log/2026-05-23 — ad-hoc daily entries

Sub-file of log — see parent for index. Holds full content for one-line pointers in log.md.

[2026-05-23] ingest | UV filter chemistry — MCE (Mexoryl 400) + PBT (TriAsorB) + Matta 2019/2020 systemic absorption

Origin: User question on La Roche-Posay Anthelios UVMune 400 — full ingredient list, study landscape, fit vs aging goals; follow-up on avobenzone absorption concerns; follow-up Triasorb comparison + “add firm findings to wiki for future reference.”

Added:

molecules/compounds/mexoryl-400.md — full compound page for MCE (methoxypropylamino cyclohexenylidene ethoxyethylcyanoacetate). PubChem CID 71226339, CAS 1419401-88-9, MW 322.4, λmax 385 nm. EU SCCS S87 (2019); EU/AU/Canada approved; not FDA-GRASE. Evidence: 4 L’Oréal-sponsored intra-individual RCTs (Marionnet 2022 n=19 RHE + in-vivo; Flament 2024 n=113 outdoor 8 wk; Mercurio 2025 n=52 outdoor 4 wk; Zhang 2024 n=60 open) + 1 independent ACD case report (Loretan 2024). verified: false + auto-extraction banner.
molecules/compounds/triasorb.md — full compound page for PBT / TriAsorB (phenylene bis-diphenyltriazine). PubChem CID 59516799, CAS 55514-22-2, MW 540.6. 1,2,4-triazine scaffold (NOT s-triazine like Tinosorb S/M — flagged in body). EU SCCS S86 (2018); EU/AU approved; NOT on Canadian Sunscreen Monograph; not FDA-GRASE. Hybrid absorber+reflector covering UVB through HEV 400–450 nm. Evidence: 7 Pierre Fabre R&D studies (Bacqueville 2021/2022 mechanism + Boyer 2023 BL-challenge headline RCT n=20+16 + Douki 2023/2024 + Le Digabel 2023 + Jacques 2022 MALDI-MS) + 1 L’Oréal cross-line methodology paper (Touti 2025) + 1 environmental persistence paper (Fagervold 2025: not biodegraded in marine sediment after 100 d). verified: false + auto-extraction banner.

Updated:

interventions/lifestyle/uv-protection.md — added § Ultra-long UVA-I filters (380–400 nm “UV-blue gap”) within § Chemical filters, with MCE and PBT subsections linking out to the new compound pages. Added § Chemical filter systemic absorption with Matta 2019/2020 JAMA plasma Cmax table (oxybenzone 209.6, homosalate 23.1, avobenzone 4.0–8.7, octocrylene 4.5–7.8, octinoxate 7.9, octisalate 5.1, ecamsule 1.5 ng/mL — all above 0.5 ng/mL FDA threshold). Added practical filter-stack risk ranking: mineral → EU non-avobenzone → EU with avobenzone → US standard → legacy with oxybenzone. Updated literature-checked-through: 2026-05-23. Updated verified-scope: to flag new sections as no-fulltext-access pending PDF cross-check (page remains verified: true with scope-bounded caveats per the schema’s partial-verification convention).

Substantive findings worth surfacing:

The “broad spectrum SPF 50+” market hides a 380–400 nm gap. Conventional UVA-I filters (avobenzone λmax 357, Tinosorb S 343, Uvinul A Plus 354, Mexoryl SX 345) drop off above ~370 nm. MCE (λmax 385) and PBT (broad through 450) are the two filters operationalizing Krutmann 2021’s expanded-spectrum recommendation as of 2026.
EU vs US filter regulatory gulf. Six of the nine filters in any UVMune 400 SKU are not FDA-approved. US Anthelios is a chemically unrelated formula. MCE/PBT/Tinosorb/Mexoryl SX/XL/Uvinul A Plus all blocked by the 2019 FDA GRASE deferral.
Sponsor-only efficacy evidence. Both MCE and PBT have no independent (non-manufacturer) efficacy replication. Both fail on biopsy CPD/8-OHdG/MMP-1 hard endpoints; both fail on head-to-head against top-tier non-MCE/PBT EU stacks. Each is supported by 3–7 sponsor-authored intra-individual hemiface/hemibody studies.
Avobenzone is in every UVMune 400 SKU; not in Avène Sunsimed KA. If avobenzone systemic absorption is the binding concern (Matta 2019/2020 Cmax 4–8.7 ng/mL), Avène Sunsimed is the avobenzone-free alternative; UVMune 400 is not.
TriAsorB environmental persistence. Fagervold 2025 documented non-biodegradation in marine sediment microcosms after 100 d — cuts against the “reef-safer than legacy filters” framing for users prioritizing aquatic ecotoxicity.
Photoaging-benefit/absorption-harm asymmetry. The Hughes 2013 OR 0.76 photoaging benefit is demonstrated; the Matta plasma absorption is unquantified for clinical harm. Declining sunscreen entirely is worse than any of the alternatives.

Gaps surfaced:

#gap/no-independent-replication on both MCE and PBT compound pages
#gap/no-biopsy-endpoint for MCE (no published CPD/8-OHdG/MMP-1 IHC in human skin with vs without MCE)
#gap/no-direct-comparator-rct for PBT (no RCT vs iron-oxide-pigmented sunscreen on visible-light endpoints)
#gap/no-fulltext-access on Matta 2019/2020, Marionnet 2022, Flament 2024, Mercurio 2025, Boyer 2023, Bacqueville 2021, Fagervold 2025 — all abstract-level pending PDF verification

Files touched: interventions/lifestyle/uv-protection.md, molecules/compounds/mexoryl-400.md (new), molecules/compounds/triasorb.md (new), log.md.

[2026-05-23] ingest | 15-PGDH gerozyme axis — protein + inhibitor + 3 study pages

Origin: User request to “look for and incorporate information about protein 15-PGDH, particularly Bhutani 2025.” The “Bhutani 2025” referent is Singla M, Wang YX,…, Blau HM, Bhutani N (last/corresponding author), Science 391(6789):1053–1062 (2025), doi:10.1126/science.adx6649, PMID 41308124 — “Inhibition of 15-hydroxy prostaglandin dehydrogenase promotes cartilage regeneration.”

Workflow: Five parallel wiki-seeder dispatches (batch of 5; per workflow-cadence memory) → propagation pass to high-priority existing pages. Today is 2026-05-23.

Added:

molecules/proteins/15-pgdh.md — central atomic page for the gerozyme HPGD (UniProt P15428, NCBI Gene 3248, HGNC:5154, Ensembl ENSG00000164120). NAD+-dependent SDR-family enzyme catalyzing first committed step of PGE2 catabolism. Body covers identity, biochemistry, aging context (gerozyme framing), tissue distribution, mechanistic consequences of inhibition (PGE2-EP4 → mitochondria/autophagy/TGF-β/UPS), cross-tissue rejuvenation table (muscle/NMJ/cartilage/heart/bone-marrow/rotator-cuff), clinical translation via Epirium Bio, explicit cancer-aging tradeoff section (tumor suppressor in colon/lung/breast — 7.6-fold colon tumor increase in HPGD-null Apc^Min/+). druggability-tier: 1 (Epirium clinical pipeline). literature-checked-through: 2026-05-23. verified: false + banner.
molecules/compounds/sw033291.md — canonical 15-PGDH small-molecule inhibitor (Zhang 2015 Science discovery; PubChem CID 3337839; MW 412.6). Used in every subsequent Blau lab aging paper. Body covers chemistry, mechanism, preclinical efficacy by tissue, PK/dosing, safety + cancer-aging tradeoff, Epirium Bio clinical translation note (derivative ≠ SW033291). New mechanism class 15-pgdh-inhibitor added to frameworks/intervention-classes.md. clinical-stage: preclinical, human-evidence-level: preclinical-only. literature-checked-through: 2026-05-23.
studies/palla-2021-15pgdh-muscle-rejuvenation.md — foundational paper. Aged C57BL/6 mice; 15-PGDH elevated in aged muscle (myofibers + macrophages); overexpression in young → atrophy (causal sufficiency); pharmacological SW033291 or genetic depletion → restored mass + grip strength + exercise performance via PGE2-EP4 axis.
studies/bakooshli-2023-15pgdh-nmj-regeneration.md — Sci Transl Med. Sciatic nerve transection/crush + aged chronically denervated mice; PGDHi → motor-axon regeneration + NMJ restoration. 15-PGDH aggregates define “target fibers” — histopathologic hallmark of human neurogenic myopathies. Local PDF available in a local paper archive after this ingest’s download.
studies/singla-2025-15pgdh-cartilage-regeneration.md — the “Bhutani 2025” paper. Aged + DMM/PTOA mice + ex-vivo human OA cartilage explants (n=11, total-knee-replacement source) — paper was incorrectly briefed as human-evidence: false, seeder corrected to human-evidence: true after PMC HTML read. Mechanism: gene-expression reprogramming of preexisting chondrocytes (↓hypertrophic-like 15-PGDH+ subset, ↑ECM-synthesizing articular subset) — NOT stem/progenitor expansion. scRNA-seq subpopulation gating + multiplexed IF (CODEX-family). scRNA-seq deposited as GEO GSE308009. PMC13127300 (released 2026-04-29) used as primary extraction source.

Updated (propagation):

phenotypes/osteoarthritis.md — added PGDHi bullet to “Interventions and clinical landscape” with Singla 2025 footnote; added altered-intercellular-communication row to hallmark-intersections table; cross-refs to 15-pgdh, sw033291, cancer-aging-tradeoffs, and the Singla 2025 study page.
phenotypes/sarcopenia.md — added PGDHi row to “Pharmacological (investigational)” table; added “15-PGDH gerozyme axis” paragraph to “Neurogenic component”; new Palla 2021 + Bakooshli 2023 footnotes; verified-scope: extended with explicit caveat that the new 15-PGDH claims are abstract-only-extracted.
tissues/skeletal-muscle.md — added 15-PGDH paragraph to “Neuromuscular junction degeneration and denervation”; added PGDHi row to “Aging-protective interventions” table; new Palla 2021 + Bakooshli 2023 footnotes; cross-refs + verified-scope: updated similarly.
frameworks/cancer-aging-tradeoffs.md — added 15-PGDH-inhibition row to the trade-off-pattern table (the gerozyme/tumor-suppressor tension is unusually explicit at the molecular level: PGE2 raised by PGDHi is the same metabolite aspirin/COX-2i pharmacology lowers for cancer prevention); cross-refs to all 5 new pages.
pathways/spm-pathway.md — new “Cross-pathway interaction with the 15-PGDH gerozyme axis” subsection making the structural tension explicit (SPMs compete against PGE2 / PGDHi raises PGE2; resolution requires (a) tissue-context, (b) dose, (c) receptor-subtype unpacking); cross-refs added.
frameworks/intervention-classes.md — new 15-pgdh-inhibitor class added by the sw033291 seeder per R16 discipline.

Slug-mismatch fix: The protein-page seeder and the Palla 2021 seeder used short study filename slugs (palla-2021-15pgdh-muscle, bakooshli-2023-15pgdh-nmj, singla-2025-15pgdh-cartilage); the actual filenames use the longer rejuvenation/regeneration suffixes. Edited 7 wikilink instances across the three files to point to canonical filenames before propagation.

UniProt typo fix: Bakooshli study page footnote cross-reference originally said “UniProt Q16836 in humans” for HPGD; corrected to P15428 (Q16836 is HSD17B6, an unrelated dehydrogenase). Memory consequence: this is a clean instance of the feedback_canonical_db_memory_unreliable.md failure mode in subagent output; the protein-page seeder (which actually pulled UniProt via WebFetch) had the correct accession.

Substantive findings worth surfacing back:

“Bhutani 2025” is the cartilage paper — Singla M (first; equal-contribution with Wang YX)… Bhutani N (last/corresponding); Blau HM is co-senior. Published online 2025-11-27, in-print 2026-03-05. PMC release 2026-04-29 (now accessible).
The mechanism in Singla 2025 is chondrocyte reprogramming, not progenitor expansion. This contradicts the natural stem-cell-recruitment model for cartilage repair and matters mechanistically — it positions 15-PGDH inhibition as a transcriptional-reprogramming therapy rather than a regenerative-medicine product.
Ex-vivo human OA cartilage data (n=11 patients) in Singla 2025 raises the translational bar above the mouse-only Palla 2021 and Bakooshli 2023.
Epirium Bio = Myoforte Therapeutics rebranded — same patent estate moved between the 2021 and 2025 papers; both lab senior author HM Blau is cofounder. Conflict-of-interest density is high on every paper in this series.
Cancer-aging tradeoff is the central unresolved question. Aspirin/COX-2 inhibitors prevent colorectal cancer by lowering PGE2; 15-PGDH inhibitors rejuvenate muscle/NMJ/cartilage by raising PGE2. Same molecule, opposite directions. Long-term human safety data does not exist.
Cross-pathway tension with SPMs: SPM-pathway page already documented “SASP-derived PGE2 competes with SPMs”; the gerozyme story flips the sign of “is PGE2 good or bad” — now explicit at pathways/spm-pathway.md and a key open question for clinical PGDHi programs.

Gaps surfaced:

#gap/proprietary-chemistry for the Epirium Bio clinical candidate (not SW033291)
#gap/long-term-unknown for systemic 15-PGDH inhibition in humans (cancer risk)
#gap/contradictory-evidence for PGE2 sign — SPM-pathway frames it as inflammatory, gerozyme frame as homeostatic
All three new study pages: verified: false pending PMC PDF cross-check (Singla 2025 is PMC HTML-extracted only; Palla 2021 + Bakooshli 2023 are abstract-extracted with Bakooshli having a local PDF for the verifier)
#gap/needs-replication — all three Blau-lab papers are single-lab; independent replication is needed for evidence weight

Files touched: molecules/proteins/15-pgdh.md (new), molecules/compounds/sw033291.md (new), studies/palla-2021-15pgdh-muscle-rejuvenation.md (new), studies/bakooshli-2023-15pgdh-nmj-regeneration.md (new), studies/singla-2025-15pgdh-cartilage-regeneration.md (new), frameworks/intervention-classes.md, phenotypes/osteoarthritis.md, phenotypes/sarcopenia.md, tissues/skeletal-muscle.md, frameworks/cancer-aging-tradeoffs.md, pathways/spm-pathway.md, log.md.

