2026-05-31

2026-05-31 — ad-hoc daily log

[2026-05-31] ingest | fisetin formulation & bioavailability (user-Q-driven)

Context: user asked to rebut an online claim (“high-dose liposomal fisetin every 3 months crushes most senescent cells”). Wiki coverage of fisetin/senolytics/senescence was deep and verified, but the page had no nano/lipid formulation bioavailability content (flagged under existing dose-response-unclear). Filled that gap.

• updated: molecules/compounds/fisetin.md — added ”# Formulation & bioavailability enhancement” subsection under Pharmacokinetics.

  • Animal-formulation PK table from Szymczak & Cielecka-Piontek 2023 review (Int J Mol Sci 24(18):14158): nanocochleates ~141-fold, nanoemulsion 11.92-fold AUC i.p., Pluronic-F127 micelles 6.3-fold AUC, DOPC/DODA-PEG liposomes 4.43-fold AUC i.p., SNEDDS 3.7-fold Cmax.
  • Only human PK study: Krishnakumar 2022 (J Nutr Sci 11:e74; PMID 36304817) — FF-20 micelle-in-fenugreek-galactomannan hydrogel (NOT a liposome), n=15 healthy volunteers single-dose crossover, AUC +26.9-fold, Cmax 238.2 vs 9.97 ng/mL. PK-only, no senolytic/efficacy endpoint.
  • Key analytic point recorded: best-in-human Cmax 238.2 ng/mL ≈ 0.83 µM, ~24-fold below the 20 µM ex-vivo human senolytic concentration (Yousefzadeh 2018). Consumer “liposomal fisetin” ≠ validated FF-20; no published human senolytic PK for it.
  • Added footnotes [^szymczak2023] and [^krishnakumar2022] (both abstract/results-table verification only, full PDFs not end-to-end verified).
  • verified-scope updated to record the addition; literature-checked-through left at 2026-05-08 (this was a targeted formulation-topic search, not a full recency refresh).

• gaps surfaced:

  • dose-response-unclear reinforced — no fisetin formulation study has measured a senolytic/clinical endpoint in humans; whether any oral formulation reaches senolytic tissue concentrations in humans is unresolved.
  • Verifier follow-up: ✓ completed — see verification entry below.
  • Possible new candidate page: FF-20 / Hybrid-FENUMAT delivery system if it recurs across compound pages.

[2026-05-31] verify | fisetin.md — Formulation & bioavailability enhancement subsection

Pages verified (targeted, subsection-only): molecules/compounds/fisetin.md (formulation subsection + two footnotes only; remainder of page previously verified 2026-05-08).

Sources checked:

• doi:10.1017/jns.2022.72 (Krishnakumar 2022) — full PDF read end-to-end. OA via PMC9574875.
• doi:10.3390/ijms241814158 (Szymczak 2023) — full PDF read end-to-end (25 pp). OA via PMC10532335.

Corrections made:

• Cmax fold-gain in body text: “~24-fold” → “23.9-fold” (paper’s Discussion states 23.9-fold; abstract says “more than twenty-three times”; 24 was a slight over-rounding)
• AUC units in body text: added “ng·h/mL” for precision (was bare numbers)
• Krishnakumar footnote: added ± SD values from Table 1; corrected fold from “~24-fold” to “23.9-fold”; added demographic detail (12M/3F, 22–55 yr, BMI 18–25); added CoI statement (Akay Natural Ingredients funded; Life Extension co-author sells FF-20 as “Bio-Fisetin”); upgraded verification tag to full-PDF-verified 2026-05-31
• Szymczak footnote: added route/species precision for each formulation entry; added internal-inconsistency flag: review body text (§4.5.4) reports nanocochleate 141-fold but review’s own Table 4 reports 13-fold for the same primary source (Bothiraja 2014 ref [97]) — wiki uses body-text figure; tagged needs-replication; upgraded verification tag to full-PDF-verified 2026-05-31
• verified-scope updated to reflect full-PDF verification of formulation subsection

Unverifiable claims: nanocochleate 141-fold figure cannot be confirmed without Bothiraja 2014 primary source (doi:10.1517/17425247.2014.848184 — not in archive, not checked for OA status).

No supersession check run (this is a targeted subsection verify, not a full-page recency refresh; literature-checked-through remains 2026-05-08).

Downstream propagation: none needed — formulation subsection is new content with no downstream citations yet.

[2026-05-31] ingest | seed senescence-escape + senescence-immune-surveillance (gap fill from fisetin Q&A)

Two wiki gaps surfaced during the fisetin/senolytic rebuttal were seeded as new type: process atomic pages (ad-hoc user-requested; NOT added to ROADMAP).

