Feminising Hormone Therapy

Evidence Review · ER-005

1. Introduction — a common first medical step, for which consent rests on knowing what it can and cannot do

Feminising hormone therapy is the use of medication to shift a person’s hormonal profile from one dominated by testosterone to one dominated by oestrogen, in order to produce feminine physical characteristics. It is a common first medical step for trans women and for some non-binary people assigned male at birth who seek bodily change. Gender-affirming treatment is uncommon at the population level — in the Netherlands in 2015, an estimated 1 in 2,800 people assigned male at birth aged 16 or over had received medical treatment for gender dysphoria, up from roughly 1 in 11,900 in 1990 by an earlier estimate the authors cite — and prevalence estimates vary by country and method.^[32] [Netherlands-specific; counts living people treated at the clinic, not everyone who is trans. The paper’s abstract gives 1 in 3,800, which its own results and table contradict; 1 in 2,800 is the tabulated figure.]

1.1 Why hormone therapy is used — to reduce distress where it is present, though it is sought either way

For some people, the incongruence between their gender and the sex characteristics their body developed at puberty is associated with clinically significant distress. ICD-11 uses the term gender incongruence, which does not require distress for diagnosis; DSM-5-TR gender dysphoria refers specifically to distress or impairment associated with that incongruence. Not everyone with gender incongruence experiences dysphoria, and FHT may be sought either way.^[1] Where distress is present, FHT aims to reduce it by changing the body so it better matches the person’s gender.

1.2 Goals of treatment — suppress testosterone-driven features and induce feminine ones, to ranges whose optimum is not established

FHT works in two directions: it suppresses testosterone-driven features (spontaneous erections, oilier skin, ongoing male-pattern changes) and induces feminine features (breast tissue, a more feminine fat distribution, softer skin). The Endocrine Society clinical practice guideline (2017) frames the treatment target as oestradiol and testosterone in the ranges typical of premenopausal women; WPATH Standards of Care, eighth version (SOC-8, 2022) uses the same numerical ranges while noting that optimal target ranges have not been established.^[1,2]

1.3 Gender dysphoria and the decision to treat — assessment has become individualised and needs-based

SOC-8 requires that gender incongruence be marked and sustained before hormone therapy is recommended, and requires a formal diagnosis only where one is needed to access care.^[1] UCSF guidance adds that primary care providers experienced with transgender patients may initiate hormones under an informed-consent model.^[18] This reflects a shift towards individualised, needs-based assessment.

1.4 Informed consent and shared decision-making — some effects persist, so understanding comes before starting

Because some effects of FHT are persistent, particularly breast development, and because fertility can be substantially suppressed with the extent and timing of any recovery uncertain, good practice centres on informed consent: understanding the expected effects, their timelines, the irreversible elements, the realistic limits, and the known risks before starting. Choosing the regimen and route together with a clinician, in light of goals and health profile, is the model endorsed by current guidelines.^[1,2] Fertility preservation is discussed before treatment (see Section 6).

What this section establishes. Feminising hormone therapy is a common first medical step, sought with or without distress. Because breast development persists and fertility can be substantially suppressed, with the extent and timing of any recovery uncertain, good practice centres on informed consent before starting.^[1,2]

2. How feminising hormone therapy works — oestradiol acts on soft tissue, and lowering testosterone is central

2.1 Oestrogen — 17β-oestradiol changes oestrogen-responsive tissue, but not the larynx

The active oestrogen used in FHT is 17β-oestradiol — the bio-identical hormone produced by the ovaries. SOC-8 notes that conjugated oestrogens have been abandoned in favour of bio-identical oestrogens; both guidelines advise against ethinylestradiol, and the Endocrine Society describes synthetic and conjugated oestrogens as undesirable.^[1,2] It acts mainly by binding oestrogen receptors (ERα and ERβ) inside target cells, where the hormone–receptor complex switches specific genes on or off. Oestrogen-responsive tissues such as the breast and adipose tissue change during adult treatment, but oestradiol does not reverse the laryngeal enlargement and vocal-fold changes established by an androgen-driven puberty. Breast development during feminising hormone therapy draws on the same oestrogen-responsive mammary tissue biology that drives pubertal breast growth, although the developmental context is different.^[21] [Population flag: cisgender developmental data.]

Oestradiol does two jobs. First, it directly stimulates feminine tissue changes. Second, it exerts negative feedback on the hypothalamus and pituitary, reducing gonadotrophin release: lower LH reduces testosterone production by the Leydig cells of the testes, and lower gonadotrophin stimulation, including of FSH, contributes to impaired sperm production. Oestradiol can suppress testosterone substantially, and in some people sufficiently; many people, however, need an additional testosterone-lowering medication to reach the female range, depending on dose, route, adherence and the target used.

2.2 Testosterone suppression — central, though higher oestradiol has not been shown to add feminisation once it is achieved

Many features people most want to change — oilier skin, spontaneous erections, ongoing male-pattern changes — are driven by testosterone and its more potent derivative dihydrotestosterone (DHT). Lowering testosterone is therefore central. UCSF guidance (2016) states that, once testosterone is adequately suppressed, there is no evidence that pushing oestradiol higher adds feminisation or breast growth.^[18] The data on oestradiol level and breast growth are mixed: in a large prospective multicentre cohort, serum oestradiol did not predict breast development at one year,^[3] whereas an exploratory analysis of a randomised trial found greater estimated breast volume at six months with higher achieved serum oestradiol (0.11 mL per 1 pmol/L, 95% CI 0.04 to 0.18) and with older age.^[49] Neither was designed to test oestradiol deliberately targeted above the usual range. A 2025 systematic review, searching to April 2023, found that the guideline range of 100–200 pg/mL lacks a primary evidence base and has not been shown to optimise feminisation or reduce adverse events.^[30] That does not mean that concentrations below and above it are interchangeable or equally safe; avoiding unnecessary exposure remains a precautionary clinical principle rather than an evidence-derived threshold.

2.3 Lowering testosterone — reduce its production, block its receptor or block its conversion to DHT

• Reduce production via pituitary suppression — a GnRH analogue, oestradiol’s own suppressive effect, or cyproterone acetate (also antigonadotrophic).
• Block the androgen receptor — spironolactone (a competitive androgen-receptor antagonist) and, partly, cyproterone (which combines central gonadotrophin suppression with peripheral receptor blockade).
• Block conversion to DHT — a 5α-reductase inhibitor (finasteride or dutasteride). This narrows testosterone’s action rather than lowering it, and its role in FHT is limited and poorly evidenced (Section 3).

Anti-androgens reduce testosterone production or block its effects. Bilateral orchidectomy removes the main source of testosterone, so anti-androgen treatment for this purpose can usually be stopped.^[39] Medication for a separate indication, such as persistent scalp hair loss, may still be appropriate.^[18]

Avoiding prolonged sex-hormone deficiency is important for bone health; the regimen should be reviewed individually rather than reduced automatically. UCSF guidance notes that some people choose a lower dose, which is appropriate if it maintains bone density, and that stopping around age 50 after gonadectomy may be considered through shared decision-making.^[18] The evidence on stopping in later life is discussed in Section 9. The Endocrine Society recommends bone-density measurement when osteoporosis risk factors are present, specifically including people who stop sex-hormone therapy after gonadectomy.^[2]

2.4 Time course — a fairly consistent order, with a magnitude that varies

Feminisation is slow and follows a fairly consistent order, though magnitude varies. The earliest changes (1–3 months) are reduced libido and fewer spontaneous erections. Softer, less oily skin and the start of fat redistribution follow over 3–6 months. Breast growth begins within the first few months, is often greatest during the first year, and may continue more gradually thereafter; fat redistribution continues for two to three years in the Endocrine Society table and up to five in SOC-8. Lean mass and muscle size fall, with guideline tables citing a maximum at one to two years; measured strength changes less consistently (Section 4). Body hair begins to change at 6–12 months and continues to change beyond three years, but hormones alone do not usually remove established facial hair. Oestradiol does not reverse the voice changes or the skeletal proportions established by an androgen-driven puberty, although sex-steroid therapy does affect bone turnover (Section 7); expectations differ where skeletal maturation or pubertal history differs. The detailed timeline is set out as a sidebar. An important caveat (Section 10): this widely reproduced timeline derives largely from expert consensus and clinical experience, not rigorous prospective measurement — breast development is the main outcome that has been measured carefully.^[1,2,3]

What this section establishes. Lowering testosterone is central, and once it is adequately suppressed, UCSF guidance states that there is no evidence higher oestradiol adds feminisation.^[18] Oestradiol changes oestrogen-responsive tissue but does not reverse the larynx or skeleton shaped by an androgen-driven puberty. Changes follow a fairly consistent order, but their size varies.

3. Medications — each option trades its effect against its own set of risks

This section is descriptive, not a recommendation of any specific drug or dose.

3.1 Oral oestradiol — inexpensive and effective, with higher clot risk than transdermal in cisgender data

Evidence. Oral 17β-oestradiol (often as oestradiol valerate) effectively raises oestradiol levels; its feminising effects are well established.

Advantages. Inexpensive, convenient, easy to titrate.

Disadvantages. Oral oestrogen passes first through the liver, which is thought to raise clotting-factor production more than non-oral routes; in cisgender postmenopausal women, oral oestrogen carries higher venous thromboembolism (VTE) risk than transdermal.^[19,31] [Best established in cisgender women.] This is separate from the very high VTE risk historically seen with ethinylestradiol, a synthetic oestrogen far more thrombogenic than 17β-oestradiol and no longer recommended for FHT.^[2,31]

3.2 Transdermal oestradiol — favoured where clot risk is higher, on evidence mainly from cisgender women

Evidence. Achieves physiological oestradiol levels and avoids first-pass liver metabolism.

Advantages. Avoids first-pass liver metabolism;^[18] in cisgender postmenopausal women, transdermal oestrogen shows little or no increase in VTE risk.^[19,46] [Population flag.] Transdermal is generally favoured where clot risk is higher, including in smokers and people with cardiovascular risk factors,^[18] and was recommended in an early Amsterdam series particularly for people over 40.^[10] In trans women, a prospective study found that thrombin generation rose less over 12 months with transdermal than with oral oestradiol (both combined with cyproterone), although levels at 12 months did not differ significantly after adjustment for baseline;^[47] the studies included in a systematic review searched to July 2022 did not compare clinical VTE outcomes by route.^[22] The comparison therefore still rests largely on cisgender data and biological plausibility.

Disadvantages. Skin irritation; patches can be visible or detach.

