Breast cancer is the most common cancer in women (Parkin et al. 2001). Incidence is highest in North America, Northern Europe, and Australia, where age-adjusted rates are 75-92 per 100,000 women (standardized to year 2000 world population), and lowest in Asia and Africa, where incidence is less than 22 per 100,000 (Parkin et al. 2001). Mortality has increased steadily from the 1960s until the late 1980s, when rates declined in many countries, including the United States (Parkin et al. 2001). Mortality continued to climb, however, for African Americans, whose mortality rates have exceeded the U.S. average since the 1980s (SEER 2002). Worldwide, breast cancer incidence continues to rise in all age groups, with an increase in U.S. age-adjusted incidence of more than 40% from the early 1970s to the late 1990s (Clegg et al. 2002; SEER 2002). An estimated 203,500 new invasive breast cancer diagnoses are expected in the United States this year, 54,300 in situ cases, and 45,000 deaths (ACS 2002). About 40% of new invasive cases are diagnosed in women younger than 60 years of age (ACS 1996), and breast cancer is the leading cause of cancer death among women 35-54 years of age (National Center for Health Statistics 1997).

The threat to women in mid life coupled with observations of substantial temporal and geographic variation and poor prediction of individual risk has prompted a search for modifiable risk factors. Because breast cancer risk changes over time and varies across geographic locations, factors associated with these variations may provide clues that can lead to prevention. Thus far, many correlates of risk have been identified, including a constellation of hormone-related reproductive factors. These factors account for a substantial portion of the variation in incidence, while also providing evidence that additional factors, probably modest in magnitude, remain to be discovered.

Taken together, epidemiologic studies of hormonal factors in breast cancer and animal studies of the hormonal activity and carcinogenic potential of certain synthetic chemicals suggest environmental pollutants as possible sources of risk. Compounds identified in laboratory studies as mammary carcinogens or hormonally active are in common commercial products and are ubiquitous pollutants to which women in industrial societies are widely exposed, so identifying effects on breast cancer has the potential for substantial public health impact, even if the relative risk associated with exposure is low.

In this article we identify promising leads in the study of environmental pollutants and breast cancer and the challenges in pursuing them. As background, we provide an overview of incidence trends and well-established and suggested breast cancer risk factors that inform environmental research. We review animal studies of chemicals that may be breast carcinogens, promote growth of breast cells and hormonally sensitive tumors, or affect mammary gland development and susceptibility. We assess current knowledge from the few epidemiologic studies of environmental pollutants, discuss the barriers to further progress, and identify research needs.

Background

Trends in incidence and mortality. The association between breast cancer risk and industrial development, historically and worldwide, is one indicator of modifiable risk. Increased access to mammography and other forms of screening is generally believed to play a role in rising incidence, particularly during the early to mid-1980s, but does not explain increases in risk before 1980 or increasing risk for younger and older women who are less likely to be screened or in developing countries with low screening rates (Ursin et al. 1994).

Currently, incidence is rising most rapidly in low-risk populations both internationally (Parkin et al. 2001) and in the United States (SEER 2002), suggesting that ongoing cultural change is a primary contributor. For example, incidence for Asian-American women at the beginning of the 1990s was 40% lower than for U.S. non-Hispanic white women but increased 19% by 1998 compared with 7% increase for non-Hispanic whites (SEER 2002).

In Los Angeles County, California, where ethnic diversity allows for more detailed analysis of trends in ethnic populations, incidence among non-Hispanic whites is 20% higher than for African Americans and roughly double the rate for Hispanics and Asian Americans; in contrast, the rates of change are highest among Asian Americans. Los Angeles County breast cancer incidence rose by 1.1% per year in 1993-1997 among non-Hispanic whites, 2.1% in Hispanics, and 4.6% in Asians, while declining by 0.3% for African Americans (Deapen et al. 2002). By the late 1990s, rates for women of Japanese and Filipino heritage were approaching rates for non-Hispanic whites.

Surveillance data for Asian-American women are consistent with studies of migrant populations showing that when women migrate from low- to high-risk countries and vice versa, their risk and the risk in successive generations change to approximate the levels in the destination country (Kliewer and Smith 1995). Further, a population-based case-control study of Asian migrants to California and Hawaii showed higher risk associated with longer residence in the United States (Ziegler et al. 1993); and for U.S.-born Asian women, the study showed higher risk for those with more U.S.-born grandparents, an indicator of acculturation. The relative risk associated with migration changed only slightly after controlling for menstrual and reproductive factors, providing evidence that other factors contribute to migration effects (Wu et al. 1996).

Although migration studies provide insight into the contribution of sociocultural factors and support the idea that heritable factors are not predominant determinants of breast cancer risk, studies of heritable genes add a complementary perspective. Mutations in the breast cancer genes BRCA1 and BRCA2 are estimated to account for fewer than 10% of cases (Claus et al. 1996), although additional genes that affect hormone synthesis and metabolism and DNA repair likely add to heritable risk (Martin and Weber 2001). The effect of the broader range of heritable genes is seen in studies of identical (monozygotic) and fraternal (dizygotic) twins. In a study of 45,000 twin pairs, 14% of monozygotic twins and 9% of dizygotic twins were concordant for breast cancer diagnosis (Lichtenstein et al. 2000), and Mack et al. (2002) reported slightly higher concordance.

Reproductive and other previously studied risk factors. The fact that reproductive characteristics affect breast cancer risk has been known since 1700, when Ramazzini reported higher incidence among nuns (Spratt et al. 1995). Factors now known to confer higher risk include older age and being female, younger at menarche, older at menopause, nulliparous, and older at a first live birth or stillbirth; whereas higher parity, longer lactation, and bilateral ovariectomy are protective (Davis et al. 1997; Kreiger et al. 1999; Parazzini et al. 1997).

Reproductive risk factors are associated with exposure to estradiol, progesterone, and other hormones; and reproductive hormones are also believed to underlie increased risk associated with alcohol consumption, lack of physical activity, higher body mass index and weight gain after menopause, and low premenopausal body mass index (Bernstein et al. 2002). In addition, recent studies provide some evidence that in utero hormonal exposures characteristic of certain pregnancies affect breast cancer risk in the offspring. Daughters exposed to lower hormone levels in pregnancies with toxemia or pre-eclampsia are at lower breast cancer risk, whereas higher hormone levels in pregnancies with twins result in higher risk (Bernstein et al. 2002). This is a new area of research with some inconsistencies within the limited number of studies completed.

Pharmaceutical hormones similarly affect risk. Both estrogen-only and estrogen-progesterone hormone replacement therapy (HRT) for postmenopausal women increase breast cancer risk. In a pooled analysis of 51 studies involving about 54,000 postmenopausal women, the relative risk of breast cancer for women with at least 5 years of recent use was 1.35 [95% confidence interval (95% CI), 1.21-1.49] (Collaborative Group on Hormonal Factors in Breast Cancer 1997). Women who stopped using HRT more than 5 years before were not at higher risk. Additional large-scale population-based epidemiologic studies show 10% increased risk after 5 years of use for estrogen alone and 40% after 15 years, and 30% increased risk for less than 5 years of use for combination HRT (Bernstein et al. 2002). In a clinical trial of combination HRT versus placebo, the Women's Health Initiative reported a hazard ratio of 1.26 (95% CI, 1.00-1.59) about 5 years after enrollment and higher risk for women with prior HRT use up to a hazard ratio of 1.81 (95% CI, 0.6-5.43) (Women's Health Initiative Investigators 2002). For oral contraceptives, recent, but not long-term, use is associated with higher risk (Bernstein 2002), with about 26% increased risk for current users (Collaborative Group on Hormonal Factors in Breast Cancer 1996). Additional information will become available as more women with long-term oral contraceptive use reach the ages of higher breast cancer risk. Diethystilbestrol (DES), a potent synthetic estrogen, has been linked to increased breast cancer risk in women who took DES during pregnancy (Colton et al. 1993; Titus-Ernstoff et al. 2001).

Diet seems very likely to affect breast cancer risk, as it does in animals, but epidemiologic studies have failed to identify specific dietary constituents that increase or decrease risk. Effects of fat and fruits and vegetables have been extensively studied, so far providing no consistent evidence of dietary risk factors (Gandini et al. 2000; Holmes et al. 1999; Hunter and Willett 1996; Michels 2002; Smith-Warner et al. 2001; Willett 1999). High soy intake in Asia has been proposed as a factor in reduced breast cancer rates there, although epidemiologic studies so far provide limited evidence of a protective effect (Adlercreutz 2002; Hilakivi-Clarke et al. 2001; Trock et al. 2000). One recent study of Asian Americans reported a protective effect for soy that was most pronounced for high soy intake beginning in adolescence (Wu et al. 2002), and this study illustrates newer approaches to diet that explore possible effects of the timing of exposure. Other new approaches focus on possible interactions of multiple aspects of diet, for example, alcohol and folate (Feigelson et al. 2003; Zhang et al. 2003), or between diet and genetic polymorphisms (Zheng W et al. 2002).

Ionizing radiation is a clearly established environmental cause of breast cancer (NRC 1990). Studies of atomic bomb survivors and women exposed to X-ray medical treatments in childhood indicate that exposures early in life impart greater risk than adult exposures. In studies of exposed Japanese women 35 years after the atomic bomb, risk of breast cancer was 4-fold greater in women younger than 4 years of age and 2-fold greater in women 10-14 years of age compared with women 20-30 years of age at the time of the bombing. Women younger than 40 years of age had a greater risk than those older than 40 at the time of bombing (Land 1995; Tokunaga et al. 1987).

Higher socioeconomic status (SES), usually measured by education level and income, is consistently associated with higher breast cancer risk, although education and income clearly are not themselves causal. This relationship is often seen even after controlling for breast cancer risk factors such as parity and age at childbearing, which are themselves associated with SES. The possibility that some part of this relationship is due to chemical exposures, for example, from use of consumer products and pesticides, warrants further study. In a small exploratory survey of breast cancer risk factors in high- and low-incidence neighborhoods, higher SES women reported significantly higher use of several different pesticides (home and lawn chemicals, repellents, and lice control) and of dry cleaning (Maxwell et al. 1999).

Role of previously studied risk factors in incidence patterns. Women diagnosed with breast cancer, as with other diseases, often ask themselves, Why me? In recent years, communities with high incidence have struggled with that question as well. A few studies have tried to address these questions at both the individual and population levels, and these studies are interesting because unexplained variation can motivate and inform studies of new hypotheses.

At the individual level, Gail et al. (1989) developed a model that predicts risk from a woman's age, age at menarche, age at first live birth, number of previous biopsies, and number of first-degree relatives with breast cancer; and this model has been used, among other things, as a basis for identifying women considered high risk as candidates for chemoprevention trials of treatments such as tamoxifen and raloxifene. Using data on breast cancer incidence and risk factors in two large national surveys, Madigan et al. (1995) estimated that 41% of breast cancer risk in the United States is explained by later childbearing, nulliparity, higher income, and family history of breast cancer.

Regarding geographic patterns within the United States, mortality is highest in the Northeast and West and intermediate in the Midwest compared with the South (National Cancer Institute et al. 1999). Sturgeon et al. (1995) reported in an ecologic analysis that recognized breast cancer risk factors accounted for nearly all regional variation in mortality among women younger than 50 years of age; however, among older women, adjustment reduced excess incidence by 50% for the Northeast and Midwest and 10% for the West compared with the South. A similar analysis of the Nurses' Health Study improved on the Sturgeon et al. method by adjusting at the individual level rather than regional level for established risk factors (Laden et al. 1997). However, little variation in breast cancer risk across regions was observed either before or after adjustment, perhaps due to the relative homogeneity in the risk-factor profile of nurses nationwide, so results are not informative.

