| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Departments of Psychiatry, Epidemiology, and Psychology (K.A.M.) and Epidemiology (Y.C., K.S.-T.), University of Pittsburgh, Pittsburgh, Pennsylvania 15213; Department of Obstetrics/Gynecology and Women’s Health (N.S.), Albert Einstein College of Medicine, Bronx, New York 10461; Department of Population Health and Reproduction (B.L.), University of California Davis and Kaiser Permanente, Davis, California 95616; Departments of Epidemiology and Biostatistics (S.C.), University of Massachusetts Medical School, Worcester, Massachusetts 01655; Massachusetts General Hospital (R.C.P.), Boston, Massachusetts 02114; and University of Michigan (M.S.), Ann Arbor, Michigan 48109
Address all correspondence and requests for reprints to: Karen A. Matthews, Ph.D., Department of Psychiatry, University of Pittsburgh, 3811 O’Hara Street, Pittsburgh, Pennsylvania 15213. E-mail: matthewska{at}upmc.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Objective: The objective of this study was to describe the relationships between menstrual cycle characteristics and daily reproductive hormone measures with CVD risk factors in middle-aged women.
Design and Setting: Cross-sectional associations were examined between CVD risk factors and urinary LH, FSH, estrone conjugates, and pregnanediol glucuronide (Pdg) measured across one menstrual cycle or 50 d.
Participants: Menstruating women (n = 500) who were free from diabetes or past stroke or heart attack enrolled in the Daily Hormone Study-Study of Women’s Health across the Nation were studied.
Main Outcome Measures: Body mass index (BMI), blood pressure, hemostatic, and metabolic factors were measured.
Results: Few differences existed in risk factors between women with evidence of luteal activity and those with no evidence of luteal activity. Associations between elevated CVD risk factors and long cycle length were reduced substantially by age and BMI adjustments. Those with lower estrone conjugate and PdG averaged across the follicular phase had higher waist circumference, triglycerides, insulin, plasminogen activator inhibitor type-1, tissue type plasminogen activator-antigen, and factor VIIc levels in age- and BMI-adjusted analyses (P < 0.05).
Conclusions: In midlife menstruating women, a longer cycle length was related to CVD risk factors, in large part through their common association with BMI. More favorable levels of metabolic and hemostatic factors were associated with higher levels of follicular-phase estrogen, a pattern consistent with a more competent ovary, and higher levels of follicular-phase PdG, perhaps of adrenal origin. Metabolic and hemostatic factors may be sensitive to hormonal variation during the early perimenopausal transition.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The present study examined associations between ovarian function during the menstrual cycle and cardiovascular risk factors in the Daily Hormone Study (DHS) from the Study of Women’s Health across the Nation (SWAN), a multiethnic cohort study of premenopausal and early perimenopausal women. Urinary levels of gonadotropins and sex steroid metabolites were assessed daily during one complete menstrual cycle or up to 50 consecutive days, whichever was shorter. Blood pressure, weight and waist circumference, metabolic factors, lipids and lipoproteins, and hemostatic and inflammatory markers were measured within a 3-month window of the hormone collection.
A priori categorization of women’s menstrual cycles allowed us to address a number of hypotheses. First, we anticipated that women who had evidence of luteal activity (ELA), presumably representing an ovulatory cycle, would have lower levels of CHD risk factors than women who experienced presumably anovulatory cycles [no ELA (NELA)]. Second, given that a longer cycle length occurs as women approach perimenopause, as well as in mild reproductive disorders, we anticipated that women with longer menstrual cycles, even if ovulatory, would have elevated risk factors. Third, we expected an inverse association between CVD risk factors and high estrogen in the follicular phase and high progesterone in the luteal phase, both being indicative of a reproductively competent ovary. We anticipated the findings to be stronger for metabolic and perhaps hemostatic factors, but not for blood pressure or weight gain, given the results of menopause cohort studies.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The sample consisted of 500 (153 Caucasian, 107 African American, 36 Hispanic, 92 Chinese, and 112 Japanese) DHS volunteers whose menstrual cycle urine collection occurred within 90 d of cardiovascular risk factor assessment (mean number of days = 37.2, SD = 26.9) and whose health history was negative for diabetes, heart attack, or stroke. They were identified from the approximately 3200 women enrolled at one of the seven clinical sites. SWAN eligibility criteria included: being from a designated minority for the given site or Caucasian; aged 42–52 yr; intact uterus and at least one ovary; not currently pregnant or breast-feeding; no use in the previous 3 months of exogenous hormone preparations, which could affect ovarian or pituitary function; and at least one menstrual period within the past 3 months. Sites used a number of sampling strategies and sampling frames to enroll women in SWAN (8). Informed consent was obtained, and all the relevant institutional review boards approved the study.
