DHEA and DHEA-S: adrenal biosynthesis, life-course trajectory, and women's health/longevity

Stage: REPORT_EN
Project: L8-1
Cutoff date: 2026-08-26
Classification: adrenal androgens, intracrinology, and female aging
Nature: scientific research report; it does not prescribe, diagnose, or propose individual use of DHEA/DHEA-S
Cumulative basis: Lua Labs scientific charter and lifetime memory; scoping.md, evidence-map.md, evidence-verification.md, mechanism.md, compute-decision.md, compute.md, hypotheses.md, adversarial-review.md, and experiment-design.md; final verification of recent primary sources.

Concrete scientific delta

This report replaces the narrative of a “youth hormone” that reaches an exact peak at age 25 and then declines inexorably with a double-partition model. First, the adrenal determines how much precursor enters the Δ5 pathway and how much is stored as sulfate through the quantity and state of the zona reticularis, CYB5A, HSD3B2, and PAPSS2–SULT2A1. Each tissue then determines how much DHEA-S enters, is desulfated by STS, is converted into androgens or estrogens, and reaches a function. The same low serum value may reflect a smaller source, less sulfation, greater tissue use, or different clearance; by itself it identifies neither active dose nor biological direction.

Human evidence demonstrates a large population decline after early adulthood, but neither an exact individual maximum nor a monotonic decline in every woman. It also demonstrates selective zona-reticularis remodeling rather than global adrenocortical insufficiency. The causal chain, however, breaks before tissue function: systemic DHEA elevation in trials changes multiple steroids without consistently restoring muscle, metabolism, or aerobic capacity. The leading systemic explanation is therefore common cause/physiological reserve. The most important new falsifiable hypothesis is that the adrenal decline results from a quantifiable combination of less reticular territory and lower per-cell function, while a second hypothesis proposes that narrow, opposing effects in trabecular bone and hormone-sensitive tissues may cancel out in female lifespan.

1. Executive scientific abstract

DHEA and DHEA-S are produced mainly in the adrenal zona reticularis. CYP17A1, supported by POR and especially CYB5A, directs 17-hydroxypregnenolone toward DHEA; PAPSS2 supplies PAPS and SULT2A1 converts DHEA into DHEA-S. Sulfation is not merely an inactivity label: it creates an abundant circulating reservoir while temporarily withdrawing DHEA from immediate activation pathways. In peripheral tissues, sulfate transporters and STS release DHEA, which can feed HSD3B, HSD17B/AKR1C3, SRD5A, or CYP19A1 to produce androstenedione, testosterone, DHT, estrone, or estradiol according to cellular context.

The phrase “peak at age 25” is unsupported as an individual estimate. The classic study by Orentreich et al. found the highest cross-sectional female mean in the 15–19-year group by RIA [1]; it did not follow women from adolescence to old age. The defensible description is a population maximum in late adolescence/early adulthood, dependent on cohort and method. The subsequent decline is real on average but not inexorable in every woman: SWAN modeled a perimenopausal inflection and a transitional rise in 84.5% of a random-effects subset, while the overall within-woman time term was small and nonsignificant (−0.33%/year, p=.06) [3].

Anatomy explains part of the selectivity. With age, zona-reticularis area and CYB5A+ territory decrease; intensity in positive cells may be preserved, the HSD3B2–CYB5A boundary becomes blurred, and glucocorticoids or 11-oxygenated C19 steroids remain relatively preserved [6–9]. Our donor-aware reanalysis of 92 female GTEx v8 adrenals found CYB5A −0.639 log2 and SULT2A1 −0.685 log2 in donors aged 60–79 versus 20–59 years after adjustment for RIN, ischemic time, and center; most of the steroidogenic panel did not decline uniformly. This is a transcriptomic signal compatible with selective remodeling, not a flux measurement.

Health does not obey a “more DHEA-S is better” rule. In cohorts of older women, associations with function, mortality, and CVD are null, linear in selected contexts, or U-shaped [15–20]. In ASPREE/SHOW, higher baseline DHEA was associated with less physical deterioration, but three-year DHEA change did not track change in grip or physical function in the paired subgroup [20]. Randomized trials that raised DHEA-S and other steroids produced mostly null results for strength, body composition, insulin sensitivity, VO₂, and quality of life [21–26]. These results falsify a sufficient systemic endocrine-deficiency model, although not a specific local DHEA-S→STS pathway.

Recent genetic evidence also does not authorize a single valence. A female MR linked genetically predicted DHEA-S to higher lumbar BMD and lower forearm-fracture risk, but not hip fracture, using limited instruments [29]. Another MR estimated higher ER+ cancer risk (OR 1.09, CI95% 1.03–1.16) [30]. A 2025 MR was null for female lifespan (0.04 years per log-µmol/L; CI95% −0.50 to 0.58), but used attained parental age rather than own survival [31]. The most defensible synthesis is that small, organ-specific effects of opposing sign may exist while the global longevity effect remains unproven.

The three final canonical hypotheses are:

1. Primary — L8-1-H4 v4: the decline in Δ5 flux decomposes into a smaller zona-reticularis fraction and lower per-cell functional state; architectural dominance is accepted only if the reticular fraction explains ≥50% of the contrast and the young:old per-cell flux is equivalent (0.80–1.25).
2. Competing — L8-1-H2 v4: for muscle and lifespan, DHEA(S) is primarily a marker of reserve/common cause; changing local flux may alter products without materially changing function.
3. Translational — L8-1-HT1 v4: a dynamic signature of adrenal topology plus organ-specific conversion will add prediction only after causality, metrology, and external incremental value are demonstrated; it remains near-killed, H0, and HUMAN_QA_REQUIRED.

Two auxiliary hypotheses preserve uncertainty: H5 v3 uses CYB5A versus PAPSS2/SULT2A1 as a benchmark for flux signatures without assuming that low sulfation is a common aging subtype; H6 v1 proposes a bone–hormone-sensitive-risk trade-off with net-null lifespan.

2. Scientific question and relevance

Question

Which mechanisms connect adrenal biosynthesis and the life-course trajectory of DHEA/DHEA-S to women's health and longevity, and which new hypotheses distinguish adrenal source, sulfation, tissue intracrinology, and common cause?

Relevance

The problem is not to decide whether a “low” concentration is good or bad. Serum DHEA-S sits between two gates and mixes processes with opposite implications. A decline caused by loss of reticular source predicts less DHEA and less peripheral supply. A decline caused by sulfation failure predicts less DHEA-S but may leave more DHEA available for androgenic or estrogenic fates. A low concentration caused by greater tissue use likewise does not equal low local exposure. Without separating these topologies, any association with strength, bone, cancer, or mortality is causally ambiguous.

Relevance to female longevity additionally requires distinguishing three levels:

• adrenal capacity: how much precursor is produced and sulfated;
• organ-specific function: whether a tissue transforms the reservoir and changes an independent function;
• net outcome: how favorable and adverse effects across organs combine under competing risks.

All-cause mortality, BMD, cancer, grip, and cognition are not interchangeable proxies for longevity. Inference must traverse each bridge in order.

3. Scope, population, and life stage

Primary population

Women undergoing natural adrenal aging from early adulthood through advanced old age. The human unit is the woman; in tissues and cultures it is the donor. Wells, sections, and cells are technical replicates.

Exclusions from the primary natural-aging analysis

Adrenal insufficiency, Cushing syndrome, congenital adrenal hyperplasia, adrenal tumor or surgery, PCOS, premature ovarian insufficiency, surgical menopause, pregnancy/lactation, hormone therapy, DHEA/androgens or systemic glucocorticoids, and advanced kidney/liver disease. These states may serve as separate quasi-experiments, but they are not mixed with the natural trajectory.

Biological domains included

Adrenal, circulation, adipose, muscle, bone, vasculature, hormone-sensitive tissue, and—with much weaker evidence—brain and immunity. Each organ retains its own entry, STS, steroid output, receptor, and function. Adipose is not generalized to muscle, nor blood to brain.

4. Background knowledge and mechanism map

4.1 Adrenal gate

The pathway begins with ACTH–MC2R/MRAP–cAMP/PKA and STAR-mediated access to cholesterol. CYP11A1 generates pregnenolone; CYP17A1 produces 17-hydroxypregnenolone and performs 17,20-lyase. CYB5A can markedly enhance this second activity in reconstituted human biochemistry [5]. HSD3B2 competes for precursors and favors the Δ4 branch; a zona reticularis with high CYB5A/SULT2A1 and low HSD3B2 favors DHEA/DHEA-S.

PAPSS2 supplies PAPS to SULT2A1. Rare human PAPSS2 deficiency demonstrates that nearly absent DHEA-S can coexist with more DHEA and active androgens [10]; human cellular studies verify functional PAPSS2–SULT2A1 cooperation [11]. This datum proves a biochemical topology, not its frequency in common aging.

A conceptual representation is:

DHEA-S flux ≈ number of ZR cells × per-cell state × CYP17A1/CYB5A/POR × (1−HSD3B2 competition) × PAPSS2/SULT2A1.

Aging may reduce any of these factors. Current evidence favors ZR loss/remodeling and selective CYB5A/SULT2A1 changes, but does not yet quantify their contributions.

4.2 Circulating reservoir

DHEA-S is more stable and abundant than DHEA, but its concentration is a state variable:

DHEA-S change = secretion + extra-adrenal sulfation − tissue entry/use − clearance.

Therefore, a single sample does not measure production rate, flux, or tissue dose. DHEA and DHEA-S must be measured separately and accompanied by downstream products, time, method, kidney/liver function, medication, stage, and body composition.

