What this is

Nine parallel research threads, roughly 250 sources, every citation carrying a DOI or PMID retrieved from PubMed rather than written from memory. Each agent was instructed to report null results, contradictions, and its own failed searches, and to say so plainly when it could not verify an identifier.

The headline is uncomfortable and worth stating first.

Executive summary

The chain that popular menopause content draws (testosterone falls at menopause, tyrosine hydroxylase drops, dopamine drops, so you get brain fog, low mood, and low libido) is broken at four of its five joints.

1. Testosterone does not fall at menopause. Androgens decline gradually with age from the twenties. Prospective data anchored to the final menstrual period show total testosterone flat across the transition.
2. There is no androgen response element in the tyrosine hydroxylase promoter. The androgen effect on TH, where it exists, is indirect.
3. TH protein level is not a readout of dopamine signaling. The enzyme is clamped by feedback inhibition from its own product and controlled by phosphorylation state, not abundance.
4. There is no human study measuring dopamine synthesis capacity across the menopause transition. None. The dopamine-and-menopause story has no direct human imaging support in menopausal women.
5. Only the last joint holds: testosterone genuinely improves sexual desire in postmenopausal women with HSDD, at high evidence quality. But nothing links that effect to a measured dopaminergic change in a human.

What survives is more interesting than what fails. The best-evidenced hormonal lever on the dopamine system is estradiol, not testosterone. The most actionable modifiable node is iron, and it reverses direction across the transition. The most useful explanatory idea for why women's experiences differ so wildly is the COMT inverted-U. And the most obviously missing study in the entire field is locus coeruleus imaging across the menopause transition.

---

Part 1: The enzyme

What tyrosine hydroxylase actually does

TH converts tyrosine to L-DOPA, which becomes dopamine, then norepinephrine, then epinephrine. It needs three things simultaneously: tetrahydrobiopterin (BH4) as cofactor, molecular oxygen, and iron in the ferrous state, held in a 17-angstrom-deep active site cleft. Ferric iron is catalytically dead, which is why oxidative conditions inactivate the enzyme.

It is expressed in substantia nigra (motor control), ventral tegmental area (motivation and reward), locus coeruleus (noradrenergic, stress and arousal), the arcuate nucleus neurons that gate prolactin, sympathetic neurons, and the adrenal medulla.

The feedback switch, finally solved

A 2022 cryo-EM structure resolved a mechanism the field had hand-waved about for decades. Dopamine binds the catalytic iron as a bidentate complex, and then a roughly 20-residue helix from the enzyme's own N-terminal tail (residues 39 to 58) folds down into the active site and physically plugs it shut. Phosphorylating Ser40 forces that helix out.

Numbers that matter:

The practically important detail: catecholamine inhibition of TH is competitive with BH4, not with tyrosine. Raising BH4 displaces bound dopamine and reactivates the enzyme without any phosphorylation. This is the mechanistic reason the BH4 arm of this story is more interesting than the tyrosine-supplement arm.

"Rate-limiting" is conditional, not absolute

The claim is well established and it is a statement about flux control under normal conditions. TH stops being the controlling step when BH4 supply falls (GCH1, PTPS, or sepiapterin reductase deficiency), when AADC is deficient, and when vesicular storage is compromised (blocking VMAT2 raises cytosolic dopamine, which feedback-inhibits TH, so storage capacity indirectly gates synthesis).

There is also a live challenge to the textbook version. A 2012 study reverse-dialyzed tyrosine into rat brain with AADC blocked and found extracellular DOPA rose linearly to 250% of control in striatum and 300% in prefrontal cortex. The authors concluded brain TH may not be near full saturation with tyrosine, and that mechanisms other than tyrosine hydroxylation may matter more for acute regulation than previously appreciated. This is unresolved and it is exactly the question that decides whether tyrosine supplementation does anything.

The isoform problem, which quietly undermines every rodent claim

Humans have four TH protein isoforms produced by alternative splicing. Rats have one. Multiple isoforms are restricted to anthropoid primates: one isoform across six subprimate and prosimian species from four families, two in all anthropoids studied, four only in humans.

The human isoforms differ in their N-terminal regulatory region, which is precisely the domain that governs phosphorylation and feedback inhibition. The regulatory architecture that this entire story is about is not conserved between the model organism and the target. Almost all in vitro TH biochemistry is done on hTH1 because it aligns with rodent TH, so isoform-specific behavior is undersampled.

---

Part 2: Testosterone and TH

There is no androgen response element in the TH promoter

Searched for directly and not found. No AR ChIP peak on TH has been reported. The TH promoter is well mapped for NURR1, PITX3, FOXA2, MEIS2, PAX6 and SIRT1. Anyone drawing a clean arrow from testosterone to the TH gene is over-reading the literature.

The anatomical substrate does exist: AR is present in 65.5% of TH-positive substantia nigra neurons in intact male rats, and AR-immunoreactive dopamine neurons exist in human substantia nigra with AR mRNA correlating with TH mRNA. But presence of receptor is not evidence of direct transcriptional control.

The cleanest experiment cuts against a simple androgen mechanism

In gonadectomized adolescent male rats:

In the same animals, DHT reproduced testosterone's effect on COMT, MAOA and MAOB perfectly. So the metabolic enzymes are cleanly androgen-receptor-driven and TH is not. If TH were a straightforward AR target, DHT should have worked at least as well as testosterone.

Two readings are possible and the original authors could not distinguish them: either TH protein requires combined AR and estrogen receptor activation (implicating aromatization), or there is an inverted-U dose-response and the DHT dose overshot it.

The mechanism appears to be indirect

The strongest dissociation available is a four-arm primate study in castrated macaques. Testosterone and DHT-plus-aromatase-inhibitor both raised locus coeruleus TH strongly; flutamide-plus-aromatase-inhibitor did not. That is a clean androgen-receptor result, aromatization-independent.

But the same paper showed LC noradrenergic neurons contain no nuclear AR at all, while over 80% contain estrogen receptors. Whatever the androgen receptor is doing, it is not doing it inside the responding cell. The signal arrives through neighboring AR-positive cells.

Other work routes androgen effects on TH through BDNF/TrkB and GDNF/iNOS trophic signaling rather than direct transcription. The one study showing coordinate regulation of TH activity, protein and mRNA by testosterone is in peripheral sympathetic ganglia, not brain, and even there the authors allowed for indirect action via trophic factors.