The DHS began the first year after baseline evaluation (9). With the exception of age, criteria were as above. A total of 867 women (259 Caucasian, 181 African American, 95 Hispanic, 155 Chinese, and 177 Japanese women) completed a daily specimen collection across the menstrual cycle or 50 d. Of these women, 840 had sufficient data to allow determination of luteal activity (see definition below). Comparisons of the 840 eligible participants, 247 eligible women who did not participate, and 113 eligible women with incomplete collections showed no differences by pre- vs. perimenopausal status, smoking history, or body mass index (BMI) or serum hormones other than FSH. DHS participants were slightly younger (by 0.5 yr), of Japanese and Chinese origin, less educated, and had lower serum FHS levels, relative to the others.
Specimen collection
Women collected specimens beginning with the first day of menstrual bleeding, if possible, and ended on the first day of bleeding of the subsequent cycle or after 50 d, whichever occurred first. They stored their samples in a nonfrost-free freezer. When collection was completed, urine specimens were transported on ice to the CLASS Laboratory at the University of Michigan for analysis (for details of the protocol, see Ref.9).
Hormone measurement
Specimens were glycerol-preserved to permit measurement of urinary (u)LH and uFSH over long storage periods (10) and to prevent interference with the estrone conjugates (E1c) and pregnanediol glucuronide (Pdg) assays (11, 12). Samples were normalized for the amount of creatinine in each specimen and expressed per milligram creatinine. uLH, uFSH, E1c, and Pdg were assayed using newly adapted chemiluminescent assays configured to be compatible with the ACS-180 Autoanalyzer (Corning Life Sciences, Acton, MA) (9). For uFSH, the inter- and intraassay coefficients of variations (CV) were 11.4 and 3.8%, respectively. For uLH, the inter- and intraassay CVs were 10.9 and 4.6%, respectively. For E1c, the minimum detectable concentration was 0.1 ng/ml, and the inter- and intraassay CVs for the E1c were 11.5 and 8.1%, respectively. For Pdg, the minimum detectable concentration was 0.0001 ng/ml, and the inter- and intraassay CVs were 17.8 and 7.7%, respectively.
Serum hormone levels were collected at the annual examinations within the early follicular phase days (13). Serum estradiol (E2) concentrations were measured with a modified, off-line ACS:180 (E2-6) immunoassay. Inter- and intraassay CVs averaged 10.6 and 6.4%, respectively, and the lower limit of detection was 1 pg/ml. Testosterone concentrations were evaluated with the ACS:180 total testosterone assay modified to increase precision in the low ranges. Inter- and intraassay CVs were 10.5 and 8.5%, respectively, and the lower limit of detection was 2 ng/dl. Serum dehydroepiandrosterone sulfate (DHEAS) concentrations involved competitive binding of dimethyl-acridinium ester with to a commercially available anti-DHEAS antibody and a solid phase conjugated to paramagnetic particles. Inter- and intraassay CVs for DHEAS were 11.3 and 7.6%, respectively, and the lower limit of detection was 2 µg/dl.
