http://www.vegetus.org/other/endocrine.htm
Alternative Explanations for Phenomena Commonly Attributed to Environmental Estrogens
by Noah Lewis
This paper was written in June 2000 to satisfy the written requirement for a BS in Chemistry at the University of Pittsburgh.
Introduction
The fact that synthetic chemicals can mimic naturally occurring estrogens has been known since the 1930s (Jordan et al. 97). In recent years, however, there has been a large increase in the discussion of environmental estrogens (also known as hormonally active agents, xenoestrogens, and endocrine disruptors) in the scientific literature as well as in the popular press. The hypothesis is that exogenous estrogen-mimicking chemicals can interfere with key stages of development and contribute to hormonally related diseases in human and non-human animals. These xenoestrogens are compounds like phytoestrogens, synthetic estrogens, dietary estrogens from meat and dairy products, and a host of chemicals used for other purposes that happen to have estrogenic properties (e.g., pesticides, alklyphenol esters, and plastics). While many things are known for sure, important pieces of evidence are weak or missing. Few, however, will deny the basic facts.
Many chemicals are capable of binding to estrogen receptors and eliciting a response. Under specific conditions where wildlife has been exposed to large amounts of xenoestrogens, endocrine disruption has occurred. This is most notable when feminization of males occurs, as in the case of male fish who produced yolk proteins that are normally produced only by females (Folmar et al. 1096). Experiments show that exposure to xenoestrogens at critical stages of development can permanently alter an animal, e.g., reversing the sex of a reptile (Guillette et al. "Organization" 157).
People are exposed to xenoestrogens through a multitude of pathways. Bisphenol-A is used in composite tooth fillings; it also leaches from the lining of canned goods and from plastic food containers when heated. Dairy products are often obtained from pregnant cows and contain high levels of estrogen (Sharpe and Skakkebæk 1393). Polychlorinated biphenyls (PCBs), DDT and other pesticides bioaccumulate and are found in fish and other meats. Water contains pesticide runoff from agriculture, and fruits and vegetables contain pesticide residues. Alkylphenol ethoxylates (APEs) are used as detergents, emulsifiers, wetting agents, and dispersing agents. Humans can be exposed to APEs through the water supply, sewage sludge used for fertilizer, seafood, food packaging, and spermicides (Nimrod and Benson 338).
Evidence of endocrine disruption in humans comes almost exclusively from evidence about children whose mothers, during pregnancy, were given diethylstilbestrol (DES), a powerful synthetic estrogen. Daughters who were exposed have higher rates of reproductive tract deformities, infertility, and a rare form of vaginal cancer, among other things (Colborn, vom Saal and Soto 379). Less research has been done on sons, but they may have higher rates of abnormal sperm, immune system problems like arthritis, undescended testicles, epididymal cysts and possibly testicular cancer (Colborn, Dumanoski and Myers 59). Even with a potent estrogen deliberately administered to pregnant women, only high doses led to effects. According to toxicologist Robert Golden, "for some reason the lowest doses were prescribed at the Mayo Clinic (in Rochester, New York [sic]) and the highest at the University of Chicago. When you look at Mayo (results), there's nothing coming out of there. Yet out of the University of Chicago, there are all sorts of reproductive problems such as small penises, decreased sperm, abnormal sperm" (qtd. in Fumento).
Despite all of the above, there is no causal evidence that widespread endocrine disruption is occurring in humans. Increases in breast cancer, testicular cancer, male genital deformities and a drop in sperm count are the most commonly cited indicators of endocrine disruption in humans. But all of the evidence is strictly correlative, sometimes wildly so when endocrine disruptors are blamed for things like a drop in SAT scores (Colborn, Dumanoski and Myers 235) and the breakdown of the traditional family (Colborn, Dumanoski and Myers 237). Not only does the lack of compelling evidence cast doubt on the belief that endocrine disruption is occurring in humans, but it also strengthens the belief that disruption is not occurring at all. If such hormone sensitive systems have not been clearly affected following the post-World War II barrage of endocrine disrupting chemicals, then it is unlikely that low doses of these chemicals are harming human health—the heart of the endocrine disruption theory. Although any hormone in the body could be subject to disruption, only estrogen mimics will be discussed here as they have by far received the most attention from both scientists and the public.
A Century of Change
The key date for the beginning of endocrine disruption is the end of WWII. This was when large quantities of known and suspected endocrine disruptors came into widespread use (Colborn, vom Saal, and Soto 378). However, there were other important changes over the past century—most notably in diet. In the US, daily carbohydrate consumption is down from 497 g in 1904 to 388 g in 1974. At the same time, total protein intake remained constant at around 100 g, but it went from being 50% animal and 50% vegetable to 70% animal and 30% vegetable (Gortner 3248). Total fat increased from 127 g to 158 g. Since 1940, pork and milk consumption dropped significantly while margarine, oil, and beef consumption increased markedly (Gortner 3251). Poultry consumption saw a "dramatic" increase with new techniques used to breed chickens on a mass scale (Gortner 3249). USDA figures also show that daily caloric intake has increased by 800 calories since the 1950s (Hunter, "America’s" 12). People are eating out more now (34% vs. 19% in the 1970s) and portions are increasing—for example, a fast food hamburger used to contain 1 oz of cooked meat in the 50s whereas today it is 6 oz (Hunter, "America’s" 12).
Another unprecedented change this century is the trend toward delayed childbirth. "For the first time in recorded history, the number of births for every thousand British women in their early 30s has recently exceeded that in women in their early 20s. The rate in older women is climbing too" (Gosden and Rutherford 1585). This paper will show that these and other lifestyle changes frequently offer a simpler and more compelling explanation for many problems that some would like to attribute to endocrine disruption.
Breast Cancer
Numerous ideas exist about how xenoestrogens may cause breast cancer. They may simply increase a woman’s overall estrogen exposure, they may act to increase the potency of endogenous estrogens, or prenatal exposure may cause permanent susceptibility (Colborn, Dumanoski and Myers 183). In the US, a 1% per year rise in breast cancer incidence has been documented since the 1940s (Feuer and Wun 1424). This is taken as evidence of endocrine disruption since women have been exposed to increasing amounts of hormonally active compounds since World War II.
However, the trend toward increasing breast cancer began well before WWII. Breast cancer incidence increased has from 1935 (with no prior data available) (Sullivan 1067). Despite this, mortality has been constant since the mid-1940s (Sullivan 1067-8). In Norway, the incidence trend has been increasing nearly steadily since 1916 (Tretli and Gaard 510). That the trend began well before the widespread use of suspect chemical indicates that other factors caused the trend initially and probably still account for the increase. Recent increases above the 1% trend can be attributed to better detection. In women over 40, rates have increased since 1982, but this is attributable to increased use of mammography (Feuer and Wun 1434; Wun, Feuer and Miller 142). In women aged 40-59, mammography may account for more than the reported rate increase; that is, there could even be a decrease in actual rates of incidence (Wun, Feuer and Miller 1433).
The idea that breast cancer can be caused by exposure to the strongly estrogenic DDT and PCBs has been disproven. Past correlations have been minimal (not even statistically significant) and based on six comparatively small epidemiological studies (Key and Reeves 1520). A much larger study (having as many women with breast cancer as all of the previous studies combined) showed no correlation between serum levels of DDE (a metabolite of DDT) or PCBs and breast cancer (Krieger et al. 589-90). Additionally, these samples were collected an average of 14 years prior to diagnosis (the time period from 1964-1971—a peak in exposure), which could better account for causality than the previous studies that collected samples at the time of diagnosis. The largest study to date, using blood samples taken from 1974 and 1989, showed no increase in risk corresponding with higher levels of DDE (Helzlsouer et al. 529). Blood samples taken from 1989-1990 also reveal that there is no correlation (Hunter, "Plasma" 1256). Recent studies of breast adipose tissue found no difference in the concentration of DDT and its metabolites in women with breast cancer and in women without breast cancer (Bagga et al. 751; Zheng et al. 455). This is particularly significant because even if studies using plasma or gluteal adipose tissue are discounted, it appears that even measuring the organochlorine concentration directly in the breast still fails to produce a correlation.
Additionally, because DDT is still used in other countries, one would expect skyrocketing rates of hormonally dependent cancers in developing nations. This is not the case. Furthermore, in a relatively small study of northern Vietnamese women, whose DDE levels are on average more than double that of US women, no correlation between DDE and breast cancer was found (Schecter et al. 454). DDT is also still in use in Mexico. A study there found no correlation between the concentration of DDE and breast cancer incidence, although DDE levels were similar to US women (López-Carrillo et al. 3730).
Another troubling point for endocrine disruption theory is that African American women have adipose DDE levels that are on average 74% higher than white Americans (Cocco et al. 3), yet rates of breast cancer are higher in whites (Zheng et al. 457).
In a large accidental exposure of humans (976 males and 1062 females) to PCBs, no increase was reported in the incidence of breast cancer, and there was actually a decrease in overall cancer mortality except perhaps for liver cancer (Hsieh et al. 417). Similarly, the results of a study on a population exposed to 2,3,7,8-tetrachlorodibenzo-para-dioxin (TCDD) as a result of an accident 15 years ago in Sevoso, Italy, show a decrease in breast, uterine and ovarian cancer (Bertazzi et al. 649). The weight of evidence concludes that increased exposure to DDT and PCBs does not increase breast cancer risk (Helzlsouer et al. 529).
In fact, women with the highest levels of DDE and PCBs consistently have the lowest risk of breast cancer (Cocco et al. 3, Dorgan et al. 7, Helzlsouer et al. 529, López-Carrillo et al. 3730, van’t Veer et al. 85). These decreases in breast cancer and the inverse association with DDE levels could ironically be explained in terms of endocrine disruption. Instead of increasing overall estrogen exposure, the less potent xenoestrogens bind to estrogen receptors without triggering transcription, thus decreasing the effect of endogenous estrogen exposure. Yet no one on either side of the question has seized upon these results as evidence of endocrine disruption. In Our Stolen Future, the authors initially point out that while DDT mimics estrogen, DDE depletes hormones in the body, including estrogen susceptibility (Colborn, Dumanoski and Myers 85), but later cited the studies available at that time that indicated increased DDE levels were associated with increased risk of breast cancer, especially in women with estrogen-responsive tumors (Colborn, Dumanoski and Myers 184). This highlights the ease with which endocrine disruption claims can be "proven" if the desired result is not found—the chemical must simply act in the opposite manner.
Proponents of endocrine disruption theory claim that even if DDE or some PCBs do not cause breast cancer, there are a myriad of other estrogen mimics that could cause breast cancer (Colborn, Dumanoski and Myers 184). But this type of dismissal belies the strength of their initial claim—that the estrogenicity of these compounds is to blame. DDE is a potent estrogen and if it does not cause breast cancer, this casts serious doubt on the idea that chemicals are dangerous simply because of their estrogenicity.
