Progesterone deficit in perimenopause: why it appears to fall before estrogen and what consequences it may have

Stage: REPORT_EN
Closing date: 2026-08-05
Project: L7-4
Core population: natural menopausal transition, STRAW+10 −3a to −2/−1
Classification: progesterone and neurosteroid metabolomics
Document type: scientific research report; not diagnosis, prescribing, or an evaluation of clinical efficacy
Global maturity: H1–H2 for loss of luteal activity/exposure; H0 for P4→E2 temporal priority, bone P4–PGR mediation, perimenopausal neurosteroid effects, and longevity

Concrete scientific delta of the report

The conclusion is not that a “cellular progesterone deficiency” has been shown to precede an “estrogen deficiency.” Human evidence supports a more precise statement: during the transition, the probability of cycles without a robust luteal phase rises and integrated exposure to pregnanediol glucuronide (PdG) slowly decreases, while estrogen exposure may retain high peaks and become erratic. This is compatible with a topological asymmetry—follicular estradiol is produced before ovulation, whereas a high-progesterone phase requires completion of an LH surge, follicular rupture, luteinization, and maintenance of a vascularized corpus luteum—but no verified study has estimated paired P4/PdG and E2/E1G change-points within the same woman at sufficient resolution.

The second delta is causal and adversarial. In bone, ovarian reserve/HPO-axis state and turnover explain the human data better than a simple low P4→low PGR→less formation pathway. Progesterone is metabolized locally in human osteoblasts and PGR function depends on cellular stage; however, metabolism does not equal receptor engagement or function. The distal bone-balance→fracture bridge has been observed in SWAN, so the concrete gap is no longer “progesterone→longevity” in the abstract, but integrated P4→intracellular P4/PGR engagement→formation–resorption balance→fracture. The first two links remain unproven.

Final review of the recent primary source identified an internal inconsistency that must not be smoothed over: the abstract of Shieh et al. 2026 reports an HR of 0.75 per SD of the bone balance index for major osteoporotic fracture, with CI95% 0.61–0.85, whereas the Results section and Table 5 report 0.61–0.95 for the same estimate. This report uses 0.61–0.95 as the interval supported by the Results/table and retains 0.61–0.85 as an apparent abstract error. The discrepancy does not change the interpretation: the result is distal, observational, and not progesterone-specific.

1. Executive scientific abstract

Cyclic progesterone and estradiol are not equivalent products of the same ovarian moment. Estradiol is generated during follicular growth. High progesterone exposure additionally requires competent positive feedback, an LH surge, follicular rupture, luteinization, and formation of a functional corpus luteum for enough days. This serial architecture provides a strong reason why progestational exposure could lose stability before estrogen peaks disappear.

Human data support the first half of that explanation. In SWAN, apparently ovulatory cycles declined from 80.9% to 64.7% across three annual assessments, and anovulatory/no-bleeding collections rose from 8.4% to 24%; PdG declined 6.6% per year in the estimand that included all collections [3]. In the adjusted analysis restricted to ovulatory cycles, decline by time in study was 5.4% per year for PdG and 1.4% for E1c, but no formal test between slopes was performed [3]. In another SWAN follow-up, 87.9% of cycles still showed evidence of luteal activity five years before the final menstrual period (FMP), and 22.8% did so in the preceding year; hormones in cycles with activity remained relatively stable until approximately three years before the FMP [4]. BIMORA likewise placed the major change in anovulation and PdG AUC in late transition rather than in a homogeneous early decline [5].

Therefore, the strong claim that “progesterone falls before estrogen” has not been demonstrated. Slope, increased mass of low-exposure cycles, increased variance, and change-point are different estimands. In addition, urinary PdG integrates production, duration, metabolism, and excretion; a single serum P4 sample correlated weakly with integrated luteal uPdG (r=0.26; total R²=0.09) [9]. Defining luteal activity with PdG and then using it to explain PdG adds circularity.

Consequences are organ-specific:

• In the endometrium, there is pharmacologic causal evidence that adding progestogenic signaling protects against hyperplasia induced by exogenous estrogen, and the PGR–HAND2–FGF/ERK axis has causal support in mice [23,24]. The endogenous perimenopausal trajectory is missing.
• In brain/VMS/sleep, HPO disruption is associated with vasomotor symptoms, but mood was null and a micronized-progesterone trial was null for its primary VMS endpoint and for the sleep-problems diary [12,13]. Peripheral P4 does not identify brain allopregnanolone or GABA-A adaptation.
• In bone, ovulation/P4 findings are mixed, whereas FSH, AMH, E2, and CTX provide more consistent human signal [14–19]. The P4–PGR subpath remains plausible but is H0.
• For vasculature, metabolism, fracture attributable to P4, healthspan, and longevity, there is no complete human chain.

