Intravaginal Testosterone Improves Sexual Satisfaction and Vaginal Symptoms Associated With Aromatase Inhibitors

Abstract

Context

Intravaginal testosterone (IVT) is a potential treatment of vulvovaginal atrophy (VVA) associated with aromatase inhibitor (AI) use.

Objective

To investigate the effects of IVT on sexual satisfaction, vaginal symptoms, and urinary incontinence (UI) associated with AI use.

Design

Double-blind, randomized, placebo-controlled trial.

Setting

Academic clinical research center.

Participants

Postmenopausal women taking an AI with VVA symptoms.

Intervention

IVT cream (300 μg per dose) or identical placebo, self-administered daily for 2 weeks and then thrice weekly for 24 weeks.

Main Outcomes and Measures

The primary outcome was the change in the sexual satisfaction score on the Female Sexual Function Index (FSFI). Secondary outcomes included vaginal symptoms and responses to the Profile of Female Sexual Function, the Female Sexual Distress Scale–Revised (FSDS-R), and the Questionnaire for UI Diagnosis. Serum sex steroids were measured.

Results

A total of 44 women were randomly assigned and 37 provided evaluable data, (mean age 56.4 years, SD 8.8 years). At 26 weeks, the mean between-group difference in the baseline-adjusted change in FSFI satisfaction scores was significantly greater for the IVT group than the placebo group (mean difference 0.73 units; 95% CI, 0.02 to 1.43; P = 0.043). IVT cream resulted in significant improvements, compared with placebo, in FSDS-R scores (P = 0.02), sexual concerns (P < 0.001), sexual responsiveness (P < 0.001), vaginal dryness (P = 0.009), and dyspareunia (P = 0.014). Serum sex steroid levels did not change. Few women had UI symptoms, with no treatment effect.

Conclusion

IVT significantly improved sexual satisfaction and reduced dyspareunia in postmenopausal women on AI therapy. The low reporting of UI among women on AI therapy merits further investigation.

Postmenopausal women with hormone receptor–positive early breast cancer are recommended treatment with an aromatase inhibitor (AI) to systemically inhibit estrogen production, either as primary therapy, or after 2 to 3 years of tamoxifen, for ≥5 years (1, 2). The profound estrogen depletion achieved with AI therapy causes vulvovaginal atrophy (VVA), manifest mostly as vaginal dryness and dyspareunia (3). Such symptoms may cause substantial personal distress and feelings of guilt (4), negatively affect relationships and quality of life (5), and often precipitate treatment discontinuation (6). Because VVA may be associated with urinary incontinence (UI), the term genitourinary syndrome of menopause (GSM) has been proposed to encompass the broader adverse genitourinary effects of estrogen depletion (7). Vaginal estrogen therapy is considered first-line therapy for VVA and GSM, but safety in women after breast cancer treated with an AI has not been established (2, 8–10).

Three clinical trials have reported alleviation of VVA symptoms with intravaginal testosterone (IVT) therapy in women taking an AI (11–13). These studies have not established efficacy because of their open label designs (11, 12, 14), lack of a control arm (12, 14), the use of supraphysiological dosages of testosterone that resulted in systemic absorption (11), and likely poor adherence to AI therapy (11). This double-blind, placebo-controlled randomized clinical trial was undertaken to establish the clinical efficacy of IVT therapy over 26 weeks in postmenopausal women with invasive breast cancer taking an AI. The predetermined primary outcome was the change in sexual satisfaction, as an index of overall sexual well-being, with secondary outcomes including dyspareunia and vaginal dryness, sexually related personal distress, other measures of sexual well-being, and vaginal health. We also hypothesized that women taking an AI would be likely to report UI because of their severe estrogen depletion, so we included a validated UI questionnaire. The main safety parameter was assessment of sex hormone levels by liquid chromatography and tandem mass spectrometry (LCMS).

Methods

Study design

This was a single-center, randomized, double-blind, placebo-controlled trial conducted in the School of Public Health and Preventive Medicine, Melbourne, Australia, with first enrolment in May 2016 and completion in December 2017. The study consisted of a 4-week screening period plus a 26-week treatment phase involving three study visits during which treatment efficacy of IVT was compared with that of identical placebo. The study was approved by the Monash University Human Research Ethics Committee, Clayton, Victoria, Australia. All participants provided written consent, which included consent to notify their family doctor and oncologist of their decision to participate. Consent from the treating oncologist was obtained in writing unless the participant was referred by her oncologist. The study was funded by an Accelerator Grant from the National Breast Cancer Foundation (Australia) grant no. NT-16-006. The study funder had no role in the study design, conduct, analysis, or reporting of the study findings.

