log/R37.md — Round 37 entries

Sub-file of log — see parent for index.

[2026-05-13] verify | pathways/type-i-interferon-signaling.md

Pages verified: 1 (pathways/type-i-interferon-signaling.md)

PDFs read: Jiménez-Loygorri 2024 (local), Jin 2022 (local), Bastard 2020 (downloaded from PMC), Stark & Darnell 2012 (downloaded), Kandhaya-Pillai 2022 (downloaded), Yu 2026 (downloaded gold OA). Yang 2026 (not_oa; kept preliminary). De Cecco 2019 (not_oa; tagged no-fulltext-access).

Corrections (7):

Jiménez-Loygorri 2024 — mouse age wrong: “18–24 months” → “24–26 months old; 6–8 months young”
Jiménez-Loygorri 2024 — rescue mechanism: body said “PINK1/Parkin pathway activation” in vivo; corrected to urolithin A 2.3 mg/kg/day i.p. pharmacological rescue; PINK1/Parkin confirmation was ARPE-19 in-vitro only
Jin 2022 — ruxolitinib fabricated: Jin 2022 uses IFNAR1/2 shRNA knockdown only; ruxolitinib not present in that paper; footnote and body corrected
Kandhaya-Pillai 2022 — cell model wrong: footnote listed IMR90/WI-38 fibroblasts; paper uses HUVECs; body corrected from “senescent fibroblasts” to endothelial cells; TMPRSS2 removed (paper reports ACE2/DPP4 only)
Bastard 2020 — age-stratified prevalence wrong attribution: “~1% <65y, ~4% 70–80y, ~7% >80y” are from Bastard 2021 Science Immunology, not Bastard 2020; 2020 paper reports 8.5% vs. 13.0% within COVID-19 cohort only; body and footnote corrected; companion paper flagged
Yu 2026 — human cohort omitted: wiki only described mouse EV transfer; primary cohort was 185 humans (20–95y); updated body and footnote
Canonical DB IDs confirmed (KEGG hsa04623, hsa04622, Reactome R-HSA-909733 all correct)

Unverifiable: De Cecco 2019 (not_oa) no-fulltext-access; Yang 2026 (not_oa) preliminary.

Supersession candidates: None identified. Bastard 2021 Science Immunology (doi:10.1126/sciimmunol.abl4247) is an important companion paper that should be added as a separate footnote citation for population-level age-stratified anti-IFN autoantibody prevalence.

Downstream propagation needed: Pages citing [^bastard2020] for “~1%/~4%/~7%” figures should be updated to Bastard 2021. Pages citing [^jin2022] for ruxolitinib as microglial senescence rescue should remove that claim.

[2026-05-13] verify | studies/bastard-2021-anti-ifn-autoantibody-age-prevalence.md

Pages verified: 1 (studies/bastard-2021-anti-ifn-autoantibody-age-prevalence.md)

PDF read: Local PMC author manuscript — — pages 1–17 (full paper). DOI confirmed via and PubMed esummary. PMC ID, PMID, journal, year all verified.

Corrections (8):

n-subjects: 38,463 → 40,016 — seeder used an undocumented aggregate; Methods section explicitly states 34,159 + 3,595 + 623 + 1,639 = 40,016. YAML comment updated with breakdown.
Deceased mean age: “70 years” → “71 years” — results section (p.9) gives “mean age: 71 years.”
“18%” deceased attribution: clarified this is the 100 pg/mL assay result (18.5%, rounded to 18% in abstract); the 10 ng/mL assay found 13.3% of deceased. Both figures now present in body.
Critical COVID-19 n for 10 ng/mL assay: 3,595 enrolled but only 3,136 tested with the 10 ng/mL assay (9.8% = 307 of 3,136); the 13.6% figure (489 of 3,595) is for the 100 pg/mL assay. Distinction now explicit.
Severe cohort n: enrolled n=623; 522 is the n available for the 100 pg/mL assay. Both now documented.
Overall prevalence in uninfected cohort (0.53%) missing — 181 of 34,159 by 10 ng/mL assay — added to TL;DR, results table, and key-findings frontmatter.
Finer age bands from results text added: <70y 0.17%, 70–75y 0.9%, 75–80y 1.6%, 80–85y >4%, >85y ~2.6%. The >4% peak at 80–85 followed by decrease after 85 was absent from seeder extract; now documented.
Assay methodology corrected: seeder described “antiviral protection assay or ISRE-luciferase reporter” — PDF confirms the functional assay is specifically ISRE-firefly-luciferase normalized against Renilla luciferase (HEK293T cells), with Gyros immunoassay as initial screen. No antiviral-protection (cell-infection) assay was used for the main cohort.

[2026-05-13] verify | molecules/compounds/amlexanox.md

Page verified: molecules/compounds/amlexanox.md (seeded 2026-05-13)

Sources checked: Reilly 2013 (10.1038/nm.3082, local PDF, full 10 pages read); Oral 2017 (10.1016/j.cmet.2017.06.006, local PDF, full 14 pages read); Zhao 2022 (10.1172/jci.insight.155552, local PDF, full 8 pages read); You 2025 (10.1038/s44355-024-00015-7, local PDF, full 8 pages read); Cho 2024 (10.1038/s12276-024-01304-0, local PDF, full 8 pages read). Gairola 2025 and Kapoor 2026 are not_oa locally — body claims for these limited to abstract-level information. PubChem CID 2161 cross-checked via REST API.

Corrections (5):

Oral 2017 dose: “200 mg TID” → “50 mg TID” — The Phase 2 placebo-controlled RCT used 50 mg amlexanox three times daily (Figure 1A, Methods). The open-label pilot also escalated from 25 mg TID → 50 mg TID. The seeder wrote “200 mg” which is incorrect. Affects body Phase 2 section, PK section, and footnote — all corrected.
Oral 2017 trial AE profile corrected — seeder wrote “no major adverse effects reported”; the paper reports 7 cases of rash across both groups (1 drug-treated patient had biopsy-confirmed perivascular inflammation that was self-limiting; 2 prior open-label cases also had skin biopsy) and 1 serious AE (non-STEMI MI in placebo arm). Safety section updated to accurately reflect the AE profile.
Footnote oral2017 n refined — corrected to show 21/21 randomized (stratified by sex; Table 1 confirms equal allocation) vs. 20/18 evaluable. Footnote updated.
Reilly 2013 sequence similarity precision — seeder wrote “65% overall sequence identity”; paper states “65% sequence similarity” (the 72% in the ATP-binding region is described as “72% identical”). Corrected: “similarity” for overall; “identity” for ATP-binding region.
Zhao 2022 mechanism augmented — seeder only cited macrophage suppression as the mechanistic basis for atherosclerosis benefit; Zhao 2022 (Results, Fig. 2) demonstrates a primary mechanism of upregulated hepatic bile acid synthesis (Cyp7a1) → increased fecal cholesterol excretion. This bile acid mechanism is distinct from and complementary to macrophage suppression. Zhao 2022 section updated to include it. Also corrected lab attribution: “Park HY group” → “Yong Sook Kim / Youngkeun Ahn lab, Chonnam National University” (Cho 2024 corresponding authors per paper masthead).

Unverifiable claims: Gairola 2025 (Eur J Pharmacol) and Kapoor 2026 (Inflammopharmacology) are not_oa locally. Body-level claims for these entries are limited to what is supported by PMID/abstract-level information; no quantitative claims from these papers are asserted that require full-text verification.

Supersession candidates (R25): None. PubMed search for amlexanox + TBK1/IKKε + obesity/diabetes filtered for RCTs or meta-analyses post-2020 returned 0 results. No confirmatory RCT has been registered. literature-checked-through: 2026-05-13 confirmed in frontmatter.

Framework check: tbk1-ikke-inhibitor class value added to frameworks/intervention-classes.md under R35 (lines 381–397 + table row at line 1264) — correctly defined; amlexanox listed as founding compound; class scope and canonical value documented. No orphan links.

Downstream propagation needed: None urgent. The Oral 2017 dose correction (200 mg → 50 mg) does not affect any other wiki page that cites this paper (only ikbke.md cites it; that page was seeded and verified today with the correct 50 mg dose already present based on the IKKε verifier pass). Main agent should confirm ikbke.md dose value is already correct.

Additional additions (not corrections, expansions from PDF):

Asymptomatic/mild COVID-19 control cohort (1,639 individuals) was absent from seeder’s cohort list; added.
Pre-existing autoantibody section: confirmed n=4 critical patients with pre-COVID samples available (+ 2 from companion 2020 paper = 6 total confirmed).
Anti-IFN-β section: added exact n=1,773 critical patients tested; 9,583 uninfected tested; overall prevalence 0.26%; Firth’s regression P=0.68 for age non-association.
DOI correction banner removed from body (DOI was already corrected in frontmatter; pathway page already updated before this verification pass).

Unverifiable claims: None — all quantitative claims were traceable to abstract, results text, or methods section of the local PDF.

Supersession candidates (R25): PubMed search (2022–2026) for anti-interferon autoantibodies + general population prevalence returned 3 papers; none supersede Bastard 2021’s prevalence figures. Fernbach 2024 (JEM, doi:10.1084/jem.20240365; confirmed via Crossref) is independent supportive evidence (longitudinal HIV cohort, n=1,876), not supersession. literature-checked-through: 2026-05-13 set.

Downstream propagation needed:

pathways/type-i-interferon-signaling.md — already has verified: true; its [^bastard2021note] footnote carries the abstract-level age-band figures (0.18%, 1.1%, 3.4%) which are all confirmed correct. The new finer-band figures and overall 0.53% prevalence are additions on the study page, not contradictions of the pathway page. Main agent should decide whether to add the overall 0.53% and finer bands to the pathway page footnote for completeness — not a factual error, low urgency.
No other pages appear to cite this study page directly yet.

[2026-05-13] verify | molecules/proteins/ikbke.md

Pages verified: 1

molecules/proteins/ikbke.md — 7 substantive corrections; verified: true

Sources checked:

Reilly 2013 (Nat Med) — local PDF, read end-to-end
Oral 2017 (Cell Metab) — local PDF, read end-to-end (downloaded during session; was pending)
Peters 2000 (Mol Cell) — local PDF, read end-to-end (downloaded during session; was pending)
Zhao 2022 (JCI Insight) — local PDF, read pages 1–3 (downloaded during session; was pending)
Cho 2024 (Exp Mol Med) — local PDF, read pages 1–3 (downloaded during session; was pending)
You 2025 (npj Gut Liver) — local PDF, read pages 1–3 (downloaded during session; was pending)
UniProt Q14164, NCBI Gene 9641, HGNC 14552, Ensembl REST — verified against live APIs

Sources not accessible (closed-access):

Chiang 2009 (Cell) — not_oa + failed download; Ikbke KO metabolic phenotype claims accepted per corroborating Reilly 2013 introduction framing; tagged in verified-scope
Fitzgerald 2003 (Nat Immunol) — not_oa; IRF3 Ser396 phosphorylation claim is canonical and accepted
Shimada 1999 (Int Immunol) — not_oa; discovery context accepted

Corrections made:

TBK1/IKKε sequence identity: “~64% kinase-domain identity” → “~65% overall / 72% in ATP-binding region” (per Reilly 2013 exact text)
Reilly 2017 primary endpoint framing (CRITICAL): the seeder framed this as a null primary endpoint + post-hoc subset signal. WRONG. The primary endpoint (HbA1c) was statistically significant in the full drug-treated group vs. placebo. The responder analysis (n=7 of 21 achieving ≥0.5% HbA1c reduction) was a pre-defined threshold analysis. Corrected throughout page body, extrapolation table, and footnote.
First author of 2017 paper: “Reilly SM et al.” → “Oral EA, Reilly SM et al.” (Reilly is second author; Elif Oral is first/lead)
JCI Insight 2022 mouse model: “ApoE−/− mice” → “Ldlr−/− mice” (per Zhao 2022 primary text; ApoE−/− was never used in that paper)
Amlexanox IC50: “low-micromolar range” → “approximately 1–2 μM” (per Reilly 2013 dose-response section)
Ensembl gap tag resolved: ENSG00000263528 confirmed by both Ensembl REST and HGNC REST APIs; #gap/needs-canonical-id removed
MASH model specificity: “MASH diet model” → “GAN diet-fed Ldlr−/− mice” with mechanism detail (bile acid pathway + Akkermansia muciniphila enrichment)

Supersession check (R25): No meta-analyses or large RCTs (n>100) identified for IKKε/amlexanox metabolic/aging indications since 2013. One 2024 neuroinflammation preclinical paper (PMID 38302598, Sci Rep) is not a clinical supersession. Aphthous-ulcer meta-analysis (2025, BMC Oral Health) covers original indication only — not relevant to metabolic/aging framing. literature-checked-through: 2026-05-13 retained.

Downstream propagation needed: None identified — this is a newly seeded page with no existing downstream pages citing it with inherited errors.

[2026-05-13] ingest | chen-2025 NMR cGAS follow-on batch (5 pages + propagation)

Spawned-from: Deferred follow-on candidates from the 2026-05-12 Chen 2025 NMR cGAS HR-repair batch. Five pages seeded in parallel with wiki-verifier passes on 4 (Liu 2018 is closed-access, permanent-blocker; no verifier).

Added (5 pages):

studies/liu-2018-nuclear-cgas-hr-suppression.md — verified: false (permanent-blocker not_oa); seeded from abstract + PubMed metadata only; all quantitative claims #gap/no-fulltext-access; Crossref author list (35 authors) used
studies/zhen-2023-trim41-cgas-l1.md — verified: true; local PDF downloaded; full primary-source verification; sister Mao-lab paper to Chen 2025
molecules/proteins/fancd2.md — verified: true; 6 primary PDFs read end-to-end (Garcia-Higuera 2001, Longerich 2014, Alcón 2020, Garaycoechea 2018, Rossi 2007, Rall-Scharpf 2025)
molecules/proteins/palb2.md — verified: true; 3 PDFs end-to-end (Reid 2007, Xia 2006, Antoniou 2014)
molecules/proteins/irf3.md — verified: true; 6 PDFs end-to-end (West 2015, Glück 2017, Yang 2017, Essers 2009, Lin 1999, Gulen 2023)

Verifier corrections of note (10 substantive across 4 verified pages):

IRF3 (highest-impact): Ser-385/Ser-386 are regulatory, NOT activation phosphoacceptor sites (per Lin 1999); the primary inducible cluster is aa 396–405 (Ser-396/Ser-398/Thr-404 the TBK1/IKKε substrates per UniProt). Essers 2009 IFN-α-driven HSC exhaustion is IRF7-driven via JAK-STAT amplification; IRF3 contribution is upstream/indirect (initiates IFN-β → IRF7 amplification loop). Gulen 2023 used both Sting−/− mice AND H-151 pharmacological inhibition (not Sting−/− alone); IRF3 KO not directly tested in that paper.
PALB2: Reid 2007 was n=7 families (not 8 as seeded); Antoniou 2014 = 154 families / 362 carriers (seeder conflated these numbers); Xia 2006 does not show BRCA1-independent HR (seeded claim removed); mr-causal-evidence promoted not-tested → partial (GWAS hits exist for PALB2 region, e.g., rs45569335, p<3e-10; no MR study leveraging instruments published).
FANCD2: Rossi 2007 Nature (doi:10.1038/nature05862) does NOT contain FA-pathway mutants — only mTR-/-, XPD-TTD, Ku80-/- mice; relevance to FANCD2 is by pathway analogy. Garaycoechea 2012 was Fanca/Aldh2 not Fancd2/Aldh2 (the Fancd2 paper is Garaycoechea 2018). Fork-protection citation Schlacher 2012 (not Garcia-Higuera 2001 as seeded). Longerich 2014: FANCD2 mono-ub is largely independent of FANCI ub (opposite of seeded “interdependent ub initiation” framing).
Zhen 2023: study-design re-coded in-vitro → in-vitro+in-vivo; senescence cell-line attribution clarified (IMR90-hTERT + HCA2-hTERT used for X-ray validation only; HeLa was etoposide model).

Propagation pass updates (Task 1 — Liu 2018 bare-DOI → study-page wikilink):

hallmarks/genomic-instability.md — [^liu2018] upgraded to wikilink form
molecules/proteins/vcp.md — [^liu2018] upgraded
studies/chen-2025-nmr-cgas-hr-repair.md — [^liu2018] upgraded
hypotheses/antagonistic-pleiotropy.md — [^liu2018] upgraded
pathways/cgas-sting.md — [^liu2018] upgraded
pathways/dna-damage-response.md — [^liu2018] upgraded
pathways/homologous-recombination.md — [^liu2018] upgraded
molecules/proteins/cgas.md — already in wikilink form (no edit needed)

Propagation pass updates (Task 1 — Zhen 2023 bare-DOI → study-page wikilink):

interventions/pharmacological/cgas-modulators.md — [^zhen2023] upgraded to wikilink form with sample sizes and full model description
molecules/proteins/trim41.md — [^zhen2023] upgraded similarly

Propagation pass updates (Task 2 — wikilink resolution check for new protein pages):

molecules/proteins/fanci.md — [[fancd2]] and [[palb2]] already in wikilink form (lines 35, 60, 65, 70, 136, 141); auto-resolve now that target pages exist; no edits needed
molecules/proteins/sting.md, molecules/proteins/tbk1.md, pathways/cgas-sting.md, processes/mtdna.md, hallmarks/mitochondrial-dysfunction.md — [[irf3]] already in wikilink form; auto-resolves now

Propagation pass updates (Task 3a — IRF3 phosphorylation cluster correction):

pathways/cgas-sting.md line 76: cites IRF3 Ser396 (correct — Ser396 is in the inducible aa 396–405 cluster); no correction needed
molecules/proteins/sting.md line 68: cites IRF3 S396 (correct); no correction needed
molecules/proteins/tbk1.md: no specific phosphosites cited; no correction needed
No instances of the erroneous “Ser-385/S385” as IRF3 activation site found on any page beyond the new irf3.md (already corrected in verifier pass)

Propagation pass updates (Task 3b — Essers 2009 IRF3-vs-IRF7 attribution):

No Essers 2009 citations found on hallmarks/stem-cell-exhaustion.md, cell-types/hematopoietic-stem-cells.md, pathways/jak-stat-pathway.md, hallmarks/chronic-inflammation.md; only the irf3.md page itself cites Essers 2009 and is already corrected. Propagation a no-op.

Propagation pass updates (Task 3c — Rossi 2007 Nature FA-pathway misattribution check):

hallmarks/stem-cell-exhaustion.md line 35: cites Rossi 2007 for HSC γH2AX foci accumulation (correct, no FA claim); no correction needed
cell-types/hematopoietic-stem-cells.md lines 113, 115, 123: cites Rossi 2007 for quiescence/DDR attenuation + repair-deficient mice (HR/NHEJ/NER) accelerating decline; correctly framed as repair-pathway-deficient strains, not FA mutants; no correction needed
interventions/stem-cell-therapy/hematopoietic-stem-cell-transplantation.md line 70: cites Rossi 2007 for DNA-damage as causal of HSC decline; no FA-pathway claim; no correction needed
tissues/bone-marrow.md line 52, 123: cites Rossi 2007 generally; no FA-pathway claim; no correction needed

Propagation pass updates (Task 4 — fanci.md per FANCD2 verifier findings):

molecules/proteins/fanci.md ID2 complex section (lines 60–62) annotated with two verifier-derived nuances: (a) Lemonidis 2023 “interdependent” framing reconciled with Longerich 2014 (FANCD2 mono-ub is largely independent of FANCI ub in original kinetic experiments) — mature consensus: FANCI ub is dispensable for initial FANCD2-ub event but required for sustaining it; (b) Alcón 2020 cryo-EM structure was reconstituted from Gallus gallus (chicken) FANCD2–FANCI orthologs (chicken K525 ≡ human K523); in that reconstituted system FANCI was NOT monoubiquitinated, so the closed-clamp conformation is established for FANCD2-ub-only ID2, not strictly doubly-ubiquitinated.

