Body size matters in breast carcinoma etiology. A consistent finding of an increased risk of breast carcinoma in tall women compared with short women in early epidemiologic studies in white women1 was believed to reflect childhood nutritional influences on breast carcinoma development. Women who were well nourished as girls (and thus at an increased risk from dietary mechanisms) were likely to attain a greater adult height. This early nutritional status was believed to influence later breast carcinoma risk. Several subsequent studies showed an increased risk for breast carcinoma in overweight or obese women,1, 2 and the association remained after correction for height. Later studies found a difference in the effect of body mass index depending on age at breast carcinoma onset; overweight or obesity increased the risk of breast carcinoma in postmenopausal women but decreased the risk in premenopausal women.1, 3 Correlation studies pointed to potential hormonal explanations for the height and body mass associations with breast carcinoma; girls who were heavier were more likely to experience early menarche and thus were exposed to reproductive hormones for longer periods of time. The finding that postmenopausal women produce estrogens in their fat tissue, through aromatization of ovarian and adrenal androgens, provided another explanation for the association between body mass index and breast carcinoma because obese postmenopausal women have higher levels of endogenous estrogens compared with lighter postmenopausal women.4, 5 Recent epidemiologic cohort studies have confirmed an association between body mass index and breast carcinoma, and have identified differences in associations depending on age at onset of the disease. Recent biologic studies have pointed to potential body mass-mediated roles of endogenous sex hormones, growth factors, and metabolic hormones in the etiology of breast carcinoma.
“Energy balance” is comprised of the interplay of diet (quantitative and qualitative), physical activity, genetics, and environmental stressors such as ambient temperature. In this country, the latter factor plays a lesser role in energy balance. Genetics influences on the predisposition toward overweight and obesity may be of importance in breast carcinoma, although studies in this area are lacking. This report focuses on the remaining pieces of the equation: body mass index and physical activity.
Few epidemiologic studies of body mass index and physical activity and breast cancer risk have included data regarding women from certain racial and ethnic backgrounds, specifically African-Americans, Hispanics, American Indian/Alaskan Natives, and Asian American/Pacific Islanders. Some studies have been conducted on Japanese women in Japan, but it is not clear if the results of such studies would pertain to women of Japanese descent born in the U.S. The biologic mechanisms hypothesized for white women may or may not pertain to women of other racial and ethnic backgrounds. Finally, the differences in body mass index across these groups will likely be reflected in the differences in attributable risk for breast carcinoma associated with body mass and physical activity across populations.
Epidemiologic Studies of Body Mass and Breast Carcinoma in Minorities in the U.S.
There have been few epidemiologic studies of body mass index and risk for breast carcinoma in U.S. minority populations. A small number of studies included some minority groups, such as African-American women, but did not report data for these women separately from white women.6, 7 The Nurses' Health Study and the Iowa Women's Health Study, two large cohort studies that have examined breast carcinoma risk according to several anthropometric factors, are comprised largely of white women.3, 8
A multiethnic cohort study of 17,628 women in Hawaii that included Japanese, Filipino, Chinese, Hawaiian, and white women found an association between increased body mass index and risk for breast carcinoma (n = 378 cases; P trend = 0.0001) (Table 1).9 The authors reported no difference in relative risk by race/ethnic group (72% of the cohort was nonwhite). In that same study, breast carcinoma patients at the 75th percentile or greater for body mass index had a 2.2 times increased risk of dying of breast carcinoma compared with lighter patients. A population-based case–control study of anthropometric effects on breast carcinoma risk was conducted in 597 Asian-American cases and 966 Asian-American controls.10 The study included Chinese, Japanese, and Filipino women living in California and Hawaii. Cases were obtained from population-based cancer registries, and controls were selected by random digit dialing (California) or a population sample (Hawaii). Among both premenopausal and postmenopausal women, increasing body mass index was associated positively with an increased risk for breast carcinoma, although the test for trend was not significant (Table 1). The authors reported that breast carcinoma risk was associated positively with usual adult relative weight in all three ethnic groups studied. However, risk was limited to women in their 40s and 50s. For women in their 50s, a gain of > 10 pounds in the preceding decade was associated with a doubling of the risk of breast carcinoma compared with no recent weight change. Furthermore, a recent loss in weight was associated with a reduced risk of breast carcinoma in all age groups, compared with stable weight.