Patient selection

Women were referred by their doctor or recruited via print and electronic news bulletins. Women were eligible for the study if they were ≥18 years old, had a diagnosis of invasive breast cancer, were taking an AI (anastrozole, exemestane, or letrozole), and were experiencing VVA symptoms such as dryness, soreness, irritation or burning, or dyspareunia for which they desired treatment. Because adjuvant endocrine therapy with an AI is given only to postmenopausal women or women on complete ovarian suppressive therapy, biochemical assessment of menopausal status was not required. Participants were required to have a vaginal pH >5 and a clinically acceptable Papanicolaou smear within the preceding 2 years, if they had a cervix. Women were ineligible if they had unsatisfactory sexual function due to another medical condition, had undiagnosed genital bleeding, or had used vaginal hormonal therapy in the past month or any systemic hormone therapy in the preceding 6 months. We excluded women who had moderate to severe acne, hirsutism, or androgenic alopecia, had any other major medical illness, had a past diagnosis of another cancer other than nonmelanocytic skin cancer, consumed more than three standard alcoholic drinks per day, or were unable to attend the study center for the required visits. Women using a serotonin-norepinephrine reuptake inhibitor (SNRI) or selective serotonin reuptake inhibitor (SSRI) were not excluded. After potential eligibility was ascertained by phone screening, serum estradiol was measured. Women were excluded if, at this point, their serum estradiol level was >10.9 pg/mL (40 pmol/L) because this level was considered indicative of possible AI poor compliance.

Randomization

The computer-generated schedule was created in random blocks of 2, 4, and 6 patients. The randomization schedules were generated and held by P.J.R., who was not involved with the day-to-day conduct of the study. The randomization code was sent to pharmacy staff, who numbered the study medication containers, and participants were sequentially assigned to the next unassigned treatment code at randomization. All participants and study staff, including outcome assessors, remained blinded to the treatment allocation until all analyses were completed.

Intervention

Participants were randomly assigned, in a 1:1 ratio, to treatment with 1 mL of an IVT cream (testosterone 300 μg/mL in VersaBase® cream) or an identical placebo of VersaBase® cream, prepared and dispensed by the Health Smart Pharmacy, Alfred Hospital, Melbourne, Australia. Participants were supplied the cream in an Adapta tip jar with five reusable syringes and instructed to insert the cream nightly for 2 weeks, and then three times a week for the study duration. They were asked to return the containers at 13 and 26 weeks, and containers were immediately weighed. A participant was considered compliant if she had used ≥75% of the study drug anticipated.

Data collection

Participants underwent a physical examination, including vital signs and pelvic examination, completed the study questionnaires, and had blood drawn for measurement of sex steroids by LCMS at baseline and at 13 and 26 weeks.

Outcome measures

The primary outcome was change from baseline in the satisfaction domain of the Female Sexual Function Index (FSFI) at 26 weeks (15). Secondary outcomes were assessed as follows. Self-reported vaginal symptoms (dryness, burning, painful intercourse, and dryness during intercourse) were scored from 0 (no symptom at all) to 4 (most severe, including refraining from penetrative sex because of dyspareunia). Physician-rated parameters included vaginal appearance and vaginal pH. Participants were requested to refrain from intercourse for 3 days before each study visit. Vaginal pH was measured with litmus paper held against the lateral vaginal wall. The vaginal surface epithelium was rated from 0 to 4 for secretions, color, epithelial integrity, and surface thickness as previously described (16). For the vaginal maturation value (MV), a vaginal wall smear was obtained. Vaginal cytology was reported by the Victorian Cytology Service Laboratory, Carlton, Australia. Smears deemed unsuitable to provide an exact 100- to 300-cell count by the cytologist (due to infection, inflammation, or cytolysis from bacteria) had an estimate made of the proportion of superficial, intermediate, and parabasal cells and were recorded as such. The vaginal MV was calculated as 1.0 × percentage superficial cells + 0.5 × percentage intermediate cells + 0 × percentage parabasal cells (score 0 to 100, with a higher score being more favorable) (17). Participants completed the FSFI and the Profile of Female Sexual Function (PFSF) (18). A higher score is more favorable for both the questionnaires. Although overlapping in some domains, the PFSF and FSFI capture complementary aspects of sexual well-being and function. The PFSF also includes a single overall rating question, “How would you rate your overall satisfaction with your sexuality?,” with options ranging from 1 (poor) to 5 (excellent). Sexually associated personal distress was assessed by the Female Sexual Distress Scale–Revised (FSDS-R), with a score of 11 indicative of distress (19).

