Women exposed to diethylstilbestrol (DES) in utero experience a greater risk of adverse reproductive events, including infertility,1 ectopic gestations, spontaneous pregnancy losses, and premature births.2 These complications may in part be mediated through teratogenic effects, namely the structural uterine and cervical abnormalities that have been associated with in utero DES exposure.3 Preeclampsia, a common pregnancy complication characterized by maternal hypertension, hyperuricemia, and proteinuria frequently involves shallow placentation. Placental establishment requires cytotrophoblast invasion of the underlying stroma and blood vessels of the maternal endometrium, a process involving immune and angiogenic mechanisms. Diethylstilbestrol-associated uterine abnormalities and possible alterations in immune function4–7 may adversely affect successful implantation.
The hypothesis that prenatal DES exposure is associated with preeclampsia risk was addressed in a small case-control study that reported a greater than two-fold risk in women who reported a history of DES exposure compared with those who did not.8 The authors investigated this association in the National Cancer Institute’s (NCI) DES Combined Cohort Follow-up, a large prospective study of U.S. individuals with documented prenatal estrogen exposure.
MATERIALS AND METHODS
Approvals for the study were obtained from the institutional review boards of the field centers and the NCI. Participants indicated informed consent by completing and returning questionnaires or by participating in a telephone interview.
The NCI DES Combined Cohort Follow-up Study started in 1992 with the aggregation of new and previously followed U.S. cohorts of individuals with documented prenatal DES exposure (or the absence of exposure). The daughters’ combined cohort derives from four individual cohorts.2 The total number of women by exposure status for each cohort is presented in Table 1. The largest cohort consists of women enrolled during the mid-1970s into the National Cooperative Diethylstilbestrol Adenosis Project (DESAD).9 Nearly half of the exposed women were identified by prenatal record review at five medical centers (Mayo Clinic, Rochester, Minnesota; Boston Lying-in Hospital, Boston, Massachusetts; Gunderson Clinic, LaCrosse, Wisconsin; Baylor College of Medicine, Houston, Texas; and the University of Southern California, Los Angeles, California). At each center a search was made for clinics or physicians’ offices where large numbers of prenatal records would be available for review. When a clinic or physician agreed, all records from 1940 to the present were reviewed to identify mothers who were given DES during pregnancy. Unexposed women were selected from the same prenatal record sources as the exposed subjects. In some cases these were unexposed sisters of the exposed participants matched by age (within 6 months). The rest were matched by age and maternal age (within 5 years) and identified from a list compiled of eligible controls (ie, women who met the same criteria as the exposed participants but had no mention of sex hormone exposure in the prenatal record or in the prenatal history from the mother). The remaining DES-exposed women came from physician referrals or self-referrals. Participants in DESAD were actively followed with annual physical examinations (through 1980), with interviews (through 1984), and by a mailed questionnaire (from 1985 through 1989).
The second cohort consists of women who participated in a randomized clinical trial of DES in 1951–1952 (Dieckmann cohort).10 Trial participants were recruited from patients registered consecutively in prenatal clinics at the University of Chicago and the Chicago Lying-in Hospital who were before 20 weeks of gestation. Every other patient was administered a placebo, and the clinicians were blinded to the DES status of participants. In 1974, attempts were made to trace these women, and 83% of the exposed and 77% of the unexposed responded to a questionnaire. There was periodic follow-up of this cohort during the 1980s; women were last contacted in 1990.
A third cohort consists of exposed daughters and their unexposed sisters whose mothers were treated with DES during pregnancy in a large private infertility practice in the Boston, Massachusetts, area (Horne cohort). All participants were offspring of mothers who took DES while under the practice’s care. Unexposed siblings of the exposed participant were identified and may have been born before or after the exposed participant. If there was more than one unexposed sibling, the mother decided which could be contacted. The Horne cohort was assembled in the mid-1970s and subjects were systematically followed through the 1980s.
