Dietary fat, serum estrogen levels, and breast cancer risk: a multifaceted story.

The hypothesis that dietary fat may increase the risk of breast cancer by increasing the availability of estrogen and related sex steroids has been explored in basic, epidemiologic, and dietaryintervention studies for many years. In this issue of the Journal, Wu et al. (1) clearly describe the debate regarding the dietary fat and breast cancer hypothesis (2,3) and have contributed to the debate by the use of meta-analysis to quantify the combined effect of fat in dietary intervention studies that have examined the influence of fat reduction on various serum estrogen levels. Their analysis focuses on dietary fat and serum estradiol, a reasonable starting point in attempting to summarize the complex interactions of diet with sex steroids. The dietary intervention studies included in the meta-analysis generally examined the effect of reducing dietary fat intake to 18%–25% of total energy in comparison with various “control” diets of 35%–40% energy from fat. However, in interpreting the results of these studies, a number of issues must be considered: 1) the changes that occur in other dietary components in association with dietary fat reduction, 2) the effect of other dietary components on serum estrogens, and 3) measurement error for both diet and serum estrogens. With respect to the first issue, examination of Table 1 in the article by Wu et al. (1) and descriptions of the diets in the controlled dietary studies that they analyzed reveals a wellknown phenomenon. Dietary fat reduction can be accompanied by changes in other dietary components—including total calories, fiber, carbohydrates, fruits and vegetables, carotenoids and other micronutrients—that may also alter sex steroid metabolism and serum estrogen levels (4,5). For example, in five of 13 studies in this meta-analysis, the intervention protocol included markedly increased amounts of dietary fiber. Some of the dietary protocols were based on Asian diets, others on vegetarian, or on the National Health Lung and Blood Institute Step II diet for cholesterol reduction. Finally, the approach to dietary intervention varied from providing counseling to change the diet, with assessment of intake by self-report, to metabolic ward studies where women were fed a defined diet and intake was recorded from trays. While many studies have included isocaloric high-fat control diets in an effort to avoid differential weight gain or loss during intervention and control periods, a number of low-fat dietary intervention studies have reported modest weight losses or avoidance of weight gain in intervention periods compared with weight gain in control periods. The domino effect that change in one dietary parameter has on other diet and related parameters demonstrates the difficulty of isolating the effect of a single dietary factor on a chronic disease, particularly when multiple dietary factors are hypothesized to be involved, often by similar mechanisms. With respect to the second issue, a number of dietary components have been examined for a possible effect on estrogen metabolism. Studies (4,5) of dietary fiber suggest that increases in fiber are inversely associated with levels of serum estradiol and other estrogens. More recently, dietary and supplement interventions have examined the effect of a variety of “phytoestrogenic” compounds, such as genistein and related soy products, for their possible estrogenic effect. The effect of these compounds varies by dose and compound and is generally ascribed to their dose-dependent activity as estrogens or antiestrogens. Severe energy restriction has long been known to alter hormonal metabolism; however, the effect of modest decreases in energy commonly seen in low-fat dietary interventions is less well understood (6). One of the mechanisms by which dietary fat is presumed to reduce estrogen levels is by lowering overall energy intake and consequently reducing adipose tissue storage and production of hormones. Among premenopausal and postmenopausal women, increased overall and central adiposity is associated with increased levels of bioavailable estrogen, and, in the case of postmenopausal women, with increased breast cancer risk. Some dietary intervention studies attempt to control for the confounding effect of energy and weight change by maintaining isocaloric conditions. However, low-fat diets that are high in fiber appear to have substantially different metabolic effects compared with isocaloric diets high in fat and low in fiber in terms of effects on weight and hormone metabolism. In addition, the variability of an individual’s response to any specific dietary alteration is a well-recognized phenomenon (7) and may explain, in part, the apparent lack of response to dietary fat reduction found in some studies (8,9). Finally, alcohol, while not commonly included in controlled dietary studies of nutrients, has been found to increase risk for breast cancer in a number of epidemiologic studies (10) and is hypothesized to alter hormone metabolism (11). With respect to the third issue, we will focus on measurement error and compatibility issues in diet and serum estrogen studies. Key issues to consider in comparing changes in serum estradiol include types of assays, lab variability, menopausal status of study participants, and, for premenopausal and perimenopausal women, timing of blood specimen collection relative to the menstrual cycle. Radioimmunoassays are the current standard for serum hormone measurement but were not commonly used in some of the earlier controlled dietary studies. Intraindividual and interindividual variability in laboratory measurement of serum hormones (12) has led to the use of a single reference laboratory performing assays for large-scale dietary intervention studies, such as the Women’s Health Intervention. Furthermore, estrogen levels decline markedly with menopause, often to the limits of