Because the immediate danger of continued exposure has long since passed, why should the effects of DES still be studied? Several compelling reasons exist. Perhaps most important, additional information is needed to appropriately counsel persons exposed to DES about the potential effect this exposure has on their risk for cancer and noncancer health outcomes and to develop strategies for risk reduction, early detection, and prevention. Second, DES exposure serves as a model for assessing the potential toxicity of a broad range of compounds that affect estrogen production or metabolism or mimic estrogen action [1, 2]. These compounds, which are widespread in the environment, include some chlorinated organic compounds, polycyclic aromatic hydrocarbons, herbicides, and pharmaceutical agents. It has been hypothesized that such compounds may contribute directly or indirectly to increasing rates of breast cancer [2] and disorders of the male reproductive tract [3]. Several of these compounds, including dichlorodiphenyltrichloroethane (DDT), dioxin, and methoxychlor (an estrogenic pesticide currently used as a substitute for DDT) have been shown to have reproductive effects similar to those of DES after in utero exposure in rodents [4-7]. The effects of these compounds in humans are less well established. The development of functional assays designed to identify compounds with activity similar to that of DES could streamline the toxicologic evaluation of these environmental estrogenic compounds [8]. Finally, the study of the health effects of DES will contribute to a better understanding of the role of estrogens in normal reproductive development, hormonal imprinting, and carcinogenesis [9].

We summarize the current information about the health effects related to DES exposure and suggest screening guidelines for persons exposed to DES.

Methods

Top Methods Author & Article Info References

We identified English-language articles through MEDLINE and CANCERLIT searches and through review of the bibliographies of identified articles. We reviewed all human studies relevant to long-term health effects of exposure to DES.

Background

Diethylstilbestrol, a synthesized stilbene, was first produced in London by Dodds and associates in 1938. Its biological properties are similar to those of naturally occurring estrogens such as estrone and estradiol-17 ß, but unlike estrogens, DES was inexpensive to manufacture [10]. Its effectiveness and potency after oral administration were particularly appealing, and DES was used to treat various gynecologic conditions with little reported toxicity [11].

On the basis of the observation that late-pregnancy toxemia, premature delivery, and fetal death in utero were preceded by a premature decrease in urinary estrogen and progesterone levels, Smith and Smith [12] proposed that the lack of hormonal support resulted in compromise of uterine vasculature. They postulated that the administration of stilbestrol would stimulate the release of estrogen and progesterone, restore hormonal balance, and avert or delay complications of late pregnancy. Early clinical reports were received enthusiastically, and DES was subsequently widely prescribed in the United States for the treatment of threatened and habitual abortions [13]. Although it was given primarily to women who had a history of problematic pregnancies, DES was also administered to normal pregnant women. An estimated 5 to 10 million Americans were prescribed DES during pregnancy (DES mothers) or were exposed to the drug in utero (DES daughters and sons) [14].

In 1953, Dieckmann and colleagues [15] published findings from a large clinical trial designed to evaluate the use of DES during pregnancy. On the basis of the analysis of 840 patients given DES and 806 women given placebo, DES was shown to be ineffective in preventing miscarriages and premature births. In a later reanalysis of these data, women exposed to DES were found to have higher risks for miscarriages, premature births, and perinatal death than women given placebo [16]. Other studies also found DES to be ineffective [17-19]. However, when these studies were done, the use of DES in high-risk pregnancies was well established, and physicians continued to prescribe DES for use in pregnancy until 1971 in the United States and until 1978 in Europe [20].

No harmful effects of DES were suspected until 1970, when vaginal clear-cell adenocarcinoma was reported in six young women aged 14 to 21 years [21]. Cancers of this site and cell type had previously been reported only rarely and in elderly women. In a case–control study of these young women, Herbst and associates [22] found a strong association with in utero exposure to DES; this finding was soon confirmed in other studies [23-25]. In November 1971, the U.S. Food and Drug Administration issued a drug bulletin that brought attention to the potential adverse effects of DES and banned its use during pregnancy [26].