The fourth cohort consists of daughters of women who participated in the Women’s Health Study,11 originally conducted during the 1970s and 1980s to determine whether women exposed to DES during pregnancy had a higher breast cancer risk. The study participants derived from four institutions: Boston Lying-in Hospital, Dartmouth-Hitchcock Medical Center, the Mayo Clinic, and one private obstetric practice in Portland, Maine. Mothers were identified from the prenatal records and were eligible if they delivered a child (while married) between 1940 and 1960. In 1992, DES-exposed mothers and a comparison group of unexposed mothers were asked for permission to contact their offspring, who were then recruited into the NCI combined cohort study. The unexposed mothers were chosen from the same records as the exposed women, from lists compiled of all women who met the definition of nonexposure. For each exposed woman, one unexposed woman was chosen at random whose date of birth was within 2 years of that of the exposed woman and who had received obstetric care at the same institution.
Review of the mother’s obstetric record allowed classification of women as prenatally DES exposed or unexposed. Data on gestational week of first DES use were available for 75% of exposed women. Individual data, available for 38% of the women, allowed the calculation of total cumulative dose. Based on differences in DES prescribing practices by U.S. region, the individual cohorts also were characterized as high or low dose (except for a portion of Women’s Health Study participants for whom prescribing practices were unknown). The individual dose data supported this classification.12
Information on pregnancies was collected by questionnaires that were mailed in 1994 and 1997. Outcome of the first pregnancy (live birth, still birth, miscarriage, induced abortion, and ectopic pregnancy) and number and delivery dates of live births were ascertained on the 1994 questionnaire, and number and delivery dates of live births were updated on the 1997 questionnaire. On the 1997 questionnaire, we queried participants about preeclampsia: “Have you EVER been diagnosed with preeclampsia or eclampsia (pregnancy-related high blood pressure) during pregnancy?” Additional questions ascertained the year of first diagnosis and treatment (with instructions to mark all that applied).
Information on education, height, weight at age 20, and smoking status was from the 1994 questionnaire. Mother’s hypertension in the pregnancy involving the participant was ascertained by chart review in the DESAD Cohort. In a subset of exposed women in the DESAD cohort (360 of 2,081), data from hysterosalpingograms were available from examinations performed routinely during the original cohort follow-up. From these data we were able to determine the presence of uterine abnormalities based on cavity shape and size and other intrauterine defects.
We mailed questionnaires to all living participants from the original cohort studies unless they had refused participation during one of the previous follow-ups or were untraceable. The proportions of participants completing questionnaires were 88% of the exposed and 84% of the unexposed in 1994 and 91% of both exposed and unexposed in 1997. The present analysis was limited to pregnancies resulting in live births, in part because details were not ascertained for pregnancies resulting in other outcomes except for the first pregnancy. The percentage of still births was low (1.84% in DES-exposed and 1.05% in unexposed women). The numbers of participants, live births, and cases are reported by exposure status and cohort in Table 2. A total of 3,672 women (2,468 exposed and 1,204 unexposed) are included in the analysis.
Person-time was accrued during each pregnancy resulting in a live birth. Women were assigned one person-year for each of these pregnancies. Thus, the number of person-years is equal to the number of births. Because the gestational age at preeclampsia diagnosis was unknown, follow-up began 1 year before and stopped at the date of birth, beginning again 1 year before the next birth date for women who had more than one pregnancy, and so on until the last reported birth before the 1997 questionnaire. Women were censored at the end of the pregnancy in which preeclampsia was first diagnosed.
We assessed the associations of several established preeclampsia risk factors, including age at index pregnancy (continuous, time-dependent), parity (time-dependent), educational level, body mass index at age 20 (BMI, calculated as weight in kilograms divided by squared height in meters), and smoking status. Poisson regression analysis13 estimated incidence relative risks and 95% confidence intervals (CI) for preeclampsia comparing DES-exposed women with unexposed women. Results were similar when repeated using Cox proportional hazards regression (data not shown). All estimates were adjusted for age at the index pregnancy. To assess potential confounding, estimates were further adjusted for the factors mentioned above (as categorical, indicator variables) and are referred to as fully adjusted. The influence of DES dose and gestational week of DES exposure on preeclampsia risk also was assessed. Tests for trend in the risk estimates were evaluated by including an ordinal variable in the regression model. Analyses were repeated, redefining cases on the basis of treatment (and including those who were not treated), with those not meeting a specific definition not counted as cases. In addition, the main analysis assessing the association of DES and preeclampsia risk was repeated, excluding sisters of DES-exposed participants and, in another analysis, using all first pregnancies regardless of birth outcome.
