Estrogen Metabolism in Postmenopausal Women Exposed In Utero to Diethylstilbestrol

This article is featured in Highlights of This Issue, p. 1113

The synthetic estrogen diethylstilbestrol (DES), to which millions were exposed, is the only known human transplacental carcinogen, as in utero exposure has been linked with cancers of the vagina (1) and breast (2), as well as cervical intraepithelial neoplasia grade 2 and higher (3). Environmental scientists also consider DES the definitive model for the impact of exposure to environmental endocrine disruptors during fetal development (4). More than three decades of studies in laboratory animals have raised multiple possible biologic mechanisms, including a possible influence of DES on hormone concentrations (5) that could be responsible for the adverse health outcomes observed in exposed individuals.

In humans, the hormones estradiol and estrone when hydroxylated at the 2-, 4-, or 16-carbon position, result in an array of metabolites that have varying carcinogenicity. Recent observations in humans show that breast cancer risk is related to specific metabolic pathways, with greater metabolism in the 2 pathway showing protection (6). This led us to assess whether hormones and their metabolite concentrations were altered in postmenopausal women with documented prenatal exposure to DES.

The data for the current analysis were collected as part of a feasibility study at Boston University (Boston, Massachusetts), 1 of 5 sites of the NCI's Combined DES Cohort Follow-up Study. The study was approved by institutional review boards at the NCI and Boston University.

In 1992, the NCI assembled extant U.S. cohorts of individuals with medical record documentation of exposure or lack of exposure to DES, including daughters identified from the National Cooperative Diethylstilbestrol Adenosis Project (DESAD; ref. 7), and a large private infertility practice in Massachusetts. A cohort of daughters of women who participated in the Women's Health Study [(WHS) Daughters’ Cohort; ref. 8] was added to the combined cohort in 1994. Questionnaires were mailed in 1994, and approximately every 5 years since to ascertain information on health outcomes and major cancer risk factors. A detailed description of the methods and findings from the main cohort study is published (9).

We sought to enroll 40 exposed and 20 unexposed women. We restricted eligibility for this substudy to women who had responded to the most recent study questionnaire (2011) and lived within driving distance of the Boston University Medical School General Clinical Research Unit (GCRU). We further restricted to postmenopausal women who had never used hormone supplements and had never been diagnosed with any cancer (excluding nonmelanoma skin cancer). History of hormone therapy use was based on any questionnaire report of use of oral medications, vaginal cream/tablet/ring, and topical creams. At the time of blood draw, women were also asked about current and recent use in case they began using hormones since their last questionnaire report. Of 2,407 Boston University participants, 793 women lived within 30 miles and among them, 300 met the other inclusion criteria. We also attempted to choose women with vaginal epithelial changes (VEC), which are correlated with a higher dose of, and earlier exposure to, DES in utero (10). We wrote to 180 of the 300 eligible women inviting them to participate in data collection and then followed up with a phone call: 16 were found to be ineligible due to (i) having moved out of the area (n = 4), (ii) death (n = 1), or (iii) not meeting other eligibility criteria (n = 11). Of 164 eligible women, 44 refused (27%), 60 (37%) were either not called or not reached by telephone, and 60 agreed to participate. The 60 participants (40 exposed, 20 unexposed) provided blood between May 2014 and July 2015.

Participants gave written informed consent, completed a short questionnaire that confirmed their eligibility and ascertained recent alcohol use, and had their height and weight measured. Body mass index (BMI) was calculated by dividing weight (kg) by height (m; ref. 2). Three 10 mL tubes of whole blood were collected and immediately processed. Serum samples were stored in the GCRU and later shipped on dry ice to the NCI's laboratory.

The DESAD study incorporated a comprehensive gynecologic examination around the time of recruitment that systematically identified VEC by means of colposcopy or iodine staining (11, 12). Vaginal epithelial changes were defined as vaginal epithelium that was glandular in nature (adenosis) or metaplastic squamous epithelium, which develops as adenosis undergoes physiologic healing. Age at menarche was ascertained on the 1994 questionnaire, parity and smoking status were ascertained on the 2006 questionnaire, and age at menopause was queried on all of the questionnaires. On the basis of knowledge of regional prescribing practices, we characterized Boston's cohorts as “high dose” (2). Among participants with complete information on cumulative dose, the median dose for Boston was 8,675 mg.