There is a simpler explanation for the long-term increase in breast cancer. An established risk factor for breast cancer is a diet high in fat, calories and protein and low in complex carbohydrates and fiber (Adlercreutz, "Diet" 281; Stephens 756). Additionally, "certainly the incidence is lower in women who adhere to a strict vegetarian diet" (Stephens 756). Clearly, the standard American diet increases women’s risk for breast cancer.
Unfortunately, this western diet is spreading worldwide, particularly to cities in developing countries where the traditional diet was low-fat vegetarian or semi-vegetarian. It is not surprising that western diseases like breast cancer would follow. Breast cancer rates in Japan increased following the introduction of a diet rich in meat, eggs and animal fat (Rose et al. 2367; Hirayama 179). A study from 1950-1975 revealed that increased breast cancer rates in Japan corresponded strongly with increases in non-fish animal fat (particularly strong with pork) (Kagawa 211; Hirayama 179).
In the past, breast cancer rates declined for Japanese women, while in the US that risk has always increased. The trend in Japan is now toward increased risk for postmenopausal women, but no data from either country indicates a change in the rates of younger women, which would suggest xenoestrogen involvement (Kodama, Murakami and Kodama 795). Regarding research back to 1970 as well as a plot of breast cancer incidence rates in Japan from 1959-1987, Kodama et al. conclude "There is no explanation feasible other than the Westernization effect" (Kodama, Murakami and Kodama 795). Data from England from the 1960s to mid-1980s reveals the same phenomenon—some increase in cancer incidence in postmenopausal women, but no change in the risk for young women (Kodama, Murakami and Kodama 795). These facts are inconsistent with a xenoestrogen explanation, but are consistent with the Westernization effect.
Numerous studies over the past 20 years have documented the phenomenon that migrants to the United States from areas of low breast cancer incidence tend to adopt the cancer rates of the new country (Dunn 3240; Stanford et al. 182; Hirayama 194). Immigrants from China and Japan, for example have breast cancer rates that are 50% lower than white American women, but only 25% lower for first generation US born Asian-Americans (Stanford et al. 182). After successive generations, the disparity between immigrant families and natives decreases even further (Stanford et al. 182), which corresponds to increasing Westernization of diet. An explanation based on chemical contaminates could certainly be devised, but their major weakness is in explaining why the rate of breast cancer continues to grow in successive generations. If the xenoestrogens are ubiquitous, the first American generation ought to be affected just as much as anyone else born here is.
Not only does the Westernization effect explain increases in breast cancer in other countries, but it also explains the steady increase in Western countries where the diet has been changing for the worse in terms of breast cancer risk factors. The marked halt in the trend of increasing breast cancer for women who experienced adolescence during WWII provides evidence for the diet-breast cancer relationship (Tretli and Gaard 511). During WWII in Norway energy intake decreased by 22%, especially with a drop in fat consumption. While the consumption of meat had increased up until the war, it "dropped substantially" during the war along with milk intake. There was increased intake of vegetables and fish as well as increased physical activity (Tretli and Gaard 511). Breast cancer is also correlated with height; the height of adolescent girls in Oslo steadily increased from 1920-1975, except for a decline during WWII (Tretli and Gaard 510). Interestingly, US breast cancer mortality declined from 1940-1946 (Sullivan et al. 1068).
Several mechanisms can explain why diet affects breast cancer. First, a high protein/carbohydrate ratio decreases the level of sex hormone binding globulin (SHBG) in the blood and enhances 2-hydroxylation of estradiol, resulting in higher estrogen concentrations than with lower protein/carbohydrate ratios (Adlercreutz, "Western" 6). The lowest protein/carbohydrate ratios are found in vegetarians and the highest in breast cancer patients (Adlercreutz, Western 6). Second, high fiber diets increase fecal weight and thus the amount of estrogens excreted while a diet higher in fat decreases the amount of estrogen found in feces. Additionally, a high fiber diet decreases the intestinal reabsorption of estrogen, and some fibers even bind directly to sex hormones (Adlercreutz, "Western" 5). Overall, a low fat-to-fiber ratio reduces the amount of biologically available estrogen, thus theoretically decreasing the risk of cancer (Adlercreutz, "Western" 6).
Finally, a separate lifestyle factor that could explain at least some of the rise in breast cancer is the trend toward delayed childbirth. The risk of breast cancer increases with age at first birth (Tretli and Haldorsen 215). Fifteen percent of Norway’s increase in breast cancer can be attributed to delayed childbirth (Tretli and Haldorsen 218).
Menarche
Xenoestrogens have been cited as the cause of the falling age of menarche, the age at first menstruation (Birnbaum 676). A potential mechanism is simply that high levels of estrogen trigger menarche in young women whose natural estrogen levels otherwise would not.
First, there is no longer even a trend toward earlier menarche; in Europe the trend has largely stopped (Tanner 160). In Denmark, for example, between 1966 and 1983, the trend toward lower ages stopped (Helm and Grolund 198). In Belgium, the average age was 13 in both 1965 and 1981 (Helm and Grolund 198). In America, the age of menarche in white girls has been stable at 12.9 since at least 1948, although there may have been a slight decline in the age for black girls (Herman-Giddens et al. 505).
Second, the claim that the trend post-WWII is due to increased exposure to xenoestrogens is refuted by the fact that the trend began by at least 1835—over 100 years before the widespread use of suspected endocrine disruptors. In Denmark from around 1835 to 1945, the age went from 17 to 14 in a linear decline. Then from 1950 to 1966 it went from 13.8 to 13.4 (Helm and Grolund 199)—the trend slowed when, according to the endocrine disruption hypothesis, it should have increased.
Finally, the trend toward earlier menarche can be best explained by rising standards of living—particularly in food quality. Prior to WWII, but less so after WWII, there was a sharp increase in height and weight of children in places like Europe, the US, Canada, and Australia (Tanner 157). Increased size alone can explain the drop in the age of menarche, given that it seems to be linked to attaining a critical weight or percent body fat. Higher levels of the hormone leptin, which increases with the amount of body fat, result in lower ages of menarche (Matkovic et al. 3243). Eating more protein and fat lowers the age, while a diet consisting mainly of plants and carbohydrates raises the age (Sanchez et al. 1341); vegetarian girls have a later menarche than their peers (Sanchez et al. 1342). In Japan, the age of menarche dropped from 15.2 in 1950 to 12.2 in 1974 (Kagawa 211). This corresponds to a time when some Westernization occurred to the Japanese diet: milk, meat, and fat intake increased respectively to 15, 7.5 and 6 times their previous levels (Kagawa 206). Adolescent girls went from being 133 cm tall and weighing 30 kg in 1900 to 148.5 cm and 41 kg in 1974, increasing slowly at first then rapidly after 1950 (Kagawa 210). Thus, it is clear that worldwide changes in diet have brought about changes in age at menarche before the introduction of xenoestrogens.
Endometriosis
Endometriosis is a disease characterized by lesions in other parts of the female body caused by cells that normally line the uterus; a major symptom is extreme pain. Estrogen is known to aid the spread and growth of endometrial cells. Factors such as smoking and exercise are associated with lowered risk of endometriosis, presumably because of the decreased estrogen levels caused by these activities (Eskenazi and Warner 253).
The supposed increasing incidence of endometriosis has been cited as an indication of endocrine disruption (Birnbaum 676, National Research Council 135). Some even went so far as to declare, "Prior to 1921, there were only twenty reports of the disease in the worldwide medical literature," absurdly implying that the disease did not even exist prior to the industrial era (Colborn, Dumanoski and Myers 181). This is false, however, as remarkably accurate and detailed reports of the symptoms and manifestations of the disease have existed throughout history, dating back to at least 1690, when even then, it was regarded as a relatively common disorder (Knapp 11).
In fact, the frequency of severe endometriosis has probably not changed since at least the 1940s (Vercellini and Crosignani 111), which, again, is particularly problematic since the vast majority of suspected xenoestrogens were not released until after 1940. The increase in mild cases of endometriosis is probably due solely to the use of naval laparoscopy, which can detect cases that would have otherwise gone undiagnosed (Vercellini and Crosignani 111).
If environmental estrogens do affect humans, one would expect to see an increase in severe endometriosis because endometrial cells require estrogen for proliferation and long-term survival (Bergqvist 33). These cells that once lined the uterus continue to react to changes in estrogen levels even once they are outside the uterus; the activation of these cells during menstruation is what causes the severe pain associated with endometriosis. Since "no association between endometriosis and prenatal DES exposure has been established" (Haney 3), there is probably no critical time during pregnancy where a dose of excess estrogens later causes endometriosis. Because lifetime estrogen (or xenoestrogen) exposure would be the cause of excess cases of endometriosis, there has been more than adequate time for a trend to emerge. Furthermore, case control studies reveal no association between incidence of endometriosis and plasma levels of 14 PCB congeners, 11 organochlorine pesticides, or dioxin (Lebel et al. 222, 225).
One explanation for any apparent or future increase in endometriosis could simply be that awareness about endometriosis on the part of both women and doctors has increased. Consequently, women are actually diagnosed with the disease instead of being dismissed as mental cases. Seventy percent of women who are ultimately diagnosed with endometriosis were told at some point that it was only psychological (Ballweg 444). Of 850 women who were referred as mental cases to a doctor who happened to be both a psychiatrist and a gynecologist, he discovered that 92% of the women actually had endometriosis (Ballweg 453).
Sperm Count
Perhaps the most important piece of evidence cited by endocrine disruption proponents is that of falling sperm counts (Sharpe and Skakkebæk 1392). Numerous mechanisms exist based on exposure to xenoestrogens in utero and up until puberty (Sharpe and Skakkebæk 1394-5). Falling sperm counts have been referred to as "the most dramatic and troubling sign that hormone disruptors may already have taken a major toll" (Colborn, Dumanoski and Myers 172). The dramatic nature of the claim has made it a key factor in influencing public opinion and public policy; Greenpeace’s slogan, "You are half the man your grandfather was," made it into Congressional debates. Because of its central role in the endocrine disruption controversy, the issue of sperm counts will be discussed in detail.
The idea that sperm counts are decreasing at all has been widely criticized. The only evidence of a global decline is a meta-analysis by Carlsen et al., which analyzed 61 studies that took place over 50 years from 1938-1990. They found a stunning decline in sperm counts—nearly 50% (Carlsen et al. 610). However, this study is fraught with uncertainty.
Many methodological problems have been noted such as publication selection bias, patient selection bias, method of sperm collection, small sample size of many studies, length between data collection and publication, use of an arithmetic mean, and use of a simple regression line (Farrow 1; Paulsen, Berman and Wang 1019; Olsen et al. 890). These variables are significant—take method of sperm collection for example. Some samples are collected at home, while others at a clinic. Stress is known to decrease sperm counts, so clearly, where the sample is collected could have a direct impact on sperm count. It is reasonable to assume that over time, as sperm banks became more popular, more of the later samples were collected in clinics. Additionally, sperm was counted using many different protocols and counting chambers (Fisch et al. 1012). Furthermore, "The men in these studies ranged in age from 17 to 64 years, the duration of abstinence was for the most part neither controlled nor recorded, and the mean sperm concentration varied threefold" (Sherins 327). Finally, sperm counts vary over course of the year, with high levels in February and March and low levels in September (Tjoa 454; Gyllenborg et al. 31). This is yet another variable that was not taken into account by the Carlsen study.
