Hormonal Treatment in Young People With Gender Dysphoria: A Systematic Review

Studies were considered eligible if participants were given hormonal treatment (GnRHas, GAHs, antiandrogens, or progestins) and if analysis of psychosocial, cognitive, and/or physical effects of these hormones were included. Participants had to be younger than 25 years of age and described as transgender or diagnosed with GD and/or GID according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition; Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition; or International Classification of Diseases criteria. This age range was selected to be consistent with the definition of adolescence used by the recent Lancet Commission on Adolescent Health.¹⁶ Studies were excluded if the effects of hormonal therapy could not be separated from gender-affirming surgery, which could cause potential issues related to interpretation of results. We included all published study designs in any language, but conference abstracts or studies in which researchers failed to report results at the group level with at least 10 individuals were excluded.

The Medline (Ovid) and Embase (Ovid) databases were searched for references from January 1, 1946, to June 10, 2017, by using thesauri and/or keywords. PubMed was searched by using keywords to retrieve electronic publications and items not indexed in Medline. The Medline search strategy was adapted for use in Embase and PubMed with the main search terms as follows: (GD or transsexualism or “sexual and gender disorders” or transgender persons or gender identity), (drug therapy or therapeutic use or [hormonal or hormone*] or *steroids or exp gestagen or exp antiandrogen), and (adolescen* or pediatric* or pediatric* or youth* or teen or teens or teenage*). Detailed search histories are available on request. Additional items were identified by manually searching reference lists of relevant retrieved articles. Two reviewers independently assessed all study titles and abstracts to determine inclusion, with the full text being subsequently retrieved for potentially eligible studies to assess final suitability. Any disagreements were resolved with discussion and consensus was reached for final articles.

Two reviewers, working independently and in duplicate, used a standardized form to extract methodological, demographic, and outcome data. Data extracted included reported youth characteristics (number of participants pre- and posttreatment, participant age range, diagnosis of GD, birth-assigned sex, and gender identity), hormonal therapy features (type, dose, route, duration of treatment), study design, and outcomes of interest (length of follow-up duration, follow-up outcome measures, and treatment effect on outcome measures).

Risk of bias in studies was assessed by 2 authors working independently using a modified version of the Quality in Prognosis Studies (QUIPS) tool from a previous study.¹⁷ The original QUIPS tool¹⁸ was modified because confounders or prognostic factors were not analyzed in this review and thus did not apply.

A detailed protocol is available at PROSPERO (identifier 42017056670).

Effect sizes were calculated for results with reported means and SDs,^19,–27 according to a previous study.²⁸ Unadjusted effect sizes using the posttest SD were calculated for the majority of studies, with an adjusted effect size using the experimental SD calculated only for 1 study with comparison between groups.²⁴ 

Meta-analysis was planned for outcomes examined by 3 or more studies but was unable to be conducted because individual outcome effect sizes were available for a maximum of 2 studies.

The study selection process is depicted in Fig 1. Eighty-three potentially relevant studies were retrieved, of which 13^{19,–27,29,–32} met the inclusion criteria and were systematically analyzed. In Table 1, we summarize the main characteristics of these 13 studies, and their key physical, psychosocial, and cognitive findings are outlined in Tables 2, 3 and 4, respectively. Because research from the same cohort was described in 2 of the studies,^26,27 they were considered as 1 study.

In all studies, there was a medium to high risk of bias (Table 5). In most studies, there were only small sample sizes (minimum of 21 and maximum of 201), with <50 participants in 38.5% of the studies. There were controls in only 2 studies, and all studies were conducted in clinical populations. There was often significant loss to follow-up, attributed partially to most studies being retrospective with missing data. Overall, the tools used to measure the specific outcomes were valid and reliable, although there was no blinding or randomization in any of the studies.

All relevant results are shown in Table 2.

GnRHas were successful in suppressing sex hormone secretion with significant decreases in gonadotropin,²⁹ estradiol, and testosterone^19,21,29 levels, although 1 study only revealed a significant decrease in transfemale adolescents (birth-assigned male individuals identifying as female individuals).¹⁹ There was decreased testicular volume in transfemale adolescents^19,21,29 and cessation of menses in transmale adolescents (birth-assigned female individuals identifying as male individuals),²¹ although the latter often occurred after a withdrawal bleed in postmenarchal individuals. Furthermore, GnRHas were shown to decrease luteinizing hormone (LH) and follicle-stimulating hormone (FSH).^19,21 

Researchers in 1 study examined the effects of the progestin lynestrenol³⁰ in transmale adolescents. Although there was no report of the efficacy of lynestrenol in stopping menses, there were significant reductions in levels of serum sex-hormone binding globulin (SHBG) and LH, in addition to a significant increase in free testosterone (fT). FSH, estradiol, testosterone, and anti-Mullerian hormone had nonsignificant decreases.³⁰ 

In one study, researchers studied the effects of the antiandrogen cyproterone acetate alone in transfemale adolescents.²² It was effective in significantly suppressing endogenous sex hormones with significant reductions in testosterone and dehydroepiandrosterone in addition to nonsignificant decreases in estradiol and fT after 12 months, with no significant changes in LH, FSH, and SHBG. Cyproterone acetate was associated with a marked increase in prolactin of ∼2.5-fold that exceeded the normal reference range after 6 months but returned to the normal range after 12 months. No clinical consequences, including galactorrhea, were reported. Furthermore, 55.6% of participants also reported decreased facial shaving frequency.²² 

Estrogen was successful in feminizing physical sex characteristics.^22,29 In 1 study, 66.7% of participants reached Tanner B3 stage (increase in breast and areola size), and 9.5% reached Tanner B4 (secondary mound created by areola and papilla) after treatment with cyproterone acetate and estrogen for at least 6 months.²² However, breast development was found to be objectively dissatisfactory and subjectively less in size than expected for the majority.²² There was a significant increase in serum estradiol after 6 months that reached the female reference range, whereas total testosterone decreased after 1 to 3 months to be outside of the male reference range.^22,32 Prolactin was unchanged.²² 

Testosterone resulted in virilization, including lower voice, clitoral enlargement, and body hair growth in a masculinized pattern.²⁹ Menses ceased in most transmale adolescents within 6 months, with an average time to cessation of 2.9 months.²⁰ Testosterone resulted in increased total testosterone and fT,^20,30,32 with most participants reaching levels within the normal male range after 6 months,^20,30 as well as significant decreases in LH and FSH.³⁰ This was accompanied by a decline in estradiol levels after 6 months,^20,30,32 which was statistically significant in 2 studies^20,30 but nonsignificant in 1 study.³² 

Hot flashes were a common side effect in transmale adolescents treated in late puberty (Tanner stages B4 and B5), although these decreased in frequency over time.²⁹ No other short-term side effects, including local reactions, were reported.

Lynestrenol was evaluated as relatively safe, with the most common side effects being initial metrorrhagia (48.7%), headaches (12.1%), hot flashes (9.8%), and acne (which increased from 14.6% to 28.6%).³⁰ 

Treatment with cyproterone acetate was evaluated to be relatively safe, with the most common side effect being fatigue (37%).²² 

Side effects reported with combined estrogen and cyproterone acetate included breast tenderness (57.1%), emotionality (28.6%), hunger (23.8%), fatigue (14.3%), and hot flashes (14.3%).²² 

Few side effects were reported with testosterone treatment, with localized injection reactions (5.6%)²⁰ and fatigue (8%)³⁰ all relatively uncommon. However, acne (37.5%) and menorrhagia (25%) were common complaints.³⁰ 

Lumbar spine bone mineral density (BMD) z scores decreased after treatment with GnRHa monotherapy,^19,29,31 and this reduction was statistically significant in all^29,31 but 1 study.¹⁹ When results were stratified by bone age, the mean reduction in z score was only significant (1.32) for individuals with a bone age <15 years.³¹ Absolute lumbar spine BMD did not change over time, and thus the decrease in z scores after GnRHas likely reflects a failure to accrue BMD compared with age-matched peers. In 2 studies, researchers also examined BMD at the hip and femoral regions, which revealed nonsignificant decreases in absolute and z scores for BMD.^19,31 However, the duration of treatment varied significantly in these studies, being unknown in 1 study³¹ and at least 1 year¹⁹ and 2 years²⁹ in the others.

There was a greater reduction in BMD in transmale adolescents treated with GnRHas than transfemale adolescents. Two studies revealed a significant decrease in absolute and z scores for lumbar spine BMD,^19,31 whereas another study revealed a significant reduction in only z scores.²⁹ In 1 study, researchers quantified the reduction in BMD z scores as being 0.79 for individuals with a bone age <14 years and 0.56 for individuals with bone ages ≥14 years.³¹ Two studies also revealed statistically significant reductions in BMD z scores at the hip and femoral regions in transmale adolescents.^19,31 

Estrogen monotherapy was associated with significant increases in both absolute BMD and z scores in the lumbar spine,^29,31 but not the hip,³¹ of transfemale adolescents previously treated with GnRHas. Furthermore, their z scores after 2 years of estrogen were still below that of age- and birth-assigned sex-matched norms.³¹ Specifically, z scores in the spine were −1.10 and −0.66 in those with younger (<15 years) and older (≥15 years) bone ages, respectively.

Testosterone monotherapy led to a significant increase in both absolute BMD and z scores in the lumbar spine^29,31 and hip³¹ of transmale adolescents, who had previously been on GnRHas. However, their z scores did not reach that of age- and birth-assigned sex-matched controls, aside from the z scores in the hip of individuals with older bone ages. Specifically, z scores in the spine and hip were −0.15 and −0.37, respectively, in those with younger (<15 years old) bone ages and −0.06 and 0.02, respectively, in those with older (≥15 years old) bone ages.

Growth velocity decreased during treatment with GnRHas²⁹ in all transgender youth compared with pubertal-matched peers.²¹ In particular, younger individuals, who had greater growth potential, had significantly lower height standard deviation scores (SDSs) after treatment.^21,29 One study revealed significantly lower height standardized values for transfemale adolescents only.¹⁹ No researchers have examined whether individuals given GnRHas achieved their predicted final height after GAHs. After 1 year on GnRHas, individuals had a significant increase in body fat percentage²⁹ and BMI,^19,21 which was accompanied by a decrease in lean body mass.²⁹ 

Lynestrenol resulted in significant increases in weight and BMI absolute and z scores during the first 6 months with a return to baseline after 12 months of treatment.³⁰ 

Cyproterone acetate resulted in a decrease in growth velocity compared with age-matched peers,²² with a final height after 12 months of treatment also being significantly lower than age-matched peers with a mean standardized score of −0.309. There were no clinically significant changes in body weight and BMI after 12 months.

It is unclear whether differences in the pubertal stage and bone age at which estrogen was commenced contributed to the variable outcomes found because these data were not collected.²⁹ Estrogen in combination with cyproterone acetate resulted in reduced growth compared with age-matched peers in 1 study,²² whereas another study revealed no change in growth velocity after estrogen.²⁹ Total BMI was significantly increased after estrogen in 1 study,^22,32 although another revealed that total BMI did not change after 6 months.³² 

Testosterone monotherapy resulted in increased growth velocity compared with age-matched peers in 1 study,²⁹ but the impact on final height (nor the age and pubertal stage at commencement) was not specified. Testosterone was also associated with weight gain,³⁰ resulting in a significantly raised absolute BMI^20,30 from an average baseline of 20.7 to 22.4 within 6 months.²⁰ This increase in BMI was less than that of age-matched male adolescents.³² 

