Reversal of relation between body mass and endogenous estrogen concentrations with menopausal status.

For more than a decade, results from published studies have indicated that the risk of breast cancer may be lower among heavy premenopausal women than among their less obese counterparts (1-8). In heavy postmenopausal women, however, this risk is either increased or unchanged in comparison with nonobese postmenopausal women (9,10). The mechanism explaining this reversal in risk is unknown, although theoretical explanations have been proposed. Obese postmenopausal women may have elevated risk from higher levels of circulating estrogens secondary to increased conversion of androgen to estrogen in adipose tissue (11) and a higher proportion of bioavailable estrogen due to the low levels of sex hormone-binding globulin (SHBG) (12,13). Obese premenopausal women may exhibit a greater degree of anovulation, resulting in lower levels of both progesterone and estradiol, lower breast cell division rates, and, consequently, a lower risk of breast cancer (13-15). We had the opportunity to evaluate the relation of body mass and hormonal profiles in a group of 195 community and 103 hospital control subjects from a study of endometrial cancer described elsewhere (16,17). Results of both control groups were similar; therefore, the two groups were combined. Body mass index (BMI) (kg/m) was calculated from reported current height and weight obtained during an interview. Fasting serum samples were generally drawn within 1 month of interview and prior to surgery for benign endometrial conditions for the hospital control subjects. Serum was extracted and separated by celite column chromatography. Estrone, estradiol, androstenedione, estrone sulfate, progesterone, and SHBG were measured by radioimmunoassay (RIA). An enzymatic hydrolysis was performed before separation for estrone sulfate. Albumin-bound estradiol was assessed using ammonium sulfate precipitation, and percent-free estradiol was estimated with an equilibrium dialysis assay. The amount of free or albumin-bound estradiol was calculated by multiplying the total estradiol concentration by the percent in the other fractions. Serum was analyzed by Nichols Institute, Inc. (San Juan Capistrano, CA). We previously documented the reliability of this laboratory for steroid hormone assays (18). Blind quality control serum evaluated with each daily batch of samples demonstrated adequate reproducibility for all assays, particularly for estradiol in postmenopausal women (overall coefficient of variation = 16.6). Written informed consent was obtained from all study participants, and procedures were approved by institutional review. Of the 477 and 253 eligible community and hospital control subjects, we were able to successfully interview 313 and 206 of these subjects, respectively, regarding demographic, medical, and reproductive factors. Following the interview, women were measured for a variety of anthropometric indices, and blood samples were obtained from 217 community and 139 hospital control subjects. As is common in population surveys, we obtained one blood sample from each subject. Clearly, multiple samples would better characterize a woman; however, one sample is generally adequate for epidemiologic purposes (19). Women were excluded from the analysis if they had no BMI data, if they had reported use of exogenous estrogens or oral contraceptives within 6 months, or if they had hormone values that indicated Premarin usage or perimenopausal status. Women were considered postmenopausal if they reported not having had a menstrual period for 6 or more months or if they reported having had a menstrual period within 6 months but exhibited a measured estradiol value of less than 20 pg/mL (laboratory cutoff for postmenopausal status). All premenopausal women reported having had a menstrual period within 6 months and all but one had experienced menses within 1 month of the interview. Analysis of covariance was used to determine age-adjusted mean values within menopausal groups, and ordinary least-squares regression was used to test the trend of BMI with hormone concentration. An interaction term was entered into the regression to test the significance of noted differences of the association of BMI with hormones by menopausal status. All P values resulted from use of two-sided statistical tests. Mean values across BMI tertiles revealed the expected decrease in SHBG with higher BMI in both menopausal groups (Table 1). As the BMI increased, however, total estradiol decreased in premenopausal women (P for trend = .11) and increased in postmenopausal women (P .0001). Free estradiol and albumin-bound estradiol (i.e., the bioavailable fraction of this hormone) were found to increase with increasing BMI in postmenopausal women; a similar tendency for albumin-bound estradiol was observed in premenopausal women. Estrone and estrone sulfate were higher in heavy postmenopausal women but showed no relation with BMI among premenopausal women. Only the relation of BMI to estradiol differed significantly by menopausal status (P = .02). When the analysis was restricted to premenopausal women with progesterone values consistent with being in the follicular phase of their menstrual cycle (50 ng/dL progesterone or less), the results were similar (Table 2) with the exception that all three estradiol fractions decreased with increasing BMI (P for trend = .03 estradiol; P = .06 free estradiol). Interestingly, the proportion of subjects in the luteal phase actually increased as BMI increased; 40%, 44%,