Testosterone Dose-Response Relationships With Cardiovascular Risk Markers in Androgen-Deficient Women: A Randomized, Placebo-Controlled Trial

Abstract

Objective:

To determine dose-dependent effects of T administration on cardiovascular risk markers in women with low T levels.

Methods:

Seventy-one hysterectomized women with or without oophorectomy with total T < 31 ng/dL and/or free T < 3.5 pg/mL received a standardized transdermal estradiol regimen during the 12-week run-in period and were then randomized to receive weekly im injections of placebo or 3-, 6.25-, 12.5-, or 25-mg T enanthate for 24 weeks. Total and free T levels were measured by liquid chromatography-tandem mass spectrometry and equilibrium dialysis, respectively. Insulin resistance and inflammatory markers were measured at baseline and 24 weeks. In a subset of women, magnetic resonance imaging of the abdomen was performed to quantify abdominal fat volume.

Results:

Fifty-nine women who completed the 24-week intervention were included in the final analysis. The five groups were similar at baseline. Mean on-treatment nadir total T concentrations were 14, 79, 105, 130, and 232 ng/dL in the placebo group and the 3-, 6.25-, 12.5-, and 25-mg groups, respectively. No significant changes in fasting glucose, fasting insulin, homeostatic model assessment of insulin resistance, high sensitivity C-reactive protein, adiponectin, blood pressure, and heart rate were observed at any T dose when compared to placebo. Similarly, no dose- or concentration-dependent changes were observed in abdominal fat on magnetic resonance imaging.

Conclusion:

Short-term T administration over a wide range of doses for 24 weeks in women with low T levels was not associated with worsening of cardiovascular risk markers.

Testosterone therapy in women has been widely promoted for the treatment of sexual dysfunction, low bone mass, and impaired cognition. Several clinical trials in naturally and surgically menopausal women have demonstrated improvement in sexual function with transdermal T replacement that increased serum T levels into the mid-high normal range for healthy young women (1, 2). Despite this increasing interest in T use, however, its cardiovascular safety remains unknown.

Available data suggest that endogenous serum T levels are adversely associated with cardiovascular risk markers. For example, in women with polycystic ovary syndrome, endogenous serum T is positively associated with fat mass, proatherogenic dyslipidemia, and insulin resistance (3). Similarly, in naturally menopausal women, endogenous T levels have been associated with increased total fat mass, abdominal obesity, metabolic syndrome, and coronary heart disease (4, 5). Data from the Study of Women's Health Across the Nation (SWAN) showed a strong direct association between bioavailable T levels and visceral fat (6). However, the effects of exogenous T administration on serum cardiovascular markers and abdominal fat in postmenopausal women remain unknown.

In female-to-male transsexuals, long-term supraphysiological im T injections showed an increase in visceral fat accumulation (7). These findings stand in contrast to those observed among women with HIV treated with a physiological dose of transdermal T for 6 months (8). It remains unclear whether adverse effects of T replacement are limited to pharmacological doses, and whether androgen replacement at physiological doses can be safely administered in postmenopausal women. Accordingly, we investigated the dose-response relationships of T administration over a wide range of doses (physiological and supraphysiological) on serum cardiovascular markers and abdominal fat in hysterectomized women with low serum T concentrations.

Subjects and Methods

Study design

The Testosterone Dose Response in Surgically Menopausal Women (TDSM) trial was a two-center, parallel-group, placebo-controlled, double-blind, randomized trial designed to determine the dose-response effects of T on a range of androgen-dependent outcomes. The eligibility criteria and design of the TDSM trial have been previously reported (9) and are described here briefly. The trial consisted of a 12-week run-in period of transdermal estradiol administration, a 24-week treatment period, and a 16-week recovery period. The study was approved by the institutional review boards of Boston University Medical Center and Charles Drew University of Medicine and Science (Los Angeles, CA). All participants provided written informed consent.