The extent to which known breast cancer risk factors account for geographic variation is a subject of particular interest in areas such as Cape Cod, Massachusetts, and Marin County, California, where incidence is higher than in a comparison population such as the entire state. Surveillance data show about 20% higher risk on Cape Cod in 1982-1994 (Silent Spring Institute 2000), and case-control data from a statewide study (the Collaborative Breast Cancer Study) show about 20% excess risk for Cape Cod women older than 50 years of age compared with others in Massachusetts, after controlling at the individual level for many recognized and hypothesized breast cancer risk factors (Silent Spring Institute 1998).

In Marin County, where elevated rates of breast cancer were first reported in the 1990s, incidence increased 6 times faster than statewide during the 1990s, rising 3.6% per year (Clarke et al. 2002). A comparison of Marin County with California census block groups that were comparable for census characteristics associated with breast cancer risk showed similar incidence rates in block groups with similar percentage white population, urban status, average parity, median household income, percentage with a college degree, percentage with a working class occupation, and percentage below the poverty line (Prehn and West 1998). Another study reached similar conclusions but relied on risk factor data for women 20-55 years of age, an age group unlikely to be representative of most women with breast cancer, who tend to be older (Robbins et al. 1997). Analysis of demographic factors is not a stopping point for analysis of rate variations, however, because the SES variables are not explanatory for disease.

Aside from the role of established breast cancer risk factors, higher rates of screening mammography could contribute to higher reported incidence in a region. For both Cape Cod and Marin County, available evidence from patterns of stage at diagnosis (based on the expectation of more early-stage diagnoses with mammography) and surveys of mammography use, although not conclusive, is on the whole not consistent with screening as an explanation for higher incidence (Clarke et al. 2002; Silent Spring Institute 1998).

An earlier experience in Marin County illustrates the public health value of drawing etiologic clues from geographic variation. Rapidly increasing incidence of endometrial cancer in Marin County and other affluent neighborhoods in the San Francisco Bay Area led to the identification in the 1970s of estrogen HRT as a causal factor (Austin and Roe 1979).

Insights from Animal Studies

Epidemiologic studies that consistently show increased risk associated with multiple sources of exposure to endogenous and pharmaceutical estrogen and other hormones strongly point to the hypothesis that hormonally active agents in commercial products and pollution also increase risk. Studies in laboratory animals, in vitro assays, and wildlife provide further evidence of mechanisms for effects of environmental pollutants on breast cancer risk through exposure to compounds that mimic or disrupt hormones that promote or inhibit tumor growth, act as breast carcinogens, or affect the development and vulnerability of the breast. Although the processes by which breast cancers develop are poorly understood, a review of the primary features of mammary gland development and the effects of hormones and chemicals on mammary gland carcinogenesis in animal models shows that the mechanisms that underlie the recognized risk factors for breast cancer in humans are also seen in animal studies. This section outlines current research related to biological mechanisms for breast cancer, including chemical and hormonal factors and the hypothesis that hormonally active chemicals--also known as endocrine disruptors--affect breast cancer. This information provides the essential scientific foundation for evaluating existing hypotheses about environmental factors in breast cancer and generating new hypotheses and directions for future research.

Mechanistic models for cancer. Historically, carcinogenesis has been characterized by three separate stages: initiation, promotion, and progression. Although the process of carcinogenesis is now recognized as more complex than this simple model suggests, the three-stage model still provides a useful paradigm by which chemicals can be described based on a potential mechanism of action (Barrett 1993; Pitot et al. 2000). Initiation is characterized as an irreversible change in a cell, very probably a genetic change or mutation, resulting in a latent neoplastic cell (Appel et al. 1990; Pitot 1993; Pitot and Dragan 1991). Promotion is the process by which an initiated cell expands clonally into a visible, benign tumor (Barrett 1993). Experimental evidence demonstrates that chemically modulated promotion of a cell requires repeated exposure; endogenous estrogen is thought to affect the process of mammary carcinogenesis primarily by this mechanism. Progression is the term used to describe the irreversible transition from a benign to malignant tumor, which involves additional genetic events, although not necessarily point mutations in DNA (Barrett 1993; Pitot 1993; Pitot and Dragan 1991).

Agents that are carcinogens are often genotoxic, or able to damage DNA. Both initiation and progression steps involve some level of genotoxicity, whereas tumor promotion more typically involves stimulation of cell proliferation. Many agents stimulate cell proliferation, and there is controversy over whether these should be considered carcinogens unless they can also induce some level of genetic damage (Alden 2000; Klaunig et al. 2000). Of course, increasing cell proliferation also increases the opportunity for spontaneous mutations, so even promoters can have some impact on DNA integrity.

Another model for carcinogenesis focuses on cell-cell interactions that maintain tissue organization in normal tissue and break down in carcinogenesis (Sonnenschein and Soto 1999). The role of stromal cells in inhibiting or promoting carcinogenic progression in breast epithelia is an ongoing area of research (Barcellos-Hoff 2001; Barcellos-Hoff and Ravini 2000; Mueller et al. 2002), and this work suggests that the study of chemical carcinogenesis must consider effects on cell signaling as well as traditional genotoxic effects.

Mammary gland development and susceptibility. The breast is one of the few organs that is not fully developed at birth. It reaches its fully differentiated state only through the hormonal stimuli induced by pregnancy and lactation, resulting in portions of the life cycle with increased susceptibility to carcinogens. Aspects of development that are known to affect gland susceptibility include rates of cell proliferation, stages of cell differentiation, and prenatal imprinting of hormonally sensitive tissues.

Greater susceptibility to genotoxic agents is expected during periods of rapid breast cell proliferation, such as prenatal, perinatal, and pubertal time periods and during pregnancy (Russo and Russo 1996; Wolff et al. 1996). Rodent studies of dimethylbenzanthracene (DMBA)-induced mammary tumors have shown a greater number of tumors and shorter latency when the carcinogen is administered to immature animals (Dunnick et al. 1995). Similar findings of increased risk for earlier age at exposure are observed in human studies of atomic bomb survivors (Tokunaga et al. 1987).

In addition to susceptibility during periods of cell proliferation, the susceptibility of the mammary gland to carcinogen exposure decreases after the first full-term pregnancy, when formerly undifferentiated cells have developed into fully differentiated cells, which are less susceptible to genetic damage and subsequent propagation of the damaged cell (Neumann et al. 1996; Russo and Russo 1996; Wolff et al. 1996). Epidemiologic studies have consistently shown that early age of first full-term pregnancy is a protective factor for breast cancer, and studies in animal models demonstrate that virgin rats are significantly more susceptible to chemically induced mammary gland cancers than are age-matched parous rats, which are relatively resistant to tumors (Brisken 2002; Russo and Russo 1998). Indeed, ductal and lobular carcinomas tend to originate from undifferentiated cells, whereas benign breast tumors tend to originate from the more differentiated cells (Russo and Russo 1996). Characterizing the specific hormonal factors that are responsible for the refractoriness of mammary glands postpregnancy is a topic of ongoing research (Brisken 2002; Sivaraman and Medina 2002).

Because the breast is particularly susceptible to carcinogen exposure up until the first full-term pregnancy, there may be an interaction between risk associated with age at first pregnancy, an established breast cancer risk factor, and risk associated with chemical exposure. In other words, in a hypothetical group of women with similar lifetime exposures to a mammary carcinogen beginning in childhood, those who were youngest at their first full-term pregnancy would experience the lowest increase in risk, and those who were oldest would experience the greatest increase in risk.

In addition, a number of studies in humans and animal models suggest that the in utero environment affects subsequent breast cancer risk in offspring (see preceding discussion of human studies). Animal studies have shown that administration of estradiol or DES during pregnancy increases breast cancer rates in female offspring (reviewed in Hilakivi-Clarke et al. 2001). One mechanism that has been proposed involves imprinting of mammary gland tissues in utero, resulting in an effect on the responsiveness of the tissues to estrogen later in life.

Hormonal factors in mammary carcinogenesis. Throughout the life cycle, the hormonal environment plays a critical role in the development of breast cancer. Removal of both ovaries reduces risk, and increased risk has been observed for women with higher levels of endogenous and pharmaceutical estrogen exposure (Henderson and Feigelson 2000). In animal studies, treatment with chemical carcinogens does not produce mammary tumors in the absence of endogenous hormones (Russo and Russo 1996, 1998). In other words, animals that have had their ovaries removed do not develop mammary tumors even after exposure to carcinogens. Supplementing animals with extra estrogens produces tumors even in the absence of specific chemical exposures (Russo and Russo 1996, 1998). These findings are consistent with the idea that estrogens are promoters of mammary tumors, which act over a long period of time by causing cell proliferation and clonal expansion of initiated cells. In addition, estrogens appear to be required for mammary carcinogenesis to occur.

Studies of normal mammary gland development and chemically induced mammary carcinogenesis in animal models have provided useful information for clarifying how the interplay of ovarian, pituitary, and placental hormones, while influencing the structure, organization, and function of the mammary gland, modulate its response to chemical carcinogens. Many hormones and growth factors have been demonstrated to affect the tumorigenic response of rats to genotoxic mammary carcinogens, including ovarian, placental, pituitary, and thyroid hormones, as well as androgens, insulin, and many growth factors (Brisken 2002; Neumann et al. 1996; Russo and Russo 1998; Sivaraman and Medina 2002; Swanson and Unterman 2002). In human studies, androgens and insulin-like growth factor 1 have been shown to be associated with risk of breast cancer (Toniolo et al. 2000; Wang et al. 2000).

Some researchers characterize certain estrogens, including the primary active endogenous estrogen 17ß-estradiol, common pharmaceutical estrogens, and the synthetic estrogen DES, as carcinogens on the basis of their significant role in hormonally mediated cancers in humans and animals (Tsutsui and Barrett 1997). Others do not consider endogenous hormones to be carcinogenic themselves but acknowledge their role as promoters of carcinogenesis because they allow neoplastically transformed cells initiated by other carcinogens to establish and grow by modifying the target tissue (Russo and Russo 1996, 1998). In addition to acting as promoters, DES, 17ß-estradiol, and certain metabolites of 17ß-estradiol, including 16ß-hydroxyestrone, have been shown to exhibit specific types of genotoxic activity under certain conditions (Liehr et al. 1990; Telang et al. 1992; Tsutsui and Barrett 1997). Steroidal estrogens are listed as known human carcinogens in the Report on Carcinogens, Tenth edition by the U.S. National Toxicology Program (NTP 2002).

Table 1

Chemical factors in mammary carcinogenesis. Experimental studies in animals offer an alternative means for identifying potential carcinogens in the environment, given that epidemiologic studies require a large number of women, a long duration, and adequate exposure information. The NTP has studied the carcinogenic potential of about 500 chemicals in animal carcinogenicity bioassays. Of these chemicals, 42 caused mammary tumors in the tests (Bennett and Davis 2002; Dunnick et al. 1995). These are listed in Table 1, along with information about their common uses. These chemicals include halogenated chemicals and solvents, including components of gasoline; aromatic amino/nitro compounds; dyes; and epoxides. Other research organizations that have conducted animal carcinogenicity bioassays on specific chemicals have identified about 160 additional chemicals as mammary carcinogens (Wolff et al. 1996). These include, for example, products of combustion [polycyclic aromatic hydrocarbons (PAHs), nitro-PAHs], ionizing radiation, common industrial solvents and other industrial chemicals (vinyl chloride, vinyl fluoride, vinylidene chloride, styrene, acrylamide), pesticides (atrazine, dichlorvos), and other substances (IARC 1999; Pinter et al. 1990). Many of the chemicals identified as mammary carcinogens in these bioassays also show evidence of genotoxicity. For example, in their review of 34 chemicals identified as mammary carcinogens by the NTP, Dunnick et al. (1995) report that 26 showed evidence of mutagenicity in the Salmonella assay.