• added: processes/senescence-escape.md (verified:false, banner) — conditional reversibility of senescence: Beauséjour 2003 p53/p21-vs-p16/Rb framework, therapy-induced-senescence (TIS) escape + stemness (Milanovic 2018), polyploid giant cells, OIS-barrier escape, sublethal-senolytic escape concern (flagged theoretical). 12 citations, all DOI-resolved via PubMed eutils; zero local PDFs (all abstract-level). literature-checked-through 2026-05-31 (R25 recency run).
• added: processes/senescence-immune-surveillance.md (verified:false, banner) — canonical home for NK/NKG2D recognition of senescent cells: MICA/MICB/ULBP ligands, DDR/p53-driven ligand induction, NK granule exocytosis, CD4-T OIS-hepatocyte surveillance (Kang 2011), macrophage efferocytosis, immune evasion (HLA-E/NKG2A — Merkt 2026; MICA shedding via ADAM10 — Zingoni 2015), age-related decline, immune-vs-pharmacological-senolysis contrast table, engineered surveillance (uPAR CAR-T Amor 2020/2024). 13 citations; Xue 2007 + Amor 2020/2024 have local PDFs. literature-checked-through 2026-05-31 (R25 recency run).
• frontmatter fixes (main agent): senescence-immune-surveillance causes: invented node [[senescent-cell-accumulation-decline]] → [[chronic-inflammation]]. ([[wnt-signaling]] on escape page confirmed valid — registered alias of [[wnt-beta-catenin]].)
• propagation (inbound links added so new pages aren’t orphans; links only, no claims altered, verification integrity preserved):
  • hallmarks/cellular-senescence.md → senescence-escape (arrest-reversibility note at definition) + senescence-immune-surveillance (×2: wound-clearance line + aging-cancer-paradox line)
  • hallmarks/disabled-adaptive-immunity.md → senescence-immune-surveillance (failed-clearance bullet)
  • interventions/pharmacological/senolytics.md → both (hit-and-run row: sublethal-escape risk + endogenous-surveillance alternative)
  • acronyms.md → NK + NKG2D entries added (NKG2D → senescence-immune-surveillance)
• gaps surfaced / follow-ups:
  • Both pages verified:false — verifier follow-up queued: foundational mouse papers (Beauséjour 2003; Krizhanovsky 2008, Kang 2011, Iannello 2013, Sagiv 2016 quantitative claims) need full-PDF cross-check; escape page has NO local PDFs (priority).
  • New protein stubs implied: nkg2d, mica, micb, ulbp1, perforin, granzyme-b, upar, rb (RB1), cdkn2a — future seeding targets (lint will surface).
  • druggability-tier:2 on both pages flagged as judgment calls (escape: MDM2/CDK4-6 probes; surveillance: NKG2A/monalizumab) — review on verification.

[2026-05-31] verify | processes/senescence-immune-surveillance

• Page verified: processes/senescence-immune-surveillance.md → verified: true
• Sources checked (10 PDFs read end-to-end): Xue 2007, Amor 2020, Amor 2024, Sagiv 2013, Sagiv 2016, Krizhanovsky 2008, Iannello 2013, Soriani 2009, Tarazona 2017, Zingoni 2015 — all LOCAL PDF available post-download
• Sources unverifiable: Kang 2011 (not_oa, closed-access); Merkt 2026 (not in archive, PMID 42127218) — both tagged no-fulltext-access

Corrections applied (8):

