Vol. 20, No. 1 Printed in Great Britain
International Journal of Epidemiology ©International Epidemiological Association 1991
Associations of Body Mass and Fat Distribution with Sex Hormone Concentrations in Postmenopausal Women Kaye S A (Division of Epidemiology, School of Public Health, University of Minnesota, 1-210 Moos Tower, 515 Delaware St, SE, Minneapolis, MN 55455, USA), Folsom A R, Soler J T, Prineas R J and Potter J D. Associations of body mass and fat distribution with sex hormone concentrations in postmenopausal women. International Journal of Epidemiology, 1991,20: 151-156. The associations of body mass and body fat distribution, as measured by waist-to-hip circumference ratio, with serum concentrations of sex hormones and sex hormone binding globulin were examined in 88 postmenopausal women. Body mass index (BMI) was significantly and negatively associated with sex hormone binding globulin (SHBG) (r = -0.41), luteinizing hormone (LH) (r - -0.4O) and follicle stimulating hormone (FSH) (r - -0.38) and was also significantly positively associated with both total and free oestradiol (r - 0.40 and 0.45, respectively). Waist-to-hip circumference ratio was significantly negatively correlated with SHBG (r - -0.53), LH (r - -0.35), and FSH (r - -0.35). After adjustment for BMI and other related factors, waist-to-hip circumference ratio was significantly and negatively associated with SHBG, LH, and FSH, and demonstrated a significant curvilinear relationship with free testosterone. These results suggest that in postmenopausal women, abdominal adiposity is associated with a relatively more androgenic sex hormone profile.
metrial cancer.1413 Differential deposition of adipose tissue in the abdomen versus the hips may be associated with postmenopausal levels of sex hormones. To date, research on body fat distribution and concentrations of sex hormones has been limited to a few studies in premenopausal women. In premenopausal women, abdominal adiposity has been positively associated with free testosterone and negatively associated with SHBG.1617 Previous research has not demonstrated an association in premenopausal women between abdominal adiposity and concentrations of other sex hormones.I61S Examining these relationships in postmenopausal women is important for two reasons. In postmenopausal women, unlike premenopausal women, plasma oestrogen is not of ovarian origin" and is therefore not subject to cyclic changes in hormone concentration. Postmenopausal plasma oestrogen is primarily from peripheral aromatization of plasma androstenedione to oestradiol in adipose tissue.20-21 Additionally, menopausal status may be related to differences in adipose tissue morphology and metabolism by anatomical site,22-23 which may also influence hormone levels.
Adiposity in postmenopausal women has been consistently associated with increased occurrence of carcinoma of the breast1"3 and endometrium.*"6 Adiposity may increase risk through its association with sex hormones. In postmenopausal women, obesity is positively correlated with concentrations of oestrone and oestradiol7-8 and conversion of androgens to oestrogens in adipose tissue.910 Obesity is also negatively correlated with sex hormone binding globulin (SHBG),"' 2 which binds a large portion of circulating oestrogens and androgens and is a marker of androgenicity.13 There is evidence to suggest that beyond therisksof general adiposity, greater abdominal adiposity, measured by the waist-to-hip circumference ratio, is associated with increased risk of breast and endo* Division of Epidemiology, School of Public Health, University of Minnesota, 1-210 Moos Tower, 515 Delaware St, SE, Minneapolis, MN 55455, USA. '•Department of Epidemiology and Public Health, School of Medicine, University of Miami, P.O. Box 016069,1550 NW 10th Ave, Miami, FL 33101, USA. Reprint requests to: Aaron R Folsom.
151
Downloaded from http://ije.oxfordjournals.org/ at University of California, San Francisco on January 23, 2015
SUSAN A KAYE\ AARON R FOLSOM\ JOHN T SOLER*, RONALD J PRINEAS" AND JOHN D POTTER*
152
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
The purpose of this study was to examine in postmenopausal women the relationship of adipose tissue distribution to blood concentrations of endogenous sex hormones and SHBG. We hypothesized that adipose tissue distribution would be significantly associated with sex hormones and SHBG after accounting for overall body mass.
