Sex Hormone–Binding Globulin and the Free Androgen Index Are Related to Cardiovascular Risk Factors in Multiethnic Premenopausal and Perimen
http://www.100md.com
循环学杂志 2005年第3期
the Department of Epidemiology (K.S.-T., S.B.), Graduate School of Public Health, University of Pittsburgh, Pittsburgh, Pa
Department of Epidemiology (R.P.W.), Tulane University School of Public Health and Tropical Medicine, New Orleans, La
Department of Psychiatry (K.A.M.), University of Pittsburgh School of Medicine, Pittsburgh, Pa
Cardiology Division (C.C., R.C.P.), Massachusetts General Hospital, and Division of Preventive Medicine, Brigham and Women’s Hospital, Boston, Mass
Department of Population Health and Reproduction (B.L.L.), University of California—Davis, Davis, Calif
Feinberg School of Medicine (D.L.-J.), Northwestern University, Chicago, Ill
Department of Epidemiology (M.F.S.), School of Public Health, University of Michigan, Ann Arbor, Mich
Department of Obstetrics, Gynecology and Women’s Health (J.L.T.), New Jersey Medical School, Newark, NJ.
Abstract
Background— Recent clinical trials have shifted attention away from estrogens and toward androgens and sex hormone–binding globulin (SHBG) as potential mediators of increasing cardiovascular (CV) risk in women at midlife.
Methods and Results— The correlation between reproductive hormones and CV risk factors was evaluated in a multiethnic (white, black, Hispanic, Chinese, and Japanese) sample of 3297 premenopausal and perimenopausal women. Testosterone and estradiol (E2) were evaluated along with SHBG and the free androgen index (FAI), the amount of testosterone not bound by SHBG. Low SHBG and high FAI were strongly and consistently related to elevated CV risk factors (higher insulin, glucose, and hemostatic and inflammatory markers and adverse lipids) even after controlling for body mass index (P<0.001 for all). Low levels of E2 were associated with elevated CV risk factors to a lesser degree. These observations were consistent across the 5 ethnic groups. Compared with whites, blacks had higher levels of SHBG and lower levels of FAI, and Chinese had lower levels of SHBG and higher levels of FAI.
Conclusions— Low SHBG and high FAI are strongly associated with CV risk factors in racially diverse women, and thus, androgens likely play a role in the CV risk profile of perimenopausal women.
Key Words: hormones ; aging ; sex ; menopause ; risk factors
Introduction
Women are known to suffer coronary heart disease 10 to 20 years later than men,1 leading to the longstanding hypothesis that endogenous ovarian hormones offer a protective effect on the development of coronary heart disease. Conversely, recent clinical trials have shown that hormone replacement therapy does not reduce the risk of coronary disease in postmenopausal women,2,3 suggesting that declining estrogen levels are not the key determinant of accelerated cardiovascular disease (CVD) risk associated with the menopausal transition.
Androgens and sex hormone–binding globulin (SHBG) have been linked to adverse CV risk factors in both premenopausal and postmenopausal women,4,5 with increased testosterone4 and decreased SHBG4,5 strongly associated with central adiposity, increased triglycerides, and decreased HDL cholesterol levels. Low levels of SHBG have also consistently been linked to higher rates of diabetes.6,7 We have recently shown that the association between hemostatic factors and insulin resistance is strongest in the presence of low SHBG.8 Recently, SHBG and the free androgen index (FAI) have also been linked with CVD events, primarily through an increase in CV risk factors.9 Thus, androgens and SHBG are reemerging as potential mediators of CV risk in women at midlife.9–11
Most of the research linking low SHBG and high androgens to CV risk has been restricted to white women and has focused on women who have reached postmenopause. The purpose of the present report is to look broadly at the association between reproductive hormones and CV risk factors in 3297 premenopausal and early perimenopausal women enrolled in the Study of Women’s Health Across the Nation (SWAN), an ethnically diverse cohort including white, black, Chinese-American, Japanese-American, and Hispanic women.
Methods
Participants
Data from the baseline visit of the SWAN were used for the present analyses. SWAN is a multicenter, multiethnic, longitudinal study designed to characterize the biological and psychosocial changes that occur during the menopausal transition in a community-based sample. Details of the study design and recruitment have been previously published.12 In brief, SWAN is being conducted at 7 sites: Boston, Mass; Chicago, Ill; the Detroit, Mich, area; Los Angeles, Calif; Newark, NJ; Pittsburgh, Pa; and Oakland, Calif. A total of 3302 women aged 42 to 52 years were enrolled from 1996 to 1997. Of these, 1550 are white, 935 black, 286 Hispanic, 250 Chinese, and 281 Japanese. The Hispanic group includes those of Central American, Mexican, and Caribbean origin.
At the time of enrollment, women had an intact uterus and at least one ovary and were not pregnant or breast-feeding. All participants were still menstruating, and women who used oral contraceptives or hormone replacement in the prior 3 months were excluded. Thus, all baseline data were for women exposed to only endogenous hormone levels. Women were classified as early perimenopausal if they reported a prior increase in cycle variability. The institutional review boards of the participating institutions approved this study, and all women signed informed consent forms before participation.
Blood Assays
A fasting blood sample was targeted to the follicular phase of the menstrual cycle (days 2 to 5). All samples were maintained at 4°C until separated and then were frozen at –80°C and shipped on dry ice to a central laboratory. Standard CV risk factors were assayed at Medical Research Laboratories (Lexington, Ky), which is certified by the National Heart, Lung, and Blood Institute, Centers for Disease Control and Prevention Part III program.13 Lipid and lipoprotein fractions were analyzed in EDTA-treated plasma. Total cholesterol and triglycerides were analyzed by enzymatic methods. HDL cholesterol was isolated after addition of heparin and 2 mol/L MnCl2. Serum insulin was measured by a radioimmunoassay (DPC Coat-a-count) procedure and monitored as part of the monthly quality assurance program by the Diabetes Diagnostic Laboratory at the University of Missouri. Glucose was measured with a hexokinase-coupled reaction (Boehringer Mannheim Diagnostics). The HOMA insulin resistance index was calculated from fasting insulin and glucose as (fasting insulinxfasting glucose)/22.5.14
Tissue-type plasminogen activator antigen (tPA) was measured in plasma with a double antibody in an ELISA (Imubind tPA ELISA, American Diagnostica). The assay uses human single-chain tPA as a standard calibrated against an international standard (NIBSAC). Plasma plasminogen activator inhibitor 1 (PAI-1) was measured by a sandwich procedure with a solid-phase monoclonal antibody and an enzyme-labeled goat second antiserum for detection (Imubind plasma PAI-1 ELISA, American Diagnostica). Fibrinogen and factor VII activity were measured in frozen, citrated plasma on an MLA Electra 1400C (Medical Laboratory Automation Inc) with a turbidometric detection system. C-reactive protein was quantified with an ultrasensitive rate-immunonephelometric method (hs-CRP on a BN 100, Dade-Behring).
Hormone assays were conducted at the University of Michigan SWAN Endocrine Laboratory with an ACS-180 automated analyzer (Bayer Diagnostics Corp). Serum follicle-stimulating hormone (FSH) concentrations were measured with a 2-site chemiluminescent immunoassay. SHBG was measured by a competitive chemiluminescent assay. Serum estradiol (E2) concentrations were measured with a modified, offline ACS-180 (E2-6) immunoassay. Testosterone concentrations were evaluated with the ACS-180 total testosterone assay modified to increase precision in the low ranges. The FAI was used estimate the amount of testosterone unbound by SHBG and thus, immediately biologically active. FAI was calculated as 100xT/(28.84xSHBG). This results in the amount of free testosterone in nanomoles per liter.