1. Sagiv 2013 — granzyme-B-KO mice fabricated: “perforin-KO or granzyme-B-KO mice” → “perforin-KO (Prf⁻/⁻) mice (60% more scarring, p<0.0001)”; granzyme-B inhibitor 3,4-DCI was in vitro only; no GzmB-KO mouse experiment in this paper.
2. Krizhanovsky 2008 — wrong mouse genotype: “p21-null mice” → “p53⁻/⁻ or INK4a/ARF⁻/⁻ mice”; paper does not use p21-null; NK depletion method was anti-AsiaGM1 antibody (added to footnote).
3. Iannello 2013 — “blocked” overstates: “CCL2 neutralization blocked NK recruitment and tumor regression” → “significantly delayed tumor elimination (p<0.001) and reduced NK recruitment”; eventual clearance still occurred; n=10–11/group added.
4. Sagiv 2016 — ERK/MICA overclaim: “ERK maintains their sustained surface expression” → ULBP2 is ERK-dependent (>50% reduction with MEKi); MICA in established senescent cells is ERK-independent (ATM inhibition further increases MICA); body and footnote corrected. NKG2D-KO fibrosis quantified: 50% more area (p<0.05) + 45% more SA-β-gal+ cells (p<0.05).
5. Amor 2020 — wrong tumor model: “lung adenocarcinoma mice” → primary survival endpoint was NRAS-driven liver tumor model (n=11/group); KP lung adenocarcinoma was a secondary validation model. CCl4 fibrosis stats added (Sirius Red p<0.0001, SA-β-gal p=0.0002).
6. Amor 2024 — quantitative numbers added: specific p-values for glucose tolerance (p=0.032), exercise duration (p=0.043), max speed (p=0.046), prophylactic sustained benefit (p=0.0026 at 15 months); dose (0.5×10⁶ cells); “proof-of-principle” qualifier from authors added.
7. Tarazona 2017 — NKG2D age-decline contradicted: NKG2D expression is reported as preserved in the elderly by this review; NKp30/NKp46/DNAM-1 decline (not NKG2D). The wiki claim that NKG2D declines is not supported by Tarazona 2017; corrected with NKG2D-preserved note and gap tag for better citation.
8. Zingoni 2015 — ADAM10 specificity: “ADAM10/ADAM17” → ADAM10 is the predominant drug-induced sheddase; ADAM17 plays minor constitutive role; ADAM10 knockdown blocked drug-induced MICB release while shADAM17 did not; body and footnote corrected.
9. Soriani 2009 footnote — design correction: “in-vitro + in-vivo” → “in-vitro + ex-vivo (patient plasma cell samples)”; no mouse in-vivo experiments in this paper.

druggability-tier:2 assessment confirmed: NKG2A/monalizumab is an advanced oncology probe (Phase 1/2 clinical trials active per Tarazona 2017 review) with no aging-validated trial. Tier 2 (high-quality probe, not aging-validated) is appropriate per CLAUDE.md aging-context convention.

Supersession check (R25): 5 recent PubMed hits reviewed — none are meta-analyses or large RCTs on senescence immune surveillance. Merkt 2026 (PMID 42127218) is a mechanistic extension already cited on the page. No supersession detected. literature-checked-through: 2026-05-31 already set.

Downstream propagation candidates: Claims from Sagiv 2013 (granzyme-B-KO correction), Krizhanovsky 2008 (p21→p53/INK4a correction), and Amor 2020 (tumor model correction) may appear on other pages citing these papers. Main agent should check: hallmarks/cellular-senescence.md, interventions/pharmacological/senolytics.md, processes/sasp.md for any inherited errors.

[2026-05-31] verify | processes/senescence-escape.md

Page verified: processes/senescence-escape.md → verified: true (with partial scope for 4 not_oa sources)

Sources checked:

• doi:10.1093/emboj/cdg417 (Beauséjour 2003) — full PDF (11pp, EMBO J Vol.22:4212–4222). Downloaded from PMC175806. VERIFIED.
• doi:10.1186/s12943-025-02310-0 (Bajtai 2025) — full PDF (31pp, Mol Cancer 24:128). Already in archive. VERIFIED.
• doi:10.1038/s41586-024-08317-9 (Gu 2025) — full PDF (10pp, Nature 637:461–469 author manuscript). Already in archive. VERIFIED.
• doi:10.1038/nature16932 (Baker 2016) — full PDF (5pp, Nature 530:184–189 author manuscript). Already in archive. VERIFIED.
• doi:10.3390/cancers12040822 (Saleh 2020) — full PDF (38pp, Cancers 12(4):822). Downloaded from PMC7226427. VERIFIED.
• doi:10.1080/15384101.2024.2364579 (Saleh 2024) — full PDF (6pp, Cell Cycle 23(6):713–721). Downloaded from PMC11229739. VERIFIED.
• doi:10.18632/aging.203495 (Pacifico 2021) — full PDF (22pp, Aging Albany NY 13(17):20962). Downloaded from PMC8457561. VERIFIED.
• doi:10.3389/fonc.2021.724781 (Song 2021) — full PDF (5pp, Front Oncol 11:724781). Downloaded from PMC8435787. VERIFIED.
• doi:10.1038/nature25167 (Milanovic 2018) — not_oa. Abstract only (Europe PMC). no-fulltext-access
• doi:10.1038/s41416-022-01787-6 (Saleh 2022) — download failed. Abstract only. no-fulltext-access
• doi:10.1016/j.semcancer.2020.11.015 (Sikora 2022) — not_oa. Abstract only. no-fulltext-access
• doi:10.1016/j.bcp.2022.115159 (Pacifico 2022) — not_oa. Abstract only (efetch confirmed). no-fulltext-access

Corrections applied:

1. Beauséjour 2003 — foundational framing corrected (substantive): Wiki said “high p16 cells re-entered cell cycle without net growth after Rb inactivation” — this was partially inaccurate. Corrected to: pRb-pathway inactivation (LgT or CDK4m) stimulates DNA synthesis in 35–70% of S-WI cells but NO net proliferation; even combined p53+pRb inactivation fails to restore growth in p16-high cells. Critical mechanism added: p16 shRNA knockdown in S-WI converts them to BJ-like phenotype where p53 inactivation (GSE-22) then fully restores growth (>70% %LN, robust proliferation). Quantitative data added: GSE-22/LgT → 70–90% %LN + >20 PDs (S-BJ); LgT → 60–70% %LN, CDK4m → 35–40% %LN, no growth (S-WI). Cell line identities added (WI-38 = fetal lung; BJ = neonatal foreskin).

2. Beauséjour 2003 — gap tag corrected: “p16 threshold above which Rb inactivation alone fails” → “combined p53+pRb inactivation fails” — the seeder had framed the gap slightly wrong.

3. Bajtai 2025 — drug resistance denominator specified: “roughly half of a panel” → “23 of 46 active compounds (of 63 total screened; 17 were non-toxic at tested concentrations), resistance defined as in ≥3/4 cell lines.” Added DOX survival rates (0.29–8.2%), bulk RNA-seq quantitation (929 DEGs, 96.5% upregulated), and the critical navitoclax in-vivo failure (DOXIL + navitoclax vs DOXIL alone, p=0.3888).

4. Gu 2025 — mechanistic axis supplemented: Correct directional axis (NRF2→FBP1 suppression→AKT→p53) confirmed, but key molecular details added: (a) FBP1 suppression mechanisms are both DNMT1-mediated methylation (62.8% of HCCs) and ERK/TRIM28-mediated proteasomal degradation; (b) FBP1→AKT connection is via PP2AC recruitment suppression; (c) AKT→p53 is via MDM2 Ser166 phosphorylation. Human n=83 HCCs (IHC; FBP1 downregulated in 71%) added to body and footnote.

5. Baker 2016 — supplemented: Genetic background detail added (two cohorts). Median lifespan extension 24–27% confirmed. Tumor-free lifespan extension 24–42% noted (not just tumor-protection effect). Core claim (tumor latency increased, incidence NOT reduced) verified accurate.

6. Pacifico 2021 — quantitative data added: SLC1A5 protein overexpression 2–6.5-fold confirmed from Fig. 5B. Five cell lines confirmed (not just “multiple”). Glutamine synthetase mechanism clarified: GS expression confers exogenous-Gln-independence (de novo synthesis from ammonia + glutamate), not elevated GS “activity” in the absence of substrate. NANOG co-overexpression and nucleoside rescue experiments added.

7. Druggability-tier: 2 → null (corrected): MDM2/CDK4-6 inhibitors are cancer drugs, not aging-context modulators of the escape process. No aging-validated drug targets the senescence-escape process itself. Null is correct per CLAUDE.md aging-context convention.

Unverifiable claims (4 not_oa sources):

• Milanovic 2018: specific limiting-dilution tumor-initiation numbers; quantitative Wnt-dependence rescue experiment details. Qualitative claims (enhanced clonogenic growth, Wnt-dependence, relapse enrichment) consistent with abstract.
• Saleh 2022: specific claims about TIS as “one of three overlapping dormancy programs.” Only abstract confirmed.
• Sikora 2022: mechanistic details linking TIS and TIP as a “single biological program.” Only abstract confirmed.
• Pacifico 2022: trabectedin mechanism details (SLC1A5 + GS downregulation confirmed from efetch abstract; no fold-changes verified).

Supersession check (R25): PubMed searched for meta-analyses on TIS escape (2022–2026). One meta-analysis found (PMID 39739833, Frey 2025, Nucleic Acids Research) was on DNA repair gene repression in senescence — not on TIS escape per se, and consistent with rather than contradicting current framing. No meta-analysis or large RCT supersedes existing evidence base. literature-checked-through: 2026-05-31 confirmed.

Downstream propagation needed (for main agent):

• hallmarks/cellular-senescence.md — if the two-layer barrier model is cited from Beauséjour 2003, the corrected framing (p16 knockdown required to make p53 inactivation sufficient in WI-38; combined p53+pRb still fails when p16 is high) should be checked and propagated if needed.
• interventions/pharmacological/senolytics.md — the Bajtai 2025 navitoclax in-vivo failure (DOXIL + navitoclax, p=0.3888) is directly relevant to any navitoclax/senolytic combination claims. Also: fisetin showed no differential TIS-cell toxicity in Bajtai 2025 Fig. 3F — relevant to molecules/compounds/fisetin.md if that page makes claims about senolytic activity against TIS cells specifically.
• molecules/proteins/p53.md and molecules/proteins/p21.md — if they cite Beauséjour 2003 for the p53-as-primary-senescence-checkpoint claim, the nuance (sufficient only in p16-low cells) should be present.