Subsequent reevaluation of Carlsen’s data by other researchers indicates that "nearly all of the observed decline in mean sperm count may be a consequence of the reduction of the lower reference value" (Bromwich et al. 21). That is, the minimum sperm count considered "normal" was lowered over the years. Earlier studies may have excluded "low" counts that were included in later studies.
Differences in recruitment of donors can also explain a decline in sperm count. As the number of prospective donors increases, as it has over the past 50 years, the sperm banks can be more selective (Fisch et al. 1012). Donor recruitment bias is significant. Over a 15-year period at the same facility, five studies requiring sperm samples were conducted. Comparisons between the mean sperm count in each study revealed statistically significant differences between the studies (although there was no overall downward trend for the time period) (Handelsman 2703). This highlights the bias of self-selected donors, rather than a true random sample of the population (which for the most part has been lacking in sperm count studies). Handelsman also stated, "It is notable that this large bias effect size within a single centre is comparable with the magnitude of the alleged effect size for decrease in sperm concentration over six decades, according to the Carlsen meta-analysis" (2703).
Adjustment for age and duration of abstinence (which were not controlled for in Carlsen’s analysis) shows no decline (Fisch et al. 1013). Abstinence is a very important variable, yet only 53% of the Carlsen studies provided data on the recommended length of abstinence (let alone the actual length of abstinence) (Olsen et al. 891). Furthermore, a large study comprising 1/3 of all the post-1971 data was conducted using men who were undergoing a vasectomy (Tjoa et al. 454). The men in this study were far less likely to comply with abstinence guidelines, since they had no interest in what their sperm count was (Olsen et al. 891). Abstinence periods could explain the decline in the 1960s. Birth control may have led to shorter abstinence periods, which is in accord with the decline in semen volume noted by Carlsen (Suominen and Vierula 1579). At least one study from 1955 noted a significantly longer abstinence period than is standard in present studies (Suominen and Vierula 1579).
If the statistical model of continuous linear decline in sperm count proposed by Carlsen is accepted, it would also indicate that sperm counts have been declining before 1940 and thus the phenomenon must be attributable to something other than or in addition to endocrine disruptors (Olsen et al. 887-93). There is also a "need for caution in using regression models to determine the statistical significance of the results, because the correlation coefficients were very small, indicating that variables other than the sperm concentration contributed to the determination of statistical significance" (Sherins 327). A stair-step model more accurately fits the data. This model indicates a rapid decline in sperm count in the mid-1960s (such as one due to shorter abstinence periods caused by the birth control pill) but no further declines (Olsen et al. 891). This leaves open the question as to what the trend looked like before 1940. Perhaps there was a spike in sperm counts due to WWII caused by any number of factors, such as longer periods of abstinence.
The Carlsen study and any attempt to analyze sperm levels suffer from the fact that there is very little data prior to 1970. Analysis of post-1970 data indicates no decline or a slight increase (Olsen et al. 891). This is very different from the linear decline that endocrine disruptor proponents promote to the public.
Geographical differences can also explain the apparent decline or at the very least, raise serious concerns about the effects of sample selection bias. New Yorkers have consistently had higher than average sperm counts (Fisch et al. 1013). In the data before 1970, 94% of the men were from the US and 87% of those from New York City. After 1970, 50% of the men were from the US and only 25% from New York City (Fisch et al. 1013). If the Carlsen data is analyzed without the large studies from New York, then there is no decline in sperm count (Lipshultz 910). An analysis of only the US data from the Carlsen data plus later US studies indicates no decline from 1938-1996 when New York and non-New York data are analyzed separately (Saidi et al. 460). If the distinction is ignored, however, a decline can be shown (Saidi et al. 460).
When well-controlled studies from one geographical location are studied, there is typically no trend toward lower counts. The downward trend seems only to emerge when a huge number of variables are introduced from many cross-geographical studies. A series of studies performed at a Los Angeles fertility clinic have revealed no differences in sperm count from samples taken in 1951, 1979, and 1994-97 (Acacio et al. 596). Similarly, a study of 20,411 infertile men from 1960 to 1996 in Northeast Spain revealed no decline in sperm count (Andolz, Bielsa, and Vila 734). A meta-analysis of six Finish studies from 1958 to 1992 revealed no decline (Suominen and Vierula 1579). Studies using sperm collected from the 1970s to the 1990s have noted an increase in four US locations and in Denmark (Paulsen, Berman and Wang 1019; Fisch et al. 1012; Gyllenborg et al. 30). A more recent Korean study revealed no decline from 1989 to 1998 (Seo et al. 194). A unique Danish study used men who were married to infertile women seeking in vitro fertilization to achieve a random sample of the general male population. The samples were collected from 1990-1996 and no associations were noted between year of birth and sperm concentration (Rasmussen 1060). For example, there was no difference in counts from men born before 1955 and those born after 1964 (Rasmussen 1060).
There are exceptions—some studies from one geographical location do reveal a decline. A study of Scottish men showed a decline in sperm count associated with year of birth (Irvine et al. 467). However, this study suffers from the fact that the older group contained men of proven fertility (they had fathered children) and the younger men contained men of unconfirmed fertility. In fact, 81% of those born before 1960 had proven fertility compared to only 5% of those born from 1970-74 (Irvine et al. 469). The younger men were more likely to contain some infertile men, which would drive down the mean sperm count for the younger men (Farrow 1). Furthermore, the results obtained were consistent with either an increase in sperm count with age or a decline in count associated with year of birth. The authors discounted the former, despite the obvious fact that older men probably have longer periods of abstinence, and thus higher sperm counts. Another study indicated that older men did have higher sperm counts (Raab 44; Joffe 44). Also, the younger group being undergraduates, were perhaps more likely to have lower sperm count due to alcohol and smoking (Eccersley 43).
A study of Parisian men reported a decline in sperm concentration from 1973 to 1992 (Auger et al. 281). Only fertile, accepted sperm donors were included, which excluded all fertile men who were rejected as sperm donors. However, the selection criteria or any changes in it over time were not noted (Lipshultz 910). "In addition, differences in age, abstinence before semen analysis, ejaculatory frequency, and the number of samples analyzed per person were not controlled for" (Sherins 327). This study too could suffer from the fact that older men in general have higher sperm concentration than younger men (Joffe 44). Rasmussen et al. note the importance of variation over even a short time period: "In our study, the average sperm concentration varies by >30% from year to year. The variation persists even after correction is made for differences in the age and number of patients, and the variation from one half-year period to another is greater than the entire difference for the period 1973-1992 in the study by Auger et al." (1062).
In case sperm counts really have fallen (or, at the very least, to add another variable that sperm researchers must control for) those who manufacture suspected endocrine disruptors have proposed their own creative explanation. Dow Chemical Company’s Health and Environmental Research Laboratory recently published a study proposing that the introduction of iodized salt is responsible for the drop in sperm count (Crissman 400). This would mean that sperm counts really have not fallen so much as returned to normal levels from abnormally high levels caused by iodine deficiency. Animal studies suggest that individuals who experienced iodine deficiency in the womb have higher sperm counts. The iodine deficiency leads to hypothyroidism, which causes a longer period of growth for sperm-producing Seritoli cells, so that ultimately those individuals possess a larger number of them. The elimination of iodine deficiency would mean that men do not develop additional Seritoli cells and therefore their sperm counts would fall. This hypothesis even explains the decline in the Parisian men because iodized salt was not introduced in France until 1952, in the middle of the years in which the men studied were born (Crissman 408). But it is also worth noting that "in 1923 William J. Hale, a chemist at the Dow Chemical Company in Midland, Michigan, made the first iodized salt in America" (Crissman 408).
Ultimately, there are an incredible number of variables associated with measuring sperm count. Even individual men experience a wide range of "normal" values. Terrifying claims of a universal, global 50% decline in semen quality are made to get people’s attention. And that is precisely what happened. Yet such claims are easily refuted. Those with an interest in preserving the status quo can use the evidence to dismiss the whole theory. Rather than making broad doomsday predictions, those who fear endocrine disruption ought to make highly conservative claims that cannot be refuted. They should want to make highly specific claims so than a very strong correlative link can be made. For example, one study of sperm counts in London noted a decline in sperm counts from 1978 to 1989. But upon closer examination, the decline was observed only among men who lived in a particular water district (Ginsburg et al. 230). Further research could lead to the identification of the responsible compound(s), which will certainly not turn out to be iodized salt.
Testicular cancer
The increasing rate of testicular cancer is widely cited as evidence of endocrine disruption (Sharpe and Skakkebæk 1392; National Research Council 168). The mechanism is not clear, but is probably the result of exposure in the womb (Sharpe and Skakkebæk 1394). Both high and low levels of estrogen during pregnancy pose an increased risk of testicular cancer. High levels of estrogen (found in older women, women with a large placenta, and women of low parity) are associated with the seminoma variety of testicular cancer. Low levels of estrogen (young women, low birth weight) are associated with non-seminoma (Ekbom and Akre 227).
One problem with claiming that the increasing rate of testicular cancer is caused by endocrine disruption is the paucity of data before WWII. In Australia for example, there is no reliable incidence data prior to 1950 (Stone et al. 211). Given that cancer rates could have very well been increasing prior to 1950 (just like menarche ages were decreasing prior to 1950), any claims based on rates of testicular cancer are inherently on questionable ground.
When there is data, it shows that the trend existed before the widespread use of the suspect chemicals. In Australia, mortality rates from testicular cancer increased from 1910 to 1969 and fell afterward (Stone et al. 211). (It is acceptable to use mortality rates for older data; it is only in the past 3 decades that improvements in treatment have allowed declines in mortality that do not correspond to declines in incidence) (Davies 930; Stone et al. 211). A study in England and Wales showed that testicular cancer mortality rates have been increasing since the beginning of the 20th century, with men born as early as 1881 being affected (Davies 930). Furthermore, it is not that there is no trend in the 19th century—there is simply a lack of data for that time period. A study of the US indicated that testicular cancer incidence has been increasing since 1935 among men aged 15-44, with no data prior to that time (Brown et al. 166). Brown actually noticed two trends, from 1935-50 there was an increase in seminoma, while in the 1970’s there was an increase in nonseminoma (169; Stone et al. 217). There were either two separate factors for the increases, or one factor has a different on men depending on when they are exposed. Thus, while testicular cancer rates are perhaps increasing due to xenoestrogens, there is definitely another factor that explains the earlier increase, and perhaps explains the present continuation of that trend as well.
In the US from 1973-1984 there was a 33% increase in testicular cancer among whites, but no increase among blacks (Van Den Eeden and Weiss 1553). This is particularly difficult to explain since blacks have higher concentrations of suspected endocrine disruptors in their bodies.