Preeclampsia risk was elevated in women who were older at index pregnancy, who had no previous live births, and who were heavier, and risk was reduced in women who ever smoked before the index pregnancy (Table 3). Preeclampsia risk was higher after 1,975, but there was no risk trend with calendar year after that time. Preeclampsia risk by calendar time was similar among the DES-exposed and unexposed women (data not shown). Preeclampsia risk was elevated (age-adjusted relative risk 1.39, 95% CI 0.68–2.84) in women whose mothers had pregnancy hypertension, but this was based on only six DES-exposed and two unexposed cases of preeclampsia. The vast majority of both exposed and unexposed women were white (96% and 92%, respectively). Comparing DES-exposed with unexposed women, there were no differences in prevalence of type 1 diabetes (0.52% and 0.50%, respectively), type 2 diabetes (2.4% and 2.3%), or hypertension (12.7% and 14.2%) diagnosed any time during the study period.
In utero DES exposure was associated with an almost 50% elevation in preeclampsia risk (Table 4). Results were only slightly attenuated with further adjustment for the risk factors in Table 3, and there was no evidence of effect modification of the DES and preeclampsia association by these factors (data not shown). The following relative risks are fully adjusted. The relative risk for DES and preeclampsia was similar with additional adjustment for twin pregnancies (relative risk 1.48, 95% CI 1.08–2.04). Of women who developed preeclampsia, a greater proportion of DES-exposed (63%) than unexposed (52%) women developed it in their first pregnancies. The relative risk for DES and preeclampsia restricted to first pregnancies resulting in any birth outcome was 1.51 (95% CI 1.03–2.24), and in those resulting in a live birth, the relative risk was 1.81 (95% CI 1.17–2.79). The excess risk of preeclampsia associated with DES was apparent only in women who were prenatally exposed before the 15th week of pregnancy. The relative risk for less than 15 weeks at exposure was 1.57 (95% CI 1.11–2.23) and for 15 or more weeks at exposure was 1.10 (95% CI 0.68–1.80). Results based on the two-dose variables (exact dose versus high or low dose) were inconsistent. There appeared to be a positive trend in preeclampsia risk with three categories of exact dose. Compared with the unexposed, the relative risks were 1.18, 1.24, and 1.37 for 2,500 mg or less, 2,500–9,999 mg, and 10,000 mg or more, respectively; the P value was 0.11 for the test for trend using an ordinal variable for the unexposed category and the three dose categories. In contrast, the dose variable based on regional prescribing practices showed relative risks of 1.66 (95% CI 1.22–2.26) for low dose and 1.04 (95% CI 0.76–1.41) for high dose compared with the unexposed (P>.50 for trend).
Further analyses were conducted to assess possible biases and biological mechanisms. The relative risk in the clinical trial participants (Dieckmann) was 1.97, but the 95% CI included one (0.69–5.67), probably owing to the smaller number of exposed cases. Among the DESAD cohort (overall DES relative risk 1.51, 95% CI 1.02–2.24), the DES risk estimate was similar when restricted to women who were identified for cohort inclusion by medical record review (1.63, 95% CI 1.01–2.63), as opposed to physician or self-referral. Excluding from the analysis women whose mothers had pregnancy hypertension, the relative risk for DES and preeclampsia was 1.51 (95% CI 1.02–2.24). The relative risk for DES and preeclampsia appeared higher for cases occurring before 1975 (2.70, 95% CI 1.04–7.05) than for those occurring in 1975 or later (1.31, 95% CI 0.94–1.82), but the interaction was not statistically significant (P=.26). There were few cases before 1971, but the relative risk was similar to that for preeclampsia occurring before 1975 (data not shown). Excluding sisters of DES-exposed participants, the relative risk for DES and preeclampsia was 1.49 (95% CI 1.04–2.13).
