Bridging epidemiology and model organisms to increase understanding of endocrine disrupting chemicals and human health effects.

Our understanding of potentially harmful effects of exposure to environmental contaminants comes from a variety of sources including wildlife life studies, in vitro and in vivo toxicology studies, epidemiologic studies, and clinical evidence. The most extensive information comes from animal studies, which are a preferable method for assessing the potential for adverse human effects and for developing strategies for prevention of harmful exposures. Environmental contaminants are not intended for human use, and it is unethical to knowingly expose humans to these chemicals under experimental conditions to assess for harmful effects. Animals have long been used to understand the effects of chemical exposure on human reproduction and development [ 21 ]. One of the first reproductive and developmental studies, from 1919, used rats to study the effects alcohol on fetal development, before it was identified in humans [ 22 , 23 ]. The reliability of experimental animal data for reproductive and developmental health has been well established and presently, there is no example of a chemical agent that has adversely affected human reproduction or development but has not caused the same or similar adverse effects in animal models [ 24 ]. Model organisms studies have many advantages, including ability to control the experimental setting to enhance ability to observe chemical effects, and shorter time from exposure to disease [ 25 ]. Multiple reviews have concluded that there is concordance of developmental and reproductive effects and that humans are as sensitive or more sensitive than the most sensitive animal species [ 23 , 26 ] [ 23 , 24 , 26 - 29 ]. Given that there is general conservation of biologic function across animal species, including humans; animal studies provide important insights into potential human harm [ 30 ]. However, limitations in the design of traditional toxicological studies have limited their utility for studying reproductive or developmental outcomes. These include: use of insensitive strains; exposures that do not cover sensitive periods of development; exposures to single chemicals at high doses rather than the mixture of low doses of chemicals more often seen by the public; and a focus on overt disease endpoints. While animal studies continue to provide important information to predict human harm, epidemiologic and clinical studies provide complimentary sources of information. Human observational studies of environmental chemicals provide the most direct evidence of the relationship between exposure and increased risk of adverse health outcomes, and are often the basis of assessments about harmful effects of EDCs. Studies are typically designed to evaluate whether the change in the risk factor of interest, or the chemical exposure, is related to the change in the incidence or prevalence of the disease of study, while at the same time accounting for factors that may influence that relationship. Model animal studies tend to narrow their methodological scope to increase their ability to detect presence of an effect from chemical exposure, while human observational studies necessarily include the factors which can influence the exposure/effect relationship as they are often impossible to control. These include pre-existing exposure to other potentially influentially chemicals, biological variability (e.g. disease status, age, gender), and typically lower levels of exposure. This makes the interpretation of the results of observational human studies challenging because these factors that influence risk (e.g. age, disease status, co-exposures), can obscure the signal for a particular chemical/outcome relationship. Features of model organisms and epidemiology are illustrated in Table 2 . Since the latter part of the 20 th century, the evolution of our understanding about the role of EDCs in human health has been accompanied by increasing insights into additional factors which influence the relationship between chemical exposure and risk of adverse health effects. These include: enhanced understanding of toxicity pathways; importance of timing of exposure; contribution of multiple chemical exposures; and low dose effects. The growth in the science in each of these areas is influencing study designs in epidemiology and model organisms, the interpretation of the data from each, and the need to integrate knowledge from both the fields. Each of these areas and their contribution to the field is discussed below. Advances in molecular biology, computational biology and information sciences is transforming our ability to interrogate and understand disease etiology on multiple levels, which in turn requires advancing our approaches to assessing potential risks from chemical exposures [ 31 , 32 ]. We are faced with expanding information at the genetic level, such as through genome wide association studies [ 33 ], epigenetic assessments [ 34 ], and metabolomics [ 35 ], providing unprecedented information