The effect of testosterone on ovulatory function in transmasculine individuals

Abstract

BACKGROUND:

An estimated 1.4 million persons in the United States identify as transgender or nonbinary, signifying that their gender identity does not correspond with their assigned sex at birth. Individuals assigned female at birth may seek gender-affirming hormone therapy with testosterone. No studies have directly examined ovulatory function in transmasculine individuals using injectable testosterone.

OBJECTIVES:

Our primary objective was to determine the effect of testosterone on ovulatory suppression in transmasculine individuals. Secondary objectives were to determine predictors of ovulation in transmasculine individuals on testosterone, and to assess the effect of testosterone on antimüllerian hormone.

MATERIALS AND METHODS:

This prospective observational study recruited participants from a community clinic that provides gender-affirming hormone therapy. Enrolled individuals were assigned female at birth and were currently using or seeking to initiate masculinizing therapy with injectable testosterone esters (transmasculine individuals). Over a 12-week study period, participants collected daily urine samples for pregnanediol-3-glucoronide testing and completed daily electronic bleeding diaries. We assessed monthly serum mid-dosing interval testosterone, estradiol and sex hormone binding globulin, and antimüllerian hormone values at baseline and study end. Ovulation was defined as pregnanediol-3-glucoronide greater than 5 μg/mL for 3 consecutive days. The primary outcome was the proportion of participants who ovulated during the study period. We examined predictors of ovulation such as age, length of time on testosterone, serum testosterone levels, body mass index, and bleeding pattern.

RESULTS:

From July to November 2018, we enrolled 32 individuals; 20 completed the study (14 continuing testosterone users, 6 new users). Median age was 23 years (range 18–37 years). Bleeding or spotting during the study period was noted by 41% of participants (13/32). Among continuing users, median testosterone therapy duration was 11 months (range 1–60 months). A single ovulation was observed out of a total of 61 combined months of testosterone use; however, several transient rises in pregnanediol-3-glucoronide followed by bleeding episodes were suggestive of 7 dysfunctional ovulatory cycles among 7 individuals. There was no difference in antimüllerian hormone from baseline to 12 weeks between participants initiating testosterone and continuing users of testosterone. We did not have the power to examine our intended predictors given the low numbers of ovulatory events, but found that longer time on testosterone and presence of vaginal bleeding over 12 weeks were associated with transient rises in pregnanediol-3-glucoronide.

CONCLUSION:

This study suggests that testosterone rapidly induces hypothalamic—pituitary—gonadal suppression, resulting in anovulation in a proportion of new users. Importantly, these data also suggest that some long-term testosterone users break through the hormonal suppression and experience an ovulatory event, thereby raising concerns pertaining to the need for contraception in transmasculine individuals engaged in sexual intercourse with sperm-producing partners. Given the small number of overall participants, this work is hypothesis generating. Larger studies are needed to confirm and to clarify these findings.

Gender dysphoria refers to the distress or discomfort an individual experiences when their assigned birth sex is incongruous with their gender identity.^1–3 Available treatment options for gender dysphoria include cross-sex hormone therapy and gender-affirming surgery. Individuals may seek out 1, both, or neither of these therapies in the course of their gender transition.^2–5 There is a broad range of identities and preferred terminology to characterize this population. For clarity and consistency, we use the term “transmasculine individuals” to refer to individuals assigned female at birth who are seeking masculinizing gender-affirming care. It is estimated that there are 1.4 million transgender or nonbinary (TNB) individuals in the United States, defined as individuals whose gender identity does not correspond to their sex assigned at birth.⁶ Data suggest that the overall number of transmasculine individuals seeking hormone therapy is increasing.⁷ Many transmasculine individuals seek hormone therapy without plans to proceed with eventual hysterectomy,^8,9 yet data on fertility and family planning in this population are exceedingly limited.

