5α-reductase Inhibitors: Finasteride and Dutasteride in Feminising Therapy

Evidence Review · ER-013

Part I — A different kind of drug

1. What the 5α-reductase inhibitors do

Mechanism — the conversion step, and only that. The other androgen-directed drugs used in feminising care act predominantly in one of two places: they suppress gonadal production of testosterone (GnRH agonists; cyproterone acetate, which also antagonises the receptor), or they antagonise the androgen receptor (spironolactone, bicalutamide) — and several do more than one thing, imperfectly. The 5α-reductase inhibitors act in a third place, downstream of both: they inhibit the enzyme that converts testosterone into dihydrotestosterone (DHT), an androgen with greater androgen-receptor affinity and generally greater transcriptional activity than testosterone in androgen-responsive tissues.^[1,4] They do not directly suppress testosterone production, and they do not occupy or antagonise the androgen receptor — testosterone and other androgens still activate it. Their androgen-directed effect is therefore partial by design, narrower than a production-suppressor or a receptor antagonist.^[3]

Two drugs, two reaches. The enzyme comes in isoforms with different tissue homes. Finasteride selectively inhibits the type II isoform — with type III activity reported in some sources — and lowers circulating DHT by roughly 70%.^[4,14] Dutasteride inhibits both type I and type II and lowers circulating DHT more completely, by around 90% or more; the exact figures vary with dose, duration and whether serum or tissue is measured, and serum DHT suppression is not the same thing as scalp-tissue suppression or clinical benefit.^[4,14] The type I isoform is expressed in the pilosebaceous unit — the sebaceous gland and its follicle in the skin — so dutasteride, by adding type I blockade, has a broader pharmacologic reach into skin and sebaceous tissue, while type II predominates in the hair-follicle outer root sheath, genital skin and prostate; the isoform map is overlapping rather than tidy.^[3,4] The two drugs also differ in a way that matters for stopping: finasteride has a half-life of a few hours, dutasteride of around five weeks, so dutasteride washes out far more slowly — relevant to assessing whether a side effect is drug-related and to any pregnancy-exposure precaution.^[14,20] In the United States, finasteride 1 mg is FDA-approved for male-pattern hair loss in men and 5 mg for benign prostatic hyperplasia; dutasteride is approved for the prostate and used off-label for hair — not FDA-approved for alopecia. All of these figures and licensed uses derive from cisgender men; their translation to a feminising regimen is assumed, not measured, and finasteride’s approval is sex-specific.

The consequence that trips people up: testosterone goes the wrong way. Because a 5α-reductase inhibitor blocks the route by which testosterone is converted to DHT, less of it undergoes that conversion — so serum testosterone can rise. In cisgender men taking finasteride for hair loss, total testosterone rises around 9%; in the small transfeminine series above, the mean rise was 92 ng/dL.^[1] This matters for interpretation more than for safety: someone whose testosterone number climbs after starting the drug has not “got worse” — total testosterone simply does not measure the fall in DHT, nor tissue-level androgen action, so the number and the tissue effect can move in opposite directions.^[1] The caveat, taken up in Section 5, is that not every individual shows a measurable rise, and a persistently unsuppressed level still warrants interpretation.

A quieter mechanism, worth flagging with care. 5α-reductase does not only make DHT. The same enzyme is the first step in converting progesterone and deoxycorticosterone into neuroactive steroids — most notably allopregnanolone, a positive allosteric modulator of the GABA_A receptor, the main mediator of fast inhibitory signalling in the brain.^[16] Allopregnanolone enhances GABAergic inhibition as benzodiazepines do, but the comparison should not be overdrawn: it binds a distinct site on the receptor from benzodiazepines and barbiturates, with its own subunit preferences and pharmacology, so it is a different molecule producing an overlapping direction of effect, not an interchangeable one.^[16] How much this matters clinically is uncertain, but the pathway is pharmacologically important: brexanolone is a pharmaceutical formulation of allopregnanolone, and although its FDA approval was withdrawn for commercial reasons in 2025, the related oral neurosteroid zuranolone remains approved for postpartum depression.^[16,21] Inhibiting 5α-reductase can alter the synthesis of allopregnanolone and related neurosteroids, and that is a biologically plausible route by which a drug taken for hair could affect mood and anxiety — one proposed mechanism, its magnitude and central consequences in treated humans not established (Section 5).^[16] It is named here because it is easy to forget these drugs touch a neurosteroid pathway at all.