Participants

Enrollment was open to healthy women, 21–60 years of age who had undergone hysterectomy with or without partial or total oophorectomy. The participants had serum total T concentrations less than 31 ng/dL or free T concentrations less than 3.5 pg/mL (less than the median for healthy young women) (10). We included women who had hysterectomy alone or partial oophorectomy if their FSH levels were ≥ 30 U/L or if they were already receiving estrogen therapy. Inclusion required a documented normal Papanicolaou test and mammogram within the last 12 months. We excluded women with major psychiatric illness, recent hospitalization, active cancers, poorly controlled diabetes mellitus (hemoglobin A1c > 8.5%), uncontrolled hypertension, severe obesity (body mass index [BMI] > 40 kg/m²), illicit drug use, alcohol dependence, and abnormal liver function. Women with a history of breast, ovarian, endometrial, or cervical cancer; hyperandrogenic disorders; cardiac disease; or thromboembolic disorders; and those taking glucocorticoids, androgens, spironolactone, and GnRH agonists were also excluded.

Randomization and study interventions

All eligible women were administered a regimen of transdermal estradiol patch applied twice a week and designed to achieve nominal delivery of 50-μg estradiol daily (Alora; Watson Pharmaceuticals) for a 12-week run-in phase. After run-in, the subjects were randomized in a double-blinded fashion to one of five groups to receive weekly im injections of placebo or 3-, 6.25-, 12.5-, or 25-mg T enanthate (ENDO Pharmaceuticals) for 24 weeks.

Hormone assays

Serum total T levels were measured by liquid chromatography-tandem mass spectrometry with sensitivity of 2 ng/dL, as described elsewhere (11). The cross-reactivity of dehydroepiandrosterone, dehydroepiandrosterone sulfate, dihydrotestosterone, androstenedione, and estradiol in the T assay was negligible at 10 times the circulating concentrations of these hormones. The interassay coefficient of variation (CV) was 15.8% at 12.0 ng/dL, 10.6% at 23.5 ng/dL, 7.9% at 48.6 ng/dL, 7.7% at 241 ng/dL, 4.4% at 532 ng/dL, and 3.3% at 1016 ng/dL, respectively. As part of the Centers for Disease Control's (CDC) Testosterone Assay Harmonization Initiative, quality control samples provided by the CDC were run every 3 months; the bias in quality control samples in the 3.47-to-34.7 nmol/L (100-to-1000 ng/dL) range was < 6.2%. Free T was measured using equilibrium dialysis with an interassay CV of 12.3% with sensitivity of 0.3 pg/mL (10, 12). SHBG levels were measured using an immunofluorometric assay with a sensitivity of 0.5 nmol/L (13). The interassay CVs were 8.3, 7.9, and 10.9%, and the intra-assay CVs were 7.3, 7.1, and 8.7%, respectively, in the low, medium, and high pools.

Metabolic and inflammatory markers, blood pressure, and heart rate

Fasting insulin was measured using a RIA and glucose by a glucose oxidase method. Insulin resistance was calculated using the homeostatic model assessment (HOMA) index (14). Adiponectin and high-sensitivity C-reactive protein (hs-CRP) were measured by ELISA assays (EMD Millipore and ALPCO Diagnostics). Blood pressure and resting heart rate were also measured.

Magnetic resonance imaging (MRI) assessment of abdominal fat volumes

In a subset of 35 patients, multislice, respiratory-triggered mixed turbo-spin echo, quantitative MRI of the abdomen (resolution, 1.65 × 1.65 × 7–8 mm³) was performed using a 1.5-Tesla MRI scanner (Intera; Philips Medical Systems of North America) at baseline and after 24 weeks of treatment. From each study, images of the upper abdomen from the L1–2 to L2–3 intervertebral levels were selected by a single operator and then analyzed using a semiautomated MathCad program. Subcutaneous and intraperitoneal anatomical compartments were automatically defined by the MathCad program. Once the fat compartments were segmented using a quantitative MRI thresholding algorithm (15), the volumes were calculated using a pixel-counting algorithm. All analyses were performed by a single radiologist who was blinded to treatment assignment.