Gaps surfaced:

Liu 2018 PDF still inaccessible (closed-access not_oa); all quantitative claims tagged #gap/no-fulltext-access; permanent-blocker status documented in study page banner
FANCD2: FA-D2 disease prevalence (3-5%) not verified against registry source; Garaycoechea 2012 not_oa; Schlacher 2012 fork-protection citation not verified
IRF3: Sharma 2003, Fitzgerald 2003, Servant 2001 (all not_oa or failed download) — TBK1-as-IRF3-kinase claim retained on independent West 2015 support; Hopfner 2020 disk path mismatch (file not at reported location)

Schema gaps escalated:

mouse-ortholog: field used on fancd2.md but not in CLAUDE.md type:protein schema — consider formalizing
pathways/type-i-interferon-signaling.md page does not exist — irf3.md substituted [[jak-stat-pathway]] for now
molecules/proteins/ikbke.md (IKKε) does not exist — referenced in IRF3 page but unresolved

Seeder-fabricates-outcomes pattern recurrences (R29):

PALB2: Reid 2007 n=8→7 families; Antoniou 2014 cohort conflation (154 families = unique probands; 362 = carriers; seeder collapsed); Xia 2006 mis-attribution (BRCA1-independent HR claim not in paper)
FANCD2: Rossi 2007 Nature doesn’t even contain FA mutants (only mTR/XPD/Ku80); Garaycoechea 2012 wrong gene (Fanca vs Fancd2); Schlacher 2012 misattributed as Garcia-Higuera 2001
IRF3: Ser-385/386 mis-attributed as activation sites (regulatory per Lin 1999)
These continue to validate the R29 feedback memory — verifier briefs must continue to emphasize “verify outcomes-not-just-targets,” including paper-content verification (does the paper even contain the mouse model claimed?)

[2026-05-13] verify | molecules/proteins/irf3.md

Pages verified: 1

molecules/proteins/irf3.md — IRF3 protein page (auto-extracted 2026-05-13)

Sources checked:

doi:10.1038/nature14156 (West 2015, Nature) — PDF read end-to-end; confirmed cGAS-STING-TBK1-IRF3 axis; ISG mRNA reduction 70–90% in STING-KO confirmed; IRF3 siRNA knockdown blocks ISGs confirmed
doi:10.1038/ncb3586 (Glück 2017, Nat Cell Biol) — PDF read end-to-end; cell models confirmed (cGAS-KO MEFs Mb21d1−/−, C57BL/6; WI-38 cells — NOT IMR90); senescence triggers confirmed (40% O2, 12 Gy IR, HRasV12, palbociclib)
doi:10.1073/pnas.1705499114 (Yang 2017, PNAS) — PDF read end-to-end; confirmed NO aged-tissue data; etoposide/irradiation/serial passage models; cGas−/− MEFs C57BL/6; BJ human fibroblasts
doi:10.1038/nature07815 (Essers 2009, Nature) — PDF read end-to-end; confirmed IFN-α specifically (mIFNα4) not generic IFN-I; IFNAR-STAT1-Sca-1 axis; chronic IFNα → HSC pool depletion in chimeras
doi:10.1128/MCB.19.4.2465 (Lin 1999, Mol Cell Biol) — PDF read end-to-end; confirmed 55-kDa protein; transactivation domain aa 134–394; two autoinhibitory domains (aa 380–427 and aa 98–240); S385/S386 are REGULATORY not direct activation sites; inducible phosphoacceptor cluster is aa 396–405
doi:10.1038/s41586-023-06373-1 (Gulen 2023, Nature) — PDF read end-to-end; confirmed H-151 + Sting−/− dual strategy; aged mice 19–20 months C57BL/6J; pTBK1 Ser172 elevated in brain; cGAMP elevated by ELISA in aged brain; microglia-specific cGasR241E model; IRF3 KO NOT directly tested
Identifiers: UniProt Q14653, HGNC:6118, NCBI Gene 3661, Ensembl ENSG00000126456 — all confirmed via UniProt REST API
Sharma 2003 (not_oa), Fitzgerald 2003 (not_oa) — TBK1/IKKε kinase identity claim retained; independently supported by West 2015 TBK1-IRF3 axis
Servant 2001 (failed download — JBC URL unavailable via PMC) — retained as low-weight citation; claim unverified from PDF
Hopfner 2020 (disk path mismatch — path reported as (stale local path) but file not found there) — used as review citation; broad mechanistic claims verified against primary sources
Decout 2021 (downloaded, not read cover-to-cover) — used as supporting review citation

Corrections made (11 substantive):

Ser-385 description corrected: was “Critical activation site; TBK1/IKKε substrate; required for dimerization” → corrected to regulatory site per Lin 1999 (S385/S386→Asp does not activate transcription; these modulate the aa 396–405 cluster); Ser-396 identified as primary TBK1/IKKε substrate per UniProt
Activation cascade step 3 rewritten: previously cited “Ser-385, Ser-386, Ser-396, Ser-398” as TBK1 substrates; corrected to “primary inducible phosphoacceptor cluster aa 396–405; Ser-386 conformational switch; Ser-385/Ser-386 regulatory” per Lin 1999
Lin 1999 footnote overhauled: now accurately describes 55-kDa protein, transactivation domain aa 134–394, dual autoinhibitory domains, and S385/S386 regulatory (not activation) role
Gulen 2023 body paragraph corrected: was “using Sting−/− mice” → corrected to both Sting−/− mice AND H-151 pharmacological inhibition; added pTBK1/cGAMP quantitative details; added note that IRF3 KO was not directly tested
Gulen 2023 footnote expanded: added aged mice details (19–20 months, C57BL/6J), H-151 dose (750 nmol), snRNA-seq cells (9,505), cognitive endpoints (p=3×10−3), pTBK1 and cGAMP data
Essers 2009 body paragraph corrected: “IFN-α activates dormant HSCs; IRF3 initiates the upstream IFN signal” → corrected to IFN-α specifically via IFNAR-STAT1-Sca-1; IRF3’s contribution is indirect (initiates IFN-β → IRF7 amplification → IFN-α); gap tag added for human evidence
Essers 2009 footnote expanded: mIFNα4 dose (10,000 units i.p.), chimera design, Sca-1 requirement, STAT1 requirement specified
Glück 2017 footnote corrected: cell models updated from implicit to explicit (cGAS-KO MEFs from Mb21d1−/−, C57BL/6; WI-38); senescence triggers listed (40% O2, 12 Gy, HRasV12, palbociclib)
Yang 2017 footnote corrected: cell lines made explicit (cGas−/− MEFs, BJ human fibroblasts, B16F10); senescence triggers listed (3 μM/10 μM etoposide MEFs, 10–20 μM BJ, 3 Gy IR); SASP genes listed; no-aged-tissue statement made bold
West 2015 footnote expanded: Tfam+/− model details (~50% mtDNA depletion), 70–90% ISG reduction in STING-KO confirmed, TBK1/IRF3 siRNA knockdown result
Other phosphorylation sites list corrected: Thr-3 (not Ser-3), Ser-123 added (was missing from wiki list); UniProt-confirmed

Unverifiable sources:

Sharma 2003 (not_oa) — TBK1/IKKε kinase identity claims; independently corroborated by West 2015
Fitzgerald 2003 (not_oa) — IKKε/TBK1 essential for IRF3 phosphorylation; independently corroborated
Servant 2001 (download failed) — signaling pathway distinction claim; low-weight citation
Hopfner 2020 (disk path mismatch; could not read PDF) — review-level claims only; primary sources verified independently

Downstream pages that may need checking:

pathways/cgas-sting.md — if it mentions IRF3 phosphorylation sites, update to reflect aa 396–405 as primary cluster
Any page citing Essers 2009 for “IRF3-driven HSC exhaustion” — the IRF3 link is indirect; IFN-α/IRF7 is the proximate driver

[2026-05-13] verify | molecules/proteins/fancd2.md

Pages verified: 1

molecules/proteins/fancd2.md — FANCD2 protein page

Sources checked (all 6 PDFs read end-to-end):

doi:10.1016/s1097-2765(01)00173-3 (Garcia-Higuera 2001, Mol Cell) — downloaded + read
doi:10.1093/nar/gku198 (Longerich 2014, NAR) — downloaded + read
doi:10.1038/s41594-020-0380-1 (Alcon 2020, NSMB) — downloaded + read
doi:10.1038/nature25154 (Garaycoechea 2018, Nature) — downloaded via PMC6047743 + read
doi:10.1038/nature05862 (Rossi 2007, Nature) — local archive + read
doi:10.1093/nar/gkaf498 (Rall-Scharpf 2025, NAR) — local archive + read
doi:10.1038/nature11368 (Garaycoechea 2012, Nature) — not_oa; not read; gap-tagged
doi:10.1016/j.cell.2012.05.003 (Schlacher 2012, Cell) — not yet archived; gap-tagged
Identifiers: NCBI Gene 2177, Ensembl ENSG00000144554, UniProt Q9BXW9 — all confirmed via APIs

Corrections made (10 substantive):

Rossi 2007 misattribution fixed — paper uses mTR-/-, XPD-TTD (NER), Ku80-/- (NHEJ) mice, NOT FA pathway mutants; body text rewritten to correct strain list; FANCD2 relevance reframed as pathway analogy with explicit note
ATM phosphorylation sites corrected — Ser-1401 added (UniProt lists Ser-222, Ser-1401, AND Ser-1404 as ATM-dependent; wiki had only Ser-222 + Ser-1404). Ser-594 added to broader phosphorylation site list (was missing from UniProt PTM enumeration)
Garcia-Higuera 2001 citation removed from ATM phosphorylation claims — that paper (2001) predates the kinase site-mapping literature and does not report Ser-222/1401/1404; corrected in body and footnote
Longerich 2014 interdependence claim corrected — wiki said “FANCI ubiquitination is needed to maintain FANCD2 ubiquitination against active deubiquitination by USP1-UAF1”; Longerich 2014 actually shows FANCD2 monoubiquitination is largely INDEPENDENT of FANCI ubiquitination in vitro; rewritten
Alcon 2020 organism caveat added — cryo-EM data from Gallus gallus orthologs (K563/K525), not human; ubD2–I channel is ~15–20 bp; FANCI K525 NOT monoubiquitinated in in-vitro reconstitution; body text and footnote updated; new limitation #8 added
Garaycoechea 2012 genotype corrected — wiki claimed “Fancd2-/-; Aldh2-/-” for the 2012 paper; the 2012 paper (not_oa) used Fanca, not Fancd2 (the Fancd2 model is the 2018 paper); corrected with not_oa gap tag
Garaycoechea 2018 p53 finding added — p53 deletion completely rescues HSC survival but not genomic instability; ~60% of double-KO metaphases showed chromosomal aberrations after ethanol; this mechanistic detail was absent
Fork protection Garcia-Higuera 2001 citation removed — that paper does not describe fork protection; Schlacher 2012 DOI retained as the correct (unarchived) source; gap tag maintained
Rall-Scharpf 2025 cohort n’s and quantitative endpoints added — FANCD2 protein ~50% decreased in old women (n=60) not men (n=54); POLη inhibitor IC50 4.5-fold lower in old women; PARP inhibitor sensitivity 3.3-fold higher in old men; full Table 1 donor demographics added to body
Schema gap flagged — mouse-ortholog: frontmatter field is non-standard per CLAUDE.md; new limitation #5

Unverifiable:

Garaycoechea 2012 (not_oa) — no-fulltext-access; Fanca genotype inferred from 2018 paper’s citation
Schlacher 2012 fork protection — not yet archived; gap tag retained
FA-D2 prevalence 3-5% — unverified against registry source; needs-verification

Downstream pages that may need checking:

molecules/proteins/fanci.md — shares Longerich 2014 and Alcon 2020 citations; verify that its framing of FANCI monoubiquitination also reflects the independence finding from Longerich 2014 and Alcon 2020’s FANCI K525 not-ubiquitinated result
Any page citing Rossi 2007 for “FA pathway mutant HSC aging” claims — Rossi 2007 does not contain FA pathway models

[2026-05-13] verify | molecules/proteins/palb2.md

Pages verified: 1

molecules/proteins/palb2.md — PALB2 protein page (auto-extracted 2026-05-13)

Sources checked:

doi:10.1038/ng1947 (Reid 2007, Nat Genet) — PDF read end-to-end (local)
doi:10.1016/j.molcel.2006.05.022 (Xia 2006, Mol Cell) — PDF read end-to-end (local after download)
doi:10.1056/NEJMoa1400382 (Antoniou 2014, NEJM) — PDF read end-to-end (local after download)
doi:10.1093/hmg/ddq207 (Rantakari 2010) — PubMed abstract only; PDF download failed
doi:10.4161/cc.6.19.4759 (Rossi 2007) — PubMed abstract only; PDF download failed
doi:10.1038/s41586-019-1363-4 (Daza-Martin 2019) — PubMed abstract only; PDF still downloading
doi:10.1200/JCO.20.02151 (Tung 2020) — unverifiable (not_oa); Tung 2020 claims left with gap tag

Corrections made:

Reid 2007 family count: “n=8 unrelated FA patients” → “n=7 unrelated FA families” (PDF Table 1 lists 7 patients; abstract explicitly states “seven families”)
Antoniou 2014 cohort description: “362 families” → “154 families / 362 female PALB2 mutation carriers” (the 362 is the carrier count, not family count; 154 families were eligible after ascertainment adjustment)
Added mean cumulative risk figure (35%, 95% CI 26–46% by age 70) from Antoniou 2014 abstract — seeder omitted this summary statistic
Coiled-coil domain coordinates corrected: “residues 1–160” → “CC proper residues 9–41; oligomerization region 1–160” per UniProt Q86YC2 Coiled coil annotation [9-41]
BRCA1-independent HR attribution corrected: removed incorrect Xia 2006 footnote attribution for “partial HR activity of PALB2 constructs that cannot bind BRCA1” — Xia 2006 does not demonstrate this; claim re-sourced to UniProt structural data with a needs-replication tag
Cisplatin sensitivity claim softened: Reid 2007 demonstrates MMC sensitivity specifically; cisplatin listed as general FA characteristic but not specifically attributed to FA-N in Reid 2007
mr-causal-evidence: “not-tested” → “partial” — GWAS Catalog shows germline hits in PALB2 region (rs45569335, p<3e-10) but no MR study for aging outcomes has been published
All footnote PDF references corrected to the right local path.
Xia 2006 and Antoniou 2014 footnotes enriched with actual quantitative findings from PDFs
⚠️ auto-extraction banner removed

Pages unverifiable (closed-access):

Tung 2020 (doi:10.1200/JCO.20.02151) — not_oa; 82% ORR and n=13 claims cannot be independently verified; retain with gap tag

Identity fields confirmed via API:

NCBI Gene 79728 ✓ (symbol: PALB2, chr: 16p12.2)
UniProt Q86YC2 ✓ (PALB2_HUMAN, Swiss-Prot reviewed, 1186 aa)
HGNC 26144 ✓ (NCBI Gene: 79728, Ensembl: ENSG00000083093)

Downstream pages that may need checking:

Any page citing … for PALB2-related claims should have footnotes re-checked against corrected Reid 2007 n (7 families) and Antoniou 2014 cohort description (154 families)

[2026-05-12] ingest | Chen et al. 2025 Science — naked-mole-rat cGAS HR-repair (cGAS dual-role + DNA-repair reframe)

User-requested ad-hoc ingestion of doi:10.1126/science.adp5056 (PMID 41066557; closed-access, NOT in PMC). The paper establishes a second, nuclear/chromatin-bound role for cGAS (homologous-recombination suppressor) parallel to its canonical cytosolic-immune-sensing role — and shows that naked mole-rat cGAS has evolutionarily lost this HR-suppressive function via four amino-acid divergences that weaken TRIM41-mediated ubiquitination and P97/VCP segregase eviction, leaving cGAS on chromatin longer → enhanced FANCI–RAD50 interaction → potentiated HR → lifespan extension. Conceptual reframings beyond the obvious naked-mole-rat page were surfaced and propagated per the paper_impact_broad feedback memory.

Caveat — verification blocked: Paper is closed-access (Science) with no PMC deposit and no PDF in a local paper archive. Per CLAUDE.md, the new study page and all derived claims carry #gap/no-fulltext-access indefinitely; specific quant claims (n’s, p-values, lifespan-extension magnitudes) are NOT in the abstract and are NOT asserted anywhere in the wiki. Re-verify when full text becomes available.

New atomic pages (9; all verified: false with auto-extracted banners and per-claim #gap/no-fulltext-access for Chen 2025 content):

studies/chen-2025-nmr-cgas-hr-repair.md — primary study page; closed-access banner + no-fulltext-access throughout
molecules/proteins/cgas.md — dual-arm protein page (cytosolic immune-sensing arm canonical + nuclear/chromatin HR-suppression arm from Chen 2025 + Liu 2018); UniProt Q8N884 verified; druggability tier 2 (aging-context — no aging-validated probe); 7 footnotes; integrates Glück 2017 CCFs, Salminen 2025 review
molecules/proteins/sting.md — ER→Golgi cycle; Gulen 2023 Nature in-vivo aged-mouse Sting−/− causal evidence integrated; UniProt Q86WV6 verified
molecules/proteins/rad50.md — MRN component; Q92878 verified; NBSLD progeroid section; FANCI–RAD50 axis as new Chen 2025 regulator
molecules/proteins/fanci.md — Q9NVI1 verified; canonical FA-pathway / ID2 / ICL-repair context + new Chen 2025 broader DSB role (with three documented mechanism ambiguities); Rall-Scharpf 2025 NAR sex-specific FA-aging integrated
molecules/proteins/trim41.md — Q8WV44; HGNC:19013 (corrected from seeder-brief’s 19014); Zhen 2023 Nat Commun (same Mao lab — prequel on TRIM41 + nuclear cGAS + ORF2p) integrated
molecules/proteins/vcp.md — P55072; multi-substrate context (ERAD, RQC, mitophagy, DDR/chromatin extraction); CB-5083/NMS-873/UPCDC30245 drug-target landscape; MSP/IBMPFD segmental aging syndrome; Lin 2026 STUB1-VCP-PINK1 mitophagy axis
interventions/pharmacological/cgas-modulators.md — NEW intervention class; Type A pan-cGAS catalytic inhibitors / Type B chromatin-retention enhancers (NMR-inspired; TRIM41 + VCP inhibitors) / Type C NMR-cGAS variant gene therapy; clinical-trials-active=0 for aging indication; added 3 new mechanism classes to frameworks/intervention-classes.md
hypotheses/dna-damage-theory-of-aging.md — NEW Mode-A evidence-aggregating hypothesis page; 5 explicit falsifiable predictions; integrates Vijg 2021, Panier 2024, Toiber & Schumacher 2026, Yang 2023 ICE, Chen 2025 NMR cGAS, Roichman 2021 SIRT6; all 27 wikilinks resolve; differentiated from somatic-mutation, information-theory, free-radical, hyperfunction in body

Propagation pass — 8 existing pages updated (verifier dispatches; all existing verified: true preserved; new content tagged #gap/no-fulltext-access):

pathways/cgas-sting.md — new “Nuclear cGAS: dual-role at chromatin” section (~270 words); De Cecco 2019 LINE-1 DOI resolved (10.1038/s41586-018-0784-9, was prior #gap/unsourced); caused-by:/causes: updated with bi-modal output
model-organisms/heterocephalus-glaber.md — new “cGAS HR-repair potentiation” subsection; new “Unifying brake-removal frame” section with 5-row table (HMW-HA / cGAS / pALT / proteostasis / non-Gompertzian mortality each mapped to specific pro-aging brake removed); literature-checked-through: 2026-05-12 set
pathways/homologous-recombination.md — new “Negative regulation by nuclear cGAS” section (~280 words); key-nodes: extended
pathways/dna-damage-response.md — new “Nuclear cGAS: an intra-nuclear brake on HR” subsection; Liu 2018 DOI corrected (10.1038/s41586-018-0195-z → 10.1038/s41586-018-0629-6) during final lint
hallmarks/genomic-instability.md — new §5 “cGAS nuclear HR-brake”; existing causes: [[chronic-inflammation]] confirmed; mechanistic-bridge note added
hallmarks/chronic-inflammation.md — new dual-role note within NF-κB driver section (~5 sentences); cross-reference to [[hallmarks/genomic-instability]] highlighting cGAS as a single molecular node bridging both hallmarks
molecules/proteins/mrn-complex.md — RAD50 HGNC corrected 9817 → 9816 (verified via UniProt + HGNC REST APIs); new “FANCI–RAD50 interaction and upstream MRN recruitment regulation” section with three documented mechanism ambiguities (#gap/needs-mechanism-clarity)
hypotheses/antagonistic-pleiotropy.md — cGAS added as new molecular AP exemplar (antiviral defense [early-life benefit] vs. HR suppression + inflammaging [late-life cost], with NMR comparative-biology natural experiment); literature-checked-through: 2026-05-12 refreshed

Conceptual reframings persisted in the wiki:

cGAS-STING dual-role — formerly framed as purely cytosolic-immune-sensing inflammaging driver; now documented with parallel nuclear/HR-suppression arm. Pure cGAS inhibitors (current autoimmune-indication pipeline) would knock out both arms; NMR-inspired chromatin-retention enhancers (TRIM41, VCP/p97 inhibitors) selectively address the HR arm.
NMR longevity portfolio = brake-removal pattern — formalized in heterocephalus-glaber.md Unifying-frame section: NMRs work by removing pro-aging brakes that exist in shorter-lived mammals (HMW-HA degradation absent, cGAS HR-suppression weakened, telomerase regulation, proteostasis robustness, non-Gompertzian mortality). Cleaner than the prior “bag of tricks” framing.
Hallmark causal-graph: new mechanistic bridge — cGAS is now a single molecular node mechanistically connecting [[genomic-instability]] (via nuclear HR-suppression arm) and [[chronic-inflammation]] (via cytosolic immune-sensing arm). The pre-existing causes-relationship genomic-instability → chronic-inflammation gains a new mechanistic route (the cGAS dual-arm route alongside Rodier 2009 SASP + Jaiswal 2014 CHIP).
Intervention modality reframing — cgas-modulators class page formalizes the direction-of-effect problem: Type B (chromatin-retention enhancers) is opposite-direction from canonical Type A pan-cGAS inhibition. Three new mechanism classes added to intervention-classes.md: innate-immune-modulation, hr-repair-modulation, ubiquitin-ligase-modulation.
Antagonistic-pleiotropy hypothesis — gained a clean molecular exemplar (cGAS evolved-for-antiviral-defense vs. late-life HR suppression + inflammaging); strengthens the conceptual-frame Mode-B status of the hypothesis.
DNA-damage theory of aging — new Mode-A hypothesis page seeded; Chen 2025 NMR cGAS positioned as a clean prediction-confirming finding (removing an HR brake → improved repair → extended lifespan). The page also positions DDT against adjacent theories (somatic-mutation as subset, information-theory as downstream, hyperfunction as competing).

