Author + information
- Received July 27, 2006
- Revision received September 19, 2006
- Accepted October 18, 2006
- Published online January 2, 2007.
- Alice Y. Chang, MD4,⁎ (, )
- Shuaib M. Abdullah, MD,
- Tulika Jain, MD,
- Harold G. Stanek, MS,
- Sandeep R. Das, MD, MPH,
- Darren K. McGuire, MD, MHSc3,
- Richard J. Auchus, MD, PhD2 and
- James A. de Lemos, MD1
- ↵⁎Reprint requests and correspondence:
Dr. Alice Y. Chang, University of Texas Southwestern Medical School, MC 9066, 5323 Harry Hines Boulevard, Dallas, Texas 75390-9066.
Objectives We sought to determine if natriuretic peptides are associated with estrogen and androgen status in a population study of young women without known cardiac disease.
Background Circulating concentrations of plasma B-type natriuretic peptide (BNP) and N-terminal proBNP (NT-proBNP) are higher in women than in men, and they may be influenced by estrogens and androgens.
Methods Cardiac magnetic resonance imaging, dual energy X-ray absorbtiometry, and measurements of BNP, NT-proBNP, follicle-stimulating hormone (FSH), total testosterone, and sex hormone-binding globulin (SHBG), were performed in 682 women (ages 35 to 49 years) participating in the Dallas Heart Study.
Results In multivariable analyses adjusting for age, race/ethnicity, body mass index (BMI), serum creatinine, left ventricular mass and left ventricular ejection fraction <55%, menopausal status, and FSH were not associated with BNP and NT-proBNP. In contrast, higher SHBG was associated with higher BNP and NT-proBNP, while the free androgen index and calculated free testosterone were inversely associated with BNP and NT-proBNP (p < 0.0001 for each). Addition of SHBG or any measure of free testosterone to the multivariable models modified the effect of BMI and lean mass, such that measures of body composition were no longer significantly associated with BNP or NT-proBNP.
Conclusions Among young women, measures of free testosterone were independently and inversely associated with BNP and NT-proBNP. These results suggest that circulating free testosterone, not estradiol, mediates gender differences in natriuretic peptides. In addition, the association between higher BMI and lean body mass with natriuretic peptides may be mediated by testosterone.
Three recent population-based studies have demonstrated that circulating concentrations of B-type natriuretic peptide (BNP) and N-terminal proBNP (NT-proBNP) are higher in women than in men (1–3). The biologic basis for this gender difference in natriuretic peptide concentrations has not been delineated. Understanding the factors contributing to gender differences in BNP and NT-proBNP may improve the discriminatory value of these measurements for screening and diagnostic purposes. The identification of mechanisms responsible for these gender differences also may provide novel insights into gender differences in cardiovascular physiology and pathophysiology that would have implications for the treatment and prevention of heart disease.
Previous studies have suggested an association between exogenous estrogen and circulating natriuretic peptide concentrations in post-menopausal women. A population-based study from Olmsted County, Minnesota, found modestly higher BNP concentrations in older women who were treated with oral estrogens versus those who were not treated (3). However, this study did not obtain other measures of hormone status aside from self-report of menopause. A subsequent experimental study in post-menopausal women demonstrated a >3-fold increase in mean BNP concentrations after 3 months of transdermal estradiol administration (4).
Observed BNP changes after puberty also suggest that gonadal steroid hormones mediate gender differences in natriuretic peptides. Mean BNP concentrations did not differ in 43 boys and girls younger than 10 years of age but were significantly higher in girls (12.1 pg/ml) compared with boys (5.1 pg/ml, p < 0.001) 10 years or older (5). No study to date has examined the influence of endogenous estrogens and androgens on circulating natriuretic peptide concentrations in pre-menopausal women, the group for whom the greatest variation would be expected.
We tested the hypothesis that plasma BNP and NT-proBNP would be associated with estrogen status in a population study of young adult women without known cardiac disease, using various measurements of reproductive hormone status: menopausal status, oral estrogen use, follicle-stimulating hormone (FSH), sex hormone-binding globulin (SHBG), total testosterone (TT), free androgen index (FAI), and calculated free testosterone (cFT). Because there is an established influence of gonadal steroids on body composition, we also evaluated whether estrogen and free testosterone status modified the previously demonstrated associations between body composition and natriuretic peptides in this population (6–9).