Once again, lifestyle factors could easily explain the rise in testicular cancer. It has been known for over 20 years that fat and meat intake corresponds to testicular cancer rates (Armstrong and Doll 627; Paulozzi, "Effects" 527; Sigurdson et al. 22). Similar to the breast cancer trend, there was a drop in testicular cancer in Denmark for men born during World War II (Møller, "Decreased" 1669). This is in accord with Paulozzi, who found a slightly higher correlation with fat consumption at age of birth, rather than at time of diagnosis ("Effects" 527). Ekbom and Arke point out that New Zealand, Switzerland and Denmark have the highest incidences of testicular cancer (226). New Zealand and Denmark have among the highest per capita meat and fat consumption rates (data on Switzerland was not provided) (Armstrong and Doll 625). Additionally, any study of testicular cancer confirms that rates increase with affluence (Stone et al. Trebling 218; Davies 931), which may or may not be related to fat consumption.
Another clear risk factor for testicular cancer is a sedentary lifestyle. Exercise decreases risk while sitting around increases the risk (Gallagher et al. 401; "Aetiology" 1397). The use of computers, televisions, and video games are all novel sedentary activities that appeared post-WWII. Physical activity has declined since 1950 (Hunter, "America’s" 13).
Perhaps most intriguing is that the risk of testicular cancer declines as the age of puberty rises (Gallagher et al. 404; "Aetiology" 1396). Also, not surprisingly, the age of puberty in places like Canada, the US, and Europe has decreased (Gallagher et al. 404), perhaps due to a similar mechanism as the decline in age of menarche, namely that higher standards of living bring about faster maturity. The lower age of puberty alone could explain the increase in testicular cancer rates ("Aetiology" 1397; Weir, Kreiger, and Marrett 257).
Male Genital Defects
Increases in hypospadias, a urethral opening on the underside of the penis, and cryptorchidism, undescended testes, are cited as evidence of endocrine disruption (Kelce and Wilson 202; Sharpe and Skakkebæk 1392). This idea is based on animal studies and a supposed increase in these disorders among sons who were exposed in utero to high levels of the synthetic estrogen DES.
There is uncertainty as to whether or not the rates of these problems in the general population are even increasing. A study that examined the rates based on groupings of countries according to gross domestic product (GDP) showed no increase anywhere in cryptorchidism since 1910 and a decrease in most countries since 1985 (Paulozzi, "International" 299-300). In the least affluent areas (China, Czech Republic, Hungary, Mexico, and South America) the rates of hypospadias have been stable since around 1974 (Paulozzi, "International" 300). In countries with the second lowest GDPs (Mediterranean nations and Ireland) there were no true increases, only an artifactual increase in Italy due to the initiation of a hypospadias study, and perhaps in Israel where the rate fluctuated wildly due to unknown reasons (Paulozzi, "International" 299). There were no increases anywhere following 1985 (Paulozzi, "International" 299). There have been increases in Scandinavia, Japan and two studies in the US, all of which are in the highest GDP grouping (Paulozzi, "International" 298, 299). However, there are no linear increases. The trend fluctuates up and down with a slight net increase. One of the US studies was stable from 1968 to 1982, there was an increase until 1985, and then it was stable again (Paulozzi, "International" 298). It seems apparent that these defects are affected by affluence, perhaps dietary as in the case of testicular cancer.
Any trend toward increasing hypospadias could be explained by a change in the definition to include more mild forms. If only severe forms, of which the definition has not changed, are examined, there is either no change or a decrease (Paulozzi, "International" 300). Also, improved documentation by physicians can account for increases in both hypospadias and cryptorchidism, especially given the emergence of the knowledge that undescended testes are more likely to become cancerous and are therefore given more attention (Paulozzi, "International" 301).
Cryptorchidism is associated with low birth weight and high maternal age (Møller and Skakkebæk 909). Thus the increasing age at which women give birth could lead to an increase in cryptorchidism. Paternal age may also be associated with increased risk but no studies have included information on paternal age.
Prostate Cancer
Because both human exposure to xenoestrogens and rates of prostate disease have increased, endocrine disruptor theorists believe there could be a causal relationship (Colborn, Dumanoski, Myers 180). In mice and rats at least, increased estrogen levels are associated with enlarged prostate glands and long-term exposure to estrogen is associated with prostate cancer (Colborn, Dumanoski, Myers 179). Furthermore, prostate cancer is a hormonally influenced cancer—endogenous hormones affect it and hormone therapy is used in treatment—therefore, it ought to be affected by exogenous estrogens.
Prostate cancer incidence rates suffer from the common problem of a lack of earlier data. There appears to be a long-term slow rise in incidence since 1935 (Sullivan et al. 1069). Mortality, however, has been constant or in slight decline since 1935 in whites, but there has been an increase for nonwhites (Hsing and Devesa 527; Sullivan et al. 1069). Better detection may have allowed more tumors to have been detected, but the lack of a corresponding increase in mortality could perhaps indicate that true incidence has remained the same. The sudden introduction of chemicals should have caused a birth year cohort effect, but there is none (Hsing and Devesa 527). Finally, the rapid increase in incidence in the early 1990s was a result of new screening programs that used prostate specific antigen (Dennis and Resnik 257).
Diet plays a role in causing prostate disease and could explain the increase in incidence. Global food consumption patterns and cancer incidence rates indicate that prostate cancer correlates also highly with total fat consumption (Armstrong and Doll 626). Following the introduction of a western diet in Puerto Rico and Japan, both countries experienced increased rates of prostate cancer (Rose et al. 2367). Cross-country comparisons show that increasing animal fat intake increases the risk of prostate cancer (Rose et al. 2366; Boyle and Zaridze 1057). Much like breast cancer rates, migrants quickly achieve the rates of the new country, whether those rates are higher or lower (Boyle and Zaridze 1048; Dunn 3244). Japanese men have the same amount of small and latent prostate tumors as men in the US, but US men have a much greater mortality from prostate cancer because the tumors grow instead of remaining dormant (Moyad 98; Armstrong and Doll 626). Even latent tumors are more proliferative in Japanese men living in Hawaii than those living in Japan are (Dunn 3242). Exposure to xenoestrogens should be comparable in Japan and America since they are both industrialized nations. The major difference between Japan and the US is that Japan still has a diet lower animal products than the US—although this is rapidly changing. Despite dramatic increases in animal product consumption beginning in the late 1940s, compared to the US in 1975, Japan’s consumption was "negligible"—1/6 the amount of meat, 1/5 the milk, and 1/2 the fat (Kagawa 206). It was studies at that time that initially revealed the differences in incidence of Western cancers.
Case-control studies, in which a group of men with prostate cancer is matched with a control group who does not, highlight any factors the groups do not have in common. Consistent with the geographical information, studies show that vegetable consumption lowers the risk of prostate cancer (Cohen, Kristal, and Stanford 63) while animal fat increases risk (Rose 2366). Furthermore, foods high in phytoestrogens are also associated with a decreased risk of prostate cancer (Strom et al. 22).
The Framingham study of 2,164 men revealed an additional factor that could contribute to the rise in incidence of prostate cancer—advanced paternal age at birth increases the risk for prostate cancer (Zhang et al. 1211). A potential explanation is that "continuous male germ cell division may increase the rate of mutation due to replication error" (Zhang et al. 1211). The trend toward later childbirth would inherently bring about an increase in prostate cancer.
Sex Ratio
Another factor cited as an example of endocrine disruption is the decline since 1950 in the sex ratio, which is the proportion of live newborn boys to girls (Møller, Trends 232; National Research Council 135). Often expressed as a percentage, the typical rate is 51% male. An increasing percentage of female infants could hypothetically be the result of higher estrogen levels. The major reason for this theory is that following the accidental exposure to high levels of dioxin in Sevoso, an unexpectedly high number of babies born were female (48 out of 74, or 35% male) (National Research Council 135). Although dioxin is not a xenoestrogen, it may interfere indirectly with the endocrine system.
A number of studies have indicated that sex ratios are falling. Data from the Netherlands from 1950-1995 shows a decline, although the author noted that values from 1950-54 were somewhat higher than the rest of the data (van der Pal-de Bruin et al. 62). Post-1950 data for England and Wales also shows a decline (Dickinson and Parker 227), as does US data from 1970-1990 (Allan et al. 40). Unfortunately, these studies, again, do not look back far enough and ignore significant historical events.
Data from Germany and the Netherlands from 1871 to 1995 revealed spikes in sex ratio immediately following the World Wars; similar patterns occurred in other combatant nations, while neutral nations experienced much smaller rises (van den Broek 805). Additional German data also revealed peaks at 1920 and 1945 with the level after 1945 slightly higher than the period from 1870-1914 (Bromen and Jöckel 804). That is, even if levels are falling today, they still have not fallen below what was the norm before 1920. Allan et al. plotted Canadian sex ratios from 1930-1990 and plotted a smoothed curve through the data. Unfortunately, the curve obscured the tremendous spike that occurred post-WWII. Additionally, by not knowing what the trend was prior to 1930, it is impossible to make judgements about the current downward trend. Canada was a major player in WWI, and undoubtedly experienced an earlier spike in sex ratios as European nations did. Plots from Denmark, Sweden, Norway and Finland from 1850 to 1996 reveal peaks at WWI and WWII (Møller, "Change" 232). The appropriate question is clearly not why is the sex ratio declining—it is why were there peaks following the World Wars from which we are seeing a return to normal levels. Claiming that the decline represents a problem is like worrying about a warming trend following an ice age—it would be a problem if the trend did not exist.
One possible explanation for the large increase in male birth following WWII is the baby boom. It has been shown that conception early in the menstrual cycle is likely to produce a boy, while fertilization at ovulation is likely to produce a girl. While individual couples cannot use this slight difference to control the sex of their child, the effect can be seen statistically in large groups (Møller, "Trends" 238). If we can assume that the baby boom was accompanied by an increase in the frequency of intercourse, then an increase in sex ratio would be the expected—and observed—result (van den Broek 805).
This same idea can be applied in reverse to the Sevoso incident. Following a major disaster, intercourse rates probably dropped dramatically, which could account for at least a portion of the high frequency of females who were born. For example, following the earthquake in Kobe, Japan, the sex ratio declined significantly 9 months later and there was a 6% decline in fertility as well (Fukuda et al. 2321). Low sex ratios have also followed floods, severe air pollution and other highly stressful events (Møller, "Trends" 235).
There is still an on-going slow decline in sex ratio that perhaps cannot be explained by the baby boom alone (although the sex ratio has not fallen below pre-WWI levels). Additional social changes may play a role in the continued stabilization/decline of the sex ratio. The use of fertility drugs that induce ovulation is known to produce a high proportion of girls (Allan et al. 40; Møller, Change 829). Additionally, women over 35 are less likely to have boys, so delaying childbirth would have the result of lowering the sex ratio (Dickinson and Parker 228).
