Abstract
Context:
Estrogen has been implicated in the development of uterine fibroids. However, the contribution of androgen in women is unknown.
Objective:
Our objective was to assess the longitudinal relations of circulating androgens and estradiol (E2) and their joint effects to the risk of developing fibroids.
Design:
This is a 13-year longitudinal study in the Study of Women's Health Across the Nation.
Setting:
This study was conducted in seven sites across the United States (1997–2013).
Participants:
At baseline, 3240 pre- or early peri-menopausal women with an intact uterus, ages 45–52 years were included; 43.6% completed the follow-up. There were 512 incident and 478 recurrent fibroid cases.
Exposures:
We measured near-annual time-varying serum levels of bioavailable E2 and bioavailable T, dichotomized at the median (high vs low).
Main Outcomes and Measures:
We estimated the conditional odds ratio (OR) of fibroids in the ensuing year using discrete-time proportional odds models adjusted for race/ethnicity/site, age, body mass index, menopausal stage, reproductive factors, smoking, timing of blood draw, and FSH.
Results:
Women with high T had a statistically significant increased risk of incident fibroids (OR, 1.33; 95% confidence interval [CI], 1.01–1.76; P = .04), but not recurrent fibroids. This risk was further elevated in those with high T and E2 (OR, 1.52; 95% CI, 1.07–2.17; P = .02). High E2 and T was associated with lower risk of recurrent fibroids (OR, 0.50; 95% CI, 0.26–0.96; P = .04).
Conclusions:
High T with high E2 was associated with an elevated risk of incident fibroids in midlife women who never reported fibroids before baseline. Conversely, the risk of recurrent fibroids was mitigated in women with high E2 and high T.
Uterine fibroids (leiomyomata) are benign neoplasms that develop from the myometrium of the uterus (1). Usually emerging during pre- or peri-menopause, fibroids are detected in 70–80% of women by age 50 years (2). Fibroids can cause irregular uterine bleeding, pelvic pain, infertility, recurrent pregnancy loss, and other reproductive complications (1). An estimated annual medical expenditure of $34.4 billion is attributed to the management of fibroids in the United States (3). Hysterectomy remains the first-line treatment for fibroids, with few alternative long-term medical treatment options (2). Myomectomy can be performed to remove individual detected fibroids; however, recurrence from underlying risk factors is still possible (4).
The strongest established risk factors for fibroids include being African-American, adiposity, and exogenous estrogen exposure (5–8). Estrogen-containing contraceptives or menopausal hormone therapy have also been associated with an increased risk of fibroids (9, 10). Further, menstruating women were found to have a higher risk of fibroids compared to postmenopausal women, most likely because of higher estrogen levels (11, 12) The bioactive form of estrogen (17β-estradiol [E2]) promotes the proliferation of fibroids via up-regulation of progesterone receptor expression (13). In vitro studies have also implicated androgen in the development of uterine fibroids (2, 14). Aromatase, which catalyzes androgen to estrogen, is overexpressed in fibroid tissue compared to normal myometrium, thus underscoring the potential relationship between androgen and estrogen in fibroid development (2, 15, 16). However, the role of circulating androgen and E2, and their interplay, in relation to the risk of fibroids is unknown. Characterizing these roles could better elucidate the pathogenesis of fibroids and aid in developing preventive strategies and nonsurgical treatment alternatives.
To address the gaps in knowledge, we leveraged data from the prospective Study of Women's Health Across the Nation (SWAN). The primary hypothesis was that higher bioavailable levels of E2 and androgens (T, and dehydroepiandrosterone-sulfate [DHEAS]) would be associated with increased risk of developing uterine fibroids. The study objectives were to determine: 1) the individual relationships and joint effects of time-varying circulating levels of bioavailable E2, bioavailable T, and DHEAS, to the risks of developing incident fibroids and first recurrence of fibroids across a 13-year follow-up period in midlife women undergoing the menopausal transition (MT); and 2) if these relationships differed among women in different racial/ethnic groups.
Materials and Methods
Study population
SWAN is a seven-site, longitudinal, community-based, multiethnic cohort study of midlife women undergoing the MT. SWAN enrolled 3302 eligible women at baseline and is currently conducting its 15th follow-up visit (17). Eligible women at baseline had to be 42–52 years old, pre- or early peri-menopausal, had a menstrual period and no exogenous hormone use in the prior 3 months, had an intact uterus and at least one ovary, and not be pregnant or lactating. Since 1996, SWAN participants were evaluated at near-annual, in-person visits using self-administered questionnaires and interviews to obtain information on demographics, medical history, reproductive characteristics, anthropometric measures, and medication use. Additionally, whole blood samples were collected and assays for serum biomarkers were performed. The institutional review boards of all participating institutions approved the SWAN core study. All participating women provided written informed consent.
The current study used a longitudinal design and excluded women who reported having hysterectomy or cancer diagnosis at baseline. Approximately 70–80% of women in the general population develop fibroids before age 50 years. Because the age range at baseline was 45–52 years, women who self-reported medical diagnosis of fibroids before baseline, but had an intact uterus, were also analyzed to enhance generalizability. Therefore, development of incident and first recurrent fibroids were of interest separately and together. The study sample consisted of 3240 women at baseline. Throughout the 13-year follow-up, women who reported having incident or recurrent fibroids, hysterectomy, died, or were permanently lost to follow-up since the last visit were censored at the time of initial report from future risk sets.
Outcome assessment: uterine fibroids
Uterine fibroids were self-reported on near-annual questionnaires throughout the follow-up. Fibroid occurrence was ascertained by asking the question, “Since your last study visit, has a doctor, nurse practitioner or other health care provider told you that you had fibroids, benign growths of the uterus or womb, or treated you for them,” from baseline to visit 3. From visits 4 through 13, fibroid occurrence was ascertained by asking “Since your last study visit, have you had fibroids?”