[2026-05-31] correction | Amor 2020 model mislabel on senescence-immune-surveillance (main-agent adjudication)

Cross-page contradiction check after the two verifications surfaced a conflict on Amor 2020 (doi:10.1038/s41586-020-2403-9): the surveillance-page verifier’s correction #5 (above) asserted the NRAS-driven liver tumor model was the primary in-vivo survival endpoint (n=11/group), with KP lung adenocarcinoma as validation-only. This contradicted the previously-verified hallmarks/disabled-adaptive-immunity.md (2026-05-07) and frameworks/intervention-by-target-immunogenicity.md, which frame the survival benefit as the lung adenocarcinoma model.

Adjudicated by reading the local Amor 2020 PDF directly (pages 1–6). The surveillance verifier was wrong; the older pages are correct:

• Abstract verbatim: “uPAR-directed CAR T cells extend the survival of mice harbouring lung adenocarcinoma treated with a senescence-inducing drug combination.”
• p.4: orthotopic KP (Kras^G12D;p53^−/−) lung adenocarcinoma + MEK/CDK4/6i → uPAR CAR-T “significantly prolonged survival.” This is the survival endpoint.
• The NRAS^G12V hepatocyte OIS model (Fig 2c–h, NSG mice) was a senescent-cell clearance readout (bioluminescence ↓ within 10 days; NRAS^+ p<0.01, SA-β-gal^+ p<0.001) — NOT a survival endpoint.
• CCl4 / diet-NASH liver fibrosis: reduced fibrosis + SA-β-gal (p<0.001), reduced suPAR/ALT/AST — homeostasis endpoint, not survival.
• The “n=11/group” and the fibrosis p-values (Sirius Red p<0.0001 / SA-β-gal p=0.0002) the verifier assigned could not be confirmed in the main text and appear mis-assigned.

Fix applied: corrected processes/senescence-immune-surveillance.md body (line ~163) + [^amor2020] footnote to reflect KP-lung = survival, NRAS-hepatocyte = clearance, CCl4/NASH = fibrosis; verified-scope updated with a 9th correction noting this. Now consistent across all three pages.

Lesson (re seeder-fabricates-outcomes + verifier reliability): a freshly “verified: true” page can still carry a verifier-introduced error; the cross-page contradiction check (newly-verified vs previously-verified citation of the same primary source) is what caught it. Worth doing routinely after verifying a page that cites papers already verified elsewhere.

Open reconciliation item (NOT yet actioned) — NKG2D age-decline

The surveillance verifier’s correction #7 found Tarazona 2017 reports NKG2D expression preserved in the elderly (NKp30/NKp46/DNAM-1 decline, not NKG2D), and tagged the new page gap for a better citation. But two existing pages still assert NKG2D declines with age:

• hallmarks/disabled-adaptive-immunity.md:90 — “Reduced activating receptors: NKG2D, NKp30, NKp46 expression declines with age”
• phenotypes/immunosenescence.md:74 — “Reduced expression of activating receptors (NKG2D, NKp30, NKp46)…” This is a contradictory-evidence (the primary literature is genuinely mixed — context/CMV-dependent NKG2D decline reported elsewhere vs Tarazona’s “preserved”). NOT resolved here: needs a check of what those two pages cite for the claim, then either substantiate or qualify. Flagged for next pass; existing verified pages left unedited pending source check.

[2026-05-31] reconcile | NKG2D age-decline (RESOLVED — cell-type conflation)

Followed up the open item above. Root cause: a cell-type conflation, not a true contradiction. Both existing pages’ “NKG2D, NKp30, NKp46 decline on NK cells” claim was unsourced (immunosenescence.md even self-tagged its NK section unsourced).