Alternative Explanations for Phenomena Commonly Attributed to Environmental Estrogens
by Noah Lewis
This paper was written in June 2000 to satisfy the written requirement for a BS in Chemistry at the University of Pittsburgh.
Introduction
The fact that synthetic chemicals can mimic naturally occurring estrogens has been known since the 1930s (Jordan et al. 97). In recent years, however, there has been a large increase in the discussion of environmental estrogens (also known as hormonally active agents, xenoestrogens, and endocrine disruptors) in the scientific literature as well as in the popular press. The hypothesis is that exogenous estrogen-mimicking chemicals can interfere with key stages of development and contribute to hormonally related diseases in human and non-human animals. These xenoestrogens are compounds like phytoestrogens, synthetic estrogens, dietary estrogens from meat and dairy products, and a host of chemicals used for other purposes that happen to have estrogenic properties (e.g., pesticides, alklyphenol esters, and plastics). While many things are known for sure, important pieces of evidence are weak or missing. Few, however, will deny the basic facts.
Many chemicals are capable of binding to estrogen receptors and eliciting a response. Under specific conditions where wildlife has been exposed to large amounts of xenoestrogens, endocrine disruption has occurred. This is most notable when feminization of males occurs, as in the case of male fish who produced yolk proteins that are normally produced only by females (Folmar et al. 1096). Experiments show that exposure to xenoestrogens at critical stages of development can permanently alter an animal, e.g., reversing the sex of a reptile (Guillette et al. "Organization" 157).
People are exposed to xenoestrogens through a multitude of pathways. Bisphenol-A is used in composite tooth fillings; it also leaches from the lining of canned goods and from plastic food containers when heated. Dairy products are often obtained from pregnant cows and contain high levels of estrogen (Sharpe and Skakkebæk 1393). Polychlorinated biphenyls (PCBs), DDT and other pesticides bioaccumulate and are found in fish and other meats. Water contains pesticide runoff from agriculture, and fruits and vegetables contain pesticide residues. Alkylphenol ethoxylates (APEs) are used as detergents, emulsifiers, wetting agents, and dispersing agents. Humans can be exposed to APEs through the water supply, sewage sludge used for fertilizer, seafood, food packaging, and spermicides (Nimrod and Benson 338).
Evidence of endocrine disruption in humans comes almost exclusively from evidence about children whose mothers, during pregnancy, were given diethylstilbestrol (DES), a powerful synthetic estrogen. Daughters who were exposed have higher rates of reproductive tract deformities, infertility, and a rare form of vaginal cancer, among other things (Colborn, vom Saal and Soto 379). Less research has been done on sons, but they may have higher rates of abnormal sperm, immune system problems like arthritis, undescended testicles, epididymal cysts and possibly testicular cancer (Colborn, Dumanoski and Myers 59). Even with a potent estrogen deliberately administered to pregnant women, only high doses led to effects. According to toxicologist Robert Golden, "for some reason the lowest doses were prescribed at the Mayo Clinic (in Rochester, New York [sic]) and the highest at the University of Chicago. When you look at Mayo (results), there's nothing coming out of there. Yet out of the University of Chicago, there are all sorts of reproductive problems such as small penises, decreased sperm, abnormal sperm" (qtd. in Fumento).
Despite all of the above, there is no causal evidence that widespread endocrine disruption is occurring in humans. Increases in breast cancer, testicular cancer, male genital deformities and a drop in sperm count are the most commonly cited indicators of endocrine disruption in humans. But all of the evidence is strictly correlative, sometimes wildly so when endocrine disruptors are blamed for things like a drop in SAT scores (Colborn, Dumanoski and Myers 235) and the breakdown of the traditional family (Colborn, Dumanoski and Myers 237). Not only does the lack of compelling evidence cast doubt on the belief that endocrine disruption is occurring in humans, but it also strengthens the belief that disruption is not occurring at all. If such hormone sensitive systems have not been clearly affected following the post-World War II barrage of endocrine disrupting chemicals, then it is unlikely that low doses of these chemicals are harming human health—the heart of the endocrine disruption theory. Although any hormone in the body could be subject to disruption, only estrogen mimics will be discussed here as they have by far received the most attention from both scientists and the public.
A Century of Change
The key date for the beginning of endocrine disruption is the end of WWII. This was when large quantities of known and suspected endocrine disruptors came into widespread use (Colborn, vom Saal, and Soto 378). However, there were other important changes over the past century—most notably in diet. In the US, daily carbohydrate consumption is down from 497 g in 1904 to 388 g in 1974. At the same time, total protein intake remained constant at around 100 g, but it went from being 50% animal and 50% vegetable to 70% animal and 30% vegetable (Gortner 3248). Total fat increased from 127 g to 158 g. Since 1940, pork and milk consumption dropped significantly while margarine, oil, and beef consumption increased markedly (Gortner 3251). Poultry consumption saw a "dramatic" increase with new techniques used to breed chickens on a mass scale (Gortner 3249). USDA figures also show that daily caloric intake has increased by 800 calories since the 1950s (Hunter, "America’s" 12). People are eating out more now (34% vs. 19% in the 1970s) and portions are increasing—for example, a fast food hamburger used to contain 1 oz of cooked meat in the 50s whereas today it is 6 oz (Hunter, "America’s" 12).
Another unprecedented change this century is the trend toward delayed childbirth. "For the first time in recorded history, the number of births for every thousand British women in their early 30s has recently exceeded that in women in their early 20s. The rate in older women is climbing too" (Gosden and Rutherford 1585). This paper will show that these and other lifestyle changes frequently offer a simpler and more compelling explanation for many problems that some would like to attribute to endocrine disruption.
Breast Cancer
Numerous ideas exist about how xenoestrogens may cause breast cancer. They may simply increase a woman’s overall estrogen exposure, they may act to increase the potency of endogenous estrogens, or prenatal exposure may cause permanent susceptibility (Colborn, Dumanoski and Myers 183). In the US, a 1% per year rise in breast cancer incidence has been documented since the 1940s (Feuer and Wun 1424). This is taken as evidence of endocrine disruption since women have been exposed to increasing amounts of hormonally active compounds since World War II.
However, the trend toward increasing breast cancer began well before WWII. Breast cancer incidence increased has from 1935 (with no prior data available) (Sullivan 1067). Despite this, mortality has been constant since the mid-1940s (Sullivan 1067-8). In Norway, the incidence trend has been increasing nearly steadily since 1916 (Tretli and Gaard 510). That the trend began well before the widespread use of suspect chemical indicates that other factors caused the trend initially and probably still account for the increase. Recent increases above the 1% trend can be attributed to better detection. In women over 40, rates have increased since 1982, but this is attributable to increased use of mammography (Feuer and Wun 1434; Wun, Feuer and Miller 142). In women aged 40-59, mammography may account for more than the reported rate increase; that is, there could even be a decrease in actual rates of incidence (Wun, Feuer and Miller 1433).
The idea that breast cancer can be caused by exposure to the strongly estrogenic DDT and PCBs has been disproven. Past correlations have been minimal (not even statistically significant) and based on six comparatively small epidemiological studies (Key and Reeves 1520). A much larger study (having as many women with breast cancer as all of the previous studies combined) showed no correlation between serum levels of DDE (a metabolite of DDT) or PCBs and breast cancer (Krieger et al. 589-90). Additionally, these samples were collected an average of 14 years prior to diagnosis (the time period from 1964-1971—a peak in exposure), which could better account for causality than the previous studies that collected samples at the time of diagnosis. The largest study to date, using blood samples taken from 1974 and 1989, showed no increase in risk corresponding with higher levels of DDE (Helzlsouer et al. 529). Blood samples taken from 1989-1990 also reveal that there is no correlation (Hunter, "Plasma" 1256). Recent studies of breast adipose tissue found no difference in the concentration of DDT and its metabolites in women with breast cancer and in women without breast cancer (Bagga et al. 751; Zheng et al. 455). This is particularly significant because even if studies using plasma or gluteal adipose tissue are discounted, it appears that even measuring the organochlorine concentration directly in the breast still fails to produce a correlation.
Additionally, because DDT is still used in other countries, one would expect skyrocketing rates of hormonally dependent cancers in developing nations. This is not the case. Furthermore, in a relatively small study of northern Vietnamese women, whose DDE levels are on average more than double that of US women, no correlation between DDE and breast cancer was found (Schecter et al. 454). DDT is also still in use in Mexico. A study there found no correlation between the concentration of DDE and breast cancer incidence, although DDE levels were similar to US women (López-Carrillo et al. 3730).
Another troubling point for endocrine disruption theory is that African American women have adipose DDE levels that are on average 74% higher than white Americans (Cocco et al. 3), yet rates of breast cancer are higher in whites (Zheng et al. 457).
In a large accidental exposure of humans (976 males and 1062 females) to PCBs, no increase was reported in the incidence of breast cancer, and there was actually a decrease in overall cancer mortality except perhaps for liver cancer (Hsieh et al. 417). Similarly, the results of a study on a population exposed to 2,3,7,8-tetrachlorodibenzo-para-dioxin (TCDD) as a result of an accident 15 years ago in Sevoso, Italy, show a decrease in breast, uterine and ovarian cancer (Bertazzi et al. 649). The weight of evidence concludes that increased exposure to DDT and PCBs does not increase breast cancer risk (Helzlsouer et al. 529).
In fact, women with the highest levels of DDE and PCBs consistently have the lowest risk of breast cancer (Cocco et al. 3, Dorgan et al. 7, Helzlsouer et al. 529, López-Carrillo et al. 3730, van’t Veer et al. 85). These decreases in breast cancer and the inverse association with DDE levels could ironically be explained in terms of endocrine disruption. Instead of increasing overall estrogen exposure, the less potent xenoestrogens bind to estrogen receptors without triggering transcription, thus decreasing the effect of endogenous estrogen exposure. Yet no one on either side of the question has seized upon these results as evidence of endocrine disruption. In Our Stolen Future, the authors initially point out that while DDT mimics estrogen, DDE depletes hormones in the body, including estrogen susceptibility (Colborn, Dumanoski and Myers 85), but later cited the studies available at that time that indicated increased DDE levels were associated with increased risk of breast cancer, especially in women with estrogen-responsive tumors (Colborn, Dumanoski and Myers 184). This highlights the ease with which endocrine disruption claims can be "proven" if the desired result is not found—the chemical must simply act in the opposite manner.
Proponents of endocrine disruption theory claim that even if DDE or some PCBs do not cause breast cancer, there are a myriad of other estrogen mimics that could cause breast cancer (Colborn, Dumanoski and Myers 184). But this type of dismissal belies the strength of their initial claim—that the estrogenicity of these compounds is to blame. DDE is a potent estrogen and if it does not cause breast cancer, this casts serious doubt on the idea that chemicals are dangerous simply because of their estrogenicity.