Exposure assessment: serum levels of sex hormones
SWAN measured serum total E2, total T, DHEAS, FSH, and SHBG levels in near-annual intervals throughout the follow-up (18). For pre- and early peri-menopausal women, timing of blood draw was defined as “in window” if between days 2 and 5 of menstrual cycle and “out of window” otherwise. Endocrine assays were performed with an Automated Chemiluminescence System (ACS)-180 analyzer (Bayer Diagnostics Corporation) using a double-antibody chemiluminescent immunoassay with a solid phase anti-IgG immunoglobulin conjugated to paramagnetic particles, anti-ligand antibody, and competitive ligand labeled with dimethyl acridinium ester, as previously described (18). Total E2 was assessed with a modified the rabbit anti-E2-6 ACS-180 immunoassay to increase sensitivity and had a lower limit of detection of 1.0 pg/ml. Total T was measured using a modified rabbit polyclonal anti-T ACS-180 immunoassay. The DHEAS and SHBG assays were developed on-site using rabbit anti-DHEAS and anti-SHBG antibodies. Duplicate E2 assays were performed and arithmetic means were reported for each participant, with a coefficient of variation of 3–12%. All other assays were single measurements. Bioavailable E2 and T levels were unitless indices representing the bioactive forms unbound to SHBG. Bioavailable E2 was calculated as the ratio of total E2 to SHBG: 100 × (total E2 [pg/ml] × [0.003671 nM per 1 pg/ml E2])/SHBG nM). Bioavailable T was calculated as the ratio of total T to SHBG: 100 × (total T [ng/dl] × [0.0347 nM per 1 ng/dl T]/SHBG nM) (17, 19).
Covariates
The following covariates were chosen based on previous literature: age at baseline, race/ethnicity (non-Hispanic white, African-American, Japanese, Chinese, and Hispanic), study site, time-varying MT stage (post- and late peri-menopause vs pre- and early peri-menopause), time-varying body mass index (BMI) (continuous; log kg/m2), time-varying sex hormone use for any reason including oral contraception, time-varying progestin use, time-varying current and former smoking status (binary) (20), time-varying passive smoke exposure (0, 1–4, >5 person-hours/week) (21), time-varying FSH (mIU/ml), parity at baseline, age at first birth, and timing of blood draw (in vs out of window) to account for short-term fluctuations in hormone levels during the menstrual cycle, while recognizing the irregularity of menstrual cycles during the MT.
Data analyses
At baseline, the distribution and normality of continuous variables were assessed using histograms, QQ plots, and Shapiro-Wilks tests. Differences in normally distributed continuous variables between racial/ethnic groups were assessed using ANOVA. Differences in non-normally distributed variables between racial/ethnic groups were assessed using Kruskal-Wallis tests. Continuous variables with right-skewed distributions were natural-log transformed to approximate normal distributions. Differences in categorical variables between racial/ethnic groups were assessed using χ2 tests. Spearman correlation coefficient matrices were used to assess the monotonic associations among independent variables.
The main exposures of interest were time-varying serum levels of bioavailable E2, bioavailable T, and DHEAS at each visit. Sex hormone levels were dichotomized: “high” levels were defined as being above the baseline median and “low” levels were below the median. Median values were from women who did not reported fibroids prior to baseline. A “combination” variable for bioavailable T and E2 was created with four categories: high T and high E2, high T and low E2, low T and high E2, and low T and low E2 (the reference category). The combination variable was used to evaluate the joint effect of bioavailable E2 and T. Similarly, a combination variable for DHEAS and bioavailable E2 was assessed.
Unadjusted and multivariable-adjusted discrete-time proportional odds models were used to estimate the conditional odds ratio (OR) of developing uterine fibroids in the ensuing year, in relation to time-varying levels of circulating sex hormones. The time scale was follow-up time in near-annual intervals (study visits). The follow-up time was from baseline to the 13th follow-up visit. Discrete-time survival models were used because the data were only observed in equally spaced discrete intervals (22). Discrete-time was specified in the models to account for tied event outcomes within intervals. The multivariable-adjusted models controlled for race/ethnicity, study site, and menopausal stage by stratification of the timescale, assuming different intercepts. The other previously mentioned covariates were parametrically adjusted. Three separate analyses were performed to estimate the risk of: 1) developing either incident or first recurrence of fibroids, in women with or without reported fibroids prior to baseline; 2) incident fibroids in women without reported fibroids at baseline; and 3) first recurrence of fibroids in women who reported fibroids prior to baseline.
Separate modeling was performed for: 1) “high” vs “low” levels of bioavailable E2, bioavailable T, and DHEAS, as dichotomized independent variables in the same model; and 2) each of the combination variables. Time-varying serum sex hormones were lagged by one follow-up visit in both the unadjusted and multivariable-adjusted analyses. Time-varying covariates were lagged by one visit in the multivariable-adjusted analyses. Each covariate was evaluated for its association with incident fibroids in separate unadjusted models. Interaction terms between bioavailable E2 and each androgen (bioavailable T and DHEAS) and between each sex hormone and race/ethnicity were included in some of the multivariable-adjusted models to assess effect measure modification. Interaction terms involving the combination sex hormone variables were not fitted in the analyses due to implied redundancy. Proportionality assumptions were assessed using interaction terms between the independent variables and time.
No blood draws occurred at the 11th follow-up visit. Women from the Newark, NJ, study site were censored after the sixth follow-up visit due to administrative issues resulting in discontinuation of data collection. Missing time-varying data throughout the follow-up period were imputed using last observation carried forward from the previous visit. All analyses were performed using SAS v9.4 (SAS Institute Inc). P values <.05 were considered statistically significant.
Results
Characteristics of the SWAN study population
Among the 3240 women at baseline, 43.6% completed the 13-year follow-up. There were 512 reported incident and 478 recurrent cases of uterine fibroids throughout the follow-up period. At baseline, 52.3% of women were pre-menopausal and 45.3% were early peri-menopausal (Table 1). The median age was 46.2 (interquartile range [IQR], 44.1–48.3) years. Nearly half of women (46.6%) self-identified as non-Hispanic white, 28.3% as African-American, 8.8% as Hispanic, 8.6% as Japanese, and 7.6% as Chinese. The median BMI was 26.6 kg/m2 (IQR, 22.9–32.1 kg/m2). Overall, 42.2% of women had ever smoked, whereas 17.2% were current smokers. Five or more person-hours/week of passive smoke exposure was reported in 28.5% of women. At baseline, the median total E2, total T, and SHBG levels were 55.1 pg/ml (IQR, 32.9–88.4 pg/ml), 41.5 ng/dl (IQR, 29.7–56.3 ng/dl), and 40.9 nM (IQR, 28.1–57.5 nM), respectively. The baseline median bioavailable E2 and T indices were 0.5 (IQR, 0.3–0.9) and 3.6 (IQR, 2.2–6.0), respectively.
Table 1.