The revised primary hypothesis is a distributive gate observed first; its rivals separate measurement/selection artifact from a common HPO clock. The translational prognostic hypothesis remains parked. The smallest experiment begins with an identifiability simulation, not another hormonal analysis: if the design cannot distinguish a true 9–12-month lead from simultaneity, no P value can rescue the question.

2. Scientific question and relevance

Primary question

During the natural menopausal transition, does loss of the probability and duration of a functional luteal phase reproducibly precede change in integrated E2/E1G exposure within woman, or are both outputs with different thresholds of the same ovarian-reserve/HPO-axis clock?

Mechanistic question

If progestational exposure decreases separately, what fraction is attributable to:

1. anovulation or failure to maintain the gate;
2. fewer days, less mass, or lower perfusion of the corpus luteum;
3. lower secretion per luteal cell at comparable state and support;
4. metabolism, distribution, and clearance;
5. phase, matrix, classification, or selection error?

Consequence question

Does any organ-specific outcome depend on P4 or a defined metabolite/receptor after separating E2, FSH, AMH, ovulation, stage, energy availability, and other axes?

Relevance to health and longevity

Relevance does not follow merely because a hormone changes. An independent exposure must precede a proximal function, that function must alter a disease domain, and the effect must persist under competing risks. In this project, bone provides the most mature distal bridge: turnover balance and BMD are related to fracture. But the P4→balance bridge does not yet exist. VMS, sleep, endometrial histology, P1NP, CTX, BMD, fracture, disability, and longevity are not interchangeable surrogates.

3. Scope, population, and life stage

The main inference is limited to women undergoing a natural menopausal transition followed from late reproductive STRAW+10 −3a through early −2 and late −1 transition. The indicative study age is 40–55 years, but biological stage and individual trajectory are more informative than chronological age alone.

The inferential unit is the woman. Cycles, days, samples, follicles, cells, and wells are repeated measures, not independent replicates.

The following must be distinguished:

• confirmed ovulatory, nonovulatory, and indeterminate cycles;
• integrated P4/PdG versus a single serum concentration;
• phase-specific E2/E1G, peak, median, and AUC;
• age, STRAW+10, chronological time, and time to FMP as separate anchors;
• ovarian production, systemic exposure, tissue metabolism, and excretion;
• bone, endometrium, brain, vasculature, and metabolism as separate domains.

Pregnancy, lactation, postpartum, primary ovarian insufficiency, surgical menopause, gonadotoxic treatment, hormonal exposure during the analytic window, and uncontrolled endocrine or hepatorenal disorders that dominate the trajectory are outside the main natural estimand. Pre-existing PCOS or chronic amenorrhea requires separate cohorts. No current Lua data, private records, or private genomes were used.

“Progesterone deficit” is used here only as a reproducible reduction in integrated luteal exposure relative to an intraindividual baseline. It is not a diagnosis or universal threshold. “Falls before” means that a prespecified gate or P4/PdG change-point precedes the E2/E1G change-point under the same calendar and with quantified uncertainty.

4. Background knowledge and mechanism map

4.1 Topological asymmetry

The minimal pathway can be represented as follows:

Lower follicular reserve → lower AMH/inhibin B → higher compensatory FSH → residual follicles still produce variable E2.

Additional steps are needed to obtain a high-P4 phase:

Sufficient E2 → positive KISS1/GnRH feedback → LH surge → follicular rupture → luteinization → vascularization and maintenance of the corpus luteum → P4.

In accounting notation:

Cycle P4/PdG AUC ≈ probability of ovulation × luteal days × luteal mass/state × secretion per cell × perfusion, modified by metabolism, clearance, excretion, and error.

Cycle E2/E1G AUC ≈ follicle number and quality × FSH response × aromatization + extraovarian sources, modified by clearance and error.

This topology predicts a hurdle distribution for P4: more mass near zero or low when the gate fails, plus a continuous distribution among cycles with a luteal phase. It does not necessarily predict uniform reduction of STAR, CYP11A1, or HSD3B2 in every cell.

4.2 Evidence levels by link

4.3 Why “uniform cellular failure” is not accepted

An older IVF study found lower P4 secretion and lower hCG response in granulosa-luteal cells from older women [28]. Another study found lower FSHR/CYP19A1/HSD17B with age but higher LHCGR/CYP11A1/PGR and stable STAR, together with more apoptosis, consistent with premature luteinization and identity change [10]. Neither measures isotopic flux in natural perimenopausal corpora lutea. Enzyme expression does not equal flux, and hormone decline can precede loss of RNA or protein.

