L6-2 — MTHFR and folate: estrogen, homocysteine, and cardiovascular risk in menopause

Stage: REPORT_EN
Date: 2026-07-15
Classification: Hormonal nutrigenomics · one-carbon metabolism · female vascular biology
Epistemic status: scientific research report; not a diagnosis, medical advice, prescription, or validated biomarker.
Core population: women during the natural menopausal transition and early postmenopause.
Continuity: integrates and corrects all verified L6-2 stages and the cumulative L6-1/L5 memory. It does not use current Lua user data, PHI, PII, or private genomes.

1. Executive scientific abstract

The relationship among MTHFR, folate, homocysteine, estrogens, and female cardiovascular risk is not a linear chain. Evidence supports with high confidence that the common MTHFR C677T variant reduces enzymatic robustness and that its phenotype depends on folate and FAD/riboflavin. In older women subjected to controlled folate depletion, total homocysteine (tHcy) increased by 44% in 677TT versus 20% in CT and 15% in CC; the genotype difference disappeared after repletion [3]. In genotype-directed trials, riboflavin reduced tHcy and increased SAM mainly in 677TT [4,5]. This demonstrates a gene–cofactor interaction, not a clinical destiny.

Evidence is much weaker for the step MTHFR → menopausal cardiovascular disease. In 12,239 postmenopausal women, 677T was not associated with higher cardiovascular mortality and showed a weak inverse direction [12]. Lowering tHcy by 18.5% over 7.3 years in 5,442 high-risk women did not reduce cardiovascular events; it also did not change inflammatory/endothelial biomarkers in a substudy, and three years of folate did not change carotid intima-media thickness or arterial stiffness [9–11]. These results refute the general sufficiency of “lowering homocysteine” to prevent events in the populations and windows studied, without proving that every homocysteine or SAH signal is inert.

The final synthesis proposes three compartments and two gates. One-carbon reserve—MTHFR, folate, riboflavin, B12, and backup pathways—determines metabolic load. Liver and kidney separate concentration from flux: the kidney extracts a large fraction of circulating SAH, and in kidney disease the associations of SAH/SAM with events disappear after renal adjustment [18,19]. The menopausal transition is better supported as a gate of vascular susceptibility: ovarian suppression, estradiol rescue, and antioxidant perturbation changed FMD in humans [15], whereas age at menopause, lipids, and mitoROS converged in another human/ex-vivo cohort [16]. Neither study measured MTHFR or SAM/SAH, so they support a parallel route, not demonstrated mediation.

The reproducible weighted analysis of NHANES 2003–2006 found adjusted tHcy 12.9% higher in natural menopause than in recent menstruation (n=825; geometric ratio 1.129, CI95% 1.016–1.253). Adjustment attenuated the coefficient by 43.9%; inverse-probability weighting (IPW) reduced it to 11.1%, and estimates were heterogeneous across cycles. Poor common support by age prevented attribution of the residual to estrogen or a causal transition effect. The result calls for a within-person test, not an endocrine conclusion.

The primary hypothesis is deliberately narrow and remains at H0: at matched intracellular SAH, reduced ERα/redox signaling will superadditively increase loss of eNOS/NO in female human arterial endothelium. The best-supported competing program separates two statements: plasma SAH is dominated by renal clearance (H1), and stage does not materially amplify the SAH–vascular relationship after confounding control (H0–H1). The translational hypothesis is still methodological (H0): a single dynamic flux should be evaluated only if it first passes blinded reliability and then adds out-of-sample performance for shear-corrected FMD.

Final scientific delta: this project replaces the “MTHFR risk gene” model with a falsifiable metabolic-load × endothelial-gain model, separates renal plasma SAH from potentially effector intracellular SAH, and reduces the COMT–estrogen bridge to a quantitative flux test that can be falsified before cohort recruitment.

2. Scientific question and relevance

The question is: what mechanisms connect functional MTHFR–folate capacity with homocysteine, estrogens, and cardiovascular health during the menopausal transition, and what predictions separate causality, effect modification, and confounding?

Its relevance to women's health and longevity does not lie in labeling a genotype. It lies in understanding why the same metabolic load may have different vascular consequences across the life course. The menopausal transition coincides with changes in E2/E1, FSH, central adiposity, lipids, inflammation, ER signaling, oxidative stress, and vascular function. At the same time, age and kidney function alter tHcy and SAH. Unless these processes are separated, “postmenopause” can act as a chronological marker and “homocysteine” as a mixture of production, distribution, and clearance.

The longevity horizon is preservation of endothelial reserve and years free of cardiovascular disease. This project does not, however, demonstrate that changing folate, tHcy, SAH, or ER increases survival or healthspan. FMD is a subclinical physiological readout; it does not replace events, mortality, or longevity.

Four causal questions organize the report:

1. Does MTHFR 677T act uniformly, or does it modify sensitivity to folate/riboflavin and backup pathways?
2. Does the transition increase one-carbon load, the vascular response to that load, both, or neither?
3. Does plasma SAH reflect endothelial methylation potential or mainly hepatorenal function?
4. Does COMT consume a quantitatively sufficient fraction of SAM to connect estrogen metabolism with this pathway?

3. Scope, population, and life stage

The primary population is women aged 42–55 years in natural late premenopause, early/late perimenopause, and early postmenopause, ideally classified with STRAW+10, bleeding pattern, and final menstrual period. Late postmenopause informs accumulation and selection but is not the initial causal contrast.

The report distinguishes:

• MTHFR rs1801133 (C677T; p.Ala222Val) and rs1801131 (A1298C), without combining them as “MTHFR positive”;
• plasma folate, red-blood-cell (RBC) folate, 5-methyl-THF, and exposure form, because these matrices are not equivalent;
• B12, functional riboflavin status, PLP/B6, and—when flux is measured—betaine/DMG;
• tHcy, methionine, SAM, SAH, and the SAM/SAH ratio in plasma and cellular compartments;
• E2, E1, FSH, progesterone, SHBG, and exogenous hormone exposure by molecule and route;
• creatinine+cystatin C, albumin, adiposity, blood pressure, lipids, glycemia, smoking, alcohol, physical activity, and inflammation;
• absolute and percentage FMD, shear stimulus, baseline diameter, and NMD.

Severe MTHFR deficiency, homocystinurias, pregnancy, neural-tube defects, fertility, and cancer are excluded as primary questions. The catechol-estrogen branch is retained as an auxiliary mechanism for continuity with L6-1, not as an oncologic risk claim. Effect sizes from White US or Chinese populations are not extrapolated to Mexican/LATAM women without replication; ancestry, fortification, and diet can modify penetrance and transportability.

4. Background and mechanistic map

4.1 One-carbon metabolism

MTHFR converts 5,10-methylene-THF to 5-methyl-THF. This molecule supports B12-dependent remethylation of homocysteine to methionine by MTR; methionine feeds SAM, and every methyltransferase reaction produces SAH. AHCY connects SAH with homocysteine and adenosine through a reversible reaction. CBS/CTH route homocysteine into transsulfuration, cysteine, and glutathione. BHMT provides a betaine-dependent remethylation route, mainly in liver and kidney, that does not require 5-methyl-THF.

The 677T variant makes MTHFR more thermolabile and FAD-sensitive [1]. A1298C reduces enzymatic activity, but its signal on tHcy is less consistent [2]. The phenotype is therefore a gene–environment slope: 677TT is expressed more clearly when folate or riboflavin is limiting and can be attenuated by cofactor sufficiency [3–5].

4.2 Three-compartment architecture

ONE-CARBON RESERVE
MTHFR + folate + FAD/riboflavin + B12
        ↓
5-methyl-THF → MTR → methionine → SAM → SAH → homocysteine
        ↑                         ↓
BHMT/betaine (liver-kidney)      CBS/CTH → cysteine/GSH

HEPATORENAL COMPARTMENT
liver: production/export of metabolites
kidney: SAH extraction and contribution to tHcy
        ↓
plasma SAM/SAH ≠ demonstrated endothelial methylation potential

MENOPAUSAL VASCULAR GATE
fluctuation/loss of ovarian signaling + adiposity/lipids/inflammation
        ↓
ER/eNOS, BH4/BH2, mitoROS, and NO bioavailability
        ↓
FMD/vascular tone/remodeling

HYPOTHETICAL CONVERGENCE
intracellular SAH/homocysteine × low ER/redox reserve
        ↓
greater eNOS/NO loss at the same metabolic dose

4.3 Auxiliary catechol-estrogen branch

E1/E2 can be converted to catechol estrogens; COMT transfers a methyl group from SAM and generates methoxy-estrogen + SAH. The chemistry is direct in recombinant systems and ex-vivo human tissue [26–28]. Its systemic relevance is uncertain because it depends on tissue, absolute substrate load, actual COMT activity, and flux magnitude relative to total SAM turnover.

The auxiliary hypothesis L6-2-AR-H3 v2 is not one of the three canonical report hypotheses. Its prior balance question is: does catechol-estrogen O-methylation move SAM/SAH beyond technical and biological variation? If it produces labeled metabolite without changing the total pool, the “COMT methyl sink” is refuted even if a genetic association persists.

5. Evidence method

This report integrates the SCOPING, EVIDENCE_MAP, EVIDENCE_VERIFICATION, MECHANISTIC_SYNTHESIS, COMPUTE_DECISION, COMPUTE_OPTIONAL, HYPOTHESIS_GENERATION, ADVERSARIAL_REVIEW, and EXPERIMENT_DESIGN artifacts. The complete memory packet was reread, and recent/decisive sources were verified against primary PubMed and DOI records, primary text, or official documentation. The search did not identify a study that simultaneously measures MTHFR, cofactors, SAM/SAH, catechol-/methoxy-estrogens, kidney function, and FMD across the transition.

Evidence was classified as human interventional, human longitudinal, human cross-sectional, human genetic/prospective, ex vivo, in vitro, animal, computational, or inferred. A link is called direct only for the relationship measured. For example, COMT using SAM is direct; MTHFR limiting COMT in menopausal endothelium is inferred.