We further evaluated the association of DES with preeclampsia, defining cases by their reported treatment (Table 5). Of 285 preeclampsia cases, 36 reported treatment with magnesium sulfate, 179 with bed rest, 87 with diet, and 71 with other treatment; 35 reported no treatment. Among those reporting other treatment, the most frequent responses were medication (n=18), hospitalization (n=15), induced delivery (n=14), cesarean delivery (n=11), and more frequent prenatal check-ups (n=3). The relative risks were most elevated among women reporting magnesium sulfate (2.10), compared with women reporting no treatment (1.11) or any treatment (1.47).
Of 360 DES-exposed women who had had a hysterosalpingogram, 79% (n=282) had evidence of a uterine abnormality. Among those with an abnormality, 12.4% developed preeclampsia in a subsequent pregnancy, compared with 7.7% of those without an abnormality. Among these DES-exposed women, the relative risk for preeclampsia associated with uterine abnormalities was 1.60 although the confidence interval was wide (95% CI 0.66–3.89).
In these data, preeclampsia risk was greater in women who were prenatally exposed to DES compared with women who were not, consistent with the findings of Mittendorf et al8 In that study, preeclampsia was verified by medical records, whereas DES exposure was self-reported. In the present study, DES-exposed women were more likely than unexposed women to develop preeclampsia in a first pregnancy. In addition, the preeclampsia risk associated with DES was more pronounced in the subgroup of women with greater certainty of diagnosis, ie, those who reported treatment, especially magnesium sulfate, than in those not reporting treatment. This greater risk might also indicate that DES is associated with more severe preeclampsia. Although the causality of this association depends on further corroboration in population or laboratory investigations, confirmation from a report of a preeclampsia case-control study with our prospective cohort data based on verified exposures indicates that this would be a worthwhile pursuit.
Preeclampsia is a complex, heterogeneous disorder of unclear etiology. Although immunological, inflammatory, and angiogenic abnormalities appear involved, the processes through which DES may affect preeclampsia are at present unclear. Older work using DES-exposed mice14 and an in vivo human model15 has shown that DES affects the developmental structure of the uterine wall, as evidenced during early pregnancy by the abnormal segregation of the stromal layers themselves. For the placenta to anchor and establish itself in the uterus, the cytotrophoblast must invade into the underlying stroma and blood vessels of the maternal endometrium. These observations imply that stromal abnormalities may influence placentation and development of preeclampsia. Perhaps stromal abnormalities that occur through mechanisms other than DES exposure might be a risk factor for preeclampsia as well.
One factor controlling the implantation process is the interaction of several genetic pathways, including the Wnt family of genes.16 In pilot studies with murine models, lack of Wnt-7a signaling produced uterine abnormalities resembling those observed in women exposed in utero to DES. Diethylstilbestrol down-regulated Wnt-7a expression in these models and possibly caused a permanent loss in its function. This type of study is beginning to unravel at a basic level the mystery of how DES may influence gene expression, but clearly further work is needed to understand the precise mechanisms involved.
The association with preeclampsia was limited to women who were DES exposed before the 15th week of pregnancy, whereas exposure that began after 15 weeks was not associated with an increased risk. These data are consistent with previous studies that show DES effects on other outcomes, including vaginal adenosis and cervical ectropion,17 and structural anomalies of the cervix and vagina18 with DES given earlier in pregnancy.
Results for dose were inconsistent for the two dose variables we assessed. Exact dose was missing for almost two thirds of the cohort (although the missing data that were abstracted from birth records would be random with regard to later preeclampsia diagnosis), and the derived measure was based on assumptions regarding regional prescribing practices. Because the overall DES association was moderate, these limitations may have obscured our ability to detect dose effects. In addition, we did not have dose estimates for different times during pregnancy to assess the dose effect for those women who received it before 15 weeks of gestation, which would seem to be the relevant exposure.