Physical Measures
Blood pressure was measured in the right arm with the participant seated after at least 5 minutes of rest. Three sequential blood pressure values were taken, and the final 2 were averaged. Height and weight were measured without shoes with participants wearing light clothing. Portable scales were calibrated weekly, and stationary clinic devices were calibrated monthly. Waist circumference was measured over undergarments or light clothing. Body mass index (BMI) was calculated as weight in kilograms divided by the square of height in meters.
Women were considered to have the metabolic syndrome if they met at least 3 of the following 5 criteria: triglycerides 150 mg/dL; HDL <50 mg/dL; blood pressure 130 mm Hg systolic, 85 mm Hg diastolic, or on hypertensive medication; fasting glucose 110 mg/dL and/or diabetes; and waist circumference 80 cm for Chinese and Japanese women and 88 cm for white, Hispanic, and black women.
Statistical Methods
All statistical procedures were performed with the maximum number of available observations for each given CV risk factor. Smoking status was defined as never, former, and current smoking. Alcohol intake was categorized into a 3-level variable, consisting of nondrinkers, drinkers below the median kilocalories of alcohol consumed per day, and drinkers at or above the median kilocalories of alcohol consumed per day. Cycle day of blood draw was dichotomized as within days 2 through 5 of the menstrual cycle or outside days 2 through 5 of the menstrual cycle. This second category included observations for which cycle day was unknown.
Numerous variables were not normally distributed, and thus, nonparametric tests were used wherever possible. Median values and interquartile ranges were calculated to describe the study sample. To assess associations between each hormone and covariate, partial Spearman correlation coefficients were calculated, after adjusting for age, ethnicity, study center, cycle day of blood draw, smoking status, alcohol intake, and BMI. Linear regression was used to evaluate race-by-hormone interaction terms. In these analyses, the natural logarithm of PAI-1, HDL, triglycerides, FSH, E2, and FAI and the square root of testosterone and SHBG was used. Because of high collinearity among the CV risk factors, it was possible that any observed correlations between SHBG and FAI and multiple CV risk factors was being driven by just one or two risk factors. This possibility was explored by using stepwise multivariable regression, with forced inclusion of the covariates just listed, and transformed hormone variables as the response. Additional variables considered in the multivariable model statement were HDL, triglycerides, PAI-1, waist circumference, glucose, insulin, and HOMA insulin resistance index. These variables had the strongest partial correlation coefficients within groupings of similar variables (lipids/lipoproteins, inflammatory/hemostatic, glucose metabolism). Model fit was assessed with model residual plots.
Because of the high number of comparisons in this analysis, we considered a probability value of 0.01 to be statistically significant. For ethnicity-specific analyses, probability values <0.05 are presented because of the smaller sample size, but these should be considered with caution. All analyses were accomplished with SAS version 8.0 for Windows.15
Results
The 3302 women enrolled in SWAN included 935 blacks, 281 Japanese, 250 Chinese, 286 Hispanics, and 1550 whites. The number of women with nonmissing data ranged from 3080 to 3297, depending on the CV risk factor being assessed (Table 1). The mean age of the women was 46.2 years, and 47% had reached early perimenopause, whereas the remaining 53% were classified as premenopausal. The CV risk factor profile for the full group and by ethnicity is presented in Table 1 along with the baseline values of FSH, testosterone, E2, SHBG, and FAI.
We have previously reported that the CV risk factors varied substantially by race, with black and Hispanic women having the highest total CVD risk.16 We have also previously reported that E2 levels did not vary by ethnicity at baseline in SWAN but that FSH levels were higher and testosterone levels lower in both black and Hispanic women compared with whites.17
The most striking associations with CV risk factors occurred for SHBG and the FAI. Both a low SHBG and a high FAI were strongly and consistently related to adverse CV risk factors. Even after adjusting for BMI, low SHBG and high FAI were associated with elevated risk for every CV risk factor (Table 2). SHBG was most strongly associated with PAI-1, triglycerides, and HDL, and FAI was most strongly associated with PAI-1, tPA, triglycerides, and glucose.
In addition to the risk factors presented in Table 2, we evaluated the association of hormone factors with the presence or absence of diabetes. There were 194 women with diabetes. These women had higher testosterone, lower SHBG, and higher FAI values than women without diabetes (P<0.001 for all). There was no difference in FSH or E2 values between the women with and without diabetes.
To determine whether the hormone–CV risk factor associations observed for the full group were consistent across ethnic groups, analyses were stratified by ethnicity (Table 3). In general, FSH values were not significantly related to CV risk factors within ethnic groups. The only exception to this was a negative correlation between FSH and BMI for whites and blacks only. For the other hormone values, ethnicity-specific results were similar to the results for the full group. Significant race-by-hormone interactions were found for each of the risk factors listed in Table 3, and these interactions primarily involved SHBG and FAI. In general, hormone–risk factor associations were in the same direction for all ethnicities, but the strength of the association varied. For example, the association of both low SHBG and high FAI with BMI was significant for all ethnicities but was particularly strong among the Japanese and white women. Likewise, the association between low SHBG and high PAI-1 was substantial for all ethnicities but was particularly strong for the Chinese and Japanese women. Chinese and Japanese woman also had particularly strong associations between SHBG and HDL. Finally, among Chinese women, the association of both low SHBG and high FAI with waist circumference was particularly strong. The only risk factor for which the direction of the hormone association differed by ethnicity was triglycerides. Testosterone was positively associated with triglycerides for blacks and negatively associated for Hispanics.
Because obesity (BMI>30) was of particular concern as a confounding variable, we also stratified the analyses in Table 2 by obesity and tested for hormone-by-obesity interactions. The associations between hormone factors and CV risk factors were similar between obese and nonobese women. The only significant interaction was an FSH-by-obesity interaction for waist circumference (P<0.001). For the stratified analyses, the Spearman correlation between FSH and waist circumference was –0.03 among the nonobese women and –0.10 among the obese women.
Because many of the CV risk factors are highly correlated with one another, it is possible that many of the risk factor associations seen with SHBG and FAI are operating through one or two primary risk factors. To determine which CV risk factors contributed independent information to lower SHBG and higher FAI, the square root of SHBG and log(FAI) were separately modeled by linear regression (Table 4). Age, site, ethnicity, cycle day of blood draw, smoking, and alcohol intake were included in each model. Risk factors that were significantly and separately related to lower SHBG levels were greater BMI, higher triglycerides, lower HDL, higher PAI-1, and ethnicity (P<0.01 for all). Blacks had higher levels of SHBG compared with whites, whereas Chinese had significantly lower levels of SHBG compared with whites. These same risk factors were related to FAI, with the addition of tPA and glucose. Blacks and Hispanics had significantly lower levels of FAI compared with whites, and Chinese had significantly higher levels of FAI compared with whites. Interestingly, for both SHBG and FAI, neither HOMA nor the presence of diabetes added further predictive value over and above the variables in the final model. When the analysis was restricted to the 2701 women without a history of diabetes, the results were similar, except that HDL and glucose were no longer related to FAI. The same was true when women with a history of the metabolic syndrome were also excluded (Table 4).
Discussion
These data suggest that among premenopausal and early perimenopausal women, hormones, particularly low SHBG and high FAI, are related to elevated CV risk factors. This observation was consistent across the 5 ethnic groups included in our population sample. Although hormone–risk factor associations were consistently present across ethnic groups, these associations were significantly stronger among some ethnic groups, often the Chinese and Japanese. These findings were consistent whether or not women with diabetes and/or the metabolic syndrome were included in the analysis.