Baseline Characteristics of the SWAN Study Population
| Continuous Variables | Race/Ethnicity |
|||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Total (n = 3240) |
White (n = 1511) |
African-American (n = 919) |
Hispanic (n = 285) |
Japanese (n = 279) |
Chinese (n = 246) |
|||||||||||||
| Median | 25th Ptl | 75th Ptl | Median | 25th Ptl | 75th Ptl | Median | 25th Ptl | 75th Ptl | Median | 25th Ptl | 75th Ptl | Median | 25th Ptl | 75th Ptl | Median | 25th Ptl | 75th Ptl | |
| Age (y) | 46.2 | 44.1 | 48.3 | 46.1 | 44.0 | 48.3 | 46.1 | 44.0 | 48.2 | 46.1 | 44.3 | 48.0 | 46.7 | 44.4 | 48.7 | 46.6 | 44.4 | 48.2 |
| BMI (kg/m2) | 26.6 | 22.9 | 32.1 | 26.1 | 22.9 | 31.3 | 30.2 | 26.1 | 36.2 | 28.3 | 25.5 | 32.2 | 22.1 | 20.4 | 24.6 | 22.4 | 20.7 | 24.7a |
| Parity | 2 | 1 | 3 | 2 | 1 | 3 | 2 | 1 | 3 | 2 | 2 | 3 | 2 | 1 | 3 | 2 | 1 | 2a |
| Age at first live birth (y) | 24 | 20 | 29 | 26 | 21 | 30 | 20 | 18 | 25 | 22 | 19 | 26 | 28 | 25 | 31 | 28 | 25 | 31a |
| Smoking pack-years in ever-smokers | 7.8 | 2.5 | 17.1 | 9.0 | 3.0 | 19.0 | 7.6 | 3.0 | 13.1 | 3.0 | 1.2 | 5.0 | 5.5 | 1.7 | 12.7 | 2.7 | 1.2 | 17.2a |
| Circulating sex hormones | ||||||||||||||||||
| Total T (ng/dl) | 41.5 | 29.7 | 56.3 | 42.9 | 30.7 | 57.4 | 41.8 | 30.5 | 57.9 | 37.4 | 26.5 | 50.8 | 38.1 | 28.3 | 52.5 | 40.1 | 28.1 | 51.6a |
| Bioavailable T index | 3.6 | 2.2 | 6.0 | 3.8 | 2.3 | 6.4 | 3.6 | 2.1 | 5.8 | 3.3 | 1.9 | 5.4 | 3.3 | 1.9 | 5.6 | 3.7 | 2.3 | 5.9b |
| Total 17β-estradiol (pg/ml) | 55.1 | 32.9 | 88.4 | 56.3 | 34.1 | 88.9 | 54.8 | 33.9 | 87.9 | 58.9 | 27.6 | 97.6 | 51.2 | 30.9 | 84.5 | 48.5 | 27.4 | 79.6b |
| Bioavailable 17β-estradiol index | 0.5 | 0.3 | 0.9 | 0.5 | 0.3 | 0.9 | 0.5 | 0.3 | 0.8 | 0.5 | 0.3 | 1.0 | 0.5 | 0.3 | 0.8 | 0.5 | 0.3 | 0.8 |
| Total DHEAS (μg/ml) | 113.9 | 74.5 | 168.9 | 122.3 | 78.5 | 178.3 | 92.2 | 60.6 | 135.7 | 107.2 | 63.9 | 144.4 | 128.7 | 87.3 | 179.2 | 155.9 | 105.2 | 216.6a |
| Total SHBG (nm) | 40.9 | 28.1 | 57.5 | 40.8 | 27.5 | 56.7 | 42.2 | 30.5 | 58.3 | 39.6 | 27.5 | 53.7 | 43.5 | 28.0 | 66.1 | 38.5 | 26.4 | 54.6b |
| FSH (mIU/ml) | 15.9 | 10.8 | 26.4 | 15.4 | 10.7 | 25.3 | 16.5 | 11.1 | 28.2 | 15.7 | 10.2 | 28.8 | 14.5 | 10.6 | 24.2 | 16.5 | 11.2 | 28.1 |
| Categorical Variables | Freq. | (%) | Freq. | (%) | Freq. | (%) | Freq. | (%) | Freq. | (%) | Freq. | (%) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Self-reported prior fibroids | 646 | 19.9 | 249 | 38.5 | 279 | 43.2 | 34 | 5.3 | 52 | 8.0 | 32 | 5.0a |
| Menopausal transition stage | ||||||||||||
| Early peri-menopause | 1468 | 45.3 | 704 | 48.0 | 456 | 31.1 | 112 | 7.6 | 104 | 7.1 | 92 | 6.3a |
| Pre-menopause | 1696 | 52.3 | 769 | 45.3 | 451 | 26.6 | 153 | 9.0 | 172 | 10.1 | 151 | 8.9 |
| Prior sex hormone use, excluding OC | 367 | 11.3 | 218 | 59.4 | 79 | 21.5 | 21 | 5.7 | 26 | 7.1 | 23 | 6.3a |
| Prior combination estrogen/progestin use | 19 | 0.5 | 9 | 47.3 | 6 | 31.6 | 0 | 0 | 2 | 10.5 | 2 | 10.5 |
| Prior progestin only use | 25 | 0.8 | 8 | 32.0 | 9 | 36.0 | 3 | 12.0 | 2 | 8.0 | 3 | 12.0 |
| Current smoking | 556 | 17.2 | 249 | 44.8 | 220 | 39.6 | 47 | 8.5 | 36 | 6.5 | 4 | 0.7a |
| Former smoking | 1367 | 42.2 | 738 | 54.0 | 421 | 30.8 | 92 | 6.7 | 100 | 7.3 | 16 | 1.2a |
| Passive smoking (person-hours/week) | ||||||||||||
| 0 | 1446 | 44.6 | 590 | 40.8 | 316 | 21.9 | 178 | 12.3 | 171 | 11.8 | 191 | 13.2a |
| 1–4 | 848 | 26.2 | 482 | 56.8 | 234 | 27.6 | 32 | 3.8 | 61 | 7.2 | 39 | 4.6 |
| ≥5 | 923 | 28.5 | 431 | 46.7 | 358 | 38.8 | 74 | 8.0 | 44 | 4.8 | 16 | 1.7 |
Includes women with intact uterus at baseline, regardless of previous self-reported fibroids at baseline. Differences in proportions between ethnicities were compared using a χ-square test. Differences in continuous variables were compared using Kruskal-Wallis tests. Discrepancy in counts is the result of missing data and rounding. No women used sex hormones, oral contraceptive, or progestin at baseline.
