DHEA, cognitive function, muscle mass, and libido: real evidence versus marketing

Stage: REPORT_EN
Date: 2026-09-02
Evidence cutoff: 2026-09-02
Primary population: adult women, with emphasis on the menopausal transition and postmenopause
Status: complete scientific report for QA; not a prescription, diagnosis, or clinical recommendation

1. Executive scientific abstract

The commercial thesis that the age-related decline in DHEA-S is a correctable deficiency and that restoring a “youthful level” rejuvenates the brain, muscle, and libido does not survive the human evidence. Across several randomized trials, oral DHEA—usually 50 mg/d for periods ranging from weeks to two years—clearly increased DHEA-S and, in women, testosterone, estrone, or estradiol, but did not produce reproducible improvements in general cognition, working memory, strength, physical performance, or sexual desire [1,2,4,6–9,14]. These trials were not designed as equivalence studies and do not exclude small effects, narrow tasks, or extreme phenotypes; they do contradict a large systemic class response.

The best-supported positive result is local, not systemic. Vaginal prasterone improves cytology, pH, dryness, and dyspareunia in postmenopausal women with genitourinary syndrome of menopause (GSM) [18]. In the same cohort of 482 participants, the FSFI desire domain increased by 0.24 units over placebo as a secondary result [19]. That signal does not demonstrate “central libido”: the population was selected for atrophy and pain, all sexual domains were assessed, temporal mediation was not measured, and the benefit could arise through tissue repair, less pain, improved local sensation, and resumed activity. An open active-comparator trial published in 2026 showed improvement with both prasterone and vaginal estradiol; the primary between-group difference for dyspareunia was at the conventional boundary (OR 2.87, CI95% 1.00–8.25, p=0.051), and estradiol improved several vaginal signs more [20]. This favors an explanation based on local steroid restoration rather than an anti-aging property unique to DHEA.

In muscle, mass and function must be separated. In women with Addison disease, DHEA modestly increased lean mass versus placebo (+869 versus +183 g; p=0.034), without a demonstrated parallel functional benefit [11]. A historical DHEA×training signal was published with the sexes pooled [12], but a women-only RCT from 2025 found that exercise+DHEA did not significantly outperform the same exercise+placebo regimen for fat-free mass (+1.5±1.7 versus +1.0±1.2 kg; d=0.33, p=0.20) among 80 participants whose intervention was not disrupted by the pandemic [13]. The positive contrast against DHEA alone mainly identifies the effect of exercise. Without a full ITT analysis, specific strength, and performance, the co-stimulus hypothesis remains imprecise and weakened.

The strongest mechanism synthesis is a three-gate model: delivery to the compartment, intracrine conversion and receptor engagement, and a susceptible functional context. Increasing the serum precursor opens the systemic exposure gate but does not guarantee free tissue dose, flux, AR/ER activation, or function. The vagina combines local delivery, intracrine capacity, and a measurable local lesion; the brain and muscle lack a closed human causal chain. Neural TrkA/p75NTR and sigma-1 pathways remain preclinical [24,25]. There is no direct evidence that DHEA reduces incident dementia, disability, falls, major events, or female mortality.

The scientific delta of this report is twofold. First, it rigorously separates lack of large systemic efficacy from DHEA-S as a marker/common cause: the former is supported by several RCTs, while the latter remains an unproven longitudinal hypothesis. Second, it converts the only consistent clinical signal into a falsifiable experiment: a human vaginal organotypic model using labeled DHEA, orthogonal gate loss, bypass, AR/ER blockade, and epithelial function must distinguish DHEA-specific intracrine dependence, nonspecific local steroid restoration, and a true functional null.

Overall judgment: moderate-to-high evidence against a large, general systemic benefit in the domains studied; moderate evidence for a short-term local vaginal benefit; low evidence for the exact mechanism of desire; very low evidence for muscle co-stimulation, direct neural action, or any extension to longevity.

2. Scientific question and relevance

The question is not whether DHEA-S declines with age, but which biological quantity—circulating DHEA-S, free DHEA, intracrine flux, or active products—actually limits a function in women. Specifically:

In adult women stratified by life stage, does DHEA/DHEA-S availability cause materially relevant changes in objective cognition, muscle mass and function, or desire with distress, through which tissue mechanism does this occur, and can any persistent response contribute to functional health or longevity?

The relevance to women’s health rests on three facts. DHEA/DHEA-S is an important source of C19 precursors after ovarian function is lost; brain, muscle, adipose tissue, and vagina have different metabolic capacities; and cognition, physical function, and sexual health are legitimate components of quality of life and healthspan. Sharing a precursor, however, does not turn these domains into a single “anti-aging” syndrome. Each requires its own causal chain and competing-risk analysis.

The marketing error comprises four substitutions:

1. an age-related population decline → an individually causal deficiency;
2. serum DHEA-S → tissue dose;
3. increased downstream steroids → function;
4. a short-term change → healthspan or longevity.

None of these equivalences has been demonstrated. This project evaluates mechanisms and experiments; it does not evaluate individual treatment decisions.