The Reynolds Women’s Study is a substudy of the Dallas Heart Study, a probability-based sample of Dallas County adults, oversampled for African Americans (n = 6,101). Details of the Dallas Heart Study design and characteristics of the enrolled cohort have been previously described (10). The study protocol was approved by the University of Texas Southwestern Medical Center Institutional Review Board, and all participants provided written informed consent to participate in the study.
Subjects and data collection
The Reynolds Women’s Study obtained additional reproductive health and menstrual cycle information by survey, hormonal measurements, and pelvic imaging in women, ages 35 to 49 years, to examine the relationship of the subclinical and clinical polycystic ovarian syndrome phenotypes to preclinical and clinical heart disease. The Reynolds Women’s Study was performed in 3 visits. Of the 1,179 women between the ages of 35 and 49 years completing the visit 1 survey, 937 returned for visit 2 to provide blood and urine samples. At visit 3, 827 women were age-eligible for an additional reproductive health survey and pelvic magnetic resonance imaging (MRI) to assess ovarian size and structure. In addition, as part of the overall study design, cardiac MRI and total body dual energy X-ray absorbtiometry (DEXA) imaging were performed at visit 3. For the purposes of this analysis, we included only those women who had complete hormonal profiles (FSH, SHBG, and TT) and albumin, BNP, and NT-proBNP measurements from visit 2, as well as DEXA and cardiac MRI results from visit 3. We excluded subjects with a self-reported history of heart failure (n = 20), which also included one woman with a creatinine >2 mg/dl, leaving a final study population of 682 women (Fig. 1).The baseline characteristics of this study population (n = 682) were similar to the 497 women ages 35 to 49 years (n = 497) who were not included, except for a significantly lower percentage of African Americans in the study group (data not shown). However, the racial composition of the study population (49% African American) approximates the parent study racial composition (50% African American) (10).
Venous blood was collected in standard blood collection tubes containing ethylene diamine tetraacetic acid for plasma and in serum separator tubes for serum. Samples were maintained at 4°C for ≤4 h and then centrifuged (1,430 gfor 15 min) at 4°C. Plasma or serum supernatant was separated and frozen at −70°C until assayed. B-type natriuretic peptide was measured on a TECAN Genesis RSP 200/8 robotic high-throughput platform (Biosite Inc., San Diego, California) (11), and NT-proBNP was measured on the Elecsys proBNP platform (Roche Diagnostics, Indianapolis, Indiana) (12). The coefficient of variation for the BNP assay averaged 11% at a concentration of 30 ng/l and 6% at concentrations >60 ng/l; the coefficient of variation for the NT-proBNP assay was 3% at a concentration of 280 ng/l and 3% at a concentration of 6,000 ng/l.
Measurements of FSH, LH, and SHBG were performed on serum samples using 2-site immunoradiometric sandwich assays from Diagnostic Systems Laboratories (Webster, Texas). For FSH, the lower detection limit was 0.1 IU/l, and the coefficient of variation was 3.6% at a concentration of 6.7 IU/l, and 3.1% at 54 IU/l. For the SHBG assay, the lower limit of detection was 3 nmol/l with a coefficient of variation of 3.7% at a concentration of 92 nmol/l. Albumin was measured by spectrophotometry.
Measurement of TT was performed on plasma samples with a competitive radioimmunoassay from Diagnostic Systems Laboratories (DSL-4100). The lower limit for detection was 0.05 ng/ml with a coefficient of variation of 7.7% at a concentration of 0.52 ng/ml. The formula TT/SHBG × 100 was used to calculate FAI. The cFT was calculated from measured values of albumin, TT and SHBG using equations derived from the laws of mass action as reported by Vermeulen et al. (13). Both derived variables of cFT and FAI have been validated as estimates of unbound circulating testosterone in women (13,14).
Cardiac MRI was performed for the assessment of cardiac structure and function as previously described in detail (15–17). Using DEXA data, we divided body composition into 3 components: lean mass, fat mass, and bone mass, as previously described (18). Scans were obtained from a bone densitometer in array mode (Delphi W, Hologic, Bedford, Massachusetts) and analyzed with Oasis software (Hologic) to quantify body composition.