Safety assessments

Serum estradiol (E₂) was measured by double-antibody precipitation radioimmunoassay (Pantex, Santa Monica, CA) before randomization and at 13 weeks. The limit of detection was 20 pmol/L (5.45 pg/mL). Testosterone, DHT, E₂, and estrone (E₁) were measured at 0, 13, and 26 weeks by LCMS in the laboratory of Dr. David Handelsman, ANZAC Institute, Sydney, NSW (20). The assay limits of detection and limits of quantification and within-run and between-run coefficients of variation (%) are total testosterone (0.01 ng/mL, 0.025 ng/mL, 2.0%, 3.9% to 6.5%), DHT (0.05 ng/mL, 0.1 ng/mL, 8%, 6.7% to 13.4%), E₂ (2.5 pg/mL, 5 pg/mL, 6.6%, 4.8% to 8.6%), and E₁ (1.25 pg/mL, 2.5 pg/mL, 4.7%, 4.6% to 7.5%).

Sample size and statistical analysis

The sample size was based on the study of Fernandes et al. (21) that reported a mean difference in FSFI satisfaction domain scores between the intervention and placebo groups at 12 weeks of 1.4 units (SD 1.4). Twenty-one participants in each group provided a power of 90% to find a difference of this magnitude (α = 0.05). The primary analysis was of the change by week 26 in the FSFI satisfaction domain by group allocation adjusted for the value at baseline. Usually studies of VVA treatments are of 13 weeks’ duration, but we elected to undertake a 26-week study to evaluate longer-term treatment effects. Where data at week 26 were missing but data had been collected at week 13, a last observation carried forward approach was used because 13 weeks is considered sufficient to capture the treatment effects of VVA therapy. For the vaginal symptoms and self-reported “overall sexuality” from question 1 of the PFSF, ordinal regression was used with the result at week 26 as the outcome variable and group allocation and the value at baseline as the explanatory variables. Linear regression analysis was used for the domains of the FSFI and the PFSF and the total score of the FSDS-R, with the outcome variable being the difference in scores between week 26 and baseline and the explanatory variables being group allocation and the value at baseline. The change from baseline for each of these variables is presented as the median (50th percentile) of the difference and interquartile range (IQR). Analysis was performed in Stata version 12.1, and a P value of 0.05 was regarded as significant.

Results

A total of 110 women contacted the research team about the study. Of these, 45 were ineligible and 21 decided not to participate (Fig. 1). Forty-four participants, aged 34 to 73 years, were randomly assigned, 22 to each treatment group. Sixteen participants allocated to placebo and 21 allocated to IVT cream provided data that could be evaluated. The treatment groups were similar in terms of age, body mass index, and mean duration and type of AI use (Table 1). Only 6 women in each group (27.3%) had any UI identified by the Questionnaire for UI Diagnosis.

Figure 1.

Flow of study participation.

Table 1.

Baseline Characteristics of Study Participants by Treatment Randomization

Variable	Placebo (n = 22)	IVT (n = 22)
Mean age ± SD, y	55.1 ± 9.7	57.7 ± 7.9
Weight ± SD, kg	68.1 ± 13.3	70.3 ± 12.0
Body mass index ± SD, kg/m2	25.1 ± 4.1	26.5 ± 4.7
Mean age of breast cancer diagnosis ± SD, y	50.2 ± 9.0	54.6 ± 7.9
	Range 32.98 to 65.60	Range 37.69 to 75.84
Mean duration of AI use ± SD, y	2.3 ± 1.6	2.4 ± 1.5
	Range 0.16 to 5.94	Range 0.22 to 6.06
Smoker (yes), n (%)	2 (9.1%)	1 (4.6%)
Alcohol use (yes), n (%)	22 (100.0%)	19 (86.4%)
Antidepressanta use at baseline (yes), n (%)	6 (27%)	6 (27%)
Type of AI
Letrozole	7 (31.8%)	8 (36.4%)
Exemestane	4 (18.2%)	0 (0.0%)
Anastrozole	11 (50.0%)	14 (63.6%)
UI
Stress UI	3 (13.6%)	3 (13.6%)
Urge UI	4 (18.2%)	5 (22.7%)
Mixed UI	1 (4.6%)	2 (9.1%)
Any UI	6 (27.3%)	6 (27.3%)

^a

Includes SNRIs and SSRIs.