Castration studies flatly contradict each other

Using unbiased stereology, castration increased TH-positive cell number in substantia nigra and VTA of both rats and mice, with testosterone reversing it. Two other groups report the exact opposite: castration reduced TH-positive neurons, reversed by DHT or testosterone.

Same manipulation, opposite direction. Age at castration (adult versus peripubertal), species and strain, counting method, and duration are plausible reconcilers but none is established. Treat the direction as unknown.

The regional split, which is the most useful single finding

Hypothalamic dopamine neurons that gate prolactin are estrogen-controlled. Midbrain and locus coeruleus populations lean androgenic.

In the tuberoinfundibular system, castrated males responded to testosterone and estradiol but not DHT, and these neurons carry ER-alpha. The NMDA-receptor regulation of these neurons is estrogen-dependent, testosterone-insensitive, and present only in females.

For a menopause audience this means falling estradiol has a defensible mechanistic route into hypothalamic dopamine that falling testosterone does not.

The prolactin loop, which is genuinely bidirectional

TIDA neurons release dopamine into the pituitary portal circulation and tonically inhibit prolactin. Less dopamine means more prolactin. Prolactin suppresses GnRH pulsatility, which lowers LH and therefore gonadal testosterone. This is the one place where the dopamine system demonstrably controls testosterone rather than the reverse.

Orchidectomy increased TIDA activity within a week and testosterone reversed it in one day. The tempting reading is that testosterone directly inhibits these neurons. That reading is wrong and the same paper disproves it: bromocriptine blocked the effect and intracerebroventricular prolactin restored it. The change was secondary to prolactin, not a direct androgen action.

---

Part 3: Estrogen and the dopamine system

Estradiol regulates TH in opposite directions depending on receptor subtype

In the same cells in the same experiment, estradiol roughly doubled TH promoter activity with ER-alpha and decreased it with ER-beta. The critical element was not a classical estrogen response element but the cAMP/calcium response element at position -45.

This single result explains most of the apparent chaos in this literature. "Does estrogen raise tyrosine hydroxylase" has no region-free answer, because the answer depends on local receptor subtype ratio and on cAMP tone.

Dose and delivery invert the result too. Short-term estradiol injections raised TH and GTP cyclohydrolase mRNA in VTA, substantia nigra, locus coeruleus and nucleus tractus solitarius, dose-dependently. Long-term injections or continuous pellets were largely ineffective, and pellets lowered TH mRNA in one region. Pulsatile and continuous estradiol are pharmacologically different drugs here, which matters directly for patch versus oral versus cyclic hormone therapy, though no human study has tested it.

The receptor-level evidence belongs to estradiol

This is the answer to the "does testosterone increase dopamine receptor sensitivity" question, arrived at from the other side.

That estradiol affinity-state finding is the single cleanest receptor-sensitivity measurement in the entire hormone-dopamine literature, and it belongs to estrogen.

The COMT inverted-U: the most useful idea here

Prefrontal cortex has sparse dopamine transporter, so COMT does a disproportionate share of dopamine clearance there. The Val158Met polymorphism sets the baseline: Val clears fast (lower prefrontal dopamine), Met clears slowly (higher). Too little and too much both impair working memory. Estradiol downregulates COMT, which moves a woman along her own curve.

The behavioral test worked and then independently replicated: in young women, within-person estradiol increases improved working memory in Val/Val women and impaired it in Met/Met women. Same hormone, opposite direction, by genotype.

Applied to perimenopause, when estradiol is high, erratic and falling, this is a real mechanistic account of why brain fog hits some women hard and skips others, and why hormone therapy helps some and does nothing for others.

Three limits that must stay attached:

• The COMT-downregulation-by-estradiol evidence comes from breast cancer cells and human myometrium. It has never been demonstrated in human brain tissue. The prefrontal claim is an inference.
• The largest postmenopausal study (n = 118) found no main effects of COMT genotype or estradiol and no performance differences. This looks like a transition phenomenon, not a post-transition one.
• No trial has ever stratified hormone therapy by COMT genotype with cognition as the endpoint. That is the obvious missing study, and it should be named as missing rather than implied to exist.

Human imaging is sparse and mostly null

No human FDOPA or VMAT2 PET study of dopamine synthesis capacity across the menopause transition or with hormone therapy exists. That is the measure closest to tyrosine hydroxylase activity and it has never been made in this population. If anyone claims "PET shows menopause lowers dopamine synthesis," ask for the citation.

The protection story is stronger in epidemiology than in treatment

If estrogen were straightforwardly protective through a dopaminergic mechanism, replacement should have helped in both human cohorts. It did not. The rodent critical-window data suggest why, and the timing hypothesis has independent human support: hormone initiation before the final menstrual period associated with better cognitive performance, after it with worse.

---

Part 4: What testosterone actually does in women

The one proven indication

The 2019 Global Consensus Position Statement, endorsed by eleven societies, concluded that the only evidence-based indication for testosterone in women is postmenopausal HSDD after formal biopsychosocial assessment (Level I, Grade A). Everything else, including cognition, mood, energy, bone and muscle, was rated Insufficient. This remains the operative guidance as of August 2026; no superseding consensus statement exists.

Meta-analysis of 36 RCTs, 8,480 women:

Roughly one extra satisfying event per month, on top of a large placebo response. Real, modest, and worth having.

The mechanism is not established as dopaminergic

No human clinical study has measured tyrosine hydroxylase, striatal dopamine synthesis, or dopamine receptor occupancy as a mediator of testosterone's effect on desire in women.

The decisive contrary datum: in the pooled RECONNECT phase 3 analysis (n = 1,202), bremelanotide, a melanocortin receptor agonist, improved desire and reduced distress across all baseline bioavailable testosterone quartiles. Desire moves pharmacologically with no reference to androgen status at all.

Three approved or studied drugs act on three different systems (androgen, serotonin/dopamine, melanocortin) and produce small effects of broadly similar magnitude. That pattern argues for convergence on a shared final motivational pathway, not that testosterone's specific effect is dopaminergic.