3. Scope, population, and life stage

The core population is women aged 40–79 years undergoing natural menopause, separated by STRAW+10: late reproductive/transition (−3b to −1), early postmenopause (+1a to +1c), and late postmenopause (+2). Women aged 25–39 years provide an exposure reference, not a universal standard of youth. Women older than 80 years require separate analysis because of multimorbidity and selective survival.

The following remain distinct phenotypes:

• primary adrenal insufficiency/Addison disease, useful for studying rescue of an extreme deficiency but not representative of physiological aging;
• surgical menopause, primary ovarian insufficiency, and hormone therapy;
• frailty/prefrailty, which may change the functional ceiling;
• GSM/dyspareunia and low desire with distress, which are not the same sexual phenotype;
• established cognitive impairment versus cognitively healthy women.

Pregnancy/lactation, PCOS/hyperandrogenism, active hormone-sensitive cancer, advanced hepatic/renal failure, systemic glucocorticoids, and treatments that materially change steroidogenesis or SHBG are excluded from the general estimand or analyzed separately.

Outcomes are not merged:

• cognition: episodic memory, working memory, executive function, attention/speed, and visuospatial performance; mood, sleep, or “brain fog” do not substitute for objective cognition;
• muscle: mass/area/quality, strength, specific strength, power, and performance; DXA-FFM is not equivalent to contractile tissue;
• sexuality: desire, distress, arousal, lubrication, pain, orgasm, satisfaction, and activity are related but distinct constructs;
• healthspan/longevity: sustained decline, disability, falls, dementia, disease-free years, and survival require their own follow-up and mediation.

4. Background knowledge and mechanism map

4.1 From the adrenal gland to the reservoir

The main adrenal pathway is:

cholesterol → pregnenolone → 17OH-pregnenolone
         --CYP17A1 + CYB5A + POR→ DHEA
DHEA --SULT2A1 + PAPSS2→ DHEA-S

DHEA-S integrates production, sulfation, distribution, tissue desulfation, and clearance. A low concentration therefore does not localize the defect. Prior L8-1/L8-3 evidence also shows that the age-related population change and the within-woman trajectory around the menopausal transition are not equivalent.

4.2 Intracrine conversion

The candidate tissue pathway is:

DHEA-S --STS→ DHEA
DHEA --HSD3B1/2→ A4 --HSD17B/AKR1C3→ testosterone --SRD5A→ DHT
DHEA --HSD17B→ androstenediol
A4 --CYP19A1→ estrone --HSD17B→ estradiol
testosterone --CYP19A1→ estradiol
products --SULT/UGT/oxidation→ conjugates or metabolites

The output is a mixture, not a single hormone. Transport, free fraction, SHBG, perfusion, multicellular architecture, reconjugation, and storage can decouple serum from tissue. Enzyme expression demonstrates possibility; only labeled carbon, mass balance, and perturbation demonstrate flux and necessity.

4.3 Receptor and function

Free products can activate AR, ERα, or ERβ. Nuclear receptor abundance, cofactors, chromatin, temporal exposure, and tissue state determine whether that signal changes a function. In female muscle, bioavailable testosterone and nuclear AR have been associated with mass/strength, whereas total testosterone has not; AR translocation in myocytes also did not guarantee Akt/mTOR activation [28]. This supports measuring engagement rather than attributing causality to DHEA.

4.4 The three-gate model

Source/sulfation/clearance
        ↓
Available DHEA-S and DHEA
        ↓  [Gate 1: tissue delivery]
Local free product
        ↓  [Gate 2: conversion + AR/ER engagement]
Tissue response
        ↓  [Gate 3: susceptible lesion/functional context]
Specific and persistent function
        ↓
Organ-specific outcome → competing-risk balance → possible healthspan

Systemic RCTs demonstrate that circulating exposure can increase; they do not demonstrate that the other gates open. The vagina is the best candidate because administration is local and an atrophic phenotype exists. The hypothesis becomes unfalsifiable if “susceptible” is defined after observing who responded; lesion, product, receptor, and function must therefore be frozen before the experiment.

5. Evidence method

All project artifacts—scope, map, verification, mechanism synthesis, compute decision, hypotheses, adversarial review, and design—were integrated with the relevant lifetime memory. Decisive claims were traced to primary publications, official registries, or both. On 2 September 2026, the primary abstract of the 2025 women-only exercise RCT, the 2026 vaginal RCT, and the official NCT03227458 and NCT03287232 records were rechecked.

Interpretive hierarchy:

1. human RCT with between-arm contrast and intention to treat;
2. small, crossover, or extreme-phenotype RCT;
3. longitudinal or cross-sectional human observation;
4. human tissue/cell evidence;
5. animal in vivo/ex vivo evidence;
6. heterologous/in-vitro system;
7. inference connecting different studies.

A nonsignificant result is described as lack of superiority, not equivalence. Positive signals, nulls, attrition, multiplicity, within-arm estimands, sex pooling, funding, and absent results were preserved. Public absence of results was not interpreted as a null result. No current Lua data or commercial sources were used as evidence.

6. Evidence map

6.1 Human cognition

Domain conclusion: randomized human evidence is predominantly null for general cognition; one visuospatial anomaly warrants task-specific replication, not a claim of neuroprotection.

6.2 Muscle and physical function

Domain conclusion: DHEA alone does not reproducibly improve strength or performance. Exercise explains the most consistent signal. Lean mass without specific strength, power, or mobility does not validate useful muscle.

6.3 Desire, sexual function, and vagina

Domain conclusion: oral DHEA has no reproducible benefit for desire. Vaginal prasterone improves GSM/dyspareunia; the desire change may be peripheral, indirect, or statistical and requires temporal ordering and distress.

6.4 Mechanistic evidence by level

7. Contradictory evidence and null findings

The central contradiction must not be smoothed over: baseline associations and preclinical mechanisms are more positive than systemic RCTs. Rival explanations rank as follows:

1. No material causality/common cause: biological age, activity, adiposity, HPA function, liver/kidney function, and disease affect both DHEA-S and function. This is the most parsimonious global explanation, but it has not been demonstrated longitudinally.
2. Compartment error: serum does not represent muscle or brain; possible, but it cannot be invoked to make every null unfalsifiable.
3. Specific lesion or window: GSM and Addison disease show that an injured tissue or extreme deficiency may respond. This does not authorize extrapolation to the general population.
4. Small imprecise effects: plausible because the RCTs lack formal equivalence; incompatible with a large universal promise.
5. Multiplicity/selection: explains visuospatial signals, within-arm findings, pooled-sex estimates, or completer analyses better than a new mechanism until replication exists.

Adversarial evidence constraining the positive signals:

• the positive visuospatial task came from n=24 and does not converge with working memory or the general battery [1–4];
• strength/SPPB in frailty was reported within arm and does not demonstrate an advantage over placebo+exercise [10];
• the lean-mass increase in Addison disease lacked concordant strength or performance [11];
• the historical DHEA×training effect was pooled by sex, and the recent women-only RCT did not confirm superiority for FFM [12,13];
• the vaginal desire domain was secondary in the same clinical cohort and had no mediation analysis [19];
• NCT03287232, a phase III trial specific to HSDD, completed 653 participants in 2019 but still has no public results at the cutoff and states that IPD will not be shared [31]. This raises selective-reporting risk; it does not justify inferring a negative result.

Systemic nulls also do not prove that DHEA is harmful or that the exact effect is zero. They show lack of detectable superiority under those populations, routes, durations, and endpoints. Long-term safety, hormone-sensitive cancer, CVD, and mortality were not adequately powered.

8. Multiscale mechanism synthesis

8.1 Cognition

The complete chain would be serum → BBB/CSF → DHEA/DHEA-S or a local metabolite → target → circuit → cognitive domain. Only some links are observed. CSF transfer and ex-vivo brain metabolism are human evidence [22,23], but there is no target engagement in women. TrkA/p75NTR and sigma-1 are preclinical [24,25]. RCTs show that raising serum and peripheral products is insufficient [1,2,4]. The direct neural hypothesis therefore remains H0 and must first demonstrate compatible exposure, independence from conversion, and target loss-rescue.

The visuospatial signal could reflect a task effect, practice/period, or a false positive. The minimum test is not a dementia trial: it is a preregistered replication with one task, parallel forms, counterbalancing, a period/carryover model, and a materiality margin. Even replication would not prove neuroprotection.

8.2 Muscle

Muscle can contain and transform steroids, but content is not production and association is not causal function [26–28]. Fat infiltration, prior activity, perfusion, and retention may generate serum–tissue decoupling. RCTs show that the systemic precursor is insufficient to reproducibly increase function [6–9].

The narrow co-stimulus version proposes that DHEA amplifies a response only when mechanical load crosses a threshold and AR/ER/IGF are engaged. Villareal suggests it [12]; Igwebuike and Jankowski weaken it [8,13]. Before new exposure, the decisive evidence is the female ITT contrast of exercise+DHEA versus exercise+placebo, with specific strength, power, and performance. If only FFM changes, the extension to useful muscle dies.

8.3 Vagina and sexual function

The most plausible local chain is:

local prasterone → active AR/ER products → epithelial/vascular/sensory repair
→ less dryness and pain ± better sensation → activity/satisfaction → reported desire

Treatment→GSM/pain is demonstrated in the short term [18]. Conversion/AR is observed in human vaginal cells [21], but not in the same participants. Pain/sensation→desire is inferred. The simplest competitor is that prasterone and estradiol restore local steroid signaling through different routes with similar function [20]. Only a design combining flux, perturbation, bypass, receptor, and function can assign specificity.

8.4 From organ to healthspan

A proximal function contributes to quality of life but does not automatically become longevity. The extension requires causal function → persistence → less decline/disability → disease-free years, while modeling cancer, CVD, and other competing risks. No verified source reaches that level. The correct statement is absence of direct evidence, not neutrality or benefit.

9. Computational layer

Decision: omitted. No bioinformatics or BioNeMo analysis was run because the decisive uncertainty requires randomized/longitudinal IPD, not sequences or baseline atlases. The NCT03227458 and NCT03287232 records contain no tabulated results and state IPD sharing: NO at the cutoff [13,31]. No public human DHEA/placebo omics dataset paired with tissue and function was found that could estimate flux or mediation.

Geneformer could describe cell states and ESM-2/Evo2 sequences, but none reconstructs missing randomization, exercise, attrition, pain→desire timing, or intracrine dose. An aggregate meta-analysis of thigh volume, FFM, and strength would create false precision by mixing estimands. Computation reopens only if IPD, complete results, function-paired omics, or a sequence-based target/variant capable of changing a perturbation becomes available.

10. Primary hypothesis

L8-4-AR-H1E v2 — absence of a large average systemic benefit

Falsifiable statement: in peri/postmenopausal women without severe adrenal insufficiency, systemic oral DHEA that raises DHEA-S and downstream steroids will not produce an average improvement greater than a prespecified functional margin in objective cognition, strength/performance, or desire with distress; any effect will be small, specific, or nonreplicable rather than a class response.

Proposed mechanism: the circulating precursor is not the dominant bottleneck. Local delivery, free fraction, intracrine flux, receptor, mechanical stimulus, lesion, and circuit/context limit each function separately.

Evidence for: multiple human RCTs with achieved exposure and null outcomes [1,2,4,6–9,14]; exercise as the dominant explanation for the recent muscle signal [13].

Evidence against: local vaginal benefit [18–20], lean mass in Addison disease [11], and small visuospatial/co-stimulus signals [3,12]. None demonstrates a systemic class effect.

Status: strengthened and narrowed.
Maturity: integrated H1; no formal equivalence.
Confidence: 0.82 for absence of a large general effect, lower for small domain-specific effects.

Associated but separate hypothesis, L8-4-AR-H1M v1: within a woman, DHEA(S) will not precede function after prior function and common causes are modeled. Status H0–H1, confidence 0.48. Restoration RCTs alone do not demonstrate this common-cause claim; bidirectional temporality is required.

11. Competing hypothesis

L8-4-AR-H2V v2 versus L8-4-AR-H2L v1 — vaginal intracrine dependence or local class restoration

H2V, falsifiable statement: in postmenopausal atrophic vaginal tissue defined before treatment, DHEA will modify a tissue function only if it forms free products, activates AR/ER, and loss of the causal gate abolishes function; a downstream product will bypass that loss.

H2L, rival: improvement will depend on restoring local steroid signaling and tissue, not on a privileged property of DHEA; DHEA, estradiol, or a downstream mixture that equalizes repair will produce equivalent function.

Why they compete: H2V predicts necessity of conversion from DHEA; H2L predicts that precursor identity is secondary once final repair/engagement is equalized. The vaginal RCT supports local function [18], Cellai supports capacity [21], and Strandberg supports a shared local class [20], but no study closes the chain.

H2V status: active, weakened/narrowed; integrated H0/H1 by links, confidence 0.30.
H2L status: priority competitor; H0–H1, confidence 0.52.

Derived sexual hypothesis, L8-4-AR-H4P v2: more than half of the small total effect on desire will be mediated by prior peripheral changes, and the direct effect will fall within triviality. Status H0 for mediation/H1 by links, confidence 0.40. This is not identical to H2V: local repair can exist even if the proposed path to desire is wrong.

12. Translational hypothesis

L8-4-AR-HT1 v2 — dynamic assay eligible only after causality

Falsifiable statement: only after H2V identifies a necessary gate, one prespecified free-product/precursor → engagement slope measured under an ex-vivo challenge must be reproducible and add external prediction of function beyond DHEA-S, age, stage, SHBG, static products, composition, and baseline function.

Predictions: valid identity and mass balance; adequate ICC/CCC/CV and preanalytical stability; sensitivity to the causal gate; better error and calibration in a second set/laboratory.

Evidence for: repeated decoupling between serum concentration and function; plausible tissue heterogeneity.
Evidence against: no standardized assay, test–retest study, second laboratory, complete causal chain, or external validation exists.

Status: parked.
Maturity: H0.
Confidence: 0.03.
Requirement: HUMAN_QA_REQUIRED before any translational assessment. It is not called a biomarker unless it passes those gates.

13. Falsifiable predictions and kill criteria

Cross-cutting criteria: failure of identity/free fraction stops flux; precursor without gate-dependent product closes intracrinology; product without engagement closes classical signaling; engagement with function inside equivalence closes relevance; a signal seen only post hoc, within arm, pooled by sex, or in one laboratory does not translate; lack of persistence blocks healthspan.

14. Discriminating experiment

14.1 D0: use already-generated human data first

D0-MUS. Reanalyze all participants randomized in NCT03227458. Primary contrast: exercise+DHEA versus the same exercise+placebo at 36 weeks; ANCOVA/mixed model with baseline, wave/site, and pandemic interruption. Primary ITT, MAR imputation, and MNAR tipping-point; per-protocol only supportive. Separate FFM from maximal strength, specific strength, power, and performance. Two distinct SESOIs are fixed before opening the data: one technical for DXA and one functional. FFM without concordant function kills “useful muscle.”

D0-SEX. Retrieve complete results/protocol/SAP and ideally IPD from NCT03287232 and the vaginal RCT. Estimate desire and distress separately. If intermediate time points exist, use longitudinal g-computation/TMLE for treatment → early pain/dryness/sensation/activity → later desire/distress; do not naively adjust for a post-treatment mediator. Without temporal ordering, report the total effect but not mediation.

D0-LAG. A cohort with at least three waves over 24–36 months and ≥400 usable women per domain, LC–MS/MS steroidome, STRAW+10, prior function, activity, adiposity, HPA, inflammation, disease, liver/kidney function, and medications. Random-intercept cross-lagged models compare DHEA_t→function_t+1 with function_t→DHEA_t+1. This phase decides temporal ordering, not definitive causality.

D0-COG. Only if resolving the anomaly is prioritized: independent double-blind crossover RCT, one primary mental-rotation task, parallel forms, and a period×order model. Planning: 96 completers (~112 enrolled) for a conservative paired effect d_z≈0.35, subject to the SD from a blinded pilot. A null CI excluding that SESOI or a period-dependent signal kills a material task improvement.

14.2 D1-VAG: the smallest decisive mechanistic experiment

Model: organotypic/vagina-on-chip model with stratified primary vaginal epithelium over stroma–smooth muscle from postmenopausal donors undergoing benign surgery for an independent indication. Donor is the unit; wells are technical replicates. GSM/atrophy is defined before treatment by histology, stratification, and permeability. Vagina-on-chip and hormone-responsive organoids support feasibility but do not yet validate DHEA/GSM [29,30].

Q0, six donors: qualify architecture, viability, benchmark E2 response, isotopic LC–MS/MS, 80–120% recovery, LOQ, adsorption, free fraction, and mass balance. Require labeled product and activity, not RNA alone. If Q0 fails, use fresh explant; do not force the model.

Within-donor arms: vehicle; [13C]-DHEA; DHEA + pharmacological HSD3B inhibition; DHEA + genetic HSD3B1/2 reduction; losses with bypass [13C]-A4 or T/DHT at measured free AUC; DHEA with AR blockade, ER blockade, and dual blockade; DHEA versus E2 and a T/DHT/E1/E2 mixture reproducing the generated products. Include cell-free, tissue-free, vehicle, positive-receptor, off-target perturbation, and viability controls.

Hierarchical endpoints:

1. J_DHEA→product flux in pmol/h/mg viable tissue;
2. AR/ER engagement by translocation/occupancy and a prespecified orthogonal readout;
3. function: recovery of permeability to a 4 kDa tracer after standardized microinjury;
4. support: blinded histological maturation, junctions, mechanics, cytotoxicity/inflammation.

Required temporal order: product in hours, engagement afterward, and repair in days.

Size: 16 analyzable atrophic donors, recruiting up to 20 for ~20% failure; 8 exploratory nonatrophic references if available. Paired design to detect a large dependence (d_z≈0.75 or loss ≥50% of the functional effect) with ~80% power; one blinded recalculation from Q0, cap 24 analyzable. Mixed model with fixed condition, random donor, and blocked batch/position. Replication in ≥6 new donors or a second laboratory.

Patterns:

• loss reduces flux and function, bypass rescues → strengthens H2V;
• DHEA/E2/mixture equalize repair, with no precursor advantage → strengthens H2L;
• product+engagement without function → functional null, kills the gate for that endpoint;
• function after conversion and AR/ER are abolished → kills the classical mechanism and opens another target only with its own loss-rescue.

14.3 D2: conditional translation

Only if D1 replicates. If H2V wins, use a double-dummy comparison of prasterone with the identified product/mixture and early exposure/engagement measures. If H2L wins, conduct an equivalence trial between strategies that equalize repair. If D1 is null, do not open D2 for that function. Calculate size from blinded SESOI/variance, 90% power, multiplicity, and attrition—not from the most favorable secondary result.

15. Biomarkers and stratification

There is no validated biomarker of “DHEA competence.” The following are candidate mechanism or stratification measures, not clinical biomarkers:

• DHEA, DHEA-S, A4, T, DHT, Adiol, E1, E2, conjugates, and SHBG by LC–MS/MS; free fraction when identifiable;
• product/precursor ratio or slope only under a validated challenge and mass balance;
• STS/HSD3B/HSD17B/SRD5A/CYP19A1 activity measured by flux, not isolated RNA;
• tissue AR/ER engagement and an independent functional response;
• for muscle: load/adherence, specific strength, power, fat infiltration, and quality, not FFM alone;
• for sexuality: GSM, pain, dryness, sensation, activity, FSFI-desire, and distress separately;
• for cognition: one objective domain, practice, sleep/mood, and baseline performance;
• common causes: STRAW+10, years since FMP, HPA, adiposity, activity, inflammation, liver/kidney function, disease, and medications.

The stratification “low DHEA-S = responder” is rejected. A future mechanistic stratum requires a pre-defined lesion, preserved conversion, and functional receptor. “High converters” can be confirmatory only if the rule is frozen and replicated out of sample.

