Senescent cells in the ovary: accumulation, SASP and paracrine effect

Scientific delta of the report

The English twin freezes, in one factual chain, the same manifest as the Spanish report. The recorded heading defect is corrected in both drafts: the background heading now contains the word mechanism, and no DOI, PMID, number, unit, hypothesis ID, confidence or limitation is moved. The title of L10-1 does not describe an ovarian SASP already measured. The accumulation visible in the human atlas is the epithelium, which rises in proportion and in cell-cycle genes, plus a fraction of CDKN1A+ nuclei that goes from 0.091 to 0.156, while the intensity of a module of 68 secreted SenMayo genes inside positive stromal nuclei does not rise (old−young delta −0.007). That module covaries with the cell-cycle score in the per-donor means of granulosa, smooth muscle and blood endothelium (Spearman 0.400, 0.943 and 0.286), and its intersection with the cycle set is empty, so the covariation is not a co-transcript. Inside the list are PLAT, PLAU and SERPINE1, the periovulatory cassette of the rhesus monkey (Liu 2004, PMID 14701676), not a senotype. The CDKN2A count (184 of 42 568 nuclei; 159 of them CDKN1A−) does not separate p16INK4a from p14ARF (Stott 1998, PMID 9724636). Q0-R2-IDENTITY remains specified and not run: the within-donor interquartile range is rejected because smooth muscle of ovary47 (11 nuclei) and granulosa of ovary51 (18) would leave 5.5 and 9 nuclei in expectation, below the floor of 10; window T is the cross-donor intersection of the 10th and 90th percentiles of genomic nFeature, written before the Spearman, and F averages 65 genes without dividing by 68 a vector that contains three zeros. L10-1-AR-H1 v1 dies only if T and F preserve rule R2 with and without ovary47. Neither that death nor that survival opens tissue, a senolytic, or organ mediation. The Pharma evaluation nominates no target: SERPINE1 is an anti-analyte. The link that remains broken is secretion, per cell and per time, from a cell that is not luteal, atretic, immune or cycling epithelium, necessary and sufficient for a neighbor function that is not p21. The bridge to women’s longevity remains at H0.

1. Executive scientific abstract

Accumulation, senescence-associated secretome and paracrine effect are not three names for one ovarian object. They are three estimands that no admissible primary has measured together in a woman, or in a female, inside natural reproductive aging.

A stable senescent state requires durable arrest and an orthogonal readout, in an identified lineage, separated from luteinization, atresia, reversible quiescence and infiltrate. The secretome is an analyte per cell and per time, with an isonutrient sham. Paracrine action is a change in the neighbor’s function that disappears when secretion is interrupted and reappears when the factor is restored, with the number of emitters matched. Longevity is an event of one organ, with mediation and competing risks. AMH, menopause, reserve, litter size and “senescent burden” do not occupy that post.

What has been measured splits into arms that the title fuses. In the human ovary of Jin et al. 2025 (Nature Aging, PMID 39578560), CDKN2A is in ~0.43% of nuclei and the authors write that human ovarian senescence requires experimental validation. CDKN1A rises as a proportion (~9.49% → ~15.56% in the article; 0.091 → 0.156 in the count of this matrix) and is the same arm that luteal exit (Jirawatnotai 2003) and an AMH-reversible quiescence (Meinsohn 2025) use for something else. The only compartment whose proportion rises is the epithelium, and it raises cell-cycle genes. At 9 months, before estropause, Isola et al. 2024 count a doubling of the immune fraction and an absolute number of CD90+ cells, not a stereology of durable arrest. The SA-β-gal stroma of Maruyama et al. 2023 already sits, in the same paper, next to multinucleated macrophages and probably atretic follicles.

The public computation on GSE202601 (42 568 nuclei, eight donors) does not turn that RNA into SASP. A module of 68 secreted SenMayo genes, without matrix and without overlapping the cycle set, accompanies the cycle score in granulosa, smooth muscle and blood endothelium. Stromal identity does not accompany it: in granulosa the rho is −0.80. The CDKN2A residue inside the CDKN1A+ gate is not estimable outside stroma, because 159 of 184 CDKN2A+ nuclei are CDKN1A−. Saul et al. 2025 (PMID 41162753) do not score a senotype until Ki67+, CD45-high and S phase are removed, and the ovary is not in that atlas. Coppé et al. 2011 (PMID 21880712) show, in human fibroblast, that ectopic p16 or p21 arrest without installing the SASP. Adversarial review withdrew the name “cell-cycle co-transcript”: the genes do not coincide. What remains is a covariation of scores across very few donors, still without identity.

The live primary hypothesis is L10-1-AR-H1 v1: that covariation stops meeting rule R2 when detection depth is equalized or when the three genes of the ovulation cassette are removed. The competitor L10-1-AR-H2 v1 activates only if both arms preserve R2, and then the residue would be secretion from cells that are still cycling, not a p16 senotype. The translational hypothesis L10-1-AR-HT1 v1 is a design rule: no index of CDKN1A fraction, unfiltered SenMayo, fused p16/p21, or SERPINE1 enters an organ study before a local analyte that is not the cassette. HUMAN_QA_REQUIRED. There is no target. Nothing is prescribed.

Executive conclusion. Maturity H2 for the covariation already measured and for the empty intersection with the cycle; H1 of literature for the identity confounders (luteal, AMH, infiltrate, macrophage); H0 for which arm explains R2, for isoform protein, for secretome per cell, for the neighbor’s function and for longevity. The smallest discriminator is Q0-R2-IDENTITY on the matrix already downloaded. A clean refutation of R2 by depth or by the cassette, or a clean preservation that forces the proliferation filter, is scientific success. Neither of the two is healthspan.

2. Scientific question and relevance

Question

What mechanisms connect the accumulation of senescent cells in the ovary, the SASP and the paracrine effect with women’s health and longevity, and which new hypotheses can be falsified?

The question decomposes into gates that must not be fused:

1. Identity. Which cell carries p16, p21, SA-β-gal or a SASP gene: granulosa, theca, stroma, endothelium, epithelium, immune, luteal cell or atretic follicle?
2. State. Is the arrest durable, or is it luteal exit, reversible quiescence or atresia?
3. Number. Does the absolute number of a durable state rise, the fraction in an atlas, or the program per cell at constant number?
4. Secretome. Is there an analyte per cell and per time, or only organ RNA, nuclear RNA or a lysate?
5. Order. Does accumulation precede the secretome, or does an intact secretome induce a new state in the neighbor with the number of emitters fixed?
6. Paracrine. Does interrupting secretion abolish a neighbor function that is not p21, and does restoring the factor restore it, with an isonutrient sham?
7. Milieu. Do E2, P4, ovarian CRH, HIF or the ovulation program explain the same readout?
8. Distal. Only after a local function: is an event of one organ mediated, with competing risks?

“Senescent cells in the ovary” is not an exposure. “SASP” is not a cytokine list. “Longevity” is not a later menopause. The roadmap feeds L10 with luteal senescence that would lower progesterone, with senescence epigenetics and with NAD+–senolytic synergy. Those three phrases are program intentions (L10-1-SCOPE-C3). L7 separated luteal architecture from reserve and did not identify senescence as the cause of the fall in P4 exposure. L9 closed healthspan at H0 and did not measure senolytics as the estimand of 10.1. Dasatinib plus quercetin, fisetin, window, NMN and an observational protocol remain in 10.2–10.6 and are not opened here.