Pharmacology and Toxicology of Diethylstilbestrol

In male humans, DES is rapidly absorbed from the proximal gastrointestinal tract; peak plasma concentrations are reached 20 to 40 minutes after an oral dose. Diethylstilbestrol is lipid-soluble and is readily distributed throughout the body compartments. Elimination is biphasic; plasma concentrations decrease rapidly, then level off after 3 to 6 hours. The half-life for the second phase is 2 to 3 days. Conjugation through glucuronidation with biliary excretion accounts for the early peak in plasma levels. Enterohepatic recirculation of DES after bacterial hydrolysis in the distal colon results in the prolonged plasma levels. Diethylstilbestrol is metabolized through the hepatic microsomal system primarily to dienestrol, and -hydroxydienestrol to epoxide and quinone intermediates, which are thought to be reactive. In rodents, DES metabolism can be increased by compounds that induce the hepatic microsomal enzymes or by DES itself. In humans, DES is primarily excreted in the urine [27-29].

Placental transfer of DES and fetal metabolism in primates have been studied in the rhesus monkey. These studies suggest that DES readily crosses the placenta and that it is conjugated and oxidatively metabolized in the fetus [27]. Metabolism of DES in the fetus has been most extensively studied in mice. In elegant toxicologic in vivo studies, Shah and McLachlan [30] showed that although the placenta appears to serve as a barrier to the passage of DES in mice, DES conjugates are accumulated in the fetal liver and gut, whereas DES is accumulated in the fetal reproductive tract. Oxidative metabolism of DES was identified in vitro in both the male and female fetal genital tract but not in the placenta or fetal liver [31]. Thus, at least in mice, the tissue-specific production of reactive DES metabolites remains a plausible explanation of DES-mediated toxicity.

The mechanism or mechanisms through which DES exerts its carcinogenic and other toxic effects remain unclear. Diethylstilbestrol is not mutagenic in the Ames test. However, investigators have observed increased chromosomal aberrations in neonatal mice that were exposed to DES in vivo and increased sister chromatid exchange in cultured human fibroblasts induced by DES [28]. The formation of abnormal or arrested mitotic spindles caused by the disruption of microtubules in embryonic hamster cells has also been seen [28]. Reactive intermediates of the oxidative metabolism bind covalently to DNA and may contribute to DES toxicity [27, 28].

Endo and coworkers [32] reported that rearrangement in the long arm of chromosome 3 (3q⁺), translocation between chromosomes 3 and 19 (t[3; 19]), and the isochromosome of chromosome 11 (i [11]) were common findings in cell lines from DES-induced uterine adenocarcinoma in neonatal mice [32]. The mechanism by which DES causes vaginal clear-cell cancer in DES daughters has not been clearly defined. Although Waggoner and colleagues [33] found altered expression of tumor-suppressor protein p53 in tumor blocks from patients with cervicovaginal clear-cell cancer and known DES exposure, others have not found mutations in the p53 tumor-suppressor gene or in the ras oncogene [34] (Boyd J. Unpublished data).

Cohorts Exposed to Diethylstilbestrol

Since the mid-1970s, adverse health effects associated with DES exposure have been defined through continued follow-up and epidemiologic study of several cohorts of persons exposed to DES (Table 1). These cohorts represent unique resources. Given the historic nature of DES use and the difficulty in retrospectively confirming exposure, reconstruction of such cohorts today would be almost impossible. All future studies will in some form stem from these existing cohorts. In interpreting existing data from these studies, it is helpful to review the manner in which these cohorts were assembled and information about the documentation of DES exposure.

For the largest cohort, the National Cooperative Diethylstilbestrol Adenosis (DESAD) project, which began in 1974, DES daughters and a comparison group of unexposed women were enrolled at five participating institutions (Baylor College of Medicine, Houston, Texas; Massachusetts General Hospital, Boston, Massachusetts; the Mayo Clinic, Rochester, Minnesota; the Gundersen Clinic, LaCrosse, Wisconsin; and the University of Southern California, Los Angeles, California). Investigators reviewed maternal prenatal records and identified 1892 women as having been exposed in utero to DES. The DESAD project also followed 2122 women who were either referred from community physicians or who walked into the centers for examinations. All of these women were required to have documented evidence of exposure to DES, through hospital, clinic, or pharmacy records or by a letter from the physician who cared for the mother during her pregnancy. A small group of patients without documented history of in utero exposure to DES but with colposcopic or histologic abnormalities consistent with DES exposure were also examined but were not included in the long-term follow-up of the cohort. A comparison group of 1033 unexposed women were selected from the same medical record sources as those of the exposed patients whose records were reviewed or were unexposed sisters of these patients. Participants in the DESAD project were followed with annual clinical examinations until 1983; between 1983 and 1989, they were followed with annual mail questionnaires.