Null results and unexpected directions were preserved. Sources with insufficient design, small samples, mixed compartments, or multiple correlations were downgraded. Metabolite reduction, FMD improvement, and event prevention were not treated as equivalent.

6. Evidence map

Reading by evidence type

• Human interventional: demonstrates gene–cofactor penetrance and a hormonal/redox route to FMD, but never in the same experiment.
• Human longitudinal: 953 women showed changes in 18 metabolites over a mean 5.0 years; decisive MTHFR–SAM/SAH–FMD nodes were not measured jointly [17].
• Human genetic/prospective: the 677TT stroke association in a low-folate population prevents universalization of null findings from fortified settings [20].
• In vitro/ex vivo: SAH/homocysteine can affect methylation and NO [23–25], but HUVEC, porcine aorta, and AHCY perturbations do not alone represent menopausal arterial endothelium.
• Computational: NHANES quantifies a residual association but also exposes heterogeneity, missingness, and poor common support.
• Inferred: the dual gate and the COMT sink are hypotheses, not observations.

7. Contradictory evidence and null findings

7.1 Biochemical genotype without event

The penetrance of 677TT on tHcy under limiting cofactors is well supported [3–5], but it did not translate into higher cardiovascular mortality in a postmenopausal cohort [12]. Fortification, late outcome measurement, survival, pleiotropy, or insufficient causal importance of tHcy could explain the discrepancy. The association with stroke—especially hemorrhagic stroke—in a low-folate Chinese population [20] prevents the conclusion that “MTHFR does not matter for CVD”; nutritional exposure, population, and outcome must be specified.

7.2 Lowering tHcy does not necessarily lower disease

WAFACS and FACIT are central adversarial evidence [9–11]. Lowering tHcy did not change events, vascular inflammation, CIMT, or stiffness. Competing explanations are that tHcy is a marker; intracellular SAH or another species is the omitted effector; intervention occurred too late; or cardiometabolic risk and the vascular gate dominate. None permits use of tHcy as a surrogate outcome for CVD.

7.3 Menopause and tHcy do not follow a simple function

Some studies find higher tHcy or an inverse E2 association; in others the difference disappears after age adjustment [7]. E2 can lower tHcy without changing flux [8]. In NHANES, cofactors attenuated the coefficient more than eGFR/albumin, but the signal differed between cycles. The endocrine explanation ranks below confounding/classification because E2/FSH and longitudinal data were absent.

7.4 FMD is not pure NO

Moreau et al. support an E2–redox–FMD path, but nitroglycerin dilation also declined after suppression in the measured subset [15]. In 100 age-matched, low-risk women, FMD did not differ significantly by menopause whereas NMD did [22]. Repeated FMD also has material variability; diameter, shear, time to peak, and NMD are therefore mandatory controls.

7.5 Plasma versus cell

Plasma SAH is strongly influenced by kidney function [18,19], but the plasma SAM/SAH ratio correlated with the lymphocyte ratio in adult women [23]. Neither identity nor decoupling can be assumed. The shared variance across plasma, PBMC, and endothelium—and which compartment retains vascular signal after eGFR—must be measured.

7.6 COMT: real reaction, unknown magnitude

COMT uses SAM and produces SAH [26–28]. Yet rs4680 explained about one fifth of hepatic activity variance [30], and the COMT×MTHFR association with tHcy did not measure flux [29]. A stoichiometrically valid reaction can be quantitatively trivial for the total pool. This is the principal threat to the bridge inherited from L6-1.

8. Multiscale mechanistic synthesis

8.1 Gene and enzyme

MTHFR 677T decreases robustness; it does not create a binary state. The effect emerges with limiting folate/FAD; A1298C must be analyzed separately. Fortification and folate form can reduce penetrance. The plasma/RBC discordance in WHI shows that “low folate” without a matrix is scientifically defective [6].

8.2 Metabolites and flux

Fasting tHcy integrates production, MTR/BHMT remethylation, transsulfuration, binding, volume, and clearance. Plasma SAM and SAH also integrate organs. A concentration does not identify flux direction. The isotopic E2 RCT directly showed that tHcy can change without altered remethylation or transmethylation [8].

BHMT is a plausible buffer: it can preserve methionine/SAM when MTHFR is limiting. Longitudinal changes in betaine/cystine/taurine around menopause are compatible with remodeling but do not demonstrate higher BHMT flux [17]. The prediction remains that the 677TT phenotype may emerge when BHMT or renal compensation is insufficient. It requires a tracer, not static betaine.

8.3 Liver and kidney

The liver can export SAH; the kidney is a major disposal site [18]. High plasma SAH can therefore indicate production, reduced clearance, or both. In CKD, Hcy, SAH, and SAM did not predict MACE independently after adjustment for age, sex, and kidney function [19]. This is strong evidence for renal influence on the plasma compartment, not for absence of intracellular action in women with normal kidney function.

8.4 Endothelium and smooth muscle

Homocysteine can inhibit DDAH, raise ADMA, and reduce NOS in cellular/ex-vivo systems [24]. Pharmacologic SAH accumulation in HUVEC changed methylation and cellular programs [25]. E2 activates eNOS through ER/PI3K–Akt in human endothelium [32]. The plausible convergence is that lower ER/BH4 signaling increases gain of SAH damage; this has not been measured at matched intracellular SAH.