Gaps surfaced:

#gap/no-fulltext-access — Chen 2025 throughout; Liu 2018 nuclear cGAS paper (closed-access) also tagged on cgas-sting.md, homologous-recombination.md, dna-damage-response.md, genomic-instability.md, cgas.md
#gap/needs-mechanism-clarity — FANCI–RAD50 interaction mode (direct binding vs ID2-complex-mediated vs cGAS-scaffolded) per Chen 2025 abstract underspecification
#gap/needs-human-replication — NMR cGAS variant lifespan extension not yet demonstrated as a knock-in in mice or human cells
#gap/no-mechanism — NMR antiviral capacity with prolonged-chromatin-retention cGAS variant: does it cost the original evolved function? Chen 2025 abstract silent
#gap/needs-canonical-id — cGAS GenAge entry (not present); STING HGNC/Ensembl re-verification pending; VCP Ensembl pending
#gap/contradictory-evidence — none surfaced; Chen 2025 extends rather than contradicts Liu 2018 and prior cGAS-STING-inflammaging literature

Follow-on seeding candidates (deferred — not in this paper-impact pass):

studies/liu-2018-nuclear-cgas-hr-suppression.md — foundational nuclear-cGAS paper; currently referenced from multiple pages as direct DOI citation; would benefit from a wiki study page
studies/zhen-2023-trim41-cgas-l1.md — Zhen et al. 2023 Nat Commun (PMID 38086852) from the same Mao lab as Chen 2025; covers TRIM41 + nuclear cGAS + ORF2p in senescent cells; prequel to Chen 2025
molecules/proteins/fancd2.md — referenced by fanci.md but missing
molecules/proteins/palb2.md — referenced by fanci.md but missing
molecules/proteins/irf3.md — referenced by sting.md but missing
frameworks/causal-graph-data.md — optional enhancement: update the existing genomic-instability → chronic-inflammation edge key-citation field to note the cGAS dual-arm mechanistic route

Verifier propagation TODO (closed-access Liu 2018 / Chen 2025): when either paper becomes accessible (PMC deposit, preprint, or institutional access), re-trigger the wiki-verifier on all pages tagged #gap/no-fulltext-access to confirm quantitative claims.

Naming conflict resolved: the batch-1 cGAS seeder created a [[p97-vcpase]] wikilink (non-standard). Propagation pass renamed to [[vcp]] across cgas.md and trim41.md (vcp.md self-aliases include p97-vcpase so the alias still resolves for back-compat). No remaining [[p97-vcpase]] wikilinks.

[2026-05-13] verify — studies/zhen-2023-trim41-cgas-l1.md

Source checked: Zhen Z et al. 2023 Nature Communications 14:8217. doi:10.1038/s41467-023-43001-y. Local PDF (3.5 MB) read end-to-end (14 pages; full text + methods + references). PubMed efetch confirmed title, authors, journal, PMC ID — all match frontmatter.

Corrections applied (3):

study-design: in-vitro → in-vitro+in-vivo (paper includes Cgas-KO mouse kidney + brain qPCR data; same correction principle as chen-2025 precedent)
organism: human-cell-lines → multiple (covers HeLa, IMR90-hTERT, HCA2-hTERT human lines + mouse)
Senescence section body: IMR90-hTERT and HCA2-hTERT were used for X-ray irradiation (15 Gy, Day 9 lysis) subcellular fractionation validation — NOT for the etoposide SIPS experiments. Etoposide (10 µg/mL, 20 min) SIPS was in HeLa cells. Corrected with accurate experimental detail per Fig. 4 caption + Methods.

Quantitative claims confirmed accurate (7 confirmed, no changes needed):

cGAS KO increased retrotransposition 2.2- and 2.5-fold in HeLa; n=6 independent experiments (Fig. 1c, d + caption)
L1 copy-number increase in Cgas-KO kidney + brain; 9 replicates from 3 independent mice per group; 3–4-month-old male and female mice (Fig. 1h, i + caption + Methods)
CHK2 phosphorylates cGAS at S120 and S305; S120A/S305A double mutant reduces inhibitory activity more than either single (Fig. 3h + body text p.3)
Mass spec: 5 candidate E3 ligases (PIAS1, ZNF598, TRIM21, TRIM41, TRIM25); only TRIM41 overexpression phenocopied ORF2p reduction (Fig. 2e, f)
Interaction surfaces: ORF2p EN domain + TRIM41 coiled-coil domain (Supplementary Figs cited in body text p.3)
ORF2p degradation MG132-sensitive; K48-linked polyubiquitination confirmed (Fig. 2a–d)
7 of 37 cancer-associated cGAS mutants abolish L1 restriction while retaining cGAMP synthesis; three mechanistic groups confirmed (Fig. 5a–g; P486L overlaps two groups — faithful to paper)

Supersession check (nuclear cGAS / TRIM41 / ORF2p / LINE-1; 2024–2026):

No meta-analyses or large RCTs on this mechanism found (basic-science mechanistic study; not an intervention RCT class)
One parallel mechanism paper found: Huang Y et al. 2025 EMBO Rep (PMID 40825873) — “STING inhibits LINE-1 retrotransposition through sorting ORF1p to lysosomes for degradation” — describes cGAS-independent STING/ORF1p lysosomal pathway, orthogonal to Zhen 2023’s nuclear-cGAS/TRIM41/ORF2p axis; does NOT contradict Zhen 2023; cited for awareness
No papers retract or significantly revise the TRIM41-ORF2p-nuclear-cGAS mechanism

Downstream propagation needed: None at this time — no entity pages currently cite studies/zhen-2023-trim41-cgas-l1 via wikilink (trim41.md, cgas.md, and genomic-instability.md reference the paper via DOI-inline footnotes on entity pages; those pages’ claim text does not inherit specific numerics from this study page that were corrected). Flag for propagation when entity pages are next verified.

Final state: verified: true; verified-date: 2026-05-13; verified-by: claude.

[2026-05-12] verify — molecules/compounds/hyaluronic-acid.md

Sources checked:

Tian 2013 (doi:10.1038/nature12234) — local PDF, full text read (8 pages); MW figures, HAS2 Asn→Ser, CD44/NF2/p16 chain, HAS2 knockdown + HYAL2 overexpression rescue, affinity assay, xenograft experiments — all confirmed
del Marmol 2021 (doi:10.1038/s41598-021-86967-9) — local PDF, full text read (15 pages); SEC peak ~2.5 MDa (lymph node/skin), agarose gel NMR skin 400 kDa–1.3 MDa (no ≥3 MDa), HABP-based localization, NMR vs guinea pig tissue comparisons, HYAL1 zymography — all confirmed
Zhang 2023 (doi:10.1038/s41586-023-06463-0) — PMC10666664 full text; n=84/91 lifespan cohort, CAG promoter + Lox-STOP + creER controls, +4.4%/+12.2% lifespan, 57%/70% cancer, IL-12p40/MIP1α/MIP1β/CCL7 inflammatory panel, FITC-dextran gut permeability, microbiome Bacteroidetes/Firmicutes, H2O2 resistance — all confirmed
Zhao 2023 (doi:10.1038/s41467-023-43623-2) — DOI confirmed via Crossref; title and findings consistent with wiki framing
Bauer 2024 (doi:10.1016/j.abb.2024.110098) — DOI confirmed via Crossref; NMR TMEM2 lacks hyaluronidase activity confirmed
Fong 2024 (doi:10.1016/j.joca.2024.02.294) — DOI confirmed via Crossref; meta-analysis of single-dose HA vs inactive controls in knee OA confirmed
DrugBank DB08818 — confirmed (entry exists for hyaluronic acid)
PubChem null — confirmed (no canonical polymer CID)
ChEMBL null — confirmed (API returned no HA polymer entry)

Corrections applied:

“NMR cells secrete ~5× more HA… per unit area” → corrected to “NMR HA is ~5× larger in MW” (Tian 2013 abstract says “over five times larger” in MW, not 5× more quantity)
“vs ~0.5–1 MDa in mouse and human” (Tian 2013 section) → “vs 0.5–3 MDa in mouse/guinea pig and 0.5–2 MDa in human, per Tian 2013 Fig. 1b”
del Marmol section rewritten: added agarose gel data (NMR skin 400 kDa–1.3 MDa spread, no ≥3 MDa HA); corrected characterization of Alcian Blue limitations (not HA-specific — confirmed by del Marmol); clarified pulse-field gel is the Tian 2013 method, SEC is del Marmol’s method; added guinea pig as primary comparison species (not mouse) for tissue measurements
“Reduced circulating IL-6, TNF-α, and MCP-1” → corrected to name actual Zhang 2023 inflammatory markers (IL-12p40, MIP-1α/β, CCL7; IL-6/TNF-α reduction was post-LPS challenge, not resting baseline)
“Improved glucose tolerance in aged animals” — removed; not supported by Zhang 2023 PMC text
“wildtype littermate controls” → corrected to “creER littermate controls” (controls are tamoxifen-treated creER, not wildtype)
“produced by mouse Has2 overexpression” → corrected to “transgene was nmrHas2 (naked mole-rat HAS2)”; interpretive point preserved
CD44/NF2 mechanism: YAP/TAZ claim qualified — Tian 2013 does not specifically demonstrate YAP/TAZ involvement; it establishes HA/CD44/NF2/p16^INK4a; YAP/TAZ is a general NF2/Hippo inference, not from Tian 2013 directly
HYAL2 “primary NMR hyaluronidase” qualifier removed — not directly supported by either Tian 2013 or del Marmol 2021 (which shows NMR HYAL2 is lower than mouse HYAL2)
Zhang 2023 footnote: n’s added (84/91 lifespan; 11/13 DMBA/TPA); CAG promoter + Lox-STOP noted; “PDF download failed” note updated to “verified against PMC10666664”
del Marmol 2021 footnote: species corrected (NMR vs guinea pig for tissue, not mouse/human); n’s per tissue added; methods specified
Body content/turnover (15 g / 5 g) — tagged unsourced with canonical attribution to Laurent 1987

Claims confirmed as correct (no change needed):

6–12 MDa reported peak for NMR HA by Tian 2013 (pulse-field gel)
Asn→Ser at two positions in NMR HAS2 vs all other mammals — confirmed
HAS2 knockdown + HYAL2 overexpression rescue experiments — confirmed
CD44/NF2/p16^INK4a chain — confirmed
del Marmol ~2.5 MDa SEC peak — confirmed
del Marmol ABUNDANCE advantage confirmed (NMR > guinea pig in skin, muscle, lymph node)
Zhang 2023 +4.4% median, +12.2% max lifespan — confirmed
Zhang 2023 cancer 57% vs 70% — confirmed
Zhang 2023 “mild” HMW-HA accumulation attributed to HYAL activity in mice — confirmed
Zhang 2023 gut organoid HMW-HA (not LMW-HA) rescue — confirmed
Zhao 2023 DOI and findings — confirmed
Bauer 2024 DOI (NMR TMEM2 no hyaluronidase activity) — confirmed
Fong 2024 DOI (meta-analysis OA) — confirmed
Intervention-classes framework: extracellular-matrix-remodeling confirmed in use on ghk-cu.md; immunomodulation confirmed in use on mesenchymal-stem-cell-therapy.md and ll-37.md
clinical-trials-active: 0 — confirmed (all active trials are cosmetic/dermal, none for systemic aging endpoints)

R25 supersession check: PubMed search for “hyaluronan AND (aging OR longevity OR healthspan OR inflammaging) AND (meta-analysis[pt] OR randomized controlled trial[pt])” 2022–2026 returned 101 results; top 10 reviewed — all are cosmetic/dermal/OA trials, none address systemic aging endpoints. No supersession of Zhang 2023 framing. literature-checked-through: 2026-05-12 confirmed.

Downstream pages for propagation (main agent):

studies/tian-2013-hmw-ha-nmr-cancer-resistance.md — if this page exists, verify HYAL2 “primary hyaluronidase” qualifier and YAP/TAZ attribution
studies/zhang-2023-nmrhas2-mouse-healthspan.md — already verified; confirmed “creER controls” and inflammatory marker specifics align with this correction pass
molecules/proteins/has2.md — already verified; confirm HYAL2 qualifier and “mouse Has2” correction are consistent

[2026-05-12] verify — molecules/proteins/cd44.md

Sources checked (all PDFs read):

Tian 2013 (doi:10.1038/nature12234) — local PDF verified (5 pages); primary mechanism source
Ponta 2003 (doi:10.1038/nrm1004) — local PDF verified (5 pages); domain structure + PTM review
Dalerba 2007 (doi:10.1073/pnas.0703478104) — local PDF verified (4 pages); CSC marker
Takasugi 2023 (doi:10.1016/j.celrep.2023.113130) — PDF downloaded + verified (6 pages)
Termeer 2002 (doi:10.1084/jem.20001858) — PDF downloaded + verified (4 pages)
Menke 2016 (doi:10.18632/oncotarget.11098) — PDF downloaded + verified (5 pages)
UniProt P16070 — REST API verified; canonical structure + PTM sites confirmed
He 2026 (PMID 41806573) — confirmed real via PubMed; DOI 10.1016/j.bbrc.2026.153576 confirmed

Corrections applied (7 factual; 2 attribution):

“CD44 knockdown via siRNA” → “CD44 blockade via CD44-blocking antibody” (Tian 2013 used antibody, not siRNA — 2 locations: numbered list item 5 and Aging context section)
“vs 0.5–2 MDa in human or mouse” → “vs 0.5–3 MDa in mouse/guinea pig and 0.5–2 MDa in human” (Tian 2013 p.2: mouse/guinea pig 0.5–3 MDa; human 0.5–2 MDa)
Hippo cascade elaboration (STK4/STK3 → LATS1/2 → YAP/TAZ) — Tian 2013 does NOT name the Hippo kinase cascade; rewritten as inference note citing broader literature
Takasugi 2023 ATF6 mechanism: “HMW-HA → CD44 engagement activates ATF6” → “CD44 activates ATF6 in an HA-independent manner” (paper explicitly states “without the requirement of HA”; mechanism is ER-localized, not extracellular-HA-dependent)
Termeer 2002 TLR2 claim: “TLR4 (and in some contexts TLR2)” → “TLR4 only” (paper explicitly showed TLR-2-deficient mice still susceptible; TLR2 not required)
Termeer 2002 cytokine list: “TNF, IL-6, IL-12” → “TNF-α + p38/p42/44 MAPK/NF-κB” (IL-6 and IL-12 not demonstrated in this paper; TNF-α specifically measured)
Dalerba 2007 “6/6 CRC specimens” → “6 distinct xenograft lines from patient specimens” (tumorigenicity tested in xenograft lines, not direct primary specimens; also clarified EpCAM^hi co-requirement)
RG7356 “hematologic malignancies” claim removed from body text (Menke 2016 is solid tumor trial only; n=65; corrected DLT doses and MTD)
Ponta 2003 PTM attribution clarified: Ponta 2003 uses variant-isoform residue numbering; UniProt P16070 canonical positions cited as authoritative

Claims verified as correct:

Canonical length 742 aa, signal peptide 1–20, link domain 32–120, TM 650–670, cytoplasmic tail 671–742 (UniProt P16070 confirmed)
NMR HA 6–12 MDa (Tian 2013 confirmed)
N-glycosylation sites Asn-25, -57, -100, -110, -120, -350, -548, -599, -636 (UniProt confirmed)
Ser-672 PKC phosphorylation (UniProt confirmed)
Takasugi 2023 cross-species longevity correlation confirmed (fibroblasts 10 sp, liver 27 sp, skin 20 rodents; PGLS p<0.0064)
Menke 2016: 21% SD (13/61), MTD 1,500 mg q2w, no objective responses — confirmed
Zhang 2023 nmrHas2 lifespan numbers (+4.4% median, +12.2% max, 57% vs 70% cancer) — confirmed (cross-checked against verified study page)
He 2026 (PMID 41806573) confirmed real; HYAL2/LMW-HA/CD44/AKT mechanism confirmed via PubMed abstract
GenAge null status — not checked against database; flagged in verified-scope

Supersession check (literature-checked-through: 2026-05-12): R25 not strictly required for type:protein. No superseding meta-analysis or RCT found in 2024–2026 CD44 aging literature.