Chemicals identified as mammary carcinogens in animal studies are priorities for follow-up study in humans. Only four of the 42 chemicals tested by the NTP (benzene, 1,3-butadiene, ethylene oxide, C.I. acid red 114) have adequate human evidence of carcinogenicity to be classified as carcinogenic in humans (NTP 2000). Although the breast is not the primary tumor site for any of these four chemicals, many of the human cohorts studied were all or predominantly male, and some limited epidemiologic evidence supports the breast as a tumor site for ethylene oxide (the sterilant) and benzene (in gasoline) (see additional discussion further below) (Hansen 2000; Petralia et al. 1998; Tompa et al. 1999). In addition, some animal mammary carcinogens identified in other testing programs also have epidemiologic evidence of breast cancers from occupational studies, including, for example, methylene chloride, PAHs, and chlorinated solvents (Hansen 1999, 2000; IARC 1999; Petralia et al. 1999).

Table 2

Potential role of hormonally active chemicals. Recent research sheds light on a class of hormonally active chemicals, referred to as endocrine disruptors, that may affect breast cancer primarily by promotional mechanisms, as well as by affecting mammary gland development and responsiveness to other carcinogens. The hypothesis has been put forward that exposure to endocrine disruptors, including chemicals that mimic estrogens, might play a role in breast cancer risk (Davis et al. 1993). To date, more than 500 chemicals have been found to be weakly estrogenic in various assays, including many chemicals in common use, such as constituents of detergents, pesticides, and plastics (Jobling et al. 1995; Nishihara et al. 2000; Soto et al. 1995). Table 2 lists selected classes of these chemicals, specific examples, and common uses. Many of these chemicals have been shown to mimic estrogen in a variety of short term in vitro assays; they bind the estrogen receptor, initiate transcription of estrogen-regulated genes, and can stimulate breast cancer cells in vitro to proliferate (Korach and McLachlan 1995; Shelby et al. 1996; Soto et al. 1995). Short-term in vivo assays, such as increase in uterine weight in rodents, are also used to demonstrate estrogenic activity (O'Connor et al. 1996). In addition, effects of these compounds have been frequently observed in wildlife; for example, widespread sexual disruption of wild fish has been reported in rivers receiving wastewater effluent, which contains a mixture of endogenous and pharmaceutical estrogens and industrial chemical endocrine disruptors (Jobling et al. 1998).

As research in this area continues to identify estrogenic compounds, significant questions are raised about how to evaluate the potential adverse health effects (Rudel 1997). These questions are far from being resolved. On the one hand, the potency of many of these endocrine-disrupting pollutants is typically much lower than the potency of endogenous estrogens, and so it has been proposed that their effects will be insignificant (Safe 1995). On the other hand, there is particular concern about the effects of endocrine-disrupting chemicals for exposures that take place when levels of endogenous hormones are very low, such as in utero or during prepubertal, or postmenopausal time periods. Also, a number of studies have demonstrated that multiple estrogenic chemicals can act together to produce an effect even when each individual component of the mixture is present below a threshold for effect, so these pollutants can act in combination (Silva et al. 2002). Finally, comparison of the in vivo estrogenic effects of a range of compounds demonstrates that estrogenic compounds exhibit diversity in both mechanism and effects (Gould et al. 1998; Rudel 1997). This diversity is attributed, at least in part, to the fact that the shape of the estrogen receptor ligand (either estradiol or an endocrine disruptor) affects the binding of the receptor-ligand complex to DNA sequences and subsequent gene expression. Current research into pharmaceutical selective estrogen response modifiers (SERMs) for menopause and breast cancer prevention is an outgrowth of this phenomenon (Emmen and Korach 2001). Recent discovery of a second estrogen receptor, ER-ß, complicates matters further because many hormonally active compounds have differential binding affinities for the two receptors, and cellular responses to such stimuli are difficult to predict (Pennie et al. 1998). Thus, just because two estrogenic chemicals cause a similar effect on one outcome (e.g., uterine weight) does not mean they will cause a similar effect on all estrogen receptor-mediated outcomes.

It is of particular interest that certain dietary constituents that have been hypothesized to be preventive of breast cancer, such as genistein in soy, are also estrogenic in many endocrine disruptor screening bioassays (Adlercreutz et al. 1995). As discussed above, the relationship between soy food intake and breast cancer risk in humans is controversial. In animal studies, genistein treatment often, but not always, reduced the rate of breast cancer, with the effect being strongest with treatment before puberty (Hilakivi-Clarke et al. 2001). It is hypothesized that the genistein treatment before puberty mimics the effect of an early pregnancy (this effect has been demonstrated with estradiol also), thus reducing the susceptibility of the mammary gland to carcinogenesis (Hilakivi-Clarke et al. 2001). Additional data from animal and in vitro studies suggest that phytoestrogens such as genistein have mixed estrogen agonist/antagonist activity and can inhibit the biological response to endogenous estrogens, although this apparent antagonist action may not take place directly via the estrogen receptor or may be due to the differential binding of genistein to ER- and ER-ß (An et al. 2001; Ford 2002; Fotsis et al. 1993; Lamartiniere et al. 1995; Markaverich et al. 1995; Po et al. 2002). This remains an active area of research.

Another new and important area of research related to hormonally active chemicals concerns imprinting of the mammary gland from in utero exposures to hormones or hormonally active chemicals. As discussed above, animal studies and limited human studies have shown that in utero exposure to estradiol or DES increases mammary tumor formation in the offspring (reviewed in Hilakivi-Clarke et al. 2001). In experiments related to dietary constituents, maternal intake of fatty acids and genistein, but not soy, increased DMBA-induced mammary carcinogenesis in the offspring (even though the soy diet increased pregnancy estrogen levels) (Hilakivi-Clarke et al. 2001). Limited research has been conducted on the effects of in utero exposures to environmental chemicals on mammary gland development and carcinogenesis (reviewed in Birnbaum and Fenton 2003). However, two studies of in utero exposure of rats to 2,3,7,8-tetrachlorodibenzo-p-dioxin (2,3,7,8-TCDD) show effects on mammary gland development, and one shows increased susceptibility to chemically induced mammary tumors (Brown et al. 1998; Fenton et al. 2002). In addition, increased susceptibility to chemically induced mammary tumors was observed in one study of a mixture of organochlorines [OCs; e.g., dichlorodiphenyltrichloroethane (DDT), dichlorodiphenyldichloroethylene (DDE), polychlorinated biphenyls (PCBs)] given neonatally to rats (Desaulniers et al. 2001), and gestational exposure to atrazine and bisphenol A have also been shown to affect mammary gland development in rodents (reviewed in Birnbaum and Fenton 2003). It is interesting to note that all of the compounds that have been shown to affect mammary gland development after gestational exposure possess some type of direct endocrine-modulating activity (e.g., estrogen agonist, androgen antagonist, etc.).

Endocrine disruptors can also act indirectly, for example, by up- or down-regulating the enzymes that metabolize endogenous estrogens or by affecting synthesis of endogenous hormones (NRC 1999). For example, effects of alcohol on breast cancer are hypothesized to be due to a variety of impacts on cellular signaling pathways, including increased circulating estrogen and androgen levels (Ginsburg et al. 1995; Singletary and Gapstur 2001). Although the focus of research in this area has been on measuring circulating serum or urinary levels of endogenous hormones, it is important to note that human breast tissue can metabolize hormones and create its own local hormonal environment independent of circulating levels (Adams 1991; Adams et al. 1992). Thus, effects of chemicals on the local hormone environment in the breast may be more relevant than effects on circulating hormone levels.

Overall, studies in lab animals, in vitro assays, and wildlife help characterize factors that influence breast development and carcinogenesis. These insights in turn inform hypothesis generation for human studies and help interpret findings in these studies. Toxicological research is a critical avenue for achieving breast cancer risk reduction because occupational epidemiology provides little information on women's cancers (see next section). Priorities for toxicologic research are outlined in the final section of this article.

Human Epidemiologic Evidence

Occupational studies. Despite the strength of toxicologic evidence for effects of certain pollutants on breast cancer risk, very little human evidence has accrued. In other areas of cancer research, leads from the laboratory often are first translated into human research in occupational studies where exposures are higher and better characterized compared with community settings, but few occupational studies have included women, so this resource is limited for evaluating breast cancer risk.

Elevated incidence has been observed repeatedly among women in white-collar jobs, due partly to reproductive risk factors, such as later childbearing, that are associated with the higher educational attainment required in these jobs and with higher SES more broadly. In some studies, associations are seen for white-collar jobs after controlling for SES and other possible confounders. For example, Band et al. (2000) observed elevated risk for teachers and medical workers. Calle et al. (1998) reported elevated risk for executives and secretaries but not teachers, librarians, or nurses, in a study that included a crude measure of physical activity, a potentially important source of confounding in studies of occupation and breast cancer. White-collar jobs do involve chemical exposures that may be related to breast cancer, including exposures to indoor pesticides, solvents, second-hand tobacco smoke, and flame retardants (Spengler et al. 2000), but these exposures are so poorly understood that most white-collar job categories are not informative with respect to questions about environmental pollutants.

Few studies have investigated breast cancer risk for women in occupations with more obvious chemical exposures, even among nurses, many of whom have substantial chemical exposures and for whom a large prospective cohort study is already in place (Nurses' Health Study 2002). Nurses are likely to have been exposed to the mammary carcinogen ethylene oxide (NTP 1998), which is used to sterilize medical equipment, and to hormonally active compounds, including nonylphenol (used in detergents and plastics) and bisphenol A (used in polycarbonate plastics) (Aschengrau et al. 1998). Two studies (Norman et al. 1995; Tompa et al. 1999) provide weak evidence of an association between ethylene oxide and breast cancer among nurses.

A few studies provide evidence of breast cancer risk associated with exposures to the mammary carcinogens benzene, PAHs, and certain organic solvents. Hansen (2000) reported higher risk of breast cancer for men exposed to gasoline and vehicular combustion products, benzene, 1,2-butadiene, 1,2-dibromoethane, 1,2-dichloroethane, and PAHs. With a lag time of at least 10 years, the odds ratio, adjusted for SES, was 2.5 (95% CI, 1.3-4.5) for exposed men, and the relative risk was more than 5-fold for men younger than 40 years of age at diagnosis (odds ratio = 5.4, 95% CI, 2.4-11.9).

Petralia et al. (1999) used interview-based lifetime job histories and a job-exposure matrix to assess women's exposure to benzene and PAHs, adjusted for breast cancer risk factors. Exposed jobs involved bus and truck operators and engine mechanics, molding and casting machine operators, and garage and service-station occupations. PAH exposures independent of benzene are also found in traffic and shipping jobs, and benzene exposures without PAHs are found among clinical laboratory technologists, painters, and sculptors. The highest risk was seen for women exposed to both benzene and PAH, with about 2-fold increased risk for women ever exposed and higher risk for women exposed for 4 or more years. Increased risk of premenopausal breast cancer was seen among women exposed to benzene. The risk of PAH exposure could not be evaluated independent of benzene because of small numbers. Results provide some evidence of higher risk with longer duration of exposure and a latency period of 20 or more years.

Organic solvents, many of which are animal mammary carcinogens, have also been associated with breast cancer in an occupational study of 7,802 Danish women diagnosed at 20-55 years of age. Breast cancer risk was increased 20-66%, adjusted for childbearing and SES, for women employed longer than a year in jobs with extensive organic solvent use (Hansen 1999). Exposed women were employed in nonadministrative jobs in industries that involved metal products, wood and furniture, printing, chemicals, and textiles. Risks were more elevated for women who worked more than 10 years in these industries and for analyses with 15 or more years lag time. A 2-fold increased risk was seen for those with more than 10 years of employment.

