Hormonally Active Agents in the Environment (1999)

Page 119

5—
Effects on Reproduction and Development

The harmful effects of exposure to environmental contaminants on reproduction and development in wildlife populations have been reported in the scientific literature for many years. Reported reproductive disorders in wildlife have included morphologic abnormalities, eggshell thinning, population declines, impaired viability of offspring, altered hormone concentrations. and changes in sociosexual behavior.

Laboratory experiments replicating the adverse effects of exposure to the potent synthetic estrogen diethylstilbestrol (DES) during critical periods in development (Newbold 1995; also see Appendix) have focused attention on the potential of chemicals with estrogenic properties to cause developmental and reproductive hazards.

The adverse consequences of prenatal exposure to DES on the female genital tract in humans have been reviewed in detail by Herbst and Bern ( 1981 ) and by Mittendorf (1995); they are the subject of continued, intensive investigation. Whether exposure to environmental hormonally active agents (HAAs) affects animals and humans similarly is not clear, but because exposure of animals to DES causes alterations in male and female offspring, the possibility must be considered that there will be adverse effects from exposure to other compounds with estrogenic, antiestrogenic, or antiandrogenic activity. There are also concerns that exposure to low doses of certain chemicals at critical stages in organ development can result in abnormalities that lead to irreversible changes in the functioning of organ systems later in life. Such damage would not occur through genetic mutations, but by processes that regulate genes during development and cell differentiation. The effects of hormones in adults are usually transient, and hormonal effects disappear when the chemical is not present. By contrast, environmental chemicals that alter gene activity during development would producecontinue

Page 120

effects much harder or impossible to reverse. Evaluating the effects of such chemicals is more difficult than evaluating the effects of chemicals on adults. Effective doses may be lower than effective doses in adults, and the effects are considerably removed in time from the exposure, which can make causal relationships more difficult to establish.

This chapter is a critical analysis of the literature on the link between exposure to HAAs and reproductive and developmental effects observed in laboratory studies, in humans, and in wildlife populations. Only a few (primarily estrogenic) HAAs are covered in this chapter, and evaluations of wildlife are limited to only a few vertebrate species. Although endocrine systems are remarkably well conserved among vertebrate groups, there are significant differences in their operation. While gonadal reversal does not occur in mammals, data from Seveso (Mocarelli et al. 1996) and occupationally exposed cohorts (Goldsmith et al. 1984: Potashnik et al. 1984; Potashnik and Porath 1995) suggest that some selective process may be involved to alter the sex ratio at birth. Hormonal control of sex differentiation is different in birds than it is in mammals, even though the same hormones (for example, estradiol) are involved. There are other important differences in the development of the reproductive systems of various vertebrate groups. Thus, HAAs could affect different vertebrate groups in different ways. For this reason, it is necessary to understand not only the effects of exposure to different HAAs on vertebrate development and reproduction but also the various effects of exposure to single HAAs on reproduction and development in different vertebrate groups.

Laboratory studies are discussed for specific HAAs, including some chemicals known to bind to estrogen receptors: insecticides (dichlorodiphenyl-trichloroethane (o,p'-DDT), methoxychlor, and chlordecone); a monomer used in plastic (bisphenol A); an alkylphenol surfactant used in detergents, cosmetics and toiletries, and other household products (octylphenol); and a plasticizer (butyl benzyl phthalate (BBP)). Other compounds are known to bind to androgen receptors: the fungicide vinclozolin and 1,1 -dichloro-2,2-bis(p-chlorophenyl)ethylene (p,p'-DDE). the persistent in vivo metabolite of DDT. Polychlorinated biphenyls (PCBs) and 2,3,7.8-tetrachlorodibenzo-p-dioxin (TCDD) also could disrupt development via several mechanisms. These compounds were selected for review because they are among the most extensively studied HAAs. The work discussed below illustrates the developmental and reproductive effects that can be caused by exposure to these estrogenic, antiestrogenic, and antiandrogenic agents in vivo. The list of compounds discussed in this chapter is by no means complete, and it might not even be representative of all HAAs.

The human studies evaluated here involve exposure to DDE, PCBs, and TCDD. Also evaluated are data on regional and temporal variations in sperm concentration in human populations. There is a discussion of adverse reproductive effects observed in wildlife populations. Where available, laboratory studies related to these findings are presented. The section on wildlife describes effectscontinue

Page 121

of exposure to HAAs in some representative vertebrate species: fish; birds; alligators; turtles; salamanders, frogs, and toads; and the Florida panther.

In evaluating the information, it is essential to examine the data that link reproductive and developmental effects to critical periods of exposure and to concentrations ordinarily found in the environment; however, for the most part this information is not available (see Chapter 1 for more details). In the cases where cause-and-effect relationships can be established or posited, known and suspected toxicologic mechanisms are discussed.

Laboratory Animal Studies of Selected HAAs

Dichlorodiphenyltrichloroethane (DDT)

Immature female rats injected intraperitoneally with a single dose of the estrogenic DDT isomer o,p'-DDT at 1 mg/kg had a significant increase in uterine wet weight (Welch et al. 1969), and newborn female rats injected subcutaneously with 1 mg/d o,p'-DDT on d 2-4 after birth had early onset of puberty and accelerated loss of fertility, referred to as ''delayed anovulatory syndrome" (Heinrichs et al. 1971). Gellert et al. (1974) report that subcutaneous injection of 0.1 mg of o,p'-DDT on d 2-4 of life led to marked impairment of fertility and reduced weight of prostate and seminal vesicles in male rats.

Male offspring from pregnant rats fed 100 mg/kg/d p.p'-DDE, the antiandrogenic metabolite of DDT. on gestation d 14-18 had reduced anogenital distance and had nipples (androgen normally blocks nipple retention in male rodents) (Kelce et al. 1995). Weanling male rats given daily doses of 100 mg/kg/ d p,p'-DDE by gavage until d 57 had delayed onset of puberty, and castrated adult rats given daily doses of 200 mg/kg/d p,p'-DDE by gavage for 4 d had decreased seminal vesicle and prostate weight (Kelce et al. 1995).

Exposure to DDT during gestation also has been shown to impair locomotor ability in mice (Tilson et al. 1979) and learning in rats (Lilienthal et al. 1990; Lilienthal and Winneke 1991) and monkeys (Schantz and Bowman 1989; Schantz et al. 1989). These studies are discussed in Chapter 8.

The mechanism by which DDT, or DDE, causes structural or functional abnormalities of the reproductive system in laboratory animals is still poorly understood. As discussed in Chapter 2, the DDT isomer o,p'-DDT has estrogenic properties: the DDT metabolite p,p'-DDE acts as an antiandrogen. Kelce et al. (1995) reported that a concentration of 3.5 µM p,p'-DDE occupied 50% of androgen receptors in rat prostate cells. This was approximately 200 times lower than the concentration of p,p'-DDE required to occupy 50% of estrogen receptors in rat uterine cells. Soto et al. (1997) also showed that p,p'-DDE is a partial agonist of estrogen. They reported that a 10-µM dose of p,p'-DDE was required to produce an increase in proliferation of MCF7 human breast-cancer cells, but the increase in proliferation was only 25% of the maximum proliferative responsecontinue

Page 122

seen at saturating doses of estradiol. Taken together, these findings show that the capacity for p,p'-DDE to bind to and interfere with the functioning of androgen receptors is considerably greater than its capacity to bind to estrogen receptors and stimulate estrogenic responses. The primary activity of p,p'-DDE as an HAA is thus as an environmental antiandrogen, not as an environmental estrogen. The antiandrogenic properties of p,p'-DDE might be of greater importance than the estrogenic properties of DDT on developing animals. p,p'-DDE persists for decades in tissues, whereas estrogenic o,p'-DDT is much less commonly detected in human serum (Stehr-Green 1989). However, exposure to o,p'-DDT, as well as to other nonpersistent pesticides (methoxychlor, for example), during critical periods in development affects fetal development in mice, and some effects, such as changes in territorial behavior, become apparent only in adulthood (vom Saal et al. 1995).

Methoxychlor

Methoxychlor is an insecticide used in home gardens and on crops and livestock (ATSDR 1994). The effects of methoxychlor on the reproductive systems of female rats and mice have been studied extensively (Cummings 1997). Methoxychlor causes adverse effects on fertility, early pregnancy, and in utero development. Accelerated pubertal ovulation, persistent vaginal cornification. accelerated loss of fertility, and abnormal cell types in the uterus and oviducts have been observed after neonatal administration of doses as low as 0.5 µg/d per neonate (Welch et al. 1969; Gellert et al. 1974: Gray et al. 1989; Eroschenko and Cooke 1990; Gray 1992). When administered to mated female rats during the peri-implantation period, a 300-mg/kg/d (approximately 100 mg/d) dose of methoxychlor completely blocked implantation of embryos (Cummings and Gray 1989; Gray et al. 1989; Cummings 1990).

Exposure to methoxychlor during development also leads to changes in the reproductive system and behavior of male rats and mice. Male offspring of mice fed 20 µg/kg/d methoxychlor in oil during the last third of pregnancy exhibited an increase in territorial marking behavior in adulthood, similar to the effect observed with a 20-µg/kg/d dose of o,p'-DDT and a 20-ng/kg/d dose of DES (vom Saal et al. 1995). Daily intraperitoneal injection of 1 mg of methoxychlor to male mice during the first week after birth led to reduced serum testosterone concentrations and to reduced DNA content in prostate and seminal vesicles in adulthood (Cooke and Eroschenko 1990). Administration of 50 mg/kg/d methoxychlor to female rats throughout pregnancy and lactation resulted in smaller testes, epididymides, and reduced sperm count in male offspring (Gray et al. 1989; Gray 1992).

The mechanism by which methoxychlor affects the reproductive system and reproductive behavior of laboratory animals is not understood. Methoxychlor has estrogenic effects in vivo only after demethylation in the liver to mono-soft

Page 123

hydroxy-methoxychlor (30% of administered dose) or bis-hydroxy-methoxychlor (23% of administered dose) (Kapoor et al. 1970); the more potent estrogenic metabolite is bis-hydroxy-methoxychlor (Welch et al. 1969; Bitman and Cecil 1970; Bulger et al. 1978; Bulger and Kupfer 1983). Unlike p,p'-DDE, methoxychlor is not persistent in vivo (most is cleared within 24 h), although it is relatively persistent (a number of months) in soil (Muir and Yarechewski 1984; ATSDR 1994).

Chlordecone

Chlordecone [decachlorooctahydro-1,3,4-metheno-2H-cyclobuta(cd)pentalen-2-one] was used in the 1960s and 1970s to control insect pests of bananas, citrus trees without fruit, tobacco, and ornamental shrubs (ATSDR 1995). The insecticide mirex, which is similar in structure to chlordecone, has been used in much greater quantities to combat fire ants.

When a 15-mg/kg/d dose of chlordecone was fed to pregnant rats on d 14-20 of gestation, 12 of 21 female offspring developed persistent vaginal estrus; the other nine rats were anovulatory at 6 mo of age (Gellert and Wilson 1979). The effects were consistent with those seen after exposure to estrogen, but no estrogenic effects were observed among male offspring. In a study of postnatal exposure, constant vaginal estrus was induced in mature female rats fed 1.5 mg/ kg/d chlordecone for 7 d (Hammond et al. 1979). Mature female mice injected with 125 µg/d chlordecone on postnatal d 1-10 developed complete cornification of the vaginal epithelium—this was similar to the effects caused by treatment with 10 µg/d estradiol (Eroschenko and Palmiter 1980). In males, spermatogenesis was completely suppressed by estradiol and was reduced by chlordecone. Injection of either 0.2 or 1 mg/kg/d chlordecone into newborn female rats on d 2 and 3 of life led to early onset of puberty and accelerated loss of cyclicity (Gellert 1978a).

In a study using quail, Eroschenko (1981) fed male quail a diet containing 200 ppm chlordecone for 3 wk and reported a significant increase in testes weight, due to edema, with dilation of the seminiferous tubules and erosion of the germinal epithelium. Abnormal sperm was also observed. With exposure for 6 wk, the testes in some animals began to atrophy.

The binding affinity of chlordecone to the estrogen receptor is approximately 0.02% (5,000-fold) lower relative to estradiol (Eroschenko and Palmiter 1980).

Vinclozolin

Vinclozolin (3-(3,5-dichlorophenyl)-5-methyl-5-vinyl oxazolidine-2,4-dione) is a dicarboximide fungicide widely used to combat damage to a variety of commodities, such as fruits, vegetables, hops, and turf. When vinclozolin was administered via gavage to pregnant and lactating rats from gestation d 14 throughcontinue

Page 124

postnatal d 3 at 100 or 200 mg/kg/d, male offspring were indistinguishable from female offspring on external examination at birth, and the males retained nipples (Gray et al. 1994). These findings indicate that the masculinizing effects of androgen on the external genitalia and the defeminizing action of androgen on the development of nipples were blocked by vinclozolin. In adulthood, due to gross abnormalities of the internal and external genitalia, the vinclozolin-treated males were infertile.

Vinclozolin is an androgen receptor antagonist (Kelce et al. 1994). After ingestion, it is degraded to metabolites that compete with endogenous androgen for binding to androgen receptors.

Polychlorinated Biphenyls (PCBs)

PCBs are no longer legally manufactured in the United States, but large quantities were produced for use in such products as electrical transformers and capacitors. Some of the more highly chlorinated of the 209 potential PCB congeners are highly persistent, and they bioaccumulate in the food chain.

Rats exposed to PCB mixtures early in life can develop reproductive effects similar to those caused by DES exposure (Bitman and Cecil 1970: Sager 1983: Sager et al. 1987, Subramanian et al. 1987; Lundkvist 1990: Jansen et al. 1993: Bergeron et al. 1994: Birnbaum 1994: Gray et al. 1995; Li and Hansen 1996). Male offspring of rats fed 32 or 64 mg/kg/d PCBs during lactation had significantly reduced seminal vesicle weight and significantly larger testes (Sager 1983; Sager et al. 1987). Those effects might be due to changes in thyroid hormone levels (Jannini et al. 1993). When treated males were mated with unexposed females, there was a significantly lower incidence of implantation, a significantly lower number of live births, and a significantly greater rate of resorption. Female offspring of dams exposed to 32 or 64 mg/kg/d PCBs during lactation had delays in puberty, vaginal opening, and first estrus (Sager and Girard 1994). At maturity, uterine wet weight was reduced at all stages of the estrous cycle, and fertility was impaired because of reduced success at the pre- or postimplantation stage. Female rats treated intrapcritoneally with PCBs also had significant increases in uterine weight (Jansen et al. 1993).

When female guinea pigs were force-fed 2.2 mg/d ( 1.8-3.2 mg/kg/d) PCBs during gestation, female offspring had delayed vaginal opening and male offspring had significantly reduced absolute and relative testis weight (Lundkvist 1990).

Pre- and postnatal exposure to coplanar PCBs can modulate thyroid hormone concentrations and uptake. When pregnant rats were orally administered 0.2, 0.6. or 1.8 mg/kg/d HCB (3,3',4,4',5.5'-hexachlo-obiphenyl) or a combination of 1 mg/kg/d TCB (3,3',4.4'-tetrachlorobiphenyl) and 0.6 mg/kg/d HCB, there were decreases in fetal, neonatal, and weanling plasma total thyroxine (T₄) and free T, concentrations, which indicated an increase in peripheral T₄ metabolism (Morsecontinue

Page 125

et al. 1993). An increase in the activity of type II thyroxine 5'-deiodTMase in brain homogenates also was observed, which suggests that local hypothyroidism occurs in the brains of fetal and neonatal rats exposed to these PCBs. The PCB mixture Aroclor 1254, administered orally at doses of 5 or 25 mg/kg/d to pregnant rats on d 10-16 of gestation, caused the selective accumulation of a hydroxylated PCB metabolite (2,3,3',4,5'-pentachloro-4-biphenylol) in fetal plasma and brain; this was believed to be the cause of the reductions of fetal plasma and brain T₄ concentrations (Morse et al. 1996).

In a reproductive toxicity study of rhesus monkeys, 80 females were fed 5-80 µg/kg/d Aroclor 1254 prior to breeding, during breeding (with untreated males), and after breeding, for 6 yr. There was a significant dose-related decreasing trend in conception rate and a significant dose-related increasing trend in fetal mortality (Arnold et al. 1995). Maternal age was not a confounding factor. It was noted during this study that while some of these animals had endometriosis the incidence or severity of the lesions could not be related to the PCB treatment. Similarly, a recent study in women led to the conclusion that exposure to PCBs and to chlorinated pesticides is not associated with endometriosis in the general population (Lebel et al. 1998).

In addition to reproductive effects, prenatal exposure to PCBs has been shown to cause deficits in neurodevelopment, such as impaired learning and altered activity levels in rats fed PCB-contaminated fish (Tilson et al. 1990; Daly 1992). These studies are described in greater detail in Chapter 6.

As discussed in Chapter 2, there is evidence that PCB mixtures and congeners and hydroxy-PCBs can have estrogenic and antiestrogenic properties. In addition, coplanar PCBs are antiestrogenic through aryl hydrocarbon-estrogen receptor crosstalk (Krishnan and Safe 1993). However, the significance of results from laboratory studies of PCBs are difficult to interpret because most PCB extracts from environmental samples do not resemble commercial PCB mixtures used in laboratory studies (WHO 1993). As PCBs are cycled through the environment, the various congeners are gradually redistributed. The most chlorinated congeners, which are typically the most toxic, accumulate preferentially.

2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD)

TCDD is a byproduct of the production of chlorinated products such as herbicides and wood preservatives, the incineration of trash containing papers and plastics, and the burning of fossil fuels (e.g., IARC 1997). A series of studies (Mably et al. 1992a,b,c) has shown effects on male rats whose mothers were given single oral doses of TCDD ranging from 0.064 µg/kg to 1 µg/kg on gestation d 15. At doses as low as 0.16 µg/kg, impaired sexual differentiation was observed in male fetuses, including a decrease in circulating testosterone and in anogenital distance at birth (Mably et al. 1992a). Effects on sexual behavior in male offspring in adulthood included changes in mounting, intromitting, numbercontinue

Page 126

of ejaculations, and latency to ejaculation, as well as in the exhibition of the female sexually receptive posture (lordosis) (Mably et al. 1992b). In addition, a dose-related decrease in the weight of the testis and epididymis was observed. There was also a decrease in daily sperm production in male offspring of pregnant rats receiving a single dose of 1 µg/kg, although there was no effect on fertility (Mably et al. 1992c).

Maternal exposure to a single oral dose of 1 µg/kg TCDD on gestation d 8 or 15 caused delayed puberty, partial clefting of the penis, and "thread" tissue development across the opening of the vagina in female offspring of rats (Gray and Ostby 1997). Ovarian weight was significantly reduced. In male offspring, the same treatment on gestation d 15 caused delayed puberty and reductions in ejaculated and epididymal sperm counts and in sex accessory gland size (Gray et al. 1995).

Chronic oral exposure to TCDD caused endometriosis in rhesus monkeys, with incidence and severity related to dose (Rier et al. 1993). Specifically, adult female monkeys were administered 2.5 x 10-7 and 1.25 x 10)-6 mg/kg/d in their food over 4 yr. Ten years later, moderate to severe endometriosis was found during laparoscopic examination in three of seven monkeys exposed to 5 ppt TCDD and in five of seven monkeys exposed to 25 ppt TCDD. None of the control monkeys showed severe disease.

The mechanism by which TCDD causes reproductive impairments and affects reproductive behavior is poorly understood. Because outcomes of prenatal TCDD exposure in rats appear similar to effects seen after treatment with DES (Peterson et al. 1993; Eskenazi and Kimmel 1995; Gray et al. 1995), the general assumption that TCDD acts as an estrogen antagonist might not apply to all effects of TCDD. In contrast, studies have shown that TCDD causes antiestrogenic responses in the rodent uterus (Gallo et al. 1986; Romkes et al. 1987: Umbreit et al. 1988, 1989; Astroff and Safe 1990; DeVito et al. 1992) and in breast cancer cells (Biegel and Safe 1990; Harris et al. 1990: Safe et al. 1991: Gierthy et al. 1993: Merchant et al. 1993) and inhibits mammary tumor growth in rodents (Kociba et al. 1978; Gierthy et al. 1993; Holcomb and Safe 1994; Tritscher et al. 1995). TCDD causes antiestrogenic effects via the aryl hydrocarbon (Ah) receptor signaling pathway, which has been characterized at the molecular level (Krishnan et al. 1995; Gillesby et al. 1997).

Bisphenol A

Bisphenol A (4,4'-(1 -methylethylidene)bisphenol) is a monomer used in the manufacture of polycarbonate plastic, resins, and some dental sealants; it is an additive in numerous other products. In a developmental toxicity study with pregnant rats and mice (Morrissey et al. 1987), gastric intubation of 1,250 mg/kg/ d bisphenol A on d 6-15 of gestation significantly decreased maternal body weight, increased maternal mortality and resorption of fetuses, and decreased thecontinue

Page 127

weight of surviving pups in mice but not in rats. There were no observable malformations in exposed fetuses.

In a subsequent study that used much lower doses of bisphenol A (Nagel et al. 1997), male offspring of pregnant mice fed 2 µg/kg/d bisphenol A on d 1 11-17 of gestation had significantly increased prostate weight as adults. In addition, 2 µg/kg/d bisphenol A produced significant enlargement of the preputial glands in male offspring, whereas the epididymides were significantly reduced. A dose of 20 µg/kg/d bisphenol A reduced daily sperm production per gram of testis by 20%, while daily sperm production uncorrected for testis weight was not significantly different (vom Saal et al. 1998). The investigators suggested that these doses could be within the range encountered by humans, as evidenced by amounts detected in the saliva of 18 patients (amounts ranged from 90 to 931 µg) treated with dental sealants (Olea et al. 1996) and in a few lacquer-coated cans of vegetables (amounts ranged from 0 to 23 µg/can) (Brotons et al. 1995). Bisphenol A does not bind to plasma-binding proteins or other components of blood to the same degree that estradiol does, and therefore, it passes more readily from blood into cells (Nagel et al. 1998). The inhibition of the uptake of steroids into cells by components of blood is particularly important during fetal life in rats, when the concentrations of gonadal and adrenal steroids are great but the bioactive fraction of steroid that can enter cells is maintained at a low concentration (vom Saal et al. 1992).

Effects of bisphenol A are seen in vitro in human breast-cancer MCF7 cells at 10-8 M (10 nM) or 2.3 ppb (molecular weight, 228) (Krishnan et al. 1993; Olea et al. 1996), and in F344 rat prolactin-secreting GH3 pituitary cells at 1 nM (Steinmetz et al. 1997). The finding of increased prostate size in male offspring of pregnant female mice fed 2µg/kg of bisphenol A (Nagel et al. 1997) is similar to that found in tests with DES at a dose of 0.02 µg/kg (vom Saal et al. 1997), which shows that bisphenol A is approximately 100-times less potent than DES when fed to pregnant mice. The same 2 µg/kg dose of bisphenol A fed to pregnant female mice also advanced the timing of puberty in female offspring, an effect seen with higher doses of other estrogenic chemicals (Howdeshell et al. 1999).

In another study (Steinmetz et al. 1997), estradiol and bisphenol A were administered to F344 and Sprague-Dawley rats via subcutaneous Silastic capsules, and serum prolactin, which is elevated by estrogen treatment, was measured (along with numerous other responses). Bisphenol A induced hyperprolactinemia in F344 rats but not in Sprague-Dawley rats. In the F344 rat, estradiol increased serum prolactin concentrations 10-fold and bisphenol A increased it 7- to 8-fold over controls. These findings are consistent with the findings of vom Saal et al. (1997) comparing bisphenol A and DES in fetal mice. Those findings suggest that bisphenol A is bioactive within the range of human exposure (Walent and Gorski 1990; Krishnan et al. 1993; Brotons et al. 1995: Olea et al. 1996; Takao et al. 1999).break

Page 128

Recently, Cagen et al. (1999) conducted a study to evaluate the effects of low doses of bisphenol A on sexual development in male mice. To the extent possible, the study protocol duplicated the studies of Nagel et al. (1997) and vom Saal et al. (1998) for all factors indicated as critical by those investigators. Some differences between the two studies include the source of the CF,, mice, the number of doses of bisphenol A, the methods used to determine sperm count, the age of mice at necropsy, and the way the animals were housed. An additional positive control group of mice was dosed with 0.2 µg/kg/d DES. No effects on testes histopathology, daily sperm production or sperm-production efficiency (i.e., daily sperm production per gram of testis), or on prostate, preputial, seminal vesicle, or epididymis weights were observed in the bisphenol A treated groups. In addition, no adverse effects were observed in any of these parameters in the group tested with DES. Thus, this study failed to replicate the results of vom Saal et al. (1997, 1998) and Nagel et al. (1997). The reason for the discrepancies in the findings is a subject of controversy and cannot be resolved at this time.

Octylphenol

Octylphenol (p-(1,1,3,3-tetramethylbutyl)phenol) is an alkylphenol used in its ethoxylated form (octylphenol ethoxylate) in a variety of products such as detergents and plastics. In one study (Sharpe et al. 1995), female rats were given drinking water containing 100 or 1,000 µg/L octylphenol before pregnancy, during gestation, and throughout lactation, and the effects on testicular size and spermatogenesis in male offspring in adulthood were investigated. Intake of octylphenol based upon water intake was calculated only for the group exposed to 1,000 µg/L and was reported to range from 129 µg/kg/d in the first 2 d after birth to 367 µg/kg/d just before weaning. There were significant decreases in absolute testis weight, in the ratio of testis-to-kidney size, in relative ventral prostate weight, and in daily sperm production in male offspring exposed to 1,000 µg/L octylphenol, and there was a significant reduction in relative prostate weight at both concentrations. When male offspring were treated postnatally only with 1,000 µg/L octylphenol on d 1-22 after birth, there was a significant reduction in average and relative testis weight (Sharpe et al. 1995). More recently, the authors reported their inability to replicate their findings (Sharpe et al. 1998). While they expressed continued confidence in their original publication, they hypothesized that the inability to replicate the work may have been due to changed biologic factors of which they were unaware and unable to control (Sharpe et al. 1998). In another study, there was no effect on prostate weight of male offspring of pregnant mice fed 2 and 20 µg/kg/d octylphenol on d 11-17 of gestation (Nagel et al. 1997), although there was a significant decrease in daily sperm production (vom Saal et al. 1998).break

Page 129

Butyl Benzyl Phthalate (BBP)

BBP is used as a plasticizer for cellulose resins, polyvinyl acetates, polyurethanes, and polysulfides and in regenerated cellulose films for packaging. When 1,000 µg/L BBP was administered in drinking water to female rats before mating and throughout lactation (nominal intake based on water intake ranged from 126 µg/kg/d in the first 2 d after birth to 366 µg/kg/d just before weaning), there was a small but significant reduction in mean testicular size and a reduction in daily sperm production in male offspring at d 90-95 (Sharpe et al. 1995). Similar findings were reported in rats treated with DES (100 µg/L in drinking water; intake was not determined), which was evaluated in the same study. The researchers note that, although these estrogenic chemicals exert similar effects on testis size and daily sperm production, there is no evidence that the effects are caused by the compounds' estrogenicity.

In a subsequent study, Ashby et al. (1997) failed to find any effects of BBP, despite very similar or identical protocols. Female rats were administered BBP (average intake 182.6 µg/kg/d) in drinking water during gestation and lactation, and their offspring were monitored for 90 d. DES (8.6 µg/kg/d) affected the sexual development of male and female pups, causing changes in anogenital distance; average day of vaginal opening and prepuce separation; weight of the uterus, testis, and accessory sex glands; and caudal epididymis sperm count and homogenization-resistant testicular sperm count. BBP had no effect other than to cause a slight advance in the average day of vaginal opening and a small increase in male anogenital distance, but those effects were attributed to the increased weight of the pups treated with BBP.

The reason for the discrepancy in findings between the study by Sharpe et al. (1995) and that of Ashby et al. (1997) is unknown.

Di-n-Butyl Phthalate

Di-n-butyl phthalate (DBP) has been characterized as a reproductive and developmental toxicant in several studies (e.g., Cater et al. 1977; Ema et al. 1994, 1995). causing fetal death and skeletal malformations (predominantly cleft palate) in rats. In addition, DBP has been shown to cause testicular atrophy, early sloughing of germ cells, and vacuolization of Sertoli cell cytoplasm (Cater et al. 1977; Fukuoka et al. 1989). Immature rats appear to be more susceptible to these effects than adult rats (Gray and Gangolli 1986; Creasy et al. 1987). Recent studies have investigated the effects in more depth.

DBP was tested in the National Toxicology Program's Reproductive Assessment by Continuous Breeding protocol using Sprague-Dawley rats (Wine et al. 1997). DBP was administered in the diet to male and female rats continuously. with an average daily intake of 52, 256, and 509 mg/kg for males and 80, 385. and 794 mg/kg for females, respectively. Breeding pairs (F₀ generation) were matedcontinue

Page 130

for an extended period, sufficient to produce five litters (F₁ generation). The last litter was raised to adulthood and allowed to mate and produce offspring (F₂ generation). In the F₀ generation, the only adverse reproductive effects observed were a 5-17% reduction in the number of live pups per litter and a decrease (<13%) in live pup weights. When crossover matings were conducted to determine the affected sex, the number of offspring was unchanged, but pups from treated females weighed significantly less, whereas offspring from treated males were unchanged. At necropsy, the high-dose females had a 14% reduction in body weight, and both males and females had a 10-15% increase in kidney and liver to body weight ratios compared with controls.

In the F₁ generation, indices of mating, pregnancy, and fertility in the high-dose group were all significantly affected. Specifically, only one live litter was delivered from 20 breeding pairs. In all dose groups, weights of the F₂ pups were 6-8% lower than controls. At necropsy, the high-dose F₁ males were found to have significantly reduced epididymal sperm counts and testicular spermatid head counts. Eight of the 10 males had degenerated seminiferous tubules and five had underdeveloped or otherwise defective epididymides. No adverse effects on ovarian or uterine development were observed in the F₁ females. The investigators concluded that DBP is a reproductive and developmental toxicant to both adult and developing rats and that DBP had greater effects on the second generation than the first generation.

Mylchreest et al. (1998) showed that similar effects on the male reproductive tract could be produced with much shorter gestational and lactational exposure. Pregnant CD rats were given DBP at 250, 500, or 750 mg/kg/d throughout pregnancy and lactation until their offspring were 20 d old. Anogenital distance was decreased in the male offspring of the mid- and high-dose groups. In adulthood, the epididymis was underdeveloped or absent in 9%, 50%, and 71% of the males of the low-, mid-, and high-dose groups, respectively. Testicular atrophy and widespread germ-cell loss were also found. Hypospadias was observed 3%, 21%, and 43% of males, and the testes were abnormally positioned or absent in 3%, 6%, and 29% of the males in the low-, mid-, and high-dose groups, respectively. In addition, small testes and seminal vesicles were found, and in some cases, the prostate glands and seminal vesicles were missing. The investigators concluded that DBP specifically impaired the androgen-dependent development of the male reproductive tract, suggesting that DBP is antiandrogenic rather than estrogenic. The investigators caution that it will be important to determine whether the reproductive toxicity of DBP is metabolite-mediated, because marked species differences in metabolism exist.

Earlier studies with DBP (Ema et al. 1993), in which exposures were conducted during organogenesis, revealed no evidence of selective effects on the developing reproductive system. Foster (1997) suggests that the difference in developmental outcomes was that the developmental toxicity study (Ema et al. 1993) did not include exposure during the critical period of sexual differentiation,continue

Page 131

which occurs after gestation d 15. Recently, Ema et al. (1998) reported data that support this. They showed that when pregnant rats were fed DBP on d 1 1-21 of pregnancy at an average daily intake of 555 and 661 mg/kg, there was a significant incidence of undescended testes and a significant decrease in anogenital distance in male fetuses.

Human Studies

DDT and its Metabolites

In a study of 722 women from North Carolina, Rogan et al. (1987) reported significantly shortened duration of lactation in relation to increasing breast milk concentration of the antiandrogenic DDT metabolite p,p'-DDE. Median duration of lactation decreased from 7.8 mo in the lowest exposed group (0-2.5 ppm p,p'-DDE in milk fat) to 3.8 mo in the most exposed group (10.0-12.5 ppm p,p'-DDE in milk fat). Adjustment for possible confounders (i.e., mother's age, race, education, occupation, smoking, and drinking) did not change these findings. Another study examined 229 women in Tlahualilo, Mexico, where DDT is used extensively for agricultural purposes (Gladen and Rogan 1995). Exposure measures were based on p,p'-DDE concentrations in breast-milk samples collected at the time of birth. As in the North Carolina study, median duration of lactation was 7.5 mo in the group with the least contaminated milk (0-2.5 ppm p,p'-DDE in milk fat) and 3 mo in the group with the most heavily contaminated milk (=12.5 ppm p,p'-DDE in milk fat). The trend was more marked among women who had breast fed children from previous pregnancies. Median duration of lactation was 8.8 mo in the least contaminated group and 2.8 mo in the most contaminated group. This cohort is being studied to determine whether there is any relationship between prenatal exposure to DDEs and adverse effects on lactation and possibly on reproductive development.

Wasserman et al. (1982) compared serum levels of DDT and its metabolites in 10 women from normal term pregnancies and 17 with premature delivery. The mean DDT serum level in women with premature deliveries (71.1 ppb) was significantly higher than that for normal controls (26.5 ppb, p < 0.005). The percentage of o,p'-DDT was unusually high in the case group.

PCBs

Several studies of the effects of consuming PCB-contaminated fish have been conducted on a variety of reproductive parameters. One of the best studied populations has been the New York State Angler Cohort, a population-based cohort of angler families from 16 counties in New York near Lake Ontario (Vena et al. 1996). Questionnaires regarding duration (number of years) and monthly frequency of consuming sport fish were completed by this cohort in 1991, and acontinue

Page 132

PCB-exposure index was calculated (a crude estimate of potential lifetime PCB exposure through fish consumption). In one study (Mendola et al. 1997). the effects of fish consumption on menstrual-cycle length was evaluated using telephone interviews conducted in 1993 with 2.223 women from the cohort. Multiple regression analyses revealed a significant reduction in the menstrual cycle length of 1.11 d with consumption of more than one fish per meal per month and with moderate to high (> 1 mg) estimated PCB index. Frequency of consumption and PCB index appeared to have a stronger relation with cycle length than the number of years of consumption. Those data must be interpreted cautiously because of the limitations of secondary data analysis, including the lack of information on potential confounders. The researchers concluded that ''While the small decreases in menstrual cycle length observed are not likely to be clinically relevant or of major public health concern per se, they may indicate potential endocrine effects on a population level."

Another study of the New York State Angler Cohort investigated the effects on fish consumption on time-to-pregnancy (TTP) (Buck et al. 1997). Telephone interviews were conducted in 1993 with 2,445 women who stated upon enrollment in the cohort that they were considering pregnancy. Among the 1,234 women who reported being pregnant, 874 had a known TTP. Multiple regression analyses of the data indicated that duration of fish consumption explained virtually none (0.5%) of the observed variance in TTP among all women with a known TTP, even after the analysis was restricted to women who reported eating fish. Thus, consumption of contaminated sport fish did not appear to have a detrimental effect on TTP.

The risk of spontaneous fetal death has also been studied in 1,820 multigravid women from the New York State Angler Cohort in relation to sport-fish consumption (Mendola et al. 1995). The reproductive histories of the women were obtained from the most recent New York State live-birth certificates found on a computerized registry. No significant increases in risk for fetal death were found in regard to lifetime estimates of PCB exposure based on species-specific PCB levels, the number of years of fish consumption, kilograms of sport fish consumed between 1990-1991, and a lifetime estimate of kilograms eaten.

Studies are being conducted to determine the effects of fish consumption on the reproductive health of a cohort of Michigan anglers (Courval et al. 1996a). In one study, the effects of fish consumption on conception is being evaluated in married couples, with conception failure being defined as inability to conceive after 12 mo. Preliminary results from 626 couples show that 15% of both men and women reported conception failure (Courval et al. 1996b). The unadjusted odds ratio for conception failure across three increasing levels of fish consumption compared with no fish consumption was 1.2, 1.3, and 2.0 for men (trend test p = 0.06) and 0.9, 1.0, and 1.4 for women (trend test p = 0.35). After adjusting for covariates, the odd ratios for conception were 1.4, 1.8, and 2.8 for men and 0.8, 0.8, and 1.0 for women. The investigators report that these data suggest a modestcontinue

Page 133

association between sports-fish consumption and the risk of conception failure for men only but these preliminary results are consistent with the frequency of infertility in the population in general (OTA 1988).

Other studies have been conducted to determine whether body burdens of PCBs could be linked to adverse effects on reproduction. For example, in a study in Rome, Italy, between 1983-1984, concentrations of PCBs in blood were measured in 120 women who had miscarried and compared with 120 controls (Leoni et al. 1989). Concentrations of tetra- and penta-isomers of PCBs, measured as Fenclor 54, were significantly greater in the blood of women who had miscarried than in controls: 8.65 ppb vs. 6.89 ppb (p < 0.05). Although this study shows a significant correlation between PCBs and miscarriage, it is only the first study to have examined this association, and further work is needed on the subject.

In a study of 170 men in which semen samples were analyzed for 74 PCB congeners (Bush et al. 1986), no association was found between PCB congener concentration and sperm count, motility, or percent normal forms. However, in samples with a low sperm count (<20 mil/mL), sperm motility was inversely associated with 3 PCB congeners (2,4,5,2',4',5'-hexachlorobiphenyl, p = 0.002, SE = 26; 2,4,5,3',4'- pentachlorobiphenyl, p = 0.002, SE = 20; and 2,4,5,2',3',4'-hexachlorobiphenyl, p < 0.01, SE = not provided).

Two episodes of accidental exposure to PCB-contaminated rice oil occurred in Yusho, Japan (1968), and Yu-Cheng, Taiwan (1978-1979). In Yusho, pregnant women exposed to contaminated oil in 1968 delivered infants with "fetal PCB syndrome." The signs included dark pigmentation of the skin and mucus membranes, gingival hyperplasia, exophthalmic edematous eye, dentition at birth, abnormal calcification of the skull, rocker bottom heel, and low birth weight (Yamashita and Hayashi 1985). A similar accident with PCBs and polychlorinated dibenzofurans (PCDFs) occurred in Yu-Cheng, Taiwan, in 1978-1979, and fetal PCB syndrome and an increased mortality rate were observed among infants of exposed women (Hsu et al. 1985; Rogan et al. 1988). Most of the affected infants were found to be shorter and had less total lean mass and soft-tissue mass than did children in a matched control group. The effects were observed only in the first children born after maternal exposure and not in subsequent children (Guo et al. 1994, 1995a). In addition, a preliminary study on sexual development was conducted on a matched cohort of 55 pairs of Yu-Cheng exposed and unexposed boys in 1993 (Guo et al. 1995b). The authors reported that exposed boys 11-14 yr had significantly shorter penis length. However, no data were provided to support this conclusion nor is it known if they corrected for the smaller body size of the children. Neurodevelopmental effects, such as delays in cognitive, psychomotor, and behavioral development, have been observed. The effects of pre- and postnatal exposure to PCBs on neurodevelopment are discussed in detail in Chapter 6.

Several developmental studies have also been conducted in the Great Lakes region of the United States of prenatal exposure to PCBs as a result of maternalcontinue

Page 134

consumption of contaminated sport fish. In the Michigan/Maternal Infant Cohort Study, Fein et al. (1984) evaluated the birth size and gestational age of 242 infants and found that maternal consumption of fish and concentrations of PCBs in cord serum were correlated with lowered birth weight, shortened gestation, and smaller head circumference. Lower weight was also observed in children from this cohort at 4 yr in a dose-dependent fashion (Jacobson et al. 1990). Children with cord serum PCB levels of 5.0 ng/mL or more weighed 1.8 kg less on average than the lowest exposed children. Prenatal exposure was also associated with deficits in neurologic development in follow-up studies of these children at up to 11 yr (Jacobson et al. 1992; Jacobson and Jacobson 1996). The details of these neurodevelopmental studies are presented in Chapter 6.

In a study of 94 Inuit infants from Hudson Bay whose mothers had high concentrations of PCBs in their breast milk, no significant differences were found between male and female newborns for birth weight, head circumference, or thyroid-stimulating hormone level. There was a negative association between male height at birth and concentration of PCBs (r values ranged from -0.23 to -0.41, p values ranged from nonsignificant to 0.04), hexachlorobenzene (r = -0.41, p = 0.006), mirex (r = -0.34, p = 0.02), and toxicity equivalency factors (TEQs) of chlorinated dibenzodioxins/chlorinated dibenzofurans (r = -0.48. p = 0.04) in milk fat. On the other hand, there was a positive association between female birth height and TEQs coplanar PCBs (r = 0.47, p = 0.05) and TEQs polychlorinated dibenzo-p-dioxins/polychlorinated dibenzofurans (r = 0.49, p = 0.04) (Dewailly et al. 1993b).

Occupational studies in New York of offspring of women exposed to PCBs during the manufacture of capacitors found a significant relation between increased maternal serum PCB levels and decreased birth weight and gestational age (Taylor et al. 1984, 1989). The decrease in birth weight was found to be at least partially related to the shorter gestational age. The investigators concluded that the magnitude of the effects was small compared with other known determinants of gestational age and birth weight, and the biologic importance of these effects was likely to be negligible.

In the North Carolina Breast Milk and Formula project, 912 children were followed to investigate the effects of prenatal exposure to environmental levels of PCBs and DDE from maternal breast milk (Rogan et al. 1986a). No relationship was found between birth weight, head circumference, and neonatal jaundice with PCB or DDE levels in milk fat. Follow-up studies of the neurodevelopment of these children have found evidence of neurobehavioral deficits that persist up to 2 yr (Gladen et al. 1988; Rogan and Gladen 1991). See Chapter 6 for the details of these neurobehavioral studies.

In an ongoing study of perinatal exposure to PCBs in the Netherlands, the birth weight and growth of 105 breast-fed infants was compared with that of 102 formula-fed infants. The investigators reported that prenatal exposure to PCBs, as measured in cord plasma, had a significant negative effect on birth weight andcontinue

Page 135

on growth from birth to 3 mo of age, but not thereafter. Postnatal exposure to PCBs and dioxin, as measured in breast milk, had no effect on growth (Patandin et al. 1998). Neurologic development in this cohort of children was found to be adversely affected in some studies (Huisman et al. 1995a; Pantandin et al. 1999) but not in others (Koopman-Esseboom et al. 1996: Lanting et al. 1998). The varied results of these studies are presented in more detail in Chapter 6.

Collectively, the data on prenatal exposure to high levels of PCBs from the Yusho and Yu-Cheng incidents and studies from the United States and the Netherlands of prenatal exposure to PCBs from maternal diet indicate that PCBs can affect birth weight and growth. Lower birth weights were not observed in the North Carolina cohort, but that might be explained by the lower exposure level of this cohort compared with the others. In addition, the Yu-Cheng children were also exposed to PCDFs, and exposure to those compounds was not noted in the other cohorts.

TCDD

In 1976, an industrial accident in Seveso. Italy, released kilograms of TCDD, the most toxic of the dioxins. Investigators are conducting studies to determine whether exposure to TCDD has caused reproductive effects. Mocarelli et al. (1996) evaluated the sex ratio among children born to heavily exposed residents from 9 mo after the accident until the end of 1984 (corresponding to one TCDD half-life in adults). Among 74 births, there was a significant deficit in the number of males (26 males vs. 48 females; c2 test; p < .001). Serum samples collected and stored from families who resided in the most heavily exposed area were analyzed after 1988. There were no males among the 12 births to the 9 parents with the greatest TCDD exposure (104-2,340 ppt in serum lipid). The exposed population in Seveso is still under study. This TCDD-associated altered sex ratio may or may not be related to recent reports of similar declines in sex ratio between the 1950s and the 1960s from the Netherlands (van der Pal-de Bruin et al. 1997) and Denmark (Moller 1996), and in Canada since 1970 (Allan et al. 1997). However, the causes of the declines in sex ratio is yet unknown.

A toxicokinetic analysis (Bois and Eskenazi 1994) using TCDD concentrations in blood from 19 highly exposed residents of Seveso showed that exposures for some of the residents were greater than were the exposures shown to cause endometriosis (5 and 25 ppt) in rhesus monkeys (Rier et al. 1993). Whether the exposures encountered in Seveso will result in an increased incidence of endometriosis is under study.

A recent epidemiology study on TCDD exposure and cancer risk in Seveso women concluded that there was a very low incidence of breast and endometrial cancers (Bertazzi et al. 1993). That study is described in more detail in Chapter 9.

Because alterations in circulating thyroid hormones in newborns might influence the maturation of the central nervous system and could thus have conse-soft

Page 136

quences for psychomotor development (Birrell et al. 1983), thyroid hormone concentrations were measured in the blood of Dutch infants exposed to elevated concentrations of dioxin in breast milk (Pluim et al. 1993). Breast-milk contamination is the result of maternal consumption of contaminated meat, dairy products, and fish oils. The total dioxin concentration in milk fat from 38 healthy mother-infant pairs was calculated as the sum of the toxic equivalence (TEQ) relative to TCDD of the 17 most toxic congeners (7 dioxins and 10 dibenzofurans). The mothers were divided into two groups based on the median dioxin concentration: low exposure (8.7-28.0 ng TEQ/kg milk fat; mean 18.6 ng TEQ/kg milk fat) and high exposure (29.2-62.7 ng TEQ/kg milk fat; mean 37.5 ng TEQ/kg milk fat). At 1 and 11 wk of age, significantly greater concentrations of total T₄ (tT₄) (3,3',5,5'-tetraiodo-L-thyronine) and tT₄/TBG (thyroxine-binding globulin) ratios were found in the high-exposure group. This was attributed to intrauterine exposure to TCDDs, although the investigators note that some exposure from contaminated breast milk could not be ruled out (Pluim et al. 1993). But in another study, the thyroid hormone status of 105 Dutch women and their infants was evaluated in relation to exposure to dioxins and PCBs. Higher levels of chlorinated dibenzo-p-dioxins, chlorodibenzofurans, and PCBs in breast milk were significantly correlated with lower plasma levels of maternal total triiodothyronine and total thyroxine, and with higher plasma levels of thyroid-stimulating hormone in infants during the second week and third month after birth. Lower plasma free thyroxine and total thyroxine levels were also observed in the second week after birth (Koopman-Esseboom et al. 1994). In another recent study of plasma thyroxin (T4) levels in 93 Dutch newborns showed that exposure to organochlorine pesticides, PCBs, dibenzodioxin, and dibenzofuran levels as measured in breast milk was correlated with lower T4 levels (whether measurements were of total or free T4 was not specified). However, multivariate analyses of the data suggest that the body-mass index of the mother and smoking during pregnancy are possible confounding factors (Fiolet et al. 1997).

PCBs and dioxin can alter binding of thyroid hormone to plasma proteins, thus resulting in a decrease in plasma thyroid hormone levels. Such binding is principally to transthyretin, which is involved in transporting thyroid and retinol in humans, rodents, and other species. Experimental evidence in rats suggests that transthyretin is particularly important with regard to transport of thyroid hormone from mother to fetus across the placenta (Brouwer et al. 1998). But Brouwer et al. (1998) did point out that humans and rats differ in major plasma thyroid hormone transporters (e.g., transthyretin (TTR) is greater than thyroxine binding globulin (TBG) in rodents and TBG is greater than TTR in humans). These data imply that thyroid hormone levels in plasma may be less effected in humans than rodents. In any case, whether this disruption of thyroid hormone homeostasis mediates the developmental effects associated with exposure to these HAAs remains to be determined.break

Page 137

Serum concentrations of testosterone, luteinizing hormone (LH), and follicle-stimulating hormone (FSH) and sperm concentration in men exposed to TCDD have been examined. Henriksen and Michalek (1996) studied veterans of the Vietnam War (Operation Ranch Hand) exposed to Agent Orange and its major contaminant, TCDD. Using regression models, the investigators found that testosterone concentrations in serum decreased with increasing TCDD exposure (slope = -0.4276, standard error = 0.0950). When adjustments were made for possible confounders. the slopes changed with military occupation. Negative slopes were observed for the three occupational strata evaluated, with the strongest among officers (slope = -1.2381, standard error = 0.2895) and weakest among enlisted ground personnel (slope = -0.3485, standard error = 0.1441). Without adjustment. FSH and LH decreased with TCDD exposure (FSH: slope = -0.0121, standard error = 0.0153; LH: slope = -0.0276, standard error = 0.0111). After adjustment, FSH increased with TCDD (slope = 0.0073, standard error = 0.0171 ) and LH decreased with TCDD (slope = -0.0242. standard error = 0.0127). The investigators note that these findings are inconsistent with earlier work (Michalek et al. 1995) showing that exposure to TCDD is greatest in enlisted ground personnel and least in officers.

An ongoing cross-sectional study of the long-term health effects of occupational exposure to chemicals contaminated with TCDD is being conducted by the National Institute for Occupational Safety and Health (Egeland et al. 1994; Sweeney et al. 1997-98). A variety of health end points have been analyzed and continue to be analyzed (Sweeney et al. 1997-98). In linear regression analyses, current serum dioxin was positively and significantly related to LH and FSH levels, and inversely related to total testosterone, after adjustment for potential confounders (p < 0.05). This suggests that as TCDD serum levels increase, testosterone decreases, and the consequence is an appropriate increase in LH and FHS. This further suggests that TCDD is operating to decrease testosterone synthesis, and that this is not occurring through an effect on the neuro-endocrine system.

Chlordecone

Chlordecone is a highly stable chlorinated hydrocarbon pesticide that was produced between 1958 and 1975 as an insecticide and fungicide, primarily for control of the banana root borer and tobacco wireworm and as bait for control of ants and cockroaches. The two principal manufacturers of chlordecone, Allied Chemical and Life Science Product Company (LSPC), were located in Hopewell. Virginia, and operated between 1968 and 1975. Allied Chemical closed in 1974. and LSPC was closed in July 1975 after a state health department inspection found severe chlordecone-related illness and contamination. An investigation by the Centers for Disease Control (CDC) followed. Of the 110 individuals who worked at the plant, more than 50% showed high blood levels of chlordecone.continue

Page 138

Semen analysis of 14 chlordecone-exposed workers showed abnormal sperm morphology, decreased sperm mobility, and oligospermia (Guzelian 1976). A recent evaluation of chlordecone by the Agency for Toxic Substances and Disease Registry (Faroon et al. 1995) concluded: "The available human data on chlordecone provide qualitative evidence to support the conclusion that intermediate or chronic-duration exposure to high concentrations of chlordecone in the workplace cause oligospermia and decrease sperm motility among male workers." That finding was consistent with reproductive toxicity demonstrated in multiple toxicologic studies, including testicular atrophy in rats, testicular abnormalities, and reduction of germinal epithelium and number of spermatozoa (Eroschenko and Wilson 1975) and a strong estrogenic effect on the oviduct of female quail (McFarland and Lacy 1969).

Agricultural Exposures

Several ecologic studies have investigated associations between exposure to agricultural chemicals, such as pesticides (insecticides, herbicides, and fungicides), and effects on fertility and development. Positive correlations have been found in some studies, and these are presented below to illustrate some of the reproductive and developmental effects observed among people exposed to agricultural chemicals. However, specific causative agents have not been identified in many of these studies, so it is not know whether they might be HAAs.

Fuortes et al. (1997) examined the industrial and occupational histories of 281 infertile women and compared them with those from 216 postpartum women. Only agriculturally related occupations were associated with infertility. However, the reliability of this association is uncertain because the sample sizes were small. Decreases in fertility were reported in a study of fruit growers exposed to pesticides, with a greater time-to-pregnancy occurring during the months when pesticides were applied (de Cock et al. 1994). Mosher and Pratt (1987) reported a higher probability of infertility among male farmers in a National Center for Health Statistics survey. A study of 32 male farm sprayers who were exposed to the herbicide 2,4-D found statistically significant levels of abnormal, motionless. and dead spermatozoa compared to 25 unexposed controls (Lerda and Rizzi 1991). Urine measurements of 2,4-D confirmed exposure status, at least for that chemical. However, the significance of this study is unclear, because the number of subjects was small. Furthermore, studies of Vietnam veterans exposed to much higher concentrations of Agent Orange, which includes 2,4-D (CDC 1989; Wolfe et al. 1992; IOM 1994), and laboratory studies with 2,4-D (Lamb et al. 1981) show no such effects.

Some studies have found no association between exposure to agricultural chemicals and adverse reproductive outcomes. For example, in a study of low-level exposure to malathion, Thomas et al. (1992) found no substantial or significant associations between maternal exposure and the occurrence of spontaneouscontinue

Page 139

abortion, intrauterine growth retardation (IUGR), or other anomalies as a group. McDonald et al. (1987) found no relationship between maternal occupation in agriculture/horticulture and spontaneous abortion, although an excess of stillbirths was noted. In another study, no significant difference was found in the incidence of abnormal semen in farmers (Gerber et al. 1988).

Garry et al. (1996) reported significant increases in circulatory/respiratory, urogenital, and musculoskeletal/integumental birth defects in 4,935 children born between 1989 and 1992 to 34,772 state-licensed male pesticide applicators in Minnesota. Those effects were more pronounced for children conceived in the spring, and was most marked in western Minnesota, where phenoxy herbicide/ fungicide use is highest. The authors noted that there is a possibility that the chemicals were contaminated with dioxin. Elevated levels of the herbicide atrazine found in municipal water supplies in Iowa were associated with excess rates of cardiovascular, urogenital, and limb-reduction deficits (Munger et al. 1992). Preliminary results from a study by Munger et al. (1992) found that communities with herbicide-contaminated water supplies had elevated IUGR rates compared with neighboring communities with different water supplies. Contaminants included atrazine, metolachlor, and cyanazine, but the authors concluded that a strong causal relationship between any specific contaminant and risk of IUGR could not be inferred.

Exposures in the studies above are to a broad range of chemically diverse pesticides. Which pesticides are responsible for the reproductive and developmental effects is unknown. Furthermore, it is difficult to isolate a single type of exposure from the wide variety of other exposures to chemicals and environmental factors to which agricultural workers are subject. Nonetheless, such ecologic studies are useful for identifying potential effects. Nurminen (1995) evaluated a number of the ecologic studies and concluded that "the published studies have given some indications of elevated reproductive risk and exposure to pesticides but, altogether, the collective epidemiologic evidence did not allow any clear inference to be drawn."

Adverse Effects on the Male Reproductive System: Regional and Temporal Variations

Some authors suggest that the incidence of several male reproductive disorders is increasing (Giwercman et al. 1993; Sharpe and Skakkabaek 1993; Toppari et al. 1996). The disorders include testicular cancer, hypospadias, cryptorchidism, and poor sperm concentration. Studies with laboratory animals have shown that prenatal exposure to some HAAs, such as methoxychlor (Gray et al. 1989; Gray 1992), TCDD (Mably et al. 1992c), and octylphenol and bisphenol A (vom Saal et al. 1998) can reduce sperm production. Other authors have questioned the validity of the trend analyses (Farrow 1994; Nieschlag and Lerchl 1996; Paulozzi 1999). Because hypotheses linking the trends to human exposure to HAAs havecontinue

Page 140

generated considerable discussion, and because such links are biologically plausible given the reported effects of exposure to HAAs in laboratory animals and wildlife species, they are discussed here.

Testicular cancer rates have been reported to be increasing in the United States (Ries et al. 1997), Canada (Weir et al. 1999), and in six European countries, particularly among men born after 1950(Bergstrom et al. 1996). The trend has been particularly marked in Denmark, where the incidence of seminomas and nonseminomas has been rising for several decades (Forman and Moller 1994). The risk factors for testicular cancer are increased with cryptorchidism (RR = 5.2) and hypospadias (RR = 4.2), suggesting a prenatal etiology (Prener et al. 1996). An increase in the incidence of hypospadias has been observed in England and Wales (Matlai and Beral 1985), Hungary (Czeizel 1985; Czeizel et al. 1986), Sweden (Kallen and Winberg 1982), Norway (WHO 1991), and Denmark (WHO 1991). One study (Paulozzi et al. 1997) in the United States shows a doubling of hypospadias rates between 1968 and 1993, from 40/10,000 to 80/ 10,000 male births. Some of those changes in incidence may be due to the reporting criteria for the disorders and other aspects of diagnosis. However, the increase is most marked among the most severe cases, which are least likely to be subject to changes in diagnosis. Hypospadias has been associated with defects in testosterone receptors and testosterone metabolism, suggesting a possible link to endocrine factors (Allen and Griffin 1984: Glatzl 1984). Genetic and other factors have also been implicated in hypospadias (Bauer et al. 1981: Harris 1990; Weidner et al. 1999). Several authors present data suggesting that cryptorchidism rates have risen in England, Wales, and Scandinavia, although the trends are less consistent, and they could be related to changes in diagnosis or treatment (Giwercman and Skakkeback 1992). Berkowitz et al. (1993) reported a rate of cryptorchidism among New York City births in 1990 that was similar to the rate reported by Scorer (1964) in London in the 1950s. Berkowitz et al. (1993) conclude there is no evidence of an increase in the cryptorchidism rate. However. their data do not directly address the question of a change in rate of the defect: 1990 New York births were not compared with births in New York in the 1950s, when rates might well have been lower. With regard to hypospadias, cryptorchidism, testicular cancer, and sperm count, there is considerable evidence for geographic variation. Studies to assess the factors contributing to this variation will be required to determine whether HAAs, in addition to or in conjunction with other factors, are related to this variability. Paulozzi ( 1999) reported that international trend rates of hypospadias and cryptorchidism increased in some geographic locations, such as the United States, Japan, and Scandinavia, while changes did not occur in other regions, most notably Canada. Since 1985, incidence rates for hypospadias have leveled off in regions which showed an increase. Incidence rates for cryptorchidism have actually declined in most regions since 1985. In any case, the relations of these trends in testicular cancer, hypospadias, or cryptorchidism to HAAs has not been examined.break

Page 141

Regional and Temporal Variation in Sperm Concentration

Sperm concentration is the best-studied male reproductive end point, and it is perhaps the measure of male reproductive function that has generated the greatest controversy. The possibility of declining sperm concentration and the environmental causes of such a decline are not new concerns.

After noting sperm concentrations in 1970-1973 that were markedly below those reported by MacLeod and Gold in 1951 (48 x 106/mL vs. 107 x 106/mL), Nelson and Bunge (1974) wrote, "The overall decrease in the sperm concentration and the semen volumes would tend to incriminate an environmental factor to which the entire population has been exposed." In 1979, Macleod and Wang noted that several studies on fertile males found a "marked reduction in spermatogenesis since 1951." These authors' analysis of their own historical data of infertile populations, however, did not support a decline. In 1980, James conducted a multinational analysis of 29 studies of sperm concentration published over 45-yr and concluded, ''There can be no reasonable doubt that these reported mean sperm counts show a decline with time of publication, at least since 1960." Dougherty et al. (1981) found a correlation between sperm concentrations taken from a sample group of university students and the presence of toxic substances, which they identified by negative-chemical-ionization screening procedures. They note that "PCB uniformly gave negative slope correlations with sperm." Murature et al. (1987) analyzed sperm concentrations listed in 45 studies published between 1929 and 1981 and found a decline in sperm concentration between 1949 and 1981. These authors argue that mean sperm concentration was correlated with several environmental exposures, including the total production of synthetic compounds within the United States. Feichtinger (1991) discusses several possible etiologies for a decline, in particular, exposure to organochlorines, such as PCBs, DDT, and hexachlorobenzene (HCB). Although there was no significant correlation between exposure to these compounds and sperm count, samples of semen that led to pregnancy after in vitro fertilization and embryo transfer (IVF/ET) contained lower average concentrations of PCBs and HCB than did samples that did not produce pregnancy (Feichtinger 1991).

Several distinct questions underlie the above controversy. It is important to separate them: (1) Has semen quality (as measured by sperm concentration, morphology, or motility) declined? (2) Does sperm concentration vary significantly according to geographic location? (3) If the answer to either of the first two questions is yes, is exposure to HAAs or other environmental toxicants causing the changes?

The question of a possible decline in sperm concentration over time has most recently been raised by an analysis (Carlsen et al. 1992) of 61 studies conducted throughout the world and published between 1938 and 1990. Carlsen's group used simple linear regression to model the changes in sperm concentration over time, concluding, "reports published worldwide indicate clearly that sperm con-soft

Page 142

centration . . . declined appreciably during 1938-1990." The analysis was followed by considerable debate, and several articles challenged its conclusions. Analyses proposing alternative, and increasingly complex, statistical models of the same data set followed (Olsen et al. 1995; Bahadur et al. 1996; Becker and Berhane 1997; Swan et al. 1997). This group of publications, the "multinational trend studies," all analyzed basically the same historical data set. "Local trend studies," on the other hand, examine data within a single country or state to assess trends in sperm concentration. While many local and multinational trend studies include data on multiple parameters; all provide data on trends in sperm concentration, the parameter for which methods have remained most comparable over time.

Carlsen et al. (1992) reviewed the literature published between 1930 and 1990 to identify studies that include data on sperm concentration in normal males. Studies that included men from infertile couples, studies that selected men on the basis of a high (or low) sperm count, and studies that used nonmanual methods for counting sperm were excluded. The authors found that mean sperm concentration had decreased from 113 x 106/mL in 1940 to 66 x 106/mL in 1990, with a slope of minus 0.93 x 106/mL/yr (p < .0001).

In response, several authors noted that mean sperm concentration in the data set appears to decline until some point in the early 1970s and then to level off or even increase. This suggested the need to use alternative (nonlinear) regression models. Olsen et al. (1995) fit three such models: a spline function (or "hockey stick"), which corresponds to a decline in sperm concentration until 1970, when the decline levels off, ceases, or is reversed; a quadratic function, which corresponds to a smoothed form of the spline with no abrupt change in sperm concentration but a gradual upturn late in the study period; or a step function. The first and second alternative models suggest that the change in sperm concentration was smooth: the step function assumes that mean sperm concentration remained constant until it dropped abruptly around 1970. Olsen et al. (1995) maintained that their three nonlinear analyses explain somewhat more of the variability in the data, as measured by the adjusted R2, than does the simple linear regression, although the incremental increase in R2 was minimal (approximately 1-3% ). Bahadur et al. (1996) recognized the need to control for geographic region, and they did so by fitting separate curves for the United States, Europe, and elsewhere. These authors reported that data from the United States fit the linear and quadratic functions equally well, although the quadratic term was not statistically significant. In their analysis, data from Europe and other countries fit neither model well. Becker and Berhane (1997) were the first to present the data set using multiple-regression analysis. Their method controlled for region and type of study population (whether men were of proven fertility or not), as categorized by Carlsen et al. (1992). Their final model, which controlled for two regions (United States and non-United States), fit the data somewhat better than had the previous models (adjusted R2 = .51, compared with .36 for Carlsen et al. ( 1992)).break

Page 143

Swan et al. (1997) abstracted data from 56 of the 61 underlying studies to control for additional variables, such as age, abstinence time, method of specimen collection, and percentage of men with proven fertility. In their analysis, studies were grouped into three regions: United States (27 studies, 1938-1988). Europe and Australia ( 16 studies, 1971-1990), and other (non-Western) countries (13 studies, 1978-1989). Multiple-regression analyses were used to fit linear and nonlinear models. Swan et al. (1997) included data from all regions in a single model, which fit separate lines (for the linear, spline, and step models) or curves (for the quadratic model) in each region. Their multiple-regression analyses considerably improved model fit. For example, the adjusted R2 was .80 for the linear multiple-regression model compared with .36 for the simple linear model. Studies from non-Western countries did not fit any of these models, perhaps reflecting the heterogeneity of the areas included, the small number of studies (13), and the short time during which these studies were published (12 yr). Nonlinear multiple-regression models fit the United States and European data nearly as well as did the linear multiple-regression model. All models show a decline, on average, in the United States and Europe, and none identifies any post-1970 increase in sperm concentration in men from these countries.

Of the studies included in these analyses, 88% were published after 1970. Because a possible decline during the last 20 yr of the study period was of particular interest, and because studies from all regions could be directly compared only after 1970, Swan and Elkin (1999) fit a multiple linear-regression model restricted to post-1970 studies. The results were comparable to those seen for the entire study period.

Becker and Berhane (1998) argued that the exclusion of five studies by Swan et al. ( 1997) had altered the statistical outcome. Two of the excluded studies did not meet the criteria for inclusion established in the original study by Carlsen et al. (1992). The other excluded studies (from Peru, Denmark, and Germany) were not written in English. The effect of these exclusions on the global average decline is minimal (overall slope = -0.95 x 106/mL, p < 0.001 vs. -0.93 x 106/mL, p < 0.001) (Swan et al. 1997). With respect to the more rigorous stratified analysis, after excluding these non-English studies the European trend was no longer statistically significant. The U.S. slope was, however, unchanged. Since the method of selecting these three non-English studies from the large non-English literature on semen quality is not known, the possible bias in the inclusion of these three studies cannot be evaluated.

These semen analysis studies were conducted in multiple laboratories using a variety of techniques, leading to concerns about the reliability of these data and possible biases that might have contributed to the observed decline. A recent analysis (Swan and Elkin 1999) examined these techniques and found no trend in variability or systematic changes in methods that would explain the observed decline. Nevertheless, controversy over the possibility of bias and the quality of the data remains that can only be resolved using carefully controlled prospective studies.break

Page 144

Although multinational trend studies are consistent with a downward trend in mean sperm concentration, this pattern may be confounded by local geographic variation (Fisch and Goluboff 1996; Saidi et al. 1999). Younglai et al. (1998) noted large differences in sperm counts in 11 centers across Canada. Eight local trend studies, published between 1979 and 1996, described trends in sperm concentration in normal men living within a single city, state, or country (Table 5-1). Only one study (Vierula et al. 1996) includes data collected before 1970, so these local trend studies, which included about one-third the number of men in the multinational trend studies (5,122 vs. 14,947), provide data only on post-1970 trends in sperm concentrations. There is considerable variability among these studies in design, population, and size; some measure mean sperm concentration at only a few time points and include a few hundred men, others include mean sperm concentration from 20 to 30 time points and more than 1,000 men. Some provide estimates of slope adjusted for age, abstinence time, and season; some are unadjusted for any covariates. Therefore, apparent geographic differences could in part be the result of these differences in method. Significant declines in sperm concentration were reported in Belgium (Van Waeleghem et al. 1996), Scotland (Irvine et al. 1996), and Paris, France (Auger et al. 1995). In the United States, significant increases were reported in Washington (Paulsen et al. 1996), as well as in Los Angeles, Minneapolis, and New York (Fisch et al. 1996). Other authors report no significant change in France (Bujan et al. 1996), Finland (Vierula et al. 1996), and Wisconsin (Wittmaack and Shapiro 1992). Differences such as those reported Fisch (114.0 x 106/mL in New York vs. 72.7 x 106/mL in California) are as large as the change in sperm concentration across the entire period studied by Carlsen et al. (1992) (113 x 106/mL to 66 x 106/mL). More recently, a study of regional differences in sperm concentration among fertile French men found that mean counts from eight regions within France ranged from 82 x 106/mL to 102 x 106/mL (Auger and Jouannet 1997).

Rasmussen et al. (1997) reported that sperm concentration had not changed in Denmark during the past 20-30 yr. Although it was noted that the method used to determine sperm counts overestimated sperm concentrations, the authors judged that it did not affect the conclusion that there was no change over time.1 In a study conducted in Australia using potential sperm donors, Handelsman (1997) found no significant difference in sperm concentration over time (1980-1995) or between years or according to year of birth. The overall median sperm concentration was 69 x 106 mL. Five studies of median sperm concentration were also conducted on potential participants in a male contraceptive study. The median sperm concentrations for two of the studies were significantly higher (103 x 106/mL and 142 x 106/mL; p < 0.05) than for the other three (63, 67, and 84 x 106/mL, respectively),continue

1 One committee member believes this finding may reflect possible selection bias as evidenced by under representation of men with low sperm counts.

Page 145

Page 146

as well as for potential sperm donors (median 69 x 106/mL). The authors suggest that this substantial variation between donor groups reflects bias from self-referred volunteers. Collectively, the studies described above suggest considerable geographic variation in sperm concentration; however, the degree to which the differences are ascribable to population selection is still not known. One of the reasons for the controversy regarding temporal trends in sperm concentration is that it is impossible to control for all confounding factors in retrospective studies due to limitations in the original data sets, and only prospective studies can clearly address the issue of the factors affecting sperm concentration. The question of whether sperm concentration has declined over the past 50 yr may not be the appropriate question for study. No analysis to date can prove or disprove a uniform global trend in sperm concentration. In fact, within single study centers and populations, considerable local variation has been demonstrated, with some studies suggesting a decline, and others no change, or even a possible increase in sperm concentration over the past 20 yr.

However, given the limited information available, one cannot assume an environmental etiology for the variability observed in human populations. Cross-sectional comparisons of sperm concentration now under way in comparably selected populations in several countries could identify areas of low (and high) sperm concentrations. Careful exposure assessments in these areas will be required to identify etiologic factors.

Joffe (1996) indirectly examined the relationship between sperm concentration and fertility by comparing time to pregnancy between Finland (high to normal sperm counts) and Britain (low sperm counts) in a pair of prenatal studies and a pair of cross-sectional studies. In both comparisons, time-to-pregnancy distributions were significantly shorter in Finnish couples (p < .05 and p < .025). The authors conclude that this supports the hypothesis that the difference in sperm counts between Finland and Britain is not artefactual. They also point out, however, that the lower smoking rates among Finnish women may be a factor. Zukerman et al. (1977) demonstrated that no statistically significant decline in fertility is observed until sperm counts reach a level of 10 x 106/mL, and even then sterility is not assured. However, a recent study of 430 couples planning their first pregnancy in which semen quality was measured at the onset of the study found that the probability of conception increased with increasing sperm concentration up to 40 x 106/mL. With higher sperm concentration, little association was seen (Bonde et al. 1998).

Wildlife Studies

Data on the reproductive and developmental effects in wild populations of fish, birds, alligators, turtles, amphibians, and panthers are summarized and evaluated below. A discussion of how these effects could influence the population dynamics of some species is presented in Chapter 10.break

Page 147

Fish

Three well-studied polluted waters have been associated with endocrine modification and altered reproductive physiology of fish: (1) waters that receive sewage treatment plant effluent (STPE), (2) contaminated waters of the North American Great Lakes, and (3) waters that receive bleached paper mill effluent (BPME).

Sewage Treatment Plant Effluent

In the United Kingdom, intersex adult roach (Rutilus rutilus) with testes containing oocyte have been observed in wild populations near sewage treatment plants. These observations led some researchers (Purdom et al. 1994) to hypothesize that sewage effluents might contain estrogenic substances. Immature male rainbow trout and common carp were placed in cages in the effluent stream of 15 sewage treatment plants throughout southern England and Wales (Purdom et al. 1994), and the fish were evaluated for estrogen-induced vitellogenin synthesis. These fish exhibited a site-dependent, 500- to 100.000-fold increase in plasma vitellogenin concentration over fish in control groups. Subsequent work has also shown that rivers downstream of STPE inputs are estrogenic to caged rainbow trout and that up to 100% of the male roach exhibit intersex conditions (Harries et al. 1997; Jobling et al. 1998). High levels of vitellogenin induction and up to 20% ovotestis have been observed in male flounder (Platichthys flesus) from industrialized estuaries in the United Kingdom (Allen et al. 1998).

Estrogenic compounds excreted by humans into wastewater, such as ethinylestradiol (the synthetic estrogen used in birth control pill formulations), have been postulated to be the active ingredients in the effluents (Purdom et al. 1994). To test this hypothesis, male rainbow trout were injected intramuscularly with ethinylestradiol at varying dosages (Purdom et al. 1994). Ethinylestradiol was shown to be more potent in trout than is naturally occurring estradiol, based on the magnitude of vitellogenin induction. Other groups of trout placed in water containing 1-10 ng/L ethinylestradiol exhibited plasma vitellogenin concentrations similar to those produced by the fish caged in effluent streams (Purdom et al. 1994). Some investigators have not detected ethinylestradiol in STPE in the United Kingdom (Purdom et al. 1994); others have detected it at <20 ng/L (Aherne and Briggs 1989). However, recent work by Desbrow et al. (1998) showed that ethinylestradiol and the natural steroids estradiol and estrone are the major contributors to estrogenicity of several STPEs.

Alkylphenol ethoxylates (APEs) and their degradation products have also been hypothesized to be responsible for observed estrogenic effects. Concentrations of APEs have been measured at hundreds of micrograms per liter in domestic sewage effluent (Ahel and Giger 1985) and at even greater concentrations in industrial effluent (Naylor et al. 1992). The nonethoxylated breakdown productscontinue

Page 148

of the APEs 4-tert-octyl-phenol ethoxylate and nonylphenol ethoxylate, octylphenol and nonylphenol. have been shown to be estrogenic in fish cells in vitro (Jobling and Sumpter 1993; Jobling et al. 1995). To examine the effects of those compounds on testicular function, male rainbow trout undergoing sexual maturation were placed in water containing 30 µg/L octylphenol, nonylphenol, or two other alkylphenols (the concentration is representative of those found in STPE of some rivers) for 3 wk. All of the alkylphenols caused a significant increase in plasma vitellogenin concentration, accompanied by a concomitant and significant decrease in the rate of testicular growth (Jobling et al. 1996). Dose-response studies, in which trout were exposed to various concentrations of alkylphenols for 3 wk at three distinct stages of testicular development, showed that the concentration of vitellogenin produced depended on the estrogenicity of the compound, the dosage, and the stage of testicular development. When testicular growth rates of alkylphenol-exposed fish were compared with those for control fish, testicular growth rate was inhibited by as much as 50% after exposure to octylphenol, the most estrogenic alkylphenol. Inhibition of testicular growth was directly correlated with the estrogenic potency of the various experimental treatments (Jobling et al. 1996). Field studies on the River Aire in Yorkshire, United Kingdom, have shown that the strong estrogenicity found below one STPE input could be largely attributed to alkylphenols derived from textile mill effluent (Harries et al. 1997). The voluntary cessation of the use of cleaning products, containing the estrogenic surfactant nonylphenol by the textile industry in England led to the amelioration of this problem in male fish in the Arne river (Matthiessen and Sumpter 1998).

Studies at other locations outside of England have confirmed the observation that male fish exposed to STPE exhibit elevated plasma concentrations of vitellogenin. In a study of male common carp from five riverine locations in Minnesota, elevated plasma vitellogenin and depressed plasma testosterone concentrations were observed in fish collected from 1 location, an effluent channel below a metropolitan sewage treatment plant on the Mississippi River (Folmar et al, 1996). In another study, Nichols et al. ( 1999) measured concentrations of estradiol and vitellogenin in plasma of fathead minnows and the common goldfish in central Michigan. Caged fish were placed directly in the effluents of seven treatment plants with a range of treatment types and sources of waste. The variability in vitellogenin measurements was great, making it difficult to demonstrate any statistically significant differences among locations. Elevated vitellogenin was observed in only one of the seven locations. Even though induction of vitellogenin was observed in both species, the degree of induction was only about 6-fold over background. The investigators concluded that the estrogenic potency of STPE was slight.

Overall, the expression of vitellogenin in male fish exposed to STPE in observational and experimental studies indicates that some STPEs have estrogenic properties that could cause the observed reproductive alterations in the wild.break

Page 149

North American Great Lakes

For nearly 2 centuries the Great Lakes have received industrial and municipal wastewater. The fish of the Great Lakes have been found to contain literally hundreds of synthetic and natural compounds (Fitchko 1986; Allan et al. 1991; IJC 1993). However, chemical contaminants such as DDE and PCBs have decreased markedly in recent years with improvements in reproductive success of several species of predatory birds (Tremblay and Gilman 1995). Two of the best-studied fish of the Great Lakes have been the introduced salmon and the native lake trout.

Pacific salmon (Oncorhynchus spp.) were introduced into the Great Lakes, mainly in the 1960s, and soon became established there (Becker 1983), although still are largely maintained through stocking. In 1980 and 1981, the rearing mortality of chinook salmon increased to 7.4% and 19.5%, respectively, from 23% before 1980 (Eadie 1983; Allan et al. 1991). Although a normal hatch rate was observed, hatched fry failed to feed, swam in circles, lost equilibrium, and died within several weeks. This upward trend in rearing mortality was alarming, especially during a period when the concentrations of many of the routinely monitored residues (PCBs, DDT, and TCDD) were decreasing in fish, and it raised the question of whether other synthetic residues might be responsible for the observed effects (Mac and Seeyle 1981a; Mac et al. 1985, 1993). Because toxic substances concentrate in the lipids of fish eggs and become more concentrated in the remaining egg yolk as the fry absorb the yolk, it is believed that the fry receive a large dose of xenobiotics at the "swim-up" stage when the last of the yolk is absorbed (Atchison 1976). A cursory investigation of the concentrations of 14 organochlorine compounds in chinook salmon fry revealed that normal fry contained 2.5 mg/kg total PCBs, whereas sick and dead fry contained 3.9 and 5.2 mg/kg total PCBs, respectively (Flagg 1982). Concentrations of p,p'-DDE were 8.0, 13.0, and 17.0 mg/kg for normal, sick, and dead fish, respectively. It should be noted that a syndrome similar to that observed in the chinook salmon was found in coho salmon fry reared in Michigan during the late 1960s and was attributed to residues of DDT in coho salmon eggs (Johnson and Pecor 1969).

Based on laboratory studies of different fish species, the concentration of DDT that is lethal to salmonid eggs is estimated at 1.0-10 mg/kg, wet weight (ww) (Macek 1968; Hopkins et al. 1969; Burdick et al. 1972). Current concentrations of DDT in salmonid eggs from the Great Lakes are in the range of 0. 1-2 mg/ kg, ww (Giesy in press), depending on species and location. In 1984, the concentration of DDT in chinook salmon eggs from Lake Michigan was 1.2 mg/kg, ww (Giesy et al. 1986). Historically, concentrations of DDT were as great as 10 mg/ kg, ww, in eggs of fish from the Great Lakes (Burdick et al. 1964; Hopkins et al.continue

Page 150

1969; Johnson and Pecor 1969; Atchison 1976). Thus, it is plausible that DDT could have caused reproductive impairment of salmon in the Great Lakes. Attempts to correlate rearing mortality in the field with DDT concentrations in the environment have been inconclusive: Some researchers have found a positive correlation (Giesy et al. 1986): others have not (Mason et al. 1967; Giesy et al. 1986).

Another major class of contaminants of the Great Lakes—PCBs—has been found to affect early and adult life stages of fish in laboratory studies (Delzell et al. 1994). When total concentrations of PCBs in fish and fish eggs from the Great Lakes were examined and compared with the results from controlled laboratory studies, the calculated hazard indices indicated that concentrations of these residues were probably sufficient to have caused egg and fry mortality (Willford et al. 1969: Halter and Johnson 1974; Lidman et al. 1975, Mauck et al. 1978; Guiney et al. 1980; FWS 1981, Binder and Lech 1984; Fisher et al. 1994). However, it has been difficult to correlate concentrations of PCBs with observed adverse effects in field studies (Gilbertson et al. 1990; Mac and Gilbertson 1990; Gilbertson 1992: Delzell et al. 1994). This is probably due to two factors: The exposure of fish to several compounds simultaneously and the great deal of variation in exposure and responses has made relating the effects to a single compound impossible (Stauffer 1979; Giesy et al. 1986).

Calculation of hazard indices indicates that currently, and possibly historically, TCDD-like compounds are the critical toxicants in the Great Lakes fish (Giesy et al. 1994a). Coho salmon exposed to 5.4 ng TCDD/g for 12 h had reduced growth and survival (Miller et al. 1979), and the threshold concentration in 24-96 h exposures was 0.54-5.4 ng TCDD/g. Dioxins and dioxin equivalents from all dioxinlike chemicals (TCDD-EQ) also have been shown to be toxic to fish at concentrations that currently occur in the eggs of salmonid fish in the Great Lakes (Giesy et al. 1986; Ankley et al. 1991; Newsted and Giesy 1992. 1993; Williams and Giesy 1992). Comprehensive reviews of the effects of TCDD on fish of the Great Lakes are given by Walker and Peterson ( 1994) and by Cook et al. (1993).

Since the early 1970s, pink, chinook, and coho salmon introduced into the Great Lakes have exhibited thyroid enlargement (goiters). These enlargements, due to both hypertrophy and hyperplasia, have ranged in size from gross lesions to histologic enlargements (Leatherland 1992, 1998). Because PCBs have been shown to be goitrogenic in various vertebrates (Brouwer et al. 1989; Leatherland 1992), it was hypothesized that these contaminants could be the cause. However, coho salmon collected from Lakes Erie, Ontario, Superior, and Michigan exhibited no correlation between the concentration of PCBs in the body and thyroid enlargement (Sonstegard and Leatherland 1976). It has been noted that rodents fed PCB-contaminated salmon (30-100% of the diet) from the Great Lakes for 2 mo showed significant thyroid enlargement, hepatomegaly, increased mixed-function oxydase activity, and a reduction in plasma thyroxine concentrationscontinue

Page 151

(Leatherland 1992). Why a relationship between PCB contamination and goiter formation is not apparent in coho salmon is not known. Other studies indicate that goiter formation in fish might be caused by bacterial agents rather than by PCBs (Gaitan et al. 1980; Leatherland 1992, 1993).

In Lake Erie, some coho salmon (Fairview stock) have displayed a series of reproductive dysfunctions (Leatherland 1993). These include reduced egg-survival-to-hatch rates, depressed steroidogenesis, lowered plasma concentrations of steroids and pituitary hormones in males and females, and lowered egg thyroid hormone concentrations. Another reproductive abnormality is the increased incidence of embryonic teratogenesis. In some years, 90% of male salmon exhibited early onset of sexual maturation and reduced development of secondary sex characteristics (Leatherland 1993). At this time, there is no evidence that goiter or any of the reproductive dysfunctions observed in coho salmon are caused by exogenous-hormone-modulating compounds.

Annual rearing mortalities in lake trout fry of as much as 97% were described for hatchery-reared fish between 1978 and 1981 (Mac et al. 1985). Mortality could not be attributed to disease or nutrition and was characterized by erratic swimming behavior and loss of equilibrium before death. Poor survival was significantly correlated with the source of eggs and sperm and not to the residue concentrations or chemistry of the water in which the eggs were reared (Mac et al. 1985). In addition, a number of fry developed blue sac disease, an edematous condition that results in fluid filling the yolk sac, which takes on a bluish color, and leads to the death of the egg. Early efforts to correlate the degree of fry mortality in lake trout with concentrations of PCBs and DDT were unsuccessful (FWS 1981; Seelye and Mac 1981).

DDT and PCBs have been shown to cause mortality in lake trout fry and eggs in laboratory studies (Berlin et al. 1981). However, the concentrations required to cause 30-50% mortality were as much as 25 times greater than were the concentrations actually found in the eggs in the Great Lakes. In addition, the concentrations of DDT and PCBs were higher during the period of normal rates of rearing success (1972-1975) than they were during the period when abnormally low rates of fry survival were observed (1978-1981 ). Subsequent studies have identified the total concentrations of TCDD-equivalents (EQ) as a more plausible causative agent (Symula et al. 1990; Walker and Peterson 1994). and TCDD and structurally similar compounds have been found to induce blue sac disease in the laboratory (Walter and Peterson 1994). Current concentrations of TCDD-EQ in fry are near the threshold for mortality in lake trout fry (Walter and Peterson 1994), thus it is likely that the concentrations of TCDD-EQ in the lake trout eggs were well above the threshold in the recent past, and that current concentrations might still impede survival.break

Page 152

Bleached Paper Mill Effluent

Since the late 1980s, a series of studies has investigated the effects of BPME on the reproductive physiology of the white sucker in the Great Lakes. BPME-exposed white suckers collected from Jackfish Bay, Lake Superior, showed decreased concentrations of plasma sex steroids, decreased egg and gonadal size, and delayed sexual maturity (Van Der Kraak et al. 1992). To examine those effects further, white suckers were collected (Munkittrick et al. 1994) from the receiving areas of 12 paper mills (excluding Jackfish Bay). BPME from paper mills that used chlorine-based and sulfite-based processes were included, as was effluent from mills that used primary or secondary treatment facilities. Fish exposed to BPME exhibited increased hepatic mixed-function oxydase (ethoxyresorufin-O-deethylase) induction, and the greatest concentrations of dioxin were found in the liver. Fish collected from sites where sulfite-based processes was used had elevated enzyme activity and decreased steroid concentrations that were not correlated with concentrations of hepatic mixed-function oxydase activity (Munkittrick et al. 1994). BPME-exposed white suckers exhibited effects on steroid synthesis pathways other than those of the sex steroids.

Male and female BPME-exposed white suckers exhibited a 30- to 50-fold decrease in gonadotropin hormone-II (GtH-II) (Van Der Kraak et al. 1992). To investigate this observation further, BPME-exposed fish were collected and given a single injection of a synthetic-gonadotropin-releasing hormone (sGnRH-A). This treatment induced an increase in plasma GtH-II, although the magnitude of response was significantly lower in the exposed population when compared with the control population (Van Der Kraak et al. 1992). In the same study, ovulation in BPME-exposed females did not occur after injection of the GnRH analogue, yet 10/10 control females ovulated within 6 h. Not only were concentrations of 17-20-progesterone lower for BPME-exposed fish, the fish did not exhibit an increase in the plasma concentrations of this hormone after sGnRH-A treatment. Plasma testosterone in females and males and 11-keto-testosterone in males were elevated in fish from the control site compared with the BPME site, but they were not increased after the GnRH injection. In contrast, BPME-exposed fish showed a transient increase in testosterone in response to the sGnRH-A injection (Van Der Kraak et al. 1992). These findings suggest an inhibitory effect on the pituitary gland.

As a follow-up to the in vivo plasma hormone studies, in vitro cultures of follicles obtained from the BPME-exposed fish and control fish were incubated with and without human chorionic gonadotropin (hCG) or forskolin, a stimulator of adenylate cyclase activity. Follicles from fish exposed to BPME had a decrease in basal secretion of testosterone and 17-20-progesterone and a reduced response to either hCG or Forskolin (Van Der Kraak et al. 1992). There was no significant difference in follicular production of prostaglandin E, either basally or after stimulation with phorbol ester or calcium ionophore a23187, when BPMEcontinue

Page 153

and control follicles were compared. Together, these data suggest that exposure to BPME affects the ovary selectively rather than generally (Van Der Kraak et al. 1992). This analysis also provides evidence that the exposure to BPME affects reproduction in the white sucker by acting at different points in the pituitary-gonadal axis.

BPME is known to be a complex mixture of artificial and naturally occurring compounds, that have been shown to affect the reproductive physiology of fish. For example, BPME contains numerous chlorinated byproducts of the bleaching process (Peterman et al. 1980) and one phytoestrogen. ß-sitosterol. is present at some mill sites at 1,200 µg/L in primary treated BPME and at 280 µg/L after secondary treatment. Experimental study has shown that injections of ß-sitosterol alter the reproductive physiology of the domesticated goldfish (MacLatchy and Van Der Kraak 1995). Testosterone and 11-keto-testosterone in male goldfish and testosterone and estradiol in female goldfish were significantly depressed after treatment; GtH-II concentrations were increased at the same time. As with virtually all toxicologic experiments, the doses used were higher than those monitored in the environment, making direct comparisons difficult to determine (see Chapter 4). In vitro cultures of testes removed from ß-sitosterol-treated male goldfish produced decreased testosterone and pregnenolone concentrations both basally and after hCG stimulation. In vitro cultures of follicles removed from ß-sitosterol-treated females produced decreased pregnenolone concentrations basally. After hCG stimulation, testosterone and pregnenolone remained depressed. However, basal testosterone was not significantly different from controls (MacLatchy and Van Der Kraak 1995).

Those data indicate that synthetic and natural HAAs released from paper mills that process raw wood products can alter the reproductive physiology of fish. How these chemicals interact with each other is not known and continues to be investigated.

Amphibians

There are many reports that salamander, toad, and frog populations are declining. Disappearances of these amphibians have been reported in areas of North America, Central and South America, Europe, Asia. Africa, and Australia (Blaustein and Wake 1990; Wake 1991: Wake and Morowitz 1990). Although there is debate about whether such declines are occurring multinationally. local declines have been documented (Blaustein and Wake 1990; Phillips 1990). For example, 80% of the Cascade frogs that have been monitored in Oregon since the mid-1970s have disappeared (Blaustein and Wake 1990), and the gastric-brooding frog of Queensland, Australia, is believed to be extinct (Tyler 1991 ). Similarly, the golden toad of the Monteverde Cloud Forest Preserve in Costa Rica has not been found in its traditional breeding sites since 1987 (Wake and Morowitzcontinue

Page 154

1990). The hypotheses proposed to explain local declines include disease and the introduction of exotic predators (Hayes and Jennings 1986; Carey and Bryant 1995), modification in exposure of eggs to ultraviolet light (Blaustein et al. 1994), and pathogens (Morell 1999). There also have been numerous reports of amphibian kills after chemical spills or agricultural spraying. Environmental contamination also might contribute to continuing declines by affecting the growth and development of young amphibians (Carey and Bryant 1995).

There have been reports of deformities, such as extra, missing, or malformed limbs, in wild populations of frogs in the United States and Canada (Schmidt 1997; Ouellet et al. 1997; Tietge 1997). The deformities could be caused by exposure to chemical pollutants, increased exposure to ultraviolet light, parasite infestations, or some combination of the three (Schmidt 1997). However, more recent studies indicate that the physical presence of parasitic trematode worms, which form cysts in the developing hind-limb regions of frogs, might be primarily responsible for supernumerary limbs in frogs (Sessions and Ruth 1990; Sessions 1997; Johnson et al. 1999; Sessions et al. 1999). The phenomenon seems to be caused by physical disturbances, because similar effects can be induced by implanting inert resin beads, similar in size to trematode cysts, into the developing limb buds of frogs (Sessions and Ruth 1990). However, results from studies conducted by the Minnesota Pollution Control Agency and the National Institutes of Environmental Health Sciences have linked gross deformities observed in frogs in northwestern Minnesota to biologically active agents in the water they inhabit (Burkhart et al. 1998). Specifically, water samples taken from ponds with high incidences of frog malformations (affected sites) were tested in FETAX assays, in which X. laevis embryos were exposed to pond water for 96 h and observed for mortality and malformations. The results of these assays were compared with those conducted with water samples taken from ponds with unaffected frog populations (reference sites). The water from affected sites induced mortality and malformations, whereas the water from reference sites did not. The observed malformations were dose dependent and reproducible. Research projects are under way to try to establish a plausible link between exposure to specific chemical contaminants and the appearance of specific deformities under laboratory conditions.

The role of hormones, such as estradiol, testosterone, and corticosterone, in the growth and metamorphosis of toads has been studied in the laboratory (Richards and Nace 1978; Gray and Janssens 1990; Hayes et al. 1993; Hayes 1995; Hayes and Wu 1995a,b). For example, the action of exogenous corticosterone on toads has been shown to closely resemble the effects of exogenous thyroid hormones, suggesting that steroids might interact with endogenous thyroid hormones (Hayes et al. 1993; Hayes 1997). In a study of DDT, male African clawed frogs were induced to synthesize the female yolk protein vitellogenin in the liver after intraperitoneal injection of 1.0 or 250.0 µg o,p'-DDT/g body weight for 7 d (Palmer and Palmer 1995). Vitellogenin synthesis is considered a hallmark of exposure to estrogens, but other studies involving DDT suggest that DDT might act as acontinue

Page 155

corticoid mimic or stressor (Hayes et al. 1997). However, studies are needed of altered hormone concentrations or modifications of the endocrine system of free-living amphibians to explore whether deformities observed in the wild are related to hormonal modification.

Reptiles

Alligators

The American alligator of Lake Apopka, Florida, is one of the most cited examples of a wildlife population affected by environmental toxicants, including HAAs. The work on the alligator in Florida's wetlands began as the population rebounded as a result of protection under the terms of the Endangered Species Act. Studies of the reproductive biology of alligator populations began in the late 1970s to determine whether this animal could sustain annual harvests for their hides. During this work, most of the study lakes showed reduced egg viability and elevated embryonic mortality (Masson 1995). Lake Apopka exhibited a massive reduction in neonatal and juvenile populations and extremely high embryonic and neonatal mortality in the early 1980s (Woodward et al. 1989, 1993). Chapter 10 discusses the population effects in detail.

Lake Apopka is Florida's most polluted lake (EPA 1979a; Schelske and Brezonik 1992). Contamination of the lake has come from extensive agricultural activities around the lake and from sewage and runoff from several municipalities. In addition, there was an accidental release of the pesticide dicofol (contaminated with up to 15% DDT and its metabolites) (Clark 1990) and sulfuric acid (EPA, unpublished report) in 1980 from the Tower Chemical Company. Water samples obtained after the spill showed DDT concentrations ranging from nondetectable to 433 µg/L and dicofol concentrations of 66-150 µg/L (EPA, unpublished report). Sediment sample concentrations of DDT were 31-1611 µg/ kg, dry weight, and of dicofol were 6,400-31,000 µg/kg, dry weight (EPA, unpublished report). One report from sampling conducted in 1993 (EPA 1994b) indicates that DDT and its breakdown products were still elevated in sediment samples, as was the pesticide toxaphene.

A study by Heinz et al. (1991) identified high concentrations of various persistent pesticides and their metabolites in alligator eggs collected between 1984 and 1985 from Lake Apopka (Table 5-2). With the exception of toxaphene, dieldrin, and chlordane, the pesticides and metabolites bind to the alligator estrogen receptor (Vonier et al. 1996). No correlation was found between elevated concentrations of organochlorine compounds and poor egg viability (Heinz et al. 1991). However, the mean concentrations of p,p'-DDE observed, 5.8 ppm wet weight (1984; range, 3.4-7.6 ppm) and 3.5 ppm wet weight (1985; range, 0.89-29 ppm), are above the concentrations known to reduce hatching success and cause deformities (Cooper 1991). Studies by EPA (1994b) have shown that juvenilecontinue

Page 156

and hatchling alligators from Lake Apopka have elevated lipid ( 1.6-8.5 ppm) and hepatic (0.013-0.17 ppm) concentrations of p.p'-DDE. Those concentrations were significantly higher in hatchlings with developmental abnormalities. Specific reproductive anomalies found in the alligators of Lake Apopka are discussed below.

Alligators hatched from eggs collected from Lake Apopka have exhibited abnormal gonadal morphology. Male juvenile alligators had poorly organized testes with unique, aberrant structures within the seminiferous tubules (Guillette et al. 1994). Some of the germ cells had clear mitotic figures. suggesting that premature spermatogenesis had begun. No mitotic or meiotic activity was seen in the testes obtained from males of the less contaminated reference lake, Lake Woodruff. Female alligators examined at 6 mo had prominent polyovular follicles, and many of the oocytes were multinucleated (Guillette et al. 1994). A normal ovarian follicle contains a single oocyte with a single nucleus. Polyovular follicles contain more than one oocyte, and in the case of the neonatal and juvenile alligators, as many as six discrete oocytes were counted in a follicle. All females hatched from the Lake Apopka eggs exhibited this condition; none of the reference females did (Guillette et al. 1994). Follow-up studies of juvenile alligators from Lake Apopka and the reference lake indicate that some oocytes are multinucleated in females from the control group, but polyovular follicles were not seen in any of the control females (Pickford 1995).

The abnormal morphologies of the ovarian follicles and oocytes in female alligators from Lake Apopka are similar to those observed in mice treated withcontinue

Page 157

DES (Iguchi et al. 1990; Guillette et al. 1994). Mouse polyovular follicles can be stimulated to ovulate and can be fertilized in vitro and in vivo, although the number of eggs fertilized is significantly less than for uniovular follicles (Iguchi and Takasugi 1986; Iguchi et al. 1986, 1990). Fertilized mouse ova from DES-induced polyovular follicles develop to implantation-stage embryos, but the frequency of embryos reaching that stage is significantly reduced compared with embryos derived from uniovular follicles (Iguchi et al. 1991). It has been hypothesized that the low egg viability observed in Lake Apopka could in part result from the greater frequency of polyovular follicles in females from this population (Guillette and Crain 1996). Egg viability on Lake Apopka has averaged less than 20% since the mid-1980s, and embryonic death occurs mainly at the zygote to late gastrula stage (Masson 1995). Additional work is needed to determine the causes and the significance of polyovular follicles in wild populations of alligators. Since Sager et al. (1991) has shown that developmental exposure of male rats to PCBs results in loss of embryos (at the blastocyst stage) they sire as adults, the importance of sperm abnormalities in embryonic loss also has to be considered.

In addition to morphologic abnormalities of the gonad, Lake Apopka male and female neonatal (6 mo old) alligators exhibited abnormal plasma sex steroid concentrations (Guillette et al. 1994, 1995a). Male alligators had greatly reduced plasma testosterone concentrations compared with males from the less-contaminated reference lake. The females from Lake Apopka had plasma estradiol concentrations that were higher than those found in reference females.

In vitro synthesis of estradiol was significantly different from that predicted by plasma concentrations: Ovaries from Lake Apopka females synthesized less estradiol than did ovaries of females from the reference lake (Guillette et al. 1995a). Testes from Lake Apopka males synthesized significantly greater concentrations of estradiol than did those of reference males. Testosterone synthesis from the testes was the same for animals from both lakes. Abnormalities in plasma hormone concentrations observed in hatchlings persisted into later prepubertal stages (Guillette et al. 1997). The current hypothesis is that modifications in gonadal synthesis, hepatic degradation, and plasma-steroid-binding protein concentrations appear to act in concert to modify plasma hormone concentrations in the Lake Apopka alligators (Guillette et al. 1997; Arnold et al. 1996).

Male alligators from Lake Apopka had significantly smaller penis size than did males from the reference lake (Guillette et al. 1997). Because penis development and size depend on elevated androgen concentrations in alligators (Raynaud and Pieau 1985), the researchers suspect that abnormal androgen concentrationscontinue

Page 158

or functioning were the cause. Although there was a positive relationship between plasma androgen concentrations and penis size in the males from the reference lake, no relationship was observed among male alligators from Lake Apopka, although penis dimensions did exhibit allometric relationships with body size. Differences between collection sites were noted on Lake Apopka. Juveniles in the area of Gourd Neck Spring near the drainage site of a pesticide spill exhibited markedly smaller penises than did juvenile males found along the northwestern shore (Guillette et al. 1996a). One hypothesis is that androgen function was modified during embryonic development and early life (Guillette et al. 1996b). Recent evidence suggests that Lake Apopka males with small penises do not have reduced plasma concentrations of 5a-dihydrotestosterone but do have reduced plasma concentrations of testosterone. Most female alligators exhibited normal secondary sex characteristics at the gross morphologic level. However, a few individuals on Lake Apopka had hypertrophied clitorises (Guillette et al. 1997).

Alligator populations on Lake Apopka exhibited a dramatic decline in numbers and recruitment in the 1980s after the pesticide spill described above. Recruitment problems at first were thought to be associated with habitat modifications—nests and nesting materials were abnormal—but this was not borne out (Masson 1995). The age of the female also can affect egg quality, and although no significant differences in female morphometrics have been noted, females from Lake Apopka are slightly larger (Rice et al. 1996). The isolation of populations also can contribute to inbreeding depression of reproduction, but alligator populations in Florida do not appear inbred. The major watersheds are extensively connected, and adult alligators move over extensive ranges (Abercrombie 1989).

Given the above observations, it has been proposed that the abnormalities observed in the alligators of Lake Apopka can be explained, in part, by hormonal disruption caused by pesticides (dicofol, DDT, and toxaphene) or their metabolites (p,p'-DDE) (Guillette et al. 1997). Experimental studies on this species (Crain et al. 1998), other reptiles (Bergeron et al. 1994), and birds (Eroschenko and Palmiter 1980; Biessmann 1982: Rattner et al. 1984; Elliott et al. 1994; Fry 1995) suggest a relationship between the developmental abnormalities observed and exposure to xenobiotics, such as p,p'-DDE, is likely.

Turtles

Snapping turtles of the Laurentian Great Lakes region of North America feed near the top of the food chain and are exposed to persistent environmental contaminants that bioaccumulate and biomagnify in the food chain. The turtles have been studied to determine whether there is a connection between environmental contamination and developmental abnormalities (Bishop et al. 1991). Eggs collected from various localities on Lake Ontario, Lake Erie, and the upper St. Lawrence River exhibit elevated concentrations of PCBs, dibenzo-p-dioxins,continue

Page 159

dibenzofurans, and various organochlorine pesticides (DDT, dieldrin) or their metabolites (p,p'-DDE) (Bishop et al. 1991; Hebert et al. 1993; Struger et al. 1993). Eggs containing the greatest concentrations exhibited significantly greater rates of embryonic mortality and deformity (deformed tails and deformed and stunted legs) when compared with eggs from reference sites. Bishop et al. (1996) reported significant increases in PCBs, p,p'-DDE, and dieldrin between 1984 and 1991 in Lake Ontario eggs, and present geographic and temporal patterns demonstrate a large variation in the contaminant load among individual turtles. PCBs and pesticides have been found in turtle embryos throughout organogenesis and muscle and skeletal development (Bishop et al. 1995). As described for the alligator eggs collected from Lake Apopka (Masson 1995), the greater rates of embryonic mortality in the more contaminated snapping turtle eggs occurred early in development (Bishop et al. 1991). Although gross anatomical deformities were noted in many hatchlings from eggs collected at the most contaminated sites, histologic examinations were not performed. Likewise, no hormone data are available. Thus, it is unknown whether the reproductive organs of these animals were abnormal or whether the observed anatomical deformities might have been hormonally mediated. Also, the sex ratio of the offspring was not recorded, which could have had important implications.

Studies of turtles in temperature-shift experiments have revealed that the duration and magnitude of incubation temperature affect gonadal development during roughly the second third of egg incubation (Bull et al. 1990; Wibbels et al. 1991). The experiments indicate that sexual determination is extremely labile during the temperature-sensitive window. Exogenous estradiol can cause gonadal feminization of turtle embryos that are incubated at male-producing temperatures. Higher doses of estradiol result in the production of significantly more female turtles (Crews et al. 1991). Exposure to exogenous estradiol alters gonadal differentiation during the same developmental period (Wibbels et al. 1991), which suggests that temperature and estradiol act in a common pathway. However, the effects of estradiol are not completely understood.

The effects of exposure to various hydroxylated PCB congeners on gonadal development of turtles also have been studied (Guillette and Crain 1996). These congeners induce sex reversal in turtles when applied to the outside of the eggshell; the effects observed are similar to those described above for estradiol. Bergeron et al. (1994) observed that 100 µg 2',4',6'-trichloro-4-biphenylol induced 100% sex reversal (based on histologic examination of gonads and internal ducts) in the red-eared turtle, whereas treatment with 100 µg 2'.3',4',5'-tetrachloro-4-biphenylol stimulated total sex reversal in 50% of the embryos and partial sex reversal (intersex) in 21% of the embryos. The hydroxy-PCBs used in this study are not found in the environment. It is still not known whether HAAs or any other contaminants might be responsible for the effects observed in wild populations of snapping turtles in the Great Lakes region.break

Page 160

Birds

Studies examining the birds of the Laurentian Great Lakes provided much of the initial data that led to the concept of hormonally active environmental pollutants (Colborn 1990). Contaminant-related effects on the reproductive health of several avian species appear to involve modifications during embryonic development (Colborn et al. 1993; Fox 1992). The studies of interest have shown alterations in sexual behavior, abnormal reproductive morphology, severe developmental abnormalities associated with growth and metabolism, and eggshell thinning.

An important aspect of understanding the potential effects of exposure to HAAs on the development of avian reproductive systems is that estrogen plays an important role in regulating the course of development and adult functioning of the gonads and accessory reproductive structures (Fry 1995). Estradiol in conjunction with other endocrine and paracrine factors, is implicated in the unilateral development of the left ovary and regression of the right ovary. Estradiol also influences whether the embryonic tissues that differentiate into oviducts and shell glands persist or regress, via the interaction of estradiol with Müllerian inhibiting hormone; the action of Müllerian inhibiting hormone is inhibited by estradiol. The active differentiation of female reproductive structures under the action of estrogen is in direct contrast to mammals, in which males are the heterogametic and differentiating sex, and estradiol is not required for differentiation of the ovary, although in mammals, estrogen and functioning estrogen receptors are required for subsequent normal functioning of the ovary (Lubahn et al. 1993). Thus, embryonic exposure to environmental estrogens or other HAAs can generate end points in birds that are different from what might be expected in mammals or other vertebrate species that do not rely on estrogen-induced sexual differentiation. This concept of varying end points among species is essential to explaining the effects observed in wildlife. Such variations do not preclude the use of data from birds or other wildlife to predict abnormalities or effects in other species. The information can be used to help determine which end points are the most likely for a given species and to explain why effects are seen in some groups and not others. Thus, an understanding of the role of estrogen in avian development leads to the prediction that exposure to environmental estrogens could alter sex ratios and feminize male birds even though different outcomes would be predicted in other vertebrates.

Sex Ratios

During the 1950s and 1960s, when organochlorine (e.g., o,p'-DDT, PCBs, dioxin) contamination was at its greatest in the United States, an increased incidence of female-female pairings in gull populations was observed in colonies in California and the Great Lakes (Schreiber 1970; Harper 1971; Hunt and Huntcontinue

Page 161

1973; Gress 1974: Hand 1980; Shugart 1980; Fitch and Shugart 1983; Fry et al. 1987). This phenomenon is usually estimated by documenting the number of nests that contain five or more eggs (supernormal clutch); a single female gull typically lays one to three eggs.

The most dramatic and well-documented example of sex skew occurred in the western gull population on Santa Barbara Island in California from 1968 to 1978 (Hunt et al. 1980). The adult sex ratio in that population was measured by laparotomy of 856 captured birds to be 0.26 males to females. The investigators also calculated the male-to-female ratio by estimating the number of nests on the island (896), the number of nonbreeding birds (200), and, based upon the number of nests with more than three eggs, the percentage of female-female pairs (15%). Using those estimates, the male-to-female ratio was 0.67. Because many birds laid fewer than normal numbers of eggs in 1978, the investigators believe that the estimate for female-female pairs might been too low. Therefore, the ratio of males to females was probably between 0.26 and 0.67.

A supernormal clutch incidence of 0.6% to 1% was documented in northeastern Lake Michigan herring gulls from 1978 to 1981 (Shugart 1980: Fitch and Shugart 1983). Both the California population and the Great Lakes gulls were exposed to great levels of organochlorine contamination, including DDT, during the 1950s to the 1970s (Fry and Toone 1981). Several historical studies have been done to investigate the occurrence of supernormal clutches in gulls using literature sources and museum specimens to determine whether incidences have actually changed in the pre- and post-DDT era. The incidence of supernormal clutches has decreased significantly for many species of terns throughout the United States (Conover 1984a). Supernormal-clutch incidence had only increased significantly in western gulls, herring gulls nesting in the Great Lakes, and Caspian terns breeding in the United States since 1950. Supernormal clutches were a regular occurrence in ring-billed and California gulls before the DDT era, and their occurrence has not changed over time (Conover and Hunt 1984a). In contrast, supernormal clutches were not found regularly in western or herring gulls until after 1950, and the sex ratio for their populations as a whole has changed dramatically for both species toward an excess of females since then. Sex skew and its effects on the population dynamics of gulls are discussed in detail in Chapter 10.

There are a number of hypotheses concerning sex skew. In general, female-female associations in gull colonies are believed to occur when there is a relative shortage of breeding male gulls available. Experimental manipulation of sex ratios in gull colonies by selectively removing males from stable colonies has demonstrated that sex ratio skew alone is sufficient to cause a proportion of the excess females to pair (Conover and Hunt 1984b). Sex skew toward females in western and herring gulls could be due to a differential mortality between males and females; however, such a differential mortality has not been well documented. It is possible that male gulls could be more susceptible to poisoningcontinue

Page 162

from persistent organochlorine contaminants. Male western gulls weigh about 25% more than females on average, and they feed higher up on the food chain (Pierotti 1981). Also, male gulls do not have the ability to excrete lipophilic contaminants by egg-laying. For these reasons, it is expected that male gulls might accumulate greater body burdens of toxicants throughout their lifetimes than females gulls do. It has also been suggested that the skewed sex ratios observed in western gulls in California and in Great Lakes herring gulls might have been caused by estrogenic contaminants, such as DDT, in the environment, either due to a differential male mortality or a feminization of male embryos which resulted in chemical sterilization and a failed recruitment into the breeding population (Fry et al. 1987). This is a plausible hypothesis, but there is no direct evidence to support it. Another hypothesis is that some females might have paired with the wrong sex due to chemically induced masculinization. However, a behavioral study of western gulls in Santa Barbara did not find significant differences in behavior between females mated with other females and those paired with males (Hunt et al. 1984).

In conclusion, there is good evidence to suggest that there has been a fundamental change in the sex ratio of several North American gull populations in the post-DDT era, such that there is an overabundance of females in some breeding colonies. The observations that the colonies most affected were in areas of great DDT contamination and that a few DDT congeners have produced abnormal gonadal development in laboratory studies support the hypothesis that environmental contaminants may have played a role in the sex ratio skew.

Alterations in Behavior

Behavioral abnormalities observed in the wild include aberrant parental behaviors, such as less inclination to sit on eggs or to defend nests, which was observed in herring gulls in Lake Ontario (Fox et al. 1978). Those alterations were sufficient to account for the high incidence of egg loss observed in this population. Because high levels of chemical contamination were found in the gulls, it was suggested that HAAs might be responsible for the behavioral alterations. Laboratory experiments with birds exposed to hormones and some environmental pollutants suggest that this hypothesis is plausible.

For example, male Japanese quail embryos injected with 1 µg estradiol or 500 µg testosterone were completely demasculinized and were behaviorally indistinguishable from females (Adkins 1979; Adkins-Regan 1987). As adults, they failed to mount, crow, or strut. This effect occurred only if treatment was given before d 12 of the 18-d incubation period (Adkins 1979), and it is believed to result from a fundamental change in the neural substrate underlying behavior that confers a differential responsiveness to the activating effects of testosterone in adulthood. Testosterone treatment restores copulation in castrated adult males but is without effect in females. Female quail treated with an antiestrogen beforecontinue

Page 163

hatching can be masculinized and as adults will mount other females (Adkins-Regan 1987).

Sexual differentiation of behavior has been extensively studied in the zebra finch, which exhibits sexual behavioral dimorphism; that is, normally only males sing, dance, and mount. The brain of this finch is sexually dimorphic. The telencephalic nuclei greater vocal center, nucleus robustus archistriatalis, nucleus magnocellularis of the anterior neostriatum, and area X of the lobus paraolfactorius are larger and more extensively connected in males than in females, and are essential for learning and production of the complex vocalizations of this species (Simpson and Vicario 1991a). The administration of estradiol to female zebra finches during the first week after hatching results in a profound organizational masculinization of brain and behavior (Gurney and Konishi 1980; Simpson and Vicario 1991 a; Adkins-Regan et al. 1994), including neural masculinization of telencephalic nuclei that sets up a functional circuit in females similar to that in males, which enables them to learn and produce complex vocalizations (Simpson and Vicario 1991b). When the treated females are stimulated as adults with testosterone, they engage in male behavior of singing and dancing (Adkins-Regan et al. 1994). Males treated with estradiol during the first week after hatching are demasculinized, and they fail to mount as adults (Adkins-Regan et al. 1994). Thus, the pattern of sexual-behavioral differentiation in the zebra finch is quite complex. It is clear from these studies and those involving quail that the process of sexual behavioral differentiation in birds is sensitive to exogenous hormones, and that hormonal manipulation can result in a profound and permanent change in reproductive behavior in both sexes.

In laboratory studies with contaminants, ring doves were fed mixtures of DDE, PCBs, mirex, and photomirex (contaminants found in salmon and gulls of Lake Ontario) during mating. The feed of the low-dose group contained 8 ppm Aroclor 1254, 1.67 ppm DDE, 0.297 ppm mirex, and 0.0954 ppm photomirex; the high-dose diet contained 29.03 ppm Aroclor 1254, 4.61 ppm DDE, 0.897 ppm mirex, and 0.324 ppm photomirex. The doves had reduced or delayed behaviorally induced increases of sex hormones, females failed to respond normally to male courtship behavior, pairs spent less time building their nests, and pairs receiving the greatest dosage spent less time feeding their young (McArthur et al. 1983). There was a marked dose-related decrease in fledgling success, and the breeding cycle was greatly asynchronous. In other studies, adult breeder doves fed PCBs exhibited aberrant incubation (Peakall and Peakall 1973) and courtship (Tori and Peterle 1983). Thus, there is evidence from laboratory studies that environmental contaminants in the Great Lakes region could cause behavioral anomalies in breeding synchrony, nest construction, incubation attentiveness, and parental care at ambient concentrations but these effects are not necessarily attributable to their hormonal activities.break

Page 164

Abnormal Reproductive Morphology

Most studies of abnormal breeding in gull populations were conducted during the mid-1970s. Fifty-seven percent of the male gull embryos collected from Scotch Bonnet Island, Canada, in 1975 and 1976 had testicular feminization (Fox 1992). Eggs at that site were contaminated with dioxins, PCBs, and mirex (Gilman et al. 1979; Fox 1992). One study of a tern colony also showed a high incidence of abnormal tests indicative of estrogenic exposure in ovo (Calambokidis et al. 1985; Nisbet et al. 1996), but the contaminants that could mediate these effects have not been identified. However, the significance of these findings is unclear because there are reports (going back to the 1800s) of apparently abnormal testes in terms as embryos or hatchlings, and this might be a normal condition that disappears as the birds age (Hart 1998).