3.3 Injectable oestradiol — peaks and troughs make levels harder to read, and no comparison of their health consequences was identified

UK context. Injectable oestradiol esters are used in some countries (notably the United States) but are not a standard NHS option and are far less common in UK practice. [Practice context; not drawn from a cited source.] Injected oestradiol esters can produce substantial variation between peak and trough, depending on the ester, dose and interval, which makes levels harder to interpret. The Endocrine Society notes that there are no data showing that this periodicity is harmful,^[2] but comparative evidence on the health consequences of different fluctuation patterns is lacking. UCSF guidance notes that a mid-cycle level is often sufficient, and that peak and trough levels may reveal wide fluctuation where symptoms are cyclical.^[18] The time since the last injection should be recorded with any result.

3.4 GnRH analogues — suppression after an initial stimulation; cyproterone’s harms are not established for them, though long-term data are lacking

GnRH agonists are long-acting analogues that, with continued exposure, desensitise the pituitary’s GnRH receptors and so suppress gonadotrophins and testicular testosterone; the Endocrine Society describes an initial increase in gonadotrophins for about ten days after the first injection (and, to a lesser degree, the second) before suppression.^[2] GnRH antagonists suppress without that initial stimulation. The Endocrine Society cites a study in which goserelin with oestrogen reduced testosterone with a low incidence of adverse reactions.^[2] GnRH agonists are commonly used in the UK.^[1] The adverse-effect pattern described for cyproterone (prolactin, HDL and meningioma) has not been established for them, but SOC-8 notes that no long-term studies have reported on meningioma or prolactinoma risk in trans women taking them,^[1] and the absence of a reported signal does not establish the absence of risk. Considerations include cost, the need for injection, and the fact that GnRH analogues can cause osteoporosis if the oestrogen given alongside is insufficient.^[1]

3.5 Spironolactone — an inconsistent effect on testosterone, with potassium and kidney function to watch

Spironolactone is principally a mineralocorticoid-receptor antagonist whose antiandrogenic effects have been used for feminisation since the 1980s. It antagonises the androgen receptor, partially inhibits the enzymes 17α-hydroxylase and 17,20-lyase in testosterone synthesis, and may raise serum oestradiol and oestrone; it does not appear to suppress the hypothalamic-pituitary-gonadal axis, and in one small study it transiently raised LH. Which of these contributes most to feminisation is not established, and its effect on circulating testosterone is inconsistent, so serum testosterone captures its antiandrogenic effect only partly.^[28] It is commonly prescribed in the United States, where cyproterone is not licensed.^[28] In a single-blinded 12-week randomised comparison of one dose pair, spironolactone 100 mg daily produced less biochemical testosterone suppression than cyproterone 25 mg daily: according to the trial abstract, about 19% of the spironolactone group reached the female testosterone range (<50 ng/dL) versus about 90% on cyproterone. The trial did not compare other doses or overall feminisation.^[15] Its main monitoring concerns are raised potassium (hyperkalaemia), kidney function and low blood pressure; as a diuretic it can also cause dehydration and postural symptoms. A systematic review of four retrospective studies found that, while several anti-androgens may lower testosterone more than spironolactone, no studies adequately compared them on the outcomes patients actually care about — breast development and hair — so the most feminising anti-androgen remains unknown; it also cautions that testosterone level is an imperfect surrogate, because serum testosterone does not capture all the antiandrogenic effects of receptor-blocking regimens.^[28] A later double-blind randomised trial addressed this directly in 63 people starting oestradiol. At six months, 22 of 32 participants allocated to cyproterone 12.5 mg daily and 14 of 31 allocated to spironolactone 100 mg daily had testosterone below 2 nmol/L (68.8% versus 45.2%). The reported odds ratio of 9.01 (95% CI 1.83 to 44.44) comes from a covariate-adjusted Firth penalised logistic regression; the crude proportions alone would give an odds ratio of about 2.7. No statistically significant difference was detected in breast–chest distance (mean difference 0.27 cm, −0.82 to 1.35) or estimated breast volume (17.26 mL, −16.94 to 51.47); both intervals are wide, and the trial does not establish equivalence. Prolactin was higher on cyproterone. Two participants in the cyproterone arm discontinued the intervention because of worsening mental state; in the spironolactone arm one discontinued because feminisation was occurring “too fast”. Mean depression scores (PHQ-9) did not differ significantly between groups. The trial was short and the authors note it may have been underpowered for breast volume; it does not show that the more suppressive drug produces more breast development over six months.^[49] [The methods describe adjustment for age, BMI and baseline testosterone; the Table 2 footnote says age and BMI. The trial abstract prints the odds-ratio interval as 1.83 to 4.44; Table 2 gives 1.83 to 44.44, which contains the estimate and is used here. Two discontinuations cannot establish that cyproterone caused the change in mental state.]

3.6 Cyproterone acetate — stronger suppression, with liver, clotting, mood and cumulative-dose meningioma warnings

Cyproterone acetate (CPA) is a potent anti-androgen and antigonadotrophin, favoured in many European countries;^[28] in a randomised comparison it suppressed testosterone more effectively than spironolactone.^[15] Its harms need separate assessment. UK product information records direct hepatic toxicity, including jaundice, hepatitis and hepatic failure, with fatal cases at dosages of 100 mg and above (most in men with advanced prostate cancer), and requires liver function tests before and regularly during treatment; it also lists thromboembolic events (though a causal relationship has not been established), temporary depressive moods, deterioration in some people with severe chronic depression, and fatigue.^[65] [Product information for the 50 mg tablet, licensed for prostate cancer; feminising use is off-label. Lower doses have not been shown to be free of serious hepatotoxicity.] Other recognised harms are raised prolactin (typically reversible on stopping, and smaller at 10 mg daily than at higher doses),^[28,50] reduced HDL cholesterol at doses above 10 mg,^[16,50] and an association between higher-dose and higher-cumulative-dose use and intracranial meningioma.^[24,28] These harms differ in kind and frequency, and cannot be ranked reliably by a single severity label. In the randomised comparison with spironolactone, the serious adverse effects reported in the cyproterone arm, as stated in the trial abstract, were drug-induced liver injury and asymptomatic hyperprolactinaemia.^[15] A commentary on the later six-month trial argued that the absence of a breast-development benefit, together with known harms, makes cyproterone less preferable for most patients.^[63] In a prospective multicentre cohort of 882 trans women also taking oestradiol, modelled mean testosterone was below 2 nmol/L at every dose from 10 to 100 mg daily, but not every individual was suppressed: of the four people who started on 10 mg, one had unsuppressed testosterone at three months and still did at twelve despite an increase to 50 mg, and some who started on 25 mg remained above the reference range after an increase to 50 mg. Doses changed during follow-up, the authors note that the 10 mg group was too small to detect differences and that indication bias was possible, and mean liver enzymes, which did not differ between doses and stayed within the reference range, cannot exclude rare serious liver injury.^[50] For meningioma the dose–response is steep. In a large French cohort of girls and women prescribed high-dose cyproterone (25 or 50 mg daily), meningioma treated by surgery or radiotherapy was more frequent in those who received at least 3 g in the first six months than in a very slightly exposed control group who received less than 3 g (adjusted hazard ratio 6.6, 95% CI 4.0 to 11.1), and risk rose with cumulative dose, to hazard ratios of 11.3 (5.8 to 22.2) at 36–60 g and 21.7 (10.8 to 43.5) at 60 g or more. Below 12 g the hazard ratios were not significantly different from 1, but the intervals were wide (at 6–12 g, 2.2, 0.9 to 5.6); this does not establish a safe cumulative threshold, and these estimates do not directly quantify risk on the lower doses now used in FHT.^[24] [Population flag: the main cohort comprised girls and women aged 7–70 on high-dose cyproterone; transgender participants were analysed separately. A systematic review of 12 case reports in trans women, mostly at 50–100 mg daily, led its authors to infer a growth-promoting role, but case reports cannot establish causation or define absolute risk.^[25] The French study’s separate transgender analysis found 3 meningiomas in 14,460 person-years (20.7 per 100,000) against none in unexposed transgender controls (P = 0.08), all at 100–150 mg daily over 3–4.5 years; the authors described this as a high risk, and the numbers are too small to estimate it precisely.^[24]] In 2020 cyproterone at daily doses of 10 mg or more was restricted in its androgen-dependent skin and hair indications, and its use to reduce sex drive in men was limited to when other options are unsuitable (use in prostate cancer was unchanged), following a European review, initiated at France’s request, that included the French cohort above: such doses should be used for hirsutism, androgenic alopecia, acne and seborrhoea only after other options including lower-dose cyproterone have failed, and reduced to the lowest effective dose once working. The EMA puts the absolute risk of meningioma at between 1 and 10 in 10,000 people depending on dose and duration, reported primarily at 25 mg daily and above and mostly after several years of exposure. Clinicians should monitor for symptoms of meningioma, and if one is diagnosed all cyproterone-containing products must be stopped permanently; a meningioma or history of meningioma is a contraindication.^[17] The MHRA advises that cyproterone should not be used for any indication in people with a meningioma or a history of one, and describes the risk as thought to be rare overall but highest at doses of 25 mg a day and above.^[64] Neither the EMA frequency band nor these dose bands is a safe threshold for current low-dose feminising use. [These are restrictions within cyproterone’s authorised indications; the review did not address transgender prescribing, which is off-label in this respect.]

3.7 Finasteride and dutasteride — a limited and poorly evidenced role, with psychiatric and sexual warnings

These block conversion of testosterone to the more potent DHT. They are sometimes added (for example, for scalp hair) but their role in FHT is limited and poorly evidenced: they do not lower testosterone itself, and once testosterone is well suppressed their added benefit is unclear. This review identified no trial evidence for their use specifically in trans women, and did not conduct a systematic search for it; the companion Evidence Note on finasteride examines the question directly.

Both carry psychiatric and sexual adverse-effect warnings.^[66,67] In 2025 the European Medicines Agency confirmed suicidal ideation as a side effect of finasteride 1 mg and 5 mg tablets, with a frequency that cannot be estimated from the available data; it found insufficient evidence to establish a causal association for dutasteride, whose product information was updated as a precaution because of a possible class effect.^[66] The MHRA’s 2026 advice states that finasteride is associated with depression, suicidal ideation and sexual dysfunction that may persist after treatment is stopped; it asks prescribers to ask about any history of depression or suicidal ideation and to review patients regularly for psychiatric and sexual side effects, and advises that people taking finasteride 1 mg who develop suicidal thoughts or depression stop it and seek advice, and that those taking finasteride 5 mg or dutasteride seek advice as soon as possible.^[67] Reports of these effects do not establish how often they occur.

3.8 Progesterone — a benefit not established, and a clot signal resting on very few events

Progesterone is one of the most contested questions in FHT and has its own companion review. Three different questions are often run together.

