Abstract
BACKGROUND:
Circulating concentrations of sex hormone– binding globulin (SHBG) have been associated with cardiovascular diseases, type 2 diabetes, metabolic syndrome, and hormone-dependent cancers; however, correlates of SHBG concentrations are not well understood.
METHODS:
We comprehensively investigated correlates of SHBG concentrations among 13 547 women who participated in the Women’s Health Initiative and who had SHBG measurements. We estimated study- and ethnicity-specific associations of age, reproductive history, usage of exogenous estrogen, body mass index (BMI), and lifestyle factors such as physical activity, smoking, alcohol consumption, coffee intake, and dietary factors with SHBG concentrations. These estimates were pooled using random-effects models. We also examined potential nonlinear associations using spline analyses.
RESULTS:
There was no significant ethnic difference in the age-adjusted mean concentrations of SHBG. Age, exogenous estrogen use, physical activity, and regular coffee intake were positively associated with SHBG concentrations, whereas BMI was inversely associated with SHBG concentrations after adjustment for potential confounding factors. Similar patterns were observed among both ever users and never users of exogenous estrogen. The spline analysis indicated nonlinear relations of regular intake of coffee, age, and BMI with SHBG concentrations. Two or more cups/day of regular coffee consumption and age of 60 years or older were associated with higher SHBG concentrations; the inverse BMI–SHBG relation was especially strong among women whose BMI was below 30.
CONCLUSIONS:
In this large sample of postmenopausal women, age, exogenous estrogen use, physical activity, regular coffee intake, and BMI were significant correlates of SHBG concentrations, presenting potential targets for interventions.
Sex hormone–binding globulin (SHBG)16 is synthesized primarily in the liver and binds with high affinity to androgens and with low affinity to estrogens (1 ). Emerging evidence indicates that SHBG is associated with the development of hormone-dependent cancers (2, 3 ), cardiovascular diseases (4 ), type 2 diabetes (5), and hip fracture (6 ).
Both genetic and environmental factors have been hypothesized to determine SHBG concentrations (7 ). Earlier studies have evaluated the potential environmental correlates of circulating SHBG concentrations (8–14 ). However, sample sizes were small, and the precision was therefore limited. Furthermore, the different factors have often been examined separately; this limits the evaluation of the relative importance of each factor. Therefore, we comprehensively investigated the relation of potential environmental correlates, both separately and jointly, with circulating SHBG concentrations in a large cross-sectional study among participants in the Women’s Health Initiative Clinical Trial (WHI-CT) and the WHI Observational Study (WHI-OS).
Materials and Methods
STUDY POPULATION
The WHI included 161 808 postmenopausal women enrolled between 1993 and 1998. The WHI has 2 major components, the WHI-CT and WHI-OS, conducted at 40 clinical centers in the US. The WHI-CT enrolled 68 132 postmenopausal women aged 50–79 years. The WHI-OS included 93 676 women not participating in the WHI-CT. The institutional review boards at all participating institutions approved the study protocols and procedures. Each participant provided written informed consent to participate in the study. A series of substudies within the WHI with various study designs (e.g., case-control study, case-cohort study, and cross-sectional study) have measured a variety of biomarkers (15 ), and 11 of them have measured plasma SHBG concentrations at baseline (see Table 1 in the Data Supplement that accompanies the online version of this report at http://www.clinchem.org/content/vol60/issue1). In total, 13 547 women with SHBG measurements were included in the current analysis. Because 1023 women were included in multiple substudies, they had multiple SHBG measurements. We compared SHBG concentrations measured using different assays on the same blood sample in these women. The median (interquartile range) intraindividual CV was 6.3% (2.9%–12.3%), suggesting that SHBG concentrations were comparable between the assays. Thus, we randomly selected one measurement for each woman. Women with missing information regarding covariates or exposures were excluded. Because stratification by disease status such as diabetes did not materially alter the estimates, participants were included in the analyses regardless of disease status at baseline. The assignment of hormone therapy in the WHI-CT was not used as a covariate in the main analyses. Sensitivity analyses with covariate adjustment for treatment assignment in the WHI-CT did not materially change the estimates.
Table 1.
Baseline characteristics according to quintiles (Q) of plasma SHBG concentrations in the WHI, 1993–1998.
| Q1 (lowest) (n = 2756) | Q2 (n = 2668) | Q3 (n = 2711) | Q4 (n = 2707) | Q5 (highest) (n = 2705) | P a | |
|---|---|---|---|---|---|---|
| Plasma SHBG concentrations, nmol/Lb | 22.9 (18.7–26.0) | 34.4 (31.8–36.9) | 45.1 (42.1–48.2) | 60.0 (55.8–65.4) | 93.2 (80.0–122.0) | |
| Age, yearsc | 63.3 (7.0) | 65.1 (7.3) | 65.7 (7.3) | 66.5 (7.4) | 65.9 (7.8) | <0.001 |
| Ethnicity, % European American | 67.0 | 68.9 | 70.2 | 71.4 | 61.1 | <0.001 |
| Ethnicity, % African American | 19.3 | 18.3 | 17.7 | 16.5 | 22.5 | |
| Ethnicity, % Hispanic | 8.0 | 7.3 | 6.5 | 7.0 | 8.3 | |
| Ethnicity, % Asian/Pacific Islander | 3.8 | 3.7 | 3.7 | 3.7 | 7.0 | |
| Ethnicity, % Native American | 1.3 | 1.0 | 1.3 | 0.7 | 0.5 | |
| BMI, kg/m2c | 32.3 (6.1) | 30.4 (5.9) | 28.9 (5.9) | 27.1 (5.5) | 25.8 (5.5) | <0.001 |
| Waist circumference, cmc | 97.7 (12.5) | 93.0 (12.9) | 89.0 (12.6) | 84.1 (12.1) | 80.7 (12.2) | <0.001 |
| Alcohol consumption, g/dayb | 0.06 (0.01–1.98) | 0.12 (0.01–3.52) | 0.50 (0.01–4.60) | 0.51 (0.01–6.00) | 0.12 (0.01–2.76) | <0.001 |
| Current smoking, % | 6.9 | 8.0 | 8.7 | 9.7 | 10.7 | <0.001 |
| Physical activity, total MET-h/weekb | 4.5 (0.5–12.1) | 5.3 (1.3–14.0) | 7.5 (1.5–16.3) | 8.3 (2.5–17.8) | 8.8 (2.5–19.2) | <0.001 |
| Past postmenopausal hormone use, % | 27.0 | 30.2 | 29.4 | 31.3 | 40.5 | <0.001 |
| Ever oral contraceptive use, % | 40.1 | 34.9 | 32.2 | 31.5 | 30.6 | <0.001 |
| Age at menopause, yearsc | 46.6 (6.8) | 46.9 (6.7) | 47.0 (6.7) | 47.6 (6.5) | 47.0 (6.5) | 0.009 |
| Time since menopause, yearsc | 16.9 (9.0) | 18.5 (9.3) | 19.0 (9.4) | 19.2 (9.1) | 19.1 (9.4) | <0.001 |
| Age at menarche, ≥13years old, % | 50.6 | 49.9 | 51.9 | 54.4 | 56.2 | <0.001 |
| Age at first pregnancy of ≥6 months, <20 years, % | 20.9 | 20.5 | 17.7 | 16.8 | 17.9 | <0.001 |
| History of hypertension, % | 49.0 | 42.2 | 39.1 | 34.8 | 33.1 | <0.001 |
| Pregnancies, ≥5, % | 20.9 | 20.8 | 19.6 | 17.5 | 16.9 | <0.001 |
| Marital status,% currently married | 56.5 | 55.1 | 51.8 | 52.7 | 51.3 | <0.001 |
| Education, % college graduates | 30.5 | 30.7 | 31.2 | 35.6 | 36.0 | <0.001 |
| Total estradiol, pg/mLb | 12.6 (8.8–18.1) (n = 2730) | 11.4 (8.0–16.5) (n = 2625) | 10.3 (7.3–15.0) (n = 2666) | 9.4 (6.8–14.0) (n = 2641) | 10.1 (6.8–21.0) (n = 2584) | <0.001 |
| Total testosterone, ng/dLb | 18.3 (12.1–25.5) (n = 1464) | 20.2 (13.1–28.4) (n = 1414) | 20.5 (13.2–29.8) (n = 1530) | 21.2 (13.4–31.5) (n = 1575) | 16.0 (0.1–28.0) (n = 1803) | <0.001 |
Baseline characteristics were compared between groups using ANOVA for continuous variables or chi-squared test for categorical variable.