The relevance to women’s health is conditional and is empty today at the causal link. If there existed a non-luteal, non-atretic, non-immune and non-proliferative-epithelial ovarian cell, with durable arrest, whose per-cell secretome was necessary and sufficient for an independent neighbor function, that trajectory could afterwards ask about an organ. Until U1 is measured, asserting that ovarian SASP shortens or lengthens women’s lives is an inference, not a mechanism. The laboratory formulates hypotheses in order to falsify that inference. It does not inherit it.

3. Scope, population and life stage

Biological population of the charter: adult women in natural reproductive aging, STRAW+10, not chronological age alone. Reference: late reproductive (−3b/−3a). Principal window: transition (−2/−1), because luteal exposure can already fall (L7-4) without that identifying luteal senescence. Replication: early postmenopause (+1c), not averaged with the transition, because the follicle-poor ovary changes the baseline composition.

The only human ovarian object this project computed does not meet that classification. Jin uses four donors aged 23–29 years and four aged 49–54, histology described as normal, without STRAW+10. Supplementary Table 1 lists ages 23, 27, 28, 29, 49, 51, 52 and 54 under ovary1–ovary8. The GEO metadata use other tokens. ovary23 (9 151 nuclei, young) matches age 23, which ST1 records as surgery, bilateral salpingo-oophorectomy, with the clinical diagnosis in the record (gender dysphoria), not autopsy. ovary47 (4 724 nuclei, old) is not in ST1: the published range of the aged donors is 49–54 and there is no row for 47 years. The young/old contrast mixes that surgical provenance with autopsies. The result is not read as life stage.

Inferential unit: the woman, or the female in the animal arm. Nuclei, follicles, sections and wells are not replicates. Donor = n.

Left out: prescription and dose; current Lua user data, PHI, PII and private genomes; features, personas and marketing; 10.2–10.6; reopening NAD, CD38, oocyte SIRT6, NMN/NR, food-log and granulosa UDP-sugars; ovarian cancer and gonadotoxic injury as if they were the natural transition; male evidence transported without a female replicate; p>0.05 as equivalence; fusion of fertility, bone, vessel, brain and mortality into a composite of “ovarian longevity”.

Wu et al. 2024 (PMID 38594460) is a contrast, not the estimand: scRNA of 9 women (3 per band of 18–28, 36–39 and 47–49 years; 92 965 cells) and spatial of 15 (5 per band), ovaries from hysterectomy for cervical or endometrial cancer without ovarian metastasis. SA-β-gal n=10/group and whole-ovary SASP qPCR n=9/group do not pass Q0. FOXP1 cKO produces a reserve phenotype (POI, litter) that HT1 is forbidden to call healthspan.

4. Background knowledge, mechanism and mechanistic map

4.1 What the object is not

p21, p16, SA-β-gal, γH2AX, lipofuscin, a SenMayo score, serum IL-6, an atlas proportion, CellChat, AMH, luteal P4 and the age at natural menopause are not, each of them, senescent state, secretome or longevity. Captured proportion is not activity per cell: that rule comes from L9-4-K4 and is inverted here. A rare cell can dominate an analyte if the analyte per cell is high. The rarity of CDKN2A does not kill that possibility. It does kill the sentence “the ovary accumulates a p16 burden”.

4.2 The chain the title invites and that this project does not assume

Ovarian senescent cell accumulates → secretes a SASP → the neighbor loses function and enters senescence → progesterone falls → women’s longevity worsens.

Today each arrow is inference or is disfavored. The p21/p27 arrest of luteal exit separates from the steroidogenic enzymes (Jirawatnotai: the double null keeps proliferating and retains them). L7-4 documents loss of luteal exposure without a causal P4→E2 change-point and without localized luteal senescence. L9-4-K10: a later menopause distributes bone and type 2 diabetes on one side and hormone-dependent cancers on the other, without a uniform lifespan. L9-4-K9: a senolytic in a small female group is not life equivalence. G8 remains closed.

4.3 Revised map: two arms, a network that falls, a matrix apart

CDKN1A / p21 ARM                            CDKN2A / locus ARM
  proportion 0.091 → 0.156                    184/42 568 nuclei (0.432%)
  luteal exit (p21+p27)                       159/184 are CDKN1A−
  AMH-reversible QASP, no β-gal               the symbol does not separate
  epithelium: proportion and cycle up         p16INK4a from p14ARF
        │                                            │
        ▼                                            ▼
  module of 68 secreted genes                 not estimable inside
  ∩ cycle = ∅                                  the CDKN1A+ gate
  Spearman of per-donor means                  (outside stroma)
  0.400 GC / 0.943 SMC / 0.286 BEC
        │
        ├─ T  nFeature not partialled      (not run)
        └─ F  without PLAT, PLAU, SERPINE1 (not run)
              if F breaks R2 → the score is the ovulation cassette
              if T and F preserve R2 → proliferation filter
                    └─ if the residue lives at low cycle → candidate,
                       still without protein and without neighbor function

NETWORK (CellChat, RNA)  total a little down; FSH/GDF empty; more input to epithelium
MATRIX                   human ECM score down; deposited murine collagen up;
                         MMP2 down; Il6 can fall with the fibrosis still there
U1                       per-cell secretion from an emitter that is not luteal,
                         atretic, immune or cycling epithelium — not measured
G8                       closed

The C/EBP node does not execute senescence. In Jin, the motif rises in most types except immune and epithelium, and CEBPD accompanies the motif in granulosa, immune and theca. Meinsohn had already raised Cebpb in a reversible secretome without a difference in β-galactosidase. Using C/EBP as a SASP signature repeats the error of using p21.

Acosta et al. 2013 (PMID 23770676) show, in oncogene-induced senescence of fibroblast, not in ovary, that IL-1 and TGFβ ligands can induce paracrine senescence via p15 and p21. Importing that arrow turns the rise in p21 of the receiving cell into a circular readout: p21 already marks luteinization and quiescence. That is why p21 of the receiving cell is not the coprimary of order.

5. Evidence method

Read, in order: the charter, the lifetime memory packet, the scoping, the map, the verification, the synthesis, the compute decision, the computation, hypothesis generation, adversarial review and the design. The report does not reopen the search and does not recompute R2. Every figure in this text comes from an artifact already written or from a primary that stage declared it had read.

Classes, preserved: H human; A female animal; V in vitro or ex vivo; C atlas or computation; I inference. “Direct” names the link the primary measured. It does not close the chain in the title. A proportion is not an absolute number. RNA of a gene annotated as secreted is not a secretome. A motif is not a protein. CellChat is not conditioned medium. p>0.05 is not equivalence. An arm that is not estimable is not a victory by elimination.

Verification corrected the map at three points that this report does not compress again. Published Isola says the immune fraction doubles at 9 months; “more than doubles” was preprint language. Maruyama already mixes CDKN2A with multinucleated macrophages and SA-β-gal with atresia, and its SASP qPCR is n=3 of breeders. Wu is oncologic surgery and does not pass Q0. The conditioned medium of PMID 42121871 is iatrogenic senescence on SKOV3 and does not fill the paracrine gate.

The computation used the public matrix, not user data. Counts SHA-256 2d9886b23f85b52eb2080c940035a98181d2016f1387cff79ed9df83a3e17852. Metadata SHA-256 58edc9fec6fba66715b3a3c273e413f73191b8c392e5ec5db55444a8a6d93423. Module of 68 genes SHA-256 899aeb8dd1888fb1f04d18f70a52340b882811d2f4c37fa9148a32884bfa7726. R 4.6.1, Matrix 1.7.5, cycle cc.genes.updated.2019 from Seurat v5.2.1 after removing FEN1. BioNeMo did not enter: there is no variant or fold to decide, and an embedding is not the collinearity statistic and does not secrete protein. The matrix was not uploaded to an external service.