Downloaded from http://ije.oxfordjournals.org/ at University of California, San Francisco on January 23, 2015
MATERIALS AND METHODS Serum samples and anthropometries were obtained from 88 postmenopausal Caucasian women, recruited from a Minneapolis, Minnesota suburb. Recruitment methods and exclusion criteria are described in detail elsewhere.24 Briefly, participants were volunteers between the ages of 55 and 69, at least one year postmenopausal, who had intact adrenal glands and at least one ovary. In addition, participants had no previous histories of cancer, diabetes mellitus, endocrine disorders, ovarian dysfunction, or liver disease, and were not taking medication known to affect hormone levels. Women were invited to participate in the study based on the distribution of self-measured waist-to-hip circumference ratio (WHR) so that a wide range of body fat distributions would be included in the investigation. Two fasting blood samples were drawn 30 minutes apart, and a third was drawn late afternoon, before the evening meal. Samples were drawn into vacuum blood tubes with minimal stasis and were allowed to clot one hour at room temperature. Serum was separated and frozen at —70°C until laboratory analysis. Each serum sample was divided and measured in duplicate by radioimmunoassay. All assays were performed within the same run. Luteinizing hormone (LH), follicle stimulating hormone (FSH), androstenedione (A), total testosterone (T), total oestradiol (E2), total oestrone (El), prolactin, and insulin were measured in each of the three samples. SHBG, free E2, and free T were measured in only the first fasting sample. Nonesterified fatty acids (NEFAs) were measured in both the first fasting sample and in the late afternoon sample. Assays were performed using commercial kits: LH and FSH (Bio-Mega Diagnostic Inc, Montreal, Canada); A and El (Wien Laboratories Inc, Succasunna, NJ, El antibody lot 03224); prolactin, E2, free and total T (Diagnostic Products Corporation, Los Angeles, CA, E2 antibody lot 023,024); SHBG (Techland, Liege, Belgium); NEFAs (Wako, Dallas, TX); insulin (Bio-RIA, Montreal, Canada). Free E2 assays were performed using the procedure outlined by MacMahon et al.a The sensitivity and specificity of all the hormone assays were carefully considered prior to use. The T assay can accurately measure T levels as low as 0.064
ng/dl. The E2 assay has a sensitivity far below 20 pg/ml and the El assay was modified by our laboratory for measurement of very low El concentrations (10 pg/ ml). For each of the assays, cross-reactivity with other steroids was considered negligible due to antibody specificity and physiological concentrations of potentially interfering steroids. The rank ordering of the morning and afternoon specimens were statistically similar, so mean hormone levels were calculated by averaging the morning hormone values with the afternoon hormone sample. With the exception of SHBG, the distributions of all hormones were normalized by natural logarithm transformation. Intra-assay coefficients of variation (calculated on the logarithm transformed values) were: prolactin: 13.7%, LH: 2.2%, FSH: 1.4%, total T: 19.1%, freeT: 24.1%, SHBG: 7.1%, total E2:16.3%, free E2: 2.5%, El: 9.5%, A: 8.3%, NEFAs: 9.0%, insulin: 8.9%. These precision results were similar to the precision reported by the manufacturers. Trained technicians measured weight, height, and the circumferences of the waist and hips. Weight (in underwear) was measured to the nearest pound and height (without shoes) was measured to the nearest quarter inch. Body mass index (BMI) was calculated as weight (kg)/height(m2). Circumferences were measured in duplicate to the nearest quarter inch. Waist circumference was measured one inch above the umbilicus and hip circumference was measured at the maximum girth between the waist and thighs. Waistto-hip circumference ratio (WHR) was calculated using the average of the duplicate measurements, dividing mean waist circumference by mean hip circumference. Participants also completed questions on cigarette smoking, physical activity, ethanol consumption, general medical history, and demographics. Univariate analyses between anthropometric variables and hormones included computation of arithmetic means and simple Pearson correlation coefficients. Analysis of variance was used to compare hormone levels by tertiles of WHR, using tertile cut points of 0.76 and 0.84. A number of possible covariates were examined using Student's t-test and analysis of variance: past exogenous oestrogen and oral contraceptive use, pregnancy history, alcohol consumption, cigarette smoking, physical activity, age at menopause, and family histories of breast and gynaecological cancers. Stepwise multivariate linear regressions of sex hormones on WHR were performed forcing BMI and age into each regression model. Regression terms tested in these models included interactions between WHR and BMI, quadratic terms for curvilinearity of BMI and WHR, and other covariates. Final models included