Values are medians (interquartile range).
Values are means (SD).
MEASUREMENTS OF SHBG CONCENTRATIONS
Blood samples were collected at baseline after at least a 12-h fast and stored at −80 °C to −70 °C. Plasma SHBG concentrations were measured using an electrochemiluminescence immunoassay (Roche Diagnostics), a solid-phase 2-site chemiluminescent immunoassay (Diagnostic Products, Siemens Healthcare Diagnostics, Siemens Medical Solutions Diagnostics, or Siemens Medical Solutions Diagnostics), or an immunoradiometric assay (Esoterix Laboratory Services) (see online Supplemental Table 1). The inter- and intraassay CVs ranged from 3.7% to 17.7% and 2.5% to 7.7%, respectively.
MEASUREMENTS OF DEMOGRAPHIC FACTORS, REPRODUCTIVE HISTORY, AND LIFESTYLE FACTORS
Height was measured by use of a wall-mounted stadiometer, and weight was measured with the women dressed in light clothing. The body mass index (BMI) was calculated as the weight in kilograms divided by the height in meters squared. Baseline information on medical history, marital status, and reproductive history was collected via several self-administered questionnaires. To assess exogenous estrogen and oral contraceptive use, a detailed interview was conducted at baseline.
As previously reported (16, 17 ), participants completed a standardized, self-administered food frequency questionnaire developed for the WHI to estimate average daily dietary intake over the previous 3-month period. This food frequency questionnaire has demonstrated reasonably good validity as a measurement of dietary intake compared with 24-h dietary recall interviews and food records (18 ). Alcohol intake (g/day) was calculated from the intake of beer, wine, and liquor, as well as foods. The validated physical activity questionnaire collected data on the frequency, duration, and pace of recreational walking and frequency and duration of other recreational activities or exercises (mild, moderate, and strenuous) and household and yard activities (19 ). Metabolic equivalent–h per week (MET-h/week) were calculated from the questionnaire.
STATISTICAL ANALYSIS
Biomarker concentrations were log-transformed to improve normality assumptions in all analyses. First, we compared baseline characteristics across quintiles of the SHBG concentrations using ANOVA for continuous variables or χ2 test for categorical variables (Table 1). Then, we cross-sectionally examined the association of each lifestyle factor separately with SHBG concentrations (Tables 2 through 4 and online Supplemental Table 2). The lifestyle factors examined were physical activity, alcohol intake, cigarette smoking, and intake of coffee, decaffeinated coffee, tea, and dietary factors. Dietary factors examined were consumptions of red meat (beef, pork, or lamb), fish, eggs, soy, fruits, vegetables, whole grains, refined grains, potatoes, and sweets. After categorizing lifestyle factors into 5 groups, we calculated geometric means and SEs of SHBG concentrations according to lifestyle factor categories. We categorized physical activity and alcohol consumption into quintiles, and we grouped other variables into meaningful categories to maintain a gradient of exposure.
Table 2.
Adjusted geometric mean plasma concentrations of SHBG by physical activity, alcohol, and smoking in the WHI, 1993–1998.
| SHBG concentrations, nmol/L | P for linear trenda | P for quadratic trenda | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| (Lowest) | 95% CI | Geometric mean | 95% CI | Geometric mean | 95% CI | Geometric mean | 95% CI | (Highest) | 95% CI | |||
| Geometric mean | Geometric mean | |||||||||||
| Physical activity | Q1 | Q2 | Q3 | Q4 | Q5 | |||||||
| MET-h/week, median (n) | 0 (2544) | 2.5 (2540) | 7.0 (2665) | 13.7 (2409) | 27.8 (2538) | |||||||
| MET-h/week, range | 0–0.5 | 0.8–4.2 | 4.2–10.0 | 10.0–18.8 | 18.8–134.2 | |||||||
| Model 1b | 46.9 | 42.2–52.1 | 44.7 | 41.5–48.1 | 46.7 | 42.5–51.4 | 47.8 | 43.6–52.5 | 49.3 | 45.3–53.6 | <0.001 | 0.95 |
| Model 2c | 46.6 | 41.8–52.0 | 44.9 | 41.6–48.4 | 46.7 | 42.4–51.5 | 47.7 | 43.2–52.6 | 49.1 | 45.1–53.5 | <0.001 | 0.95 |
| Alcohol | Q1 | Q2 | Q3 | Q4 | Q5 | |||||||
| Median (g/day) (n) | 0.00 (2,708) | 0.02 (2699) | 0.13 (2701) | 1.99 (2703) | 13.9 (2702) | |||||||
| Range | 0.00–0.01 | 0.01–0.03 | 0.03–0.99 | 0.99–6.49 | 6.49–191.9 | |||||||
| Model 1b | 45.8 | 41.8–50.2 | 46.5 | 42.2–51.3 | 47.3 | 43.1–51.8 | 48.5 | 44.2–53.3 | 46.5 | 43.0–50.3 | 0.33 | 0.04 |
| Model 2c | 45.8 | 41.5–50.6 | 46.4 | 41.9–51.3 | 47.2 | 42.9–52.0 | 48.3 | 44.0–53.1 | 46.4 | 43.0–50.0 | 0.27 | 0.04 |
| Years of smoking (n) | Never (6997) | <10 years (1301) | 10–29 years (2545) | 30–49 years (2048) | ≥50 years (301) | |||||||
| Model 1b | 47.0 | 42.9–51.5 | 47.8 | 44.3–51.5 | 47.5 | 43.2–52.2 | 46.4 | 42.6–50.5 | 45.6 | 41.5–50.2 | 0.29 | 0.48 |
| Model 2c | 47.0 | 42.8–51.6 | 48.0 | 44.1–52.1 | 47.1 | 42.7–52.0 | 46.5 | 42.6–50.7 | 46.3 | 41.6–51.6 | 0.35 | 0.67 |
| Cigarettes per day (n) | Never or past (4918) | <5 (196) | 5–24 (786) | 25–34 (127) | ≥35 (63) | |||||||
| Model 1b | 46.5 | 42.6–50.8 | 44.8 | 41.6–48.3 | 49.1 | 45.2–53.3 | 50.8 | 45.7–56.6 | 41.8 | 37.1–47.2 | 0.002 | 0.001 |
| Model 2c | 46.5 | 42.5–50.8 | 45.5 | 42.2–49.0 | 49.6 | 45.1–54.4 | 50.4 | 46.2–55.0 | 47.6 | 45.3–50.0 | 0.002 | 0.002 |
Based on median values in categories of the participants with the exception of smoking. For increasing categories of years of smoking, 0, 5, 20, 40, and 60 years of smoking were assigned, and for increasing categories of cigarettes per day, 0, 1, 15, 30, and 45 cigarettes per day were assigned to compute P values for trend.