Adversarial review read Ding et al. 2020 (PMID 32341560), Stott et al. 1998 (PMID 9724636) and Liu et al. 2004 (PMID 14701676) in order to attack the identity that the hypothesis stage had assigned to the module. The design froze the window and the denominator and did not execute the recalculation. A design is not a result.

6. Evidence map

Reading of the map. Q0-IDENTITY does not pass. Q0-SECRETOME does not pass. G1, the absolute number of a durable state, is empty: the only absolute that rises with counts is CD90+. G2 and G3, functional paracrine action and order, have no factorial in non-tumor ovary. Scope H1 and H4 remain at H0 because of a gap, not because of a null inside the estimand. Gao’s null dissociates marker and structure in a young injured female; it does not kill the hypothesis in the natural transition and is not used to rank dasatinib, quercetin or fisetin.

7. Contradictory evidence and null findings

1. p21 does not diagnose senescence. It rises at luteal exit and in a quiescence that reverses when AMH is withdrawn, without differential β-gal. The p21 attenuator does not rescue the Meinsohn phenotype. Treating CDKN1A+ as ovarian senescent burden contradicts those two primaries.
2. The p21/p27 double null does not switch off steroidogenesis. The bridge “senescent luteal cell → less progesterone” is disfavored. The exposure deficit, if it is studied, is studied as mass, perfusion and state of the corpus luteum (L7-1, L7-4), not as SASP.
3. The cell that accumulates in the woman is in cycle. Only the epithelium rises in proportion, and it raises cell-cycle genes. Granulosa, theca and endothelium fall. That is the opposite sign of a durable arrest that accumulates.
4. The predicted network is not a SASP field. CellChat falls a little in total, empties FSH/GDF toward granulosa and theca, and increases input to the epithelium. A propagating inflammatory field should do the opposite in first order. Ligand–receptor RNA does not refute a minority; it does forbid drawing the field.
5. Fibrosis does not follow the IL-6 module. The human ECM score falls. In the mouse, deposited collagen rises, fibroblast collagen RNA does not, and MMP2 falls: the nearest tissue mechanism is less degradation. In doxorubicin, Il6 and Mcp1 can fall with the fibrosis and Tgfβ1 still there. “Il6 fell” does not prove that Il6 caused the matrix.
6. Isola is not “>2×”. The published paper says doubling. Flow does not confirm the rise in B or myeloid cells that the scRNA suggested. Lipofuscin and giant cells are not accepted as classical senescence.
7. Maruyama does not pass Q0 from the inside. CDKN2A around multinucleated macrophages; weak SA-β-gal in probably atretic follicles; qPCR n=3. Reproductive history moves the western (breeders versus virgins, n=3).
8. The module is neither cycle nor senotype. Intersection with 96 cycle genes: none. Spearman of means, not of 42 568 nuclei. BEC enters at 0.286, below 0.30, only by sign. The median is 0.400 because SMC carries 0.943 with 6 donors and granulosa 0.400 with 4.
9. SERPINE1 inside SenMayo is periovulatory PAI-1 in the monkey, not Saul’s lung senotype. Using it as an analyte of senescent burden confuses ovulation, luteinization and senescence.
10. 159/184 is not the p16 senotype. It is a locus count without isoform. In Stott’s lines, p14ARF raises p21; in this matrix most CDKN2A+ nuclei are CDKN1A−. That does not demonstrate p16, or ARF, or that the zero is biological. It demonstrates that the RNA disjunction instantiates neither loop.
11. There is no admissible conditioned medium. PMID 42121871 is cancer. Meinsohn measures chemokines in a pooled lysate. Battaglia measures vesicles from IVF follicular fluid, not a senescent emitter.
12. The confidence 0.56 that a fused index “does not mediate” an organ was inflated. There is no event, no mediator, no competing risks. What is kept is the rule not to freeze the index. Mediation, as a fact, remains at H0.
13. ovary47 does not reorder the call already made. Removing that token leaves the SMC cycle rho at 0.90 and the median at 0.40. A future call that depends on assigning it 49–54 years is invalid.
14. Du 2026 does not promote the p21 arm. It reanalyzes the same GEO, without Saul’s filter, and the mouse arm is 14–18 months, not the factorial. An IL-1 edge can rise inside a total network that Jin published as down.

These nulls lower maturity. They do not excuse stopping the formulation. The uncertain link is named in section 8.

8. Multiscale mechanistic synthesis

Molecular

Two CDK inhibitors are not one state. The arm that rises in the atlas is CDKN1A. It is shared with luteal exit, with AMH quiescence and with the human epithelium that also raises cell-cycle genes. The CDKN2A arm is rare. The count does not say which of the two proteins of the locus is there. The module of 68 secreted genes shares no genes with the cycle and does contain the program with which the monkey follicle ruptures (PLAT, PLAU, SERPINE1). SERPINE2, INHA and PAPPA are on the list and this project assigns them no ovarian mechanism: they do not enter the pre-registered subtraction. Eight matrix genes, MMP2 included, were removed from the module on purpose: fibrosis is another branch. Six SenMayo symbols are not in the matrix (ANG, CCL1, CCL7, CSF2, CST4, IL13). IGFBP5 was removed because it overlaps the stromal marker.

mTOR and insulin rise as pathway scores in the same atlas, with RICTOR by ISH. They remain an aging program the authors highlight. They are not reopened as NAD, CD38 or a senolytic. HIF-1 is higher in CDKN1A+ nuclei and higher still with age; Jin’s own discussion recalls that in-vitro hypoxia can brake senescence at oxygen tensions that are not those of the tissue. H3 is not closed.

Cellular

The emitter the title seeks — not luteal, not atretic, not immune, not cycling epithelium, with durable arrest — has no clean lineage in any admissible primary. The epithelium is the compartment that increases and the one that least resembles an arrest. Aged granulosa, immune and theca lose identity genes and gain a stromal signature (Extended Data Fig. 8f of Jin): a “stromal” signature does not locate a fibroblast that became senescent. It can be another cell that resembles stroma. In the mouse, the CD90+ infiltrate rises in absolute number and Cdkn1a rises in type-17 lymphoid cells. Maruyama’s SA-β-gal stroma was not separated from the macrophage or from atresia.

Tissue

The predicted communication has the opposite sign of a propagation field and does not refute it, because it is RNA. Exhausted FSH and GDF, with fewer receptors and fewer steroidogenic enzymes in granulosa, are follicular endocrinology. That lane is L10-1-SCOPE-H3 v1 for follicle function, not a senescent cargo. The matrix competes only for the fibrosis outcome: deposited collagen and MMP2 protein, with per-cell IL-6 and CD90+ number as covariates. It can survive even if the secretome hypothesis dies, and it can die even if a secretome moves another function. It is not fused.

Systemic

There is no bridge. Jirawatnotai separates proliferative arrest from the enzymes. L7-4 forbids converting the fall in P4 exposure into a P4→E2 change-point. Jin integrates age-at-menopause variants: ANM is the outcome the translational hypothesis is forbidden to use as longevity. Systemic inflammation does not substitute for the per-cell analyte. L6-2 already separated lowering homocysteine from lowering events. A serum IL-6 is not the secretome of an ovarian emitter.