Abbreviations: Freq., frequency; OC, estrogen-containing oral contraceptives; Ptl, percentile.
P < .0001.
P values <.05 were considered statistically significant.
Associations of potential confounders with risks of incident and recurrent fibroids
African-American women had the highest unadjusted risk of incident fibroids (OR, 1.37; 95% confidence interval [CI], 1.12–1.68; P < .01), whereas Chinese women had the lowest (OR, 0.60; 95% CI 0.40–0.88; P < .01), compared to non-Hispanic white women (Table 2). Women in natural postmenopause had half the odds of incident fibroids (OR, 0.52; 95% CI, 0.34–0.78; P < .01), compared to premenopausal women. Age at first birth was inversely associated with risk of incident fibroids (OR, 0.96; 95% CI, 0.95–0.98; P < .0001), but parity was not significantly associated. Current use of exogenous sex hormone therapy was associated with nearly double the odds of incident fibroids (OR, 2.46; 95% CI, 1.97–3.07; P < .0001), as was progestin use (OR, 1.77; 95% CI, 1.26–2.50; P < .01).
Table 2.
Unadjusted Associations of Serum Sex Hormones and Pertinent Characteristics With Risk of Developing Uterine Fibroids
| Characteristics | Risk of Incident Fibroids |
Risk of First Recurrent Fibroids |
Risk of Incident and Recurrent Fibroids |
|||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Unadjusted Modelsa |
Unadjusted Modelsb |
Unadjusted Modelsc |
||||||||||
| OR | 95% CI Lower | 95% CI Upper | P Value | OR | 95% CI Lower | 95% CI Upper | P Value | OR | 95% CI Lower | 95% CI Upper | P Value | |
| Serum sex hormones | ||||||||||||
| Bioavailable T, high vs low | 1.00 | 0.83 | 1.19 | .95 | 1.06 | 0.86 | 1.30 | .60 | 0.97 | 0.85 | 1.11 | .65 |
| DHEAS, high vs low | 0.93 | 0.78 | 1.11 | .40 | 1.04 | 0.84 | 1.29 | .72 | 0.89 | 0.78 | 1.02 | .10 |
| Bioavailable 17β-estradiol, high vs low | 1.31 | 1.09 | 1.58 | <.01d | 0.97 | 0.78 | 1.20 | .75 | 1.14 | 1.00 | 1.31 | .06 |
| FSH (mIU/ml) | 0.99 | 0.99 | 1.00 | <.0001d | 1.00 | 1.00 | 1.00 | .11 | 1.00 | 0.99 | 1.00 | <.0001d |
| Demographic/anthropometric | ||||||||||||
| Age at baseline (y) | 0.98 | 0.95 | 1.01 | .22 | 0.96 | 0.93 | 1.00 | .05 d | 1.01 | 0.99 | 1.03 | .50 |
| Time-varying BMI (log kg/m2) | 1.19 | 0.83 | 1.71 | .34 | 1.59 | 0.99 | 2.54 | .05 d | 1.37 | 1.06 | 1.77 | .01d |
| Race/ethnicity | ||||||||||||
| African-American | 1.37 | 1.12 | 1.68 | <.01d | 1.73 | 1.34 | 2.23 | <.0001 d | 1.93 | 1.68 | 2.22 | <.0001d |
| Hispanic | 0.87 | 0.59 | 1.28 | .48 | 0.73 | 0.40 | 1.33 | .31 | 0.76 | 0.57 | 1.00 | .05d |
| Japanese | 0.77 | 0.55 | 1.08 | .13 | 1.30 | 0.89 | 1.89 | .17 | 1.05 | 0.82 | 1.33 | .72 |
| Chinese | 0.60 | 0.40 | 0.88 | <.01d | 1.03 | 0.62 | 1.71 | .92 | 0.72 | 0.54 | 0.95 | .02d |
| Non-Hispanic white | Ref | Ref | Ref | |||||||||
| Reproductive factors | ||||||||||||
| Time-varying menopausal transition stage | ||||||||||||
| Surgical post-menopause | 22.05 | 12.75 | 38.15 | <.0001d | 10.86 | 2.62 | 45.11 | <.001d | 21.75 | 14.98 | 31.57 | <.0001d |
| Natural post-menopause | 0.52 | 0.34 | 0.78 | <.01d | 0.66 | 0.38 | 1.14 | .14 | 0.62 | 0.47 | 0.82 | <.001d |
| Late peri-menopause | 0.80 | 0.51 | 1.26 | .34 | 0.83 | 0.47 | 1.46 | .51 | 0.88 | 0.64 | 1.21 | .44 |
| Early peri-menopause | 1.13 | 0.83 | 1.55 | .44 | 1.27 | 0.89 | 1.82 | .18 | 1.30 | 1.13 | 1.49 | <.001d |
| Unknown because of hormone use | 2.46 | 1.70 | 3.55 | <.0001d | 1.50 | 0.91 | 2.47 | .12 | 2.23 | 1.76 | 2.84 | <.0001d |
| Pre-menopause | Ref. | Ref. | Ref. | |||||||||
| Parity at baseline | 0.99 | 0.93 | 1.05 | .73 | 0.97 | 0.90 | 1.04 | .35 | 0.96 | 0.92 | 0.99 | .02d |
| Age at first birth (y) | 0.96 | 0.95 | 0.98 | <.0001d | 0.98 | 0.96 | 1.00 | .06 | 0.97 | 0.96 | 0.98 | <.0001d |
| Time-varying current sex hormone use, including OC | 2.46 | 1.97 | 3.07 | <.0001d | 1.72 | 1.25 | 2.37 | <.01d | 2.17 | 1.83 | 2.58 | <.0001d |
| Time-varying progestin-only use | 1.77 | 1.26 | 2.50 | <.01d | 1.43 | 0.88 | 2.35 | .15 | 1.65 | 1.25 | 2.18 | <.001d |
| Smoking/smoke exposure | ||||||||||||
| Time-varying current smoking | 0.82 | 0.63 | 1.07 | .15 | 1.08 | 0.81 | 1.43 | .61 | 0.92 | 0.77 | 1.10 | .33 |
| Time-varying former smoking | 0.80 | 0.62 | 1.03 | .09 | 1.34 | 1.02 | 1.77 | .04d | 0.96 | 0.83 | 1.10 | .52 |
| Cumulative smoking pack-years at baseline | ||||||||||||
| Q1 vs never smoking | 0.90 | 0.57 | 1.42 | .65 | 0.71 | 0.42 | 1.20 | .20 | 1.05 | 0.78 | 1.43 | .74 |
| Q2 vs never smoking | 1.23 | 0.90 | 1.69 | .19 | 0.77 | 0.50 | 1.19 | .24 | 1.03 | 0.82 | 1.29 | .83 |
| Q3 vs never smoking | 0.92 | 0.67 | 1.25 | .59 | 0.90 | 0.63 | 1.27 | .54 | 1.04 | 0.85 | 1.28 | .72 |
| Q4 vs never smoking | 1.00 | 0.76 | 1.32 | .99 | 0.90 | 0.63 | 1.28 | .56 | 0.85 | 0.70 | 1.05 | .13 |
| Time-varying passive smoke exposure (person hours/week) | ||||||||||||
| 1–4 | 1.27 | 1.03 | 1.57 | .03d | 1.20 | 0.93 | 1.55 | .17 | 1.19 | 1.03 | 1.39 | .02d |
| ≥5 | 1.24 | 1.00 | 1.54 | .05d | 1.18 | 0.91 | 1.53 | .21 | 1.30 | 1.12 | 1.51 | <.001d |
| 0 | Ref. | Ref. | Ref. | |||||||||
High bioavailable testosterone levels were above the median value of 3.6 nmol/liter. High DHEAS levels were above the median value of 116.6 ng/ml. High bioavailable estradiol levels were above the median value of 0.5 nmol/liter. Median baseline values were from women without fibroids at baseline. Sex hormones were lagged by one follow-up visit. Discrete-time proportional odds models were used. The outcome of interest was developing incident or recurrent fibroids.