After 1 year of GnRHa, there were no changes in carbohydrate or lipid metabolism as measured by fasting glucose, insulin, cholesterol, low-density lipoprotein (LDL), and high-density lipoprotein (HDL) levels.²⁹ In 1 study, researchers observed that alkaline phosphatase (ALP) was decreased as a likely secondary result of decreased bone turnover, whereas all other liver enzymes were unchanged.²¹ In this same study, researchers also reported lower levels of creatinine and hypothesized that this might be due to reduced muscle mass but found that there was no correlation between change in muscle mass and creatinine.²¹ 

Progestins were associated with an adverse lipid profile, with a significant decrease in HDL cholesterol by an average of 0.46 mmol/L and an elevation of LDL cholesterol by 0.37 mmol/L after 1 year.³⁰ There were no significant changes in hemoglobin A1c (HbA1c), glucose levels, insulin levels, or homeostasis model assessment (HOMA) index.³⁰ Alanine aminotransferase (ALT) increased after 12 months, but this was not clinically significant.³⁰ Mean hemoglobin (Hb) and hematocrit (Hct) levels increased in the first 6 months and subsequently remained stable.³⁰ 

Cyproterone acetate was associated with a significant reduction in only triglycerides, but total cholesterol, LDL cholesterol, HDL cholesterol, HbA1c, glucose, insulin, and HOMA index were unaffected.²² There was no change in liver enzymes or thyrotropin.²² There was a slight decrease in Hb and Hct after 12 months, but this was not clinically significant.²² 

Apart from 1 study in which a significant decrease in HDL after 4 to 6 months was observed,³² estrogen had no effect on carbohydrate and lipid metabolism.^22,29,32 Similarly, no significant changes in liver enzymes,^22,32 thyrotropin, or free thyroxine (fT4)²² were noted with estrogen treatment. A significant increase in serum creatinine was seen after 12 months with combined estrogen and cyproterone acetate treatment.²²Hb and Hct were found to have decreased initially but returned to baseline after approximately 6–12 months.³². However, when estrogen was used in combination with a progestin, Hb and Hct levels did not change any further after 12 months.²² Blood pressure (BP) was unchanged after 6 months of estrogen.³² 

Testosterone had no significant effect on carbohydrate and lipid metabolism.^29,30,32 Although in 1 study researchers observed raised liver enzymes (aspartate aminotransferase [AST] and ALT) after a year,³⁰ another study revealed no significant change in AST and a decrease in ALT after 4 to 6 months.³² Testosterone treatment decreased thyrotropin and fT4 to be outside of the normal reference ranges, although these changes were not clinically relevant because there was no clinical or biochemical hypo- or hyperthyroidism in participants.³⁰ Serum creatinine increased after 6 months of testosterone with no subsequent change thereafter and was thought to reflect an increase in muscle mass.³⁰ Hb²⁰ and Hct³⁰ were increased after 6 months but remained stable during the next 6 months within the male reference ranges,^20,30 whereas another study revealed no significant change in these parameters at any stage.³² Systolic BP was elevated in treated individuals, with an average rise of 5 mm Hg after 6 months.²⁰ 

All relevant results are shown in Table 3.

GnRHa treatment was associated with significant improvements in multiple psychological measures, including global functioning,^25,–27 depression,^26,27 and overall behavioral and/or emotional problems.^26,27 The effects of GnRHas on anger and anxiety remain unclear with conflicting results.^26,27 Moreover, GnRHa treatment had no significant effect on symptoms of GD,^26,27 with researchers in 1 study observing a nonsignificant increase in GD and body image difficulties.²⁶ 

Critically, no researchers have examined the psychosocial effects of these hormonal therapy types in transgender youth.

All relevant results are shown in Table 4.