about the pathways from exposure to overt disease. Concurrently, the technological capacity to test multiple chemicals and their effects on biological pathways rapidly through high throughput vitro assays expands the amount of knowledge available for interpretation [ 36 ]. The advancement of scientific ability to evaluate early perturbations in the disease process increases the opportunity to identify potentially harmful chemicals using early biological markers of disease [ 4 ], and is a feature which can be integrated across the interpretation data from human and model organism studies. Human development is vulnerable to biological disruption particularly when these changes occur during critical windows of development, which typically occur periconceptually, during pregnancy, infancy, childhood, puberty, pregnancy and lactation. Perturbations from chemical exposures can lead to important functional deficits and increased risks of disease and disability in infants, children and across the entire span of human life [ 4 , 30 , 37 ]. The prenatal period has become a particular area of focus of environmental chemical exposure for two primary reasons. First, over the past 60 years it has become clear that the placenta does not protect the fetus from damaging chemicals, [ 38 ], that chemicals can be found in the fetus, and in some cases may be biomagnified. For example, analysis of second trimester amniotic fluid samples from 51 women found the presence of at least one environmental contaminant. [ 39 ], and for mercury, fetal exposures to environmental contaminants may be higher than maternal. [ 40 - 42 ]. Second, while the effects of prenatal exposure on immediate outcomes, such as birth defects or childhood illness has been relatively well known [ 15 ], newer science find that in utero disturbances from the external environment can increase risk of adult disease (in particular the central nervous system, the cardiovascular system, the endocrine system and the immune system), referred to as the Developmental Basis of Adult Disease/Dysfunction [ 2 ]. Numerous animal studies now find that co-exposures to chemicals can enhance risks of adverse health effects compared to risks from exposure to individual chemicals, even if the chemicals act through different mechanisms of action [ 43 - 46 ]. The National Academy of Sciences has recently recommended that chemical exposures that can affect the same common adverse outcome should be considered, and in many cases are additive [ 47 ]. Studies of exposure pathways show the population is in constant contact with environmental chemicals in air, food, water and consumer products [ 48 - 51 ]. Now we have extensive biomonitoring data (measurements of chemicals in biological matrices), which show ubiquitous exposure to many EDCs in the general population found in homes, communities and workplaces [ 8 ]. Further, new analysis of CDC population-based biomonitoring data among pregnant women in the U.S. found virtually all pregnant women have body burdens of a number of EDCs in their body include certain pesticides, perchlorate, Bisphenol-A (BPA), phthalates, perfluorochemicals (PFCs), polychlorinated biphenyls (PCBs) and polybrominated diphenyl ethers (PBDEs) ( Table 3 ). Background health status, as influenced by age, pre-existing disease, genetics, and other factors, can influence the effect of chemical exposure on subsequent health risks. For example, individuals with preexisting immune suppression, such as organ transplant patients, those who are HIV positive, and those at early or late lifestages, might experience disproportionately higher risk of adverse health effects due to chemical exposure compared to more healthy individuals. A vast number of biological processes that are ongoing throughout a lifetime can contribute to disease etiology. For example, neurological deficits and cancer, are relatively prevalent in the population, and have multiple additional factors that contribute to their disease etiology. Thus, we would expect that other ongoing biological processes are already occurring that can contribute independently to disease risk, and EDC exposure would be in addition to these ongoing biological processes [ 52 ]. Exposure to a chemical can influence disease processes through multiple mechanisms of action. Hence, studies that focus on a particular pathway or individual biological event may provide an incomplete picture of the range of effects produced by a particular chemical [ 53 ]. Traditionally, the evaluation of noncancer effects from chemical assumes that there is an exposure below which exposures are anticipated to not contribute to increase risk of disease (referred to as a “threshold”). However, scientific evidence now finds that adverse health effects can occur from exposures at common population levels (e.g. lead and mercury), and threshold cannot be identified in the human population. Further, background chemical exposures, pre-existing vulnerabilities in the population, such as from age and pre-existing disease, can contribute to population background risks such that