Model 1, geometric mean and CI of SHBG concentrations and P values for linear trend and quadratic trend were estimated using a random-effects model with adjustment for age (continuous), ethnicity (categorical), BMI (continuous), cigarette smoking (never, past, or current), alcohol consumption (none/past drinkers, <1 drink/week, or ≥1 drink/week), physical activity (MET-h/week) (continuous), and total energy intake (continuous) that were not examined as the primary independent variable.
Model 2, further adjusted for reproductive history: use of exogenous estrogen, use of oral contraceptives, age of menarche (<13 vs ≥13 years), age of first pregnancy (<20 vs ≥20 years), parity (<5 vs ≥5 live births), and marital status (currently married vs not married).
Table 4.
Adjusted geometric mean plasma concentrations of SHBG by categories of red meat, fish, egg, dairy, and soy intakes in the WHI, 1993–1998.
| SHBG concentrations, nmol/L | P for linear trenda | P for quadratic trenda | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| (Lowest) | 95% CI | Geometric mean | 95% CI | Geometric mean | 95% CI | Geometric mean | 95% CI | (Highest) | 95% CI | |||
| Geometric mean | Geometric mean | |||||||||||
| Red meat | ||||||||||||
| Median, servings/week (n) | 0 (1974) | 0.5 (7588) | 2.0 (2027) | 3.5 (1718) | 7.6 (206) | |||||||
| Range, servings/week | 0 | <1 | 1–2 | 2–7 | ≥7 | |||||||
| Model 2b | 48.2 | 44.6–53.4 | 47.4 | 43.3–51.9 | 46.2 | 42.4–50.4 | 44.7 | 40.3–49.6 | 46.6 | 39.3–55.2 | 0.04 | 0.03 |
| Fish | ||||||||||||
| Median, servings/week (n) | 0 (725) | 0.6 (4748) | 1.5 (3825) | 3.0 (3925) | 8.5 (290) | |||||||
| Range, servings/week | 0 | <1 | 1–2 | 2–7 | ≥7 | |||||||
| Model 2b | 46.9 | 42.3–51.8 | 47.1 | 43.2–51.4 | 47.2 | 42.6–52.2 | 46.9 | 42.8–51.3 | 48.4 | 43.1–54.5 | 0.44 | 0.14 |
| Egg | ||||||||||||
| Median, servings/week (n) | 0 (2368) | 0.5 (7919) | 2.0 (2042) | 3.4 (1074) | 3.5 (110) | |||||||
| Range, servings/week | 0 | <1 | 1–2 | 2–7 | ≥7 | |||||||
| Model 2b | 47.3 | 42.7–52.3 | 47.2 | 43.3–51.5 | 47.0 | 42.2–52.4 | 45.9 | 41.5–50.7 | 44.9 | 38.9–51.8 | 0.03 | 0.57 |
| Dairy | ||||||||||||
| Median, servings/week (n) | 0.5 (419) | 1.6 (590) | 4.6 (3750) | 11.9 (6803) | 26.7 (1951) | |||||||
| Range, servings/week | <1 | 1–2 | 2–7 | 7–21 | ≥21 | |||||||
| Model 2b | 51.2 | 44.3–59.2 | 48.2 | 43.0–53.9 | 47.1 | 42.6–52.2 | 47.3 | 43.5–51.3 | 45.5 | 41.3–50.0 | 0.11 | 0.56 |
| Soy | ||||||||||||
| Median, servings/week (n) | 0 (11 397) | 0.3 (1526) | 2.0 (247) | 3.5 (268) | 7.6 (75) | |||||||
| Range, servings/week | 0 | <1 | 1–2 | 2–7 | ≥7 | |||||||
| Model 2b | 46.8 | 43.0–50.9 | 47.4 | 42.7–52.6 | 49.4 | 43.0–56.8 | 53.4 | 46.2–61.7 | 47.9 | 39.2–58.5 | 0.007 | 0.18 |
Based on median values in categories of the participants.
Model 2, geometric mean and CI of SHBG concentrations and P values for linear trend and quadratic trend were estimated using the random-effects model with adjustment for age (continuous), ethnicity (categorical), BMI (continuous), cigarette smoking (never, past, or current), alcohol consumption (none/past drinkers, <1 drink/week, or ≥1 drink/week), physical activity (metabolic equivalent (MET)-hours per week) (continuous), total energy intake (continuous), and reproductive history: use of exogenous estrogen, use of oral contraceptives, age of menarche (<13 vs ≥13 years), age of first pregnancy (<20 vs ≥20 years), parity (<5 vs ≥5 live births), and marital status (currently married vs not married).
