At the present level of knowledge… the idea of endocrine disruption is still in the hypothetical realm, and the scientific and regulatory community is still polarized between believers and detractors.
—A. C. VIDAEFF AND L. E. SEVER
In 1992 the
British Medical Journal published an article entitled “Evidence for Decreasing Quality of Semen During the Past 50 Years.”
1 The authors, from the University of Copenhagen, reviewed sixty-one papers from a wide range of countries that reported on semen quality from 1938 to 1991. Although previous studies had raised the issue of a decline in semen quality in selected populations, the
BMJ paper was the first to take a systematic approach by considering all published studies of men without a history of infertility. Their “meta-analysis” (that is, an averaging of the results of sixty-one individual studies) showed that average “sperm count” had declined globally from 113 × 10
6 to 66 × 10
6/milliliter, or by 42 percent, from 1940 to 1990. Ruling out methodological variation and selection bias as possible explanations, the authors judged that their results reflected a “true biological phenomenon.” They went on to make two additional leaps, first, suggesting that the sperm count results might reflect a decline in male fertility—the male’s ability to father a child—and second, linking the decline to an increase in testicular cancer and other male genitourinary abnormalities, including undescended testes and hypospadias, a birth defect in which the opening of the urethra is not at the tip of the penis.
In their opening sentence, the authors referred to increasing concern “about the impact of the environment on public health, including reproductive ability.” They closed with the words, “Whether oestrogens or compounds with oestrogen-like activity or other environmental or endogenous factors damage testicular function remains to be determined.”
Since 1992 the paper has been cited 2,707 times in the scientific literature, an astonishing number for a scientific paper, and has been widely reported in the media. It appeared at a critical moment of mounting concern about the possible effects of environmental pollution on wildlife and human health, including increasing cancer rates, and particularly breast cancer. A particular focus of concern was possible harmful effects of exposure to synthetic chemicals, including pesticides, like DDT, aldrin, and dieldrin; industrial pollutants, like polychlorinated biphenyls (PCBs), dioxins, and heavy metals; and compounds released by the burning of fossil fuels. The Danish paper, with its implication of a drastic decrease in male fertility—conjuring up a possible end to human procreation—appeared to add one more piece of evidence that supported the mounting concern that exposure to chemicals accumulating in the environment from industrial and agricultural practices was having a wide range of effects throughout the ecosystem and was threatening the most basic biological processes.
A number of observations over the preceding decades had galvanized ecologists, reproductive specialists, and epidemiologists.
2 For example, male alligators in Lake Apopka, Florida, had become feminized following exposure to pesticides released into the lake from runoff and effluent from a sewage treatment plant. Trout had disappeared from the Great Lakes, possibly as a result of exposure to dioxin-like pollutants from industrial runoff. Children whose mothers had consumed sport fish from the Great Lakes scored lower on intelligence tests at age seven compared to children whose mothers had not consumed fish from that source. The decrement was associated with prenatal exposure to PCBs, which were found at high levels in sport fish from the Great Lakes. The babies’ in utero exposure, indicated by umbilical cord blood levels, showed a significant association with lower scores on a visual recognition test.
3But by far the most credible and significant finding concerning the potential effects of exposure to chemicals on development did not involve an environmental exposure at all. Rather, it was the result of what is referred to as a “natural experiment.” Starting in the 1940s and continuing through the 1960s, pregnant women with a history of bleeding or of prior miscarriage were given the synthetic estrogen diethylstilbestrol (DES) to prevent miscarriage. It is estimated that during this period three million women were given DES in the United States. In 1971 Arthur Herbst, a gynecologist at Massachusetts General Hospital, and colleagues published a landmark paper in the
New England Journal of Medicine reporting that daughters born to women who had been prescribed DES during pregnancy were at increased risk of developing an extraordinarily rare cancer, clear-cell adenocarcinoma of the vagina, when they reached maturity.
4 This demonstrated that DES taken by the mother during pregnancy could cross the placenta and affect the cells of the vagina of the developing fetus in ways that resulted in the development of cancer decades later. In the language of biologists, DES was a
transplacental carcinogen. The discovery of the effects of DES therapy was to serve as a model for research into the effects of exposure to chemicals in the environment on development.
5Rachel Carson gave powerful expression to concern about the potential impact of exposure to environmental pollution on the ecosystem and on human health in her best seller
Silent Spring (1992). The years following the Second World War saw an enormous expansion of industrial production and modern agriculture, with the introduction of thousands of new synthetic compounds, and, in response to increasing evidence of far-reaching impacts of these developments, a multifaceted environmental movement evolved throughout the 1960s and 1970s.
6 This new environmental consciousness led to the creation of the Environmental Protection Agency in 1970 and the enactment of the Clean Air and Clean Water Acts. Also in 1970, President Richard Nixon declared a “War on Cancer” and expanded the mission and responsibilities of the National Cancer Institute.
Another, rarely cited, factor that was to influence the new environmental awareness dates from the late 1960s, when influential epidemiologists posited that the vast majority of cancer was due to “environmental factors.” The term “environment” was used in the broadest sense to include lifestyle exposures and behaviors, such as smoking, diet, alcohol consumption, infectious agents, and chemical exposures, as opposed to genetics. The public widely, if understandably, misunderstood this axiom to mean mainly exposure to pollution, including trace amounts of chemicals in the external environment.
7 This unfortunate misreading of the concept of the “environment” as it relates to the development of disease has been widespread and persistent and has been the source of much confusion surrounding the role of contaminants in food, air, and water and their contribution to disease.
Throughout the 1970s and 1980s interest in the possible impact of environmental pollution on disease focused overwhelmingly on cancer. This focus was amply fueled by industrial accidents (Times Beach, Missouri; Seveso, Italy; Three Mile Island, Pennsylvania; Chernobyl, Soviet Union); the Love Canal incident in upstate New York; the identification of thousands of toxic waste sites and contaminated wells; reports of drinking water contaminated by low levels of chlorinated compounds and other industrial chemicals, pesticides, and oral contraceptives; and cancer clusters, like those in Toms River, New Jersey, and Woburn, Massachusetts.
This is where things stood in the early 1990s, when the many varied observations in wildlife and a number of apparent trends in human disease were to provide the basis for an ambitious and provocative new theory positing a linkage between a wide range of exposures and an equally wide range of health outcomes. The report concerning declining sperm counts was one example—although a prominent one—of the many possible impacts of environmental exposures that this new theory put on the agenda. The theory, known initially as the “environmental estrogen hypothesis” and later as the “endocrine disruption hypothesis,” was formulated independently in the early 1990s on both sides of the Atlantic.
In the summer of 1991 a small group of scientists was brought together by Theo Colborn, a zoologist and ecologist who had spent years compiling data on the effects of environmental pollutants on wildlife and particularly documenting the effects of pollution of the Great Lakes. Meeting in Racine, Wisconsin, the group produced a consensus document, known as the “Wingspread statement,” which highlighted the many findings underlying the hypothesis as well as its implications for human health.
8 Under the heading “We Are Certain of the Following,” the authors cited the extensive evidence regarding alterations in sexual development in wildlife associated with exposure to chemical pollutants in the environment, including decreased fertility, gross birth deformities, metabolic abnormalities, behavioral abnormalities, and changes in sexual characteristics. Many of these changes had been observed across a wide range of wildlife, including birds, fish, and mammals. The fact that some of these observations in wildlife appeared to be explained by biological mechanisms that had been identified in laboratory studies examining the effects of exposure to specific chemicals seemed to support a causal relationship.
A crucial aspect of the formulation was that the adverse effects of exposure could vary dramatically depending on the timing of exposure: whether this involved the embryo, the fetus, the newborn, or the adult. The effects, Colborn and colleagues noted, were most often manifested in the offspring rather than in the exposed parent. Thus exposure during embryonic development could result in birth defects but could also have delayed effects on reproductive ability, development, or metabolism that appeared only later in life. This was the import of the reported effects of in utero exposure to PCBs from contaminated fish on cognitive ability in children and, even more strikingly, of the DES experience.
