Breast cancer remains the most frequently diagnosed malignancy in American women and the second most common cause of cancer death (Wingo et al. 1995). Although breast cancer is a common disease, the perception among many women is that their personal risk is even greater than the estimated 1 in 8 lifetime chance of being diagnosed with breast cancer or the 1 in 28 lifetime risk of dying from breast cancer. American women in increasing numbers participate in screening programs involving breast self-exam and mammography in the hope that early detection will prevent mortality. However, a significant portion of breast cancer cases elude early detection due to the expression of invasive and metastatic biological phenotypes prior to the development of a palpable tumor or radiographic changes on a mammogram. Many concerned women desire and seek information regarding diet and nutritional approaches for breast cancer prevention. The rapid growth of a multibillion dollar "health food" and supplement industry and the emergence of many "alternative" health care practitioners focusing upon nutrition in American communities is a direct result of a failure of the established medical and nutritional sciences profession to provide appropriate education and guidance to the public. The news media often compromises nutrition education by sensationalizing individual studies without consideration of the much larger body of data. The tabloid press frequently promotes stories of unsubstantiated and miraculous cures of human cancers by poorly characterized products and approaches related to diet and nutrition. Breast cancer patients, and the public in general, increasingly report that information in the media changes week to week and seems contradictory or confusing. Indeed, a similar perception by non-nutritional scientists and health care practitioners further weakens efforts to obtain funding for nutrition education, intervention and research in the area of breast cancer.

The importance of developing and implementing effective, nontoxic and health-promoting strategies for prevention of breast cancer depends upon a greater understanding of etiologic factors. This symposium is an effort by the American Society for Nutritional Sciences to provide an overview of a complex and often controversial area of nutritional and breast cancer. The goals are to provide a summary of current knowledge derived from laboratory and epidemiologic studies, discuss the strengths and limitations of current investigative tools, and foster collaborative multidisciplinary and integrative research approaches to test novel hypotheses. The following is a brief historical perspective on the development of major concepts in the area of dietary lipids and energy, anthropometry, genetics and breast carcinogenesis.

HISTORICAL PERSPECTIVES: EXPERIMENTAL NUTRITION AND BREAST CARCINOGENESIS

In parallel to the progress in the nutritional sciences, many fundamental principles of carcinogenesis were also established during this period. The identification of pure chemical carcinogens from environmental substances associated with human cancer rapidly led to the characterization of many animal models of carcinogenesis. The concept that specific chemical substances act as tumor initiators, co-carcinogens or tumor-promoting agents by Berenblum (1941) provided the framework for many future studies of nutrition and carcinogenesis. The development in the 1930s and 1940s of genetically homogeneous inbred strains of mice that exhibited various susceptibilities to breast tumor development provided early clues into a role for genetic factors (Strong 1935). The possibility that viral infections contributed to mammary carcinogenesis was indicated in early studies showing the transmission of a tumor-promoting substance via the milk in mice (Bittner 1935). By 1940 the nutrient requirements of rodents were well defined, and purified components of foods were available for the preparation of carefully controlled experimental diets in studies using the new models of mammary carcinogenesis. The meticulous studies of Albert Tannenbaum and colleagues published in the 1940s clearly documented the relationships of dietary fat concentration, energy intake and mammary carcinogenesis that have been expanded upon by many subsequent investigators. His studies using female mice developing "spontaneous" mammary tumors (Fig. 1) illustrated the independent and additive tumor-promoting effects of dietary fat concentration and energy intake (Tannenbaum 1942 and 1945). The profound effect of energy intake on mammary tumorigenesis has been often overlooked by many scientists evaluating agents for tumor prevention or therapy in animal models. Although the relevance of these and many similar animal studies to human cancer was frequently discussed during this period, the epidemiologic methodology needed to investigate nutrition and cancer hypotheses in humans had not been adequately developed in the years immediately following World War II. During the 1950s, the possibility that nutrients indirectly modify human cancer incidence was overshadowed by the perceived hazard of additives and environmental contaminants in the food supply.

The landmark investigations of Charles Huggins at the University of Chicago during the early 1960s provided a new rodent model of chemically induced mammary carcinogenesis that is used by investigators around the globe (Welsch 1985). The polycyclic aromatic hydrocarbon 7,12-dimethylbenz()anthracene (DMBA) provides a model of modest cost that is consistent and reproducible over time by many investigators and mimics many hormonal relationships observed in human breast cancer (Huggins 1979). Kenneth Carroll from the University of Western Ontario established that both dietary lipid concentration and source had a striking effect on mammary tumorigenesis in the DMBA model (Carroll and Kohr 1971). A key question, still debated among investigators, concerns the ability of dietary fat concentration to stimulate mammary tumorigenesis independently of energy intake. Data from our own laboratory (Clinton et al. 1984) illustrates one of many studies that addressed these issues. Table 1 shows the effects of diets containing 12, 24 or 48% of energy from corn oil on energy intake, growth and breast tumor incidence in DMBA-treated rats. Dietary fat concentration had no significant influence on energy intake or growth. In contrast, the effect of dietary lipid on tumorigenesis is best described as a linear effect (P < 0.001) with the odds of a rat developing a pathologically confirmed tumor (odds of a tumor/probability of no tumor) multiplied by about 2.15 for each successive doubling of corn oil. The food intake of each individual rat was carefully collected throughout the study. Figure 2 shows the frequency distribution of mean daily self-selected energy intakes for the 351 rats in the study. Superimposed upon the intake distribution is a line graph showing the risk of breast cancer according to the self-selected energy intake. We observed that the odds of developing an adenocarcinoma, adenoma or tumor of any type are multiplied by 1.10, 1.14 and 1.09, respectively, for each 1-kcal increase in self-selected intake. On the basis of our calculations, a drop of average energy consumption of 12-13% is associated with about a 30% reduction in risk of a tumor at necropsy. The effects of lipid and energy observed in our study are typical of many other experiments conducted by investigators employing a diverse array of breast cancer-inducing agents, including direct and indirect chemical carcinogens, hormones, irradiation and viruses (Clinton et al. 1995, Freedman et al. 1990, Rogers and Longnecker 1988). The relevance of these studies to human cancer has been strengthened by more recent studies describing increased growth rates of human breast carcinoma transplants in immune-deficient mice fed high fat diets (Blank and Ceriani 1989, Borgeson et al. 1989, Gabor et al. 1990, Gonzalez et al. 1991). The strength and overall consistency of results with regards to dietary fat and energy in the laboratory studies cannot be easily dismissed, particularly because these same models mimic many aspects of human breast cancer biology and have proven useful for the testing of anti-cancer therapies.