Primary-literature reconciliation (citations pulled from PubMed eutils, not memory):

• Almeida-Oliveira 2011 (Hum Immunol 72:319-29, PMID 21262312) — DECISIVE, abstract directly read: “a decrease in the expression of activating receptors (NKp30 and NKp46) was observed in NK cells in elderly… NKG2D expression was decreased in T cells of elderly subjects.” → NCRs down on NK; NKG2D down on T cells, not NK cells.
• Tarazona 2017 (already local/verified) — NK-cell NKG2D preserved; NCR/DNAM-1 decline.
• Campos 2014 (Exp Gerontol 54:130-7, PMID 24440462, not_oa) — CD57⁺/NKG2C⁺ expansion tracks CMV, not age; most NK receptors incl. NKG2D show little CMV effect; age vs CMV separable.
• Le Garff-Tavernier 2010 (Aging Cell 9:527-35, PMID 20477761, archive-pending) — CD56bright decline + LIR-1 increase; NK function recovers with IL-2 in very old. Hazeldine & Lord 2013 (Ageing Res Rev 12:1069-78, PMID 23660515) — review of per-cell cytotoxicity/NCR decline.

Reconciled position: on NK cells, NCRs (NKp30/NKp46) reliably decline and NKG2D is preserved/inconsistent; the robust NKG2D decline is on aged T cells. The widespread “NKG2D declines with age” line most plausibly originates from the T-cell finding mis-applied to NK cells. CMV serostatus is a major confounder for any NK-receptor aging claim.

Edits:

• processes/senescence-immune-surveillance.md — rewrote § Age-related decline (fixed internal Tarazona contradiction; added “cell-type conflation” subsection + bottom-line); added 4 footnotes (almeida2011, legarfftavernier2010, campos2014, hazeldine2013) — ALL abstract/metadata-level → needs-fulltext-verification (Almeida abstract directly read; Campos not_oa; Le Garff-Tavernier archive-pending). verified-scope updated (10th correction).
• hallmarks/disabled-adaptive-immunity.md — corrected the NK activating-receptor bullet (NCRs down / NK-NKG2D preserved / T-cell-NKG2D distinction); added [
• phenotypes/immunosenescence.md — same correction; removed the unsourced tag (now sourced); added [^almeida2011nk] + [

Follow-up (download attempts run 2026-05-31): Hazeldine 2013 → downloaded (LOCAL PDF now available; verifier to read end-to-end, flip needs-fulltext-verification). Le Garff-Tavernier 2010 + Almeida-Oliveira 2011 → download FAILED (no OA URL) → both now no-fulltext-access; Almeida-Oliveira’s cited finding is abstract-verbatim (decisive line quoted above). Campos 2014 → not_oa (#gap/no-fulltext-access). Residual contradictory-evidence retained on the surveillance page for NK-cell NKG2D direction (no large CMV-stratified cohort has correlated NK NKG2D density with senescent-cell burden in aged humans).

[2026-05-31] ingest | tirzepatide + SURPASS-CVOT (GLP-1 video review, user-requested ad-hoc)

Triggered by user review of a longevity-MD YouTube video advocating low-dose tirzepatide. Verified the video’s load-bearing claims against primary sources (SELECT/FLOW already in wiki; STEP 9 n=407 NOT 47 [video transcription slip], WOMAC −41.7 vs −27.5 ✓; SURPASS-CVOT primary 3-pt MACE HR 0.92, NI met p=0.003, superiority NOT met p=0.09; post-hoc JAMA Cardiology 2026 6-component HR 0.84; Qin pilot n=20 +17% vs <1% cartilage thickness).

• added: molecules/compounds/tirzepatide.md (verified:false, R25 recency run, literature-checked-through 2026-05-31). Canonical IDs pulled from primary DBs (PubChem CID 166567236, formula C225H348N48O68, MW 4813 Da; ChEMBL CHEMBL4297839; DrugBank null gap). SURPASS-2 head-to-head weight-loss deltas (treatment-diff only from abstract; absolute per-arm needs-pdf), retatrutide vs tirzepatide heart-rate contrast (bpm values abstract-absent → gap), Locatelli 2024 lean-mass. New mechanism class incretin-receptor-agonist added to frameworks/intervention-classes.md. Forward-ref stubs: glp1r, gipr, pomc-neurons. clinical-trials-active 119 (all-indication scope noted).
• added: studies/surpass-cvot-2025.md (verified:false, no-fulltext-access — NEJM paywalled, not in archive). Biblio verified via Crossref+PubMed: Nicholls SJ et al., NEJM 2025;393(24):2409–2420, PMID 41406444, DOI 10.1056/NEJMoa2505928. Post-hoc DOI 10.1001/jamacardio.2026.0767 (Nissen SE et al., JAMA Cardiol 2026). Framed post-hoc 6-component composite explicitly as hypothesis-generating, not superiority on the pre-registered endpoint.
• updated: studies/qin-2026-semaglutide-oa.md — added secondary-source corroboration block (ScienceAlert + Science News) for human pilot (n=20, +17% vs <1% MRI cartilage thickness) and pair-fed mouse dissociation; kept verified:false (press, not PDF); 2 new footnotes flagged SECONDARY SOURCE.
• updated: interventions/pharmacological/glp1-agonists.md — class table SURPASS-CVOT “(ongoing)” → published result + surpass-cvot-2025 link; tirzepatide cross-ref de-stubbed.
• verifier follow-up queued: tirzepatide SURPASS-2 Table 2 absolute weights + retatrutide phase-2 bpm-by-dose; SURPASS-CVOT primary HR CI + post-hoc CI (both PDFs); Qin 2026 PDF when archive-indexed; confirm DrugBank ID.
• gaps surfaced: glp1r, gipr, pomc-neurons pages still unseeded (now referenced by 2+ pages).

