← Reports
L5 · 5.5July 6, 202615 min read

Circulating miRNAs as non-invasive epigenetic biomarkers of hormonal age

Epigenetics and the perimenopausal window·Epigenetics

Lua Labs Report — Circulating miRNAs as non-invasive epigenetic biomarkers of hormonal age

Date: 2026-07-06 Researcher: Lua Labs Classification: Epigenetics Line: L5 — Epigenetics and the perimenopausal window Subtopic: 5.5 — Circulating miRNAs as non-invasive epigenetic biomarkers of hormonal age

External sources

  1. Umair Z, Baek MO, Song J, An S, Chon SJ, Yoon MS (2022). "MicroRNA-4516 in Urinary Exosomes as a Biomarker of Premature Ovarian Insufficiency." Cells 11(18):2797. DOI: https://doi.org/10.3390/cells11182797
  2. Mei Q, et al. (2024). "Follicular fluid-derived exosomes rejuvenate ovarian aging through miR-320a-3p-mediated FOXQ1 inhibition." Life Medicine 3(1):lnae013. DOI: https://doi.org/10.1093/lifemedi/lnae013
  3. Battaglia R, Musumeci P, Ragusa M, Barbagallo D, Scalia M, Zimbone M, Lo Faro JM, Borzì P, Scollo P, Purrello M, Vento EM, Di Pietro C (2020). "Ovarian aging increases small extracellular vesicle CD81+ release in human follicular fluid and influences miRNA profiles." Aging (Albany NY) 12(12):12324-12341. DOI: https://doi.org/10.18632/aging.103441. PMID: 32554857
  4. Abu-Halima M, Becker LS, Ayesh BM, Baus SL, Hamza A, Fischer U, Hammadeh M, Keller A, Meese E (2021). "Characterization of micro-RNA in women with different ovarian reserve." Scientific Reports 11:13351. DOI: https://doi.org/10.1038/s41598-021-92901-w
  5. Kuiper LM, Mens MMJ, Wu JW, et al.; Ghanbari M (2025). "Plasma microRNA signatures of aging and their links to health outcomes and mortality: findings from a population-based cohort study." Genome Medicine. DOI: https://doi.org/10.1186/s13073-025-01437-5
  6. Fulzele S, Mendhe B, Khayrullin A, Johnson M, Kaiser H, Liu Y, Isales CM, Hamrick MW (2019). "Muscle-derived miR-34a increases with age in circulating extracellular vesicles and induces senescence of bone marrow stem cells." Aging (Albany NY) 11(6):1791-1803. DOI: https://doi.org/10.18632/aging.101874. PMID: 30910993
  7. Ledderose C, Möhnle P, Limbeck E, Schütz S, Weis F, Rink J, Briegel J, Kreth S (2012). "Corticosteroid resistance in sepsis is influenced by microRNA-124-induced downregulation of glucocorticoid receptor-α." Critical Care Medicine 40(10):2745-2753. PMID: 22846781

Background knowledge

miRNAs are non-coding RNA molecules of ~22 nucleotides that silence genes post-transcriptionally: they bind the 3'UTR of target mRNAs within the RISC complex (with Argonaute/AGO2 as the catalytic core) and repress translation or degrade the transcript. Their biogenesis proceeds through a pri-miRNA transcribed by Pol II, nuclear processing by the Drosha/DGCR8 microprocessor, export via exportin-5, and cytoplasmic maturation by Dicer/TRBP before loading into RISC. A single miRNA can regulate dozens to hundreds of genes; that is why they function as "master regulators" of entire programs (differentiation, inflammation, senescence, steroidogenesis) rather than single-gene switches.