Details on the DES dosage and the duration of exposure in DESAD participants were incomplete [35]. The approximate date on which DES exposure began was known for about 70% of participants, but the duration of exposure was known for only 29% of the exposed participants. The total DES dosage could be calculated for only 26% of the cohort. Among participants for whom complete information on dosage was available, the median total dosage was 4190 mg; however, there was a 10-fold difference between the median total dosage administered at Baylor College of Medicine (1625 mg) and that administered at the Massachusetts General Hospital (10 424 mg). The latter hospital had the highest median total dosage and used the Smith and Smith regimen most frequently.

A second cohort (the DES Mothers Study) consists of 3033 mothers exposed to DES who were identified at the Mayo Clinic; the Boston Lying-In Hospital, Boston, Massachusetts; Dartmouth-Hitchcock Medical Center, Hanover, New Hampshire; and a private obstetric practice in Portland, Maine [36]. Prenatal records were reviewed for documentation that DES or other nonsteroidal estrogens were prescribed during pregnancy. A comparison group of 3033 unexposed women was chosen from the same record sources and was matched with each woman in the cohort within 2 years of age.

A third cohort consists of mothers and offspring from the original Dieckmann clinical trial that was done during the early 1950s [15]. This study is particularly valuable because it was a randomized, double-blind trial; therefore, effects found in the exposed persons can be attributed to the exposure itself and not to some predisposing factor associated with its use. Dosages were well documented and were similar to those of the Smith and Smith regimen used at the Massachusetts General Hospital.

Several other small cohorts have been followed by individual investigators [37-39]. Two clinical trials of the efficacy of DES were done in the United Kingdom. The first was a small randomized, placebo-controlled study of pregnant diabetic women receiving DES that was conducted by the British Medical Research Council [18]. This study was remarkable for the high doses administered to study participants. The second British study [40] was a large clinical trial similar to that done by Dieckmann and colleagues [15].

The Registry for Research on Hormonal Transplacental Carcinogenesis [41] at the University of Chicago has provided much of the data on clear-cell adenocarcinoma of the cervix and vagina among women exposed to DES in utero, including risk factors, pathologic findings, survival, and recurrence rates. Cases are voluntarily reported by physicians. Of the 586 patients currently enrolled, 352 have a confirmed history of exposure to DES [42].

To foster further study of health effects related to DES exposure, the National Cancer Institute (NCI) has recently funded two major initiatives. The first provides support for the follow-up of identified cohorts of mothers, daughters, and sons exposed to DES and will include new participants identified through familial links within the cohorts. The second initiative will continue follow-up of patients with cervicovaginal clear-cell adenocarcinoma to provide information about survival, recurrence rates, and the rate of second malignancies. Related studies will better define risks, etiologic factors, and optimal treatment strategies in these women.

Health Risks Associated with Diethylstilbestrol

Diethylstilbestrol exposure has been associated with several health effects, including cancer, reproductive abnormalities, immune dysfunction, and alteration of psychosexual development (Table 2). These associations are reviewed below.

Cancer

Diethylstilbestrol Mothers

Bibbo and colleagues [43] were the first to study the association between DES exposure in pregnancy and cancer risk in exposed mothers. The investigators reported 32 incident cases of breast cancers among 693 mothers exposed to DES compared with 21 cases of breast cancer among 668 controls from Dieckmann and colleagues' cohort. Sixty-three percent of the original participants were included in this study, and a standardized risk ratio (standardized to the Connecticut Tumor Registry) of 1.46 was reported. Beral and Colwell [44] reported results from 27 years of follow-up in diabetic mothers from the Medical Research Council study. With estimated average dosages of DES and ethisterone of 16.3 g and 13.8 g, respectively, this study was noteworthy for the high dosages of hormones administered. With a follow-up rate of 97% for 80 women randomly assigned to receive hormonal treatment and 76 controls, the investigators on this study reported four cases of breast cancer among the treated women and no cases among the controls (P = 0.06; Fisher exact test). All cancers were diagnosed more than 15 years after treatment. Vessey and colleagues [40] reported results after 20 years of follow-up in women pregnant for the first time who were enrolled in a randomized, placebo-controlled trial done at the University College Hospital, London. Diethylstilbestrol was administered according to the Smith and Smith regimen. Vessey and colleagues achieved a follow-up rate of 64% for 813 women enrolled in this study between 1952 and 1953. No difference in the rate of breast cancer was observed between the treated and control groups.