The median change in the FSFI satisfaction domain was 0.6 units (IQR 0 to 1.4) in the placebo group and 1.2 (IQR 0.4 to 2.0) in the IVT group (Table 2). This between-group difference in the change in score was statistically significant in favor of the IVT group, according to a nonparametric test (P = 0.046), and remained significant when the mean between-group difference in the change in scores was adjusted for baseline in a linear regression analysis (mean difference 0.73 units; 95% CI, 0.02 to 1.43; P = 0.043) in favor of the IVT group. A between-group difference in the change in lubrication scores [median change 0 (IQR −0.15 to 1.35) in the placebo group and 0.6 (IQR 0 to 2.4) in the IVT group] was statistically significant according to a nonparametric test (P = 0.046), but the mean between-group difference in the change in scores (0.76; 95% CI, −0.21 to 1.72; P = 0.12) did not remain statistically significant when adjusted for baseline. There were no other between-group differences for the change in scores adjusted for baseline for any of the other domains of the FSFI.

Table 2.

Baseline and Week 26 FSFI, PFSF, and FSDS-R Questionnaire Outcomes by Treatment Group, Change in FSFI and PFSF Domain Scores, and FSDS-R Total and Results of Linear Regression Modeling for the Change in Score Adjusted for Baseline

Variable	Week 0 Median (IQR)		Week 26a Median (IQR)		Differenceb (Week 26a − Week 0) Median (IQR)			Linear Regression of Difference (Week 26a − Week 0)
	Placebo (n = 22)	IVT (n = 22)	Placebo (n = 16)	IVT (n = 21)	Placebo (n = 16)	IVT (n = 21)	P	β Treatment, 95% CI	P
PFSF domains
Desire	18.9 (8.9 to 26.7)	24.4 (13.3 to 40.0)	20.0 (10.0 to 28.9)	33.3 (20.0 to 48.9)	3.33 (−4.4 to 5.6)	4.44 (0 to 11.1)	0.602	5.66	0.222
								−3.58 to 14.90
Arousal	33.3 (20.0 to 46.7)	33.3 (13.3 to 60.0)	23.3 (13.3 to 60.0)	53.3 (26.7 to 66.7)	0 (−16.7 to 23.3)	13.3 (0 to 26.67)	0.268	13.38	0.100
								−2.69 to 29.45
Concerns	40.0 (26.7 to 53.3)	53.3 (33.3 to 73.3)	40.0 (23.3 to 70.0)	80 (60.0 to 93.3)	6.7 (−6.7 to 13.3)	13.3 (6.7 to 40.0)	0.033	26.40	< 0.001
								13.14 to 39.66
Responsiveness	45.7 (34.3 to 60.0)	58.6 (34.3 to 68.6)	44.3 (22.9 to 61.4)	77.1 (62.9 to 82.9)	−5.7 (−14.3 to 12.9)	17.1 (8.6 to 25.7)	0.003	21.88	< 0.001
								11.64 to 32.11
Pleasure	21.4 (2.9 to 31.4)	31.4 (2.9 to 51.4)	27.1 (14.3 to 32.9)	40.0 (31.4 to 51.4)	2.9 (−7.1 to 17.1)	0 (0 to 20)	0.424	8.90	0.080
								−1.12 to 18.93
Orgasm	30.0 (10.0 to 65.0)	37.5 (20 to 75)	50.0 (15.0 to 72.5)	45.0 (35.0 to 80.0)	7.5 (0 to 22.5)	5.0 (0 to 25)	0.902	2.23	0.799
								−15.45 to 19.92
Self-image	30.0 (20.0 to 40.0)	40.0 (25.0 to 55.0)	37.5 (17.5 to 50.0)	60.0 (35.0 to 65.0)	5 (−2.5 to 15)	10.0 (0 to 20)	0.370	7.98	0.174
								−3.69 to 19.65
FSFI domains
Desire	1.5 (1.2 to 2.4)	2.4 (1.8 to 3.0)	2.4 (1.5 to 3.0)	3.0 (2.4 to 3.6)	0.3 (0 to 1.2)	0 (0 to 1.2)	0.320	0.02	0.948
								−0.60 to 0.64
Arousal	1.95 (1.2 to 3.0)	2.7 (1.2 to 3.6)	3.0 (1.7 to 3.6)	3.9 (2.4 to 4.5)	0.3 (−0.6 to 1.1)	0.9 (0.3 to 1.2)	0.243	0.58	0.138
								−0.20 to 1.37
Lubrication	1.35 (1.2 to 2.1)	1.2 (1.2 to 2.7)	1.8 (1.2 to 3.6)	3.3 (1.8 to 4.8)	0 (−0.15 to 1.35)	0.6 (0 to 2.4)	0.046	0.76	0.120
								−0.21 to 1.72
Orgasm	1.6 (1.2 to 4.0)	2.2 (1.2 to 5.2)	3.4 (1.8 to 4.8)	4.0 (2.8 to 4.4)	0.4 (−0.4 to 1.2)	0.4 (0 to 1.6)	0.587	0.27	0.530
								−0.58 to 1.11
Satisfaction	2.4 (1.6 to 3.2)	2.6 (1.6 to 3.2)	2.6 (2.4 to 4.4)	4.0 (3.2 to 4.8)	0.6 (0 to 1.4)	1.2 (0.4 to 2.0)	0.046	0.73	0.043
								0.02 to 1.43
Pain	2.4 (0.0 to 3.2)	1.2 (0.0 to 2.4)	2.8 (0.8 to 4.6)	3.2 (0.0 to 5.2)	0.4 (0 to 1.4)	1.2 (0 to 3.2)	0.382	0.76	0.222
								−0.48 to 2.00
FSDS-R total score	33 (28 to 43)	23.5 (17 to 34)	34 (17 to 42)	12 (5 to 21)	−4 (−8 to 0)	−9 (−13 to −1)	0.225	−10.28	0.021
								−18.89 to −1.67

^a

Or observation at week 13 carried forward.

^b

Median of differences.

The median change in the PFSF sexual concerns domain was 6.7 units (IQR −6.7 to 13.3) in the placebo group and 13.3 units (IQR 6.7 to 40.0) in the IVT group, and according to a nonparametric test this difference was statistically significant (P = 0.033) in favor of the IVT group. When the mean between-group change in score was adjusted for baseline, in a linear regression analysis, the difference of 26.40 (95% CI, 13.14 to 39.66) remained significant (P < 0.001). Similarly, for the PFSF sexual responsiveness domain, the median change in the placebo group was −5.7 (IQR −14.3 to 12.9) and in the IVT group was 17.1 (IQR 8.6 to 25.7). According to a nonparametric test, this between-group difference in the change in score was significant (P = 0.003) in favor of the IVT group and remained so when the mean between-group difference of 21.88 (95% CI, 11.64 to 32.11) was adjusted for baseline (P < 0.001).

At baseline 13 of 22 participants in the placebo group rated their overall satisfaction with their sexuality as poor, along with 10 of 22 participants in the IVT group. At 26 weeks, 8 of 16 participants in the placebo group still rated their overall sexual satisfaction as poor, whereas only 2 of 21 in the IVT group did so. At 26 weeks, only 1 woman in the placebo group, compared with 5 in the IVT group, rated their overall satisfaction with their sexuality as at least very good. In the ordinal regression for this PFSF item at 26 weeks, the IVT group was more likely to have a higher score (ordinal regression = 7.52 per unit; 95% CI, 1.86 to 30.41; P = 0.005) adjusted for the baseline score.

All 22 participants in the placebo group and 20 of the 22 participants in the IVT group had a baseline FSDS-R score of ≥11. The mean difference between groups in the change in the FSDS-R total score, adjusted for baseline, was −10.28 (95% CI, −18.89 to −1.67) in favor of the IVT group (P = 0.02).

We tested the impact of including terms for each of age and SSRI or SNRI use in our analyses of each of the domains of the PFSF, the FSFI, and the FSDS-R total score. In no case did either age or SSRI or SNRI use make a statistically significant contribution to the outcome. Furthermore, the inclusion of each of these variables did not change the treatment effect for any of the domains in which a benefit was observed.