Much of the research assessing fertility in TNB communities has focused on fertility-sparing options for those desiring a biological child. Contraception for those individuals not currently desiring fertility is often overlooked, frequently due to the assumption that transmasculine individuals do not have sexual relationships that could result in pregnancy. However, it is common for transmasculine individuals to have cisgender male partners.^10–14 Limited data suggest that condoms are the most common form of contraception in this population, although some transmasculine individuals do use hormonal contraceptives.^13,15,16 In addition to potential dysphoria associated with pregnancy in a transmasculine individual, testosterone can cause virilization of female fetuses and is categorized as a teratogen. Therefore, gaining a better understanding of contraceptive needs of this population is an important goal. Despite nearly universal amenorrhea within months of initiating testosterone therapy, unplanned pregnancies have been documented, and some individuals who stop testosterone to become pregnant have conceived prior to return of menses.¹⁵

To our knowledge, no prior studies in transmasculine individuals have assessed ovulation during testosterone use. Fertility potential during testosterone use and after cessation is not well characterized. Assessment of hormonal cycles through measurement of urinary hormone metabolites has been validated in a wide variety of medical conditions.^17–21 Daily measurement of urinary pregnanediol-3-glucoronide (PdG) specifically for evidence of luteal activity has shown excellent utility for retrospective confirmation of ovulation.²² This noninvasive method charts hormonal cycles and assesses individuals longitudinally for evidence of ovulation. This prospective cohort study of transmasculine individuals used daily urinary PdG measurements over a 12-week period to monitor ovulatory events in new and continuous transmasculine individuals using injectable testosterone. Findings of this study provide insight into the effect of exogenous testosterone on ovulatory function and the need to counsel transmasculine individuals about the risks for unintended pregnancy and the need for an effective contraceptive method.

Materials and Methods

Study population

Participants were recruited from a community clinic with 3 sites in the Seattle, Washington, metropolitan area. Participants assigned female at birth currently using or planning to initiate masculinizing therapy with injectable testosterone were eligible for the study if they had not undergone hysterectomy or oophorectomy, were currently on or initiating injectable testosterone therapy, and had monthly menses prior to the initiation of hormone therapy. Exclusion criteria included use of any form of hormonal contraception in the past 3 months, treatment with a gonadotropin-releasing hormone (GnRH) agonist, current use of finasteride, or use of topical testosterone because of inconsistency with absorption and variable dosing. All individuals used weekly injectable testosterone cypionate at a dose of 50–100 mg either intramuscularly, subcutaneously, or a combination, based on current guidelines for gender-affirming care.²³ Given the lack of previous data, the sample size was based on precision estimates. We estimated that to detect 4 ovulatory events, 30 participants would be required to detect an ovulation rate of 13.3% (95% confidence interval 5.1–30.6%). Written informed consent was obtained from all participants. Study procedures were approved by the University of Washington Institutional Review Board (HSD 4261).

Data collection

Baseline surveys collected demographic data, medical, sexual, and menstrual histories, fertility plans, and experience with fertility and contraception counseling. The survey was developed in conjunction with a TNB health expert and community member with experience developing gender-affirming care guidelines for multiple health systems. Gender and sexual identity categories were free text and then collapsed into categories for data analysis. Participants’ medical records were abstracted to confirm body mass index (BMI), medical history, current medications, and testosterone dosing. At the enrollment visit, participants had their blood drawn for testosterone, estradiol, sex hormone binding globulin (SHBG), and antimüllerian hormone (AMH). Participants were provided with urine collection kits with detailed instructions for collection and storage. Participants collected daily first-morning urine specimens for the entire 12-week study period. If this was not possible, urine collection occurred as early as possible in the day. Participants stored urine in 2-mL tubes in their home freezer. They received daily text message reminders and were given the opportunity to report any irregularities in their urine collection for that day, for example, a late collection. For new users, they were instructed to begin urine collection when they began testosterone injections, and this was considered day 1 of participation. Continuing users of testosterone were instructed to begin urine collection immediately following study entry. Data on daily bleeding patterns were collected via text message; a paper bleeding diary served as back-up. Definitions of bleeding and spotting were based on World Health Organization (WHO) guidelines developed for contraceptive studies.²⁴

Follow-up visits were scheduled on the participant’s mid-dosing interval day (either 3 or 4 days after injection) at 4 weeks, 8 weeks, and 12 weeks at a site that was convenient for the participant. At each follow-up visit, blood was collected for testosterone, estradiol, and SHBG, and a short online survey about health updates and testosterone dosing was completed (at home or in clinic). In addition, participants turned in urine specimens collected since the last visit or retained them until study completion. The 12-week visit also included a blood collection for AMH and a final survey.

Data management

Data were collected and managed using Research Electronic Data Capture (REDCap) electronic data capture tools hosted at the Institute of Translational Health Science at the University of Washington. REDCap is a secure, Web-based application designed to support data capture for research studies.²⁵

Laboratory analysis

Serum samples were sent to LabCorp (Burlington, NC). Testosterone, estradiol, and SHBG were analyzed via electrochemiluminescence immunoassay per LabCorp protocol. AMH was analyzed via electrochemiluminescence.