5. Evidence method

This report integrates the verified artifacts from SCOPING, EVIDENCE_MAP, EVIDENCE_VERIFICATION, MECHANISTIC_SYNTHESIS, COMPUTE_DECISION, COMPUTE_OPTIONAL, HYPOTHESIS_GENERATION, ADVERSARIAL_REVIEW, and EXPERIMENT_DESIGN. Decisive statements were traced to primary PubMed/PMC or DOI sources.

The applied hierarchy was:

1. human longitudinal cohorts with daily or repeated measurement;
2. human interventions for tissue perturbability, without confusing treatment with etiology;
3. human ex-vivo/in-vitro evidence for molecular proximity;
4. animal models for possibility or necessity in that model;
5. explicitly labeled mechanistic inference.

Design, population, stage, size, matrix, exposure, comparator, outcome, effect, and limitations were extracted. Nulls and contradictions were preserved. Several SWAN publications share participants and infrastructure and were not counted as independent replications.

Final directed verification confirmed in recent primary sources:

• Shieh et al. 2026: 436 early-perimenopausal women; lower AMH and higher CTX predicted fast imminent bone loss, with apparent improvement when combined [19].
• Shieh et al. 2026: 451 women, 2,667 observations, 78 fractures over 15.1 years; a bone balance index was associated with fracture, whereas CTX and PINP alone were null [26].
• Zhong et al. 2017: Pgr in progenitors affected trabecular acquisition, but age- or gonadal-deficiency-related loss was similar in knockout and controls [21].
• Santoro et al. 2008: confirmed PdG decline and more non-ELA collections, but not a formal comparison of change-points [3].

6. Evidence map

6.1 Human endocrine trajectory

6.2 Human consequences

6.3 Cellular and animal evidence

7. Contradictory evidence and null findings

7.1 Does P4 fall early?

Santoro 2008 shows a PdG slope and more non-ELA collections [3], but Santoro 2017 shows relative stability until near FMP [4], BIMORA places the major change in late peri [5], and the Swiss study is null by stage [8]. These findings should not be averaged away: the system may show a deteriorating tail, greater mass of low cycles, and late median change. Those patterns do not define the same change-point.

7.2 Does architecture dominate?

More cycles without a luteal phase support architecture [3–5]. However, PdG also falls within ovulatory cycles, and IVF cellular capacity may decline [28]. The probability×duration×amplitude decomposition is accounting, not mediation. The attributed percentage depends on decomposition order; it should be symmetric/Shapley and use ovulation measured independently of PdG.

7.3 Does P4 cause bone loss?

Waugh found no mean luteal P4 effect [16]; Grewal was null for ovulation/luteal length after adjustment [14]; PeKnO was positive for ovulation but confounded by all hormones and null for turnover markers [17]. The mouse knockout weighs against extrapolating PGR to menopausal loss [21]. Local possibility remains; demonstrated mediation does not.

7.4 Does P4 cause VMS, insomnia, or mood changes?

SWAN associates the LH/HPO pattern with VMS, not independent P4, and preserves the null mood result [12]. The oral P4 trial did not meet the VMS primary, and the sleep-problems diary was null [13]. A subjective secondary does not demonstrate endogenous P4→brain Allo→GABA-A.

7.5 Is there a bridge to longevity?

There is no chain of integrated endogenous P4→specific mediator→disease/healthspan. Bone balance is associated with fracture [25,26], but this neither completes the route from P4 nor extends it to global longevity.

8. Multiscale mechanistic synthesis

8.1 Ovarian molecular and cellular levels

Declining reserve lowers inhibins and raises FSH. Residual follicles can maintain aromatization and variable E2. The P4 phase requires the system to complete a discrete event. Failure can lie in positive feedback/LH surge, rupture, luteinization, LH support, vascularization, duration, or cellular capacity. Human evidence does not identify the first molecular node.

The STAR–CYP11A1–HSD3B2 machinery is necessary, but there is no evidence of a uniform acquired loss. IVF data are compatible both with lower secretory response and with identity change and apoptosis [10,28].

8.2 Organ and systemic levels

The corpus luteum is a transient organ. Losing an entire event abruptly reduces the P4 integral, whereas a preceding follicle may already have produced E2. This is the strongest topological explanation for the observed asymmetry.

Urinary PdG adds metabolism and excretion. Hepatic/renal function, body composition, hormonal exposure, thyroid, prolactin, and energy availability can modify both ovulation and measurement. Matrix is therefore not a minor technical choice.

8.3 Endometrium

The most perturbable pathway is E2/ESR1→stromal FGF→epithelial FRS2/ERK→proliferation, counterregulated by P4/PGR→HAND2→lower FGF [24]. PEPI confirms that progestogenic signaling modifies histology under CEE [23]. Integrated endogenous P4/E2, tissue PGR/HAND2, and longitudinal natural histology are missing.

8.4 Brain

The P4→5α-DHP→allopregnanolone→GABA-A pathway is biochemically plausible, but every arrow requires independent measurement. Peripheral P4 is not brain Allo; Allo/P4 does not measure conversion; GABA-A does not imply inhibition without the Cl− gradient; a concentration does not represent oscillatory history or adaptation. L7-4 does not elevate this branch beyond the constraints already established in L7-2/L7-3.

8.5 Bone

The leading systemic pathway is:

reserve/HPO changes → AMH/FSH/E2/ovulation and turnover change together → formation–resorption balance deteriorates → BMD/fracture.

The experimental subpath is:

free P4 → local metabolism by AKR1C1/SRD5A1/AKR1C2 → intracellular P4 and metabolites → stage-dependent PGR → PINP/collagen/mineralization.

The sign is not fixed. In mice, early Pgr loss increased trabecular acquisition and loss in mature osteoblasts was null; loss from age/hormone deficiency was similar [21]. In human osteoblasts, metabolizing P4 does not prove changed intracellular P4 or PGR engagement [20]. Testing must proceed through gates.

8.6 Life course

The BBI→fracture association shows that a distal intermediate can be clinically relevant [26]. It does not show that P4 determines BBI, that modifying the route prevents fractures, or that doing so increases longevity. The appropriate healthspan outcome would be years free of fracture/frailty or disability, under competing risks, only after validation of the proximal chain.

9. Computational layer

The decision was to use conventional computation on BIMORA/ICPSR 4452 V2, not BioNeMo. The bundle was not obtained: official endpoints required Researcher Passport authentication and were blocked by Cloudflare; no connected visible session was available. No change-point, AUC, odds ratio, contribution, or interval was calculated. This failure is epistemic, not biological.

Adversarial review added a larger barrier: even with access, BIMORA is not confirmatory for an intraindividual lead of approximately one year. It samples six months per year, mean participation was about 21 months, only 53 women completed five years, and published curves share age-based shape [6]. It may describe distributions and heterogeneity, but must first pass an identifiability simulation.

BioNeMo was excluded because the relevant inputs are hormonal trajectories, ovulation, time, and missingness. Sequence embeddings, folding, or transcriptomic perturbation do not estimate that order or correct circularity.

10. Primary hypothesis

L7-4-AR-H1 v1 — distributive ovulatory-luteal gate observed first

Statement: during natural transition, one independently validated measure of ovulation probability and corroborative luteal duration will deteriorate before phase-aligned E2/E1G exposure; a symmetric decomposition will attribute a material fraction of PdG reduction to frequency and duration, not only to within-ovulatory-cycle amplitude.

Status: retained and narrowed.
Maturity: H1 for links; H0 for order.
Confidence: 0.38.

Evidence for: more non-ELA collections, lower PdG, and serial topology [3–6].
Evidence against: major late changes, circular ELA, incomplete sampling, residual cellular capacity, and a possible common HPO cause [4,5,8,10,28].
What it no longer claims: an independently causal luteal lesion or universal enzyme decline.

11. Competing hypothesis

The prior compound hypothesis was split because “common cause or measurement or selection” could not lose.

L7-4-AR-H2M v1 — measurement/selection lead

Statement: the apparent gate lead will shrink to equivalence or reverse sign when using ovulation independent of PdG, continuous sampling, phase-aligned hormones, serum and urine analyzed separately, explicit censored cycles, and MNAR sensitivity.

Status/maturity/confidence: leader; H1; 0.64.

Critical prediction: a naive analysis will show a larger lead than the corrected analysis.
Kill: the lead persists with ultrasound, phase, multiple matrices, MNAR, and center replication.

L7-4-AR-H2C v1 — sufficient common HPO clock

Statement: a latent state built only from AMH, inhibin B, FSH, stage, and cycle length will predict both gate and E2/E1G outside woman and center; adding a gate-specific process will not materially improve calibration or log-score.

Status/maturity/confidence: retained; H0–H1; 0.54.

Critical prediction: the gate may cross an observed threshold first without requiring a separate causal process.
Kill: the gate-specific term adds reproducible external performance under orthogonal measurement. That would demonstrate added structure, not molecular causality by itself.

12. Translational hypothesis

L7-4-AR-HT1 v1 — dynamic endocrine signature for bone change

Statement: after demonstrating metrology and a P4-specific bone contribution, a frozen signature of ovulation probability, luteal days, PdG per luteal day, and E1G will improve prediction of continuous lumbar change in an external center over covariates+AMH+CTX and over an expanded comparator with FSH/E2/ovulation, without using future FMP, outcomes, or ratios in construction.