A validated and reported stable isotope dilution high-performance liquid chromatography–tandem mass spectrometry (HPLC-MS/MS) assay was employed to measure 15 estrogens and estrogen metabolites, including estrone, estradiol, 2-hydroxyestrone, 2-methoxyestrone, 2-hydroxyestradiol, 2-methoxyestradiol, 2-hydroxyestrone-3-methyl ether, 4-hydroxyestrone, 4-methoxyestrone, 4-methoxyestradiol, 16α-hydroxyestrone, 16-ketoestradiol, estriol, 17-epiestriol, and 16-epiestriol, in serum as described previously (13, 14) with updated instrumentations and additional stable isotope–labeled estrogen metabolites. Briefly, HPLC-MS/MS analysis was performed using a Thermo TSQ Vantage triple quadrupole mass spectrometer (Thermo Fisher Scientific) coupled with a Prominence LC system (Shimadzu Scientific Instruments). Both HPLC and MS were controlled by Xcalibur software (Thermo Fisher Scientific). Twelve stable isotopically labeled estrogens and estrogen metabolites were used to account for losses during sample preparation and HPLC-MS/MS analyses, which included deuterated estriol, (C/D/N Isotopes, Inc.); deuterated 16-epiestriol (Medical Isotopes, Inc.); and 13C-labeled estrone, estradiol, 2-hydroxyestrone, 2-methoxyestrone, 2-hydroxyestradiol, 2-methoxyestradiol, 2-hydroxyestrone-3-methyl ether, 4-hydroxyestrone, 4-methoxyestrone, and 4-methoxyestradiol (Cambridge Isotope Laboratories). Measurements were made at the Laboratory of Proteomics and Analytical Technologies, Cancer Research Technology Program, Leidos Biomedical Research, Inc. The stable isotope dilution HPLC-MS/MS was used to quantify 15 estrogens and estrogen metabolites that circulate primarily as sulfated and/or glucuronidated conjugates (14). Five (estrone, estradiol, estriol, 2-methoxyestrone, and 2-methoxyestradiol) were measured in unconjugated forms. The serum sample was split into two aliquots to measure the combined concentration of each of the 15 estrogens (sum of conjugated plus unconjugated forms) and the unconjugated forms. To measure the combined parent estrogen or estrogen metabolite level, an enzymatic hydrolysis with sulfatase and glucuronidase activity was added to the sample preparation to cleave any sulfate and glucuronide groups. To measure the unconjugated forms, the enzymatic hydrolysis was not included in sample preparation. For metabolites with both combined and unconjugated measurements, the concentration of the conjugated form was calculated as the difference between the combined and unconjugated estrogen measurements. The calibration curves were linear over a concentration range of 1 pg/mL to 1,000 pg/mL for all estrogens and estrogen metabolites. The assay limit of detection (LOD) providing estrogen signal-to-noise ratio greater than 3-to-1 was 100 fg/mL (∼0.33–0.37 pmol/L). The assay lower limit of quantitation (LLOQ) was 1 pg/mL for each estrogen and estrogen metabolite with intra- and interbatch coefficients of variation <15% and assay accuracy between 85% and 115% of known targeted values at LLOQ. Laboratory coefficients of variation from 10 blinded replicates across batches were <3.0% for all hormones measured.