Our investigation included only pregnancies resulting in live births, thereby limiting the generalizability of our findings. This was mainly due to the lack of detailed information on pregnancies with other birth outcomes, except in the first. A higher proportion of DES-exposed women than of unexposed women experience outcomes other than live births, predominately in the form of early miscarriages.2 In pregnancies that terminate early, preeclampsia is less likely to develop and be recognized due to the generally late presentation of the condition. Thus, theoretically, including early pregnancy losses in the analysis should lead to an underestimate of the DES-preeclampsia association. This would be true of still births as well, but there were very few in these data. Indeed, repeating the analyses and including all of the participants’ first pregnancies, regardless of birth outcome, gave a relative risk of 1.51 compared with a relative risk of 1.81 based on first pregnancies resulting in a live birth only.
There are limitations of our study. The preeclampsia diagnosis was self-reported and not validated by obstetric records. It is possible that some of the women reporting preeclampsia actually had isolated pregnancy-induced hypertension (PIH) and that DES-exposed women were more likely to report PIH as preeclampsia, biasing the relative risk toward a positive association between DES and preeclampsia. However, the prevalence of preeclampsia in all live births among the unexposed women (2.9%) was similar to that among white women from the period 1988–1999 based on the U.S. National Health Discharge Survey (2.6%).19 Furthermore, in one study,20 agreement between medical records and women’s report of toxemia in any previous pregnancy was high (95%, κ=0.90). Random diagnostic misclassification would tend to underestimate the effect of DES on preeclampsia risk, but more accurate reporting by DES-exposed women, who tend to have other reproductive and pregnancy complications, could bias toward a spuriously elevated relative risk. We attempted to address this potential bias by restricting the analysis to women who reported medical treatment for preeclampsia and who thus might be more likely to recall their diagnosis regardless of DES exposure. In these analyses, results were similar to those found overall, with the highest relative risk in women who were treated with magnesium sulfate. An earlier study21 in DESAD of mother’s questionnaire report of reproductive history validated by medical records collected 10 years earlier, which showed no difference in agreement between DES-exposed and unexposed women, is reassuring. It remains possible that DES-exposed women were more likely to receive a preeclampsia diagnosis. In the early 1970s, the adverse pregnancy outcomes associated with DES were not yet recognized.22 The relative risks for DES and preeclampsia actually appeared higher before 1975 (2.70) than in subsequent time periods (1.31), suggesting that diagnostic bias does not account for our findings. If mothers with a history of preeclampsia were more likely to receive DES during a subsequent pregnancy, then DES daughters, who might inherit a greater tendency to develop preeclampsia, might appear to be at higher risk. Mother’s history of pregnancy hypertension was associated with an increased preeclampsia risk in the daughters, although this estimate was based on only eight cases and was not statistically significant. Results for DES and preeclampsia were similar when analyses were restricted to daughters whose mothers did not report a history of pregnancy hypertension. Furthermore, the association between DES and preeclampsia was similar in the cohort of daughters whose mother participated in a clinical trial of DES to that shown in the other cohorts.
In summary, the present findings suggest a slightly elevated preeclampsia risk in the pregnancies of women prenatally exposed to DES. Possible biological mechanisms to explain this effect include uterine abnormalities and alterations in immune mechanisms.
1.Palmer JR, Hatch EE, Rao RS, Kaufman RH, Herbst AL, Noller KL, et al. Infertility among women exposed prenatally to diethylstilbestrol. Am J Epidemiol 2001;154:316–21.
2.Kaufman RH, Adam E, Hatch EE, Noller K, Herbst AL, Palmer JR, et al. Continued follow-up of pregnancy outcomes in diethylstilbestrol-exposed offspring. Obstet Gynecol 2000;96:483–9.
3.Kaufman RH, Noller K, Adam E, Irwin J, Gray M, Jefferies JM, et al. Upper genital tract abnormalities and pregnancy outcome in diethylstilbestrol-exposed progeny. Am J Obstet Gynecol 1984;148:973–84.
4.Kalland T. Long-term effects on the immune system of an early life exposure to diethylstilbestrol. In: Hunt VR, Smith MK, Worth D, editors. Environment factors in human growth and development (Banbury Report no. 11). Cold Spring Harbor (NY): Cold Spring Harbor Laboratory; 1982. p. 217–39.