Downstream pages to check (not propagated here):

molecules/compounds/hyaluronic-acid.md — may cite “TLR2” or the “HMW-HA → CD44 → ATF6” framing; both now corrected here
model-organisms/heterocephalus-glaber.md — cites Tian 2013 CD44 mechanism; siRNA vs antibody correction may need propagation
studies/tian-2013-nmr-hmw-ha-cancer.md — no study page yet; flagged as seeder candidate (cited from 3+ atomic pages)

[2026-05-12] verify — molecules/proteins/has2.md

Pages verified: 1 — molecules/proteins/has2.md

Sources checked:

Tian 2013 (doi:10.1038/nature12234) — local PDF verified; main text read; Asn→Ser positions confirmed absent from main text (gap tag updated)
Zhang 2023 (PMC10666664 / doi:10.1038/s41586-023-06463-0) — PMC HTML verified; survival n’s, lifespan %, cancer %, epigenetic age method confirmed
del Marmol 2021 (doi:10.1038/s41598-021-86967-9) — local PDF verified; ~2.5 MDa NMR HA max, no ≥3 MDa, Alcian Blue critique confirmed; NMR fibroblast culture conditions (32.5°C, 5% O2) confirmed
Camenisch 2000 (doi:10.1172/JCI10272) — PubMed abstract (PMID 10930438) verified; correct PMID added; lethality stage E9.5–10 confirmed (wiki had “E9.5–10.5”)
UniProt Q92819 — 552 aa, 7-TM helices, Mg²⁺, PTM sites Ser-221/Thr-110/Thr-328/Lys-190 all confirmed
HGNC 4819, NCBI Gene 3037, Ensembl ENSG00000170961, mouse ortholog P70312 — all confirmed

Corrections applied:

Lethality stage: “E9.5” (opening paragraph) and “lethal at E9.5” (isoform table) → “E9.5–10” (per Camenisch 2000 abstract: “E9.5–10”)
Lethality mechanism: “cardiac jelly absent; endocardial cushion defects; vascular failure” → “lack HA; severe cardiac and vascular abnormalities; failure of cardiac endothelial-to-mesenchyme transformation” (more precise per abstract)
Camenisch 2000 footnote: wrong PMID removed; PMID 10930438 added; lethality description rewritten to match abstract; DOI case corrected to JCI10272
Tian 2013 Asn→Ser text: changed from “[positions TBD — needs verification]” framing to accurate statement that positions are not given in main text but the active-site cytoplasmic loop location is confirmed
“≥4 MDa” threshold (del Marmol section) → “≥3 MDa” (matching del Marmol language and agarose gel result)
del Marmol 2021 body text: added “NMR fibroblasts cultured under physiological conditions (32.5°C, 5% O2)” specificity
Zhang 2023: survival n’s filled in (n=84/91); gap/needs-verification markers removed from lifespan % (now verified); cancer >27-mo data added (49% vs 83%); sex-specific lifespan data added; epigenetic age delta (~−0.2 yr) added; inflammation analytes named
Zhang 2023 footnote: n’s added; PMID added; “lifespan n not confirmed” note removed
del Marmol 2021 footnote: n’s per tissue added; method detail expanded; culture conditions added
⚠️ banner removed

Supersession check: PubMed 2023–2026 search for “HAS2 aging longevity” returned 1 result = Zhang 2023 itself (PMID 37612507). No superseding meta-analysis or RCT found.

Downstream pages to check (not propagated here):

studies/zhang-2023-nmrhas2-mouse-healthspan.md (already verified separately; cross-check the Camenisch lethality stage if cited there)
studies/tian-2013-nmr-hmw-ha-cancer.md (to be created — seeder candidate)
Any page citing has2 or hyaluronic-acid that quotes embryonic lethality stage

[2026-05-12] verify — studies/zhang-2023-nmrhas2-mouse-healthspan.md

Source checked: PMC10666664 (HTML; no local PDF — OA-package unavailable). Tian 2013 (doi:10.1038/nature12234) verified against local PDF. del Marmol 2021 (doi:10.1038/s41598-021-86967-9) verified against local PDF. Author list verified via PubMed efetch PMID 37612507.

Corrections applied:

Cancer >27-month subanalysis n’s added: n=74 nmrHas2, n=81 creER (previously omitted; full cohort n=84/91 stated only)
Tamoxifen route: “oral gavage” → “route not specified in paper methods” (paper says “received 80 mg/kg” without specifying route)
DMBA/TPA per-sex n’s added from Extended Data: females n=4 nmrHas2 vs n=5 creER; males n=7 vs n=8 (previously pooled-only)
del Marmol 2021 footnote rewritten: original text claimed it “confirmed NMR plasma HA is predominantly high-molecular-mass and ~5× more abundant than in rats” — corrected to: del Marmol compared NMR vs guinea pig (not rats), found NMR HA peak MW ~2.5 MDa (not 6–12 MDa), found no ultra-HMW HA ≥3–4 MDa, and partially challenges Tian 2013
Follow-up section corrected from “Two closely related papers” to “Three” (Zhao 2023, Takasugi 2023, McGuire 2024 — all three confirmed via Crossref)

Claims verified as correct (no change needed):

Lifespan: +4.4% median (n=84 vs 91), +12.2% maximum; females +9% median (n=50 vs 54), males +16% maximum (n=34 vs 37); two-tailed log-rank p<0.05
Cancer overall: 57% vs 70% (full cohort); >27mo: 49% vs 83% (two-tailed χ²)
Epigenetic clock: HorvathMammalMethylChip40, n=9 per group, −0.2 yr age acceleration, 165/6,553 differentially methylated CpGs (145 gain-sites lower, 20 loss-sites higher)
Inflammation panel: IL-12p40, MIP1α, MIP1β, CCL7; old females n=7 nmrHas2, n=8 creER; young n=11 each
Gut barrier: FITC-dextran 4 kDa; old n=12 per group; HMM-HA rescue at 20 µg/mL; LMW-HA no effect
Microbiome: n=9 nmrHas2, n=10 creER; Bacteroidetes↑ Firmicutes↓ old groups only; Muribaculaceae↑, Deferribacteraceae/Streptococcaceae/Lachnospiraceae↑ in controls
Frailty index: young creER n=17, nmrHas2 n=16; old creER n=13, nmrHas2 n=14
Transgenic construct: pCALNL-GFP (Addgene #13770), CAG promoter, Lox-STOP cassette, C57BL/6 background, R26-CreERT2 cross
Mouse Has2 control: primary skin fibroblasts; H₂O₂ at 100/200/400 µM; equivalent protection confirmed
Author list: 16 authors confirmed; Zhang Z and Tian X as co-first authors; Seluanov A and Gorbunova V as co-corresponding

R25 supersession check: Gorbunova/Seluanov “hyaluronan 2024-2026” PubMed search — no superseding meta-analysis or large RCT identified. Three follow-ups confirmed (Zhao 2023, Takasugi 2023, McGuire 2024) all extend rather than contradict Zhang 2023.

Downstream pages that may need propagation review:

molecules/compounds/hyaluronic-acid.md — footnote [^zhang2023] cites “wildtype littermate controls” (incorrect; controls are creER-only, not wildtype); cancer n’s may need updating
molecules/proteins/cd44.md — cancer/lifespan numbers cited correctly; no critical corrections needed
molecules/proteins/has2.md — footnote notes “lifespan n not confirmed from abstract” — now confirmed n=84/91

[2026-05-09] verify — molecules/compounds/patisiran.md

Result: verified: true (partial scope — HELIOS-B local PDF is a misfetched DI-fusion help file; HELIOS-B verified against PubMed abstract PMID 39213194; canonical-DB identity fields not re-verified).

Corrections made (14):

Infusion duration: “60-min infusion” → “approximately 80-minute infusion” (APOLLO paper: “period of approximately 80 minutes”)
TTR knockdown precision: “~80% reduction” → “median 81% reduction (range −38 to 95%)” (APOLLO paper states “median…was 81%”)
Peripheral edema rate: “29%” → “30%” (44/148 patisiran per APOLLO Table 3)
Trial geography: “49 countries” → “44 sites across 19 countries” (APOLLO paper methods; “49” was fabricated)
Norfolk QoL-DN CI added: difference −21.1; 95% CI −27.2 to −15.0 (from APOLLO text)
APOLLO-B randomization ratio: “2:1 patisiran:placebo” → “1:1 (181 patisiran, 179 placebo)” (APOLLO-B paper explicitly states 1:1)
APOLLO-B wtATTR %: “~56% wtATTR-CM” → “~80% wtATTR-CM” (Table 1: 144/181=80% patisiran, 144/178=81% placebo wild-type)
APOLLO-B 6MWT primary result: “patisiran −8.7 m vs placebo −21.7 m; difference +13.0 m (95% CI 0.5–25.4; p=0.040)” → “Hodges-Lehmann estimate +14.69 m (95% CI 0.69–28.69; P=0.02); medians −8.15 m patisiran vs −21.35 m placebo” (all three prior numbers were wrong)
HELIOS-B randomization ratio: “2:1 vutrisiran:placebo” → “1:1 (326 vutrisiran, 329 placebo)”
HELIOS-B primary endpoint statistic: “RR 0.72” → “HR 0.72” (hazard ratio, not rate ratio)
HELIOS-B all-cause mortality: “RR 0.64” → “HR 0.65 (through 42 months)” (point estimate and statistic type corrected)
HELIOS-B 6MWT: “difference +24.7 m” → “LS mean difference +26.5 m (95% CI 13.4–39.6)” (abstract gives 26.5 m)
5-year OLE enrollment: “212 patients who completed APOLLO were enrolled” → “211 of 212 eligible patients (from APOLLO and phase 2 OLE); 138 completed” (Adams 2025 JAMA Neurol)
Aimo 2025 NMA summary corrected: “Acoramidis and tafamidis ranked higher” — acoramidis was NOT significant in the NMA (HR 0.76, 95% CrI 0.56–1.03); only tafamidis and vutrisiran mono were significant vs placebo. NMA section rewritten with actual SUCRA values and HR/CrI data.

Archive note: HELIOS-B local PDF (10.1056_nejmoa2409134.pdf) is a 3-page DI-fusion library help guide fetched from a Belgian institutional repository redirect — not the actual paper. Filed for re-download with correct URL.

Downstream propagation needed: None identified (patisiran.md is the canonical compound page; no other pages currently cite these specific numerics per grep check).

[2026-05-09] verify — molecules/compounds/acoramidis.md

Result: verified: true (partial scope — Gillmore 2024 NEJM primary abstract-only; Maurer 2025, Masri 2026, Alexander 2026, Facin 2025, Briasoulis 2026 abstract-only; Judge 2025 JACC + Judge 2025 Circulation OLE full PDFs read).

Corrections made (3):

Lede win ratio CI added + attribution split: “win ratio of 1.8 (P<0.001)” → “1.8 (95% CI 1.4–2.2; P<0.001)” with separate citation for the 35.9% vs 50.5% figures now correctly attributed to Judge 2025 JACC (secondary analysis) rather than implicitly to Gillmore 2024.
Annualized CVH frequency CI added: RRR 50% row had ”—” in CI column → “95% CI 36–64%” (from Judge 2025 JACC Table 2: RR ratio 0.50, 95% CI 0.36–0.70).
Footnotes upgraded: Judge 2025 JACC and Judge 2025 Circulation OLE footnotes updated from abstract-level to full-PDF-verified with complete CIs and secondary outcomes.

Supersession check: No superseding RCTs or meta-analyses found. Two new papers (real-world meta-analysis PMID 41311351; medication use analysis PMID 40750720) are derivative. All 8 PMIDs from 2024–2026 PubMed search already cited or not superseding. literature-checked-through: 2026-05-09 confirmed.

ClinicalTrials.gov check: All 6 NCT numbers confirmed (NCT06563895, NCT07306949, NCT07235462, NCT07112066, NCT06465810, NCT04988386) with matching statuses.

[2026-05-09] verify — molecules/compounds/tafamidis.md

Result: verified: true (partial scope — Damy 2025, Judge 2025, Fontana 2025 not independently verified; Hussain 2025 abstract-only; Bulawa 2012 abstract-only; DrugBank/InChIKey not re-queried).

Corrections made (6):

Coelho 2012 n per group corrected: “64 tafamidis, 64 placebo” → “65 tafamidis, 63 placebo”; ITT population n=125 (not 128) added.
Coelho 2012 co-primary endpoint added: TQOL was a co-primary alongside NIS-LL; both missed in ITT (TQOL p=0.116). Wiki only mentioned NIS-LL.
Coelho 2012 ITT NIS-LL responder rates added: 45.3% vs 29.5% (p=0.068) — omitted from body text.
ATTR-ACT primary win ratio added to table: 1.695 (95% CI 1.255–2.289; p<0.001) — was absent.
CV hospitalization RRR precision corrected: “~32% reduction” → RRR 0.68 (95% CI 0.56–0.81); 0.48 vs 0.70/yr — exact figures from paper.
Facin 2025 attribution clarified: the RR 0.71/0.81 figures are for the TTR stabilizer subgroup (tafamidis + acoramidis combined), not tafamidis alone; this nuance was absent.

Supersession check: Fan & Coleman 2026 (AJHP, PMID 42084327) meta-analysis of 2 RCTs (n=1,052) confirms NYHA I/II benefit concentration (HR 0.64 mortality, RR 0.46 CV hospitalizations) and no benefit in NYHA III — consistent with wiki framing. No direct supersession of tafamidis evidence; the page’s treatment-stage gradient claim is supported.

Downstream pages that may need updates: none identified (no other wiki pages currently cite tafamidis study-page cross-links to the corrected figures).

[2026-05-09] verify — molecules/compounds/donanemab.md

Result: verified: true (partial scope — Zimmer 2025 ARIA, Zimmer 2026 LTE, Lu 2025, and Yaari 2025 are abstract-only; Nonino 2026 Cochrane verified via PubMed abstract PMID 41985900).

Corrections made (7):

ARIA-E symptomatic rate corrected: 5.8% → 6.1% (52/853; Sims 2023 Table 3).
ARIA-H rate corrected: 31.3% → 31.4% (268/853; Sims 2023 Table 3).
ARIA-H placebo rate corrected: 13.0% → 13.6% (119/874; Sims 2023 Table 3).
Amyloid clearance claim made precise: “>75% achieved clearance” → 80.1% low/medium tau and 76.4% combined at 76 weeks, with CIs, attributed to Sims 2023 (not Zimmer 2026 LTE as originally sourced).
Cochrane antibody count corrected: “9 anti-amyloid antibodies” → “7”; conclusion language aligned with abstract.
Combined iADRS LSM values made precise: -10.2/-13.1 → -10.19/-13.11 (Sims 2023 Table 2).
ARIA source discrepancy documented in-text: 24.0% (Sims 2023 TBZ-ALZ 2 only) vs 24.4% (Zimmer 2025 combined trials) explained.

Confirmed correct (representative): n per arm (860/876), all iADRS/CDR-SB differences + 95% CIs + p-values, all TBZ-ALZ4 clearance rates and median time to clearance, ARIA rates in TBZ-ALZ4, dosing regimen, limited-duration paradigm cutoff (<24.1 CL).

Supersession candidates: Nonino 2026 Cochrane is the primary supersession candidate (abstract verified; full text not accessible); rated “trivial” cognitive benefit across the class. No additional post-Sims-2023 RCTs or meta-analyses identified beyond Cochrane 2026.

[2026-05-09] verify — molecules/compounds/lecanemab.md

Result: verified: true (partial — van Dyck 2023 NEJM primary paper verified end-to-end against local PDF; OLE alz.70905 and safety meta trc2.12377 remain abstract-level only; brain volume claim and QoL percentages removed pending sourcing).

Corrections made (7):

ARIA-H placebo rate corrected: 2.7% → 9.0% (81/897 per Table 3). The 2.7% figure was the placebo ARIA-E rate, not ARIA-H.
ARIA-H lecanemab rate corrected: 13.0% → 17.3% (155/898 per Table 3). The 13.0% was “ARIA with microhemorrhages or hemosiderin deposits” (subset label from Table 3 ≠ total ARIA-H).
Falls rate corrected: “11%” → 10.4% (93/898) lecanemab; 9.6% (86/897) placebo — both arms now reported.
Symptomatic ARIA-E rate added: 2.8% (25/898) lecanemab vs 0% placebo; symptoms specified per paper text.
APOE4-stratified ARIA-E rates added from Table 3: noncarrier 5.4%, heterozygote 10.9%, homozygote 32.6% (lecanemab); symptomatic breakdown also added.
Brain volume paradox claim corrected: Van Dyck 2023 explicitly states MRI/volumetric results were not analyzed in the primary paper. Claim updated to flag as needing a separate source; gap tags added.
QoL secondary percentages (EQ-5D 49%, QOL-AD 56%, ZBI 38%) removed: Not present in van Dyck 2023 primary paper. Flagged unsourced pending identification of source (likely a supplementary analysis or conference presentation).

Schema issue: translation-gap: effect-size-modest was non-canonical. Changed to human-evidence-strong with inline comment explaining the choice and flagging for CLAUDE.md schema discussion (a ‘clinically-contested’ enum value may be warranted for FDA-approved drugs with Cochrane-disputed clinical meaningfulness).

Supersession candidate (Cochrane 2026, Nonino et al.): PMID 41985900, doi:10.1002/14651858.CD016297 — Cochrane meta-analysis of 17 RCTs (n=20,342; 9 anti-amyloid antibodies). Concludes effects on cognition are “trivial” and on function “small at best” — discordant with FDA-approval framing. Added as supersession section and footnote. Flagged for main-agent decision on AmyloSENS synthesis revision.

Downstream pages to update: sens-damage-categories § AmyloSENS (Cochrane discordance should be noted); any page that inherits the incorrect ARIA-H 13%/2.7% figures from this page.

[2026-05-09] verify — hypotheses/longevity-escape-velocity.md

Result: verified: true (partial — Oeppen-Vaupel 2002, Olshansky 2002, and Lewis-de Grey 2024 are closed-access; claims attributed to them remain abstract-level / secondary-description only).

Corrections made (4):

Hayflick-as-Warner-2005-signatory error: Wiki body listed Hayflick among Warner 2005 EMBO Reports signatories — he is NOT in the 28-name byline. Corrected to Miller and Melov (who are actual signatories). Hayflick co-signed the separate Olshansky 2002 position statement only.
De Grey 2004 AEV threshold corrected: Wiki said “perhaps one additional year of life expectancy per year” — not in the paper. De Grey 2004 actually states AEV requires ~10% per year sustained decline in age-specific mortality rates, and that a 30% healthy lifespan gain from first-generation therapies gives ~20 extra years to reach second-generation therapies. Body and footnote updated to match.
Palmer 2022 three-tier framing corrected: Wiki mischaracterized as “regenerative medicine + damage repair + prevention”. Actual tiers: Tier 1 (lifestyle/pharma), Tier 2 (full-spectrum surveillance/diagnostics), Tier 3 (genetic engineering/reprogramming). Body and footnote updated.
De Grey 2004 footnote publication type corrected: “commentary” → “book review / opinion” (PLoS Biology classified it “Book Review/Science in the Media”). Local PDF status updated from “pending” to “a local paper archive”.

Sources checked: de Grey 2004 (PDF verified); Warner 2005 (PDF verified, 28 signatories confirmed); Palmer 2022 (PDF verified); Oeppen-Vaupel 2002 (not_oa, abstract only); Olshansky 2002 (not_oa, no abstract available); Lewis-de Grey 2024 (not_oa).

Supersession check: PubMed “longevity escape velocity OR actuarial escape velocity OR Methuselarity” 2024–2026 returned 4 results. None are meta-analyses or RCTs. PMID 41205028 (GeroScience 2025 open-problems roadmap) does not mention LEV. PMID 40156933 (rodent aging interventions database) and PMID 38477630 (Lewis & de Grey 2024 — already cited) are the most relevant. No supersession detected. literature-checked-through already set to 2026-05-09.

[2026-05-09] verify — processes/glucosepane.md

Result: verified: true (partial — abstract-level; no local PDFs available for any of the 14 sources).

Corrections made (4):

Sell 2005 tissue scope corrected: Wiki said “skin and tendon collagen” → corrected to “skin collagen (n=110) and glomerular basement membrane collagen (n=28)”. The Sell 2005 paper measured skin + GBM, not tendon. Tendon appears in other AGE literature but not in Sell 2005.
Sell 2005 quantitative context expanded and qualified: Added actual units (pmol/mg), values at age 90 (~2000 nondiabetic, ~5000 diabetic), and clarification that the ~10–100× vs pentosidine ratio is a review-synthesis figure (Monnier 2005, Sjöberg 2009) rather than a direct Sell 2005 abstract statement.
Monnier 2022 specimen type corrected: Wiki body implied skin collagen; Monnier 2022 measured plasma protein-bound AGEs (n=466, 3 time points). Body text updated to specify plasma source and correct neuropathy p-value (p<0.003, not ≤0.02). Also clarified CSME association is driven by CML, not glucosepane specifically.
Hartog 2011 footnote enriched: Added primary endpoint (peak VO₂, P=0.06), secondary endpoint p-values from abstract, and corrected author list (Hartog JW not Hartog JWL).

Footnotes updated: All 14 footnotes enriched with PMIDs where found, volume/page data from Crossref/PubMed, and archive-status notes.

Supersession check (Sell 2005): PubMed 2009–2026 search for “glucosepane skin collagen aging” returned 6 results. No meta-analysis or large RCT challenges the foundational Sell 2005 abundance finding. Voziyan 2026 extends to bone site-mapping. Monnier 2014 and 2022 extend to clinical associations. No supersession detected.

literature-checked-through: confirmed 2026-05-09 (set by seeder).

