log/R34.md — Round 34 entries

Sub-file of log — see parent for index.

[2026-05-13] verify | molecules/proteins/irf9.md

Page verified: molecules/proteins/irf9.md

Sources checked:

Rengachari 2018 (doi:10.1073/pnas.1718426115) — PDF downloaded and read end-to-end. DBD residues 9–116 confirmed; IAD 182–399 confirmed; STAT2 CCD docking confirmed. No numeric corrections needed.
Platanitis 2019 (doi:10.1038/s41467-019-10970-y) — PDF downloaded and read. Key finding: ISGF3 assembly on DNA (not cytoplasm) per their model; classical cytoplasmic model is labeled “dogmatic view” they challenge. Cascade step 3 corrected and Platanitis finding explicitly noted. STAT2-IRF9 pre-complex confirmed.
Fink & Grandvaux 2013 (doi:10.4161/jkst.27521) — PDF downloaded and read. IRF9 non-phosphorylation feature and ISRE vs GAS specificity confirmed accurate.
Meyts & Casanova 2021 (doi:10.1002/eji.202048793) — PMC full-text XML (PMC8900014) accessed; download failed (bronze OA, KU Leuven link 404’d). Critical correction: wiki had “Irf9-/- mice show profound susceptibility including VSV with dramatically shortened survival curves” — NOT supported. Meyts 2021 explicitly states in-vivo challenge data is sparse; only LCMV in-vivo model exists (Irf9-/- survive, unlike Stat1-/-); VSV data was from in-vitro cells only. Also: wiki claimed “lymphopenia” in IMD65 — not stated in Meyts 2021. Both corrected.
Odama 2023 (doi:10.1021/acs.jafc.2c08563) — PDF downloaded and read. Correction: wiki said BA “suppressed IRF9 transcript levels” — paper showed BA reduced IRF9 protein (western blot Figure 4); transcript-level downregulation of IRF9 itself not directly shown. siRNA result (Figure 7) confirmed accurate: siIRF9 reduced IFIG mRNAs in senescent NHDFs. SA-β-gal quantitative data (47.1% → 24.2% at 10 µM) added to body.
Lei 2025 (doi:10.1038/s41467-025-66368-6) — PDF downloaded and read. IRF9 is NOT mentioned in this paper. Mechanism is NRF1→TBK1/IRF3→type-I IFN production (upstream of IFNAR signaling). Wiki had over-attributed this paper to “tonic IFN flowing through IRF9.” Body text and footnote corrected to clarify the NRF1-TBK1/IRF3 axis and that IRF9 connection is an inference, not a Lei 2025 finding.

Corrections (6):

ensembl: null → ENSG00000213928 (confirmed via UniProt REST)
STAT2 Tyr689 → Tyr690 in signaling cascade (confirmed UniProt P52630 and Rengachari 2018)
ISGF3 assembly description updated: Platanitis 2019 challenges cytoplasmic assembly model (assembly on DNA)
Irf9-/- phenotype: removed unsupported in-vivo VSV survival claim; corrected to in-vitro data + LCMV in-vivo model
Human IMD65: removed unsupported “lymphopenia” claim; corrected to influenza pneumonia + MMR complications
Odama 2023: “IRF9 transcript levels” → “IRF9 protein levels”; siRNA result detail and SA-β-gal quantitative data added
Lei 2025: body text and footnote corrected to note IRF9 not directly studied; NRF1-TBK1/IRF3 axis described accurately; IRF9 connection flagged as pathway inference

Supersession check (R25): type:protein page; supersession check not mandated per CLAUDE.md. literature-checked-through: 2026-05-13 confirmed (set by seeder).

Downstream propagation needed: None — irf9.md newly seeded; no other entity pages currently cite it.