Two-sided P values for linear trend were computed by assigning a category score for each category and including the variables as continuous variables in the models. We also computed 2-sided P values for quadratic trends by further including a quadratic term in each linear trend model. In model 1, the study-specific geometric means and SEs of SHBG concentrations were estimated using a generalized linear model with adjustment for age (continuous), ethnicity (categorical), BMI (continuous), cigarette smoking (never, past, or current), alcohol consumption (non-/past drinkers, <1 drink/week, or ≥1 drink/week), physical activity (MET-h/week) (continuous), and total energy intake (continuous). Then, we pooled estimates using the DerSimonian and Laird random-effects model (20 ). In model 2, we further adjusted for reproductive history: hormone use, oral contraceptive use, age of menarche (categorical, ≥13 years old or not), age of first pregnancy (categorical, ≥20 years old or not), parity (categorical, ≥5 or not), and marital status (categorical, currently married or not). To examine a possible bias introduced by pooling studies with different study designs, we further examined the associations after excluding participants who served as cases in the studies included in the analysis.
To jointly examine the associations between age, BMI, and lifestyle factors and SHBG concentrations, we further included these factors in the model simultaneously (model 3, Table 5) (n = 8712). We estimated study- and ethnicity-specific means and SEs of log-transformed SHBG concentrations per 5-year increment in age, BMI unit, 7.5 MET-h/week for physical activity (e.g., brisk walking for 30 min, 5 days per week), 10 cigarettes smoked/day, 10 g alcohol ingested/day, 1 serving/day for each dietary factor with adjustment for total energy intake (continuous), and reproductive history [use of exogenous estrogen, use of oral contraceptives, age of menarche (<13 vs ≥13 years), age of first pregnancy (<20 vs ≥20 years), parity (<5 vs ≥5 live births), and marital status (currently married vs not married)]. Then we pooled the results using the DerSimonian and Laird random-effects model. We also examined the associations between those factors and SHBG concentrations according to ethnicity. Heterogeneity between studies and ethnicity was evaluated using the χ2 test (21 ). We further stratified studies according to SHBG assay types and study outcomes (cardiovascular diseases or type 2 diabetes, cancer, fractures, and others). Additionally, because sex hormones are also considered to regulate SHBG synthesis in the liver (22 ), we further stratified according to exogenous estrogen use. To provide a visual representation of the dose–response curve, we fitted unrestricted quadratic spline models by including transformed variables of each factor to the regression models (with knots at the 20th, 40th, 60th, and 80th percentiles) (23 ).
Table 5.
Adjusted mean difference in log-transformed plasma SHBG concentrations in relation to age, BMI, lifestyle factors, reproductive history, and use of exogenous estrogen in the WHI, 1993–1998.
| Variable | Difference in log-SHBG concentrations (n = 8712) | |
|---|---|---|
| Pooled meana | 95% CIa | |
| Age, per 5-year increment | 0.037 | 0.026–0.047 |
| BMI, per unit | −0.028 | −0.033 to −0.024 |
| Physical activity, per 7.5 MET-h/week | 0.034 | 0.012 to 0.056 |
| Alcohol, per 10 g/day | −0.0003 | −0.0008 to 0.0003 |
| Cigarettes, per 10/day | −0.010 | −0.021 to 0.000 |
| Regular coffee, per cup/day | 0.025 | 0.018 to 0.032 |
| Red meat, per serving/day | −0.077 | −0.171 to 0.016 |
| Fish, per serving/day | 0.026 | −0.032 to 0.085 |
| Eggs, per serving/day | 0.000 | −0.201 to 0.202 |
| Dairy, per serving/day | −0.016 | −0.036 to 0.003 |
| Soy, per serving/day | −0.013 | −0.018 to 0.082 |
| Fruits, per serving/day | 0.011 | −0.021 to 0.043 |
| Vegetables, per serving/day | 0.002 | −0.010 to 0.013 |
| Whole grains, per serving/day | 0.001 | −0.014 to 0.017 |
| Refined grains, per serving/day | −0.009 | −0.017 to 0.000 |
| Potatoes, per serving/day | −0.015 | −0.053 to 0.024 |
| Sweets, per serving/day | 0.013 | −0.002 to 0.027 |
| Oral contraceptive use, ever vs never | −0.013 | −0.037 to 0.011 |
| Exogenous estrogen use, current vs past/never | 0.225 | 0.043 to 0.408 |
| Exogenous estrogen use, ever vs never | 0.083 | 0.012 to 0.153 |
| Years since menopause, per 5-year increment | 0.001 | −0.007 to 0.010 |
| Menarche age, ≥13 vs <13 years | −0.000 | −0.021 to 0.021 |
| Number of parities, ≥5 vs <5 | −0.028 | −0.062 to 0.006 |
| Marital status, currently married vs not married | −0.033 | −0.062 to −0.003 |
Random-effects models were used to estimate adjusted means and CIs of log-transformed SHBG concentrations per 5-year age increment, BMI unit, 7.5 MET-h/week for physical activity, 10 cigarettes smoked/day, 10 grams of alcohol ingested/day, and 1 serving/day for dietary factors and according to exogenous estrogen use, oral contraceptive use, and reproductive history (years since menopause, menarche age, number of parities, and marital status) with adjustment for age (continuous), BMI (continuous), physical activity [MET-h/week (continuous)], cigarettes smoked per day (continuous), alcohol ingestion per day [in grams (continuous)], total energy intake (continuous), and all dietary factors and indices of reproductive history listed in the table that were not examined as the primary independent variable.
Finally, we estimated pooled means and SEs of log-transformed SHBG concentrations according to exogenous estrogen use, oral contraceptive use, years since menopause (per 5-year interval), menarche age (≥13 vs <13 years), number of parities (≥5 vs <5), and marital status (currently married vs not married) using the same model (model 3, Table 5) (n = 8712). We conducted statistical analyses using SAS (version 9.3; SAS institute) and Stata (version 11.2; StataCorp).
Results
BASELINE CHARACTERISTICS
The baseline characteristics according to quintiles of SHBG concentrations are presented in Table 1. Compared to women with lower SHBG concentrations, women with higher SHBG concentrations tended to be older, have low adiposity measures, be current smokers, engage in physical activity, and have higher education levels, but less likely to have a history of hypertension. Furthermore, they tended to be past users of postmenopausal hormones and never users of oral contraceptives, had more years since menopause, and underwent menarche at an older age. African-Americans or Asians/Pacific Islanders appeared to have higher SHBG concentrations in the crude model (Table 1). However, there was no significant ethnic difference in the age-adjusted geometric mean concentrations of SHBG (P for heterogeneity = 0.35; European Americans, mean, 46.4 nmol/L, 95% CI, 43.0–50.2; African-Americans, mean, 45.0, 95% CI, 39.6–51.3; Hispanics, mean, 40.4, 95% CI, 33.7–48.4; Asians/Pacific Islanders, mean, 50.4, 95% CI, 42.6–59.6) (not shown).