Estrogens and estrogen metabolites were analyzed individually, in groups representing metabolic pathways, and as ratios of metabolic pathways. Estrone and estradiol, the parent estrogens, are irreversibly hydroxylated at the C-2, C-4, or C-16 positions of the steroid ring, leading to a cascade of metabolites (Supplementary Fig. S1). The combined metabolites in each of the C-2, C-4, and C-16 pathways were used to investigate whether prenatal DES exposure was associated with altered patterns of estrogen metabolism. These included metabolites in the 2-hydroxylation pathway (Path2, i.e., 2-hydroxyestrone, 2-methoxyestrone, 2-hydroxyestradiol, 2-methoxyestradiol, and 2-hydroxyestrone-3-methyl ether); metabolites in the 4-hydroxylation pathway (Path4, i.e., 4-hydroxyestrone, 4-methoxyestrone, and 4-methoxyestradiol); and metabolites in the 16-hydroxylation pathway (Path16, i.e., 16α-hydroxyestrone, estriol, 17-epiestriol, 16-ketoestradiol, and 16-epiestriol). These pathways were also examined as ratios—for example, path2:path16—to assess relative metabolism down each of the pathways. Total parent estrogens were the sum of estrone and estradiol; total metabolites were the sum of Path2, Path4, and Path16; and total estrogens were defined as the sum of the parent estrogens and metabolites.

Values for the estrogens were examined for outliers and those with 1.5 interquartile ranges above the 75% percentile or below the 25% percentile were excluded for the analysis of that estrogen. Values were log transformed to improve normality. Pearson correlations were calculated among the 15 estrogens. Standard linear models (Proc GLM; SAS) with log estrogen (or log ratio) as the dependent variable and DES status as the independent variable evaluated the effect of DES exposure. Models were adjusted for age (continuous variable). The coefficient, β, for DES was exponentiated and estimates the ratio of the geometric means between the two groups. The percentage change is defined as 100 × (exp(β)-1). A positive percent change indicates that the mean is higher in the DES exposed compared with the unexposed, and a negative percent change indicates that the mean is lower in the exposed compared with the unexposed. The reported age-adjusted geometric means for each of the DES groups were estimated using the least-square means procedure (lsmeans statement) in SAS. In separate models, we further adjusted for years since menopause, BMI, parity, and recent alcohol use. Forest plots were created to visually present the percent change in the means for the parent estrogens and metabolites between the exposed and unexposed with 95% confidence intervals (CIs).

Mean age was 60.8 years (range, 51.6–67.7) for exposed women and 61.7 years (55.9–67.8) for unexposed women, and all participants identified themselves as white. Reproductive, anthropometric, and lifestyle characteristics are presented in Table 1. The exposed and unexposed women were generally similar in these characteristics. Thirty-six of the 40 exposed women had a history of VEC.

In the exposed and unexposed combined, parent estrogens, Path2, Path4, and Path16 metabolites comprised 38.3% (estrone 34.6% and estradiol 3.7%), 16.3%, 1.7%, and 43.6% of total estrogens (sum of all 15 concentrations), respectively (Supplementary Fig. S2). The parent hormones, estrone and estradiol, were highly correlated (r = 0.84), whereas the other metabolites showed less correlation (Supplementary Table S1).

Age-adjusted geometric mean serum concentrations of the parent estrogens and estrogen metabolites by prenatal DES exposure are presented in Table 2. In general, total estrogens and parent estrogens were greater in the DES-exposed women. The ratios of path2:total estrogens, path2:parent estrogens, and path2:path16 (i.e., less metabolism in the 2 vs. 16 pathway) were all lower in the DES-exposed women, whereas the ratio of path16:total metabolites was greater in the DES exposed. Supplementary Figure S3 shows the same information presented in Table 2, age-adjusted percent change in geometric means between the prenatally DES exposed and the unexposed in parent estrogens, estrogen metabolites, metabolite pathways, and ratios. Associations were similar with additional adjustment for years since menopause, BMI, parity, and recent alcohol intake (Fig. 1).

In this study, we found evidence that women who were prenatally exposed to DES compared with those who were not had less metabolism of parent estrogens in the 2 pathway; this association remained after adjustment for total estrogens and adjustment for some other factors that may influence estrogen metabolism, such as BMI, parity, years since menopause, or recent alcohol use. In the early 1970s, young women who had been exposed prenatally to DES demonstrated a strikingly high relative risk of vaginal cancer (1). With long-term follow-up, multiple other adverse health outcomes were linked to this exposure, including precancerous lesions of the cervix (3), and a modest increase in breast cancer (2). These findings stimulated extensive laboratory investigations pursuing possible underlying biologic mechanisms.