[2026-05-23] verify | 15-PGDH gerozyme axis — 5-page verification pass

Same-day verifier dispatch on all 5 newly-seeded pages. Five wiki-verifier runs (3 studies in parallel → propagation → 2 atomic pages in parallel). Substantive corrections worth surfacing:

Palla 2021 — verified against PMC7938328 HTML:

Mouse ages corrected: seeder said “18–22 mo” aged + “2–3 mo” young — actual paper uses >24 mo aged and 2–4 mo young throughout
SW033291 mg/kg dose is NOT stated in this paper (only referenced via Zhang 2015)
Human microarray source is the Raue 2012 vastus lateralis dataset (mean aged 78 ± 6 yr vs young 25 ± 3 yr); seeder’s “human IHC” framing was fabricated
Sex of mice not stated anywhere in the paper (permanent gap)

Bakooshli 2023 — verified against local PDF:

Both seeder background-citation DOIs were wrong. Deschenes 2010 10.1152/japplphysiol.00773.2010 returns 404; correct is 10.1016/j.exger.2010.03.007. Gonzalez-Freire 2014 10.1093/gerona/glu072 resolves to an unrelated co-trimoxazole paper; correct is 10.3389/fnagi.2014.00208. Clean second instance of feedback_seeder_brief_doi_unreliable.md.
Mechanism is motor-neuron-centric (EP4→cAMP→CREB in spinal-cord motor neurons), NOT Schwann-cell-centric. Schwann cells were never directly tested. Seeder’s tSC framing was wrong; propagated correction to sw033291.md (lines 106, 111, 115) and the protein page.
Primary cellular source of muscle 15-PGDH is denervated myofibers (snRNA-seq DN myonuclei), not Schwann cells.
SW033291 dose confirmed: 5 mg/kg i.p. once daily (14 d for crush cohort; 1 month for aged cohort).
Quantified: Hpgd mRNA rises ~20× by d90 post-SNT (10× at d14), protein 4×. Aged EDL 20.2% myofibers denervated vs 4.4% young. Plantar flexor force +37.2 ± 4.9% at 14 dpi; motor axon counts 1.9× higher; aged motor neuron apoptosis 11.6%→4.2% with PGDHi.
Human biopsy corrected: n=10 patients (9/10 positive), diseases are axonal neuropathies / myositis / motor neuron disease (ALS, SMA) / lumbar radiculopathy / neurogenic amyloidosis — NOT inclusion-body myositis, NOT limb-girdle MD, NOT post-polio syndrome (seeder fabrications).
LC-MS/MS measures PGEM (15-keto-PGE2 breakdown product) as activity proxy, NOT PGE2 directly.

Singla 2026 — verified against PMC13127300:

Year corrected: 2025 → 2026 in-print (online 2025-11-27). Slug stays singla-2025-* (filename); citations now use 2026.
n=11 human OA explants is the FACS characterization cohort only. PGDHi treatment used n=5/group (GAG/stiffness) and n=3/group (CD200+ readout). Propagated to osteoarthritis.md, sw033291.md, 15-pgdh.md.
Aged-mouse n=9/group; PTOA n=7/group; baseline young n=3 / aged n=5 (p=0.036).
CatWalk pain p-values corrected: vehicle vs PGDHi is p=0.010 (paw contact) / p=0.028 (mean intensity) — original “p≤0.032” is Ctrl vs Vehicle (injury effect), not treatment effect.
Chondrocyte cluster markers corrected: hypertrophic-like = CD200, Ihh, Mef2c, Wnt5b, Spp1, Pth1r, Alpl, Runx2 (NOT Col10a1/Mmp13); articular = Bmp5, Fgf2, Stat5a, Hdac9 (NOT Col2a1/Acan/Sox9). All seeder guesses on this were wrong.
scRNA-seq deposited as GEO GSE308009.

15-PGDH protein page — companion-page cross-check + UniProt re-validation:

druggability-tier: 1 → 2. No PGDHi clinical trial on ClinicalTrials.gov. The Epirium Bio NCT04386304 is an unrelated epicatechin/Becker MD program. Tier 1 requires “a clinical drug exists”; SW033291 is a tier-2 high-quality probe. Propagated the “Epirium clinical pipeline” framing softening across sarcopenia, osteoarthritis, skeletal-muscle, cancer-aging-tradeoffs → “IP held by Epirium Bio; no registered PGDHi trial as of 2026-05-23”.
Isoform count 7 → 5; RefSeq NM_000860.6 → NM_000860.5; Zhang 2015 PMID added (26068857); Chaudhary 2025 second-author fix.

SW033291 compound page — companion-page cross-check + R25 supersession sweep:

Ki = 0.1 nM corrected (was “IC50 ≈ 0.1 µM / Ki ~3-30 nM”). Propagated to frameworks/intervention-classes.md § 15-pgdh-inhibitor where the same “~3 nM” claim appeared.
Palla 2021 duration corrected: “14 days” → 1 month (14 days is Bakooshli’s crush cohort).
Palla 2021 dose claim removed (paper doesn’t state mg/kg).
Palla 2021 mechanism qualifier corrected: depleted satellite cells → EP4 conditional KO in myofibers.
Bakooshli 2023 author list previously listed “Bhutani N” — Bhutani is NOT an author on Bakooshli 2023 (only Singla 2026). Corrected.
R25 supersession sweep (2025–2026): 13 preclinical hits, no Phase 1/2 RCTs or meta-analyses. literature-checked-through: 2026-05-23 confirmed current.

Cumulative meta-pattern surfaced: This ingest produced clean instances of two known seeder failure modes: (a) feedback_seeder_brief_doi_unreliable.md — both Bakooshli background-citation DOIs (Deschenes 2010 + Gonzalez-Freire 2014) wrong; (b) feedback_seeder_fabricates_outcomes.md — chondrocyte cluster markers all wrong; Palla 2021 “human IHC ~3-fold elevation in ≥65 yr biopsies” fabricated (actual data is Raue 2012 public microarray of 78±6 vs 25±3 yr, no fold-change in text); inclusion-body-myositis / limb-girdle-MD / post-polio fabricated as Bakooshli human-biopsy diseases.

Files touched: molecules/proteins/15-pgdh.md, molecules/compounds/sw033291.md, studies/palla-2021-15pgdh-muscle-rejuvenation.md, studies/bakooshli-2023-15pgdh-nmj-regeneration.md, studies/singla-2025-15pgdh-cartilage-regeneration.md, frameworks/intervention-classes.md, phenotypes/osteoarthritis.md, phenotypes/sarcopenia.md, tissues/skeletal-muscle.md, frameworks/cancer-aging-tradeoffs.md, log.md.

[2026-05-23] ingest | Klotho-FGF23-phosphate axis seeding cluster

Trigger: User question surfaced phosphate-additive aging-relevance and called out the absence of a vascular-calcification page as a significant gap. Ad-hoc seed (skip ROADMAP queue per ad-hoc-seed convention).

Pages seeded (3 in parallel; standard wiki-seeder workflow; all verified: false):

molecules/proteins/fgf23.md — 205 lines / 8 footnotes. UniProt Q9GZV9 confirmed via live API. Druggability-tier 2 documented (burosumab is FDA-approved for XLH not aging-indicated; convention applied per CLAUDE.md aging-context rule). Includes dedicated “MR discordance” section flagging Henry 2022 Circulation MR showing FGF23 protective against HF (opposite to observational direction) — R25 discipline. literature-checked-through: 2026-05-23. Verifier-priority HIGH: Gutiérrez 2008 NEJM specific OR/CI (bronze OA accessible); Sharma/Ix 2021 JAGS cause-specific HRs (PMC OA).
processes/vascular-calcification.md — 269 lines / 6 footnotes (all archive-confirmed via DOI lookup). Includes load-bearing intimal-vs-medial framework distinction (the orchestration with existing atherosclerosis + arterial-stiffening). Discordance framing on MK-7 vit K2 evidence (Li 2023 meta-analysis positive vs Murali 2023 review null; root-caused to CKD-population vs non-CKD differences). literature-checked-through: 2026-05-23. Druggability-tier 3 with documented rationale (no aging-context clinical drug; K2 is dietary-vitamin grade).
interventions/dietary/phosphate-additive-reduction.md — Bioavailability differential framing (~80-100% inorganic vs ~40-60% organic) as core insight. Honest evidence-grading: cohort-strong (Dhingra 2007 Framingham ~1.55× CV risk), FGF23-mechanism partial RCT support, hard-endpoint RCT in non-CKD absent. human-evidence-level: limited, translation-gap: phase-3-rct-needed, clinical-trials-active: 0 (non-CKD scope). literature-checked-through: 2026-05-23. New mechanism class added to frameworks/intervention-classes.md § dietary-phosphate-reduction / klotho-fgf23-axis-modulation with canonical values phosphate-load-reduction, fgf23-modulation, klotho-axis-preservation.

Propagation pass completed:

molecules/proteins/klotho.md — #stub tag removed from FGF23 cross-link line; added vascular-calcification reference noting Klotho’s FGF23-independent anti-calcification mechanism
molecules/proteins/fgf21.md — added sibling-FGF cross-reference noting FGF19-subfamily endocrine-FGF architecture (FGF15/19, FGF21, FGF23 all Klotho-coreceptor-requiring; alpha-Klotho for FGF23 vs beta-Klotho for FGF21)
pathways/fgf-signaling.md — already comprehensive on FGF23, no edits required
phenotypes/arterial-stiffening.md — added wikilink in medial-calcification subsection pointing at the new vascular-calcification mechanism page (was a no-mechanism tag)
phenotypes/atherosclerosis.md — added intimal-vs-medial distinction in calcification bullet with wikilink to vascular-calcification
interventions/dietary/mediterranean-diet.md — added two cross-reference entries (phosphate-additive-reduction as complementary; vascular-calcification as partial-but-incomplete MedDiet target)
Hallmark pages (cellular-senescence, chronic-inflammation): NO edits required — they already reference atherosclerosis + arterial-stiffening, which both now point to vascular-calcification; redundant entries would clutter the downstream-phenotype matrices.

Gaps surfaced (#gap/needs-page):

[[matrix-gla-protein]] (MGP) — load-bearing for MK-7 mechanism on vascular-calcification page; highest-priority next stub
[[runx2]] — master osteogenic TF for VSMC transdifferentiation; second-priority
[[bmp-2]], [[osteopontin]] — osteogenic signaling for VSMC switch
[[bone]] (tissue page absent), [[osteocytes]] (cell-type page absent) — load-bearing for FGF23 secretion biology
[[arteries]] (tissue page absent), [[vsmc]] (cell-type page absent)
[[wnt-signaling]] (pathway page absent) — referenced in vascular-calcification mechanism

Schema-history escalation pending: vascular-calcification seeder flagged a question about whether selective-variants: on type:process pages should allow free-text strings (when variants live as subsections on the same page) vs. mandate wikilinks (autophagy precedent — but autophagy variants like mitophagy are genuine separate pages). Decision pending user input; if formalized, document in schema-history.md.

Verification pass (2026-05-23): molecules/proteins/sost.md — flipped verified: true. ARCH (Saag 2017) and FRAME (Cosman 2016) PDFs read end-to-end. 9 corrections: (1) ARCH CV p-value “p=0.07 by Fisher’s exact” removed — not in paper; replaced with OR 1.31 (95% CI 0.85–2.00) + sub-event breakdown (cardiac ischemic 0.8% vs 0.3%; cerebrovascular 0.8% vs 0.3%). (2) ARCH vertebral fracture RRR ~50% → 48% (RR 0.52; 95% CI 0.40–0.66; P<0.001). (3) ARCH + FRAME footnotes fully expanded with per-arm n, doses, exact rates, CIs, PDF-verified flag. (4) Hallmarks corrected: deregulated-nutrient-sensing + loss-of-proteostasis → altered-intercellular-communication + stem-cell-exhaustion; rationale section added to body. (5) mr-causal-evidence: partial → yes (Zheng 2023 GWAS + two-sample MR; n=33,961; MI OR 1.35, T2DM OR 1.32, hypertension OR 1.09 for lower sclerostin). (6) New footnote [^zheng2023mr] (PMID 37096546) added + integrated into ARCH CV discordance discussion. (7) New footnote [^chen2026meta] (PMID 41635538; n=12,384 RCT meta-analysis; no significant CV excess overall) added as counterpoint. (8) Retraction check: PMID 41723626 (Tominaga et al.) not cited — confirmed safe. (9) ARCH framing already consistent with corrected wnt-beta-catenin.md framing — no cross-cluster correction needed. Unverifiable: Ellies 2006 (download failed), Bonewald 2007 (download failed), Bhattoa 2013 (not_oa), Stelmaszek 2025 (not_oa). Downstream propagation needed: processes/vascular-calcification.md (add Zheng 2023 MR), phenotypes/osteoporosis.md (verify ARCH language).

[2026-05-23] verify | phenotypes/osteoporosis.md

Verifier: claude (wiki-verifier subagent)

Sources checked:

Farr 2024 Nature Medicine (doi:10.1038/s41591-024-03096-2) — full PDF verified
Farr 2023 JCI (doi:10.1172/JCI162519) — full PDF verified
ICD-10-CM codes — verified against NLM ICD-10-CM API (2025 release)
Händel 2023 BMJ NMA, Singh 2022, Xiao 2022, Ayers 2023 — not_oa or download-pending; abstract-level only; not fully verified

Corrections made (5):

ICD-10 frontmatter label — M81.0 mislabeled: "M81.0 (postmenopausal)" → "M81.0 (age-related osteoporosis without current pathological fracture)". In ICD-10-CM, M81.0 is the age-related code; there is no separate postmenopausal-specific code at M81 in the US clinical modification (NLM API confirms only M81.0, M81.6, M81.8 as valid M81 codes).
Farr 2024 intervention table — P1NP and CTx time points wrong: The table stated “P1NP ↑ +34%, CTx ↓ −11%, radius BMD ↑ +2.7% at 20 weeks” for the T3 exploratory tertile. Per the PDF (pages 5–6, Fig. 4a,b): P1NP +34% (p=0.035) and CTx −11% (p=0.049) were measured at 2 weeks post-dosing; radius BMD +2.7% (p=0.004) was at 20 weeks. The erroneous “at 20 weeks” applied to all three was corrected to distinguish the 2-week biomarker signal from the 20-week BMD signal. The overall P1NP secondary result (+16% at 2 wks and 4 wks, p=0.020/0.024; not significant at 20 wks −9%, p=0.149) was also added to the table for completeness.
Farr 2024 footnote — time points omitted, dose/design detail missing: The footnote gave no time points for the tertile numerics, enabling the table’s timing error. Expanded to document dosing schedule (D 100 mg + Q 1000 mg × 3 days/cycle, monthly × 6), the tertile size (~10/arm), and the time-point split between biomarker and BMD results. Clarified open-label, single-site design.
Farr 2023 JCI footnote — expanded with precise study design: Added AP20187 dose (10 mg/kg twice-weekly), n per group for local-model arm (n=15 females + 10 males/group), transplant-arm n (n=11/group; young adult 4-month-old syngeneic male C57BL/6 WT mice), and clarified that systemic model comparison data are from the prior p16-INK-ATTAC publication (not from this paper’s primary arm). “Local PDF available” updated to “local PDF verified end-to-end 2026-05-23.”
Romosozumab CV safety language — inconsistent with wnt-beta-catenin verified framing: "higher MACE in ARCH" → "higher rate of serious cardiovascular adverse events in ARCH year 1" (aligning with the corrected framing on the verified wnt-beta-catenin page).