PHYSICAL ACTIVITY
Physical activity was positively associated with SHBG concentrations. In model 2, compared with women in the lowest quintile of physical activity, women in the highest quintile had 5.4% higher SHBG concentrations (P for linear trend <0.001; Table 2). In model 3, the pooled adjusted mean difference in log-SHBG concentrations per increment of 7.5 MET-h/week was 0.034 (log-transformed SD, 0.061; 95% CI, 0.012–0.056; Table 5). Physical activity was positively associated with SHBG concentrations regardless of exogenous estrogen use (not shown). Further excluding the case subjects resulted in similar findings (not shown). Stratification according to the study outcome or SHBG assay type did not explain the heterogeneity across studies (not shown).
ALCOHOL CONSUMPTION
After multivariable adjustment, we did not observe a clear linear association between alcohol consumption and SHBG concentrations (P for linear trend = 0.27), but a nonlinear association was suggested in models 1 and 2 (P for quadratic trend = 0.04; Table 2). However, model 3 (Table 5) and spline curves (not shown) did not indicate any clear alcohol–SHBG association.
CIGARETTE SMOKING
Although we did not observe any significant association between years of smoking and SHBG concentrations, a weak positive association between number of cigarettes smoked per day and SHBG concentrations was suggested in models 1 and 2 (Table 2). However, in model 3, no apparent association was observed (Table 5). Moreover, spline curves for number of cigarettes smoked indicated no clear association (not shown).
REGULAR COFFEE, DECAFFEINATED COFFEE, AND TEA CONSUMPTION
Higher regular coffee intake, but not decaffeinated coffee or tea intake, was associated with higher SHBG concentrations. In model 2, compared with no cups of coffee per day, intake of 6 or more cups of regular coffee per day was associated with 12.0% higher SHBG concentrations (P for linear trend <0.001; P for quadratic trend = 0.002; Table 3). Further excluding the case subjects did not materially change the results (not shown). Model 3 also indicated that regular coffee intake was moderately and positively associated with SHBG concentrations [log-SHBG concentrations per cup/day = 0.025 (SD, 0.045); 95% CI, 0.018–0.032; Table 5]. Similar patterns were observed among both ever users and never users of exogenous estrogen (not shown). We did not observe any substantial heterogeneity across the studies or ethnicities (not shown). Spline curves suggested that heavy drinkers of regular coffee (>2 cups/day) had higher concentrations of plasma SHBG (Fig. 1C).
Table 3.
Adjusted geometric mean plasma concentrations of SHBG by regular coffee, decaffeinated coffee, and tea intakes in the WHI, 1993–1998.
| SHBG concentrations, nmol/L | P for linear trenda | P for quadratic trenda | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| (Lowest) | 95% CI | Geometric mean | 95% CI | Geometric mean | 95% CI | Geometric mean | 95% CI | (Highest) | 95% CI | |||
| Geometric mean | Geometric mean | |||||||||||
| Regular coffee, n | 1646 | 2135 | 4175 | 1245 | 393 | |||||||
| Range, cups/day, n | 0 | 1 | 2–3 | 4–5 | ≥6 | |||||||
| Model 1b | 47.7 | 43.3–52.6 | 46.2 | 41.9–50.9 | 47.2 | 43.7–51.0 | 50.4 | 46.3–54.8 | 52.8 | 47.7–58.5 | <0.001 | 0.002 |
| Model 2c | 47.4 | 43.0–52.3 | 45.9 | 41.5–50.8 | 47.3 | 43.6–51.3 | 50.9 | 46.6–55.5 | 53.1 | 47.8–58.9 | <0.001 | 0.002 |
| Decaffeinated coffee, n | 2215 | 859 | 645 | 121 | 22 | |||||||
| Range, cups/day, n | 0 | 1 | 2–3 | 4–5 | ≥6 | |||||||
| Model 1b | 51.5 | 44.3–60.0 | 50.9 | 43.2–60.1 | 52.0 | 45.2–59.9 | 49.4 | 43.4–56.3 | 51.6 | 41.4–64.3 | 0.62 | 0.82 |
| Model 2c | 51.6 | 44.0–60.4 | 50.8 | 42.5–60.6 | 51.8 | 44.9–59.7 | 49.3 | 43.4–55.9 | 53.2 | 42.4–66.7 | 0.64 | 0.57 |
| Tea, n | 4444 | 706 | 581 | 98 | 28 | |||||||
| Range, cups/day, n | 0 | 1 | 2–3 | 4–5 | ≥6 | |||||||
| Model 1b | 51.1 | 43.5–59.9 | 49.5 | 43.9–55.8 | 50.2 | 44.6–56.5 | 52.6 | 42.7–64.8 | 51.1 | 41.2–63.3 | 0.70 | 0.47 |
| Model 2c | 50.9 | 43.2–60.0 | 49.9 | 43.5–57.2 | 50.2 | 44.2–56.9 | 54.6 | 42.6–70.1 | 51.5 | 41.8–63.3 | 0.46 | 0.44 |
For increasing categories, 0, 1, 2.5, 4.5, and 6 cups/day were assigned to compute P values for trend.
Model 1, geometric mean and CI of SHBG concentrations and P values for linear trend and quadratic trend were estimated using the random-effects model with adjustment for age (continuous), ethnicity (categorical), BMI (continuous), cigarette smoking (never, past, or current), alcohol consumption (none/past drinkers, <1 drink/week, or ≥1 drink/week), physical activity (MET-h/week) (continuous), and total energy intake (continuous).
Model 2, further adjusted for reproductive history: use of exogenous estrogen, use of oral contraceptives, age of menarche (<13 vs ≥13 years), age of first pregnancy (<20 vs ≥20 years), parity (<5 vs ≥5 live births), and marital status (currently married vs not married).
Fig. 1. Estimated plasma SHBG concentrations using spline analyses according to age (A), BMI (B), and regular coffee consumption (C).
Unrestricted quadratic spline models with the inclusion of transformed variables for each factor in the regression models (with knots at the 20th, 40th, 60th, and 80th percentiles) with adjustment for age (continuous), BMI (continuous), physical activity [MET-h/week (continuous)], cigarettes smoked per day (continuous), alcohol ingestion per day [in grams (continuous)], total energy intake (continuous), and all dietary factors and indices of reproductive history included in Table 5 that were not examined as the primary independent variable were used to plot curves (solid curve, point estimates; dashed curves, point-wise 95% confidence limits).
RED MEAT, FISH, EGG, DAIRY, AND SOY INTAKES
In model 2, red meat and egg intakes appeared to be inversely associated with SHBG concentrations (Table 4), while after simultaneous adjustment for other lifestyle factors, we did not find any apparent association in model 3 (Table 5) or spline curves (not shown). In addition, a weak positive association between soy intake and SHBG concentrations was suggested in model 2 (Table 4), but model 3 (Table 5) and a spline model (not shown) did not indicate any clear soy–SHBG association. Fish and dairy intakes were not associated with SHBG concentrations (Tables 4 and 5).