Some evidence in mice suggest that prenatal DES exposure may affect hormone concentrations later in life. An in vitro study of total estrogen, testosterone, and progesterone in mouse ovarian tissue showed significantly higher secretion in the DES-exposed animals (5). These hormonal changes may be caused by epigenetic alterations. Recent data in mice (15) show that maternal exposure during pregnancy to environmentally relevant doses of BPA (chemically similar to DES) induces sex-specific, dose-dependent (linear and curvilinear), and brain region–specific changes in expression of genes encoding estrogen receptors and estrogen-related receptor-γ in juvenile offspring. In the same study (15), BPA altered mRNA levels of the epigenetic regulators DNA methyltransferase (DNMT) 1 and DNMT3A in the juvenile cortex and hypothalamus, paralleling changes in estrogen-related receptors. Importantly, changes in ERα and DNMT expression in the female hypothalamus were associated with DNA methylation changes in the ERα gene. There are no direct data in animals on the effect of DES on estrogen metabolism.

Data on the effect of prenatal DES exposure on hormone concentrations in humans are scant. Two small studies of DES and hormones (16, 17) were conducted in the 1970s with inconsistent results and using hormone assay methods far inferior to those now available. More recently, data from a small sample of premenopausal women in the Harvard Study of Moods and Cycles showed that those who reported prenatal DES exposure had lower estradiol concentrations than those who did not report DES exposure (18). In the absence of medical record confirmation, DES exposure status is likely to be misclassified, although the magnitude of the effect of this on previous study findings is unknown. In our study, which includes confirmation of DES exposure status for all participants, results suggest that parent estrogen concentrations, including estradiol, are somewhat greater in DES-exposed postmenopausal women.

This study was not a random sample of the entire combined cohort study. The women who were included were likely exposed to a high cumulative dose of DES in utero, and the majority had VEC, which is associated with high DES dose and early exposure in pregnancy (10). It was also limited in sample size. Nonetheless, it demonstrated a pattern of results consistent with a relative decrease in levels of 2-hydroxylation pathway metabolites and in the ratio of these to total parent estrogens and to 16-hydroxylation pathway metabolites in the prenatally DES exposed compared with unexposed women. This profile has been associated with an increased risk of breast cancer in postmenopausal women (6). Although the association between DES and estrogen metabolism is in the direction of higher breast cancer risk for the prenatally DES exposed and could therefore possibly explain some of their modestly increased risk compared with the unexposed, it will take a larger study to quantify the magnitude of the effect (2). Because the number of participants studied was small, we tried to account only for the major influences on estrogen metabolism. We did not collect information on dietary intake or physical activity, and we think it unlikely that there would be a strong association between these factors and whether the participant's mother took DES during the daughter's pregnancy. Our hypothesis was that prenatal exposure to DES effects estrogen metabolism and that alterations in estrogen metabolism persist after menopause. Because we believe the effects of DES on estrogen metabolism would occur prenatally and therefore precede the adverse reproductive events we have chronicled (9), we did not adjust for those intervening events in the analyses of DES exposure and estrogen metabolism.

This area of research has great potential to improve our understanding of the biological mechanisms associated with endocrine disruption in humans during the prenatal developmental period. In this study, we had the unique opportunity to assess hormone concentrations in women with documented exposure to an established endocrine disrupting chemical in utero. Our preliminary findings suggest that prenatal exposure to endocrine disruptors may influence estrogen metabolism many years later.

E.E. Hatch is a consultant/advisory board member for UpToDate (author on DES). No potential conflicts of interest were disclosed by the other authors.

Conception and design: R. Troisi, E.E. Hatch, J.R. Palmer, L. Titus, R.N. Hoover

Development of methodology: R. Troisi, L. Titus

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E.E. Hatch, J.R. Palmer, L. Titus, X. Xu, R.N. Hoover

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R. Troisi, J.N. Sampson, R.N. Hoover

Writing, review, and/or revision of the manuscript: R. Troisi, E.E. Hatch, J.R. Palmer, L. Titus, X. Xu, R.N. Hoover

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): R. Troisi

Study supervision: E.E. Hatch, L. Titus

The work was funded through contracts with the Division of Cancer Epidemiology and Genetics, NCI.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.