Statistical analysis

Analyses were performed on subjects who had baseline and on-treatment data on metabolic parameters and inflammatory markers. A secondary analysis was performed on the subset of women who underwent MRI at baseline and 24 weeks. Tabular and graphical displays were used to compare baseline characteristics of groups. Mean change in outcomes was compared across treatment doses by linear regression incorporating adjustment for baseline outcome measurements. Response at each dose category was estimated using a treatment contrast and 95% confidence interval. Evidence in favor of an overall dose effect was assessed using a likelihood ratio test comparing this model to a similar model excluding dose. Generalized additive models were used to evaluate the association between change in each outcome measure with changes in total and free T levels (16). These models allow for curvilinearity in associations, if it was present, with preference for a simpler model if it was consistent with the data. Analyses were conducted using R version 2.14.2 (R Foundation for Statistical Computing).

Results

Flow of participants through the study

Of the 850 women who underwent telephone screening, 218 met eligibility criteria, 85 entered the estrogen run-in-period, 71 were randomized, and 59 who had baseline and end-of-treatment outcome data constituted the completer sample: placebo (n = 11), 3 mg (n = 12), 6.25 mg (n = 12), 12.5 mg (n = 14), or 25 mg (n = 10). Abdominal MRIs were available for evaluation in 35 women.

Baseline characteristics

Baseline characteristics across the five treatment groups are displayed in Table 1. Mean age of women enrolled in the study was 53 years, and average BMI was 29.9 kg/m². Participants across the dose groups were similar in terms of metabolic and inflammatory markers. Seven women reported a diagnosis of diabetes, but none was on insulin therapy. Baseline abdominal fat volumes are shown for a subset of patients who had MRI performed.

Table 1.

Baseline Characteristics of Analytic Sample by Randomized Assignment (n = 59)

Dose of T Enanthate, mg/wk	Placebo	3	6.5	12.5	25
n	12	11	12	14	10
Demographics
Age, y	54 ± 5	54 ± 5	51 ± 5	52 ± 6	54 ± 4
Weight, lb	185 ± 22	186 ± 47	168 ± 35	166 ± 34	169 ± 40
BMI, kg/m2	33 ± 3	32 ± 6	28 ± 6	29 ± 5	28 ± 7
Diabetes history, n (%)	2 (17)	1 (9)	0 (0)	3 (21)	1 (10)
Systolic blood pressure, mm Hg	124 ± 18	128 ± 15	115 ± 11	127 ± 16	125 ± 12
Diastolic blood pressure, mm Hg	74 ± 9	79 ± 11	70 ± 9	80 ± 9	79 ± 9
Hysterectomy alone, n (%)	3 (25)	3 (27)	0 (0)	1 (7)	4 (40)
Partial oophorectomy, n (%)	1 (8)	1 (9)	0 (0)	0 (0)	0 (0)
Bilateral oophorectomy, n (%)	8 (67)	7 (64)	12 (100)	13 (93)	6 (60)
Baseline hormone levels
Total T, ng/dLa
Screening	9 ± 4	16 ± 15	18 ± 16	19 ± 20	15 ± 11
Post-estrogen run-in	15 ± 9	13 ± 5	14 ± 13	13 ± 10	16 ± 10
Free T, pg/mLa
Screening	1.0 ± 0.4	1.0 ± 0.9	1.0 ± 1.2	1.0 ± 1.1	1.7 ± 2.7
Post-estrogen run-in	2.6 ± 1.7	2.1 ± 0.6	2.4 ± 2.5	2.5 ± 2.4	2.1 ± 1.3
SHBG, nmol/L (post-estrogen run-in)	63 ± 25	68 ± 32	58 ± 25	64 ± 35	95 ± 40
Baseline metabolic profile
Total cholesterol, mg/dL	196 ± 30	214 ± 49	217 ± 42	204 ± 38	205 ± 47
LDL, mg/dL	113 ± 23	119 ± 47	105 ± 30	125 ± 34	104 ± 32
HDL, mg/dL	58 ± 17	64 ± 24	69 ± 16	56 ± 15	74 ± 18
Triglycerides, mg/dL	139 ± 61	95 ± 36	131 ± 46	176 ± 125	84 ± 37
Hemoglobin A1c	5.8 ± 0.3	5.7 ± 0.4	5.5 ± 0.2	5.8 ± 0.8	5.9 ± 0.2
Insulin, μIU/mL	11 ± 11	9 ± 7	11 ± 15	12 ± 20	8 ± 13
Fasting glucose, mg/dL	97 ± 18	90 ± 10	97 ± 20	100 ± 27	96 ± 12
HOMAIR	3 ± 3	2 ± 2	3 ± 5	3 ± 5	2 ± 4
Adiponectin, μg/mL	12 ± 6	16 ± 12	18 ± 14	15 ± 12	14 ± 7
hs-CRP, mg/mL	3 ± 2	2 ± 2	4 ± 4	4 ± 4	2 ± 2
Abdominal fat volumes by MRI (n = 35)
Subcutaneous, mL	1748 ± 639	1484 ± 1303	1621 ± 872	1844 ± 912	927 ± 575
Visceral, mL	1496 ± 516	1025 ± 590	1061 ± 528	1444 ± 543	995 ± 405