Hadjimichael and associates [37] reviewed cancer risk in a cohort of 3139 obstetric patients delivering babies between 1946 and 1965 in 11 obstetrics practices in Fairfield and New Haven counties in Connecticut. For 1706 mothers, DES exposure was confirmed through review of medical records; a comparison group of 1405 women was selected and matched by age, race, sex of offspring, and date of index pregnancy. Cancer cases were identified through linkage to the Connecticut Tumor Registry. Dosages of DES administered in this study were relatively low, with a mean total dosage of about 2 g. A nonsignificant excess of cases of breast cancer was observed among exposed women, with a relative risk of 1.37. A risk in excess of 2.28 was excluded with 95% confidence.

The largest and most informative study of breast cancer risk in women who received DES has recently been updated by Colton and coworkers [45]. This study, which now has more than 30 years of follow-up, has tracked 5494 (91%) of the total cohort of 6058 mothers identified in the DES Mothers Study. Continued follow-up shows a small but statistically significant increase in the relative risk for breast cancer (relative risk, 1.35; 95% CI, 1.05 to 1.74); this risk has not, as was initially feared [46], increased markedly over time.

Thus, most evidence Figure 1 suggests that the increased risk for breast cancer among DES mothers is real but small, with less than a twofold increased risk. The risk for no other cancer has been shown to be significantly elevated in DES mothers, but these risks have not been adequately explored.

View larger version (19K): [in this window] [in a new window]   Figure 1. Relative risk estimates and 95% CIs for breast cancer among mothers exposed to diethylstilbestrol (DES) during pregnancy. A relative risk greater than 1.0 indicates that DES increases the risk for breast cancer. The reference number for individual studies is given in parentheses.

Diethylstilbestrol Daughters

In contrast to the large number of exposed mothers that was needed to establish a slight risk for breast cancer in DES mothers, the landmark studies that convincingly showed the association between in utero DES exposure and the development of cervicovaginal clear-cell adenocarcinoma involved surprisingly few patients. This reflects the rarity of the outcome and the strength of the association (Table 3). In a classic case–control study, Herbst and coworkers [22] reported the unusual occurrence of vaginal adenocarcinoma in eight young women whose diagnosis was established at the Massachusetts General Hospital. In utero exposure to DES could be confirmed in seven of these cases. None of the 32 matched controls had a history of in utero DES exposure [22]. Not surprisingly, a significant association was also found between a maternal history of bleeding during the index pregnancy and a history of previous pregnancy loss, conditions for which DES was typically prescribed. Several months later, Greenwald and colleagues [23] confirmed this observation in a second case–control study of young women with vaginal adenocarcinoma who were identified through the New York State Cancer Registry. Four of these women had a history of in utero exposure to DES. The fifth had in utero exposure to dienestrol and estrone. None of eight control patients had in utero exposure to estrogens [23]. Hill [24] described a series of six patients with cervicovaginal clear-cell carcinoma from the University of California, San Francisco. Three of these women had a documented history of DES exposure, two had a history of hormone exposure that could not be confirmed, and one had no medical records [24]. Sharp and Cole [47] later reviewed maternal histories from patients with confirmed DES exposure from the Herbst registry and from exposed and unexposed participants from the DESAD project; they convincingly showed that clear-cell adenocarcinoma was associated with DES exposure and not with the underlying conditions for which DES was prescribed.