Hormone levels across the lifespan, precisely

Testosterone declines with age from the twenties, not at menopause. In 1,423 community-recruited women aged 18 to 75, total testosterone, free testosterone, DHEAS and androstenedione all declined steeply with age, with the steeper fall in the earlier decades, and examining year by year in women aged 45 to 54 showed no independent effect of menopausal status.

Prospectively, in 172 women followed annually through natural menopause, testosterone "remains unchanged during the menopausal years," with means not varying by time relative to the final menstrual period.

What does move is SHBG, and it moves in the direction most people get backwards. Across the natural transition, SHBG falls about 43% from four years before to two years after the final menstrual period, so with total testosterone flat, the free androgen index rises about 80%. In free-hormone terms the natural transition is a relative androgen increase.

The opposite mechanism, SHBG rising and free testosterone falling, is what oral estrogen therapy does through first-pass hepatic effect. Transdermal estradiol does not. This is the most under-explained reason a woman starts oral hormone therapy and feels her libido drop. Two separate mechanisms that are frequently conflated.

Caveat kept deliberately: free androgen index is a calculated ratio, not a measured hormone, testosterone assays in the female range are genuinely unreliable, and the consensus panel argues research should focus on total testosterone because evidence that free testosterone is the biologically active fraction is lacking.

Surgical menopause is the real androgen event. Women 55 and older with bilateral oophorectomy had significantly lower total and free testosterone than intact women the same age. The postmenopausal ovary is an ongoing site of testosterone production. Conflating surgical and natural menopause is the commonest error in this space.

What is marketed and contradicted

Susan Davis, first author of both the Global Consensus and the meta-analysis behind it, at The Menopause Society 2025 Annual Meeting: there is no justification in 2025 for treating women with testosterone to prevent muscle loss, fracture, or heart disease, or to treat depression or cognitive decline. She also stated there is no such thing as a testosterone insufficiency syndrome and no blood level below which testosterone associates with any syndrome or symptoms.

The Endocrine Society had already found that endogenous testosterone levels did not predict response to therapy. That finding should end the "test her level and treat the low ones" workflow, and has not.

The pellet problem

384 women on compounded pellet hormone therapy versus 155 on FDA-approved therapy:

Nine women exceeded 400 ng/dL testosterone and four had estradiol above 1,000 pg/mL. One in five women on pellets ended up having a hysterectomy. The consensus statement explicitly recommends against any preparation producing supraphysiologic levels, naming pellets and injections.

Testosterone prescribing to US women rose 2.6-fold from 2016 to 2025, concentrated in women aged 45 to 64, who accounted for 62.2% of prescriptions in 2025. There is no FDA-approved female testosterone product, so the entire market is off-label or compounded. Australia's TGA registered AndroFeme 1 in November 2020, the first such approval worldwide.

---

Part 5: Does testosterone raise dopamine receptor sensitivity?

No, as a general claim, and the question has largely not been asked properly.

No study has measured what sensitivity means pharmacologically under androgen manipulation: no shift in receptor affinity (Kd), no change in the D2 high-affinity state fraction, no agonist-stimulated GTPgS binding. The androgen literature measures receptor counts, and those move in several directions at once.

That last row is the dissociation that makes "more receptors equals more sensitivity" unsafe.

The anatomy explains why. In mouse brain, accumbens AR staining was weak and only occasional with little AR-driven reporter activity, and caudoputamen staining was intermittent, while cortex showed strong nuclear staining in roughly 80% of layer 2 to 3 cells in several areas and hypothalamus was strongest. AR is abundant on dopamine-producing neurons and cortex, patchy on the striatal medium spiny neurons that receive dopamine. Testosterone has good access to the sending side of that synapse and poor access to the receiving side.

Consistent with that, androgen effects on the D2L/D2S splice ratio (D2S being the presynaptic autoreceptor isoform) appeared in pituitary, hypothalamus, substantia nigra and olfactory tubercle, and left the striatal ratio completely unchanged.

Structural effects are real and directionally messy: three independent groups found androgens decrease dendritic spine density in nucleus accumbens shell, while a 2026 study found castration reduced mature spines in dorsal striatum with testosterone replacement preventing it via BDNF/TrkB.

Human data: effectively none. No PET or SPECT study measures dopamine receptor binding before and after testosterone manipulation in humans, in any population. The closest is a 12-woman placebo-controlled crossover in which a single sublingual dose increased ventral striatal BOLD to reward-anticipation cues, with the effect largest in women who started with the least appetitive motivation. That is a real, human, women-specific finding, and BOLD cannot distinguish more dopamine release from more receptor sensitivity from a downstream change.

And the experiment the menopause question needs does not exist: no study has given testosterone to ovariectomized female animals with dopamine receptor binding as the outcome. In postmortem human caudate, menopausal women and age-matched men did not differ in D1 or D2 receptor expression.

---

Part 6: Does testosterone reduce willingness to accept unpredictable rewards?

False as a general statement about humans. True in one narrow, well-defined place, and the distinction between "unpredictable reward" and "risk" is exactly where the answer lives.

The human evidence is a clean null

A meta-analytic r of -0.002 is not a small effect. It is an absence.

The construct map, which is the whole answer

Probability discounting and punishment-risk are both called "risk taking" in popular writing, and testosterone moves them in opposite directions. That is the fault line.

The rodent case that supports the claim, and its dopamine mechanism

Adolescent male rats on chronic 7.5 mg/kg testosterone chose the large uncertain reward significantly less than controls, while the same animals chose the large high-effort reward significantly more. The authors' summary: testosterone increases aversion to uncertainty but decreases sensitivity to the cost of effort.

That uncertainty aversion is dopamine-mediated in the hypofunction direction. A D2 agonist (quinpirole) restored large-uncertain preference in testosterone-treated rats only; a D1 agonist eliminated the group difference. The interpretation is that chronic androgen exposure reduces D1 and D2 receptor function in nucleus accumbens shell.

This is the exact opposite of the folk model in which testosterone raises dopamine and therefore raises appetite for uncertain reward.

Limits: one lab, adolescent male rats, anabolic-abuse doses, not independently replicated elsewhere.

The dopamine-uncertainty link itself is real and underused

Fiorillo, Tobler and Schultz showed phasic dopamine varies monotonically with reward probability (prediction error), but that a second, separate response covaries with uncertainty: a gradual ramping until the potential moment of reward, maximal at p = 0.5 where uncertainty peaks. That sustained uncertainty signal is the substrate for why variable-ratio schedules and gambling are sticky. Incentive salience is amplified by unpredictability, not just by reward size.

