression in Women

Abstract

Diethylstilbestrol (DES) is an estrogenic endocrine disruptor with long-term health effects, possibly including depression, following exposure in utero. Understanding the relation between in utero DES exposure and depression will provide insight to the potential adverse effects of bisphenol A, a functionally similar and ubiquitous endocrine disruptor. The association between in utero DES exposure and depression was assessed among participants in the Nurses’ Health Study II who first reported their history of antidepressant use in 1993 and lifetime history of depressive symptoms in 2001. DES exposure was reported by 1,612 (2.2%) women. A history of depression at baseline was higher among women exposed to DES in utero compared with those not exposed (age-adjusted odds ratio (OR) = 1.47, 95% confidence interval (CI): 1.26, 1.72) (P < 0.001). Incident depression (first use of antidepressants among women who also reported depressive symptoms) during follow-up (1995–2005) was reported by 19.7% of women exposed to DES and 15.9% unexposed (age-adjusted OR = 1.41, 95% CI: 1.22, 1.63) (P < 0.001). Adjustment for risk factors of depression and correlates of DES exposure moderately attenuated the association (multivariable-adjusted OR = 1.30, 95% CI: 1.13, 1.51) (P = 0.0004). These results suggest that the neurophysiologic effects of in utero exposure to DES could lead to an increased risk of depression in adult life. Further research should assess whether in utero exposure to bisphenol A has similar adverse effects.

The fetal environment has been implicated in adult-onset depression in a number of epidemiologic studies (1–4). In particular, the results of some investigations have suggested that the daughters and sons of women prescribed diethylstilbestrol (DES) during the index pregnancies may be at increased risk of anxiety and depression (5–9). One possible explanation is that depression among the exposed may be due to their awareness or experiences of the potential adverse outcomes associated with exposure in utero, including vaginal adenocarcinoma, reproductive organ dysfunction, and abnormal pregnancies (10, 11). However, in a retrospective analysis of adult offspring who were unaware of the treatment (DES or placebo) to which their mothers were randomly assigned during a clinical trial of the efficacy of DES in preventing toxemia, the risk of psychiatric disorders, including depression and anxiety, was about 2-fold in the treated group compared with the placebo group (5).

Although use of DES in pregnancy has been discontinued because of its carcinogenic effects, bisphenol A (BPA), a structurally and functionally similar compound, is widely used as a component of polycarbonate plastic in water bottles, baby bottles, dental sealants, and epoxy resin in the lining of food cans. Two recent reports summarized the effects of pre- and perinatal BPA exposure on neural and behavioral outcomes in animals (12, 13). Thus, even if DES is no longer prescribed to pregnant women, the possible link between DES exposure and depression is of public health importance. We therefore have examined prospectively whether participants of the Nurses’ Health Study II exposed prenatally to DES had increased risk of depression.

MATERIALS AND METHODS

The Nurses’ Health Study II cohort began in 1989 when 116,671 female US nurses aged 25–42 years completed and returned an initial questionnaire on their lifestyle and diseases. Every 2 years since inception, participants have returned questionnaires to update exposure information and report newly diagnosed diseases. All participants were born (1946–1965) when DES was available to pregnant women. The 1993 questionnaire, completed by 94,503 women, included questions on their mother's use of DES or “other hormones” during her pregnancy with the participant and on participants’ history of antidepressant use. Of the 94,503 women who answered the 1993 questionnaire, 77,802 also answered the lifetime history of depressive symptoms question in 2001; 1,562 women were excluded for missing exposure or use of antidepressants without symptoms of depression, leaving 76,240 women eligible for the analyses.