Measurement of menstrual cycle characteristics
Women were categorized into ELA and NELA groups, with the latter group further divided into those who ended the cycle collection with a menstrual bleed (NELA-bleed) vs. no bleed (NELA-no bleed). The dichotomous determination of luteal activity was based on having the algorithm developed by Kassam et al. (14) and modified for use in SWAN (9). The algorithm defined a cycle-specific baseline of Pdg using a 5-d moving average and considered a 3-fold increment over the nadir 5-d average Pdg that was sustained for a minimum of 3 d to be consistent with ovulation. The Kassam algorithm was validated against weekly serum progesterone determination from a test sample of midreproductive, regularly cycling women. The probable day of the luteal transition in ELA cycles was assessed using a modified version (10) of previously developed algorithms of Waller et al. (15) and Baird et al. (16) and was based on an increase in the ratio of E1c to Pdg followed by an immediate decrease in the ratio. These assessments allowed us to divide cycles into follicular (all days before the day of the luteal transition, d 0) and luteal (all days after the day of the luteal transition) phases. The modified Kassam et al. (9, 14) algorithm had 98% sensitivity and 96% specificity for the detection of ELA status, when compared with a panel of six expert raters, and the algorithmic-derived day of luteal transition agreed with the experts to within 3 d in 92% of the ELA collections.
Ninety-seven percent of women provided at least 80% of all possible cycle days and were considered to have adequate data for characterization of their menstrual cycle. Daily hormone concentrations were averaged within the follicular and luteal phases, using the trapezoidal rule. Cycles missing hormones for 2 or more consecutive days were omitted from the analyses of integrated hormones, resulting in 7% missing data. Among ELA women, cycle length was categorized into 24 d or less, 25–32 d, and 33 d or more.
Cardiovascular risk factor measurement
Risk factors were measured if they were predictors of coronary events in women or if they were influenced by hormonal factors. The blood draw occurred after a minimum 10-h fast between d 2 and 5 of the follicular phase to allow for a reasonable standardized hormonal milieu among women still menstruating. Samples were maintained at 4 C until separated, frozen at –80 C, and shipped on dry ice to the central laboratory (Medical Research Laboratories, Highland Heights, KY), which is certified by the Centers for Disease Control Lipid Standardization Part III program (17). All lipid and lipoprotein fractions were analyzed on EDTA-treated plasma. Total cholesterol and triglycerides were analyzed by enzymatic methods, and high-density lipoprotein-cholesterol (HDL-C) was isolated using heparin-2M manganese chloride (18). Serum insulin was measured using RIA (DPC Coat-a-count, Los Angeles, CA) procedure and monitored as part of the monthly quality assurance program by the Diabetes Diagnostic Laboratory at the University of Missouri. Glucose was measured using a hexokinase-coupled reaction on a Hitachi 747–200 (Roche Molecular Biochemicals Diagnostics, Indianapolis, IN). Lipoprotein (a) [Lp(a)] was quantified by competitive ELISA (19). Fibrinogen and factor VII were measured in frozen citrated plasma using a clot-based turbidimetric detection system, with factor VII assay using factor VII-deficient plasma in preparing the standard curve. Tissue type plasminogen activator-antigen (tPA-ag) was measured in plasma using a double antibody in an ELISA (American Diagnostica, Greenwich, CT), with a human single-chain tPA-ag as a standard calibrated against an international standard (NIBSAC, Hertfordshire, UK). Plasminogen activator inhibitor type-1 (PAI-1) was measured using a solid-phased monoclonal antibody and an enzyme-labeled goal second antiserum for detection (American Diagnostica). C-reactive protein-high sensitivity (CRP-hs) was measured using an ultrasensitive rate immunonephelometry (Dade-Behring, Marburg, Germany).
Waist circumference (narrowest part of torso) was measured over undergarments. Seated blood pressure was measured twice using a standard mercury column and averaged. Technicians were certified for their performance and compliance with standard SWAN protocol before collecting the physical measures. Current smoking status (yes/no) was assessed by self-report. Height in centimeters and weight in kilograms were measured, and BMI was calculated by dividing weight by the square of height in meters.