Table 1. Body Mass Index and Risk of Breast Carcinoma in Minority Women in the U.S.
Type of study
Case ages (yrs)
No. of cases
No. of noncases
RR (95% CI), highest vs. lowest level
RR: relative risk; 95% CI: 95% confidence interval; Pre: premenopausal; Post: postmenopausal; CASH: Cancer and Steroid Hormone.
Two studies of anthropometrics have been conducted in non-American Asians. A breast carcinoma case–control study in Singaporean Chinese women ages 45–69 years (n = 204 cases and 882 controls) found that although obesity did not predict risk for breast carcinoma, central obesity was strongly and significantly related to risk.11 Women in the highest quintile of waist-to-hip ratio were > 9 times (95% confidence interval [95% CI], 4.6–17.5) more likely to develop breast carcinoma compared with women in the lowest quintile (P = 0.0001). In Japanese women in Japan, women who were in the highest quintile for body mass index had double (95% CI, 1.49–2.03) the risk for breast carcinoma compared with women in the lowest quintile.12 Furthermore, they found that weight gain in later life was associated positively with risk for breast carcinoma, regardless of body mass index in early adulthood. Women at greatest risk from adult weight gain tended to be postmenopausal and age ≥ 60 years.
A small number of case–control studies of breast carcinoma have focused on African-American women.13 In a small, hospital-based case–control study in African-American women, cases were less likely to be overweight or obese compared with controls (Table 1).14 However, the controls in that study may not have been comparable to cases; the report did not document the total number of eligible cases and controls from which the samples were selected. Another hospital-based case–control study (in which cases were female African-American patients in seven New York hospitals and controls were female African-American noncancer patients) found no association between Quetelet's index (an index of weight corrected for height) and breast carcinoma risk; the means for this index were nearly identical for cases and controls.15 A larger hospital-based case–control study of 528 African-American breast carcinoma patients and 589 African-American controls found an increased risk for breast carcinoma in postmenopausal women only, after multivariate adjustment (Table 1).16 The Cancer and Steroid Hormone (CASH) case–control study included 490 African-American breast carcinoma cases and 485 controls ages 20–54 years.17 In that study, body mass index either at age 18 years or as an adult was not found to be related to breast carcinoma risk in African-American women (Table 1), and the effect did not vary with multivariate adjustment.
A case–control study in Mexican-American women in New Mexico included data regarding body mass index in cases and controls (Table 1).18 From raw data provided in the report, Mexican-American women in the highest tertile of body mass index had a 2.6 times increased risk of developing breast carcinoma compared with women in the lightest tertile. After multivariate logistic regression adjustment, the results remained positive (F. Gilliland, personal communication). To the authors' knowledge no studies have been reported on breast carcinoma risk and anthropometrics or physical activity in American Indian or Alaskan Eskimo women.
