Over 20 y ago, de Waard (1975) reviewed the epidemiologic research on weight, height and breast cancer in search of clues about the role of nutrition in breast cancer etiology. He postulated that the underlying mechanisms would involve endogenous hormones and metabolism. Since that time, however, epidemiologic and experimental research on breast cancer has focused on specific dietary factors, such as elevated fat intake, that might cause breast cancer. The importance of weight and height in human breast carcinogenesis is still controversial. Excess weight has been presented to the medical community as a weak and clinically unimportant determinant of breast cancer in postmenopausal women, and height as a risk factor that is not yet firmly established (Harris et al. 1992). However, recent reviews have suggested that height and adiposity (weight adjusted for height), along with weight change and patterns of fat deposition, merit more careful consideration (Ballard-Barbash 1994, Hunter and Willett 1993).

HEIGHT

Table 1. Relative risks1 (RR) of breast cancer by height in women participating in a U.S. multicenter breast cancer screening program2 [View Table]

In our population-based case-control study of breast cancer among Asian-American women, height was strongly predictive of breast cancer risk, even after adjustment for accepted breast cancer risk factors (Ziegler et al. 1996). Risk doubled (RR = 2.0) over the 18-cm range in height (from 151 cm to 166 cm) (P for trend = 0.003), with comparable effects in premenopausal and postmenopausal women. This study, conducted among women of Chinese, Japanese, and Filipino ethnicity, 20 to 55 y of age and living in San Francisco-Oakland and Los Angeles (California) and Oahu (Hawaii), was designed to take advantage of the diversity in lifestyle and breast cancer risk in these migrant populations. Breast cancer incidence rates have historically been four to seven times higher in the United States and many other Western countries than in Asia. When Chinese, Japanese or Filipino women migrate to the United States, their risk of breast cancer rises over several generations and reaches that for U.S. white women (Ziegler et al. 1993), indicating that modifiable exposures, related to lifestyle or environment, are involved. Within our study population, we have demonstrated a sixfold gradient in breast cancer risk by migration history, comparable to the international differences in breast cancer incidence rates (Ziegler et al. 1993).

It is frequently assumed that height is related to breast cancer risk only in those populations in which inadequate intake of energy and nutrients in childhood and adolescence limits growth, such as developing countries and European nations coping with food shortages during World War II (London et al. 1989, Tretli 1989). This explanation is consistent with numerous animal experiments demonstrating that energy intake restriction inhibits both spontaneous and induced mammary carcinogenesis (Tannebaum and Silverstone 1953). Furthermore, because events during childhood and adolescence determine adult height, early exposures, possibly affecting mammary mass (Albanes and Winick 1988), may also be critical in breast carcinogenesis (Hunter and Willett 1993, Swanson et al. 1988). Recently, it has been postulated that height reflects the total number of ductal stem cells that develop in the breast in utero and thus the importance of prenatal exposures (Trichopoulos and Lipman 1992).

However, an enhanced risk of breast cancer among taller women has been demonstrated in populations, such as the United States population, in which energy or nutrient deficiency is not evident and tallness is determined primarily by genetic factors (Ballard-Barbash 1994, Hunter and Willett 1993, Ziegler et al. 1996). Thus it is possible that inherited patterns in endogenous hormones and growth factors contribute to the height attained prior to epiphyseal closure at puberty and also to the promotion of breast carcinogenesis, either at puberty when breast tissue is rapidly developing or at a later stage in the life cycle. Estrogen and progesterone are generally believed to be the hormones that determine breast cancer risk (Bernstein and Ross 1993), but androgens, growth hormones, insulin and insulin-like growth factors deserve consideration. In addition, dietary exposures other than energy deprivation may influence height; an overabundance of energy and fat and variation in macronutrient intake in the years prior to puberty may play a role.

ADIPOSITY

The inverse association between adiposity and breast cancer in younger women was initially attributed to earlier detection of breast tumors in leaner women, but further analyses that considered tumor stage did not support this hypothesis (Brinton and Swanson 1992, London et al. 1989). At present, hormonal mechanisms are being evaluated, specifically whether heavier premenopausal women have more anovulatory menstrual cycles, resultant decreases in estrogen and progesterone exposure, and lower luteal phase progesterone levels in ovulatory cycles (Key and Pike 1988).

In our breast cancer study in Asian-American women, the dominant pattern was increasing risk with increasing adiposity (Ziegler et al. 1996). Among women in their 40s and in their 50s, risk more than doubled when extreme deciles of usual adult adiposity³ were compared (>31.3 kg/m^1.5 vs. <22.9 kg/m^1.5) (P for trend = 0.05 for both age groups). Only in the heaviest of the youngest women (>29 kg/m^1.5 and <40 y) was breast cancer risk substantially reduced. Thus the positive association with adiposity observed among the Asian-American women was stronger and apparent at younger ages than in other U.S. studies of breast cancer, in which RR generally range from 1.0 to 1.5 in postmenopausal women (Ballard-Barbash 1994, Hunter and Willett 1993, Ziegler et al. 1996). Our findings are consistent with a meta-analysis of case-control data from countries at high, moderate and low risk for breast cancer, which demonstrated that breast cancer incidence rates consistently increased with adiposity among both premenopausal and postmenopausal women except for premenopausal women from high risk countries, where an inverse relationship was noted (Pathak and Whittemore 1992).