Unverified claims needing full PDF: Sell 2005 pentosidine ratio table; Genuth 2015 n per complications arm; Gautieri 2014 “14 Lys–Arg pairs” specific count; Bakris 2004 Phase 2 trial n and p-values; Sjöberg 2009 alagebrium mechanistic-critique claim (not_oa, cannot verify).

[2026-05-09] verify — processes/lipofuscin.md

Result: verified: false maintained — all 7 cited primary sources inaccessible.

5 papers not_oa (Brunk & Terman 2002 Free Rad Biol Med; Terman & Brunk 2004 Int J Biochem Cell Biol; Nakano 1995; Moore 1995; Terman & Welander 1999).
2 OA papers (Brunk & Terman 2002 Eur J Biochem [bronze]; Wu & Sparrow 2010 JACS [green]) triggered download — both failed (no valid candidate URL resolved for archive ingest). Tagged #gap/no-fulltext-access.
Bibliographic metadata (title, authors, journal, pages, year) confirmed via Crossref for all 7 — no attribution errors found.
Terman 2004 citation count 439 vs 430 (Crossref current) — snapshot-timing difference; left as-is.

Corrections made (3):

CR dose-dependence added (Moore 1995 abstract): Only severe 52% CR (not 25% mild CR) retarded lipofuscin accumulation in C57BL/6 mouse brain — threshold/dose-dependent effect. Wiki previously said “at least one controlled study” without capturing this nuance.
Greenland shark supersession note added to Limitations: PMID 42024652 (Aging Cell 2026) — massive lipofuscin in cardiomyocytes coexists with preserved cardiac function in 300-year-old sharks, directly challenging the causal-driver framing. Added contradictory-evidence.
[^brunk2002axis] OA label corrected: Changed “OA bronze” to “download failed” since both download attempts returned no valid URL.

Supersession search: ran PubMed 2020–2026 for lipofuscin+aging+mechanism (127 results). No single meta-analysis or large RCT supersedes the Brunk-Terman model. PMID 32298788 (Prog Retin Eye Res 2020) extends the model in AMD-specific context (adds mitophagy/NFE2L2/PGC-1α/NLRP3 nodes). PMID 42024652 (Aging Cell 2026) challenges the pathological-consequence claim (flagged).

literature-checked-through: set to 2026-05-09.

[2026-05-09] ingest + propagate — bempedoic acid / ezetimibe / mevalonate / MTP (batch 4, 4 pages)

Trigger: continuation of batches 1–3. Batch 4 closes the most-load-bearing remaining stubs: bempedoic acid (statin-intolerant alternative), ezetimibe (LDL-lowering add-on), mevalonate pathway (statin pleiotropy framework + downstream of HMGCR), MTP (ApoB-particle assembly factor + lomitapide target).

Pages added (4, all verified: true partial-scope):

molecules/compounds/bempedoic-acid.md (~229 lines) — ATP-citrate lyase inhibitor; liver-specific bioactivation by ACSVL1 (skeletal-muscle-absent → no SAMS); CLEAR Outcomes 2023 NEJM HR 0.87 (95% CI 0.79–0.96, p=0.004); Pinkosky 2016 mechanism; Banach 2020 LDL-C −17.8%; Hamayal 2024 confirmatory meta-analysis; PubChem CID 10472693 (seeder corrected from wrong CID 24779522 in original brief).
molecules/compounds/ezetimibe.md (~? lines) — NPC1L1 internalization-block (Ge 2008 Cell Metab); IMPROVE-IT 2015 NEJM (n=18,144, HR 0.936 [0.89–0.99] at 7 yr) first non-statin LDL-lowering CV-outcomes RCT; SHARP 2011 renal cohort; Bach 2019 elderly subgroup; Ez-PAVE 2026 NEJM intensive LDL targeting.
pathways/mevalonate-pathway.md (~196 lines) — KEGG hsa00900 / Reactome R-HSA-191273. 9-step pathway from acetyl-CoA → IPP → branch points (FPP→cholesterol vs FPP→GGPP→prenylation, dolichol, ubiquinone). Statin pleiotropy framework (5 mechanisms: cholesterol, GGPP/prenylation, CoQ10/SAMS, dolichol, selenoprotein/T2D). Drug classes: statins (HMGCR), bempedoic acid (ACLY), bisphosphonates (FDPS).
molecules/proteins/mtp.md (~? lines) — Heterodimer (MTTP large subunit + PDI/P4HB); ER lumen lipid-loading onto nascent ApoB; abetalipoproteinemia (homozygous LOF); lomitapide target. Cuchel 2013 Lancet Phase 3 (n=29 HoFH, mean LDL-C −50%, hepatic steatosis dose-limiting).

Verification corrections (notable across all 4):

bempedoic-acid.md (5): CRITICAL: tendon rupture is a “Warning and Precaution,” NOT a boxed warning — the seeder mischaracterized FDA labeling. Boxed warnings are a much stronger regulatory signal; this matters for clinical context. AMPK mechanism corrected — ETC-1002-CoA directly activates AMPK β1 in an energy-charge-INDEPENDENT manner per Pinkosky 2016 (not via AMP/ATP ratios). Banach 2020 LDL-C −17.4% → −17.8%. Gout incidence quantified (3.2% vs 2.2%). Primary prevention HR 0.70 (95% CI 0.55–0.89) with concrete numerics added.
ezetimibe.md (7): CRITICAL: Bach 2019 elderly subgroup n=1,260 → 2,798 (off by ~2.2×; the ≥75 subgroup of IMPROVE-IT comprised 2,798 patients per the actual paper). Zhou 2026 DOI fabricated → corrected (10.1186/s12872-026-04250-x → 10.1186/s12872-026-05541-1). Mahmoud 2025 safety mischaracterized — seeder said “comparable” but paper found significantly higher severe AEs with combination (p=0.03). SHARP percentage LDL reduction not in abstract (only absolute 0.85 mmol/L); corrected. Ez-PAVE 2026 DOI confirmed real via Crossref + PMID 41910315.
mevalonate-pathway.md (7): CRITICAL: Dohlmann 2018 study design fabricated — seeder claimed “RCT, n=50, 400 mg × 8 wk” but the 2018 paper is cross-sectional observational (n=84). The 400 mg × 8 wk RCT is the Dohlmann 2022 Antioxidants paper (n=37, the muscle-biopsy primary outcome study). Two papers conflated. Both now correctly distinguished. (Note: coq10.md correctly cites Dohlmann 2022 — no propagation needed there.) Sever 2003 E3-ligase attribution removed (Hrd1/RNF145 not in Sever 2003). HMGCR half-life claim corrected. Pinkosky 2016 AMPK mechanism corrected (LDL-lowering is AMPK-independent, ACLY-driven). Ference 2019 specifics added (OR 0.823 per 10 mg/dL LDL-C reduction).
mtp.md (7): “Median LDL-C reduction” → “Mean LDL-C reduction” (Cuchel 2013 reports means ± SD throughout; primary endpoint is mean −50%, 95% CI −62, −39, p<0.001). ApoB reduction quantified (−49%, p<0.0001). MTP large subunit MW ~97 → ~99 kDa (UniProt P55157 sequence mass = 99,351 Da). Trial duration: 78 weeks total (26-week efficacy + 52-week safety, was just “26 weeks”). Hepatic steatosis NMRS data quantified (mean liver fat 1.0% → 8.6% at week 26 → 8.3% at week 78, n=20 evaluable). Transaminase AE rates quantified (10/29 ALT/AST >3× ULN; 4/29 >5× ULN).

Propagation pass:

apob.md: cleaned [[mtp]] (does not exist; forward reference) annotation (mtp now seeded); added pointer to lomitapide-target context.
coq10.md: updated [[mevalonate-pathway]] (stub — page does not yet exist) annotation to active link with FPP-derived ubiquinone framing.
lipoprotein-metabolism.md: cleaned [[bempedoic-acid]] (stub) and [[molecules/compounds/statins]] (stub) annotations; therapeutic-targeting table now has direct entity-page links + condensed mechanism summaries.
hmgcr.md: cleaned [[mevalonate-pathway]] forward-ref, no page yet annotation; added FPP/GGPP branch-point context.
phenotypes/atherosclerosis.md: ezetimibe + bempedoic acid sections now linked to entity pages; IMPROVE-IT and CLEAR Outcomes effect sizes inline.
familial-hypercholesterolemia.md: HeFH treatment ladder all wikilinked (statins/ezetimibe/bempedoic-acid); HoFH lomitapide line linked to mtp with Cuchel 2013 numerics.
interventions/pharmacological/statins.md: opening paragraph extended with mevalonate-pathway + ezetimibe + bempedoic-acid adjacent-class context.
No propagation needed: PCSK9-AD claim was localized to hmgcr.md (corrected in batch 2); Dohlmann 2018-vs-2022 distinction correctly handled in coq10.md (no propagation needed); Bach 2019 n=2,798 only on ezetimibe.md (no other pages cite it).

Mechanistic chain coverage as of end-of-batch-4 (16 wiki-resident entities):

Dietary fat (palmitic-acid vs PUFA; mediterranean-diet) → hepatic srebp-2 sensing → ldlr surface density (post-translationally regulated by pcsk9; pharmacologically by hmgcr inhibition + SREBP-2 feedback via statins / bempedoic-acid / ezetimibe) → plasma apob-particle clearance + Lp(a) residence + apoe-mediated remnant clearance + LPL-mediated TG hydrolysis + mtp-mediated ApoB-particle assembly → atherosclerosis → CV events. mevalonate-pathway is the upstream cholesterol-biosynthesis framework (statin pleiotropy lives here). familial-hypercholesterolemia is the Mendelian-extreme endpoint. lipoprotein-metabolism is the integrated pathway view.

Coverage: 13 atomic protein/compound pages + 1 phenotype + 2 pathways + 2 interventions = 18 wiki-resident entities for the dietary-fat → CV-event chain.

Stubs surfaced for future seeding: abca1 (Tangier disease, RCT), apoa1 (HDL anchor), cetp (HDL→VLDL exchange), lcat (HDL maturation), apoc3 (LPL inhibitor; volanesorsen target), apoc2 (LPL activator), angptl3 (evinacumab target), gpihbp1 (LPL transporter; FCS variant), lomitapide (compound page; MTP-inhibitor monograph), abetalipoproteinemia (phenotype), npc1l1 (ezetimibe target protein), acly (bempedoic target protein), lrp1 (chylomicron remnant receptor), insig-1/insig-2 (SCAP retention), fdps/fdft1 (mevalonate branch enzymes), bisphosphonates (FDPS-targeting class).

[2026-05-09] verify — pathways/mevalonate-pathway.md

Sources checked:

Sever 2003 Mol Cell (doi:10.1016/s1097-2765(02)00822-5) — local PDF verified
Peterson 2011 Cell (doi:10.1016/j.cell.2011.06.034) — local PDF verified; cross-checked against verified srebp-2.md
Pinkosky 2016 Nat Commun (doi:10.1038/ncomms13457) — local PDF verified
Ference 2019 NEJM (doi:10.1056/NEJMoa1806747) — local PDF verified (freshly downloaded)
Dohlmann 2018 J Clin Endocrinol Metab (doi:10.1210/jc.2018-01185) — PDF download failed; verified via PubMed abstract (PMID 30299473)
Dohlmann 2022 Antioxidants (doi:10.3390/antiox11091698) — new footnote added; verified via PubMed abstract (PMID 36139772)
Sattar 2010 Lancet — closed-access; values retained unverified
KEGG hsa00900 (“Terpenoid backbone biosynthesis”), Reactome R-HSA-191273 (“Cholesterol biosynthesis”) — confirmed via REST APIs

Corrections (6):

MAJOR — Dohlmann 2018 study design: Wiki had “double-blind RCT, n=50 (30 statin / 20 controls), 400 mg CoQ10 × 8 wk.” Paper is cross-sectional observational, n=84 (64 statin + 20 controls), no CoQ10 supplementation arm. Corrected. Added [^dohlmann2022] for the actual CoQ10 RCT (2022 Antioxidants, n=37, 400mg × 8wk).
Sever 2003 E3-ligase attribution: “recruiting E3 ligases Hrd1 and RNF145” removed — Sever 2003 does not identify these ligases. Half-life numbers corrected (~1h with INSIG-1+sterols vs ~4–6h).
Sever 2003 cell lines: “CHO + HEK293” → “CHO-K1 + HEK-293S.”
Pinkosky 2016 AMPK: “AMPK activation secondary” → ETC-1002-CoA directly activates AMPK β1 independently of ACLY inhibition; LDL-C lowering is AMPK-independent.
Peterson 2011 footnote: Corrected cell lines; no direct lipin-1–SREBP physical interaction noted; S106/S472 confirmed correct.
Ference 2019 footnote: Added OR 0.823 (95% CI 0.78–0.87, P=4.0×10⁻¹⁴).

Downstream propagation needed: coq10.md may cite Dohlmann 2018 with wrong study design — check before next lint pass.

[2026-05-09] verify — molecules/compounds/ezetimibe.md

Page: molecules/compounds/ezetimibe.md

Sources checked:

IMPROVE-IT 2015 (doi:10.1056/NEJMoa1410489) — PDF verified end-to-end (local archive, 570 KB)
Ge 2008 (doi:10.1016/j.cmet.2008.04.001) — PDF verified (local archive, pages 1-3)
Altmann 2004 (doi:10.1126/science.1093131) — abstract-only (PMID 14976318); not_oa
SHARP 2011 (doi:10.1016/s0140-6736(11)60739-3) — abstract-only (PMID 21663949); not_oa
Bach 2019 (doi:10.1001/jamacardio.2019.2306) — abstract-only (PMID 31314050); download failed
Ez-PAVE 2026 (doi:10.1056/NEJMoa2600283) — DOI confirmed via Crossref; PMID 41910315 abstract verified
Zhou 2026 (PMID 41555252) — abstract verified
Mahmoud 2025 (PMID 40557229) — abstract verified
PubChem API: CID 150311, C24H21F2NO3, MW 409.4, InChIKey OLNTVTPDXPETLC-XPWALMASSA-N — all confirmed

Corrections made (7):

CRITICAL — Bach 2019 n: 1,260 → 2,798 (15.4% of 18,144 total IMPROVE-IT enrollees). Previous figure was ~55% of actual number — substantive error.
Bach 2019 p-value clarified: Added “p=0.02 for interaction” qualifier; added 8.7 pp ARR figure for ≥75 subgroup.
Zhou 2026 DOI corrected: 10.1186/s12872-026-04250-x → 10.1186/s12872-026-05541-1 (seeder fabricated the DOI; confirmed via Crossref).
Mahmoud 2025 safety mischaracterization corrected: Seeder stated “safety comparable” — paper found severe adverse events were significantly higher with ezetimibe+statin combination (p=0.03). Body text and footnote updated to reflect this signal.
SHARP LDL reduction claim corrected: “~34% from baseline” → “average LDL-C difference of 0.85 mmol/L (~33 mg/dL)” as stated in the abstract; the percentage was not source-stated.
Ez-PAVE DOI status: Seeder flagged DOI as unconfirmed — DOI is real and resolves correctly. Footnote note updated; added achieved median LDL values (56 vs 66 mg/dL).
IMPROVE-IT footnote expanded: Added arm-level n (9,077 / 9,067), explicit median follow-up vs 7-year KM estimate distinction, secondary endpoint HRs (MI 0.87, ischemic stroke 0.79), CV death HR 1.00.

Unverifiable claims: SHARP LDL percentage from baseline (not stated in abstract; full PDF not_oa). Altmann 2004 group-level n (not in abstract; not_oa). Mahmoud 2025 specific adverse event category driving p=0.03 (not in abstract).

Supersession candidates: Recency search 2023–2026 (PubMed eutils). J Clin Lipidol 2026 meta-analysis (PMID 41775617; n=74,466) confirms 30% RRR per 1 mmol/L LDL reduction including ezetimibe as agent — consistent with existing framing. No supersession of IMPROVE-IT benefit claim found. Ez-PAVE 2026 is already incorporated. literature-checked-through: 2026-05-09.

Downstream propagation needed: molecules/proteins/apob.md and molecules/compounds/statins.md should be checked — both are noted in the seeder brief as having partially verified IMPROVE-IT figures. If they cite n=1,260 for the Bach 2019 elderly subgroup, correct to n=2,798.

Final state: verified: true

[2026-05-09] verify — molecules/compounds/bempedoic-acid.md

Page: molecules/compounds/bempedoic-acid.md Sources checked: Pinkosky 2016 (10.1038/ncomms13457, PDF verified), Banach 2020 (10.1001/jamacardio.2020.2314, PDF verified), Nissen 2023 JAMA primary prevention (10.1001/jama.2023.9696, PDF verified), Hamayal 2024 (10.1080/14796678.2024.2388478, PDF verified), CLEAR Outcomes NEJM (10.1056/NEJMoa2215024, not_oa — headline figures cross-verified via FDA openFDA API + secondary sources), PubChem API (CID 10472693), ChEMBL API (CHEMBL3545313).

Corrections made (6):

CRITICAL — Boxed warning → Warning and Precaution: Wiki incorrectly stated an “FDA boxed warning” for tendon rupture. FDA prescribing information (openFDA) confirms Warning and Precaution only. Added tendon rupture incidence by population.
AMPK mechanism corrected: ”↑ AMP/ATP ratio → AMP-dependent activation” → ETC-1002-CoA activates AMPKα1β1γ1 by direct allosteric mechanism independent of cellular energy charge (Pinkosky 2016 Fig. 1e/f).
Banach 2020 LDL-C −17.4% → −17.8%: Primary efficacy figure for ASCVD/HeFH pool corrected from rounded Key Points value to primary results figure (95% CI −19.5% to −16.0%). Table updated to show both pools.
Gout incidence corrected: “~3–4% vs ~2%” → “3.2% vs 2.2% (CLEAR Outcomes per FDA label)”; earlier Phase 3 data 1.4% vs 0.4% cited from Banach 2020.
Primary prevention subgroup expanded: Added HR 0.70 (95% CI 0.55–0.89; p=0.002); LDL-C −30.2 mg/dL (−21.3%); n per arm; noted this is numerically stronger than overall trial HR 0.87.
Pravastatin AUC: ~1.5-fold → ~2-fold per FDA label.

Supersession candidates: None — 23 post-CLEAR papers reviewed; all consistent with existing framing. literature-checked-through: 2026-05-09.

[2026-05-09] verify — molecules/proteins/mtp.md

Sources checked:

Cuchel 2013 Lancet (doi:10.1016/s0140-6736(12)61731-0) — downloaded + verified against full PDF (15 pages)
UniProt P55157 (MTP large subunit) — confirmed via REST API: 894 aa, 99,351 Da, HGNC:7467, NCBI Gene 4547, ENSG00000138823, locus 4q23
UniProt P07237 (PDI / P4HB) — confirmed via REST API: 508 aa, 57,116 Da
Wetterau 1992 Science (doi:10.1126/science.1439810) — not_oa; metadata confirmed via Crossref
Sharp 1993 Nature (doi:10.1038/365065a0) — not_oa; metadata confirmed via Crossref (cloning + ABL mutation analysis)
Wetterau 1985 Chem Phys Lipids (doi:10.1016/0009-3084(85)90068-4) — not_oa; metadata confirmed via Crossref
Wetterau 1990 JBC (doi:10.1016/s0021-9258(19)38742-3) — not_oa; metadata confirmed via Crossref (PDI component identification)
Narcisi 1995 Am J Hum Genet (PMID:8533758) — confirmed via PubMed efetch; title + journal + authors verified; DOI not resolvable via Crossref
Ricci 1995 JBC (doi:10.1074/jbc.270.24.14281) — download failed (HTTP 403); metadata confirmed via archive

Corrections (6):

LDL-C reduction characterization (CRITICAL): “Median LDL-C reduction: −50%” → “Mean LDL-C reduction: −50% (95% CI −62, −39; p<0.001)”. The paper reports mean ± SD values throughout (Table 1); there is no median reported for the primary endpoint. The abstract reports “reduced by 50%” consistent with mean, not median.
ApoB reduction at week 26: “~50% from baseline” → “−49% from baseline at week 26 (p<0.0001)” per Table 1 of Cuchel 2013.
MTP large-subunit MW: “~97 kDa” → “~99 kDa” in both the Identity section and the heterodimer table. UniProt P55157 sequence mass = 99,351 Da.
Trial duration and completers added: Added “78-week total treatment” and “23/29 completed” to the efficacy section (from Cuchel 2013 Methods + Results).
Hepatic steatosis data added: Added NMRS figures (baseline 1.0% → week 26 8.6%, stable to week 78 8.3%; n=20 evaluable) from Cuchel 2013 Results/Discussion to the adverse effects paragraph.
Transaminase AE detail corrected: Vague “ALT/AST elevations common” → precise “10/29 >3× ULN; 4/29 >5× ULN; no permanent discontinuations” from Cuchel 2013.
Gap section: trial duration: “~26 weeks active treatment + follow-up” → “78 weeks total (26-week efficacy + 52-week safety)” with note that fibrosis endpoint was not assessed.

Unverifiable claims (closed-access PDFs):

Wetterau 1985: lipid-transfer substrate hierarchy (TG > CE >> PL) — claimed based on paper title + canonical knowledge; full-text not accessible. no-fulltext-access on footnote.
Wetterau 1990: PDI obligate heterodimer claim — abstract/title confirm PDI component but full mechanistic detail unverified. no-fulltext-access on footnote.
Wetterau 1992: “absence of MTP in intestinal biopsies” claim — Crossref confirms paper exists with matching title; full methodology unverified.
Sharp 1993: MTTP cloning + ABL mutation analysis — Crossref title confirms subject matter; exact mutation data unverified.
Ricci 1995: “30-aa truncation disrupts PDI interaction” — archive confirms paper exists; PDF download failed (403). no-fulltext-access added to footnote.
Narcisi 1995: “over 30 distinct MTTP mutations” claim — PMID confirmed, abstract confirms mutation characterization but count not verifiable from abstract alone.

Supersession candidates (R25): None materially superseding Cuchel 2013 for lomitapide HoFH efficacy data. Four recent RCTs/meta-analyses identified (2022–2026) — all are for different agents (obicetrapib/CETP, plozasiran/APOC3, alirocumab/PCSK9i, lerodalcibep/PCSK9i) and do NOT compete with or supersede the lomitapide-HoFH evidence base. Lomitapide remains the only FDA-approved direct MTP inhibitor and no newer Phase 3 lomitapide trial exists. Literature-checked-through: 2026-05-09.

Final verified state: verified: true (partial scope — Cuchel 2013 fully verified; five closed-access foundational papers verified by metadata only; Ricci 1995 PDF inaccessible).

[2026-05-09] ingest + propagate — Lp(a)/statins/ApoE/LPL (batch 3, 4 pages)

Trigger: continuation of batches 1–2. Batch 3 picks the highest-leverage stubs prioritizing clinical relevance and chain completeness.

Pages added (4, all verified: true partial-scope):

molecules/proteins/lpa.md (~231 lines) — LPA/apo(a)/Lp(a). KIV-2 CNV heritability (61–69% Eu / 19–44% Af per Schmidt 2016). MR causality (Burgess 2018; Björnson 2024 6.6× per-particle). Pelacarsen Lp(a)HORIZON, olpasiran OCEAN(a)-Outcomes, muvalaplin oral Phase 2.
interventions/pharmacological/statins.md — class page, all 7 approved statins; CTT 2010 RR 0.78 (95% CI 0.76–0.80) per mmol/L; ACC/AHA 2018 intensity tiers; CTT 2024 T2D update; SAMSON N-of-1 nocebo finding; cerivastatin withdrawal context; ClinicalTrials.gov 180 active.
molecules/proteins/apoe.md (~218 lines) — Three-allele biology (ε2 7%, ε3 78%, ε4 14%); Farrer 1997 ε4 AD OR 3.2/14.9; Schächter 1994 centenarian ε2 enrichment; Montagne 2020 BBB sPDGFRβ; Shi 2017 tau-independent neurodegeneration; Bejaoui 2025 epigenetic-aging.
molecules/proteins/lpl.md (~231 lines) — Capillary-endothelial TG hydrolase (NOT plasma-secreted; GPIHBP1-anchored). ApoC-III/ANGPTL3 inhibitors. REDUCE-IT vs STRENGTH discordance preserved unresolved. Olezarsen BRIDGE-TIMI 73a, plozasiran PALISADE Phase 3. FCS treatment landscape.

Verification corrections (notable):

lpa.md (8): CRITICAL: Burgess 2018 cohort — wiki said “n=431,167 UK Biobank” but Burgess 2018 actually used CHD Exome+ Consortium (n=48,333 IPD); UK Biobank NOT used. Kamstrup 2009 instruments — only KIV-2 CNV (NOT rs10455872). Kamstrup 2009 statistic HR not OR. CRITICAL: ApoB “overcounts → underestimates” — Lp(a) inclusion in ApoB underestimates total atherogenic risk (Lp(a) particles are ~6.6× more atherogenic; logically inverted in seed). Wu 2026 PMID corrected.
statins.md (5): CRITICAL: CTT 2010 CI corrected — wiki had “0.74–0.82” (older 2005 14-trial CTT figure); correct 2010 26-trial figure is RR 0.78 (95% CI 0.76–0.80, p<0.0001). IMPROVE-IT DOI added (Cannon 2015). SAMSON DOI corrected (Wood 2020). ClinicalTrials.gov active count 9→180 (off by 20×).
apoe.md (6): Schächter 1994 n 338→325 genotyped; Montagne 2020 cohort 245→435 with proper sub-counts; CSF marker name PDGFR-β → sPDGFRβ; Bejaoui 2025 cohort distinguished (88 with methylation, not 14,669 screening); ε2/ε2 amyloid resilience claim flagged unsourced (Christchurch was APOE3 Christchurch, NOT ε2/ε2).
lpl.md (8): CRITICAL: STRENGTH citation — wiki cited secondary analysis for HR 0.99; corrected to primary JAMA 2020 paper. CRITICAL: APOC3 LOF DOI wrong (resolved to correction notice; corrected to NEJMoa1307095). CRITICAL: APOC3 LOF design mislabeled “mendelian-randomization” → exome-sequencing/LOF-variant association. CRITICAL: LPL Ser447Stop CAD claim — Xie 2017 reports OR 1.04 (p=0.90, I²=87%) per allele, NOT the seeder’s “~10% lower CHD risk.” PALISADE dosing Q6M→Q3M; volanesorsen thrombocytopenia rates corrected.

Propagation pass:

phenotypes/atherosclerosis.md: mechanistic chain extended to lpa (parallel high-atherogenicity particle), apoe (chylomicron remnant clearance + AD risk), lpl (capillary TG hydrolysis), statins; APOE row in non-modifiable risks linked; Lp(a) biomarker section linked + Björnson 2024 6.6× claim added; LDLR^-/-^/ApoE^-/-^ mouse-model footnote linked; statins section linked with CTT 2024 T2D framing.
Forward-refs already in place in apob.md, lipoprotein-metabolism.md, trem2.md (apoe-TREM2 ligand) all now resolve.

Mechanistic chain coverage as of end-of-batch-3:

Dietary fat (palmitic-acid vs PUFA; mediterranean-diet) → hepatic srebp-2 sensing → ldlr surface density (post-translationally by pcsk9; pharmacologically by hmgcr inhibition + SREBP-2 feedback via statins) → plasma apob-particle clearance + Lp(a) residence + apoe-mediated remnant clearance + LPL-mediated TG hydrolysis → atherosclerosis → CV events. familial-hypercholesterolemia is the Mendelian-extreme endpoint; lipoprotein-metabolism is the integrated pathway view.

End-to-end coverage now: 12 atomic protein/compound pages + 1 phenotype + 1 pathway + 2 interventions (Mediterranean diet, statins). The dietary-fat → CV-event chain is fully wiki-resident.

Stubs for future seeding: mtp, abca1, apoa1, cetp, lcat, apoc3, apoc2, angptl3, gpihbp1, mevalonate-pathway, bempedoic-acid, ezetimibe, fibrates, icosapent-ethyl, lrp1, insig-1/insig-2.

[2026-05-09] verify — molecules/proteins/lpl.md

Sources checked:

UniProt P06858 — confirmed via REST API: HGNC 6677, NCBI Gene 4023, Ensembl ENSG00000175445, signal peptide 1–27, mature chain 28–475 (448 aa), catalytic triad Ser159/Asp183/His268, EC 3.1.1.34 + 3.1.1.32
NCBI Gene 4023 — chromosomal location 8p21.3 confirmed
REDUCE-IT (doi:10.1056/NEJMoa1812792; PMID 30415628) — closed-access; verified via PubMed abstract: n=8,179, HR 0.75 (0.68–0.83), p<0.001, event rates 17.2% vs 22.0%
STRENGTH primary (doi:10.1001/jama.2020.22258; PMID 33190147) — verified via PubMed abstract: n=13,078, HR 0.99 (0.90–1.09), p=0.84, baseline TG ~240 mg/dL
STRENGTH secondary analysis (doi:10.1001/jamacardio.2021.1157) — PDF verified; n=13,078 enrolled (10,382 with EPA/DHA levels); HR for top EPA tertile 0.98 (0.83–1.16), p=0.81
BRIDGE-TIMI 73a / olezarsen (doi:10.1056/NEJMoa2402309; PMID 38587249) — closed-access; verified via PubMed abstract: n=154, TG reductions 49.3 pp (50 mg) and 53.1 pp (80 mg) vs placebo
PALISADE / plozasiran (doi:10.1056/NEJMoa2409368; PMID 39225259) — closed-access; verified via PubMed abstract: n=75, dosing Q3M (not Q6M), TG −80% (25 mg) and −78% (50 mg) vs −17% placebo, pancreatitis OR 0.17
APOC3 LOF primary (doi:10.1056/NEJMoa1307095) — PDF verified: n=110,970 for CHD analysis (not ~75,000), exome-seq + LOF association (not MR), TG −39% combined, CHD OR 0.60 (0.47–0.75)
APPROACH / volanesorsen (PMID 31390500) — PubMed abstract: n=66, TG −77%, thrombocytopenia 45% with plt 50–100K, 6% severe (25K), 0% below 50K after enhanced monitoring
Xie 2017 meta-analysis (doi:10.3390/ijerph14010084) — PDF verified: Ser447X X allele carrier model OR 1.04 (ns); XX homozygote associated with increased (not decreased) CAD risk OR 2.37 — does NOT support per-allele cardioprotection
Young 2011 GPIHBP1 review (doi:10.1194/jlr.r018689) — PDF verified: GPIHBP1 expressed exclusively in capillary endothelial cells; LPL transported from interstitial space across endothelium via GPIHBP1
EMA Waylivra entry — confirmed conditional approval 2019 for FCS
ClinicalTrials.gov v2 — olezarsen: 3 active trials; plozasiran: 3 active trials

Corrections (8):

STRENGTH footnote citation (CRITICAL): seeder cited doi:10.1001/jamacardio.2021.1157 (secondary analysis paper) for HR 0.99 (0.90–1.09); corrected to doi:10.1001/jama.2020.22258 (primary JAMA 2020 paper). Secondary analysis reference preserved in footnote note-field.
BRIDGE-TIMI 73a TG reductions: “50 mg reduced TG 53%” and “80 mg reduced TG 60%” → “49.3 percentage points (50 mg) and 53.1 percentage points (80 mg) vs placebo” per PMID 38587249.
PALISADE dosing interval: “Q6M” → “Q3M (every 3 months)” per PMID 39225259; also added 50 mg arm data (−78%) and pancreatitis OR.
APOC3 LOF footnote DOI (CRITICAL): doi:10.1056/nejmx140063 (a correction notice) → doi:10.1056/NEJMoa1307095 (the actual primary paper).
APOC3 LOF footnote study design: “mendelian-randomization” → “exome-sequencing + LOF-variant association study (NOT MR)”.
APOC3 LOF footnote n: “~75,000” → “110,970 for CHD analysis (34,002 cases + 76,968 controls)”.
Ser447X CAD-risk claim (CRITICAL): wiki claimed Xie 2017 supports “~10% lower CHD risk per allele” — Xie 2017 does NOT support this; X allele model is null (OR 1.04, p=0.90); only XX homozygote shows increased CAD risk (OR 2.37). Claim rewritten; per-allele benefit attributed to GWAS consortia data (different source). Gap tag added.
Volanesorsen thrombocytopenia rate: “platelet counts fell below 50×10⁹/L in ~25%” → accurate APPROACH trial data: 45% had plt 50–100K, 6% had severe (~25K), 0% below 50K after enhanced monitoring per PMID 31390500.

Unverifiable claims (closed-access abstracts only):

REDUCE-IT median baseline TG ~216 mg/dL: not in abstract; flagged in body with “(per full paper — not confirmable from abstract alone)”
REDUCE-IT LDL-C increase in mineral oil placebo arm (~11 mg/dL): not verifiable from abstract

Supersession candidates (R25): None found. The PALISADE (plozasiran) and BRIDGE-TIMI 73a (olezarsen) Phase 2/3 data are the most recent for these specific agents. Ongoing Phase 3 trials (SHASTA-3, MUIR-3, CORE-OLE) have not yet reported CV outcomes. No post-PALISADE Phase 3 readouts identified through 2026-05-09. Literature-checked-through confirmed 2026-05-09.

Downstream propagation needed: None identified (lpl.md is not yet cited from other wiki pages per inbound-link count; propagation not applicable for initial verification pass).

[2026-05-09] verify — molecules/proteins/apoe.md

Sources checked:

Schächter 1994 (doi:10.1038/ng0194-29) — local PDF verified
Montagne 2020 (doi:10.1038/s41586-020-2247-3) — downloaded + verified (PMC7250000)
Shi 2017 (doi:10.1038/nature24016) — downloaded + verified (Duke repository)
Bejaoui 2025 (doi:10.1007/s11357-025-01548-2) — downloaded + verified
Farrer 1997 (doi:10.1001/jama.1997.03550160069041) — closed-access; OR values confirmed via PubMed abstract (PMID:9343467)
UniProt P02649 — confirmed via REST API; GenAge HAGR ID 138 confirmed

Corrections (7):

Schächter 1994 n (CRITICAL): body + footnote had “n=338 centenarians” — Table 1 shows APOE-typed centenarian group n=325 (338 = total enrolled; 325 = APOE-genotyped). Corrected body + footnote. Added allele frequencies from Table 1.
Montagne 2020 n + cohort: wiki said “n=245 cognitively normal” — enrolled n=435 (normal + mildly impaired); n=245 DCE-MRI; n=350 CSF sPDGFRβ. Corrected.
Montagne 2020 marker name: “PDGFR-β” → “sPDGFRβ” per paper.
Montagne 2020 mechanism: added CypA-MMP9 pathway detail from paper results.
Bejaoui 2025 cohort: clarified that n=88 had methylation arrays (not 14,669); Qatar BioBank origin noted; GrimAge p=0.043 added.
Unsourced ε2/ε2+amyloid+resilience claim: Christchurch case is APOE3 Christchurch (not ε2/ε2); claim corrected + unsourced tag added.
Farrer 1997 footnote: added PMID:9343467 + exact case/control counts.

Supersession check (R25): literature-checked-through: 2026-05-09; no superseding evidence found. Final state: verified: true

[2026-05-09] verify — molecules/proteins/lpa.md

Sources checked: Burgess 2018 (PDF, local), Schmidt 2016 (PDF, local), Lampsas 2023 (PDF, local), Bjornson 2024 (PDF, downloaded + read), Kamstrup 2009 (PubMed abstract, not_oa), Wu 2026 (PubMed abstract via PMID 41937094).

Corrections (8):

Burgess 2018 cohort + n (CRITICAL): wiki had “n=431,167 UK Biobank + CARDIoGRAMplusC4D” — paper used CHD Exome+ Consortium (n=48,333 individual participant data, 5 studies) + CARDIoGRAMplusC4D summarized data (62,240 CHD + 127,299 controls). UK Biobank was NOT used. Corrected body text and footnote.
Burgess 2018 CI added: 95% CI 71.0–137.0 for the 101.5 mg/dL estimate was in the paper but missing from wiki body; added.
Kamstrup 2009 n (MEDIUM): wiki footnote had “n=9,330 Copenhagen City Heart Study” — CCHS had n=8,637; full multi-cohort study n≈40,486. Corrected to multi-cohort total.
Kamstrup 2009 instruments (MEDIUM): wiki body listed “KIV-2 repeat size and rs10455872 SNP” as instruments — Kamstrup 2009 used only KIV-2 repeat-size polymorphism; rs10455872 was NOT an instrument in that paper. Corrected.
Kamstrup 2009 OR → HR: wiki said “odds ratio 1.22” — Kamstrup 2009 is a prospective cohort; correct statistic is HR (hazard ratio) 1.22. Corrected.
Bjornson 2024 n: wiki had “>430,000” — actual n=502,413 per paper. Corrected body and footnote. Added specific OR data (1.28 vs 1.04 per 50 nmol/L).
KIV-2 copy minimum: wiki had “2–40+ copies” — Schmidt 2016 states 1 to >40 copies. Corrected to “1 to >40”.
ApoB “overcounts” language: wiki said ApoB assay “overcounts atherogenic particle burden” — incorrect direction; ApoB blends Lp(a)-ApoB (more atherogenic) with LDL-ApoB equally, meaning it underestimates total atherogenic risk for high-Lp(a) individuals. Corrected to “underestimates.”

Additional corrections:

OxPL mechanism: MCP-1 is secreted by endothelial cells, not macrophages; IL-8 is from macrophages. Corrected attribution in mechanism paragraph.
Wu 2026 PMID: wiki had 41264184 (wrong, returns different paper) → corrected to 41937094. Added CIs to drug reduction percentages.
Bjornson 2024 footnote: updated “download pending” → PDF now available locally; added specific OR data.

Supersession check (R25): No Phase 3 cardiovascular outcomes trial results for Lp(a)HORIZON (pelacarsen) or OCEAN(a)-Outcomes (olpasiran) found as of 2026-05-09. Wiki statement “None has reported outcomes data as of 2026-05-09” confirmed accurate. literature-checked-through: 2026-05-09.

Downstream propagation needed: pathways/lipoprotein-metabolism.md cites Burgess 2018 n and may inherit the wrong UK Biobank cohort description — verify. apob.md may reference Lp(a) ApoB-overcounting language — verify direction.

[2026-05-09] ingest + propagate — lipid-lowering target triangle completed (batch 2, 3 pages)

Trigger: continuation of batch 1 (apob/palmitic-acid/srebp-2/ldlr/mediterranean-diet). Three pages flagged in batch 1 as missing-but-load-bearing: HMGCR (statin target), familial hypercholesterolemia (clinical Mendelian endpoint of the chain), lipoprotein-metabolism (canonical pathway integration).

Pages added (3, all verified: true partial-scope):

molecules/proteins/hmgcr.md (~221 lines) — rate-limiting cholesterol biosynthesis; statin target; mevalonate pathway branch point; SCAP/INSIG/ERAD; AMPK Ser872; CTT meta-analysis; Swerdlow 2015 HMGCR-MR T2D adverse-effect.
phenotypes/familial-hypercholesterolemia.md (~198 lines) — autosomal codominant LDL clearance disorder; HeFH 1:313 (Beheshti 2020); HoFH 1:160k–300k (EAS); Goldstein-Brown 5-class LDLR mutation schema; cumulative-LDL-exposure framework.
pathways/lipoprotein-metabolism.md (~224 lines) — KEGG hsa04979, Reactome R-HSA-174824. 5 sub-pathways (exogenous/dietary, endogenous/hepatic, RCT, Lp(a), FFA flux). Therapeutic-targeting overview integrating all approved + pipeline lipid drugs.

Verification corrections (substantial across all 3):

hmgcr.md (7): active-site residues confirmed Glu559/Lys691 via UniProt sequence (the seeder had over-corrected the brief’s training-memory residues to all-Asp, which was wrong); T2D OR 1.12 (95% CI 1.06–1.18) per Swerdlow 2015 (was “9–11%”); Ference 2015 n 108,376 (was ~48k); Benn 2017 n 111,194; CRITICAL: Benn 2017 PCSK9-AD claim was factually inverted — wiki had “PCSK9 shows AD signal” but Benn 2017 found BOTH HMGCR and PCSK9 MR null for AD; corrected with #gap/contradictory-evidence. CTT 2024 dose-intensity T2D update integrated.
familial-hypercholesterolemia.md (6): Beheshti 2020 scope corrected (104-study meta, was “44 studies”); HoFH 1:160k–300k attribution clarified (EAS via Abifadel 2023, not Tromp 2022 which says 1:300k); evinacumab citation upgraded from review to Raal 2020 NEJM ELIPSE-HoFH primary trial (n=65, between-group −49.0 pp, p<0.001); Santos 2014 reframed as three-group analysis.
lipoprotein-metabolism.md (7): TANDEM n=407, LDL-C 48.6%; CRITICAL: anacetrapib REVEAL re-classified — wiki had grouped all 4 CETP inhibitors as failures, but REVEAL (n=30,449) showed RR 0.91, p<0.004 (attributed to LDL-C, not HDL-C — Merck declined to submit); Schmidt 2016 KIV-2 variance corrected to population-specific (61–69% Europeans / 19–44% Africans, was “~50%”); ODYSSEY/FOURIER enriched.

Schema decision deferred: seeder added non-schema icd-10-hofh: E78.02 field on FH page to handle two ICD codes (E78.01 HeFH, E78.02 HoFH). Recommendation for user: formalize via icd-10: accepting list, or add icd-10-variants: field.

Propagation pass:

apob.md: cleaned “(forward reference — no page yet)” annotations; added pointer to lipoprotein-metabolism in Pathway context.
pcsk9.md: pathways: ["[[lipoprotein-metabolism]]"] added; FH GOF section now references familial-hypercholesterolemia for genetic-architecture context.
ldlr.md: cleaned (stub; forward-ref) and #stub annotations on 5 newly-resolved wikilinks; added lipoprotein-metabolism to cross-references.
srebp-2.md: added familial-hypercholesterolemia and lipoprotein-metabolism to cross-references.
phenotypes/atherosclerosis.md: mechanistic chain in opening paragraph extended to hmgcr + lipoprotein-metabolism + familial-hypercholesterolemia; statins section enriched (CTT 22%/mmol/L, T2D OR 1.12, geranylgeranyl-PP pleiotropy); risk-factor table wikilinked.
PCSK9-AD propagation check: confirmed pcsk9.md does NOT contain the fabricated PCSK9-AD claim (it was localized to hmgcr.md and corrected in-place by verifier).

Stubs surfaced for future seeding: mtp (lomitapide target), lpa / apo(a) (highly atherogenic genetic CV-risk factor), abca1, lpl, apoa1, apoe, cetp, lcat, mevalonate-pathway, bempedoic-acid, angptl3, statins (class page), insig-1 / insig-2.

Mechanistic chain coverage as of end-of-batch-2: dietary fat (palmitic-acid vs PUFA; mediterranean-diet) → hepatic srebp-2 sensing → ldlr surface density (regulated post-translationally by pcsk9; pharmacologically by hmgcr inhibition + SREBP-2 feedback) → plasma apob-particle clearance → atherosclerosis → CV events. familial-hypercholesterolemia is the Mendelian-extreme endpoint. lipoprotein-metabolism is the integrated pathway view. End-to-end chain is now wiki-resident with 8 protein/compound atomic pages + 1 phenotype + 1 pathway + 1 intervention.

[2026-05-09] verify — lipoprotein-metabolism pathway page

Page verified: pathways/lipoprotein-metabolism.md

Sources checked (8 total):

doi:10.1056/NEJMoa1615664 (FOURIER) — local PDF verified
doi:10.1056/NEJMoa1801174 (ODYSSEY) — local PDF verified (downloaded during session)
doi:10.1093/eurheartj/ehab551 (Ginsberg EAS 2021) — local PDF verified (downloaded during session)
doi:10.1001/jamacardio.2018.1470 (Burgess MR 2018) — local PDF verified (downloaded during session)
doi:10.1194/jlr.R067314 (Schmidt Lp(a) 2016) — local PDF verified (downloaded during session)
doi:10.1161/CIRCRESAHA.117.311978 (Tall & Rader CETP 2018) — local PDF verified (downloaded during session)
doi:10.1016/s0140-6736(25)00721-4 (TANDEM 2025) — not_oa; n and LDL-C% confirmed via PMID 40347969 abstract
doi:10.1016/j.ecl.2022.02.008 (Feingold 2022) — not_oa; sub-pathway mechanics not independently re-verified

Corrections made (7):

TANDEM n: “n=not verified” → n=407 (PMID 40347969)
TANDEM LDL-C reduction: “~50%” → 48.6% (PMID 40347969)
Anacetrapib REVEAL characterization: reframed from “failed to reduce cardiovascular outcomes” to correctly note REVEAL showed significant reduction (rate ratio 0.91, p<0.004, n=30,449); Merck declined FDA submission due to adipose accumulation concerns
Schmidt 2016 KIV-2 variance claim: “explains ~50% of heritable variation” → population-specific: 61–69% in Europeans, 19–44% in Africans
FOURIER footnote: added group sizes (13,784/13,780), key secondary endpoint HR 0.80 (95% CI 0.73–0.88)
ODYSSEY footnote: corrected p<0.0001 → p<0.001; added 95% CI 0.78–0.93 and all-cause mortality HR with CI; updated download status
Lp(a) trial status: pelacarsen trial name corrected to Lp(a)HORIZON; olpasiran updated to Phase 3 OCEAN(a)-Outcomes; both described as ongoing as of 2026-05
All footnote download statuses updated to reflect newly available local PDFs

Supersession check (R25): No superseding meta-analyses or RCTs found. PCSK9 inhibitor evidence reinforced by 2023–2026 literature. literature-checked-through: 2026-05-09.

Unverifiable: Feingold 2022 sub-pathway mechanics (not_oa); Wang 2022 compounding analysis (not_oa); Brown & Goldstein 1986 (not_oa).

[2026-05-09] ingest + propagate — dietary-fat → ApoB → CV-event mechanistic chain (5-page batch)

Trigger: a coverage audit surfaced that the wiki had no canonical home for the upstream half of the dietary-fat → atherogenic-particle → atherosclerosis chain. Atherosclerosis itself, PCSK9, and senescence-driven plaque biology were well covered; ApoB, LDLR, SREBP-2, palmitic acid, and Mediterranean diet were not.

Pages added (5, all verified: true partial-scope after this batch):

molecules/proteins/apob.md — central apolipoprotein on every atherogenic particle; MR-causality hierarchy ApoB > non-HDL-C > LDL-C; ~200 lines
molecules/compounds/palmitic-acid.md — dominant dietary SFA; 7-hallmark mechanistic page (NLRP3 palmitoylation, ER stress, mTORC1, autophagy block, endothelial senescence, β-oxidation overload, SASP propagation); LDL/ApoB-raising via SREBP-2/LDLR
molecules/proteins/srebp-2.md — sterol-regulatory TF; SCAP-INSIG-S1P/S2P sensor; the dietary-fat → LDLR transcriptional node
molecules/proteins/ldlr.md — LDL receptor; clathrin-mediated endocytosis cycle; PCSK9 dual-route degradation; FH 5-class mutation schema
interventions/dietary/mediterranean-diet.md — RCT-validated dietary intervention; PREDIMED 2018 + CORDIOPREV 2022 + Lyon 1999; Bozack 2025 epigenetic-aging signature

Verification summary: all 5 verified end-to-end against primary sources where PDFs available. Substantial corrections — see per-page log entries below. Notable:

ldlr.md: ALL 7 domain residue ranges seeder-fabricated (TM helix off by 74 residues); Poirier 2009 mechanism inverted
srebp-2.md: Ser-372 was on SREBP-1c, not SREBP-2 (fabricated residue); PCSK9 not in Horton 2002 (PCSK9 discovered 2003)
apob.md: Richardson 2020 instrument count fabricated (“118” → actual 255 SNPs / 531 multivariable)
mediterranean-diet.md: ClinicalTrials.gov active-trial count off by ~80× (4 → 325); first author of CRP umbrella review wrong (Kasiri → Tran)
palmitic-acid.md: DHHC enzyme overspecification; “equivalent to NLRP3 KO” claim removed

Propagation pass:

phenotypes/atherosclerosis.md: added body-prose mechanistic chain anchoring apob/ldlr/srebp-2/palmitic-acid/mediterranean-diet; added dietary-fat-composition row to Modifiable risk-factor table; updated statins section to cite SREBP-2/LDLR upregulation as primary mechanism; added ApoB to biomarker list.
molecules/proteins/pcsk9.md: backlinks to apob/ldlr/srebp-2; cleaned [[atherosclerosis]] (stub) reference to active link; added palmitic-acid + mediterranean-diet in dietary-modulation framing.
hallmarks/chronic-inflammation.md: NLRP3 inflammasome section now cites palmitic acid as a saturated-fatty-acid DAMP with mechanistic detail; mediterranean-diet linked as dietary lever.
Other 5 hallmark pages (cellular-senescence, deregulated-nutrient-sensing, disabled-macroautophagy, loss-of-proteostasis, altered-intercellular-communication, mitochondrial-dysfunction): no manual edits needed — Dataview queries auto-include both new pages via frontmatter hallmarks: / target-hallmarks:.

Frameworks updates (made by seeders during initial drafting):

frameworks/intervention-classes.md: 5 new mechanism classes — pro-inflammatory, ldl-raising, lipotoxic, senescence-inducing (palmitic-acid seed), dietary-fat-composition (mediterranean-diet seed).

Stubs surfaced for future seeding (highest priority): hmgcr (statin target, completes the lipid-lowering triangle), familial-hypercholesterolemia (FH phenotype page), lipoprotein-metabolism (canonical pathway page).

[2026-05-09] verify — molecules/proteins/ldlr.md

Pages verified: 1

molecules/proteins/ldlr.md — corrections applied:
1. Domain architecture — all seven boundary residue ranges corrected against UniProt P01130 feature table (accessed 2026-05-09). Every single domain boundary in the seeder’s table was wrong. Corrected values: LBD class A repeats 25–313 (was 22–292); EGF-like 1 (EGF-A) 314–353 (was 293–332); EGF-like 2 (EGF-B) 354–393 (was 333–371); β-propeller/YWTD 397–658 (was 372–541); EGF-like 3 (EGF-C) 663–712 (was 542–581); O-glycosylation region 721–768 (was 582–714); transmembrane helix 789–810 (was 715–736); cytoplasmic/NPVY motif 823–828 (was 737–860, NPVY confirmed correct). Added note that residue numbering includes 21-aa signal peptide.
2. PCSK9 mechanism attribution corrected — Poirier 2009 (J Biol Chem) demonstrates that endogenous PCSK9 degrades LDLR primarily via the intracellular trans-Golgi→lysosome route, not the extracellular co-endocytosis route described in the wiki. CLC knockdown (disrupts Golgi trafficking) increased LDLR ~2.9-fold; Dynasore (blocks surface endocytosis) had no significant effect. The extracellular pathway requires 100 nM exogenous PCSK9 (supra-physiological). Body text rewritten to accurately reflect this; footnote updated accordingly.
3. HeFH prevalence updated — ~1:313 per the most recent meta-analysis (104 studies, 11M subjects, cited in Abifadel 2023 [ref. 21/22]); prior wiki value was “~1:250–300” (an older EAS Consensus Panel estimate). HoFH 1:160,000–300,000 confirmed.
4. Statin LDL-C reduction figures mis-cited — ~50%/~30% figures were cited to Rashid 2005 (a mouse PCSK9-KO study), which is the wrong paper. Tagged #gap/unsourced and flagged that the correct citation is a clinical statin meta-analysis (e.g., Cholesterol Treatment Trialists).
5. Mensink 2003 mechanistic claim flagged — Mensink 2003 is a clinical meta-analysis of dietary fat trials; the molecular mechanism (dietary palmitate → INSIG-SCAP-SREBP-2 → LDLR suppression) was attributed to it but is not present in a clinical dietary meta-analysis. Flagged #gap/needs-replication with note that a molecular primary source is needed.
6. Brown 1979 ~150 recycling cycles confirmed absent — read full PDF; no quantitative recycling-cycle count appears. The 1979 paper labels recycling as ”? recycling” (question mark) in Fig. 7, indicating it was hypothetical at the time. Gap tag already present; footnote updated.
7. FOURIER n/HR/CI/p verified against PDF — n=27,564, HR 0.85 (95% CI 0.79–0.92), p<0.001 for primary composite endpoint confirmed exactly.
8. Closed-access tags added — Etxebarria 2015, Defesche 2017, Reeskamp 2021, Rashid 2005, Mensink 2003 all tagged #gap/no-fulltext-access; their quantitative claims are unverified against full text.

Pages unverifiable (closed-access or download failure): Etxebarria 2015, Defesche 2017, Reeskamp 2021, Rashid 2005, Mensink 2003.

[2026-05-09] verify — phenotypes/familial-hypercholesterolemia.md

Pages verified: 1

phenotypes/familial-hypercholesterolemia.md — corrections applied:
1. Beheshti 2020 study scope corrected — the paper is a 104-study meta-analysis across 4 population categories; the seeder described it as “44 studies” throughout. 44 studies is the general-population subanalysis only. Prevalence figure (1:313, 95% CI 1:384–1:256) and n=10.9M confirmed via PubMed abstract.
2. HoFH prevalence attribution corrected — Wiki said “1:160,000–300,000 [^tromp-2022]” but Tromp 2022 PDF only states “~1 in 300,000.” The 1:160,000–300,000 range derives from the EAS Consensus Panel via Hardy-Weinberg calculation, cited in Abifadel 2023. Body text and footnote updated to correctly attribute the range and clarify Tromp 2022 uses ~1:300,000.
3. Evinacumab citation replaced — seeder cited ELIPSE HoFH trial data to Doggrell SA (Expert Opin Biol Ther 2021), a review. Replaced with primary source: Raal FJ et al. NEJM 2020 (doi:10.1056/NEJMoa2004215). Trial parameters verified: n=65, 2:1 randomization, 47.1% LDL-C reduction vs +1.9% placebo, between-group difference −49.0 pp (95% CI −65.0 to −33.1, p<0.001). Old doggrell-2021 footnote replaced with raal-2020 footnote.
4. Santos 2014 three-group description corrected — body text framed the comparison as “null vs Class 5,” which is not what Santos 2014 measures. Paper compares null (22.5%), defective (27.1%), and no identified mutation (47.4%); the 47.4% group is “no identified mutation” not “Class 5.” Body text rewritten with all three groups, sample sizes (n=40/59/57), and OR (9.07) confirmed via PubMed abstract.
5. Abifadel 2003 PDF path corrected — seeder footnote pointed to non-existent (stale local path) path; corrected to. S127R (family HC92) and F216L (family HC60) mutation assignments confirmed against full PDF.
6. Tromp 2022 archive status updated — footnote said “archive status: pending”; PDF downloaded during this verification pass. Path updated to. n=751, 9% ASCVD at diagnosis, median age first MACE 24.5 yr (non-HIC) vs 37.0 yr (HIC) confirmed against full PDF. Note: abstract uses 37.0 yr; results body text shows 35.0 yr for MACE-free survival figure — abstract headline used as primary reference.
7. Ibrahim 2024 footnote enriched — added n=90 FH + 45 controls, exact OR 3.66 (95% CI 1.62–8.27), and mean cumulative LDL-C values (181 ± 54 vs 105 ± 33 mmol/L·yr); confirmed 75 mmol/L·yr doubling claim via PubMed abstract.
8. schema note: seeder added non-schema icd-10-hofh: E78.02 field to handle FH’s two ICD-10 codes (E78.01 HeFH, E78.02 HoFH). Left in place per verification SOP — this is a schema decision for the user, not a verifier action.

Pages unverifiable (PDF download failed): Beheshti 2020 (green OA, fetch failed — verified via PubMed abstract); Ibrahim 2024 (hybrid OA, fetch failed — verified via PubMed abstract); Santos 2014 (pending, then failed — verified via PubMed abstract); Etxebarria 2015 (not downloaded); Sturm 2018 (not_oa).

Supersession check (R25): searched PubMed for FH meta-analyses/RCTs 2023–2026. No supersession identified for core prevalence or treatment claims. One 2025 pediatric FH meta-analysis (European Heart Journal Open) updates treatment-response data in children but does not supersede Beheshti 2020 prevalence estimate.

Downstream propagation needed: molecules/proteins/ldlr.md references HoFH prevalence (1:160,000–300,000) — the attribution clarification (Tromp 2022 = ~1:300,000; 1:160,000–300,000 from EAS Consensus Panel via Abifadel 2023) should be propagated there. Also: molecules/proteins/pcsk9.md may cite Abifadel 2003 — the pedigree labels (HC92=S127R, HC60=F216L) are now more precisely documented.

Supersession check (R25 — type: protein, not strictly required; checked given active drug development context):

PubMed search “LDL receptor LDLR therapy aging” 2024–2026: 7 results. No meta-analysis or large RCT supersedes the FOURIER trial (HR 0.85) or the core PCSK9-LDLR biology. PMID 40082904 (Jee 2025 MR study, Korean/Japanese GWAS, n=~2M) provides additional MR support for LDLR as a causal CAD target — consistent with existing wiki framing.
literature-checked-through: 2026-05-09

[2026-05-09] verify — molecules/proteins/apob.md

Pages verified: 1

molecules/proteins/apob.md — corrections applied:
1. Richardson 2020 instrument count fabricated: Wiki claimed “118 genetic instruments.” The paper used 255 SNPs for the ApoB GWAS and 531 combined SNPs in the multivariable MR. The “118” figure is not stated anywhere in the paper; it was fabricated. Corrected to 255 ApoB SNPs / 531 multivariable-MR SNPs with exact OR statistics (ApoB: 1.92; 95% CI 1.31–2.81; P=7.47×10⁻⁴).
2. Isoform length precision: Wiki stated “4,563 aa (ApoB-100)” in identity section without distinguishing precursor from mature protein. UniProt P04114 chain annotation: precursor = 4,563 aa (signal peptide aa 1–27); mature ApoB-100 = 4,536 aa (residues 28–4563); mature ApoB-48 = 2,152 aa (residues 28–2179). Corrected in identity section and isoform table.
3. **Ference 2024 ~3× claim tagged no-fulltext-access: Closed access (not_oa); cannot PDF-verify the quantitative claim.
4. **Mensink 2003 0.036 mmol/L coefficient tagged no-fulltext-access: Bronze OA; download failed (Cloudflare); specific LDL-C coefficient and ApoB-specific effects are not in the public abstract. Body text updated to distinguish what the abstract confirms vs. what is unverified.
5. Sniderman 2019 footnote updated: PMC7369156 cross-checked; “one ApoB per particle” stoichiometry confirmed from PMC text. Confirmed the review does NOT contain LDLR site B residue numbers — already correctly tagged unsourced in body. Download note corrected from “pending” to “archive failed (0 candidate URLs).”
6. Richardson 2020 n-range corrected: Wiki said “n=441,016”; the GWAS n varies by trait (393,193–441,016; ApoB in 439,214). Corrected to range.
7. Canonical identity fields re-verified: UniProt P04114, NCBI Gene 338, HGNC 603, Ensembl ENSG00000084674 all confirmed via live APIs.

Pages unverifiable (closed-access): Ference 2024 (not_oa) — no-fulltext-access added. Mensink 2003 (bronze OA download failure) — no-fulltext-access added.

literature-checked-through: 2026-05-09 — PubMed search “apolipoprotein B mendelian randomization” last 12 months returned 41 results; no meta-analysis or large RCT directly supersedes Richardson 2020’s core finding that ApoB is the predominant causal CHD lipid signal. Supersession candidate noted: PMID 40089107 (JLR 2025, MR study) raises ApoB-independent HDL subparticle risk profile — complementary rather than contradictory. PMID 41568673 (EHJ 2026, “risk-weighted apoB”) extends rather than supersedes by adding TRL and Lp(a) weighting to the ApoB metric. Neither overturns the Richardson 2020 causal-attribution claim.