To disseminate the endocrine disruption hypothesis to the widest possible audience, Colborn teamed up with a science writer and an environmental scientist to write the best-selling book Our Stolen Future: Are We Threatening Our Fertility, Intelligence, and Survival? A Scientific Detective Story (1996). The book carried a foreword written by then vice president Al Gore. By 1999 sixty-two thousand copies had been sold in the United States, and the book had been translated into sixteen languages.
Responding to the same body of seminal findings, scientists in Europe had independently formulated a version of the endocrine disruption hypothesis. They were aware of discussions in the United States on “estrogens and the environment” as well as the Wingspread statement. In 1993 Richard Sharpe of the University of Edinburgh and Niels Skakkebaek (the senior author on the paper showing declining sperm counts worldwide) wrote a hypothesis article in the
Lancet that proposed that in utero exposure of males to estrogens (from various sources) might underlie common male reproductive disorders.
9 They vividly conjured up a possible role of xenoestrogens—that is, estrogenic compounds in the environment—pointing out that “humans now live in an environment that can be viewed as a virtual
sea of estrogens.” In 2001 Skakkebaek formulated another variant of the endocrine disruption hypothesis, referred to as the “testicular dysgenesis syndrome hypothesis,” which posited that fetal exposure to endocrine disrupting chemicals plays a role in malformations of the male reproductive organs, impaired sperm production, reduced androgen production, and testicular cancer.
10The endocrine disruption hypothesis articulated a new paradigm that was enormously successful in galvanizing the research community, environmental agencies, and the public to take seriously the hither-to neglected but potentially far-reaching and varied effects of environmental pollution on wildlife and human health. There had been a steady, low-level output of research papers on “environmental estrogens” from the late 1960s to the early 1990s, and these increased sharply in the early 1990s. The number of scientific publications on “environmental endocrine disruptors” listed in the National Library of Medicine’s online bibliographic database surged from 4 in 1995 to 427 in 2013.
11 As scientific and regulatory attention to endocrine disruption grew, the topic began to receive regular coverage in the media. According to one observer, a review of print media mentioning endocrine disrupters revealed “the exponential rise of media attention to the issue from the early to mid-1990s.”
12Responses to the hypothesis and its presentation in
Our Stolen Future and in the scientific literature varied greatly. Without doubt, one of the more enthusiastic endorsements of the hypothesis came from Sheldon Krimsky, a scholar who focuses on science and public policy, in his book devoted to the endocrine disruption hypothesis, entitled
Hormonal Chaos (2000). Krimsky called it a “bold and unorthodox hypothesis” that brought together results from different disciplines that no one had hitherto considered. He likened the potential importance of what he termed the “environmental endocrine hypothesis” to the discovery of chemical mutagenesis and the discovery that chlorofluorocarbons were depleting the protective ozone layer in the atmosphere.
13 In fact, Krimsky basically took Colborn and colleagues’ own estimation of the hypothesis at face value. While the scientific community widely accepted the endocrine disruption hypothesis in a programmatic sense as defining an agenda for research, there were those who early on questioned some of the key assumptions behind the hypothesis or noted that the theory had little in the way of factual underpinnings.
14 Others appeared to accept the framework provided by the hypothesis—or rather, the bundle of questions subsumed under it—but never lost sight of the enormous difficulty of elucidating the effects of low-level environmental exposures on normal development.
15 Some scientists simultaneously carried out research studies but, at the same time, were severe critics of methodologically weak studies and one-sided claims.
16The success of the endocrine disruption hypothesis would depend on identifying an important contribution of a common, widespread exposure in the general population—or some substantial segment of the population—to the development of some disease or pathologic condition. For example, if it could be demonstrated that the increase in breast cancer, testicular cancer, male reproductive malformations such as cryptorchidism or hypospadias, or obesity was in a substantial way associated with, and preceded by, exposure to a particular chemical or group of chemicals in drinking water or food, or by other modes of exposure, and if removing or reducing this exposure led to a decrease in these conditions, this would provide solid and important evidence for the theory. After twenty years of research, regulatory attention, and abundant media coverage, it is reasonable to ask how the endocrine disruption hypothesis has fared, what new knowledge it has yielded, and what happens to science when it addresses a question that is both difficult to study and at the same time evokes powerful emotions and preconceptions. This chapter will examine what happened when research focused on specific disease trends and specific exposures to test the endocrine disruption hypothesis and how this research led to a tangled scientific and public controversy that shows no sign of abating. In view of the vast scope of the endocrine disruption hypothesis, I will confine my discussion to effects on human health, since the question of effects on wildlife involves different disciplines and different methods of study. Furthermore, it can be argued that the issues involved in these two aspects of the endocrine disruption hypothesis are so different that their conflation has contributed to the confusion surrounding the question of the effects of “endocrine disrupting chemicals” on humans.
* * *
Hormones are chemical messengers secreted by ductless glands and travel through the bloodstream to affect distant organs. Hormones play a crucial role in orchestrating the body’s growth, maintaining physiologic balance, and sexual functioning and development. Estrogen and testosterone influence the development and the functioning of the reproductive organs; insulin regulates the body’s level of blood sugar; thyroid hormones are important in regulating the metabolic rate. Hormones also orchestrate the development and functioning of many other tissues. The network of glands, hormones they produce, and receptors they bind to are collectively referred to as the endocrine system.
17Once secreted, a hormone must be transported via the bloodstream to the target organ by a carrier protein. Once there it binds to a receptor, and the hormone-receptor unit binds to a specific region of a cell’s DNA to activate particular genes. Different synthetic compounds can influence hormonal activity in a number of ways. Some endocrine disruptors can mimic a natural hormone and bind to the hormone receptor, producing the same response as the natural hormone, or strengthening or weakening its effect. Other compounds can stimulate the production of more hormone receptors, thereby amplifying the effect. Still other compounds can block the action of a hormone simply by occupying the hormone’s site on the receptor.
18A fundamental insight of the endocrine disruption hypothesis was that synthetic compounds such as the synthetic estrogen DES, pesticides including DDT, and industrial chemicals, such as bisphenol-A, could influence the body’s hormonal pathways, even though their chemical structure differed from that of the natural hormones estrogen and testosterone. The latter have a distinctive four-ring structure, whereas the former have a two-ring configuration (
fig. 5.1).
The endocrine disruption hypothesis brought together a number of observations that raised the question whether exposure to chemicals in everyday life could be contributing to a wide variety of diseases and conditions, some of which appeared to be becoming more frequent. The scope of the hypothesis was vast, encompassing thousands of chemicals and a daunting number of different pathways and mechanisms—mostly unknown—by which these chemicals might affect biological development and functioning. Before discussing what has come of major lines of research on this question, I need to make some preliminary points that rarely get attention.
When considering the hormonal effects of different substances, it is crucial to keep in mind that estrogenic (and other hormonal) substances have different potencies, determined by their ability to bind to receptors and thereby elicit cellular responses. The hormonal effect of a substance will depend on its potency and concentration.
What research in this area demonstrates most vividly is how difficult it is to identify a specific causal factor when we are dealing with low- and very low-level environmental exposures in free-living populations. Much of what we know about the effects of exposure to chemicals comes from studies of occupationally exposed workers or from industrial accidents or accidental contamination of food or drinking water. Another source of information about the effects of exposure to chemicals comes from epidemiologic studies of exposures, including smoking, intake of alcoholic beverages, use of oral contraceptives, and postmenopausal hormone therapy, and studies of treatment with therapies like tamoxifen. We have solid knowledge about the effects of these exposures because they involve prolonged, habitual exposure (in the case of smoking and drinking) and relatively high levels of exposure (in the case of occupational exposures). When it comes to lower levels of exposure to contaminants in food, air, and water, the situation is very different.