At week 26 vaginal dryness and painful intercourse were significantly less severe in the IVT group than in the placebo group (P = 0.009 and P = 0.031, respectively) (Table 3). Of the clinical examination parameters, the only between-group difference that was statistically significant was vaginal surface thickness, which was in favor of the IVT group (P = 0.009) (22). The median vaginal MVs were similar at baseline in the placebo and IVT groups (14, IQR 5 to 28; and 11, IQR 5 to 34, respectively). At 26 weeks, the median vaginal MVs increased in the placebo group (24, IQR 10 to 47) and the IVT group (29, IQR 19 to 45). In a linear regression analysis of change in MV adjusted for baseline, the between-group difference was not statistically different (β = 3; 95% CI, −9 to 15; P = 0.62). There were no between-group differences for the reporting of any UI symptoms at 26 weeks, adjusted for baseline status.

Table 3.

Vaginal Symptoms at Baseline and Week 26 by Treatment Group

Variable	Week 0 (n = 44)		Week 26 (Last Observation Carried Forward) (n = 37)		Ordinal Regressiona
	Placebo (n = 22)	IVT (n = 22)	Placebo (n = 16)	IVT (n = 21)	Ordinal Regression for Treatment	95% CI	P
Vaginal dryness (n = 42)					0.14	0.03–0.61	0.009
No	3 (14.3%)	1 (4.8%)	1 (6.3%)	12 (57.1%)	(n = 34)
Somewhat	3 (14.3%)	3 (14.3%)	6 (37.5%)	3 (14.3%)
Moderately	8 (38.1%)	6 (28.6%)	6 (37.5%)	4 (19.1%)
Severely	6 (28.6%)	11 (52.4%)	3 (18.8%)	2 (9.5%)
Didn’t answer question	1 (4.8%)	0
Vaginal burning (n = 43)					0.014	0.01–1.41	0.096
No	11 (52.4%)	12 (54.6%)	11 (68.8%)	19 (90.5%)	(n = 35)
Somewhat	3 (14.3%)	5 (22.7%)	2 (12.5%)	1 (4.8%)
Moderately	4 (19.1%)	3 (13.6%)	2 (12.5%)	1 (4.8%)
Severely	2 (9.5%)	2 (9.1%)	1 (6.3%)	0 (0.00%)
Didn’t answer question	1 (4.8%)	0
Currently sexually active (n = 42)	21 (95.5%)	18 (90.0%)	15 (93.8%)	15 (79.0%)
Painful intercourse (n = 40)			(n = 34)		0.21	0.05–0.87	0.031
No	1 (4.8%)	0 (0.00%)	0 (0.00%)	6 (33.3%)	(n = 34)
Somewhat	1 (4.8%)	2 (10.5%)	6 (37.5%)	4 (22.2%)
Moderately	8 (38.1%)	3 (15.8%)	4 (25.0%)	3 (16.7%)
Severely	11 (52.4%)	14 (73.7%)	6 (37.5%)	5 (27.8%)
Dryness during intercourse (n = 38)			(n = 34)		0.40	0.11–1.49	0.171
No	0 (0.00%)	0 (0.00%)	0 (0.0%)	3 (16.7%)	(n = 32)
Somewhat	1 (5.00%)	1 (5.6%)	5 (31.3%)	3 (16.7%)
Moderately	6 (30.0%)	1 (5.6%)	4 (25.0%)	7 (38.9%)
Severely	13 (65.0%)	16 (88.9%)	7 (43.8%)	5 (27.8%)

^a

Ordinal regression for the result at 26 wk by treatment group adjusted for the result at baseline. An ordinal regression < 1 means the treatment has improved the symptom.

There were no between-group differences for the serum levels of any of the sex steroids at baseline and no statistically significant between-group differences in any sex steroid at week 26 (Table 4).

Table 4.

Medians and Ranges for Each Sex Steroid Measured at Baseline and Week 26

	Baseline (n = 37)			Week 26a (n = 37)
Variable	Placebo (n = 16)	IVT (n = 21)	P	Placebo (n = 16)	IVT (n = 21)	P
E2, pg/mL			0.61			0.16
Median	3	3		3	3
Range	2.58 to 7.97	3 to 8.45		2.86 to 6.43	3 to 8.07
E1, pg/mL			0.81			0.21
Median	1	1		1	1
Range	1 to 4.68	1 to 2.5		1 to 1	1 to 69.5
Testosterone, ng/dL			0.40			0.30
Median	11.65	12.7		11.5	12.8
Range	5.61 to 19.8	7.47 to 43.8		5.53 to 19.9	8.56 to 32.4
DHT, ng/dL
Median	5.0	5.0	0.25	5.00	5.0	0.88
Range	5.0 to 6.0	5.0 to 5.0		5.0 to 6.0	5.0 to 9.0

To convert E₂ to pmol/L, multiply by 3.67; to convert E₁ to pmol/L, multiply by 3.7. To convert testosterone to nmol/L, multiply by 0.0347; and to convert DHT to nmol/L, multiply by 0.0344.