There is a simpler explanation for the long-term increase in breast cancer. An established risk factor for breast cancer is a diet high in fat, calories and protein and low in complex carbohydrates and fiber (Adlercreutz, "Diet" 281; Stephens 756). Additionally, "certainly the incidence is lower in women who adhere to a strict vegetarian diet" (Stephens 756). Clearly, the standard American diet increases women’s risk for breast cancer.
Unfortunately, this western diet is spreading worldwide, particularly to cities in developing countries where the traditional diet was low-fat vegetarian or semi-vegetarian. It is not surprising that western diseases like breast cancer would follow. Breast cancer rates in Japan increased following the introduction of a diet rich in meat, eggs and animal fat (Rose et al. 2367; Hirayama 179). A study from 1950-1975 revealed that increased breast cancer rates in Japan corresponded strongly with increases in non-fish animal fat (particularly strong with pork) (Kagawa 211; Hirayama 179).
In the past, breast cancer rates declined for Japanese women, while in the US that risk has always increased. The trend in Japan is now toward increased risk for postmenopausal women, but no data from either country indicates a change in the rates of younger women, which would suggest xenoestrogen involvement (Kodama, Murakami and Kodama 795). Regarding research back to 1970 as well as a plot of breast cancer incidence rates in Japan from 1959-1987, Kodama et al. conclude "There is no explanation feasible other than the Westernization effect" (Kodama, Murakami and Kodama 795). Data from England from the 1960s to mid-1980s reveals the same phenomenon—some increase in cancer incidence in postmenopausal women, but no change in the risk for young women (Kodama, Murakami and Kodama 795). These facts are inconsistent with a xenoestrogen explanation, but are consistent with the Westernization effect.
Numerous studies over the past 20 years have documented the phenomenon that migrants to the United States from areas of low breast cancer incidence tend to adopt the cancer rates of the new country (Dunn 3240; Stanford et al. 182; Hirayama 194). Immigrants from China and Japan, for example have breast cancer rates that are 50% lower than white American women, but only 25% lower for first generation US born Asian-Americans (Stanford et al. 182). After successive generations, the disparity between immigrant families and natives decreases even further (Stanford et al. 182), which corresponds to increasing Westernization of diet. An explanation based on chemical contaminates could certainly be devised, but their major weakness is in explaining why the rate of breast cancer continues to grow in successive generations. If the xenoestrogens are ubiquitous, the first American generation ought to be affected just as much as anyone else born here is.
Not only does the Westernization effect explain increases in breast cancer in other countries, but it also explains the steady increase in Western countries where the diet has been changing for the worse in terms of breast cancer risk factors. The marked halt in the trend of increasing breast cancer for women who experienced adolescence during WWII provides evidence for the diet-breast cancer relationship (Tretli and Gaard 511). During WWII in Norway energy intake decreased by 22%, especially with a drop in fat consumption. While the consumption of meat had increased up until the war, it "dropped substantially" during the war along with milk intake. There was increased intake of vegetables and fish as well as increased physical activity (Tretli and Gaard 511). Breast cancer is also correlated with height; the height of adolescent girls in Oslo steadily increased from 1920-1975, except for a decline during WWII (Tretli and Gaard 510). Interestingly, US breast cancer mortality declined from 1940-1946 (Sullivan et al. 1068).
Several mechanisms can explain why diet affects breast cancer. First, a high protein/carbohydrate ratio decreases the level of sex hormone binding globulin (SHBG) in the blood and enhances 2-hydroxylation of estradiol, resulting in higher estrogen concentrations than with lower protein/carbohydrate ratios (Adlercreutz, "Western" 6). The lowest protein/carbohydrate ratios are found in vegetarians and the highest in breast cancer patients (Adlercreutz, Western 6). Second, high fiber diets increase fecal weight and thus the amount of estrogens excreted while a diet higher in fat decreases the amount of estrogen found in feces. Additionally, a high fiber diet decreases the intestinal reabsorption of estrogen, and some fibers even bind directly to sex hormones (Adlercreutz, "Western" 5). Overall, a low fat-to-fiber ratio reduces the amount of biologically available estrogen, thus theoretically decreasing the risk of cancer (Adlercreutz, "Western" 6).
Finally, a separate lifestyle factor that could explain at least some of the rise in breast cancer is the trend toward delayed childbirth. The risk of breast cancer increases with age at first birth (Tretli and Haldorsen 215). Fifteen percent of Norway’s increase in breast cancer can be attributed to delayed childbirth (Tretli and Haldorsen 218).
Menarche
Xenoestrogens have been cited as the cause of the falling age of menarche, the age at first menstruation (Birnbaum 676). A potential mechanism is simply that high levels of estrogen trigger menarche in young women whose natural estrogen levels otherwise would not.
First, there is no longer even a trend toward earlier menarche; in Europe the trend has largely stopped (Tanner 160). In Denmark, for example, between 1966 and 1983, the trend toward lower ages stopped (Helm and Grolund 198). In Belgium, the average age was 13 in both 1965 and 1981 (Helm and Grolund 198). In America, the age of menarche in white girls has been stable at 12.9 since at least 1948, although there may have been a slight decline in the age for black girls (Herman-Giddens et al. 505).
Second, the claim that the trend post-WWII is due to increased exposure to xenoestrogens is refuted by the fact that the trend began by at least 1835—over 100 years before the widespread use of suspected endocrine disruptors. In Denmark from around 1835 to 1945, the age went from 17 to 14 in a linear decline. Then from 1950 to 1966 it went from 13.8 to 13.4 (Helm and Grolund 199)—the trend slowed when, according to the endocrine disruption hypothesis, it should have increased.
Finally, the trend toward earlier menarche can be best explained by rising standards of living—particularly in food quality. Prior to WWII, but less so after WWII, there was a sharp increase in height and weight of children in places like Europe, the US, Canada, and Australia (Tanner 157). Increased size alone can explain the drop in the age of menarche, given that it seems to be linked to attaining a critical weight or percent body fat. Higher levels of the hormone leptin, which increases with the amount of body fat, result in lower ages of menarche (Matkovic et al. 3243). Eating more protein and fat lowers the age, while a diet consisting mainly of plants and carbohydrates raises the age (Sanchez et al. 1341); vegetarian girls have a later menarche than their peers (Sanchez et al. 1342). In Japan, the age of menarche dropped from 15.2 in 1950 to 12.2 in 1974 (Kagawa 211). This corresponds to a time when some Westernization occurred to the Japanese diet: milk, meat, and fat intake increased respectively to 15, 7.5 and 6 times their previous levels (Kagawa 206). Adolescent girls went from being 133 cm tall and weighing 30 kg in 1900 to 148.5 cm and 41 kg in 1974, increasing slowly at first then rapidly after 1950 (Kagawa 210). Thus, it is clear that worldwide changes in diet have brought about changes in age at menarche before the introduction of xenoestrogens.
Endometriosis
Endometriosis is a disease characterized by lesions in other parts of the female body caused by cells that normally line the uterus; a major symptom is extreme pain. Estrogen is known to aid the spread and growth of endometrial cells. Factors such as smoking and exercise are associated with lowered risk of endometriosis, presumably because of the decreased estrogen levels caused by these activities (Eskenazi and Warner 253).
The supposed increasing incidence of endometriosis has been cited as an indication of endocrine disruption (Birnbaum 676, National Research Council 135). Some even went so far as to declare, "Prior to 1921, there were only twenty reports of the disease in the worldwide medical literature," absurdly implying that the disease did not even exist prior to the industrial era (Colborn, Dumanoski and Myers 181). This is false, however, as remarkably accurate and detailed reports of the symptoms and manifestations of the disease have existed throughout history, dating back to at least 1690, when even then, it was regarded as a relatively common disorder (Knapp 11).
In fact, the frequency of severe endometriosis has probably not changed since at least the 1940s (Vercellini and Crosignani 111), which, again, is particularly problematic since the vast majority of suspected xenoestrogens were not released until after 1940. The increase in mild cases of endometriosis is probably due solely to the use of naval laparoscopy, which can detect cases that would have otherwise gone undiagnosed (Vercellini and Crosignani 111).
If environmental estrogens do affect humans, one would expect to see an increase in severe endometriosis because endometrial cells require estrogen for proliferation and long-term survival (Bergqvist 33). These cells that once lined the uterus continue to react to changes in estrogen levels even once they are outside the uterus; the activation of these cells during menstruation is what causes the severe pain associated with endometriosis. Since "no association between endometriosis and prenatal DES exposure has been established" (Haney 3), there is probably no critical time during pregnancy where a dose of excess estrogens later causes endometriosis. Because lifetime estrogen (or xenoestrogen) exposure would be the cause of excess cases of endometriosis, there has been more than adequate time for a trend to emerge. Furthermore, case control studies reveal no association between incidence of endometriosis and plasma levels of 14 PCB congeners, 11 organochlorine pesticides, or dioxin (Lebel et al. 222, 225).
One explanation for any apparent or future increase in endometriosis could simply be that awareness about endometriosis on the part of both women and doctors has increased. Consequently, women are actually diagnosed with the disease instead of being dismissed as mental cases. Seventy percent of women who are ultimately diagnosed with endometriosis were told at some point that it was only psychological (Ballweg 444). Of 850 women who were referred as mental cases to a doctor who happened to be both a psychiatrist and a gynecologist, he discovered that 92% of the women actually had endometriosis (Ballweg 453).
Sperm Count
Perhaps the most important piece of evidence cited by endocrine disruption proponents is that of falling sperm counts (Sharpe and Skakkebæk 1392). Numerous mechanisms exist based on exposure to xenoestrogens in utero and up until puberty (Sharpe and Skakkebæk 1394-5). Falling sperm counts have been referred to as "the most dramatic and troubling sign that hormone disruptors may already have taken a major toll" (Colborn, Dumanoski and Myers 172). The dramatic nature of the claim has made it a key factor in influencing public opinion and public policy; Greenpeace’s slogan, "You are half the man your grandfather was," made it into Congressional debates. Because of its central role in the endocrine disruption controversy, the issue of sperm counts will be discussed in detail.
The idea that sperm counts are decreasing at all has been widely criticized. The only evidence of a global decline is a meta-analysis by Carlsen et al., which analyzed 61 studies that took place over 50 years from 1938-1990. They found a stunning decline in sperm counts—nearly 50% (Carlsen et al. 610). However, this study is fraught with uncertainty.
Many methodological problems have been noted such as publication selection bias, patient selection bias, method of sperm collection, small sample size of many studies, length between data collection and publication, use of an arithmetic mean, and use of a simple regression line (Farrow 1; Paulsen, Berman and Wang 1019; Olsen et al. 890). These variables are significant—take method of sperm collection for example. Some samples are collected at home, while others at a clinic. Stress is known to decrease sperm counts, so clearly, where the sample is collected could have a direct impact on sperm count. It is reasonable to assume that over time, as sperm banks became more popular, more of the later samples were collected in clinics. Additionally, sperm was counted using many different protocols and counting chambers (Fisch et al. 1012). Furthermore, "The men in these studies ranged in age from 17 to 64 years, the duration of abstinence was for the most part neither controlled nor recorded, and the mean sperm concentration varied threefold" (Sherins 327). Finally, sperm counts vary over course of the year, with high levels in February and March and low levels in September (Tjoa 454; Gyllenborg et al. 31). This is yet another variable that was not taken into account by the Carlsen study.