Updates on these patients have been reported as cases in the Herbst registry have accumulated [28, 34, 48]. The risk of DES daughters for developing clear-cell adenocarcinoma has been estimated on the basis of cases reported to the Registry using data on DES sales and U.S. births between 1947 and 1961. This risk from birth through age 34 years is estimated to be 1 in 1000 to 1 in 10 000 women exposed to DES in utero. On the basis of these estimates, cases of clear-cell adenocarcinoma related to DES exposure could be expected to develop through the year 2000. Rates per 1 000 000 women have markedly decreased over successive birth cohorts since 1951, a trend that parallels sales of DES. The rates generally increase at puberty and level off around age 30 to 35 years [49]. Because of the relatively young age of the oldest cohorts, few data are available to assess the risk for clear-cell adenocarcinoma in women older than 40 years. Vaginal clear-cell adenocarcinoma has been diagnosed as a second primary cancer in a 42-year-old DES daughter [50]. It is unclear whether the risk for vaginal clear-cell adenocarcinoma will again increase in these women as they approach the age at which this cancer becomes more prevalent in the general population.

Several case–control studies have attempted to define risk factors for the development of cervicovaginal clear-cell adenocarcinoma among women exposed to DES. Herbst and coworkers [48] compared Registry patients with DESAD participants who were exposed in utero to DES and found that daughters exposed to DES in utero within the first trimester of pregnancy were at increased risk. Daughters exposed in utero in the sixth week of gestation or earlier appear to have the highest risk, whereas those exposed in later stages of gestation have progressively decreasing risk. Herbst and colleagues could not establish whether the timing of exposure or the total DES dose determined cancer risk [48, 51]. A maternal history of previous spontaneous abortion was also associated with an increased risk for clear-cell adenocarcinoma. No difference was noted in the age at which menarche occurred [51]. Sharp and Cole [47] compared Registry patients with DES daughters from Dieckmann and colleagues' cohort and the DESAD project. They found that greater height and adiposity at the age of 14 to 15 years were directly associated with risk for clear-cell adenocarcinoma. This is interesting because these are also risk factors for endometrial cancer [52].

Estimates of the number of cases of clear-cell adenocarcinoma are likely to be low. This may partly represent the misdiagnosis of women with this rare disorder [53]. According to Registry data and a recent survey of members of the Society of Gynecologic Oncologists (Trimble T. Unpublished data), cases of cervicovaginal clear-cell adenocarcinoma are probably under-reported. With reports from the registry and Society of Gynecologic Oncologists survey combined, 25 new cases were reported in 1990 and 23 were reported in 1991. Cancers recurred in approximately 10 patients per year. Although these women presumably have been exposed to hormones, in utero DES exposure has been documented in approximately 60% of Registry patients [49]. Herbst and Anderson [42] have reported that 90% of initially diagnosed women are found to have stage I or stage II disease. Ten-year survival in women with stage I disease has approached 90%, with survival decreasing markedly as the disease stage at diagnosis increases.

Vaginal adenosis, dysplasia, or squamous metaplasia has frequently been observed in DES daughters [54, 55]. Robboy and colleagues [56] found that adenosis was documented in the medical records of 3% of the participants in the DESAD project. These changes were more common among women exposed to large in utero DES doses or those exposed in an early gestation stage. Although clear-cell adenocarcinoma typically arises in areas of adenosis, progression from adenosis to cancer is not usually observed [57]. Prospective studies of women exposed in utero have suggested that the prevalence of adenosis decreases with age [58, 59].

The relation between cervicovaginal dysplasia and squamous-cell carcinoma of the cervix and vagina in DES daughters is more controversial. Although initial data did not indicate a increase in dysplasia [55, 60], a more recent review of daughters from the DESAD project has shown a twofold increase in the risk for squamous-cell carcinoma in situ [61]; invasive squamous-cell carcinoma of the vagina has also been reported in a DES daughter [62].

In one study [63], ovarian germ-cell cancer in young women has also been associated with DES; the odds ratio for in utero exposure to DES or other hormones was 3.6 (CI, 1.2 to 13.1). Peritoneal papillary serous carcinoma has been reported in a 34-year-old nulliparous woman who had in utero DES exposure [64].

Further follow-up of exposed daughters is clearly needed to define the risk not only for cervicovaginal cancers but also for cancer of the breast, ovary, and endometrium. Moreover, the effect, if any, of exogenous estrogens (oral contraceptives, drugs to induce ovulation, and replacement estrogens) in modifying cancer risk in these women is still unclear.