What makes miRNAs unique as biomarkers is not just that they reflect gene activity: it is that they circulate stably in blood, urine, and other fluids despite an extracellular environment rich in RNases. They achieve this by being packaged in extracellular vesicles (EVs): exosomes (30-150 nm, originating from multivesicular bodies) and microvesicles (100-1000 nm, budding directly from the plasma membrane), or bound to protein complexes (free AGO2) and lipoproteins (HDL). This packaging is not passive: EVs are active vehicles of intercellular communication — they transfer their miRNA cargo to recipient cells and reprogram their gene expression, which turns some circulating miRNAs into causal mediators, not just correlational markers, of the phenotype of the cell that receives them.

For the female hormonal axis, this opens a layer that none of the previous L5 approaches can offer. The functional epigenetic clock described in L5.1 reads age-function discordance via DNAm methylation — a relatively slow signature that changes over months/years. The methylation reading of ESR1/ESR2 promoters from L5.2 is also relatively stable. The chromatin/histone reading in adipose tissue from L5.3 is inaccessible without a biopsy. The exposure history from L5.4 is retrospective, not dynamic. Circulating miRNAs are the only layer in this line that is simultaneously liquid (non-invasive, theoretically measurable via dried blood spot or even urine) and dynamic (able to change within days-weeks in response to an inflammatory, metabolic, or acute stress event). This positions them as the natural bridge between a fixed epigenetic mark and a liquid biomarker capable of changing within a matter of weeks.

Relevant tissue sources for the hormonal axis: (1) granulosa/oocyte — follicular miRNAs reflect oocyte quality and follicular environment; (2) adipose tissue (especially visceral) — direct inheritance from L5.3, adipose tissue secretes EVs carrying miRNA cargo that acts as a systemic endocrine signal; (3) senescent cells of any tissue — the senescence-associated secretory phenotype (SASP) includes specific miRNAs, not just cytokines; (4) the HPA axis/hypothalamus-pituitary — some miRNAs directly modulate glucocorticoid receptor sensitivity; (5) vascular endothelium — endothelial miRNAs (e.g., miR-126) decline with vascular age and loss of estrogenic protection.

Findings from recent papers

The 2019-2025 evidence confirms that the circulating miRNA pool is not noise: it has identifiable tissue and functional structure. Umair et al. 2022 showed that miR-4516 in urinary exosomes is significantly elevated in women with premature ovarian insufficiency (POI), independent of Turner syndrome, and that this miRNA suppresses STAT3 in the ovary in a cyclophosphamide/busulfan-induced murine model of POI — the first evidence of a non-invasive miRNA (urine, not blood) with a functional mechanism demonstrated in the ovary. It is notable that the fluid is urine: this further lowers the invasiveness barrier compared with venipuncture.

Battaglia et al. 2020 characterized EVs from human follicular fluid in younger vs. older women undergoing IVF. The central finding is twofold: the number of small CD81+ EVs increases 2.1x in older women (p<0.01), and after normalizing for vesicle number, 46 miRNAs are differentially expressed by age, with consistent downregulation of miR-16-5p, miR-214-3p, and miR-449a, and upregulation of miR-125b, miR-155-5p, and miR-372. The mechanistic interpretation the authors propose — via TP53 — is that the aged ovary secretes MORE vesicles in response to follicular oxidative stress, but those vesicles carry a cargo that reflects worse oocyte quality, not a protective compensation. This directly parallels the L5.1 finding: more activity is not the same as better function.

Mei et al. 2024 contribute the most disruptive finding of the session: follicular fluid exosomes from young mice, rich in miR-320a-3p, rescued ovarian function in old mice by inhibiting FOXQ1 in granulosa cells, improving proliferation and mitochondrial function. It is the molecular equivalent of a parabiosis experiment (young-old blood) but isolated to a single miRNA transported by exosome. The protein hnRNPA2B1 actively controls the entry of miR-320a-3p into the exosome — meaning selective miRNA packaging into EVs is not random, it is regulated.