Subsequent reevaluation of Carlsen’s data by other researchers indicates that "nearly all of the observed decline in mean sperm count may be a consequence of the reduction of the lower reference value" (Bromwich et al. 21). That is, the minimum sperm count considered "normal" was lowered over the years. Earlier studies may have excluded "low" counts that were included in later studies.
Differences in recruitment of donors can also explain a decline in sperm count. As the number of prospective donors increases, as it has over the past 50 years, the sperm banks can be more selective (Fisch et al. 1012). Donor recruitment bias is significant. Over a 15-year period at the same facility, five studies requiring sperm samples were conducted. Comparisons between the mean sperm count in each study revealed statistically significant differences between the studies (although there was no overall downward trend for the time period) (Handelsman 2703). This highlights the bias of self-selected donors, rather than a true random sample of the population (which for the most part has been lacking in sperm count studies). Handelsman also stated, "It is notable that this large bias effect size within a single centre is comparable with the magnitude of the alleged effect size for decrease in sperm concentration over six decades, according to the Carlsen meta-analysis" (2703).
Adjustment for age and duration of abstinence (which were not controlled for in Carlsen’s analysis) shows no decline (Fisch et al. 1013). Abstinence is a very important variable, yet only 53% of the Carlsen studies provided data on the recommended length of abstinence (let alone the actual length of abstinence) (Olsen et al. 891). Furthermore, a large study comprising 1/3 of all the post-1971 data was conducted using men who were undergoing a vasectomy (Tjoa et al. 454). The men in this study were far less likely to comply with abstinence guidelines, since they had no interest in what their sperm count was (Olsen et al. 891). Abstinence periods could explain the decline in the 1960s. Birth control may have led to shorter abstinence periods, which is in accord with the decline in semen volume noted by Carlsen (Suominen and Vierula 1579). At least one study from 1955 noted a significantly longer abstinence period than is standard in present studies (Suominen and Vierula 1579).
If the statistical model of continuous linear decline in sperm count proposed by Carlsen is accepted, it would also indicate that sperm counts have been declining before 1940 and thus the phenomenon must be attributable to something other than or in addition to endocrine disruptors (Olsen et al. 887-93). There is also a "need for caution in using regression models to determine the statistical significance of the results, because the correlation coefficients were very small, indicating that variables other than the sperm concentration contributed to the determination of statistical significance" (Sherins 327). A stair-step model more accurately fits the data. This model indicates a rapid decline in sperm count in the mid-1960s (such as one due to shorter abstinence periods caused by the birth control pill) but no further declines (Olsen et al. 891). This leaves open the question as to what the trend looked like before 1940. Perhaps there was a spike in sperm counts due to WWII caused by any number of factors, such as longer periods of abstinence.
The Carlsen study and any attempt to analyze sperm levels suffer from the fact that there is very little data prior to 1970. Analysis of post-1970 data indicates no decline or a slight increase (Olsen et al. 891). This is very different from the linear decline that endocrine disruptor proponents promote to the public.
Geographical differences can also explain the apparent decline or at the very least, raise serious concerns about the effects of sample selection bias. New Yorkers have consistently had higher than average sperm counts (Fisch et al. 1013). In the data before 1970, 94% of the men were from the US and 87% of those from New York City. After 1970, 50% of the men were from the US and only 25% from New York City (Fisch et al. 1013). If the Carlsen data is analyzed without the large studies from New York, then there is no decline in sperm count (Lipshultz 910). An analysis of only the US data from the Carlsen data plus later US studies indicates no decline from 1938-1996 when New York and non-New York data are analyzed separately (Saidi et al. 460). If the distinction is ignored, however, a decline can be shown (Saidi et al. 460).
When well-controlled studies from one geographical location are studied, there is typically no trend toward lower counts. The downward trend seems only to emerge when a huge number of variables are introduced from many cross-geographical studies. A series of studies performed at a Los Angeles fertility clinic have revealed no differences in sperm count from samples taken in 1951, 1979, and 1994-97 (Acacio et al. 596). Similarly, a study of 20,411 infertile men from 1960 to 1996 in Northeast Spain revealed no decline in sperm count (Andolz, Bielsa, and Vila 734). A meta-analysis of six Finish studies from 1958 to 1992 revealed no decline (Suominen and Vierula 1579). Studies using sperm collected from the 1970s to the 1990s have noted an increase in four US locations and in Denmark (Paulsen, Berman and Wang 1019; Fisch et al. 1012; Gyllenborg et al. 30). A more recent Korean study revealed no decline from 1989 to 1998 (Seo et al. 194). A unique Danish study used men who were married to infertile women seeking in vitro fertilization to achieve a random sample of the general male population. The samples were collected from 1990-1996 and no associations were noted between year of birth and sperm concentration (Rasmussen 1060). For example, there was no difference in counts from men born before 1955 and those born after 1964 (Rasmussen 1060).
There are exceptions—some studies from one geographical location do reveal a decline. A study of Scottish men showed a decline in sperm count associated with year of birth (Irvine et al. 467). However, this study suffers from the fact that the older group contained men of proven fertility (they had fathered children) and the younger men contained men of unconfirmed fertility. In fact, 81% of those born before 1960 had proven fertility compared to only 5% of those born from 1970-74 (Irvine et al. 469). The younger men were more likely to contain some infertile men, which would drive down the mean sperm count for the younger men (Farrow 1). Furthermore, the results obtained were consistent with either an increase in sperm count with age or a decline in count associated with year of birth. The authors discounted the former, despite the obvious fact that older men probably have longer periods of abstinence, and thus higher sperm counts. Another study indicated that older men did have higher sperm counts (Raab 44; Joffe 44). Also, the younger group being undergraduates, were perhaps more likely to have lower sperm count due to alcohol and smoking (Eccersley 43).
A study of Parisian men reported a decline in sperm concentration from 1973 to 1992 (Auger et al. 281). Only fertile, accepted sperm donors were included, which excluded all fertile men who were rejected as sperm donors. However, the selection criteria or any changes in it over time were not noted (Lipshultz 910). "In addition, differences in age, abstinence before semen analysis, ejaculatory frequency, and the number of samples analyzed per person were not controlled for" (Sherins 327). This study too could suffer from the fact that older men in general have higher sperm concentration than younger men (Joffe 44). Rasmussen et al. note the importance of variation over even a short time period: "In our study, the average sperm concentration varies by >30% from year to year. The variation persists even after correction is made for differences in the age and number of patients, and the variation from one half-year period to another is greater than the entire difference for the period 1973-1992 in the study by Auger et al." (1062).
In case sperm counts really have fallen (or, at the very least, to add another variable that sperm researchers must control for) those who manufacture suspected endocrine disruptors have proposed their own creative explanation. Dow Chemical Company’s Health and Environmental Research Laboratory recently published a study proposing that the introduction of iodized salt is responsible for the drop in sperm count (Crissman 400). This would mean that sperm counts really have not fallen so much as returned to normal levels from abnormally high levels caused by iodine deficiency. Animal studies suggest that individuals who experienced iodine deficiency in the womb have higher sperm counts. The iodine deficiency leads to hypothyroidism, which causes a longer period of growth for sperm-producing Seritoli cells, so that ultimately those individuals possess a larger number of them. The elimination of iodine deficiency would mean that men do not develop additional Seritoli cells and therefore their sperm counts would fall. This hypothesis even explains the decline in the Parisian men because iodized salt was not introduced in France until 1952, in the middle of the years in which the men studied were born (Crissman 408). But it is also worth noting that "in 1923 William J. Hale, a chemist at the Dow Chemical Company in Midland, Michigan, made the first iodized salt in America" (Crissman 408).
Ultimately, there are an incredible number of variables associated with measuring sperm count. Even individual men experience a wide range of "normal" values. Terrifying claims of a universal, global 50% decline in semen quality are made to get people’s attention. And that is precisely what happened. Yet such claims are easily refuted. Those with an interest in preserving the status quo can use the evidence to dismiss the whole theory. Rather than making broad doomsday predictions, those who fear endocrine disruption ought to make highly conservative claims that cannot be refuted. They should want to make highly specific claims so than a very strong correlative link can be made. For example, one study of sperm counts in London noted a decline in sperm counts from 1978 to 1989. But upon closer examination, the decline was observed only among men who lived in a particular water district (Ginsburg et al. 230). Further research could lead to the identification of the responsible compound(s), which will certainly not turn out to be iodized salt.
Testicular cancer
The increasing rate of testicular cancer is widely cited as evidence of endocrine disruption (Sharpe and Skakkebæk 1392; National Research Council 168). The mechanism is not clear, but is probably the result of exposure in the womb (Sharpe and Skakkebæk 1394). Both high and low levels of estrogen during pregnancy pose an increased risk of testicular cancer. High levels of estrogen (found in older women, women with a large placenta, and women of low parity) are associated with the seminoma variety of testicular cancer. Low levels of estrogen (young women, low birth weight) are associated with non-seminoma (Ekbom and Akre 227).
One problem with claiming that the increasing rate of testicular cancer is caused by endocrine disruption is the paucity of data before WWII. In Australia for example, there is no reliable incidence data prior to 1950 (Stone et al. 211). Given that cancer rates could have very well been increasing prior to 1950 (just like menarche ages were decreasing prior to 1950), any claims based on rates of testicular cancer are inherently on questionable ground.
When there is data, it shows that the trend existed before the widespread use of the suspect chemicals. In Australia, mortality rates from testicular cancer increased from 1910 to 1969 and fell afterward (Stone et al. 211). (It is acceptable to use mortality rates for older data; it is only in the past 3 decades that improvements in treatment have allowed declines in mortality that do not correspond to declines in incidence) (Davies 930; Stone et al. 211). A study in England and Wales showed that testicular cancer mortality rates have been increasing since the beginning of the 20th century, with men born as early as 1881 being affected (Davies 930). Furthermore, it is not that there is no trend in the 19th century—there is simply a lack of data for that time period. A study of the US indicated that testicular cancer incidence has been increasing since 1935 among men aged 15-44, with no data prior to that time (Brown et al. 166). Brown actually noticed two trends, from 1935-50 there was an increase in seminoma, while in the 1970’s there was an increase in nonseminoma (169; Stone et al. 217). There were either two separate factors for the increases, or one factor has a different on men depending on when they are exposed. Thus, while testicular cancer rates are perhaps increasing due to xenoestrogens, there is definitely another factor that explains the earlier increase, and perhaps explains the present continuation of that trend as well.
In the US from 1973-1984 there was a 33% increase in testicular cancer among whites, but no increase among blacks (Van Den Eeden and Weiss 1553). This is particularly difficult to explain since blacks have higher concentrations of suspected endocrine disruptors in their bodies.