16. Individual variability

Variability can arise at every gate:

• source: age, adrenal architecture, HPA, and sulfation;
• circulation: SHBG, liver, kidney, time of day, and disease;
• tissue: transport, STS/HSD/SRD5A/aromatase, composition, perfusion, and storage;
• receptor: abundance, localization, cofactors, and chromatin;
• context: ovarian stage, lesion, mechanical load, sleep, mood, activity, pain, and relational context;
• measurement: immunoassay versus LC–MS/MS, cognitive practice, DXA error, and sexual scales.

There are insufficient data to estimate magnitude or heterogeneity in Mexican/LATAM women. Differences in ancestry, body composition, comorbidity, hormonal exposure, and sociocultural context could affect transport, but must not be invented. Mechanism/metrology comes first; transportability then requires explicit replication.

17. Pharma relevance and maturity

The opportunity is not “anti-aging DHEA.” At most, it is a local mechanism de-risking program.

1. Differential vaginal intracrine gate. If H2V survives, a conversion target or local formulation with defined function may exist. Maturity integrated H0; not ready for partnering.
2. Local steroid restoration as a class. If H2L wins, value depends on demonstrating a material advantage in exposure, repair, function, or safety over active comparators. Equivalence without advantage reduces differentiation. Maturity conceptual H0–H1.
3. Dynamic assay. Eligible only after causality, variance components, second laboratory, and external gain. Maturity H0, parked, HUMAN_QA_REQUIRED.

Early disqualifiers: equivalent function despite flux/engagement; mechanism reproduced by a comparator without advantage; signal only at supraphysiological exposure, high passage, or one donor; unresolved systemic/hormone-sensitive tissue exposure; no replication or persistence. No regulatory, clinical, commercial, or partnership readiness is claimed.

18. Limitations

1. Systemic RCTs were not designed for equivalence; the synthesis excludes large effects better than small ones.
2. Several trials have small female samples, multiple outcomes, completer analyses, or sex-pooled estimates.
3. The Jankowski 2025 RCT is an abstract, not a full manuscript; the 152 enrolled/144 randomized/80 analyzed discrepancy requires a flow diagram and ITT [13].
4. Absence of NCT03287232 results does not determine direction; it documents nonreporting risk only [31].
5. Positive vaginal evidence largely comes from a sponsored program and related cohorts; it requires independent replication/mediation.
6. Cellai demonstrates capacity in passage 7–8 cells, not the complete clinical chain [21].
7. Muscle steroid content may reflect infiltration, uptake, or activity rather than causal conversion [26,27].
8. TrkA/p75NTR, sigma-1, LTP, and BDNF are preclinical systems that do not model exposure or the adult female brain [24,25].
9. There are insufficient dementia, disability, or mortality outcomes and insufficient rare long-term safety data.
10. There is no basis for transporting magnitudes to Mexico/LATAM.
11. The proposed organotypic model represents barrier/repair, not lived pain, desire, systemic safety, or longevity.
12. The search targeted decisive sources rather than constituting a systematic review with dual screening and meta-analysis.

19. Conclusions

1. Serum DHEA-S is neither a tissue dose nor a causal deficiency by definition. Its age-related decline may integrate production, sulfation, use, and clearance.
2. Systemic restoration of DHEA-S and sex-steroid products does not reproducibly improve general cognition, strength/performance, or female desire in the studied populations. This refutes a large class promise, not an exactly zero effect.
3. Vaginal prasterone is the real clinical exception: it improves GSM, dryness, and dyspareunia in the short term. The small desire signal does not demonstrate central libido or a DHEA-specific mechanism.
4. Mass is not function. The Addison and exercise-FFM signals do not by themselves extend to useful muscle, disability, or healthspan.
5. Direct brain mechanisms remain preclinical. There is no female neural target engagement or bridge to dementia.
6. The favored explanation is a systemic precursor that is rarely the average functional bottleneck, with possible effects only in defined local contexts. The common-cause hypothesis, although plausible, requires within-woman temporal evidence.
7. The smallest decisive experiment is vaginal and donor-aware: tracer, two losses, bypass, receptor, and function in the same system. It can adjudicate intracrine dependence, local class restoration, or a functional null.
8. Longevity remains closed. No proximal improvement permits claims about survival or disability-free years without persistence, organ-specific mediation, and competing risks.

Scientific claims from REPORT_EN

• L8-4-REPORT-EN-C1: multiple women’s RCTs separate achieved systemic exposure from function and support that oral DHEA does not produce a large, reproducible average benefit in cognition, useful muscle, or desire.
• L8-4-REPORT-EN-C2: the vaginal benefit is local and real for GSM/dyspareunia, but its mechanism does not yet distinguish DHEA-specific intracrine dependence from local steroid restoration as a class.
• L8-4-REPORT-EN-C3: the 2025 women-only RCT weakens muscle co-stimulation because the relevant contrast against the same exercise+placebo regimen was nonsignificant and limited to uninterrupted completers.
• L8-4-REPORT-EN-C4: the completed, unreported HSDD trial prevents treating the positive desire literature as complete evidence, without turning nonpublication into a null result.
• L8-4-REPORT-EN-C5: there is no verified clinical bridge from DHEA to dementia, disability, or survival; any relevance to longevity is presently inferred and extremely immature.

