Levels of Sex Steroid and Cardiovascular Disease Measures in Premenopausal and Hormone-Treated Women at Midlife: Implications for the “Timing Hypothesis” | Cardiology | JAMA Internal Medicine | JAMA Network
[Skip to Navigation]
Table 1. 
Age, BMI, and Hormones Values According to Menopause Status or Hormone Therapy Group, SWAN 5-Year Follow-upa
Age, BMI, and Hormones Values According to Menopause Status or Hormone Therapy Group, SWAN 5-Year Follow-upa
Table 2. 
Cardiovascular Intermediate Markers According to Menopause Status or Hormone Therapy Groupa
Cardiovascular Intermediate Markers According to Menopause Status or Hormone Therapy Groupa
Table 3. 
Partial Correlations of Cardiovascular Measures With Serum Sex Hormone–Binding Globulina
Partial Correlations of Cardiovascular Measures With Serum Sex Hormone–Binding Globulina
Original Investigation
October 27, 2008

Levels of Sex Steroid and Cardiovascular Disease Measures in Premenopausal and Hormone-Treated Women at Midlife: Implications for the “Timing Hypothesis”

Author Affiliations

Author Affiliations: Department of Epidemiology, School of Public Health (Drs Sowers and McConnell and Ms Jannausch), and Division of Cardiology, Department of Internal Medicine (Dr Jackson), University of Michigan, and Department of Obstetrics and Gynecology, University of Michigan Health Sciences System (Dr Randolph), Ann Arbor; and Center for Health and the Environment, University of California, Davis (Dr Lasley).

Arch Intern Med. 2008;168(19):2146-2153. doi:10.1001/archinte.168.19.2146

Background  The “timing hypothesis,” in addressing findings from the Women's Health Initiative trial, suggests that hormone therapy (HT) should be initiated within 6 years of the menopause transition to extend a favorable estrogenic environment after menopause.

Methods  We compared sex steroid and cardiovascular profiles at the 5-year follow-up visit in a community-based, longitudinal study of the menopause transition (Study of Women's Health Across the Nation). Women aged 47 to 57 years were in 1 of 4 groups: premenopausal, women using conjugated equine estrogen with or without progestin, or postmenopausal (<5 years) without HT use. Cardiovascular assays included low-density lipoprotein cholesterol, oxidized low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglycerides, apolipoproteins A-I and B, F2a-isoprostanes, C-reactive protein, and lipoprotein (a)-1. Sex steroid assays were performed for estradiol, estrogen receptor ligand load, 2-hydroxyestrone, 16α-hydroxyestrone, total testosterone, and sex hormone–binding globulin.

Results  Users of HT had 50% higher levels of sex hormone–binding globulin (P < .001 for both HT groups), which limits binding of sex steroids to their receptors, and higher excreted estrone metabolites (more than 60%; P < .001 for both HT groups) than premenopausal or postmenopausal women. These findings were, in turn, associated with higher levels of F2a-isoprostanes, an oxidative stress measure, than in premenopausal women. The HT users had a more favorable ratio of high-density to low-density lipoprotein cholesterol than did premenopausal or postmenopausal women (P < .01), but higher triglyceride levels (P < .01).

Conclusion  Although HT users had some more favorable lipid profiles than premenopausal and postmenopausal women, there was evidence of adverse HT effects even in women free of atherosclerosis studied within the approximate 6-year period proposed with the timing hypothesis.

Studies to understand underlying cardioprotection in women relative to men at midlife1 have focused on the importance of estrogens and their relation to lipids, especially high-density lipoprotein cholesterol (HDL-C). Exogenous hormone therapy (HT) was the long-time paradigm thought to demonstrate that maintenance of estrogen levels after menopause contributed to heart health.2,3 However, when the Women's Health Initiative (WHI) and the Heart and Estrogen/Progestin Replacement Study demonstrated that widely prescribed exogenous hormone products were not cardioprotective,4,5 as had been originally hypothesized, alternative explanations were proposed.

One proposal, the “timing hypothesis,” is that a positive impact of HT on cardiovascular status is dependent on extending a favorable estrogenic environment after menopause without a substantial time discontinuity.6 A substantial interval between menopause and HT use could permit the development of atherogenic lesions and vascular compromise that would be less responsive to positive effects of HT use. In instances of HT initiation around menopause, it could also be hypothesized that HT use would confer sex steroid status that would help sustain favorable lipid or cardiovascular profiles.