Confidence: 🟢 High (mechanism, isozyme distribution, DHT reduction, and that these drugs do not lower — and can raise — testosterone) / 🟡 Moderate (the size of any of these effects in transfeminine people specifically, where the data are near-absent; the clinical significance of the neurosteroid pathway).

Part II — The core argument

2. Why a large additive effect is unlikely

The case for a 5α-reductase inhibitor in feminising care rests on a simple chain: feminisation means lowering androgen action; DHT is an androgen; so blocking DHT should help. The chain is intact but the arithmetic weakens it. DHT is made from testosterone, so when suppression is reliable and testosterone sits in the intended feminising range, the principal substrate for DHT formation is reduced, and a large additive effect from blocking the conversion becomes biologically less likely.^[1] Two honest caveats sit alongside that. Serum testosterone is not a direct measure of intrafollicular or intracutaneous androgen availability, and tissue androgen metabolism is not captured by serum assays; no study has measured whether conversion is truly negligible in skin or scalp during adequately suppressive treatment. So the claim is not that DHT blockade does nothing — it is that the incremental value is uncertain and, on the mechanism, likely small.

This is not a fringe reading. The most direct appraisal of the question — Irwig’s 2021 commentary — states that it is unclear finasteride will add any clinical benefit once testosterone has already been lowered with standard treatment.^[1] Note the exact claim: additive benefit is unclear, not disproved. No transfeminine trial has tested the addition, so the case for added feminisation is inference resting on inference: that DHT blockade matters, and that it still matters when its substrate is already scarce. Neither step has been demonstrated in this population.

Put the mechanism from Section 1 alongside it and the argument sharpens. A 5α-reductase inhibitor lowers DHT but can raise testosterone; when testosterone is already well suppressed, the clinical importance of that reciprocal change is unknown and may be small. Where testosterone is not well suppressed — the drug used as a lone or partial agent — more substrate remains, but so does more testosterone available to the receptor, and testosterone is itself an androgen. The drug is at its most defensible neither as an add-on to good suppression nor as a stand-in for it, but only where a specific DHT-driven, tissue-local process persists despite low testosterone. That is the exception, and it is the subject of the next section.

Lowering testosterone also lowers the principal substrate for DHT — so the incremental value of blocking the conversion is uncertain, and on the mechanism, likely small.

Confidence: 🟡 Moderate. The argument is mechanistically sound and supported by the one authoritative appraisal, but it is reasoned inference, not a trial result: no study has directly tested whether adding a 5α-reductase inhibitor to a suppressive regimen changes any feminising outcome, or measured tissue DHT during treatment.^[1]

3. The narrow, indication-specific uses

Uncertain in general is not useless in particular. There are a few situations where a 5α-reductase inhibitor has a real, if modest, rationale — and naming them precisely, with their evidence gaps, is more useful than a blanket yes or no.

Scalp-hair loss (androgenetic alopecia). This is the strongest use, and it is a dermatological one rather than a feminising one. Androgenetic alopecia is driven by DHT acting locally at genetically susceptible scalp follicles — a matter of follicular androgen sensitivity and local metabolism — so it can progress even when serum androgens are low, and serum levels alone do not fully predict its course. That is exactly the circumstance where a DHT-specific blocker still has something to do. Two prior points, though. First, presumed androgenetic alopecia should be confirmed, not assumed: telogen effluvium, iron deficiency, thyroid disease, traction and scarring alopecias and medication-related shedding can mimic or coexist with it, and warrant a scalp assessment before a drug is started. Second, testosterone suppression itself may slow or stabilise androgenetic hair loss, which makes any additional drug effect hard to isolate. Reviews of hair-loss therapeutics in transgender and gender-diverse people place topical minoxidil as a common first-line or foundational treatment and list finasteride and dutasteride as options selected by pattern, goals, contraindications and shared decision-making rather than a fixed sequence.^[5] Expectations should be set honestly: the benefit is usually maintenance — slowing or halting loss and preventing further miniaturisation — more than regrowth of advanced loss, and it builds over months. In cisgender men that efficacy is well established — the landmark randomised trial of finasteride 1 mg showed slowed progression and increased scalp hair count over two years.^[17] The efficacy evidence is overwhelmingly from cisgender men with male-pattern hair loss; the transgender-specific literature is small and largely observational, and finasteride’s licensing is for men.