Previous data have shown SHBG to be positively associated with HDL18 and physical fitness19 and negatively associated with obesity,7,18,20 upward fat distribution,21 triglycerides,22 insulin resistance,21,23,24 and diabetes.6,7 The data presented here are consistent with these findings and extend them to a racially diverse population sample. In addition, we have shown that SHBG and FAI are correlated with hemostatic factors, particularly PAI-1 and tPA. We have previously reported in this population sample that both low SHBG levels and high PAI-1 levels are related to greater insulin resistance and that the association between fibrinolytic markers and insulin resistance is strongest when SHBG is also low.8
The present study provides the unique opportunity to compare SHBG and FAI values across race. When data were adjusted for BMI and other CV risk factors, blacks had significantly higher SHBG and lower FAI values than did whites, and Chinese had significantly lower SHBG and higher FAI values than did whites. The association of SHBG and FAI with risk factors was consistent across racial groups. However, there is evidence that these associations are particularly strong for certain ethnic groups, most often Japanese and Chinese women. Racial differences in BMI within the SWAN population sample have been presented and discussed previously.17 It should be noted that the majority of the Japanese and Chinese women were in the lower tertile of the BMI distribution, whereas the majority of Hispanic and black women were in the upper tertile of the BMI distribution. This may be part of the underlying reason for some of the race-by-hormone interactions observed. The lower prevalence of obesity among the Chinese and Japanese women may have allowed the association between hormones and other risk factors to be more clearly observed in these groups. Though speculative, another possible explanation may relate to the observation that Chinese and Japanese women maintain relatively high levels of adrenal androgens. We have previously reported higher levels of dehydroepiandrosterone sulfate (DHEAS) early in the menopausal transition among Chinese and Japanese women relative to the other ethnic groups in SWAN.25 Although DHEAS is a weak androgen, it can be converted peripherally to more bioactive compounds. It is possible, therefore, that Japanese and Chinese women have higher levels of total circulating bioactive androgens.
There are a number of potential mechanisms for the strong association between SHBG and CV risk factors. The positive correlation between HDL and SHBG is thought to be due to the influence of SHBG on the metabolism and/or production of HDL,5 although the precise mechanism is unknown. Insulin has a direct inhibitory effect on SHBG,26,27 and SHBG may thus be a marker of insulin resistance and/or the metabolic syndrome. We found that low SHBG was more highly associated with higher weight, triglycerides, and PAI-1 and low HDL than with elevated insulin or insulin resistance. This is consistent with the fact that SHBG has been found to predict the development of type 2 diabetes in women, over and above the effects of glucose and insulin concentrations.6 Likewise, PAI-1 has also been shown to predict type 2 diabetes after controlling for insulin resistance.28 Thus, the constellation of higher weight, triglycerides, and PAI-1 and lower HDL and SHBG may mark a process separate from but related to insulin resistance.
SHBG may be linked to CV risk factors through its association with androgens. Because SHBG binds preferentially to androgens,29 a decrease in SHBG has a greater impact on the amount of circulating free androgens than free estrogens, and SHBG has been viewed as an index of androgenicity.6 High androgen levels are associated with alterations in glucose metabolism, an adverse lipid profile and higher blood pressure,30,31 higher rates of subclinical atherosclerosis,31,32 higher levels of coronary atherosclerosis,30 and higher CV event rates.30 Testosterone specifically has been linked to coronary atherosclerosis on angiography and was found to be more predictive of atherosclerosis than total cholesterol in postmenopausal women.33 In the data presented here, higher levels of free testosterone were correlated with risk factors in a similar fashion to low levels of SHBG. Because the FAI is calculated with SHBG, it is difficult to determine which of these is the variable more closely linked to CV risk. A direct measure of free testosterone would be more useful in determining the relative effect of these 2 factors on CV risk.
Finally, it is possible that SHBG has metabolic actions of its own. SHBG has been identified to be part of a signal transduction pathway that mediates androgen and estrogen signaling at the cell membrane.34,35 Future research may reveal actions of SHBG that link it to CV risk factors directly. Future data from SWAN will include a prospective documentation of changes in hormones over time and how this relates to changes in risk factors. This will help determine the relative importance of these hormone factors with respect to changes in CV risk with the menopausal transition.
One limitation of the aforementioned studies as well as this one is the failure to take into account the total effect of the combined adrenal androgens, including DHEAS, androstenediol, and androstenedione. These hormones and their peripheral conversion products likely contribute to the total circulating bioactive androgen level and act with testosterone to further reduce SHBG levels. An FAI that incorporates these other values might prove to be most useful in linking reproductive hormones to CV risk.
There are a number of general limitations to this analysis. First is the cross-sectional nature of the data. Longitudinal CV risk factor data over a 5-year period will soon be available from SWAN. The prospective data will allow observations of how the hormone–risk factor associations change as women progress through the transition and reach their final menstrual period. The present data are limited to the period of time when these women were all still cycling. Because of this, it was necessary to choose a consistent time during the menstrual cycle when blood would be drawn. The interval from days 2 to 5 was chosen, but this may not be optimum for all hormones that were evaluated. Finally, categorization of premenopausal versus perimenopausal status is challenging. It is likely that not all women were correctly categorized with regard to menopausal stage.
In conclusion, low levels of SHBG and high levels of free testosterone are strongly associated with CV risk factors during the perimenopause, and this is consistent across racial groups. The evaluation of CV risk in women at midlife should shift away from solely focusing on declining endogenous estrogen levels and should include markers of androgens, such as SHBG, and measures of free androgen activity.
Acknowledgments
The SWAN was funded by the National Institute on Aging, the National Institute of Nursing Research, and the National Institutes of Health (NIH) Office of Research on Women’s Health.
Clinical Center: University of Michigan, Ann Arbor—Mary Fran Sowers, principal investigator (PI; U01 NR04061); Massachusetts General Hospital, Boston, Mass—Robert Neer, PI, 1995–1999; Joel Finkelstein, PI 1999–present (U01 AG012531); Rush University, Rush-Presbyterian–St. Luke’s Medical Center, Chicago, Ill—Lynda Powell, PI (U01 AG012505); University of California, Davis/Kaiser—Ellen Gold, PI (U01 AG012554); University of California, Los Angeles—Gail Greendale, PI (U01 AG012539); University of Medicine and Dentistry–New Jersey Medical School, Newark—Gerson Weiss, PI, 1995–2004; Nanette Santoro, PI, 2004–present (U01 AG012535); and the University of Pittsburgh, Pittsburgh, Pa—Karen Matthews, PI (U01 AG012546).
NIH Program Office: National Institute on Aging, Bethesda, Md—Sherry Sherman, 1994–present; Marcia Ory, 1994–2001; National Institute of Nursing Research, Bethesda, Md.
Central Laboratory: University of Michigan, Ann Arbor—Daniel McConnell, 2000–present (U01 AG012495, Central Ligand Assay Satellite Services).
Coordinating Center: New England Research Institutes, Watertown, Mass—Sonja McKinlay, PI (U01 AG012553), 1995–2001 and/or the University of Pittsburgh, Pittsburgh, Pa—Kim Sutton-Tyrrell, PI (U01 AG012546), 2001–present.
Steering Committee: Chris Gallagher, Chair 1995–1997; Susan Johnson, Chair 2002–present.
We thank the study staff at each site and all the women who participated in SWAN.
References
Kannel WB. The Framingham Study: historical insight on the impact of cardiovascular risk factors in men versus women. J Gend Spec Med. 2002; 5: 27–37.
Hulley S, Grady D, Bush T, Furberg C, Herrington D, Riggs B, Vittinghoff E. Randomized trial of estrogen plus progestin for secondary prevention of coronary heart disease in postmenopausal women; Heart and Estrogen/progestin Replacement Study (HERS) Research Group. JAMA. 1998; 280: 605–613.
Writing Group for the Women’s Health Initiative Investigators. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results from the Women’s Health Initiative randomized controlled trial. JAMA. 2002; 288: 321–333.
Haffner SM, Valdez RA. Endogenous sex hormones: impact on lipids, lipoproteins, and insulin. Am J Med. 1995; 98: 40S–47S.