153
FAT DISTRIBUTION AND SEX HORMONES
these higher order terms only when p
International Journal of Epidemiology ©International Epidemiological Association 1991
Associations of Body Mass and Fat Distribution with Sex Hormone Concentrations in Postmenopausal Women Kaye S A (Division of Epidemiology, School of Public Health, University of Minnesota, 1-210 Moos Tower, 515 Delaware St, SE, Minneapolis, MN 55455, USA), Folsom A R, Soler J T, Prineas R J and Potter J D. Associations of body mass and fat distribution with sex hormone concentrations in postmenopausal women. International Journal of Epidemiology, 1991,20: 151-156. The associations of body mass and body fat distribution, as measured by waist-to-hip circumference ratio, with serum concentrations of sex hormones and sex hormone binding globulin were examined in 88 postmenopausal women. Body mass index (BMI) was significantly and negatively associated with sex hormone binding globulin (SHBG) (r = -0.41), luteinizing hormone (LH) (r - -0.4O) and follicle stimulating hormone (FSH) (r - -0.38) and was also significantly positively associated with both total and free oestradiol (r - 0.40 and 0.45, respectively). Waist-to-hip circumference ratio was significantly negatively correlated with SHBG (r - -0.53), LH (r - -0.35), and FSH (r - -0.35). After adjustment for BMI and other related factors, waist-to-hip circumference ratio was significantly and negatively associated with SHBG, LH, and FSH, and demonstrated a significant curvilinear relationship with free testosterone. These results suggest that in postmenopausal women, abdominal adiposity is associated with a relatively more androgenic sex hormone profile.
metrial cancer.1413 Differential deposition of adipose tissue in the abdomen versus the hips may be associated with postmenopausal levels of sex hormones. To date, research on body fat distribution and concentrations of sex hormones has been limited to a few studies in premenopausal women. In premenopausal women, abdominal adiposity has been positively associated with free testosterone and negatively associated with SHBG.1617 Previous research has not demonstrated an association in premenopausal women between abdominal adiposity and concentrations of other sex hormones.I61S Examining these relationships in postmenopausal women is important for two reasons. In postmenopausal women, unlike premenopausal women, plasma oestrogen is not of ovarian origin" and is therefore not subject to cyclic changes in hormone concentration. Postmenopausal plasma oestrogen is primarily from peripheral aromatization of plasma androstenedione to oestradiol in adipose tissue.20-21 Additionally, menopausal status may be related to differences in adipose tissue morphology and metabolism by anatomical site,22-23 which may also influence hormone levels.
Adiposity in postmenopausal women has been consistently associated with increased occurrence of carcinoma of the breast1"3 and endometrium.*"6 Adiposity may increase risk through its association with sex hormones. In postmenopausal women, obesity is positively correlated with concentrations of oestrone and oestradiol7-8 and conversion of androgens to oestrogens in adipose tissue.910 Obesity is also negatively correlated with sex hormone binding globulin (SHBG),"' 2 which binds a large portion of circulating oestrogens and androgens and is a marker of androgenicity.13 There is evidence to suggest that beyond therisksof general adiposity, greater abdominal adiposity, measured by the waist-to-hip circumference ratio, is associated with increased risk of breast and endo* Division of Epidemiology, School of Public Health, University of Minnesota, 1-210 Moos Tower, 515 Delaware St, SE, Minneapolis, MN 55455, USA. '•Department of Epidemiology and Public Health, School of Medicine, University of Miami, P.O. Box 016069,1550 NW 10th Ave, Miami, FL 33101, USA. Reprint requests to: Aaron R Folsom.
151
Downloaded from http://ije.oxfordjournals.org/ at University of California, San Francisco on January 23, 2015
SUSAN A KAYE\ AARON R FOLSOM\ JOHN T SOLER*, RONALD J PRINEAS" AND JOHN D POTTER*
152
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
The purpose of this study was to examine in postmenopausal women the relationship of adipose tissue distribution to blood concentrations of endogenous sex hormones and SHBG. We hypothesized that adipose tissue distribution would be significantly associated with sex hormones and SHBG after accounting for overall body mass.