4.3 Tissue gate

The reservoir becomes functional only if the following occurs:

DHEA-S → transporter-mediated entry → STS → DHEA → downstream enzyme → T/DHT or E1/E2 → AR/ER → function.

Human adipose expresses transporters and STS, converts DHEA-S, and showed greater ex vivo STS activity in postmenopausal samples [12–13]. But the decisive study measured homogenate capacity, not maintenance of product or function. In muscle, local steroids and partial machinery are detected, but fat/vessels may contribute, and exercise may improve function while intramuscular steroids decrease [14]. The labeled DHEA-S→product→STS-dependent function chain has not been demonstrated in female muscle.

4.4 From organ to longevity

Attributing longevity requires six links:

1. a real reduction in adrenal output;
2. lower active exposure in a tissue;
3. a product-dependent functional change;
4. within-woman temporal precedence;
5. replicated organ-specific mediation;
6. an effect on years free of disability/disease under competing risks.

The evidence stops before the third link. No reviewed study demonstrates that manipulating DHEA/DHEA-S lengthens women's lives.

5. Evidence method

Four layers were integrated:

1. Primary human sources: longitudinal cohorts, histology, ex vivo tissue, RCTs, and Mendelian randomization.
2. Human biochemistry/in vitro evidence: to localize CYB5A, PAPSS2–SULT2A1, and STS gates without automatically transporting them to aging or function.
3. Reproducible human computation: donor-aware reanalysis of female adrenal GTEx v8 and muscle GSE303107; RNA was interpreted as a prioritization signal, not activity.
4. Adversarial reasoning: explicit search for nulls, U-shaped curves, reverse causality, selection, pleiotropy, invalid proxies, and organ-specific signs.

Each decisive claim was classified as human, ex vivo/in vitro, animal, computational, or inferred. Denominators, actual comparison, method, uncertainty, and limitations were preserved. No new meta-analysis was performed. Some historical sources could be verified only through their original registry/abstract; their numbers are not transported between RIA and LC–MS/MS.

Final verification of 2024–2025 evidence confirmed:

• Iwahashi et al. analyzed a broad GTEx bulk set, but their age-related cellular layer contains one 27-year-old woman and two men aged 64/74; it does not identify female cellular aging [9].
• Quester et al.'s MR used a female exposure sample (n=8,565) and female outcomes; its bone signal was site-specific [29].
• Nounu et al.'s MR found OR 1.09 (CI95% 1.03–1.16) for ER+, but the DHEA-S exposure came from a mixed-sex GWAS and instruments may act through downstream steroids [30].
• Schooling and Zhao found a female null for attained parental age, not own survival [31].

6. Evidence map

Trajectory and numbers that must not be conflated

• Orentreich's cross-sectional maximum does not define an “individual physiological peak.”
• The 84.5% in SWAN is a random-effects estimate among 1,423 women, not visual inspection of 1,202 monotonic curves.
• The ~30% with a late increase in PAQUID refers to the mixed-sex sample; it must not be attributed only to women [4].
• Oral DHEA is not equivalent to endogenous DHEA-S: it bypasses initial desulfation, undergoes first-pass metabolism, and changes multiple hormones.
• Baseline DHEA and change in DHEA are distinct exposures; ASPREE suggests that the first may mark reserve without the second mediating deterioration.

7. Contradictory evidence and null findings

7.1 Population decline coexists with individual increases

Age is strongly associated with lower concentrations across women, but within a woman the transition may produce an inflection. Rival explanations are adrenal production, sulfation change, kidney/liver clearance, tissue use/release, HPA/ACTH, cohort, or model error. The rise after BSO in 14/20 women indicates that the ovary is not necessary, but does not demonstrate greater adrenal production [32].

7.2 Less area does not demonstrate preserved per-cell capacity

Dharia found less territory with similar CYB5A intensity in positive cells [6]. Immunohistochemical intensity, however, does not measure 17,20-lyase, precursor, PAPS, or product. Iwahashi suggested CYB5A↓/HSD3B2↑ within older ZR, but age and sex are completely confounded in three donors [9]. The correct conclusion is architecture×state, not a pure mosaic model.

7.3 Intracrine capacity does not equal a functional buffer

Greater postmenopausal adipose STS activity may reflect adaptation, cellular composition, or surgical selection [13]. To qualify as a buffer it must show that, under lower substrate, absolute flux and function are maintained; that was not measured.

7.4 Local product does not imply function

Female muscle contains steroids and transcripts for STS, transporters, AKR1C3, SRD5A1, AR, and ESR1; no age slope in the GSE303107 panel survived BH. In tissue, training improved function while intramuscular DHEA/E2/T decreased [14]. The expected adversarial result is conversion without functional consequence.

7.5 Cohorts and RCTs do not converge on a global valence

• CHS All Stars associated concurrent DHEA-S declines with deterioration, but did not establish precedence [15].
• CHS mortality was mostly null in women; WHAS showed a U-shaped curve [16–17].
• WISE lost significance after adding CAD severity and showed no nonfatal events [18].
• SWAN was null for elevated depressive symptoms [19].
• Nair and Dhatariya are strong nulls for a broad muscle/metabolism theory [21–22].
• DAWN permits a narrow bone effect, not systemic rejuvenation [26].
• DHEA reduced HDL/large particles in a hypoadrenal cohort, reminding us that intermediate markers may diverge [27].

7.6 MR does not resolve the trade-off by itself

The bone, ER+, and lifespan studies did not use exactly the same exposure/instruments/outcomes. The bone MR was dominated by one locus and two correlated markers; the cancer MR could not calculate variance explained/F for DHEA-S because allele frequencies were absent from the source GWAS; the lifespan MR used parental age. Apparent convergence could be a real trade-off, pleiotropy, incompatible instruments, or chance. The smallest discriminant is a harmonized reanalysis, not an intervention.

8. Multiscale mechanism synthesis

Molecular

CYB5A modulates CYP17A1 efficiency; HSD3B2 competes with the Δ5 pathway; PAPSS2–SULT2A1 controls entry into the sulfated reservoir. Loss of CYB5A reduces total entry into DHEA. Loss of sulfation can lower DHEA-S without lowering DHEA and, if downstream capacity exists, can increase active products. These perturbations should produce different isotopic signatures.

Cellular

Aging may reduce the number of ZR cells, alter their identity, or both. Bulk RNA does not distinguish composition from within-cell regulation. The critical variable is flux per viable ZR cell, measured in the same tissue in which f_ZR is quantified.

Tissue

Each organ applies a second partition. Adipose has demonstrated capacity; muscle remains unidentified; bone may respond at trabecular sites; breast/endometrium may convert precursors toward hormone-sensitive exposure. There is no single “DHEA-S effect” outside tissue context.

Systemic

Biological age, disease, activity, nutrition, inflammation, HPA, adiposity, and clearance can simultaneously cause low DHEA-S and worse function. This topology explains why baseline associations appear informative while systemic restoration and longitudinal change are null.

Longevity

Two nonexclusive mechanisms produce a net-null effect:

1. DHEA-S is a reserve marker without material mediation.
2. Narrow causal effects of opposite sign across organs cancel out.

Current evidence favors the first for muscle/lifespan and retains the second as an H0–H1 hypothesis.

9. Computational layer

Female adrenal GTEx v8

We reconstructed 92 unique donors: 59 aged 20–59 and 33 aged 60–79 years. The voom-limma model adjusted for RIN, log-ischemic time, and center, with BH within a prespecified panel. Core results:

Residual signatures covaried (rho=.538), compatible with shared zonal composition. This does not demonstrate coregulation, protein, activity, or flux.

Muscle GSE303107

Among 57 women aged 19–80 years, STS, SLCO2B1, SLCO3A1, AKR1C3, SRD5A1, AR, and ESR1 were detectable in bulk. No age slope survived BH. HSD3B1/2, SRD5A2, and CYP19A1 were not modelable because of low coverage. GSE167186, intended for cellular localization, derives from 72 men; the female arm was declared not_identifiable and nuclei were not pseudoreplicated.

Decision

The computation moderately strengthens a selective adrenal process and prioritizes CYB5A/architecture over isolated low sulfation. It does not raise causal maturity and does not justify pure myotubes as the first muscle model. BioNeMo was rejected because cross-sectional expression cannot be transformed into flux or function.

10. Primary hypothesis

L8-1-H4 v4 — Architecture × reticular-state decomposition

Falsifiable statement: in non-neoplastic female adrenal tissue, the age-related decline in total Δ5 flux will be explained by an estimable combination of lower ZR fraction and lower per-cell flux; architectural dominance will be accepted only if f_ZR explains ≥50% of the total contrast and the geometric young:old ratio of per-cell flux lies within 0.80–1.25.

Mechanism: loss of CYB5A^high/SULT2A1^high territory, greater contact/overlap with HSD3B2, and possible within-cell state change reduce DHEA/DHEA-S output without shutting down the entire cortex.

Predictions:

• f_ZR will decrease with age and associate with total flux.
• If architecture dominates, introducing f_ZR will attenuate the age coefficient by ≥50%.
• If cell state dominates, older-cell flux will lie outside equivalence.
• Cortisol/fasciculata branches will remain relatively preserved; a parallel decline weakens selectivity.

Evidence for: human histology, product profiles, and female-only GTEx.
Evidence against: intensity is not activity; female single-cell aging is unidentifiable; the perimenopausal inflection does not fit a simple monotonic anatomical model.
Status/maturity: weakened_reframed; H1 for separate links, H0 integrated; confidence 0.40.

11. Competing hypothesis

L8-1-H2 v4 — Common cause/reserve for muscle and lifespan

Falsifiable statement: after prior function and common causes are modeled, DHEA(S) will not explain a material subsequent change in muscle function or female lifespan; a physiological perturbation may change local products while function remains within equivalence.

Mechanism: age/disease/nutrition/activity/inflammation/HPA/clearance reduce the adrenal pool and function in parallel. DHEA-S reports context but is not the dominant functional bottleneck.

Predictions:

• baseline/cross-sectional associations will be larger than lagged changes;
• adjustment for prior function, activity, disease, kidney/liver function, and composition will attenuate the signal;
• STS-dependent muscle product will coexist with equivalent glucose uptake;
• common-cause models will match DHEA(S)-augmented models out of sample.

Evidence for: ASPREE change-null, female PAQUID/CHS findings, null systemic RCTs, and null female lifespan MR.
Evidence against: concurrent CHS signal, a narrow lumbar-bone effect, and the possibility that oral DHEA is an unsuitable perturbation of a local pathway.
Status/maturity: strengthened_but_scoped; human H1, H0 for complete causal equivalence; confidence 0.73 for muscle/lifespan.

12. Translational hypothesis

L8-1-HT1 v4 — Dynamic flux-topology signature

Falsifiable statement: only if an adrenal benchmark distinguishes source from sulfation, a tissue demonstrates functional conversion, and each component passes metrology will a minimal dynamic signature add external prediction beyond age, stage, static DHEA-S/DHEA, HPA, clearance, and baseline function.

Candidate research components:

1. an adrenal flux signature that classifies CYB5A versus PAPSS2/SULT2A1 perturbation;
2. organ-specific DHEA-S→product conversion;
3. one proximal outcome causally linked to that conversion.

Predictions: adequate ICC/CCC/CV; ≥80% blinded classification; improvement in external error/calibration; organ specificity.

Kill criteria: death of any parent, metrology failure, classification no better than chance, no incremental value, signal driven by batch/composition/kidney/liver function, or need for biopsy without a justifiable context.

Status/maturity: parked_near_kill, H0, confidence 0.02, HUMAN_QA_REQUIRED. It is not a validated biomarker.

Necessary auxiliary hypotheses

L8-1-H5 v3 — Source-versus-sulfation benchmark

Orthogonal perturbations of CYB5A and PAPSS2/SULT2A1 must produce distinguishable isotopic signatures. CYB5A loss reduces total C19, DHEA, and DHEA-S; loss of sulfation reduces DHEA-S and raises DHEA:DHEA-S when input is adequate. Only then will those signatures be sought spontaneously by age. Maturity H1 gates/H0 aging.

L8-1-H6 v1 — Multiorgan trade-off

Persistently higher DHEA(S)/downstream exposure will preferentially favor trabecular bone/forearm while increasing at least one hormone-sensitive outcome, with a small net effect on lifespan. The hypothesis dies if harmonized instruments do not colocalize, one branch depends on one locus, MVMR explains it through T/E2/SHBG, or own lifespan shows a robust net direction. Maturity H1 links/H0 integrated.

L8-1-H1 v4 — Local muscle gate

Intact female muscle must generate labeled product from physiological DHEA-S before myocyte versus paracrine localization is debated. Without product it is parked; product with equivalent function kills the functional extension; only STS-dependent product+function opens cellular localization. Maturity H0.

13. Falsifiable predictions and kill criteria

14. Discriminating experiment

Overall sequence

M0 → E0/E0b determines why adrenal output declines. A0 attacks the trade-off using public data. M0 → D−1 determines whether a functional muscle branch exists at all. D0, human temporal analyses, and translation remain closed until positive gates.

M0 — Metrology

Isotopic LC–MS/MS for 13C-17OH-pregnenolone, DHEA, DHEA-S, A4, 11OHA4, and cortisol; PAPS/PAP for sulfation. Criteria: CV ≤15% (≤20% at LLOQ), bias ≤15%, recovery 80–120%, low carryover, and a stable unexplained balance fraction across conditions. An unvalidated ratio does not replace a failed analyte.

E0 — Architecture versus per-cell state

Model: fresh organotypic sections of non-neoplastic adrenal tissue from women aged 20–40 and 60–80 years; donor as the unit.
Same tissue: multiplex CYB5A/SULT2A1/HSD3B2/CYP17A1 imaging, independent ZR count, viability, and 13C-17OH-pregnenolone pulse.
Primary outcomes: total Δ5 flux per viable mass, flux per million ZR cells, and f_ZR.
Pilot: 8+8 for SD/viability, not hypothesis testing.
Confirmatory: 22–30 per stratum, blinded adaptation; TOST 0.80–1.25; replication in a second laboratory.
Decision: per-cell equivalence + ≥50% attenuation supports architectural dominance; lower cellular flux + <50% supports cell state; both produce a mixed model.

E0b — Causal benchmark

In a qualified human adrenal system, perturb CYB5A, SULT2A1, and PAPSS2 with two reagents and resistant rescue. Inputs are 13C-17OH-pregnenolone and 13C-DHEA; the frozen vector comprises total C19, DHEA, DHEA-S, ratio, A4, and 11OHA4. Go: leave-one-donor-out classification ≥80% and specific rescue. Only then apply it blindly to E0.

D−1 — Muscle gate

Model: fresh vastus-lateralis explants from postmenopausal women aged 55–75.
Conditions: low/high physiological free DHEA-S, 13C-DHEA-S, corroborative STS inhibition, 13C-DHEA bypass, and controls.
Mechanistic primary: first labeled product above LLOQ in tissue+medium.
Conditional functional primary: insulin-stimulated 2-deoxyglucose uptake.
Size: 10 for qualification; 24 confirmatory, maximum 32 for 0.85–1.15 equivalence.
Stop: no product or product with equivalent function; do not build a coculture to rescue the branch.

A0 — Harmonized MR

Freeze the same female DHEA-S instrument set; harmonize LS/heel/forearm/hip, ER+/ER−, own lifespan if available, and parental age secondarily. Require locus colocalization, Steiger, leave-one-locus-out, robust MR, and MVMR with T/E2/SHBG. Materiality margins: |β|<0.10 SD for BMD, OR 0.90–1.10 for ER+, and ±0.5 years for lifespan per harmonized exposure unit. Without power for the margin, the result is unidentifiable.

Smallest experiment that decides the uncertain link

The single highest-value experiment is E0: simultaneously quantify architecture and per-cell flux in the same tissue. Without it, CYB5A/SULT2A1↓ may mean fewer cells, a worse cell, or both. D−1 is the smallest test of muscle health: product without function would strengthen common cause without denying intracrinology.

15. Biomarkers and stratification

There is no validated clinical biomarker in this project. The following are research measures:

Minimum stratification: age, STRAW+10, years since FMP, hormone therapy, kidney/liver function, time, method, HPA/cortisol, adiposity/composition, activity, disease, medication, ancestry, center/ischemia, and baseline function. Post hoc subgroups will not be used to rescue null primary outcomes.

16. Individual variability

Variability is not residual noise; it is part of the mechanism:

• trajectory: mean decline, perimenopausal inflection, and late increases may coexist;
• adrenal: ZR fraction and per-cell state may vary independently;
• sulfation: PAPS/SULT2A1 and entry into DHEA determine different signs;
• tissue: STS, transporters, downstream enzymes, and receptors vary by organ and composition;
• stage: age and the HPO clock are not equivalent;
• clearance: kidney/liver function, time, and medications alter the pool;
• survival: ASPREE and GTEx select populations different from WHAS/WISE;
• ancestry: European/White MR magnitudes do not automatically transport to Mexico/LATAM.

The inferential unit will always be the donor/woman. Pseudoreplicating cells or wells would inflate certainty. LATAM transportability is studied after causality and metrology; no current Lua data are used.

17. Pharma relevance and maturity

Research opportunities

Current Pharma value is target discovery, not readiness. No hypothesis exceeds H1 in isolated links or is eligible for partnering/a human trial. The main risk is benefit in one organ with cost in another. Any advance requires HUMAN_QA_REQUIRED, selective pharmacology, tissue exposure, and multiorgan safety. H5 is not assigned.

18. Limitations

1. No female longitudinal trajectory spans adolescence to old age and locates the individual maximum.
2. Histology, RNA, steroids, and function rarely come from the same woman.
3. GTEx is postmortem bulk; composition, cause of death, terminal illness, and ischemia persist despite adjustment.
4. The recent adrenal cellular layer confounds sex and age; it does not decide female aging.
5. Historical RIA/immunoassays and LC–MS/MS do not share an absolute scale.
6. Young female adrenal tissue is difficult to obtain and may undergo surgical selection.
7. Oral DHEA does not reproduce endogenous DHEA-S, transport, or the tissue microenvironment.
8. Historical trials are small or too short for distal outcomes; nulls without equivalence may be imprecise.
9. Muscle evidence may be contaminated by fat/vessels and lacks a causal tracer.
10. Rare PAPSS2 genetics proves chemistry, not age-related prevalence.
11. MR depends on relevance, independence, exclusion restriction, LD, ancestry, and lifelong exposure; parental age is not own survival.
12. H6 integrates studies with different instruments and has not been harmonized.
13. We conducted neither a new meta-analysis nor individual-level cohort access.
14. There is no causal evidence of female life extension or a validated clinical biomarker.

19. Conclusions

1. An exact female maximum at age 25 has not been demonstrated. Evidence places the highest population mean in late adolescence/early adulthood and cannot locate an individual peak.
2. The age-related decline is large on average but not inexorable within each woman. The menopausal transition may include an inflection and estimated rise.
3. The adrenal ages selectively: the zona reticularis remodels and CYB5A/SULT2A1 decline without a uniform transcriptomic collapse of the cortex. We still do not know how much reflects fewer cells versus worse per-cell function.
4. Serum DHEA-S is a causally ambiguous reservoir. By itself it does not measure source, sulfation, entry, active product, or tissue signaling.
5. PAPSS2 genetics demonstrates that low DHEA-S may coexist with high downstream activation, but not that this topology is common in aging.
6. Human intracrinology exists in adipose; in female muscle the conversion→function chain remains unproven. Product without function would be a decisive negative result.
7. RCTs falsify a broad sufficient systemic-deficiency theory. Common cause/reserve is the leading explanation for muscle and lifespan.
8. Trabecular bone, ER+ cancer, and null lifespan suggest organ-specific effects of different sign, not a globally good or bad hormone. This hypothesis requires harmonized instruments and colocalization.
9. The experimental priority is to measure f_ZR, total flux, and per-cell flux in the same tissue. Only then does organ-specific translation make sense.
10. No branch justifies prescription, a clinical biomarker, systemic intervention, or Pharma readiness.