FRUIT, VEGETABLE, GRAIN, POTATO, AND SWEET INTAKES
We did not observe any significant association of fruit, vegetable, whole grain, refined grain, or potato intake with SHBG concentrations (Table 5 and online Supplemental Table 2). A moderate positive association between sweets and SHBG concentrations was suggested in model 2 (see online Supplemental Table 2), but we observed no clear association in model 3 (Table 5) or a spline model (not shown).
AGE
In model 3, age was moderately and positively associated with SHBG concentrations; log-SHBG concentration per 5-year increment in age was 0.037 (SD, 0.066; 95% CI in log-SHBG concentrations, 0.026–0.047; Table 5). The association was observed regardless of the use of exogenous estrogen. No substantial heterogeneity was seen across studies or ethnicities (not shown). Spline curves indicated that age of 60 years or older was associated with higher SHBG concentrations (Fig. 1A).
BODY MASS INDEX
In model 3, BMI was inversely associated with SHBG concentrations [log-SHBG concentrations per 1-U increase in BMI, −0.028 (SD, −0.050); 95% CI, −0.033 to −0.024; Table 5]. Similar patterns were observed for the association between waist circumference and SHBG concentrations (not shown). The BMI–SHBG association was observed among both ever users and never users of exogenous estrogen (not shown). Heterogeneity was suggested across studies, but stratification according to SHBG assay type or study outcomes did not explain the heterogeneity across studies (not shown). Spline curves indicated an inverse association of BMI with SHBG concentrations, especially at BMIs below 30 (Fig. 1B).
EXOGENOUS ESTROGEN USE AND REPRODUCTIVE HISTORY
The use of exogenous estrogen was positively associated with SHBG concentrations. Current estrogen users had higher SHBG concentrations than past or never users of exogenous estrogen [log-SHBG, 0.225 (SD, 0.401); 95% CI, 0.043–0.408; Table 5]. Heterogeneity was suggested across studies, but stratification according to SHBG assay type or according to study outcomes did not explain the heterogeneity (not shown). In addition, compared with never users of exogenous estrogen, ever users had higher SHBG concentrations [log-SHBG, 0.083 (SD, 0.149); 95% CI, 0.012–0.153; Table 5]. In addition, compared with women who were never or previously married, women who were currently married had lower concentrations of SHBG [log-SHBG, −0.033 (SD, −0.059), 95% CI, −0.062 to −0.003; Table 5].
Discussion
In this large cross-sectional study of postmenopausal women, positive associations with SHBG concentrations were observed for age, use of exogenous estrogen, physical activity, and regular coffee intake, whereas BMI was inversely associated with SHBG concentrations. Our comprehensive assessment supports the notion that in addition to sex hormones, age, adiposity, physical activity, and regular coffee intake may be associated with SHBG concentrations.
Regular coffee intake, but not decaffeinated coffee or tea intake, was positively associated with SHBG concentrations. These findings are in line with the findings of earlier studies (11, 24 ). One recent study found that caffeinated coffee and caffeine intakes, but not decaffeinated coffee intake, were positively associated with SHBG concentrations (11 ). Because few decaffeinated coffee or tea drinkers were included in our study, we lacked statistical power to detect an association with SHBG concentrations. The molecular mechanisms underlying the relation between coffee and SHBG are largely unknown. Caffeine and other major components of coffee (cafestol and kahweol) alter liver enzyme expression and activity (25 ). Because SHBG is synthesized and metabolized primarily in the liver (1 ), coffee intake may affect SHBG metabolism in the liver to influence SHBG concentrations (24 ). Importantly, caffeine increases the resting metabolic rate (26 ), and SHBG production is regulated by the metabolic state of the liver via hepatocyte nuclear factor-4α (27 ).
The positive association between age and SHBG concentrations observed in this study is in line with earlier findings (8, 28 ); several mechanisms have been proposed to explain this association. The age-related decrease in growth hormone and insulin-like growth factor 1 concentrations may be responsible for the apparent age-associated increase in SHBG concentrations. In addition, the age-related decline in insulin secretion or insulin sensitivity may result in increased SHBG concentrations. The age-related weight change may also explain the age-associated increase in SHBG concentrations (8 ). However, there is as yet no conclusive mechanistic understanding of the apparent positive age–SHBG association.
Our findings indicated that exogenous estrogen use was positively associated with SHBG concentrations. This observation is consistent with the findings from clinical studies that exogenous estrogen administration increases SHBG concentrations (29–31 ). Our results did not indicate a relation between oral contraceptive use and SHBG concentrations among postmenopausal women, suggesting that past use of oral contraceptives may have little impact on SHBG concentrations later in life, whereas exogenous estrogen use may influence the regulation of SHBG in postmenopausal women.
Additionally, our findings indicated that currently married women had lower SHBG concentrations than single women, whereas an earlier study reported that married nulliparous postmenopausal women had higher SHBG concentrations than single nulliparous postmenopausal women (14 ). Because few nulliparous women were included in our study, future work is needed to better understand this relation. Consistent with previous studies, we observed little or no association between number of parities or age of menarche and SHBG concentrations (14 ). Number of parities and age of menarche have been associated with SHBG concentrations among premenopausal but not among postmenopausal women. Although previous research reported that the number of years since menopause was positively associated with SHBG concentrations (14 ), we observed no association between years since menopause and SHBG concentrations after multivariable adjustment for potential confounding factors including age. However, this finding is not surprising in light of the fact that age and years since menopause are highly correlated (r = 0.71 in our study). In fact, we observed a positive association between age and SHBG concentrations even after adjustment for the number of years since menopause. This finding suggests that age, but not years since menopause, may be an important predictor of SHBG concentrations.
The positive association between physical activity and SHBG concentrations is consistent with earlier reports (32, 33 ). In a post hoc analysis of the Estrogen in the Prevention of Atherosclerosis Trial examining the effect of estradiol vs placebo on atherosclerosis among postmenopausal women, the mean levels (averaged over the trial) of physical activity were positively associated with mean SHBG concentrations (32 ). Also, a randomized clinical trial of 170 sedentary, overweight, postmenopausal women showed a greater increase in SHBG concentrations among exercisers compared with a control group (33 ). However, little is known about the mechanisms underlying the possible effect of exercise on SHBG metabolism. Improved insulin sensitivity or loss of adiposity induced by physical activity may influence SHBG production in the liver.
The inverse association between BMI and SHBG concentrations observed in this study is consistent with previous studies (34–36 ). The Study of Women’s Health Across Nation reported inverse associations of BMI, waist circumference (35 ), and visceral fat (36 ) with SHBG concentrations in premenopausal and early perimenopausal women. The mechanisms responsible for the BMI-SHBG association are not well understood. Decreased SHBG concentrations have been associated with insulin resistance (37 ). Adiposity is also known to be associated with higher blood insulin concentrations resulting from insulin resistance, and insulin has been reported to inhibit SHBG production in hepatoma cells (38 ). These findings may suggest that compensatory hyperinsulinemia in women with higher BMI levels can lead to lower SHBG concentrations. Alternatively, insulin resistance itself might play a major role in the regulation of SHBG concentrations.