Abbreviations: OC, estrogen-containing oral contraceptives; Q, quartile; Ref., reference.
Risk of incident fibroids among those free of fibroids and having an intact uterus at baseline.
Risk of recurrent fibroids among those who self-reported medical diagnosis of fibroids before baseline, but still have an intact uterus.
Risk of incident and recurrent fibroids among all women with an intact uterus at baseline, with and without previously reported fibroids.
P values <.05 were considered statistically significant.
African-American women had a 73% (95% CI, 34–123; P < .0001) higher odds of fibroid recurrence compared to non-Hispanic white women, among those who reported fibroids before baseline, but had an intact uterus (Table 2). Age at baseline was inversely associated with risk of recurrent fibroids (OR, 0.96; 95% CI, 0.93–1.00; P < .05). Increased BMI was associated with an increased risk of fibroid recurrence (OR, 1.59; 95% CI, 0.99–2.54; P < .05). Women who formerly smoked had a significant increased risk of fibroid recurrence compared to never smokers (OR, 1.34; 95% CI, 1.02–1.77; P = .04).
In women with an intact uterus at baseline, with and without prior report of fibroid occurrence, increased BMI, African-American race/ethnicity, early peri-menopause, sex hormone use, progestin use, and passive exposure to smoke were associated with increased odds of developing incident or recurrent fibroids (Table 2). FSH was inversely associated with developing fibroids (OR, 0.995; 95% CI, 0.993–0.997; P < .0001). Increased parity and later age at first birth were also associated with decreased risk of developing either incident or recurrent fibroids (OR, 0.96; 95% CI, 0.92–0.99; P = .02) and (OR, 0.97; 95% CI, 0.96–0.98; P < .0001), respectively).
Associations of serum sex hormone levels with risks of incident and recurrent fibroids
Women with high bioavailable T had a significant 1.33 (95% CI, 1.01–1.76; P = .04) times adjusted odds of incident fibroids, compared to women with low bioavailable T (Table 3). Women with both high bioavailable T and E2 had an even further elevated risk of incident fibroids compared to women with low levels of both hormones (adjusted OR, 1.52; 95% CI, 1.07–2.17; P = .02). However, T was not significantly associated with fibroid recurrence, among women with a history of fibroids at baseline.
Table 3.
Serum Sex Hormones in Relation to Risk of Incident and Recurrent Fibroids in Midlife Women
| Risk of Incident Fibroids |
Risk of First Recurrent Fibroids |
Risk of Incident and Recurrent Fibroids |
||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Multivariable-Adjusted Modelsa |
Multivariable-Adjusted Modelsb |
Multivariable-Adjusted Modelsc |
||||||||||
| OR | 95% CI Lower | 95% CI Upper | P Value | OR | 95% CI Lower | 95% CI Upper | P Value | OR | 95% CI Lower | 95% CI Upper | P Value | |
| Model I: dichotomized bioavailable T, DHEAS, and bioavailable 17β-estradiol | ||||||||||||
| T, high vs low | 1.33 | 1.01 | 1.76 | .04d | 0.86 | 0.53 | 1.37 | .52 | 1.19 | 0.94 | 1.51 | .14 |
| DHEAS, high vs low | 0.91 | 0.70 | 1.18 | .47 | 1.14 | 0.75 | 1.75 | .54 | 0.92 | 0.74 | 1.14 | .44 |
| 17β-estradiol, high vs low | 1.17 | 0.87 | 1.56 | .30 | 0.61 | 0.38 | 0.98 | .04d | 0.99 | 0.78 | 1.27 | .96 |
| Model II: combination bioavailable T and 17β-estradiol | ||||||||||||
| High T, high 17β-estradiol | 1.52 | 1.07 | 2.17 | .02d | 0.50 | 0.26 | 0.96 | .04d | 1.14 | 0.85 | 1.54 | .38 |
| High T, low 17β-estradiol | 1.31 | 0.91 | 1.89 | .15 | 1.01 | 0.57 | 1.78 | .97 | 1.24 | 0.92 | 1.68 | .15 |
| Low T, high 17β-estradiol | 1.17 | 0.80 | 1.72 | .41 | 0.71 | 0.39 | 1.29 | .25 | 1.06 | 0.78 | 1.45 | .70 |
| Low T, low 17β-estradiol | Ref | Ref | Ref | |||||||||
| Model III: combination DHEAS and bioavailable 17β-estradiol | ||||||||||||
| High DHEAS, high 17β-estradiol | 1.19 | 0.81 | 1.75 | .38 | 0.70 | 0.36 | 1.31 | .26 | 0.99 | 0.72 | 1.36 | .94 |
| High DHEAS, low 17β-estradiol | 1.00 | 0.71 | 1.41 | .99 | 0.99 | 0.58 | 1.70 | .98 | 0.95 | 0.72 | 1.24 | .69 |
| Low DHEAS, high 17β-estradiol | 1.32 | 0.90 | 1.93 | .15 | 0.50 | 0.26 | 0.96 | .04d | 1.03 | 0.76 | 1.41 | .83 |
| Low DHEAS, low 17β-estradiol | Ref. | Ref. | Ref. | |||||||||
Discrete-time proportional odds models were adjusted for race/ethnicity/study site, age at baseline, time-varying menopausal status, sex hormone use, progestin use, current passive smoking exposure, menstrual cycle within 2–5 days, current smoking status, former smoking status, time-varying FSH, parity at baseline, and age at first birth. Models I, II, and III were also fit with an interaction between the main effect and race/ethnicity/site. Each of the interactions between the main effect and race/ethnicity/site was not statistically significant. Model I was also fit with interactions between estradiol × testosterone and estradiol × DHEAS, the interaction terms were not statistically significant. All sex hormone concentrations were lagged by 1 y. High bioavailable testosterone levels were above the median value of 3.6 nmol/liter. High DHEAS levels were above the median value of 116.6 ng/ml. High bioavailable estradiol levels were above the median value of 0.5 nmol/liter. Median baseline values were from women without fibroids at baseline.