In a case-control study of 995 incident breast cancers in British Columbia, Band et al. (2000) reported elevated risk among women in job titles associated with exposure to solvents and pesticides. In a study of Shanghai Cancer Registry data, Petralia et al. (1998) found breast cancer standardized incidence ratios (SIRs) were most elevated for women in professional jobs, but SIRs were also 40% higher for women with high probability of exposure to organic solvents and elevated for exposure to benzene and medium and high probability of pesticide exposure, based on a small number of cases. On the basis of "usual occupation" in mortality records for 33,509 cases and 117,794 controls in 24 states in the United States, Cantor et al. (1995) reported higher risk associated with higher probability and level of exposure to styrene; the widely used organic solvents methylene chloride, carbon tetrachloride, and formaldehyde; acid mists; and several metals.

Among 115 earlier studies of occupation and breast cancer reviewed by Goldberg and Labreche (1996), a few notable associations were seen. Two cohort studies reported evidence of higher risk for women in pharmaceutical manufacturing, and higher risk was also reported for women employed as cosmetologists or beauticians. Pollan and Gustavsson (1999) similarly reported elevated incidence for pharmacists, hairdressers, and beauticians with SES controlled in a cohort of women employed in 1970. Both historical and current risk among hairdressers is of interest because the mammary carcinogen vinyl chloride was used in hairspray until the early 1970s. Knowledge of workplace practices, more generally, may lead to better understanding of potentially informative inconsistencies among occupational studies.

Elevated risk was observed in other chemical-exposed jobs among metal platers and coaters (Pollan and Gustavsson 1999), whereas Goldberg and Labreche (1996) found little support for higher breast cancer risk for women in textile production (with exposure to dyes), dry cleaning (with exposure to organic solvents), or the nuclear industry. The negative finding in the nuclear industry despite clear evidence that ionizing radiation increases risk could mean that most workers were not actually exposed, or it could be due to protective characteristics of the workforce in that setting. For example, some jobs may attract or require women with high levels of physical activity, or sensitive workers may develop acute effects such as dermatitis and central nervous system symptoms that cause them to leave the workplace. This well-known phenomenon, referred to as the "healthy worker effect," complicates interpretation of negative occupational studies.

Similarly, breast cancer risk among farm women is of interest because of possible exposure to pesticides, but in general, observed breast cancer risk is lower among U.S. farm women, perhaps due to greater levels of physical activity or patterns in other established risk factors. Consistent with other studies, the Carolina Breast Cancer Study found that women who lived or worked on a farm had lower risk, but among those who did not wear protective clothing when applying pesticides, a 2-fold higher risk of breast cancer was observed (Duell et al. 2000). Research under way in the Agricultural Health Study will provide much better information about farm-related risk (Alavanja et al. 1994).

Overall, occupational studies provide fairly consistent evidence that elevated risk independent of SES is associated with a few specific exposures--benzene, organic solvents, and PAHs--especially for younger workers, and it is interesting to note that the chemicals with the most consistent human evidence have also been identified as animal mammary carcinogens (Table 1). Leads from previous occupational findings and new directions based on animal studies are priorities for further research, although follow-up studies will be challenging. Some of the challenges are typical of occupational studies; for example, workers are typically exposed to mixtures of chemicals, so specific exposures and exposure histories are difficult to reconstruct. In addition, using surveillance methods that are common in occupational studies makes it hard to separate out the effects of chemical exposures in populations that have protective characteristics, such as higher physical activity or lower-risk reproductive patterns. Other challenges arise from women's typical work histories, with exposed women likely to move into and out of the workforce and to be employed in dispersed, small-scale settings such as beauty shops. Goldberg and Labreche (1996) identify a number of weaknesses common in the studies they reviewed: reliance on administrative data and broad job categories as an indicator of exposure; lack of information on confounders, including childbearing and SES; use of mortality as an outcome rather than incidence, which limits the relevance to etiology; and low statistical power. Concerted efforts to overcome these limitations are important because occupational studies are the primary means by which chemicals become identified as human carcinogens (IARC 1998).

In future studies, possible confounding by work-related physical activity could be assessed using job matrix methods that parallel the assessment of chemical exposures. However, studies that contact workers to assess a broader range of established breast cancer risk factors concurrently with workplace exposures are needed to deal with other potential confounders. These studies will be most useful in evaluating chemical exposures that result in cancers diagnosed during women's working years, and longitudinal follow-up will be required to pick up effects among older women. Studies of health outcomes that are known or suspected to be related to breast cancer risk, including breast density, fertility outcomes, and age at menopause, also provide avenues to learn about breast cancer through occupational studies without waiting for workers to reach the older years when breast cancers are typically diagnosed. The likelihood, based on effect sizes for established breast cancer risk factors, that effects of occupational exposures may be modest in size means that large sample sizes or meta-analysis of multiple studies will be needed to discern effects. As more women move into jobs with substantial chemical exposure, assessment of occupational risks will become even more important.

Population-based studies. Population-based studies have investigated a narrow range of the compounds identified in the toxicologic literature as plausibly relevant to breast cancer. Certain OC compounds (DDT, PCBs) have been most studied; because they are persistent and lipophilic, residues can be measured in adipose tissue and blood years after exposure. Most studies to date have measured residues at the time of diagnosis or interview and assumed that these recent measures can be used as proxies for historical exposures. A few studies have assessed PAHs, some of which are potent mammary carcinogens in animals, and tobacco smoke, mixtures with complex toxicologic properties. Accidental exposures have led to studies of dioxin (TCDD) and perchloroethylene (PCE, also called tetrachloroethylene).

The largest recent report is from the Long Island Breast Cancer Study Project case-control study that assessed PAHs and certain OCs, based on blood samples drawn near the time of diagnosis (cases) or interview (controls) (Gammon et al. 2002a, 2002b). PAH exposure was assessed by measuring PAH-DNA adducts, a measure of DNA damage from exposure over the previous months to a few years. Results showed 49% higher risk, adjusted for breast cancer risk factors, for the highest compared with the lowest quintile of adducts (95% CI, 1.00-2.21), with no evidence of a dose-response relationship (Gammon et al. 2002a). Although the authors expected grilled food and tobacco smoke to be the primary sources of PAH, the lack of relationship between these exposures and PAH-DNA adducts suggests that other sources, for example, air pollution, may be more important. PAH-DNA adducts represent combined effects of intake and individual response, so the lack of dose response could mean that this measure is a better indicator of individual response than exposure (within the range of exposures in this study).

The Long Island study showed no significantly elevated risk associated with lipid-adjusted blood levels of the OC compounds DDE (the primary metabolite of DDT), chlordane, dieldrin, or the sum of the four most common PCB congeners, although small increases in risk were observed for the highest compared with the lowest exposure groups, with no dose-response trend, for DDE, DDT, and dieldrin (Gammon et al. 2002b). No consistent associations were seen for subgroups defined by reproductive risk factors, body size, years of residence on Long Island, or tumor estrogen- or progesterone-receptor status.

The results for DDE are consistent with scientific evidence that accumulated over the years during which the Long Island study took place. Although a few early studies reported an association with breast cancer, only 6 of 27 studies reviewed by Snedeker (2001) reported statistically significant positive associations. In her review, Snedeker offers a potential explanation for the many negative studies. She points out that most studies rely on DDE as an indicator of previous exposure to DDT because DDT is not currently detectable in blood in countries where DDT was banned years ago. However, diet (especially meat, fish, and dairy) is a major ongoing route of exposure to DDE, so DDE levels in blood represent exposure from diet as well as DDE metabolized from previous DDT exposure. DDE is much less hormonally active, so it may be that DDT, but not DDE, contributes to breast cancer, and if exposure to DDT is poorly measured by current blood levels of DDE, studies that rely on DDE are not informative. In fact, a recent study by Hoyer et al. (2000a) showed a significant relationship, with dose response, for breast cancer risk and p,p´-DDT measured prospectively in the late 1970s and early 1980s but no association for DDE. In addition, preliminary results from a California study using blood drawn during active DDT use showed increased risk of breast cancer diagnosed before age 50. Serum levels were measured prospectively in 131 case-control pairs. The odds ratio was 3.9 (95% CI, 1.4-10.9) for the second versus first tertile of DDT and 10.4 (95% CI, 2.5-43.2) for the third versus first tertile, with a highly statistically significant p-value for trend (Cohn et al. 2002). Additional studies of DDT levels in women currently exposed around the world or in blood drawn during years when DDT was in use in the United States may be informative.

A series of analyses of the association between breast cancer and blood levels of the pesticide dieldrin in Danish women have shown significant associations and dose-response trends for 1970s blood levels and breast cancer incidence (Hoyer et al. 1998) and mortality (Hoyer et al. 2000b). Mortality was increased more than 5-fold for women with the highest dieldrin levels averaged across two measurements from the 1970s and early 1980s (relative risk = 5.76; 95% CI, 1.86-17.92) (Hoyer et al. 2000b). Subgroup analyses showed the strongest associations with breast cancer risk for estrogen-receptor-negative tumors (Hoyer et al. 2001) and for tumors with p53 mutations (Hoyer et al. 2002). One potential explanation for these positive findings compared with other OC results is that blood measures were taken closer to the time of dieldrin use, which ended in the late 1970s, so they are better indicators of exposure.

Given the many difficulties of measuring historical exposures and characterizing variation among individuals in community settings, studies of unusual accidental exposures are a valuable resource. In a study of dioxin in women who were infants to 40 years of age at the time of a 1976 industrial accident in Seveso, Italy, Warner et al. (2002) reported a 2-fold increase in breast cancer risk among women with a 10-fold increase in serum level of dioxin (hazard ratio = 2.1; 95% CI, 1.0-4.6). Aschengrau et al. (2002) reported small to moderate increases in risk for women on Cape Cod, Massachusetts, exposed to PCE that leached from vinyl-lined water distribution pipes (adjusted odds ratios = 1.5-1.9 for > 75th percentile with 0-15 years of latency). Both of these studies have significance beyond the accidental exposure scenarios because dioxin and PCE are common exposures in everyday settings that could be reduced through changes in public policy. Dioxin is a widespread environmental contaminant, for example, from waste incineration. PCE is a solvent commonly used in industry and in dry cleaning, leading to both worker and consumer exposures.

Studies of breast cancer and tobacco smoke, including active smoking or passive exposure to environmental smoke from spouses or co-workers or in commercial and leisure settings, are more numerous than for other environmental pollutants, in part because exposure can be easily and inexpensively measured in interviews. Many early studies found no increased risk among smokers, and a recent meta-analysis of 53 studies comparing "ever" to "never" smokers found no association with breast cancer risk (Collaborative Group on Hormonal Factors in Breast Cancer et al. 2002). However, recent studies that separate active from passive exposure, consider a woman's age at exposure, and take into account genetic polymorphisms that affect the mechanism for ridding the body of smoke provide some evidence for an association, although the data are still inconsistent (Band et al. 2002; Bartsch et al. 2000; Dunning et al. 1999; Kropp and Chang-Claude 2002; Perera 2000).

In general, studies of genetic polymorphisms and breast cancer have focused on genes related to PAH and steroid metabolism (e.g., CYP, GST, NAT2), and studies of interaction between genetic polymorphisms and environmental pollutants have focused on tobacco smoke, with two studies of PCBs. Overall, results of these studies have been inconsistent (Bartsch et al. 2000; Basham et al. 2001; Dunning et al. 1999), with some evidence of effects of CYP, GST, and NAT2 polymorphisms and smoking on breast cancer risk, particularly in subgroup analyses (Ambrosone et al. 1996; Bartsch et al. 2000; Chang-Claude et al. 2002; Firozi et al. 2002; Hunter et al. 1997; Morabia et al. 2000; Zheng W et al. 2002; Zheng T et al. 2002, 2003), and two positive reports for PCBs and CYP polymorphisms in postmenopausal women (Laden et al. 2002; Moysich et al. 1999).