^a

Including week 13 carried forward.

Discussion

This double-blind, randomized, placebo-controlled trial demonstrates that low-dose IVT significantly improves sexual satisfaction and sexual responsiveness, alleviates vaginal dryness and dyspareunia, and reduces sexual concerns and sexually associated personal distress in women with breast cancer taking an AI. IVT also resulted in a significant improvement over placebo in overall sexual satisfaction. IVT was not associated with increased sexual desire or arousal above that of placebo. Estrogenic effects of lowering of vaginal pH and increasing vaginal epithelial superficial cells were not seen, and UI symptoms did not change. There was no evidence of systemic testosterone absorption, as evidenced by no increase in testosterone, E₂, or E₁ levels.

Sexual satisfaction was selected as the study primary outcome for this study, because it captures both sexually associated physical discomfort and psychological well-being. The parameters that exhibited statistically significant improvement with IVT above placebo provide a clear pattern of treatment response. Reduced vaginal dryness and dyspareunia, less sexually associated distress and improved responsiveness, translated into increased sexual satisfaction and lessening of sexual concerns.

Androgen receptors (ARs) are widespread throughout the vagina and lower urinary tract (23, 24), and the current study offers important insights into testosterone in these tissues. Specifically, the use of IVT with concurrent AI therapy, and with a placebo control, has enabled us to examine the direct effects of testosterone on the urogenital tract. Studies of postmenopausal women not taking AI therapy have demonstrated estrogen-mediated effects of both systemic and vaginal testosterone consistent with local aromatization of testosterone to E₂ (25). These estrogen-mediated effects include an increased proportion of vaginal superficial cells and their glycogen content (26, 27), lowering of vaginal pH, and improved vaginal health scores (28). Increased vaginal cell maturation was observed only when IVT was coadministered with vaginal estrogen (29).

In the current study IVT was not more effective than placebo in restoring vaginal epithelial maturation and vaginal pH, indicating that aromatization is needed to achieve these effects. Similarly, Witherby et al. (12), in a study of women using AIs, compared two doses of IVT (150 μg and 300 μg) and found no effect on vaginal pH. Although more superficial cells were seen in vaginal smears from the higher-dose group, there was no control group, so this may not have been a testosterone-induced effect (12).

Both oral and topical vaginal estrogen increase vaginal vascularization, but vaginal estrogen has been associated with greater improvement in vaginal blood flow than oral estrogen therapy, despite oral estrogen resulting in much higher circulating E₂ levels (30). We hypothesize that IVT alleviates vaginal dryness and dyspareunia by increasing local blood flow. Because our study participants were all taking an AI, local blood flow effects would probably be mediated directly by the AR. ARs have been identified in vaginal mucosa and vascular endothelium (31). Testosterone is a vasodilator and increases vaginal blood flow (32–34). In the current study, all physician-reported indices of vaginal health, including vaginal color, moved in a favorable direction in the IVT group. The lack of statistically significant between-group differences for the clinical vaginal assessments may have resulted from insufficient study power for these parameters. In addition, vaginal AR density is lower after menopause and increases with testosterone exposure (23, 31). An even longer treatment period may be needed for complete vaginal blood flow effects to be achieved. Increased self-reported vaginal lubrication was also significantly greater in the IVT group, although it did not remain significant after adjustment for baseline differences.

An unexpected finding was the low prevalence of UI symptoms in the study participants, compared with women in the general population (35), despite their profound estrogen depletion. A lower-than-expected frequency of UI symptoms, assessed by the Questionnaire for UI Diagnosis, was also found in an observational study of women on AI therapy (36). Estrogen receptors are present throughout the lower urinary tract and pelvic floor muscles (37), yet clinical evidence that estrogen deficiency directly contributes to UI is lacking. In contrast, there is increasing evidence that testosterone may be important for lower urinary tract function. In a rat model of stress UI, testosterone has both preventive and curative effects (24, 38), which persist with concurrent aromatase inhibition (39). Furthermore, a recent analysis of data from the National Health and Nutrition Examination Survey found that women in the lowest quartile of serum testosterone were significantly more likely to have stress or mixed UI, after adjustment for a number of variables (40). Hence, there is a compelling case for additional studies of IVT as a local treatment of postmenopausal UI.