Unverifiable claims:

Händel 2023 BMJ NMA fracture-reduction claims (not_oa; abstract only) — #gap/no-fulltext-access implied; footnote notes “DOI lookup pending”
Singh 2022 romosozumab meta-analysis fracture reduction magnitude (download pending via PMC9003152)
Xiao 2022 global prevalence 19.7% / 108-study count (not_oa; abstract only — numbers consistent with published abstract)
Ayers 2023 Annals IM (not_oa; abstract only)
Prevalence “~30% women / ~16% men 65+” — widely-cited NHANES/IOF consensus; not traced to single primary source in this pass
ICD-11 FB83.1 — WHO ICD-11 API requires bearer token; could not independently verify

Supersession candidates (R25):

PubMed search 2023–2026 for osteoporosis fracture RCTs/meta-analyses returned 240 hits; top 10 reviewed. None represent a paradigm-changing supersession of the current framing. Notable: Cipolloni et al. 2026 meta-analysis (Medicina) confirms anabolic-first strategy; Huang & Watt 2026 (Annals IM) shows 3-month romosozumab noninferior to 12-month for total hip BMD — consistent with current framing, does not contradict.
No supersession requiring editorial revision found. literature-checked-through: 2026-05-23 confirmed.

Downstream propagation needed:

interventions/pharmacological/senolytics.md — the Farr 2024 footnote ([^farr2024-bone]) on that page states “P1NP +34%, CTx −11%, radius BMD +2.7% at 20 weeks” (line 179), omitting the 2-week time point for P1NP/CTx. Same error as osteoporosis.md table; needs same correction.
tissues/bone.md — the Farr 2024 footnote ([^farr2024-phase2-rct]) line 228 says “D+Q increased P1NP (+34%, p=0.035), reduced CTx (−11%, p=0.049), increased radius BMD (+2.7%, p=0.004) at 20 weeks.” This also incorrectly attributes all three results to “at 20 weeks.” Needs same correction. (Body text line 164 says “P1NP ↑ (+34%), CTx ↓ (−11%), and radius BMD ↑ (+2.7%) at 20 weeks” — same timing error in the table body.)
studies/farr-2024-senolytic-bone-rct.md — if this study page exists and is verified, check whether it has the correct time-point split. If not yet seeded, seed priority elevated given multiple pages citing this paper.

Final verified state: verified: true (partial scope — Farr 2024 and Farr 2023 primary sources verified end-to-end; ICD-11, closed-access NMA sources, and prevalence primary sources not independently verified).

[2026-05-23] verify | pathways/wnt-beta-catenin.md — merged-section verification

Verifier: claude (wiki-verifier subagent)

Scope: Merged sections only (per verified-scope frontmatter). Original R20 sections (Brack 2007 / Liu 2007 / Florian 2013 / Clevers 2006 / Logan & Nusse 2004) NOT re-verified.

Sources checked:

Yun 2023 (doi:10.1038/s41467-022-35487-9, Nature Communications) — full PDF verified (downloaded + 8 pages read end-to-end). Confirms: first author Jina Yun (not “Yum MK”); senescent-MEF conditioned medium → mouse small intestinal organoid model (jeunum crypts); sPtk7 identified via MS as a SASP factor in medium fraction; sPtk7 → FZD7 binding → Wnt/Ca²⁺ non-canonical → Ca²⁺ oscillations → nuclear YAP → ISC differentiation disruption; n=3–5 biological replicates (organoids per individual mouse); cystic organoid phenotype reversible. Citation count 56, FWCI 13.9 — confirmed via DOI lookup.
Kocełak 2024 [as khalid2024] (doi:10.17219/acem/169567, Adv Clin Exp Med 2024;33(5):519–532) — full PDF verified (downloaded + all 14 pages read). Confirms: first authors Kocełak P, Puzianowska-Kuźnicka M, Olszanecka-Glinianowicz M, Chudek J (no “Khalid” in author list); PRISMA systematic review of 160 studies; ARCH trial CV signal cited as “higher frequency of severe cardiovascular adverse events in romosozumab vs. alendronate (2.5% vs. 1.9%)” from Saag 2017; FRAME trial separately cited as not finding increased CV risk vs placebo; paper does NOT label the CV events as “MACE” — they are “severe cardiovascular adverse events”; paper notes events were myocardial infarctions and stroke.
Zhu/Chen 2024 [as chen2024osteoblast] (doi:10.1038/s41421-024-00689-6, Cell Discovery 2024;10:71) — full PDF verified (first author Siyu Zhu; corresponding: Wei Chen + Yi-Ping Li; 39-page review covering Wnt, BMP, TGF-β, Hedgehog, PTH, FGF, Ephrin, Notch, Hippo, Piezo1/2; citation count 257, FWCI 123 — both confirmed via DOI lookup).
Saag 2017 [ARCH trial] (doi:10.1056/NEJMoa1708322, NEJM) — verified via PubMed abstract (PMID 28892457). Confirms n=4,093 (2046 romosozumab / 2047 alendronate); 12-month blinded phase then open-label alendronate; “serious cardiovascular adverse events” were a monitored secondary endpoint; 48% lower vertebral fracture risk and 27% lower clinical fracture risk confirmed. Full PDF bronze OA (not downloaded; abstract-level verification sufficient for the claims on this page).
Xu 2024 [as oglcnac2024] (doi:10.1096/fj.202401649RR, FASEB J 2024;38(24)) — abstract-level only (not_oa; cannot read full text). First author Xu Lin (not “Zhang Y” — confirmed via Crossref). Abstract confirms: OGT and O-GlcNAcylation upregulated in artery tissues of mouse calcification models + CKD patients; mechanism: OGT promotes VSMC osteogenic transdifferentiation via Wnt/β-catenin. Tagged no-fulltext-access.
Nuzzo 2026 [as romosozumab2026] (doi:10.1007/s00198-026-07932-8) — DOI not in archive (too recent). Title and year confirmed via Crossref: “Romosozumab is effective in patients with variants in LRP5, LRP6, or WNT1.” LRP5/LRP6/WNT1 variant claim on wiki page consistent with confirmed title.
Yeon 2026 [as fwci-sclerostin2026] (doi:10.1007/s11883-025-01377-w) — DOI not in archive. Title, authors, and journal confirmed via Crossref: Yeon S, Seto SW, Bhuyan JD, Chang D, Li CG; Current Atherosclerosis Reports 2026; “Sclerostin in Vascular Calcification: Hypoxia-Driven Regulation and Therapeutic Modulation by Natural Products.” Author attribution corrected from placeholder to actual authors.

Corrections made (7):

[^yum2023] first author WRONG. “Yum MK et al.” → “Yun J, Hansen S, Morris O et al.” — Jina Yun is first author per Crossref and PDF title page.
[^yum2023] ISC mechanism description imprecise. “senescent-cell-induced Ptk7 expression in ISC neighbors” → “the N-terminal extracellular domain of Ptk7, secreted as a SASP factor by senescent fibroblasts.” Ptk7 is secreted BY senescent cells (not induced in ISC neighbors); receptor is FZD7 (not mentioned in wiki); pathway is Wnt/Ca²⁺ (not merely “non-canonical Wnt/YAP”). Body text and footnote both corrected.
[^khalid2024] first author WRONG. “Khalid A et al.” → “Kocełak P et al.” — No “Khalid” author in the paper. Four actual authors named.
[^khalid2024] CV signal terminology corrected. Wiki body said “excess major adverse cardiac events (MACE).” Source (Kocełak 2024 citing Saag 2017) uses “severe cardiovascular adverse events” — not “MACE” specifically. Added specific rates (2.5% vs 1.9%) and note that FRAME trial did not find increased CV risk vs placebo. Updated footnote with full ARCH reference chain.
[^chen2024osteoblast] first author WRONG. “Chen X et al.” → “Zhu S, Chen W, Masson A, Li YP.” First author is Siyu Zhu; Wei Chen is second author and co-corresponding author.
[^chen2024osteoblast] scope description imprecise. “comprehensive review of Wnt/BMP/PTH axes” understates scope — paper covers 9 signaling pathways (Wnt, BMP, TGF-β, Hedgehog, PTH, FGF, Ephrin, Notch, Hippo, Piezo1/2) and multiple transcription factor families. Body text updated from “Wnt/BMP/PTH axes” to full scope.
[^oglcnac2024] first author WRONG. “Zhang Y et al.” → “Xu L et al.” — Xu Lin is first author per Crossref.

Unverifiable claims:

[^oglcnac2024] Xu 2024: specific molecular mechanism details (OGT post-translational modification sites, in vivo model details beyond abstract) — not_oa; abstract-level only. Tagged no-fulltext-access.
[^romosozumab2026] Nuzzo 2026: full case series results (which LRP5/LRP6/WNT1 variants, what response size). DOI not in archive.
[^fwci-sclerostin2026] Yeon 2026: specific hypoxia mechanism details, which natural products studied. DOI not in archive.

Cross-link integrity: All 8 cross-linked pages from new sections confirmed to exist: runx2, matrix-gla-protein, bmp-2, osteopontin, fgf23, osteocytes, vascular-calcification, arterial-stiffening. sost and dkk1 confirmed absent (both tagged #gap/needs-page in body — correct per existing annotation).

Supersession check (R25): PubMed 2018–2026 search for “romosozumab cardiovascular safety” returned 77 results. Most recent large real-world study (PMID 41961089, 2026, Osteoporosis Int, propensity-matched cohort): romosozumab linked to lower all-cause mortality vs teriparatide, no added CV risk — does NOT supersede existing framing (which correctly presents the ARCH vs FRAME discordance as unresolved). The existing #gap/contradictory-evidence tag is appropriate. No meta-analysis of RCTs changes the ARCH/FRAME discordance framing. literature-checked-through: 2026-05-23 confirmed accurate.

Downstream propagation needed:

No downstream atomic pages cite the now-corrected footnotes directly (the wnt-beta-catenin page is the only consumer of [^yum2023], [^chen2024osteoblast], [^khalid2024], [^oglcnac2024], [^romosozumab2026], [
If a studies/yun-2023-ptk7-isc-senescence.md study page is seeded in a future round, use “Yun J” (not “Yum MK”) as first author.

[2026-05-23] verify | molecules/proteins/osteopontin.md

Page flipped: verified: true (partial scope — see verified-scope frontmatter).

Sources checked:

Steitz 2002 Am J Pathol (doi:10.1016/S0002-9440(10)64482-3) — LOCAL PDF read end-to-end (12 pp). Verified.
Wada 1999 Circ Res (doi:10.1161/01.res.84.2.166) — PubMed abstract verified; full PDF still downloading at verification date.
Lehallier 2019 Nat Med (doi:10.1038/s41591-019-0673-2) — LOCAL PDF read end-to-end (27 pp). Verified.
Soraci 2024 Clin Kidney J (doi:10.1093/ckj/sfae336) — PubMed abstract verified.
Quesnel 2024 Alzheimer’s Dementia (doi:10.1002/alz.14065) — PubMed abstract verified.
Yu 2023 Front Immunol (doi:10.3389/fimmu.2023.1175490) — PubMed abstract verified.
Yuan 2025 IUBMB Life (doi:10.1002/iub.70038) — PubMed abstract verified. Remains not_oa.
Giachelli 2005 Circ Res (doi:10.1161/01.RES.0000161997.24797.c0) — not_oa; not verified.
Shapses 2003 Calcif Tissue Int (doi:10.1007/s00223-002-1090-x) — not_oa; not verified.
UniProt P10451, NCBI Gene 6696, HGNC 11255, Ensembl ENSG00000118785 — confirmed via live UniProt REST API.

Corrections applied (7):

Lehallier 2019 “top hit” overstatement corrected. Wiki claimed OPN was “one of the top hits” / “one of the proteins exhibiting a pronounced age-associated increase” in Lehallier 2019’s main-text findings. After reading the full PDF: OPN/SPP1 is NOT named in the main text; top-named proteins are SOST, GDF15, ARFIP2, CHRDL1, MMP12, PTN, SELL, TFPI, ADAMTS13. Body text corrected to state OPN is in the supplementary tables but not among the main-text named top hits; a gap tag added for the specific DE-SWAN rank.
Soraci 2024 sex-direction WRONG → corrected. Wiki stated “stronger associations in men.” PubMed abstract confirms pOPN associated with all-cause mortality in WOMEN (HR 1.84, 95% CI 1.20–2.89) but not men; associated with eGFR decline in all patients. Body text, analysis paragraph, and footnote all corrected. n=2,013 added (wiki had only “SCOPE cohort (older adults ≥75)”).
Quesnel 2024 plasma → CSF corrected. Wiki framed as “plasma proteomics.” Paper studies CSF OPN across PREVENT-AD (n=109) and ADNI (n=167 CU + 399 MCI). Body text and footnote both corrected.
Steitz 2002 mouse strain corrected. Wiki implied C57BL/6; paper uses 129/SvJ × Black Swiss background mice. Corrected in body and footnote.
Steitz 2002 n added. “n=not stated” → “n=4–8 per group” (paper states “n = 4 to 8”). Corrected in body and footnote.
Steitz 2002 calcification values added. Body text and footnote now include the quantitative Ca content data: OPN+/+ ~3000±600, OPN-/- ~6900±250 mmol Ca/L/g protein.
Wada 1999 species corrected. Wiki said “rat/bovine VSMC”; paper uses bovine aortic smooth muscle cells (BASMCs) only. Dose range (0.05–5 µg/mL) added to body and footnote.