To determine whether there were unappreciated changes in the endogenous sex steroid hormone environment among users of conjugated equine estrogen (with and without progestin) that might compromise lipids or other cardiovascular measures, we evaluated estradiol (E2) levels and novel estrogen measures in 4 groups: (1) premenopausal women; (2) users of conjugated equine estrogen; (3) users of conjugated equine estrogen with progestin; and (4) women postmenopausal for less than 5 years without HT use. Novel measures included the amount of estrogen acting as a ligand to the estrogen receptor (estrogen receptor ligand load [ERLL]) and an estimate of E2 bioavailability (free estrogen index). Selected estrogen metabolites have been hypothesized to contribute biological activity,7 so we assayed 2-hydroxyestrone (2-OHE1) and 16α-hydroxyestrone (16α-OHE1) on the premises that 2-OHE1s may act as antioxidants8 and that 16α-OHE1 activity may include covalent binding to the estrogen receptor.9 We characterized the potential antioxidant associations of estrogens by means of F2a-isoprostanes, a product of arachidonic acid oxidation and degradation.10,11 We further characterized the endogenous sex steroid environment with measures of total testosterone and sex steroid bioavailability with sex hormone–binding globulin (SHBG).

Sampling and study population

Data are from the Study of Women's Health Across the Nation (SWAN), a multicenter, multiethnic longitudinal study of the menopausal transition (see boxed copy on page 2151). Data and specimens were from the fifth annual follow-up examination, which took place in the year before the release of the WHI cardiovascular findings.4

At baseline, eligibility criteria for the SWAN study included age 42 to 52 years, presence of an intact uterus and at least 1 ovary, no use of exogenous hormones, menses in the 3 months before enrollment, and self-identification with a site's designated racial/ethnic group. Therefore, Caucasian and non-Caucasian samples were recruited including African-American women in Boston, Massachusetts; Chicago, Illinois; the Detroit, Michigan area; and Pittsburgh, Pennsylvania; and Japanese, Chinese, and Hispanic women in Los Angeles, California; Oakland, California; and Newark, New Jersey, respectively.12

A total of 2606 women participated in the 5-year follow-up visit (78.9% of the 3302 baseline participants). However, data from 2 of the 7 sites (Chicago and Newark) were excluded from these analyses because their site protocols did not include urine collection for assay of estrogen metabolites. Also excluded from data analyses were women with diabetes mellitus, heart disease, nonestrogen hormone use, or surgical menopause without HT use. Women in the early (n = 670) and late (n = 196) perimenopause stages were excluded to allow a more clear delineation of hormone and cardiovascular measures in premenopausal women, postmenopausal women (without HT use), and women using 2 types of conjugated equine estrogen therapy, with and without a progestin. Institutional review board approval for the study protocol and repository storage was obtained at each study site.

Menopause stage and ht use

Premenopause was defined as the presence of menses within the previous 3-month period, with no decrease in predictability. Early perimenopause was the presence of menses within the previous 3 months and less predictable menstrual frequency; late perimenopause was defined as having 3 to 11 months of amenorrhea.13-15 Twelve consecutive months of amenorrhea without alternative explanation indicated postmenopause.15 Medication and HT use were self-reported with corroborative visualization of the prescription container, when possible. Of 782 women available for this report, 98 (12.5%) were premenopausal without HT use, 53 (6.8%) used conjugated equine estrogen only, 243 (31.1%) used conjugated equine estrogen–progestin combinations, and 370 (47.3%) were postmenopausal without HT use. Eighteen women using transdermal formulations were excluded from analyses.

Specimens and assays

Annual morning blood and urine collections followed an overnight fast. Two attempts were made to obtain specimens in days 2 to 5 of a spontaneous menstrual cycle. If timed specimens could not be obtained, random fasting specimens were collected. Blood was refrigerated 1 to 2 hours after phlebotomy, centrifuged, aliquotted, frozen, and then sent on dry ice to laboratories for assay.

Specimens were analyzed for levels of hormones and SHBG at the University of Michigan (Clinical Laboratory Improvement Amendments certified) laboratory on an automated analyzer (ACS-180; Bayer Diagnostics, Tarrytown, New York). The E2 was assayed in duplicate by means of a rabbit anti–E2-6 immunoassay that was modified to reduce the lower limit of detection to 1.0 pg/mL (to convert to picomoles per liter, multiply by 3.671) and a coefficient of variation (CV) of 3% to 12%. The testosterone competitive immunoassay used testosterone labeled with dimethylacridinuimester, a polyclonal rabbit antitestosterone antibody, and a monoclonal mouse antirabbit antibody coupled to paramagnetic particles. The testosterone reporting range was 10 to 100 ng/dL (to convert to nanomoles per liter, multiply by 0.0347), and assay range was 2 to 478 ng/dL. The testosterone assay was standardized analytically and confirmed by gas chromatography–mass spectrometry. Interassay and intra-assay CVs were 13.8% and 6.6%, respectively. The competitive SHBG assay was developed on-site with rabbit anti-SHBG antibodies, with a lower limit of detection of 1.95 nmol/L (to convert to micrograms per milliliter, divide by 8.896). Total testosterone level was indexed to SHBG to calculate the free androgen index ([100 × testosterone in nanograms per deciliter]/[28.84 × SHBG in nanomoles per liter]). Total E2 was indexed to SHBG as the free E2 index ([100 × total E2]/[272.11 × SHBG]) to estimate nonbound E2 bioavailability; (10×E2)/testosterone was used to make the resulting ratio unit-free.