20. References

1. Kritz-Silverstein D, von Mühlen D, Laughlin GA, Bettencourt R. Effects of dehydroepiandrosterone supplementation on cognitive function and quality of life: the DHEA and Well-Ness (DAWN) Trial. J Am Geriatr Soc. 2008;56(7):1292–1298. DOI: 10.1111/j.1532-5415.2008.01768.x. PMID: 18482290.
2. Barnhart KT, Freeman E, Grisso JA, et al. The effect of dehydroepiandrosterone supplementation to symptomatic perimenopausal women on serum endocrine profiles, lipid parameters, and health-related quality of life. J Clin Endocrinol Metab. 1999;84(11):3896–3902. DOI: 10.1210/jcem.84.11.6153. PMID: 10566625.
3. Stangl B, Hirshman E, Verbalis J. Administration of dehydroepiandrosterone (DHEA) enhances visual-spatial performance in postmenopausal women. Behav Neurosci. 2011;125(5):742–752. DOI: 10.1037/a0025151. PMID: 21942436.
4. Merritt P, Stangl B, Hirshman E, Verbalis J. Administration of dehydroepiandrosterone increases serum levels of androgens and estrogens but does not enhance short-term memory in post-menopausal women. Brain Res. 2012;1483:54–62. DOI: 10.1016/j.brainres.2012.09.015. PMID: 22985672.
5. Davis SR, Shah SM, McKenzie DP, Kulkarni J, Davison SL, Bell RJ. Dehydroepiandrosterone sulfate levels are associated with more favorable cognitive function in women. J Clin Endocrinol Metab. 2008;93(3):801–808. DOI: 10.1210/jc.2007-2128. PMID: 18073302.
6. Nair KS, Rizza RA, O'Brien P, et al. DHEA in elderly women and DHEA or testosterone in elderly men. N Engl J Med. 2006;355(16):1647–1659. DOI: 10.1056/NEJMoa054629. PMID: 17050889.
7. Percheron G, Hogrel JY, Denot-Ledunois S, et al. Effect of 1-year oral administration of dehydroepiandrosterone to 60- to 80-year-old individuals on muscle function and cross-sectional area: a double-blind placebo-controlled trial. Arch Intern Med. 2003;163(6):720–727. DOI: 10.1001/archinte.163.6.720. PMID: 12639206.
8. Igwebuike A, Irving BA, Bigelow ML, et al. Lack of dehydroepiandrosterone effect on a combined endurance and resistance exercise program in postmenopausal women. J Clin Endocrinol Metab. 2008;93(2):534–538. DOI: 10.1210/jc.2007-1027. PMID: 18029465.
9. Dhatariya KK, Bigelow ML, Nair KS. Dehydroepiandrosterone replacement therapy in hypoadrenal women: protein anabolism and skeletal muscle function. Mayo Clin Proc. 2008;83(11):1218–1225. DOI: 10.4065/83.11.1218. PMID: 18990320.
10. Kenny AM, Boxer RS, Kleppinger A, et al. Dehydroepiandrosterone combined with exercise improves muscle strength and physical function in frail older women. J Am Geriatr Soc. 2010;58(9):1707–1714. DOI: 10.1111/j.1532-5415.2010.03019.x. PMID: 20863330. Registry: NCT00664053.
11. Gurnell EM, Hunt PJ, Curran SE, et al. Long-term DHEA replacement in primary adrenal insufficiency: a randomized, controlled trial. J Clin Endocrinol Metab. 2008;93(2):400–409. DOI: 10.1210/jc.2007-1134. PMID: 18000094.
12. Villareal DT, Holloszy JO. DHEA enhances effects of weight training on muscle mass and strength in elderly women and men. Am J Physiol Endocrinol Metab. 2006;291(5):E1003–E1008. DOI: 10.1152/ajpendo.00100.2006. PMID: 16787962.
13. Jankowski CM, Schmiege SJ, Carpenter D, et al. DHEA Augmentation Of The Musculoskeletal Effects Of Exercise In Older Women: A Randomized Controlled Trial. Med Sci Sports Exerc. 2025;57(10S):565. DOI: 10.1249/01.mss.0001159624.19768.17. Registry: NCT03227458.
14. Panjari M, Bell RJ, Jane F, Wolfe R, Adams J, Morrow C, Davis SR. A randomized trial of oral DHEA treatment for sexual function, well-being, and menopausal symptoms in postmenopausal women with low libido. J Sex Med. 2009;6(9):2579–2590. DOI: 10.1111/j.1743-6109.2009.01381.x. PMID: 19619146.
15. Arlt W, Callies F, van Vlijmen JC, et al. Dehydroepiandrosterone replacement in women with adrenal insufficiency. N Engl J Med. 1999;341(14):1013–1020. DOI: 10.1056/NEJM199909303411401. PMID: 10502590.
16. Løvås K, Gebre-Medhin G, Trovik TS, et al. Replacement of dehydroepiandrosterone in adrenal failure: no benefit for subjective health status and sexuality in a 9-month, randomized, parallel group clinical trial. J Clin Endocrinol Metab. 2003;88(3):1112–1118. DOI: 10.1210/jc.2002-020769. PMID: 12629093.