Status: parked, purely prognostic.
Maturity: H0.
Confidence: 0.08.
Requirement: HUMAN_QA_REQUIRED.

A signature could predict without P4 falling first; endocrine priority and prognostic utility are therefore not conflated. A P4-specific interpretation first requires mechanistic gate L7-4-AR-H3A:

Mechanistic subhypothesis L7-4-AR-H3A: under physiologic free P4, perturbing the dominant prereceptor enzyme must change intracellular P4 and PGR engagement in human osteoprogenitors before changing function. Metabolites without engagement kill the P4–PGR route for that context.

13. Falsifiable predictions and kill criteria

The 9-month, ±6-month, 50%, and 20% thresholds are decision SESOIs, not biological constants. They require human QA approval and must be shown estimable by simulation.

14. Discriminating experiment

E0 — identifiability simulation

Simulate 10,000 datasets per scenario with true leads of 0, 6, 9, 12, or 18 months; analytic error, ovulation error, long cycles, continuous or BIMORA-like sampling, and MAR/MNAR dropout. The estimand is singular: δ=CP_E−CP_gate. Using the minimum of ovulation CP and luteal-days CP is prohibited because it selects early noise.

Gate: type I error 4–6%, CI95% coverage 93–97%, absolute bias <3 months, and sign recovery ≥80% for 9 months. If BIMORA-like fails, it remains descriptive. If a 180-woman/48-month design fails, it is not recruited as such.

E1A — gate calibration

Eighty women, with a minimum target of 60 ovulatory and 60 nonovulatory cycles, up to 300 cycles. Serial transvaginal ultrasound with blinded readers adjudicates rupture. The primary classifier uses LH, FSH, E1G, bleeding, and time, but not PdG. PdG measures exposure and serves as an adversarial circular comparator.

Gate: point sensitivity and specificity ≥0.90, lower CI95% bounds ≥0.80, median lag ≤2 days, and indeterminate cycles ≤15%.

E1B — continuous endocrine cohort

Provisional maximum 180 enrolled, 144 completers, 48 months, only if E0 passes. Daily urine LH/FSH/E1G/PdG; phase-aligned serum LC–MS/MS every six months; AMH/inhibin B/FSH every six months; repeated ultrasound in a subcohort; competing causes measured.

One CP_gate and CP_E are estimated, classification error propagated, descriptive Shapley applied, and common-clock versus gate-specific performance evaluated outside woman/center. Future FMP does not enter as a feature.

E2A/E2B — ex-vivo bone gate

Primary cells from 8 donors in feasibility and up to 16 in confirmation, paired as osteoprogenitor and mature osteoblast. Defined medium without inadvertent steroids; isotopic P4; measured free P4; PGR-A/B by proteomics; PGR loss/rescue; one dominant AKR1C1 or SRD5A1 enzyme with orthogonal loss and rescue.

Sequence:

1. PGR is present and signalable.
2. Flux changes intracellular P4.
3. Flux changes PGR occupancy/signature.
4. Engagement changes PINP/procollagen I.
5. Only then is stage dependence tested.

Wells are technical; donor is n. Failure of one gate stops the next.

E2C — human bone module

Only after positive E2A/B, nested within E1B: quarterly CTX and PINP, DXA at baseline/24/48 months. Primary outcome is bivariate PINP+CTX; BBI and continuous lumbar BMD are secondary. Integrated P4 must precede balance, add external performance over the HPO block, and agree with the cellular mechanism.

Smallest truly discriminating experiment

The smallest immediate step is E0. It requires no recruitment, does not presuppose BIMORA access, and can show that the central estimand is structurally unrecoverable. If E0 passes, E1A is the smallest biological experiment: without an ovulation measure independent of PdG, no decomposition of PdG can validate the gate.

15. Candidate biomarkers and stratification

No variable in this section is called a validated biomarker.

Exposure/architecture

• ovulation probability calibrated against ultrasound;
• luteal days;
• P4/PdG AUC and PdG per luteal day;
• E2/E1G AUC, peak, and phase;
• change-point difference with uncertainty.

Reserve/HPO

• AMH;
• inhibin B;
• FSH and LH;
• cycle length and variability;
• STRAW+10 as context, not mechanism.

Bone

• paired PINP and CTX;
• BBI as a candidate, not a universal standard;
• continuous lumbar/hip BMD;
• intracellular P4 and PGR engagement as experimental mechanistic markers.

Required strata

STRAW+10 stage, ovulatory/nonovulatory/indeterminate, BMI/composition, activity/energy availability, thyroid function, prolactin, hepatic/renal function, incident hormone exposure, site/batch, and self-reported race/ethnicity/ancestry. Strata address heterogeneity and transport; they do not create “progesterone subtypes.”

16. Individual variability

Perimenopause is not a uniform slope. One woman may retain high-P4 cycles between cycles without a luteal phase; another may maintain ovulation with shorter duration or lower amplitude; another may show estrogen-dominant change. A single sample confuses this architecture with “normal/low.”

Sources of variability include:

• age and reserve at entry;
• HPO-clock speed;
• stochastic probability of ovulation;
• corpus-luteum duration, mass, perfusion, and support;
• adiposity and extraovarian estrogen sources;
• energy, activity, stress, sleep, and intercurrent disease as bidirectional causes;
• thyroid, prolactin, liver, and kidney;
• matrix, phase, batch, LOD, and sampling adherence;
• differential dropout and incident hormone use;
• stage and cellular composition in the target organ.

Inference should show individual trajectories and indeterminate results, not select post-hoc “responders.” Mexico/LATAM remains a transportability question; no effect size is assumed universal.

17. Pharma relevance and maturity

Opportunity map

Pharma gates

Lead discovery, partnering, or clinical evaluation is not justified until there is:

1. target engagement and rescue in primary human cells;
2. reproducible function by donor and stage;
3. an exposure window without toxicity/adverse proliferation;
4. selectivity against the steroidome and relevant endometrial, breast, hepatic, and neural tissues;
5. coherence with the human trajectory;
6. tractability, tissue exposure, and IP;
7. human QA.

If metabolism does not change engagement, engagement does not change function, or the signal does not translate to human balance, the bone P4–PGR/enzyme opportunity closes even if the chemistry exists. There is no ready target, candidate formulation, or basis for prescribing.

18. Limitations

1. SWAN dominates the evidence; several publications are not independent.
2. ELA uses PdG and may circularly explain PdG.
3. One cycle per year or six months per year loses heterogeneity, long cycles, and continuity.
4. Observed FMP, hormone exclusion, and attrition induce selection.
5. Urine, serum, saliva, tissue, and brain are not interchangeable matrices.
6. PdG/E1G are metabolites/excretion, not ovarian production or free tissue exposure.
7. There are no paired human change-points with independent anatomic ovulation.
8. Cellular luteal evidence comes from IVF/infertility and does not separate age from stimulation or identity.
9. Bone P4–PGR evidence combines small cultures and mice; the chain does not exist in one population.
10. The 2026 BBI was associated with fracture but requires external validation and is not P4-specific.
11. The oral P4 RCT tests a pharmacologic state, not endogenous etiology.
12. PEPI is postmenopausal and uses exogenous CEE.
13. The neural branch lacks a blood–brain bridge, stereochemistry, and GABA-A function during perimenopause.
14. E0–E2 are proposed designs, not results.
15. SESOIs are operational decisions requiring simulation and QA.
16. BMD, balance, fracture, disability, healthspan, and longevity remain distinct levels.
17. Treatment, efficacy, safety, dose, and clinical utility were not studied.

19. Conclusions

Progesterone may appear to fall before estrogen because a high-P4 phase requires a transient organ and a gate that follicular estradiol does not require. Human evidence confirms heterogeneous loss of luteal activity and exposure; it does not yet confirm that the P4/PdG change-point precedes the E2/E1G change-point within woman or that the mechanism is uniform cellular failure.

The leading adversarial explanation is that part of the apparent lead is measurement or selection; the second is that a common HPO clock generates outputs with different thresholds. The distributive gate remains plausible but must be earned with independent ovulation measurement, continuous follow-up, and a single estimand.

Consequences do not form one syndrome. Endometrium offers the strongest perturbability under exogenous exposure; VMS/sleep retains important nulls; bone has a robust human bridge from turnover/balance to BMD/fracture, but not from P4. The priority bone hypothesis is no longer “P4 is anabolic,” but whether prereceptor metabolism changes intracellular P4, PGR engagement, and then functional balance in a stage-dependent manner.

The longevity claim remains H0. The concrete scientific advance is to locate the gap and design a sequence that can close it or kill it early:

E0 identifiability → E1A orthogonal ovulation → E1B order/common HPO → E2A intracellular P4/PGR → E2B function → E2C human balance → prognostic validation → Pharma assessment.

Stable claims added by REPORT_EN

• L7-4-REPORT-C1: human evidence demonstrates loss of luteal activity/exposure, not a P4 change-point preceding E2.
• L7-4-REPORT-C2: the P4–E2 asymmetry is topological before it is enzymatic, but dominant architecture remains H0.
• L7-4-REPORT-C3: one isolated P4 sample and integrated uPdG answer different questions; neither alone identifies production or tissue signaling.
• L7-4-REPORT-C4: a common HPO/reserve cause leads the bone explanation; P4–PGR remains a mechanistic gate.
• L7-4-REPORT-C5: osteoblastic P4 metabolism does not equal PGR engagement or function; the chain must be tested sequentially.
• L7-4-REPORT-C6: bone balance is associated with fracture, but no P4→balance or global-longevity bridge yet exists.
• L7-4-REPORT-C7: BIMORA is descriptive unless simulation demonstrates identifiability; microdata access does not correct a structural limitation.
• L7-4-REPORT-C8: prediction and mechanism are separate; a dynamic signature might predict without demonstrating that P4 falls first.
• L7-4-REPORT-C9: for the HR 0.75 for major osteoporotic fracture per SD of BBI, the Results section and Table 5 report CI95% 0.61–0.95, whereas the abstract reports 0.61–0.85; the inconsistency is preserved and 0.61–0.95 is prioritized as the table-supported interval.

20. References

1. Santoro N, et al. Assessing menstrual cycles with urinary hormone assays. Am J Physiol Endocrinol Metab. 2003;284:E521–E530. DOI: 10.1152/ajpendo.00381.2002. PMID: 12441312.
2. Santoro N, et al. Body size and ethnicity are associated with menstrual cycle alterations in women in the early menopausal transition: the Study of Women’s Health Across the Nation. J Clin Endocrinol Metab. 2004;89:2622–2631. DOI: 10.1210/jc.2003-031578. PMID: 15181033.
3. Santoro N, Crawford SL, Lasley WL, et al. Factors related to declining luteal function in women during the menopausal transition. J Clin Endocrinol Metab. 2008;93:1711–1721. DOI: 10.1210/jc.2007-2165. PMID: 18285413. PMCID: PMC2386686.
4. Santoro N, Crawford SL, El Khoudary SR, et al. Menstrual Cycle Hormone Changes in Women Traversing Menopause. J Clin Endocrinol Metab. 2017;102:2218–2229. DOI: 10.1210/jc.2016-4017. PMID: 28368525. PMCID: PMC5505186.
5. O’Connor KA, Brindle E, Miller RC, et al. Progesterone and ovulation across stages of the transition to menopause. Menopause. 2009;16:1178–1187. DOI: 10.1097/gme.0b013e3181aa192d. PMID: 19568209. PMCID: PMC2783957.
6. Ferrell RJ, et al. Monitoring reproductive aging in a 5-year prospective study: aggregate and individual changes in steroid hormones and menstrual cycle lengths with age. Menopause. 2005;12:567–577. DOI: 10.1097/01.gme.0000172265.40196.86. PMID: 16145311.
7. Burger HG, Dudley EC, Hopper JL, et al. Prospectively measured levels of serum follicle-stimulating hormone, estradiol, and the dimeric inhibins during the menopausal transition in a population-based cohort of women. J Clin Endocrinol Metab. 1999;84:4025–4030. DOI: 10.1210/jcem.84.11.6158. PMID: 10566644.
8. Grub J, Süss H, Willi J, Ehlert U. Steroid Hormone Secretion Over the Course of the Perimenopause: Findings From the Swiss Perimenopause Study. Front Glob Womens Health. 2021;2:774308. DOI: 10.3389/fgwh.2021.774308. PMID: 34970652. PMCID: PMC8712488.
9. Santoro N, et al. Daily luteal serum and urinary hormone profiles in the menopausal transition. Menopause. 2020;27:127–133. DOI: 10.1097/GME.0000000000001453. PMID: 31794501. PMCID: PMC7050767.
10. Wu YG, et al. Aging-related premature luteinization of granulosa cells is avoided by early oocyte retrieval. J Endocrinol. 2015;226:167–180. DOI: 10.1530/JOE-15-0246. PMID: 26264981.
11. Lederman MA, Lebesgue D, Gonzalez VV, et al. Age-related LH surge dysfunction is associated with reduced responsiveness of kisspeptin neurons to estradiol positive feedback in middle-aged rats. Neuropharmacology. 2010;58:314–320. DOI: 10.1016/j.neuropharm.2009.06.015. PMID: 19559035. PMCID: PMC2901500.
12. McConnell DS, Crawford SL, Gee NA, et al. Lowered progesterone metabolite excretion and a variable LH excretion pattern are associated with vasomotor symptoms but not negative mood in the early perimenopausal transition: Study of Women’s Health Across the Nation. Maturitas. 2021;147:26–33. DOI: 10.1016/j.maturitas.2021.03.003. PMID: 33832644. PMCID: PMC8091174.
13. Prior JC, et al. Oral micronized progesterone for perimenopausal night sweats and hot flushes: a phase III Canada-wide randomized placebo-controlled 4-month trial. Sci Rep. 2023;13:9082. DOI: 10.1038/s41598-023-35826-w. PMID: 37277418. PMCID: PMC10241804.
14. Grewal J, Sowers MFR, Randolph JF Jr, Harlow SD, Lin X. Low bone mineral density in the early menopausal transition: role for ovulatory function. J Clin Endocrinol Metab. 2006;91:3780–3785. DOI: 10.1210/jc.2006-0544. PMID: 16835283.
15. Sowers MFR, Jannausch M, McConnell D, et al. Hormone predictors of bone mineral density changes during the menopausal transition. J Clin Endocrinol Metab. 2006;91:1261–1267. DOI: 10.1210/jc.2005-1836. PMID: 16403818.
16. Waugh EJ, Polivy J, Ridout R, et al. A prospective investigation of the relations among cognitive dietary restraint, subclinical ovulatory disturbances, physical activity, and bone mass in healthy young women. Am J Clin Nutr. 2007;86:1791–1801. DOI: 10.1093/ajcn/86.5.1791. PMID: 18065600.
17. Starrach T, et al. Perimenopausal Bone Loss Is Associated with Ovulatory Activity—Results of the PeKnO Study. Diagnostics. 2022;12:305. DOI: 10.3390/diagnostics12020305. PMID: 35204396. PMCID: PMC8871419.
18. Karlamangla AS, Shieh A, Greendale GA, et al. Anti-Mullerian Hormone as Predictor of Future and Ongoing Bone Loss During the Menopause Transition. J Bone Miner Res. 2022;37:1224–1232. DOI: 10.1002/jbmr.4525. PMID: 35373854. PMCID: PMC9283201.
19. Shieh A, Karlamangla AS, Gossiel F, Eastell R, Burnett-Bowie SA, Greendale GA. Anti-Mullerian Hormone and Collagen Type I C-telopeptide Predict Fast, Imminent Bone Loss in Early Perimenopause: SWAN. J Clin Endocrinol Metab. 2026;111:e1653–e1661. DOI: 10.1210/clinem/dgaf658. PMID: 41399290.
20. Quinkler M, et al. Progesterone is Extensively Metabolized in Osteoblasts: Implications for Progesterone Action on Bone. Horm Metab Res. 2008;40:679–684. DOI: 10.1055/s-2008-1078718. PMID: 18537080.
21. Zhong ZA, Kot A, Lay YA, et al. Sex-Dependent, Osteoblast Stage-Specific Effects of Progesterone Receptor on Bone Acquisition. J Bone Miner Res. 2017;32:1841–1852. DOI: 10.1002/jbmr.3186. PMID: 28569405. PMCID: PMC5611815.
22. Liu X, Lian X, Liu X, et al. Aldo-keto reductase family 1 member C1 regulates the osteogenic differentiation of human ASCs by targeting the progesterone receptor. Stem Cell Res Ther. 2021;12:383. DOI: 10.1186/s13287-021-02425-3. PMID: 34233738. PMCID: PMC8261971.
23. PEPI Writing Group. Effects of hormone replacement therapy on endometrial histology in postmenopausal women. JAMA. 1996;275:370–375. DOI: 10.1001/jama.1996.03530290040035. PMID: 8569016.
24. Li Q, Kannan A, DeMayo FJ, et al. The antiproliferative action of progesterone in uterine epithelium is mediated by Hand2. Science. 2011;331:912–916. DOI: 10.1126/science.1197454. PMID: 21330545.
25. Cauley JA, et al. Bone resorption and fracture across the menopausal transition: the Study of Women’s Health Across the Nation. Menopause. 2012;19:1200–1207. DOI: 10.1097/gme.0b013e31825ae17e. PMID: 22850443. PMCID: PMC3483443.
26. Shieh A, Karlamangla AS, Gossiel F, Eastell R, Burnett-Bowie SA, Greendale GA. An index of net bone formation relative to resorption is associated with incident fracture during the menopause transition and postmenopause. JBMR Plus. 2026;10:ziag092. DOI: 10.1093/jbmrpl/ziag092. PMID: 42311599. PMCID: PMC13271363.
27. Winkles JF, Colvin A, El Khoudary SR, et al. Using a Composite Summary of Daily Sex Hormones to Gauge Time Until Menopause: A Focus on Pregnanediol Glucuronide (PDG). J Clin Endocrinol Metab. 2025;110:2832–2844. DOI: 10.1210/clinem/dgae895. PMID: 39847409. PMCID: PMC12448593.
28. Pellicer A, et al. Effects of aging on the human ovary: the secretion of immunoreactive alpha-inhibin and progesterone. Fertil Steril. 1994;61:663–668. DOI: 10.1016/S0015-0282(16)56643-6. PMID: 8150108.