Abbreviation: Ref., reference.
Risk of incident fibroids among women free of fibroids and having an intact uterus at baseline.
Risk of recurrent fibroids among women who self-reported medical diagnosis of fibroids before baseline, but still have an intact uterus.
Risk of incident and recurrent fibroids among all women with an intact uterus at baseline, regardless of previously reported fibroids.
P values <.05 were considered statistically significant.
Women with high bioavailable E2 had a lower adjusted odds of first fibroid recurrence compared to women with low bioavailable E2 (OR, 0.61; 95% CI, 0.38–0.98; P = .04). Those with both high bioavailable T and E2 had an even lower risk of fibroid recurrence compared to women with low levels of both hormones (adjusted OR, 0.50; 95% CI, 0.26–0.96; P = .04). However, the association between bioavailable E2 and risk of incident fibroids was not statistically significant.
Among women with an intact uterus at baseline, with and without reported prior fibroids, bioavailable T, DHEAS, bioavailable E2, and their joint effects were not associated with developing either incident or recurrent fibroids.
Discussion
To our knowledge, this study is the first longitudinal investigation of the relationship between serum androgen and estrogen levels and the development of uterine fibroids. Our findings suggest that circulating T and E2 have opposite effects on the risks of incident and recurrent fibroids. Higher T levels were found to be associated with an increased risk of incident fibroids, but seemingly unrelated to risk of recurrent fibroids. The increased risk of incident fibroids was further elevated in women with both high T and E2 levels compared to those with low levels of both. Conversely, higher E2 levels were found to be associated with lower risk of recurrent fibroids in women who reported fibroids before study baseline. Further, women with both high E2 and T levels had an even greater protective effect against risk of fibroid recurrence. However, a significant relationship was not found between E2 levels and risk of incident fibroids. When examining all eligible participants who had an intact uterus at baseline, T, DHEAS, and E2 were not found to be associated with developing incident or recurrent fibroids. This finding was probably because their diametrically opposing effects on incident and recurrent fibroids drowned each other out when including prevalent cases at baseline.
Findings from this study raise the question as to why T and E2 have opposing relationships with risk of incident and recurrent fibroids. It should be noted that 70–80% of women in the general population develop fibroids by age 50 years. The baseline age range of SWAN was 45–52 years. Therefore, restricting the study sample to those who were free of fibroids at baseline would create a form of selection bias that limits generalizability to older women, and would only be relevant to late-onset fibroids. However, women who had previously reported fibroids before baseline may have a greater susceptibility of recurrence. Therefore, there may be heterogeneity in fibroid risk in the study sample when including both women free of fibroids and prevalent cases at baseline. As such, we performed three separate analyses restricting to women at baseline with an intact uterus who: 1) did not have fibroids, 2) reported prior fibroids, and 3) did and did not report prior fibroids as a means to isolate the risk of incident and recurrent fibroids separately and together. From a biological perspective, differences in the effects of T and E2 on incident and recurrent fibroids may be due to differences the expression of estrogen and androgen receptors in the myometrium of women with and without prior fibroids (1). Higher expression of estrogen and androgen receptors have been found in fibroids compared to normal myometrium (23–25). Previously having fibroids, which were removed by myomectomy, may alter receptor expression patterns in the vicinity. Furthermore, T and E2 maybe acting upon receptors on undetected fibroids, whose distributions are different between those with and without prior fibroids at baseline. However, previous studies found that estrogen promotes, rather than inhibits, growth of fibroids (1). Caution is recommended when interpreting the results. These findings warrant further investigation of the effect of androgens and E2 on fibroid risk.
The most prominent strength of this study was its longitudinal design, which established temporality among time-varying sex hormone levels, covariates, and outcome. SWAN is a unique community-based cohort with extensive participant characterization and near-annual repeated measurements of serum sex hormone measurements. The repeated measurements allowed the long-term longitudinal trajectories of sex hormone levels and pertinent factors over time to be evaluated in relation to fibroid risk. Additionally, the multiethnic composition of SWAN improves generalizability of the findings to the US population.
However, this investigation had limitations. First, the assessment of fibroids relied on self-report of previous medical diagnosis. SWAN did not assess fibroid size, number, or classification. Many uterine fibroids are asymptomatic; therefore, most of the reported cases of fibroids were likely to be symptomatic. Differential misclassification of the outcome may occur if higher sex hormone levels affects fibroid size, number, or classification, thus altering detection rates (26). However, undersampling asymptomatic women with higher T or E2 would likely bias the estimates toward the null and lead to conservative estimates of risk. Second, random mismeasurement of sex hormone levels would likely conservatively underestimate the effects. Third, near-annual sex hormone measurements, especially of E2, may not fully capture the short-term dynamic changes during the MT, and the timing of blood sampling may be challenging because of irregular menstrual cycling that is common to women with fibroids and women undergoing the MT. However, timing of blood collection related to the menstrual cycle was controlled in the analyses. Fourth, unmeasured confounders may have introduced bias even though major known confounders from previous literature were controlled. In particular, LH and serum progesterone may be important factors related to fibroids and E2, but were not measured in SWAN (27, 28). However, time-varying progestin use was controlled in the analysis.