Assessment of outcome

History of antidepressant medication use was ascertained in 1993 when participants were asked whether they had ever taken any of the following medications: 1) tricyclic antidepressant (e.g., Elavil (AstraZeneca Pharmaceuticals LP, Wilmington, Delaware; brand discontinued in the United States); Norpramin (Sanofi-Aventis U.S., Bridgewater, New Jersey); Tofranil (Mallinckrodt Pharmaceuticals, Covidien, Mansfield, Massachusetts); Pamelor (Mallinckrodt Pharmaceuticals); Sinequan (Pfizer, Inc., New York, New York; brand discontinued in the United States); Vivactil (Merck & Company, Inc., West Point, Pennsylvania); or Surmontil (Duramed Pharmaceuticals, Inc., Cincinnati, Ohio)) or 2) Prozac (Eli Lilly and Company, Indianapolis, Indiana) (fluoxetine) or Zoloft (Pfizer, Inc.) (sertraline). The initiation date was not asked. On each biennial questionnaire from 1997 onward, regular use of antidepressant medications during the previous 2 years was ascertained. The list of drugs was updated to include Paxil (GlaxoSmithKline, Research Triangle Park, North Carolina) in 1997 and Celexa (Forest Pharmaceuticals, Inc., St. Louis, Missouri) in 2001. History of depressive symptoms was assessed on the 2001 questionnaire in which subjects were asked, “In your lifetime have you ever had 2 weeks or longer when nearly every day you felt sad, blue, or depressed for most of the day?” and “Did you ever tell a doctor or mental health specialist that you were feeling depressed?”. On the 2003 and 2005 questionnaires, participants were asked whether in the past 2 years they had used antidepressants or “had 2 weeks or longer when nearly every day you felt sad, blue, or depressed for most of the day.” Because antidepressants can be prescribed for conditions other than depression, women who used antidepressants but did not report depressive symptoms were excluded.

History of depression.

Women who reported in 1993 having ever used antidepressants were considered to have a history of depression if they also answered questions on the lifetime history of depressive symptoms positively on the 2001 questionnaire.

Incident depression.

Women were considered to have an episode of incident depression if 1) between 1995 and 2001 they initiated antidepressant use and answered positively questions on the lifetime history of depressive symptoms on the 2001 questionnaire, or 2) they reported both initiating antidepressant use and having depressive symptoms in the previous 2 years on either the 2003 or the 2005 questionnaire.

Assessment of exposure

A supplementary questionnaire was mailed to 2,742 women who reported in 1993 that they were exposed to DES in utero, and 2,317 responded. Among the respondents, 2,032 reported that they were certain/somewhat certain of their exposure, 123 that they were not certain, and 162 that they were not exposed. We considered those who were certain/somewhat certain as exposed. In 2001, 29,070 participants’ mothers each completed a questionnaire pertaining to her pregnancy with, and early life exposures of, the participant. The agreement between DES exposure as reported by daughters in the supplementary questionnaire and DES exposure as reported by mothers was k = 0.74.

Assessment of covariates

Information about age, smoking, menopausal status, current weight, and multivitamin use was ascertained on the baseline questionnaire and updated every 2 years. In follow-up questionnaires, participants also reported income, history of infertility and miscarriages, physical activity, alcohol intake, parental smoking during pregnancy and childhood, their birth weight, and whether they were a premature birth.

Statistical analysis

Logistic regression was used to assess the association between DES exposure and history of antidepressant use reported in 1993. Pooled logistic regression was used to assess the association between DES and incident depression during follow-up. Potential confounders were risk factors of depression (smoking, alcohol use, history of infertility and miscarriages, physical activity, self-reported cancer, multivitamin use, menopausal status, income, and body mass index) and the pre- and perinatal correlates of DES exposure (participants’ birth weight, premature delivery, exposure to other hormones in utero or to smoke in utero or during childhood). The multivariable-adjusted analyses restricted to the subset of participants with mother-reported DES were further adjusted for potential confounders (maternal use of prenatal vitamins, antinausea medications, or sleeping medications) that were ascertained only from the mother’s questionnaire. All P values presented are 2 tailed; P < 0.05 was considered statistically significant, and analyses were performed by using SAS, version 9.1, software (SAS Institute, Inc., Cary, North Carolina).

RESULTS

There were 76,240 women eligible for inclusion in this study. Those who reported that they were certain/somewhat certain of their exposure to DES in utero (n = 1,612) had higher rates of infertility and miscarriages, and they were more likely to have been born premature or with low birth weight or to have been exposed to other hormones and cigarette smoke in utero. Physical activity, multivitamin use, alcohol consumption, and being postmenopausal were slightly higher among the exposed, while smoking rates, income, and body mass index did not vary (Table 1).