Overall, the population-based studies of breast cancer and environment represent a very sparse literature. Particularly notable is the focus on smoking and a small number of persistent OCs. Even for the most-studied chemicals, the number of studies is relatively small. In comparison, the recent meta-analyses of pharmaceutical estrogens and breast cancer are based on nearly twice as many studies as have been reported for DDT/DDE.

Challenges and Priorities

A variety of challenges in conducting studies about breast cancer and the environment may have discouraged work in this area, and these challenges define areas where future study will likely have the greatest impact. In particular, lack of exposure assessment tools and lack of toxicologic studies to develop hypotheses limit the scope of epidemiologic studies. In addition, issues of timing with respect to latency and periods of breast vulnerability, and individual differences in genetic susceptibility are challenges in research design that require attention. A substantial investment is needed in basic areas that are the foundation of successful human research--exposure assessment, toxicology, and susceptibility--before we can expect a pay-off from large epidemiologic studies of breast cancer and environment.

Exposure assessment. Multiple aspects of exposure assessment present methodological challenges. As in other cancer studies, latency means that exposures must be assessed for a time period long before diagnosis. For breast cancer specifically, evidence from both animal and epidemiologic studies suggests that there may be vulnerable periods, perhaps during gestation or adolescence or between menarche and birth of a first child, when exposure is most important. In addition, effects of environmental exposures may differ before and after menopause, as seen with some previously studied risk factors (e.g., body mass index and a recent report on smoking; Band et al. 2002). These multiple timing considerations are a particular challenge in studying exposures, such as air and water pollutants, that women cannot report retrospectively, in contrast with exposures, for example, childbearing history, that comprise the recognized risk factors. As yet, none of the available biomarkers can assess exposure dating back many years, let alone decades, and it is a particular challenge to characterize exposures for specific periods of the life span (e.g., during puberty). The complexity of mixtures in both occupational and community settings is another difficulty, along with simultaneous exposure to poorly understood degradation products and metabolites of pollutants.

Recent studies include efforts to improve exposure assessment in light of these challenges. Thus, the Long Island study and new research on tobacco smoke have included a relatively novel measure of PAH-DNA adducts. The Cape Cod Breast Cancer and Environment Study, now under way, defined development of new exposure assessment methods as a core goal (Brody et al. 1996). The study developed a geographic information system (GIS), a computer-mapping database, designed first to generate hypotheses and conduct ecologic analyses and later to assess exposures to wide-area pesticide use and drinking water contamination at individual addresses of 2,100 women in a case-control study (Brody et al. 2002). GIS is also being used in exposure reconstruction in several other epidemiologic studies (Beyea and Hatch 1999; Lynberg et al. 2001; Stellman et al. 2003; Ward et al. 2000). Capitalizing on geographically based research makes sense in studies of pollutants because many exposures vary geographically in relation to sources. Examples of nationally available data include the Toxics Release Inventory (http://www.epa.gov/tri), which documents point sources of pollutants, and records generated under the Safe Drinking Water Act (1974) for every public drinking water supply (Caldwell et al. 1998). Although some exposure data are available nationally, developing additional GIS exposure data is often more practical in a geographically limited area.

Because of enormous gaps in previous research about breast cancer and environmental pollutants, beginning with a lack of basic knowledge about the frequency and level of exposure to compounds identified as hormonally active or as animal mammary carcinogens, exposure studies that investigate these questions without yet tackling the link to breast cancer are an efficient way to proceed. For example, the Cape Cod Study developed an environmental sampling program for hormonally active compounds and mammary carcinogens in groundwater and drinking water, household air and dust, and women's urine. Results documented a potential pathway of exposure to endocrine disruptors that travel from septic systems to groundwater and drinking water, and identified 72 different hormonally active target compounds in homes, showing substantial opportunity for exposure (Rudel et al. 1998, 2001, 2002). Compounds for which frequent or high exposures have been identified and methods for measuring exposures developed might then be targeted in toxicologic and epidemiologic studies.

Considering that the ideal exposure assessment would provide information about the agent, dose, exposure pathway, timing in relation to latency, and timing in relation to life-cycle development, no one measurement technique is likely to provide a "gold standard." Self-report is vulnerable to response bias and cannot assess pollutant exposures unknown to the study participant. GIS offers a new approach to historical exposures and is independent of knowledge or bias among study participants, but it is vulnerable to missing data and faulty models of relationships between indicators and individual exposures. Environmental and biological sampling methods also may not accurately reconstruct individual historical exposure. Further, measurement methods have been developed for only a limited range of compounds, and measurements are expensive and sometimes intrusive to collect, resulting in small sample sizes with low statistical power and possible bias from nonparticipation. Analyses of relationships among environmental, biological, self-report, and GIS measures can help inform interpretation of studies using each of these exposure assessment methods and help identify sources of exposure. Studies to characterize environmental and biological exposures can also help identify populations or settings with high exposures that may provide unique opportunities for study.

Toxicology and mammary gland biology. Among 70,000 chemicals in commerce, fewer than 1,000 have been tested in cancer bioassays, and there has been no systematic testing for hormonal activity (U.S. EPA 1999). The challenge of analyzing mixtures and the idiosyncratic dose-response relationships (e.g., U-shaped) for hormones and hormonally active pollutants adds another layer of complexity. In addition, the biological and hormonal regulation of mammary gland development and carcinogenesis is poorly understood, so forming hypotheses about how chemicals will affect these processes is difficult.

Although standard animal bioassays for identifying carcinogens provide important direction for study in humans, improvements are needed in the development and application of animal models for mammary tumors specifically. For example, current protocols may not adequately address increased susceptibility to carcinogens for early-life exposures because dosing typically begins in pubertal animals (Bennett and Davis 2002). In addition, the rodent strains typically used for carcinogenesis bioassays may not be optimal for identifying mammary carcinogens, either because of a reduced susceptibility to such tumors (B6C3F₁ mice), because a high background rate of mammary tumors makes results difficult to interpret (Fischer 344 rats), or because hormonal regulation of the rodent mammary gland differs from that in humans (Bennett and Davis 2002; Dunnick et al. 1995; Snedeker 2001).

Another important issue for animal models is that, although it is important to identify chemical carcinogens that are genotoxic, which the current protocols are designed to do, it may also be important to identify chemicals that effectively promote the growth of cells after they have been initiated by some other carcinogen. The powerful role of endogenous hormones in promoting breast tumor development suggests that environmental chemicals that act as promoters could play an important role in breast cancer. Assays to look for tumor-promoting activity involve treating with a single dose of an initiator and then following with the promoter. In an assay like this, DDT was found to accelerate the rate of mammary tumor formation in male rats (females were not tested), suggesting that it could be active as a tumor promoter (Scribner and Mottet 1981), and wheat bran was shown to decrease the incidence of DMBA-initiated mammary tumors (Zile et al. 1998). Finally, it is also a priority to develop animal models that characterize the effects of in utero chemical exposures on development and susceptibility of the mammary gland in the offspring because in utero hormonal environments have been shown to affect later susceptibility to carcinogens (Hilakivi-Clarke et al. 2001).

Individual susceptibility and intermediate outcomes. Consideration of individual susceptibility is another area where limitations in previous research have led to recent innovation. Although high-risk breast cancer genes account for a small fraction of cases, lower risk, more common genetic polymorphisms that affect metabolism of endogenous estrogen and other chemicals are promising directions for study, as discussed above. However, studies to date have yielded conflicting results, in part because of the need for large sample sizes to achieve adequate statistical power and because of limited information on specific functional outcomes of the polymorphisms in relation to mechanisms of breast carcinogenesis (Dunning et al. 1999; Friedberg 2001; Perera 2000; Pharoah et al. 2002). This is another aspect of basic biology that could advance our ability to study breast cancer.

The difficulties of linking exposures with disease may also be remedied by studies of intermediate outcomes and of interactions or effect modification associated with recognized breast cancer risk factors. Studies of effects of chemical exposures on puberty, breast density, and in situ disease--all recognized risk factors for breast cancer--reduce the time lag between exposure and outcome measurement. Research to identify new intermediate outcomes, such as hallmarks of mammary gland development, will add to tools available for addressing breast cancer etiology.

Conclusion

Although journalistic reports have recently implied that scientific evidence shows that environmental pollutants are unrelated to breast cancer (Associated Press 2002; Kolata 2002), a review of research in this area reveals a much different picture of major knowledge gaps, difficult challenges in research design, and contrasting bodies of evidence from toxicologic and epidemiologic studies. Strong toxicologic evidence points to a large number of ubiquitous pollutants that are plausibly linked to breast cancer because they mimic or disrupt hormones known to affect breast cancer risk, initiate mammary tumors in animals, or permanently alter breast development, affecting susceptibility. Epidemiologic research is far more limited because very few of the compounds identified as endocrine disruptors or animal mammary carcinogens have ever been targeted in a human breast cancer study. A small but interesting body of occupational studies that link higher risk with jobs involving likely exposures to organic solvents and PAHs is generally consistent with animal studies. The relatively few population-based epidemiologic studies have been mostly negative overall, with positive results often limited to subgroups. Many plausible reasons for null epidemiologic results have been advanced in this article and elsewhere, including poor historical exposure measurement, restriction to a small number of pollutants, failure to study compounds in current use, low statistical power to detect modest effects, and failure to take into account genetic susceptibility or life-cycle effects. Limited study of women in occupational settings where exposures are relatively high and well defined is another barrier to understanding chemical risks. Given the modest relative risks associated with the recognized breast cancer risk factors, an integrated research agenda for study of environmental pollutants in both laboratory and human settings has great potential. Even if the relative risks of environmental factors are modest, discovery of a risk that can be modified would save many thousands of lives.

References

ACS. 1996. Breast Cancer Facts and Figures, 1996. Atlanta, GA:American Cancer Society.

------. 2002. Breast Cancer Facts and Figures, 2002. Atlanta, GA:American Cancer Society.

Adams JB. 1991. Enzymatic regulation of estradiol-17ß concentrations in human breast cancer cells. Breast Cancer Res Treat 20:145-154.

Adams JB, Vrahimis R, Phillips N. 1992. Regulation of estrogen sulfotransferase by estrogen in MCF-7 human mammary cancer cells. Breast Cancer Res Treat 22:157-161.

Adlercreutz CHT, Goldin BR, Gorbach SL, Hockerstedt KAV, Watanabe S, Hamalainen EK, et al. 1995. Soybean phytoestrogen intake and cancer risk. J Nutr 125:757S-770S.

Adlercreutz H. 2002. Phytoestrogens and breast cancer. J Steroid Biochem Mol Biol 83:113-118.

Alavanja MCR, Akland G, Baird D, Blair A, Bond A, Dosemeci M, et al. 1994. Cancer and noncancer risk to women in agriculture and pest control: the Agricultural Health Study. J Occup Med 36:1247-1250.

Alden CL. 2000. Safety assessment for non-genotoxic rodent carcinogens: curves, low-dose extrapolations, and mechanisms in carcinogenesis. Hum Exp Toxicol 19:557-560.

Ambrosone CB, Freudenheim JL, Graham S. 1996. Cigarette smoking, N-acetyltransferase 2 genetic polymorphisms, and breast cancer risk. JAMA 276:1494-1501.