Uncertain link U1

Does an ovarian cell that is not a luteal cell, not an atretic follicle, not an infiltrating immune cell and not cycling epithelium, with durable arrest, release per cell and per time a factor necessary and sufficient to change a neighbor function that is not p21?

Until U1 is measured, the chain in the title is not built. The site where it breaks has been located. The computation did not close U1. It showed that the candidate RNA, inside the gate that rises, is a covariation of scores. Review added that this cell also cannot be named with the CDKN2A count or with SERPINE1, and that the module may be reading ovulation or sequencing depth. The design added that the first spend is not to look for the cell in a section.

The rank orders what is distinguished first. It is not a causal probability.

9. Computational layer

Decision: use. The question was whether, in CDKN1A+ nuclei that are not immune or epithelium, a secreted SASP module that does not overlap stroma, cycle or matrix is explained by stromal identity and by cycle at donor level, or whether a residue remains in high-CDKN2A nuclei of a lineage that is not stroma. That could reorder the draw. It could not take scope H1 or H4 out of H0, because RNA is not a secretome.

Input. GSE202601, integer dgCMatrix, 27 315 genes × 42 568 nuclei. Eight donors: ovary23, ovary27, ovary28, ovary29 (young, 23 534 nuclei) and ovary47, ovary51, ovary52, ovary54 (old, 19 034). Human SenMayo, sheet human, 125 genes; 83 annotated as secreted. After removing eight matrix genes, six absent and IGFBP5, 68 remain. The tautology threshold was 15 genes. It did not fire.

Score. Mean of log1p(count / library × 10 000) per nucleus. Positive = raw count > 0, threshold not optimized against the module. Accompaniment: Spearman of per-donor means, usable row ≥10 CDKN1A+ nuclei. A covariate accompanies if the median rho is ≥ 0.30 and at least three lineages have rho > 0. A p from 42 568 nuclei is not used.

Census. CDKN1A: 5 105 nuclei (fraction 0.120); young 0.0910 (2 142); old 0.1557 (2 963). The article publishes ~9.49% in the young. The raw count is reported; the threshold is not moved to close 9.10 against 9.49. CDKN2A: 184 nuclei (0.432%). Double positives: 25. CDKN2A+ and CDKN1A−: 159. Epithelium: 0 CDKN2A+ nuclei.

R2, usable rows. GC 4 donors, SMC 6, BEC 7. Theca and lymphatic endothelium do not reach 4 rows.

Stroma: median 0.464, only 2 of 3 positive signs. It does not accompany. Cycle: median 0.400, three positive signs. It accompanies. The weight is SMC. Without ovary47, SMC remains at 5 donors, cycle rho 0.90, and the call does not change.

CDKN2A residue inside the gate. GC, TC, SMC, BEC and LEC do not assemble four donors with both classes. That emptiness is not estimable, not equivalence. Stroma assembles five donors, mean of deltas +0.010, with opposite signs across donors. The rule forbids naming stroma as the first tube.

What the computation changed. Scope H2 received a public cycle signal, with an H2 ceiling: the fraction rises and the intensity of the module inside positive stromal nuclei does not (delta −0.007). Bulk CDKN1A+ is not the first tube of a secretome. The CDKN2A minority did not win draw priority with this RNA. Scope H1 and H4 remained at H0.

What did not change, and what review took away. The computation did not say that the module “is cycle”. The later hypothesis turned the Spearman into that identity. Review withdrew it: disjoint list, a statistic of means, an ovulation cassette inside the numerator, and a score that scales by library without partialling the number of genes detected. Ding limits the platform (more intron in cortical nuclei than in cells) and does not measure depletion of these 68 genes. It is not asserted that Jin’s matrix includes introns.

What cannot be inferred. Protein, nucleoside, SA-β-gal, medium, sham, clamp, STRAW+10, paracrine action, healthspan. A rho of 0.40 in four granulosa is an order of ranks, not an interval. SenMayo was trained on bone, marrow and adipose. In ovary it can read inflammation even with the immune cluster left out.

10. Primary hypothesis

L10-1-AR-H1 v1 — the SASP–cycle covariation is not identified

Role. Primary. It replaces the identity of L10-1-HG-H1 v1, which called a cell-cycle co-transcript a module with no cycle genes. It does not assert that a secretome moves the neighbor.

Falsifiable statement. In the CDKN1A+ nuclei of granulosa, smooth muscle and blood endothelium of GSE202601, rule R2 (per-donor Spearman median ≥ 0.30, three lineages with positive rho, and also SMC ≥ 0.30 in the death criterion) stops being met when the score is recomputed in the shared window of detected genes, or when PLAT, PLAU and SERPINE1 are removed from the module before the rho is looked at.

Mechanism. The score and the cycle set share no genes. What moves together, across donors, can be (T) a nucleus with fewer zeros that raises the mean of any long module, or (F) the cassette with which monkey granulosa and theca execute follicular rupture. If T or F break R2, the identities “it is cycle” and “it is SASP” fall together for this score. If both preserve R2, the object passes to AR-H2. It is not called a co-transcript again.

Predictions.

1. In the detection window, SMC falls below 0.30, or fewer than three estimable lineages remain with positive rho. Four granulosa do not carry the call alone.
2. Subtracting the three genes breaks R2 even if the window does not break it. Then the module was reading the cassette. SCOPE-H3 wins that score. It does not win neighbor function.
3. If both arms break R2, no explanation is unique and SCOPE-H3 does not keep the score exclusively.
4. SMC leave-one-out is reported. It does not kill H1. It is not compared against the jump 0.943 → 0.90 as if that jump were a threshold.
5. SERPINE2, INHA and PAPPA are not subtracted. If someone subtracts them and the rho changes, that change does not count.
6. None of this is an analyte or a neighbor function.

Evidence for. C, this project: empty intersection; Spearman 0.400 / 0.943 / 0.286; written sensitivity limited to ovary47. C, list: the three genes are in the 68. A, female primate, not senescence: Liu 2004. Method, not ovary: Ding 2020.

Evidence against. Neither arm has been run. R2 may survive, and then this primary dies. Three genes among 68 may not move a mean. With 4 donors, the window can leave the contrast not estimable, and not estimable is not a victory. Coppé predicts biological covariation of cytokines with proliferation: if T and F fail to break R2, that reading recovers the sign. The −0.007 intensity delta in stroma does not depend on this identity and is not used as proof of H1.

Kill criteria. Both arms remain preserved in the call with ovary47 and in the call without it: three estimable lineages, three rho > 0, median ≥ 0.30, SMC ≥ 0.30. H1 dies. The object passes to AR-H2. If one arm is not estimable and the other preserves R2, H1 stays inconclusive on the empty arm and the other is not identified by elimination. Exception frozen in the design: if F is broken, the cassette is enough to explain the score even if T is not estimable, because removing three genes is a fact of the score, not a residue of the window. If the subtraction is relaxed after the rho is seen, the call is invalid.

Status. proposed_primary. Maturity. H2 for the covariation and the empty intersection; H0 for which arm explains it. Confidence. 0.72 that “cell-cycle co-transcript” is an over-reading; 0.38 that T or F break R2. The design does not update those figures: the experiment was not run.

Lineage. HG-H1 v1 weakened; CO-C2; CO-C3; AR-C1; AR-C2; AR-C3; Liu 2004; Ding 2020 as a platform limit.

11. Competing hypothesis

L10-1-AR-H2 v1 — what would survive T and F is secretion from cells that are still cycling, not a p16 senotype

Role. Competing. Contrary to declaring that every residue is an artifact, and contrary to HG-H2, which named the emitter with the CDKN2A count. It is the Coppé reading that the hypothesis stage had absorbed inside the word “cycle”. It is not scored if H1 wins.

Falsifiable statement. If Q0-R2-IDENTITY preserves R2, the difference of the module between CDKN1A+ nuclei and their CDKN1A− sisters of the same lineage and the same donor is restricted to nuclei with a high cycle score or MKI67 > 0, and disappears, inside the design rule, in nuclei with a low cycle score and MKI67 = 0. That cycling cell is not promoted to a senotype. The CDKN2A+ minority does not enter this statement.

Mechanism. In human fibroblast, damage pushes cytokines while the cell is still proliferating, and the CDK inhibitor does not install the SASP. If the ovarian score survives depth and the cassette, the place to look for it is the proliferation gate, not the p16 cell. Saul requires Ki67-negative and out of S phase before calling a senotype: this cell does not meet that definition. SERPINE1 stays out. The locus without isoform stays out.

Predictions.

1. After T and F, the filter empties the CDKN1A+ versus CDKN1A− difference in at least two non-epithelial and non-immune lineages. If the difference remains at low cycle, with the same sign in two donors and two lineages, H2 dies and a residue candidate reappears. That candidate is not yet a senotype: the protein is missing and the function is missing.
2. The cycling cell does not change a neighbor function above its proliferative CDKN1A− sister. If it does, the object is a secretome of a proliferating cell, still without durable senescence, and it requires the factorial. If it does not, the RNA was not function.
3. A protein class that is p16INK4a-exclusive, p14ARF-negative, with its own analyte distinct from the cassette, revives weakened HG-H2. It neither saves nor kills AR-H2 by itself.
4. The epithelium stays excluded. It is not turned into the emitter because it is the compartment that increases.

Evidence for. V, fibroblast: Coppé. C, other tissues: Saul does not score SenMayo in S phase or Ki67+. C, this project: R2 exists, and H2 inherits it only if T and F do not break it.

Evidence against. It is conditional. The same Coppé shows that medium from cells arrested by p16 does not move epithelium: expressing cytokines is not being necessary for the neighbor. Luteal-exit p21 and AMH quiescence can explain a CDKN1A+ without a durable state. Stott forbids a failure of H2 from being reread as a victory of the CDKN2A minority.

Kill criteria. T or F break R2: H2 does not activate. After the filter, the difference remains at low cycle, same sign, two donors, two lineages of {GC, SMC, BEC}, in the report with ovary47 and without it: H2 dies. The only analyte that moves is SERPINE1, receiver p21 or a luteinization marker: it does not count. B does not abolish the function, or the isonutrient sham reproduces it.

Status. proposed_competing. Maturity. H1 of literature in fibroblast for the sign; H0 in ovary. The unfiltered covariation is computational H2 and is not yet this hypothesis. Confidence. 0.27, conditional on R2 surviving.

What remains below, and is not the live competitor

L10-1-HG-H2 v1 — the paracrine effect, if it exists, comes from a p16-exclusive cell that the p21 gate did not score — remains weakened. It revives only as p16INK4a protein, p14ARF-negative, p21-negative, Ki67-negative, not macrophage, not luteal cell, not epithelium, with an analyte that is not SERPINE1, PLAT, PLAU or SERPINE2. Coppé is against treating rarity as high secretion. Maruyama is against it if the p16+ cell in the section is the macrophage. CellChat downward forbids the field and does not refute the minority.

L10-1-SCOPE-H1 v1 and L10-1-SCOPE-H4 v1 remain at H0. The hiding place in the RNA CDKN2A+ minority was closed. Receiver p21 remains forbidden as the coprimary of order: the coprimary, if the factorial is ever opened, is a function that is not p21 or luteinization, and a durable arrest of the receiving cell (negative nucleoside in a fixed window plus an orthogonal readout), not a rise in CDKN1A.

12. Translational hypothesis

L10-1-AR-HT1 v1 — no parameter of this ovary enters an organ study before a local analyte that is not the cassette

Role. Experimental translational. It splits L10-1-HG-HT1 v1, which had put confidence 0.56 on an unmeasured mediation. It is not a clinical biomarker. It is not a product. HUMAN_QA_REQUIRED. It is not marked H5.

Falsifiable statement. In adult women, outside any user cohort, an index of CDKN1A fraction, a SenMayo score without a proliferation filter, a fused p16/p21, or an analyte reduced to SERPINE1/PLAT/PLAU, will not be accepted as a mediator of an event of a single organ. The only parameter that could reach that study is one that has survived Q0-R2-IDENTITY, the proliferation filter and a factorial in which the neighbor function is not p21 or luteinization. Until then, asserting that it “predicts worse than STRAW+10” is not a result.

Mechanism of the restriction. The arm that rises in the atlas is composition. The module that correlates with the cycle contains the ovulation program. Fusing it with a CDKN2A count without isoform mixes stage, follicle, detection and a locus. A distal organ is not mediated with that mixture. This is a rule. Mediation, as a fact, remains at H0.

Predictions.

1. Breaking the rule invalidates the study. An events study is not required for the rule to be in force.
2. If AR-H1 wins, there is no split parameter to send to mediation.
3. If AR-H2 wins only as RNA of cycling cells, that RNA is not promoted either: the neighbor function is missing.
4. If a p16INK4a-exclusive class survives and its analyte, distinct from the cassette, moves the function, the fused index stays forbidden and the split parameter can pass to human evaluation. This report does not mark it H5.
5. Fibrosis stays apart. It is not folded into HT1.

Evidence for. The fraction rises and the intensity of the module inside the positives does not. The cassette is inside the score. L9-4-K10: ANM is not longevity. Coppé: a p16 without SASP is not a counted secretion. Saul: two senotypes are not added. Neither of those two is ovary; both suffice to forbid the shortcut.

Evidence against. There is no mediation study that this hypothesis can win empirically. That is why confidence 0.56 was withdrawn. A future parameter that passes Q0 could deserve that study: HT1 does not forbid it forever; it forbids opening it now. A serum cytokine can beat any ovarian index for a systemic reason. That does not show that the ovary is irrelevant to another outcome, and it also does not authorize the index.

Kill criteria. Using the fused index, the unfiltered fraction or SERPINE1 as a mediator invalidates the study; it is not a biological refutation. If Q0 and the factorial deliver a split parameter that, in a study without Lua data, predicts an event of one organ better than stage and than systemic inflammation, with competing risks, and that parameter mediates the effect, then the form “no ovarian parameter will ever be able to” dies and passes to human evaluation. Using menopause, AMH, POI, litter, mortality or user data invalidates the result.

Status. parked_translational_test. Maturity. H2 for “the fraction is not the intensity” and for “the module contains the cassette”; H0 for mediation. Confidence. 0.74 in the rule not to freeze the index or the cassette; 0.31 that no split parameter ever reaches mediation. The second figure is low because it is not measured. Requirement: HUMAN_QA_REQUIRED.