On breast development, WPATH SOC-8 reviewed the evidence and was unable to find quality evidence that progestogens improve breast or areolar development.^[1] A 2022 prospective case–control study added 100 mg oral micronised progesterone to established oestradiol therapy in 23 people, against 19 continuing standard care, and found no statistically significant difference in self-reported Tanner stage at three months (P = 0.42) — though both groups were already at Tanner 3.3 at baseline after some two years of hormone therapy, Tanner stage is a coarse endpoint for this purpose, and the study was powered for sleep rather than breast development. Participants self-selected by requesting progesterone, which, the authors suggest while calling it speculative, would if anything have biased the study towards finding a benefit.^[36] An open-label randomised trial from Amsterdam has measured breast volume by 3D imaging. Its design is published;^[13] its results have so far been reported at conferences, in an investigator-authored abstract and in the centre’s institutional announcement, not in a peer-reviewed paper.^[14,41] The abstract reports estimates for each arm referenced to the control group on baseline oestradiol, in per cent: 13 (95% CI −7 to 33) with 200 mg progesterone, 20 (0 to 40) with 400 mg, 6 (−13 to 24) with doubled oestradiol alone, 37 (18 to 57) with doubled oestradiol plus 200 mg progesterone, and 27 (8 to 47) with doubled oestradiol plus 400 mg.^[41] The abstract’s wording (“compared with group E”) most plausibly denotes differences from the control group rather than growth from baseline, but it does not say so explicitly, nor whether the figures are percentage-point differences in percentage volume change or another contrast; the centre’s announcement described increases of up to 30%, which does not reconcile cleanly with these figures.^[14,41] The units, model, time point, analysis population, missing data and handling of multiple comparisons await the full report. The larger estimates in the combined arms do not by themselves establish an interaction between progesterone and oestradiol. Side effects included tiredness (44%), breast tenderness (27%) and mood changes (22%).^[41] The trial was unblinded and placebo-free, with about 15 people per arm, all after orchidectomy or vaginoplasty, at 200–400 mg daily, and its authors described it as able to serve as a pilot.^[13] No peer-reviewed outcome paper had been identified as of September 2026, and none was found in a targeted web search on 12 October 2026; in September 2025 the centre said it was now able to prescribe progesterone, in a trial setting, to people who had taken oestradiol for at least a year.^[14] A separate placebo-controlled trial of 200 mg micronised progesterone, NCT04534881, was terminated for lack of recruitment after enrolling five participants (three on progesterone, two on placebo). Its only registered outcome, and only posted result, is the change in testosterone at 24 weeks (mean +1.5 ng/dL on progesterone, −6.4 ng/dL on placebo); no breast-development results were posted, and no serious or other adverse events were reported in either arm. Its size does not resolve whether progesterone improves breast development.^[44]

On sleep and psychological distress, the same 2022 study found no statistically significant difference in sleep quality (Pittsburgh Sleep Quality Index, P = 0.35) or psychological distress (Kessler K10, P = 0.64) over three months, with one participant deteriorating significantly and improving after stopping progesterone.^[36]

On safety, a 2019 review concluded that data in both cisgender and transgender groups suggest added VTE risk with progestins.^[31] In a US clinic cohort of 2,509 trans women with 19 thrombotic events (a count that included arterial and superficial events), a recent progestin prescription was associated with higher odds of an event (adjusted odds ratio 2.95, 95% CI 1.02 to 8.57). The estimate for medroxyprogesterone acetate was much larger (20.0, 2.14 to 187.2) but rested on one or two events (the paper’s text and tables give different counts), and the estimate for micronised progesterone was imprecise (2.27, 0.72 to 7.16). The authors suggested that the association may be stronger for medroxyprogesterone acetate than for progesterone.^[51] Neither source isolates micronised progesterone at the doses now being trialled.

The current position: for breast development, a benefit of progesterone is plausible and under investigation, but not established by peer-reviewed outcome data. For sleep and psychological distress, the one controlled study identified, of 100 mg for three months, detected no between-group difference; it does not exclude effects at other doses, over longer treatment or in other populations. For safety, a signal of added thrombotic risk with progestins, possibly stronger with medroxyprogesterone acetate, rests on a mixed thrombotic endpoint with very few events; it does not establish a risk, or its absence, specific to micronised progesterone.

See companion review: Progesterone in Feminising Hormone Therapy.

3.9 Formulation and interaction cautions — avoid ethinylestradiol, and know what distorts a level

• Avoid ethinylestradiol and conjugated equine oestrogens for FHT; use 17β-oestradiol, which is less thrombogenic and whose serum level can be measured directly, although interpretation still depends on the assay, route and sampling time.^[2,31]
• Oral versus sublingual oestradiol: taking oestradiol sublingually produces higher, shorter-lived peaks, which makes level-based monitoring less reliable;^[2,18] UCSF guidance describes only weak evidence that sublingual use might reduce clot risk, and it has not been shown to share the lower clot risk associated with transdermal use;^[18] route should be agreed with a clinician.
• Biotin supplements (high-dose, not ordinary dietary amounts) can interfere with susceptible hormone immunoassays and produce misleading results. Interference and the appropriate withholding interval are assay- and platform-dependent, so the testing laboratory’s guidance should be followed rather than a general rule. [General laboratory practice; no trans-specific source.]
• Spironolactone interactions: UK product information warns that use with other potassium-sparing diuretics, ACE inhibitors, angiotensin II antagonists, non-steroidal anti-inflammatory drugs, heparin, potassium supplements or potassium-rich salt substitutes may lead to severe hyperkalaemia, and specifically that trimethoprim/sulfamethoxazole (co-trimoxazole), a common short-course antibiotic, may cause clinically relevant hyperkalaemia with spironolactone; hyperkalaemia may occur with impaired renal function. Spironolactone also reduces lithium clearance, increasing the risk of lithium toxicity, and lithium levels should be monitored if the two are used together.^[70] [Product information for licensed indications, not feminising therapy.]
• Cyproterone acetate: meningioma risk rises with cumulative exposure; regulatory advice is to use the lowest effective dose, to monitor for symptoms of meningioma, and to stop all cyproterone-containing products permanently if one is diagnosed.^[17,64] Current or previous meningioma is a contraindication for all indications;^[64] product information also lists liver disease and existing thrombosis or embolism as contraindications, and requires liver function tests before and during treatment.^[65] In a large cohort, mean testosterone was suppressed at 10 mg daily as at higher doses, though only four people started on 10 mg and not every individual was suppressed.^[50]

What this section establishes. Each medication brings its own trade-off. Transdermal oestradiol is associated with lower clot risk than oral in cisgender data, and with a smaller rise in a clotting marker in trans women; the 2024 systematic review did not analyse clinical outcomes by route.^{[19,22,46,47]} Cyproterone suppresses testosterone more than spironolactone, and carries liver, thromboembolic, mood and cumulative-dose meningioma warnings.^[49,64,65] The role of finasteride and dutasteride is limited and poorly evidenced, and a benefit of progesterone has not been established.^[1]

4. Expected physical changes — variability is the rule, and no variable examined has consistently predicted breast development

This is the section readers ask about most, and where realistic expectations matter most. The single most important message is that variability is the rule: timelines and especially the degree of change differ substantially, and for breast development, the best-measured outcome, no variable examined in a large prospective cohort reliably predicted the result.^[3]

4.1 Breast development — modest on average and unpredictable, with most measured change in the first year

Expected change. Growth of glandular tissue and fat, producing an anatomically feminine breast. Volumes are modest on average: in three-year prospective follow-up, mean measured volume reached about 100 cc per breast, and 71% of participants were below the A-cup category the study derived from volume (taking each 150 cc as one cup size), which is not a reliable guide to commercial bra size or to an individual’s result.^[4] The comparison with cisgender development has limits: according to the review’s abstract, in cisgender women early mammary tissue forms before birth and stays quiescent until oestrogen drives breast growth at puberty, and breast tissue continues to change during pregnancy, lactation and menopause.^[21] Adult feminising hormone therapy therefore occurs in a different developmental context rather than simply reproducing cisgender puberty at a later age. [Population flag: cisgender developmental data.]

Timeline. Measurable growth is present within the first three months in prospective data;^[3] guideline tables cite onset at 3–6 months.^[2] In prospective data, the largest measured changes occur in the first year, most within the first six months; growth then slows but does not stop. Three-year prospective follow-up found that the breast–chest (tape-measure) difference plateaued after about nine months, while breast volume measured by 3D imaging continued to increase, more gradually, over the three years (a decrease between 24 and 36 months, seen only in people who had had a vaginoplasty, was no longer observed after adjustment for age and hormone levels).^[3,4] Guideline tables cite a maximum at 2–3 years (Endocrine Society) or 2–5 years (SOC-8), and UCSF guidance a typical Tanner stage of 2–3.^[1,2,18]

Evidence quality. Moderate — the best-measured feminising outcome. In a large prospective multicentre study (229 participants), the mean breast–chest difference increased by about 3.7 cm (95% CI 3.3 to 4.2) over one year, from 4.1 cm at baseline to 7.9 cm, and nearly half (48.7% of the 197 with cup-size data) had less than an AAA cup at one year.^[3]

Variability. High and largely unexplained. In that cohort, no clinical or laboratory variable examined, including measured serum oestradiol, predicted the amount of breast development at one year;^[3] in an exploratory analysis of a later randomised trial, estimated breast volume at six months was greater with older age and with higher achieved serum oestradiol.^[49] One caveat bounds even this finding: these analyses used total serum oestradiol. They did not test whether free oestradiol (the unbound fraction) or bioavailable oestradiol (conventionally, free plus albumin-bound hormone) predicts breast development better, and no validated treatment target for either is established. Average measured development over the first one to three years was modest in these cohorts, and those observations do not establish ultimate size.^[3,4] Some people later seek breast augmentation, which is a realistic part of the conversation, not a treatment failure. Satisfaction does not track volume closely: in the three-year cohort, 58% were satisfied with the size of their breasts, irrespective of the volume gained, with 63–92% satisfied with symmetry and 51–83% with shape across visits — though the authors note that satisfaction may be inflated, because participants who had breast augmentation were excluded from that point, probably removing the least satisfied.^[4]

4.2 Body-fat redistribution — more fat, deposited differently, over years

Expected change. FHT commonly increases total fat mass,^[29] and guideline tables describe redistribution of body fat.^[1,2] This is a change in where fat is deposited, often with relatively more on the hips and thighs, rather than existing abdominal fat moving elsewhere; abdominal fat need not decrease, and the visible response varies.

Timeline. Onset ~3–6 months; maximum ~2–5 years.^[1]

Evidence quality. Low–Moderate — the direction is well supported: a narrative review of imaging studies found most studies associating oestrogen with increased fat mass and decreased lean mass in trans women, though the specific timeline figures come largely from expert-consensus tables.^[2,29]

Variability. High; influenced by genetics, baseline body composition and weight.

4.3 Muscle mass and strength — mass falls, and strength changes less consistently

Expected change. Lean mass and muscle size commonly decrease. Measured strength changes less consistently: a 2021 systematic review of 24 studies found that strength, lean body mass and muscle area decreased, but that strength loss after 12 months ranged from non-significant to 7%, and concluded that strength may be well preserved during the first three years.^[61]

Timeline. Guideline tables give onset at ~3–6 months and a maximum at ~1–2 years for muscle mass and strength together;^[2] a time to maximum effect has not been established by measurement.

Evidence quality. Low–Moderate. Variability. Affected by activity and resistance training, which partly offset the loss.

4.4 Skin — less oily, though new acne is not rare

Expected change. Skin typically becomes softer and less oily; sebum (oil) production falls as androgen activity decreases; existing acne often improves and pores may look less prominent. [Clinical observation, not drawn from a cited source.] New acne is not rare, however. In a large US matched cohort of people without acne at baseline, incident acne diagnoses after starting oestradiol were more frequent than in matched cisgender men and less frequent than in matched cisgender women (hazard ratios in the Results 1.64, 95% CI 1.37 to 1.96, and 0.54, 0.46 to 0.64; the abstract gives 1.56 and 0.53, a discrepancy the paper does not explain), while moderate-to-severe acne, defined by isotretinoin or at least 30 days of oral antibiotics, did not differ from cisgender men.^[52] Comparisons with cisgender populations do not isolate the effect of oestradiol or show whether pre-existing acne improves, and they rest on recorded diagnoses and prescriptions rather than examined severity.

Timeline. Within the first few months (~3–6 months).^[2]

Evidence quality. Low–Moderate — direction is consistent and mechanistically clear (sebum is androgen-driven), but much support is self-report and clinical observation rather than objective measurement. [See companion review: Skin Effects of Oestrogen.]

Variability. Moderate; some people find skin becomes too dry.

4.5 Body hair and facial hair — body hair softens, and established facial hair often persists

Expected change. Body hair becomes finer, lighter and slower-growing. Established facial hair is reduced much less and often persists. For people who want further treatment, laser reduction selectively targets dark, coarse hair, while electrolysis can permanently remove successfully treated follicles and can be used on all hair colours and skin types; both need multiple sessions, and UCSF guidance notes that lifelong treatment is often required for a sustained effect.^[18] That further treatment is for regrowth after laser reduction, or for hairs not yet or incompletely treated; a follicle permanently destroyed by electrolysis does not need retreatment. Suitability depends on hair and skin characteristics, device selection and operator expertise; repeated sessions, residual hair and adverse effects are common considerations. [Procedural practice; detail belongs in a dedicated hair-removal review.]

Timeline. Body-hair changes over ~6–12 months, continuing beyond three years; hormones alone do not usually remove established facial hair.^[2]

Evidence quality. Low–Moderate for body hair. A 2023 systematic review found that feminising therapy may reduce facial and body hair growth (prospective and retrospective cohorts and case reports, very few with objective measures), and that some facial hair commonly persists.^[53] Variability. High; likely influenced by genetic background, among other factors.^[18]

4.6 Scalp hair — loss often slows, but lost hair is not reliably regrown

Expected change. Existing male-pattern loss often slows or stabilises; hormones do not reliably regrow lost hair.

Timeline. UCSF guidance describes arrest and possible reversal of scalp hair loss;^[18] a 2023 systematic review found low-level evidence that feminising therapy may improve androgenetic alopecia.^[53] Neither the onset nor the extent has been well quantified.

Evidence quality. Low–Moderate. Variability. High; depends on the stage of loss at the start.

4.7 Fertility — sperm production is suppressed incompletely, so hormone therapy is not reliable contraception

FHT suppresses sperm production, causing higher rates of abnormal semen parameters or azoospermia.^[38] Suppression is incomplete and variable, so FHT is not reliable contraception: SOC-8 notes that azoospermia is not complete and that contraception needs to be considered if pregnancy is to be avoided, and UCSF guidance states that hormone therapy alone is not a reliable form of contraception.^[1,18] A systematic review of testicular histology after feminising therapy found spermatogenesis ranging from complete, through arrest at various stages, to no germ cells at all, with no consistent relationship to how long treatment had been given.^[38] In 214 orchidectomy specimens from one centre, some tubules still showed complete spermatogenesis in 4.7%, all in people who had started treatment at Tanner stage 4 or later; neither stopping hormones before surgery nor treatment duration was associated with preserved germ cells.^[58] [All participants used oestrogen with cyproterone or triptorelin.]

Recovery after stopping has been documented. A 2025 systematic review concluded that spermatogenesis is restored in most people after stopping, though the time needed may be long; its evidence is small studies, including one series of nine, a study in which 80% of people could cryopreserve semen after stopping treatment for three to six months, and a study in which azoospermia persisted in 1 of 11 people four months after stopping.^[38] For an individual, the probability, timing and completeness of recovery remain uncertain. Nine trans women who chose to stop treatment for reproductive reasons all eventually produced viable spermatozoa: six on their first semen analysis, two more at eight and ten months after stopping, having first been documented azoospermic, and the ninth only by surgical testicular sperm extraction at seventeen months, having still been azoospermic at thirteen. Five of the nine had impaired semen quality after stopping: four whose semen never reached normal values, and the one who needed surgical extraction; four produced at least one normal analysis.^[37] [Nine participants, self-selected by electing to stop treatment; no systematic semen data from before or during treatment (one participant had banked sperm beforehand), so pre-existing impairment cannot be excluded. Azoospermia was documented before recovery in only three of them.]

Recovery therefore cannot be assumed to be prompt, complete, or achievable without a procedure. This is covered as a counselling topic in Section 6.

4.8 Libido, erectile function and emotional effects — early reductions, with a later course that varies

Reduced sexual desire and fewer spontaneous erections are among the earliest effects; guideline tables give onset at 1–3 months and place the largest early change by 3–6 months, which is a conventional estimate of early change rather than a permanent plateau.^[2] Erectile function commonly decreases.

Sexual desire and erectile function may change again when the hormone regimen changes. As a physiological expectation, if the testes remain and testosterone production recovers after treatment is reduced or stopped, reduced desire and fewer spontaneous erections may improve; direct evidence is insufficient to predict how much or how quickly, or whether previous function will return. Separately, where preserving erectile function is a treatment goal, UCSF guidance considers reducing androgen blockade to allow testosterone to rise, depending on the person’s goals.^[18] After bilateral orchidectomy, stopping medication does not restore testicular testosterone production.

Emotional and mental-health effects should be considered separately: their course after treatment changes cannot be predicted from androgen levels alone, and a return to the previous emotional state should not be assumed. Many people report improved mood, wellbeing and quality of life and reduced dysphoria; this evidence is mostly observational and uncontrolled, so it supports the direction of benefit more than its precise size. FHT should not be presented as a substitute for appropriate mental-health care where that is needed. [Evidence: Low–Moderate. This rates the evidence on effects during treatment. Direct evidence predicting recovery of sexual or emotional function after treatment is reduced or stopped is insufficient, and no rating is offered for it.]

What this section establishes. Most adults see feminising change, but its degree varies widely, and for breast development no variable examined has consistently predicted it.^[3,49] Breast development is usually modest, begins within the first few months and is often greatest in the first year.^[3,4] Lean mass falls and strength changes less consistently; established facial hair often persists; and sperm production is suppressed incompletely, so hormone therapy is not reliable contraception.^[1,18,53,61]

5. What hormones cannot change — after an androgen-driven puberty, skeletal proportions and voice

One of the most useful things to be clear about is what FHT does not do. Hormones change soft tissue (fat, skin, glandular breast) and physiology, but once an androgen-driven puberty is complete they do not reverse its skeletal proportions or voice changes, although sex steroids continue to affect bone turnover. Expectations differ where skeletal maturation or pubertal history differs. The features below are not reliably reversed by hormones and, if a person wishes to change them, require other interventions.

What this section establishes. After an androgen-driven puberty, hormones do not reverse skeletal proportions or voice, and changing those needs other interventions.

6. Before starting — most conditions call for individual assessment, not automatic exclusion

The brief picture before treatment matters as much as monitoring during it. The specifics below reflect common guideline-based practice (chiefly the Endocrine Society 2017 guideline and standard hormone-prescribing practice); exact processes vary between services.^[2]

6.1 Baseline assessment — history, examination and regimen-specific blood tests

A pre-treatment review typically covers:

• History: general medical history; personal and family history of venous thromboembolism or thrombophilia; migraine (and specifically whether with aura); hypertension; smoking and other cardiovascular risk factors; liver disease; kidney disease; and current medications (for interactions).
• Fertility intentions and a discussion of preservation (below).
• Examination: baseline blood pressure and weight/BMI.
• Baseline blood tests: oestradiol and testosterone; prolactin, which the Endocrine Society advises measuring at baseline in everyone starting oestrogen;^[2] potassium and kidney function where spironolactone is planned;^[1] and liver enzymes where cyproterone is planned, given reports of drug-induced liver injury;^[15,28] alongside general health bloods as clinically indicated.

6.2 Contraindications and precautions — an active oestrogen-sensitive cancer excludes oestrogen, and most else calls for individual care

These are best understood as situations that require specialist discussion and individualised care. Two call for particular care. UCSF guidance treats an active oestrogen-sensitive cancer as a contraindication to oestrogen.^[18] An acute venous thromboembolism needs urgent treatment and an individual plan for hormones: UCSF guidance (2016) advises stopping oestrogen for the duration of treatment,^[18] whereas a 2024 haematology review suggests that people who develop a clot on feminising oestrogen and wish to continue may do so if full-dose anticoagulation is continued concurrently, with treatment individualised; that suggestion rests on data from cisgender women, there being no randomised trials in this population.^[60] Whether oestrogen is withheld or continued during therapeutic anticoagulation is therefore decided with the treating team, weighing recurrence and bleeding risk, the formulation and the consequences of interruption; neither source supports continuing oestrogen without anticoagulation. UCSF guidance does not recommend routine screening for thrombophilia in the absence of risk factors.^[18] The following are commonly treated as precautions requiring individual risk assessment rather than automatic exclusion: a past history of venous thromboembolism (though UCSF guidance advises against oestrogen where a past VTE or other significant VTE risk is combined with continued smoking^[18]); known thrombophilia; a past history of oestrogen-sensitive cancer; liver disease, where UCSF guidance notes that non-oral routes may be preferred though no specific evidence supports this; significant cardiovascular disease; migraine with aura; severe uncontrolled hypertension; and the combination of smoking with older age. [Uncontrolled hypertension and smoking combined with older age are general clinical practice; neither cited guideline lists them in these terms.] In several of these, choosing a transdermal route and the lowest effective dose, optimising the underlying risk factor and involving the relevant specialist allow treatment to proceed in many people.^[2,18]

Suitability is assessed separately for oestradiol and for each adjunct, and a route chosen to reduce vascular risk does not override an adjunct’s contraindications. Cyproterone is contraindicated in liver disease, existing thrombosis or embolism, and current or previous meningioma;^[64,65] spironolactone is contraindicated in hyperkalaemia, Addison’s disease, anuria, acute renal insufficiency or significant renal compromise, and with eplerenone or other potassium-sparing diuretics, so suitability is assessed before treatment as well as monitored during it.^[70] A history of hormone-sensitive cancer calls for tumour-specific oncology advice. Current product information and specialist input are needed where these issues arise.