Once again, lifestyle factors could easily explain the rise in testicular cancer. It has been known for over 20 years that fat and meat intake corresponds to testicular cancer rates (Armstrong and Doll 627; Paulozzi, "Effects" 527; Sigurdson et al. 22). Similar to the breast cancer trend, there was a drop in testicular cancer in Denmark for men born during World War II (Møller, "Decreased" 1669). This is in accord with Paulozzi, who found a slightly higher correlation with fat consumption at age of birth, rather than at time of diagnosis ("Effects" 527). Ekbom and Arke point out that New Zealand, Switzerland and Denmark have the highest incidences of testicular cancer (226). New Zealand and Denmark have among the highest per capita meat and fat consumption rates (data on Switzerland was not provided) (Armstrong and Doll 625). Additionally, any study of testicular cancer confirms that rates increase with affluence (Stone et al. Trebling 218; Davies 931), which may or may not be related to fat consumption.
Another clear risk factor for testicular cancer is a sedentary lifestyle. Exercise decreases risk while sitting around increases the risk (Gallagher et al. 401; "Aetiology" 1397). The use of computers, televisions, and video games are all novel sedentary activities that appeared post-WWII. Physical activity has declined since 1950 (Hunter, "America’s" 13).
Perhaps most intriguing is that the risk of testicular cancer declines as the age of puberty rises (Gallagher et al. 404; "Aetiology" 1396). Also, not surprisingly, the age of puberty in places like Canada, the US, and Europe has decreased (Gallagher et al. 404), perhaps due to a similar mechanism as the decline in age of menarche, namely that higher standards of living bring about faster maturity. The lower age of puberty alone could explain the increase in testicular cancer rates ("Aetiology" 1397; Weir, Kreiger, and Marrett 257).
Male Genital Defects
Increases in hypospadias, a urethral opening on the underside of the penis, and cryptorchidism, undescended testes, are cited as evidence of endocrine disruption (Kelce and Wilson 202; Sharpe and Skakkebæk 1392). This idea is based on animal studies and a supposed increase in these disorders among sons who were exposed in utero to high levels of the synthetic estrogen DES.
There is uncertainty as to whether or not the rates of these problems in the general population are even increasing. A study that examined the rates based on groupings of countries according to gross domestic product (GDP) showed no increase anywhere in cryptorchidism since 1910 and a decrease in most countries since 1985 (Paulozzi, "International" 299-300). In the least affluent areas (China, Czech Republic, Hungary, Mexico, and South America) the rates of hypospadias have been stable since around 1974 (Paulozzi, "International" 300). In countries with the second lowest GDPs (Mediterranean nations and Ireland) there were no true increases, only an artifactual increase in Italy due to the initiation of a hypospadias study, and perhaps in Israel where the rate fluctuated wildly due to unknown reasons (Paulozzi, "International" 299). There were no increases anywhere following 1985 (Paulozzi, "International" 299). There have been increases in Scandinavia, Japan and two studies in the US, all of which are in the highest GDP grouping (Paulozzi, "International" 298, 299). However, there are no linear increases. The trend fluctuates up and down with a slight net increase. One of the US studies was stable from 1968 to 1982, there was an increase until 1985, and then it was stable again (Paulozzi, "International" 298). It seems apparent that these defects are affected by affluence, perhaps dietary as in the case of testicular cancer.
Any trend toward increasing hypospadias could be explained by a change in the definition to include more mild forms. If only severe forms, of which the definition has not changed, are examined, there is either no change or a decrease (Paulozzi, "International" 300). Also, improved documentation by physicians can account for increases in both hypospadias and cryptorchidism, especially given the emergence of the knowledge that undescended testes are more likely to become cancerous and are therefore given more attention (Paulozzi, "International" 301).
Cryptorchidism is associated with low birth weight and high maternal age (Møller and Skakkebæk 909). Thus the increasing age at which women give birth could lead to an increase in cryptorchidism. Paternal age may also be associated with increased risk but no studies have included information on paternal age.
Prostate Cancer
Because both human exposure to xenoestrogens and rates of prostate disease have increased, endocrine disruptor theorists believe there could be a causal relationship (Colborn, Dumanoski, Myers 180). In mice and rats at least, increased estrogen levels are associated with enlarged prostate glands and long-term exposure to estrogen is associated with prostate cancer (Colborn, Dumanoski, Myers 179). Furthermore, prostate cancer is a hormonally influenced cancer—endogenous hormones affect it and hormone therapy is used in treatment—therefore, it ought to be affected by exogenous estrogens.
Prostate cancer incidence rates suffer from the common problem of a lack of earlier data. There appears to be a long-term slow rise in incidence since 1935 (Sullivan et al. 1069). Mortality, however, has been constant or in slight decline since 1935 in whites, but there has been an increase for nonwhites (Hsing and Devesa 527; Sullivan et al. 1069). Better detection may have allowed more tumors to have been detected, but the lack of a corresponding increase in mortality could perhaps indicate that true incidence has remained the same. The sudden introduction of chemicals should have caused a birth year cohort effect, but there is none (Hsing and Devesa 527). Finally, the rapid increase in incidence in the early 1990s was a result of new screening programs that used prostate specific antigen (Dennis and Resnik 257).
Diet plays a role in causing prostate disease and could explain the increase in incidence. Global food consumption patterns and cancer incidence rates indicate that prostate cancer correlates also highly with total fat consumption (Armstrong and Doll 626). Following the introduction of a western diet in Puerto Rico and Japan, both countries experienced increased rates of prostate cancer (Rose et al. 2367). Cross-country comparisons show that increasing animal fat intake increases the risk of prostate cancer (Rose et al. 2366; Boyle and Zaridze 1057). Much like breast cancer rates, migrants quickly achieve the rates of the new country, whether those rates are higher or lower (Boyle and Zaridze 1048; Dunn 3244). Japanese men have the same amount of small and latent prostate tumors as men in the US, but US men have a much greater mortality from prostate cancer because the tumors grow instead of remaining dormant (Moyad 98; Armstrong and Doll 626). Even latent tumors are more proliferative in Japanese men living in Hawaii than those living in Japan are (Dunn 3242). Exposure to xenoestrogens should be comparable in Japan and America since they are both industrialized nations. The major difference between Japan and the US is that Japan still has a diet lower animal products than the US—although this is rapidly changing. Despite dramatic increases in animal product consumption beginning in the late 1940s, compared to the US in 1975, Japan’s consumption was "negligible"—1/6 the amount of meat, 1/5 the milk, and 1/2 the fat (Kagawa 206). It was studies at that time that initially revealed the differences in incidence of Western cancers.
Case-control studies, in which a group of men with prostate cancer is matched with a control group who does not, highlight any factors the groups do not have in common. Consistent with the geographical information, studies show that vegetable consumption lowers the risk of prostate cancer (Cohen, Kristal, and Stanford 63) while animal fat increases risk (Rose 2366). Furthermore, foods high in phytoestrogens are also associated with a decreased risk of prostate cancer (Strom et al. 22).
The Framingham study of 2,164 men revealed an additional factor that could contribute to the rise in incidence of prostate cancer—advanced paternal age at birth increases the risk for prostate cancer (Zhang et al. 1211). A potential explanation is that "continuous male germ cell division may increase the rate of mutation due to replication error" (Zhang et al. 1211). The trend toward later childbirth would inherently bring about an increase in prostate cancer.
Sex Ratio
Another factor cited as an example of endocrine disruption is the decline since 1950 in the sex ratio, which is the proportion of live newborn boys to girls (Møller, Trends 232; National Research Council 135). Often expressed as a percentage, the typical rate is 51% male. An increasing percentage of female infants could hypothetically be the result of higher estrogen levels. The major reason for this theory is that following the accidental exposure to high levels of dioxin in Sevoso, an unexpectedly high number of babies born were female (48 out of 74, or 35% male) (National Research Council 135). Although dioxin is not a xenoestrogen, it may interfere indirectly with the endocrine system.
A number of studies have indicated that sex ratios are falling. Data from the Netherlands from 1950-1995 shows a decline, although the author noted that values from 1950-54 were somewhat higher than the rest of the data (van der Pal-de Bruin et al. 62). Post-1950 data for England and Wales also shows a decline (Dickinson and Parker 227), as does US data from 1970-1990 (Allan et al. 40). Unfortunately, these studies, again, do not look back far enough and ignore significant historical events.
Data from Germany and the Netherlands from 1871 to 1995 revealed spikes in sex ratio immediately following the World Wars; similar patterns occurred in other combatant nations, while neutral nations experienced much smaller rises (van den Broek 805). Additional German data also revealed peaks at 1920 and 1945 with the level after 1945 slightly higher than the period from 1870-1914 (Bromen and Jöckel 804). That is, even if levels are falling today, they still have not fallen below what was the norm before 1920. Allan et al. plotted Canadian sex ratios from 1930-1990 and plotted a smoothed curve through the data. Unfortunately, the curve obscured the tremendous spike that occurred post-WWII. Additionally, by not knowing what the trend was prior to 1930, it is impossible to make judgements about the current downward trend. Canada was a major player in WWI, and undoubtedly experienced an earlier spike in sex ratios as European nations did. Plots from Denmark, Sweden, Norway and Finland from 1850 to 1996 reveal peaks at WWI and WWII (Møller, "Change" 232). The appropriate question is clearly not why is the sex ratio declining—it is why were there peaks following the World Wars from which we are seeing a return to normal levels. Claiming that the decline represents a problem is like worrying about a warming trend following an ice age—it would be a problem if the trend did not exist.
One possible explanation for the large increase in male birth following WWII is the baby boom. It has been shown that conception early in the menstrual cycle is likely to produce a boy, while fertilization at ovulation is likely to produce a girl. While individual couples cannot use this slight difference to control the sex of their child, the effect can be seen statistically in large groups (Møller, "Trends" 238). If we can assume that the baby boom was accompanied by an increase in the frequency of intercourse, then an increase in sex ratio would be the expected—and observed—result (van den Broek 805).
This same idea can be applied in reverse to the Sevoso incident. Following a major disaster, intercourse rates probably dropped dramatically, which could account for at least a portion of the high frequency of females who were born. For example, following the earthquake in Kobe, Japan, the sex ratio declined significantly 9 months later and there was a 6% decline in fertility as well (Fukuda et al. 2321). Low sex ratios have also followed floods, severe air pollution and other highly stressful events (Møller, "Trends" 235).
There is still an on-going slow decline in sex ratio that perhaps cannot be explained by the baby boom alone (although the sex ratio has not fallen below pre-WWI levels). Additional social changes may play a role in the continued stabilization/decline of the sex ratio. The use of fertility drugs that induce ovulation is known to produce a high proportion of girls (Allan et al. 40; Møller, Change 829). Additionally, women over 35 are less likely to have boys, so delaying childbirth would have the result of lowering the sex ratio (Dickinson and Parker 228).
Last edited: