Peripheral aromatase: how adipose tissue converts DHEA-derived precursors into estrogens after menopause

Stage: REPORT_EN
Evidence cutoff: 2026-08-28
Index population: women with natural postmenopause; nonmalignant abdominal subcutaneous adipose tissue (SAT) and omental/visceral adipose tissue (VAT)
Question: What mechanisms connect the peripheral conversion of DHEA-derived precursors into estrogens with women's health and longevity, and what new hypotheses can be falsified?
Nature of this document: scientific research report; it contains no clinical recommendation or prescription.

Scientific delta of the report

This report replaces the linear narrative “more fat → more aromatase → more estrogen → more or less longevity” with a causal model whose estimands remain separate. DHEA is not a direct substrate of CYP19A1: the candidate adipose pathway requires transport and desulfation of DHEA-S, conversion of DHEA into androstenedione or testosterone, aromatization into E1/E2, partitioning between free and conjugated forms, and local or systemic disposition. Human evidence demonstrates these links in different preparations, but it still does not demonstrate the complete chain in intact postmenopausal abdominal adipose tissue.

The principal advance is to distinguish the capacity of a cell population from its contribution to whole-depot flux. Two independent human atlases locate transcript coverage compatible with the pathway in VAT mesothelial/premesothelial and ASPC populations, but the decisive genes are sparse, were measured in class-level pseudobulk, and the atlases do not record menopause. The finding therefore does not prove autonomy. The positive hypothesis is narrowed: an omental TM4SF1+ population will be causal only if its selective perturbation materially reduces intact-tissue DHEA-S→E1/E2 flux and a specific add-back restores it. Its priority rival predicts isolated capacity without a material contribution.

The second correction is that E1-S is a verified competing source. DHEA-S and E1-S share transport/STS, but the direction of their competition cannot be inferred from the number of steps: it depends on free availability, transport, kinetics, depot, and cell state. The third correction is functional: inhibiting CYP19A1 may accumulate A4/T and reroute signaling toward AR; consequently, a metabolic null is interpretable only if the steroidome is measured and ER is separated from AR. The link to healthspan or survival remains undemonstrated and requires an organ-specific chain with competing risks.

1. Executive scientific abstract

After menopause, loss of ovarian production makes peripheral steroidogenesis relatively more important. Adipose tissue contains machinery that can take up and transform sulfated steroids, aromatize immediate androgens, and interconvert E1/E2. Calling this process “conversion of DHEA by aromatase,” however, conceals two biochemical facts: CYP19A1 does not aromatize DHEA, and adipose E1 can also arise directly from E1-S.

The candidate chain is:

DHEA-S → DHEA → A4/T → E1/E2.

Its minimal gates are sulfate transport and STS; 3β-HSD activity for DHEA→A4; CYP19A1/POR for A4→E1 or T→E2; HSD17B for E1↔E2; and then conjugation, sequestration, or export. The competing E1-S→E1 route bypasses 3β-HSD and CYP19A1. Therefore, an STS measurement that does not specify substrate, a high E1 concentration, or CYP19A1 expression does not attribute the carbon to DHEA-S.

The most direct human evidence is fragmentary but coherent:

• A whole-body tracer study in seven postmenopausal women demonstrated peripheral DHEA→E1 without localizing the tissue.
• SAT/VAT homogenates from 18 premenopausal and seven postmenopausal women converted DHEA-S→DHEA; higher postmenopausal activity demonstrates STS capacity, not preservation of E1/E2 or function.
• Human preadipocytes/adipocytes took up and desulfated DHEA-S, but did not produce detectable A4/T/androstenediol at six hours in the direct study.
• Human breast stroma from six premenopausal donors converted DHEA→A4→E1 under confluence- and dexamethasone-dependent conditions.
• SAT and VAT from 37 nononcologic postmenopausal women converted E1-S→E1; SAT showed greater E1→E2 conversion than VAT.

Public computation adds localization, not causality. In two independent atlases, STS was more abundant in adipocytes; CYP19A1 and HSD3B shifted toward stroma depending on depot. VAT mesothelium met the STS–HSD3B–CYP19A1 class-level gate in 6/7 women in GSE176171 and 7/8 in Lazarescu; VAT ASPC met it in 3/6 and 4/8. This favors a depot-dependent mosaic, but it does not show that the three genes coexist in one cell, produce active protein, or make a material contribution to total flux.

The conclusion about health is calibrated. Animal models prove that adipose estrogen signaling can modify metabolism, but they do not identify DHEA-S carbon or transport effect magnitude or direction to postmenopausal women. Systemic DHEA trials lasting 6–24 months were largely null for body composition, strength, VO₂, insulin sensitivity, well-being, and several skeletal sites; they retained small, site-specific bone signals. They refute a broad benefit, not a narrow local function. There is no direct evidence that DHEA-derived adipose flux prolongs healthspan or survival.

The primary hypothesis, L8-2-HG-H1 v2, predicts a material contribution of an omental TM4SF1+ population to VAT flux. Competitor L8-2-AR-H1N v1 predicts isolated capacity without a material contribution. Translational hypothesis L8-2-HG-HT1 v2 remains parked: a flux vector will be eligible for evaluation only after causality and function are shown. The smallest discriminating program is Q0 metrology → E0 omental depletion/add-back → E1 dual DHEA-S/E1-S tracing → F0 adipose function with ER–AR separation.

Overall judgment: separated human biochemical links, H1; replicated class-level transcript localization, computational H2; integrated adipose chain, H0; human function specific to the flux, H0; healthspan/longevity, not established.

2. Scientific question and relevance

The question is not whether adipose tissue “has aromatase,” but:

1. what proportion of adipose estrogenic carbon comes from DHEA-S/DHEA versus E1-S or other precursors;
2. which cells and depots execute each step;
3. whether a competent population contributes materially to intact-tissue flux;
4. what fraction of product remains free, acts locally, is reconjugated, or is exported;
5. whether that dose modifies an independent function through ER after controlling androgen rerouting toward AR;
6. and only then, whether an organ-specific trajectory is relevant to disease-free years or survival.

This decomposition matters because postmenopause combines two potentially opposing trends: adrenal DHEA(S) supply falls with age, whereas some studies find greater STS or aromatase capacity in selected tissues or states. Compensation at one enzyme does not preserve flux if another gate or the precursor is limiting. Likewise, greater fat mass can increase whole-body capacity through cell number without raising per-cell capacity, while simultaneously changing SHBG, clearance, inflammation, and lipid sequestration.

Relevance to female longevity does not have a single valence. Local E1/E2 could support a metabolic or skeletal function and, at the same time, increase proliferative signaling in hormone-sensitive tissues. A possible benefit in one organ and cost in another must be analyzed with explicit mediators and competing risks; they cannot be summed into a post hoc composite.

3. Scope, population, and life stage

Primary population

The primary estimand concerns women aged 50–75 years with natural postmenopause documented by STRAW+10 and years since the final menstrual period, undergoing abdominal surgery for an indication independent of the research. The primary tissue is human omentum with an identifiable peritoneal surface and paired abdominal SAT, both nonmalignant.

The donor is the biological unit. Fragments, wells, time points, nuclei, and analytical replicates are nested units. Age, years since FMP, fat mass, adipocyte size, inflammation, DHEA-S, E1-S, SHBG, hepatic/renal function, anesthesia, and ischemia are retained as distinct determinants.

Exclusions from the causal core

Tumor or tumor-adjacent tissue, surgical or induced menopause, current hormone therapy, DHEA/prasterone, aromatase inhibitors, SERMs, antiandrogens, systemic glucocorticoids, and major adrenal, hepatic, or renal disease are excluded. These contexts may be separate comparators; they will not define basal physiology.

Transportability

Direct evidence comes from small European/North American cohorts and surgical samples. There is no basis for extrapolating magnitudes to Mexico/LATAM. Transportability across ancestry, adiposity, surgical practice, and stage is assessed only after mechanism and metrology are validated.

Out of scope

This project does not test interventions, doses, supplements, or clinical recommendations. It does not use current Lua user data. It proposes no product, digital variables, content, or marketing. Bone, vasculature, breast, endometrium, frailty, and survival remain later endpoints, not outputs of an adipose explant.

4. Background knowledge and mechanism map

4.1 Correct biochemistry

Aromatase CYP19A1 uses A4 and T, not DHEA or DHEA-S. The candidate long pathway is:

1. DHEA-S enters through sulfate transporters and STS releases DHEA.
2. A 3β-HSD activity, not yet localized by isoform in each adipose class, converts DHEA into A4.
3. CYP19A1, with POR, converts A4 into E1; a branch through T produces E2.
4. HSD17B7/HSD17B12 and HSD17B2/HSD17B14 are candidates for the direction of E1↔E2, but no isoform has been assigned causally in the complete chain.
5. SULT activity, esterification, binding, lipid droplets, and efflux determine whether product is sequestered, reconjugated, or reaches ER.

The short E1-S→E1 pathway shares transport/STS but bypasses 3β-HSD and CYP19A1. Therefore:

J_D ≤ the minimum of uptake+STS, 3β-HSD, and CYP19A1, multiplied by the recovered fraction.

J_E depends on uptake+STS and subsequent E1↔E2/conjugation partitioning.

These are topological expressions, not fitted kinetic models. They explain why the bottleneck can change with concentration, preparation, and depot.

4.2 Cellular architecture

Three models remain open:

• VAT boundary autonomy: a mesothelial/premesothelial population contains enough transport, STS, 3β-HSD, and CYP19A1 to close the pathway.
• Intercellular relay: adipocytes/preadipocytes desulfate DHEA-S and transfer DHEA/A4 to ASPC, fibroblasts, or mesothelium with greater downstream activity.
• Redundant mosaic: several fractions can close parts or all of the pathway; architecture, abundance, and substrate access determine net contribution.

The evidence favors mosaic as a description, not a single causal architecture. Aromatase was localized predominantly to the stromal/vascular fraction in historical studies; desulfation was demonstrated in adipocytes/preadipocytes; and DHEA→A4→E1 was demonstrated in breast stroma. No study observed transfer of a labeled intermediate in postmenopausal SAT/VAT.

4.3 Depot and organism scale

Depot flux is the sum of competent-cell number multiplied by cell-specific flux, modified by viable mass, architecture, interstitial availability, and retention/export. A rare cell layer can have high per-cell capacity yet contribute little to the tissue. Conversely, a surface layer might receive precursor directly and create a material microgradient despite low mass. Only perturbation of intact tissue separates these options.

Tissue content, effluent, free fraction, and nuclear signaling are different doses. Tissue E1/E2 concentrations exceeding serum demonstrate compartmentalization, not net synthesis. Serum E1/E2 mixes other tissues, SHBG, mass, volume, and clearance.

4.4 Conditional bridge to health

If free product is demonstrated:

• in adipose tissue, ER could modify insulin responsiveness, lipolysis, adipogenesis, or inflammation;
• in bone, downstream steroids could affect remodeling;
• in vasculature, ER could modulate endothelium or smooth muscle;
• in breast/endometrium, E1/E2 and inflammatory context could favor survival or expansion.

None of these links is currently attributed to adipose DHEA-S carbon in women. Ligand identity matters: E1 and E2 must not be pooled as “total estrogen,” and aromatase blockade may increase androgens and AR signaling.

5. Evidence method

All previous project artifacts were integrated: SCOPING, EVIDENCE_MAP, EVIDENCE_VERIFICATION, MECHANISTIC_SYNTHESIS, COMPUTE_DECISION, COMPUTE_OPTIONAL, HYPOTHESIS_GENERATION, ADVERSARIAL_REVIEW, and EXPERIMENT_DESIGN, together with the lifetime memory packet and structured project state.

Decisive sources were audited for design, population, stage, tissue, sample size, substrate, comparator, outcome, uncertainty, and primary identifier. Evidence was classified as follows:

The adversarial search prioritized evidence able to refute universal menopause induction, E1-S dominance from pathway length, RNA-inferred mesothelial autonomy, estrogen-attributed metabolic function without AR control, and broad DHEA benefit. Recent claims about adipose atlases, omental populations, and DHEA-S metabolism in cell lines were reverified against 2024–2025 primary publications.

Absence of an end-to-end study is stated as “not identified in the verified search,” not as proof of nonexistence.

6. Evidence map

6.1 Direct human links

6.2 Scale and compartment determinants

Historical studies related body weight or fat mass to whole-body A4→E1 conversion and production per gram. In stromal cells, per-cell activity increased modestly with age and not with weight; oophorectomy in young women did not reproduce the increase. This supports separating age, menopause, total mass, and cell capacity.

Tissue and serum concentrations can decouple by depot, hormone therapy, and tissue subtype. A high tissue concentration may reflect uptake, storage, interconversion, or low efflux—not local production. SHBG and clearance add systemic confounding.

6.3 Functional and distal evidence

No study was found that measures adipose DHEA-S→E1/E2 flux, organ exposure, and healthspan or survival.

7. Contradictory evidence and null findings

7.1 Menopause is not a universal aromatase switch

Paatela found greater STS capacity in only seven postmenopausal women, whereas historical studies separated age, weight, and menopause and did not identify universal induction. Inflamed breast adipose tissue from high-risk cohorts shows elevated aromatase, but this does not automatically transport to healthy abdominal SAT/VAT. The most parsimonious model is heterogeneity by age, depot, composition, and inflammation—not a single postmenopausal program.

7.2 Capacity is not contribution

The VAT mesothelial signal rests on class-level pseudobulk with sparse HSD3B/CYP19A1. Even if an isolated fraction makes E1/E2, low abundance could leave intact VAT flux unchanged. Different genes may also belong to different cells within the same label. The adversarial test is not another atlas, but selective depletion/perturbation and add-back.

7.3 A short pathway is not dominance

E1-S requires fewer steps, but one MCF-7 study found that DHEA-S 1 μM reduced apparent hydrolysis of E1-S 2 nM by about 70% without a reciprocal change in DHEA-S. This historical tumor context does not decide adipose physiology; it does refute inferring direction from topology. Recombinant human STS hydrolyzes both sulfates. Free availability, transport, and substrate ratio must be measured.

7.4 Product is not function

Systemic DHEA RCTs changed several steroids without broad benefit. They do not prove that a narrow intracrine effect is zero. Conversely, a change in E1/E2 or proximal ER signaling does not demonstrate glucose uptake, lipolysis, or systemic health. A null after CYP19A1 blockade is ambiguous if A4/T rises and AR changes.

7.5 One organ is not longevity

Small skeletal signals coexist with metabolic nulls and plausible hormone-sensitive risks. Sizes, instruments, and outcomes cannot be merged. There is no direct evidence of a net balance for disease-free years or personal survival.

8. Multiscale mechanistic synthesis

Molecular

DHEA-S→E1/E2 flux is a serial chain. Higher STS does not necessarily compensate for lower DHEA-S; higher CYP19A1 does not raise product if 3β-HSD or supply is limiting. The bottleneck can change with concentration: historical human placental microsomes showed STS limitation at low concentration and aromatase limitation at higher concentration, reminding us that the gate is not fixed.

Cellular

Human triangulation suggests partial specialization: adipocytes/preadipocytes desulfate, stroma aromatizes, and some VAT populations cover the entire pathway. Cell separation, however, changes identity and microgradients. Autonomy will be demonstrated only by temporal order, sequential blocks, and bypasses in an intact fraction.

Tissue

SAT and VAT differ in anatomy and E1/E2 partitioning. Mesothelium/premesothelium is specific to omentum; comparing it with SAT confounds depot and presence of a boundary layer. The decisive estimand is the change in omental-explant J_D after a selective manipulation, accompanied by abundance, injury, and reconstruction.

Systemic

Declining adrenal DHEA(S), fat mass, SHBG, and clearance modify circulating concentrations. Greater adipose capacity may act locally without raising serum, or high serum may reflect another source. Export requires tissue–effluent mass balance and free-fraction measurement.

Health and longevity

A causal adipose function requires flux → free E1/E2 mixture → ER → independent endpoint, with AR controlled. Other organs require new chains. A net healthspan effect will be identifiable only after human temporality by organ and competing risks. Today the hypothesis of a null or small net effect follows from uncertainty and possible compensation, not from demonstrated survival equivalence.

9. Computational layer

Question and data

The analysis asked whether transcript gates fell within one class or complementary classes. Public datasets were used:

• GSE176171: 137,684 human nuclei, 13 donors, 10 women and 3 men; SAT/VAT with donor, sex, age, and BMI.
• GSE281356/Lazarescu: 37,879 SAT nuclei and 83,731 VAT nuclei; 15 samples from 10 donors.

Neither atlas records menopause. Age was not used as a substitute. The unit was the female donor; counts were aggregated by donor×depot×class without imputation or pseudoreplication.

Result

STS was ubiquitous and higher in adipocytes. CYP19A1 was enriched in stroma, especially SAT ASPC and VAT mesothelial/perivascular classes. HSD3B was the sparse gate. The strict class-level STS∧HSD3B1/2∧CYP19A1 gate replicated in VAT mesothelium and VAT ASPC. Using AKR1C3 as a substitute for 3β-HSD made the pathway artificially ubiquitous and was rejected.

Interpretation

The classification is “depot-dependent mosaic with autonomy-compatible stromal VAT foci.” This is replicated human observational computational evidence, H2 for class localization. It does not demonstrate cellular coexpression, protein, flux, competition, postmenopause, function, or longevity.

Experimental decision

Prioritize omental mesothelial/premesothelial and ASPC fractions, preserve paired SAT–VAT, and retain reconstitution/transwell as a falsifier. Computation changed the order of validation, not causal maturity.

10. Primary hypothesis

L8-2-HG-H1 v2 — material, abundance-weighted contribution of omental TM4SF1+

Falsifiable statement: in intact postmenopausal omental VAT, a TM4SF1+ population—classical MSLN-high or transitional MSLN-low/IGFBP2+—will sequentially close DHEA-S→DHEA→A4→E1/E2, and its selective depletion or perturbation will reduce total J_D by at least a prespecified material fraction; SAT will not show an equivalent autonomous contribution and will depend more on multicellular architecture.

Mechanism: a boundary population with precursor access expresses functional transport/STS, HSD3B, and CYP19A1; product diffuses into parenchyma or effluent. Ferrero et al. identified an omental TM4SF1+/MSLN-low population enriched for IGFBP2 and distinct from classical TM4SF1+/MSLN-high mesothelium, so the two must be separated.

Evidence for: replicated class-level gate in VAT mesothelium; human adipose STS and aromatase activity; DHEA→A4→E1 in human stroma.

Evidence against: sparse genes, pseudobulk, missing menopause, possible low mass, contamination or intracategory mixing; direct sources derive from other depots/stages.

Core predictions:

1. An isolated TM4SF1+ fraction will form labeled DHEA, A4, and E1/E2 in temporal order.
2. STS blockade will reduce downstream products and DHEA will bypass it; 3β-HSD blockade will accumulate DHEA and A4 will bypass it; CYP19A1 blockade will accumulate A4/T and reduce E1/E2.
3. Valid depletion will reduce explant J_D by ≥20%; specific add-back will restore the component.
4. The effect will persist after weighting abundance, surface area, and viability.
5. SAT will show a greater excess from adipocyte–stromal reconstitution than omentum.

Kill criteria:

• no labeled E1/E2 with valid Q0, entry, viability, and positive controls;
• isolated product without a material change in intact explant;
• lack of specific rescue;
• signal explained by contamination or injury;
• equivalent SAT/VAT architecture;
• dependence on one donor or supraphysiological exposure.

Status: weakened and narrowed by adversarial review.
Maturity: integrated H0; H2 only for transcript class localization.
Confidence: 0.18.

11. Competing hypothesis

L8-2-AR-H1N v1 — boundary capacity without material contribution

Falsifiable statement: an omental TM4SF1+ population may show DHEA-S→E1/E2 capacity in isolation, but its removal or perturbation will leave intact VAT flux within 0.80–1.25 equivalence because it is rare, has limited access, or because STS, HSD3B, and CYP19A1 reside in different cells within the label.

Mechanism: high per-cell capacity does not compensate for low abundance; parenchyma, ASPC, or another source such as E1-S dominates flux. Computational coverage may represent intracategory complementarity.

Evidence for: sparse HSD3B/CYP19A1; no direct coexpression or flux; mesothelial layer without a SAT equivalent; tissue content and capacity do not predict export.

Evidence against: a rare surface population may directly receive precursor and create microgradients; sparse RNA may underestimate protein.

Core predictions:

1. Per-cell TM4SF1+ capacity will not predict explant J_D across donors.
2. Abundance weighting will reduce its estimated contribution below 20%.
3. Valid depletion will leave the 90% CI of depleted:sham J_D within 0.80–1.25.
4. Multiplex RNAscope/protein will show cellular separation or insufficient triple-positive frequency.
5. TM4SF1+ add-back will not materially change reconstructed microtissue.

Kill criteria:

• depletion reproducibly reduces J_D and specific add-back restores it;
• abundance-weighted contribution is material in multiple donors;
• the same triple-positive cell demonstrates the functional sequence.

Status: proposed, priority competitor.
Maturity: H0.
Confidence: 0.52.

Secondary falsifiers that this competition must preserve

L8-2-AR-H2R v1 predicts that DHEA-S:E1-S competition changes with free ratio, depot, and transport, without universal dominance. L8-2-HG-H3 v2 predicts that even with valid J_D, adipose function will remain equivalent after ER and AR are controlled. These rivals do not replace the H1 versus H1N comparison; they prevent extending a positive contribution to dominant source or health.

12. Translational hypothesis

L8-2-HG-HT1 v2 — causal research vector, not a biomarker

Falsifiable statement: only if a causal adipose function replicates may a frozen minimal vector of J_D, J_E, E1/E2 identity, and free/exported fraction predict that same function out of sample better than static DHEA-S, E1/E2, SHBG, mass/depot, stage, and clearance.

Mechanism: a dynamic response would separate source, capacity, and fate that a static concentration mixes. Its validity depends completely on the causal parent hypothesis.

Predictions:

1. Each component will pass selectivity, LOD/LOQ, recovery, stability, test–retest, ICC/CCC, and interlaboratory concordance.
2. The vector will classify causal topology in donors held out from fitting.
3. It will reduce error and improve external calibration of the primary function over static comparators.
4. Gain will be depot- and function-specific; it will not be extrapolated to bone, cancer, healthspan, or survival.

Kill criteria:

• death of parent causality or equivalent function;
• failed metrology or concordance;
• no external gain;
• signal explained by batch, isolation, composition, adiposity, liver/kidney, or therapy;
• need for post hoc retuning or high dimensionality.

Status: parked; HUMAN_QA_REQUIRED before any translational evaluation.
Maturity: H0.
Confidence: 0.03.

13. Falsifiable predictions and kill criteria

Global stopping rule: if material J_D is not detected despite valid isotope identity, entry, viability, positive controls, and bypasses, the physiological adipose extension is refuted in that model/range and function or distal organs are not studied.

14. Discriminating experiment

Q0 — metrology and stability

Validate LC–MS/MS for DHEA-S, DHEA, androstenediol, A4, T, DHT, E1-S, E1, E2, and conjugates in medium, free fraction, and tissue. Use 13C-enriched DHEA-S and nonoverlapping deuterated E1-S, with positions fixed after purity and fragmentation review.

Gates: CV ≤15% outside the LLOQ and ≤20% at the LLOQ; recovery 80–120% or stable correction with CV ≤15%; carryover <20% of the LLOQ; identity by two transitions or transition+retention. Measure adsorption, stability, natural abundance, cross-talk, and free fraction.

The biological window is defined at 0, 30, 60, 120, and 360 minutes. Use the first linear segment above LOQ that maintains viability ≥85%, LDH/ATP within 20% of sham, and no >2× increase in IL8, IL1B, TNF, HIF1A, or PTGS2. Explant inflammatory drift will not count as basal physiology.

E0 — capacity versus contribution

Preparations per donor:

1. intact omental explant;
2. SAT explant;
3. mature adipocytes;
4. classical TM4SF1+/MSLN-high;
5. transitional TM4SF1+/MSLN-low/IGFBP2+;
6. TM4SF1− ASPC;
7. total SVF;
8. adipocyte+ASPC and adipocyte+TM4SF1+ reconstitutions;
9. transwell if yield is sufficient.

Surface depletion is interpretable only if it reduces the target population by ≥80%, loses ≤20% of ASPC/endothelium/adipocytes, and maintains injury/viability within 0.80–1.25 of sham. If it fails, no null is declared: intact contribution remains unresolved, and microtissue depletion/add-back plus weighting is used.

Primary endpoint: donor-level molar slope of labeled E1+E2 from DHEA-S. Confirmatory contrast: depleted:sham J_D ratio. HG-H1 requires a ≥20% reduction with a 95% CI excluding a smaller effect and specific rescue; H1N requires a 90% CI fully within 0.80–1.25 and an upper contribution bound <20%.

E1 — source and competition

E1-A titrates DHEA-S and E1-S separately until the intracellular AUC of desulfated precursor is matched. Endpoint: labeled E1+E2 AUC / intracellular precursor AUC.

E1-B uses a 3×3 matrix of 0.5×, 1×, and 2× donor-level free concentrations in intact explant and paired homogenate. STS blockade affects both routes; DHEA and A4 localize upstream/downstream competition. Equivalent interaction kills the selector; sign reversal favors AR-H2R.

F0 — function with ER–AR separation

It opens only if J_D explains ≥20% of estrogenic flux or a valid perturbation materially changes free E1/E2. The sole endpoint is insulin-stimulated 2-deoxyglucose uptake in explant.

Matrix: vehicle; gate blockade; blockade+autologous E1:E2 mixture; rescue+ER blockade; AR control; viability control. Measure free A4, T, DHT, E1, and E2 simultaneously.

HG-H3 v2 dies only if two perturbations change J_D and function; E1/E2 rescues; ER removes the rescue; AR does not explain the effect; viability/composition/inflammation are comparable; and the result replicates in another donor set and laboratory.

Sample size and analysis

An external pilot of six donors, excluded from confirmation, estimates donor-level variance, valid-tissue rate, LLOQ, and within-donor correlation. Fixed confirmatory n=12–24, set by simulation: 90% power for a reduction ≥20%, at least 80% for equivalence 0.80–1.25, tolerating 15% nonevaluable. If >24 donors are required, the assay is declared infeasible in that form; the SESOI is not reduced.

Mixed model on log scale with fixed condition, depot, and preparation; donor intercept; prespecified batch, ischemia, and composition. 95% CIs for materiality; TOST and 90% CIs for equivalence. Intervals allowing both equivalence and materiality are inconclusive.

15. Measurement candidates and stratification

There is no validated biomarker. The following are research measurement candidates:

Prespecified stratification: SAT versus omentum; years since FMP; age; mass and adipocyte size; inflammation/fibrosis; free DHEA-S/E1-S; SHBG; hepatic/renal function; prior hormonal exposure; ischemia time. Interactions remain exploratory until replication.

16. Individual variability

Variation among women may arise from six layers:

1. Adrenal source: DHEA(S) trajectory, sulfation, and HPA.
2. Entry: transporters, free fraction, albumin, and microperfusion.
3. Composition: proportions of adipocytes, ASPC, mesothelium/premesothelium, vasculature, and immune cells.
4. Cell state: inflammation, glucocorticoids, confluence, fibrosis, and hypoxia.
5. Fate: HSD17B, reconjugation, lipid droplets, and efflux.
6. Response: ESR1/ESR2/GPER, AR, insulin signaling, and prior health.

BMI does not represent these layers. Postmenopause is also not homogeneous: age and years since FMP must be modeled separately. Surgical selection may enrich obesity, diabetes, or inflammation. Technical variability—ischemia, batch, isolation, inflammatory drift—can mimic biology and is blocked/measured.

Lack of LATAM data prevents estimation of heterogeneity by ancestry or environment. Quantitative effects are not extrapolated; the chain is validated first.

17. Pharma relevance and maturity

Judgment

There is no readiness for a drug, indication, partnering, or H5. The current opportunity is target de-risking through system pharmacology and human tissue. HUMAN_QA_REQUIRED before external assessment.

Kill Pharma: a target is eliminated for this pathway if it changes metabolites without function, requires supraphysiological exposure, redistributes the steroidome without selectivity, or does not permit multiorgan risks to be decoupled.

18. Limitations

1. Direct postmenopausal adipose DHEA-S evidence uses only seven women in the decisive STS study.
2. DHEA→A4→E1 was demonstrated in premenopausal breast stroma, not postmenopausal omentum.
3. No study integrates transport, STS, HSD3B, CYP19A1, product, export, and function in the same tissue.
4. The atlases do not record menopause and key genes are sparse; pseudobulk is not coexpression.
5. TM4SF1/MSLN/IGFBP2 markers separate useful populations, not absolute exclusivity.
6. Surface depletion may injure tissue or miss interlobular populations.
7. Interstitial free fraction is not precisely established; the explant lacks perfusion and clearance.
8. STS/HSD3B/CYP19A1 inhibitors affect other pathways; balance, rescue, and orthogonality are required.
9. Labeled E1/E2 may fall below the LLOQ; raising dose outside the range invalidates physiology.
10. Systemic DHEA RCTs do not reproduce the local pathway and are short for events.
11. Causal functional evidence comes mainly from rodents or breast context.
12. There is no chain to frailty, multimorbidity, healthspan, or survival.
13. There are insufficient data for Mexico/LATAM magnitudes.
14. An ex vivo adipose function does not represent systemic health.

19. Conclusions

1. The phrase “aromatase converts DHEA” is biochemically incomplete. DHEA-S requires transport/STS, DHEA→A4/T, and only then CYP19A1→E1/E2.
2. Human adipose possesses all capacities separately, but the end-to-end chain in intact postmenopausal abdominal tissue has not been demonstrated.
3. E1-S is a direct competing source. The dominant source cannot be inferred from pathway length, total concentrations, or STS without substrate.
4. Computation identifies a depot mosaic and stromal VAT foci, not cellular autonomy. Class capacity and depot contribution are different estimands.
5. The most useful hypothesis pits a material omental TM4SF1+ contribution against rare capacity without contribution. The decisive experiment is tracer-based depletion/add-back with mass balance.
6. A functional null is interpretable only after free A4/T/E1/E2 are measured and ER is separated from AR.
7. Systemic RCTs refute a broad DHEA benefit, but not a narrow local function; nor do they demonstrate longevity.
8. Healthspan and survival remain unestablished. Only an organ-specific causal chain with temporality and competing risks would permit the net balance to be addressed.

The most valuable scientific result may be negative: demonstrating cell capacity without tissue contribution, competition without a universal direction, or chemistry without function would close three shortcuts that currently confound adiposity, aromatase, and longevity.

20. References

1. Longcope C, Bourget C, Flood C. The production and aromatization of dehydroepiandrosterone in post-menopausal women. Maturitas. 1982;4:325–332. DOI: 10.1016/0378-5122(82)90065-2. PMID: 6221177. https://pubmed.ncbi.nlm.nih.gov/6221177/
2. Paatela H, et al. Steroid sulfatase activity in subcutaneous and visceral adipose tissue: a comparison between pre- and postmenopausal women. Eur J Endocrinol. 2016;174:167–175. DOI: 10.1530/EJE-15-0831. PMID: 26553725. https://pubmed.ncbi.nlm.nih.gov/26553725/
3. Dalla Valle L, et al. Adipose tissue acts as a steroidogenic organ by expressing steroid sulfatase and 17β-hydroxysteroid dehydrogenases. J Endocrinol. 2006. DOI: 10.1677/joe.1.06811. PMID: 16837617. https://pubmed.ncbi.nlm.nih.gov/16837617/
4. Killinger DW, et al. Estrone formation from dehydroepiandrosterone in cultured human breast adipose stromal cells. J Steroid Biochem Mol Biol. 1995. DOI: 10.1016/0960-0760(94)00164-H. PMID: 7873453. https://pubmed.ncbi.nlm.nih.gov/7873453/
5. Ackerman GE, et al. Aromatization of androstenedione by human adipose tissue stromal cells in monolayer culture. J Clin Endocrinol Metab. 1981. DOI: 10.1210/jcem-53-2-412. PMID: 7251819. https://pubmed.ncbi.nlm.nih.gov/7251819/
6. Cleland WH, Mendelson CR, Simpson ER. Effects of aging and obesity on aromatase activity of human adipose cells. J Clin Endocrinol Metab. 1985;60:174–177. DOI: 10.1210/jcem-60-1-174. PMID: 3964790. https://pubmed.ncbi.nlm.nih.gov/3964790/
7. MacDonald PC, et al. Effect of obesity on conversion of plasma androstenedione to estrone in postmenopausal women with and without endometrial cancer. Am J Obstet Gynecol. 1978;130:448–455. DOI: 10.1016/0002-9378(78)90287-9.
8. Forney JP, et al. Aromatization of androstenedione to estrone by human adipose tissue in vitro. J Clin Endocrinol Metab. 1981;53:192–199. DOI: 10.1210/jcem-53-1-192.
9. Longcope C, Baker R. Androgen and estrogen dynamics: relationships with age, weight, and menopausal status. J Clin Endocrinol Metab. 1993;76:601–604. DOI: 10.1210/jcem.76.3.8445016.
10. Hetemäki N, et al. Estrogen Metabolism in Abdominal Subcutaneous and Visceral Adipose Tissue in Postmenopausal Women. J Clin Endocrinol Metab. 2017;102:4588–4595. DOI: 10.1210/jc.2017-01474. PMID: 29029113. https://pubmed.ncbi.nlm.nih.gov/29029113/
11. Kinoshita T, et al. An innovative LC–MS/MS-based method for determining CYP17 and CYP19 activity in adipose tissue using 13C-labeled steroid substrates. J Clin Endocrinol Metab. 2014. DOI: 10.1210/jc.2013-3715. PMID: 24456285. https://pubmed.ncbi.nlm.nih.gov/24456285/
12. Iyengar NM, et al. Menopause Is a Determinant of Breast Aromatase Expression and Its Associations With BMI, Inflammation, and Systemic Markers. J Clin Endocrinol Metab. 2017;102:1692–1701. DOI: 10.1210/jc.2016-3606. PMID: 28323914. https://pubmed.ncbi.nlm.nih.gov/28323914/
13. Qureshi R, et al. The Major Pre- and Postmenopausal Estrogens Play Opposing Roles in Obesity-Driven Mammary Inflammation and Breast Cancer Development. Cell Metab. 2020;31:1154–1172.e9. DOI: 10.1016/j.cmet.2020.05.008. PMID: 32492394. https://pubmed.ncbi.nlm.nih.gov/32492394/
14. Emont MP, et al. A single-cell atlas of human and mouse white adipose tissue. Nature. 2022;603:926–933. DOI: 10.1038/s41586-022-04518-2. PMID: 35296864. https://pubmed.ncbi.nlm.nih.gov/35296864/
15. Lazarescu O, et al. Human subcutaneous and visceral adipocyte atlases uncover classical and nonclassical adipocytes and depot-specific patterns. Nat Genet. 2025;57:413–426. DOI: 10.1038/s41588-024-02048-3. PMID: 39856219. https://pubmed.ncbi.nlm.nih.gov/39856219/
16. Ferrero R, et al. A human omentum-specific mesothelial-like stromal population inhibits adipogenesis through IGFBP2 secretion. Cell Metab. 2024. DOI: 10.1016/j.cmet.2024.04.017. PMID: 38729152. https://pubmed.ncbi.nlm.nih.gov/38729152/
17. Karakuş E, et al. Intracrine Formation of Steroid Hormones in Breast Cancer, Epidermal Keratinocyte, Dermal Fibroblast, and Adipocyte Cell Lines Measured by LC-MS/MS. Int J Mol Sci. 2025;26:1188. DOI: 10.3390/ijms26031188. PMID: 39940958. https://pubmed.ncbi.nlm.nih.gov/39940958/
18. MacIndoe JH, Woods G, Jeffries L, Hinkhouse MM. The Hydrolysis of Estrone Sulfate and Dehydroepiandrosterone Sulfate by MCF-7 Human Breast Cancer Cells. Endocrinology. 1988;123:1281–1287. DOI: 10.1210/endo-123-3-1281. PMID: 2969800. https://pubmed.ncbi.nlm.nih.gov/2969800/
19. Purohit A, Williams GJ, Howarth NM, Potter BVL, Reed MJ. The hydrolysis of oestrone sulphate and dehydroepiandrosterone sulphate by human steroid sulphatase expressed in transfected COS-1 cells. J Steroid Biochem Mol Biol. 1995;52:101–104. DOI: 10.1016/0960-0760(94)00177-C.
20. Gibb W, Lavoie JC. Kinetic Studies on the Formation of Estrogens from Dehydroepiandrosterone Sulfate by Human Placental Microsomes. Endocrinology. 1984;114:2323–2329. DOI: 10.1210/endo-114-6-2323.
21. Van Sinderen ML, et al. Effects of Estrogens on Adipokines and Glucose Homeostasis in Female Aromatase Knockout Mice. PLoS One. 2015;10:e0136143. DOI: 10.1371/journal.pone.0136143. PMID: 26317527. https://pubmed.ncbi.nlm.nih.gov/26317527/
22. Ohlsson C, et al. Increased adipose tissue aromatase activity improves insulin sensitivity and reduces adipose tissue inflammation in male mice. Am J Physiol Endocrinol Metab. 2017;313:E450–E462. DOI: 10.1152/ajpendo.00093.2017. PMID: 28655716. https://pubmed.ncbi.nlm.nih.gov/28655716/
23. Nair KS, et al. DHEA in elderly women and DHEA or testosterone in elderly men. N Engl J Med. 2006;355:1647–1659. DOI: 10.1056/NEJMoa054629. PMID: 17050889. https://pubmed.ncbi.nlm.nih.gov/17050889/
24. von Mühlen D, et al. Effect of dehydroepiandrosterone supplementation on bone mineral density, bone markers, and body composition in older adults: the DAWN trial. Osteoporos Int. 2008;19:699–707. DOI: 10.1007/s00198-007-0520-z. PMID: 18084691. https://pubmed.ncbi.nlm.nih.gov/18084691/
25. Jankowski CM, et al. Effects of dehydroepiandrosterone replacement therapy on bone mineral density in older adults. J Clin Endocrinol Metab. 2006;91:2986–2993. DOI: 10.1210/jc.2005-2484.
26. Panjari M, et al. 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. DOI: 10.1111/j.1743-6109.2009.01381.x. PMID: 19619146. https://pubmed.ncbi.nlm.nih.gov/19619146/
27. Fain JN, et al. The inflammatory response seen when human omental adipose tissue explants are incubated in primary culture is not dependent upon albumin and is primarily in the nonfat cells. J Inflamm. 2010;7:4. DOI: 10.1186/1476-9255-7-4.
28. Stylianou E, et al. Optimal method for isolation of human peritoneal mesothelial cells from clinical samples of omentum. J Invest Surg. 2006. PMID: 17153120. https://pubmed.ncbi.nlm.nih.gov/17153120/
29. Friede T, Kieser M. Blinded sample size reassessment in non-inferiority and equivalence trials. Stat Med. 2003;22:995–1007. DOI: 10.1002/sim.1456.