Abu-Halima et al. 2021 studied circulating (non-follicular) miRNA directly in 159 women stratified by AMH (low/normal/high reserve) and found 29 miRNAs significantly correlated with AMH status after validation, highlighting miR-100-5p for predicting high AMH (AUC 0.756) and miR-21-5p associated with reduced reserve. This is the most direct human study of "peripheral blood miRNA as a proxy of ovarian reserve" available today — without requiring follicular fluid or biopsy.

At the systemic aging layer, Kuiper/Ghanbari et al. 2025 (Rotterdam Study, Genome Medicine) quantified 2,083 extracellular miRNAs in plasma from 2,684 participants and built composite biomarkers: mirAge (108 miRNAs, predicts chronological age), mirPA (153 miRNAs, predicts PhenoAge — a multi-system biological age metric), mirFI (81 miRNAs, predicts frailty index), and mirMort (50 miRNAs, predicts 10-year mortality) independent of chronological age. This is, in essence, the first "miRNA clock" replicated and validated at population scale — the liquid-dynamic analogue of the Horvath (DNAm) clock that anchors L5.1. Fulzele et al. 2019 add the specific mechanism for a key piece of that clock: muscle-derived miR-34a, transported in circulating EVs, increases with age and induces senescence in recipient bone marrow stem cells by suppressing SIRT1 — direct demonstration of senescent phenotype transfer between distant tissues via circulating miRNA, with a positive miR-34a-IL6 correlation in subjects aged 20 to 90.

Finally, Ledderose et al. 2012 establish the HPA bridge: in sepsis (and replicated in other stress contexts), exposure to glucocorticoids induces miR-124, which in turn directly silences the glucocorticoid receptor alpha (NR3C1/GR-α), generating functional cortisol resistance at the receptor level — a molecular negative-feedback mechanism that operates over hours-days, much faster than any NR3C1 methylation change.

Full molecular/endocrine mechanism

The circulating miRNA pool in a woman is not a single signature: it is the weighted sum of EV secretions from multiple organs, each contributing its own signature. For the female hormonal axis, five sources dominate the relevant signal:

OVARY/GRANULOSA             VISCERAL ADIPOSE           IMMUNE/SENESCENCE         HPA/HYPOTHALAMUS         ENDOTHELIUM
miR-320a-3p (↓ FOXQ1)        miR-34a, miR-122           miR-146a (↓ NF-κB)        miR-124 (↓ NR3C1/GR-α)   miR-126 (vascular
miR-4516 (↑ STAT3, POI)      (via VAT, see L5.3)        miR-21-5p (SASP,          (induced by cortisol,    repair, ↓ with age)
miR-16-5p/miR-449a (↓ with                              inflamm-aging)            silences cortisol
age, oocyte quality)                                                              sensitivity)
        |                        |                          |                        |                       |
        v                        v                          v                        v                       v
                              CIRCULATING miRNA POOL (exosomes + free AGO2 + HDL-bound)
                                              |
                                              v
                          Reprograms distant recipient cells: granulosa receives
                          miR-34a from senescent adipose/muscle -> local SIRT1
                          suppression; hypothalamus receives cortisol-induced
                          miR-124 -> lower GR sensitivity -> HPA axis "deaf"
                          to its own feedback

The central mechanistic point of L5.5 is that the circulating miRNA pool functions as a cross-messaging system between the axes the lab has already modeled separately. Until now, L1 (microbiome), L2 (HPA), L4 (circadian), and L5.1-5.4 (epigenetics) were treated as systems connected by classical biochemical pathways (hormones, metabolites, neurotransmitters). Circulating miRNAs are an additional, faster connection layer: an inflamed visceral adipocyte (see L5.3) can literally send miR-34a via exosome to the ovary and silence SIRT1 in granulosa within hours — without any classical hormone mediating. This is direct post-transcriptional communication, not endocrine in the traditional sense.

