Dioxin exposure, from infancy through puberty, produces endocrine disruption and affects human semen quality.

The biological and socioeconomic characteristics of the two study groups were similar, except for a higher education level and lower occupation in the furniture-manufacturing industry in the comparison group ( Table 1 ). These differences did not affect the comparison between exposed and unexposed groups; the effect of the inclusion of these variables as covariates in the model was negligible. The incidence of self-reported varicocele or cryptorchidism were not statistically different in the exposed and comparison groups; however, we excluded these men from the analyses. The 1976 serum TCDD concentrations of eligible men who did and did not participate in the study were similar ( Figure 1 ). The TCDD concentrations ( Figure 2A, 2B ) were also comparable among exposed age groups. Median serum TCDD levels in 1998 ( Figure 2C, 2D ) were higher in males exposed in 1976 as adults than in males who were exposed as children. This is in agreement with the much shorter TCDD half-life in children ( Aylward et al. 2005 ; Kreuzer et al. 1997 ), but this observation did not explain any statistically significant effects. We assumed that serum TCDD concentrations for the comparison groups were ≤ 15 ppt in 1976–1977 ( Needham et al. 1997 ) and < 6 ppt in 1998/2002 on the basis of serum results for residents of uncontaminated areas around Seveso (Mocarelli P, unpublished data). Because the only dioxin-like chemical involved with the Seveso incident was TCDD, we focused on TCDD for these analyses. If TCDD acts in concert with other dioxin-like chemicals in affecting sperm quality, the total dioxin toxic equivalency (TEQ) should be considered. In nine serum pools from females residing in the uncontaminated area in 1976, Eskenazi et al. (2004) found an average TEQ of 100 ppt. In 71 men exposed at 1–9 years of age (mean, 6.2 years), serum TCDD concentrations (median, 210 ppt) had a significant effect on semen quality measured 22 years later. Indeed, significant decreases in sperm count ( p = 0.025), progressive sperm motility ( p = 0.001), and total number of motile sperm ( p = 0.01) were observed relative to the comparison group ( Table 2 ). Quartile distribution ( Figure 3A, 3C ) shows that serum TCDD concentrations ≤ 113 ppt (median of first quartile, 68 ppt) adversely affected sperm concentration and total motile sperm count. In contrast to the observed effects on men exposed at 1–9 years of age, exposure (median TCDD serum concentration, 164 ppt) at 10–17 years of age (mean, 13.2 years) resulted in effects that appeared to be stimulatory to semen parameters ( Table 2 and Figure 3B, 3D ). In 20 men 40–47 years of age who were exposed to TCDD (serum concentration, 15.5–1,310 ppt; median, 123 ppt) at 18–26 years of age (mean, 21.5 years), we found no statistically significant differences for any of the sperm variables compared with the 32 men in the comparison group. Moreover, no trends in these variables were related to different TCDD serum concentrations in 1998. Also, we observed no statistically significant differences for sperm morphology between exposed and comparison groups. Men exposed to TCDD at 1–9 and 10–17 years of age had lower serum E 2 concentrations ( p < 0.001) and higher serum FSH concentrations ( p = 0.055 and p = 0.038, respectively) than the comparison groups ( Table 2 ). We found differences in E 2 at TCDD concentrations < 53 ppt ( Figure 3E, 3F ). In contrast, subjects exposed at 18–26 years of age (mean, 21.5 years) showed no differences in concentrations of E 2 ( p = 0.248) or other hormones relative to the comparison group. Exposure status had no effect on testosterone or inhibin B concentrations in any group. The semen of men exposed to TCDD at 1–9 years of age presented significantly greater effects relative to their respective comparisons than semen from men exposed at 10–17 years of age ( Table 2 ). Indeed, the former group showed statistically significant lower sperm concentrations ( p = 0.008), total sperm counts ( p = 0.004), progressive motility ( p = 0.005), and total motile sperm counts ( p < 0.001) than the latter. No statistically significant differences were noted among the comparison groups ( Table 2 ).