FMD reflects endothelium, shear stimulus, autonomic tone, baseline diameter, and smooth muscle. Therefore, an interaction only in FMD% without a proximal eNOS/NO readout, shear, and NMD will not confirm the hypothesis.

8.5 Endocrine context, adipose tissue, and lipids

The transition is not reducible to E2. FSH, progesterone, SHBG, adiposity, local aromatization, lipids, and inflammation change on different timelines. Darvish et al. linked age at menopause, FMD, MitoQ response, serum, TG(16:0), and mitoROS [16]; the human design was observational and perturbations were ex vivo, so a lipid pathway can explain part of the signal without MTHFR or SAH.

8.6 From mechanism to longevity

Lower endothelial reserve over decades is a plausible path to vascular disease. This project does not demonstrate that the proposed interaction persists, causes plaque, stroke, IHD, or mortality. Null trials require separation of metabolic target engagement from clinical benefit. Longevity maturity remains inferred/H0.

9. Computational layer

9.1 Design

Public NHANES 2003–2004 and 2005–2006 microdata were analyzed using combined MEC weights, strata, and PSUs. Women aged 40–65 years with recent menstruation were compared with a conservative natural-menopause group, excluding pregnancy/lactation, hysterectomy, bilateral oophorectomy, and identified current hormone use. The outcome was log(tHcy).

Sequential models added flexible age/cycle/race-ethnicity; eGFR and albumin; serum and RBC folate, B12, and PLP; and cardiometabolic factors. Analysis and independent validation are documented in stages/compute.md.

9.2 Results

The coefficient attenuated by 43.9%. The renal/albumin block contributed 7.6 percentage points of attenuation; cofactors contributed another 25.7; and the cardiometabolic block another 10.5. These are not proportions of causal mediation.

The estimate was 1.201 (1.053–1.369) in 2003–2004 and 1.067 (0.890–1.280) in 2005–2006. There were no recently menstruating women aged 60–65 and only nine aged 55–59 in the analytic cohort. Only ages 50–54 had approximately symmetric support.

9.3 Interpretation

The result met neither the prespecified rule for strengthening a fully renal/nutritional explanation nor the opposing rule for elevating an endocrine explanation. It is intermediate and heterogeneous. The next minimum contrast is within-person change in tHcy across stage, adjusted for renal and cofactor change. NHANES lacks MTHFR, SAM/SAH, functional riboflavin, harmonizable E2/FSH, FMD, and longitudinality; it does not update the maturity of the dual-gate model.

10. Primary hypothesis

L6-2-AR-H1 v2 — Endothelial SAH×ER/redox gain

Falsifiable statement: at a matched intracellular SAH concentration within the range observed in human cells, reducing ERα signaling or BH4/redox reserve will superadditively increase loss of eNOS/NO activity in female human arterial endothelium; the interaction must reproduce with two orthogonal SAH perturbations.

Lineage: L6-2-SCOPE-H1 → L6-2-EVM-H1 v1–v3 → L6-2-MECH-H6 → L6-2-HG-H1 → L6-2-AR-H1 v2.

Proposed mechanism: intracellular SAH inhibits methyltransferases and alters redox programs; lower ERα/BH4 signaling reduces eNOS/NO reserve. The interaction—not the main effects—would produce greater NO loss at a fixed metabolic dose. Human translation predicts a different cellular-SAH–FMD slope across the transition even if plasma tHcy is stable.

Predictions: (1) the SAH×ERα term exceeds the sum of main effects on L-NAME-dependent eNOS; (2) ERα or BH4/redox rescue reduces the interaction without changing SAH; (3) the effect appears in NO/eNOS, not only ROS/viability; (4) plasma SAH relates more strongly to eGFR and the cellular proxy more strongly to FMD; (5) MTHFR×cofactor predicts metabolic load, not FMD directly.

Evidence for: separate human gene–cofactor [3–5] and E2/redox–FMD links [15,16]; cellular Hcy/SAH–NO/methylation mechanisms [24,25,32].

Evidence against: no joint measurement; NMD changes; noisy FMD; null event trials; no higher postmenopausal mortality with MTHFR; PBMC is not endothelium.

Kill criteria: an interaction equivalent to zero with two perturbations; a signal explained by adenosine/homocysteine, ATP, toxicity, or viability; an additive-only effect; or redox rescue improving NO without changing SAH sensitivity.

Status/maturity: weakened/reformulated, H0, operational confidence 0.30. Cellular confirmation could raise the cellular nucleus to H4, not the menopause–CVD chain.

11. Competing hypothesis

The competing program remains a single contrary program with two subclaims that must not inherit strength from one another.

L6-2-AR-H2R v1 — Renal capture of SAH

Statement: eGFR from creatinine+cystatin C will explain most longitudinal plasma-SAH variation in peri/postmenopausal women; cellular SAH will add little independent vascular signal.

Predictions: eGFR change accompanies or precedes SAH change; renal adjustment reduces plasma SAH–FMD; plasma and PBMC may correlate without PBMC predicting FMD.

Kill criteria: plasma or cellular SAH retains a reproducible longitudinal association with FMD after precise kidney-function measurement, and a perturbation changes SAH/FMD without changing kidney function.

Status/maturity: strengthened, H1 for plasma/kidney; H0 for absence of cellular signal; confidence 0.55.

L6-2-AR-H2S v1 — No amplification by stage

Statement: within the same woman and under comparable cofactor conditions, crossing stage will not materially modify the slope between cellular SAH and vascular function; age, adiposity, pressure, lipids, and kidney function will explain the association.

Predictions: the SAH×stage interaction falls within an equivalence region; NMD and FMD change consistently with a nonspecific vascular effect; MTHFR×stage is null although MTHFR×cofactor persists for metabolites.

Kill criteria: a reproducible cellular-SAH×stage-change interaction that is specific to FMD/NO, independent of kidney function, and responsive to orthogonal rescue.

Status/maturity: proposed/respecified, H0–H1, confidence 0.45.

Ranking: H2 is currently more probable than H1 by parsimony and renal evidence, but it is not demonstrated. The MTHFR–stroke association under low folate [20] constrains any totalizing “everything is confounding” account.

12. Translational hypothesis

L6-2-AR-HT1 v2 — Feasibility and incremental utility of one flux

Falsifiable statement: after passing prespecified reliability, one primary one-carbon flux will add out-of-sample performance for absolute shear-corrected FMD beyond a clinical+renal+concentration model.

Lineage: L6-2-SCOPE-HT → L6-2-EVM-HT v1–v3 → L6-2-HG-HT1 → L6-2-AR-HT1 v2.

Utility mechanism: static concentrations mix production and clearance; a reproducible flux may better approximate functional exposure. This does not make flux a clinical biomarker.

Predictions: the selected flux has acceptable ICC/CV; improves error/calibration in nested validation; the gain persists when blocking on eGFR and batch; and replicates externally.

Evidence for: tHcy can decouple from flux [8]; plasma/RBC folate diverges [6]; SAH depends on kidney function [18,19].

Evidence against: complexity, cost, renal leakage, FMD variability, and no validation.

Kill criteria: inadequate ICC/CV; batch drift; improvement only in-sample; gain eliminated by eGFR; or an irreproducible protocol.

Status/maturity: proposed method, H0. It is not a biomarker, companion diagnostic, or clinical tool. HUMAN_QA_REQUIRED.

13. Falsifiable predictions and kill criteria

The most uncertain link is the intracellular SAH×ER/redox interaction. Explanations rank as: (1) renal/temporal/compartment signal; (2) no stage amplification; (3) endothelial gain; (4) COMT sink. The minimum cellular experiment can eliminate the third before a cohort; the COMT balance can eliminate the fourth in parallel.

14. Discriminating experiment

14.1 Step A — Cellular nucleus of H1

Model: primary human coronary or aortic arterial endothelium from female donors; donor is the biological unit. HUVEC may serve as a bridge, not the sole confirmatory model.

Factorial: measured basal/high intracellular SAH × intact/reduced ERα. Two SAH perturbations: pharmacologic AHCY inhibition and CRISPRi/siRNA with rescue. Calibration must match achieved SAH and measure adenosine, homocysteine, ATP, viability, and apoptosis.

Primary endpoint: L-NAME-dependent eNOS activity, ideally isotopic arginine→citrulline conversion, with an orthogonal NO confirmation. BH4/BH2, eNOS phosphorylation/dimerization, mitoROS, and methylation are hierarchical secondary endpoints.

Size: calibration in six donors; planned confirmation in 24, with at least 21 passing QC. The prospective d_z=0.65 threshold is a prioritization rule, not an observed effect.

Stop: interaction equivalent to zero with both methods, toxic signal, or disappearance after matching SAH. Go: superadditive, eNOS/NO-specific, reproduced, rescuable interaction.

14.2 Step B — COMT budget

Use primary human hepatocytes/a validated hepatic model with intact/reduced COMT, low/high catechol-estrogen load in a justified range, and [methyl-13C]methionine. Measure labeled and total SAM/SAH, methoxy-product, absolute catechol, and mass balance.

The pragmatic gate is a COMT fraction whose CI95% is not below 5% of total flux plus a pool change above the minimum detectable change. If it fails, stop the branch before introducing MTHFR or menopause.

14.3 Step C — Human reproducibility

Thirty women in a stable STRAW+10 stage, with two visits 2–4 weeks apart at the same time of day and under standardized preanalytics. Gates: shear-corrected FMD ICC ≥0.75 and CV ≤12%; LC-MS/MS analytic CV ≤10%; plasma/PBMC SAH ICC ≥0.60; leukocyte composition must not dominate PBMC SAH. If a gate fails, correct the method or stop; increasing n does not repair an unstable assay.

14.4 Step D — Within-person longitudinal study

Only if A and C pass: women aged 42–55 years in late premenopause/early perimenopause, with three visits over 18–24 months. Provisional evaluable target: 177 (118 stage changers, 59 calendar controls); with 15% attrition, approximately 209 recruited/evaluable according to blinded variance re-estimation. The number is not final before the pilot.