Additional clarifications (not factual errors, but nuance improvements):

Steitz 2002 “OPN administration promotes regression” reframed: the paper did not administer exogenous OPN to test regression — it observed that natural OPN accumulation correlated with regression in OPN+/- mice at day 30. New mechanism (carbonic anhydrase II induction → local acidification → mineral dissolution) documented.
Wada 1999 model updated to “GFAV subcutaneous implant” (not generically “vasculature”).
Yu 2023 context clarified as colorectal cancer (not aging per se) in body text and footnote.
Lehallier 2019 footnote expanded with the paper’s actual top-named proteins and DE-SWAN three-wave-crests finding (ages 34, 60, 78).

Unverifiable claims:

Giachelli 2005 review claims (OPN structure, isoform biology, phosphorylation details) — not_oa; cannot verify from primary source.
Shapses 2003 bone mineral phenotype — not_oa; cannot verify.
Yuan 2025 SASP-regulatory p53 mechanism — not_oa; verified against abstract only; mechanism (SPP1 knockdown reduces p53 activation / SASP in ALI model) confirmed from abstract.

Supersession check (R25):

PubMed 2024–2026 search for OPN-aging meta-analyses and large RCTs: no meta-analysis or large RCT contradicting the page’s primary claims found.
Emerging papers (2025–2026): OPN continues to appear in SASP, osteoarthritis pain (Abbas 2026, Mol Pain, PMID 41622147), and CAA (Ersözlü 2026, Neurobiol Aging, PMID 42033997) contexts — consistent with aging-biomarker framing. No supersession.
The OPN-calcification meta-analysis hit (PMID 38431384) was a kidney-stone/CVD meta-analysis referencing OPN in passing; not an OPN-calcification supersession.
literature-recency-verified: 2026-05-23 added to frontmatter.

Cross-link integrity:

runx2, bmp-2, matrix-gla-protein pages: FOUND (seeded in same round).
macrophages, bone pages: MISSING (noted as needs-page in cross-links section — already flagged correctly on the page).
sasp, vascular-calcification, arterial-stiffening pages: FOUND.

Downstream propagation needed (for main agent):

processes/vascular-calcification.md — if it cites OPN claims independently, verify the anti-calcification / context-dependent framing is consistent with the corrections above (particularly the GFAV model context vs. direct vascular model).
Any page that inherited the “OPN is a top hit in Lehallier 2019” framing should be corrected (search: lehallier2019 + OPN or osteopontin).
processes/sasp.md — if OPN is listed there, confirm the SASP-component claim notes the cancer (CRC) context of the strongest scRNA-seq evidence (Yu 2023).

Files touched: molecules/proteins/fgf23.md (new), processes/vascular-calcification.md (new), interventions/dietary/phosphate-additive-reduction.md (new), molecules/proteins/klotho.md, molecules/proteins/fgf21.md, phenotypes/arterial-stiffening.md, phenotypes/atherosclerosis.md, interventions/dietary/mediterranean-diet.md, frameworks/intervention-classes.md, log.md.

[2026-05-23] verify | molecules/proteins/runx2.md

Verifier: claude (wiki-verifier subagent)

Sources checked:

Komori 1997 (Cell, doi:10.1016/s0092-8674(00)80258-5) — local PDF verified end-to-end (10 pages).
Otto 1997 (Cell, doi:10.1016/s0092-8674(00)80259-7) — local PDF verified end-to-end (7 pages).
Katano 2026 (J Vasc Res, doi:10.1159/000550336) — local PDF downloaded and verified (pages 1–3; key abstract + methods + Table 1 confirmed).
Voicu 2026 (J Transl Med, doi:10.1186/s12967-026-07686-1) — local PDF downloaded and verified (pages 1–3; abstract + methods confirmed).
Al Akhdar 2026 (Cells, PMID 41827899) — DOI resolved via PubMed efetch: 10.3390/cells15050466.
Lu 2026 (J Mol Histol, doi:10.1007/s10735-026-10749-8, PMID 41758410) — PubMed abstract verified.
Kim 2026 (Aging Cell, doi:10.1111/acel.70476, PMID 41992429) — PubMed abstract verified.
UniProt Q13950 — live REST API query: 521 aa confirmed; Runt domain 101–229 confirmed; PST-rich region 336–521 (not 340–460 as drafted).

Corrections made (8): (1) PST region residues ~340–460 → ~336–521 per UniProt live annotation; separate “C-terminal disordered” domain row merged into PST row; (2) Komori 1997 mouse strain corrected: “C57BL/6 × 129” → 129-derived E14 ES cells × C57BL/6 background; footnote expanded with osteoblast staging and cartilage nuance; (3) Otto 1997 mouse strain corrected: “CBA/Ca × C57BL/6” → 129-derived GK129/R1.3 ES cells × C57BL/6 (CBA/Ca not in paper); quantitative n values added (152 embryos, 170 pups); CCD-causal-gene demonstration added; (4) VSMC wikilink corrected: vascular-smooth-muscle-cells → vascular smooth muscle cells; (5) Voicu 2026 corrected: dual-targeting design (shRunx2 + shSmad3 in separate arms, not combined) clarified in body and footnote; (6) Al Akhdar 2026 DOI resolved: 10.3390/cells15050466; miR-222/miR-30/β-catenin mechanism detail added; (7) needs-page tags removed for bmp-2, osteopontin, osteocytes (those pages now exist); (8) Lu 2026 and Kim 2026 footnotes enriched with PMIDs and pathway-inhibitor details; “archive: pending” removed.

Supersession check (R25): 0 meta-analyses or large RCTs on RUNX2 + vascular calcification found in 2023–2026 PubMed search. No supersession. literature-checked-through: 2026-05-23 confirmed.

Downstream propagation needed: (a) processes/vascular-calcification.md — check for vascular-smooth-muscle-cells wikilink; correct to vsmc; (b) study pages komori-1997-cbfa1-null-bone.md and otto-1997-cbfa1-osteoblast-ccd.md referenced but not yet seeded — flag as seeding candidates.

[2026-05-23] verify | molecules/proteins/fgf23.md

Verifier: claude (wiki-verifier subagent)

Sources checked:

Urakawa 2006 (Nature, doi:10.1038/nature05315) — local PDF verified end-to-end. Confirms Klotho converts FGFR1c into high-affinity FGF23 receptor; FGF23 signaling reconstituted by Klotho + FGFR1(IIIc) co-expression. No corrections required for mechanism claims.
Henry 2022 (Circulation, doi:10.1161/CIRCULATIONAHA.121.056663) — local PDF verified end-to-end. Study is a broad 90-protein proteomics + MR screen; FGF23 was one of 8 proteins with robust multiverse-concordant MR signal and one of 5 with protective MR direction. Cross-trait MR: higher genetically-predicted FGF23 → lower eGFR. Footnote updated with full consortium attribution and corrected framing.
Liu 2022 (Front CV Med, doi:10.3389/fcvm.2022.989574) — local PDF verified. n=135,576, 29 prospective studies confirmed. Per-doubling CIs added: CV mortality RR 1.43 (95% CI 1.09–1.88); all-cause mortality RR 1.37 (95% CI 1.15–1.62). All other RRs confirmed accurate.
Gutierrez 2008 (NEJM, doi:10.1056/NEJMoa0706130) — verified against PubMed abstract (bronze OA PDF blocked by Cloudflare). ORs confirmed. Q4 CI (2.6–12.6) added to footnote. Full PDF unverifiable — tagged no-fulltext-access.
Sharma 2021 (JAGS, doi:10.1111/jgs.16910, PMC8175094) — verified against PMC full text. Correction: n=2,496 → 2,763. Median follow-up 8.3 yr. HRs populated: all-cause mortality HR 1.24 (95% CI 1.12–1.37)/2-fold iFGF23; CV mortality HR 1.31 (95% CI 1.11–1.54)/2-fold; Q4 vs Q1 all-cause HR 1.31 (1.05–1.62); CV mortality HR 1.54 (1.08–2.18).
Parker 2010 (Ann Intern Med) — closed-access; full text inaccessible. Tagged no-fulltext-access.
Kurosu 2006 (JBC) — download failed (hybrid OA, no candidate URLs). Tagged no-fulltext-access.

Corrections made: (1) Sharma 2021 n: 2,496 → 2,763; (2) Sharma 2021 body: HRs populated from full text; (3) Liu 2022 per-doubling CIs added to footnote; (4) Gutierrez 2008 Q4 CI (2.6–12.6) added to footnote; (5) Henry 2022 framing expanded (broad 90-protein screen; eGFR cross-trait MR noted; consortium attribution corrected in footnote); (6) Liu 2022 footnote: local PDF path added, all endpoint RRs enumerated; (7) Druggability section: removed “verifier should…” call-to-actions; (8) Limitations: updated gap tags, removed seeder’s verifier instructions; (9) verified flag flipped to true with partial-scope documented.

Supersession check (R25): PubMed 2023–2026 search returned 1 result (cinacalcet hyperparathyroidism; unrelated). No superseding FGF23-mortality meta-analysis found. literature-checked-through: 2026-05-23 retained.

Downstream propagation needed:

studies/sharma-2021-fgf23-cause-mortality-habc.md — if seeded, update n from 2,496 to 2,763 and populate HRs from PMC full text
molecules/proteins/klotho.md — check whether Kurosu 2006-specific SPR binding claims appear; those are unverified pending Kurosu full text

[2026-05-23] verify | interventions/dietary/phosphate-additive-reduction.md → verified:true

Sources verified:

Calvo 2023 (doi:10.3390/nu15163510) — full 14-page PDF read. Downloaded and verified.
Gutiérrez 2010 (doi:10.1093/ndt/gfq316) — full 8-page PDF read. Downloaded and verified.
Dhingra 2007 (doi:10.1001/archinte.167.9.879) — PubMed abstract only (not_oa). n=3,368, HR=1.55 (95% CI 1.16–2.07, p=0.004) confirmed. Quartile mg/dL cutoffs NOT in abstract; tagged unsourced.
Foley 2008 ARIC (doi:10.1016/j.ahj.2008.05.016) — PubMed abstract only (not_oa). n=15,732 confirmed, follow-up 12.6 yr confirmed.
Ritz 2012 (doi:10.3238/arztebl.2012.0049) — PubMed abstract only (not_oa). Policy/public-health framing confirmed.
Karp 2007 (doi:10.1007/s00223-007-9011-7) — PubMed abstract only (not_oa). n=16 confirmed; no FGF23 measured confirmed.
Isakova 2011 (doi:10.1093/ndt/gfq419) — DOI lookup failed. Abstract verified via PubMed (PMID 20631407, CKD pilot RCT, n=39, dietary restriction + binders reduced FGF23).
Calvo & Tucker 2013 (doi:10.1111/nyas.12300) — failed to download; not_oa via archive. Could not be directly verified.

Corrections made (6):

Gutiérrez 2010 misattribution — CRITICAL. Wiki claimed this paper “demonstrated that a high-phosphate meal produces measurable acute increases in serum phosphate and FGF23 within hours.” FALSE. The paper’s key finding was racial differences in urinary phosphate excretion; postprandial FGF23 did NOT significantly change (paper explicitly states this); serum phosphate also did not significantly differ over time. Corrected both in body text and footnote.
FGF23-elevates-within-hours mechanism claim — removed as stated; replaced with accurate framing that acute FGF23 response is modest and sustained loading likely required (per Calvo 2023 citing Volk 2022 and Nishida 2006).
Foley ARIC n=13,340 → 15,732. Body text and footnote corrected.
Foley ARIC endpoint specificity corrected. Wiki said “increased CV events” without distinction; ARIC found significant association for stroke (HR 1.15) and mortality (HR 1.15) but NOT coronary heart disease (HR 1.03, NS). Corrected to reflect actual endpoint specificity.
Bioavailability percentages (80-100% vs 40-60%) not explicitly stated in Calvo 2023. Calvo 2023 uses qualitative language only (“rapidly and more completely absorbed”). Corrected to reflect this; added gap marker noting these are review-level synthesis estimates.
Karp 2007 n added to footnote: n=16 women (crossover design, 5 sessions each as own control). Only phosphate supplement (not whole foods) significantly increased PTH; FGF23 not measured.

Unverifiable claims:

Dhingra 2007 quartile cutoff values (~3.5–4.5 vs <3.0 mg/dL) — paper is closed-access; abstract does not provide; tagged unsourced.
Per-serving phosphate-additive magnitudes in the sources table (Kraft Singles, cola, processed meats) — these are review-level estimates, not directly verifiable against wet-chemistry primary data; documented as such throughout.
Calvo & Tucker 2013 content (320–470 mg/day additive estimate; 17–30% rise) — paper failed to download; cited in wiki footnote but not independently verified.

Supersession check (R25): PubMed eutils search (2023–2026) for phosphate additives + cardiovascular + RCT/meta-analysis in non-CKD adults returned 0 results. No superseding trials found. ClinicalTrials.gov v2 API confirmed no active non-CKD primary-prevention trials on dietary phosphate additive reduction. literature-checked-through: 2026-05-23 confirmed accurate.

Cross-link integrity: fgf23 and vascular-calcification sibling pages confirmed to exist at correct paths. intervention-classes.md section at line 1837 confirmed present and accurate (note: the 80-100%/40-60% percentages also appear in intervention-classes.md definition text — see downstream propagation note below).

Downstream propagation needed:

frameworks/intervention-classes.md line 1841 — contains the uncaveated “~80–100% efficiency” / “approximately 40–60% efficiency” language as stated fact. Should be qualified the same way as the corrected intervention page (review-level synthesis estimates, not explicitly stated in Calvo 2023). Main agent should propagate.
molecules/proteins/fgf23.md — the “FGF23 elevates within hours after a high-phosphate meal” claim cited with [^gutierrez2010] may appear on the FGF23 protein page; should be checked and corrected there to reflect the actual Gutiérrez 2010 findings (racial differences in excretion; no acute FGF23 response). Main agent should check and propagate.