Diethylstilbestrol Sons

The association between in utero DES exposure and the development of testicular cancer is controversial. An increased risk for testicular tumors was observed in studies of mice exposed to DES [65, 66], and an increased incidence of cryptorchidism, a strong risk factor for testicular cancer, has been observed in DES sons [67, 68]. Several cases of testicular cancer in men exposed to DES have been reported [68, 69], and two cohort studies have each reported one case of testicular teratoma [40, 70]. The findings from six case–control studies that have examined the association between DES exposure in utero (or exposure to any exogenous hormone) and testicular cancer have been mixed. On the basis of interviews with mothers of 79 men with testicular cancer that was diagnosed between 1972 and 1974 and interviews with 79 matched controls, Henderson and associates [71] found a relative risk of 5.0 (P = 0.11; CI not given) for any hormone use during the pregnancy. Schottenfeld and colleagues [72] reported that drug use was associated with a relative risk of 1.96 (CI, 0.8 to 4.3) for bleeding, spotting, or threatened miscarriage during the pregnancy; however, 45% of case mothers and 63% of control mothers could not recall the specific drug taken. In a study of 108 cases diagnosed in California between 1973 and 1979 and 108 matched controls, Depue and coworkers [73] found that any exogenous hormone use during the first trimester was associated with an eightfold increased relative risk for testicular cancer; however, only two case mothers reported receiving DES. In contrast to these suggestive studies, three studies have found no effects of DES exposure on the risk for testicular cancer [74-76]. Moss and colleagues [75] reported an odds ratio of 0.9 for use of exogenous hormones during pregnancy; DES exposure, reported by four case mothers and two control mothers, resulted in a nonsignificant odds ratio of 2.0. Similarly, Brown and associates [74] found an odds ratio of 0.8 for hormone use, and Gershman and Stolley [76] reported an odds ratio of 0.8 for the receipt of drugs used to prevent miscarriage.

Problems common to these studies include the lack of documentation to confirm exposure and the difficulty of maternal recall of exposures during pregnancy. The relatively young age of the cohort at the time of study could also have precluded detection of an increased risk for testicular cancer, which in the general population peaks in the fourth decade of life. Although these studies probably rule out a high level of risk, they are based on fairly small numbers, which may preclude the detection of a low to moderate increase in risk associated with DES. No other cancer sites in men have been implicated.

Reproductive Dysfunction

Diethylstilbestrol Daughters

Developmental abnormalities in the female reproductive tract frequently occur after DES exposure. Among DESAD participants at the Baylor College of Medicine who were exposed to DES, 50 of 282 (18%) were found to have gross anatomical changes of the cervix (absent pars vaginalis, coxcomb, hypoplastic cervix collar, or pseudopolyp) [77-79]. Among a subgroup of DESAD participants recruited for a fertility study, 154 of 293 (53%) were found to have abnormal hysterosalpingograms. These abnormalities included T-shaped and hypoplastic uteri; constriction of the upper, middle, or cornual regions; and irregular uterine margins [79].

Data from the Dieckmann and colleagues' cohort have consistently shown reductions in fertility in DES daughters [80, 81]. On the basis of data analyzed until 1986, 33% of the exposed women compared with 14% of the unexposed women reported primary infertility. Secondary infertility was also reported significantly more often among the exposed women. Vaginal epithelial changes and cervicovaginal ridges were found more often among the exposed women with primary infertility. In contrast, an early analysis of data from the DESAD cohort [82] found that exposed and unexposed daughters were similar in the number achieving pregnancy, the total number of pregnancies, and age at first pregnancy. However, these women may have been studied too early in their reproductive life span to detect major differences in fertility.

Kaufman and associates [79] found no difference in the proportion of women with normal and abnormal hysterosalpingograms who had difficulty with conception, suggesting that structural abnormalities of the uterus alone did not account for failure to conceive. Some clinical studies and case reports have suggested that hormonal changes in DES daughters occur, including elevated testosterone and prolactin levels [83-85]. However, a prospective cohort study suggested that although in utero DES exposure was related to a reduction in the duration and amount of menstrual bleeding, exposure did not affect cycle length and variability of cycle length. This suggests that gross endocrine function was not disturbed [86]. Failure of implantation [87] and alterations in ovarian steroidogenesis have also been postulated as possible causes of infertility in these women [88].