When participants returned urine specimens to study staff, they were stored at the University of Washington at −80°C until all specimens were collected and ready for analysis. They were shipped on dry ice to the Oregon National Primate Research Center where they were analyzed for urinary PdG using enzyme-linked immunosorbent assay (ELISA; Arbor Assays). The criterion for ovulation was defined as urinary PdG greater than 5 μg/mL for 3 days.²¹

Statistical methods

The primary outcome was the proportion of patients who ovulated, defined as urinary PdG greater than 5μ g/mL for 3 days,²¹ which has been shown to have a sensitivity of 81% and specificity of 100% when compared to transvaginal ultrasound confirmation of ovulation. Secondary outcomes were the (1) proportion of months during which participants ovulated, (2) AMH change from baseline to 12 weeks, and (3) correlation between ovulation and age, duration of testosterone use, BMI, and the presence of vaginal bleeding.

We calculated the proportion of transmasculine individuals ever experiencing ovulation over the study period and the proportion of months in which ovulation was observed. We performed univariable logistic regression to determine the relationship between ovulation and secondary outcomes and multivariable logistic regression analysis to identify predictors of ovulation such as serum hormone levels or BMI. The change in AMH at baseline and 12 weeks was calculated for each study participant. Change in AMH among new users vs continuing users was compared using independent, 2-sample t tests.

Data analysis was performed using R statistical software version 3.4.1 (2017 R Foundation for Statistical Computing, Viena, Austria).

Results

Baseline characteristics

Of the 109 individuals approached, 32 were enrolled; 46 of 109 were ineligible, 26 of 109 declined participation, and 5 of 109 were excluded for other reasons (Figure 1). In all, 22 participants completed at least 4 weeks and 20 completed 12 weeks of data collection. Of the 10 participants who discontinued the study with a known reason, 1 individual discontinued because of starting twice-weekly testosterone injections; the remaining participants discontinued because of an inability to make study visits or to complete study activities.

FIGURE 1.

Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) study flow diagram

Baseline characteristics (Table 1) varied between new users and continuing users of testosterone. Continuing users were more likely to be of white ethnicity than new users (P = .49). Among continuing users, median duration of testosterone therapy was 11 months (range 1–60 months), and median time to amenorrhea after starting testosterone was 3 months (range 0–20 months). The median age was 23 years (range 18–37 years). Median BMI was 28 (range 18–64). The majority of participants were white (91%, n = 29).

TABLE 1.

Baseline characteristics of study participants (N = 32)

Characteristic	Value (n = 32)
Time on testosterone, mo	6 (0, 60)
Age, y	23 (18, 37)
Body mass index	28 (18, 64)
Race (check all that apply)
White	29(91%)
African American	2 (6%)
Asian	2 (6%)
Hawaiian or Pacific Islander	1 (3%)
Other	1 (3%)
Multiracial	2 (6%)
Laboratory measures at baseline
Testosterone, ng/dL	483 (19, 1162)
AMH, ng/mL	3.3 (1.3, 20.3)
Estradiol, pg/mL	45 (12, 368)
SHBG, nmol/L	32 (10, 84)

Data are median (range) or count (%).

AMH, antimUllerian; SHBG, sex hormone binding globulin.

Half of participants identified their gender identity as male (n = 16); transmasculine was the second-most common gender identity (31%, n = 10). The most common sexual identity among participants was bisexual/pansexual/queer (41%, n = 13), with an equal number of participants identifying as straight and gay/lesbian (28%, n = 9). Approximately one-third of participants (31%, n = 10) reported having used contraception in the past, most commonly the oral contraceptive pill (28%, n = 9). Nearly one-third reported having sex that could put them at risk for pregnancy within the past 6 months (32%, n = 10). Although a large proportion of participants did report a desire to parent in the future (44%, n = 14), most indicated that they planned to adopt or have a partner carry the child; none reported a desire to carry a pregnancy.