Inflamed VAT (see L5.3)
  -> p53+ senescent cells secrete EVs loaded with miR-34a
  -> circulation -> ovarian granulosa captures EV -> miR-34a suppresses SIRT1 3'UTR
  -> reduced deacetylation of p53/FOXO3/PGC-1α in granulosa
  -> reduced oocyte mitochondrial resilience (direct bridge to L9 NAD+/SIRT1, not yet opened)

Chronic cortisol (L2.1-L2.4)
  -> induces miR-124 (Ledderose 2012, demonstrated in human lymphocytes under stress)
  -> miR-124 silences NR3C1/GR-α in hypothalamus, pituitary, and peripheral tissues
  -> the HPA axis keeps secreting cortisol, but tissues "hear less"
  -> L2.1 paradox: chronic high cortisol without proportional reproductive suppression
  -> possible explanation not considered in L2.1-L2.4: acquired receptor resistance, not just GR saturation

Cross-synthesis with previous findings

  • The functional epigenetic clock of L5.1 gains its dynamic counterpart. That approach relies on functional discordance inferred without a direct molecular clock. Circulating miRNAs are the first real candidate for a "liquid clock" testable in a dried-blood-spot sub-study — while the L5.1 reading changes slowly (months), a miRNA panel could capture week-scale fluctuations, functioning as the "high-frequency sensor" of the same underlying variable that L5.1 measures at low frequency.
  • L5.2 and L5.3 gain a circulating mechanism, not just a tissue-level one. L5.2 described ESR1/ESR2 methylation in situ; L5.3 described adipose chromatin in situ — neither is measurable without a biopsy. Battaglia 2020 and Mei 2024 show that the ovary and follicular fluid already secrete these signals into circulation via EVs. It is plausible (not demonstrated) that circulating miRNA profiles correlate with the epigenetic state described in L5.2/L5.3 without requiring tissue.
  • L5.4 gains a plausible transmission pathway for "multigenerational memory that activates late." If early exposure alters the number of senescent cells or the follicular oxidative stress threshold across the lifespan, the adult circulating miRNA pool (miR-34a, miR-21, miR-146a) could be the closest available reading of that "countdown" before symptoms appear.
  • L2 (HPA) gains a new molecular mechanism not considered in L2.1-L2.6: acquired resistance to the glucocorticoid receptor via miR-124, distinct from the four HPA→HPO suppression pathways already documented in L2.1 (GR-KNDy, CRH-PVN-GABA, GR-pituitary, RFRP-3). This is a fifth candidate pathway: not suppression, but receptor desensitization by a miRNA induced by chronic cortisol itself.
  • L1 (progesterobolome/estrobolome) and L5.5 share the concept of the extracellular vesicle as a vehicle. L1 established microbiome-host communication via soluble metabolites (SCFAs, sulfates). L5.5 adds that the host itself uses inter-organ vesicular communication with the same logic of "packaged molecular cargo, transported, delivered to a specific receptor."
  • Direct bridge to L9 (NAD+/SIRT1, not yet opened): the miR-34a→SIRT1 mechanism (Fulzele 2019) is the first concrete, mechanistic link between L5 (circulating epigenetics) and L9 (NAD+/sirtuins). When L9 opens, this finding should be cited as precedent: a tissue's NAD+/SIRT1 load may be determined in part by miRNA signals imported from other tissues, not only by local NAD+ biosynthesis.

Lua Labs Hypotheses

Hypothesis 58: Liquid convergence of this line's epigenetic biomarkers

Statement: A candidate panel of 8-12 circulating miRNAs spanning the five axes (ovarian/granulosa, adipose, immune/senescence, HPA, endothelial) would covary significantly with the four digital epigenetic biomarkers described in L5.1-L5.4 — designed independently across separate sessions — suggesting that these four readings are partial fragments of the same underlying allostatic-endocrine load, of which the circulating miRNA pool would be the common liquid integrator.