No human study combines testosterone manipulation with dopamine measurement or a dopaminergic drug.

What the famous findings look like now

• Coates and Herbert 2008 (London trading floor): 17 male traders over 8 days, measuring profitability rather than risk preference, no causal manipulation.
• Sapienza 2009 (PNAS): drew a formal statistical rebuttal in the same journal.
• Stanton 2011 U-shaped finding: the same first author later ran three further studies and reported no consistent relationship, with effects failing family-wise error correction.
• Dual-hormone hypothesis (cortisol moderates testosterone): meta-analytic interaction r = -.061 at n = 8,538, with the meta-analysts themselves reporting publication bias and analytic flexibility.
• 2D:4D digit ratio: a 54-study meta-analysis found no relationship between 2D:4D and prenatal testosterone, adult testosterone, or testosterone reactivity; a preregistered replication at n > 2,100 was null. Actively contested rather than settled, and far too weak to carry any claim.

The better explanation: status, not risk

Testosterone administration reduced aversion to advantageous inequality while increasing aversion to disadvantageous inequality, and increased prosocial learning when prosociality was the status-appropriate move. Its effect on aggression was mediated by the subjective reward derived from aggression and moderated by androgen receptor CAG repeat length.

The status account predicts exactly what the field found: context-free lab gambles show nothing, because a lottery against a computer carries no status. It survives the null results rather than being falsified by them.

A second competing account, reduced punishment sensitivity, explains the rodent punishment-risk results cleanly and predicts no change in pure probability discounting, which is what the human data show.

The gonadal-hormone study that matters most here

In the cleanest gonadectomy-plus-replacement design: orchiectomy decreased risky choice in male rats, testosterone replacement had no effect in either orchiectomized or sham males, and estradiol decreased risky choice in both sexes while ovariectomy increased it.

If a hormone is moving risk behavior across the menopause transition, this study says the candidate is loss of estradiol, not testosterone.

---

Part 7: What actually moves the pathway

Tyrosine supplementation

The mechanism is more interesting than the usual "the enzyme is saturated" shorthand. TH is held in a low-activity state by end-product feedback inhibition and unlocked by activity-dependent phosphorylation, so it only becomes substrate-responsive in actively firing neurons. Elevating brain tyrosine stimulates catecholamine production, and that effect is exclusive to neurons that are already firing.

Everything downstream follows from that one sentence.

That older-adult trial is a single n = 17 crossover and should be described as such. It is also the only trial in an age-relevant cohort and it points the opposite way from the supplement marketing.

The depletion studies show the asymmetry cleanly: acute tyrosine/phenylalanine depletion reliably lowers phasic dopamine release in rodents and degrades human response selection, and AMPT (a direct TH inhibitor) caused relapse in 9 of 9 seasonal-affective patients in summer remission. But the definitive meta-analysis of 45 tryptophan-depletion and 8 APTD studies found monoamine depletion does not lower mood in healthy humans, only in those with a personal or family history of depression. You can break the system by subtracting. That does not mean adding helps an intact one.

BH4: better mechanism, no outcome trial

BH4 is the obligate cofactor for TH, tryptophan hydroxylase, phenylalanine hydroxylase, and all three nitric oxide synthases. GCH1 is rate-limiting. Two features make it more interesting than tyrosine.

First, it is oxidation-labile. Peroxynitrite reacts with BH4 six to ten times faster than with ascorbate or thiols, and ascorbate can recycle the resulting radical back to BH4 while thiols largely cannot. That gives vitamin C a specific mechanistic role, never tested against a brain dopamine endpoint.

Second, inflammation moves BH4 legibly. Interferon-gamma upregulates GCH1 (raising neopterin) while the accompanying oxidative burst destroys BH4. Because BH4 is also the cofactor for phenylalanine hydroxylase, functional insufficiency shows up as a rising phenylalanine/tyrosine ratio. This pairing has been demonstrated in humans: patients with active inflammatory disease had elevated neopterin and elevated Phe/Tyr ratio.

No adequately powered RCT of sapropterin or BH4 for mood or cognition exists outside phenylketonuria. The closest proxy is adjunctive L-methylfolate 15 mg outperforming placebo in SSRI non-responders, but only in a post hoc analysis stratified by inflammatory and metabolic markers, from a manufacturer-linked program. Feature BH4 as mechanism-with-biomarker-support, never as an intervention with outcome data.

Iron: the actionable node, and it reverses direction

TH is a non-heme iron enzyme. Iron deficiency reduces striatal D1 and D2 receptor density, with striatal iron loss correlating specifically with D2 loss and D2 mRNA unchanged, indicating a receptor trafficking defect rather than a transcriptional one. Brain iron deficiency also downregulates adenosine A1 receptors alongside D2, the current best explanation for restless legs syndrome.

The reversal is real: in HFE C282Y homozygous women serum ferritin rose 3.6-fold in the first 10 years postmenopause, and in postmenopausal women serum ferritin independently predicts marrow iron accumulation on MRI. A ferritin that meant deficiency at 45 may be unremarkable or rising at 58. Test, re-test across the transition, and do not carry forward a premenopausal assumption.

Everything else

---

Part 8: The frontier, and the missing study

What Mosconi's group actually found

Their tracers are 18F-FES (estrogen receptor), 18F-FDG (glucose metabolism) and PiB (amyloid). None of this measures dopamine.

Their 2021 multimodality paper found menopause-stage differences in gray matter, connectivity and brain energy metabolism, with partial gray matter recovery after menopause and greater amyloid deposition specifically in APOE4-carrying peri- and postmenopausal women versus genotype-matched men.

Their 2024 18F-FES paper found brain estrogen receptor density is higher after menopause in estrogen-regulated networks, classified every woman correctly as pre- or postmenopausal, and, counterintuitively, higher receptor density predicted worse memory and more self-reported mood and cognitive symptoms. A 2025 methods paper cut acquisition to a 30 to 60 minute static window, making it usable outside a research PET center.

The locus coeruleus is the missing bridge

It is TH-expressing, it degenerates early in Alzheimer disease, it is imageable non-invasively today with neuromelanin-sensitive MRI, and it carries a documented APOE signal.