Androgens and androgen receptors are expressed in fibroid tissue (14). Androgens are direct precursors of estrogens via conversion by aromatase; the mechanism by which T may increase the risk and growth of uterine fibroids could be via its conversion to E2 (29). Moreover, the overexpression of aromatase in fibroid tissue suggests that androgen might play an important role in fibroid development (15, 16). The lack of association between DHEAS and risk of fibroids may be due to its weak androgenic effects and lack of aromatization compared to T (30). Exogenous and endogenous estrogen are established factors in the development of uterine fibroids (5, 10, 12). Endogenous E2 levels dramatically decrease in menopause (31). Consistent with previous studies, we found that late peri- and post-menopausal women had lower risk of fibroids compared to pre-menopausal women (1).
African-American women had a substantially higher risk for uterine fibroids compared to non-Hispanic white women in our study. This finding was consistent with prior studies, which found that African-Americans were more likely to have a clinical diagnosis of uterine fibroids, with more severe related symptoms than white women (7, 8, 32–34). Higher passive smoke exposure was associated with an increased risk of fibroids. Further investigation into the relationship between passive smoke exposure and fibroid risk may be warranted.
In summary, findings from this longitudinal study of midlife women suggest that higher levels of both bioavailable T and E2 is related to an increased risk of incident fibroids in the ensuing year in women who never previously reported them, but is related to a decreased risk of recurrent fibroids in women who previously reported them. Considering the individual and joint effects of circulating T and E2 levels on fibroid risk could help in identifying high-risk midlife women and developing personalized nonsurgical therapeutic options. Future studies may benefit from focusing on both the short-term and long-term dynamics of sex hormone balance in the development of fibroids. Furthermore, a longer follow-up time beginning earlier in adulthood could improve the understanding of the relationship between sex hormone dynamics and fibroid risk across the life-course of women.
Acknowledgments
The authors thank Dr Po-Yin Chang and Dr Elaine Waetjen, the study staff at each site, and all the women who participated in Women's Health Across the Nation (SWAN).
SWAN has grant support from the National Institutes of Health (NIH), Department of Health and Human Services, through the National Institute on Aging (NIA), the National Institute of Nursing Research (NINR), and the NIH Office of Research on Women's Health (ORWH) (Grants U01NR004061, U01AG012505, U01AG012535, U01AG012531, U01AG012539, U01AG012546, U01AG012553, U01AG012554, U01AG012495, U01AG017719). The content of this article is solely the responsibility of the authors and does not necessarily represent the official views of the NIA, NINR, ORWH, or the NIH. This publication was supported in part by the National Center for Research Resources and the National Center for Advancing Translational Sciences, NIH, through UCSF-CTSI Grant UL1 RR024131. Supplemental funding from the NIA (7R21AG040568) is also gratefully acknowledged.
Clinical Centers. University of Michigan, Ann Arbor: Siobán Harlow, principal investigator (PI), 2011–present, MaryFran Sowers, PI, 1994–2011; Massachusetts General Hospital, Boston, MA: Joel Finkelstein, PI, 1999–present; Robert Neer, PI, 1994–1999; Rush University, Rush University Medical Center, Chicago, IL: Howard Kravitz, PI, 2009–present, Lynda Powell, PI, 1994–2009; University of California, Davis/Kaiser: Ellen Gold, PI; University of California, Los Angeles: Gail Greendale, PI; Albert Einstein College of Medicine, Bronx, NY: Carol Derby, PI, 2011–present, Rachel Wildman, PI, 2010–2011; Nanette Santoro, PI, 2004–2010; University of Medicine and Dentistry, New Jersey Medical School, Newark: Gerson Weiss, PI, 1994–2004; and the University of Pittsburgh, Pittsburgh, PA: Karen Matthews, PI. NIH Program Office: National Institute on Aging, Bethesda, MD: Winifred Rossi, 2012–present, Sherry Sherman, 1994–2012, Marcia Ory, 1994–2001; National Institute of Nursing Research, Bethesda, MD, Program Officers. Central Laboratory: University of Michigan, Ann Arbor: Daniel McConnell (Central Ligand Assay Satellite Services). SWAN Repository: University of Michigan, Ann Arbor, MI: Siobán Harlow, 2013; Dan McConnell, 2011–2013; MaryFran Sowers, 2000–2011. Coordinating Center: University of Pittsburgh, Pittsburgh, PA: Maria Mori Brooks, PI, 2012–present, Kim Sutton-Tyrrell, PI, 2001–2012; New England Research Institutes, Watertown, MA: Sonja McKinlay, PI, 1995–2001. Steering Committee: Susan Johnson, current chair. Chris Gallagher, former chair.
All authors have contributed to study design, analyses, data interpretation, and manuscript composition. We declare no conflicts of interest.
Disclosure Summary: The authors have nothing to disclose.
Footnotes
- ACS
- Automated Chemiluminescence System
- BMI
- body mass index
- CI
- confidence interval
- DHEAS
- dehydroepiandrosterone-sulfate
- E2
- estradiol
- IQR
- interquartile range
- MT
- menopausal transition
- OR
- odds ratio
- SWAN
- Study of Women's Health Across the Nation.