Most epidemiologic studies of breast cancer have utilized a single static measure of adult weight. However, in many women, weight fluctuates during adult life, and the timing and magnitude of such changes may be crucial in understanding their importance in cancer etiology (Ballard-Barbash 1994). In our study of Asian-American women, we asked about usual weight during each decade of adult life, excluding the most recent 3 y. Therefore, we could examine the influence of adiposity at different periods over a lifetime and at different stages of carcinogenesis (Ziegler et al. 1996). Recent adiposity influenced breast cancer risk more than earlier adiposity. For women in their 50s (Table 2) and in their 40s (data not shown), positive trends in risk with increasing adiposity became more striking as the decade in which breast cancer was diagnosed was approached. We wanted to determine whether the remarkable effect of excess weight in the years immediately preceding breast cancer diagnosis that we observed was suggested in other studies and other populations, but we have been unable to find other breast cancer studies that evaluated recent adiposity in older women.

Table 2. Relative risks1 (RR) of breast cancer by adiposity during specific decades of adult life in Asian-American women in their 50s [View Table]

The critical importance of excess weight in the years preceding breast cancer diagnosis is consistent with the many recent reports showing that adiposity at the time of breast cancer diagnosis is associated with an increased probability of recurrence and a decreased survival time, even after adjusting for stage and treatment (Ballard-Barbash 1994, Senie et al. 1992, Tretli et al. 1990). If adiposity can enhance tumor growth after diagnosis, it should be expected to promote tumor development and growth also in the late stages of breast carcinogenesis prior to clinical detection.

WEIGHT CHANGE

In our study of Asian-American women, adult weight gain was associated with increased breast cancer risk for women in their 40s and 50s; however, further analyses suggested that it was primarily important as a determinant of adiposity in the years preceding diagnosis (Ziegler et al. 1996). Recent weight change, (between the current and preceding decade) was more predictive of breast cancer risk than adult weight change. For women in their 50s, a recent gain of more than 10 pounds was associated with a doubling of risk (RR = 2.3), relative to no recent weight change (P for trend = 0.002). Recent weight loss was associated with a reduced risk of breast cancer in all age groups (RR approximately 0.7), relative to no change in weight. Although the effects of adiposity and weight change in the decade preceding diagnosis were not totally independent, neither measure fully explained the effect of the other.

The importance of recent weight gain and recent adiposity in these Asian-American women suggests that excess weight may function as a late stage promoter of breast carcinogenesis. Whether our findings in Asian-American women can be generalized is not a simple question (Ziegler et al. 1996). Asian-American women are leaner than other U.S. women. They adhere to distinctive dietary and physical activity patterns. The protective influence of adiposity in early adult life is minimal in this population. However, genetics is not the explanation, because these Asian-American women acquire the high breast cancer rates of U.S. whites after several generations of acculturation.

FAT DEPOSITION PATTERNS

Mechanisms that might account for the relationship of adult adiposity, adult weight gain, abdominal fat deposition and breast cancer in older, postmenopausal womena relationship demonstrated in many epidemiologic studieshave been proposed but not proven. Estrogen production in adipose tissue from circulating androgens is elevated in heavy women, may promote tumor growth, and becomes increasingly important as ovarian estrogen production diminishes with age (Bernstein and Ross 1993, Key and Pike 1988). The effect could be localized, with especially high estrogen levels near breast adipose tissue. In addition, the decreased sex hormone-binding globulin levels and the increased triglyceride levels associated with excess weight increase the bioavailability of estrogen (Ballard-Barbash 1994). Alternatively, the elevated levels of insulin and growth factors associated with adiposity may promote tumor growth, either directly or by modulating steroid activity (Bruning et al. 1992, Stoll and Secreto 1992).

Relatively consistent associations between anthropometric measures, specifically height, adiposity, weight gain and body fat deposition, and breast cancer risk have been emerging in recent years. The patterns suggest that lifestyles leading to a positive energy balance are involved in the etiology of breast cancer and that energy intake and physical activity may be more critical than dietary fat. Many unanswered questions about anthropometric exposures remain for the epidemiologists: whether there are especially susceptible periods in the lifespan or in breast carcinogenesis, how body size and shape interact with a family history of breast cancer and with reproductive and menstrual risk factors, over what ranges of body size and shape are effects on breast cancer seen. However, experimentalists will need to identify the hormonal and metabolic mechanisms underlying the epidemiologic associations. Experimental studies ranging in focus from the molecular level to the metabolic laboratory will be required. The challenge will be to enhance communication and collaboration between experimental and epidemiologic researchers so that the findings of each approach can be rigorously evaluated by the other.