First, any effects of such exposures may be subtle, transient, or nonexistent. Just because we can measure the presence of a compound in blood or urine using powerful modern technology does not mean that it is having a detectable effect. Second, lifestyle behaviors and exposures, such as smoking, drinking, diet, weight gain, physical activity, and breast-feeding, may overwhelm or modify any effects of environmental exposures. Third, an individual’s genetic makeup is likely to influence his or her ability to metabolize and detoxify these exposures. Fourth, many environmental exposures involve mixtures rather than a single substance, and exposures are likely to change over time as a person’s life circumstances change, making it difficult to obtain a complete record of exposure over the relevant decades. Finally, the types of studies that are done may be capable of picking up a strong effect, but if the effect is subtle or confined to a subgroup with particular vulnerability, few studies will have the ability to pick this up.
Beyond the difficulties inherent in establishing clear-cut effects of such environmental exposures on human health, there is a wider social context in which certain findings are disseminated and attract attention. And the existence of interested parties in the form of a concerned public, environmental advocates, the legal profession, government agencies, and scientists themselves can play a major role in how a scientific question is framed and perceived. When studies are done, they often appear to show an intriguing, novel, and important result, and such a result will inevitably generate excitement among researchers, who are looking for evidence of a relationship. Such results are also of great interest to regulatory agencies and, needless to say, the media. All too often, however, initial findings that appear to furnish evidence of an effect are not borne out when larger and more rigorous studies are carried out.
19 Moreover, it is basic reality that positive findings get more attention and, one could even say, are more psychologically satisfying and convincing than studies that find no effect. Such problems affect many areas of research into factors influencing our health, but when it comes to the endocrine disruption hypothesis, they appear—if one may put it this way—to be “on steroids.”
* * *
From early on there were scientists who voiced skepticism regarding the endocrine disruption hypothesis. One of these was Stephen Safe, a toxicologist at Texas A&M University. In 1995 and again in 2000 he articulated a number of fundamental points that argued against exposure to industrially derived endocrine disrupting chemicals being responsible for a global decrease in sperm counts, decreased male reproductive capacity, or breast cancer in women.
20 Among his key points were the following. First, it is difficult to sort out causality in many of the alleged effects of environmental exposures on wildlife. In any event, the most striking instances of changes observed in wildlife were associated with unusually heavy exposure to pollution. Second, levels of exposure to synthetic estrogens in the environment are extremely low compared to concentrations of naturally occurring endocrine-active compounds in our diet (isoflavones). For example, according to Safe, levels of “estrogen equivalents” from organochlorine pesticides in food are on the order of one-thousandth that found in a standard portion of red wine or beans and closer to one-ten thousandth that found in cabbage. Third, alleged changes in male reproductive capacity are not correlated with differences in exposure to industrial pollution. Safe also pointed out that chemicals in the environment could have antiestrogenic as well as estrogenic activity and, at the same time, androgenic and antiandrogenic activity. This was an early formulation of the idea that we are exposed to very low levels of compounds that are likely to have a variety of endocrine actions, some reinforcing one another and others working in opposite directions.
Other early critics dismissed the evidence from animal studies, which involved much higher levels of exposure than humans would encounter under normal conditions.
21In assessing whether the endocrine disruption hypothesis was strong enough to include in an ambitious study of the effects of early life exposures being planned by several government agencies, Matthew Longnecker of National Institute of Environmental Health Sciences concluded that “overall the evidence supporting endocrine disruption in humans is not sufficiently strong that endocrine disruption studies should be a primary motivating factor in the NCS [National Children’s Study].”
22* * *
In the early 1990s the increasing prominence of the idea that estrogenic substances in the environment might be having important effects on health coincided with intense concern and activism regarding breast cancer. Decades of research on breast cancer indicated that the main factors influencing a woman’s risk—for the majority of women who lack a family history of the disease—were her age and her reproductive history. However, breast cancer advocates pointed to the increasing incidence rates of the disease in preceding decades and lobbied for research into chemical exposures that might have played a role. One of the initial targets of research was the organochlorine pesticide dichlorodiphenyltrichloroethane, or DDT, which had been widely used following World War II but which was banned for agricultural use in the United States in 1972. The most common type of breast cancer is fueled by the body’s own estrogen, and it was reasoned that exposure to estrogenic compounds in the environment could also stimulate breast cancer. Interest in DDT was justified on the basis that it accumulates in fat tissue, bears a structural resemblance to DES, and exerts hormonal effects.
23 Furthermore, the International Agency for Research on Cancer had classified DDT as a “possible human carcinogen.” Tests in animals indicated, however, that DDT and its analogs are much weaker estrogens compared to the body’s natural estrogen—between one thousand- and one million-fold weaker.
24 It was also known that since DDT was banned for most uses in 1972, levels of the compound had decreased markedly in food and in human tissues.
So it is important to realize that the evidence in favor of DDT as a compound that might be contributing to breast cancer was relatively weak. However, a number of small epidemiologic studies had been published by the early 1990s spurring interest in this question. Then, in 1993, an article appeared in the prestigious Journal of the National Cancer Institute that had an enormous impact.
The study made use of stored blood samples from New York University’s Women’s Health Study, a prospective study designed to investigate the role of diet and hormones in the development of cancer, to measure DDT, its main metabolite dichlorodiphenyldichloroethylene (DDE), and PCBs in women diagnosed with breast cancer and women free of the disease.
25 The analysis showed that, after controlling for potential confounding factors, women with the highest blood level of DDE were nearly four times more likely to have developed breast cancer compared to women with the lowest blood level. No association was found with PCB levels. In their discussion the authors referred to the “strong association” of DDE with breast cancer risk and cautioned that, “given the widespread dissemination of organochlorines in the environment, these findings have immediate and far-reaching implications for public health intervention worldwide.” An editorial accompanying the paper referred to it as a “wake-up call for further urgent research,”
26 and the National Cancer Institute and the National Institute of Environmental Health Sciences set up special programs to encourage research on DDT and other organochlorine compounds and breast cancer.
From the vantage point of twenty years, it is easier to see the
JNCI paper in perspective, but at the time it was seized on by some as providing tangible evidence that an environmental exposure might indeed play a role in breast cancer. In reality, the paper had a number of weaknesses that should have tempered the response to it.
27 These included the small number of women who developed breast cancer (only fifty-eight cases); the fact that these cancers were diagnosed shortly after enrollment (and therefore some women may already have had breast cancer when the study began); and, finally, the fact that the dose-response relationship between DDE level and breast cancer was somewhat ambiguous and unstable due to the small numbers involved. All this should have led to a more guarded assessment of the paper.
Spurred by the
JNCI paper, under the auspices of the National Cancer Institute and the National Institute of Environmental Health Sciences, new studies were initiated, and existing datasets were analyzed. As a result, over the following decade, several dozen analyses of the DDT/DDE and breast cancer association were published. These new studies—many of them larger and more rigorous—carried out in different populations showed no hint of an association of DDT exposure and risk of breast cancer. In 2004 a meta-analysis of the studies was published showing that the initial result—the fourfold increase in risk—appeared to be an anomaly and was not borne out by the subsequent studies.
28 In fact, when the studies were compiled and analyzed together, there was no evidence of an increased risk due to increased blood levels of DDT.
29The point is that the DDT–breast cancer hypothesis was never strongly supported. It was pursued because DDT/DDE could be measured in blood and because blood levels were believed to indicate something about long-term exposure in the past. The early results got a lot of attention and reinforced the belief of advocates and some members of the scientific/regulatory community that the environment must be playing a role in breast cancer. In retrospect, however, the DDT–breast cancer story can be seen as an instance of looking under the lamppost for one’s keys, not because one dropped them there but because that is where the light is. At the same time, the search for effects of endocrine disrupting chemicals was to broaden out to encompass a wide range of other chemicals and potential biological effects.