Primary endpoint: within-person change in absolute shear-adjusted FMD. Primary interaction: SAH_PBMC×stage change on FMD change, with eGFR, cofactors, diameter, shear, pressure, adiposity, and time prespecified. NMD controls smooth muscle. MTHFR×cofactor is tested on metabolic load, not directly on FMD.

Smallest discriminating design: before the complete chain, two measurements within the same woman can determine whether Δstage retains an association with ΔtHcy/ΔSAH after ΔeGFR/cofactors. If it does not, the endocrine metabolic explanation weakens even if cross-sectional comparisons remain positive.

15. Biomarkers and stratification

There are no newly validated biomarkers. The following are candidate research measures:

Minimum stratification must preserve stage, age, kidney function, cofactors, fortification, hormone exposure, adiposity, pressure, lipids, smoking, and ancestry. Genotype does not replace biochemical phenotype, and biochemical phenotype does not replace vascular outcome.

16. Individual variability

Variability is expected along at least eight axes:

1. Genetic: 677TT and A1298C have different effects; rs4680 explains only part of COMT activity.
2. Nutritional: folate, riboflavin, B12, B6, and betaine condition penetrance; fortification changes the population background.
3. Renal/hepatic: clearance and production alter plasma without necessarily revealing the cell.
4. Endocrine: fluctuating perimenopause, early postmenopause, and late postmenopause are not equivalent; adipose E1 and exogenous exposures alter load.
5. Tissue: plasma, RBC, PBMC, liver, kidney, endothelium, breast, and urine have different pools.
6. Cardiometabolic: adiposity, lipids, pressure, glycemia, and inflammation can dominate FMD.
7. Population: ancestry, linkage disequilibrium, diet, and fortification change frequency and expression; Mexican/LATAM calibration is insufficient.
8. Analytic/temporal: FMD, SAM/SAH, and estrogen metabolites have preanalytic, batch, and within-person variability.

This variability does not justify premature clinical personalization. It justifies stratified designs, repeated measures, and equivalence regions.

17. Pharma relevance and maturity

The immediate opportunity is target validation, not partnering or clinical development. There is no basis for dose selection, cardiovascular prevention, combination therapy, a companion diagnostic, or an event trial. If AR-H1 dies, the combined one-carbon/endothelial intervention thesis disappears. If AR-H3 dies, COMT ceases to be a target in this chain.

Overall portfolio maturity: H0–H1. No hypothesis is H5. Any partnering assessment requires HUMAN_QA_REQUIRED and later evidence of selectivity, reproducibility, therapeutic window, safety, and a relevant outcome.

18. Limitations

1. No human study jointly measures MTHFR, cofactors, SAM/SAH, kidney function, hormones, catechol estrogens, and FMD across the transition.
2. The best gene–cofactor data do not adequately stratify by stage; the best vascular data do not measure one-carbon metabolism.
3. Much cellular SAH evidence comes from HUVEC and pharmacologic AHCY perturbation.
4. Plasma and PBMC do not validate the endothelial compartment; correlation across compartments is not equivalence.
5. FMD has test–retest variability and is not a clinical event; NMD can also change with hormonal suppression.
6. NHANES is cross-sectional, fortified, bleeding-classified, incomplete, and has poor age support.
7. Vitamin trials test specific regimens and windows; they do not exclude early injury or tHcy-independent mechanisms.
8. CKB includes both sexes and does not demonstrate menopausal specificity; its effect may depend on folate and stroke type.
9. COMT evidence comes from placenta, a tumor line, recombinant systems, ex-vivo liver, and a small young cohort; it does not demonstrate menopausal vascular biology.
10. Cohorts are predominantly non-LATAM; allele frequencies and penetrance do not transport automatically to Mexico.
11. Experimental effect, ICC, and 5% flux thresholds are prospective R&D decision rules, not physiological constants or clinical thresholds.
12. Cultured endothelium loses part of the donor's hormonal, epigenetic, and hemodynamic history.
13. The longevity branch is inferred; there is no evidence that changing these nodes increases CVD-free years or survival.
14. This report does not assess individual safety or efficacy of folates, vitamins, hormones, or drugs and does not authorize prescribing.

19. Conclusions

1. MTHFR C677T is a modifier of cofactor sensitivity, not a switch or diagnosis. Its biochemical effect is reproducible; its menopausal cardiovascular effect is unproven.
2. Folate must be measured by matrix and species. Plasma/RBC discordance shows that a single label can reverse interpretation.
3. tHcy is neither flux nor a causal surrogate for events. Its reduction was insufficient to change CVD, vascular inflammation, or structure in decisive trials.
4. Plasma SAH is strongly conditioned by kidney function. Possible intracellular SAH action must be tested separately and by direct measurement.
5. The menopausal transition is better supported as a change in redox/NO vascular vulnerability than as a demonstrated amplifier of homocysteine or MTHFR.
6. The most informative primary hypothesis is an intracellular SAH×ERα/redox interaction on eNOS/NO. It is H0 and can die in a small orthogonal cellular factorial.
7. The renal/temporal competing explanation remains first by parsimony but cannot be universalized: under low folate, 677TT is associated with stroke and not IHD.
8. The COMT bridge should continue only if isotopic balance demonstrates sufficient magnitude. Chemistry without a flux budget is not systemic causality.
9. Reliability gates must precede a longitudinal cohort. Only then can a within-person design separate load, stage, kidney, and vascular function.
10. Pharma relevance remains target validation; there is no clinical readiness, validated biomarker, or longevity claim.