References
- 1. Flake GP, Andersen J, Dixon D. Etiology and pathogenesis of uterine leiomyomas: a review. Environ Health Perspect. 2003;111:1037–1054. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Taylor DK, Leppert PC. Treatment for uterine fibroids: searching for effective drug therapies. Drug Discov Today Ther Strateg. 2012;9:e41–e49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Cardozo ER, Clark AD, Banks NK, Henne MB, Stegmann BJ, Segars JH. The estimated annual cost of uterine leiomyomata in the United States. Am J Obstet Gynecol. 2012;206:211:e211–e219. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Lethaby A, Vollenhoven B. Fibroids (uterine myomatosis, leiomyomas). BMJ Clin Evid. 2015;2015. [PMC free article] [PubMed] [Google Scholar]
- 5. Blake RE. Leiomyomata uteri: hormonal and molecular determinants of growth. JAMA. 2007;99:1170–1184. [PMC free article] [PubMed] [Google Scholar]
- 6. Bulun SE. Uterine fibroids. N Engl J Med. 2013;369:1344–1355. [DOI] [PubMed] [Google Scholar]
- 7. Baird DD, Dunson DB, Hill MC, Cousins D, Schectman JM. High cumulative incidence of uterine leiomyoma in black and white women: ultrasound evidence. Am J Obstet Gynecol. 2003;188:100–107. [DOI] [PubMed] [Google Scholar]
- 8. Marshall LM, Spiegelman D, Barbieri RL, et al. Variation in the incidence of uterine leiomyoma among premenopausal women by age and race. Obstet Gynecol. 1997;90:967–973. [DOI] [PubMed] [Google Scholar]
- 9. Sabry M, Al-Hendy A. Medical treatment of uterine leiomyoma. Reprod Sci. 2012;19:339–353. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Templeman C, Marshall SF, Clarke CA, et al. Risk factors for surgically removed fibroids in a large cohort of teachers. Fertil Steril. 2009;92:1436–1446. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Stewart EA. Clinical practice. Uterine fibroids. N Engl J Med. 2015;372:1646–1655. [DOI] [PubMed] [Google Scholar]
- 12. Terry KL, De Vivo I, Hankinson SE, Missmer SA. Reproductive characteristics and risk of uterine leiomyomata. Fertil Steril. 2010;94:2703–2707. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Moravek MB, Yin P, Ono M, et al. Ovarian steroids, stem cells and uterine leiomyoma: therapeutic implications. Human Reprod Update. 2015;21:1–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Ke LQ, Yang K, Li J, Li CM. Danazol for uterine fibroids. Cochrane Database System Rev. 2009:CD007692. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Bulun SE, Imir G, Utsunomiya H, et al. Aromatase in endometriosis and uterine leiomyomata. J Steroid Biochem Molec Biol. 2005;95:57–62. [DOI] [PubMed] [Google Scholar]
- 16. Folkerd EJ, Newton CJ, Davidson K, Anderson MC, James VH. Aromatase activity in uterine leiomyomata. J Steroid Biochem. 1984;20:1195–1200. [DOI] [PubMed] [Google Scholar]
- 17. Gallicchio L, Schilling C, Romani WA, Miller S, Zacur H, Flaws JA. Endogenous hormones, participant characteristics, and symptoms among midlife women. Maturitas. 2008;59:114–127. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Sowers MR, Finkelstein JS, Ettinger B, et al. , Study of Women's Health Across the Nation. The association of endogenous hormone concentrations and bone mineral density measures in pre- and perimenopausal women of four ethnic groups: SWAN. Osteoporos Int. 2003;14:44–52. [DOI] [PubMed] [Google Scholar]
- 19. Sodergard R, Backstrom T, Shanbhag V, Carstensen H. Calculation of free and bound fractions of testosterone and estradiol-17 beta to human plasma proteins at body temperature. J Steroid Biochem. 1982;16:801–810. [DOI] [PubMed] [Google Scholar]
- 20. Ferris BG. Epidemiology Standardization Project (American Thoracic Society). The American review of respiratory disease. 1978;118:1–120. [PubMed] [Google Scholar]
- 21. Coghlin J, Hammond SK, Gann PH. Development of epidemiologic tools for measuring environmental tobacco smoke exposure. Am J Epidemiol. 1989;130:696–704. [DOI] [PubMed] [Google Scholar]
- 22. Singer JD, Willett JB. It's about time: using discrete-time survival analysis to study duration and the timing of events. J Educ Stat. 1993;18:155–195. [Google Scholar]
- 23. Bakas P, Liapis A, Vlahopoulos S, et al. Estrogen receptor alpha and beta in uterine fibroids: a basis for altered estrogen responsiveness. Fertil Steril. 2008;90:1878–1885. [DOI] [PubMed] [Google Scholar]
- 24. Sakaguchi H, Fujimoto J, Aoki I, Tamaya T. Expression of estrogen receptor alpha and beta in myometrium of premenopausal and postmenopausal women. Steroids. 2003;68:11–19. [DOI] [PubMed] [Google Scholar]
- 25. Di Luigi G, D'Alfonso A, Patacchiola F, Di Stefano L, Palermo P, Carta G. Leiomyosarcoma: a rare malignant transformation of a uterine leiomyoma. Eur J Gynaecol Oncol. 2015;36:84–87. [PubMed] [Google Scholar]
- 26. Myers SL, Baird DD, Olshan AF, et al. Self-report versus ultrasound measurement of uterine fibroid status. J Womens Health (Larchmt). 2012;21:285–293. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Williams AR, Critchley HO, Osei J, et al. The effects of the selective progesterone receptor modulator asoprisnil on the morphology of uterine tissues after 3 months treatment in patients with symptomatic uterine leiomyomata. Hum Reprod. 2007;22:1696–1704. [DOI] [PubMed] [Google Scholar]
- 28. Baird DD, Kesner JS, Dunson DB. Luteinizing hormone in premenopausal women may stimulate uterine leiomyomata development. J Soc Gynecol Invest. 2006;13:130–135. [DOI] [PubMed] [Google Scholar]
- 29. Moroni R, Vieira C, Ferriani R, Candido-Dos-Reis F, Brito L. Pharmacological treatment of uterine fibroids. Ann Med Health Sci Res. 2014;4:S185–S192. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Rehman KS, Carr BR. Sex differences in adrenal androgens. Semin Reprod Med. 2004;22:349–360. [DOI] [PubMed] [Google Scholar]
- 31. Anichini C, Calamai G, Pedemonte E, Moroni M, Tozzini S, Nesi G. Intravenous leiomyoma with cardiac involvement. Int Angiol. 2001;20:345–347. [PubMed] [Google Scholar]
- 32. Marsh EE, Bulun SE. Steroid hormones and leiomyomas. Obstet Gynecol Clin North Am. 2006;33:59–67. [DOI] [PubMed] [Google Scholar]
- 33. Baird DD, Hill MC, Schectman JM, Hollis BW. Vitamin D and the risk of uterine fibroids. Epidemiology. 2013;24:447–453. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Weiss G, Noorhasan D, Schott LL, Powell L, Randolph JF, Jr., Johnston JM. Racial differences in women who have a hysterectomy for benign conditions. Women's Health Issues. 2009;19:202–210. [DOI] [PMC free article] [PubMed] [Google Scholar]