Table 1. Effects of dietary fat concentration as corn oil on energy intake, body weight and tumor incidence in female Sprague-Dawley rats given 7,12-dimethylbenz()anthracene at 56 d of age1 [View Table]

HISTORICAL PERSPECTIVES: NUTRITIONAL EPIDEMIOLOGY OF HUMAN BREAST CANCER

A second component of the puzzle was provided when food availability data and estimated nutrient contents of foods in nations around the globe were compiled. A report by Carroll and Kohr (1975) showed a strong and direct correlation between the per capita availability of dietary fat and national breast cancer rates, which supported the many published animal studies. The hypothesis that dietary fat played a critical role in human breast cancer became a predominant theory of breast cancer etiology for many years. Causal inference based upon ecologic studies alone is impossible because so many other characteristics differ between the populations that could be contributing to cancer risk and the intake of dietary fat. Although nonspecific, ecologic studies allow the evaluation of nutrition intake or a food component across a much wider range than can be conducted within a nation and provide a fertile source of new hypotheses.

Case control studies are based on comparisons of retrospective dietary data between paired groups with and without breast cancer. Many of the past studies were of such limited size that the ability to detect an effect of fat over the narrow range of intake observed was very limited. The major concern, however, is the issue of differential recall of diet by persons with breast cancer compared with controls (Giovannucci et al. 1993). Breast cancer has been the subject of many case control investigations, and recent meta-analyses of published studies suggest a modest increase in risk associated with high fat dietary patterns (Boyd et al. 1993, Howe et al. 1990).

Prospective or cohort studies define a population that is monitored for incidence of disease and exposure to potential risk factors over time. Large cohort studies in the area of nutrition became possible with the gradual evolution and refinement of the food-frequency questionnaire over the last three decades (Willett 1990). The studies avoid inaccuracies of estimating dietary intake retrospectively, and the prospective assessment of diet is unbiased by the cancer experience. The main disadvantage is the enormous cost associated with large studies having long periods of follow-up. To the surprise of many, the prospective studies have not provided data supportive of the dietary fat and breast cancer relationship (Hunter et al. 1996). Indeed, these highly publicized studies are often quoted as proving that a relationship between fat concentration and breast cancer is extremely unlikely. The limitations of current assessment tools to accurately measure exposure and of statistical approaches to correctly dissect independent effects of dietary fat concentration, lipid source, energy balance or body size while adjusting for the other variables should be recognized and efforts directed towards improving our methodology.

Intervention studies offer obvious scientific advantages compared with observational epidemiology and would be most analogous to the laboratory studies in animal models. However, intervention studies will remain few in number due to costs associated with the very large sample population necessary to generate the statistical power needed to evaluate the effects of a dietary risk factor on outcomes that are infrequent in the average population. Furthermore, the inability to ensure or carefully document compliance, as well as ethical considerations that limit major interventions over long time periods, prevents implementation of studies.

HISTORICAL PERSPECTIVES: GENETICS AND BREAST CANCER

Anecdotal references to familial clustering of breast cancer have been described by medical scholars since ancient Rome. The tendency of family members and health practitioners to assume that genetics plays a key role in these observations ignores some alternative contributing factors that may be involved. For example, families and siblings often share the same geographically limited environmental exposures to potential carcinogens. Furthermore, shared cultural and socioeconomic factors may also define dietary patterns or other risk factors such as age of first birth. The recent identification of highly penetrant mutant tumor suppressor genes (BRCA1, BRCA2, p53) associated with breast cancer in young women is an initial step in understanding breast cancer genetics. Family studies to date suggest that members with inherited mutations of these genes have a high probability of developing breast cancer, but the frequencies of these mutations in the general population seem to be low and can account for only a small minority of the observed cases. The majority of breast cancers are not due to the presence of rare highly penetrant genes such as these but are more likely to be related to a combination of more common genes of lower penetrance and those genes modifying host-environment interactions. Polymorphisms in genes controlling lipid and energy metabolism, carcinogen activation and detoxification, hormone synthesis and clearance, signal transduction pathways and immune function are beginning to be characterized, and these may potentially interact with diet, nutrition and other environmental variables to modulate breast cancer risk.

SUMMARY