Menopausal status was defined by a combination of survey variables, FSH, and adjudication by pelvic MRI when necessary (Fig. 2).Oral estrogen use was ascertained by questionnaire as use of oral contraceptives or hormone-replacement therapy.
Statistical analyses were performed using SAS software (Version 9.1, SAS Corporation, Cary, North Carolina). In univariable analyses, Spearman correlation coefficients were used to assess the association between natriuretic peptides and gonadal steroid hormones and SHBG. Because of their non-normal distributions, NT-proBNP, BNP, FSH, SHBG, and TT were log-transformed when performing multivariable analysis. Multiple linear regression analysis was performed with log BNP or log NT-proBNP as the dependent variables and the independent variables of post-menopausal status, FSH, body mass index (BMI), creatinine, age, race, left ventricular (LV) mass, reduced LV ejection fraction (<55%) and oral estrogen use to develop the baseline model. Subsequent models added log SHBG, log TT, log FAI, and log cFT individually to the baseline model. For the last analysis, standardized estimates to compare the relative effect size of the variables in the predictive model were calculated by standardizing all variables to a mean of 0 and a variance of 1. The Cochran-Armitage trend test was used to compare BNP and NT-proBNP across increasing quartiles of SHBG, TT, and cFT, whereas the Scheffe post hoc test was used to analyze pair-wise comparisons.
The study population comprised 682 women. Baseline characteristics stratified by NT-proBNP above versus below the median value of 37.2 ng/l are shown in Table 1.Women with NT-proBNP concentrations above the median value were older, more often white, had a lower BMI, and were more likely to have left ventricular hypertrophy (LVH). No differences in BNP or NT-proBNP concentrations were observed between pre- and post-menopausal women despite significant differences in age and oral estrogen use. Among post-menopausal women, BNP or NT-proBNP concentrations were similar among oral estrogen users and nonusers (for BNP: median 5.4 interquartile range [0, 15.5] vs. median 2.7 interquartile range [0, 11.4], p = 0.24; for NT-proBNP: median 38.8 interquartile range [22.0, 79.0] vs. median 38.4 interquartile range [18.4, 74.1], p = 0.70). Thus, we were unable to demonstrate an association between BNP or NT-proBNP and estrogens in this cohort.
Univariable associations between hormones and natriuretic peptides
Spearman correlation coefficients between hormone concentrations and natriuretic peptide concentrations are presented in Table 2.Both BNP and NT-proBNP were positively correlated with SHBG and inversely correlated with TT, FAI, and cFT. In contrast, no significant correlations were observed between FSH and LH and either natriuretic peptide measurement. Concentrations of BNP and NT-proBNP increased significantly across quartiles of SHBG and decreased significantly across quartiles of cFT (Fig. 3).These data suggest that androgens are inversely associated with BNP and NT-proBNP.
Multivariable analysis evaluating natriuretic peptides and hormonal status
In multiple linear regression models adjusting for potential confounders, post-menopausal status and FSH were not associated with either log transformed BNP or NT-proBNP (Table 3,Model 1). In contrast, higher levels of SHBG were independently associated with higher BNP and NT-proBNP concentrations (Model 2), whereas higher TT, FAI, and cFT were independently associated with lower BNP and NT-proBNP concentrations (Table 3, Models 3 to 5).
The addition of log SHBG, log TT, log FAI, or log cFT to the multivariable models modified the effect of BMI, such that BMI was no longer significantly associated with BNP or NT-proBNP (Table 4,Models 1 to 4). The modifying effect of free testosterone and SHBG also was seen when lean body mass was substituted for BMI in the models (Table 4, Models 6 to 9). Moreover, comparing the standardized parameter estimates, measurements of free testosterone status contributed more to the model than BMI or lean body mass (Table 4). In summary, these analyses confirm a strong influence of androgens on natriuretic peptide concentrations in young women.
In a population study of young adult women without known cardiac disease, we did not demonstrate an association between estrogen status, as measured by post-menopausal status, oral estrogen use or FSH concentrations, and plasma BNP and NT-proBNP concentrations. In contrast, we observed a positive association between SHBG and natriuretic peptides and inverse associations between all measures of circulating free testosterone (FAI and cFT) and natriuretic peptides even after adjusting for other factors known to influence natriuretic peptide concentrations, including structural and functional cardiac parameters, renal function, and BMI. In addition, we found that after accounting for the free testosterone status or SHBG, the influence of body composition on natriuretic peptides was markedly attenuated.