An J, Tzagarakis-Foster C, Scharschmidt TC, Lomri N, Leitman DC. 2001. Estrogen receptor beta-selective transcriptional activity and recruitment of coregulators by phytoestrogens. J Biol Chem. 276:17808-17814.

Appel KE, Furstenberger G, Hapke HJ, Hecker E, Hildebrandt AG, Koransky W, et al. 1990. Chemical carcinogenesis: definitions of frequently used terms. J Cancer Res Clin Oncol 116:232-236.

Aschengrau A, Coogan PF, Quinn MM, Cashins LJ. 1998. Occupational exposure to estrogenic chemicals and the occurrence of breast cancer: an exploratory analysis. Am J Ind Med 34:6-14.

Aschengrau A, Rogers S, Ozonoff MD. 2002. Tetrachloroethylene-contaminated drinking water and the risk of breast cancer: additional result from Cape Cod, Massachusetts. Environ Health Perspect 111:175-178.

Associated Press. 2002. Study: no pollution, cancer link. Available: http://www.nytimes.com/aponline/national/AP-Cancer-Mapping.html [accessed 6 August 2002].

Austin DF, Roe KM. 1979. Increase in cancer of the corpus uteri in the San Francisco-Oakland standard metropolitan statistical area, 1960-1975. J Natl Cancer Inst 62:13-16.

Band PR, Le ND, Fang R, Deschamps M, Gallagher RP, Yang P. 2000. Identification of occupation cancer risks in British Columbia: a population-based case-control study of 995 incident breast cancer cases by menopausal status, controlling for confounding factors. J Occup Environ Med 42:284-310.

Band PR, Nhu DL, Fang R, Deschamps M. 2002. Carcinogenic and endocrine disrupting effects of cigarette smoke and risk of breast cancer. Lancet 360:1044-1049.

Barcellos-Hoff MH. 2001. It takes a tissue to make a tumor: epigenetics, cancer and the microenvironment. J Mammary Gland Biol Neoplasia 6:213-221.

Barcellos-Hoff MH, Ravini SA. 2000. Irradiated mammary gland stroma promotes the expression of tumorigenic potential by unirradiated epithelial cells. Cancer Res 60:1254-1260.

Barrett JC. 1993. Mechanisms of multistep carcinogenesis and carcinogen risk assessment. Environ Health Perspect 11:9-20.

Bartsch H, Nair U, Risch A, Rojas M, Wikman H, Alexandrov K. 2000. Genetic polymorphism of CYP genes alone or in combination, as a risk modifier of tobacco-related cancers. Cancer Epidemiol Biomarkers Prev 9:3-28.

Basham VM, Pharoah PDP, Healey CS, Luben RN, Day NE, Easton DF, et al. 2001. Polymorphisms in CYP1A1 and smoking: no association with breast cancer risk. Carcinogenesis 22:1797-1800.

Bennett LM, Davis BJ. 2002. Identification of mammary carcinogens in rodent bioassays. Environ Mol Mutagen 39:150-157.

Bernstein L. 2002. Epidemiology of endocrine-related risk factors for breast cancer. J Mammary Gland Biol Neoplasia 7:3-15.

Bernstein L, Allen M, Anton-Culver H, Deapen D, Horn-Ross PL, Peel D, et al. 2002. High breast cancer incidence rates among California teachers: results from the California Teachers Study (United States). Cancer Causes Control 13:625-635.

Beyea J, Hatch M. 1999. Geographic exposure modeling: a valuable extension of geographic information systems use for environmental epidemiology. Environ Health Perspect 107:181-190.

Birnbaum L, Fenton S. 2003. Cancer and developmental exposure to endocrine disruptors. Environ Health Perspect 111:389-394.

Brisken C. 2002. Hormonal control of alveolar development and its implications for breast carcinogenesis. J Mammary Gland Biol Neoplasia 7:39-48.

Brody JG, Rudel RA, Maxwell NI, Swedis SR. 1996. Mapping out a search for environmental causes of breast cancer. Public Health Rep 6:494-507.

Brody JG, Vorhees DJ, Melly SJ, Swedis SR, Drivas PJ, Rudel RA. 2002. Using GIS and historical records to reconstruct residential exposure to large-scale pesticide application. J Expo Anal Environ Epidemiol 12:64-80.

Brown N, Manzolillo P, Zhang J, Wang J, Lamartiniere C. 1998. Prenatal exposure to TCDD and predisposition to mammary cancer in the rat. Carcinogenesis 19:1623-1629.

Budavari S. 1996. The Merck Index: An Encyclopedia of Chemicals, Drugs and Biologicals, 12th ed. Whitehouse Station, NJ:Merck & Co.

Caldwell J, Woodruff T, Morello-Frosch R, Axelrad D. 1998. Application of health information to hazardous air pollutants modeled in EPA's cumulative exposure project. Toxicol Ind Health 14:429-454.

Calle EE, Murphy TK, Rodriguez C, Thun MJ, Heath CW. 1998. Occupation and breast cancer mortality in a prospective cohort of US women. Am J Epidemiol 148:191-197.

Cantor K, Stewart P, Brinton L, Dosmeci M. 1995. Occupational exposures and female breast cancer mortality in the United States. J Occup Environ Med 37:336-348.

Chang-Claude J, Kropp S, Jager B, Bartsch H, Risch A. 2002. Differential effect of NAT2 on the association between active and passive smoke exposure and breast cancer risk. Cancer Epidemiol Biomarkers Prev 11:698-704.

Clarke CA, Glaser SL, West DW, West DW, Ereman RR, Erdmann CA, et al. 2002. Breast cancer incidence and mortality trends in an affluent population: Marin County, California, USA, 1990-1999. Breast Cancer Res Treat 4:R13.

Claus EB, Schildkraut JM, Thompson WD. 1996. The genetic attributable risk of breast and ovarian cancer. Cancer 77:2318-2324.

Clegg L, Feuer E, Midthune D, Fay M, Hankey B. 2002. Impact of reporting delay and reporting error on cancer incidence rates and trends. J Natl Cancer Inst 94:1537-1545.

Cohn BA, Wolff MS, Cirillo PM, Sholtz RI, Christianson R, van den Berg BJ, et al. 2002. Timing of DDT exposure and breast cancer before age 50 [Abstract]. In: ISEE/ISEA Linking Exposure and Health: Innovations and Interactions, 10-15 August 2002, Vancouver, British Columbia, Canada. Epidemiology 13(4):S197.

Collaborative Group on Hormonal Factors in Breast Cancer. 1996. Breast cancer and hormonal contraceptives: collaborative reanalysis of individual data on 53,297 women with breast cancer and 100,239 women without breast cancer from 54 epidemiological studies. Lancet 347:1713-1727.

------. 1997. Breast cancer and hormone replacement therapy: collaborative reanalysis of data from 51 epidemiological studies of 52,705 women with breast cancer and 108,411 women without breast cancer. Lancet 350:1047-1113.

Collaborative Group on Hormonal Factors in Breast Cancer, Hamajima N, Hirose K, Tajima K, Rohan T, Calle E, et al. 2002. Alcohol, tobacco and breast cancer--collaborative reanalysis of individual data from 53 epidemiological studies, including 58,515 women with breast cancer and 95,067 women without the disease. Br J Cancer 87:1234-1245.

Colton T, Greenberg ER, Noller K, Resseguie L, Van Bennekom C, Heeren T, et al. 1993. Breast cancer in mothers prescribed diethylstilbestrol in pregnancy: further follow-up. JAMA 269:2096-2100.

Davis DL, Axelrod D, Osborne MP, Telang NT. 1997. Environmental influences on breast cancer risk. Sci Med 4:56-63.

Davis DL, Bradlow HL, Wolff M, Woodruff T, Howl DG, Anton-Culver H. 1993. Medical hypothesis: xenoestrogens as preventable causes of breast cancer. Environ Health Perspect 101:372-377.

Deapen D, Liu L, Perkins C, Bernstein L, Ross RK. 2002. Rapidly rising breast cancer incidence rates among Asian-American women. Int J Cancer 99:747-750.

Desaulniers D, Leingartner K, Russo J, Perkins G, Chittim B, Archer M, et al. 2001. Modulatory effects of neonatal exposure to TCDD, or a mixture of PCBs, p,p'-DDT, and p,p'-DDE, on methylnitrosourea-induced mammary tumor development in the rat. Environ Health Perspect 109:739-747.

Duell EJ, Millikan RC, Savitz DA, Newman B, Smith JC, Schell MJ, et al. 2000. A population-based case-control study of farming and breast cancer in North Carolina. Epidemiology 11:523-531.

Dunnick JK, Elwell MR, Huff J, Barrett JC. 1995. Chemically induced mammary gland cancer in the National Toxicology Program's carcinogenesis bioassay. Carcinogenesis 16:173-179.

Dunning AM, Healy CS, Pharoah PD, Teare MD, Ponder BA, Easton DF. 1999. A systematic review of genetic polymorphisms and breast cancer risk. Cancer Epidemiol Biomarkers Prev 8:843-854.

Emmen JM, Korach KS. 2001. Developing animal models for analyzing SERM activity. Ann NY Acad Sci 949:36-43.

Feigelson HS, Jonas CR, Robertson AS, McCullough ML, Thun MJ, Calle EE. 2003. Alcohol, folate, methionine, and risk of incident breast cancer in the American Cancer Society Cancer Prevention Study II nutrition cohort. Cancer Epidemiol Biomarkers Prev 12:161-164.

Fenton SE, Hamm JT, Birnbaum LS, Youngblood GL. 2002. Persistent abnormalities in the rat mammary gland following gestational and lactational exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD). Toxicol Sci 67:63-74.

Firozi PF, Bondy ML, Sahin AA, Chang P, Lukmanji F, Singletary ES, et al. 2002. Aromatic DNA adducts and polymorphisms of CYP1A1, NAT2, and GSTM1 in breast cancer. Carcinogenesis 23:301-306.

Ford D. 2002. Mechanistic explanations for the chemopreventive action of soyabean isoflavones: reducing the possibilities. Br J Nutr 88:439-441.

Fotsis T, Pepper M, Adlercreutz H. 1993. Genistein, a dietary-derived inhibitor of in vitro angiogenesis. Proc Natl Acad Sci USA 90:2690-2694.

Friedberg T. 2001. Cytochrome P450 polymorphisms as risk factors for steroid hormone-related cancers. Am J Pharmacogenomics 1:83-91.

Gail MH, Brinton LA, Byar DP, Corle DK, Green SB, Schairer C, et al. 1989. Projecting individualized probabilities of developing breast cancer for white females who are being examined annually. J Natl Cancer Inst 81:1879-1886.

Gammon MD, Santella RM, Neugut AI, Eng SM, Teitelbaum SL, Paykin A, et al. 2002a. Environmental toxins and breast cancer on Long Island. I. Polycyclic aromatic hydrocarbon DNA adducts. Cancer Epidemiol Biomarkers Prev 11:677-685.

Gammon MD, Wolff MS, Neugut AI, Eng SM, Teitelbaum SL, Britton JA, et al. 2002b. Environmental toxins and breast cancer on Long Island. II. Organochlorine compound levels in blood. Cancer Epidemiol Biomarkers Prev 11:686-697.

Gandini S, Merzenich H, Robertson C, Boyle P. 2000. Meta-analysis of studies on breast cancer risk and diet: the role of fruit and vegetable consumption and the intake of associated micronutrients. Eur J Cancer 36:636-46.

Ginsburg E, Walsh B, Shea B, Gao X, Gleason R, Barbieri R. 1995. The effects of ethanol on the clearance of estradiol in postmenopausal women. Fertil Steril 63:1227-30.

Goldberg MS, Labreche F. 1996. Occupational risk factors for female breast cancer: a review. Occup Environ Med 53:145-156.