This study on men from Seveso provides evidence of a permanent disruptive effect of TCDD on the human male reproductive system, depending on the age at exposure. Prepubertal children (< 9 years of age) are very sensitive to TCDD, with a reduction of sperm concentration and motility observed at serum levels < 68 ppt (equivalent to a body burden of about 12 ng/kg body weight). In contrast, exposure to TCDD during puberty causes an increase of these semen parameters. If men who are first exposed at 1–9 years of age continued to be exposed at 10–17 years of age to higher than background levels, we would expect the effects to balance out. However, this is not the case. One possible explanation is the much shorter TCDD half-life (months, not years) in young children. Therefore, some of the children exposed at 1–9 years of age may have had a low dose of TCDD (> 15 ppt background level) still present at puberty, which did not determine a stimulatory effect; the other possibility is the presence of a higher dose at puberty, which nevertheless did not produce a stimulatory effect. Therefore, this contributes to the hypothesis of a permanent effect ( Mocarelli et al. 2000 ); note the striking differences between men exposed at 1–9 years of age compared with those exposed at 10–17 years as shown in Table 2 . Indeed, exposure to endocrine-disrupting chemicals during the period when “programming” of the endocrine system is in progress may result in a permanent change of function or sensitivity to stimulatory/inhibitory signals ( Damstra et al. 2002 ). One consequence of these opposite effects in infancy compared with puberty could be that the action of dioxin and similar pollutants in the general male population is obscured because the two effects could cancel out each other to give an average normal appearance. However, in both age groups, TCDD exposure results in a significant reduction in serum E 2 levels in adulthood. It is important to note that the TCDD body burden and serum levels of these men were within background levels for that time period, demonstrating a permanent effect of the original low dose they received ( Figure 2 ). No effect was observed at all when exposure to TCDD occurred during adulthood. This study has several strengths. First, we have clearly and directly related original exposure levels of the ubiquitous environmental endocrine disruptor TCDD to reproductive outcomes years after exposure. Second, the participants of the study are fully representative of the available eligible population. Third, the TCDD concentrations affecting children, particularly boys, is similar to the maternal TCDD body burden that has been shown to induce a reduction in sperm numbers in adult rats exposed in utero and/or during lactation to TCDD ( Faqi et al. 1998 ; Mably et al. 1992 ; Roman and Peterson 1998 ; Theobald et al. 2003 ) and to serum concentration shown to decrease the sex ratio ( Mocarelli et al. 1996 and 2000 ) in offspring of TCDD-exposed men at Seveso. These serum concentrations are also lower than the concentrations shown at Seveso to induce a slight, nonstatistically significant increased risk of endometriosis ( Eskenazi et al. 2002 ) and breast cancer ( Warner et al. 2002 ) in women, and developmental dental aberrations in men exposed at ages younger than 5 years ( Alaluusua et al. 2004 ). However, the study may be weak because of sampling problems involving voluntary sperm analysis. A bias, mainly due to low compliance (∼ 20–40%), has been recorded using sampling as a representation of the general population ( Jorgensen et al. 2001 ). To deal with such bias, we chose healthy blood donors from a nearby area; this group of men showed a high compliance (∼ 60%) ( Table 1 ) and may be considered representative of the general healthy male population. In any case, we were able to overcome a possible bias by the observation of very significant differences between the 22- to 31-year-old and the 32- to 39-year-old exposed groups ( Table 2 ), whereas no differences were seen between the equivalent comparison groups. A possible role of chronic exposure to solvents or other toxic substances used in the furniture-manufacturing industry has been ruled out by similar exposure ( Table 1 ) and by multivariate statistical analysis. Currently, no data directly relate TCDD exposure at a young age with human sperm quality. The only similar data are those on effects of PCBs and phthalates: Hauser et al. (2005) reported a decrease of sperm motility after exposure to PCBs and phthalates as in our case; and Guo et al. (2000) observed alterations in sperm morphology after prenatal exposure to PCBs/PCDFs (unlike our data) after an incident in Taiwan in 1979. Also, in Taiwan, men exposed to PCBs/PCDFs at 18–30 years of age showed abnormal sperm morphology ( Huang et al. 2003 ). These researchers, however, did not measure PCB/PCDF concentrations at exposure and did not show modification in sperm number, as is the case in the present study. This effect, present in experimental animals ( Roman and Peterson 1998 ; Theobald et al. 2003 ), could have gone unnoticed because of the absence of exposure data on children in those studies. The contrasting effects of infant