Are gender differences in natriuretic peptides mediated by androgens rather than estrogens?
Our findings do not support the notion that estrogens are primarily responsible for gender differences in BNP, as has been previously suggested (3,4). Rather, the present findings suggest that free testosterone contributes more to the BNP variability in women than estrogens. Consistent with our observations, testosterone suppressed volume-stimulated release of atrial natriuretic peptide from isolated perfused rat atria (19). The proposed primary role of testosterone can still be consistent with previous reports of an estrogen effect. Since higher circulating estrogens will lower free testosterone (by increasing SHBG), the effect of exogenous estrogen may have been mediated through lowering free testosterone. Similarly, the differences between pubertal boys and girls might represent the “BNP-lowering” effects of rising testosterone production in boys rather then “BNP-raising” effects associated with higher estrogen concentrations at menarche in girls. Men and women differ more by circulating testosterone concentrations than by estrogen concentrations. Total testosterone ranges in men are an order of magnitude larger than those in women, and free testosterone is almost 40 times greater in men than in women. In contrast, estradiol concentrations are usually only 1 to 2 times greater in women than in men (20).
The present findings lead to a rephrasing of the question, from “why do women have higher natriuretic peptides?” to “why do men have lower natriuretic peptides?” Men may have a natriuretic peptide handicap, or impaired natriuretic peptide response, as has been previously proposed with regard to obesity (6,21). Viewing gender differences in BNP in this light, we speculate that modestly higher BNP concentrations in young women may contribute to cardioprotection before menopause. For example, BNP has established antiproliferative and vasodilatory effects and has been shown to act as a counter-regulatory hormone in the renin-angiotensin-aldosterone and adrenergic axes (22). Higher pre-menopausal natriuretic peptide concentrations may in part mediate the relative protection against LVH in young women. The relative decrease in natriuretic peptides with age may help to explain the marked increase in LVH after menopause observed in healthy women from the Framingham study (3,23). Although menopause is traditionally considered a time of falling estrogens, it is also a time in which free testosterone concentrations rise significantly. In a prospective study of women during the menopause transition, the FAI increased with time relative to the final menstrual period by as much as 80% during the course of 6 years (24).
Body composition, natriuretic peptides, and free testosterone status
Previous studies have reported an inverse association between BMI and natriuretic peptides, an observation that was thought to result from increased BNP clearance by the natriuretic peptide clearance receptor in adipose tissue (25). However, we and others have recently reported similar associations for N-terminal-pro-atrial natriuretic peptide and NT-proBNP (6,18), which do not bind the clearance receptor, suggesting that the association between obesity and lower natriuretic peptide concentrations is more likely mediated by reduced natriuretic peptide production and/or release than by enhanced clearance. Moreover, using DEXA measurements of body composition, we discovered that lean mass rather than fat mass primarily accounted for the inverse association between BMI and natriuretic peptides.
We extend these previous findings here with the observation that measures of free testosterone status modified the effect of BMI and lean mass on BNP and NT-proBNP, such that BMI and lean mass were no longer significantly associated with BNP or NT-proBNP after adjusting for measures of free testosterone. These findings suggest that free testosterone, which can increase lean mass and may directly decrease natriuretic peptide synthesis, may be an important mediator of the association between body composition and natriuretic peptides. These results highlight the significance of lean mass, even in obese subjects, and suggest that important confounders or effect-modifiers, including free testosterone, may contribute to the complex relationship described between BMI and cardiovascular disease (26).
SHBG and testosterone measurements in women
SHBG is a transport protein for estrogens, testosterone, and other androgens that has a greater affinity for testosterone than for estradiol, such that SHBG gives a relative indication of the ratio of free estradiol to free testosterone. In women, SHBG is present in large excess; the ratio of SHBG:testosterone is 10 to 100:1. When estradiol increases, steady-state SHBG concentrations increase and subsequently lower the amount of unbound or free testosterone. With higher circulating testosterone, SHBG production is suppressed and even more unbound circulating free testosterone is available. Therefore, in women, SHBG has a strong inverse relationship with free testosterone concentrations such that high SHBG correlates with low free testosterone and low SHBG correlates with high free testosterone (20).