The 2-OHE1 and 16α-OHE1 were assayed by enzyme immunoassay (Estramet; Immuna Care Corp, Blue Bell, Pennsylvania) in triplicate.16 Interassay and intra-assay CVs were less than 10% for each analyte. Because urinary 2-OHE1 and 16α-OHE1 are found as 3-glucuronides or 3,3,16-glucuronides, these sugars were removed to achieve recognition sites for the monoclonal antibodies by means of a mixture of β-glucuronidase and arylsulfatase enzyme isolated from Helix pomatia. The assay range was 0.6 to 40.0 ng/mL.

At the University of California, Davis, archival serum specimens were analyzed for ERLL, a cell-based bioassay for measuring signal transduction activity of total circulating bioactive estrogens and detecting activation of estrogen-dependent gene expression.17,18 Procedures are available from the authors.

Specimens were assayed for cardiovascular measures at Medical Research Laboratory/Global Central Lab Services, which is certified by the National Heart, Lung, and Blood Institute, Centers for Disease Control Part III program. Lipid fractions were determined from plasma samples with EDTA. Total cholesterol and triglycerides concentrations were determined by enzymatic methods on an analyzer (Hitachi 747; Boehringer Mannheim Diagnostics, Indianapolis, Indiana). The HDL-C was isolated by means of heparin-2M manganese chloride.19,20 Low-density lipoprotein cholesterol (LDL-C) level was calculated by the Friedewald equation.21 Lipoprotein (a)-1 (Lp[a]-1) was quantified by competitive enzyme-linked immunosorbent assay.22 Plasma oxidized LDL-C concentrations were measured by sandwich enzyme-linked immunosorbent assay. Apolipoproteins (apo) A-I and B were measured by immunonephelometry (BN1A-100; Behring Diagnostics, Westwood, Massachusetts) calibrated with a World Health Organization traceable standard.23 C-reactive protein (CRP) was quantified by an ultrasensitive rate immunonephelometric method (high-sensitivity CRP on a BN100 nephelometer; Dade Behring, Marburg, Germany). The lower limit of detection of the CRP assay was 0.03 mg/L (to convert to nanomoles per liter, multiply by 9.524); the CV was 10% to 12% (0.05 mg/dL) and 5% to 7% (2.2 mg/dL). Plasma plasminogen activator inhibitor 1 (PAI-1) was measured with a sandwich procedure using a solid-phase monoclonal antibody and enzyme-labeled goat antiserum (IMUBIND plasma PAI-1 enzyme-linked immunosorbent assay; American Diagnostica, Greenwich, Connecticut). Monthly interassay CVs were 5% to 9% and 4% to 9% at mean concentrations of 7 and 22.5 ng/dL, respectively (to convert to picomoles per liter, multiply by 1.923).

Samples analyzed for F2a-isoprostanes were prepurified on affinity columns (Cayman Chemical, Ann Arbor, Michigan), washed, eluted with 95% ethanol, evaporated, dried, and then diluted 1:10 with 0.1M phosphate buffer, assayed by means of an F2a-isoprostane enzyme immunoassay (Cayman Chemical), and read at 405 nm. The standard curve range was 3.9 to 500 pg/mL. Postextraction intra-assay and interassay CVs were 14.4% (51.2 pg/mL, n = 83 pairs) and 17.5% (51.2 pg/mL, n = 85), respectively; the intra-assay CV was 5.8% (n = 1707 pairs).

Physical and interview-based measures

Heights and weights were measured to estimate body mass index (calculated as weight in kilograms divided by height in meters squared). Waist circumference in centimeters was measured about 3 cm above the umbilicus after a relaxed expiration.

On the basis of self-report of race/ethnicity, the analytical sample was 17.0% African American (n = 133), 61.0% Caucasian (n = 477), 12.1% Japanese (n = 94), and 10.1% Chinese (n = 94). Exogenous hormone users in the analytical sample were 17.2% African American (n = 51), 62.2% Caucasian (n = 184), 12.2% Chinese (n = 36), and 8.4% Japanese (n = 25). Type of hormone delivery system did not vary according to race/ethnicity. Smoking and physical activity data were from self-administered questionnaires incorporating American Thoracic Society questions24 and a modified Baecke instrument,25,26 respectively.

Statistical methods

On the basis of examination of distributions, log or square root transformations were applied to measures of lipids and hormones for use in statistical modeling and backtransformed (including variances with a Taylor series) for presentation. Primary analytical approaches included analysis of covariance, correlation analyses, and multivariable regression modeling, testing main effects and interactions (eg, the interaction between SHBG and group status or interactions between 5-year difference in cardiovascular disease [CVD] measures and group status). Analysis of covariance was used to evaluate whether there were significant mean differences in lipid or hormone levels according to menopause stage or HT group designation, after adjusting for site, age, body mass index, physical activity, race/ethnicity, and smoking status. Duration of HT use and difference in CVD measures between baseline and 5-year follow-up were evaluated and did not change the reported relationships. Pearson and partial correlations (rYX) were obtained before and after adjustment for covariates, respectively. Type III sums of squares were used to estimate the adjusted partial correlations. Data are shown with 95% confidence intervals (CIs) or P values (2-sided with α < .01). The statistical software used was SAS version 9.1 (SAS Institute Inc, Cary, North Carolina).