Gould JC, Leonard LS, Maness SC, Wagner BL, Conner K, Zacharewshi T, et al. 1998. Bisphenol A interacts with estrogen receptor alpha in a distinct manner from estradiol. Mol Cell Endocrinol 142:203-214.

Hansen J. 1999. Breast cancer risk among relatively young women employed in solvent-using industries. Am J Ind Med 36:43-47.

------. 2000. Elevated risk for male breast cancer after occupational exposure to gasoline and vehicular combustion products. Am J Ind Med 37:349-352.

Harris CA, Henttu P, Parker MG, Sumpter JP. 1997. The estrogenic activity of phthalate esters in vitro. Environ Health Perspect 105:802-811.

Henderson BE, Feigelson HS. 2000. Hormonal carcinogenesis. Carcinogenesis 21:427-433.

Hilakivi-Clarke L, Cho E, deAssis S, Olivo S, Ealley E, Bouker KB, et al. 2001. Maternal and prepubertal diet, mammary development and breast cancer risk. J Nutr 131:154S-157S.

Holmes MD, Hunter DJ, Colditz GA, Stampfer MJ, Hankinson SE, Speizer FE, et al. 1999. Association of dietary intake of fat and fatty acids with risk of breast cancer. JAMA 281:914-920.

Hoyer A, Gerdes A, Jorgensen T, Rank F, Hartvig H. 2002. Organochlorines, p53 mutations in relation to breast cancer risk and survival. A Danish cohort-nested case-control study. Breast Cancer Res Treat 71:59-65.

Hoyer A, Jorgensen T, Grandjean P, Hartvig H. 2000a. Repeated measurements of organochlorine exposure and breast cancer risk (Denmark). Cancer Causes Control 11:177-184.

Hoyer A, Jorgensen T, Rank F, Grandjean P. 2001. Organochlorine exposures influence on breast cancer risk and survival according to estrogen receptor status: a Danish cohort-nested case-control study. BMC Cancer 1:8.

Hoyer AP, Grandjean P, Jorgensen T, Brock JW, Hartvig HB. 1998. Organochlorine exposure and risk of breast cancer. Lancet 352:1816-1820.

Hoyer AP, Jorgensen T, Brock JW, Grandjean P. 2000b. Organchlorine exposures and breast cancer survival. J Clin Epidemiol 53:323-330.

Hunter D, Hankinson S, Hough H, Gertig D, Garcia-Closas M, Spiegelman D, et al. 1997. A prospective study of NAT2 acetylation genotype, cigarette smoking, and risk of breast cancer. Carcinogenesis 18:2127-2132.

Hunter DJ, Willett WC. 1996. Nutrition and breast cancer. Cancer Causes Control 7:56-68.

IARC. 1998. IARC Monographs Programme on the Evaluation of Carcinogenic Risks to Human. Lyon:International Agency for Research on Cancer. Available: http://193.51.164.11/ [accessed 12 December 1998].

------. 1999. Overall Evaluations of Carcinogenic Risks to Humans. IARC Monographs. Lyon:International Agency for Research on Cancer. Available: http://193.51.164.11/monoeval/crthall.html [accessed 15 October 2002].

Illinois Environmental Protection Agency. 1997. Endocrine Disruptors Strategy. Springfield, IL:Illinois Environmental Protection Agency.

Jobling S, Nolan M, Tyler CR, Brighty G, Sumpter JP. 1998. Widespread sexual disruption in wild fish. Environ Sci Technol 32:2498-2506.

Jobling S, Reynolds T, White R, Parker MG, Sumpter JP. 1995. A variety of environmentally persistent chemicals, including some phthalate plasticizers, are weakly estrogenic. Environ Health Perspect 103:582-587.

Klaunig JE, Kamendulis LM, Xu Y. 2000. Epigenetic mechanisms of chemical carcinogenesis. Hum Exp Toxicol 19:543-555.

Kliewer EV, Smith KR. 1995. Breast cancer mortality among immigrants in Australia and Canada. J Natl Cancer Inst 87:1154-1161.

Kolata G. 2002. What causes cancer: can science find the missing link? New York Times (New York, NY), 11 August, D1.

Korach KS, McLachlan JA. 1995. Techniques for detection of estrogenicity. Environ Health Perspect 103:5-8.

Kreiger N, Sloan M, Cotterchio M, Kirsch V. 1999. The risk of breast cancer following reproductive surgery. Eur J Cancer 35:97-101.

Kropp S, Chang-Claude J. 2002. Active and passive smoking and risk of breast cancer by age 50 years among German women. Am J Epidemiol 156:616-626.

Laden F, Ishibe N, Hankinson SE, Wolff MS, Gertig DM, Hunter DJ, et al. 2002. Polychlorinated biphenyls, cytochrome P450 1A1, and breast cancer risk in the Nurses' Health Study. Cancer Epidemiol Biomarkers Prev 11:1560-1565.

Laden F, Spiegelman D, Neas LM, Colditz GA, Hankinson SE, Manson JE, et al. 1997. Geographic variation in breast cancer incidence rates in a cohort of U.S. women. J Natl Cancer Inst 89:1373-1378.

Lamartiniere CA, Moore J, Holland M, Barnes S. 1995. Neonatal genistein chemoprevents mammary cancer. Proc Soc Exp Biol Med 208:120-123.

Land CE. 1995. Studies of cancer and radiation dose among atomic bomb survivors: the example of breast cancer. JAMA 274:402-407.

Lichtenstein P, Kaprio J, Hemminki K. 2000. Cancer, genes and the environment [Letter]. N Engl J Med 343:1495-1946.

Liehr JG, Roy D, Ari-Ulubelen A, Bui QD, Weisz J, Strobel HW. 1990. Effect of chronic estrogen treatment of Syrian hamsters on microsomal enzymes mediating formation of catecholestrogens and their redox cycling: implications for carcinogenesis. J Steroid Biochem 35:555-560.

Lynberg M, Nuckols JR, Langlois P, Ashley D, Singer P, Mendola P, et al. 2001. Assessing exposure to disinfection by-products in women of reproductive age living in Corpus Christi, Texas, and Cobb County, Georgia: descriptive results and methods. Environ Health Perspect 109:597-604.

Mack TM, Hamilton AS, Press MF, Diep A, Rappaport EB. 2002. Heritable breast cancer in twins. Br J Cancer 87:294-300.

Madigan MP, Ziegler RG, Benichou J, Byrne C, Hoover RN. 1995. Proportion of breast cancer cases in the United States explained by well-established risk factors. J Natl Cancer Inst 87:1681-1685.

Markaverich BM, Webb B, Densmore CL, Gregory RR. 1995. Effects of coumestrol on estrogen receptor function and uterine growth in ovariectomized rats. Environ Health Perspect 103:574-581.

Martin AM, Weber BL. 2001. Genetic and hormonal risk factors in breast cancer. J Natl Cancer Inst 92:1126-1135.

Maxwell N, Polk R, Melly S, Brody J. 1999. Newton Breast Cancer Study--Key Findings. Newton, MA:Silent Spring Institute. Available: http://www.silentspring.org/newweb/projects/newton/findings.html [accessed 20 May 2003].

Michels K. 2002. The contribution of the environment (especially diet) to breast cancer risk. Breast Cancer Res 4:58-61.

Morabia A, Bernstein MS, Bouchardy I, Kurtz J, Morris MA. 2000. Breast cancer and active and passive smoking: the role of the N-acetyltransferase 2 genotype. Am J Epidemiol 152:226-232.

Moysich K, Shields P, Freudenheim J, Schisterman E, Vena J, Kostyniak P, et al. 1999. Polychlorinated biphenyls, cytochrome P4501A1 polymorphism, and postmenopausal breast cancer risk. Cancer Epidemiol Biomarkers Prev 8:41-44.

Mueller SO, Clarke JA, Myers PH, Korach KS. 2002. Mammary gland development in adult mice requires epithelial and stromal estrogen receptor alpha. Endocrinology 143:2357-2365.

National Cancer Institute. 1999. Atlas of Cancer Mortality in the United States 1950-1994. NIH Publ no 99-4564. Bethesda, MD:National Cancer Institute.

National Center for Health Statistics. 1997. Multiple Cause-of-Death File 1993. Datafile. SETS Version 1.22a. Atlanta, GA:Centers for Disease Control and Prevention.

Neumann DA, Crisp TM, Olin SS. 1996. Mammary gland neoplasia. Environ Health Perspect 104:912-914.

Nishihara T, Nishikawa J, Kanayama T, Dakeyama F, Saito K, Imagawa M, et al. 2000. Estrogenic activities of 517 chemicals by yeast two-hybrid assay. J Health Science 46:282-298.

Norman SA, Berlin JA, Soper KA, Middendorf BF, Stolley PD. 1995. Cancer incidence in a group of workers potentially exposed to ethylene oxide. Int J Epidemiol 24:276-284.

NRC (National Research Council). 1999. Hormonally Active Agents in the Environment. Washington, DC:National Academy Press.

------. 1990. Health Effects of Exposure to Low Levels of Ionizing Radiation (BEIR V). Washington, DC:National Academy Press.

NTP. 1998. Abstracts of NTP Long-Term Cancer Studies. Research Triangle Park, NC:National Toxicology Program.

------. 2000. Report on Carcinogens, Ninth edition: Research Triangle Park NC:National Toxicology Program.

------. 2002. Report on Carcinogens, Tenth edition. Research Triangle Park, NC:National Toxicology Program.

Nurses' Health Study. 2002. Nurses' Health Study History. Available: http://www.channing.harvard.edu/nhs/hist.html [accessed 8 October 2002].

O'Connor JC, Cook JC, Craven SC, Pelt CSV, Obourn JD. 1996. An in vivo battery for identifying endocrine modulators that are estrogenic or dopamine regulators. Fundam Appl Toxicol 33:182-195.

Parazzini F, Braga C, Vecchi CL, Negri E, Acerboni S, Franceschi S. 1997. Hysterectomy, oophorectomy in premenopause, and risk of breast cancer. Obstet Gynecol 90:453-456.

Parkin DM, Bray FI, Devesa SS. 2001. Cancer burden in the year 2000. The global picture. Eur J Cancer 37:S4-S66.

Pennie WD, Aldridge TC, Brooks AN. 1998. Differential activation by xenoestrogens of ER alpha and ER beta when linked to different response elements. J Endocrinol 158:R11-R14.

Perera FP. 2000. Molecular epidemiology: on the path to prevention? J Natl Cancer Inst 92:602-612.

Petralia SA, Chow WH, McLaughlin J, Jin F, Gao YT, Dosemeci M. 1998. Occupational risk factors for breast cancer among women in Shanghai. Am J Ind Med 34:477-483.

Petralia SA, Vena JE, Freudenheim JL, Dosemeci M, Michalek A, Goldberg MS, et al. 1999. Risk of premenopasual breast cancer in association with occupational exposure to polycyclic aromatic hydrocarbons and benzene. Scand J Work Environ Health 25:215-221.

Pharoah P, Antoniou A, Bobrow M, Zimmern R, Easton D, Ponder B. 2002. Polygenic susceptibility to breast cancer and implications for prevention. Nat Genet 31:33-36.

Pinter A, Torok G, Borzonyi M, Surjan A, Calk M, Kelecsenvi Z, et al. 1990. Long-term carcinogenicity bioassay of the herbicide atrazine in F344 rats. Neoplasma 37:533-544.

Pitot HC. 1993. The dynamics of carcinogenesis: implications for human risk. Chemical Industry Institute of Toxicology (CIIT) Activities 13:1-7.

Pitot HC, Dragan BD. 1991. Facts and theories concerning the mechanisms of carcinogenesis. FASEB J 5:2280-2286.