[2026-05-23] verify | processes/vascular-calcification.md → verified:true (partial scope)

Sources checked:

Li/Wang 2023 (doi:10.3389/fnut.2023.1115069) — full 12-page gold OA PDF downloaded and verified (local path confirmed). 14-RCT systematic review + meta-analysis.
Murali 2023 (doi:10.1161/jaha.123.031676) — full 17-page gold OA PDF downloaded and verified. 49-RCT systematic review.
de Vries 2025 (doi:10.3390/nu17050815) — full 11-page PDF verified from local DOI lookup.
Liang 2024 (doi:10.7717/peerj.18063) — first 6 pages (mechanism sections) verified from downloaded gold OA PDF.
Knapen 2015 (doi:10.1160/th14-08-0675) — not_oa; verified against Crossref abstract and as cited in Li/Wang 2023 Table 2 (n=244, 180 µg MK-7, 36 months).
Campos/Faul 2025 (doi:10.1093/ndt/gfaf001) — download failed twice (status: failed after retry); mechanistic claims cross-checked against Liang 2024 which covers the same phosphate→PiT-1/2→ERK/RUNX2 literature; tagged no-fulltext-access in footnote.
Clayton 2023 (doi:10.1161/HYPERTENSIONAHA.123.21392) — previously verified on arterial-stiffening.md; consistency check only (no re-read).

Corrections made (6 substantive):

de Vries 2025 population scope wrong — CRITICAL. Wiki claimed “n=165 postmenopausal women.” PDF confirms: 165 women total = 78 pre/peri-menopausal + 87 post-menopausal. Corrected throughout body and footnote.
de Vries 2025 BP result misattributed to overall MK-7 group. Wiki stated “BP at brachialis decreased −3.0% ± 9.0 (p=0.007) in the MK-7 group.” PDF (Tables 2 and 4) shows the overall post-menopausal MK-7 group BP result was p=0.91 (NS). The −3.0% / p=0.007 result was specific to the post-menopausal HIGH stiffness index subgroup (SI >9.83; n=26). Body text and footnote corrected to accurately represent this as a subgroup result.
Li/Wang 2023 n attribution misleading. Wiki said “14 RCTs, n=1,533” for the CAC-specific result. The n=1,533 is total enrolled across all 14 RCTs. The CAC-specific meta-analysis used only 4 studies with n=424 analyzed (219 experimental, 205 control). Corrected in body and footnote; dominant weight of Shea 2009 (94.4%) now documented.
Li/Wang 2023 dp-ucMGP CI wrong in footnote. Wiki had “p=0.0001 vs control” without CI. Corrected to full result: MD −243.31 (95% CI −366.08 to −120.53); p=0.0001; I²=71% (7 RCTs, n=578).
Murali 2023 vitamin K classification overstated as negative. Wiki said “did not show consistent benefit” and framed this as effectively negative. Murali 2023 Table 8 actually classifies vitamin K as “possibly reduces progression” — same tier as several other agent classes. Vitamin K was NOT placed in the “unlikely to reduce” category. Corrected framing; exact Murali classification language now stated.
“CKD-driven positivity” discordance framing incorrect. The seeder framed the Li 2023 positive CAC result as “likely driven by CKD/hemodialysis trial populations.” PDF analysis reveals Shea 2009 (community-dwelling adults) contributed 94.4% of the weight in the fixed-effects CAC model. The three CKD-range trials had negligible weight. The discordance with Murali 2023 is better attributed to different inclusion criteria, endpoint heterogeneity, and effect concentration in a single dominant trial — not a CKD population artifact. Discordance framing section substantially revised.

Additional data verified as correct (no change needed):

CAC result: MD −17.37, 95% CI −34.18 to −0.56, p=0.04 — confirmed from Li 2023 Figure 3A.
Knapen 2015 n=244 (120 MK-7 / 124 placebo) — confirmed from Crossref and Li 2023 Table 2.
de Vries 2025 Young’s modulus result (placebo +49.1% ± 77.4 vs MK-7 +9.4% ± 67.1; p=0.035) — confirmed from Table 2.
Murali 2023: 49 RCTs, n=9,901, median 104, median 12 months — confirmed from abstract.
Liang 2024 mechanism statements (RUNX2, PiT-1/PiT-2, ERK1/2, BMP-2, MSX2, Osterix, NF-κB, Wnt/β-catenin) — all confirmed from PDF pages 4–6.
Mönckeberg spelling: consistent with diacritic in body; both forms retained in aliases (intentional — both spellings standard in literature).
CAC cost ~ $100-400, r a d ia t i o n 1 m S v, M K - 7 cos t$ 15/month, PP >50 thresholds — sanity-checked as clinically reasonable; no correction needed.
de Vries 2025 COI added to footnote: industry-funded (Gnosis by Lesaffre); multiple corresponding authors employed by sponsor.

Unverifiable claims:

Campos/Faul 2025 (doi:10.1093/ndt/gfaf001): phosphate ≥2 mM threshold for VSMC osteogenic transdifferentiation; myocardial calcification as understudied pathology. Download failed; claims consistent with Liang 2024 mechanism sections but not independently verified against this specific review. Tagged no-fulltext-access.

Supersession check (R25):

PubMed 2024–2026 search: vitamin K2 + vascular calcification or arterial stiffness + RCT/meta-analysis — 5 results. None is a superseding large RCT or meta-analysis. AVADEC substudy 2025 (PMID 41100911, n=388, Hasific et al., Atherosclerosis) — evaluated epicardial adipose inflammation, NOT calcification; not a supersession candidate. Other hits: dialysis RCT (n=123), T2DM RCT (n=68), hemodialysis RCT (n=96) — all smaller than Li 2023/Murali 2023, non-CKD-specific or null results consistent with existing framing.
SNF472 search 2024–2026 — 10 results; no Phase 3 CAC results in non-CKD aging population. Two 2026 reviews reference CaLIPSO/CALCIPHYX CKD/calciphylaxis studies only.
No supersession candidates found. literature-checked-through: 2026-05-23 confirmed current.

Downstream propagation needed:

interventions/pharmacological/ or wherever MK-7/vitamin K2 supplementation is discussed: the corrected discordance framing (not CKD-driven; Shea 2009 community trial dominant weight; Murali classifies as “possibly reduces” not negative) should propagate if mentioned.
hallmarks/chronic-inflammation.md, hallmarks/cellular-senescence.md — if they cite vascular-calcification.md, the corrected mechanism summaries are now authoritative.
molecules/proteins/klotho.md — references Klotho-null calcification phenotype; cross-check that the “~40–50% Klotho decline between young adulthood and age 65–70” claim on the vascular-calcification page is consistent with citations on the Klotho page (claim was not changed here; the klotho.md verified page is the canonical source).
No propagation needed for the clayton2023 senolytic claim (unchanged from arterial-stiffening.md verified version).

[2026-05-23] verify | molecules/proteins/matrix-gla-protein.md → verified:true (partial scope)

Sources checked:

Luo 1997 (doi:10.1038/386078a0, Nature) — full 3-page PDF + methods read end-to-end from local archive.
de Vries 2025 (doi:10.3390/nu17050815, Nutrients) — full 11-page PDF re-read from local archive (same PDF used in vascular-calcification.md verification earlier today).
Naiyarakseree 2023 (doi:10.3390/nu15112422, Nutrients) — full 8-page PDF read from local archive.
Berlot 2025 (doi:10.1161/JAHA.124.036459) — archive: pending download; not verified.
Beulens 2009 (doi:10.1016/j.atherosclerosis.2008.07.010) — not_oa; not verified.
Cranenburg 2008 (doi:10.1159/000124863) — not_oa; not verified.
Schurgers 2013 (doi:10.1038/ki.2013.26) — pending download; not verified.
Canonical IDs (UniProt P08493, HGNC 7060, NCBI Gene 4256) confirmed against live UniProt REST and HGNC APIs.

Corrections made (8):

Luo 1997 time-to-death phrase: “Die at ~2 months” → “Die within two months” (paper’s exact phrasing; the “~2 months” approximation was slightly imprecise to the source).
Luo 1997 arterial territory understated: body text said “aorta, coronary, carotid.” PDF (Fig. 2 legend) shows 10 named arterial sites: carotid, cervical trunk, axillary, aortic arch, aorta, intercostal, coeliac, renal, iliac — all elastic and muscular arteries. Paper explicitly states “not arterioles, capillaries or veins.” Body text expanded to reflect the full documented scope.
Luo 1997 cartilage + skeletal phenotype added: Paper documents growth-plate disorganisation, tracheal calcification, manubrium sterni calcification, short stature, osteopenia, and fractures in Mgp-/- mice. These were not mentioned in the body text. Added.
de Vries 2025 population scope — CRITICAL (propagation alert correction applied): body text said “12-month RCT of 165 post-menopausal women.” PDF confirms: 165 women total = 78 pre/peri-menopausal + 87 post-menopausal (not post-menopausal only). Corrected in body text.
de Vries 2025 post-hoc framing added: Paper title and abstract explicitly state “The aim of the post hoc analysis of this one-year study is to investigate…”. Page had called it an RCT without flagging post-hoc nature. Body text and footnote now both label it as a “post-hoc analysis” of the parent NCT02404519 RCT.
de Vries 2025 primary efficacy metric corrected: page said “vascular stiffness progression vs placebo (stiffness increase +9.4% vs +49.1%, p=0.035).” The correct metric is Young’s modulus (Table 2 of PDF; not cfPWV or a generic “stiffness” measure). Added metric name.
de Vries 2025 dose added: MK-7 dose of 180 µg/day was not stated in the body text or footnote. Added (confirmed from Methods §2.2 of PDF).
de Vries 2025 BP result subgroup attribution — CRITICAL (propagation alert correction applied): footnote previously implied BP benefit applied to the overall MK-7 post-menopausal group. PDF (Table 4) confirms the −3.0% / p=0.007 brachial BP result was specific to post-menopausal women with HIGH stiffness index (SI >9.83; n=26 MK-7). Corrected in body text and footnote with explicit subgroup attribution.
de Vries 2025 industry-funding COI added: Gnosis by Lesaffre / Lesaffre International employees among co-authors (Bittner, Gåserød, Jeanne, Machuron); de Vries, Schurgers received research funding from funder. Added to footnote.
Naiyarakseree 2023 control arm cfPWV change corrected: body text said “−6.8%”; PDF (Table 2) reports control ITT −6.7% (−18.8, 7.8). Corrected. Full ITT interval notation added to body text.
Naiyarakseree 2023 diabetic subgroup n added: n=16 MK-7 / n=14 control (Table 2). Added to body text.
Luo 1997 footnote rewritten: removed wikilink to non-existent study page [[studies/luo-1997-mgp-knockout-calcification]]; replaced with full citation including background strain, actual phenotype scope, and note that n per group not specified in paper.

Unverifiable claims:

Berlot 2025 MESA cohort (dp-ucMGP HR 3.05 / HR 2.73; age-stratified discordance between midlife and >65 adults) — archive pending download; body text and footnote marked archive: pending download; contradictory-evidence tag retained as appropriate.
Beulens 2009 menaquinone + coronary calcification association — not_oa; accepted as stated (abstract-level claim consistent with mechanism).
Cranenburg 2008 ucMGP ELISA development — not_oa; body text is low-specificity; no correction possible.
Schurgers 2013 warfarin review claims — pending; mechanism claims are consistent with biology on other verified pages.

Supersession check (R25): PubMed 2023–2026 search for “matrix gla protein OR dp-ucMGP OR menaquinone-7 vascular calcification” + meta-analysis/RCT returned 10 hits. Triaged:

PMID 41100911 (Hasific 2025, AVADEC substudy, n=388, Atherosclerosis): evaluated epicardial adipose inflammation NOT calcification; dp-ucMGP reduction confirmed with supplementation but no significant EAT/PCAT change. Consistent with existing wiki framing (biomarker endpoint ≠ hard outcome). Not a supersession.
PMID 40258362 (Poulikakos 2025, Blood Purif, n=20): MCO vs HFD dialysis membrane; dp-ucMGP used as secondary biomarker. Not a supersession of any MGP page claim.
Other 8 results: small HD/CKD trials or secondary biomarker analyses; none contradicts the wiki framing. literature-checked-through: 2026-05-23 retained as accurate.

Cross-link integrity check: Confirmed existing: vascular-calcification.md, klotho.md, fgf23.md, runx2.md, bmp-2.md, osteopontin.md, chronic-inflammation.md, altered-intercellular-communication.md, arterial-stiffening.md, atherosclerosis.md. Confirmed MISSING (correctly flagged needs-page): ggcx.md, vkor.md, vitamin-k.md, vitamin-k-cycle.md. Note: studies/luo-1997-mgp-knockout-calcification.md does not exist — wikilink removed from footnote; flag as study-page-seeding candidate (Luo 1997 is cited from 2+ pages).

Downstream propagation needed:

processes/vascular-calcification.md — already corrected for de Vries 2025 population scope + BP subgroup by earlier verify pass today. No further action needed.
molecules/compounds/vitamin-k2.md or equivalent (if seeded) — should use the corrected de Vries 2025 framing (post-hoc; 78+87 total; Young’s modulus endpoint; subgroup BP only).
studies/luo-1997-mgp-knockout-calcification.md — does not exist; flag for study-page seeder as Luo 1997 is the canonical MGP loss-of-function paper and load-bearing across vascular-calcification.md.

[2026-05-23] verify | cell-types/vsmc.md → verified:true (partial scope)

Sources checked:

Clayton 2023 (doi:10.1161/HYPERTENSIONAHA.123.21392) — full text read via PMC10530538. Confirmed: ABT-263 arm n=14-16/group; aged C57BL/6N mice (27 months); PWV 446±9 → 356±11 cm/s p<0.0001 (not p<0.05 as seeded); EDD 82±3% → 96±1% p=0.0004; dose 50 mg/kg/day oral gavage 1 wk on/2 wks off/1 wk on; p16-3MR GCV arm: PWV 477±10 → 382±7 cm/s (p<0.05); journal is Hypertension 2023;80(10):2072–2087.
Liang 2024 (doi:10.7717/peerj.18063) — full 22-page PDF read from local archive. Confirmed: BMP2 signals via SMAD1/5 (not SMAD1/5/8 as seeded); PiT-1/PiT-2 mediated phosphate sensing confirmed; high-Pi VSMC cytokine production is TNF-α, IL-1β, IL-6 + BMP-2 (not just TNF-α, IL-6).
Venkatasubramanian 2025 (doi:10.1161/HYPERTENSIONAHA.125.25408) — full text read via PMC12327803. CRITICAL: this study uses young adult C57BL/6 mice (4-6 months) treated with doxorubicin, NOT aged mice. Seeded description “aged C57BL/6 mice” was wrong. Correct ABT-263 comparison: doxo-veh post 404±15 vs doxo-ABT263 341±7 cm/s (p<0.0001). GCV arm: doxo-veh 425±6 vs doxo-GCV 348±4 cm/s (p<0.0001). The “425±6 vs 342±7” values seeded mix the two different study cohorts.
Sá Ferreira 2026 (PMID:41967384) — DOI confirmed: 10.1016/j.biomaterials.2026.124117 (Biomaterials 2026 Oct;333:124117). Seeded as “doi:not-confirmed.” Study includes both in-vitro (decellularized ECM, 2-week EC culture) and in-vivo (S-ECM disc implant in young mice) components — seeded as “in-vitro” only, corrected.
Weng 2026 (PMID:42137359) — DOI confirmed: 10.3389/fendo.2026.1839111 (Front Endocrinol 2026;17:1839111; PMCID:PMC13167474). Seeded as “doi:not-confirmed.” Content description accurate.
Lao 2019, Murtada 2023 — not locally verified (bronze OA / green OA download pending). Cited claims are consistent with their described scope; flagged in verified-scope.