13. Falsifiable predictions and kill criteria

Preserved, broken and not estimable are the only three outcomes of each arm. Preserved requires three estimable lineages (≥4 donors with ≥10 nuclei in that arm’s set), three rho > 0, median ≥ 0.30 and SMC ≥ 0.30. BEC does not have to exceed 0.30: in the original call it enters at 0.286 by sign, and that slack is kept. The median of three numbers is the middle one. It is not replaced by a mean or by an interval. With 4, 6 and 7 donors an interval does not distinguish 0.30 from 0.

rho_det — Spearman between the module mean and the nFeature mean — may be reported and does not vote. Promoting it to a third arm after seeing it invalidates the call. A zero sign in the proliferation filter does not count as persistence. An arm that is not estimable is not declared biological equivalence.

14. Discriminating experiment

The smallest experiment that can falsify the primary fits in the matrix already downloaded. It was not run in the design stage or in this report. L10-1-ED-C1.

Q0-R2-IDENTITY

Same matrix, same fingerprints, same donors, same lineages GC, SMC and BEC. ovary23 stays. ovary47 enters one call and leaves the other, without relabeling it as 49–54. Surgery versus autopsy is not stratified inside the rule: the frozen rule did not ask for it. The sex of the object is a woman’s ovary; the result is not transported to mouse or to man.

Arm T. Inside each lineage, on donors that already have ≥10 CDKN1A+ nuclei: 10th percentile and 90th percentile of nFeature (quantile type 7). nFeature is the number of genes, among the 27 315, with raw count > 0. It is not the number of non-zeros inside the 68. The shared window is the interval between the maximum of the 10th percentiles and the minimum of the 90th percentiles. It is written to disk, with the per-donor count, before cor. The call without ovary47 recomputes the window from scratch. A donor stays inside if ≥10 of its nuclei fall in the window.

Why not each donor’s IQR. Retaining the central half of 11 nuclei (SMC of ovary47) leaves 5.5 in expectation. Half of 18 (GC of ovary51) leaves 9. Both fall below the floor of 10 before a rho is looked at. The 10th–90th percentile is the narrowest central band that, in that row of 18, still leaves ≥10 in expectation (0.8 × 18 = 14.4; 0.5 × 18 = 9). For the 11 nuclei, 0.8 × 11 = 8.8: they stay inside only if they fall in the range of the others. Losing ovary47 does not empty SMC (five rows remain). Losing the granulosa of 18 does empty the lineage, because there is no spare donor, and then T is not estimable. The window is not widened after the rho is seen or after seeing that it came out empty.

Arm F. PLAT, PLAU and SERPINE1 are removed. The score is the mean over the remaining 65 genes. Library size is not recomputed. Setting the three genes to zero and continuing to divide by 68 is forbidden: it lowers the score of whoever expresses them and changes the denominator of whoever does not. F does not drop nuclei. With the rows already usable it is estimable (SMC 6 and 5). T and F are not combined into a single arm.

Controls that do not vote. SMC leave-one-out on the 68-gene score and on each preserved arm. rho_det. Input fingerprints: if one does not match, the turn stops.

What does not open if H1 does not die

Q0-FILTER only after the predicate has killed H1. Score of 65 genes, not of 68. High gate: cycle score ≥ the median of the CDKN1A+ of that donor and lineage, or MKI67 > 0. Low gate: cycle score below that median and MKI67 = 0. Delta against the CDKN1A− sister. The donor votes if each class has ≥10 nuclei. A lineage is estimable with at least two donors. AR-H2 dies if the low-gate delta keeps the same sign in two donors and two lineages, with and without ovary47. If the low gate does not assemble two donors, the statement is inconclusive. It is not read as “then they are cycling cells”.

Q0-PROTEIN is not opened to confirm that R2 was cycle. It opens if the filter left a low-cycle residue, or if a decision is made to spend section in an attempt to revive HG-H2. Population: adult women, non-tumor ovary, donor = n, two donors with the same call. STRAW+10 is recorded if the chart carries it; if it is absent, the result is not read as stage. It is not Wu’s set. They are not Lua users. Assay: immunofluorescence on the same section, p16INK4a (exon 1α, without recognizing p14ARF), p14ARF (exon 1β), p21, Ki67, plus markers to exclude macrophage, luteal cell and epithelium. This report does not name a clone. A clone enters only with a selectivity citation read before counting. If both locus antibodies light up in the same control, the run is invalid. Majority only with ≥20 p16INK4a+ cells and the lower bound of the two-sided 95% Clopper–Pearson interval > 0.5. In the design it was calculated that 15/20 gives 0.509 and 14/20 gives 0.457. That count is the 0.025 quantile of Beta(k, n−k+1), not an ovarian frequency. The 0.43% of nuclei is not converted into “one must look at 4 650 nuclei”.

Q0-SECRETOME and the factorial D0-SASP-ORDER stay closed until there is a class and an analyte that is not the cassette, SERPINE2, INHA or PAPPA. Arms, when they exist: A, cell present and secretion intact; B, secretion interrupted and number of emitters matched; C, factor returned without the cell; isonutrient sham; clamp of E2 and of P4. The concrete assay of the neighbor function is not named until the lineage exists, and changing it is forbidden if only p21 moves. Absence of the factorial is not equivalence of the analyte.

There is no power calculation with a variance this ovary has not measured. Inventing an n to detect rho = 0.3 would be a false number. The n of the identity is already fixed: 4, 6 and 7 donors. A nucleus is not a replicate. Two donors in the protein is the replication criterion the project already uses, not a prevalence. The factorial, if it arrives, uses donor or female as n, two units, the same sign, not the same batch of medium.

15. Biomarkers and stratification

No analyte of this project is promoted to a biomarker. The word stays forbidden for the following objects, which are proxies of something else:

The only stratification the design authorizes today is internal to the experiment: donor, lineage (GC, SMC, BEC), proliferation gate, and the pair of calls with and without ovary47. It is not a patient stratification. There is no indication.

16. Individual variability

Eight ovaries are not the transition of women. The variability this project can name, and that a young/old average erases, is this:

• Provenance. ovary23 is surgery with the diagnosis ST1 records. The other young donors in the supplement, and the aged donors of 49–54, are autopsies in the table. The contrast mixes indication and age. ovary23 is not removed after the fact: the rule did not ask for it. Calling the contrast “normal ovary of the transition” is forbidden.
• Age token. ovary47 carries 4 724 nuclei inside group=old and has no row in ST1. A result that lives only in that token is not published as 49–54 years. The R2 call already made did not depend on it. The future T window is rewritten without it, from scratch.
• Leverage. SMC holds the 0.943 with 6 donors, and one of those rows has 11 nuclei. Granulosa has 4 donors, and one of the usable aged rows has 18. A single point can move a Spearman of four ranks. That is why the outcome is a rule, not a population parameter, and why leave-one-out is reported without voting.
• Ovulatory history, in the mouse. Maruyama: more CDKN2A by western in breeders than in virgins, n=3, similar age. Isola uses virgins and does not model that effect. Chronological age is not the estimand if reproductive history moves the marker.
• Stage not measured. No identity primary brings STRAW+10. Human Meinsohn is fertility-preservation cortex, prepubertal and postpubertal, n=3 samples, not n of adult women. Wu brings age bands and group AMH (3.76, 1.35 and <0.06 ng/mL cited by the authors) in ovaries from oncologic surgery. Those bands are not reclassified as the natural transition.
• Isoform and dropout. A zero in snRNA is not absence of protein. Even if the UMI were perfect, CDKN2A does not name p16. The 159/184 disjunction fits Saul’s pattern (rare doubles, in other tissues) and fits a technical defect. Both readings stay open until the stain, and the stain is not the first spend.
• Sex. Jin’s atlas is a woman’s ovary. Saul does not include ovary, and the sex of those mice was not extracted: it is neither corrected nor invented. Coppé is not ovary. Liu is a female monkey in induced ovulation. None of those sources transports a senotype to the postmenopausal woman.