Important evidence note: much of the contraindication and precaution framework is extrapolated from guidance on combined hormonal contraception and menopausal hormone therapy in cisgender women, and the trans-specific evidence behind it is limited. For venous thromboembolism and stroke by route, menopausal hormone-therapy guidance is the main source.^[20] Migraine with aura warrants assessment of vascular risk and hormone-related symptom changes. Among people with migraine, those with aura have a higher risk of stroke, and combined hormonal contraception is contraindicated in current migraine with aura at any age (UKMEC category 4; a history of aura five or more years ago is category 3);^[43] but that restriction should not be applied automatically to 17β-oestradiol used for feminising therapy: UCSF guidance notes that it is not clear whether the stroke risk seen with oral contraceptives translates to bio-identical oestradiol.^[18] Menopausal guidance does not regard aura alone as a contraindication to HRT and favours transdermal oestrogen;^[42] its application to feminising therapy is extrapolative, with limited direct evidence.

6.3 Fertility counselling — recovery is documented, but cannot be promised to an individual

FHT commonly suppresses sperm production but is not reliable contraception; where sexual activity could result in pregnancy, contraception should be discussed irrespective of hormone use.^[1,18] Recovery has been documented; its probability, timing and completeness remain uncertain: in a published series of nine people who stopped for reproductive reasons, six of nine already had sperm at their first analysis, taken between three and twenty-seven months after stopping, while the three with documented azoospermia produced sperm at eight, ten and seventeen months — the last by surgical extraction. Five of the nine had impaired semen quality after stopping. The authors of that series set out to test the untested assumption that feminising therapy causes permanent infertility, and they nonetheless recommend banking sperm beforehand for anyone who might want to be a genetic parent — in part because the months off treatment, as testosterone returns, are likely to carry physical and psychological costs.^[37] Counselling on fertility-preservation options before starting treatment is also a strong guideline recommendation;^[2] UCSF guidance describes sperm cryopreservation before starting hormones as the most successful fertility-preservation option.^[18] Reproductive wishes should be assessed and reassessed during care,^[38] and SOC-8 recommends that surgeons confirm reproductive options have been discussed before gonadectomy.^[1] Collecting sperm after stopping treatment, or by surgical retrieval, may be possible later but cannot be assumed;^[37,38] bilateral orchidectomy permanently removes testicular sperm production.

What this section establishes. An active oestrogen-sensitive cancer is a contraindication to oestrogen, and each anti-androgen has its own contraindications; most other conditions call for individual assessment rather than automatic exclusion.^{[18,64,65,70]} Fertility preservation is discussed before starting, because recovery cannot be promised to an individual.^[37,38]

7. Monitoring — recommended for effect and safety, though evidence that any schedule prevents major events is limited

Clinical review and regimen-specific laboratory monitoring are recommended to assess desired changes, hormone exposure and adverse effects; evidence that a particular schedule reduces major clinical events is limited. The following reflects common guideline-based practice.^[2] Serious complications can occur between normal blood results, and routine results do not exclude them. Sudden focal neurological symptoms, severe unexplained breathlessness, concerning acute chest pain or collapse need emergency assessment; other symptoms suggesting a pulmonary embolism, or new swelling or pain in one leg, need urgent assessment. On cyproterone, jaundice or other signs of liver injury, and symptoms or signs of meningioma, need prompt clinical review.^[64,65] [Triage categories are general clinical practice.]

7.1 Hormone levels — the usual range lacks a primary evidence base, so clinical progress weighs more than a single number

The two core tests are oestradiol and testosterone. The usual aims are testosterone in the female range (commonly cited as below ~50 ng/dL, roughly <1.7 nmol/L) and oestradiol within the premenopausal female range (commonly ~100–200 pg/mL, about 367–734 pmol/L) rather than higher.^[2] The Endocrine Society’s recommendation 4.1 suggests clinical evaluation and laboratory monitoring of sex-steroid levels every three months during the first year of hormone therapy, then once or twice yearly; its monitoring table separately lists three-monthly testosterone and oestradiol measurement without repeating the later interval in that row, so the two should be read together.^[2] Because oestradiol fluctuates and assays vary between laboratories, levels are interpreted alongside clinical progress rather than chased in isolation.^[18] This caution now has direct support: a 2025 systematic review of 49 studies, searching to April 2023, found that the commonly cited 100–200 pg/mL oestradiol range lacks a primary evidence base and has not been shown to optimise feminisation or reduce adverse events — so it is best treated as a pragmatic guide; the review’s authors suggest that monitoring individual goals and patient satisfaction likely provides more clinical value than targeting the range.^[30] Part of the difficulty is technical: the same review notes high imprecision and variability between oestradiol assays at lower concentrations,^[30] and total oestradiol does not directly measure the unbound (free) fraction, which changes in sex hormone-binding globulin (SHBG) can shift; bioavailable oestradiol conventionally means free plus albumin-bound hormone. Whether either predicts feminisation better than total oestradiol is not established, and no validated alternative treatment target exists, so this is a reason to weight clinical progress over any single number, not to test these fractions routinely. [General laboratory knowledge, not drawn from a cited source.]

Interpreting a result. The formulation, route, dose interval and time since the last dose should be recorded with any hormone result. Peaks after injection or sublingual use,^[2,18] contamination of the sampling site by topical oestradiol gel, and differences between assays can produce misleading results. An unexpected value is interpreted alongside symptoms, adherence and clinical progress, and repeated or confirmed before a dose change; if testosterone is unexpectedly unsuppressed, missed doses, administration problems, timing and assay issues are considered before oestradiol is simply increased. [General laboratory and prescribing practice except where cited.]

7.2 Regimen-specific safety bloods — potassium on spironolactone, liver enzymes on cyproterone

• On spironolactone: monitor potassium and kidney function (hyperkalaemia risk). The Endocrine Society states that electrolytes, particularly potassium, should be monitored every three months in the first year and annually thereafter.^[2] In a single-centre US review of 318 people on spironolactone, which excluded people with hypertension, renal failure, diabetes, heart failure or medications affecting the renin–angiotensin–aldosterone system, 8 (2.5%) had at least one potassium above 5.1 mmol/L, none severe: 8.9% of those over 45 against 1.5% of those 45 or younger. The authors propose that frequent monitoring may be unnecessary in healthy people aged 45 or under, and should be reserved for older people and those with relevant conditions or medications.^[54] That selected retrospective cohort does not establish that monitoring can be omitted; monitoring is reassessed after dose changes, new interacting medicines, kidney impairment or acute illness.
• On cyproterone acetate: monitor liver enzymes. Prolactin-monitoring practice varies by guideline, regimen and symptoms. Cyproterone is associated with a larger rise in prolactin than spironolactone,^[16,49] and the rise was smaller at 10 mg daily than at higher doses;^[50] markedly or progressively elevated levels warrant further assessment, including for prolactinoma. The Endocrine Society suggests periodic prolactin measurement (a weak recommendation), while SOC-8 makes no recommendation for or against and favours judgement informed by regimen and symptoms.^[1,2] In the Amsterdam cohort, the authors considered routine screening for prolactinomas unnecessary because these tumours were rare — although prolactinomas were more frequent than in the general female population (standardised incidence ratio 4.3, 95% CI 2.1–7.9) and four of the nine were found only through routine prolactin checks. That is an inference from observational data, not a trial showing that monitoring offers no benefit.^[12] Drug-induced liver injury was among the serious adverse effects reported, in the trial abstract, in the cyproterone arm of the randomised comparison with spironolactone, which is the trial evidence behind the liver-enzyme monitoring above.^[15]
• Blood count: haematocrit and haemoglobin fall on feminising therapy. In a large US cohort, anaemia by male thresholds was about six times as common as in cisgender men (HR 5.9, 95% CI 4.6–7.5) and modestly more common than in cisgender women by female thresholds (1.4, 1.1–1.8); the authors support interpreting haematological results against reference ranges for the affirmed gender.^[55] That applies to haemoglobin and haematocrit after sustained treatment; it does not imply that a single gender-based reference interval suits every laboratory test.

7.3 Blood pressure, weight and bone — inadequate replacement after gonadectomy is a recognised bone risk

Blood pressure and weight are checked routinely, and cardiovascular risk-factor assessment including lipid profile and diabetes screening follows the same standards as for anyone else, as the Endocrine Society guideline suggests.^[2] Inadequate hormone replacement, especially after gonadectomy, is a recognised bone risk.^[2,56] Trans women often have lower bone density than cisgender comparators even before treatment, as the fracture study notes from earlier work.^[34] After gonadectomy, adequate replacement matters: in a cross-sectional Italian study of 57 trans women at least two years after surgery, 40% had low bone mass by male reference ranges (30% by female ranges); low bone mass was associated with lower oestradiol and with collecting less than 80% of prescribed oestrogen, though the design cannot show cause and 93% also had low vitamin D.^[56] A narrative review of imaging studies found no consistent change in bone density on oestrogen therapy, though most studies followed people for only about a year.^[29] Fracture data do exist, and they follow the same pattern as the cardiovascular evidence: the comparison group decides the answer. In a nationwide cohort using emergency-department diagnoses over three years, 4.4% of trans women aged 50 or over sustained a fracture, against 2.4% of age-matched reference men (odds ratio 1.90, 95% CI 1.32–2.74) but 4.2% of age-matched reference women (1.05, 0.75–1.49). Below 50, 2.4% fractured, against 3.0% of reference men and 1.6% of reference women, differing significantly from neither (against reference women, odds ratio 1.49, 0.96–2.32, which the authors describe as a tendency to higher risk). Older trans women therefore had higher fracture odds than reference men and an estimate close to that of reference women, though that interval does not establish equivalence; fractures were ascertained from emergency-department diagnoses of all types over three years, and the design does not isolate the effect of hormones.^[34] Bone-density scanning (DXA) is used selectively, guided by risk factors, rather than universally.^[2] Relevant factors include previous fragility fracture, prolonged low sex-steroid exposure (for example after gonadectomy without adequate replacement), low weight, glucocorticoid use, nutrition and physical activity, and, where applicable, previous puberty suppression. [General clinical practice; not drawn from a cited source.]

7.4 Cancer screening — it follows the organs a person has, and never delays assessment of symptoms

Screening and diagnostic assessment are different. Screening should follow the organs a person has and their hormone-exposure history, rather than a single rule, and should take account of age, duration of oestrogen exposure, family history and local programmes. Breast-screening considerations apply to trans women who have developed breast tissue; the prostate is retained after vaginoplasty, so it remains relevant — though, as noted in Section 8, PSA interpretation is complicated by androgen suppression.^[5,9] A new breast lump or other suspicious breast change needs prompt assessment whatever a person’s screening eligibility.

Administrative records affect invitations. In England, people registered with their GP as male are not invited for breast screening, but their GP can arrange a referral for mammograms; trans women registered as female are not invited for abdominal aortic aneurysm screening and should consider arranging it at 65.^[68] This guidance applies to England only.^[68] For the prostate, a 2025 systematic review found that the available PSA data, from four studies with 290 participants, cannot be used to make clinical recommendations and that there are no clinical PSA thresholds to guide interpretation in this population; it concluded that not everyone over 40 should be offered PSA testing, but that people with a genetic predisposition, a family history or symptoms of prostate cancer may request or be offered it.^[69] Decisions about PSA testing without symptoms therefore rest on individual discussion of age, family history, genetic predisposition and the difficulty of interpreting PSA during androgen suppression. Symptoms or signs suggesting prostate cancer need diagnostic assessment, which neither screening eligibility nor a low PSA result should delay. [Distinction between screening and symptomatic assessment is general clinical practice.]

What this section establishes. Clinical review and regimen-specific laboratory monitoring are recommended, though evidence that a particular schedule reduces major clinical events is limited.^[1,2] The usual oestradiol range lacks a primary evidence base,^[30] so clinical progress weighs more than any single level. Serious complications can occur between normal results, which routine results do not exclude.

8. Risks — venous thromboembolism is the prominent signal, and the arterial picture depends on the comparison

Two distinctions run through this section:

• Absolute versus relative risk. A “doubling” (relative) can still be a small change if the baseline (absolute) risk is low. Most risks below are relative increases on a low base.
• Old oestrogens versus modern oestradiol. Much alarming historical data comes from ethinylestradiol, far more thrombogenic than today’s 17β-oestradiol; applying those figures to modern regimens overstates risk.^[2,31]

The FHT evidence base is dominated by a few observational cohorts, and they are less independent of one another than the reference list makes them look. Eight of the citations in this review draw on the Amsterdam gender clinic registry: breast cancer, prostate cancer, mortality, acute cardiovascular events, benign brain tumours, population prevalence, the 2026 cardiovascular cohort, and fracture outcomes.^{[5,6,8,9,12,23,32,34]} Seven more draw on the same clinic’s patients outside the registry: the three-year breast development study, the progesterone trial (three citations), the trial of stopping oestradiol in older women, and two fertility studies (one including a single participant from an Australian clinic).^{[4,13,14,37,41,57,58]} Four are multicentre ENIGI studies in which Amsterdam is one of several sites: one-year breast development, sexual desire, clotting markers and cyproterone dose.^[3,40,47,50] A second, smaller cluster comes from one Melbourne group.^{[27,28,36,39,49,53]} The Amsterdam studies differ in outcome, period and method, and several are the strongest studies available, but they are substantially one clinic’s population followed over five decades. This should be known before the studies are counted: agreement between them is weaker evidence than agreement between independent cohorts would be, and a systematic feature of that clinic’s practice or population would propagate through most of this review. The cohorts also cannot fully separate hormones from confounders such as smoking, HIV and socioeconomic factors.

8.1 Venous thromboembolism — a prominent signal whose size on a specified modern regimen is not known

• The very high historical VTE rates were seen chiefly with oral ethinylestradiol, now largely abandoned for FHT. The Endocrine Society guideline cites a twenty-fold increase in a large Dutch cohort, which it says may have been associated with ethinylestradiol, and notes that incidence fell when clinicians stopped prescribing it;^[2] in an early Amsterdam series of 816 trans women, 41 had a venous thromboembolism, 40 of them while taking oral ethinylestradiol.^[31] That series’s own abstract reports venous thromboembolism as the major complication with oral oestrogens.^[10]
• Historical ethinylestradiol regimens carried particularly high risk.^[31] Contemporary cohorts support continued concern about VTE but do not supply a single risk estimate for any specified modern 17β-oestradiol regimen: they mix regimens, earlier ethinylestradiol or cyproterone exposure, follow-up periods and ascertainment. In the 2024 meta-analysis, 1.6% of trans women (265 of 16,626) had a venous thromboembolism; that is a proportion pooled across cohorts with differing follow-up, so it should not be read as an annual risk, a cumulative incidence over a defined period, or the risk on a modern regimen. The most recent Dutch cohort observed 2.4 events per 1,000 person-years (95% CI 1.8–3.1), a rate specific to that cohort and its treatment histories.^[22,23] In a US single-centre cohort (1996–2025), 16 of 1,173 adults prescribed oestrogen (1.4%) had a VTE; six followed surgery and six were unprovoked, and most occurred alongside other established risk factors, but without a comparator group or standardised follow-up the authors caution that this is not an estimate of incidence or relative risk.^[62] A 2021 prevalence meta-analysis of 18 studies (11,542 people) pooled prevalence at 2.0% (95% CI 1.0–3.0) with very high heterogeneity (I² 89.2%); in study-level meta-regression, which does not isolate an individual effect of treatment duration, prevalence rose with mean age and with duration of oestrogen therapy, and was 0.0% (0.0–2.0) in series with a mean age under 37.5 years. Data were too sparse to compare routes.^[45]
• In the comparator studies pooled in the 2024 meta-analysis, VTE incidence was higher in trans women, not all of whom were using hormones, than in cisgender comparison groups. The 2024 meta-analysis pooled this to a relative risk of 2.2 versus cisgender men, but the interval is wide (1.1–4.5) and heterogeneity between the pooled studies was very high (I² 93.5%), so that figure should not be quoted on its own.^[6,7,22]
• Route matters: transdermal oestrogen is associated with lower clot risk than oral — well supported in cisgender postmenopausal women, in a case–control study and in a large UK nested case–control analysis (oral versus no HRT, odds ratio 1.58, 1.52–1.64; transdermal 0.93, 0.87–1.01), whose authors caution about residual confounding.^[19,46] [Population flag.] In trans women the evidence is indirect: a 2019 review reported that transdermal oestradiol has the least thrombogenic profile while noting that there are no head-to-head studies,^[31] and thrombin generation rose less with transdermal than oral oestradiol;^[47] the 2024 systematic review did not analyse outcomes by route, noting that variation in administration routes limits current knowledge,^[22] and the one cohort identified that compared routes for clinical thrombotic events had too few events to be informative (transdermal patch: adjusted odds ratio 1.71, 0.18–16.5).^[51] No adequately powered comparison of clinical VTE outcomes by route in trans women has been identified.
• Surgery and immobility add transient VTE risk, while older age, previous VTE, thrombophilia, smoking and other conditions affect baseline risk;^[31,45] VTE also occurs outside the peri-operative period, including unprovoked events.^[62]
• Routine cessation of oestrogen before surgery developed without evidence that it reduces harm. Three retrospective cohorts covering roughly 900 vaginoplasties report two venous thromboembolisms in total, and a meta-analysis of two of them gives a pooled odds ratio of 0.77 — with a confidence interval from 0.04 to 14.01, compatible with a large reduction and a large increase alike.^[35] The honest reading is that events are too rare in this setting for these studies to detect a difference, not that continuation has been shown to be safe. Management is therefore individualised: UCSF guidance (2016) states that there is no evidence that trans women without specific risk factors (smoking, personal or family history, excessive doses or synthetic oestrogens) must stop oestrogen around surgery, particularly with appropriate prophylaxis and an informed-consent discussion, and suggests a lower dose or a switch to transdermal as alternatives;^[18] the general principle is that hormone risks are weighed against the harms of withholding treatment, such as poor mental health.^[31] A later single-centre retrospective study (primary vaginoplasty, 2016–2023) compared 238 people who stopped oestrogen for two weeks before surgery with 91 who continued it: within 90 days, one VTE occurred in the group that stopped and none in the group that continued, and a further event in the continued group, on day 97, fell outside the outcome window. All received inpatient enoxaparin prophylaxis, and the groups differed in oestrogen route and formulation, surgeon and date of surgery (continuation became the centre’s policy in 2021); events are too few, and the groups too different, to show that continuation is safe.^[71]

8.2 Arterial events — myocardial infarction not shown to be raised against cisgender men, and stroke findings mixed

Both the comparison group and the study design materially affect the estimates. Comparisons with cisgender men and with cisgender women provide different benchmarks, and neither isolates a causal effect of FHT.

Against cisgender men, the arterial signal has been mixed. The Amsterdam cohort (2,517 trans women, 22,830 person-years) found stroke raised (SIR 1.80, 95% CI 1.23–2.56) but myocardial infarction not (0.79, 0.54–1.11).^[6] A US cohort (2,842 transfeminine people, mean follow-up 4.0 years) found no statistically detectable increase in either outcome overall — ischaemic stroke 1.2 (0.9–1.7), myocardial infarction 0.9 (0.6–1.5) — but among the 853 who started oestrogen during follow-up — the subgroup with observed oestrogen initiation — ischaemic stroke was raised overall (2.3, 1.2–4.3); the statistically detectable excess was in follow-up beyond six years (9.9, 3.0–33.1; 8 events), while the interval for the first six years (1.3, 0.6–2.9; 9 events) did not exclude an increase — a statistical pattern, not evidence of a biological six-year threshold; the authors concluded oestrogen is probably a risk factor for ischaemic stroke.^[7] [Nota’s “stroke” included haemorrhagic stroke and transient ischaemic attack; Getahun and the 2026 cohort report ischaemic stroke.] A 2024 systematic review of 22 studies pooled six of them to a stroke relative risk of 1.3 (1.0–1.8), and five to a myocardial infarction relative risk of 1.0 (0.8–1.2).^[22]

Against cisgender women, both cohorts found excess arterial events — Nota reporting stroke 2.42 and myocardial infarction 2.64, Getahun 1.9 and 1.8.^[6,7] The meta-analysis does not bear on this comparison: it analysed trans women against cisgender men only.^[22] The 2026 cohort, which did make the comparison, found no statistically detectable increase in either outcome after adjustment for socioeconomic status — myocardial infarction 1.30 (0.84–1.85) and stroke 1.20 (0.92–1.52) — though before additional adjustment for SES, with standardisation for age, sex and calendar year only, both were raised (1.57 and 1.31).^[23] Comparisons against men and against women use different baseline incidence rates and answer different benchmarking questions; neither isolates a causal effect of hormone therapy.

The most recent analysis added socioeconomic adjustment and uniform national-registry ascertainment for both groups. A 2026 Dutch registry study (2,714 trans women, 23,907 person-years, events captured 2012–2022, adjusted for socioeconomic status) found myocardial infarction at half the rate of general-population men (SIR 0.50, 0.32–0.71) and no statistically detectable difference in stroke incidence (0.94, 0.72–1.19).^[23] The authors suggest that the earlier Amsterdam analysis’s higher ratios may partly reflect its Norwegian control population and outcome ascertainment that differed between groups (medical records versus registry), and they attribute its higher VTE ratios partly to years when ethinylestradiol was still prescribed. [One study; a lower point estimate this large may still reflect residual confounding, which the authors call a major limitation, and it does not establish that FHT reduces myocardial-infarction risk. The earlier analysis states that its reference populations were Dutch or Norwegian.]

A Danish national register cohort (1,401 people assigned male at birth, matched to control men and control women) found any cardiovascular disease — a composite that included hypertension, hyperlipidaemia and cardiovascular medication — more common than in control men (adjusted HR 1.92, 95% CI 1.59–2.33) and control women (1.73, 1.46–2.05), while rates of major arterial events did not differ significantly from either. In mediation analysis, recorded hormone treatment did not explain the excess; only a minority had recorded hormone use, and the authors note possible surveillance bias.^[48]

Across the cited studies, myocardial-infarction incidence was not consistently higher than in cisgender men. Findings for ischaemic stroke are mixed: some cohorts, including the US oestrogen-initiation cohort, found increased incidence, whereas the recent Dutch registry study found no statistically detectable increase after socioeconomic adjustment. Differences in populations, treatment histories, follow-up and methods limit direct comparison. These observational findings do not establish a single causal risk estimate for contemporary feminising hormone therapy.^{[6,7,22,23,48]} VTE shows a more consistent adverse association than the arterial outcomes (above), but its regimen-specific causal magnitude also remains uncertain. In one cohort’s published abstract, current (not past) ethinylestradiol use was associated with an independent three-fold increase in cardiovascular death — another reason it is no longer used.^[11]

8.3 Mortality — higher than in the general population, with hormones neither implicated nor excluded

The longest follow-up, from the Amsterdam cohort (2,927 trans women over five decades), found overall mortality higher than expected compared with general-population men (standardised mortality ratio ~1.8) and women (~2.8), with cause-specific excess in cardiovascular disease, lung cancer, HIV-related disease and suicide. Excess mortality did not decline over the five decades and was highest among those who started treatment in 2010–18, and the study could not compare hormone regimens. The authors concluded that this pattern gives no indication of a specific effect of hormone treatment, and that monitoring, optimising and, if necessary, treating comorbidities and lifestyle factors (such as tobacco use) remain important.^[8] With no untreated transgender comparison group, the study could neither establish nor exclude a causal contribution from hormones. [Against general-population men the cardiovascular excess is borderline (SMR 1.4, 1.0–1.8); the authors attribute the 2010–18 peak partly to changes in case-mix.]

8.4 Smoking and age — both bear on clot risk, and continued smoking after a clot is where one guideline draws the line

UCSF guidance states that tobacco use with oestrogen therapy is associated with an increased risk of venous thromboembolism;^[18] the size of any interaction between smoking and modern oestradiol in trans women has not been reliably quantified. VTE prevalence rises with age.^[31,45] Smokers should be counselled on cessation at every visit, but smoking alone is not an absolute contraindication: a harm-reduction approach using transdermal oestrogen is preferred,^[18] and transdermal oestradiol has long been recommended for people over 40.^[10] That does not extend to continued smoking with a previous VTE or other substantial thrombotic risk: UCSF guidance (2016) states that, regardless of the circumstances, oestrogen should not be given to people with significant risk factors for, or a history of, VTE who continue to smoke.^[18]

8.5 Cancer — breast-cancer incidence lies between that of cisgender men and women, and PSA reads low

Breast. In a Dutch nationwide cohort of 2,260 trans women, 15 invasive breast cancers were identified after a median 18 years of hormone treatment. Incidence was 46 times that of cisgender men (SIR 46.7, 95% CI 27.2–75.4) and lower than in cisgender women (SIR 0.3, 0.2–0.4).^[5] The lower incidence relative to cisgender women was an overall finding. In the 30–50-year age group, the standardised incidence ratio was 1.0 (95% CI 0.5–1.8), based on nine cases; this imprecise subgroup estimate does not establish equivalence.^[5] The authors concluded that following cisgender breast-screening guidelines seems sufficient. Screening in practice also depends on age, duration of oestrogen exposure, family history and local guidelines.

Prostate. The prostate is retained after genital surgery. In a cohort of 2,281 trans women (median follow-up 14 years), only six prostate cancers occurred, a lower incidence than in reference males (standardised incidence ratio 0.20, 95% CI 0.08–0.42) — consistent with the authors’ hypothesis that androgen deprivation is protective; the cohort mostly used cyproterone.^[9] One screening caveat: androgen suppression substantially lowers PSA. In a Veterans Health Administration cohort of 210 trans women receiving oestrogen, median PSA across 852 tests was 0.02 ng/mL (IQR 0–0.2), the 95th percentile 0.6 ng/mL and the highest value recorded 2.21 ng/mL; in similarly aged cisgender men without known prostate cancer the median is about 1 ng/mL. PSA was undetectable in 36% of patients.^[33] The authors conclude that the conventional 4 ng/mL threshold is likely far too high in this population, while noting that the optimal cut point, and whether PSA screening benefits trans women at all, remain unknown; a 95th percentile in people without known cancer is a reference-distribution figure, not a validated cancer-detection threshold. Spironolactone was being taken at 89% of the PSA measurements and finasteride, which independently lowers PSA, at 30%, and 49% of measurements followed bilateral orchidectomy, so the figures reflect a mixed regimen rather than oestrogen alone. [No comparison group; PSA testing indications unknown; the cohort was 86% white and 2% Black, against roughly 16% Black in the wider VHA population.] Fixed numerical correction factors borrowed from cisgender men on 5α-reductase inhibitors have not been validated in trans women on feminising therapy, so a specific multiplier cannot be recommended. [Editorial judgement; no cited source evaluates such factors in trans women.]

8.6 Bone health — fracture risk depends on the comparison group and on adequate replacement

Inadequate oestrogen replacement, especially after gonadectomy if hormones are not continued adequately, is a recognised bone risk.^[2,56] Direct fracture outcomes are given in Section 7: over three years of emergency-department ascertainment, trans women aged 50 or over had higher fracture odds than age-matched reference men, and an estimate close to that of reference women that does not establish equivalence; below 50, neither comparison was statistically significant. The study does not isolate the effect of hormone treatment.^[34]

8.7 Liver — in the cited evidence, the main concern is cyproterone

The main liver consideration in the cited evidence is cyproterone acetate: drug-induced liver injury was among the serious adverse effects reported, in the trial abstract, in the cyproterone arm of the randomised comparison with spironolactone, supporting liver-enzyme monitoring on CPA;^[15] product information records hepatitis and hepatic failure, with fatal cases at 100 mg and above, mostly in men with advanced prostate cancer, and requires liver function tests before and regularly during treatment;^[65] a systematic review likewise reports severe hepatotoxicity at doses of 100 mg and above in the prostate-cancer literature;^[28] in a large cohort mean liver enzymes did not differ across doses of 10–100 mg,^[50] which cannot exclude rare serious injury at lower doses.

8.8 Mental health — improvement is commonly reported, but its size and cause are uncertain

A broad 2023 systematic review of 46 studies found that gender-affirming hormone therapy was consistently associated with reductions in depressive symptoms and psychological distress, while evidence for broad quality-of-life gains was inconsistent and the risk of bias across studies was highly variable.^[26] A prospective controlled study (35 people starting feminising therapy, against cisgender controls) found a statistically significant reduction in gender dysphoria within six months — smaller than the scale’s 11-point minimum clinically significant difference, on a measure the authors note is not validated for change over time — though quality-of-life domains did not shift significantly over that window.^[27] A 2025 WHO-commissioned systematic review of adults concluded that gender-affirming hormones may improve depression, quality of life and suicidality; its health and quality-of-life component included 13 studies published to March 2024. It rated its only randomised trial, which concerned testosterone, as low-certainty evidence and judged the uncontrolled longitudinal studies to be at serious to critical risk of bias, so it included no randomised trial of starting feminising therapy for these outcomes;^[59] randomised comparisons of particular feminising regimens exist,^[49] but answer different questions. Within the outcomes, populations and follow-up it covered, mainly psychosocial, it identified no substantive harms; that does not exclude uncommon or long-latency adverse effects and does not supersede the dedicated vascular and cancer evidence above.^[59] So observational studies commonly report improvement in depression and distress during gender-affirming hormone therapy; confidence in that reported association is greater than confidence in the size, persistence or causal attribution of benefit specifically to FHT, which bias and mixed-treatment populations limit. For dysphoria the controlled evidence is one short study. FHT should not be presented as a substitute for appropriate mental-health care where that is needed.

This section discusses serious health risks, including suicide as a cause of death in cohort data. It is general information, not personal medical advice; the relevance of each risk differs greatly between individuals and should be discussed with your own clinician. (See the support note at the end of this article.)

What this section establishes. Venous thromboembolism is a prominent serious vascular safety signal, far higher historically with ethinylestradiol, and mixed-regimen cohorts give no single risk for a specified modern regimen.^{[2,22,23,31,45]} Myocardial-infarction incidence has not been shown to be higher than in cisgender men, stroke findings are mixed, and excess mortality in one cohort could neither be attributed to hormones nor excluded.^{[6,7,8,22,23]} Breast-cancer incidence lies between that of cisgender men and cisgender women.^[5]

9. Special situations — older age and partial feminisation call for individual plans

9.1 Older adults — route and dose matter more, and stopping in later life has been studied only on a small scale

Older people more often have co-occurring conditions that may raise the risk of adverse effects, so the route and dose of oestrogen matter more. Transdermal oestradiol and the lowest effective dose are commonly preferred, cardiovascular risk factors are optimised, and treatment is individualised.^[18,20] UCSF guidance notes that transdermal oestrogen may be preferred where cardiovascular risk factors are present and supports informed decisions with existing cardiovascular disease;^[18] NICE menopause guidance in cisgender women advises considering transdermal rather than oral HRT where VTE risk is raised, and that for people with a history of coronary heart disease or stroke HRT is discussed with them and offered, if appropriate, by a clinician with expertise in menopause.^[20] [Largely extrapolated from cisgender HRT data.]