While many studies focused on exposures measured in midlife in relation to breast cancer risk, at the same time a reassessment was taking place regarding environmental exposures and their contribution to breast cancer and other diseases. Several observations served as touchstones for this reassessment. First was the well-established fact that an earlier age at menarche was associated with an increased risk of breast cancer. This is usually explained by the fact that the earlier the onset of menarche and the later the onset of menopause, the longer a woman’s breasts are exposed to the proliferative effects of ovarian hormones (principally estrogen). A striking demographic trend is the decline in the average age at menarche in the United States over the past 150 years from 17 to about 12 years, a trend that has accompanied improvements in living standards and nutrition. This trend and the trend toward having fewer children are themselves correlated with increasing breast cancer rates. A second observation was that, among women exposed to radiation from the atomic bomb dropped on Hiroshima, the greatest increase in breast cancer was seen in those who had been in their teens, whereas those who were adults had a much more modest increase in risk. A third, seminal, finding was the ability of DES therapy given to pregnant women to cause cancer in their daughters when they reached maturity. These observations were part of a growing recognition of the possibility that diseases occurring in adulthood may have important roots in early life. They pointedly suggested that environmental exposures may have their greatest impact during critical periods of development, including the prenatal period and puberty.
30 These insights have led to a new generation of studies following cohorts of girls through menarche to examine both lifestyle and environmental exposures in relation to breast development as well as experimental studies in animal models to understand how the timing of exposure to specific chemicals influences the mammary gland and the development of mammary tumors. Much of this work is being conducted under the auspices of the National Institute of Environmental Health Sciences.
31Since no studies have followed girls from birth or prepuberty for decades to see who developed breast cancer—an undertaking that would require unimaginable resources—it remains very much an open question whether exposures in utero, during puberty, or in adolescence—and, if so, which ones—influence a woman’s risk of breast cancer.
In the present state of knowledge a major source of information concerning the health effects of synthetic estrogens comes from the DES experience. Although this experience is routinely cited by researchers interested in the health effects of chemicals in the environment, its real significance is rarely brought out. DES is a highly potent synthetic estrogen that is structurally similar to, and as strong as, the natural hormone estradiol. DES was administered to pregnant women as a drug to prevent miscarriage in the middle decades of the twentieth century in doses in the micrograms per kilogram of body weight per day range. These doses were escalated during the course of the pregnancy. DES was subsequently shown to cause breast cancer in exposed women as well as congenital malformations of the reproductive tract in their male and female offspring exposed in utero, including adenocarcinoma of the vagina in daughters. This contrasts with exposures to environmental contaminants measured in nanograms per kilogram of body weight—that is, to trace-level exposures that are far weaker in binding to the estrogen receptor. When the difference between these two very different exposure situations is ignored or suppressed, a critical opportunity for understanding is lost.
* * *
Because breast cancer is a disease that, for the most part, occurs in older women, the attempt to link early exposures to the development of the disease is extremely challenging. In contrast, a number of male reproductive disorders occur earlier in life. These include the relatively common birth defects known as “cryptorchidism” and “hypospadias.” The former refers to a condition in which one or both testes have not descended (i.e., remain within the body cavity); the latter refers to the displaced opening of the urethra along the shaft of the penis rather than at the tip. In addition to these conditions, testicular cancer tends to occur in young male adults. Finally, as noted in the opening of this chapter, variations in sperm number and quality had prompted questions about a possible decline in male fertility. The four anomalies of the male reproductive system were sometimes grouped under the label “testicular dysgenesis syndrome,” or TDS.
If, in fact, sperm numbers and quality were undergoing a drastic decline worldwide, this might provide hard evidence that some exposure that had accompanied modern life—and possibly exposure to estrogenic compounds in the environment—was the culprit. Responding to the
BMJ meta-analysis of 1992 showing declining sperm counts and, more generally, to the endocrine disruption hypothesis, researchers attempted to clarify trends in male reproductive anomalies and identify their causative factors. Over the past fifteen years, as more data have become available, the picture has changed dramatically.
32 It turned out that the participation rate in studies of semen quality, relied on in the Copenhagen paper, was quite low (“30% is regarded as good”
33), casting doubt on the representativeness of the findings from these studies. When trends in sperm count were analyzed in studies from centers with higher-quality data, it was found that there was substantial variation in different places and differences in the trend over time. For example, data from Paris indicated that semen quality had deteriorated between 1973 and 1992, and similar evidence came from Ghent and Edinburgh. However, no evidence of a decline was found in data from Toulouse, France, or Finland, or five areas in the United States. Thus, rather than supporting the pattern of a universal decline worldwide from some common baseline, later studies indicated that there was wide variation from place to place, even within the same country.
A fundamental problem with the
BMJ meta-analysis was that the researchers had compared data obtained from one country at one time with data from other countries obtained at other points in time.
34 The data for the early years were heavily weighted by data from New York City, whereas later studies were largely from less developed countries and from Europe.
35 In other words, the researchers were not comparing like with like. Commenting on the paper, Larry Lipshultz, a professor of urology at the Baylor College of Medicine in Houston, said that the comparison performed in the meta-analysis “would be okay if there were no such thing as geographic variation in sperm counts.”
36 In the words of Michael Joffe, a specialist in human fertility at Imperial College London, “The idea of a simultaneous decline, with similar levels across wide swaths of the globe, needs to be abandoned.”
37But the problem with examining trends in sperm number and quality goes much deeper. As Harry Fisch of Columbia University pointed out in a penetrating analysis of the issue, sperm number, semen volume, and sperm morphology vary not only by geographic region and between individuals but also within individuals.
38 Sperm count and quality are influenced by the following factors: time since last ejaculation, scrotal temperature, prolonged sitting, season, smoking, and drug use. These factors were not controlled for and received little attention in discussions of declining sperm count.
Fisch went on to emphasize the biased sampling in the studies included in the
BMJ meta-analysis and itemized six “major weaknesses.” One of these is the comparison of data from different countries at different times, as mentioned above. Fisch demonstrated that when the data from Carlsen and colleagues were reanalyzed accounting for geographic variation, no decline in sperm counts was seen. Furthermore, according to Fisch, thirty-one studies that were published following the Carlsen study attempted to address important methodological problems. Of these newer studies, six showed clear evidence of a decline in sperm counts; sixteen studies (including ten times as many subjects as in the studies showing evidence of a decline) showed either no change or an increase; and the remaining studies showed ambiguous results.
Fisch commented that, “far from being a worldwide and well-proved phenomenon, declines in semen quality are, at best, a highly local phenomenon with an unknown cause and, at worst, a collective artifact arising from the observation of a highly variable physical attribute (sperm counts) with a relatively low-resolution tool (retrospective analysis of non-randomized study populations).” In view of its many flaws, he argued that the
BMJ meta-analysis “warrants its exclusion from any review of data supporting a decline.”
39Attempts to demonstrate links between other aspects of male reproductive capacity and exposure to endocrine disrupting chemicals proved similarly problematic.
Whatever the vagaries of sperm concentration and quality and the many factors (climate, lifestyle, exposure to infectious organisms, environmental exposures, genetics, in addition to those mentioned by Fisch) that may influence them, sperm number and quality are only weakly associated with male fertility.
40 Therefore any impact of a decline in sperm quality on male fertility is likely to be small. Like trends in sperm concentration, trends in fertility also show variation by place. Studies in Europe and the United States actually indicated an overall rising trend in fertility, casting added doubt on the significance of the sperm count data.
41 Rather than drawing any conclusions, Joffe concluded his discussion by emphasizing the urgent need for research that sheds light on behavioral factors that influence fertility.