Epidemiologic Studies of Physical Activity and Breast Carcinoma in U.S. Minorities
There have been more than two dozen studies of physical activity in relation to breast carcinoma in women; these have been reviewed in depth elsewhere.19, 20 Few of these studies included U.S. women other than whites, and of those few, data were not reported separately by race or ethnicity. In white women, on average high levels of physical activity have been associated with an approximately 30% reduction in risk for breast carcinoma,19, 20 although some large cohort studies have found no effect of physical activity on breast carcinoma risk.21 Effects differed variably within studies by menopausal status, parity, and other breast carcinoma risk factors. Two studies were conducted in Japanese women in Japan.22, 23 A hospital-based case–control study of 1045 cases and 21,034 controls found a 26% reduced risk of breast carcinoma in women at the highest level of recreational physical activity compared with women at the lowest level (95% CI, 0.55–0.99) in premenopausal women.22 Nearly identical estimates were found in postmenopausal women. A Japanese population-based case–control study of 154 cases and 361 screened controls found a 47% reduced risk of breast carcinoma in women reporting to be in the highest level of recreational physical level compared with women in the lowest level (95% CI, 0.19–1.52).23
Potential Etiologic Mechanisms
Fat tissue storage
Fat is a living, functioning organ system, comprised of cells, supporting structures, and blood vessels. It stores toxins, medications, and certain vitamins.24 Stored toxins could serve as a continuous source of carcinogens; thus women with more fat tissue might be exposed to a continuous source of carcinogens. Storage of carcinogens could be of particular importance to women in some racial or ethnic groups who might be exposed to high levels of carcinogens (for example, Hispanic migrant agricultural workers). Alternatively, stored vitamins (such as fat-soluble antioxidant vitamins A, D, and E) could be cancer-protective,25 which might decrease the observed risk estimate for obesity and breast carcinoma risk.
Reproductive system and sex hormone effects
Events of early and late menstrual and reproductive life may be important in the induction or promotion of breast carcinoma.26 Girls who are overweight or obese or who are sedentary on average experience menarche and regular ovulation at an earlier age than girls who are thin or who participate in vigorous exercise.27 An early age at menarche (before age 12 years) increases the risk for breast carcinoma by approximately 30–50%.26 Exercise during reproductive life has been shown to decrease the number of fertile menstrual cycles,28 alter the concentrations of sex hormones,29 and, if particularly vigorous, lead to amenorrhea.30 Another plausible mechanism for the effects of obesity and exercise on breast carcinoma risk is through effects on circulating sex hormones in postmenopausal women. As mentioned earlier, postmenopausal women may continue to produce estrogen through the peripheral conversion (mainly in fat cells) of adrenal or ovarian androgens to estrogens.4 Elevated circulating estrogen levels in postmenopausal women have been found in several recent studies to increase the risk of breast carcinoma,31–33 In a multivariate model, obesity and a sedentary lifestyle have been found to independently predict higher serum concentrations of estradiol, estrone, and androgens and lower levels of sex hormone-binding globulin in postmenopausal women.5 The trends in increasing prevalence of overweight are paralleled by the increasing rate of incidence of breast carcinoma in African-American and white women in this country (Fig. 1). Although many risk factors take several years to impact the risk of breast carcinoma, overweight and obesity could have a more immediate effect if they cause an increase in endogenous levels of estrogens, which in turn can promote breast carcinoma growth.
Metabolic and growth hormones
Increased blood insulin concentrations have been associated with breast carcinoma risk. Women who are lighter or who exercise regularly have significantly lower insulin, glucose, and triglyceride levels and higher high density lipoprotein cholesterol levels.34 Thus, weight control and physical activity could exert a protective effect against breast carcinoma through a metabolic pathway. Insulin-like growth factors (IGFs) stimulate cell turnover in the majority of body tissues and have been associated with an increased risk of breast carcinoma.35 IGF is down-regulated by increased production of its binding protein (IGFBP-1), which can result from increased exercise, decrease caloric intake, and decreased body weight. Decreased IGF activity may increase the hepatic synthesis of sex hormone-binding globulin, resulting in diminished availability of free sex hormones. Thus, low body weight and increased exercise could result in lowered biologically available endogenous sex hormones via a cascade of metabolic events, and thus a lowered risk of breast carcinoma.