Mean ages of women in the premenopausal group, the conjugated equine estrogen group, the conjugated equine estrogen plus progestin group, and the postmenopausal group were 48.9, 52.1, 51.9, and 52.6 years, respectively; age was not different according to race/ethnicity. The numbers of women and duration of their use of conjugated equine estrogen alone were as follows: 5 women for 1 year, 11 for 2 years, 12 for 3 years, 17 for 4 years, and 8 for 5 years. The numbers and duration of conjugated equine estrogen plus progestin use were as follows: 40 women for 1 year, 55 for 2 years, 58 for 3 years, 50 for 4 years, and 40 for 5 years.

Mean E2 values were similar in premenopausal women and the 2 HT groups and more than double the value in postmenopausal women (Table 1). There were no statistically significant mean differences by group in ERLL levels. Also, there were no differences in mean levels of total testosterone in the premenopausal women compared with the 2 HT groups, although testosterone levels were significantly higher in the postmenopausal women (Table 1).

There were pronounced estrone metabolite differences according to group (Table 1). Although mean 2-OHE1 levels were significantly higher in the premenopausal group than in the postmenopausal group, values in HT users were more than double those in either premenopausal or postmenopausal women (Table 1). Furthermore, mean 16α-OHE1 levels were more than 60% higher in HT users than in premenopausal or postmenopausal women.

Mean SHBG was almost 50% higher among the HT users than among the premenopausal or postmenopausal women (Table 1).

Intermediate cardiovascular markers

Mean oxidized LDL-C levels were similar in HT users and premenopausal women (Table 2). However, the LDL-C values in both premenopausal women and HT users were, on average, 11% to 12% lower than the mean value in postmenopausal women. The mean HDL-C values were approximately 10% lower in premenopausal women than in the other groups (Table 2). Collectively, there was a more favorable HDL-C:LDL-C ratio in women using HT (Table 2). Furthermore, the apo B to apo A-I ratio was more favorable in premenopausal and postmenopausal women than in HT users.

The triglyceride profile was less favorable in HT users than in either premenopausal or postmenopausal women (Table 2). Furthermore, mean Lp(a)-1 levels were approximately 10% higher in women using HT than in premenopausal and postmenopausal women.

Mean CRP level was significantly higher in women using the conjugated equine estrogen preparations than in postmenopausal women (Table 2). Mean level of F-isoprostanes, a measure of oxidative stress, was about 10% higher in women using the conjugated equine estrogen plus progestin preparation than in the other 3 groups, a difference significant at the P < .05 level. In contrast, the mean level of PAI-1, an inhibitor of fibrinolysis, was almost 50% lower in HT users than mean values in premenopausal and postmenopausal women (Table 2). Adjusting for duration of conjugated equine estrogen product use did not alter these associations.

Relating sex steroid measures to cvd markers

In exogenous hormone users, the partial correlations between E2 and HDL-C and apo A-I were 0.17 (95% CI, 0.05-0.29) and 0.19 (0.06-0.31), respectively. There were no other notable associations of CVD measures with E2, free estrogen index, or ERLL.

In premenopausal women, estrone metabolites were positively associated with LDL-C levels and with apo B. Partial correlations between LDL-C and the 2-OHE1 and 16α-OHE1 metabolites were 0.25 (95% CI, 0.02-0.47) and 0.33 (0.09-0.54), respectively. Partial correlations between apo B and 2-OHE1 and 16α-OHE1 metabolites were 0.20 (95% CI, 0.05-0.43) and 0.32 (0.08-0.53), respectively. However, in HT users, estrone metabolites and lipid measures were not statistically associated although mean metabolite levels were almost 50% higher in HT users.

Higher estrone metabolite levels were associated with higher levels of F2a-isoprostanes in all groups, indicating that estrone metabolites were associated with greater oxidative stress. In premenopausal women, the partial correlations between the logF2a-isoprostanes and log2-OHE1 and log16α-OHE1 metabolites were 0.57 (95% CI, 0.38-0.72) and 0.63 (0.46-0.76), respectively. In users of conjugated equine estrogen only, rY/X values between logF2a-isoprostanes and log2-OHE1 and log16α-OHE1 metabolites were 0.30 (95% CI, 0.05-0.58) and 0.43 (0.10-0.68), respectively. In users of conjugated equine estrogen plus progestin, rY/X values between logF2a-isoprostanes and log2-OHE1 and log16a-OHE1 metabolites were 0.44 (95% CI, 0.31-0.55) and 0.49 (0.37-0.59), respectively.