Table 1.

Age-adjusted Characteristics of Cohort Participants According to DES Exposure in Utero Among US Women Aged 28–48 Years in 1993^a

	Source of Information on DES Exposure in Utero
	Cohort Participant (n = 76,240)		Mother (n = 29,070)
	Unexposed (n = 74,628)	Exposed (n = 1,612)	Unexposed (n = 28,275)	Exposed (n = 795)
Mean age (range), years	38 (28–48)	39 (28–47)	38 (28–48)	38 (28–47)
Annual income, %
<$40,000	6.0	6.1	5.9	5.8
$40,000–<$50,000	7.8	6.4	7.7	5.9
$50,000–<$75,000	23.2	22.1	23.4	25.4
$75,000–<$100,000	17.9	18.8	18.7	17.1
$100,000–<$150,000	18.7	21.7	19.4	20.0
≥$150,000	10.5	11.0	11.3	10.2
Mean age of mother at participant's birth, yearsb			26	27
Mother's education, %b
Less than high school			13.5	9.4
High school			49.4	43.9
College degree or higher			36.9	46.3
Father's education, %b
Less than high school			17.9	14.2
High school			36.6	31.5
College degree or higher			41.2	51.5
Parents own their home at the time of participant's birth, %b			52.1	47.1
Smoking status in 1993, %
Never	66.1	64.4	68.1	69.4
Past	23.5	25.0	22.6	23.2
Current	10.3	10.6	9.2	7.3
Mean alcohol intake in 1993, g/day	3.2	3.5	3.2	3.2
Mean body mass index in 1993, kg/m2	25.2	25.1	24.9	25.1
Physical activity in 1991, METs/week	20.7	22.8	21.1	20.4
Postmenopausal in 1993, %	4.7	7.0
Participant experienced infertility, %c	21.7	35.6	21.0	29.8
Participant experienced miscarriage, %	22.9	31.2	22.6	27.7
Pregnancy-related characteristic
Mother smoked during pregnancy, %d	23.3	32.5	26.1	33.0
Parents smoked at home during participant's childhood, %
Neither	33.7	29.3	38.1	34.9
Mother only	7.5	9.2	6.8	9.6
Father only	26.9	24.6	28.8	26.1
Both	26.5	32.7	24.7	27.6
Mother used hormones other than DES during pregnancy, %	0.8	1.7	0.6	5.5
Mother used prenatal vitamins during pregnancy, %b			62.1	69.4
Mother used sleeping medication during pregnancy, %b			0.4	4.0
Mother used antinausea medication during pregnancy, %b			6.3	18.9
Birth weight <5.5 pounds, %d,e	6.7	14.5	5.4	11.1
Premature by ≥2 weeks, %d	6.8	17.8	11.8	19.6
Never breastfed, %d	58.3	60.3	53.7	52.6
Delivery by cesarean section, %b			3.2	3.6
Use of forceps during delivery, %b			19.8	22.7
Multiple birth, %b			6.5	5.4
Mother living with biologic father at time of participant's birth, %b			98.7	99.2

Abbreviations: DES, diethylstilbestrol; MET, metabolic equivalent.

^aPercentages may not total to 100% because of missing data.

^bVariables were derived from a questionnaire sent to participants’ mothers in 2001; not available for all participants.

^cExperienced infertility defined as “tried to become pregnant for more than 1 year without success.”

^dVariables included in both the participants’ questionnaire (used in columns labeled “cohort participant”) and in the mothers’ questionnaire (used in columns labeled “mother”).

^eMetric equivalent = 2.5 kg.

A past history of depression was reported by 13.8% (n = 222) of the women exposed to DES in utero and by 10.8% (n = 8,079) of those not exposed (age-adjusted odds ratio (OR) = 1.47, 95% confidence interval (CI): 1.26, 1.72) (P < 0.001) (Table 2). Adjustment for risk factors for depression, including smoking, alcohol use, income, physical activity, self-reported cancer, multivitamin use, menopausal status, history of infertility and miscarriages, and body mass index, and for pre- and perinatal correlates of DES exposure, including participants’ birth weight, premature delivery, and exposure to other hormones or to smoke in utero, only slightly attenuated the results (Table 2). Participants’ history of infertility did not modify the association between DES exposure in utero and lifetime history of depression (P = 0.36).