Regarding testicular cancer, Joffe has argued that the epidemiology of the disease is not consistent with exposure to chemicals in the environment beginning in the post–World War II period. Reliable statistics are available regarding testicular cancer in developed countries. This cancer typically occurs in young men, between the ages of 20 and 45. Its incidence has increased dramatically (between three- and fourfold) in European and certain other populations. In England and Wales, the rise in incidence started in 1920, and in northern Europe around midcentury. Since there is clear evidence that testicular cancer is initiated early in life and possibly in utero, this suggests strongly that early life events at the beginning of the twentieth century or the late nineteenth century are germane to the rise in incidence. This makes the hypothesis that exposure to chemicals in the post–World War II environment rather unconvincing. Joffe concluded that “clearly the factors(s) responsible for the rise in testicular cancer in the 20th century do(es) not explain all of the observed variation in male reproductive system impairment.”
42Unlike the situation regarding testicular cancer, reliable data on trends in hypospadias and cryptorchidism are scarce, making inferences about their causes difficult. A review of data from twenty-nine registries that monitor a total of four million births per year around the world revealed wide intercountry variation in rates of these conditions.
43 There was a suggestion of an increase in hypospadias in more affluent countries, which appeared to end in the mid-1980s. The author pointed out that a number of artifacts might account for the apparent increase, including changes in the definition of hypospadias and changes in physician registration practices. There was no indication of an increase in cryptorchidism since 1970. Thus, in spite of common claims that the rates of these conditions are increasing, more systematic examinations do not support this impression. Furthermore, there is no clear evidence that low-level environmental exposures contribute to these conditions.
44Joffe concluded his assessment with the words, “In summary, a thorough review of the evidence leads to the conclusion that the endocrine disruption hypothesis cannot explain the main features of the rise in testicular cancer or more broadly of TDS [testicular dysgenesis syndrome].”
45 As we shall see, others who have tried to take a broad view, putting the diverse research findings in perspective, have reached similar conclusions.
* * *
After more than two decades of research and thousands of scientific papers devoted to endocrine disruption, the field has become embroiled in a bewildering scientific and political controversy focused on an unlikely culprit—a compound that has been in wide use for over fifty years. Bisphenol A, or BPA, is a carbon-based compound first synthesized by a Russian chemist in 1891. Since the late 1950s it has been widely used in the manufacture of polycarbonate plastic bottles and in the epoxy resins used to line food and beverage containers. The latter use has proved highly effective in preventing illness due to food spoilage. In recent years BPA has found its way into a wide variety of products, including medical equipment, bike helmets, reading glasses, CDs, bullet-proof glass, smart phones, flat-screen televisions, and thermal sales receipt paper.
In the early 1930s the British biochemist Edward Charles Dodds had observed that, owing to its resemblance to the natural estrogen estradiol, BPA had the ability to mimic estrogen, triggering estrogen pathways within the body; however, he determined that BPA was thirty-seven thousand times weaker than estradiol.
46 But it was not until the 1990s that the first scientific papers investigating possible health effects of BPA exposure started appearing, and the chemical became a major focus of scientific research only in the late 1990s. Another ten years elapsed before BPA seized the attention of the public owing to reports that it could leach out of plastic bottles, food containers, and “sippy cups” used by infants, leading to calls for regulating and banning BPA in consumer products.
47One paper in particular had sparked scientific interest in BPA as a chemical that could disrupt normal development. In 1997 Susan C. Nagel and colleagues at the University of Missouri reported that exposing fetal mice to a low dose of BPA resulted in estrogenic activity.
48 Specifically, prostate weight was increased in male mice at six months of age following exposure of the pregnant dams to BPA during gestation. The paper was from the laboratory of Frederick vom Saal, a biologist who had done important work on fetal exposure to hormones and who was to become a leading figure asserting the dangers of BPA to human health. The Nagel paper was to stimulate a cascade of papers from vom Saal and other groups, which appeared to show evidence of a wide variety of adverse effects from exposure to low doses of BPA that had been assumed to be safe. From the 1960s until the late 1990s only a handful of scientific papers had appeared each year on the health effects of BPA. In 1997 the number increased to 10; in 2005, there were 65 publications; and in 2013, 199.
These studies have involved many different test systems and mechanisms of hormonal action at different stages of development. In addition to experimental studies in animals and cell culture, there have been many epidemiologic studies examining associations of BPA levels (usually measured in urine) and health outcomes including diabetes, obesity, breast cancer, heart disease, and behavioral abnormalities.
49Since the late 1990s there has been a pointed controversy concerning the interpretation of the results of both the experimental and the epidemiological studies on BPA. As research findings have accumulated, rather than resolving key differences, the controversy has only intensified, and the opposing positions have become more entrenched. This has led to the existence of two camps with drastically divergent interpretations of the same body of evidence. For the purposes of identifying the two groups, I will refer to them as the “proponents” or “advocates” of BPA as an endocrine disruptor and “opponents” or “critics” of the hypothesis, respectively. Some of the leading figures among the proponents are Frederick vom Saal and his group at the University of Missouri; Ana Soto, Carlos Sonnenschein, and Laura Vandenberg of Tufts University; Thomas Zoeller of the University of Massachusetts at Amherst; Niels Skakkebaek of Copenhagen’s Royal Hospital; and Linda Birnbaum, the director of the National Institute of Environmental Health Sciences. Scientists in the opposition include Stephen Safe of Texas A&M University, Daniel Doerge of the FDA’s National Center for Toxicological Research, Justin Teeguarden of Pacific Northwest National Laboratory, Rochelle Tyl of Research Triangle Institute, and Richard Sharpe of the University of Edinburgh.
The two camps differ on technical issues, the overall interpretation of the available evidence, the most fundamental principles of toxicology, and, finally, philosophy regarding regulation and the basis for regulatory action (i.e., invocation of the precautionary principle). It would be hard to imagine a more complete disjunction between two groups examining the same body of evidence.
Key methodological issues dividing the two camps include (1) the level of BPA exposure at which effects are observed and the pattern of effects at different exposure levels and whether there is a threshold below which no effects are observed; (2) the importance of different routes of exposure (dermal and inhalation versus ingestion); (3) metabolism and excretion of BPA following exposure by different routes; (4) the relevance of animal models to the human exposure situation; and (5) the validity of measuring BPA in blood and the issue of contamination. Underlying these specific points of disagreement are the questions of whether humans are exposed to truly significant levels of BPA and whether any biological effects can be reliably attributed to this exposure.
A fundamental principle of toxicology is that “the dose makes the poison.” This principle lies behind the “dose-response relationship” that is seen for most toxins, that is, the higher the dose to which humans or test animals are exposed, the higher likelihood of observing an effect. Examples of the dose-response are seen in the effects of smoking cigarettes (the greater the number of cigarettes smoked per day, the greater the risk of lung cancer and other diseases caused by smoking), exposure to ionizing radiation, and exposure to lead. The assumption of a dose-response underlies environmental regulation. Advocates of endocrine disruption argue that this model does not apply to the action of hormones, and they posit the existence of what they call a “nonmonotonic dose-response” model or “U-shaped dose-response,” meaning that effects can occur at low levels of exposure, whereas there may be no observable effect or a weaker effect at an intermediate or higher level of exposure. The advocates argue that there are plausible mechanisms that can explain this unorthodox dose-response.
50 If this contention were correct, it would require a major rethinking of toxicology.
51Regarding the significance of BPA concentrations measured in human populations, the advocates argue that the levels of the compound measured in urine and in blood in certain studies represent significant exposure, and that these levels correspond to exposure levels at which detrimental effects are observed in experimental animal studies.