The project's generative contribution is an architecture that can be falsified in parts. If cellular H1 fails, the endothelial interaction is removed. If kidney function explains SAH and SAH does not predict FMD, H2 wins. If COMT does not move the pool, that branch closes. If human measures are not reproducible, the panel stops. This capacity to kill explanations is the main advance over the initial narrative.

20. References

1. Frosst P, et al. A candidate genetic risk factor for vascular disease: a common mutation in methylenetetrahydrofolate reductase. Nature Genetics. 1995. DOI: 10.1038/ng0595-111. PMID: 7647779.
2. Chango A, et al. The effect of 677C→T and 1298A→C mutations on plasma homocysteine and 5,10-methylenetetrahydrofolate reductase activity in healthy subjects. British Journal of Nutrition. 2000. DOI: 10.1017/S0007114500000751. PMID: 10911766.
3. Kauwell GPA, et al. Methylenetetrahydrofolate reductase mutation (677C→T) negatively influences plasma homocysteine response to marginal folate intake in elderly women. Metabolism. 2000. DOI: 10.1053/meta.2000.16555. PMID: 11092508.
4. McNulty H, et al. Riboflavin lowers homocysteine in individuals homozygous for the MTHFR 677C→T polymorphism. Circulation. 2006. DOI: 10.1161/CIRCULATIONAHA.105.580332. PMID: 16380544.
5. Rooney M, et al. Impact of the MTHFR C677T polymorphism on one-carbon metabolites: evidence from a randomised trial of riboflavin supplementation. Biochimie. 2020. DOI: 10.1016/j.biochi.2020.04.004. PMID: 32330571.
6. Cheng TYD, et al. Associations between genetic variants and blood biomarkers of one-carbon metabolism in postmenopausal women from the Women's Health Initiative Observational Study. Journal of Nutrition. 2022. DOI: 10.1093/jn/nxab444. PMID: 34967850.
7. Bruschi F, et al. Age, menopausal status and homocysteine levels in women around menopause. European Journal of Obstetrics & Gynecology and Reproductive Biology. 2005. DOI: 10.1016/j.ejogrb.2004.11.016. PMID: 15925051.
8. Smolders RGV, et al. Oral estradiol decreases plasma homocysteine, vitamin B6, and albumin in postmenopausal women but does not change the whole-body homocysteine remethylation and transmethylation flux. Journal of Clinical Endocrinology & Metabolism. 2005. DOI: 10.1210/jc.2004-1021. PMID: 15671107.
9. Albert CM, et al. Effect of folic acid and B vitamins on risk of cardiovascular events and total mortality among women at high risk for cardiovascular disease: a randomized trial. JAMA. 2008. DOI: 10.1001/jama.299.17.2027. PMID: 18460663.
10. Christen WG, et al. Effect of combined treatment with folic acid, vitamin B6, and vitamin B12 on plasma biomarkers of inflammation and endothelial dysfunction in women. Journal of the American Heart Association. 2018. DOI: 10.1161/JAHA.117.008517. PMID: 29776960.
11. Durga J, et al. Effect of 3 y of folic acid supplementation on the progression of carotid intima-media thickness and carotid arterial stiffness in older adults. American Journal of Clinical Nutrition. 2011. DOI: 10.3945/ajcn.110.006429. PMID: 21430116.
12. Roest M, et al. Methylenetetrahydrofolate reductase 677 C/T genotype and cardiovascular disease mortality in postmenopausal women. American Journal of Epidemiology. 2001. DOI: 10.1093/aje/153.7.673. PMID: 11282795.
13. Spijkerman AMW, et al. S-adenosylmethionine and 5-methyltetrahydrofolate are associated with endothelial function after controlling for confounding by homocysteine: the Hoorn Study. Arteriosclerosis, Thrombosis, and Vascular Biology. 2005. DOI: 10.1161/01.ATV.0000157981.57694.d2. PMID: 15692102.
14. Zawada AM, et al. S-adenosylhomocysteine is associated with subclinical atherosclerosis and renal function in a cardiovascular low-risk population. Atherosclerosis. 2014. DOI: 10.1016/j.atherosclerosis.2014.02.002. PMID: 24589563.
15. Moreau KL, et al. Decline in endothelial function across the menopause transition in healthy women is related to decreased estradiol and increased oxidative stress. GeroScience. 2020. DOI: 10.1007/s11357-020-00236-7. PMID: 32770384.
16. Darvish S, et al. Preservation of vascular endothelial function in late-onset postmenopausal women. Circulation Research. 2025;136:455–469. DOI: 10.1161/CIRCRESAHA.124.325639. PMID: 39886766.
17. Miyake A, et al. Longitudinal study of plasma metabolites during menopause and their associations with later onset of metabolic syndrome. Journal of Clinical Endocrinology & Metabolism. 2026;111:e1672–e1680. DOI: 10.1210/clinem/dgaf666. PMID: 41389027.