Analysis of ovulation and bleeding

Of the 22 participants who submitted at least 4 weeks of urine collection, 6 were new users of testosterone and 16 were continuing users. Serum hormone profiles can be found in Table 2. A total of 1758 daily specimens contributing to 251 participant-weeks of urine for PdG testing was collected. PdG values of new users varied from those who were continuing users (Figure 2). Only 1 participant (1.6%) met ovulatory criteria (PdG >5 μg/mL for 3 consecutive days) (Table 2); this was a new user who ovulated 2 weeks after study initiation. However, we noticed that several other new users had transient rises in PdG values not meeting ovulation criteria, followed by an episode of withdrawal bleeding 2–5 days after PdG drop, suggestive of ovulatory or corpus luteum dysfunction. Given the desire in contraceptive studies to maximize sensitivity of ovulation detection, we then reanalyzed the data using a very liberal criterion of PdG greater than 3 μg/mL for 2 days. This increased the ovulatory months to 8 participant-months (13%). In bivariate analysis using the more liberal definition for ovulation, baseline measures for age and BMI were not associated with ovulation. Time since testosterone initiation was associated with ovulation, with participants ovulating having on average 14 fewer months on testosterone prior to the start of the 12-week observation period (P = .051). Chance of ovulation decreased by total duration of testosterone use (Figure 3). Bleeding or spotting was reported by 37.5% of participants during the study period, on 5.6% of days. Those who ovulated were more likely to have vaginal bleeding over the 12 weeks of observation, with bleeding documented in 28% (4/14) of those who did not have biochemical evidence of ovulation and 88% (7/8) of those who ovulated under our liberalized criteria (P = .008) (data not shown).

TABLE 2.

Hormonal profiles and ovulation data of study participants

Characteristic	All participants (N = 32)	New users (n = 7)	Continuing users (n = 25)	P valuea
Change in AMH, baseline to 12 wk	0.0 (−2.8, 0.5)	−0.2 (−2.2, 0.2)	0.0 (−2.8, 0.5)	.90
Bleeding baseline to 4 wk	9/30 (30%)	6/7 (86%)	3/23 (13%)	.0002
Bleeding baseline to 12 weeksb	12/30 (40%)	6/7 (86%)	6/23 (26%)	.0048
PdG rise in 12 weeks
Ovulation, >5 μg × 3 days	1/22(5%)	1/6(17%)	0/16(0%)	.095
Subthreshold PdG elevation, > 4 μg × 3 days	4/22 (18%)	4/6 (67%)	0/16(0%)	.0003
Subthreshold PdG elevation, >3 μg × 2 days	8/22 (36%)	6/6(100%)	2/16(13%)	<.0001

Data are median (range) or count (%).

AMH, antimUllerian; PdG, pregnanediol-3-glucoronide; SHBG, sex hormone binding globulin.

^aP value is either for ttest or for χ² test comparing participants not using testosterone before the study to those using testosterone

^bThe number 30 is used for the denominator here, as the numerator includes all 9 participants who noted bleeding at 4 weeks, although we did not have full 12-week data on all of these participants.

FIGURE 2.

a, Pregnanediol-3-glucoronide (PdG) levels among new users of testosterone (n = 6). b, PdG levels among continuing users of testosterone (n = 16)

FIGURE 3.

Ovulation by testosterone use. Orange lines show time on testosterone. Dots show ovulation: light blue indicates ≥2 consecutive pregnanediol-3-glucoronide (PdG) measures greater than 3 μg/mL; blue indicates ≥3 consecutive PdG measures greater than 4 μg/mL (with 2 of them greater than 5 μg/mL); and black indicates ≥3 consecutive PdG measures greater than 5 μg/mL

Mean change in AMH over 12 weeks among new users was 0.2 and was 0.0 among continuing users (P = .90).

Comment

Principal findings

This study supports a small but growing body of literature documenting the effects of testosterone on ovulatory function, and to our knowledge is the first study to look prospectively over 12 weeks at ovulation in transmasculine individuals currently using or newly initiating testosterone. We found that testosterone may cause rapid hypothalamic—pituitary—ovarian dysfunction and suppression in most individuals. Bleeding, although uncommon after 6 months, may reflect breakthrough ovulatory events or endometrial atrophy. Testosterone did not significantly affect AMH levels over the study period. New users of testosterone were more likely to experience ovulatory events during the first 4 weeks of use. However, attenuated PdG levels in the luteal phase may suggest abnormal corpus luteal function. By week 4–8 of testosterone use, no elevations of PdG were observed in new testosterone users suggesting anovulation. Ovulatory rates in continuous testosterone use were rare; 2 users experienced a potential ovulatory event after 6 and 17 months of testosterone, respectively. However, similar to new users, continuous users of testosterone also had dysfunctional ovulatory events characterized by attenuated PdG levels in the luteal phase.