Nigral and LC neuromelanin signal track different cognitive domains: nigral predicted reinforcement learning, LC independently predicted attention, working memory and executive function, each controlling for the other. The executive-and-attention profile women describe as brain fog maps onto LC function more cleanly than onto striatal dopamine.

No LC neuromelanin study stratified by menopause stage or hormone therapy use exists. That is the single most obviously missing study in this entire area.

APOE4 and the catecholamine system: what is honestly known

Established: - APOE4 is the strongest common genetic risk factor for late-onset Alzheimer disease, and the risk is greater in women. - Amyloid deposition was more pronounced in peri- and postmenopausal APOE4-carrying women than genotype-matched men, in vivo, in humans.

Emerging, in mice: - Humanized APOE4 female mice show accelerated endocrine aging, failure of bioenergetic adaptation, mitochondrial decline, neuroimmune activation and demyelination across the perimenopause transition. One allele impairs adaptation; two alleles are worse. A 2026 metabolomics paper adds postmenopausal amino acid depletion and reduced TCA cycle intermediates. - Observationally, APOE4 associates with earlier menopause, and APOE4 women with early menopause had the highest Alzheimer risk.

Emerging, in human tissue (preprint, not peer reviewed): - Spatial transcriptomics of 85 locus coeruleus sections from 33 neurotypical middle-aged donors, balanced by sex, ancestry and APOE genotype, found APOE-genotype-associated astrocytic expression differences adjacent to LC neurons, and higher APOE expression correlating with reduced neuromelanin content.

In living humans: - APOE genotype interacts with sleep duration to shape LC functional connectivity in 692 amyloid-positive, cognitively unimpaired adults, and LC connectivity partially mediated age-related cognitive decline.

Not established, and must not be claimed: - There is no evidence APOE4 directly reduces tyrosine hydroxylase expression or activity in humans. - There is no evidence APOE4 lowers dopamine synthesis capacity in humans. No FDOPA study stratified by APOE genotype in women exists. - The chain from APOE4 to mitochondrial and amino acid failure to a bioenergetically expensive, iron- and BH4-dependent enzyme is a chain of individually supported steps. The chain itself has not been demonstrated.

The actionable APOE4 signal in this literature is metabolic and mitochondrial, concentrated in the perimenopausal window. It is not a dopamine signal.

Trials worth watching

Not one registered testosterone-in-women trial has a cognitive primary endpoint. The libido question is being studied. The brain fog question has not been asked.

---

What can be said with confidence

1. TH is the rate-limiting enzyme of catecholamine synthesis under normal conditions, and it is feedback-inhibited by its own product and regulated by phosphorylation at four serines rather than by abundance.
2. Catecholamine synthesis is genuinely sensitive to tyrosine availability, but only in already-firing neurons.
3. Testosterone works for hypoactive sexual desire disorder in postmenopausal women. Thirty-six RCTs, 8,480 participants. This is the strongest claim in the domain and deserves to be stated without hedging.
4. Testosterone therapy for women is off-label everywhere in the US, and long-term cardiovascular, cancer and cognitive safety data do not exist.
5. Non-oral, physiologically dosed testosterone is what the evidence covers. Injections and pellets are not.
6. Androgens decline with age from early adulthood; natural menopause is not the inflection point; surgical menopause is.
7. SHBG falls about 43% across the natural transition, raising the free androgen index roughly 80% even with total testosterone flat. Oral estrogen does the reverse.
8. Perimenopausal cognitive change is real, small, stays within normal limits for nearly all women, presents as a failure to improve rather than a decline, and rebounds postmenopause.
9. Sleep, vasomotor symptoms, anxiety and depressive symptoms independently affect midlife cognition. These are the mechanisms with actual evidence.
10. Dopamine mediates wanting, not liking.
11. Humans have four TH protein isoforms; rats have one; multiple isoforms are restricted to anthropoid primates.
12. Estradiol has substantially stronger receptor-level evidence than testosterone: density, affinity state, G protein coupling and receptor-subtype ratio have all been measured for estradiol and none for testosterone.

What must not be said

• "Testosterone crashes at menopause."
• "Your testosterone is low, that is why you have brain fog." Endogenous levels do not predict treatment response, and there is no recognized androgen deficiency syndrome in healthy women.
• "TH is the rate-limiting step, so raising TH raises dopamine."
• "Testosterone makes your dopamine receptors more sensitive." Nobody measured that, and receptor counts move in opposite directions in the same brain.
• "Testosterone raises dopamine, so it improves brain fog, memory, mood or energy in menopausal women."
• "Menopausal brain fog is a dopamine deficiency."
• "Dopamine is the pleasure chemical."
• "A dopamine detox resets your receptors."
• "Testosterone will make her bolder or more willing to bet on herself." Two 1,000-person preregistered trials say no.
• Anything derived from finger-length ratios or from the 17-trader London study.
• Any claim that a male rodent study at supraphysiologic dose tells you what a transdermal cream does in a 52-year-old woman.

Gaps the agents flagged and refused to paper over

• The commonly quoted Km of TH for tyrosine, and any published half-life for TH protein. Neither could be verified from a primary source.
• Quantified evidence that vesicular packaging, release probability, receptor sensitivity and reuptake are the binding constraints downstream of TH.
• Stereotype threat and expectation effects in menopausal cognitive testing.
• A dedicated methodological critique of ovariectomy as a menopause model.
• A demonstrated funnel-plot analysis of the testosterone-in-women trial literature. The main meta-analysis was prospectively registered and searched EMA and FDA files for unpublished data, which is the correct method, and is a point in its favor.
• Any primary study demonstrating estradiol-driven COMT downregulation in human brain tissue.
• Direct allopregnanolone-to-dopamine data in females.
• Human evidence that copper status limits dopamine beta-hydroxylase in non-deficient adults.
• Hormone receptor expression mapped onto the ten transcriptionally distinct human dopamine neuron subtypes. Until that exists, hormone-to-dopamine claims stay at the level of bulk tissue.

The one-line version

The dopamine story about menopause is not wrong so much as unmeasured, the hormone with the real receptor-level evidence is estradiol rather than testosterone, and the two things a woman can actually act on today are her iron status and her sleep.