Table 2.

DES Exposure in Utero and Depression in US Women Aged 28–48 Years in 1993^a

Antidepressants, Depressive Symptoms	DES		Age Adjusted			Multivariable Adjusted
	Unexposed, no.	Exposed, no.	Odds Ratio	95% Confidence Interval	P Value	Odds Ratio	95% Confidence Interval	P Value
	History of Depression by 1993 (n = 76,240)
Nonuser, no symptoms reported	36,592	655	Referent
Nonuser, symptoms reported	29,957	735	1.35	1.21, 1.50	<0.001	1.28	1.15, 1.43	<0.001
User, symptoms reported	8,079	222	1.47	1.26, 1.72	<0.001	1.40	1.19, 1.65	<0.001
Total	74,628	1,612
	Incident Depression Between 1995 and 2005 (n = 60,684)
Nonuser, symptoms reported	20,919	492	1.24	1.11, 1.39	<0.001	1.19	1.06, 1.33	0.004
User, symptoms reported	9,438	252	1.41	1.22, 1.63	<0.001	1.30	1.13, 1.51	0.0004
Total	30,357	744

Abbreviation: DES, diethylstilbestrol.

^aParticipants were considered to have symptoms if they answered yes to the question: “In your lifetime have you ever had 2 weeks or longer when nearly every day you felt sad, blue, or depressed for most of the day?” or “Did you ever tell a doctor or mental health specialist that you were feeling depressed?”.

Exclusion of women with a history of depression in 1993 or with incomplete data left 60,684 women to be followed between 1995 and 2005. Incident depression was reported by 19.7% (n = 252) of the exposed women and by 15.9% (n = 9,438) of those not exposed (age-adjusted OR = 1.41, 95% CI: 1.22, 1.63) (P < 0.001). This association remained after adjustment for potential confounders (Table 2). The significantly increased risk of incident depression among women exposed to DES in utero was apparent when the analyses were performed separately for the periods 1995–2001 (OR = 1.45, 95% CI: 1.24, 1.69) (P < 0.001) for DES exposed compared with nonexposed and 2001–2005 (OR = 1.35, 95% CI: 1.05, 1.75) (P < 0.001). Exposure to “other hormones” in utero was also strongly associated with incident depression (multivariable OR = 1.71, 95% CI: 1.37, 2.15) (P < 0.0001).

In 2001, a subset of participants’ mothers (n = 29,070) returned questionnaires pertaining to their pregnancy with, and the early life of, their daughter. Findings for the association between DES exposure and depression were similar to those in the full cohort, although power was reduced. The age- and multivariable-adjusted odds ratios of a history of depression in 1993 were 1.40 (95% CI: 1.11, 1.76) (P = 0.004) and 1.34 (95% CI: 1.05, 1.77) (P = 0.02) comparing DES exposed, as reported by the mother, with nonexposed participants. Similarly, the age- and multivariable-adjusted odds ratios for incident depression between 1995 and 2005 were 1.28 (95% CI: 1.04, 1.58) (P = 0.02) and 1.18 (95% CI: 0.95, 1.47) (P = 0.13).

DISCUSSION

In this prospective study, we found a higher risk of depression among middle-aged women exposed to DES and/or other hormones in utero. The findings persisted 10 years after women reported their exposure. The strengths of our study include the partially prospective cohort design, large sample size, and long follow-up. However, information on depression was based on self-reported use of antidepressants and symptoms of depression, rather than a physician diagnosis of depression following criteria from the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders. DES exposure was also self-reported by participants, but among those whose mothers also reported DES exposure, agreement was good. We cannot rule that some mothers, and thus participants, are unaware that they were prescribed DES, but this would not explain our findings.