Low concentrations of SHBG have been associated with increased cardiovascular risk in both pre- and post-menopausal women (27,28). In the Women’s Health Study, SHBG was inversely related to C-reactive protein in post-menopausal women who developed clinical cardiovascular disease (29). Our data associate lower SHBG concentrations with lower concentrations of natriuretic peptides. Given the cardioprotective effects of natriuretic peptides, including antiproliferative and vasodilatory effects, as well as antagonism of the renin-angiotensin-aldosterone and adrenergic axes (22), we speculate that lower natriuretic peptide concentrations may explain, in part, the excess cardiovascular risk associated with low SHBG concentrations. Future studies should evaluate whether the same relationships among SHBG, free testosterone, and natriuretic peptide levels persist in older women as well as in men.
Blood samples were not timed to the menstrual cycle, and analyte measurements were not duplicated. We did not perform direct measures of estrogen or free estrogen concentrations, nor did we perform measurements of free testosterone using the traditional equilibrium dialysis method. However, cFT correlates well with free testosterone measured by equilibrium dialysis (13). Because these are cross-sectional data, we were not able to determine the clinical significance of hormone-related differences in BNP and NT-proBNP.
Among a population-based sample of young women, measures of free testosterone were independently and inversely associated with circulating BNP and NT-proBNP, whereas various measures of estrogen status were not. We found that BMI and lean body mass were no longer significantly associated with BNP or NT-proBNP after adjustment for free testosterone measurements. These data suggest that gender differences in natriuretic peptides, and potentially the previously observed inverse relationship of BNP and NT-proBNP with BMI, reflect an inhibitory effect of testosterone in men rather than a stimulatory effect of estradiol in women. Additional investigation is needed to test this hypothesis in older women, in men, and in subjects with normal and overweight BMI. Future studies modeling natriuretic peptide regulation should include a parameter to control for the influence of free testosterone.
↵1 Dr. de Lemos has received grant support from Biosite Inc. and Roche Diagnostics, speaker honoraria from Biosite Inc., and consulting fees from Bayer Diagnostics and Inverness.
↵2 Dr. Auchus is on the speakers’ bureau for Columbia Laboratories.
↵3 Dr. McGuire is on the speakers’ bureau for Pfizer and Takeda.
↵4 Dr. Chang is on the speakers’ bureau for Scios.
Grant support was provided by the Donald W. Reynolds Foundation, by USPHS GCRC grant #M01-RR00633 from NIH/NCRR-CR, and by Roche Diagnostics, Indianapolis, Indiana. Reagents for NT-proBNP were provided by Roche Diagnostics and reagents for BNP by Biosite Inc., San Diego, California.
- Abbreviations and Acronyms
- body mass index
- B-type natriuretic peptide
- calculated free testosterone
- dual energy X-ray absorbtiometry
- free androgen index
- follicle-stimulating hormone
- left ventricular hypertrophy
- magnetic resonance imaging
- N-terminal pro-B-type natriuretic peptide
- sex hormone-binding globulin
- total testosterone
- Received July 27, 2006.
- Revision received September 19, 2006.
- Accepted October 18, 2006.
- American College of Cardiology Foundation
- Abdullah S.M.,
- Khera A.,
- Das S.R.,
- et al.
- Redfield M.M.,
- Rodeheffer R.J.,
- Jacobsen S.J.,
- Mahoney D.W.,
- Bailey K.R.,
- Burnett J.C. Jr.
- Koch A.,
- Singer H.
- Wang T.J.,
- Larson M.G.,
- Levy D.,
- et al.
- Mehra M.R.,
- Uber P.A.,
- Park M.H.,
- et al.
- Barnes S.C.,
- Collinson P.O.,
- Galasko G.,
- Lahiri A.,
- Senior R.
- Chung A.K.,
- Das S.R.,
- Leonard D.,
- et al.
- Das S.R.,
- Drazner M.H.,
- Dries D.L.,
- et al.
- Larsen P.,
- Kronenberg H.M.,
- Melmed S.,
- Polonsky K.S.
- Rexrode K.M.,
- Manson J.E.,
- Lee I.M.,
- et al.
- Sutton-Tyrrell K.,
- Wildman R.P.,
- Matthews K.A.,
- et al.
- Joffe H.V.,
- Ridker P.M.,
- Manson J.E.,
- Cook N.R.,
- Buring J.E.,
- Rexrode K.M.