Residual androgenic skin and hair once testosterone is suppressed. Where oily skin, acne or androgen-pattern hair persists despite verified suppression, there is a theoretical rationale for adding DHT blockade — and dutasteride’s type I (pilosebaceous) reach is the pharmacologic argument for it.^[3] But these are biologically distinct endpoints — sebum, acne, facial hair and body hair differ in their androgen dependence and treatment response — and no direct evidence establishes that 5α-reductase inhibition improves any of them in adequately suppressed transfeminine patients. Each should first be assessed on its own terms and treated with established measures before a drug trial. The frequency and magnitude of such residual features after verified suppression have not been well quantified. And established terminal facial hair is not reliably eliminated by any androgen-directed drug, 5α-reductase inhibitors included; effects on newly emerging transfeminine facial hair specifically have not been quantified, and durable removal needs electrolysis or laser, as set out in the anti-androgens review.

Single-agent or partial use — an inadequately studied practice, not a “milder” option. Some older clinical guidance describes off-label single-agent use when only partial reduction of DHT-mediated effects is wanted, or when other agents are unsuitable.^[3] This should not be read as a gentler or lower-strength version of testosterone suppression or receptor blockade: it produces a different, incomplete pattern of androgen modification, leaving testosterone itself unopposed. The supporting guidance is from 2016, predates SOC8 and the newer safety reviews, and there are no transfeminine outcome data. It is better treated as an under-studied practice to discuss honestly than as a “defensible use.”

Confidence: 🟡 Moderate (scalp androgenetic alopecia as a DHT-driven, tissue-local process a 5α-reductase inhibitor can address, on cisgender-dominant evidence) / 🔴 Low (benefit for residual skin or hair features, and single-agent use, in transfeminine people specifically — theoretical, essentially unstudied).

Part III — Evidence and safety

4. The efficacy evidence, graded

Stated plainly: there is no controlled trial of a 5α-reductase inhibitor for feminising outcomes in transfeminine people. What exists is a commentary appraising the question,^[1] usage data showing the drugs are prescribed anyway,^[1] dermatological reviews of hair-loss therapeutics that include transgender populations largely by extension from cisgender data,^[5] and a large body of cisgender evidence on DHT reduction, prostate and hair.^[4] None of it establishes that adding a 5α-reductase inhibitor to a feminising regimen improves breast development, body or facial hair, or any other feminising endpoint.

The one adjacent piece of controlled evidence comes from the transmasculine side. A small randomised, double-blind, placebo-controlled pilot in trans men found that adding dutasteride to testosterone did not impair testosterone’s anabolic effect on muscle over the study’s limited follow-up — evidence that conversion to DHT was not required for that particular outcome.^[12] It is out of scope as a feminising result, and it cannot show that blocking DHT “changes little” generally — it says nothing about hair, skin, genital tissue or sexual function. Read narrowly, it is one more reminder that DHT is one channel of androgen action rather than the whole of it.

The usage–evidence gap is the striking feature. At one endocrinology clinic, 30% of 166 trans women were on finasteride^[1] — frequent prescribing at that centre despite the absence of direct transfeminine efficacy evidence, though how representative it was of other centres or current care is unknown. That gap is the honest headline of this section: a practice resting on theory more than data, at least where it is used.

Confidence: 🔴 Low. There is no transfeminine efficacy trial; the grade reflects near-absence of direct evidence, not evidence of no effect.

5. Safety and uncertainties

The safety profile is better characterised than the efficacy, but almost entirely from cisgender men — and between 2024 and 2026 the regulatory picture moved substantially. The distinctions below matter: what a regulator lists as an associated effect, what has been confirmed as causal, and what is a precautionary class extension are three different things, and they differ by drug and by jurisdiction.