[2026-05-13] verify | molecules/proteins/stat2.md

Page verified: molecules/proteins/stat2.md

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

Hambleton 2013 (doi:10.1073/pnas.1220098110) — PDF downloaded and read. n updated; splice-site mutation c.381+5 G>C specified; cell type corrected to skin fibroblasts.
Yamagami 2018 (doi:10.1038/s41514-018-0030-6) — PDF downloaded and read. Cell type specified as NHDFs; ISG target list added; JAK1-independence of ISG expression confirmed and added.
Arimoto 2017 (doi:10.1038/nsmb.3378) — PDF downloaded and read. Critical: IFNAR2 interaction domain corrected “via STAT2 TAD” → “via STAT2 CC+DB domains.”
Kok 2020 (doi:10.15252/msb.20198955) — PDF downloaded and read. Key claims confirmed; model details added to footnote.
Takahashi 2018 (doi:10.1038/s41467-018-03555-8) — PDF downloaded and read. Critical: cell line “human IMR90 cells” corrected → “human TIG-3 fetal lung diploid fibroblasts.”
Levy & Darnell 2002 (doi:10.1038/nrm909) — local PDF read end-to-end. DOI confirmed correct.

Corrections (5): See verified-scope in frontmatter.

Supersession check (R25): type:protein page; supersession check not mandated. No superseding meta-analyses found. literature-checked-through: 2026-05-13.

Downstream propagation needed: None — stat2.md newly seeded; not yet cited by other entity pages.

[2026-05-13] verify | molecules/proteins/ifnar1.md

Page verified: molecules/proteins/ifnar1.md

Sources checked:

Stark & Darnell 2012 (Immunity; 10.1016/j.immuni.2012.03.013) — local PDF read end-to-end. Confirmed: TYK2→IFNAR1 and JAK1→IFNAR2 constitutive pre-associations, ISGF3 assembly. Confirmed: does NOT describe FNIII/SD1–SD4 ectodomain architecture or anifrolumab (perspective published 2012). Misattribution of anifrolumab SD1-SD2 structural claim to this paper corrected.
TULIP-2 Morand 2020 (NEJM; 10.1056/NEJMoa1912196) — local PDF read end-to-end. All BICLA/n/CI/p-value data confirmed consistent with verified IFNAR2 page: 180/182 mITT, 86/180 vs 57/182, 16.3 pp, 95% CI 6.3–26.3, P=0.001, HZ 7.2%, bronchitis 12.2%, one pneumonia death in anifrolumab arm. Treatment duration corrected 52→48 weeks.
Bhattacharya 2010 (J Biol Chem; 10.1074/jbc.M109.071498) — local PDF read end-to-end (downloaded this session). Ligand-induced pathway kinase corrected (PKD → TYK2-dependent). Ligand-independent pathway mechanism nuanced: PERK activates an intermediate kinase (not PERK directly) → Ser532 priming.
Bastard 2021 (Sci Immunol; 10.1126/sciimmunol.abl4340) — cross-checked against verified studies/bastard-2021-anti-ifn-autoantibody-age-prevalence.md (PDF not re-read). All body claims confirmed consistent.
Roy 2022 (Immunity; 10.1016/j.immuni.2022.03.018) — local PDF read end-to-end (downloaded this session). Cell-type specificity corrected and expanded (microglial MKO→post-synaptic; neural NKO→pre-synaptic+plaque; blockade→behavioral rescue).
Hou 2022 (Aging Cell; 10.1111/acel.13582) — DOI MISMATCH. Archive resolves to Uhrlaub et al. (CD8/WNV paper), not the claimed IFNAR1-COVID paper. Claim tagged unsourced.

Corrections (6):

Anifrolumab structural claim attribution: removed from Stark & Darnell 2012 → new footnote pointing to structural publication needed.
Ligand-induced degron kinase: PKD → TYK2-catalytic-activity-dependent.
Ligand-independent degron: PERK-direct → PERK-activates-intermediate-kinase → Ser532 priming.
TULIP-2 treatment duration: “52 weeks” → “48 weeks.”
Roy 2022 cell-type specificity corrected in body.
Hou 2022 DOI flagged as wrong paper; dependent claim tagged unsourced.

Supersession candidates (R25): No supersessing RCT or meta-analysis found. literature-checked-through: 2026-05-13 confirmed.

Downstream propagation needed:

pathways/type-i-interferon-signaling.md — check TULIP-2 treatment duration if stated.
molecules/compounds/anifrolumab.md (when seeded) — SD1 structural claim needs correct source.
Any page citing 10.1111/acel.13582 as Hou IFNAR1-COVID paper.

[2026-05-13] verify | molecules/proteins/irf7.md

Page verified: molecules/proteins/irf7.md

Sources checked:

Ciancanelli 2015 (doi:10.1126/science.aaa1578; PMID 25814066) — PDF download failed (archive status: failed, green OA). Abstract verified via PubMed efetch. Key correction: wiki said “homozygous or compound-heterozygous” — paper reports compound-heterozygous null only (single case); body text corrected. Paper also tested fibroblasts and iPSC-derived pulmonary epithelial cells in addition to leukocytes/pDCs; body text updated.
Zhang 2020 (doi:10.1126/science.abd4570) — PDF downloaded and read pages 1–8 (13-page paper). IRF7 confirmed as specifically listed gene; n=8 IRF7-variant patients in Table 1, ages 37–69 (body expanded with specifics). pDC IFN-α/λ production assay in AR IRF7-deficient patient confirmed. n=659/534 confirmed.
Sugihara 2026 (doi:10.1186/s12979-026-00555-x) — PDF downloaded and read end-to-end. Critical direction error corrected: wiki claimed “excess basal IFN-α activity” associated with frailty — SOURCE SHOWS THE OPPOSITE. pDC IFN-α production DECLINES with frailty (OR 0.212, 95% CI 0.051–0.895, p=0.04 in fully adjusted model). n=141 confirmed. Age (79.5±5.3 yr) confirmed. Frail n=13 noted.
Wu 2025 (doi:10.1007/s11357-025-01892-3; PMID 40987991) — PDF download failed (hybrid OA). Abstract verified via PubMed efetch. Mouse details confirmed: C57BL/6J, 12-wk young vs 70-wk old. Key nuance added: at 7 dpi, Irf7 mRNA is ELEVATED in aged mice (late-phase), contradicting simple “loss of IRF7” framing.
Yang 2025 (doi:10.1186/s13018-025-06518-0) — PDF downloaded and read pages 1–12 (21-page paper). Confirmed IRF7 and SPHK1 downregulated in aging meniscus tissue; n=15+15 confirmed; IHC p<0.0001 confirmed; ROC-AUC values (SVM=0.73, ANN=0.79) confirmed.

Corrections (5):

“Homozygous or compound-heterozygous” → “compound-heterozygous null” for Ciancanelli patient genotype
Ciancanelli cell types expanded to include fibroblasts and iPSC-derived pulmonary epithelial cells
Aging context section completely rewritten: direction of pDC IFN-α change with frailty corrected from “increased basal tone” to “decreased stimulus-evoked IFN-α production” (Sugihara 2026)
Wu 2025 nuance added: Irf7 mRNA elevated at 7 dpi in aged mice; early IFN deficit is a timing/kinetics defect, not simple IRF7 loss
Yang 2025 body text expanded with precise n, IHC p-value, ROC-AUC metrics, and age cutoffs

Supersession check (R25): No meta-analyses or large RCTs found. Wu YX et al. Aging Cell 2025 (PMID 40583123) confirms IRF7 signaling impairment in aged CD11b+ cells during viral challenges — corroborates rather than supersedes. Fernandes 2025 IJMS (resveratrol/IRF7 in aging) is methodologically unrelated. literature-checked-through: 2026-05-13 (already set at seed time).

[2026-05-13] verify | pathways/tlr3-trif-pathway.md

Page verified: pathways/tlr3-trif-pathway.md

Sources checked:

Alexopoulou 2001 (doi:10.1038/35099560) — local PDF read end-to-end. Confirmed TLR3 dsRNA receptor, NF-κB + IFN activation, C57BL/6 TLR3-/- model. Paper does NOT quantify minimum dsRNA length in bp; corrected attribution of 40-48 bp claim to Choe 2005.
Srivastava 2025 (doi:10.3390/pathogens14070624) — PDF downloaded and read (gold OA). Critical error corrected: paper studies lung/airway epithelial cells (AECs), NOT corneal epithelial cells. Cytokine corrected (IFN-α not IFN-β), n corrected (5/group), age clarified (68 wk), DC/macrophage findings added.
Wang 2026 (doi:10.3389/fimmu.2025.1650375) — PDF downloaded and read (gold OA). Body claims verified correct. Footnote expanded with full design (bioinformatics + human tissue + chondrocyte cell lines + ACLT rat n=7-8/group).
Fitzgerald 2003 DOI confirmed: doi:10.1038/ni921 (PMID 12692549). Meylan 2004 DOI confirmed: doi:10.1038/ni1061 (PMID 15064760) — RIP1/NF-κB paper, not IRF3 phosphosites. Ser386/Ser396 residue attribution still gap/unsourced.
Reactome R-HSA-168164 confirmed valid. KEGG hsa04620 accepted.
Yamamoto 2003, Oshiumi 2003, Choe 2005, Jin 2026, Town 2006 — not_oa, cannot verify.

Corrections (5): dsRNA length attribution moved to choe2005; Srivastava 2025 cell type “corneal” → “airway/lung AEC”; cytokine IFN-β → IFN-α; irf3-sites PMIDs corrected; Wang 2026 footnote expanded.

Supersession check (R25): No meta-analyses or large RCTs on TLR3/TRIF in aging (2022-2026 search); consistent with page’s own assessment.

[2026-05-13] verify | pathways/rig-i-mavs-pathway.md

Page verified: pathways/rig-i-mavs-pathway.md

Sources checked:

Hou 2011 (doi:10.1016/j.cell.2011.06.041) — local PDF read end-to-end. 4 corrections applied.
Seth 2005 (doi:10.1016/j.cell.2005.08.012) — local PDF read end-to-end. 2 corrections applied.
Dhir 2018 (doi:10.1038/s41586-018-0363-0) — local PDF read end-to-end. 1 correction applied.
Victorelli 2025 (doi:10.1038/s41467-025-66159-z; PMID 41398033) — local PDF downloaded and read end-to-end. Claims confirmed; in vivo model terminology corrected.
Zhang 2026 (doi:10.1038/s41422-026-01224-w) — not in archive (DOI not found); unverifiable from PDF. Claims tagged preliminary on page.
KEGG hsa04622 — confirmed via REST (RIG-I-like receptor signaling pathway).
Reactome R-HSA-168928 — confirmed via web (DDX58/IFIH1-mediated induction of interferon-alpha/beta; 5-star curation).
Reikine 2014 review (doi:10.3389/fimmu.2014.00342) — review paper; no primary quantitative claims sourced exclusively to it; not re-read.

Corrections (5):

Hou 2011 body — IRF3 activation: “sub-nanomolar concentration” → “as low as 16 ng/ml” (exact paper language; sub-nanomolar is a derived calculation not stated in paper).
Hou 2011 body — added quantitative prion propagation datum: ~1 ng PK-MAVS converts ~16 ng endogenous MAVS within 30 min (stated in paper, not previously on page).
Seth 2005 body — peroxisome mislocalization: removed “or peroxisome” attribution to Seth 2005; paper tested plasma membrane (CAAX) and ER (VAMP-2), NOT peroxisome. Tagged unsourced for Dixit 2010 peroxisome claim.
Seth 2005 footnote — wrong cell lines: “HEK293T + Vero cells” → “HEK293 cells + multiple human cell lines (Jurkat, U937, Huh7, A549, HeLa)”; Seth 2005 uses HEK293 (not HEK293T); Vero cells not in Seth 2005.
Dhir 2018 footnote — patient model: “PNPT1-mutation patient blood” → “primary fibroblasts from 4 PNPT1-mutation patients (ISG upregulation measured in peripheral blood); hepatocyte-specific PNPase KO mice.”
Victorelli 2025 body — in vivo model: “metabolic liver disease” → “FFC-diet model of metabolic dysfunction-associated steatohepatitis (MASH)” with miMOMP mechanism added.