practically, a threshold will not exist in the population and we expect risks even at low exposures [ 52 ]. Our increasing appreciation of the influence of biological background (i.e. age and disease status) and chemical background to EDC and disease etiology comes from integrating findings from human and model organisms studies. In turn, this base of knowledge guides the interpretation of insights from studies from both disciplines. The following case studies illustrate the importance of each and how scientific concepts aid in interpreting and expanding our knowledge base. This case study of thyroid hormone (TH) perturbations during pregnancy and neurological outcomes illustrates how human studies provided initial evidence of the link between exposure and outcome that was later enhanced by studies in model organisms. THs, thyroxine (T 4 ) and triiodothyrine (T 3 ), are essential to neurological development. Insufficient levels of TH during development can lead to mild to severe cognitive impairment, neurobehavioral disorders, hypomyelination and attendant physical impairments [ 54 , 55 ]. First insights into effects of TH decrements during pregnancy on neurological development comes from studies during the first half of the 20 th century of human populations with severe endemic goiter, caused by insufficient TH, where children were found to be increased risk of mental retardation and other neurocognitive deficits [ 56 ]. While clinically relevant decrements in T 4 during pregnancy were recognized as harmful, later epidemiological studies observed neurobehavioral delays in children born to women with mild to moderately low levels of T 4 , not indicative of maternal hypothyroidism, during the first trimester of pregnancy [ 57 - 59 ] [ 57 , 59 - 61 ]. Other epidemiologic studies found that small deficits in circulating levels of TH are associated with decreased cognitive performance at various times during development and adulthood [ 62 - 68 ]. The fetal thyroid gland does not begin to concentrate iodide until the 12th gestational week [ 69 ] and does not produce significant quantities of TH until about week 20 [ 69 ]. Thus, until about the 20th week the fetus is entirely dependent on maternal THs [ 70 ], and T4 decrements during this time period can have a greater impact of subsequent neurological risks. While human studies have shown the importance of the relationship between maternal TH and neurocognitive development in the child, mechanistic studies of the role of chemical exposures in TH changes has been an important complement to the field. They have enhanced our understanding of the mechanisms involved, the changing TH levels during fetal development, role of multiple chemical exposures, and the importance of different THs to neurodevelopment. The effects of TH on the developing brain are directly related to serum concentrations of T 4 in laboratory animal studies, with animal studies also finding that the developing brain is sensitive to small reductions in serum TH [ 71 - 79 ]. For example, declining serum T 4 has been shown to be associated with linear declines in the number of oligodendrocytes and increases in the number of astrocytes in white matter of rat pups[ 80 ], and effects on the developing hippocampus [ 77 ]. A number of environmental chemicals are capable of disrupting TH levels, often through different mechanisms (illustrated in Table 4 ) [ 81 , 82 ]. For example, perchlorate inhibits the uptake of iodine, resulting in decreased synthesis of TH. PCBs and the pesticide acetochlor activate enzymes in the liver that increase excretion of TH, thus reducing circulating levels of TH [ 83 ]. Crofton et al. evaluated to--rats exposed to a mixture of 18 TDCs (dioxins, dibenzofurans and PCBs), at doses comparable to human exposures, for effects on serum T 4 [ 43 ]. Components of this mixture affect T 4 through two different mechanisms of action: the dioxins, dibenzofurans and dioxin-like PCBs activate one set of liver enzymes, and the non-dioxin-like PCBs activate a separate set of liver enzymes. Crofton et al. found the mixture had a dose-additive effect on T 4 at environmentally-relevant doses and a 2-3 fold greater than dose-additive effect on T 4 at higher doses [ 43 ], demonstrating that exposures to TDCs acting on different mechanisms can have cumulative effects. Clinical sciences first relies on changes in TSH to indicate thyroid related condition [ 84 ]. However, there are situations where serum T 4 and T 3 levels may not be associated with serum TSH as they are under normal conditions, such as in the case of liver disease, or if there are defects in the proteins responsible for transporting T 4 and T 3 in the circulation or in the receptors mediating negative feedback of T 4 on TSH [ 85 , 86 ]. Likewise, there are chemicals, such as PCBs, that can cause a decrease in serum total and free T 4 in animals without causing a concomitant increase in serum TSH [ 87 , 88 ]. Surveillance of the human population finds significant portions are already at risk from additional T 4 decrements. About one third of