In one study, researchers examined the effect of GnRHas on executive functioning using the Tower of London test, which is used to assess mental planning ability.²⁴ After GnRHa treatment, there was significantly reduced accuracy in transfemale adolescents.²⁴ There was also exaggerated regional brain activation typical of birth-assigned sex on functional magnetic resonance imaging (fMRI).²⁴ However, given the small sample size (8 participants), these results should be interpreted cautiously.

In another study, researchers examined the effect of GnRHa treatment on mental rotation in transmale adolescents,²³ exploring whether rotated pairs of three-dimensional shapes were identical images of each other.^35,36 Because men significantly perform better on this task compared with women, this result has also previously been suggested as evidence for the classic theory of the organizational and activational effects of sex hormones on the brain.^37,38 Interestingly, GnRHa suppression in transmale adolescents was associated with male brain activation patterns, with reduced activity in the right frontal area.²³ 

No researchers have examined the cognitive effects of these treatments.

In the same study, researchers also examined the effects of testosterone in transmale adolescents on mental rotation tasks, in which they observed moderate to strong improvements in accuracy and reaction time.²³ Similar to control boys, treated transmale adolescents also demonstrated increased activation of brain regions implicated in mental rotation on fMRI.²³ 

This is the first systematic review of the effects of hormonal treatment in transgender youth; authors of previous systematic reviews in this field, including those commissioned by the recent Endocrine Society Clinical Practice Guidelines, focused on the use of GAHs in adults.^8,–11,15 

GnRHas successfully suppressed endogenous puberty, consistent with the primary objective of this treatment, although there was only a single study in which researchers actually recorded these data.²⁹ GnRHas were observed to be associated with significant improvements in global functioning,^25,–27 depression,^26,27 and overall behavioral and/or emotional problems^26,27 but had no significant effect on symptoms of GD. The latter is probably not surprising, because GnRHas cannot be expected to lessen the dislike of existing physical sex characteristics associated with an individual’s birth-assigned sex nor satisfy their desire for the physical sex characteristics of their preferred gender. Like GnRHas, the antiandrogen cyproterone acetate effectively suppressed testosterone in transfemale adolescents,²² but its potential psychosocial benefits remain unclear. Meanwhile, GAHs increased estrogen and testosterone levels and thus induced feminization and masculinization, respectively, of secondary sex characteristics.^22,29 However, in the case of breast development, the outcomes were subjectively less in size than expected in the majority of recipients,²² and the potential psychosocial benefits of GAHs remain unknown. Finally, although the use of the progestin (lynestrenol) has been studied in transmale adolescents,³⁰ its effects were predominantly examined in the context of potential adverse effects, so the therapeutic impact of progestins for menses suppression and psychosocial outcomes cannot be understood from the current literature.

Overall, hormonal treatments for transgender youth were observed to be relatively safe but not without potential adverse effects. For GnRHas, a significant concern in clinical practice is their potential effects on BMD accrual; their use was associated with a significant reduction in BMD,^19,29,31 which appeared to be worse for transmale adolescents^19,31 and is consistent with previous studies of nontransgender youth^39,40 and adults⁴¹ who received GnRHas. However, given the relatively short follow-up duration of the studies reviewed here, it will be important for future researchers to better establish if this reduction in bone density is long-lasting or transient, as observed in nontransgender youth after GnRHa cessation.^39,40 It is notable that BMD increased after estrogen and testosterone, which suggests potential compensation by GAHs. However, for estrogen treatment, the BMD of those who had previously received GnRHas still remained lower than age-matched peers 2 years after estrogen treatment,³¹ so compensation may only be partial. Furthermore, there is a lack of reporting of pubertal stage at treatment commencement, which makes interpretation of some changes difficult, especially BMD.