[2026-05-31] ingest | Corley semaglutide epigenetic-aging RCT (GLP-1 maintenance-benefit Q, user-requested)

Triggered by user question: “at equal body comp/nutrition/exercise, is there a benefit to low-dose GLP-1?” Best human evidence FOR a weight-independent aging effect.

• added: studies/corley-2025-semaglutide-epigenetic-aging.md (verified:false; drafted from medRxiv/PMC full text of the now-peer-reviewed paper). Corley MJ et al., Nat Commun 2026, DOI 10.1038/s41467-026-72861-3 (preprint 10.1101/2025.07.09.25331038, PMC12338914), NCT04019197. Phase 2b RCT, semaglutide n=45 / placebo n=39, 32 wk, HIV-associated lipohypertrophy. Epigenetic aging = SECONDARY/exploratory endpoint (primary = CT visceral/SAT + DEXA fat). After ANCOVA adjustment for age/sex/BMI/hsCRP/sCD163: PCGrimAge −3.08 (p=0.007), GrimAge V2 −2.26 (p=0.008), PhenoAge −4.90 (p=0.004), DunedinPACE −0.09 ≈9% slower (p=0.01), OMICmAge/SystemsAge/RetroAge concordant; organ clocks inflammation/brain/heart most affected. No multiple-comparison correction (authors: exploratory).
• Material COI captured in body + footnote: 3/7 authors TruDiagnostic-affiliated (Corley advisor; Dwaraka, Smith employees); TruDiagnostic performed all clock analyses → clock vendor co-authored study showing their own clocks respond to drug. McComsey (PI) consults Merck/ViiV/Gilead.
• Interpretive frame: closest human approximation to the “weight-independent aging benefit” question (BMI+inflammation-adjusted), BUT diseased/lipodystrophic population ≠ healthy fixed-comp; BMI-adjusted ≠ optimal body comp; secondary endpoint; COI. Does NOT establish benefit in a lean optimized person.
• updated: molecules/compounds/semaglutide.md (verified:true, minimal edits) — replaced the “no trial has measured a validated biological-age clock” gap line with the Corley finding + caveats; [^corley2025] footnote (Nat Commun, COI signal); gap reframed to needs-healthy-population-replication (dedicated aging RCT in healthy older adults on an INDEPENDENT epigenetic platform still needed).
• verifier follow-up queued: Nat Commun PDF not yet archive-indexed; fill missing individual-clock 95% CIs from published tables; confirm printed COI; then flip verified:true.
• side-note: GLP-1 has NO NIA ITP lifespan data (ITP page lists incretins as future test case only); contrast rapamycin/acarbose — relevant to “longevity drug at fixed body comp” framing.

[2026-05-31] ingest | low-back-pain + intervertebral-disc-degeneration (ad-hoc)

Trigger: Question of whether the wiki covered low back pain — it did not (confirmed gap; only fragmentary disc-senolytic data on navitoclax.md/quercetin.md). Two atomic phenotype pages seeded in parallel (wiki-seeder ×2).

Added:

• phenotypes/intervertebral-disc-degeneration.md — mechanism anchor. NP dehydration/aggrecan loss, collagen-II→I shift, cartilaginous-endplate calcification → nutrient deprivation; NP-cell senescence (p16/p21/SASP), ECM catabolism (MMP-3/-13, ADAMTS-4/-5), NP-progenitor exhaustion. Senolytic thread: Novais 2026 (Bone Research, SM/J mice) navitoclax-negative / D+Q-positive → cell-type-specific SCAP (NP cells need ephrin/PI3Kδ, not BCL-xL); Zhu 2015 (Ercc1 progeroid, D+Q ↑disc GAG); Ghazizadeh 2026 (o-vanillin/RG-7112 pain prevention, recency hit). Hallmark map; verified:false; literature-checked-through 2026-05-31.
• phenotypes/low-back-pain.md — clinical-syndrome anchor. Mechanical/non-specific (~85–90%) vs specific (radicular/facet/SI/fracture); aging substrates (disc degeneration, facet OA, paraspinal atrophy + multifidus fatty infiltration, vertebral/osteoporotic change); radiographic-pain disconnect (Brinjikji 2015: 37%@20 → 96%@80 asymptomatic); GBD 2021 #1 YLD (619M → projected 843M by 2050); exercise as best-evidenced mitigation (Hayden 2021 Cochrane, 249 RCTs); verified:false; literature-checked-through 2026-05-31.