18. Garibotto G, et al. The kidney is the major site of S-adenosylhomocysteine disposal in humans. Kidney International. 2009. DOI: 10.1038/ki.2009.117. PMID: 19357721.
19. Emrich IE, et al. Association of homocysteine, S-adenosylhomocysteine and S-adenosylmethionine with cardiovascular events in chronic kidney disease. Nutrients. 2025;17:626. DOI: 10.3390/nu17040626. PMID: 40004955.
20. Bennett DA, et al. MTHFR and risk of stroke and heart disease in a low-folate population: a prospective study of 156,000 Chinese adults. International Journal of Epidemiology. 2023;52:1862–1869. DOI: 10.1093/ije/dyad147. PMID: 37898918.
21. Keller AC, et al. Elevated plasma homocysteine and cysteine are associated with endothelial dysfunction across menopausal stages in healthy women. Journal of Applied Physiology. 2019;126:1533–1540. DOI: 10.1152/japplphysiol.00819.2018. PMID: 30896357.
22. Sharad P, et al. Assessment of endothelial dysfunction by brachial artery flow-mediated dilatation in postmenopausal women at low risk for cardiovascular disease. Journal of Mid-life Health. 2021;12:193–198. DOI: 10.4103/jmh.jmh_111_21. PMID: 34759700.
23. Melnyk S, et al. Measurement of plasma and intracellular S-adenosylmethionine and S-adenosylhomocysteine utilizing coulometric electrochemical detection: alterations with plasma homocysteine and pyridoxal 5′-phosphate concentrations. Clinical Chemistry. 2000;46:265–272. PMID: 10657384.
24. Stühlinger MC, et al. Homocysteine impairs the nitric oxide synthase pathway: role of asymmetric dimethylarginine. Circulation. 2001. DOI: 10.1161/hc4601.098514. PMID: 11714652.
25. Castro R, et al. Intracellular S-adenosylhomocysteine increased levels are associated with DNA hypomethylation in HUVEC. Journal of Molecular Medicine. 2005. DOI: 10.1007/s00109-005-0679-8. PMID: 15976919; and Yu X, et al. Relationship of impairment induced by intracellular S-adenosylhomocysteine accumulation with DNA methylation in HUVEC treated with 3-deazaadenosine. International Journal of Experimental Pathology. 2009. DOI: 10.1111/j.1365-2613.2009.00687.x. PMID: 19958400.
26. Dawling S, et al. In vitro model of mammary estrogen metabolism: structural and kinetic differences between catechol estrogens 2- and 4-hydroxyestradiol. Chemical Research in Toxicology. 2004. DOI: 10.1021/tx0498657. PMID: 15377160.
27. Jiang H, et al. Human catechol-O-methyltransferase down-regulation by estradiol. Neuropharmacology. 2003. DOI: 10.1016/S0028-3908(03)00286-7. PMID: 14573393.
28. Zhu BT, et al. O-methylation of catechol estrogens by human placental catechol-O-methyltransferase: interindividual differences in sensitivity to heat inactivation and to inhibition by dietary polyphenols. Drug Metabolism and Disposition. 2010. DOI: 10.1124/dmd.110.033548. PMID: 20606002.
29. Tunbridge EM, et al. The role of catechol-O-methyltransferase polymorphisms in modulating homocysteine levels. American Journal of Medical Genetics Part B. 2008. DOI: 10.1002/ajmg.b.30700.
30. Zhang J, et al. Catechol O-methyltransferase pharmacogenomics: human liver genotype-phenotype correlation and proximal promoter studies. Pharmacogenetics and Genomics. 2009. DOI: 10.1097/FPC.0b013e32832c15c6. PMID: 19641441.
31. Venter G, et al. Oral contraceptives containing ethinyl estradiol and drospirenone increase hydroxylation and methylation of endogenous estrogen but not genotoxic estrogen DNA-adduct formation. Scientific Reports. 2025. DOI: 10.1038/s41598-025-16892-8. PMID: 40858744.
32. Haynes MP, et al. Membrane estrogen receptor engagement activates endothelial nitric oxide synthase via the PI3-kinase–Akt pathway in human endothelial cells. Circulation Research. 2000. DOI: 10.1161/01.RES.87.8.677. PMID: 11029403.
33. van Mil ACCM, et al. Impact of volunteer-related and methodology-related factors on the reproducibility of brachial artery flow-mediated vasodilation: analysis of 672 individual repeated measurements. Journal of Hypertension. 2016;34:1738–1745. DOI: 10.1097/HJH.0000000000001012.

Internal computational provenance

Lua Labs. L6-2 — Reproducible computation with NHANES 2003–2006. Artifact memoria agentes/Lua Labs/autonomous/projects/L6-2/stages/compute.md, executed and validated on 2026-07-14 with public CDC/NCHS microdata, input hashes, software versions, commands, and prespecified decision rules.