Clinically, patients who receive GnRHas and still have significant growth potential are counseled about the risk of the treatment affecting their final height. Although researchers in 2 studies have now examined growth and height characteristics in transgender youth receiving GnRHas,^21,29 their relatively short follow-up times (≤3 years) precluded determination of the effects of GnRHas on final height, and future researchers should address this knowledge gap. Another clinical concern in the use of GnRHas is the induction of menopausal-like symptoms due to the withdrawal of sex steroids, especially in postpubertal individuals. GnRHas were commonly observed to cause hot flashes in transmale adolescents in late puberty, but these decreased in frequency over time.²⁹ For potentially similar reasons, one of the main complaints after cyproterone acetate administration in transmale adolescents was fatigue.

Hormonal treatment of transgender adults is known to be associated with various metabolic and cardiovascular effects.^10,–12 GnRHas significantly increased both body fat percentage²⁹ and BMI^19,21 while decreasing lean body mass.²⁹ Similarly, testosterone significantly increased both body fat and BMI.^20,29 Although lynestrenol also increased BMI, this was transient, with BMI returning to baseline after 12 months.³⁰ Cyproterone acetate was not associated with any changes in BMI.²² In terms of lipid metabolism, neither testosterone nor estrogen had any observable impact, but lynestrenol was associated with lower HDL and higher LDL cholesterol after 1 year,³⁰ whereas cyproterone acetate significantly reduced triglycerides.²² 

The findings from this review are subject to limitations. Firstly, the current literature has a limited number of studies in which the different hormonal treatments in transgender youth is examined. Secondly, for any given class of hormonal treatments, there is a variety of different agents, formulations, and administration routes that are being used clinically in transgender youth. For example, the physical effects of 1 antiandrogen and 1 progestin have been studied in only 1 study each, with no confirmation of results or further exploration. Thirdly, in existing studies there is a medium to high risk of bias, given small sample sizes, retrospective nature, and lack of long-term follow-up. In this regard, although randomized controlled trials are often considered gold standard evidence for judging clinical interventions, it should be noted that, in the context of GD in which current guidelines highlight the important role of hormonal treatments,¹⁵ conducting such trials would raise significant ethical and feasibility concerns. Fourthly, authors of existing studies have neglected several key outcomes. These include the following: psychological symptoms related to GD, which is a critical knowledge gap given the high rates of mental health problems observed in transgender youth and justification of these treatments as treating GD⁴²; the impact of hormonal treatments on fertility, which is an integral part of the counseling recommended by current guidelines¹⁵; and potential adverse effects such as arterial hypertension, which was reported in a recent case series in association with GnRHas.⁴³ Finally, there are no known studies to date in which researchers have reported the rates and circumstances under which transgender youth cease their hormonal therapy in an unplanned manner or the risk of subsequent regret, which would be of great clinical utility.

Notwithstanding these limitations, collectively, the studies reviewed provide qualified support for the use of GnRHas, GAHs, cyproterone acetate and, to a lesser extent, lynestrenol in transgender youth. Overall, these hormonal treatments appear to provide some therapeutic benefits in terms of physical effects and are generally well-tolerated on the basis of current evidence.

Looking ahead, it will be essential for future researchers to reassess and expand on the findings of the existing studies. Large, prospective longitudinal studies, such as have been recently initiated,⁴⁴ with sufficient follow-up time and statistical power and the inclusion of well-matched controls will be important, as will the inclusion of outcome measures that investigate beyond the physical manifestations.

ALT

alanine aminotransferase

AST

aspartate aminotransferase

BMD

bone mineral density

BP

blood pressure

fMRI

functional magnetic resonance imaging

FSH

follicle-stimulating hormone

fT

free testosterone

fT4

free thyroxine

GAH

gender-affirming hormone

GD

gender dysphoria

GID

gender identity disorder

GnRHa

gonadotropin-releasing hormone analog

Hb

hemoglobin

HbA1c

hemoglobin A1c

Hct

hematocrit

HDL

high-density lipoprotein

HOMA

homeostasis model assessment

LDL

low-density lipoprotein

LH

luteinizing hormone

QUIPS

Quality in Prognosis Studies

SDS

standard deviation score

SHBG

sex-hormone binding globulin