Pugeat M, Moulin P, Cousin P, Fimbel S, Nicolas MH, Crave JC, Lejeune H. Interrelations between sex hormone-binding globulin (SHBG), plasma lipoproteins and cardiovascular risk. J Steroid Biochem Mol Biol. 1995; 53: 567–572.
Haffner SM, Valdez RA, Morales PA, Hazuda HP, Stern MP. Decreased sex hormone-binding globulin predicts noninsulin-dependent diabetes mellitus in women but not in men, J Clin Endocrinol Metab. 1993; 77: 56–60.
Goodman-Gruen D, Barrett-Connor E. Sex hormone–binding globulin and glucose tolerance in postmenopausal women: the Rancho Bernardo Study. Diabetes Care. 1997; 20: 645–649.
Sowers M, Derby C, Jannausch ML, Torrens JI, Pasternak R. Insulin resistance, hemostatic factors, and hormone interactions in pre- and perimenopausal women: SWAN. J Clin Endocrinol Metab. 2003; 88: 4904–4910.
Rexrode KM, Manson JE, Lee IM, Ridker PM, Sluss PM, Cook NR, Buring JE. Sex hormone levels and risk of cardiovascular events in postmenopausal women. Circulation. 2003; 108: 1688–1693.
Reinecke H, Bogdanski J, Woltering A, Breithardt G, Assmann G, Kerber S, von Eckardstein A. Relation of serum levels of sex hormone binding globulin to coronary heart disease in postmenopausal women. Am J Cardiol. 2002; 90: 364–368.
Golden SH, Maguire A, Ding J, Crouse JR, Cauley JA. Zacur H, Szklo M. Endogenous postmenopausal hormones and carotid atherosclerosis: a case-control study of the Atherosclerosis Risk in Communities cohort. Am J Epidemiol. 2002; 155: 437–445.
Sowers MF, Crawford S, Sternfeld B, Morganstein D, Gold EB, Greendale GA, Evans D, Neer R, Matthews R, Sherman S, Lo A, Weiss G, Kelsey J. Design, survey sampling and recruitment methods of SWAN: a multi-center, multi-ethnic, community-based cohort study of women and the menopausal transition. In: Lobos R, Marcus R, Kelsey JL, eds. Menopause. New York, NY: Academic Press; 2000: 175–188.
Myers GL, Cooper GR, Winn CL, Smith SJ. The Centers for Disease Control–National Heart, Lung, and Blood Institute Lipid standardization program: an approach to accurate and precise lipid measurements. Clin Lab Med. 1989; 9: 105–135.
Haffner SM. Progress in population analyses of the insulin resistance syndrome. Ann N Y Acad Sci. 1997; 8271–12.
SAS,version 8.02. Cary, NC: SAS, Inc; 1999.
Matthews K, Sowers MF, Derby C, Stein E, Miracle-McMahill H, Crawford S. Ethnic differences in total cardiovascular risk factor burden among middle-aged women: Study of Women’s Health Across the Nation. Am Heart J. In press.
Randolph JF Jr, Sowers M, Gold EB, Mohr BA, Luborsky J, Santoro N, McConnell DS, Finkelstein JS, Korenman SG, Matthews KA, Sternfeld B, Lasley BL. Reproductive hormones in the early menopausal transition: relationship to ethnicity, body size, and menopausal status. J Clin Endocrinol Metab. 2003; 88: 1516–1522.
Haffner SM, Moss SE, Klein BE, Klein R. Sex hormones and DHEA-SO4 in relation to ischemic heart disease mortality in diabetic subjects: the Wisconsin Epidemiologic Study of Diabetic Retinopathy. Diabetes Care. 1996; 19: 1045–1050.
Semmens JB, Rouse IL, Beilin LJ, Masarei JR. Relationships between age, body weight, physical fitness and sex-hormone–binding globulin capacity, Clin Chim Acta. 1983; 133: 295–300.
Soler JT, Folsom AR, Kaye SA, Prineas RJ. Associations of abdominal adiposity, fasting insulin, sex hormone binding globulin, and estrone with lipids and lipoproteins in post-menopausal women. Atherosclerosis. 1989; 79: 21–27.
Evans DJ, Hoffmann RG, Kalkhoff RK, Kissebah AH. Relationship of androgenic activity to body fat topography, fat cell morphology, and metabolic aberrations in premenopausal women, J Clin Endocrinol Metab. 1983; 57: 304–310.
Lapidus L, Lindstedt G, Lundberg PA, Bengtsson C, Gredmark T. Concentrations of sex-hormone binding globulin and corticosteroid binding globulin in serum in relation to cardiovascular risk factors and to 12-year incidence of cardiovascular disease and overall mortality in postmenopausal women. Clin Chem. 1986; 32 (pt 1): 146–152.
Haffner SM, Katz MS, Stern MP, Dunn JF. The relationship of sex hormones to hyperinsulinemia and hyperglycemia, Metab Clin Exp. 1988; 37: 683–688.
Haffner SM, Dunn JF, Katz MS. Relationship of sex hormone-binding globulin to lipid, lipoprotein, glucose, and insulin concentrations in postmenopausal women. Metab Clin Exp. 1992; 41: 278–284.
Lasley BL, Santoro N, Randolf JF, Gold EB, Crawford S, Weiss G, McConnell DS, Sowers MF. The relationship of circulating dehydroepiandrosterone, testosterone, and estradiol to stages of the menopausal transition and ethnicity. J Clin Endocrinol Metab. 2002; 87: 3760–3767.
Plymate SR, Matej LA, Jones RE, Friedl KE. Inhibition of sex hormone–binding globulin production in the human hepatoma (Hep G2) cell line by insulin and prolactin. J Clin Endocrinol Metab. 1988; 67: 460–464.
Loukovaara M, Carson M, Adlercreutz H. Regulation of production and secretion of sex hormone–binding globulin in HepG2 cell cultures by hormones and growth factors. J Clin Endocrinol Metab. 1995; 80: 160–164.
Festa A, D’Agostino R Jr, Tracy RP, Haffner SM, and Insulin Resistance Atherosclerosis Study Investigators. Elevated levels of acute-phase proteins and plasminogen activator inhibitor-1 predict the development of type 2 diabetes: the Insulin Resistance Atherosclerosis Study. Diabetes. 2002; 51: 1131–1137.
Rosenfield RL. Studies of the relation of plasma androgen levels to androgen action in women. J Steroid Biochem. 1975; 6: 695–702.
Birdsall MA, Farquhar CM, White HD. Association between polycystic ovaries and extent of coronary artery disease in women having cardiac catheterization. Ann Intern Med. 1997; 126: 32–35.
Talbott EO, Zborowski JV, Sutton-Tyrrell K, McHugh-Pemu KP, Guzick DS. Cardiovascular risk in women with polycystic ovary syndrome. Obstet Gynecol Clin North Am. 2001; 28: 111–133.
Talbott EO, Guzick DS, Sutton-Tyrrell K, McHugh-Pemu KP, Zborowski JV, Remsberg KE, Kuller LH. Evidence for association between polycystic ovary syndrome and premature carotid atherosclerosis in middle-aged women. Arterioscler Thromb Vasc Biol. 2000; 20: 2414–2421.
Phillips GB, Pinkernell BH, Jing TY. Relationship between serum sex hormones and coronary artery disease in postmenopausal women. Arterioscler Thromb Vasc Biol. 1997; 17: 695–701.
Rosner W, Hryb DJ, Khan MS, Nakhla AM, Romas NA. Androgen and estrogen signaling at the cell membrane via G-proteins and cyclic adenosine monophosphate. Steroids. 1999; 64 (1–2): 100–106.