Variability is not filled with user data. 10.6 is not opened from here.

17. Pharma relevance and maturity

There is no readiness. There is no nominated target. There is no modality. There is no indication. There is no patient stratification that this report authorizes. Maturity does not reach H5. HUMAN_QA_REQUIRED before any partnership conversation. L10-1-ED-C6.

The opportunity is negative, and it is the operation of AR-HT1, not a development program:

1. The only molecular handle the module already contains, and that someone might want to turn into an asset, is the cassette PLAT / PLAU / SERPINE1. Liu 2004 measured it in induced ovulation of the monkey, with granulosa tPA and theca PAI-1, not in senescence and not in the postmenopausal woman. Using it as an analyte of senescent burden would aim at the follicular-rupture program. It stays forbidden. That is not an evaluation of a compound, and it does not open 10.2–10.5.
2. If T breaks R2, the handle was detection depth. There is no asset.
3. If H1 dies and AR-H2 holds, the RNA lives in cells that are still cycling. A modality that kills arrested cells — the premise of a senolytic — is not the object of that statement. This report does not rank senolytics. It records the modality mismatch.
4. If exclusive p16INK4a protein, p14ARF-negative, revives, the target remains intracellular and the secretome remains unmeasured. It is not a secretion asset. The count is not fused with p21 or with the CDKN1A fraction.
5. Fibrosis, if it is ever pursued, is not a senolytic. It is MMP2 and deposited collagen, with IL-6 and CD90+ matched. It is not ready either.
6. What would have to exist before a human evaluation is a per-cell analyte, outside the cassette and outside SERPINE2 / INHA / PAPPA, necessary and sufficient for a neighbor function that is not p21, in a cell that is not luteal, immune, atretic or cycling epithelium. That is H0.

18. Limitations

• T and F are not run. The report freezes a protocol. It does not report a new rho. If the nFeature values do not overlap, T comes out not estimable, and that was not measured.
• The 10th–90th percentile can leave granulosa below four donors. The protocol refuses to widen the window to avoid that. That outcome is “not estimable”, not a demonstrated technical identity.
• rho_det can show the confounder and, by rule, cannot decide.
• Liu 2004 is induced ovulation in the monkey, not human aging. Animal n was not extracted. A fold of tPA or of PAI-1 is not asserted.
• Stott 1998 is tumor lines. p14ARF → p21 there does not predict the ovary; it only forbids reading the symbol as p16.
• Ding 2020 is not ovary and does not measure the 68 genes.
• Saul 2025 does not include ovary. Skin, lung, brain, muscle, bone and liver do not transport the senotype by analogy. The skin immunofluorescence does not bring an overlap percentage extracted in this project.
• Coppé is human fibroblast, ectopic p16 and p21, not the endogenous cell of the ovary.
• Du 2026 was not reanalyzed number by number. The PMID was not located. Folds that were not read are not cited.
• The SMC rho 0.943 rests on 6 donors. Granulosa has 4. BEC does not reach 0.30.
• The young CDKN1A fraction in this matrix is 0.091, not the article’s ~0.0949. The threshold was not reconciled.
• There is no stereology, no admissible conditioned medium, no STRAW+10 in the atlas. The interval 15/20 = 0.509 is not a frequency observed in ovary.
• No antibody clone is named, and no concrete neighbor function is named. Without a cited selectivity the count is invalid. Without a surviving lineage the neighbor outcome would be a choice with the catalog open.
• Confidence is a range from the review, not a posterior probability.
• It is not asserted that removing a cell, raising or lowering a cytokine, or moving the age at menopause lengthens women’s lives.

19. Conclusions

1. The title of L10-1 splits three objects. No admissible experiment, human or female, shows that the per-cell secretome of an ovarian lineage with a stable senescent state is necessary and sufficient for a neighbor function. Scope H1 and H4 remain at H0 because of that gap.
2. The visible accumulation is not that state. In the woman, the epithelium rises in proportion and in cell-cycle genes; CDKN1A rises as a fraction; CDKN2A is rare. In the mouse, before estropause, what doubles is the immune fraction, and the absolute that flow confirms is CD90+. p21 marks luteal exit and an AMH-reversible quiescence. Maruyama’s “senescent” stroma is already mixed with macrophage and atresia.
3. The RNA this project scored inside CDKN1A+ nuclei is not a Saul senotype and not a cell-cycle co-transcript. It is a covariation of per-donor means (0.400, 0.943, 0.286) of a module of 68 genes that does not intersect the cycle set and that contains the monkey’s periovulatory cassette. The p16-exclusive cell is not named by that count.
4. The live primary is L10-1-AR-H1 v1. The discriminator is Q0-R2-IDENTITY, with the 10th–90th percentile window written before the Spearman and with the mean over 65 genes. H1 dies only if T and F preserve R2 with and without ovary47. If F breaks R2, the cassette suffices for the score and SCOPE-H3 does not win neighbor function. Tissue is not opened except by that death of H1.
5. The live competitor, L10-1-AR-H2 v1, is conditional: secretion from cells that are still cycling, not a Ki67-negative senotype and not the CDKN2A minority. The translational hypothesis, L10-1-AR-HT1 v1, forbids fusing fraction, SenMayo, p16/p21 and PAI-1 into an organ mediator. HUMAN_QA_REQUIRED. There is no Pharma target. There is no H5.
6. The bridge to women’s health and longevity remains at H0. An ovary–SASP–longevity claim would be transported only if it survived the identity of R2, the proliferation filter, the protein disjunction of isoforms, a secretome factorial on a function that is not p21, and mediation to an event of one organ with competing risks. None of that is measured. The first spend is not to look for that cell in a section. It is to stop the current score from being read as if it already were her.