Statistical analyses
Skewed variables were log transformed before analyses, i.e. integrated hormones, triglycerides, Lp(a), glucose, insulin, PAI-1, and CRP-hs. Five women on lipid-lowering medications were excluded from the lipid analyses, and 51 women on hypertensive medication were excluded from the blood pressure analyses. Study hypotheses were analyzed with a series of Student’s t tests and ANOVAs for continuous variables and 
2 tests for categorical variables comparing the cardiovascular risk factors of ELA and NELA women; the NELA-bleed and NELA-no bleed women; and three cycle length categories among ELA women. Age-adjusted partial correlation coefficients described the associations between the daily integrated hormone measures and risk factors. Additional partial correlation adjusted for age, BMI, and smoking; 17 women without BMI data were excluded from this analysis. The interval between measurements of menstrual cycle characteristics and risk factors was not associated with any cardiovascular risk factors. To test whether the significant age- and BMI-adjusted associations between risk factors and menstrual cycle characteristics and hormones varied by ethnic group, we also conducted regression analyses including the interaction of ethnicity and the predictor variable in the model along with age, BMI, ethnicity, and predictor variable. Unless noted otherwise, the results of the full covariate analysis and that of age and BMI were the same, and associations did not vary by ethnic group. P values 
0.05 were considered statistically significant.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
shows the characteristics of the study sample. By design, women were from diverse ethnic backgrounds. Most women had more than a high school education, were early perimenopausal (self reports of bleeding within the last 3 months and irregularity in the past year), and had ELA cycles.
|
The proportions of women in each ethnic group did not vary between ELA and NELA status, P = 0.31. Women classified as having ELA cycles had lower BMI, waist circumference, diastolic blood pressure, and fibrinogen, compared with women classified as having NELA cycles (Table 2; age-adjusted). Further adjustment for BMI did not alter the results for diastolic blood pressure and fibrinogen.
|
Among women with ELA cycles, 65 had short cycles, 51 had long cycles, and 304 had intermediate length cycles. Cycle length varied by ethnicity, with more Hispanics having long cycles relative to Chinese or Japanese (30.0 vs. 6.3 and 7.1%, respectively), with African Americans and Caucasians intermediate (16.7 and 12.6%, respectively). Longer cycle length was associated with higher BMI, waist circumference, blood pressure, triglycerides, glucose, insulin, fibrinogen, tPA-ag, factor VII, and CRP-hs levels and lower HDL-C levels, after age adjustments (Table 3). When adjusted for BMI, triglycerides and CRP-hs remained elevated among the long cyclers. Length of the luteal phase was positively associated with only fibrinogen (r = 0.11, P = 0.03).
|
E1c concentrations.
Among women classified as having an ELA cycle, higher E1c concentrations during the follicular phase were associated with lower BMI, waist circumference, apoB, triglycerides, Lp(a), glucose, insulin, PAI-1, tPA-ag, and CRP-hs, with the associations remaining significant for waist circumference, triglycerides, Lp(a), insulin, and PAI-1 in the age- and BMI-adjusted analyses (Table 4).
|
The daily E1C concentrations averaged within the luteal phase were only associated with lower PAI-1 in the age- and BMI-adjusted analyses.
PdG concentrations. As expected, higher daily Pdg concentrations averaged within the luteal phase were associated with lower BMI, waist circumference, blood pressure, triglycerides, Lp(a), insulin, PAI-1, tPA-ag, factor VII, and CRP-hs in age-adjusted analyses. The association with systolic blood pressure was strongest for Hispanic women. However, higher daily Pdg concentrations were only associated with lower tPA-ag in the age- and BMI-adjusted analyses.
Unexpectedly, lower Pdg concentrations averaged within the follicular phase were associated with all but one risk factor. Age- and BMI-adjusted analyses showed that higher Pdg measured during the follicular phase was associated with lower waist circumference, triglycerides, insulin, PAI-1, tPA-ag, and factor VII (Table 4
).