There are several reasons why significant increases in sexual desire and arousal were not observed in our study. Our study was not powered to assess all domains of sexual function, notably desire and arousal. Therefore, the lack of any statistical improvement in these other domains could reflect our small study size. AIs inhibit conversion of testosterone to E₂, so the participants did not have notably low serum testosterone levels at baseline. Therefore, there was no increase in circulating sex steroid levels to mediate a central effect on desire or arousal. We also expected that the women most likely to participate would be those interested in sexual activity but limited by discomfort, not women with low sexual desire or an impaired capacity to become aroused. The improvements in self-reported sexual function in previous uncontrolled studies of IVT were probably due to a placebo effect, which was also present in our study and is known to be substantial for sexual outcomes (41). Melisko et al. (11) reported that IVT increased sexual interest in women on AI therapy. However, the IVT dosage used in their study was 5000 μg, three times per week. This resulted in supraphysiological total testosterone levels, as measured by LCMS (mean 171 ng/dL and median 67 ng/dL), such that a central effect on libido was not surprising (11).

The strengths of this study are the double-blind, placebo-controlled design, use of validated questionnaires, and measurement of sex steroids by LCMS. The main limitation was a greater number of discontinuations in the control group. This is not uncommon in studies of sexual function and difficult to avoid when study participants perceive no treatment benefit (42, 43). Recruitment was limited by potential participants and their doctors being apprehensive about a study involving “hormone therapy” after a diagnosis of invasive breast cancer.

Studies of transdermal testosterone therapy, of ≤2 years’ duration, have shown no differences from placebo in adverse events or serious adverse events, including on metabolic variables (41). Transdermal testosterone therapy has been associated with increased hair growth but not with increased facial hair, alopecia, or voice change (41). Although there was no evidence of absorption with the IVT dosage used in this study, higher dosages can result in increased serum testosterone. Therefore, if women taking an AI with VVA elect to be treated with IVT, they should be monitored with regular serum sex steroid levels measured by a sensitive technique.

In conclusion, testosterone has important direct urogenital effects in women, and low-dose IVT offers a therapeutic approach to alleviate vaginal dryness and dyspareunia and improve sexual well-being in women taking an AI. This finding warrants confirmation in a larger clinical trial. The potential use of IVT for management of UI merits additional investigation.

Abbreviations:

Abbreviations:
 
• AI

  aromatase inhibitor

 
• AR

  androgen receptor

 
• E₁

  estrone

 
• E₂

  estradiol

 
• FSDS-R

  Female Sexual Distress Scale–Revised

 
• FSFI

  Female Sexual Function Index

 
• GSM

  genitourinary syndrome of the menopause

 
• IQR

  interquartile range

 
• IVT

  intravaginal testosterone

 
• LCMS

  liquid chromatography and tandem mass spectrometry

 
• MV

  maturation value

 
• PFSF

  Profile of Female Sexual Function

 
• SNRI

  serotonin-norepinephrine reuptake inhibitor

 
• SSRI

  selective serotonin reuptake inhibitor

 
• UI

  urinary incontinence

 
• VVA

  vulvovaginal atrophy

Acknowledgments

We thank the participants for their contributions, the Breast Cancer Network of Australia for informing women about the study, and Ms. Corallee Morrow, Ms. Ashley Baring, and Dr. Farwa Rizvi for their assistance with the study.

Financial Support: The study was supported by an Accelerator Grant of the National Breast Cancer Foundation (Australia) (grant no. NT-16-006 to S.R.D.); S.R.D. is an NHMRC senior principal research fellow (grant no. 1135843).

Clinical Trial Information: Australian New Zealand Clinical Trials registry no. ACTRN12615000083594 (registered 2 February 2015).

Disclosure Summary: S.R.D. has received honoraria from Besins Healthcare, Pfizer Australia, and Lawley Pharmaceuticals Australia. S.W. has received conference sponsorship from Novartis, Roche, and Bristol-Myers Squibb and is a member of the advisory boards of Novartis, Roche, and Bristol-Myers Squibb.