Although it is generally agreed that greater than 90 percent of BPA exposure comes from ingesting food that has absorbed the chemical from packaging, the advocates argue that other “routes of exposure” may be important, including inhalation of air or dust containing BPA, dermal absorption of BPA from thermal paper receipts, and bathing in contaminated water. Furthermore, they argue that these alternative routes of exposure are likely to result in significant levels of the active compound since they bypass the liver, where most BPA is deactivated. This would mean that the burden of BPA in the body has been underestimated owing to the focus on ingestion as the main route of exposure. Advocates also emphasize that BPA in the mother’s circulation is transferred to the developing fetus through the placenta, thereby potentially posing a serious danger.
In their position papers, advocates of endocrine disruption make what appear to be plausible and cogent arguments. However, reading through their papers claiming significant adverse effects of BPA exposure, one is struck by a number of features of their style of argumentation. First, there is little concern for, or attention to, overall quality or methodology of the different studies for its own sake. Rather, methodology seems to become an issue only when the proponents are defending the results of studies that are in line with their hypothesis. Thus little or no attention is paid to experimental design, adequate sample size, appropriateness of the experimental system to real-world exposure, or replication of results.
Second, not surprisingly, there is a tendency to cite work by the authors themselves and like-minded scientists from other groups. It is striking, however, that there is virtually no acknowledgment of any results that go counter to their hypothesis. It is highly unusual in science that all findings line up perfectly in support of one’s hypothesis, and it is essential to attempt to understand the reasons for apparent contradictions, in order to make progress.
Third, there is a narrowness of focus, by which I mean that virtually no recognition is given to the fact that in comparison to low-level exposures to environmental contaminants, there are other exposures that are likely to greatly outweigh them, namely, obesity, the body’s natural hormones, and phytoestrogens in the diet, which are many orders of magnitude more potent than what is being investigated. These factors, as well as others, such as maternal smoking and use of certain medications, would be likely to dwarf the effects of what is being studied. However, there is no attempt to put exposure to endocrine disruptors in a broader context.
Fourth, the proponents’ publications tend to refer to the increasing frequency of various diseases or conditions and imply that endocrine disruption is playing a role. In fact, as we have seen, in spite of decades of research there is no firm evidence to support a role of endocrine disrupting chemicals in these diseases.
Essentially, it appears that the proponents are expressing their
belief in the endocrine disruption hypothesis very much in the vein of the Wingspread statement of 1991. But all they have to point to are questionable results that have not been replicated, and they studiously avoid acknowledging that there is no firm or consistent evidence of an effect. Finally, there is a tendency to favor certain types of research (particularly academic research funded by the National Institute of Environmental Health Sciences) and to impugn the motives of those whose results they disagree with, implying that research funded by industry or government agencies such as the EPA and the FDA is automatically flawed.
52 Once one becomes aware of these stylistic features, one begins to suspect that the proponents of endocrine disruption have a predetermined goal, and that, rather than judging studies on their merits and taking only the best evidence into account, they are wedded to results that support their position.
If one’s investment in a hypothesis is so powerful that one loses the ability to assess studies dispassionately and objectively—independent of whether the results conform to one’s hypothesis—it is easy to be misled. There are many points at which bias can creep into the design of an experiment or its interpretation. For example, the earliest experiment to raise the question of “low-dose” effects of BPA by Nagel and vom Saal in 1997 involved giving BPA in drinking water to pregnant mice during the prenatal and immediately postnatal periods. To examine effects on prostate weight in male offspring, they randomly selected one adult male from each of seven litters. Studies attempting to replicate these findings have shown that there is substantial variability in prostate weight within a single litter. In one such study, when all members of the litter were included in the analysis, no association was found between BPA exposure and prostate weight.
53 In fact, the study by Nagel and colleagues has never been successfully repeated; however, this does not deter the advocates from citing it as important evidence of endocrine disruption. This is just one example that drives home a point that is fundamental to research studies but virtually never gets attention when results are presented to the public. That is, what data were collected and how they were collected can determine the result obtained. We saw something similar in the claims of a drastic decline in sperm counts, which was most likely due to a biased selection of data included in the meta-analysis published in
BMJ in 1992. Many similar pitfalls affecting studies of endocrine disruption have been pointed out in the literature.
54As basic research studies focused on particular test systems and mechanisms and studies attempting to assess BPA exposure and its effects in human populations have piled up, some very high-quality studies have appeared, raising damaging questions about BPA as a “model endocrine disruptor.” These well-designed studies provide data on BPA exposure, metabolism, and excretion in rats and monkeys in utero and postnatally, and on heavy BPA exposure in humans. And they indicate that consumer products contain little BPA, and leaching from the container or packaging into the food is minimal.
55 Moreover, although most U.S. residents are exposed, actual exposures are very low—more than 99 percent of ingested BPA is efficiently metabolized and excreted. And this is true even in newborns.
56 Crucially, these studies measured both free and bound BPA—only the free compound can have biological effects.
In a real-world experiment to determine the impact of heavy BPA exposure on blood and urine levels, Justin Teeguarden of Pacific Northwest National Laboratory and colleagues had twenty volunteers eat meals rich in canned foods and analyzed BPA in blood and urine samples collected over a twenty-four-hour period.
57 The experimental diet was designed to put the subjects in the ninety-fifth percentile for BPA exposure in the United States. In spite of their high exposure, free BPA was not detectable in any of 320 blood samples using a highly sensitive method, indicating that the compound was rapidly absorbed and rapidly excreted. This confirmed how low actual human exposures are, even under high-dose conditions.
To make sense of the contradictory findings and conflicting interpretations bedeviling the field, Teeguarden and colleagues carried out a reevaluation of published studies that had reported serum BPA concentrations to determine whether they could plausibly be causing estrogen-mediated effects.
58 Their analysis included data from ninety-three published studies of more than thirty thousand individuals in nineteen countries across all life stages. The authors used four different methods to calculate serum BPA concentrations. These methods took into account what is known about the correlation between urinary and blood BPA levels, different routes of exposure, and levels predicted by a validated human pharmacokinetic model (that modeled how BPA is metabolized and excreted). The different methods gave a remarkably consistent picture of the range of active and inactive BPA serum levels. In the authors’ words, “Typical serum BPA concentrations are orders of magnitude lower than levels measurable by modern analytical methods and below concentrations required to occupy more than 0.0009%” of major estrogen binding sites. They concluded: “Our results show limited or no potential for estrogenicity in human, and question reports of measurable BPA in human serum.”
Given this impressive consistency, how was one to explain reports in the literature of blood BPA concentrations three orders of magnitude higher than those they had obtained? After ruling out a number of possible explanations, Teeguarden and colleagues concluded that the high serum BPA concentrations reported in some studies could plausibly be explained by their having been conducted in a hospital or clinic setting, where contamination of the blood samples by BPA in medical devices, including intravenous lines, could have resulted in higher BPA concentrations than what is consistently measured in the general population.
59This is the kind of analysis that examines all available data in order to identify the reasons behind the conflicting results cited by different groups. If the analysis by Teeguarden and colleagues is correct, this would indicate that the study results that the endocrine disruption proponents take as the cornerstone of their case (that human exposures to BPA and their effects are being missed) are actually outliers, that is, results that are anomalous, probably due to poor methodology, and therefore of questionable relevance to public health.
The proponents have come back stronger than ever, however, attacking the results and the arguments of their opponents and defending their own studies against all criticisms.
60 They claim that circulating levels of unconjugated BPA are higher than what is predicted by models that assume that the only relevant route of exposure is oral. They flatly reject the criticism that the high levels of BPA in blood in their favored studies are due to sample contamination. And they stress that effects of BPA exposure on the fetus and in early life have been demonstrated and are explained by the much less developed capacity of the fetus and the developing animal to metabolize the chemical.
Finally, we should note that the advocates’ position is at odds with the conclusions of many national health agencies, including the U.S. FDA, the EPA, Health Canada, the European Food Safety Authority, Food Standards Australia New Zealand, and the German Federal Institute for Risk Assessment, which, after thoroughgoing assessments of the evidence, have found BPA to be safe at levels to which the general population is exposed.