Mood and suicidality — signal, epidemiology, and regulation, kept separate. There is a biologically plausible route from these drugs to low mood via the neurosteroid pathway of Section 1, and pharmacovigilance has long shown disproportionate reporting of depression and suicidality, concentrated in younger men on finasteride for hair loss.^[9] Pharmacovigilance generates or strengthens a signal; it cannot measure incidence or establish cause. The controlled epidemiology is weak and inconsistent: a large cohort found increased self-harm and depression in the first 18 months, though not completed suicide;^[6] a meta-analysis returned a pooled estimate compatible with anything from no increase to a modest one, its confidence interval crossing the null (hazard ratio 1.23, 95% CI 0.99–1.54) — imprecise evidence, not evidence of absence;^[7] and a nationwide comparison found finasteride no worse than dutasteride, with prior mood disorder the stronger predictor.^[8] These cohorts are also vulnerable to confounding by indication, since androgenetic alopecia itself affects psychological wellbeing. What changed is the regulatory reading. Following a UK review in 2024 — which strengthened warnings and introduced patient alert cards^[22] — a European review concluded in 2025 that suicidal ideation is a confirmed side effect of finasteride 1 mg and 5 mg tablets, frequency unknown; it did not establish a causal link for dutasteride, but added the finasteride mood-change information to dutasteride’s product information as a precautionary class measure.^[15] In May 2026 the UK’s MHRA strengthened its guidance further, identifying depression and suicidal ideation as psychiatric risks of finasteride and warning that associated sexual dysfunction may persist after stopping, and advising prescribers to ask about a history of depression or suicidal ideation before starting and to review patients regularly.^[13] The honest reading: a plausible mechanism, a regulatory conclusion that suicidal ideation should be listed as an adverse effect of finasteride tablets (frequency unknown) with precautionary wording for dutasteride, and cohort evidence whose magnitude remains uncertain. Whether any of this transfers to transfeminine people — who differ in baseline hormonal milieu, in the direction of the intended change, and often in why they take the drug — is entirely unstudied.

A caution on the reporting counts. UK Yellow Card data from 1994 to 31 May 2025 included 170 reports of suicidal-ideation-related terms and 19 fatal suicide reports for finasteride, against five and none for dutasteride.^[13] These spontaneous-report counts have no exposure denominator and are shaped by prescribing volume, market duration, publicity-driven reporting and differential recognition; they cannot be used to estimate incidence or compare risk between the drugs, and a suspected report does not by itself establish cause.

Sexual effects — which land differently here. In cisgender men, 5α-reductase inhibitors are associated with reduced libido, erectile difficulty and reduced ejaculate — reported in roughly 3.4–15.8% depending on dose, indication and how they were ascertained — and reports of symptoms persisting after stopping, now reflected in the strengthened labelling.^[10,13] In a feminising context this reads differently: reduced erectile function and libido may be unwanted, neutral, or aligned with the person’s goals, and the framing should be individual rather than imported wholesale from a cisgender-male harm list. On persistence, the regulators are more specific than the advocacy literature: the UK guidance explicitly acknowledges that sexual dysfunction associated with finasteride may persist after discontinuation, while whether psychiatric symptoms persist because of prior exposure is less clearly established; the frequency and causality of any persistent symptoms remain uncertain.^[11,13]

Fertility. Because 5α-reductase inhibitors affect the male reproductive tract, they can reduce semen volume and, in some men, sperm parameters. For a transfeminine patient who retains testes and may wish to preserve fertility, that is a relevant, easily-overlooked consideration. Fertility goals and any wish for gamete preservation are best discussed before starting or changing the overall feminising regimen — the stronger urgency usually attaches to the oestrogen and testosterone suppression, with any additional effect of 5α-reductase inhibition on semen parameters folded into that conversation.^[19,20]

Teratogenicity, and handling — drug-specific. This one is not contested in substance. 5α-reductase inhibitors can disrupt development of the male external genitalia — hypospadias is one possible abnormality, not the whole concern — so product information contraindicates exposure in pregnancy, though the exact wording (“contraindicated,” “not for use in women,” and so on) varies by drug and jurisdiction.^[10,14] Handling is drug-specific: someone who is or may be pregnant should avoid contact with crushed or broken finasteride tablets and with leaking dutasteride capsules, since the active drug can be absorbed through the skin, while intact coated finasteride tablets are designed to prevent contact in ordinary handling.^[14,19,20] For most transfeminine patients this is a handling, partner and household consideration rather than a personal-pregnancy one, but it belongs in any honest account.