Proposed mechanism: Each prior L5 approach was designed to capture a specific tissue-level mechanism (systemic DNAm clock, receptor methylation, adipose chromatin, exposure history). But the five axes that produce circulating miRNA (ovary, adipose, immune, HPA, endothelium) respond to the same upstream triggers — chronic low-grade inflammation, cortisol, insulin resistance, oxidative stress — that feed the four epigenetic readings of L5.1-L5.4. If this is correct, a liquid miRNA panel would not be a fifth isolated biomarker but the latent variable that explains why the other four tend to move together in the same woman.

Confidence level: Medium — each individual mechanistic piece (organ-specific miRNA, EV transfer, senescence) is well documented; the specific cross-covariance between the four digital biomarkers of L5.1-L5.4 and a real miRNA panel has never been tested because those four readings are original digital constructs of this laboratory, not replicated by any other research group.

How to validate:

  • With a formal study: sub-cohort of 100-150 women stratified by quartiles of a composite index of the digital epigenetic biomarkers from L5.1-L5.4, dried blood spot or plasma collection, targeted RT-qPCR panel of 8-12 miRNAs (miR-320a-3p, miR-4516, miR-16-5p, miR-449a, miR-34a, miR-146a-5p, miR-21-5p, miR-124, miR-126, miR-100-5p), correlation and regression analysis.

Limitations: Circularity risk: the four prior biomarkers already share input variables (symptoms, sleep, age, weight) by design, so they could covary with each other without any real biological covariance with miRNA. The miRNA panel must predict INCREMENTAL variance over age/weight/symptoms before the hypothesis can be accepted.

Hypothesis 59: miR-34a as a direct molecular bridge to L9 (NAD+/SIRT1)

Statement: Elevated circulating miR-34a — transported in EVs derived from inflamed visceral adipose tissue or senescent muscle — directly suppresses SIRT1 in distant tissues, including ovarian granulosa, such that women with high circulating miR-34a would have lower tissue "functional SIRT1 reserve" even before a measurable NAD+ decline is evident, functioning as an anticipatory biomarker for the future L9.

Proposed mechanism: miR-34a is induced by p53 in senescent cells, packaged into EVs, and transferred to distant cells — demonstrated directly in the muscle→bone marrow axis (Fulzele 2019). SIRT1 is a NAD+-dependent deacetylase critical for p53, FOXO3, and PGC-1α, and its suppression reduces mitochondrial resilience. If the inflamed visceral adipose tissue already characterized in L5.3 secretes EVs rich in miR-34a into circulation, and if these EVs are taken up by ovarian granulosa (a mechanism demonstrated for other tissue-tissue pairs, not yet for adipose-ovary specifically), this would produce a form of "imported senescence": the ovary would age functionally faster not from autonomous depletion, but from an exogenous signal received from a distant tissue.

Confidence level: Low-Medium — the miR-34a→SIRT1 mechanism is solidly established for the muscle→bone marrow pair; the specific adipose→ovary transfer in humans is a reasoned extrapolation, not a direct finding from the reviewed literature.

How to validate:

  • With a formal study: once L9 opens, design a sub-study with paired measurement of plasma miR-34a + blood NAD+/NADH + a composite proxy of visceral adiposity (L5.3) + body composition (waist/DXA if available), testing whether miR-34a predicts functional NAD+ decline ahead of chronological age alone.

Limitations: This is the most speculative hypothesis of the session because it connects two lines (L5 and L9) that have not yet been studied together by any known laboratory; the risk is over-extending a mechanism valid in one tissue pair (muscle-bone marrow) to an untested pair (adipose-ovary).

Hypothesis 60: Window of functional cortisol resistance via miR-124 (bridge to L2)

Statement: High circulating miR-124 levels — or their behavioral correlate of discordance between perceived cortisol and symptoms already explored in L2.4-L2.5 — would mark a state of acquired functional cortisol resistance at the receptor level (NR3C1/GR-α downregulation), in which the HPA axis keeps secreting elevated cortisol but target tissues — including the hypothalamic KNDy network and ovarian granulosa — respond proportionally less, which could reconcile the paradox observed in L2.1: chronically elevated cortisol without always-proportional reproductive suppression.