Once pregnancy is achieved, DES daughters are at high risk for an unfavorable pregnancy outcome. In a review of English-language articles, Swan [89] estimated that, overall, DES daughters are 8.6 times more likely to have an ectopic pregnancy, 1.8 times more likely to have a miscarriage, and 4.7 times more likely to have a premature birth than unexposed women. Among women with an abnormality of the cervix, vagina, or uterus, the relative risks for ectopic pregnancy, miscarriage, and premature birth are even higher (13.5, 2.6, and 9.6, respectively).

Diethylstilbestrol Sons

Some investigators have reported abnormalities of the urogenital system in DES sons [39, 90-92], whereas others have found no increase in such abnormalities compared with men who were not exposed to DES [40]. Gill and coworkers [68] examined 308 men exposed to DES and 307 men receiving placebo who were traced from Dieckmann and colleagues' cohort and found that the prevalence of epididymal cysts and hypotrophic testes was four times greater among exposed men. In men with testicular hypoplasia, cryptorchidism was observed in 65% (17 of 26) of men exposed to DES compared with 17% (1 of 6) of controls. No significant differences were found in mean circulating follicle-stimulating hormone, luteinizing hormone, or testosterone levels in the two groups. Spermatozoa were analyzed in 134 men (44%) exposed to DES and in 84 men (27%) who received placebo. The average sperm density of the group exposed to DES was lower than that of the placebo group (91 sperm cells x 10^6/mL compared with 115 sperm cells x 10^6/mL; P = 0.05). Semen quality was compared using the average Eliasson score, a scoring system that assesses sperm concentration, percentage of motile sperm, motility, and morphology. A score of 1 is classified as normal; a score of 5 to 10 is classified as pathologic; and a score of greater than 10 is classified as severely pathologic. The average Eliasson score was higher in the group exposed to DES than in the group exposed to placebo (4.9 compared with 2.5; P = 0.01); more men exposed to DES than controls had severe semen pathologic disorders (an Eliasson score > 10) (24 of 134 men exposed to DES [18%] compared with 7 of 87 controls [8%]; P = 0.05). In contrast, a study done by the Mayo Clinic found no significant differences between men who were and were not exposed to DES in the proportion of testicular or penile anomalies, sperm density or Eliasson score, or the number of pregnancies attained by their wives [70]. These conflicting results may be related to differences in the maternal DES dose levels, heterogeneous hormone (non-DES) exposures, or different methods of recruiting study participants in the reported studies.

Alterations of Immune Function

Speculation that prenatal exposure to DES may lead to an altered immune status is based on observations that mice exposed perinatally to DES show reversible thymic and splenic atrophy, decreased T-helper cell subpopulations, and impaired natural killer-cell activity [93]. Interestingly, two small case series have shown altered T-cell and natural killer-cell function in women exposed to DES in utero [94, 95]. In reviewing data from the DESAD cohort, Noller and associates [96] observed that reporting of a lifetime history of autoimmune diseases was increased among persons exposed to DES in utero compared with unexposed controls. Overall, the rate of any autoimmune disease was 28.6 cases per 1000 persons among the exposed daughters and 16.9 cases per 1000 persons among the unexposed women, resulting in a relative prevalence rate of 1.8 cases per 1000 persons (CI, 0.99 to 3.1). Although no individual autoimmune disease was significantly associated with exposure, Hashimoto thyroiditis was reported by 10 of the 1711 women exposed to DES compared with 1 of 922 controls. No other clinical manifestations of immune dysfunction have been established. On the basis of the results of animal studies, however, Blair [97] has hypothesized that immune dysfunction in exposed offspring may only become clinically manifest with aging.

Psychosexual Effects

Several animal studies have suggested that DES exposure may affect the developing brain and central nervous system [98-100], resulting in masculinization of the female brain and the reverse effect in the male brain. In normal rodent development, most of the circulating ovarian estrogens are bound by -fetoprotein and do not reach the brain. In studies of mice, DES and its metabolites had a weak affinity for -fetoprotein and therefore may affect the developing brain [101]. In a study in infant rhesus monkeys, patterns of gonadotrophin secretion persistently changed after perinatal exposure to DES [102].