versus pubertal TCDD exposure on sperm count and the lack of effect in adults may have a physiologic explanation related to differences in the hormonal regulation of Sertoli cell proliferation with age ( Sharpe et al. 2003 ). Final Sertoli cell number is the main determinant (other than abstinence period) of sperm count in men ( Sharpe et al. 2003 ). Proliferation of these cells in humans occurs during three periods: fetal, postnatal (0–8 months of age), and probably prepubertal. Thus, in the present study, a similar exposure during the prepubertal period (average age, 6.2 years) suppresses Sertoli cell number, whereas exposure during the peripubertal period stimulates Sertoli cell number. This differential action may reflect diversities in the mechanisms that regulate Sertoli cell proliferation at these two time points. Androgens may be the primary stimulator of perinatal and prepubertal proliferation ( Atanassova et al. 2005 ), whereas peripubertal proliferation is driven principally by FSH ( Johnston et al. 2004 ). Differential effects of TCDD on androgen and FSH action in infancy compared with puberty may provide a ready explanation for the observed differences in sperm count. TCDD and other dioxin-like chemicals produce their effects primarily through the aryl hydrocarbon receptor (AhR). Activation of AhR by dioxin, therefore, could be a mechanism by which androgen action is reduced; this could explain the observed decrease in sperm count in adults who were exposed to TCDD as young children (i.e., when Sertoli cell development is more testosterone dependent). This hypothesis is supported by the observation that in utero exposure of human males to maternal smoking causes reduced sperm counts in the offspring at adulthood; this probably is a result of reduced Sertoli cell number ( Jensen et al. 2004 ; Storgaard et al. 2003 ) due to the action of polycyclic aromatic hydrocarbons present in cigarette smoke on AhR. In contrast, when TCDD contamination occurs at puberty, Sertoli cell proliferation is primarily FSH dependent. E 2 is a potent negative regulator of FSH secretion, and studies have shown that E 2 suppression of FSH can reduce Sertoli cell proliferation and number ( Johnston et al. 2004 ). TCDD-induced reduction of E 2 levels (and corresponding elevation of FSH levels), as shown in adults exposed during infancy or puberty (present study), may indicate that increased FSH levels during puberty may lead to increased Sertoli cell proliferation, and hence, to higher sperm counts in adulthood. Although a similar change may have occurred in boys exposed during infancy, the effect of FSH on Sertoli cell proliferation at this age may be insignificant and/or it may be counteracted by the negative repercussions related to suppression of androgen action. Exposure to TCDD after puberty (i.e., after completion of the reproductive system) would not modify estrogen concentration or semen quality, which is consistent with our results. Taken together, our data are consistent with the untested hypothesis that TCDD exposure during sensitive developmental “windows” may affect expression of responsive genes (with or without the effects of estrogens and/or androgens), permanently altering the programming of the primordial germ cells. The effect of AhR signaling could be stimulatory or inhibitory, depending on the interplay of factors that include the level of dioxin exposure, the period of sensitivity and/or development of the target cells, and the actual level of key regulatory molecules, including the androgen-estrogen balance. It could also explain the lack of effect of TCDD on spermatogenesis of the mature reproductive system and the “normal morphology” of sperm of exposed men.

Our results directly demonstrate a reduction in E 2 and a permanent effect on semen quality in human males as a result of the disruptive action of low concentrations of TCDD on the endocrine system. This occurs after exposure especially in infancy/prepuberty, less in puberty, and not in adulthood, at levels, until recently, that were seen in the general population of many industrialized countries. Our data could explain, at least in part, the reported reduction ( Andersen et al. 2000 ; Menchini-Fabris et al. 1996 ; Van Waeleghem et al. 1996 ) of semen quality of the youngest populations in Western countries. In fact, these data demonstrate that serum concentrations of about 100 TEQ are border limits; however, at these levels, effects on E 2 concentration and on the developing male reproductive system begin to be produced. Certain human populations, especially children during breast-feeding ( Link et al. 2005 ), may have a total body burden of dioxin-like chemicals close to this limit. Sensitive children can also be affected at lower concentrations; it will be of interest to see if, as a result of public health efforts in decreasing dioxin levels [from a TEQ level in children in Seveso in 1976 of about 100 ppt ( Eskenazi et al. 2004 ) to about 10 ppt in Germany in 2002/2003 ( Link et al. 2005 )], there will be a reversal in the reported reduction of semen quality. One remaining significant question will be to determine whether in utero exposure will affect human sperm quality.