In premenopausal women, higher SHBG levels were associated with lower LDL-C, apo B, triglyceride, CRP, and PAI-1 levels (Table 3), suggesting a potential protective effect for SHBG. However, these relationships were not replicated in HT users, whose SHBG levels were remarkably higher than levels in premenopausal women. In women using conjugated equine estrogen only, SHBG levels were positively correlated with levels of HDL-C, apo A-I, and CRP; in women using conjugated equine estrogen plus progestin, associations were mixed (negative associations with LDL-C and PAI-1 levels, but positive associations with CRP level). The SHBG level was negatively associated with CRP levels in premenopausal women but positively associated with CRP levels in HT users.


The WHI and the Heart and Estrogen/Progestin Replacement Study demonstrated that widely prescribed conjugated equine estrogen–based products were not identified as cardioprotective4,5 in women remote from their final menstrual period, thereby generating debate about explanatory mechanisms. This debate is currently centered about the timing hypothesis,6 which speculates that HT use should be initiated within 6 years of the menopause transition (on the basis of data extrapolated from monkeys subjected to ovariectomy)27 so that optimal sex steroid and CVD profiles observed in the premenopause stage can be sustained before CVD progresses to an ir reversible state. We related a panel of usual and more novel estrogen and androgen measures with cardiovascular markers in HT users, premenopausal women, and postmenopausal women under the theory that cardiovascular values observed in premenopausal women were an expression of the desirable CVD environment during HT use.

Traditionally, studies of HT use have focused on the HDL-C levels, citing the greater cardioprotective aspects with this lipid fraction.2,3 We too found that HDL-C:LDL-C ratios and their apolipoprotein ratios were favorable in both conjugated equine estrogen groups. However, triglyceride levels were much less favorable in the HT users than in premenopausal women. This previously reported finding motivated the recommendation for using HT agents that do not require “first-pass” hepatic metabolism.28 The markedly higher SHBG levels, which affect sex steroid binding, may be a reflection of altered hepatic metabolism that indirectly affect cardiovascular measures. There were higher Lp(a)-1 and PAI-1 levels among HT users than in the other 2 groups. Lipoprotein(a) is thought to recruit inflammatory cells through interaction with Mac-1 integrin.29 The Lp(a) structure is similar to plasminogen and tissue plasminogen activator and it competes with plasminogen for its binding site, leading to reduction in fibrinolysis.

Although the effect of HT use on some, but not all, lipids appeared favorable, users had a less favorable oxidative environment and more pronounced inflammatory response. Premarin or the Premarin-progestin combination Prempro (Wyeth Pharmaceuticals Inc, Madison, New Jersey), which compose the preponderance of formulations used by SWAN (and all WHI) enrollees, include conjugated estrogens of which more than 50% are estimated to be estrone. Consistent with this, we identified much greater estrone metabolite excretion in HT users. Recent work suggests that higher estrone metabolite levels were highly associated with higher F2a-isoprostane levels,30 a product of the free radical–induced peroxidation of arachidonic acid.31,32 F2a-isoprostanes have been characterized as a superior marker of the in vivo oxidative stress response.10,33-35 The mean F2a-isoprostane level was highest in women using the conjugated equine estrogen plus progestin combination. Notably, this treatment arm of the WHI was terminated early because of excess cardiovascular risk.4

There were higher levels of CRP, an inflammation marker, in women using conjugated equine estrogen formulations than in postmenopausal women. Elevated CRP levels in hormone users have been reported previously by our group.36 In the Postmenopausal Estrogen/Progestin Interventions Study, hormone use was associated with 50% higher levels of CRP than in nonusers.37 The CRP level was higher in each treatment group, including estrogen alone or in combination with micronized progesterone or medroxyprogesterone acetate.37 The investigators in that study suggested that a mechanism for adverse cardiovascular events with HT use is through increasing inflammatory response, with the potential for accelerated atherosclerosis, plaque destabilization, or thrombosis.

Markedly higher SHBG levels were observed in women using HT. Sex hormone–binding globulin is synthesized in the liver with concentrations being regulated by the androgen-estrogen balance, thyroid hormones, and insulin. Other publications have identified that SHBG levels were significantly correlated with intermediate markers of CVD risk,36,38 including positive associations with HDL-C concentrations and negative associations with LDL-C, oxidized LDL-C, triglyceride, and apo B levels during the menopausal transition.39 Traditionally, descriptions of SHBG activity have focused on its binding capacity with testosterone and E2 in determining bioavailability so that less than 2% to 3% of these sex steroids are available for hormone receptor–mediated action.40 Expanded actions of SHBG are also being identified.40 It remains to be determined whether the elevated 2-OHE1 and 16α-OHE1 metabolites bind with SHBG and disrupt the favorable role of SHBG seen in premenopausal women. It also remains to be determined whether alterations in liver metabolism alter the impact of sex steroids on CVD metabolism via the increased SHBG production.

We observed no significant associations of the ERLL assay. Potentially, estrogen receptor binding is a minor mechanism in the lipid synthesis and degradation system relative to other relevant receptors such as the apo B receptor.