* * *
In the summer of 2013 the long-standing scientific debate over endocrine disruption erupted in a new forum when a European Commission (EC) proposal to regulate endocrine disruptors was leaked. The commission’s framework for its proposed regulation was based on a document drawn up under the auspices of the United Nations Development Program and the World Health Organization entitled
Endocrine Disrupting Chemicals 2012: The State of the Science.
61 Citing the report, the European Commission document called for “appropriate policy action on the basis of the precautionary principle” to regulate endocrine disruptors as a distinct category of chemicals, even though they are already covered by existing laws concerning toxic substances. News of the EC proposal provoked an immediate and scathing response in the form of an open letter to the commission’s chief scientific adviser, Anne Glover, by Daniel Dietrich of the University of Konstanz in Germany and signed by eighteen toxicology journal editors. The letter, which was published as an editorial in fourteen toxicology journals, charged the commission with planning to regulate “so-called endocrine disrupting chemicals” within a framework that was “based on virtually complete ignorance of all well-established and taught principles of pharmacology and toxicology.”
62 The authors questioned why “endocrine disrupting chemicals” should be treated as a distinct category and judged by different standards from those routinely applied to any chemical. They stressed the need to take into account real-world exposure and to accept the principle of a threshold below which adverse effects are not observed. They criticized the EC framework for failing to distinguish between transient perturbations and truly adverse effects and stressed the need to base its judgments on data from human studies and whole animal experiments, rather than on data from artificial test systems (in vitro tests). The editorial further charged that the EC framework inexplicably ignored the conclusions of its own expert authority, the European Food Safety Authority, as well as those of other bodies and societies that had determined that BPA was not a hazard at levels to which people are normally exposed. Finally, they stressed the harm to science and society that will be caused by allowing the complex process of evaluating the science to be influenced by political pressures. In the following months nearly one hundred scientists signed the Dietrich editorial.
Proponents of endocrine disruption responded with a defense of the EC framework, which was also published in multiple journals. They rejected each of Dietrich and colleagues’ criticisms and accused them of ignoring evidence that disruption of the endocrine system during development can lead to irreversible effects.
63 The endocrinologist Andrea Gore charged that the Dietrich editorial “seeks to foment doubt on the relevance of EDCs” and reflected “unrelenting pressure from individuals and corporations with stakes in the status quo to keep doubt alive.” She characterized the events over the summer and early fall as “one of the most remarkable experiences in my career.” While acknowledging that it was vital that the two communities work together on this issue, she admitted that, “It’s hard to imagine these two groups sitting down and having a pleasant conversation.”
64Owing to the unfortunate experience of pregnant women who were given the synthetic estrogen DES in the middle of the last century, we know something about the long-term effects of this compound at high doses. It took decades of following these women and their children to document the health effects from that exposure. Let’s take a moment to make an obvious comparison, which, tellingly, is virtually never made by proponents of endocrine disruption. The average exposure of Americans to BPA is about 0.02 micrograms per kilogram of body weight.
65 DES, on the other hand, was given to pregnant women at doses as high as 2,000 micrograms per kilogram of body weight
66—a difference of about five orders of magnitude. But, in addition, BPA has approximately ten-thousand-fold lower estrogenic potency compared to DES. So if we take into account both dose and potency, we are talking about a difference of nine orders of magnitude!
At the heart of this heated debate are not only starkly divergent readings of the evidence, as we have seen, but also two very different philosophies. Invocation of the precautionary principle by proponents of the endocrine disruption hypothesis sounds eminently reasonable. It simply argues that if the potential harmful effects of an action or an exposure are not fully known, one should act in such a way as to minimize any possible adverse effects. The problem is that, when applied to the question of endocrine disruption, this seemingly reasonable approach ignores the extensive scientific evidence that has accumulated over the past two decades. And, as we have seen, in order to support their contention of a hazard, proponents are forced to ignore the high-quality studies and defend results that have never been successfully replicated. An important shortcoming of the proponents’ position is that it fails to distinguish between transient effects (due to the body’s compensatory response to endocrine perturbations) and irreversible effects.
67 Ignoring such crucial distinctions, the proponents rely on “hazard identification,” placing emphasis on the toxicity of synthetic chemicals, while ignoring “weight-of-the-evidence” assessments that take into account actual exposure in real-world situations and what is known from experiments in whole animals. Rather than “hazard identification,” which ignores some of the most crucial information available (i.e., human exposure data), the critics argue that what is needed is a “risk assessment” approach that takes into account all relevant scientific information available.
68Doing justice to the complexity of the science on endocrine disruption and the difficulty of establishing clear-cut effects is in itself a daunting task. However, once a scientific question gets catapulted into the public arena, the science and its interpretation no longer occupy a protected realm, where the standards of scientific discourse are supposed to hold sway. Rather, the science becomes hopelessly overlaid with personal, political, and ideological associations. The French sociologist of science Bruno Latour has referred to these composite objects as hybrids. These overtones and associations can become as important as the science, and many will react to the narrative of the science they find most convincing and congenial based on these overtones and associations.
* * *
It is important to realize that this intense controversy, which pits two camps with dramatically divergent interpretations of the same body of evidence, can exist because of the low levels of exposure; the complexity of biology, with the possibility of different effects at different stages of development; the large number of factors and exposures that can distort normal development; the differences between animal models and humans, making extrapolation from the former to the latter perilous; and the difficulty of establishing causal associations between a low-level exposure and health effects in humans. Given the depth and intensity of the controversy, its political overtones, and its far-reaching policy implications, it is difficult to see how the current impasse will get resolved any time soon.
Rather than looking for an imminent resolution of the controversy, it is more rewarding to step back from the narrow focus on “endocrine disruption” and take a wider view of the issue. It comes as a breath of fresh air when one moves from individual studies and the literature of debate to attempts to reflect critically on the experience of the past two to three decades and ask what has been learned. A number of academic researchers have been prompted by the endocrine disruption experience to write overviews that attempt to put the issue in perspective. These commentaries may bring us as close to the truth of the matter as we can get at present. Their authors do not rule out the possibility that major discoveries will be made in the future, but, for the most part, they are confident enough about the high-quality work that has been done that they are able to say that, for now, maybe enough attention has been devoted to BPA and to endocrine disruption in general and maybe it is time to look at new hypotheses. The impulse behind these overviews is to learn from the experience of the past twenty years in order to turn the page.
The critical reflections of these scientists on the endocrine disruption saga help make sense of the almost impenetrable confusion surrounding the issue. They explain how science can get hijacked and diverted by nonscientific agendas and how an issue like BPA can take on a life of its own, siphoning off enormous amounts of scarce research funds and regulatory attention.
69 By critically examining the claims and the findings regarding endocrine disruption and placing them in a broader context, these overviews take stock of what has been learned and raise the question of where one should look in the future for more productive hypotheses regarding exposures that affect human health and development. In addition to published reflections by figures involved in research on endocrine disruption, I also refer below to interviews that I conducted with a number of these figures.
Allen Wilcox, who is a senior scientist at the National Institute of Environmental Health Sciences (NIEHS), has a valuable perspective on the endocrine disruption hypothesis. His career has been devoted to the study of environmental exposures on human reproduction. In the early 1990s, motivated by the finding that DES was a transplacental carcinogen, Wilcox and colleagues formulated a number of other hypotheses about effects of DES on human health based on findings in animals and basic science.
70 He and his colleagues had what they considered the perfect research design to test these hypotheses, namely, follow-up of the randomized DES clinical trial done at the University of Chicago in the early 1950s. (Wilcox commented to me that “random allocation of the exposure of interest is a luxury few epidemiologists ever get.”) None of their hypotheses was borne out by the data, “and we published six (!) negative papers.” He acknowledged that there are of course many differences between DES and environmental estrogens, and “there may be important ED [endocrine disruption] effects yet to be discovered—but I have to admit that my own negative studies made me a bit skeptical about ED effects reported in less rigorous studies.”