Kahn SM, Hryb DJ, Nakhla AM, Romas NA, Rosner W. Sex hormone–binding globulin is synthesized in target cells. J Endocrinol. 2002; 175: 113–120.(Kim Sutton-Tyrrell, DrPH;)
Department of Epidemiology (R.P.W.), Tulane University School of Public Health and Tropical Medicine, New Orleans, La
Department of Psychiatry (K.A.M.), University of Pittsburgh School of Medicine, Pittsburgh, Pa
Cardiology Division (C.C., R.C.P.), Massachusetts General Hospital, and Division of Preventive Medicine, Brigham and Women’s Hospital, Boston, Mass
Department of Population Health and Reproduction (B.L.L.), University of California—Davis, Davis, Calif
Feinberg School of Medicine (D.L.-J.), Northwestern University, Chicago, Ill
Department of Epidemiology (M.F.S.), School of Public Health, University of Michigan, Ann Arbor, Mich
Department of Obstetrics, Gynecology and Women’s Health (J.L.T.), New Jersey Medical School, Newark, NJ.
Abstract
Background— Recent clinical trials have shifted attention away from estrogens and toward androgens and sex hormone–binding globulin (SHBG) as potential mediators of increasing cardiovascular (CV) risk in women at midlife.
Methods and Results— The correlation between reproductive hormones and CV risk factors was evaluated in a multiethnic (white, black, Hispanic, Chinese, and Japanese) sample of 3297 premenopausal and perimenopausal women. Testosterone and estradiol (E2) were evaluated along with SHBG and the free androgen index (FAI), the amount of testosterone not bound by SHBG. Low SHBG and high FAI were strongly and consistently related to elevated CV risk factors (higher insulin, glucose, and hemostatic and inflammatory markers and adverse lipids) even after controlling for body mass index (P<0.001 for all). Low levels of E2 were associated with elevated CV risk factors to a lesser degree. These observations were consistent across the 5 ethnic groups. Compared with whites, blacks had higher levels of SHBG and lower levels of FAI, and Chinese had lower levels of SHBG and higher levels of FAI.
Conclusions— Low SHBG and high FAI are strongly associated with CV risk factors in racially diverse women, and thus, androgens likely play a role in the CV risk profile of perimenopausal women.
Key Words: hormones ; aging ; sex ; menopause ; risk factors
Introduction
Women are known to suffer coronary heart disease 10 to 20 years later than men,1 leading to the longstanding hypothesis that endogenous ovarian hormones offer a protective effect on the development of coronary heart disease. Conversely, recent clinical trials have shown that hormone replacement therapy does not reduce the risk of coronary disease in postmenopausal women,2,3 suggesting that declining estrogen levels are not the key determinant of accelerated cardiovascular disease (CVD) risk associated with the menopausal transition.
Androgens and sex hormone–binding globulin (SHBG) have been linked to adverse CV risk factors in both premenopausal and postmenopausal women,4,5 with increased testosterone4 and decreased SHBG4,5 strongly associated with central adiposity, increased triglycerides, and decreased HDL cholesterol levels. Low levels of SHBG have also consistently been linked to higher rates of diabetes.6,7 We have recently shown that the association between hemostatic factors and insulin resistance is strongest in the presence of low SHBG.8 Recently, SHBG and the free androgen index (FAI) have also been linked with CVD events, primarily through an increase in CV risk factors.9 Thus, androgens and SHBG are reemerging as potential mediators of CV risk in women at midlife.9–11
Most of the research linking low SHBG and high androgens to CV risk has been restricted to white women and has focused on women who have reached postmenopause. The purpose of the present report is to look broadly at the association between reproductive hormones and CV risk factors in 3297 premenopausal and early perimenopausal women enrolled in the Study of Women’s Health Across the Nation (SWAN), an ethnically diverse cohort including white, black, Chinese-American, Japanese-American, and Hispanic women.
Methods
Participants
Data from the baseline visit of the SWAN were used for the present analyses. SWAN is a multicenter, multiethnic, longitudinal study designed to characterize the biological and psychosocial changes that occur during the menopausal transition in a community-based sample. Details of the study design and recruitment have been previously published.12 In brief, SWAN is being conducted at 7 sites: Boston, Mass; Chicago, Ill; the Detroit, Mich, area; Los Angeles, Calif; Newark, NJ; Pittsburgh, Pa; and Oakland, Calif. A total of 3302 women aged 42 to 52 years were enrolled from 1996 to 1997. Of these, 1550 are white, 935 black, 286 Hispanic, 250 Chinese, and 281 Japanese. The Hispanic group includes those of Central American, Mexican, and Caribbean origin.
At the time of enrollment, women had an intact uterus and at least one ovary and were not pregnant or breast-feeding. All participants were still menstruating, and women who used oral contraceptives or hormone replacement in the prior 3 months were excluded. Thus, all baseline data were for women exposed to only endogenous hormone levels. Women were classified as early perimenopausal if they reported a prior increase in cycle variability. The institutional review boards of the participating institutions approved this study, and all women signed informed consent forms before participation.
Blood Assays
A fasting blood sample was targeted to the follicular phase of the menstrual cycle (days 2 to 5). All samples were maintained at 4°C until separated and then were frozen at –80°C and shipped on dry ice to a central laboratory. Standard CV risk factors were assayed at Medical Research Laboratories (Lexington, Ky), which is certified by the National Heart, Lung, and Blood Institute, Centers for Disease Control and Prevention Part III program.13 Lipid and lipoprotein fractions were analyzed in EDTA-treated plasma. Total cholesterol and triglycerides were analyzed by enzymatic methods. HDL cholesterol was isolated after addition of heparin and 2 mol/L MnCl2. Serum insulin was measured by a radioimmunoassay (DPC Coat-a-count) procedure and monitored as part of the monthly quality assurance program by the Diabetes Diagnostic Laboratory at the University of Missouri. Glucose was measured with a hexokinase-coupled reaction (Boehringer Mannheim Diagnostics). The HOMA insulin resistance index was calculated from fasting insulin and glucose as (fasting insulinxfasting glucose)/22.5.14
Tissue-type plasminogen activator antigen (tPA) was measured in plasma with a double antibody in an ELISA (Imubind tPA ELISA, American Diagnostica). The assay uses human single-chain tPA as a standard calibrated against an international standard (NIBSAC). Plasma plasminogen activator inhibitor 1 (PAI-1) was measured by a sandwich procedure with a solid-phase monoclonal antibody and an enzyme-labeled goat second antiserum for detection (Imubind plasma PAI-1 ELISA, American Diagnostica). Fibrinogen and factor VII activity were measured in frozen, citrated plasma on an MLA Electra 1400C (Medical Laboratory Automation Inc) with a turbidometric detection system. C-reactive protein was quantified with an ultrasensitive rate-immunonephelometric method (hs-CRP on a BN 100, Dade-Behring).
Hormone assays were conducted at the University of Michigan SWAN Endocrine Laboratory with an ACS-180 automated analyzer (Bayer Diagnostics Corp). Serum follicle-stimulating hormone (FSH) concentrations were measured with a 2-site chemiluminescent immunoassay. SHBG was measured by a competitive chemiluminescent assay. Serum estradiol (E2) concentrations were measured with a modified, offline ACS-180 (E2-6) immunoassay. Testosterone concentrations were evaluated with the ACS-180 total testosterone assay modified to increase precision in the low ranges. The FAI was used estimate the amount of testosterone unbound by SHBG and thus, immediately biologically active. FAI was calculated as 100xT/(28.84xSHBG). This results in the amount of free testosterone in nanomoles per liter.
Physical Measures
Blood pressure was measured in the right arm with the participant seated after at least 5 minutes of rest. Three sequential blood pressure values were taken, and the final 2 were averaged. Height and weight were measured without shoes with participants wearing light clothing. Portable scales were calibrated weekly, and stationary clinic devices were calibrated monthly. Waist circumference was measured over undergarments or light clothing. Body mass index (BMI) was calculated as weight in kilograms divided by the square of height in meters.