Strengths of this report include infrequently characterized lipid measures (oxidized LDL-C and the apo B:apo A-I ratio) for a more complete depiction of how hormone measures relate to CVD measures in HT users and nonusers. An expanded battery of sex steroid markers, especially the estrone metabolites, was also associated with CVD markers. Women were all premenopausal or in early perimenopause at study inception, so that transitions to postmenopause occurred within the 5 years of study observation. All study enrollees were free of hormone use at study inception, so that hormone therapy was initiated early in the menopausal transition. Women with heart disease at baseline were excluded from analysis. Data also reflect relationships observed before the release of WHI findings. However, there are limitations. Although we evaluated 2 specific formulations (conjugated equine estrogen with and without progestin) and duration of use, we did not specifically address dose. This cross-sectional evaluation precludes experimentally assigned HT use. Furthermore, SWAN includes measures of cardiovascular intermediates but not hard end points.

Collectively, our findings demonstrate positive and negative aspects of conjugated equine estrogen use relative to cardiovascular intermediate end points observed in women during the menopausal transition. The evidence suggests that conjugated equine estrogen use may provide positive effects via some, but not all, lipids; however, there were less favorable aspects of the use of conjugated equine estrogen, including higher estrone metabolite and SHBG levels and increased propensity for greater oxidative stress, thrombotic activity, and inflammation even in middle-aged women free of heart disease. From a research perspective, this study indicates that there should be a much closer examination of liver metabolism of SHBG and the estrone metabolites during conjugated equine estrogen use in relation to potential CVD risk. From a practice perspective, these results suggest very thoughtful consideration in the clinical use of HT because there was evidence of adverse HT effects even in women free of atherosclerosis who were studied in a time frame that was well within the approximate 6-year period proposed by the timing hypothesis.

Box Section Ref ID

The Study of Women's Health Across the Nation

Clinical Centers

University of Michigan, Ann Arbor: MaryFran Sowers, PhD, principal investigator; Massachusetts General Hospital, Boston: Robert Neer, MD, principal investigator 1995 to 1999, Joel Finkelstein, MD, principal investigator 1999 to present; Rush University, Rush University Medical Center, Chicago, Illinois: Lynda Powell, PhD, principal investigator; University of California, Davis/Kaiser: Ellen Gold, PhD, principal investigator; University of California, Los Angeles: Gail Greendale, MD, principal investigator; University of Medicine and Dentistry–New Jersey Medical School, Newark: Gerson Weiss, MD, principal investigator 1995 to 2004, Nanette Santoro, MD, principal investigator 2004 to present; and University of Pittsburgh, Pittsburgh, Pennsylvania: Karen Matthews, PhD, principal investigator.

National Institutes of Health Program Office

National Institute on Aging, Bethesda, Maryland: Marcia Ory, PhD, 1994 to 2001, Sherry Sherman, PhD, 1994 to present; National Institute of Nursing Research, Bethesda: program officers.

Central Laboratory

University of Michigan: Daniel McConnell, PhD (Central Ligand Assay Satellite Services).

SWAN Repository

University of Michigan: MaryFran Sowers, PhD.

Coordinating Center

New England Research Institutes, Watertown, Massachusetts: Sonja McKinlay, PhD, principal investigator 1995 to 2001; University of Pittsburgh: Kim Sutton-Tyrrell, PhD, principal investigator 2001 to present.

SWAN Steering Committee

Chris Gallagher, MD, chair; Susan Johnson, MD, chair.

Correspondence: MaryFran R. Sowers, PhD, Department of Epidemiology, School of Public Health, University of Michigan, 109 Observatory, Room 1851A, Ann Arbor, MI 48109.

Accepted for Publication: May 1, 2008.

Author Contributions: All authors had full access to the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Sowers. Acquisition of data: Sowers and McConnell. Analysis and interpretation of data: Sowers, Randolph, Jannausch, Lasley, and Jackson. Drafting of the manuscript: Sowers and Jannausch. Critical revision of the manuscript for important intellectual content: Sowers, Randolph, Lasley, Jackson, and McConnell. Statistical analysis: Sowers and Jannausch. Obtained funding: Sowers. Administrative, technical, and material support: Sowers, Randolph, and McConnell. Study supervision: Lasley and Jackson.

Financial Disclosure: None reported.

Funding/Support: The SWAN and the SWAN Repository have grant support from the National Institutes of Health, Department of Health and Human Services, through the National Institute on Aging, the National Institute of Nursing Research, and the National Institutes of Health Office of Research on Women's Health (grants AG17104, AG017719, NR004061, AG012505, AG012535, AG012531, AG012539, AG012546, AG012553, AG012554, AG012495, and AG17104). The oxidized LDL-C, ERLL, estrogen metabolites, and isoprostane data are from an estrogen metabolism grant using specimens from the SWAN Repository (AG17104).