Patterns of ovulatory dysfunction observed among new users in our study are similar to those observed in cisgender women in the perimenopausal period.^18,26 Because of the small numbers of ovulation events by our PdG criteria, we were unable to fully assess factors that might affect the associations between ovulatory dysfunction and testosterone initiation and duration of use. In the menopause transition, ovulatory dysfunction is associated with a suboptimal luteinizing hormone (LH) surge and reduced luteal phase progesterone.^18,26,27 Thus it is possible that abnormal PdG secretion may be due to altered hypothalamic-pituitary axis function and submaximal LH surges or irregular LH pulses.^27,28 We hypothesize that the anovulation seen in the majority of our participants after initiation of testosterone may be due to suppression of the hypothalamic—pituitary—ovarian (HPO) axis either by testosterone itself or by estrogen resulting from aromatized testosterone.

Clinical implications

Although no participants in our study indicated interest in future fertility, other recent studies suggest that up to a quarter of transmasculine individuals are interested in bearing children in the future.²⁹ Whether interested in future fertility or in preventing pregnancy, greater knowledge of testosterone’s effect on ovulatory function is critical to providing proper care to transmasculine individuals initiating and using testosterone, including the ability to counsel accurately on fertility as well as need for contraception. Although our study showed ovulation inhibition, the lack of change in AMH in our participants indicates that fertility and female physiology may return after cessation of testosterone, along with the possibility of spontaneous conception or good response to ovarian stimulation for in vitro fertilization. As testosterone is considered a teratogen,³⁰ and given the possibility of breakthrough ovulatory events, transmasculine individuals having sex that confers a risk of pregnancy should be counseled to use a reliable contraceptive method.

Research implications

Larger-scale studies designed to assess the effect of dose and duration of testosterone use on ovulation while controlling for confounding factors are necessary to evaluate the reproducibility of these findings. In addition, the clinical significance of sub-threshold elevations in PdG is unknown: specifically, whether they may result in a viable pregnancy. The mechanism of ovulation suppression by testosterone warrants further study. It is possible that central or peripheral aromatization of testosterone to estradiol, which disrupts GnRH pulsatility and gonadotropin release, is a possible mechanism, although serum estradiol levels do not appear to be elevated in transmasculine individuals.^31,32

Strengths and limitations

Strengths of this study include its prospective longitudinal design, use of well-established methods to detect ovulation in a novel population, and the richness of the data (>1750 blood and urine samples were collected). Limitations include the small sample size and relatively short follow-up (12 weeks). Because of our small sample size, we grouped together new initiators and continuing users, although these individuals may have different hormonal profiles. If participants were oligo-ovulatory, we may not have captured an ovulatory event during the study period, and we could not discern the effect of testosterone dose or route of administration (intramuscular vs subcutaneous) on ovulation. There are also limitations to our laboratory data. Hormone data obtained by immunoassay are less accurate than those derived from liquid chromatography—mass spectrometry.³³ In addition, we cannot know the true adherence to urine sample collection, raising the possibility of errors in the data. Finally, we did not quantify LH and FSH levels and we did not assess follicular development. It is possible that there is no difference in gonadotropin release, and that a dominant follicle develops but fails to ovulate or luteinize.

Conclusion

This study should be seen as hypothesis generating, and may guide future larger-scale studies that can better characterize testosterone’s effect on ovulation. Our study suggests that testosterone may cause rapid suppression of ovulation in transmasculine individuals on testosterone. Further research is needed to confirm these findings, to elucidate physiologic mechanisms that affect ovulation, and to improve the reproductive healthcare of this underrepresented and understudied patient population.

AJOG at a Glance.

Why was this study conducted?

• The effect of testosterone on ovulation is unknown, and the direct effect of testosterone has not been examined in transmasculine individuals

Key findings

• Testosterone leads to ovulatory dysfunction and rapidly induces anovulation.

• Testosterone levels did not significantly affect antimüllerian hormone (AMH).

• Bleeding is uncommon after 6 months. However vaginal bleeding may indicate persistent breakthrough ovulatory events.

What does this add to what is known?

• This is the first study to directly assess ovulatory status of transmasculine individuals on testosterone.

• These results suggest that rapid ovulatory suppression after testosterone initiation is common.

• More data are needed to understand the timeline and mechanism of ovulatory suppression, and the clinical significance with regard to fertility and long-term health consequences.