---

Key sources

The load-bearing citations. Every DOI and PMID below was returned by PubMed during research, not written from memory. Full per-thread source lists (roughly 250 references) are in the session record.

The enzyme

1. Daubner SC, Le T, Wang S (2011). Tyrosine hydroxylase and regulation of dopamine synthesis. Arch Biochem Biophys 508(1):1-12. 10.1016/j.abb.2010.12.017. PMID 21176768.
2. Bueno-Carrasco MT, Cuellar J, Flydal MI, et al. (2022). Structural mechanism for tyrosine hydroxylase inhibition by dopamine and reactivation by Ser40 phosphorylation. Nat Commun 13(1):74. 10.1038/s41467-021-27657-y. PMID 35013193.
3. Dunkley PR, Dickson PW (2019). Tyrosine hydroxylase phosphorylation in vivo. J Neurochem 149(6):706-728. 10.1111/jnc.14675. PMID 30714137.
4. Haycock JW (2002). Species differences in the expression of multiple tyrosine hydroxylase protein isoforms. J Neurochem 81(5):947-953. 10.1046/j.1471-4159.2002.00881.x. PMID 12065606.
5. Brodnik Z, Bongiovanni R, Double M, Jaskiw GE (2012). Increased tyrosine availability increases brain regional DOPA levels in vivo. Neurochem Int 61(7):1001-1006. 10.1016/j.neuint.2012.07.012. PMID 22841861.
6. Fernstrom JD, Fernstrom MH (2007). Tyrosine, phenylalanine, and catecholamine synthesis and function in the brain. J Nutr 137(6 Suppl 1):1539S-1547S. 10.1093/jn/137.6.1539S. PMID 17513421.
7. Lindgren N, Xu ZQ, Herrera-Marschitz M, et al. (2001). Dopamine D2 receptors regulate tyrosine hydroxylase activity and phosphorylation at Ser40 in rat striatum. Eur J Neurosci 13(4):773-780. 10.1046/j.0953-816x.2000.01443.x. PMID 11207812.

Androgens and TH

8. Purves-Tyson TD, Handelsman DJ, Double KL, et al. (2012). Testosterone regulation of sex steroid-related mRNAs and dopamine-related mRNAs in adolescent male rat substantia nigra. BMC Neurosci 13:95. 10.1186/1471-2202-13-95. PMID 22867132.
9. Purves-Tyson TD, Owens SJ, Double KL, et al. (2014). Testosterone induces molecular changes in dopamine signaling pathway molecules in the adolescent male rat nigrostriatal pathway. PLoS ONE 9(3):e91151. 10.1371/journal.pone.0091151. PMID 24618531.
10. Johnson ML, Day AE, Ho CC, et al. (2010). Androgen decreases dopamine neurone survival in rat midbrain. J Neuroendocrinol 22(4):238-247. 10.1111/j.1365-2826.2010.01965.x. PMID 20136692.
11. Bethea CL, Belikova Y, Phu K, et al. (2016). Reproductive steroid receptors and actions in the locus coeruleus of male macaques. Prog Neuropsychopharmacol Biol Psychiatry 71:210-222. 10.1016/j.pnpbp.2016.04.002. PMID 27083854.
12. Toney TW, Lookingland KJ, Moore KE (1991). Role of testosterone in the regulation of tuberoinfundibular dopaminergic neurons in the male rat. Neuroendocrinology 54(1):23-29. 10.1159/000125846. PMID 1922674.
13. Ribeiro AB, Leite CM, Kalil B, et al. (2015). Kisspeptin regulates tuberoinfundibular dopaminergic neurones and prolactin secretion in an oestradiol-dependent manner in male and female rats. J Neuroendocrinol 27(2):88-99. 10.1111/jne.12242. PMID 25453900.
14. Bitar MS, Ota M, Linnoila M, Shapiro BH (1991). Modification of gonadectomy-induced increases in brain monoamine metabolism by steroid hormones in male and female rats. Psychoneuroendocrinology 16(6):547-557. 10.1016/0306-4530(91)90038-u. PMID 1725825.

Estrogen and the dopamine system

15. Maharjan S, Serova L, Sabban EL (2005). Transcriptional regulation of tyrosine hydroxylase by estrogen: opposite effects with estrogen receptors alpha and beta and interactions with cyclic AMP. J Neurochem 93(6):1502-1514. 10.1111/j.1471-4159.2005.03142.x. PMID 15935066.
16. Serova LI, Maharjan S, Huang A, et al. (2004). Response of tyrosine hydroxylase and GTP cyclohydrolase I gene expression to estrogen in brain catecholaminergic regions varies with mode of administration. Brain Res 1015(1-2):1-8. 10.1016/j.brainres.2004.04.002. PMID 15223360.
17. Levesque D, Di Paolo T (1988). Rapid conversion of high into low striatal D2-dopamine receptor agonist binding states after an acute physiological dose of 17 beta-estradiol. Neurosci Lett 88(1):113-118. 10.1016/0304-3940(88)90324-2. PMID 2969467.
18. Le Saux M, Morissette M, Di Paolo T (2006). ERbeta mediates the estradiol increase of D2 receptors in rat striatum and nucleus accumbens. Neuropharmacology 50(4):451-457. 10.1016/j.neuropharm.2005.10.004. PMID 16309717.
19. Xie T, Ho SL, Ramsden D (1999). Characterization and implications of estrogenic down-regulation of human catechol-O-methyltransferase gene transcription. Mol Pharmacol 56(1):31-38. 10.1124/mol.56.1.31. PMID 10385681.
20. Jacobs E, D'Esposito M (2011). Estrogen shapes dopamine-dependent cognitive processes: implications for women's health. J Neurosci 31(14):5286-5293. 10.1523/JNEUROSCI.6394-10.2011. PMID 21471363.
21. Louis CC, Jacobs E, D'Esposito M, Moser J (2023). Estradiol and the catechol-o-methyltransferase gene interact to predict working memory performance: a replication and extension. J Cogn Neurosci 35(7):1144-1153. 10.1162/jocn_a_02001. PMID 37159230.
22. Dumas JA, Makarewicz JA, Bunn J, et al. (2018). Dopamine-dependent cognitive processes after menopause: the relationship between COMT genotype, estradiol, and working memory. Neurobiol Aging 72:53-61. 10.1016/j.neurobiolaging.2018.08.009. PMID 30212711.
23. Petersen N, Rapkin AJ, Okita K, et al. (2021). Striatal dopamine D2-type receptor availability and peripheral 17beta-estradiol. Mol Psychiatry 26(6):2038-2047. 10.1038/s41380-020-01000-1. PMID 33420479.
24. Rocca WA, Smith CY, Gazzuola Rocca L, Savica R, Mielke MM (2022). Association of premenopausal bilateral oophorectomy with parkinsonism and Parkinson disease. JAMA Netw Open 5(10):e2238663. 10.1001/jamanetworkopen.2022.38663. PMID 36287560.
25. Huo N, Smith CY, Gazzuola Rocca L, Rocca WA, Mielke MM (2021). Association of premenopausal bilateral oophorectomy with restless legs syndrome. JAMA Netw Open 4(2):e2036058. 10.1001/jamanetworkopen.2020.36058. PMID 33523190.
26. Jacobs EG, Weiss B, Makris N, et al. (2017). Reorganization of functional networks in verbal working memory circuitry in early midlife. Cereb Cortex 27(5):2857-2870. 10.1093/cercor/bhw127. PMID 27178194.