Additional Information: This report is based on samples from the SWAN Core Repository. If scientists are interested in developing studies based on this resource, a description of the SWAN Core and DNA Repositories and how to obtain access to the resources can be found at http://www.swanrepository.org.

Tunstall-Pedoe  HKuulasmaa  KAmouyel  PArveiler  DRajakangas  AMPajak  A Myocardial infarction and coronary deaths in the World Health Organization MONICA Project: registration procedures, event rates, and case-fatality rates in 38 populations from 21 countries in four continents.  Circulation 1994;90 (1) 583- 612PubMedGoogle ScholarCrossref
Barrett-Connor  EBush  TL Estrogen and coronary heart disease in women.  JAMA 1991;265 (14) 1861- 1867PubMedGoogle ScholarCrossref
Barrett-Connor  EGrady  D Hormone replacement therapy, heart disease, and other considerations.  Annu Rev Public Health 1998;1955- 72PubMedGoogle ScholarCrossref
Rossouw  JEAnderson  GLPrentice  RL  et al.  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 (3) 321- 333PubMedGoogle ScholarCrossref
Grady  DWenger  NKHerrington  D  et al.  Postmenopausal hormone therapy increases risk for venous thromboembolic disease: the Heart and Estrogen/Progestin Replacement Study.  Ann Intern Med 2000;132 (9) 689- 696PubMedGoogle ScholarCrossref
Barrett-Connor  E Hormones and heart disease in women: the timing hypothesis.  Am J Epidemiol 2007;166 (5) 506- 510PubMedGoogle ScholarCrossref
Seeger  HMueck  AOLippert  TH Effect of estradiol metabolites on the susceptibility of low density lipoprotein to oxidation.  Life Sci 1997;61 (9) 865- 868PubMedGoogle ScholarCrossref
Bradlow  HLHershcopf  REFishman  JF Oestradiol 16 alpha-hydroxylase: a risk marker for breast cancer.  Cancer Surv 1986;5 (3) 573- 583PubMedGoogle Scholar
Swaneck  GEFishman  J Covalent binding of the endogenous estrogen 16 alpha-hydroxyestrone to estradiol receptor in human breast cancer cells: characterization and intranuclear localization.  Proc Natl Acad Sci U S A 1988;85 (21) 7831- 7835PubMedGoogle ScholarCrossref
Morrow  JD Quantification of isoprostanes as indices of oxidant stress and the risk of atherosclerosis in humans.  Arterioscler Thromb Vasc Biol 2005;25 (2) 279- 286PubMedGoogle ScholarCrossref
Roberts  LJMorrow  JD Measurement of F2-isoprostanes as an index of oxidative stress in vivo.  Free Radic Biol Med 2000;28 (4) 505- 513PubMedGoogle ScholarCrossref
Sowers  MFCrawford  SSternfeld  B  et al.  Design, survey sampling and recruitment methods of SWAN: a multi-center, multi-ethnic, community-based cohort study of women and the menopausal transition. Wren  JLobo  RAKelsey  Jmarcus  R Menopause: Biology and Pathobiology. San Diego, CA Academic Press2000;175- 188Google Scholar
Brambilla  DJ McKinlay  SMJohannes  CB Defining the perimenopause for application in epidemiologic investigations.  Am J Epidemiol 1994;140 (12) 1091- 1095PubMedGoogle Scholar
Dudley  ECHopper  JLTaffe  JGuthrie  JRBurger  HGDennerstein  L Using longitudinal data to define the perimenopause by menstrual cycle characteristics.  Climacteric 1998;1 (1) 18- 25PubMedGoogle ScholarCrossref
 Research on the menopause in the 1990s: report of a WHO Scientific Group.  World Health Organ Tech Rep Ser 1996;8661- 107PubMedGoogle Scholar
Klug  TLBradlow  HLSepkovic  DW Monoclonal antibody-based enzyme immunoassay for simultaneous quantitation of 2- and 16 alpha-hydroxyestrone in urine.  Steroids 1994;59 (11) 648- 655PubMedGoogle ScholarCrossref
Rogers  JMDenison  MS Recombinant cell bioassays for endocrine disruptors: development of a stably transfected human ovarian cell line for the detection of estrogenic and anti-estrogenic chemicals.  In Vitr Mol Toxicol 2000;13 (1) 67- 82PubMedGoogle Scholar
Natarajan  KOverstreet  JWRogers  JM  et al.  Detection of xenoestrogens in serum after immunoprecipitation of endogenous steroidal estrogens.  Environ Health Perspect 2002;110 (8) 791- 795PubMedGoogle ScholarCrossref
Steiner  PMFreidel  JBremner  WFStein  EA Standardization of micro-methods for plasma cholesterol, triglyceride and HDL-cholesterol with the Lipid Research Clinics' methodology [abstract].  J Clin Chem Clin Biochem 1981;19 (8) 850Google Scholar