The prostate, and PSA. Most transfeminine patients retain a prostate, including after standard vaginoplasty, and 5α-reductase inhibitors act on it — reducing serum prostate-specific antigen (PSA) by around 50% on sustained treatment, though the response varies between individuals and over time.^[14,19] Anyone on one of these drugs who has a PSA measured needs it interpreted with that in mind: the conventional practice of doubling the value after about six months is a cisgender-male, BPH-derived approximation and has not been validated in transfeminine populations, so a treatment-specific baseline and any confirmed rise from the on-treatment nadir matter more than a single arithmetic correction. Oestradiol and androgen suppression may already lower PSA, and screening thresholds in this population are not well established. The practical point stands: read PSA carefully on these drugs, or a real problem can be missed.

Other. Gynaecomastia and breast tenderness are recognised in cisgender men; in transfeminine patients on oestradiol, attribution differs and breast growth is expected, but a new or unilateral breast mass still warrants assessment on its own merits. Dutasteride is extensively hepatically metabolised, and some product information contraindicates or advises against its use in severe hepatic impairment — the applicable local label should be consulted.^[14]

Monitoring wrinkle. The testosterone change from Section 1 is the practical trap. Total testosterone may rise modestly on a 5α-reductase inhibitor and should not be expected to fall because of the drug; not every individual shows a measurable rise, but a rise, where it occurs, does not by itself signal treatment failure. DHT — not testosterone — is the analyte the drug acts on, and it is not routinely measured. A persistently unsuppressed testosterone still needs interpreting in context — regimen, adherence, absorption, goals and symptoms — rather than waved away; the narrower point is that the change attributable to the drug itself is not failure.^[1]

A note on screening. Asking about a history of depression or suicidal ideation before prescribing, as regulators now advise, is a sensible precaution — but it is a risk-minimisation measure, not a validated predictor, and a clear baseline history does not exclude risk in someone with no prior psychiatric illness.^[13]

Confidence: 🟢 High (teratogenicity; the direction of the testosterone effect; that these drugs lower PSA; that regulators concluded suicidal ideation should be listed for finasteride and added precautionary wording for dutasteride) / 🟡 Moderate (the plausible neurosteroid mechanism; sexual-effect frequencies and the possibility that sexual dysfunction persists after stopping, from cisgender data) / 🔴 Contested or uncertain (the epidemiological magnitude of the mood/suicidality risk; whether any psychiatric effect persists after discontinuation; and whether any of it applies in transfeminine people, which is unstudied).

Part IV — The verdict

6. Guideline stance and practice

The major guideline is explicit. WPATH’s Standards of Care version 8 recommends against the routine use of 5α-reductase inhibitors as part of feminising hormone therapy, grouping them with other strategies it judges unsupported by data, on the basis of lack of evidence and potential risks.^[2] Read it precisely: it is a recommendation against routine use as a feminiser, built on absence of efficacy evidence rather than evidence of harm to feminisation — a “not shown to help, and not without risk” verdict, not a “shown to harm” one, and not a blanket bar on every indication. Earlier Endocrine Society guidance predates much of this evidence base and does not address the 5α-reductase inhibitors in comparable detail, so WPATH is the guideline that speaks to the question directly.^[18]

Set against that is the reality that the drugs are used — the 30% figure — and that a specific, defensible use exists, chiefly scalp-hair loss.^[1,5] The two are reconcilable, and the distinction the evidence supports is the indication, not the drug: routine use as a feminiser is hard to justify, while treating genuine androgenetic alopecia in a transfeminine patient — an established dermatologic indication for finasteride in cisgender men, extrapolated to transfeminine care and often off-label under sex-specific labelling — is a separate, defensible dermatological decision, provided the diagnosis is confirmed and the limits of the evidence are stated. Preferring dutasteride over finasteride for residual skin or hair on mechanism alone goes beyond what outcome evidence supports.