20. References

1. Orentreich N, Brind JL, Rizer RL, Vogelman JH. Age changes and sex differences in serum dehydroepiandrosterone sulfate concentrations throughout adulthood. J Clin Endocrinol Metab. 1984;59:551–555. DOI: 10.1210/jcem-59-3-551; PMID: 6235241.
2. Davison SL, Bell R, Donath S, Montalto JG, Davis SR. Androgen levels in adult females: changes with age, menopause, and oophorectomy. J Clin Endocrinol Metab. 2005. DOI: 10.1210/jc.2005-0212; PMID: 15827095.
3. Crawford S, Santoro N, Laughlin GA, et al. Circulating dehydroepiandrosterone sulfate concentrations during the menopausal transition. J Clin Endocrinol Metab. 2009. DOI: 10.1210/jc.2009-0386; PMID: 19470626.
4. Mazat L, Lafont S, Berr C, et al. Prospective measurements of dehydroepiandrosterone sulfate in a cohort of elderly subjects: relationship to gender, subjective health, smoking habits, and 10-year mortality. Proc Natl Acad Sci USA. 2001. DOI: 10.1073/pnas.121177998; PMID: 11427700.
5. Katagiri M, Kagawa N, Waterman MR. The role of cytochrome b5 in the biosynthesis of androgens by human P450c17. Arch Biochem Biophys. 1995. DOI: 10.1006/abbi.1995.1173; PMID: 7893148.
6. Dharia S, Slane A, Jian M, et al. Effects of aging on cytochrome b5 expression in the human adrenal gland. J Clin Endocrinol Metab. 2005. DOI: 10.1210/jc.2005-0017; PMID: 15870126.
7. Nanba AT, Rege J, Ren J, et al. 11-oxygenated C19 steroids do not decline with age in women. J Clin Endocrinol Metab. 2019. DOI: 10.1210/jc.2018-02527; PMID: 30753518; PMCID: PMC6525564.
8. Tezuka Y, Turcu AF, et al. The age-dependent changes of the human adrenal cortical zones are not congruent. J Clin Endocrinol Metab. 2021. DOI: 10.1210/clinem/dgab007; PMID: 33524149.
9. Iwahashi N, Umakoshi H, Fujita M, et al. Single-cell and spatial transcriptomics analysis of human adrenal aging. Molecular Metabolism. 2024;84:101954. DOI: 10.1016/j.molmet.2024.101954; PMCID: PMC11101872.
10. Noordam C, Dhir V, McNelis JC, et al. Inactivating PAPSS2 mutations in a patient with premature pubarche. N Engl J Med. 2009. DOI: 10.1056/NEJMoa0810489; PMID: 19474428.
11. Mueller JW, Gilligan LC, Idkowiak J, et al. Human DHEA sulfation requires direct interaction between PAPS synthase 2 and DHEA sulfotransferase SULT2A1. J Biol Chem. 2018. DOI: 10.1074/jbc.RA118.002248; PMID: 29743239.
12. Dalla Valle L, Toffolo V, Nardi A, et al. Tissue-specific transcriptional initiation and activity of steroid sulfatase complementing dehydroepiandrosterone sulfate uptake and intracrine steroid activations in human adipose tissue. J Endocrinol. 2006. DOI: 10.1677/joe.1.06811; PMID: 16837617.
13. Paatela H, Wang F, Vihma V, et al. Steroid sulfatase activity in subcutaneous and visceral adipose tissue: a comparison between pre- and postmenopausal women. Eur J Endocrinol. 2016. DOI: 10.1530/EJE-15-0831; PMID: 26553725.
14. Sipilä S, et al. Intramuscular sex steroid hormones are reduced after resistance training in postmenopausal women, but not affected by estrogen therapy. Steroids. 2022; PMID: 35809683.
15. Sanders JL, Boudreau RM, Cappola AR, et al. Concurrent change in dehydroepiandrosterone sulfate and functional performance in the oldest old. J Gerontol A. 2010. DOI: 10.1093/gerona/glq072; PMID: 20466773.
16. Cappola AR, O'Meara ES, Guo W, et al. Trajectories of dehydroepiandrosterone sulfate and mortality in older adults. J Gerontol A. 2009. DOI: 10.1093/gerona/glp129; PMID: 19713299; PMCID: PMC2773814.
17. Cappola AR, Xue QL, Walston JD, et al. DHEAS and mortality in disabled older women: a U-shaped association. J Gerontol A. 2006. DOI: 10.1093/gerona/61.9.957; PMID: 16960027.
18. Shufelt C, Bretsky P, Almeida CM, et al. DHEA-S levels and cardiovascular disease mortality in postmenopausal women: results from the National Institutes of Health—National Heart, Lung, and Blood Institute-sponsored Women's Ischemia Syndrome Evaluation (WISE). J Clin Endocrinol Metab. 2010. DOI: 10.1210/jc.2010-0143; PMID: 20739385.
19. Bromberger JT, et al. Longitudinal change in reproductive hormones and depressive symptoms across the menopausal transition: results from the Study of Women's Health Across the Nation (SWAN). Arch Gen Psychiatry. 2010; PMID: 20530009.
20. Wang Y, Islam RM, Hussain SM, et al. Associations between blood concentrations of sex hormones and physical function in community-dwelling older women. J Gerontol A. 2024. DOI: 10.1093/gerona/glad287; PMID: 38157294; PMCID: PMC10919882.
21. 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. DOI: 10.1056/NEJMoa054629; PMID: 17050889.
22. Dhatariya K, Bigelow ML, Nair KS. Dehydroepiandrosterone replacement therapy in hypoadrenal women: protein anabolism and skeletal muscle function. Mayo Clin Proc. 2008. DOI: 10.4065/83.11.1218; PMID: 18990320; PMCID: PMC2753533.
23. Dayal M, Sammel MD, Zhao J, et al. Supplementation with DHEA: effect on muscle size, strength, quality of life, and lipids. J Womens Health. 2005. DOI: 10.1089/jwh.2005.14.391; PMID: 15989411.
24. 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;93:534–538. DOI: 10.1210/jc.2007-1027; PMID: 18029465; PMCID: PMC2729150.
25. 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. DOI: 10.1111/j.1532-5415.2010.03019.x; PMID: 20863330.
26. von Mühlen D, Laughlin GA, Kritz-Silverstein D, Bergstrom J, Bettencourt R. Effect of DHEA supplementation on bone mineral density, bone markers, and body composition in older adults: the DAWN trial. Osteoporos Int. 2008. DOI: 10.1007/s00198-007-0520-z; PMID: 18084691; PMCID: PMC2435090.
27. Srinivasan M, Irving BA, Dhatariya K, et al. Effect of dehydroepiandrosterone replacement on lipoprotein profile in hypoadrenal women. J Clin Endocrinol Metab. 2009. DOI: 10.1210/jc.2008-1774; PMID: 19066301; PMCID: PMC2681274.
28. Nunes-Souza E, et al. From adrenarche to aging of adrenal zona reticularis: precocious female adrenopause onset. Endocrine Connections. 2020. DOI: 10.1530/EC-20-0416; PMID: 33112833; PMCID: PMC7774755.
29. Quester J, Nethander M, Eriksson AL, Ohlsson C. Endogenous DHEAS is causally linked with lumbar spine bone mineral density and forearm fractures in women. J Clin Endocrinol Metab. 2022;107:e2080–e2086. DOI: 10.1210/clinem/dgab915; PMID: 34935937; PMCID: PMC9016453.
30. Nounu A, Kar SP, Relton CL, Richmond RC. Sex steroid hormones and risk of breast cancer: a two-sample Mendelian randomization study. Breast Cancer Research. 2022;24:66. DOI: 10.1186/s13058-022-01553-9; PMID: 36209141.
31. Schooling CM, Zhao JV. Dehydroepiandrosterone sulfate on lifespan in men and women using Mendelian randomization. Nutr Metab Cardiovasc Dis. 2025;35:104128. DOI: 10.1016/j.numecd.2025.104128; PMID: 40450423.
32. Lasley BL, et al. Circulating dehydroepiandrosterone sulfate levels in women who underwent bilateral salpingo-oophorectomy during the menopausal transition. Menopause. 2011. DOI: 10.1097/gme.0b013e3181fb53fc; PMID: 21178790.

Scientific close: DHEA/DHEA-S contributions to women's health will not be resolved by normalizing a serum concentration. We must measure how much zona reticularis exists, how much flux each cell retains, which topology produces low sulfate, and whether a tissue converts that reservoir into a material function. Until then, DHEA-S is a biologically rich but causally ambiguous indicator, and female longevity cannot be attributed to its decline.