Women were considered to have the metabolic syndrome if they met at least 3 of the following 5 criteria: triglycerides 150 mg/dL; HDL <50 mg/dL; blood pressure 130 mm Hg systolic, 85 mm Hg diastolic, or on hypertensive medication; fasting glucose 110 mg/dL and/or diabetes; and waist circumference 80 cm for Chinese and Japanese women and 88 cm for white, Hispanic, and black women.
Statistical Methods
All statistical procedures were performed with the maximum number of available observations for each given CV risk factor. Smoking status was defined as never, former, and current smoking. Alcohol intake was categorized into a 3-level variable, consisting of nondrinkers, drinkers below the median kilocalories of alcohol consumed per day, and drinkers at or above the median kilocalories of alcohol consumed per day. Cycle day of blood draw was dichotomized as within days 2 through 5 of the menstrual cycle or outside days 2 through 5 of the menstrual cycle. This second category included observations for which cycle day was unknown.
Numerous variables were not normally distributed, and thus, nonparametric tests were used wherever possible. Median values and interquartile ranges were calculated to describe the study sample. To assess associations between each hormone and covariate, partial Spearman correlation coefficients were calculated, after adjusting for age, ethnicity, study center, cycle day of blood draw, smoking status, alcohol intake, and BMI. Linear regression was used to evaluate race-by-hormone interaction terms. In these analyses, the natural logarithm of PAI-1, HDL, triglycerides, FSH, E2, and FAI and the square root of testosterone and SHBG was used. Because of high collinearity among the CV risk factors, it was possible that any observed correlations between SHBG and FAI and multiple CV risk factors was being driven by just one or two risk factors. This possibility was explored by using stepwise multivariable regression, with forced inclusion of the covariates just listed, and transformed hormone variables as the response. Additional variables considered in the multivariable model statement were HDL, triglycerides, PAI-1, waist circumference, glucose, insulin, and HOMA insulin resistance index. These variables had the strongest partial correlation coefficients within groupings of similar variables (lipids/lipoproteins, inflammatory/hemostatic, glucose metabolism). Model fit was assessed with model residual plots.
Because of the high number of comparisons in this analysis, we considered a probability value of 0.01 to be statistically significant. For ethnicity-specific analyses, probability values <0.05 are presented because of the smaller sample size, but these should be considered with caution. All analyses were accomplished with SAS version 8.0 for Windows.15
Results
The 3302 women enrolled in SWAN included 935 blacks, 281 Japanese, 250 Chinese, 286 Hispanics, and 1550 whites. The number of women with nonmissing data ranged from 3080 to 3297, depending on the CV risk factor being assessed (Table 1). The mean age of the women was 46.2 years, and 47% had reached early perimenopause, whereas the remaining 53% were classified as premenopausal. The CV risk factor profile for the full group and by ethnicity is presented in Table 1 along with the baseline values of FSH, testosterone, E2, SHBG, and FAI.
We have previously reported that the CV risk factors varied substantially by race, with black and Hispanic women having the highest total CVD risk.16 We have also previously reported that E2 levels did not vary by ethnicity at baseline in SWAN but that FSH levels were higher and testosterone levels lower in both black and Hispanic women compared with whites.17
The most striking associations with CV risk factors occurred for SHBG and the FAI. Both a low SHBG and a high FAI were strongly and consistently related to adverse CV risk factors. Even after adjusting for BMI, low SHBG and high FAI were associated with elevated risk for every CV risk factor (Table 2). SHBG was most strongly associated with PAI-1, triglycerides, and HDL, and FAI was most strongly associated with PAI-1, tPA, triglycerides, and glucose.
In addition to the risk factors presented in Table 2, we evaluated the association of hormone factors with the presence or absence of diabetes. There were 194 women with diabetes. These women had higher testosterone, lower SHBG, and higher FAI values than women without diabetes (P<0.001 for all). There was no difference in FSH or E2 values between the women with and without diabetes.
To determine whether the hormone–CV risk factor associations observed for the full group were consistent across ethnic groups, analyses were stratified by ethnicity (Table 3). In general, FSH values were not significantly related to CV risk factors within ethnic groups. The only exception to this was a negative correlation between FSH and BMI for whites and blacks only. For the other hormone values, ethnicity-specific results were similar to the results for the full group. Significant race-by-hormone interactions were found for each of the risk factors listed in Table 3, and these interactions primarily involved SHBG and FAI. In general, hormone–risk factor associations were in the same direction for all ethnicities, but the strength of the association varied. For example, the association of both low SHBG and high FAI with BMI was significant for all ethnicities but was particularly strong among the Japanese and white women. Likewise, the association between low SHBG and high PAI-1 was substantial for all ethnicities but was particularly strong for the Chinese and Japanese women. Chinese and Japanese woman also had particularly strong associations between SHBG and HDL. Finally, among Chinese women, the association of both low SHBG and high FAI with waist circumference was particularly strong. The only risk factor for which the direction of the hormone association differed by ethnicity was triglycerides. Testosterone was positively associated with triglycerides for blacks and negatively associated for Hispanics.
Because obesity (BMI>30) was of particular concern as a confounding variable, we also stratified the analyses in Table 2 by obesity and tested for hormone-by-obesity interactions. The associations between hormone factors and CV risk factors were similar between obese and nonobese women. The only significant interaction was an FSH-by-obesity interaction for waist circumference (P<0.001). For the stratified analyses, the Spearman correlation between FSH and waist circumference was –0.03 among the nonobese women and –0.10 among the obese women.
Because many of the CV risk factors are highly correlated with one another, it is possible that many of the risk factor associations seen with SHBG and FAI are operating through one or two primary risk factors. To determine which CV risk factors contributed independent information to lower SHBG and higher FAI, the square root of SHBG and log(FAI) were separately modeled by linear regression (Table 4). Age, site, ethnicity, cycle day of blood draw, smoking, and alcohol intake were included in each model. Risk factors that were significantly and separately related to lower SHBG levels were greater BMI, higher triglycerides, lower HDL, higher PAI-1, and ethnicity (P<0.01 for all). Blacks had higher levels of SHBG compared with whites, whereas Chinese had significantly lower levels of SHBG compared with whites. These same risk factors were related to FAI, with the addition of tPA and glucose. Blacks and Hispanics had significantly lower levels of FAI compared with whites, and Chinese had significantly higher levels of FAI compared with whites. Interestingly, for both SHBG and FAI, neither HOMA nor the presence of diabetes added further predictive value over and above the variables in the final model. When the analysis was restricted to the 2701 women without a history of diabetes, the results were similar, except that HDL and glucose were no longer related to FAI. The same was true when women with a history of the metabolic syndrome were also excluded (Table 4).
Discussion
These data suggest that among premenopausal and early perimenopausal women, hormones, particularly low SHBG and high FAI, are related to elevated CV risk factors. This observation was consistent across the 5 ethnic groups included in our population sample. Although hormone–risk factor associations were consistently present across ethnic groups, these associations were significantly stronger among some ethnic groups, often the Chinese and Japanese. These findings were consistent whether or not women with diabetes and/or the metabolic syndrome were included in the analysis.