[2026-05-09] verify — interventions/pharmacological/statins.md

Pages verified: 1

interventions/pharmacological/statins.md — corrections applied:
1. CTT 2010 CI corrected (CRITICAL) — footnote [^ctt2010] stated “RR ~0.78 (95% CI 0.74–0.82)”. PDF Figure 1 (all 26 trials combined) shows RR 0.78 (95% CI 0.76–0.80, p<0.0001). The CI 0.74–0.82 is from the earlier 2005 CTT analysis; the 2010 paper’s combined estimate is 0.76–0.80. Body text updated and footnote corrected. Also updated n to “169,138 (≈170,000)”.
2. IMPROVE-IT DOI confirmed; unsourced removed — doi:10.1056/NEJMoa1410489 (Cannon et al., NEJM 2015;372:2387–97, n=18,144, median 7-yr) confirmed via PubMed PMID 26039521 and Crossref. New [^improveit2015] footnote added.
3. SAMSON trial added with correct DOI — doi:10.1056/NEJMc2031173 (Wood et al. NEJM 2020;383:2182–84; PMID 33196154) confirmed. N-of-1 crossover, ~90% nocebo. New [^samson2020] footnote added; body text rewritten to correctly identify SAMSON and Kristiansen as separate concordant trials.
4. Kristiansen 2020 footnote corrected — published volume/issue 7(6):507–16 added; mean VAS difference 0.31 (95% CI −0.24 to 0.86) added; “71 completed” (of 77 randomized) noted.
5. ClinicalTrials active count updated — 9 → 180 (ClinicalTrials.gov v2 API, 2026-05-09).
6. CTT 2024 T2D data confirmed — low/mod RR 1.10 (1.04–1.16), high-intensity RR 1.36 (1.25–1.48), ~62% in top glycaemia quartile all confirmed against PDF.
7. Landmark RCT numerics confirmed via PubMed abstracts — 4S, WOSCOPS, CARE, HPS, PROVE-IT, TNT, JUPITER figures all match wiki exactly.

Pages unverifiable (no local PDF): Ference 2015 JACC (status: failed); landmark RCT PDFs (pending/not_oa — verified via abstracts only); SAMSON (pending); Kristiansen (pending); IMPROVE-IT (not checked).

Supersession candidates (R25): PubMed 2024–2026 search (307 hits) — no meta-analysis supersedes the core CTT 22%/mmol/L framing. 2026 elderly meta-analysis (Atherosclerosis, n=38,089) consistent with existing framing. literature-checked-through: 2026-05-09.

Downstream propagation needed: none — hmgcr.md and apob.md already have the correct CI 0.76–0.80; incorrect 0.74–0.82 was localized to statins.md footnote only.

[2026-05-09] verify — molecules/proteins/srebp-2.md

Pages verified: 1

molecules/proteins/srebp-2.md — corrections applied:
1. Ser-372 misattribution corrected: Wiki claimed “AMPK phosphorylates SREBP-2 at Ser-372.” Li 2011 (Cell Metabolism) identifies Ser-372 as the SREBP-1c residue; SREBP-2 is also a direct AMPK substrate (in vitro kinase assay) but the equivalent site on SREBP-2 is not explicitly numbered in the paper. Corrected to reflect SREBP-1c Ser-372 as the characterized site; SREBP-2 equivalent noted as unspecified.
2. PCSK9 removed from Horton 2002 target list: Horton 2002 JCI does not list PCSK9 (PCSK9 was not discovered until 2003). PCSK9’s SRE-mediated SREBP-2 regulation is now attributed to Jeong 2008 + Dubuc 2004 only.
3. Cleavage residue numbers qualified: Brown & Goldstein 1997 describes the S1P cleavage sequence as RSVL↓S (Leu-Ser bond) without specifying exact residue numbers for SREBP-2. Residue numbers (~Leu-522 for S1P, ~Leu-484 for S2P) attributed to Duncan 1997 (J Biol Chem 272:12778) and framed as approximate.
4. Peterson 2011 mechanism corrected: “excluding SREBP from chromatin” → “reducing nuclear SREBP protein abundance”; no direct lipin-1–SREBP physical interaction detected in Peterson 2011.
5. Identity fields confirmed: UniProt Q12772, NCBI Gene 6721, HGNC 11290, Ensembl ENSG00000198911, 1141 aa all verified via live API.

Pages unverifiable (closed-access): Dubuc 2004 (not_oa) — verified via PubMed abstract only; PCSK9-as-SREBP-2-target claim confirmed by abstract.

Downstream pages to check: pcsk9, hmgcr, ldlr, ampk — may cite SREBP-2 Ser-372 claim or the PCSK9-SREBP2 mechanism; main agent should check footnotes.

literature-checked-through: 2026-05-09 — no supersession candidates found (no meta-analysis or large RCT post-2023 directly addressing SREBP-2 mechanism in aging context).

[2026-05-09] R37 close — remaining R35-Stage2 + R36 forward queue (20 pages)

Continuation round following the R35-Stage2 + R36-followup close. User-requested “continue with remaining items” — seeded + verified the remaining forward-queue items surfaced by R35-Stage2. 20 atomic content + MOC pages seeded + verified in single-day round. Three waves of 8 + 8 + 4 seeders, with matched verifier waves; 60 agent invocations.

Pages added (all 20)

Wave 1 (8):

molecules/proteins/il-6.md (IL-6 cytokine — surprisingly absent before this round)
molecules/proteins/pomc.md (proopiomelanocortin parent precursor)
pathways/melanocortin-system.md (navigational MOC for MC1R-MC5R + ligands)
molecules/compounds/tocilizumab.md (anti-IL-6R Actemra; FDA 2010 RA + 6 other indications)
molecules/compounds/setmelanotide.md (Imcivree; MC4R agonist; FDA 2020 monogenic obesity + 2022 BBS)
pathways/neurotrophin-signaling.md (NGF/BDNF/NT-3/NT-4 + TrkA/B/C/p75NTR)
pathways/trkb-pathway.md (BDNF-TrkB sub-pathway)
pathways/spm-pathway.md (specialized pro-resolving mediators)

Wave 2 (8):

molecules/proteins/mc3r.md (Lam 2021 linear-growth/puberty + Patel 2025 hepatic autophagy)
molecules/proteins/mc5r.md (Chen 1997 KO exocrine phenotype)
molecules/compounds/colchicine.md (COLCOT + LoDoCo2 cardiovascular benefit)
processes/melanogenesis.md (eumelanin/pheomelanin chemistry; McSC exhaustion)
pathways/camp-signaling.md (Gαs/Gαi → AC → cAMP → PKA + Epac → CREB)
pathways/pomc-processing.md (PC1/3 + PC2 prohormone-convertase cascade)
biomarkers/glycanage-2017.md (IgG glycan aging clock; rapid intervention response)
molecules/proteins/hgf.md (Koike 2025 satellite cell aging mechanism)

Wave 3 (4):

molecules/proteins/c-met.md (HGF receptor; Paik 2020 tepotinib NEJM)
molecules/proteins/vegfr2.md (VEGF receptor; vascular aging + capillary rarefaction)
molecules/proteins/enos.md (NOS3; eNOS uncoupling + Nisoli 2005 CR-mito axis)
molecules/proteins/growth-hormone-receptor.md (Laron syndrome longevity paradox; canonical GH-IGF1 axis)

Mechanism classes added to frameworks/intervention-classes.md:

IL-6R antagonist (founding compound: tocilizumab)
microtubule-polymerization-inhibitor (founding: colchicine)

Major verifier findings — fabricated/inverted/wrong claims caught

This round continued the high-yield-correction pattern of the prior rounds. Every page had at least one substantive correction; many had multiple.

Direction-inverted claims (most consequential):

Lian 2020 (VEGFR2 page) — wiki said angiogenesis “reversed by miR-25-3p inhibition”; paper actually shows miR-25-3p is DOWNregulated in aging and restoration is achieved by AGOMIR (upregulation). Direction reversed.
Picca 2022 (GHSR page; carryover from prior round) — already documented; paper says ghrelin levels do NOT vary with aging (opposite of seeder claim).
Dehghan 2019 (cAMP-signaling page) — wiki said hydralazine INHIBITS PKA; paper actually shows hydralazine ACTIVATES PKA (binds + stabilizes catalytic subunit, releases regulatory subunit, then activates SIRT1). Direction backwards.
Kelly 2018 prefrontal cAMP (cAMP-signaling) — wiki conflated prefrontal + hippocampal cAMP decline; paper shows prefrontal cAMP INCREASES with age (opposite direction).

Fabricated cohort numbers / statistics:

IL-6 protein page — “~30% lower CHD odds” (Swerdlow 2012) was the most consequential factual error of the round; actual OR 0.95 (95% CI 0.93-0.97) per allele = 5% reduction; 30% nowhere in paper. Cohort n: “~40,000 cases” was actually 25,458/100,740. MR instrument: rs2228145 was actually rs7529229 (rs2228145 is functional correlate). sIL-6R plasma level: 300-400 ng/mL was actually for sgp130 (different soluble protein); sIL-6R is ~75 ng/mL — two soluble proteins swapped.
Colchicine page — multiple fabricated COLCOT figures: GI 7.7%/5.3% (not in paper; actual 17.5%/17.6% NS); non-CV death 0.9%/0.5% p=0.09 (fabricated); LoDoCo2 GI discontinuation 9.7%/6.4% (paper doesn’t report). Pneumonia signal misattributed COLCOT↔LoDoCo2 swap (LoDoCo2 directionally opposite). COPPS-2 p=0.002 → actual 0.046. clinical-trials-active 7 → 68.
GHR page — Guevara-Aguirre 2011 “~5-fold cancer reduction” removed (paper actually shows 1/99 GHRD vs 17% relatives — near-complete absence, not “fold reduction”). Multiple lifespan-extension percentages (Coschigano 35-40%; Bartke 50-60%; Brown-Borg 30-50%) removed because primary papers don’t state them.
GlycanAge page — Jurić 2020 framing wrong: wiki said “surgical menopause cohort with bilateral oophorectomy”; paper is GnRH agonist (leuprolide) pharmacological suppression in 36 healthy premenopausal women — completely different intervention.

Major attribution errors:

Nillni 2016 (POMC-processing page) — wiki described it as POMC/SIRT1 paper; paper is actually about CRH processing in PVN (PVN and arcuate are anatomically distinct).
Yan 2007 vs Enns 2010 (cAMP-signaling) — AC5-KO ~30% lifespan extension was attributed to Enns 2010; that paper is about PKA RIIβ-KO/Cβ-KO. AC5-KO data is from Yan 2007.
Latina 2017 (neurotrophin-signaling) — TrkA decline in human MCI/AD was attributed to this in vitro rat study; actual claim belongs to Counts & Mufson 2005.
Jang 2010 (TrkB pathway) — aging-rodent memory rescue misattributed; paper has no aging arm. Correctly attributed to Zeng 2012.
Webber 2024 GDF-15 PMID/DOI — already documented; same pattern.

Receptor-ligand mis-assignments:

GPR32 (SPM pathway) — wiki said GPR32 binds RvD2; actual GPR32/DRV1 mediates RvD1, AT-RvD1, RvD3, RvD5. RvD2 is the GPR18/DRV2 ligand, not GPR32.
MC5R ligand affinity (MC5R) — wiki had ACTH > β-MSH; actual order is α-MSH > ACTH = β-MSH >> γ-MSH (ACTH and β-MSH equal-affinity).
MC3R γ-MSH affinity — “Highest” → “High (comparable to α-MSH)”; Ki values overlap.

Fabricated phosphorylation sites:

NEFL (carryover; documented in prior round) — already corrected.
c-MET Ser985 — not in UniProt features; replaced with correct Tyr1003 (Cbl docking, eliminated by METex14).
VEGFR2 Tyr1054 — Zeng 2001 actually showed Y1054F = null effect; only Tyr1059 functionally essential.
eNOS Tyr657 nitrosylation — unsourced from cited paper; removed.

Critical gene/Ensembl ID errors:

MC5R Ensembl ID — ENSG00000198822 → ENSG00000176136 (old ID resolved to GRM3 — completely different gene, glutamate receptor). Most consequential canonical-DB-ID error of the round.

Mechanism overclaims hidden:

Tocilizumab Cupido 2022 pneumonia signal MISSING — IL6R instrument associates with INCREASED pneumonia risk (OR 1.17, 95% CI 1.09-1.27) — major safety-relevant finding for aging-indication framing.
Sosne 2015a (carryover; TMSB4X) — Phase 2 RCT failed both co-primary endpoints; documented in prior-round verifier.
Setmelanotide Collet 2017 — described as “case series”; actually Phase 1b randomised double-blind RCT.
Bonner 2013 (VEGFR2) — wiki had only ~60% capillary loss; cardiac was ~50%, skeletal ~60% — different magnitudes.
Wang 2014 SPM-receptor expression in AD (FPR2; carryover from prior round) — already documented.

Wrong species / strains / mouse models:

Howard 1996 (GHSR; carryover) — already documented.
Wang 2021 (cAMP-signaling) — strain C57BL/6 → CD1/ICR.
Koike 2025 (HGF + c-MET) — strain C57BL/6 → C57BL/6J male only.
Jang 2010 (TrkB) — “rat/mouse” → mouse only.
Joshi 1995 (POMC-processing) — ages 3/12 mo → 4-5/12-13 mo.

Mechanism overclaims corrected:

Eclipse of Ser1177 phosphorylation as Akt-only (eNOS) — multiple kinases now correctly attributed.
Fontana 2010 CR/HOMA-IR (carryover) — already documented.
Kovacs 2010 fibroblast cytokines (melanogenesis) — wiki said “bFGF”; actual SCF/HGF/KGF; “senescent” → “photoaged”.
Nishimura 2005 McSC depletion (melanogenesis) — wiki listed 3 mechanisms; paper shows only 2 (ectopic differentiation + BCL2 apoptosis); “DNA damage” + “Notch” removed as unsupported.
Niroula 2021 L-CHIP↔CAD null finding (carryover from prior round) — already documented.

Cross-page propagation passes

interventions/lifestyle/exercise.md (carryover) — Erickson 2011 framing already corrected in prior round.
molecules/compounds/mots-c.md (carryover) — Zempo 2021 kinesio-genomic p-value already corrected in prior round.

This round’s corrections were all caught at verification time on the newly-seeded pages; no carryover propagation surfaced because the pages are new.

Schema gaps surfaced

Carrying forward from prior rounds + new:

type: biomarker modality: enum still needs glycomic (GlycanAge precedent in this round) and somatic-genetic (CHIP precedent from prior round).
type: biomarker mendelian-randomization: enum still needs receptor-axis vs direct-protein-pQTL distinction.
clinical-stage enum still needs abandoned-phase-1 value (Melanotan II precedent from prior round).
GlycanAge filename convention question: glycanage-2017.md (commercial) vs glycanage-2014.md (Kristic foundational) — flagged for user.

Forward queue surfaced — next-round seeding

Implicit stubs from this round (lower priority; many are downstream sub-pathway pages):

Protein/receptor pages:

[[ngf]], [[ntrk1]], [[ntrk2]], [[ntrk3]], [[ngfr]] (neurotrophin signaling)
[[il-6r]] (IL-6 receptor; used by tocilizumab + IL-6 pathway)
[[pcsk1]], [[pcsk2]], [[cpe]], [[pam]] (POMC-processing convertases)
[[agrp]], [[asip]], [[mc2r]] (melanocortin system completion)
[[adcy]] family, [[prkaca]], [[rapgef3]], [[gnas]], [[creb1]], [[crtc1]], [[pde4]] (cAMP signaling)
[[gpr32]], [[chemr23]], [[lgr6]], [[alox15]], [[alox5]], [[alox12]], [[ptgs2]] (SPM pathway)
[[vegfa]], [[mitf]], [[tyr]], [[tyrp1]], [[dct]] (VEGFR2 + melanogenesis)

Pathway pages:

[[hgf-met-signaling]] (parent for HGF + c-MET)
[[hpa-axis]] (referenced from POMC + cortisol biology)
[[no-cgmp-pkg-pathway]] (eNOS downstream)
[[gh-jak-stat-pathway]] (GHR-specific JAK-STAT)
[[arachidonic-acid-pathway]], [[omega-3-metabolism]], [[phospholipase-a2]] (SPM upstream)

Phenotype/process:

[[laron-syndrome]] (referenced from GHR)
[[neurodegeneration]] (referenced from NEFL + neurotrophin pages)

Total agents dispatched (this round)

20 seeders (8 wave 1 + 8 wave 2 + 4 wave 3) + 20 verifiers (8 + 8 + 4) = 40 invocations across the day. All 20 atomic content pages now verified: true with partial-scope notes.

🧬 Aging Wiki

Explorer

R37

log/R37.md — Round 37 entries

[2026-05-13] verify | pathways/type-i-interferon-signaling.md

[2026-05-13] verify | studies/bastard-2021-anti-ifn-autoantibody-age-prevalence.md

[2026-05-13] verify | molecules/compounds/amlexanox.md

[2026-05-13] verify | molecules/proteins/ikbke.md

[2026-05-13] ingest | chen-2025 NMR cGAS follow-on batch (5 pages + propagation)

[2026-05-13] verify | molecules/proteins/irf3.md

[2026-05-13] verify | molecules/proteins/fancd2.md

[2026-05-13] verify | molecules/proteins/palb2.md

[2026-05-12] ingest | Chen et al. 2025 Science — naked-mole-rat cGAS HR-repair (cGAS dual-role + DNA-repair reframe)

[2026-05-13] verify — studies/zhen-2023-trim41-cgas-l1.md

[2026-05-12] verify — molecules/compounds/hyaluronic-acid.md

[2026-05-12] verify — molecules/proteins/cd44.md

[2026-05-12] verify — molecules/proteins/has2.md

[2026-05-12] verify — studies/zhang-2023-nmrhas2-mouse-healthspan.md

[2026-05-09] verify — molecules/compounds/patisiran.md

[2026-05-09] verify — molecules/compounds/acoramidis.md

[2026-05-09] verify — molecules/compounds/tafamidis.md

[2026-05-09] verify — molecules/compounds/donanemab.md

[2026-05-09] verify — molecules/compounds/lecanemab.md

[2026-05-09] verify — hypotheses/longevity-escape-velocity.md

[2026-05-09] verify — processes/glucosepane.md

[2026-05-09] verify — processes/lipofuscin.md

[2026-05-09] ingest + propagate — bempedoic acid / ezetimibe / mevalonate / MTP (batch 4, 4 pages)

[2026-05-09] verify — pathways/mevalonate-pathway.md

[2026-05-09] verify — molecules/compounds/ezetimibe.md

[2026-05-09] verify — molecules/compounds/bempedoic-acid.md

[2026-05-09] verify — molecules/proteins/mtp.md

[2026-05-09] ingest + propagate — Lp(a)/statins/ApoE/LPL (batch 3, 4 pages)

[2026-05-09] verify — molecules/proteins/lpl.md

[2026-05-09] verify — molecules/proteins/apoe.md

[2026-05-09] verify — molecules/proteins/lpa.md

[2026-05-09] ingest + propagate — lipid-lowering target triangle completed (batch 2, 3 pages)

[2026-05-09] verify — lipoprotein-metabolism pathway page

[2026-05-09] ingest + propagate — dietary-fat → ApoB → CV-event mechanistic chain (5-page batch)

[2026-05-09] verify — molecules/proteins/ldlr.md

[2026-05-09] verify — phenotypes/familial-hypercholesterolemia.md

[2026-05-09] verify — molecules/proteins/apob.md

[2026-05-09] verify — interventions/pharmacological/statins.md

[2026-05-09] verify — molecules/proteins/srebp-2.md

[2026-05-09] R37 close — remaining R35-Stage2 + R36 forward queue (20 pages)

Pages added (all 20)

Major verifier findings — fabricated/inverted/wrong claims caught

Cross-page propagation passes

Schema gaps surfaced

Forward queue surfaced — next-round seeding

Total agents dispatched (this round)

Graph View

Table of Contents

Backlinks