Clinical evidence in women

27. Davis SR, Baber R, Panay N, et al. (2019). Global Consensus Position Statement on the Use of Testosterone Therapy for Women. J Clin Endocrinol Metab 104(10):4660-4666. 10.1210/jc.2019-01603. PMID 31498871.
28. Islam RM, Bell RJ, Green S, Page MJ, Davis SR (2019). Safety and efficacy of testosterone for women: a systematic review and meta-analysis of randomised controlled trial data. Lancet Diabetes Endocrinol 7(10):754-766. 10.1016/S2213-8587(19)30189-5. PMID 31353194.
29. Wierman ME, Arlt W, Basson R, et al. (2014). Androgen therapy in women: a reappraisal: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab 99(10):3489-3510. 10.1210/jc.2014-2260. PMID 25279570.
30. Davison SL, Bell R, Donath S, Montalto JG, Davis SR (2005). Androgen levels in adult females: changes with age, menopause, and oophorectomy. J Clin Endocrinol Metab 90(7):3847-3853. 10.1210/jc.2005-0212. PMID 15827095.
31. Burger HG, Dudley EC, Cui J, Dennerstein L, Hopper JL (2000). A prospective longitudinal study of serum testosterone, dehydroepiandrosterone sulfate, and sex hormone-binding globulin levels through the menopause transition. J Clin Endocrinol Metab 85(8):2832-2838. 10.1210/jcem.85.8.6740. PMID 10946891.
32. Simon JA, Kingsberg SA, Portman D, et al. (2022). Prespecified and integrated subgroup analyses from the RECONNECT phase 3 studies of bremelanotide. J Womens Health 31(3):391-400. 10.1089/jwh.2021.0225. PMID 35230162.
33. Greendale GA, Huang MH, Wight RG, et al. (2009). Effects of the menopause transition and hormone use on cognitive performance in midlife women. Neurology 72(21):1850-1857. 10.1212/WNL.0b013e3181a71193. PMID 19470968.
34. Greendale GA, Wight RG, Huang MH, et al. (2010). Menopause-associated symptoms and cognitive performance: results from SWAN. Am J Epidemiol 171(11):1214-1224. 10.1093/aje/kwq067. PMID 20442205.
35. Maki PM, Jaff NG (2024). Menopause and brain fog: how to counsel and treat midlife women. Menopause 31(7):647-649. 10.1097/GME.0000000000002382. PMID 38888619.
36. Jiang X, Bossert A, Parthasarathy KN, et al. (2021). Safety assessment of compounded non-FDA-approved hormonal therapy versus FDA-approved hormonal therapy in treating postmenopausal women. Menopause 28(8):867-874. 10.1097/GME.0000000000001782. PMID 33973545.
37. Avivi I, Stuenkel CA, Sampath-Kumar R, Ben-Yehuda O (2026). Accelerating testosterone prescribing for U.S. women: implications for cardiovascular safety. JACC Adv 5(6 Pt 1):102724. 10.1016/j.jacadv.2026.102724. PMID 42095794.

Receptor sensitivity

38. Watanabe H, et al. (1989). Castration increases striatal D-2 dopamine receptors in mid-life rats. Jpn J Pharmacol 50(1):79. 10.1254/jjp.50.79. PMID 2724703.
39. Kindlundh AMS, Lindblom J, Bergstrom L, Wikberg JE, Nyberg F (2001). The anabolic-androgenic steroid nandrolone decanoate affects the density of dopamine receptors in the male rat brain. Eur J Neurosci 13(2):291-296. 10.1046/j.0953-816x.2000.01402.x. PMID 11168533.
40. Andersen SL, et al. (2002). Pubertal changes in gonadal hormones do not underlie adolescent dopamine receptor overproduction. Psychoneuroendocrinology 27(6):683-691. 10.1016/s0306-4530(01)00069-5. PMID 12084661.
41. Guivarc'h D, Vernier P, Vincent JD (1995). Sex steroid hormones change the differential distribution of the isoforms of the D2 dopamine receptor messenger RNA in the rat brain. Neuroscience 69(1):159-166. 10.1016/0306-4522(95)00228-b. PMID 8637614.
42. Hermans EJ, Bos PA, Ossewaarde L, et al. (2010). Effects of exogenous testosterone on the ventral striatal BOLD response during reward anticipation in healthy women. Neuroimage 52(1):277-283. 10.1016/j.neuroimage.2010.04.019. PMID 20398773.
43. Meitzen J, Perry AN, Westenbroek C, et al. (2013). Enhanced striatal beta-1 adrenergic receptor expression following hormone loss in adulthood is programmed by both early sexual differentiation and puberty: a study of humans and rats. Endocrinology 154(5):1820-1831. 10.1210/en.2012-2131. PMID 23533220.

