Varicocele, urethral stenosis, and structural abnormalities of the penis, including hypospadias, were not associated with prenatal DES exposure (Table ​(Table1).1). In contrast, the RRs for exposed versus unexposed men were 1.9 (95 percent CI 1.1-3.4) for cryptorchidism and 2.5 (95 percent CI 1.5-4.3) for epididymal cyst. Only four men reported both conditions. Elevated RRs for these outcomes were observed in all three cohorts (data not shown).

DES exposure was not associated with the occurrence of benign prostatic hypertrophy or with inflammation or infection of the urethra, epididymus, or prostate. There was a positive association with inflammation or infection of the testes (RR = 2.5, 95 percent CI 1.5-4.4).

As shown in Table ​Table2,2, associations of DES exposure with cryptorchidism, epididymal cyst, and infection/inflammation of the testes were stronger for men who were first exposed before the 11^th week of gestation than for men first exposed later in gestation. Stronger associations were also observed for cumulative exposures of five or more grams of DES.

A previous investigation of participants in the Mayo Clinic study reported null findings regarding a possible association of prenatal DES exposure with risk of epididymal cyst, cryptorchidism, or other urogenital abnormalities [6]. DES doses were generally low in the population served by the Mayo Clinic, and the timing of first exposure varied greatly [8]. We carried out separate analyses among participants from the Mayo cohort to assess whether timing of DES exposure would explain the earlier results. Whereas there were no statistically significant associations for DES exposure overall, epididymal cyst and cryptorchidism were significantly associated with prenatal DES exposure that began before the 11^th week of gestation (Table ​(Table3).3). Cumulative dose information was available for only 20 percent of Mayo participants, and the cumulative dose was 5 grams or higher for only one case of cryptorchidism and one case of epididymal cyst.

Studies of rats [11] and mice [12] given DES during pregnancy found that the male offspring had a higher than expected prevalence of cryptorchidism and epididymal cysts. The occurrence of epididymal cysts may be due to the persistence of mullerian remnants. It has been hypothesized that exogenous female sex hormones, acting transplacentally, might interfere with the action of mullerian inhibiting factor [13]. Normal descent of the testes is, at least in part, under hormonal control [14]. The first stage, transabdominal descent, occurs during the first trimester of pregnancy under the influence of insulin-like factor 3, which is inhibited by 17 β-estradiol in mice embryonic cells. The second stage, inguinal-scrotal descent, is induced by androgens and usually occurs during weeks 26-35. In a case-control study of boys with cryptorchidism, Bernstein et al. found that the mothers of cases had higher levels of free and albumin-bound estradiol during pregnancy as compared with controls, but there was no difference in total estradiol concentration [15]. A similar study found higher levels of total estradiol in the mothers of cases overall but did not find a difference for levels during the first trimester of pregnancy [16]. A third study also found no differences between cases and controls in mother's total estradiol in the first trimester of pregnancy [17]. DES is a more potent estrogen than estradiol and may have other chemical effects as well.

An examination study of 308 DES-exposed and 307 unexposed sons of women who participated in the University of Chicago randomized trial indicated a higher prevalence of epididymal cysts and of hypoplastic testes in exposed relative to unexposed men [4]. Sixteen years later, a questionnaire study of the same cohort of men also indicated an association between DES exposure and these outcomes [5]. This study found the observed associations to be stronger among men whose mothers DES treatment had started less than 11 weeks after their last menstrual period as compared with men whose mothers had begun taking DES later in the pregnancy. Subsequent to the examination study from the Chicago cohort, a separate examination study was carried out in 265 exposed and 274 unexposed men who had been delivered at the Mayo Clinic between 1939 and 1962 [6]. Contrary to findings from the University of Chicago study, the Mayo Clinic study reported no association between prenatal DES exposure and epididymal cyst or anomalies of the testes (including cryptorchidism) [6].

The present findings help to resolve previous discrepancies in results. The weaker association observed in the Mayo Clinic cohort in the present study and the conflicting results from earlier studies can be explained by differences in timing and cumulative dose of DES. Women treated at the Mayo Clinic typically took DES for a few weeks to a few months and their median cumulative dose was 720 mg [6]. For the University of Chicago trial, women were instructed to begin taking DES at entry into the trial, usually within the first trimester of pregnancy, and continue with increasing doses until a week or two before delivery [7]. The typical cumulative dose for this cohort was around 11 grams. The third cohort making up the present study consisted of offspring of women from an infertility practice in which the usual protocol was to take DES throughout the pregnancy, resulting in a high cumulative dose. Therefore, the Mayo Clinic cohort differs from the other cohorts in two major ways: only 48 percent were exposed before the 11^th week of gestation (as shown in Table ​Table3)3) and very few were exposed to 5 grams or more. When the analysis of Mayo Clinic sons was stratified on timing of first exposure, it became clear that there was indeed an association of DES exposure with occurrence of cryptorchidism and epididymal cysts if exposure began before the 11^th week of pregnancy, and that the magnitude of the association was similar to that observed in the overall study data.

Interestingly, timing of in utero exposure to DES has also been shown to be a predictive factor for structural anomalies of the lower genital tract in exposed daughters. In a study of structural anomalies of the cervix and vagina in DES-exposed daughters, investigators found a higher prevalence of anomalies (44 percent) in women who were prenatally exposed to DES before the 15^th week of gestation, compared to prevalences of 22 percent for weeks 15-22 and 5 percent for weeks 23 and greater [18].

The chief limitation of our study is reliance on self-reports for all outcomes. It is certainly possible that urogenital abnormalities may be more likely to come to diagnosis among DES-exposed men, and that DES-exposed men may be more likely than unexposed men to remember and report such abnormalities or inflammation. The fact that associations were observed for only three of ten conditions suggests an explanation other than reporting bias. Furthermore, the stronger associations observed for exposure that began early in pregnancy and for high cumulative dose of DES would not be explained by reporting bias since it is highly unlikely that there would be differential reporting of outcomes according to dose or timing.

The association with inflammation/infection of the testes may be a chance finding since we examined a number of outcomes and did not have an a priori hypothesis for finding an association with this endpoint. In addition, we did not have information on the specific conditions or when they occurred. It is possible that minimal structural abnormalities, such as minor obstructions, which may be associated with prenatal DES exposure, could explain this association. Further follow-up is needed to establish whether the association with inflammation/infection of the testes is due to chance or is causal. It is reassuring, however, to find no evidence of an increase risk of benign prostatic hypertrophy in the DES-exposed group.

ALH, KLN, RT, LTE, EEH, and RNH participated in the design of the study, acquisition of data, interpretation of results, and had critical input into the manuscript.

DAB and LAW participated in data analysis, interpretation of results, and had critical input into the manuscript.