Few human studies have evaluated the possibility of psychological and sexual effects of DES exposure. A few small studies reported slightly more masculine behavior patterns among the exposed daughters [103], including an increased incidence of homosexuality and bisexuality; however, these studies had many methodologic problems. Another study reported that men who had been exposed to high doses in utero in a clinical trial of DES in diabetic mothers were significantly less likely to be married than untreated men. The proportion of men who were married was also significantly related to the timing and dose of DES exposure [104]. In contrast, Vessey and colleagues [40] found no difference in the marital status of offspring who had been exposed to DES as part of a randomized clinical trial in England during the 1950s. However, analyses of data from medical records found that psychiatric disorders, most notably depression and anxiety, were reported twice as frequently among the exposed daughters and sons. Because the offspring in this study were unaware of their exposure status, the result was not caused by anxiety about the exposure. Gustavson and associates [105] reported that the prevalence of eating disorders and extreme weight loss was approximately four times higher among the exposed persons in the DESAD cohort.

In summary, several reports of altered psychosexual development in persons exposed to DES have been published; however, these studies are far from conclusive. Most are based on small numbers, and many have major methodologic problems, including inappropriate comparison groups and possible response bias. In addition, most of the studies are plagued by the possibility that knowledge of and anxiety about the exposure itself, in combination with the physical changes to the reproductive tract that are associated with DES, may have resulted in psychological effects in some offspring.

Third-Generation Effects

The children born to men and women exposed in utero to DES are now in their midteens to twenties. Currently, no direct evidence exists for a transgenerational effect of DES in humans. However, in a potentially groundbreaking lawsuit, a man exposed in utero has alleged that his exposure caused the death of his 13-year-old daughter from clear-cell adenocarcinoma [106]. No reports of this case have appeared in the medical literature. Although some animal studies have suggested the potential for third-generation effects [107, 108], the relevance of these studies to humans is unclear. Tomatis [109] has reviewed the experimental and epidemiologic evidence for transgenerational carcinogenesis.

Recommended Follow-up of Persons Exposed to Diethylstilbestrol

Most DES mothers are in their mid-60s, and most DES daughters and sons are in their late 30s. However, because DES was used during a 30-year period, some mothers may be in their 40s and some daughters and sons, in their early 20s. It is important for health providers to query patients born in the late 1940s through the early 1970s and women who were of child-bearing age during this period about possible DES exposure. Verification of exposure should be obtained from medical records whenever possible. Because many mothers who were prescribed DES during pregnancy are unaware of their exposure [110], their children who were exposed in utero will also be unaware of their exposure. However, a provider may suspect in utero DES exposure in women on the basis of characteristic findings in the cervix or vagina that are observed during a pelvic examination. These persons should be followed in the same manner as those with known exposure.

On the basis of current knowledge, the American College of Obstetricians and Gynecologists proposed the following guidelines for the follow-up of women exposed to DES [111].

1) Diethylstilbestrol mothers: In light of the slight increased risk for breast cancer, the committee recommends that women who received DES during pregnancy be encouraged to do monthly breast self-examinations, have an annual breast physical examination, and follow established recommendations for screening mammography.

2) Diethylstilbestrol daughters: Careful inspection of the cervix and vagina using one-half strength aqueous Lugol (iodine) solution and palpation of the entire vaginal wall is recommended because most cancers in women under surveillance are detected in this manner. The committee recommends that in addition to an annual Papanicolaou test done in the cervix, women exposed in utero to DES should also have vaginal cytologic tests and colposcopy if the results are suspicious; biopsies should be done as indicated. Noller [112] has reviewed in detail the appropriate examination of these women and the role of colposcopy.

Because of the increased risk for spontaneous abortions, ectopic pregnancies, early cervical effacement, and premature labor, DES daughters should be followed as obstetric high-risk patients. Most women can be conservatively managed during pregnancy with good outcome [113]. Practitioners are encouraged to refer women with known in utero exposure to centers that have focused on the follow-up, treatment, and education of these women. Currently, data are insufficient to adequately advise DES daughters on the potential risk of hormonal contraceptives, ovulatory agents, and replacement estrogen therapy.

Although in utero DES exposure and subsequent development of cancer in men have not been clearly associated, periodic testicular self-examination should be encouraged. Like all other men, those exposed to DES should receive a testicular examination as part of a routine physical examination. This is especially true for men with a history of cryptorchidism.

Recommendations for medical care and follow-up of patients with DES-induced clear-cell adenocarcinoma are topics beyond the scope of this review. It is recognized, however, that additional clinical research is needed to optimize therapeutic approaches to increase survival, reduce morbidity, and preserve organs.