The practical posture that follows is careful, informed shared decision-making. Where someone requests or is using a 5α-reductase inhibitor, the appraisal that best fits the evidence is Irwig’s: in the absence of efficacy and safety data in this population, discuss it honestly — what it can and cannot do, the testosterone paradox, the psychiatric and sexual risks, the fertility and teratogenicity points, and the PSA effect — and decide together.^[1] Current regulatory advice adds concrete steps: ask about a history of depression or suicidal ideation before prescribing, counsel on the psychiatric adverse effects and on sexual dysfunction (including the possibility that sexual dysfunction may persist after stopping), and review during treatment — bearing in mind these are precautions, not guarantees.^[13] None of this is a reason to start or stop one alone; it is a reason to be clear about how little is settled.

Confidence: 🟢 High (that WPATH SOC8 recommends against the routine use of 5α-reductase inhibitors as feminising hormone therapy; that regulators strengthened psychiatric/sexual warnings across 2024–2026) / 🟡 Moderate (the indication-specific reconciliation offered here, which is a reasoned position rather than a guideline one).

7. Evidence quality

The grading, gathered in one place. The mechanism is firm and the safety directions are firm; the feminising efficacy is essentially unstudied; the mood magnitude is contested even as the regulatory listing is now clear. Ratings use the series scheme — 🟢 reasonably robust, 🟡 moderate, 🔴 limited — as editorial judgements, not formal GRADE. See Reading the Evidence.

8. Frequently asked questions

Does finasteride help me feminise? There is no evidence that adding it to a regimen already suppressing testosterone improves feminisation, and the main guideline recommends against its routine use as a feminiser. Its clearest evidence-supported indication is scalp androgenetic alopecia. (Sections 2, 3, 6)

My testosterone went up after starting finasteride — is that bad? A modest rise can occur and is pharmacologically expected — these drugs block the conversion of testosterone to DHT, so the number can climb even as DHT falls, and that is not failure. Not every patient shows one. A persistently unsuppressed level is still worth interpreting with your clinician, but testosterone does not measure what this drug is doing. (Sections 1, 5)

Finasteride or dutasteride? Dutasteride lowers DHT more completely and also blocks the type I enzyme in the skin’s sebaceous unit, so it has a broader pharmacologic effect; finasteride is the one licensed for hair loss (in men). No transfeminine outcome evidence shows one works better than the other, and dutasteride lingers far longer after stopping. (Sections 1, 3)

Will it get rid of my facial hair? No. No androgen-directed drug reliably eliminates established terminal facial hair; that needs electrolysis or laser. A 5α-reductase inhibitor’s effect on transfeminine facial hair specifically has not been quantified. (Section 3)

Is it dangerous for mood? There is a plausible mechanism and a reporting signal, and across 2024–2026 regulators strengthened the warnings — European authorities concluded suicidal ideation should be listed as a side effect of finasteride tablets (frequency unknown), and UK guidance identifies depression, suicidal ideation and sexual dysfunction as risks and warns that the sexual dysfunction may persist after stopping, advising a check for a depression or suicidal-ideation history before prescribing. How large the risk actually is remains uncertain from the cohort evidence, and none of it is in transfeminine people. Worth knowing about, screening for, and watching. (Section 5)

9. Clinical bottom line

1. A 5α-reductase inhibitor is not a feminising agent and not a substitute for a testosterone-lowering agent; it blocks only the DHT conversion step.^[1,4]
2. When testosterone is reliably suppressed, the substrate for DHT is reduced and a large additive benefit is unlikely — though this is inference, not a measured result.^[1]
3. Expect the total-testosterone number to rise or hold rather than fall because of the drug; the drug-related rise is not failure, though a persistently unsuppressed level still warrants interpretation.^[1]
4. The clearest indication-specific use is confirmed scalp androgenetic alopecia (with minoxidil commonly first-line). Additional use for residual acne, seborrhoea, facial or body hair after verified suppression remains theoretical, and established condition-specific treatments should take priority. There is insufficient evidence to prefer dutasteride over finasteride on clinical-outcome grounds.^[3,5]
5. Counsel on the strengthened psychiatric and sexual warnings (screen for a depression or suicidal-ideation history first), the fertility, teratogenicity and handling points, and the PSA effect; WPATH SOC8 recommends against routine use as a feminiser, and careful informed shared decision-making is the honest posture.^[2,10,13]