A total of 397 Caucasian males (of the eligible 415) from the highly TCDD-contaminated A zone ( Di Domenico et al. 1990 ; Needham et al. 1997 ) and from nearby contaminated areas, all of whom were 1–26 years of age in 1976, were invited to participate in the study conducted in 1997–1998 ( Figure 1 ). Frozen serum samples (generally ≤ 1 mL in volume) from blood collected in 1976–1977 from these subjects were available for TCDD measurements. A total of 372 consecutive healthy volunteer blood donors the same age as the exposed men, but not living in TCDD-contaminated areas (i.e., they were not exposed to TCDD by the Seveso explosion) were also invited to participate ( Figure 1 ). All participants were first screened for any hidden disease by clinical laboratory tests for liver, bone marrow, kidney, and pancreatic functions. All participants with specific diseases or conditions ( Table 1 ) were excluded. The participants completed a questionnaire on health and socioeconomic status and donated blood and semen samples (samples were collected the morning after having been sexually abstinent for at least 3 days). The study protocol was approved by the Institutional Human Subjects Committee. All study participants gave written informed consent. Participants collected a postmasturbatory semen sample at home. Each sample was transported at approximately body temperature to the Desio Hospital laboratory and kept at 37°C until examination, which occurred within 1 hr after ejaculation and tests were performed in blind by the same two technicians according to the World Health Organization ( WHO 1982 ) recommendation. Ejaculate volume was calculated gravimetrically. Sperm motility was assessed at 400× magnification on a microscope heating stage (37°C) in duplicate, and the average value was recorded. Sperm concentration was measured using a Bürker-Türk chamber at phase contrast (400× magnification). Morphology was evaluated by the same observer on 300 Papanicolaou-stained sperm per slide ( David et al. 1975 ; Jouannet et al. 1988 ). Fasting blood samples were obtained on the same morning as semen collection. An aliquot of serum was stored at −80°C and analyzed for hormone levels in large batches to reduce interassay variability. Serum 17β-estradiol (E 2 ), follicle-stimulating hormone (FSH), inhibin B, and luteinizing hormone (LH) were measured according to established immunofluorimetric methods, and testosterone was measured by radioimmunoassay. Quality control protocols were applied with strict criteria for all tests. Vials containing 0.6–1.0 mL serum samples stored frozen since 1976–1977 were analyzed for TCDD by isotope-dilution mass spectrometry at the Centers for Disease Control and Prevention ( Patterson et al. 1987 ). Serum TCDD concentrations (parts per trillion on serum lipid basis) were also determined in samples drawn in 1997–1998 from all individuals whose 1976–1977 serum TCDD value exceeded 15 ppt [then the “background level” ( Needham et al. 1997 )] and in pooled samples of men from uncontaminated areas to assess the background levels in 1998/2002. The samples with concentrations less than the detection limit were assigned a value half of that limit. We established and maintained a general database using SAS software (version 8.2; SAS Institute Inc., Cary, NC, USA). The exposed and comparison groups were divided, according to the developmental stage of the reproductive system ( Sharpe et al. 2003 ), into three 1976 age classes: infancy/prepuberty, puberty, and young adult (1–9, 10–17, and 18–26 years of age, respectively). Sensitivity analyses were performed to test the cutoff among the age groups. Sperm and hormone data were fitted with a general linear model including group, age class, interaction group × age class as terms, and with abstinence length (not considered for hormone analysis), smoking status (total number of cigarettes smoked per day during months of habitual smoking), body mass index (BMI), and chronic exposure to solvents and other toxic substances in the furniture-manufacturing industry as covariates. We applied scale transformations to approximate normal distribution and homoscedasticity: sperm concentration, total sperm count, progressive motile sperm count, and concentrations of E 2 , testosterone, and FSH were log-transformed; semen volume and concentrations of LH and inhibin B were square-root–transformed. Results were expressed as back transformation of least squares means (i.e., the means adjusted for all the terms in the model). Two families of comparisons were considered: “among groups within age-class” and “among age classes within group.” According to the Bonferroni principle, a 0.025 comparison-wise risk of type I error ensures a family-wise risk of type I error ≤ 0.05.