U.S. women, including pregnant women, have low iodine intake (iodine is required for TH synthesis) [ 89 ]). In addition, between 1999-2002, 7.3% of the U.S. population aged 12 years and older reported that they had thyroid disease or were taking thyroid medication [ 90 ]. In addition, among women receiving TH replacement medication for treatment of hypothyroidism, 14% had TSH and T 4 measurements indicating continued hypothyroidism [ 90 ]. Pregnancy causes an increased demand on the thyroid gland and hypothyroidism is more than twice as common among pregnant women than among non-pregnant women ages 12-49 [ 90 ]. A number of environmental chemicals are capable of disrupting TH levels, including decreasing circulating levels of T 4 . Compensatory mechanisms are likely insufficient to counteract the potential adverse health effects of these T 4 decrements, and thus the fetus is at increased risk for neurological effects from exposure to low levels of TDCs. This conclusion is supported by complimentary findings from the animal and human literature. First, findings in both animals and humans indicate that even small T 4 decrements can have adverse neurodevelopmental consequences in the offspring. Second, fetuses and infants do not have stored TH, and thus have limited capacity to respond to TH decrements during critical stages of development. Third, human evidence find a substantial prevalence of TH and iodine insufficiency in the population of U.S. women, putting them an increased risk of exposure to chemicals that affect TH levels. Fourth, animal studies find that exposure to TDCs can act together to affect TH levels and that human studies show people are currently exposed to multiple TDCs. A number of different chemicals can affect androgen action. Phthalates are a relatively well characterized class of chemicals in this group and provide a case example of how model organisms played a pivotal role in our understanding of the effects on anti-androgenic substances on male reproductive development that lead to complimentary human studies, significantly increasing our overall knowledge in this field. Phthalates are a group of industrial chemicals used as plasticizers, and as solubilizing and stabilizing agents. They are found in numerous products including personal care products, medical devices, various plastics include flooring, polyvinyle chloride type plastics and tubings [ 8 ]. It was discovered in 2000 that phthalates were also ubiquitiously measured in people through reports from the first CDC national biomonitoring report and in publication [ 91 ]. Proper male reproductive tissue development is dependent on a transient peak in testosterone levels during fetal development. Disruption to androgen activity during this critical developmental window can result in a number of male reproductive tract abnormalities, including retained female structures, such as nipples (in animals), and malformations of male reproductive structures, such as hypospadias (an abnormal location of the urethral opening) and cryptorchidism [ 53 ]. Early studies of phthalates focused on adult exposures and primarily effects other than reproductive, though some standard reproductive studies noted effects on male rat testicles [ 92 ]. Differential effects by age were noted in early studies, with early ages being more susceptible, though it was not until later studies when effects specific to the fetal exposures were identified through multigeneration study of DBP in the rat [ 93 ]. Follow up studies in rats identified a critical window of exposure during fetal developmental [ 94 - 98 ]. Additionally, studies identified a sensitive period of exposure during gestational days 15-17 for dibutyl phthalate exposures in the rat, about gestational weeks 8 to 14 in humans [ 99 ]. Phthalates appear to interfere with androgen activity by inhibiting testosterone synthesis. Specifically, phthalates with four to six carbon side chains interfere with the production of testosterone by inhibiting the uptake of cholesterol into the mitochondria by Steroidogenic Acute Regulatory protein (StAR) protein and by inhibiting some, but not all, of the enzymes in the steroidogenic pathway [ 100 ]. These phthalates have been most consistently shown to reduce testosterone production in animal research to date, though other phthalates may also be important [ 101 , 102 ]. Studies in rats have demonstrated that a phthalate-induced decrease in testosterone predictably results in a syndrome of anti-androgenic reproductive abnormalities characterized by: malformations of the epididymis, vas deferens, seminal vesicles, prostate, hypospadias, cryptorchidism and testicular injury; permanent changes (feminization) in the retention of nipples and areolae (sexually dimorphic structures in rodents); and demasculinization of the growth of the perineum, resulting in a reduced anogenital distance [ 98 ]. This constellation of effects has been referred to as the “phthalate syndrome” and, as with other anti-androgenic chemicals, the severity of effects increases with the dose [ 99 ]. The spectrum of effects