Gaps surfaced: #gap/needs-human-replication (senolytic IVDD interventions); #gap/no-mechanism (what converts structural degeneration → symptomatic pain); stub wikilinks [[tissues/intervertebral-disc]], [[facet-joints]], [[spinal-ligaments]] (valid future stubs). Schema note: type: phenotype block in CLAUDE.md lacks a literature-checked-through: field though both pages carry it (optional per cadence table) — candidate minor schema add.

Verifier queued: Brinjikji 2015, Novais 2026, Ghazizadeh 2026 PDFs pending DOI lookup; Roberts 2006 disc-structure quantitative claims flagged for PDF cross-check.

[2026-05-31] verify | phenotypes/low-back-pain.md

PDFs checked: GBD 2021 (doi:10.1016/s2665-9913(23)00098-x) + Hayden 2021 Cochrane (doi:10.1002/14651858.CD009790.pub2). Brinjikji 2015 still pending download — age-stratified asymptomatic-degeneration table remains unverified.

Corrections applied:

• Hayden 2021 effect-size metric: SMD ~0.3–0.4 → MD −15.2 (pain) / MD −6.8 (function) on 0–100 scale. The paper uses mean differences (MD), not standardised mean differences (SMD). Certaintly level (moderate-certainty GRADE) and comparator framing also corrected.
• Hayden 2021 participant count: ~24,000 → 24,486 (exact figure from PDF).
• GBD 2021 footnote enriched with 95% UIs (554–694M; 759–933M), age-standardised prevalence rate (7460/100,000), YLD rate (832/100,000), and risk-fraction (38.8% occupational/smoking/BMI).
• Banner removed; verified: true set with verified-scope documenting partial coverage.

Page: phenotypes/low-back-pain.md → verified: true (partial scope; Brinjikji 2015 and secondary sources unverified).

[2026-05-31] verify | low-back-pain + intervertebral-disc-degeneration → verified:true (scoped)

Same-day scoped verification of both freshly-seeded pages (wiki-verifier ×3; Brinjikji PDF landed mid-pass).

intervertebral-disc-degeneration.md → verified:true. Zhu 2015 (10.1111/acel.12344) read end-to-end: dose/model/disc-GAG (Fig 6E p<0.05)/kyphosis (Fig 6A–B) all confirmed; corrected footnote to disambiguate aged-C57BL/6 n=6–9 (fat-depot senolysis) from Ercc1−/Δ progeroid n=7–8 (the arm carrying the disc-GAG + kyphosis outcomes) — the seed had implied disc outcomes came from the C57BL/6 arm. Brinjikji 2015 (10.3174/ajnr.A4173) anchors (37%@20→96%@80, bulge 30→84, protrusion 29→43, fissure 19→29) confirmed vs Table 2. Novais 2026 cross-checked against verified navitoclax.md (PDF unavailable) — faithful restatement, no drift. Roberts 2006 / Ghazizadeh 2026 remain PDF-unverified (#gap intact).

low-back-pain.md → verified:true. GBD 2021 (10.1016/s2665-9913(23)00098-x) read: 619M (UI 554–694) → 843M (UI 759–933) by 2050, #1 YLD, UIs/rates added. Hayden 2021 Cochrane (10.1002/14651858.CD009790.pub2) read: material correction — page attributed SMD ~0.3–0.4 to exercise effect; paper reports mean differences on 0–100 scale (pain MD −15.2 [−18.3,−12.2]; function MD −6.8 [−8.3,−5.3]), moderate-certainty; participant count 24,000→24,486 (249 RCTs confirmed). Brinjikji 2015 re-verified after PDF landed: 6 interior-decade cell corrections (disc degen @60 91→88; height-loss @60 73→67; bulge @80 77→84; protrusion @40 36→33; fissure @40 20→22; fissure @60 22→25); age-20/80 anchors were already correct.

Propagation: quercetin.md [^zhu2015] footnote + table rows checked — already correctly disambiguates the C57BL/6 vs Ercc1 arms; no fix needed. No other downstream pages carry these claims.