Corrections made (7):

Clayton 2023 p-value: ABT-263 PWV reduction p<0.05 → p<0.0001; body text + footnote + Therapeutic Relevance section all updated.
Clayton 2023 n and strain added: “aged mice” → “aged C57BL/6N mice (n=14-16/group, 27 months old)” in body text; footnote updated with full cohort details.
Clayton 2023 EDD p-value added: 96±1% vs 82±3% p=0.0004 (not previously stated); p16-3MR GCV arm details added to footnote.
Venkatasubramanian 2025 mouse model corrected: “aged C57BL/6 mice” → “doxorubicin-treated young adult C57BL/6 mice (4-6 months; single 10 mg/kg i.p.)” throughout body text, footnote, and gap tags.
Venkatasubramanian 2025 PWV values corrected: “425±6 vs 342±7 cm/s” (mixed-cohort) → “404±15 vs 341±7 cm/s (ABT-263 arm) and 425±6 vs 348±4 cm/s (GCV arm)”; Therapeutic Relevance section now states both arms with correct numbers.
Liang 2024 SMAD1/5/8 → SMAD1/5: BMP-2 pathway description corrected in “Triggers of osteogenic transdifferentiation” section; footnote updated.
Confirmed DOIs added: ferreira2026 (doi:not-confirmed → 10.1016/j.biomaterials.2026.124117) and weng2026 (doi:not-confirmed → 10.3389/fendo.2026.1839111) updated with full bibliographic details and PMCID.

Alias added: vascular-smooth-muscle-cells added to frontmatter aliases: to fix cross-link from molecules/proteins/runx2.md (which uses [[vascular-smooth-muscle-cells]]).

Supersession check (R25): PubMed 2026 search for VSMC senescence/aging returned 10 hits (Jan–May 2026). None are meta-analyses or large RCTs (n>100). No supersession candidates identified. literature-checked-through: 2026-05-23 confirmed.

Unverifiable claims:

Lao 2019 scRNA sub-state claims (contractile/synthetic/osteogenic proportional shifts in aged human aorta) — bronze OA pending download; claims are consistent with the broader VSMC scRNA literature and are hedged with #gap/needs-replication.
Murtada 2023 progeria scRNA claims (osteochondrogenic sub-state, RUNX2/SOX9/COL2A1) — green OA pending download; the progeria model caveat is explicitly documented in body + footnote.
Calponin CNN1 isoform specificity — accepted per Liang 2024 (which uses CNN1/calponin as a VSMC contractile marker) and the VSMC literature consensus; not independently re-checked against CellMarker 2.0 (CellMarker website returned connection error during verification).

Downstream propagation needed:

molecules/proteins/runx2.md — uses [[vascular-smooth-muscle-cells]] wikilink; now resolved by the alias addition to vsmc.md. No edit needed on runx2.md.
phenotypes/arterial-stiffening.md — already carries the correct Clayton 2023 numbers (verified 2026-05-19); consistent with corrections applied here. No propagation needed.
processes/vascular-calcification.md — SMAD1/5 correction (from SMAD1/5/8) should be checked if this page independently cites the BMP2→SMAD signaling axis.
molecules/proteins/bmp-2.md (if seeded) — same SMAD1/5 correction may be needed.

[2026-05-23] verify | molecules/proteins/bmp-2.md → verified:true (partial scope)

Sources checked:

Liang 2024 (doi:10.7717/peerj.18063, PeerJ) — local PDF at ; pages 1–11 read end-to-end (abstract, introduction, survey methodology, all mechanism sections, conclusion, additional information). Load-bearing reference for mechanistic content.
Hao 2023 (doi:10.1161/JAHA.122.027222, J Am Heart Assoc) — downloaded via PMC mirror (PMC9973608); pages 1–8 read. IL-29/JAK2/STAT3/BMP2 claim confirmed.
Zhu 2024 (doi:10.1186/s12964-024-01873-7, Cell Commun Signal) — downloaded via PMC mirror (PMC11468037); pages 1–4 read. Nesfatin-1 mechanism verified.
Wu 2023 (doi:10.1515/med-2023-0861, Open Medicine) — downloaded via PMC mirror (PMC10751896); pages 1–4 read. ENPP1/PPi/BMP-2 claim verified.
Yang 2026 (doi:10.1016/j.athplu.2026.02.002, Atherosclerosis Plus) — NOT in archive; Crossref confirms existence (CC BY-NC-ND, Elsevier tdm-only, no OA); NOT verified.
Canonical IDs (UniProt P12643, HGNC 1069, NCBI Gene 650, Ensembl ENSG00000125845) confirmed via live UniProt REST and HGNC REST APIs.

Corrections made (6):

Propeptide cleavage enzyme: “furin-family proteases” → “PCSK5” per UniProt P12643 feature annotation. Propeptide boundaries made explicit (residues 24–282; mature chain 283–396, 114 aa).
Disulfide bond count — CRITICAL: “seven conserved cysteines form four intramolecular disulfide bonds plus one interchain bond” → “three intramolecular disulfide bonds plus one interchain bond.” UniProt P12643 lists exactly 3 intramolecular disulfides ([296–361], [325–393], [329–395]) + 1 interchain ([360–360]).
Glycosylation sites: residue numbers added (Asn-135, Asn-163, Asn-164, Asn-200, Asn-338) to anchor the “Five N-linked sites” claim against UniProt.
TAK1 attribution to Liang 2024 removed: Liang 2024 does not mention TAK1 anywhere. p38-MAPK phosphorylation of RUNX2 is described in that paper, but is not attributed to a BMP2→TAK1 arm. Body text corrected; TAK1 arm now noted as belonging to the broader BMP-signaling literature (not Liang 2024).
Nesfatin-1 mechanism — CRITICAL: “potentiating BMP-2 receptor signaling” → BMP-2 protein stabilization via inhibition of E3 ubiquitin ligase SYTL4 → downstream pSMAD1/5/9 + HDAC4 phosphorylation → RUNX2 activation + MSX2 upregulation. The mechanism is at the protein-stability level, not the receptor level (Zhu 2024 abstract).
MGP–BMP-2 sequestration sourcing: Liang 2024 does not describe direct MGP–BMP-2 binding. Attribution to [^liang2024] removed; unsourced tag added. Needs a primary citation (Price 2001 / Zebboudj 2002 are typical sources in the field).

Unverifiable claims:

Yang 2026 G9A/EHMT2–BMP2 epigenetic axis — closed-access; tagged no-fulltext-access on claim and footnote.
SMAD9 C-terminal phospho-serine residue numbers — UniProt phospho annotations confirmed for SMAD1 (Ser-463, Ser-465) and SMAD5 (Ser-463, Ser-465); SMAD9 API call returned incorrect data; residues documented as per-homology.

Supersession check (R25): PubMed 2024–2026 “BMP2 vascular calcification” — 45 hits; meta-analysis/RCT filter returned 0. No meta-analysis or large RCT on BMP2 in vascular calcification exists. literature-checked-through: 2026-05-23 confirmed.

Downstream propagation needed:

processes/vascular-calcification.md — the SMAD1/5/8 → SMAD1/5/9 rename should be checked; note that vsmc.md verify pass earlier today (line 1321 above) already applied this correction to that page’s BMP2→SMAD description citing Liang 2024. No further action needed on vascular-calcification.md.
molecules/proteins/matrix-gla-protein.md — if it describes direct MGP–BMP-2 sequestration, it should carry a dedicated primary source citation rather than relying on Liang 2024. Main agent should check.
[[runx2]], [[osteopontin]] — sibling stubs; when seeded, corrected BMP2→SMAD1/5/9 cascade and nesfatin-1/SYTL4 mechanism should inform content.

[2026-05-23] verify | tissues/arteries.md → verified:true (partial scope)

Sources checked:

Mammoto A, Matus K, Mammoto T 2022 (doi:10.3389/fcell.2022.822561) — full 8-page PDF downloaded and verified end-to-end (local path confirmed). Gold OA mini review in Front Cell Dev Biol. Covers three-layer anatomy, elastic fiber fragmentation and matrikine release, collagen crosslinking, endothelial senescence, VSMC synthetic switch and senescence, MMP upregulation, AGE accumulation. Does NOT state a numeric elastin half-life.
Lakatta EG, Levy D 2003 (doi:10.1161/01.cir.0000048892.83521.58, PMID 12515756) — closed-access (not_oa per a local paper archive). Full text unverifiable. Claims attributed to this source retained with no-fulltext-access on the footnote.
Shapiro 1991 (doi:10.1172/JCI115204) — not re-read (closed-access; verified previously on eln.md per R40); claims cross-checked against verified eln.md canonical extraction.
Clayton 2023 senolytic claim — cross-checked against verified arterial-stiffening.md (no re-read of PDF); consistent.

Corrections made (3 substantive):

Elastin half-life claim — CRITICAL attribution + precision error. Wiki stated: “Isotopic tracer studies estimate the half-life of aortic elastin at approximately 70–75 years” (cited to Mammoto 2022 + Lakatta 2003). Mammoto 2022 PDF confirms it does NOT state a numeric half-life. The correct primary source is Shapiro 1991 (doi:10.1172/JCI115204), which measured a mean carbon residence time of ~74 yr (95% CI 40–174 yr) in human LUNG parenchyma (n=14 post-mortem specimens) — not aortic elastin. Aortic elastin metabolic stability is supported by Powell 1992 (PMID 1466664) D-aspartate racemization data but no numeric half-life is available for aortic tissue from any identified primary source. Corrected to: “~74 yr (95% CI 40–174 yr) in human lung parenchyma [^shapiro1991]; analogous aspartate racemization in aortic elastin confirms comparable metabolic stability, though a directly measured half-life for aortic elastin has not been separately radiocarbon-dated.” New [^shapiro1991] footnote added. needs-replication tag added.
Lakatta 2003 footnote updated: Full title (Part I confirmed via PubMed PMID 12515756), PMID added, no-fulltext-access tag added, “archive confirmed (doi verified; PDF not locally downloaded)” → explicit closed-access statement.
Mammoto 2022 footnote corrected: Author list corrected to Mammoto A, Matus K, Mammoto T (three authors, not two); “PDF at Frontiers; not locally downloaded” → “PDF verified”; “ECM composition percentages” removed from scope description (paper does not give specific percentage figures); “does NOT state a numeric elastin half-life” added explicitly.
Limitations section gap tag updated: “elastin half-life estimate (~70–75 years) is from radiocarbon isotope tracer approaches in rodents and extrapolation” → corrected to reflect Shapiro 1991’s actual population (14 human lung specimens, not rodents) and the specific gap (aortic elastin not directly radiocarbon-dated; Powell 1992 confirms non-renewal without numeric half-life).

Data confirmed as correct (no change):

Three-layer anatomy (intima/media/adventitia) — consistent with Mammoto 2022 and structural review consensus.
Matrikine biology (elastin-derived peptides → macrophage M1/M2 + osteoblast-like VSMC) — confirmed in Mammoto 2022 Discussion pp.4–5.
Collagen I and III as primary structural collagens; collagen type-localisation (I around VSMCs, III alongside external elastic lamina) — confirmed in Mammoto 2022 p.3.
Endothelial senescence markers (ICAM-1, VCAM-1, ET-1, ↓NO) — confirmed in Mammoto 2022 Discussion p.4.
Vessel-type taxonomy table (elastic/muscular/arteriole classification) — review-level structural consensus; remains tagged unsourced (no primary histomorphometry citation identified).
VSMC layer count (~50–60 aortic lamellar units) — structural consensus; tagged unsourced; no primary source identified during this pass.
Clayton 2023 ~20% PWV reduction — consistent with arterial-stiffening.md verified values.

Unverifiable claims:

Any specific claims supported only by Lakatta 2003 — closed-access; cannot distinguish Lakatta-sourced from review-consensus statements without full text.

Supersession check (R25): This is a type: tissue page (canonical-data-stable type per CLAUDE.md); formal R25 supersession check is not required. Spot-check PubMed 2025–2026 “arterial ECM aging aorta” returned no new meta-analyses or large RCTs contradicting the structural/mechanistic framing. literature-checked-through: 2026-05-23 confirmed current.

Downstream propagation needed:

molecules/proteins/eln.md — no propagation needed; eln.md is the canonical source; arteries.md now correctly cross-references it.
phenotypes/arterial-stiffening.md — elastin half-life is mentioned qualitatively there (“biological half-life of decades”). If it independently cites a numeric figure, align with Shapiro 1991 lung-specific ~74 yr framing and add needs-replication. Main agent should check.
phenotypes/cardiovascular-aging.md — line currently reads “elastin has a biological half-life of decades”; no numeric value stated; no correction needed but cross-link to eln could be added for precision.