17. Hackbert L, Heiman JR. Acute dehydroepiandrosterone effects on sexual arousal in postmenopausal women. J Womens Health Gend Based Med. 2002;11(2):155–162. DOI: 10.1089/152460902753645290. PMID: 11975863.
18. Labrie F, Archer DF, Koltun W, et al. Efficacy of intravaginal dehydroepiandrosterone on moderate to severe dyspareunia and vaginal dryness, symptoms of vulvovaginal atrophy, and of the genitourinary syndrome of menopause. Menopause. 2016;23(3):243–256. DOI: 10.1097/GME.0000000000000571. PMID: 26731686. Registry: NCT02013544.
19. Labrie F, Derogatis L, Archer DF, et al. Effect of Intravaginal Prasterone on Sexual Dysfunction in Postmenopausal Women with Vulvovaginal Atrophy. J Sex Med. 2015;12(12):2401–2412. DOI: 10.1111/jsm.13045. PMID: 26597311.
20. Strandberg M, Cockin A, Lindén Hirschberg A. Effects of vaginal dehydroepiandrosterone and estradiol on dyspareunia, a symptom of vulvovaginal atrophy in postmenopausal women—a randomized controlled trial. Maturitas. 2026;208:108924. DOI: 10.1016/j.maturitas.2026.108924. PMID: 41903372. Registry: NCT05586711.
21. Cellai I, Di Stasi V, Comeglio P, et al. Insight on the Intracrinology of Menopause: Androgen Production within the Human Vagina. Endocrinology. 2021;162(2):bqaa219. DOI: 10.1210/endocr/bqaa219. PMID: 33247714.
22. Guazzo EP, Kirkpatrick PJ, Goodyer IM, Shiers HM, Herbert J. Cortisol, dehydroepiandrosterone, and DHEA sulfate in the cerebrospinal fluid of man: relation to blood levels and effects of age. J Clin Endocrinol Metab. 1996;81(11):3951–3960. DOI: 10.1210/jcem.81.11.3951. PMID: 8923843.
23. Weill-Engerer S, David JP, Sazdovitch V, et al. In vitro metabolism of DHEA to 7α-hydroxy-DHEA and Δ5-androstene-3β,17β-diol in specific regions of the aging brain from Alzheimer's and non-demented patients. Brain Res. 2003;969(1–2):117–125. DOI: 10.1016/S0006-8993(03)02288-1.
24. Lazaridis I, Charalampopoulos I, Alexaki VI, et al. Neurosteroid dehydroepiandrosterone interacts with nerve growth factor receptors, preventing neuronal apoptosis. PLoS Biol. 2011;9(4):e1001051. DOI: 10.1371/journal.pbio.1001051.
25. Chen L, Dai XN, Sokabe M. Chronic administration of DHEA-S primes for facilitated induction of long-term potentiation via sigma-1 receptor: optical imaging study in rat hippocampal slices. Neuropharmacology. 2006;50(3):380–392. DOI: 10.1016/j.neuropharm.2005.10.015. PMID: 16364377.
26. Pöllänen E, Sipilä S, Alen M, et al. Differential influence of peripheral and systemic sex steroids on skeletal muscle quality in pre- and postmenopausal women. Aging Cell. 2011;10(4):650–660. DOI: 10.1111/j.1474-9726.2011.00701.x. PMID: 21388496.
27. Pöllänen E, Kangas R, Horttanainen M, et al. Intramuscular sex steroid hormones are associated with skeletal muscle strength and power in women with different hormonal status. Aging Cell. 2015;14(2):236–248. DOI: 10.1111/acel.12309. PMID: 25645687.
28. Alexander SE, et al. Bioavailable testosterone and androgen receptor activation, but not total testosterone, are associated with muscle mass and strength in females. J Physiol. 2025;603(18):5181–5208. DOI: 10.1113/JP286803. PMID: 39393048.
29. Mahajan G, et al. Vaginal microbiome-host interactions modeled in a human vagina-on-a-chip. Microbiome. 2022;10:201. DOI: 10.1186/s40168-022-01400-1.
30. Zeng B, et al. Sex hormone-responsive human vaginal epithelial organoids: a novel in vitro platform for studying Chlamydia trachomatis infection. Cell Commun Signal. 2026;24:121. DOI: 10.1186/s12964-026-02685-7. PMID: 41580857.
31. ClinicalTrials.gov. Prasterone (DHEA) for the Treatment of Hypoactive Sexual Desire Disorder (HSDD). NCT03287232. Status checked 2026-09-02: completed, 653 participants, no public results, IPD sharing: NO.
32. Sultana A, et al. Effect of dehydroepiandrosterone therapy on cognitive performance among postmenopausal women: a systematic review of randomized clinical trial data. Menopause. 2023;30(11):1167–1173. DOI: 10.1097/GME.0000000000002251. PMID: 37788418. (Used for completeness; decisive inferences rely on primary RCTs.)