Whether to continue oestradiol in later life after gonadectomy has now been studied directly, though only on a small scale. In a single-group trial, 19 trans women aged 59–78, all after gonadectomy and a median 23 years on oestradiol, stopped transdermal oestradiol for 12 weeks and then restarted. Off treatment, 37% reported hot flushes and 26% mood swings; total menopausal-symptom scores fell substantially only after restarting (β −3.0, 95% CI −4.8 to −1.2), and mean depression, anxiety, quality of life and body satisfaction did not change. One participant chose to stop permanently. The trial was small, unblinded and placebo-free, reported no bone or cardiovascular outcomes (cardiometabolic data are to be reported separately), and 30 eligible women declined because they did not want to stop. The size and direction of changes varied between individuals for every outcome. Its authors concluded that short-term discontinuation produced oestradiol-deprivation symptoms but did not indicate psychological or body-image barriers for those who wish to stop, and that the symptoms may support continuing treatment in older age.^[57] In selected volunteers over twelve weeks, that does not predict how an individual would tolerate permanent cessation, or establish long-term bone or cardiovascular safety. UCSF guidance (2016) notes that there is no evidence to support either continuation or cessation in older trans women, and considers it reasonable to consider stopping around age 50 after gonadectomy, by analogy with the age of menopause, through shared decision-making;^[18] there is no established age at which oestradiol must be stopped. A decision to stop takes in symptoms, personal goals, vascular and cancer considerations and a plan for bone health, including bone-density measurement, which the Endocrine Society recommends for people who stop sex-hormone therapy after gonadectomy when osteoporosis risk factors are present.^[2] Bone protection after gonadectomy otherwise depends on adequate replacement (Section 7).

9.2 Non-binary people and partial feminisation — effects cannot be selected independently

Not everyone wants full feminisation. SOC-8 explicitly recognises non-binary people, who may seek partial or specific changes.^[1] Regimens can be adjusted to such goals — for example, lower doses, oestrogen without an anti-androgen, or an anti-androgen without oestrogen — but their effects cannot be selected independently: SOC-8 notes that the possibility of breast growth cannot be avoided by someone who wants other effects of oestrogen, and that its extent is highly variable;^[1] this means accepting that the evidence base for partial or non-standard regimens is thinner than for standard full-dose feminisation, so expectations and monitoring should be discussed individually. Regimens that markedly suppress androgen production without providing adequate replacement sex steroid can produce hypogonadal symptoms and, if sustained, may compromise bone health; partial-feminisation regimens therefore need attention to adequate sex-steroid exposure as well as to the desired physical effects.^[1,2]

What this section establishes. In older adults, route and dose matter more, and stopping oestradiol in later life has been studied only on a small scale.^[18,20,57] In partial feminisation, effects cannot be selected independently.^[1]

10. Evidence quality — confidence varies by topic, and an observed association is rated apart from its cause

A core aim of this review is to be explicit about how strong the evidence is for each claim. The ratings use a simple traffic-light scheme — 🟢 reasonably robust, 🟡 moderate, 🔴 limited — assigned by judging four things: the type of evidence (randomised trial, prospective cohort, expert consensus), its consistency, how directly it applies to trans women, and the size of the effect relative to its uncertainty. In the text and table, Low corresponds to 🔴, Moderate to 🟡 and High to 🟢; a judgement that falls between two levels is written as a range (for example, “Low to Moderate”) and shown with both marks. These are editorial judgements, not formal GRADE ratings. In this version the judgements also take account of risk of bias, applicability to the population and regimen, imprecision, consistency, overlap between cohorts and possible publication bias, though no formal instrument was applied to any of them. Confidence that an association was observed is not confidence that hormones caused it, and a clinical recommendation may rest partly on consensus; where these differ, the table rates them separately. For more on how we weigh evidence in this field — why randomised trials are scarce, why cohort studies dominate, and how we treat figures borrowed from cisgender populations — see Reading the Evidence.

What this section establishes. Confidence varies by topic. An observed association is rated separately from confidence that hormones caused it, and these are editorial judgements, not formal GRADE ratings.

11. Frequently asked questions — what the evidence answers, and what it leaves open

Can I choose patches over tablets? Often, yes — a shared decision with your clinician. Transdermal oestradiol avoids first-pass liver metabolism and is associated with lower clot risk in cisgender women, so it is frequently preferred for people at higher clot risk, including smokers;^[18,19,46] in trans women, clotting-marker data point the same way,^[47] but no adequately powered comparison of clinical VTE outcomes by route in trans women has been identified, and the one observational comparison found, of thrombotic events of all kinds, had too few events to be informative.^[51]

Does progesterone improve breast growth? Not established. SOC-8 found no quality evidence of benefit, and a small controlled study of 100 mg added to established therapy detected no difference from controls in Tanner stage, sleep quality or distress at three months. An open-label randomised trial in people on long-term therapy after gonadectomy, measuring breast volume by 3D imaging, has reported results only in conference reports. Its largest control-referenced estimates were in the arms combining doubled oestradiol with progesterone, but the abstract does not establish an interaction, a preferred regimen or long-term safety, and does not fully specify its units or analysis. No peer-reviewed outcome paper had been identified as of September 2026, and none was found in a targeted web search on 12 October 2026, so this remains preliminary.^{[1,13,14,36,41]} A signal of added clot risk has been reported with progestogens, possibly stronger with medroxyprogesterone acetate, but it rests on very few events.^[31,51]

Can hormones change my facial structure? Hormones soften the face by redistributing fat but do not reverse the facial bone structure established by an androgen-driven puberty; bony change requires surgery (Section 5).

Can fertility return after stopping? It can: recovery has been documented, but its probability, timing and completeness remain uncertain. In a published series of nine people, all eventually produced sperm — most on the first test, one only after a surgical procedure at seventeen months — and five of the nine had impaired semen quality after stopping. Nine self-selected people is a thin basis for a general expectation, and the authors still recommend banking sperm before starting, because recovery can take many months off treatment, during which rising testosterone is likely to carry physical and psychological costs, and because semen quality is often reduced after treatment.^[37] FHT is not reliable contraception while the testes remain,^[1,18] and bilateral orchidectomy removes testicular sperm production permanently.

How long before I see skin changes? Usually softer, less oily skin within the first few months (~3–6 months).^[2]

Why do results differ so much? Largely unexplained; genetic variation is one suggested explanation.^[3] For breast development, a large prospective study found that no variable examined, including oestradiol level, predicted the outcome at one year,^[3] though a later trial found exploratory associations with age and oestradiol level.^[49]

Will higher oestrogen doses give more feminisation? UCSF guidance (2016) states that, once testosterone is well suppressed, there is no evidence that higher oestradiol adds feminisation or breast growth;^[18] a 2025 systematic review, searching to April 2023, found that the guideline range of 100–200 pg/mL lacks a primary evidence base and has not been shown to optimise feminisation or reduce adverse events, which does not make higher concentrations equally safe.^[30]

What this section establishes. The answers above restate findings set out in Sections 2 to 9, with one addition: genetic variation is named as one suggested explanation for differences in breast development.^[3]

12. Clinical bottom line — feminisation is reliable in direction and variable in degree

1. Oestradiol reliably produces feminising changes in most adults; the degree varies widely.^[2,3]
2. 17β-oestradiol is the appropriate oestrogen; ethinylestradiol is no longer recommended and has largely been abandoned in FHT because of its higher clot risk.^[2,31]
3. Reducing testosterone action is central, though greater suppression has not been shown to add breast development;^[49] once testosterone is suppressed, higher oestradiol has not been shown to add feminisation,^[18] and the guideline range of 100–200 pg/mL lacks a primary evidence base, though that does not make concentrations above and below it equivalent.^[30]
4. Breast development is usually modest, begins within the first few months, is often greatest during the first year, and may continue more gradually thereafter; no variable examined has consistently predicted it.^[3,4,49]
5. Route may matter for clot risk: transdermal oestradiol is associated with lower clot risk in cisgender data and with smaller rises in clotting markers in trans women, though no adequately powered comparison of clinical VTE outcomes by route in trans women has been identified, and the one observational comparison found, of thrombotic events of all kinds, had too few events to be informative.^{[19,46,47,51]}
6. Tobacco use with oestrogen is associated with increased clot risk, and smoking is weighed in choosing the route; cessation is worth raising at every visit, though continued smoking is not an absolute bar to treatment.^[18]
7. VTE is a prominent serious vascular safety signal in treated transfeminine cohorts, on a low absolute base, but mixed-regimen cohorts do not give a single risk for a specified modern regimen. Myocardial-infarction incidence has not been shown to be higher than in cisgender men, and one recent analysis reported it lower, which does not establish a protective effect; stroke estimates range from a clear increase in some earlier data, particularly among people with documented oestrogen initiation followed beyond six years, to no statistically detectable increase in the most recent cohort adjusted for socioeconomic status. Comparisons against cisgender women give different estimates because they use different background rates. The authors judged the pattern of excess mortality not to indicate a direct hormonal effect, but the study could neither establish nor exclude one, and could not compare hormone regimens.^{[6,7,8,22,23]}
8. Breast cancer risk is higher than in cisgender men but, overall in this cohort, lower than in cisgender women; prostate cancer is rare under androgen suppression; screening follows anatomy and exposure.^[5,9]
9. Cyproterone acetate suppresses testosterone effectively, and mean suppression at 10 mg daily was similar to higher doses in one cohort, though few started on 10 mg and not everyone was suppressed. Its harms — liver toxicity, thromboembolism, depressive mood and fatigue, raised prolactin, lower HDL and cumulative-dose meningioma — need separate assessment; current or previous meningioma is a contraindication, and regulators advise that the lowest effective dose should be used.^{[15,17,24,50,64,65]}
10. Evidence for a progesterone benefit remains limited and is evolving, and a signal of added clot risk with progestogens, possibly stronger with medroxyprogesterone acetate, rests on very few events;^{[1,13,14,41,51]} baseline assessment and regular clinical and laboratory monitoring are recommended, though evidence that a particular schedule reduces major events is limited.^[1,2]
11. In 19 selected older volunteers after gonadectomy, stopping oestradiol for twelve weeks produced menopause-like symptoms in some, easing on restarting, while mean depressive-symptom, anxiety, quality-of-life and body-satisfaction scores did not change, although happiness fell slightly. The study does not predict individual tolerance of permanent cessation or establish its long-term safety, and there is no established age at which oestradiol must be stopped.^[18,57]
12. FHT is not reliable contraception. Fertility preservation is discussed before treatment, and reproductive goals are revisited during care and before gonadectomy.^[1,2,18]

What this section establishes. This section gathers the findings of Sections 2 to 10, adding one detail of the later-life trial: happiness fell slightly.^[57]