Circulating Testosterone and SHBG Concentrations Are Heritable in Women: The Framingham Heart Study

Abstract

Context:

Many factors influence the concentration of circulating testosterone and its primary binding protein, SHBG. However, little is known about the genetic contribution to their circulating concentrations in women, and their heritability in women is not well established.

Objective:

Our objective was to estimate the heritability of circulating total testosterone (TT), free testosterone (FT), and SHBG in women in families from the Framingham Heart Study.

Methods:

Women in the Framingham Heart Study who were not pregnant, had not undergone bilateral oophorectomy, and were not using exogenous hormones were eligible for this investigation. TT was measured using liquid chromatography tandem mass spectrometry and SHBG using an immunofluorometric assay (Delfia-Wallac), and FT was calculated. Heritability estimates were calculated using variance-components methods in Sequential Oligogenic Linkage Analysis Routines (SOLAR) and were adjusted for age, age², body mass index (BMI), BMI², diabetes, smoking, and menopausal status. Bivariate analyses were done to assess genetic correlation between TT, FT, and SHBG.

Results:

A total of 2685 women were studied including 868 sister pairs and 688 mother-daughter pairs. Multivariable adjusted heritability estimates were 0.26 ± 0.05 for FT, 0.26 ± 0.05 for TT, and 0.56 ± 0.05 for SHBG (P < 1.0 × 10⁻⁷ for all). TT was genetically correlated with SHBG [genetic correlation coefficient (ρG) = 0.31 ± 0.10] and FT (ρG = 0.54 ± 0.09), whereas SHBG was inversely correlated with FT (ρG = −0.60 ± 0.08).

Conclusion:

Circulating TT, FT, and SHBG concentrations in women are significantly heritable, underscoring the importance of further work to identify the specific genes that contribute significantly to variation in sex steroid concentrations in women. The strong shared genetic component among pairs of TT, FT, and SHBG concentrations suggests potential pleiotropic effects for some of the underlying genes.

Sex steroids are essential for normal sexual differentiation and reproductive health across the lifespan. Sex steroids influence many age-related chronic diseases in women including osteoporosis (1), metabolic syndrome (MetS) (2, 3), type 2 diabetes (T2DM) (2, 4, 5), and cardiovascular disease (CVD) (6, 7). The relationship between sex steroids and chronic disease is affected by changes in hormone profiles in women, particularly with menopause.

Testosterone and its primary binding protein, SHBG, mediate sex-hormone-sensitive phenotypes in women. Higher circulating testosterone has been associated with incident T2DM (4) and CVD (8). Lower SHBG has been associated with MetS and T2DM (4, 9, 10), whereas higher SHBG has been associated with greater risk for hip fracture (1).

Although circulating testosterone and SHBG are affected by many environmental factors including obesity, smoking, insulin resistance, and T2DM (11), they are believed to be at least partially heritable. In men, heritability estimates range from 25–75% for testosterone and 30–50% for SHBG (12–14). In contrast, little is known about the genetic influences of circulating sex hormones in women. One study estimated the heritability of total testosterone at 39% and SHBG at 56% in postmenopausal sisters and twins (15). The purpose of this investigation was to estimate the heritability of circulating total and free testosterone and SHBG in adult women in families from the Framingham Heart Study (FHS), a multigenerational population-based study.

Subjects and Methods

Study population

The FHS design and participants have been described in detail previously (16). In brief, the FHS recruited a population-based sample of men and women residing in Framingham, MA, in 1948 (n = 5209) with the purpose of prospectively studying CVD. In 1971, children of the original cohort and their spouses were recruited as a second generation (the Offspring Study). In 2002, children of the Offspring cohort were recruited as a third generation. Women from Generation 2 and Generation 3 aged 19 and above with sex steroid hormone measurements were eligible for this study. Women taking estrogens, progestins, and androgens or who were pregnant or had undergone bilateral oophorectomy were excluded. All participants signed informed consent before participation in the FHS. The study was approved by the institutional review board at Boston University Medical Center.

All blood samples were collected in the morning, usually between 0730 and 0930 h after participants fasted approximately 10 h overnight. Samples were aliquoted, frozen, and stored at −80 C. Samples used for testosterone and SHBG measurements were not thawed previously.

Assay measurements

Total testosterone was measured by liquid chromatography tandem mass-spectrometry (17). Mass spectrometry was performed using TSQ Thermo-Finnigan Quantum Ultra (Thermo Fisher Scientific, Waltham, MA). The functional limit of detection, defined as the lowest concentration detected with less than 20% coefficient of variation, was 2 ng/dl. Cross-reactivity with other steroids including dehydroepiandrosterone/dehydroepiandrosterone sulfate, androstenedione, and estradiol was negligible. SHBG was measured with a two-site directed immunofluorometric assay with sensitivity of 0.5 nm and less than 0.1% cross-reactivity with other circulating proteins (Delfia-Wallac, Inc., Turku, Finland). Free testosterone was calculated from total testosterone and SHBG using the law of mass action equations (18).