Other urinary hormones. Women with lower risk factor levels had higher uFSH and uLH concentrations measured during either phase. However, after age and BMI adjustments, few risk factors were associated with uFSH and uLH concentrations measured in both phases.
Other relevant data
Among women who had ELA cycles, there were strong associations between each of the urinary concentrations across phases (r 
0.57). Urinary E1c concentrations in both the follicular and luteal phase were similarly correlated with serum DHEAS (r = 0.17 and 0.15, P < 0.002, respectively), but only follicular phase E1c was associated with serum E2 (r = 0.17, P < 0.001). Pdg concentrations measured in the follicular phase were more strongly associated with serum DHEAS than when measured during the luteal phase (r = 0.31, P < 0.001 vs. r = 0.09, P = 0.09). Luteal phase Pdg concentrations were associated negatively with serum testosterone (r = –0.14, P = 0.005). Age- and BMI-adjusted correlations showed that serum E2 was associated significantly with higher HDL-C (r = 0.13, P = 0.004) and fibrinogen (r = 0.14, P = 0.003) and lower fasting glucose (r = –0.09, P = 0.05), testosterone with higher HDL-C (r = 0.11, P = 0.02) and tPA (r = 0.10, P < 0.05), and DHEAS with higher fasting glucose (r = 0.13, P = 0.004).
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
There was partial support for a third hypothesis. Pdg concentrations were inversely associated with elevated risk factor levels but unexpectedly in both the luteal and follicular phase. Age- and BMI-adjusted associations were statistically significant for more risk factors when PdG was measured in the follicular than in the luteal phase (six vs. one risk factor associations). The source of Pdg concentrations during the follicular phase is likely to be adrenal rather than ovarian (21). In support of this interpretation, our results showed that urinary Pdg concentrations during the follicular phase were more related to serum DHEAS than those measured during the luteal phase. Alternatively, Pdg in the very early follicular phase is the result of delayed excretion from the prior luteal phase. Thus, high rates of progesterone production in the prior luteal phase or a foreshortened follicular phase could increase the average follicular phase Pdg levels.
As anticipated, higher daily E1c concentrations during the follicular phase were associated with lower levels of BMI, adjusted for age, and waist circumference, triglycerides, Lp(a), insulin, and PAI-1, adjusted for BMI and age. The findings imply that women who have a reproductively competent ovary in midlife, as evidenced by high levels of E1c during the follicular phase and ELA, are at lower CVD risk, primarily through lower levels of metabolic and hemostatic factors.
An advantage of the daily assessment of urinary hormone metabolites is its better approximation of women’s overall exposure to hormones during a menstrual cycle, in contrast to a single serum hormone measurement during the very early follicular phase when estrogens and progesterone are known to be particularly low and have less variability than later in the cycle. Our results showed stronger and more theoretically consistent associations with urinary hormone metabolites than with serum hormone measures, although the serum hormones were measured concurrently with the cardiovascular risk factors. It should be noted that an index of free circulating testosterone (not measured in DHS) was also associated with elevated risk factors, especially the metabolic factors in the full SWAN cohort (22, 23).
Obesity was an important factor in the influence of cycle length. Obesity may lead to longer cycles and suppressed reproductive hormones, which, in turn, lead to alterations in other CHD risk factors. Adipose tissue secretes a number of adipocytokines, such as leptin, which are known to influence hypothalamic-pituitary-ovarian function (24). Thus, increased adiposity per se may have an adverse effect on reproductive hormones and menstrual cycles (25, 26). Lower reproductive hormone levels in midlife may lead to increasing levels of risk factors, including weight, such that obesity is not the initiating event. This model is unlikely because the perimenopausal transition is not associated with weight gain, beyond the effect of chronological aging (2), and weight reduction interventions lead to improvement in other cardiovascular risk factors (27) and menstrual cyclicity (25, 26). Obesity in midlife may influence the reproductive and metabolic systems independently. The fact that low E1c levels observed during the follicular phase were associated with metabolic risk factors and some hemostatic factors, independent of BMI, is consistent with this possibility. Longitudinal measures of hormones, cardiovascular risk factors, and menstrual cycle characteristics will permit us to evaluate these models as we continue to follow the enrolled women.