[2026-05-23] verify | molecules/proteins/lrp5-lrp6.md → verified:true

Sources checked:

Gong 2001 Cell (doi:10.1016/s0092-8674(01)00571-2) — full PDF verified (downloaded bronze OA 2026-05-23; 8 pages read end-to-end). 28 families, 17 OPPG patients; lumbar spine BMD Z-score: OPPG patients −4.7 (SD 0.9), obligate carriers −1.3 to −1.4. Ocular phenotype: leukocoria and retrolental membrane from failed regression of fetal tunica vasculosa lentis / primary hyperplastic primary vitreous — NOT “blindness” as seeded.
Zeng 2005 Nature (doi:10.1038/nature04185) — full PDF verified (locally available; 5 pages read). Confirmed as canonical primary source for PPPSP-Axin docking mechanism: GSK3β phosphorylates site I (priming); CK1 phosphorylates site II; dual sequential phosphorylation promotes Axin recruitment to LRP6. Replaces unsourced [^kim2008] placeholder throughout.
Davidson 2005 Nature (doi:10.1038/nature04170) — full PDF verified (locally available; 3 pages read). Confirms CK1γ as the membrane-associated CK1 isoform coupling Wnt receptor activation to LRP6 phosphorylation; complementary to Zeng 2005.
Little 2002 NEJM (doi:10.1056/nejm200209193471216; PMID 12239268) — not_oa; abstract-level via PubMed. Published as a letter (not full research article). Confirms p.G171V LRP5 missense mutation as cause of autosomal dominant HBM. Authors: Little RD, Recker RR, Johnson ML. Framing on page is accurate.
Mani 2007 Science (doi:10.1126/science.1136370; PMID 17332414) — not_oa; full abstract via PubMed efetch. Single-family (Iranian pedigree) LRP6 p.R611C variant associated with early CAD + metabolic syndrome. NOT replicated in population GWAS. Seeded page framing was too permissive (“have been associated with”) without noting single-family status.
Balemans 2008 (not_oa), Levasseur 2005 (not_oa), Yorgan 2015 (download failed), Tanaka 2020 (not_oa) — DOI lookup confirmed; framing retained at abstract level.
Canonical DB IDs confirmed via UniProt REST API: LRP5 O75197 (gene LRP5, NCBI Gene 4041, HGNC:6697, ENSG00000162337, chr11); LRP6 O75581 (gene LRP6, NCBI Gene 4040, HGNC:6698, ENSG00000070018, chr12).

Corrections made (5):

Ocular phenotype in OPPG corrected — “congenital/infancy-onset blindness” → “vision-threatening ocular involvement from failed regression of fetal vitreoretinal vasculature; phenotype ranges from leukocoria to retrolental membrane, severity variable.” Added quantitative BMD Z-score data from Gong 2001 Table 1 (OPPG patients −4.7 ± 0.9).
OPPG family count added — “multiple kindreds” → “28 families, 17 OPPG patients” in footnote and body.
PPPSP mechanism citation corrected — [^kim2008] (#gap/unsourced placeholder) removed throughout; replaced with [^zeng2005] (Zeng et al. 2005 Nature, doi:10.1038/nature04185, dual-kinase mechanism confirmed end-to-end). Mechanism description updated to note dual-sequential GSK3β + CK1 phosphorylation of PPPSPxS motifs (site I then site II).
ADCAD2 / LRP6-CAD framing corrected — seeded text implied established population association; corrected to single-family report (28-member Iranian pedigree, p.R611C in EGF-like domain, Mani 2007 Science). Added needs-replication. Mutation residue specified (R611C, not just “missense”).
PPPSP repeat position numbers corrected — “~1500–1612” → “~1480–1613” per Figure 1a of Zeng 2005 showing motif positions in LRP6.

Unverifiable claims:

Balemans 2008 SOST–LRP5 binding affinity / HBM mutation binding abolition — not_oa; claim plausible and consistent with SOST page content; no-fulltext-access retained on footnote.
Yorgan 2015 Lrp5 HBM G171V knock-in SOST blunting data — download failed; claim framing is abstractly consistent with the broader mouse-model Wnt literature.
Tanaka 2020 romosozumab clinical review — not_oa; citation is for a review paper, low verification priority.

Supersession candidates (R25): No meta-analyses or large RCTs on LRP5/6 specifically as drug target have been published since 2023. Romosozumab (anti-SOST, restoring LRP5/6 engagement) is the current approved endpoint — SOST page already carries the FRAME + ARCH RCT data (verified 2026-05-23). The electroacupuncture-Wnt/LRP5 rat study (PMID 39460675, Acupuncture in Medicine 2024) is not supersession-relevant. No contradicting meta-analysis found. literature-checked-through: 2026-05-23 confirmed.

Cross-cluster consistency: SOST steric-blockade vs DKK1 Kremen-internalization mechanism distinction confirmed consistent across all three pages (lrp5-lrp6.md, sost.md, dkk1.md). Note: sost.md and dkk1.md both carry #gap/needs-page for lrp5-lrp6 — this tag is now stale on both sibling pages (lrp5-lrp6.md exists). Main agent should remove stale gap tags on those two pages.

Downstream propagation needed:

molecules/proteins/sost.md — #gap/needs-page for lrp5-lrp6 is stale; remove.
molecules/proteins/dkk1.md — #gap/needs-page for lrp5-lrp6 is stale; remove (dkk1.md already verified 2026-05-23; this is a clean one-line edit).
hallmarks/altered-intercellular-communication.md and hallmarks/stem-cell-exhaustion.md — if they cite LRP5/6 in the context of bone or stem-cell-niche Wnt signaling, check whether the ADCAD2 framing is more conservative (single-family) than what they state.

[2026-05-23] verify | cell-types/endothelial-cells.md → verified:true (partial scope)

Verifier: claude (wiki-verifier subagent)

Sources checked:

Hwang 2022 Int J Mol Sci (doi:10.3390/ijms231710135) — full 15-page PDF read end-to-end
Janaszak-Jasiecka 2023 Cell Mol Biol Lett (doi:10.1186/s11658-023-00423-2) — PDF pages 1–12 of 28 read (local); full abstract + eNOS uncoupling mechanism sections verified; gold OA
Dobner 2024 Angiogenesis (doi:10.1007/s10456-023-09904-6) — PDF pages 1–10 of 17 read (local); abstract, brain/muscle/lung/heart/kidney/skin tissue sections and Table 1 verified; hybrid OA
Bloom 2023 Nat Rev Cardiol (doi:10.1038/s41569-022-00739-0) — verified via PMC full text PMC10026597 (local PDF download failed HTTP 403; green OA PMC version used as primary)
Rossman 2017 AJP Heart Circ Physiol (doi:10.1152/ajpheart.00416.2017) — key quantitative data from PubMed efetch abstract (PMID 28971843; bronze OA; local PDF downloading at time of verification — Cloudflare-blocked); n-per-group, % increases, r-values all confirmed from abstract
Cell Ontology CL:0002138 / CL:0002139 — verified via EBI OLS4 API 2026-05-23

Corrections made (8):

CRITICAL: Cell Ontology ID error — CL:0002138 vs CL:0002139. Seeder wrote CL:0002138 ("endothelial cell of vascular tree") in both frontmatter comment and body. Verified via EBI OLS4: CL:0002138 is “endothelial cell of lymphatic vessel” (LEC synonyms). The correct term for “endothelial cell of vascular tree” is CL:0002139. Corrected in frontmatter comment, Cell Ontology body section, and Limitations gap tag. CL:0000115 retained as primary frontmatter cell-ontology-id: per design decision.
eNOS Ser1177 attribution misattributed to Hwang 2022. Wiki cited [^hwang2022] for “reduced Akt-mediated Ser1177 phosphorylation” in aged ECs. Hwang 2022 full PDF confirms: Ser1177 appears only in Figure 2 caption as framework context; no primary quantitative aging-vs-young Ser1177 measurement is made. The phosphorylation-site mechanism (Ser1177/Ser633/Ser615 activating; Thr495/Ser114 inhibiting; Akt/PKA/AMPK as kinases) is sourced to Janaszak-Jasiecka 2023 p.3. Attribution corrected to [^janaszak2023]; gap tag added for human quantification.
EC SASP cytokine list corrected (Bloom 2023). Wiki listed “IL-6, IL-8, ICAM-1, VCAM-1, MCP-1, and MMPs.” Bloom 2023 PMC text states: IL-1β, IL-6, IL-8, CXCL11, PAI-1, ICAM-1, VCAM-1, and reduced IL-10. MCP-1 and MMPs are not Bloom 2023’s primary EC SASP list. Citation corrected from [^hwang2022] to [^bloom2023]; cytokine list updated.
eNOS uncoupling BH4 oxidation agents clarified. Original: “oxidized by peroxynitrite.” Janaszak 2023 specifies: “rapidly oxidized by superoxide anion or, especially strongly, by peroxynitrite.” Both agents involved; body text updated to reflect both and the vicious-cycle mechanism; BH2 competitive inhibition at heme oxygenase domain clarified (not dimer disruption).
Rossman 2017 quantitative data added to body and footnote. Exact n-per-group: venous ECs — young sedentary n=9, older sedentary n=12, older exercising n=13; arterial ECs — young sedentary n=9, older sedentary n=18, older exercising n=14. Marker increases (older sedentary vs young): venous p53 +116%, p21 +119%, p16 +128% (all P<0.05); arterial p53 +26%, p21 +23% (P<0.05). FMD correlations (venous EC): p53 r=−0.49 p=0.003; p21 r=−0.38 p=0.03; p16 r=−0.58 p=0.002.
Dobner 2024 pages and authors wrong. Footnote: “Angiogenesis 2024;27:1–22” → “27:129–145.” Authors: “Dobner S, Schaper J, Goswami N, et al.” → “Dobner S, Tóth F, de Rooij LPMH” (CeMM Vienna).
“Clayton 2023” navitoclax EC-dilation citation removed — unsourced. Intervention table cited “Clayton 2023” for navitoclax improving EC-dependent dilation in aged vessels. Bloom 2023 does not cite this. Citation removed; text rewritten to Bloom 2023’s actual content (dasatinib+quercetin → carotid ED-dilation + reduced uncapped telomeres; navitoclax → selectively kills senescent ECs in vitro). Gap tag added.
Dobner 2024 single-cell section rewritten to reflect actual tissue-by-tissue Table 1 data. Original section overstated ICAM-1/VCAM-1 as “primary scRNA finding” (these are protein markers; Dobner 2024 reports gene transcriptomic signatures). Replaced with tissue-specific gene signature data from Table 1: brain (Vcam1/B2m/Cxcl12/IFN genes↑, Cldn5/Ocln↓), heart (coronary ECs most senescence-prone, FOXO3A↓), muscle (CCL2+ veins↑, circadian TFs↓), kidney (glomerular ECs most prone, Serpine1/PAI-1↑), lung (Apln↓), skin (NOTCH3-HES1↓).

Unverifiable claims:

FMD rate “0.2–0.5%/decade” — not stated in Rossman 2017 abstract; appears to be field estimate; retained with hedge and needs-replication, Rossman attribution removed for this specific rate.
“Clayton 2023” navitoclax claim — source not identified; removed per correction #7.

Supersession check (R25): type: cell-type is canonical-data-stable per CLAUDE.md — formal R25 check not required. Spot-check: 10 PubMed hits 2024–2026 for “endothelial senescence eNOS aging vascular”; all mechanistic/in-vitro; no meta-analyses or large RCTs contradicting page framing. literature-checked-through: 2026-05-23 confirmed current.

Downstream propagation needed:

cell-types/vsmc.md — if its SASP cytokine list includes MCP-1 as EC-derived (citing Bloom 2023), update to corrected list. Cross-cluster SASP consistency check requested by main agent.
molecules/proteins/enos.md (stub/existing) — Ser1177/Ser633/Ser615 attribution should be Janaszak 2023, not Hwang 2022.
hallmarks/cellular-senescence.md and phenotypes/atherosclerosis.md — check for MCP-1 attribution to Bloom 2023 EC SASP list; correct if present.
Any page citing “Clayton 2023” navitoclax EC dilation — review for independent sourcing.

Final verified state: verified: true with scope — “Hwang 2022, Janaszak 2023, Dobner 2024 PDFs read locally; Bloom 2023 verified via PMC10026597; Rossman 2017 key quantitative data from PubMed abstract (local PDF pending); Cell Ontology IDs verified via EBI OLS4.”