Proposed mechanism: Glucocorticoids induce miR-124 in a molecular negative-feedback loop (demonstrated in human lymphocytes under stress/sepsis, Ledderose 2012); high miR-124 silences GR-α, generating relative cortisol resistance. This could explain why some women with chronically elevated cortisol documented in L2.1-L2.4 do not show the "expected" reproductive suppression according to the four pathways already described (GR-KNDy, CRH-PVN-GABA, GR-pituitary, RFRP-3): their elevated miR-124 has already desensitized the receptor, simultaneously attenuating both the reproductive harm and any physiological regulatory benefit of cortisol.

Confidence level: Low — the mechanism is demonstrated in sepsis and the animal hippocampus under acute stress; it has never been specifically studied in the chronic human HPA-HPO reproductive context that is the focus of L2.

How to validate:

  • With a formal study: paired measurement of plasma miR-124 + diurnal salivary cortisol + ex vivo GR sensitivity (dexamethasone suppression in lymphocytes) in a cohort of women with FHA/PCOS vs. controls, already identified in L2.1 as the two poles of the same HPA-HPO continuum.

Limitations: This is the hardest hypothesis to operationalize without a biological sample because there is no validated behavioral proxy for "GR resistance" — the cortisol-symptom discordance could be explained by dozens of confounding variables (subjective stress perception, social support, sleep, NR3C1/FKBP5 variants already documented in L2.1) besides miR-124.

Candidate formulation (if applicable)

I am not proposing a pharmacological supplement formulation — the object of this session is a diagnostic panel, not an intervention. However, given that the general nutrigenomics literature suggests certain dietary components modulate the expression of specific inflammatory/senescent miRNAs (quercetin and curcumin reported as modulators of miR-34a in preclinical models; omega-3 associated with modulation of miR-21/miR-146a in chronic inflammation studies), I include a low-risk note for future research: a watchlist of dietary candidates (quercetin, curcumin, omega-3, red-fruit polyphenols) whose association with modulation of specific inflammatory/senescent miRNAs should be reviewed against solid human evidence before any claim, and which must NOT be communicated as a validated intervention under any circumstance — current evidence is mostly preclinical/in vitro.

Regulatory status: Not applicable — a research note, not a formulation for product or communication.

Requires validation: Specific human RCT evidence on dietary modulation of miRNA before considering any public mention.

Individual variability

The concentration and composition of the circulating miRNA pool varies among women across multiple layers, beyond chronological age:

  • EV secretion rate by tissue: women with more inflamed VAT (see L5.3) or greater senescent cell burden likely secrete more EVs carrying miR-34a/miR-21; this is not uniform even at the same chronological age.
  • miRNA biogenesis efficiency: polymorphisms in DROSHA, DGCR8, DICER1, TARBP2, and AGO2 can alter overall miRNA processing efficiency, affecting the absolute magnitude of any circulating signal independent of the stimulus.
  • Selective packaging into EVs: proteins such as hnRNPA2B1 (which controls miR-320a-3p entry into exosomes, Mei 2024) may vary in activity between individuals, determining how much of an intracellularly produced miRNA actually reaches circulation versus is degraded locally.
  • Recipient tissue sensitivity: the same circulating dose of miR-34a would have a different effect on granulosa depending on the recipient tissue's baseline SIRT1/NAD+ level — women with already-compromised NAD+ reserve (age, diet, sleep) would be more vulnerable to the same signal.
  • Baseline inflammatory context: women with chronic low-grade inflammaging (diet, dysbiotic microbiome inherited from L1, fragmented sleep from L4) would have a circulating pool skewed toward proinflammatory miRNAs (high miR-21, relatively low miR-146a) independent of ovarian age.

Notice. Lua Labs is a scientific research laboratory. Reports are literature syntheses, not medical advice. Any clinical decision should be made with a health professional.