Uncertainty and reward

44. Dreber A, Johannesson M, Nave G, et al. (2025). Investigating the effects of single-dose intranasal testosterone on economic preferences in a large randomized trial of men. PNAS 122(39):e2508519122. 10.1073/pnas.2508519122. PMID 40986339.
45. Sanchez Rodriguez I, Bailo L, Panizza F, Ricciardi E, Bossi F (2026). No relationship between testosterone and risk aversion: a meta-analytic review. Neurosci Biobehav Rev 184:106575. 10.1016/j.neubiorev.2026.106575. PMID 41638539.
46. Zethraeus N, Kocoska-Maras L, Ellingsen T, et al. (2009). A randomized trial of the effect of estrogen and testosterone on economic behavior. PNAS 106(16):6535-6538. 10.1073/pnas.0812757106. PMID 19366676.
47. Wallin KG, Alves JM, Wood RI (2015). Anabolic-androgenic steroids and decision making: probability and effort discounting in male rats. Psychoneuroendocrinology 57:84-92. 10.1016/j.psyneuen.2015.03.023. PMID 25900595.
48. Wallin-Miller KG, Kreutz F, Li G, Wood RI (2018). Anabolic-androgenic steroids increase sensitivity to uncertainty by inhibition of dopamine D1 and D2 receptors. Psychopharmacology 235(4):959-969. 10.1007/s00213-017-4810-7. PMID 29242988.
49. Orsini CA, Blaes SL, Dragone RJ, et al. (2021). Regulation of risky decision making by gonadal hormones in males and females. Neuropsychopharmacology 46(3). 10.1038/s41386-020-00827-0. PMID 32919406.
50. Fiorillo CD, Tobler PN, Schultz W (2003). Discrete coding of reward probability and uncertainty by dopamine neurons. Science 299(5614):1898-1902. 10.1126/science.1077349. PMID 12649484.
51. Berridge KC (2007). The debate over dopamine's role in reward: the case for incentive salience. Psychopharmacology 191(3):391-431. 10.1007/s00213-006-0578-x. PMID 17072591.

Practical modulation

52. van de Rest O, Bloemendaal M, de Heus R, Aarts E (2017). Dose-dependent effects of oral tyrosine administration on plasma tyrosine levels and cognition in aging. Nutrients 9(12):1279. 10.3390/nu9121279. PMID 29168741.
53. Coull N, Chrismas B, Watson P, Horsfall R, Taylor L (2016). Tyrosine ingestion and its effects on cognitive and physical performance in the heat. Med Sci Sports Exerc 48(2):277-286. 10.1249/MSS.0000000000000757. PMID 26285023.
54. Ruhe HG, Mason NS, Schene AH (2007). Mood is indirectly related to serotonin, norepinephrine and dopamine levels in humans: a meta-analysis of monoamine depletion studies. Mol Psychiatry 12(4):331-359. 10.1038/sj.mp.4001949. PMID 17389902.
55. Fanet H, Capuron L, Castanon N, Calon F, Vancassel S (2021). Tetrahydrobiopterin (BH4) pathway: from metabolism to neuropsychiatry. Curr Neuropharmacol 19(5):591-609. 10.2174/1570159X18666200729103529. PMID 32744952.
56. Erikson KM, Jones BC, Hess EJ, Zhang Q, Beard JL (2001). Iron deficiency decreases dopamine D1 and D2 receptors in rat brain. Pharmacol Biochem Behav 69(3-4):409-418. 10.1016/s0091-3057(01)00563-9. PMID 11509198.
57. Low MSY, Speedy J, Styles CE, De-Regil LM, Pasricha SR (2016). Daily iron supplementation for improving anaemia, iron status and health in menstruating women. Cochrane Database Syst Rev 4:CD009747. 10.1002/14651858.CD009747.pub2. PMID 27087396.
58. Avni T, Reich S, Lev N, Gafter-Gvili A (2019). Iron supplementation for restless legs syndrome: a systematic review and meta-analysis. Eur J Intern Med 63:34-41. 10.1016/j.ejim.2019.02.009. PMID 30798983.
59. Volkow ND, Tomasi D, Wang GJ, et al. (2012). Evidence that sleep deprivation downregulates dopamine D2R in ventral striatum in the human brain. J Neurosci 32(19):6711-6717. 10.1523/JNEUROSCI.0045-12.2012. PMID 22573693.

Frontier

60. Mosconi L, Berti V, Dyke J, et al. (2021). Menopause impacts human brain structure, connectivity, energy metabolism, and amyloid-beta deposition. Sci Rep 11(1):10867. 10.1038/s41598-021-90084-y. PMID 34108509.
61. Mosconi L, Nerattini M, Matthews DC, et al. (2024). In vivo brain estrogen receptor density by neuroendocrine aging and relationships with cognition and symptomatology. Sci Rep 14(1):12680. 10.1038/s41598-024-62820-7. PMID 38902275.
62. Wang T, Mao Z, Shang Y, et al. (2025). Accelerated midlife endocrine and bioenergetic brain aging in APOE4 females. Front Aging Neurosci 17:1632877. 10.3389/fnagi.2025.1632877. PMID 40900992.
63. Sun S, Madge V, Djordjevic J, et al. (2025). Selective effects of substantia nigra and locus coeruleus degeneration on cognition in Parkinson's disease. Mov Disord 40(5):844-854. 10.1002/mds.30148. PMID 39945211.
64. Mulvey B, Divecha HR, Tippani M, et al. (2025). Impact of Alzheimer's disease risk factors and local neuromelanin content on the transcriptomic landscape of the human locus coeruleus. bioRxiv. PREPRINT, NOT PEER REVIEWED. 10.1101/2025.10.29.685354. PMID 41278831.
65. Kamath T, Abdulraouf A, Burris SJ, et al. (2022). Single-cell genomic profiling of human dopamine neurons identifies a population that selectively degenerates in Parkinson's disease. Nat Neurosci 25(5):588-595. 10.1038/s41593-022-01061-1. PMID 35513515.
66. Nerattini M, Williams S, Andy C, et al. (2025). Sex-specific associations of serum testosterone with gray matter volume and cerebral blood flow in midlife individuals at risk for Alzheimer's disease. PLoS ONE 20(1):e0317303. 10.1371/journal.pone.0317303. PMID 39804890.
67. Walsh MJM, Gibson K, Hynd M, et al. (2023). Perimenopausal Effects of Estradiol on Anhedonia and Psychosis Study (PEEPs): study protocol. Trials 24(1):150. 10.1186/s13063-023-07166-7. PMID 36855177.