20. References

1. Jin C, Wang X, Yang J, et al. (2025). Molecular and genetic insights into human ovarian aging from single-nuclei multi-omics analyses. Nature Aging. DOI 10.1038/s43587-024-00762-5. PMID 39578560. PMC PMC11839473. snRNA and snATAC of the same ovary; 4 donors aged 23–29 years and 4 aged 49–54; 42 568 RNA nuclei. CDKN2A ~0.43%; CDKN1A ~9.49% → ~15.56%. The authors require experimental validation of senescence. Data: GSE202601.
2. Isola JVV, Ocañas SR, Hubbart CR, et al. (2024). A single-cell atlas of the aging mouse ovary. Nature Aging. DOI 10.1038/s43587-023-00552-5. PMID 38200272. PMC PMC10798902. Virgin C57BL/6J, 3 vs 9 months. The immune proportion doubles. Flow: number and percentage of CD90+ rise; B and myeloid do not. Deposited collagen rises; MMP2 falls.
3. Maruyama N, Fukunaga I, Kogo T, et al. (2023). Accumulation of senescent cells in the stroma of aged mouse ovary. Journal of Reproduction and Development. DOI 10.1262/jrd.2023-021. PMID 37926520. C57BL/6N, 2–4 vs 8–10 months. SASP qPCR n=3, breeders. CDKN2A next to multinucleated macrophages; SA-β-gal in probably atretic follicles.
4. Ansere VA, Ali-Mondal S, Sathiaseelan R, et al. (2021). Cellular hallmarks of aging emerge in the ovary prior to primordial follicle depletion. Mechanisms of Ageing and Development. DOI 10.1016/j.mad.2020.111425. PMID 33383072. PMC PMC8279026. qPCR n=10/age; lipofuscin n=3; follicles n=5. No genome-wide methylation change.
5. Meinsohn MC, Cleverdon J, Mermin-Bunnell AM, et al. (2025). Anti-Müllerian hormone induces a reversible quiescence-associated secretory phenotype in preantral follicles. Biology of Reproduction. DOI 10.1093/biolre/ioaf230. PMID 41074767. PMC PMC12643070. Reversible QASP, no differential β-gal. Human cortex ex vivo, 4 days, 10 μg/mL, qPCR n=3 samples. It is not STRAW+10.
6. Wu M, Tang W, Chen Y, et al. (2024). Spatiotemporal transcriptomic changes of human ovarian aging and the regulatory role of FOXP1. Nature Aging. DOI 10.1038/s43587-024-00607-1. PMID 38594460. PMC PMC11031396. scRNA of 9 women and spatial of 15, surgery for uterine or cervical cancer. Outside the natural estimand.
7. Battaglia R, Musumeci P, Ragusa M, et al. (2020). Ovarian aging increases small extracellular vesicle CD81+ release in human follicular fluid and influences miRNA profiles. Aging. DOI 10.18632/aging.103441. PMID 32554857. 25 women >38 and 25 <35 years, ICSI for male factor. Vesicle class, not SASP.
8. Gao Y, Wu T, Tang X, et al. (2023). Increased cellular senescence in doxorubicin-induced murine ovarian injury: effect of senolytics. GeroScience. DOI 10.1007/s11357-023-00728-2. PMID 36648735. Females of 8 weeks, 10 mg/kg. Marker–structure dissociation: Cdkn2a, Cdkn1a, Il6 and Mcp1 fall; fibrosis and γH2AX are not reversed. Does not rank drugs. It is not the transition.
9. Jirawatnotai S, Moons DS, Stocco CO, et al. (2003). The cyclin-dependent kinase inhibitors p27Kip1 and p21Cip1 cooperate to restrict proliferative life span in differentiating ovarian cells. Journal of Biological Chemistry. DOI 10.1074/jbc.M301206200. PMID 12609976. Double null: proliferation maintained and steroidogenic enzymes retained.
10. Acosta JC, Banito A, Wuestefeld T, et al. (2013). A complex secretory program orchestrated by the inflammasome controls paracrine senescence. Nature Cell Biology. DOI 10.1038/ncb2784. PMID 23770676. Fibroblast OIS, not ovary. IL-1 and TGFβ can induce p21 in the neighbor. That is why receiver p21 is not the ovarian coprimary.
11. Coppé JP, Rodier F, Patil CK, Freund A, Desprez PY, Campisi J. (2011). Tumor suppressor and aging biomarker p16INK4a induces cellular senescence without the associated inflammatory secretory phenotype. Journal of Biological Chemistry. DOI 10.1074/jbc.M111.257071. PMID 21880712. PMC PMC3196093. Human fibroblast. Ectopic p16 and p21 arrest without SASP.
12. Saul D, Doolittle ML, et al., Passos JF. (2025). Distinct senotypes in p16- and p21-positive cells across human and mouse aging tissues. The EMBO Journal 44:7295–7325. DOI 10.1038/s44318-025-00601-2. PMID 41162753. PMC PMC12669595. Ki67 / CD45-high / S-phase filter. The ovary is not there.
13. Saul D, Kosinsky RL, Atkinson EJ, et al. (2022). A new gene set identifies senescent cells and predicts senescence-associated pathways across tissues. Nature Communications. DOI 10.1038/s41467-022-32552-1. PMID 35974106. SenMayo. List from which the 68 genes come. It is not ovary.
14. Liu YX, Liu K, Feng Q, Hu ZY, et al. (2004). Tissue-type plasminogen activator and its inhibitor plasminogen activator inhibitor type 1 are coordinately expressed during ovulation in the rhesus monkey. Endocrinology 145:1767–1775. DOI 10.1210/en.2003-1327. PMID 14701676. Female rhesus monkey, induced ovulation. tPA, uPA, PAI-1. It is not senescence. The n was not extracted.
15. Stott FJ, Bates S, James MC, et al. (1998). The alternative product from the human CDKN2A locus, p14ARF, participates in a regulatory feedback loop with p53 and MDM2. The EMBO Journal 17:5001–5014. DOI 10.1093/emboj/17.17.5001. PMID 9724636. Human tumor lines. p16INK4a and p14ARF. It is not ovary.
16. Ding J, Adiconis X, Simmons SK, et al. (2020). Systematic comparison of single-cell and single-nucleus RNA-sequencing methods. Nature Biotechnology 38:737–746. DOI 10.1038/s41587-020-0465-8. PMID 32341560. Lines, PBMC and mouse cortex. Platform limit. It is not ovary.
17. Du Y, Zhou C, Tian Y, Li Z, Huang X. (2026). Single-cell atlas reveals a Wilms' tumor 1-mediated axis driving granulosa cell senescence in human ovarian aging. International Journal of Molecular Medicine. Online 3 July 2026. DOI 10.3892/ijmm.2026.5916. Reanalysis of GSE202601. 12 mice of 3–4 months and 12 of 14–18 months. It is not the factorial of 10.1. PMID not located in this project.
18. Lori G, Mancini C, Paffetti C, et al. (2026). Senescent stroma-derived glutamine: a driver of aggressiveness in prostate and ovarian cancer cells. Cells. DOI 10.3390/cells15090770. PMID 42121871. Therapy-induced senescence and SKOV3 invasion. Rejected for paracrine action of the natural estimand. Not used to rank glutamine.

Laboratory memory cited as a restriction, not re-audited number by number in this report: L7-1 (luteal architecture separated from reserve and from acquired enzymatic injury); L7-4 (loss of progesterone exposure without a P4→E2 change-point); L9-4-K4 (captured proportion is not activity per cell); L9-4-K9 (a senolytic in 15 females is not life equivalence); L9-4-K10 (late menopause is not a uniform lifespan). Mei et al. 2024, cited in L5.5, was not re-audited and does not enter.

Claims in this report

• L10-1-RES-C1 — The report freezes that accumulation, SASP and paracrine action are not one measured object. The CDKN1A fraction goes from 0.091 to 0.156 and the intensity of the module of 68 genes inside CDKN1A+ SC does not rise (delta −0.007). The module intersects the cycle set in the empty set and contains PLAT, PLAU and SERPINE1.
• L10-1-RES-C2 — The three live hypotheses are L10-1-AR-H1 v1 (covariation without identity), L10-1-AR-H2 v1 (secretion from cells that are still cycling, conditional) and L10-1-AR-HT1 v1 (no parameter enters an organ before a local analyte that is not the cassette). HG-H1 as co-transcript and HG-H2 as an RNA p16 cell remain weakened.
• L10-1-RES-C3 — Q0-R2-IDENTITY is specified and not executed. H1 dies only if T and F remain preserved with and without ovary47. If F breaks, SCOPE-H3 wins the score and not the function. Neither the death nor the survival of H1 opens tissue, a senolytic or mediation.
• L10-1-RES-C4 — The Pharma evaluation is negative. There is no target and no H5. SERPINE1 is an anti-analyte. HUMAN_QA_REQUIRED. U1 remains unmeasured. The bridge to women’s longevity remains at H0.

Nothing is written into memoria agentes/Lua Labs/reportes/. Laboratory memory is not consolidated at this stage. Nothing is prescribed.