Data represent mean ± SD or number (percentage). LDL, low-density lipoprotein; HDL, high-density lipoprotein; HOMA_IR, HOMA of insulin resistance.

^aT concentrations are provided in SI units in Supplemental Table 1. To convert T in ng/dL to nmol/L, multiply T concentration in ng/dL by 0.0347. To convert free T from pg/mL to pmol/L, multiply free T concentration in pg/mL by 3.467.

Hormone levels

Baseline mean total and free T concentrations were 16.0 ng/dL and 1.0 pg/mL, respectively, well below the range for healthy, menstruating women (9). Serum nadir total and free T levels, measured during week 24, 1 week after the previous injection, increased from baseline in a dose-dependent fashion, resulting in levels that were in the physiological to supraphysiological range. Mean on-treatment nadir total T concentrations were 14, 79, 105, 130, and 232 ng/dL, and free T concentrations were 2.7, 13, 18, 23, and 44 pg/mL at the 0, 3, 6.25, 12.5, and 25-mg doses, respectively.

Insulin resistance

Neither fasting glucose nor insulin concentrations changed significantly across groups. Insulin resistance, measured by HOMA of insulin resistance, did not significantly change at any dose group when compared to placebo and was not related to increases in T concentrations (Figure 1A and Supplemental Figure 2).

Figure 1.

A, Change in cardiometabolic measures by treatment group (n = 59). B, Change in abdominal sc and visceral fat volumes by treatment group (n = 35). In the bar graphs on the left, data represent absolute mean changes (±SE) from baseline for each treatment group. The significance level for the overall dose effect (by likelihood ratio test) is shown. Scatterplots on the right superimpose estimates and 95% confidence regions for the generalized additive model (GAM) of change in outcome as a function of change in total T levels. TE, T enanthate; HOMA_IR, HOMA of insulin resistance.

Inflammatory markers

There were no significant changes in adiponectin or hs-CRP levels from baseline to the end of treatment when compared to placebo, and these changes were not related to increases in serum T concentrations (Figure 1A and Supplemental Figure 2).

Blood pressure and heart rate

No significant dose-dependent changes were observed in heart rate or systolic or diastolic blood pressure after the 6-month intervention (Supplemental Figure 1).

Abdominal fat volumes

There were no significant changes in abdominal sc or visceral fat volumes when compared to placebo. Changes in sc and visceral fat were not related to increases in T concentrations (Figure 1B and Supplemental Figure 2). Adjustment for baseline BMI did not alter the results.

Discussion

In our trial of hysterectomized women with and without oophorectomy, short-term T administration for 6 months was not associated with worsening of cardiovascular risk markers. Data from epidemiological studies in specific patient populations have suggested that T levels in women may be related to cardiovascular health. For example, women with hirsutism and polycystic ovary syndrome have an unfavorable cardiovascular risk profile and premature atherosclerosis (17, 18). Similarly, higher T levels in postmenopausal women have been associated with insulin resistance, metabolic syndrome, visceral fat accumulation, and coronary heart disease (4, 6). Unlike these cross-sectional and observational studies, our 24-week dose-response interventional trial did not show worsening in cardiovascular risk parameters, as measured by serum markers and abdominal fat, over a wide range of T doses.