Statistical analysis

Serum hormones were transformed by rank normalization to minimize skew. Covariates considered in the analysis included age, age², body mass index (BMI), BMI², T2DM, smoking, and menopausal status. Covariates were defined as follows: T2DM, fasting blood sugar of at least 126 mg/dl or use of diabetes medications; current smoking, yes/no; and postmenopause, at least 1 yr without menses. SAS version 9.1 (SAS Inc., Cary, NC) was used. SOLAR (Sequential Oligogenic Linkage Analysis Routines) statistical software was used to estimate heritability (19). All family relationships including first-degree relatives (mother-daughter, sister-sister pairs) and extended family relationships (aunt-niece, cousins) were analyzed. Heritability estimates were adjusted for the same covariates across all models. Heritability estimates for free and total testosterone were also adjusted for SHBG to assess for genetic influences independent of SHBG. Bivariate analyses were used to examine the genetic correlation between circulating testosterone and SHBG and were adjusted for the above covariates.

Results

The 2685 women studied included first-degree relatives (868 sister-sister, 688 mother-daughter pairs) and extended family relationships (Table 1). The older, predominantly postmenopausal Generation 2 women had higher BMI (28 ± 6 vs. 26 ± 6 kg/m²) and more T2DM (10 vs. 2%) than the younger, predominantly premenopausal Generation 3. SHBG was higher in Generation 3, whereas free testosterone was slightly higher in Generation 2.

Table 1.

Characteristics of women from the Offspring and Generation 3 cohorts of the FHS

	Offspring Generation 2	Generation 3
Sample (n)	1071	1614
Age (yr)	62 ± 10	41 ± 8
BMI (kg/m2)	28 ± 6	26 ± 6
Postmenopause (%)	80	10
Smoker (%)	12	16
Diabetes (%)	10	2
Total testosterone (ng/dl)	32.1 ± 21.8	27.9 ± 15.3
Free testosterone (pg/ml)	3.6 ± 2.4	2.9 ± 1.9
SHBG (nmol/liter)	74.2 ± 38.8	87.2 ± 46.6

Heritability estimates (Tables 2 and 3)

Table 2.

Heritability of total and free testosterone and SHBG in adult women from the FHS Generations 2 and 3: univariate analysis

	Total testosterone	SHBG	Free testosterone
n	2671	2677	2671
h2 (se)	0.263 (0.053)	0.556 (0.052)	0.259 (0.052)
P value	1.00 × 10−7	7.29 × 10−30	3.25 × 10−8
SHBG-adjusted analysisa
n	2671		2671
h2 (se)	0.246 (0.053)		0.203 (0.052)
P value	3.00 × 10−7		1.24 × 10−5

Covariates included age, age², BMI, BMI², menopausal status, diabetes, and current smoking. h², Heritability estimate.

^aTotal and free testosterone heritability estimates adjusted for circulating SHBG in addition to age, age², BMI, BMI², menopausal status, diabetes, and current smoking.

Table 3.

Heritability of total and free testosterone and SHBG in adult women from the FHS Generations 2 and 3: bivariate analysis

	Total testosterone and SHBG	Total testosterone and free testosterone	SHBG and free testosterone
n	2677	2671	2677
ρG (se)	0.31 (0.10)	0.54 (0.09)	−0.60 (0.08)
P value
Ho: ρG = 0	3.32 × 10−3	1.69 × 10−3	1.93 × 10−8
Ho: ρG = 1	3.00 × 10−7	1.54 × 10−9	1.29 × 10−5
ρE (se)	0.10 (0.06)	0.85 (0.017)	−0.37 (0.053)
P value
Ho: ρE = 0	0.14	2.23 × 10−8	9.78 × 10−44
ρP	0.171	0.769	−0.440

Covariates included age, age², BMI, BMI², menopausal status, diabetes, and current smoking. ρE, Environmental correlation; ρG, genetic correlation; ρP = phenotypic correlation; Ho, null hypothesis.

Total and free testosterone as well as SHBG showed strong heritability after adjusting for age, BMI, T2DM, current smoking, and menopausal status (Table 2). Both total and free testosterone showed heritability estimates of 0.26 (se = 0.05). SHBG heritability was estimated at 0.56 (se = 0.05). Heritability estimates of circulating free and total testosterone remained significant after adjustment for SHBG: free testosterone heritability was 0.20 (se = 0.05) and total testosterone heritability was 0.25 (se = 0.05).

Bivariate analyses (Table 3) showed significant genetic correlation between total testosterone and SHBG [genetic correlation coefficient (ρG) = 0.31; se = 0.10; P = 3.32 × 10⁻³] after adjustment for age, BMI, T2DM, current smoking, and menopausal status. Total and free testosterone were also highly genetically correlated (ρG = 0.54; se = 0.09; P = 1.69 × 10⁻³) as would be expected because free testosterone, a small fraction of total testosterone concentrations, is calculated from total testosterone and SHBG. SHBG was inversely genetically correlated with free testosterone (ρG = −0.60; P = 1.93 × 10⁻⁸), suggesting that specific genes may have opposite effects on their circulating concentrations. Estimates of environmental influences on circulating SHBG and free testosterone were inversely correlated as well (environmental correlation, ρE = −0.37; se = 0.05; P = 2.23 × 10⁻⁸), suggesting that environmental influences may affect SHBG and free testosterone in opposite directions.