5.Ford CD, Johnson GH, Smith WG. Natural killer cells in in utero diethylstilbestrol-exposed patients. Gynecol Oncol 1983;16:400–4.
6.Ways SC, Mortola JF, Zvaifler NJ, Weiss RJ, Yen SS. Alterations in immune responsiveness in women exposed to diethylstilbestrol in utero. Fertil Steril 1987;48:193–7.
7.Holladay SD, Blaylock BL, Comment CE, Heindel JJ, Fox WM, Korach KS, et al. Selective prothymocyte targeting by prenatal diethylstilbestrol exposure. Cell Immunol 1993;152:131–42.
8.Mittendorf R, Lain KY, Williams MA, Walker CK. Preeclampsia: a nested, case-control study of risk factors and their interactions. J Reprod Med 1996;41:491–6.
9.Labarthe D, Adam E, Noller KL, O’Brien PC, Robboy SJ, Tilley BC, et al. Design and preliminary observations of the National Cooperative Diethylstilbestrol Adenosis (DESAD) Project. Obstet Gynecol 1978;51:453–8.
10.Bibbo M, Gill WB, Azizi F, Blough R, Fang VS, Rosenfield RL, et al. Follow-up study of male and female offspring of DES-exposed mothers. Obstet Gynecol 1977;49:1–8.
11.Greenberg ER, Barnes AB, Resseguie L, Barrett JA, Burnside S, Lanza LL, et al. Breast cancer in mothers given diethylstilbestrol in pregnancy. N Engl J Med 1984;311:1393–8.
12.Palmer JR, Wise LA, Hatch EE, Troisi R, Titus-Ernstoff L, Strohsnitter W, et al. Prenatal diethylstilbestrol exposure and risk of breast cancer. Cancer Biomark Epidemiol Prev 2006;15:1509–14.
13.Preston DL, Lubin JH, Pierce DA, McConney ME. EPICURE user’s guide. Seattle (WA): Hirosoft International Co; 1993.
14.Taguchi O, Cunha GR, Robboy SJ. Experimental study of the effect of diethylstilbestrol on the development of the human female reproductive tract. Biol Res Pregnancy Perinatol 1983;4:56–70.
15.Robboy SJ, Toguchi O, Cunha GR. Normal development of the human female reproductive tract and alterations resulting from experimental exposure to diethylstilbestrol. Hum Pathol 1982;13:190–8.
16.Sassoon D. Wnt
genes and endocrine disruption of the female reproductive tract: a genetic approach. Mol Cell Endocrinol 1999;158:1–5.
17.Robboy SJ, Kaufman RH, Prat J, Welch WR, Gaffey T, Scully RE, et al. Pathologic findings in young women enrolled in the National Cooperative Diethylstilbestrol Adenosis (DESAD) project. Obstet Gynecol 1979;53:309–17.
18.Jefferies JA, Robboy SJ, O’Brien PC, Bergstralh EJ, Labarthe DR, Barnes AB. Structural anomalies of the cervix and vagina in women enrolled in the Diethylstilbestrol Adenosis (DESAD) project. Am J Obstet Gynecol 1984;148:59–66.
19.Tucker MJ, Berg CJ, Callaghan WM, Hsia J. The Black-White disparity in pregnancy-related mortality from 5 conditions: differences in prevalence and case-fatality rates. Am J Public Health 2007;97:247–51.
20.Githens PB, Glass CA, Sloan FA, Entman SS. Maternal recall and medical records: an examination of events during pregnancy, childbirth, and early infancy. Birth 1993;20:136–41.
21.Tilley BC, Barnes AB, Bergstralh E, Labarthe D, Noller KL, Colton T, et al. A comparison of pregnancy history recall and medical records. Am J Epidemiol 1985;121:269–81.
22.Barnes AB, Colton T, Gundersen J, Noller KL, Tilley BC, Strama T, et al. Fertility and outcome of pregnancy in women exposed in utero to diethylstilbestrol. N Engl J Med 1980;302:609–13.