An association between DES exposure and increased psychiatric and anxiety disorders was first reported by Vessey et al. (5) in 1983. In this study, general practitioners were contacted directly for the medical history of offspring of women randomized to placebo or DES in a double-blind controlled trial conducted in London during the 1950s. DES was started at 5 mg/day as close to the sixth week of pregnancy as possible, but not after the 16th week, and continued until the 35th week when the dose had incrementally reached 125 mg/day. Although study documentation from the original trial was lost, and it was only known if women had been assigned to group A or group B, the authors wrote that they were fairly certain which group was in fact the exposed group. Offspring of the exposed group had twice as many reports of psychiatric disorders, including depression and anxiety, as the unexposed group (15.1% in the exposed, 7.4% in the unexposed). These findings cannot be attributed to distress at learning of DES exposure because that was unknown at the time of the Vessey study, or to confounding by indication because of the randomized design. Further, the reports of illnesses of the breast and urogenital system were similar in both groups, and therefore the somatic effects of DES were unlikely to be the cause of the increased incidence of psychiatric conditions. Two small case-control studies were published thereafter. Lifetime occurrence of major depressive disorder was similar among 30 DES-exposed women and 30 women with abnormal Papanicolaou smears (7). Fifty DES-exposed women had significantly higher odds of major depressive disorder and recurrent major depression (based on criteria from the third edition of the Diagnostic and Statistical Manual of Mental Disorders assessed with the Diagnostic Interview Schedule) and of amphetamine abuse when compared with their control sister, whereas no difference was found when they were compared with controls with abnormal Papanicolaou smears (6). In both studies, the authors concluded that an increase in major depressive disorder was a result of concern about gynecologic problems. However, recently a large study comprising 4 cohorts assembled retrospectively, including 8,000 men and women who had documented DES exposure in utero, found similar proportions of self-reported diagnosis and treatment for mental illness across exposure status (14). Finally, one partially prospective study of offspring exposed to DES found no association between prenatal DES exposure and serious psychiatric outcomes (suicide or hospitalization) or serious psychiatric outcomes plus psychiatric/psychological consultation, as reported by their mother (15). We had insufficient suicides to examine the association in the current analysis.

It is possible that the increased initiation of antidepressants and depressive symptoms among the DES-exposed women in our study results from their knowledge of the long-term adverse reproductive effects of exposure in utero, but this interpretation would not explain the results of Vessey et al. (5) reported above. The weaker association found in our study as compared with the study by Vessey et al. may be explained by lower doses and shorter duration of DES exposure among women in our cohort. Although we do not have direct information on dose, the dose and early administration of DES were likely to have been more consistent among women in a randomized trial.

Two recent reports summarize the neural and behavioral effects of BPA when administered to experimental animals in doses considered safe in humans (12, 13), including reduced novelty-seeking (16), increased anxiety (17, 18), and “depression-like” behavior (19). BPA is an estrogenic compound, similar to DES, used in polycarbonate plastic bottles, dental sealants, and epoxy resin in the lining of food cans. It has been detected in the urine at levels of 0.1 μg/L in 95% of the adults and 92.6% of the participants aged 6 years or more in US reference populations (20, 21). Children had higher concentrations than adolescents who had higher concentrations than adults.

Recent research implicates endocrine disruptors and DNA methylation in the etiology of psychiatric disorders (22, 23). In animal models, exposure to DES, BPA, and other estrogenic compounds during development and early life results in altered methylation patterns in CpG islands (in promoter genes, thus changing tissue-specific gene expression) (24). For example, it has been shown that estradiol lowered catechol-O-methyl transferase (COMT) transcription (25). COMT is one of a number of genes associated with bipolar disorder and schizophrenia when its promoter region is hypomethylated (22, 26). In mice experiments, the hypomethylating effects of exposure to BPA were negated by dietary supplementation with folate, which is a coenzyme in the methylation reactions (27).

In summary, we found that women who were exposed to DES in utero had a higher risk of depressive symptoms and use of antidepressants, and that this increase in risk extended into their middle age. Although the possibility that some women became depressed because of awareness of their exposure status cannot be ruled out, these results are consistent with the hypothesis that the physiologic effects of in utero exposure to DES lead to higher rates of depression during adult life. It remains to be established whether prenatal exposures to ubiquitous environmental chemicals that are structurally similar to DES and have similar estrogenic effects also increase the risk of depression.

Glossary

Abbreviations

BPA

bisphenol A

CI

confidence interval

DES

diethylstilbestrol

OR

odds ratio