Another possible mechanism of the association between exercise and breast carcinoma may be via immunologic effects. The immune system is capable of recognizing and eliminating neoplastic cells, and the effects of exercise on the immune system have long been postulated.20 It has been suggested that the relation between exercise and immune functions follows a “J”-shaped curve, with the lowest risk observed among individuals who undertake regular moderate exercise.36 Long term effects of endurance training with potential protective effects against cancer include increases in the number and/or activity of macrophages, natural killer cells, and lymphokine-activated killer cells and their regulating cytokines.20 Furthermore, endurance training can result in increased mitogen-induced lymphocyte proliferation rates. In randomized controlled trials37, 38 moderate exercise training over 8-week or 12-week periods was not found to be associated with T-cell or natural killer cell function; however, the number of days with symptoms of upper respiratory tract infection was reduced in women in the exercise groups compared with those in the nonexercising groups. Last, a potential effect of exercise on immune function may be mediated via sex hormones. Sex hormones exert immunoregulatory effects in both in vivo and in vitro situations.39 T-cell subsets, macrophages, immunoglobulin synthesis, and various cytokines are targets for sex hormones. Progesterone and androgens appear to suppress the immune system, whereas estrogens can be either stimulatory or suppressive.
Body fat distribution
Exercise may have a beneficial effect against cancer development through a reduction in the highly metabolic abdominal fat mass. Abdominal fat, particularly visceral (intraabdominal) fat, appears to be the most metabolically active of fat deposits.40 Changes in visceral fat of as little as 3–4 pounds can have profound effects on glucose tolerance, fasting insulin, and lipids.40 Aerobic exercise preferentially causes a reduction in visceral fat.34 Conversely, weight loss from caloric restriction causes diffuse fat and fat free mass loss, sometimes with little effect on visceral fat.40
Prevalence of Obesity and Sedentary Life-Style in the U.S. by Race and Ethnicity
The prevalence of obesity and a sedentary life-style are high in women from some racial and ethnic groups in the U.S. Data from the National Health and Nutrition Examination Surveys (NHANES) indicate that 48.6% of non-Hispanic black women and 46.7% of Mexican-American women are overweight as defined by a body mass index of ≥ 27.3.41 The prevalence of overweight is highest in women in these 2 groups between ages 40–69 years, during which as many as 50–60% of women have a body mass index of ≥ 27.3. An examination of NHANES surveys over time indicates that the prevalence of overweight has increased. Among African-American women, 41.6% were overweight in 1960–1962 compared with 49.2% in 1988–1991. The prevalence of overweight in Mexican-American women increased from 41.5% in 1982–1984 to 48.1% in 1988–1991.
Another analysis of NHANES data showed that the trends in increasing overweight are reflective of increases in the highest levels of obesity, especially in the proportion of women with a body mass index of ≥ 30.42 Between 1976–1984 and 1988–1994, the prevalence of class 2 and 3 obesity (a body mass index ≥ 35.0) increased by approximately 33% in non-Hispanic black and Mexican-American women.
A telephone survey study of American Indians in 1985–1996 found a similar prevalence of overweight or obesity.43 This survey, which included 2575 women from Alaska, North and South Dakota, Oklahoma, New Mexico, Arizona, Washington, and Oregon, found that between 31.1–46.4% of women had a body mass index of ≥ 27.8. Furthermore, the prevalence of overweight increased over the study survey period in all areas except Alaska; on average, the adjusted body mass index increased significantly by 0.1–0.2 units per year.
The majority of women in the U.S. are physically inactive, especially in their older years.44 A small survey of women from diverse racial and ethnic groups indicated that very few (approximately 10%) participated in regular or vigorous physical activities at levels that might be protective against breast carcinoma.45 Data from NHANES and the Behavioral Risk Factor Surveys support these data.46, 47
If overweight, obesity, or a sedentary life-style are related causally to the risk of breast carcinoma in postmenopausal women, the significant increases in overweight or obesity prevalence and the low levels of physical activity among these racial and ethnic groups may result in increases in breast carcinoma incidence or mortality rates in the future.