Pitot HC, Hikita H, Dragan Y, Sargent L, Haas M. 2000. Review article: the stages of gastrointestinal carcinogenesis--application of rodent models to human disease. Aliment Pharmacol Ther 14:153-160.

Po L, Wang T, Chen Z, Leung L. 2002. Genistein-induced apoptosis in MCF-7 cells involves changes in Bak and Bcl-x without evidence of anti-oestrogenic effects. Br J Nutr 88:463-9.

Pollan M, Gustavsson P. 1999. High-risk occupations for breast cancer in the Swedish female working population. Am J Public Health 89:875-881.

Prehn AW, West DW. 1998. Evaluating local differences in breast cancer incidence rates: a census-based methodology (United States). Cancer Causes Control 9:511-517.

Robbins AS, Brescianini S, Kelsey JL. 1997. Regional differences in known risk factors and the higher incidence of breast cancer in San Francisco. J Natl Cancer Inst 89:960-965.

Routledge EJ, Parker J, Odum J, Ashby J, Sumpter JP. 1998. Some alkyl hydroxy benzoate preservatives (parabends) are estrogenic. Toxicol Appl Pharmacol 153:12-19.

Rudel R. 1997. Predicting health effects of exposures to compounds with estrogenic activity: methodological issues. Environ Health Perspect 105:655-663.

Rudel RA, Geno P, Melly SJ, Sun G, Brody JG. 1998. Identification of alkylphenols and other estrogenic phenolic compounds in wastewater, septage, and groundwater on Cape Cod, Massachusetts. Environ Sci Technol 32:861-869.

Rudel RA, Geno PW, Sun G, Yau A, Spengler JD, Vallarino J, et al. 2001. Identification of selected hormonally active agents in animal and mammary carcinogens in commercial and residential air and dust samples. J Air Waste Manage Assoc 51:499-513.

Rudel RA, Swartz CH, Brody JG, Camann DM, Yau A, Zuniga M, et al. 2002. Residential indoor air and dust measurements for pesticides, alkylphenols, phthalates, and other endocrine disruptors [Abstract]. In: Proceedings of Annual Conference of International Society for Environmental Epidemiology (ISEE)/ International Society of Exposure Analysis (ISEA) 10-15 August 2002, Vancouver, British Columbia, Canada. Epidemiology 13(4):S198.

Russo IH, Russo J. 1996. Mammary gland neoplasia in long term rodent studies. Environ Health Perspect 104:938-967.

------. 1998. Role of hormones in mammary cancer initiation and progression. J Mammary Gland Biol Neoplasia 3:49-61.

Safe Drinking Water Act of 1974. 1974. 42 U.S.C. s/s 300f et seq.

Safe SH. 1995. Environmental and dietary estrogens and human health: is there a problem? Environ Health Perspect 103:346-351.

Scribner JD, Mottet NK. 1981. DDT acceleration of mammary gland tumors induced in the male Sprague-Dawley rat by 2-acetamidophenanthrene. Carcinogenesis 2:1235-1239.

SEER: Surveillance Epidemiology and End Results. 2002. Cancer Statistics Review, 1973-1990. Incidence: Breast Cancer. Bethesda, MD:National Cancer Institute. Available: http://seer.cancer.gov/faststats/html/inc_breast.html [accessed 15 October 2002].

Shelby MD, Newbold RR, Tully DB, Chae K, Davis VL. 1996. Assessing environmental chemicals for estrogenicity using a combination of in vitro and in vivo assays. Environ Health Perspect 104:1296-1300.

Silent Spring Institute. 1998. The Cape Cod Breast Cancer and Environment Study: Results of the First Three Years of Study. Newton, MA:Silent Spring Institute. Available: http://www.silentspring.org/newweb/publications/SilentSpring97report.pdf [accessed 20 May 2003].

------. 2000. Cape Cod Breast Cancer and Environment Atlas. Newton, MA:Silent Spring Institute. Available: http://www.silentspring.org/newweb/atlas/index.html [accessed 20 May 2003].

Silva E, Rajapakse N, Kortenkamp A. 2002. Something from "nothing"--eight weak estrogenic chemicals combined at concentrations below NOECs produce significant mixture effects. Environ Sci Technol 36:1751-1756.

Singletary K, Gapstur S. 2001. Alcohol and breast cancer: review of epidemiologic and experimental evidence and potential mechanisms. JAMA 286:2143-2151.

Sivaraman L, Medina D. 2002. Hormone-induced protection against breast cancer. J Mammary Gland Biol Neoplasia 7:77-92.

Smith ER, Quinn MM. 1992. Uterotropic action in rats of amsonic acid and three of its synthetic precursors. J Toxicol Environ Health 36:13-25.

Smith-Warner S, Spiegelman D, Yaun S, Adami H, Beeson W, van den Brandt PA, et al. 2001. Intake of fruits and vegetables and risk of breast cancer: a pooled analysis of cohort studies. JAMA 285:769-776.

Snedeker SM. 2001. Pesticides and breast cancer risk: a review of DDT, DDE, and dieldrin. Environ Health Perspect 109:35-47.

Sonnenschein C, Soto AM. 1999. The Society of Cells: Cancer and Control of Cell Proliferation. New York:Bios Scientific Publishers.

Soto AM, Sonnenschein C, Chung KL, Fernandez MF, Olea N, Serrano FO. 1995. The E-SCREEN assay as a tool to identify estrogens: an update on estrogenic environmental pollutants. Environ Health Perspect 103:113-122.

Spengler JD, Samet JD, McCarthy JF, eds. 2000. Indoor Air Quality Handbook. New York:McGraw Hill.

Spratt J, Donegan W, Sigdestad C. 1995. Epidemiology and etiology. In: Cancer of the Breast (Donegan W, JS Spratt, eds). 4th edition. Philadephia:W.B. Sanders Company, 116-141.

SRI International. 1995. Chemical Economics Handbook. Menlo Park, CA:SRI International.

Stellman SD, Stellman JM, Weber T, Tomasallo C, Stellman AB, Christian R. 2003. A geographic information system for characterizing exposure to agent orange and other herbicides in Vietnam. Environ Health Perspect 111:321-328.

Sturgeon SR, Schairer C, Gail M, McAdams M, Brinton LA, Hoover RN. 1995. Geographic variation in mortality from breast cancer among white women in the United States. J Natl Cancer Inst 87:1846-1853.

Swanson SM, Unterman TG. 2002. The growth hormone-deficient spontaneous dwarf rat is resistant to chemically induced mammary carcinogenesis. Carcinogenesis 23:977-982.

Telang NT, Suto A, Wong GY, Osborne MP, Bradlow HL. 1992. Induction by estrogen metabolite 16 alpha-hydroxyestrone of genotoxic damage and aberrant proliferation in mouse mammary epithelial cells. J Natl Cancer Inst 84:634-638.

Titus-Ernstoff L, Hatch EE, Hoover RN, Palmer J, Greenberg ER, Ricker W, et al. 2001. Long-term cancer risk in women given diethylstilbestrol (DES) during pregnancy. Br J Cancer 84:126-133.

Tokunaga M, Land CE, Yamamoto T, Asano M, Tokuoka S, Ezaki H, et al. 1987. Incidence of female breast cancer among atomic bomb survivors, Hiroshima and Nagasaki, 1950-1980. Radiat Res 112:243-272.

Tompa A, Major J, Jakab MG. 1999. Is breast cancer cluster influenced by environmental and occupational factors among hospital nurses in Hungary? Pathol Oncol Res 5:117-121.

Toniolo P, Bruning PF, Akhmedkhanov A, Bonfrer JM, Koenig KL, Lukanova A, et al. 2000. Serum insulin-like growth factor-I and breast cancer. Int J Cancer 88:828-832.

Trock B, Butler LW, Clarke R, Hilakivi-Clarke L. 2000. Meta-analysis of soy intake and breast cancer risk. In: Third International Symposium on the Role of Soy in Preventing and Treating Chronic Disease, 31 October-3 November 1999. Washington, DC. J Nutr 130:653S-680S.

Tsutsui T, Barrett JC. 1997. Neoplastic transformation of cultured mammalian cells by estrogens and estrogenlike chemicals. Environ Health Perspect 105:619-624.

Ursin G, Bernstein L, Pike MC. 1994. Breast cancer. In: Cancer Surveys (Doll R, Fraumeni JF, Muir CS, eds). Plainview, NY:Cold Spring Harbor Laboratory Press, 241-264.

U.S. EPA. 1999. Chemical Hazard Data Availability Study. Washington, DC:U.S. Environmental Protection Agency. Available: http://www. epa.gov/opptintr/chemtest/hazchem.htm#Master [accessed 11 October 1999].

Wang DY, Allen DS, Stavola BLD, Fentiman IS, Brussen J, Bulbrook RD, et al. 2000. Urinary androgens and breast cancer risk: results from a long-term prospective study based in Guernsey. Br J Cancer 82:1577-1584.

Ward MH, Nuckols JR, Weigel SJ, Maxwell SK, Cantor KP, Miller RS. 2000. Identifying populations potentially exposed to agricultural pesticides using remote sensing and a geographic information system. Environ Health Perspect 108:5-12.

Warner M, Eskenazi B, Mocarelli P, Gerthoux PM, Samuels S, Needham L, et al. 2002. Serum dioxin concentrations and breast cancer risk in the Seveso Women's Health Study. Environ Health Perspect 110:625-628.

Willett WC. 1999. Dietary fat and breast cancer. Toxicol Sci 52:127-146.

Wolff MS, Collman GW, Barrett JC, Huff J. 1996. Breast cancer and environmental risk factors: epidemiological and experimental findings. Annu Rev Pharmacol Toxicol 36:573-596.

Women's Health Initiative Investigators. 2002. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results from the Women's Health Initiative randomized controlled trial. JAMA 288:321-333.

Wu AH, Wan P, Hankin J, Tseng CC, Yu MC, Pike MC. 2002. Adolescent and adult soy intake and risk of breast cancer in Asian-Americans. Carcinogenesis 23:1491-1496.

Wu AH, Ziegler RG, Pike MC, Nomura AM, West DW, Kolonel LN, et al. 1996. Menstrual and reproductive factors and risk of breast cancer in Asian-Americans. Br J Cancer 73:680-686.

Zhang SM, Willett WC, Selhub J, Hunter DJ, Giovannucci EL, Holmes MD, et al. 2003. Plasma folate, vitamin B6, vitamin B12, homocysteine, and risk of breast cancer. J Natl Cancer Inst 95:373-380.

Zheng T, Holford T, Zahm S, Owens P, Boyle P, Zhang Y, et al. 2002. Cigarette smoking, glutathione-S-transferase M1 and t1 genetic polymorphisms, and breast cancer risk (United States). Cancer Causes Control 13:637-45.

------. 2003. Glutathione S-transferase M1 and T1 genetic polymorphisms, alcohol consumption and breast cancer risk. Br J Cancer 88:58-62.

Zheng W, Wen W, Gustafson D, Gross M, Cerhan J, Folsom A. 2002. GSTM1 and GSTT1 polymorphisms and postmenopausal breast cancer risk. Breast Cancer Res Treat 74:9-16.

Ziegler RG, Hoover RN, Pike MC. 1993. Migration patterns and breast cancer risk in Asian-American women. J Natl Cancer Inst 85:1819-1827.

Zile MH, Welsch CW, Welsch MA. 1998. Effect of wheat bran fiber on the development of mammary tumors in female intact and ovariectomized rats treated with 7,12-dimethylbenz(a)anthracene and in mice with spontaneously developing mammary tumors. Int J Cancer 75:439-443.

Last Updated: June 11, 2003