seen in phthalate syndrome parallels a spectrum of human diseases that has been termed “testicular dysgenesis syndrome” (TDS): infertility, cryptorchidism, hypospadias, and testicular cancer, which has been hypothesized to have a fetal origin associated with male development of the testis [ 103 ]. In 2000, with publication of the first CDC biomonitoring results in the peer reviewed literature, Blount et al. showed that there was widespread exposure in the US population to a number of different phthalates, finding metabolites of DBP, DEP and BzBP in over 95% of the population [ 91 ]. They also found that women of reproductive age had significantly higher levels of metabolites of DBP than other age and gender groups. While phthalates had been measured in the various species since the 1970s, including deep-sea jellyfish [ 104 ], fish caught in various parts of North America [ 105 ], and bovine tissue [ 106 ], (this was the first measurements in humans of wide spread common exposure). Further, the higher exposures among women of reproductive age combined with findings in animals suggested that male reproductive development could be at risk. Given ubiquitous exposure in the human population, identified male reproductive effects, and analytic ability to measure body burdens, human epidemiologic studies could evaluate potential health effects at exposure levels experienced by the general population, which are lower than the doses in the animal studies. Several epidemiological studies have evaluated the relationship between adult exposures and male reproductive effects, in particular semen quality – with some studies finding an association between increased phthalate levels and decreases in semen quality in males [ 107 ], (though not in every study [ 108 ]). Other human studies evaluated prenatal exposures, previously identified by animal studies as a sensitive period for exposure. In particular, the study by Swan et al. found associations between certain prenatal phthalate exposure and male developmental reproductive outcomes, in particular shortened ano-genital distance [ 109 ]. This study was notable for several reasons. First, it used a marker of male reproductive effects reflecting feminization that was common in toxicological studies, but not applied in epidemiologic studies – ano-genital distance. Subsequent human studies have observed ano-genital distance to be sexual dimorphic and reflective of penile length and growth in males [ 110 , 111 ]. This novel application was possible only by cross fertilization across model organisms and human observation studies. Other studies also identified effects from developmental exposures including changes in reproductive hormone levels in male infants exposed to phthalates in breast milk [ 112 ]. Studies find that exposure to a mixture of chemicals with both similar and different mechanisms of action results in dose-additive effects [ 46 , 113 ]. Rats exposed to a mixture of androgen receptor antagonists, vinclozolin, procymidone, and flutamide, at doses that would not have caused hypospadias alone, resulted in over 50% of animals with hypospadias [ 113 ]. A second study analyzed the effects of prenatal exposure to a mixture of seven different anti-androgens (phthalates and pesticides) with differing mechanisms of action (i.e., androgen receptor antagonist or inhibition of androgen synthesis) [ 46 ]. All combinations produced cumulative, dose additive outcomes in the androgen-dependent tissues. Swan found that a summary measure of phthalates, integrating exposure to five phthalate metabolites associated with ano-genital distance, was more predictive of the ano-genital distance than the individual metabolite concentrations [ 114 ]. The prenatal exposure of males to anti-androgenic chemicals illustrates how findings from animal and human studies together enhance our understanding of effects: 1) animal studies identified critical periods of development sensitive to exposures; 2) animal and human studies propose that perturbations early in the pathway of male reproductive development result in a wide array of permanent and irreversible adverse outcomes; and 3) animal studies show that exposure to different chemicals that affect male reproductive development can have a cumulative effect even if through different mechanisms of action.

The complexity of human disease requires that we draw broadly upon available information to understand the potential impacts of EDCs. Our knowledge of the EDC/disease relationship is strengthened and data gaps reduced when we integrate findings from animal and human studies. Our evolving understanding of the relationship between exposure and disease and the factors that can influence risk, including chemical background exposures, disease status and susceptibility can put a portion of the population at risk from small, incremental exposures, as illustrated in Table 5 . Thus, future research efforts should build upon these findings to improve the utility of new research and to contribute to the efforts to understand, and prevent, exposures to harmful chemicals.

Concerning temporal trends in human reproductive health has prompted concern about the role of environmental mediated risk factors. Studies report increases in reproductive diseases and decline in reproductive function since the mid-20 th century among certain locations and populations (primarily in the developed world), previously reviewed and illustrated in Table 1 [ 1 ]. The relatively short time frame over which decline in reproductive health and function has been observed cannot be explained by genetic changes. Environmental chemicals have been identified as one of the potential risk factors that may be contributing to observed changes in reproductive health [ 2 - 4 ]. Over roughly the same period, manufacture and use of both natural and synthetic chemicals has increased by over 20 fold [ 5 ]. In the US, there are approximately 87,000 chemical substances registered for use in commerce as of 2006, and about 3,000 chemicals manufactured or imported in excess of 1 million pounds each. [ 6 ]. The population is exposed to chemicals present in air, water, food and in a variety of consumer and personal care products. In the United States, nationally representative samples of population through the National Health and Nutrition Examination Survey find that every individual has measured levels of multiple environmental chemicals in his/her body [ 7 , 8 ]. There are similar findings from studies in Europe [ 9 ], and populations in the Arctic far from pollution sources [ 10 ]. Thus it is expected that all human populations are exposed to some level of synthetic chemicals. Previous studies have demonstrated environmental chemicals can adversely impact human health. Poisoning incidents with mercury in Japan and polychlorinated biphenyl (PCB) in Taiwan, produced neurological, reproductive, and developmental effects, even when the mother was asymptomatic [ 11 - 13 ]. Occupational exposures to 1,2-dibromo-3-chloropropane (DBCP), produced male infertility [ 14 ]. Science has evolved to study health effects of lower level exposures among the population. For example, reviews of the epidemiological evidence find that contemporary exposures to PCBs are associated with a decrease in semen quality, specifically reduced sperm motility [ 2 ] and prenatal exposures to methylmercury and PCBs at contemporary concentrations can increase risk of neurological deficits in children [ 15 - 18 ]. Environmental contaminants can adversely affect reproductive health through diverse biological mechanisms. Historically, much of the scientific inquiry focused on genotoxic or mutagenic chemicals and cancer effects [ 19 , 20 ]. Since the latter part of the 20 th century there has increased emphasis on an important class of chemicals called endocrine disrupting chemicals (EDCs) that interfere with the production, release, transport, metabolism, binding, action, or elimination of natural hormones in the body and are responsible for maintenance of homeostasis and regulation of developmental processes and their potential effects on reproductive and developmental effects [ 2 ]. Recent reviews and scientific consensus statements find that “the evidence for adverse reproductive outcomes (infertility, cancers, malformations) from exposure to endocrine disrupting chemicals is strong, and there is mounting evidence for effects on other endocrine systems, including thyroid, neuroendocrine, obesity and metabolism, and insulin and glucose homeostasis.” [ 2 ]. However, there are still many gaps in our understanding of the relationship between EDC exposures and reproductive and developmental effects. First, the number of chemicals that have been evaluated for effects on human health remains limited [ 6 ]. Second, many reproductive and developmental health conditions, such as female reproductive effects (e.g. fibroids, endometriosis), male reproductive effects (e.g. cryptchordism, prostate cancer), and childhood diseases (e.g. obesity, cancer), which are likely to be influenced by EDC exposures have not been fully evaluated for their contribution. As new efforts are initiated to address knowledge gaps the links between EDCs and reproductive and development outcomes, understanding how findings from human and model organisms can each contribute to filling the gaps is critical. Findings from human observational studies provide direct evidence of the relationship between exposure to environmental chemicals and subsequent adverse health effects. However, limitations in the epidemiologic data, including scientific and ethical, require the use of findings from mammalian and other model organisms to compliment human epidemiologic evidence. This paper reviews some of the key scientific concepts relevant to integrating information from human epidemiologic and model organisms to understand the relationship between EDC exposure and adverse human health effects. This is followed by two cases studies: first on thyroid disrupting chemicals and second on anti-androgens chemicals, illustrating how the interplay between the findings from two disciplines can enhance our understanding of the relationship between exposure and disease.