Downloaded from http://ije.oxfordjournals.org/ at University of California, San Francisco on January 23, 2015
MATERIALS AND METHODS Serum samples and anthropometries were obtained from 88 postmenopausal Caucasian women, recruited from a Minneapolis, Minnesota suburb. Recruitment methods and exclusion criteria are described in detail elsewhere.24 Briefly, participants were volunteers between the ages of 55 and 69, at least one year postmenopausal, who had intact adrenal glands and at least one ovary. In addition, participants had no previous histories of cancer, diabetes mellitus, endocrine disorders, ovarian dysfunction, or liver disease, and were not taking medication known to affect hormone levels. Women were invited to participate in the study based on the distribution of self-measured waist-to-hip circumference ratio (WHR) so that a wide range of body fat distributions would be included in the investigation. Two fasting blood samples were drawn 30 minutes apart, and a third was drawn late afternoon, before the evening meal. Samples were drawn into vacuum blood tubes with minimal stasis and were allowed to clot one hour at room temperature. Serum was separated and frozen at —70°C until laboratory analysis. Each serum sample was divided and measured in duplicate by radioimmunoassay. All assays were performed within the same run. Luteinizing hormone (LH), follicle stimulating hormone (FSH), androstenedione (A), total testosterone (T), total oestradiol (E2), total oestrone (El), prolactin, and insulin were measured in each of the three samples. SHBG, free E2, and free T were measured in only the first fasting sample. Nonesterified fatty acids (NEFAs) were measured in both the first fasting sample and in the late afternoon sample. Assays were performed using commercial kits: LH and FSH (Bio-Mega Diagnostic Inc, Montreal, Canada); A and El (Wien Laboratories Inc, Succasunna, NJ, El antibody lot 03224); prolactin, E2, free and total T (Diagnostic Products Corporation, Los Angeles, CA, E2 antibody lot 023,024); SHBG (Techland, Liege, Belgium); NEFAs (Wako, Dallas, TX); insulin (Bio-RIA, Montreal, Canada). Free E2 assays were performed using the procedure outlined by MacMahon et al.a The sensitivity and specificity of all the hormone assays were carefully considered prior to use. The T assay can accurately measure T levels as low as 0.064
ng/dl. The E2 assay has a sensitivity far below 20 pg/ml and the El assay was modified by our laboratory for measurement of very low El concentrations (10 pg/ ml). For each of the assays, cross-reactivity with other steroids was considered negligible due to antibody specificity and physiological concentrations of potentially interfering steroids. The rank ordering of the morning and afternoon specimens were statistically similar, so mean hormone levels were calculated by averaging the morning hormone values with the afternoon hormone sample. With the exception of SHBG, the distributions of all hormones were normalized by natural logarithm transformation. Intra-assay coefficients of variation (calculated on the logarithm transformed values) were: prolactin: 13.7%, LH: 2.2%, FSH: 1.4%, total T: 19.1%, freeT: 24.1%, SHBG: 7.1%, total E2:16.3%, free E2: 2.5%, El: 9.5%, A: 8.3%, NEFAs: 9.0%, insulin: 8.9%. These precision results were similar to the precision reported by the manufacturers. Trained technicians measured weight, height, and the circumferences of the waist and hips. Weight (in underwear) was measured to the nearest pound and height (without shoes) was measured to the nearest quarter inch. Body mass index (BMI) was calculated as weight (kg)/height(m2). Circumferences were measured in duplicate to the nearest quarter inch. Waist circumference was measured one inch above the umbilicus and hip circumference was measured at the maximum girth between the waist and thighs. Waistto-hip circumference ratio (WHR) was calculated using the average of the duplicate measurements, dividing mean waist circumference by mean hip circumference. Participants also completed questions on cigarette smoking, physical activity, ethanol consumption, general medical history, and demographics. Univariate analyses between anthropometric variables and hormones included computation of arithmetic means and simple Pearson correlation coefficients. Analysis of variance was used to compare hormone levels by tertiles of WHR, using tertile cut points of 0.76 and 0.84. A number of possible covariates were examined using Student's t-test and analysis of variance: past exogenous oestrogen and oral contraceptive use, pregnancy history, alcohol consumption, cigarette smoking, physical activity, age at menopause, and family histories of breast and gynaecological cancers. Stepwise multivariate linear regressions of sex hormones on WHR were performed forcing BMI and age into each regression model. Regression terms tested in these models included interactions between WHR and BMI, quadratic terms for curvilinearity of BMI and WHR, and other covariates. Final models included
153
FAT DISTRIBUTION AND SEX HORMONES
these higher order terms only when p