Proposed Research Agenda
Few data are available regarding the etiologic relations among body mass index, physical activity, and risk of breast carcinoma in minority women in the U.S. A research agenda should be established to obtain high quality data in specific racial and ethnic populations. There are ongoing cohort studies in the U.S. that include large numbers of racial and ethnic minority women. For example, the Women's Health Initiative has enrolled > 28,000 African-American, Hispanic, American Indian, and Asian-American women into long term clinical trials and a cohort study.48 Excellent data are being collected in that study, including measured height, weight, waist-to-hip ratio, and self-reported physical activities. These variables are collected at baseline and during follow-up so that changes in these factors also can be assessed and related to breast carcinoma risk. These studies will provide important information regarding breast carcinoma risk in women from specific groups.
Case–control studies specifically aimed at determining the effects of lifetime physical activity and weight patterns in particular racial and ethnic minority groups would be helpful. There are several ongoing case–control studies that include these variables in women of all races and ethnicities in their particular study areas. Investigators might be encouraged to pool their data regarding specific minority groups because there may be too few cases and controls of a particular group to make inferences from one study alone.
The measurement of weight, body mass index, and physical activity may differ in different racial and ethnic groups. For example, many white, obese women in the U.S. underreport weight and overreport physical activity,49, 50 possibly because of cultural attitudes regarding weight in this group. In groups such as some African-American or Hispanic women, for whom overweight may be culturally acceptable, the reporting of weight may be more accurate. Needs for measurement of physical activity vary by racial and ethnic group.51 Common physical activities vary by group, such as household chores, dances, religious activities, and sports and recreation.51 Furthermore, women from some groups may not report some activities as being physical activities. Research to determine culturally acceptable and validated means of collecting physical activity information is needed.
The role of body fat mass and distribution in breast carcinoma risk should be defined in different groups of women. African-American women, for example, may have less abdominal fat and more subcutaneous fat compared with white women with similar body mass indices.52 In the NHANES 1988–1994 surveys, 9209 women had measures of bioelectrical impedance, from which total body fat mass and lean mass can be calculated.53 Approximately 2500 each of non-Hispanic black and Mexican-American women were included in these measurements. The impact of body fat mass on breast carcinoma biomarkers such as benign breast disease, mammographic density, and endogenous hormones could be assessed in this very important cohort.
Small-scale clinical trials of the effects of weight loss in overweight or obese women, or of physical activity in sedentary women, should be conducted to examine the effects of these changes on biomarkers of breast carcinoma risk.54, 55 For example, the effect of weight loss on endogenous sex and metabolic hormones in obese postmenopausal women from diverse racial and ethnic groups could be tested in randomized controlled trials of weight loss interventions. Similarly, clinical trials concerning exercise could be conducted in overweight, sedentary women of specific racial and ethnic backgrounds to test exercise effects on hormones, immune function, mammographic density, and other potential biomarkers.54
The cultural acceptance of weight loss and physical activity may differ by race or ethnicity. If weight control or physical activity are to be promoted as public health interventions to reduce the impact of cancer, it will be important to learn how various groups approach these issues, whether they find the concepts acceptable, and what kinds of interventions will work in different groups. This research area might benefit from a series of focus groups, followed by survey research studies.
The prevention of weight gain in adolescence and young adulthood may be important in reducing breast carcinoma risk.3 African-American girls and young women are at high risk of gaining weight in young adulthood.56 Furthermore, young African-American women are twice as likely as young white women to experience a decline in fitness after 7 years, as measured by a symptom-limited exercise tolerance test.57 Research is needed to determine methods of preventing weight gain and fitness declines in women in all racial and ethnic groups.
Overweight, obesity, and a sedentary life-style all are potentially modifiable risk factors for breast carcinoma. Data are needed regarding the relations between measurement of body mass index, physical activity, and risk for breast carcinoma in specific minority populations in the U.S. The biology of breast carcinoma in relation to energy balance and physical activity in racial and ethnic minority groups must be defined. Finally, interventions to reduce or prevent overweight, obesity, and sedentary life-styles are needed, as well as studies to measure the effect of changes in these factors on biomarkers of breast carcinoma.