No clear association was seen between soy intake and SHBG concentrations in this study. Our results are consistent with findings from a recent metaanalysis, suggesting that soy or isoflavone intake may not play a major role in SHBG metabolism (39 ).
The strengths of our study include its large sample size and multiethnic study design, with detailed measurement of variables. However, several limitations of this study should be recognized. First, the major limitation is its inability to establish a temporal relationship between exposure (e.g., lifestyle factors) and SHBG concentrations, although SHBG concentrations are unlikely to alter exposure. Second, the studies pooled in the current investigation differed in their SHBG measurement methods, the laboratories where the measurements were performed, and the study outcomes that were selected, which may have introduced sources of heterogeneity. Third, although SHBG is produced by the liver, we were unable to examine the relation between liver function and SHBG concentrations. Fourth, we cannot exclude the possibility of residual confounding arising from incompletely measured or unmeasured covariates. Finally, biases due to measurement errors in exposure variables, covariates, and SHBG concentrations may exist and may vary across the studies. However, because such biases are likely to be nondifferential, the estimates would tend to attenuate toward the null.
In conclusion, in this large multiethnic study of postmenopausal women, we identified age, exogenous estrogen use, physical activity, and regular coffee intake to be positively associated with SHBG concentrations, whereas BMI was inversely associated with SHBG concentrations. Based on the emerging significance of SHBG as an early biomarker for vascular outcomes, some cancers, and fracture risk, these findings may have important implications for developing preventive measures against various diseases.
Supplementary Material
Acknowledgments:
We are indebted to all participants and the entire staff of the WHI for their dedicated and conscientious collaboration and assistance. We are grateful for the invaluable contributions of the following investigators and staff: Sara A. Chacko, Kei-Hang Chan, Jennifer Yu, Xuyang Lu, and Nai-chieh Yuko You at the Program on Genomics and Nutrition and Center for Metabolic Disease Prevention, UCLA; Mitsuhiko Noda and Maki Goto at the National Center for Global Health and Medicine; and Yasuo Terauchi at Yokohama City University.
Research Funding:
The WHI is funded by the National Heart, Lung, and Blood Institute (NHLBI), NIH, R01 DK066401, and US Department of Health and Human Services through contracts N01WH22110, 24152, 32100-2, 32105-6, 32108-9, 32111-13, 32115, 32118-32119, 32122, 42107-26, 42129-32, and 44221. B.H. Chen, the Burroughs Wellcome Fund Interschool Training Program in Metabolic Diseases; R.D. Jackson, NIH; S. Liu, the Burroughs Wellcome Fund Interschool Training Program in Metabolic Diseases.
Role of Sponsor:
The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, or preparation or approval of manuscript.
Footnotes
Authors’ Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest:
Employment or Leadership: None declared.
Consultant or Advisory Role: R.D. Jackson, Merck.
Stock Ownership: None declared.
Honoraria: None declared.
Expert Testimony: None declared.
Patents: S. Liu, patent number US 20120157378.
Nonstandard abbreviations: SHBG, sex hormone–binding globulin; WHI-CT, Women’s Health Initiative Clinical Trial; WHI-OS, WHI Observational Study; BMI, body mass index; MET-h/week, metabolic equivalent-hours per week.
References
- 1.Anderson DC. Sex-hormone-binding globulin. Clin Endocrinol 1974;3:69–96. [DOI] [PubMed] [Google Scholar]
- 2.Key T, Appleby P, Barnes I, Reeves G; Endogenous Hormones and Breast Cancer Collaborative Group. Endogenous sex hormones and breast cancer in postmenopausal women: reanalysis of nine prospective studies. J Natl Cancer Inst 2002;94:606–16. [DOI] [PubMed] [Google Scholar]
- 3.Allen NE, Key TJ, Dossus L, Rinaldi S, Cust A, Lukanova A, et al. Endogenous sex hormones and endometrial cancer risk in women in the European Prospective Investigation into Cancer and Nutrition (EPIC). Endocr Relat Cancer 2008;15:485–97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Haffner SM, Moss SE, Klein BE, Klein R. Sex hormones and DHEA-SO4 in relation to ischemic heart disease mortality in diabetic subjects. The Wisconsin Epidemiologic Study of Diabetic Retinopathy. Diabetes Care 1996;19:1045–50. [DOI] [PubMed] [Google Scholar]
- 5.Ding EL, Song Y, Malik VS, Liu S. Sex differences of endogenous sex hormones and risk of type 2 diabetes: a systematic review and meta-analysis. JAMA 2006;295:1288–99. [DOI] [PubMed] [Google Scholar]
- 6.Lee JS, LaCroix AZ, Wu L, Cauley JA, Jackson RD, Kooperberg C, et al. Associations of serum sex hormone-binding globulin and sex hormone concentrations with hip fracture risk in postmenopausal women. J Clin Endocrinol Metab 2008;93:1796–803. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Coviello AD, Haring R, Wellons M, Vaidya D, Lehtimäki T, Keildson S, et al. A genome-wide association meta-analysis of circulating sex hormone–binding globulin reveals multiple Loci implicated in sex steroid hormone regulation. PLoS Genet 2012;8:e1002805. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Maggio M, Lauretani F, Basaria S, Ceda GP, Bandinelli S, Metter EJ, et al. Sex hormone binding globulin levels across the adult lifespan in women–the role of body mass index and fasting insulin. J Endocrinol Invest 2008;31:597–601. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Chan MF, Dowsett M, Folkerd E, Bingham S, Wareham N, Luben R, et al. Usual physical activity and endogenous sex hormones in postmenopausal women: the European Prospective Investigation into Cancer and Nutrition (EPIC). Cancer Epidemiol Biomarkers Prev 2007;16:900–5. [DOI] [PubMed] [Google Scholar]
- 10.McTiernan A, Wu L, Chen C, Chlebowski R, Mossavar-Rahmani Y, Modugno F, et al. Relation of BMI and physical activity to sex hormones in postmenopausal women. Obesity 2006;14:1662–77. [DOI] [PubMed] [Google Scholar]
- 11.Goto A, Song Y, Chen BH, Manson JE, Buring JE, Liu S. Coffee and caffeine consumption in relation to sex hormone-binding globulin and risk of type 2 diabetes in postmenopausal women. Diabetes 2011;60:269–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.McTiernan A, Wu L, Barnabei VM, Chen C, Hendrix S, Modugno F, et al. Relation of demographic factors, menstrual history, reproduction and medication use to sex hormone levels in postmenopausal women. Breast Cancer Res Treat 2008;108:217–31. [DOI] [PubMed] [Google Scholar]
- 13.Gower BA, Nyman L. Associations among oral estrogen use, free testosterone concentration, and lean body mass among postmenopausal women. J Clin Endocrinol Metab 2000;85:4476–80. [DOI] [PubMed] [Google Scholar]
- 14.Moore JW, Key TJ, Bulbrook RD, Clark GM, Allen DS, Wang DY, Pike MC. Sex hormone binding globulin and risk factors for breast cancer in a population of normal women who had never used exogenous sex hormones. Br J Cancer 1987;56:661–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.The Women’s Health Initiative. https://cleo.whi.org/ (Accessed November 2012).