71Earlier I cited Michael Joffe’s views on the possibility that exposure to low-level estrogenic chemicals in the environment could be playing a role in male reproductive abnormalities. At the end of one of his critical reviews, Joffe commented on the impact of attention to the endocrine disruption hypothesis on the ability to conduct original science in his area:
One hypothesis to explain the deterioration in the aspects of male reproductive health grouped together as TDS [testicular dysgenesis syndrome] is that it is due to some form of endocrine disruption or modulation. This has been highly influential since the early 1990s, to the extent that it has not only driven out discussion of other hypotheses, but has also dominated the debate about the male reproductive system—major research programs on endocrine disruption were initiated in the USA and Europe, and in most countries it was only possible to carry out research on the epidemiology of male reproduction if the project could be presented as a test of the endocrine disruption hypothesis.
72Richard Sharpe, who specializes in male reproductive function, occupies an unusual position in the endocrine disruption controversy. Since the early 1990s he has taken seriously the possibility that significant exposure to chemicals in the environment could play a role in human disease, but at the same time he has been uncompromising as a scientist and has been a severe critic of the poor quality of much on the work on endocrine disruption. In a sense he has taken both sides in the controversy. For these reasons, his perspective on the science and on the factors that led to the inflation of the issue is especially worth hearing. In an e-mail he wrote:
In my opinion the big problem in this area is that it has become an “industry,” sucking in huge amounts of funding in ways that have become self-serving and self-supporting. So many people’s labs, careers and funding are dependent on the “threat” from EDCs being real and important, that they do not look in any other direction for explanations. In science that is a danger that I always teach students/young scientists about, because it can cost you your career if it turns out you are wrong (and most of the time we are). To borrow a term used all too often by the EDC researchers, there is a huge influence of “vested interest,” not from industry in this case (although that is omnipresent), but from the EDC research community.
73Based on his own experience, he can see how the pressures on scientists (publications, publicity, getting grants, career advancement) can distort the path they take. He wrote that “in retrospect I consider that circumstances helped me because I ended up disproving my own hypothesis/ideas (on the potential impact of environmental oestrogens on male reproductive disorders) early on in the ED saga.” And he went on to make a crucial distinction: “plus I was lucky that the question that drove me was ‘what causes these disorders?’ not ‘how do EDCs cause these disorders?’ Such a simple difference, but it takes your thought processes in a very different direction.”
74Perhaps the strongest and most penetrating comments on the mechanisms that allowed anomalous research results on BPA to obscure more rigorous research findings come from Daniel Doerge of the Food and Drug Administration’s National Center for Toxicological Research. In response to a question from me, Doerge wrote:
As long as investigators see that others can successfully use a strategy that is fundamentally uncritical (i.e., by only pursuing evidence of toxicity and disregarding evidence against their chosen hypothesis, disregarding human exposure, inadequate experimental design), why shouldn’t they pile on the bandwagon? I think the BPA episode is instructive in how a well-disciplined, but unprincipled, group of academic investigators spun a distorted version of reality. In this alternate reality, the tools of science are given over to the political realm, where distorted concepts and inflammatory rhetoric become weapons to destroy individuals espousing opposing conclusions. When such a self-interested group gains access to the levers of a large national funding agency, the chaos can continue for an extended time. This model is obviously applicable to any other chemical entity that a fearful pubic can be driven to focus on.
75If BPA has been a costly distraction/diversion that has consumed hundreds of millions of research dollars and massive regulatory attention and generated widespread public concern and confusion, what targets should researchers interested in environmental contributions (in the broadest possible sense) to disease focus on?
Doerge emphasized that, unlike BPA, which is a very weak estrogen and is rapidly metabolized, more persistent toxins that entail long-lasting exposure and accumulation in the body are more plausible targets of research. He pointed out that there are clearly dangerous chemicals in the environment, like inorganic arsenic, that should receive more attention. In addition, he feels that TCDD (dioxin) is “problematic” because of its highly persistent nature in the body and its toxic potency in many experimental animals. Doerge has devoted much of his career to the toxicity and carcinogenicity of compounds in the diet (this is what motivated him to turn his attention to BPA, whose principal source is the diet). He thinks that acrylamide—a compound that is formed when starchy foods such as potato chips and French fries are heated higher than 248° Fahrenheit—is “interesting” since animal studies provide strong evidence of carcinogenicity. And he thinks that it is probably a human carcinogen. He notes, however, that because exposure to acrylamide is so widespread and is essentially unavoidable in the general population (factors that also make it difficult to study epidemiologically), there is probably little that can be done to reduce the risk at the population level.
76When I asked Richard Sharpe what he thought are more worthwhile research questions in the field of reproductive disorders, he said he would answer in a “roundabout way.” He told me about John Sumpter, a professor of aquatic ecotoxicology in Britain, who did seminal work on estrogens in sewage effluent and the induction of intersex in fish. Sumpter, Sharpe told me, gives a superb talk in which he tells of his research starting in the early 1990s devoted to alkylphenols, which had been identified as being weakly estrogenic and which were clearly released into river water. It was only after ten years of work and large amounts of money that he and his colleagues finally “nail[ed] the culprit, and of course it turned out to be ethinyl estradiol/other potent synthetic estrogens” (i.e., from oral contraceptives). Sharpe said that Sumpter would end his talk by saying that in light of this discovery, “You would think that we would have indulged in some intelligent thinking, but sadly we appear not to have done so across the ED world.”
77Sharpe continued:
With this example in mind, I often in talks pose the question, “if we were to start from scratch in looking for what sorts of exogenous factors might impact reproductive development via hormonal perturbations, where would be the most logical place to start looking? Should we start with very weakly active compounds that are present in the environment (i.e., in low/very low levels) and which are readily metabolized or should we start with compounds that are designed to be bioactive and resistant to metabolism/breakdown and which we are exposed to intentionally in high amounts (I’m talking about pharmaceuticals or, less likely perhaps, certain components of diet)?” As you know, the answer is that only the unlikely option has been investigated in the main, even though it is illogical.
Sharpe told me that he is now taking the logical route and for the past three years has been examining the effects of pharmaceuticals taken by pregnant women on the fetus and the developing child. This work is producing evidence of real effects including cryptorchidism and neurobehavioral effects. He doesn’t claim that these effects explain all reproductive disorders, but “what it does shout out is that we have been looking in very much the wrong direction, the lack of intelligent thinking once again.”
The endocrine disruption story and particularly the BPA saga are cautionary tales of what happens when science is hijacked by people who use the power and the prestige of science to scare the public, work the media, and pressure health agencies to pile on the bandwagon and fund work that stands little chance of advancing our knowledge about the complex processes involved in normal development and disease. The appeal to the public and the media—science by press release—in an area as conflicted and as contested as the endocrine disruption hypothesis short-circuits the crucial process of working out the science. We need fewer but better studies that can help elucidate the key underlying issues, and eventually we need to reach a scientific consensus on what the evidence shows. Until then, we need to be aware of both the scientific reasons for the impasse as well as all the extrascientific agendas that get imposed on the science. Certainly, as a number of people pointed out to me, some responsibility lies with the media, which is always ready to retail scary reports about a threat lurking in the recesses of our daily lives. But ultimately, as Doerge and all the scientists who insist on maintaining a critical stance stressed, this issue has to do with the conduct of science in the area of the “environment” and human health. Doerge, along with scientists at the highest reaches of the National Institutes of Health, see the need for a total overhaul of the process by which scientific research is conceived, evaluated, funded, reviewed for publication, and published. As Doerge commented to me, “When we are not the most critical of our own data we defer that obligation to others.”
78