The derived phenotypic correlation, ρP, between SHBG and free testosterone is −0.44 (Table 3), which is consistent with the correlation of the two hormones circulating concentrations (Pearson correlation = −0.41, P < 0.0001).

Discussion

We have shown that total and free testosterone and SHBG are moderately to highly heritable (26–56%) in white women of European ancestry in the FHS. Furthermore, circulating total testosterone and SHBG are genetically correlated, suggesting that the two are influenced by common genes or that genetic influences on SHBG may indirectly influence total testosterone given that SHBG is testosterone's primary binding protein. Free testosterone and SHBG are inversely genetically correlated, suggesting that genes may influence them in opposite directions, which is consistent with the inverse correlation of their circulating concentrations. Given the importance of testosterone and SHBG for both reproductive and nonreproductive health in women, further analyses of the genetic influences of circulating testosterone and SHBG are needed to elucidate risk factors for the diseases influenced by these hormones, particularly osteoporosis, MetS, T2DM, and CVD.

Women experience a rapid decline in estrogen production at the time of menopause. In contrast, testosterone declines at a relatively stable rate across the menopause (20). The menopausal transition is associated with changes in metabolism and body fat distribution consistent with the MetS. Furthermore, there is an increase in osteoporosis (1) and CVD (21) after menopause traditionally thought to be due to estrogen deficiency. However, estrogen and estrogen/progestin treatment in clinical trials have not shown a benefit in CVD risk or mortality and have suggested possible increased risk for CVD and stroke (7). The role of testosterone independent of estrogens in osteoporosis, CVD, and metabolic disorders is not clear. Some studies (including the FHS) (22) have shown that higher testosterone levels are associated with increased prevalence of CVD and T2DM (2, 4). Prospective studies have found higher testosterone to be associated with increased risk of T2DM (4) and CVD (8). Postmenopausal women in the WISE (Women's Ischemia Syndrome Evaluation) study being evaluated for ischemia with a history of irregular menses and elevated free testosterone had a higher burden of angiographic coronary artery disease and were more likely to have a myocardial infarction or cardiovascular-related death (23). There are very limited data on the genetic influences of circulating testosterone in otherwise healthy women.

SHBG tightly binds testosterone in women. SHBG production and clearance are affected by many factors including age, adiposity, smoking, hormone use, liver disease, insulin resistance, and T2DM, resulting in variable circulating SHBG levels (11). SHBG was thought to function only as a carrier protein for sex hormones, but it has been proposed that SHBG binds to its own receptor and exerts biological effects as well (24, 25). Lower SHBG has recently been shown to predict incident T2DM in women (9–10). SHBG concentrations are higher in premenopausal women compared with postmenopausal women due to the stimulation of SHBG production by the liver in response to higher estrogen levels. Thus, the lower SHBG levels coincident with lower estrogen levels after menopause may be related to the greater metabolic and cardiovascular risk observed after the menopausal transition.

Our adjusted heritability estimate for SHBG of 56% suggests a significant genetic component and is in the range estimated in the San Antonio Heart Study in men and women and extends the observation in the Australian study of postmenopausal sisters to premenopausal women as well (12, 15). Additionally, SHBG had a strong inverse genetic correlation with free testosterone, suggesting that the two share common genes that influence their circulating levels. The candidate gene approach has been used to identify regions of interest that may influence circulating hormone levels in conjunction with disease states. A SHBG single-nucleotide polymorphism, rs179994, has been associated with T2DM (odds ratio = 0.94; 95% confidence interval = 0.91–0.97; P = 2 × 10⁻⁵) (10). Further study of the genetic factors influencing circulating SHBG levels are warranted to determine whether these genes influence susceptibility to chronic diseases, particularly T2DM and CVD.

The use of mass spectrometry, a state of the art method for measuring testosterone in the low concentrations prevalent in women, is a significant strength of this study. The multigenerational design of the FHS is also a significant strength. Our findings are based on the FHS population, which is Caucasian of European ancestry. Future studies of the genetic influences of circulating sex steroid profiles in women should be conducted in populations of varied racial and ethnic backgrounds.

Summary

Circulating testosterone and SHBG levels are highly heritable in women of white, European ancestry, suggesting strong genetic influences. Further study to elucidate the genetic loci that contribute to the determination of circulating hormone profiles is important given that circulating sex hormone profiles are associated with increased risk for significant chronic disease in women including osteoporosis, T2DM, and CVD. The candidate gene approach is limited by the assumptions underlying the choice of the genes to be analyzed. Genome-wide association scans may be a more fruitful approach for identifying genetic regions that influence sex hormone and SHBG levels that are not encumbered by the bias of a priori assumptions about which genes will be important in mediating their synthesis or action. There is evidence of potential pleiotropic genetic effects with single genes affecting both testosterone and SHBG. Understanding pleiotropic genetic effects on circulating hormone concentrations may have important implications for the development of hormone-targeted therapy for chronic diseases.