Previous data have shown SHBG to be positively associated with HDL18 and physical fitness19 and negatively associated with obesity,7,18,20 upward fat distribution,21 triglycerides,22 insulin resistance,21,23,24 and diabetes.6,7 The data presented here are consistent with these findings and extend them to a racially diverse population sample. In addition, we have shown that SHBG and FAI are correlated with hemostatic factors, particularly PAI-1 and tPA. We have previously reported in this population sample that both low SHBG levels and high PAI-1 levels are related to greater insulin resistance and that the association between fibrinolytic markers and insulin resistance is strongest when SHBG is also low.8
The present study provides the unique opportunity to compare SHBG and FAI values across race. When data were adjusted for BMI and other CV risk factors, blacks had significantly higher SHBG and lower FAI values than did whites, and Chinese had significantly lower SHBG and higher FAI values than did whites. The association of SHBG and FAI with risk factors was consistent across racial groups. However, there is evidence that these associations are particularly strong for certain ethnic groups, most often Japanese and Chinese women. Racial differences in BMI within the SWAN population sample have been presented and discussed previously.17 It should be noted that the majority of the Japanese and Chinese women were in the lower tertile of the BMI distribution, whereas the majority of Hispanic and black women were in the upper tertile of the BMI distribution. This may be part of the underlying reason for some of the race-by-hormone interactions observed. The lower prevalence of obesity among the Chinese and Japanese women may have allowed the association between hormones and other risk factors to be more clearly observed in these groups. Though speculative, another possible explanation may relate to the observation that Chinese and Japanese women maintain relatively high levels of adrenal androgens. We have previously reported higher levels of dehydroepiandrosterone sulfate (DHEAS) early in the menopausal transition among Chinese and Japanese women relative to the other ethnic groups in SWAN.25 Although DHEAS is a weak androgen, it can be converted peripherally to more bioactive compounds. It is possible, therefore, that Japanese and Chinese women have higher levels of total circulating bioactive androgens.
There are a number of potential mechanisms for the strong association between SHBG and CV risk factors. The positive correlation between HDL and SHBG is thought to be due to the influence of SHBG on the metabolism and/or production of HDL,5 although the precise mechanism is unknown. Insulin has a direct inhibitory effect on SHBG,26,27 and SHBG may thus be a marker of insulin resistance and/or the metabolic syndrome. We found that low SHBG was more highly associated with higher weight, triglycerides, and PAI-1 and low HDL than with elevated insulin or insulin resistance. This is consistent with the fact that SHBG has been found to predict the development of type 2 diabetes in women, over and above the effects of glucose and insulin concentrations.6 Likewise, PAI-1 has also been shown to predict type 2 diabetes after controlling for insulin resistance.28 Thus, the constellation of higher weight, triglycerides, and PAI-1 and lower HDL and SHBG may mark a process separate from but related to insulin resistance.
SHBG may be linked to CV risk factors through its association with androgens. Because SHBG binds preferentially to androgens,29 a decrease in SHBG has a greater impact on the amount of circulating free androgens than free estrogens, and SHBG has been viewed as an index of androgenicity.6 High androgen levels are associated with alterations in glucose metabolism, an adverse lipid profile and higher blood pressure,30,31 higher rates of subclinical atherosclerosis,31,32 higher levels of coronary atherosclerosis,30 and higher CV event rates.30 Testosterone specifically has been linked to coronary atherosclerosis on angiography and was found to be more predictive of atherosclerosis than total cholesterol in postmenopausal women.33 In the data presented here, higher levels of free testosterone were correlated with risk factors in a similar fashion to low levels of SHBG. Because the FAI is calculated with SHBG, it is difficult to determine which of these is the variable more closely linked to CV risk. A direct measure of free testosterone would be more useful in determining the relative effect of these 2 factors on CV risk.
Finally, it is possible that SHBG has metabolic actions of its own. SHBG has been identified to be part of a signal transduction pathway that mediates androgen and estrogen signaling at the cell membrane.34,35 Future research may reveal actions of SHBG that link it to CV risk factors directly. Future data from SWAN will include a prospective documentation of changes in hormones over time and how this relates to changes in risk factors. This will help determine the relative importance of these hormone factors with respect to changes in CV risk with the menopausal transition.
One limitation of the aforementioned studies as well as this one is the failure to take into account the total effect of the combined adrenal androgens, including DHEAS, androstenediol, and androstenedione. These hormones and their peripheral conversion products likely contribute to the total circulating bioactive androgen level and act with testosterone to further reduce SHBG levels. An FAI that incorporates these other values might prove to be most useful in linking reproductive hormones to CV risk.
There are a number of general limitations to this analysis. First is the cross-sectional nature of the data. Longitudinal CV risk factor data over a 5-year period will soon be available from SWAN. The prospective data will allow observations of how the hormone–risk factor associations change as women progress through the transition and reach their final menstrual period. The present data are limited to the period of time when these women were all still cycling. Because of this, it was necessary to choose a consistent time during the menstrual cycle when blood would be drawn. The interval from days 2 to 5 was chosen, but this may not be optimum for all hormones that were evaluated. Finally, categorization of premenopausal versus perimenopausal status is challenging. It is likely that not all women were correctly categorized with regard to menopausal stage.
In conclusion, low levels of SHBG and high levels of free testosterone are strongly associated with CV risk factors during the perimenopause, and this is consistent across racial groups. The evaluation of CV risk in women at midlife should shift away from solely focusing on declining endogenous estrogen levels and should include markers of androgens, such as SHBG, and measures of free androgen activity.
Acknowledgments
The SWAN was funded by the National Institute on Aging, the National Institute of Nursing Research, and the National Institutes of Health (NIH) Office of Research on Women’s Health.
Clinical Center: University of Michigan, Ann Arbor—Mary Fran Sowers, principal investigator (PI; U01 NR04061); Massachusetts General Hospital, Boston, Mass—Robert Neer, PI, 1995–1999; Joel Finkelstein, PI 1999–present (U01 AG012531); Rush University, Rush-Presbyterian–St. Luke’s Medical Center, Chicago, Ill—Lynda Powell, PI (U01 AG012505); University of California, Davis/Kaiser—Ellen Gold, PI (U01 AG012554); University of California, Los Angeles—Gail Greendale, PI (U01 AG012539); University of Medicine and Dentistry–New Jersey Medical School, Newark—Gerson Weiss, PI, 1995–2004; Nanette Santoro, PI, 2004–present (U01 AG012535); and the University of Pittsburgh, Pittsburgh, Pa—Karen Matthews, PI (U01 AG012546).
NIH Program Office: National Institute on Aging, Bethesda, Md—Sherry Sherman, 1994–present; Marcia Ory, 1994–2001; National Institute of Nursing Research, Bethesda, Md.
Central Laboratory: University of Michigan, Ann Arbor—Daniel McConnell, 2000–present (U01 AG012495, Central Ligand Assay Satellite Services).
Coordinating Center: New England Research Institutes, Watertown, Mass—Sonja McKinlay, PI (U01 AG012553), 1995–2001 and/or the University of Pittsburgh, Pittsburgh, Pa—Kim Sutton-Tyrrell, PI (U01 AG012546), 2001–present.
Steering Committee: Chris Gallagher, Chair 1995–1997; Susan Johnson, Chair 2002–present.
We thank the study staff at each site and all the women who participated in SWAN.
References
Kannel WB. The Framingham Study: historical insight on the impact of cardiovascular risk factors in men versus women. J Gend Spec Med. 2002; 5: 27–37.
Hulley S, Grady D, Bush T, Furberg C, Herrington D, Riggs B, Vittinghoff E. Randomized trial of estrogen plus progestin for secondary prevention of coronary heart disease in postmenopausal women; Heart and Estrogen/progestin Replacement Study (HERS) Research Group. JAMA. 1998; 280: 605–613.
Writing Group for the Women’s Health Initiative Investigators. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results from the Women’s Health Initiative randomized controlled trial. JAMA. 2002; 288: 321–333.
Haffner SM, Valdez RA. Endogenous sex hormones: impact on lipids, lipoproteins, and insulin. Am J Med. 1995; 98: 40S–47S.