Some studies have evaluated the reproductive morphology of adult birds. For example, 31 adult female glaucous-winged gulls collected in 1984 from Tacoma, Washington, adjacent to the Commencement Bay, Puget Sound (a PCB-and heavy-metal-contaminated Superfund site), were trapped on their nests and killed for gonadal inspection (Calambokidis et al. 1985). The right oviducts of these gulls were found to be persistent and large. The length of the right oviduct was correlated with the estimated chemical contamination (Fry et al. 1987). However, the significance of these data are unclear, as all birds were successfully incubating clutches. Furthermore, the most severe category of oviduct enlargement was rated as greater than 10 mm long; the literature indicates that a vestigial right oviduct of 9-10 mm is normal in the herring gull (Boss and Witschi 1947). An attempt to correlate alterations in testes in male birds with organochlorine contamination in this gull population was inconclusive (Fry et al. 1987).

Ovotestis formation in male embryos and retention of the right oviduct in female embryos also were observed in experimental studies of gull eggs injected with hormones such as estradiol (Fry and Toone 198 1) and DES (Boss and Witschi 1947) and with environmental contaminants such as methoxychlor and DDT (Fry and Toone 1981). Chicken and quail eggs injected with DDT showed similar effects (Lutz-Ostertag and David 1973). Because concentrations of DDT (2-100 ppm) found in the eggs of wild gulls caused effects consistent with those induced by estradiol and DES (Fry and Toone 1981), it is plausible that DDT or other estrogenic contaminants could be responsible for the effects observed in the wild.

Most morphologic abnormalities in the wild have been found near areas identified as hot spots of organochlorine contamination. Residues of PCBs, TCDD-EQ, and DDT are approximately 10-fold greater than those in other locations (Giesy et al. 1994b). Concentrations of many of the residues are declining in the Great Lakes, but are still among the highest. Green Bay and Saginaw Bay also have hot spots, where concentrations of organohalogen compounds are significantly greater than the Great Lakes as a whole (Giesy et al. 1994b).break

Page 165

Growth and Development Abnormalities

A group of embryonic abnormalities directly related to contaminant exposure in some fish-eating birds has been defined as a specific syndrome. GLEMEDS (Great Lakes embryo mortality, edema, and deformity syndrome) (Gilbertson and Fox 1977; Gilbertson et al. 1991). GLEMEDS involves a consistent pattern of subcutaneous edema, beak malformations, cardiac edema, and skeletal malformations. The expression of this syndrome in bald eagle, cormorant, gull, and tern chicks is correlated with dioxin toxic equivalents of some PCB congeners that are primarily the result of maternal bioaccumulation from eating contaminated fish and resultant deposition of coplanar PCB congeners in the eggs (Gilbertson et al. 1991). Adults from populations in which chicks have GLEMEDS have shown abnormal plasma thyroid hormone concentrations and thyroid morphology (Fox 1992), but no relationship between thyroid hormonal disfunction and GLEMEDS has been found. Several sources of organochlorines have been controlled in response to regulatory action in the early 1970s, and concentrations of DDT and PCBs in fish tissues decreased approximately 20-fold in the late 1970s and early 1980s. However, no change was found between 1985 and 1992 in chinook salmon (Miller 1994), and these contaminants continue to persist in tissues and the environment today.

Eggshell Thinning

During the 1960s and 1970s, when the pesticide DDT and its metabolite DDE were present at higher concentrations than today in North America, it was observed that populations of several bird species declined when individuals were unable to successfully incubate eggs because of abnormally thin eggshells (Cooke 1973). Many of these species, such as the double-crested cormorant, have experienced dramatic population increases since DDT was banned from use in the United States (Ludwig 1984; Weseloh and Ewins 1994). It is now well established that the DDT metabolite, DDE, and to a lesser extent other organochlorines, causes eggshell thinning (for a review, see Cooke 1973). Research into the mechanism of DDE-induced eggshell thinning has been extensive (Gould 1972; Peakall et al. 1975; Cooke et al. 1976; Miller et al. 1976; Eastin and Spaziani 1978; Cooke 1979; Lundholm 1980, 1982, 1984a,b,c, 1985, 1987, 1988, 1993, 1994; Lundholm and Mathson 1983; Lundholm and Bartonek 1991, 1992; Haynes and Murad 1985). Some of the postulated mechanisms include premature termination of shell formation, premature oviposition, effects on the protein matrix of the shell, effects on initiation sites of shell formation, enhancement of shellgrowth inhibitors, decrease in carbonate availability for shell formation, effects on progesterone binding in the shell-gland mucosa, and alteration in calcium metabolism of the shell gland. The current hypothesis regarding the mechanism of DDE-induced eggshell thinning is an inhibition of prostaglandins by the shell-soft

Page 166

gland mucosa (Lundholm and Bartonek 1992). Many of the biochemical end points described above are interrelated, and it has been difficult to determine which end points are the direct targets of DDE and which are merely coinfluenced by its action. However, it does not appear that eggshell thinning is a result of DDE acting as a hormone-receptor agonist or antagonist. The situation is complicated further because sensitivities to DDE-induced eggshell thinning vary among avian species, suggesting that different mechanisms cause eggshell thinning in different species.

Florida Panther

The possibility that exposure to HAAs affects reproduction in the endangered Florida panther has generated considerable interest. The population size is estimated at 30-50 animals of two genetic strains (O'Brien et al. 1990). Most of the panthers exhibit developmental abnormalities (including congenital heart defects) and defects of the reproductive system (cryptorchidism, low sperm density, and sperm defects) (Barone et al. 1994). Reproductive abnormalities had been attributed to genetic inbreeding (Miththapala et al. 1991; Roelke et al. 1993), but a study by Facemire et al. (1995) examining contaminant loading in female panthers has led investigators to conclude that persistent, bioaccumulated contaminants, such as organochlorines, also could contribute to the problems observed. Because the Florida panther is an endangered species, tissue samples for analysis are rare. Three females were examined after death for concentrations of mercury and several bioaccumulated organochlorine compounds. Concentration ranges of various contaminants found in the muscle (µg/g lipid fresh weight) of the animals were 5.45-57.65 µg/g p,p'-DDE; 7.32- 27.06 µg/g Aroclor-1254; <0.0098-2.00 µg/g oxychlordane; and <0.0098-4.82 µg/g trans-nonachlor.

The most frequent developmental abnormality in the Florida panther population is cryptorchidism or testicular nondescent. Cryptorchidism has increased exponentially in male cubs since 1975 (Roelke 1990), and 70% of wild Florida panthers are at least unilaterally cryptorchid, as compared with 20% in the mid-1980s (Roelke et al. 1993). Most of the male panthers exhibiting cryptorchidism have the testis in an inguinal location. A comparative study of free-ranging and captive panthers from Florida, Texas, Colorado, Latin America, and North America indicates that the incidence of cryptorchidism in the Florida panther is more than ten times that found in other populations (Barone et al. 1994). Only two cases of cryptorchidism have been reported in captive populations of North American panthers; the condition has never been reported in any other large felid (Roelke et al. 1993). Cryptorchidism is a heritable trait in some inbred domesticated species (McPhee and Buckley 1934; Claxton and Yeates 1972) that could be a response to an abnormal hormonal environment during embryonic development (Hezmall and Lipshultz 1982; Sharpe and Skakkebaek 1993: Hutson et al. 1994).break

Page 167

Müllerian-inhibiting hormone has been implicated in testicular descent (Hutson et al. 1994), but no data are available on the influence of HAAs on the synthesis of that hormone during gonadal development. In studies with rats, prenatal exposure to the antiandrogenic metabolites of the fungicide vinclozolin caused hypospadias, cleft penis, and suprainguinal (cryptorchid) ectopic testes (Gray et al. 1993; Kelce et al. 1994); exogenous treatment of rats with the androgen dihydrotestosterone stimulates testicular descent (Frey et al. 1983). Treatment of male mice with DES on d 9-16 in utero led to an increase in the incidence of cryptorchidism (McLachlan et al. 1975). Male mammals lacking either adequate concentrations of androgen or androgen receptors exhibit a high incidence of cryptorchidism (Wilson and Foster 1985). Those data indicate that compromised (or altered) androgen receptors-because of receptor abnormalities or because of the presence of an antiandrogen or potent estrogen-preclude normal testicular descent during development in mammals. The elevated concentrations of p,p'-DDE present in the tissue of three female Florida panthers (Facemire et al. 1995) and the knowledge that this metabolite of DDT is known to exhibit antiandrogenic activity (Kelce et al. 1995) suggest a relationship between contaminant exposure and cryptorchidism in Florida panther cubs. However, the tissue of panthers has been shown to be contaminated with a variety of toxic substances, including mercury (Roelke et al. 1991).

Electroejaculation studies of 12 male Florida panthers have shown low sperm density, poor sperm motility, and elevated numbers of sperm defects (Facemire et al. 1995). Sperm density (concentration/milliliter) averaged 4.8 ± 1.4 x 106 sperm in Florida panthers compared with 15.4 ± 4.4 x 106 for males from a Texas population and 22.5 ± 9.2 x 106 in Latin American populations (Barone et al. 1994). The male panthers studied had 24-50% more sperm abnormalities than were found in Texas panthers and significantly smaller testicular volumes (Barone et al. 1994). Furthermore, males had abnormal sex-steroid ratios, exhibiting higher concentrations of estradiol-17ß than testosterone in plasma (Facemire et al. 1995). As yet, there are no exposure data for other populations against which the contaminant concentrations in Florida panthers can be compared.

Although the available evidence suggests that the reproductive anomalies could be the consequence of environmental contaminants, the role of extensive inbreeding in this small population cannot be discounted.

Summary and Conclusions

Several reproductive and developmental disorders have been observed in wildlife and human populations exposed to environmental contaminants, including HAAs. Laboratory studies using male and female rats, mice, and guinea pigs, and female rhesus monkeys have shown that exposure of these animals during development to certain HAAs (e.g., DDT, methoxychlor, PCBs, dioxin, bisphenol A, octylphenol, BBP, DBP, chlordecone, and vinclozolin) can produce structuralcontinue

Page 168

and functional abnormalities of the reproductive tract. Some of these studies, according to the investigators, were conducted using doses at or near levels encountered in the environment, but in most instances the environmental relevance of the dose used is unknown, because of lack of data concerning the level of environmental contamination.

With the exception of PCBs, TCDD, and DDT and its metabolite DDE, there are few human studies on the reproductive and developmental effects of exposure to HAAs. The effects of prenatal exposure to PCBs, DDE, and other contaminants from maternal consumption of contaminated fish or other food products has been studied in several populations in the United States and abroad. Collectively, these studies indicate that prenatal exposure to PCBs can cause lower birth weight and shorter gestation, and have also been correlated with IQ and memory deficits as well as delayed neuromuscular development. Pre- and post-natal exposure to PCBs and PCDFs from accidental contamination of rice oil in Yusho, Japan and Yu-Cheng, Taiwan have resulted in various developmental defects.

Exposure of men to environmental HAAs has been suggested as the cause of worldwide increases in hypospadias, cryptorchidism, testicular cancer, and declines in sperm concentration. Studies examining these trends show considerable variation, both temporally and geographically. The degree to which the results reflect differences in the populations selected for study (fewer men of proven fertility, or men with concerns about their sperm concentration), diagnostic practice, or other methodologic differences, is the subject of continued controversy. With respect to the end point most closely studied, sperm concentration, retrospective analyses of trends over the past half-century remain controversial. When the data from large regions are combined together and analyzed, some data sets indicate a statistically significant trend consistent with declining sperm concentrations. However, aggregation of data over larger geographic regions may not be an appropriate spatial scale for this analysis given significant geographic heterogeneity in genetic and environmental factors. The current data are inadequate to assess the possibility of trends within more appropriately defined small regions. Acquiring data at smaller regional scales is critical to assessing the significant geographic variation in sperm concentration which is the subject of collaborative studies currently being conducted in the United States (funded by NIEHS), Europe (funded by the European Union), and Japan (funded by the Japanese EPA).

Many wildlife studies show associations between reproductive and developmental defects and exposure to environmental contaminants, some of which are HAAs. One of the best established linkages between exposure to an environmental contaminant and reproductive effects in birds has been the correlation of DDT, and its metabolite DDE, with eggshell thinning. Many potential mechanisms for DDE-induced eggshell thinning have been described. The most current hypothesis is that the mechanism involves an inhibition of prostaglandin by the shell-soft

Page 169

gland mucosa. However it does not appear likely that this is a result of DDE acting as a hormone receptor agonist or antagonist.

Reproductive and developmental abnormalities have also been observed in several populations of fish exposed to effluents from sewage treatment plants and paper mills and polluted waters of the Great Lakes. Effects observed include intersexes in trout exposed to sewage treatment plant effluent (STPE); increased egg and fry mortality in Great Lakes trout and salmon; thyroid enlargement in Great Lakes salmon; and changes in plasma sex-steroid concentrations, decreased egg and gonad size, and delayed sexual maturity in whiter suckers exposed to effluents from paper mills along Lake Superior.

Laboratory experiments with specific HAAs found in those effluents and polluted waters have produced effects consistent with these wildlife observations. For example, certain HAAs found in STPEs induce estrogenic responses in male trout. Specifically, ethinylestradiol and alkylphenol ethoxylates have been shown to induce vitellogenin synthesis, a hallmark of estrogen exposure, and to decrease the rate of testicular growth in male fish in tests that duplicate concentrations found in some effluents. Dioxin and structurally related compounds have been shown to induce blue sac disease in trout and reduced growth and survival of salmon. Thyroid enlargement in salmon of the Great Lakes is hypothesized to be caused by exposure to PCBs, which also have been shown to induce goiter formation in laboratory rodents fed PCB-contaminated salmon. Finally. B-sitosterol found in paper-mill effluent has been shown to alter the reproductive physiology of goldfish under experimental conditions.

Laboratory studies are also consistent with some reproductive and developmental abnormalities (e.g., skewed sex ratios, behavioral modifications, and morphologic abnormalities of the gonads) observed among North American gull populations. Specifically, it has been shown that gull eggs injected with DDT at concentrations found in wild gull eggs induce gonadal abnormalities that are similar to those observed in contaminated gulls. Also, doves fed mixtures containing DDE and PCBs exhibit abnormal breeding behavior.

Similarly, defects seen in alligators from Lake Apopka (the site of a chemical spill containing dicofol and DDT) including small penis size and abnormal testes in males and abnormal ovaries in females, are consistent with structural and functional reproductive abnormalities that occur following perinatal exposure of laboratory rodents to estrogenic and antiandrogenic chemicals.

It has also been suggested that cryptorchidism. the most common reproductive anomaly found in male Florida panthers, is the result of exposure to p,p'-DDE. Because testicular descent is in part androgen dependent, and because antiandrogens and potent estrogens have induced cryptorchidism in rats and mice. it is plausible that exposure to contaminants with antiandrogenic or estrogenic properties could be causing the effects in male panthers. However, the Florida panther population is exposed to many other contaminants, including methoxy-break

Page 170

chlor, PCBs, and mercury, and the role of extensive inbreeding in this small population cannot be discounted.

Recommendations

Based on evaluation of reproductive and developmental effects observed in humans, laboratory animals, and wildlife exposed to HAAs, the committee recommends that wildlife and human populations continue to be monitored for adverse developmental and reproductive effects. Specifically, the committee recommends the following:

—Studies of wildlife that exhibit population declines, abnormal sociosexual behavior, or deformities should be designed to investigate those phenomena in light of specific environmental factors, including chemical contamination and environmental degradation.

—Prospective and cross-sectional studies, using common protocols and strict quality control, be conducted in human populations suspected of being affected by HAAs. Serum hormone concentrations, body burdens of HAAs, and sperm concentration in seminal fluid should be measured, especially in relation to any adverse effects, or banked for later exposure assessment. Prospective and cross-sectional studies are particularly needed on cohorts tracked from conception through adulthood on female and male reproductive end points such as sperm concentration, cryptochidism, and hypospadias.

—Regional differences in male reproductive end points such as sperm count and rates of hypospadias, cryptorchidism, and testicular cancer should be examined prospectively to determine whether the differences can be associated with genetic and environmental factors. Such prospective analyses should be accompanied by quantitative sensitivity analyses.

—Free range farm animals should be studied for potential effects of environmental contaminants on fertility.break