- 16.Anderson GL, Manson J, Wallace R, Lund B, Hall D, Davis S, et al. Implementation of the Women’s Health Initiative study design. Ann Epidemiol 2003;13:S5–17. [DOI] [PubMed] [Google Scholar]
- 17.The Women’s Health Initiative Study Group. Design of the Women’s Health Initiative clinical trial and observational study. Control Clin Trials 1998;19:61–109. [DOI] [PubMed] [Google Scholar]
- 18.Patterson RE, Kristal AR, Tinker LF, Carter RA, Bolton MP, Agurs-Collins T. Measurement characteristics of the Women’s Health Initiative food frequency questionnaire. Ann Epidemiol 1999;9:178–87. [DOI] [PubMed] [Google Scholar]
- 19.Meyer AM, Evenson KR, Morimoto L, Siscovick D, White E. Test-retest reliability of the Women’s Health Initiative physical activity questionnaire. Med Sci Sports Exerc 2009;41:530–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7:177–88. [DOI] [PubMed] [Google Scholar]
- 21.Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002;21:1539–58. [DOI] [PubMed] [Google Scholar]
- 22.Lee IR, Dawson SA, Wetherall JD, Hahnel R. Sex hormone-binding globulin secretion by human hepatocarcinoma cells is increased by both estrogens and androgens. J Clin Endocrinol Metab 1987;64:825–31. [DOI] [PubMed] [Google Scholar]
- 23.Greenland S. Dose-response and trend analysis in epidemiology: alternatives to categorical analysis. Epidemiology 1995;6:356–65. [DOI] [PubMed] [Google Scholar]
- 24.Kotsopoulos J, Eliassen AH, Missmer SA, Hankinson SE, Tworoger SS. Relationship between caffeine intake and plasma sex hormone concentrations in premenopausal and postmenopausal women. Cancer 2009;115:2765–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Huber WW, Scharf G, Rossmanith W, Prustomersky S, Grasl-Kraupp B, Peter B, et al. The coffee components kahweol and cafestol induce gamma glutamylcysteine synthetase, the rate limiting enzyme of chemoprotective glutathione synthesis, in several organs of the rat. Arch Toxicol 2002;75:685–94. [DOI] [PubMed] [Google Scholar]
- 26.Astrup A, Toubro S, Cannon S, Hein P, Breum L, Madsen J. Caffeine: a double-blind, placebo-controlled study of its thermogenic, metabolic, and cardiovascular effects in healthy volunteers. Am J Clin Nutr 1990;51:759–67. [DOI] [PubMed] [Google Scholar]
- 27.Selva DM, Hammond GL. Thyroid hormones act indirectly to increase sex hormone-binding globulin production by liver via hepatocyte nuclear factor-4alpha. J Mol Endocrinol 2009;43:19–27. [DOI] [PubMed] [Google Scholar]
- 28.Elmlinger MW, Kühnel W, Wormstall H, Döller PC. Reference intervals for testosterone, androstenedione and SHBG levels in healthy females and males from birth until old age. Clin Lab 2005;51:625–32. [PubMed] [Google Scholar]
- 29.Edlefsen KL, Jackson RD, Prentice RL, Janssen I, Rajkovic A, O’Sullivan MJ, Anderson G. The effects of postmenopausal hormone therapy on serum estrogen, progesterone, and sex hormone-binding globulin levels in healthy postmenopausal women. Menopause 2010;17:622–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Karim R, Hodis HN, Stanczyk FZ, Lobo RA, Mack WJ. Relationship between serum levels of sex hormones and progression of subclinical atherosclerosis in postmenopausal women. J Clin Endocrinol Metab 2008;93:131–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Stomati M, Hartmann B, Spinetti A, Mailand D, Rubino S, Albrecht A, et al. Effects of hormonal replacement therapy on plasma sex hormone-binding globulin, androgen and insulin-like growth factor-1 levels in postmenopausal women. J Endocrinol Invest 1996;19:535–41. [DOI] [PubMed] [Google Scholar]
- 32.Choudhury F, Bernstein L, Hodis HN, Stanczyk FZ, Mack WJ. Physical activity and sex hormone levels in estradiol- and placebo-treated postmenopausal women. Menopause 2011;18:1079–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.McTiernan A, Tworoger SS, Ulrich CM, Yasui Y, Irwin ML, Rajan KB, et al. Effect of exercise on serum estrogens in postmenopausal women: a 12-month randomized clinical trial. Cancer Res 2004;64:2923–8. [DOI] [PubMed] [Google Scholar]
- 34.Haffner SM, Katz MS, Dunn JF. Increased upper body and overall adiposity is associated with decreased sex hormone binding globulin in post-menopausal women. Int J Obes 1991;15:471–8. [PubMed] [Google Scholar]
- 35.Santoro N, Torrens J, Crawford S, Allsworth JE, Finkelstein JS, Gold EB, et al. Correlates of circulating androgens in mid-life women: the Study of Women’s Health Across the Nation. J Clin Endocrinol Metab 2005;90:4836–45. [DOI] [PubMed] [Google Scholar]
- 36.Janssen I, Powell LH, Kazlauskaite R, Dugan SA. Testosterone and visceral fat in midlife women: the Study of Women’s Health Across the Nation (SWAN) fat patterning study. Obesity 2010;18:604–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Nestler JE. Sex hormone-binding globulin: a marker for hyperinsulinemia and/or insulin resistance? J Clin Endocrinol Metab 1993;76:273–4. [DOI] [PubMed] [Google Scholar]
- 38.Plymate SR, Matej LA, Jones RE, Friedl KE. Inhibition of sex hormone-binding globulin production in the human hepatoma (Hep G2) cell line by insulin and prolactin. J Clin Endocrinol Metab 1988;67:460–4. [DOI] [PubMed] [Google Scholar]
- 39.Hooper L, Ryder JJ, Kurzer MS, Lampe JW, Messina MJ, Phipps WR, Cassidy A. Effects of soy protein and isoflavones on circulating hormone concentrations in pre- and post-menopausal women: a systematic review and meta-analysis. Hum Reprod Update 2009;15:423–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.