The study has a number of limitations and strengths. One single menstrual cycle may not represent cycles in the months surrounding risk factor assessment, thereby accounting for the few differences between ELA and NELA women’s risk factors. Although the modified Kassam algorithm determined cycles likely to be ovulatory, the cycles classified as having no evidence of clear luteal activity may be a heterogeneous mixture of ovulatory and anovulatory menstrual cycles. We did, however, compare the risk factors of NELA women according to whether their specimen collections ended with bleeding or not, and this distinction was not related to cardiovascular risk factors. Among the strengths of the study are the large number of cycles that were characterized, the large number of established and emerging cardiovascular risk factors measured, and the multiethnic nature of the sample.
In summary, evidence of ovulation in a single cycle had minimal association with cardiovascular risk factor levels in midlife women. Longer cycle length was related to cardiovascular risk factors, in large part through their common association with BMI. Estrone urinary metabolites and Pdg concentrations during the follicular phase were associated with metabolic and hemostatic factors, independent of age and BMI, suggesting that metabolic and hemostatic factors may be most sensitive to changes during the perimenopausal transition. Characterization of hormonal exposure across the menstrual cycle phases provides unique information about risk factors associated with later CVD.
|
Acknowledgments |
|---|
Central Laboratory: University of Michigan (Ann Arbor, MI), Rees Midgley, PI, 1995–2000; Dan McConnell, PI, 2000 to present (U01 AG12495, Central Ligand Assay Satellite Services).
Coordinating Center: New England Research Institutes (Watertown, MA), Sonja McKinlay, PI, 1995–2000 (U01 AG12553); University of Pittsburgh (Pittsburgh, PA), Kim Sutton-Tyrrell, PI, 2000-present (U01 AG12546).
Steering Committee Chairs: Chris Gallagher, 1995–1996; Jenny Kelsey, 1996–2002; Susan Johnson, 2002 to present.
National Institutes of Health Project Offices: National Institute on Aging (Bethesda, MD), Sherry Sherman; National Institute of Nursing Research (Bethesda, MD), Carole Hudgings and Janice Phillips.
|
Footnotes |
|---|
First Published Online February 21, 2006
Abbreviations: BMI, Body mass index; CHD, coronary heart disease; CRP-hs, C-reactive protein-high sensitivity; CV, coefficient(s) of variations; CVD, cardiovascular disease; DHEAS, dehydroepiandrosterone sulfate; DHS, Daily Hormone Study; E2, estradiol; E1c, estrone conjugate; ELA, evidence of luteal activity; HDL-C, high-density lipoprotein-cholesterol; Lp(a), lipoprotein (a); NELA, no ELA; PAI-1, plasminogen activator inhibitor type-1; Pdg, pregnanediol glucuronide; tPA-ag, tissue type plasminogen activator-antigen; SWAN, Study of Women’s Health across the Nation; u, urinary.
Received May 12, 2005.
Accepted February 13, 2006.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  A. M. Z. Jukic, C. R. Weinberg, A. J. Wilcox, D. R. McConnaughey, P. Hornsby, and D. D. Baird Accuracy of Reporting of Menstrual Cycle Length Am. J. Epidemiol., January 1, 2008; 167(1): 25 - 33. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Keck, M. J. Romero-Aleshire, Q. Cai, P. B. Hoyer, and H. L. Brooks Hormonal status affects the progression of STZ-induced diabetes and diabetic renal damage in the VCD mouse model of menopause Am J Physiol Renal Physiol, July 1, 2007; 293(1): F193 - F199. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