Warnick  GRAlbers  JJ A comprehensive evaluation of the heparin-manganese precipitation procedure for estimating high density lipoprotein cholesterol.  J Lipid Res 1978;19 (1) 65- 76PubMedGoogle Scholar
Friedewald  WTLevy  RIFredrickson  DS Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge.  Clin Chem 1972;18 (6) 499- 502PubMedGoogle Scholar
Stein  EKumbla  LMiller  JSrivastiva  LKashyap  M Development and evaluation of a competitive ELISA for Lp(a) [abstract 0566].  Clin Chem 1992;38 (6) 1067Google Scholar
Myers  GLCooper  GRWinn  CLSmith  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 (1) 105- 135PubMedGoogle Scholar
Ferris  BG Epidemiology Standardization Project (American Thoracic Society).  Am Rev Respir Dis 1978;118 (6, pt 2) 1- 120PubMedGoogle Scholar
Sternfeld  BAinsworth  BEQuesenberry  CP Physical activity patterns in a diverse population of women.  Prev Med 1999;28 (3) 313- 323PubMedGoogle ScholarCrossref
Baecke  JABurema  JFrijters  JE A short questionnaire for the measurement of habitual physical activity in epidemiological studies.  Am J Clin Nutr 1982;36 (5) 936- 942PubMedGoogle Scholar
Mikkola  TSClarkson  TB Estrogen replacement therapy, atherosclerosis and vascular function.  Cardiovasc Res 2002;53 (3) 605- 619PubMedGoogle ScholarCrossref
Turgeon  JLCarr  MCMaki  PMMendelsohn  MEWise  PM Complex actions of sex steroids in adipose tissue, the cardiovascular system, and brain: insights from basic science and clinical studies.  Endocr Rev 2006;27 (6) 575- 605PubMedGoogle ScholarCrossref
Sotiriou  SNOrlova  VVAl-Fakhri  N  et al.  Lipoprotein(a) in atherosclerotic plaques recruits inflammatory cells through interaction with Mac-1 integrin.  FASEB J 2006;20 (3) 559- 561PubMedGoogle Scholar
Sowers  M McConnell  DJannausch  ML  et al.  Oestrogen metabolites and their relation to isoprostanes as a measure of oxidative stress.  Clin Endocrinol (Oxf) 2008;68 (5) 806- 813PubMedGoogle ScholarCrossref
Neele  DMde Wit  ECPrincen  HM Insulin suppresses apolipoprotein(a) synthesis by primary cultures of cynomolgus monkey hepatocytes.  Diabetologia 1999;42 (1) 41- 44PubMedGoogle ScholarCrossref
Rainwater  DLHaffner  SM Insulin and 2-hour glucose levels are inversely related to Lp(a) concentrations controlled for LPA genotype.  Arterioscler Thromb Vasc Biol 1998;18 (8) 1335- 1341PubMedGoogle ScholarCrossref
Kikuchi  TOnuma  TShimura  M  et al.  Different change in lipoprotein(a) levels from lipid levels of other lipoproteins with improved glycemic control in patients with NIDDM.  Diabetes Care 1994;17 (9) 1059- 1061PubMedGoogle ScholarCrossref
Heesen  BJWolffenbuttel  BHLeurs  PB  et al.  Lipoprotein(a) levels in relation to diabetic complications in patients with non-insulin-dependent diabetes.  Eur J Clin Invest 1993;23 (9) 580- 584PubMedGoogle ScholarCrossref
Rainwater  DLMacCluer  JWStern  MPVandeBerg  JLHaffner  SM Effects of NIDDM on lipoprotein(a) concentration and apolipoprotein(a) size.  Diabetes 1994;43 (7) 942- 946PubMedGoogle ScholarCrossref
Sowers  MRJannausch  MRandolph  JF  et al.  Androgens are associated with hemostatic and inflammatory factors among women at the mid-life.  J Clin Endocrinol Metab 2005;90 (11) 6064- 6071PubMedGoogle ScholarCrossref
Cushman  MLegault  CBarrett-Connor  E  et al.  Effect of postmenopausal hormones on inflammation-sensitive proteins: the Postmenopausal Estrogen/Progestin Interventions (PEPI) Study.  Circulation 1999;100 (7) 717- 722PubMedGoogle ScholarCrossref
Sutton-Tyrrell  KWildman  RPMatthews  KA  et al. SWAN Investigators, Sex hormone–binding globulin and the free androgen index are related to cardiovascular risk factors in multiethnic premenopausal and perimenopausal women enrolled in the Study of Women's Health Across the Nation (SWAN).  Circulation 2005;111 (10) 1242- 1249PubMedGoogle ScholarCrossref
Berg  GMesch  VBoero  L  et al.  Lipid and lipoprotein profile in menopausal transition: effects of hormones, age and fat distribution.  Horm Metab Res 2004;36 (4) 215- 220PubMedGoogle ScholarCrossref
Khosla  S Sex hormone binding globulin: inhibitor or facilitator (or both) of sex steroid action [editorial]?  J Clin Endocrinol Metab 2006;91 (12) 4764- 4766PubMedGoogle ScholarCrossref