Higher T levels also have been shown to predict insulin resistance and incident type 2 diabetes in older women (19, 20). However, the effects of T replacement in postmenopausal women on insulin resistance have not been extensively studied in clinical trials. A randomized-controlled trial in postmenopausal women reported a small decrease in insulin sensitivity in postmenopausal women assigned to combination oral T undecanoate and estradiol than in those assigned to estradiol alone (21), suggesting that T therapy may lead to insulin resistance. In our study, we did not find significant changes in insulin resistance with 24 weeks of exogenous T administration, even at doses that achieved highly supraphysiological T concentrations.

Some of the potential concerns of T therapy in women include unfavorable changes in lipoproteins (22). Davis et al (1) found no significant changes in serum lipid profile in both naturally and surgically menopausal women who were receiving physiological doses of a transdermal T patch (without estrogen) for 6 months. We recently confirmed these findings over a wide range of T doses and have reported this in the primary publication of this trial (9) (see Supplemental Table 2). Although the high-density lipoprotein levels decreased in all dose groups, these changes were small and not significantly different compared to placebo.

The relationship between endogenous androgen levels in women and blood pressure have been inconsistent (22). Our study did not demonstrate significant changes in blood pressure or heart rate, confirming the findings of previous clinical trials of physiological T replacement (23). We demonstrate that short-term T replacement with pharmacological doses did not worsen these parameters, consistent with the findings of trials using high-dose T replacement in female-to-male transsexuals (24, 25).

In cross-sectional studies in postmenopausal women not on estrogen therapy, circulating T levels are positively correlated with CRP and negatively with adiponectin levels (26, 27). However, our intervention study did not confirm these findings. One study demonstrated that physiological and supraphysiological T may exert anti-inflammatory effects by suppressing cytokine expression in monocyte-derived macrophage cells of postmenopausal women (28), suggesting that T may protect against the progression of atherosclerosis. Long-term studies are needed to conclusively address this question.

We have previously shown that supraphysiological T administration in these women resulted in significant gains in lean body mass but had no effects on total fat mass, measured by dual-energy x-ray absorptiometry (9). In this study, T administration was not associated with significant changes in either the abdominal sc or the visceral fat compartments. Our results are consistent with data from HIV-infected women treated with physiological T replacement (8), but are in contrast to those obtained from female-to-male transsexuals (7), in whom long-term supraphysiological T doses were associated with increases in visceral fat. It is possible that larger T doses than those used in this trial and/or longer treatment duration may be necessary to exert significant effects on fat distribution.

Our study has notable strengths and some limitations. The trial had many features of a good trial design: concealed randomization, placebo control, blinding, and oversight by an independent data and safety monitoring board. Total and free T levels were measured using liquid chromatography-tandem mass spectrometry and equilibrium dialysis, respectively, both of which are widely considered the reference methods with the highest sensitivity and specificity. Testosterone injections were effective in raising T concentrations in a dose-dependent fashion over a wide range. However, measurement of cardiovascular disease risk markers was not the primary outcome of the trial, and the trial was not designed to detect a difference in changes in these cardiovascular disease risk parameters. Our analysis was therefore limited by small sample size that may have reduced our power to detect small effects. Most subjects in each group at study entry were either overweight or obese, which may have precluded us from demonstrating small changes in fat distribution. However, controlling for baseline BMI did not alter the results. Finally, the 6-month duration of intervention may not have been long enough to demonstrate a significant change in cardiovascular risk profile and atherosclerosis progression.

In conclusion, short-term T administration over a wide range for 24 weeks in hysterectomized women was not associated with worsening of cardiovascular risk markers. The cardiovascular safety of T therapy in women needs further investigation in long-term, adequately powered trials. Based on the findings of this trial, short-term use of T could be considered in select populations of women with androgen deficiency.