Pugeat M, Moulin P, Cousin P, Fimbel S, Nicolas MH, Crave JC, Lejeune H. Interrelations between sex hormone-binding globulin (SHBG), plasma lipoproteins and cardiovascular risk. J Steroid Biochem Mol Biol. 1995; 53: 567–572.
Haffner SM, Valdez RA, Morales PA, Hazuda HP, Stern MP. Decreased sex hormone-binding globulin predicts noninsulin-dependent diabetes mellitus in women but not in men, J Clin Endocrinol Metab. 1993; 77: 56–60.
Goodman-Gruen D, Barrett-Connor E. Sex hormone–binding globulin and glucose tolerance in postmenopausal women: the Rancho Bernardo Study. Diabetes Care. 1997; 20: 645–649.
Sowers M, Derby C, Jannausch ML, Torrens JI, Pasternak R. Insulin resistance, hemostatic factors, and hormone interactions in pre- and perimenopausal women: SWAN. J Clin Endocrinol Metab. 2003; 88: 4904–4910.
Rexrode KM, Manson JE, Lee IM, Ridker PM, Sluss PM, Cook NR, Buring JE. Sex hormone levels and risk of cardiovascular events in postmenopausal women. Circulation. 2003; 108: 1688–1693.
Reinecke H, Bogdanski J, Woltering A, Breithardt G, Assmann G, Kerber S, von Eckardstein A. Relation of serum levels of sex hormone binding globulin to coronary heart disease in postmenopausal women. Am J Cardiol. 2002; 90: 364–368.
Golden SH, Maguire A, Ding J, Crouse JR, Cauley JA. Zacur H, Szklo M. Endogenous postmenopausal hormones and carotid atherosclerosis: a case-control study of the Atherosclerosis Risk in Communities cohort. Am J Epidemiol. 2002; 155: 437–445.
Sowers MF, Crawford S, Sternfeld B, Morganstein D, Gold EB, Greendale GA, Evans D, Neer R, Matthews R, Sherman S, Lo A, Weiss G, Kelsey J. Design, survey sampling and recruitment methods of SWAN: a multi-center, multi-ethnic, community-based cohort study of women and the menopausal transition. In: Lobos R, Marcus R, Kelsey JL, eds. Menopause. New York, NY: Academic Press; 2000: 175–188.
Myers GL, Cooper GR, Winn CL, Smith SJ. The Centers for Disease Control–National Heart, Lung, and Blood Institute Lipid standardization program: an approach to accurate and precise lipid measurements. Clin Lab Med. 1989; 9: 105–135.
Haffner SM. Progress in population analyses of the insulin resistance syndrome. Ann N Y Acad Sci. 1997; 8271–12.
SAS,version 8.02. Cary, NC: SAS, Inc; 1999.
Matthews K, Sowers MF, Derby C, Stein E, Miracle-McMahill H, Crawford S. Ethnic differences in total cardiovascular risk factor burden among middle-aged women: Study of Women’s Health Across the Nation. Am Heart J. In press.
Randolph JF Jr, Sowers M, Gold EB, Mohr BA, Luborsky J, Santoro N, McConnell DS, Finkelstein JS, Korenman SG, Matthews KA, Sternfeld B, Lasley BL. Reproductive hormones in the early menopausal transition: relationship to ethnicity, body size, and menopausal status. J Clin Endocrinol Metab. 2003; 88: 1516–1522.
Haffner SM, Moss SE, Klein BE, Klein R. Sex hormones and DHEA-SO4 in relation to ischemic heart disease mortality in diabetic subjects: the Wisconsin Epidemiologic Study of Diabetic Retinopathy. Diabetes Care. 1996; 19: 1045–1050.
Semmens JB, Rouse IL, Beilin LJ, Masarei JR. Relationships between age, body weight, physical fitness and sex-hormone–binding globulin capacity, Clin Chim Acta. 1983; 133: 295–300.
Soler JT, Folsom AR, Kaye SA, Prineas RJ. Associations of abdominal adiposity, fasting insulin, sex hormone binding globulin, and estrone with lipids and lipoproteins in post-menopausal women. Atherosclerosis. 1989; 79: 21–27.
Evans DJ, Hoffmann RG, Kalkhoff RK, Kissebah AH. Relationship of androgenic activity to body fat topography, fat cell morphology, and metabolic aberrations in premenopausal women, J Clin Endocrinol Metab. 1983; 57: 304–310.
Lapidus L, Lindstedt G, Lundberg PA, Bengtsson C, Gredmark T. Concentrations of sex-hormone binding globulin and corticosteroid binding globulin in serum in relation to cardiovascular risk factors and to 12-year incidence of cardiovascular disease and overall mortality in postmenopausal women. Clin Chem. 1986; 32 (pt 1): 146–152.
Haffner SM, Katz MS, Stern MP, Dunn JF. The relationship of sex hormones to hyperinsulinemia and hyperglycemia, Metab Clin Exp. 1988; 37: 683–688.
Haffner SM, Dunn JF, Katz MS. Relationship of sex hormone-binding globulin to lipid, lipoprotein, glucose, and insulin concentrations in postmenopausal women. Metab Clin Exp. 1992; 41: 278–284.
Lasley BL, Santoro N, Randolf JF, Gold EB, Crawford S, Weiss G, McConnell DS, Sowers MF. The relationship of circulating dehydroepiandrosterone, testosterone, and estradiol to stages of the menopausal transition and ethnicity. J Clin Endocrinol Metab. 2002; 87: 3760–3767.
Plymate SR, Matej LA, Jones RE, Friedl KE. Inhibition of sex hormone–binding globulin production in the human hepatoma (Hep G2) cell line by insulin and prolactin. J Clin Endocrinol Metab. 1988; 67: 460–464.
Loukovaara M, Carson M, Adlercreutz H. Regulation of production and secretion of sex hormone–binding globulin in HepG2 cell cultures by hormones and growth factors. J Clin Endocrinol Metab. 1995; 80: 160–164.
Festa A, D’Agostino R Jr, Tracy RP, Haffner SM, and Insulin Resistance Atherosclerosis Study Investigators. Elevated levels of acute-phase proteins and plasminogen activator inhibitor-1 predict the development of type 2 diabetes: the Insulin Resistance Atherosclerosis Study. Diabetes. 2002; 51: 1131–1137.
Rosenfield RL. Studies of the relation of plasma androgen levels to androgen action in women. J Steroid Biochem. 1975; 6: 695–702.
Birdsall MA, Farquhar CM, White HD. Association between polycystic ovaries and extent of coronary artery disease in women having cardiac catheterization. Ann Intern Med. 1997; 126: 32–35.
Talbott EO, Zborowski JV, Sutton-Tyrrell K, McHugh-Pemu KP, Guzick DS. Cardiovascular risk in women with polycystic ovary syndrome. Obstet Gynecol Clin North Am. 2001; 28: 111–133.
Talbott EO, Guzick DS, Sutton-Tyrrell K, McHugh-Pemu KP, Zborowski JV, Remsberg KE, Kuller LH. Evidence for association between polycystic ovary syndrome and premature carotid atherosclerosis in middle-aged women. Arterioscler Thromb Vasc Biol. 2000; 20: 2414–2421.
Phillips GB, Pinkernell BH, Jing TY. Relationship between serum sex hormones and coronary artery disease in postmenopausal women. Arterioscler Thromb Vasc Biol. 1997; 17: 695–701.
Rosner W, Hryb DJ, Khan MS, Nakhla AM, Romas NA. Androgen and estrogen signaling at the cell membrane via G-proteins and cyclic adenosine monophosphate. Steroids. 1999; 64 (1–2): 100–106.
Kahn SM, Hryb DJ, Nakhla AM, Romas NA, Rosner W. Sex hormone–binding globulin is synthesized in target cells. J Endocrinol. 2002; 175: 113–120.(Kim Sutton-Tyrrell, DrPH;)