2026-05-23 verify — molecules/proteins/dkk1.md flipped verified:true (partial scope — 2 primary sources not_oa). PDFs read: Purro 2014 J Mol Cell Biol (review, 6 pp), Mace 2022 JBMR Plus (original article, 8 pp), Fang 2025 J Cell Sci (original article, 4 pp of 10+), Gau 2022 IJMS (review, 4 pp of 18). 6 corrections: (1) Purro 2014 re-attributed as review: wiki cited it as originating DKK1-AD elevation data and antibody-rescue endpoint; it is a narrative review. Primary sources are Purro et al. 2012 J Neurosci 32:3492–3498 (synapse rescue) and Caricasole et al. 2004 J Neurosci (human AD brain DKK1 elevation). Footnote updated accordingly. (2) Fang 2025 mechanism corrected: wiki stated “Aβ activates WNT5A-ROR2 pathway involving elevated DKK1, triggering JNK signaling” — this conflates two parallel toxic stimuli. The paper shows Aβ and DKK1 each independently activate WNT5A-ROR2→JNK; DKK1 is not an intermediary in the Aβ→JNK pathway. Body text and footnote corrected with quantitative JNK C:N ratios. (3) Mace 2022 model precision added: “CKD rat” → “5/6-nephrectomized Wistar rat”; “primary osteoblasts” implied → UMR-106 osteoblast-like cell line specified; DKK1 secretion quantities added (119 [78–551] pg/mL at 24h; 353 [110–686] pg/mL at 48h); complete abolition of mineralization quantified. (4) DKN-01 trial status resolved against ClinicalTrials.gov v2 API: 2 ACTIVE_NOT_RECRUITING trials confirmed (NCT04166721 phase 1/2 gastric/esophageal; NCT05761951 phase 2 endometrial); MM phase 1 trials (NCT01457417, NCT01711671) are completed. (5) Stale gap tags removed: #gap/needs-page for sost and lrp5-lrp6 removed — both pages now exist. (6) Unverifiable sources tagged: Bayod 2015 (not_oa, no-fulltext-access) and Carrillo 2021 (not_oa, no-fulltext-access). Canonical IDs confirmed live: UniProt O94907, NCBI Gene 22943, HGNC 2891, Ensembl ENSG00000107984. Supersession: 5 PubMed hits 2024–2026 on DKK1+aging+Alzheimer; all mechanistic/in-vitro; no meta-analyses or large RCTs — no supersession. Downstream propagation needed: (a) any page citing Purro 2014 as a primary data source should be updated to cite Purro 2012 J Neurosci or Caricasole 2004; (b) hallmarks/chronic-inflammation.md and hallmarks/deregulated-nutrient-sensing.md if they cite DKK1-specific data; (c) phenotypes/alzheimers-disease.md if it cites Purro 2014 as primary for DKK1 elevation data. literature-checked-through: 2026-05-23.
2026-05-23 verify — tissues/bone.md flipped verified:true (partial scope). PDFs read: Farr 2023 JCI (15 pp, end-to-end), Ding 2022 eLife (5 pp of 25 + abstract confirmed), Fang 2023 Calcified Tissue International (6 pp of full review, concept-level). 8 corrections: (1) Garcia-Gomez 2020 footnote — DOI: 10.1016/j.arteri.2020.01.002 → 10.1016/j.arteri.2019.03.008 (PubMed PMID 31221532 returns the 2019.03.008 suffix; journal-online date is 2020 Jan-Feb). (2) Liu 2026 footnote — DOI: 10.1016/j.bone.2026.117400 → 10.1016/j.bone.2026.117890 (PubMed PMID 41962789 returns 117890 suffix). (3) Chen 2026 footnote — DOI added: 10.7150/ijms.130435 (PubMed PMID 42158825 confirmed). (4) Farr 2023 body — local vs systemic senolysis framing corrected: Wiki said “local senolysis partially prevented bone loss; systemic required for full benefit including spine + femur.” Corrected to reflect that local senolysis improved lumbar SPINE trabecular BV/TV (via bone formation increase) but had no effect on FEMUR parameters, bone resorption, or marrow adiposity. Systemic senolysis improved both spine and femur AND reduced resorption and marrow adipocyte numbers. Rankl reduction was systemic-only. (5) Farr 2023 footnote expanded: model details, n per group (15F + 10M per group), AP20187 dosing, transplantation arm n (11/group), and TAF osteocyte senescence finding at 60 days post-transplant added. (6) Ding 2022 body — “selective osteocyte depletion” corrected: paper uses PARTIAL ablation (DTAhet heterozygotes); complete ablation is perinatally lethal. Also added scRNA-seq-confirmed myeloid-biased hematopoiesis finding and ~20-40 week lifespan detail. Footnote updated with precise model name (DMP1cre × Rosa26-DTA heterozygotes), lifespan figure, and scRNA-seq detail. (7) Ding 2022 footnote — archive status updated: “pending download” → “downloaded and verified 2026-05-23”. (8) Fang 2023 and Farr 2020 footnote — archive status updated: Fang downloaded and verified; Farr 2020 PMC7541777 confirmed (download failed, PMC available as fallback). SUPERSESSION CANDIDATE flagged: Farr 2024 Nature Medicine (PMID 38956196; doi:10.1038/s41591-024-03096-2) — Phase 2 RCT of D+Q in 60 postmenopausal women; primary endpoint (CTx) negative overall (p=0.611); exploratory high-senescent-burden tertile shows P1NP ↑34%, CTx ↓11%, radius BMD ↑2.7% at 20 weeks. Added as new footnote [^farr2024-phase2-rct] and updated the senolytics row in the interventions table with full RCT result + supersession flag. Schema extension: literature-checked-through: on type:tissue pages documented as R47 in schema-history.md (optional, 18mo cadence, for tissue pages with active intervention landscapes). Downstream propagation needed: hallmarks/cellular-senescence.md (senolytics bone section if it cites the Farr 2023 preclinical claim without the 2024 RCT); any page citing Garcia-Gomez 2020 with the wrong DOI; interventions/pharmacological/senolytics.md (if it cites bone-specific preclinical evidence without the 2024 Phase 2 RCT result). literature-checked-through: 2026-05-23.
2026-05-23 verify — cell-types/osteocytes.md flipped verified:true. 7 corrections: (1) Doolittle 2023 body text + footnote — senolytic agents: “ABT-263 or dasatinib + quercetin” → AP20187 (INK-ATTAC p16ink4a-driven caspase 8 transgene) or dasatinib + quercetin (D+Q); ABT-263/navitoclax was not used in this paper. (2) Doolittle 2023 footnote — n: “not specified in abstract” → n=88 total (stated in paper body, p.2). (3) Tiede-Lewis 2019 footnote — pages: “101-107” → “101-113” (confirmed via Crossref API + PMC6638547 full-text). (4) Tiede-Lewis 2019 body — fracture-BMD independence: claim that “LCN degradation predicts fracture risk independently of bone mineral density” was stronger than the paper’s actual language; softened to “likely contributor rather than formally quantified independent predictor” + added needs-replication. (5) Zhao 2024 footnote — issue number: “9(24)” → “9(23)” (confirmed via PMID 39641271 PubMed record; DOI 10.1172/jci.insight.177557 confirmed correct). (6) Takeuchi 2025 footnote — DOI and pages: DOI 10.1007/s00774-024-01574-4 (invalid) → 10.1007/s00774-025-01580-4; pages “1-10” → “26-32” (confirmed via PMID 39825110). (7) Jiang 2024 footnote + body — Galnt3-KO FGF23 reduction: “~40%” → “~50-60%” (Figure 9A shows ~40-50 pg/ml in KO vs ~100 pg/ml WT at 8 wk). He 2024 footnote: delivery vehicle description expanded to “(AspSerSer)6-modified sEVs encapsulating DM1-Gal” (verified via PMID 38063283 abstract). Tiede-Lewis 2019 PDF download failed; verified via PMC6638547 full-text XML. Doolittle 2023 (full PDF, 20 pp) and Zhao 2024 (full PDF) read end-to-end. Takeuchi 2025 PDF downloaded under corrected DOI. CL:0000137 confirmed via OLS4. Literature recency spot-checked: 6 PubMed hits 2024-2026 for “osteocyte senescence aging bone senolytic”; most notable is Doolittle et al. 2026 JCI 136(7):e204645 (companion editorial to an expanded Doolittle senotype study — does not contradict existing page framing). Supersession: none — no meta-analysis or large RCT on osteocyte senescence or bone-targeted senolytics exists as of 2026-05-23. literature-checked-through: 2026-05-23.
2026-05-23 verify — molecules/proteins/15-pgdh.md flipped verified:true. Cross-checked against three verified study pages (Palla 2021, Bakooshli 2023, Singla 2025/2026); canonical IDs re-confirmed via UniProt REST + NCBI eutils; 8 unverified DOIs sanity-checked via Crossref/PubMed. 7 corrections: (1) isoform count 7→5 (UniProt “Named isoforms=5”); (2) canonical RefSeq NM_000860.6→NM_000860.5 (per UniProt MANE-Select ENST00000296522.11); (3) human tissue method for Palla 2021 data: “immunohistochemistry” → “microarray gene expression analysis (Raue 2012 vastus lateralis dataset)” — Palla 2021 used microarray, not IHC; (4) druggability-tier 1→2 — no active PGDHi clinical trial registered on ClinicalTrials.gov (API verified); Epirium Bio’s only ClinicalTrials.gov entry is NCT04386304 (epicatechin/Becker MD, unrelated); compound identity proprietary; tier 2 (high-quality probe) is accurate per CLAUDE.md tier definitions; (5) Chaudhary 2025 second-author corrected: “Desai AB”→“Cordova BA, Hong M, Klein BR” (confirmed via Crossref); (6) Zhang 2015 PMID added: 26068857 (confirmed via PubMed eutils); (7) PMID 38904907 (recency table) identified as Ahmad 2024 biocomputational screening paper (Molecular Diversity 2024). No Schwann cell mechanism errors found on this protein page — seeder correctly flagged “NOT Schwann cells” in the Bakooshli footnote. No “18–22 mo” mouse age errors found — protein page correctly uses “>24 mo / 2–4 mo” per verified Palla 2021 corrections. Downstream propagation: molecules/compounds/sw033291.md (if seeded) should use correct isoform count; frameworks/cancer-aging-tradeoffs.md can note no active clinical trials; phenotypes/sarcopenia.md should confirm microarray (not IHC) for the Palla 2021 human data citation if it independently cites it.
2026-05-23 verify — studies/bakooshli-2023-15pgdh-nmj-regeneration.md flipped verified:true. Full 18-page PDF + methods read. 10 corrections: (1) Two background-citation DOIs wrong — Deschenes 2010 DOI 10.1152/japplphysiol.00773.2010 returns HTTP 404 (invalid); corrected to 10.1016/j.exger.2010.03.007 (PMID 20226849, Experimental Gerontology). Gonzalez-Freire 2014 DOI 10.1093/gerona/glu072 resolves to an unrelated paper on co-trimoxazole/sulfonylurea hypoglycemia (Tan et al.); corrected to 10.3389/fnagi.2014.00208 (PMID 25157231, Frontiers in Aging Neuroscience). (2) 15-PGDH upregulation quantified: seeder said “marked increase”; PDF states ~20-fold Hpgd mRNA increase at day 90, fourfold protein increase by Western blot in denervated vs contralateral GA. (3) SW033291 dose confirmed: 5 mg/kg i.p. once daily (seeder had left this TBD). (4) Aged mouse age confirmed: 24–26 months old (NIA colony); seeder left this as “approximate, to be confirmed.” (5) Sex confirmed: male C57BL/6 mice exclusively (seeder did not specify sex). (6) Young mouse age confirmed: 2–4 months (Jackson Laboratory). (7) Human biopsy n confirmed: n=10 patients; 9/10 had 15-PGDH aggregates composing the target bullseye. (8) Force recovery quantified: 37.2 ± 4.9% increase in tetanic force at 14 dpi (seeder said “approximate, to be confirmed”); motor axon counts: 797 ± 69.3 (PGDHi) vs 415.2 ± 52.5 (vehicle); NMJ innervation: 97.4 ± 2.1% vs 84.5 ± 2.2%. (9) Motor neuron viability quantified: aged mice 11.6 ± 2.5% Cl-Casp3+ ChAT+ → 4.2 ± 0.6% with PGDHi. (10) Mechanism corrected: seeder emphasized Schwann cells as “likely” primary neural target; paper’s mechanistic experiments focus on motor neuron CREB phosphorylation via EP4→cAMP→CREB axis (Schwann cells not tested); LC-MS/MS measures PGEM (breakdown product) not PGE2 directly. Target fiber disease spectrum broadened: paper covers axonal neuropathies, myositis, lumbar radiculopathy, ALS, SMA — not specifically post-polio syndrome. Downstream propagation needed: molecules/proteins/15-pgdh.md, molecules/compounds/sw033291.md, phenotypes/sarcopenia.md, tissues/skeletal-muscle.md — check for any reliance on wrong DOIs, unquantified 15-PGDH fold-change, Schwann-cell mechanism framing, or aged-mouse age left as TBD.
2026-05-23 verify — studies/singla-2025-15pgdh-cartilage-regeneration.md flipped verified:true. Verified against PMC13127300 full-text HTML (journal PDF closed-access; PMC release 2026-04-29). 9 corrections: (1) year: 2025 → 2026 (Crossref published-print: 2026-03-05; Science 391(6789) is a March 2026 issue); (2) n=11 human explant claim corrected throughout — n=11 is the FACS characterization cohort; PGDHi treatment experiments used n=5 per group (and n=3 for CD200+ change readout); (3) aged-mouse OARSI cohort n added: n=9 per group; (4) PTOA OARSI cohort n added: n=7 per group; (5) baseline 15-PGDH comparison n added: n=3 young, n=5 aged; (6) CatWalk p-values corrected: “p≤0.032” → p=0.010 (paw contact, vehicle vs PGDHi) and p=0.028 (mean intensity, vehicle vs PGDHi); the 0.032 value belongs to Ctrl vs Veh comparison, not PGDHi vs Veh; (7) Cluster 1 hypertrophic-like markers corrected: Col10a1 and Mmp13 were not defining markers — confirmed markers are CD200, Ihh, Mef2c, Wnt5b, Spp1, Pth1r, Alpl, Runx2; (8) Cluster 3 articular chondrocyte markers corrected: Col2a1/Acan/Sox9 were not the defining marker set — confirmed markers are Bmp5, Fgf2, Stat5a, Hdac9; (9) COI updated with patent 62/983,421 (multiplex microscopy images; Epirium licensing not stated) and note that consulting-fees recipient appears truncated in PMC text. PTOA timeline clarified: histology cohort evaluated at 6 weeks post-injury; pain cohort evaluated at 8 weeks (separate cohorts). Downstream propagation needed: molecules/proteins/15-pgdh.md, molecules/compounds/sw033291.md, phenotypes/osteoarthritis.md (if any of these cite n=11 as the treatment cohort), frameworks/cancer-aging-tradeoffs.md.
2026-05-23 verify — studies/palla-2021-15pgdh-muscle-rejuvenation.md flipped verified:true. Verified against PMC7938328 HTML (a local paper archive PDF download failed — 0 candidate URLs after filtering despite oa_status=green; PMC author-manuscript HTML used as verification source). 3 corrections: (1) mouse ages: “young (2–3 months) vs. aged (18–22 months)” → “young (2–4 months) vs. aged (>24 months)” per consistent PMC figure legend and methods statements — this is the most significant error; (2) SW033291 dose: clarified that mg/kg dose is NOT stated anywhere in this paper (not merely unclear from HTML extraction) — paper cites Zhang 2015 for original characterization; (3) human microarray source identified: Raue U et al. J Appl Physiol 2012; cohort n not stated in Palla 2021. Additional confirmations: p62/colchicine autophagy direction confirmed (increased p62 = enhanced flux, correct); EP4-specificity via ONO-AE3-208 antagonist confirmed; all per-experiment n values confirmed against figure legends. Unverifiable: sex of mice not stated in paper anywhere; specific % fold-change effect sizes not in paper text (figures only). Downstream pages needing age-range correction: molecules/proteins/15-pgdh.md, molecules/compounds/sw033291.md, phenotypes/sarcopenia.md, tissues/skeletal-muscle.md (if they state the 18–22 mo age range).

🧬 Aging Wiki

Explorer

2026-05-23

log/2026-05-23 — ad-hoc daily entries

[2026-05-23] ingest | UV filter chemistry — MCE (Mexoryl 400) + PBT (TriAsorB) + Matta 2019/2020 systemic absorption

[2026-05-23] ingest | 15-PGDH gerozyme axis — protein + inhibitor + 3 study pages

[2026-05-23] verify | 15-PGDH gerozyme axis — 5-page verification pass

[2026-05-23] ingest | Klotho-FGF23-phosphate axis seeding cluster

[2026-05-23] verify | phenotypes/osteoporosis.md

[2026-05-23] verify | pathways/wnt-beta-catenin.md — merged-section verification

[2026-05-23] verify | molecules/proteins/osteopontin.md

[2026-05-23] verify | molecules/proteins/runx2.md

[2026-05-23] verify | molecules/proteins/fgf23.md

[2026-05-23] verify | cell-types/endothelial-cells.md → verified:true (partial scope)

Graph View

Table of Contents

Backlinks