Supersession candidates (R25): None. López-Polo et al. 2024 (PMID 39191740, doi:10.1038/s41467-024-51363-0, Nature Communications) is a converging independent replication by the Serrano lab reaching the same conclusion (mt-dsRNA drives SASP via RNA sensing), not a supersession — Victorelli 2025 explicitly acknowledges it. No meta-analyses or large RCTs identified. literature-checked-through: 2026-05-13 confirmed.

Unverifiable claims:

Zhang 2026 (doi:10.1038/s41422-026-01224-w) — not indexed in archive; PDF not obtainable. SEC61A1 axis, Mavs-ablation cognitive rescue, and AD patient pathology claims remain unverified against full text; tagged preliminary on page.
PINK1-MAVS direct phosphorylation claim (line 144) — already tagged unsourced on page; not addressed in any PDF read here.

Downstream propagation needed: The Seth 2005 cell line correction (HEK293T → HEK293) may propagate to molecules/proteins/mavs.md if that page’s footnote uses the same erroneous “HEK293T” description — main agent should check.

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

Page verified: molecules/proteins/mavs.md

Sources checked:

Seth 2005 (doi:10.1016/j.cell.2005.08.012) — local PDF read end-to-end. All claims confirmed.
Meylan 2005 (doi:10.1038/nature04193) — local PDF read end-to-end. All claims confirmed.
Hou 2011 (doi:10.1016/j.cell.2011.06.041) — local PDF read end-to-end. All claims confirmed.
Dhir 2018 (doi:10.1038/s41586-018-0363-0) — local PDF read end-to-end. All claims confirmed including ~90-fold IFN-β induction, 4 PNPT1 patients, MDA5 (not RIG-I) as primary sensor, MAVS knockdown abrogation.
Xu 2005 (doi:10.1016/j.molcel.2005.08.014) — DOI resolved via PubMed efetch (PMID 16153868 confirmed) and Crossref; PDF downloaded (camoufox, 640 KB) and read. DOI confirmed correct. All claims confirmed.
Zhang 2026 (doi:10.1038/s41422-026-01224-w) — verified via PubMed efetch (PMID 41692872; PMC13092635; Vol 36(5):322-339; published 2026-02-16). PDF not in archive (DOI not indexed; PMC OA release 2027-05-01). Abstract cross-checked; all claims on page consistent.
Wang 2025, Dixit 2010, Liu 2015, Rai 2021 — PDFs not locally available; not PDF-verified.
Victorelli 2025 (doi:10.1038/s41467-025-66159-z; PMID 41398033) — added to senescence section (consistent with companion rig-i-mavs-pathway.md; PDF not independently verified on this pass).

Corrections (3):

[^xu2005] footnote: “local PDF: pending download” → updated to confirmed local path
[^zhang2026] footnote: added Vol 36(5):322-339, PMC13092635, 5xFAD mice context, clarified OA release timeline; updated archive note to reflect DOI not indexed (not a download failure).
MAVS senescence section: added Victorelli 2025 paragraph (mt-dsRNA → RIG-I/MDA5 → MAVS aggregation → SASP in normal senescent cells; BAX/BAK leakage regulation) with [^victorelli2025] footnote; Wang 2025 oncology paragraph retained with clarified framing.

Supersession check: Skipped — type: protein page (canonical-data-stable type per SOP).

Final state: verified: true (scope updated to reflect Xu 2005 PDF now verified; Victorelli 2025 added; Zhang 2026 abstract-verified; Wang 2025/Dixit 2010/Liu 2015/Rai 2021 remain PDF-unverified per scope note).

[2026-05-13] verify | molecules/proteins/stat1.md

Page verified: molecules/proteins/stat1.md

Sources checked:

Au-Yeung N, Mandhana R, Horvath CM 2013 (doi:10.4161/jkst.23931) — local PDF downloaded and read end-to-end. Corrections applied.
Cancado de Faria R et al. (Gonzalo lab) 2023 (doi:10.59368/agingbio.20230009) — local PDF downloaded and read end-to-end. Corrections applied.
Hu X et al. 2021 (doi:10.1038/s41392-021-00791-1) — local PDF downloaded and read end-to-end. Corrections applied.
Li X et al. 2023 (doi:10.1038/s41392-023-01502-8) — local PDF read (pre-existing). Used for HSC aging context.
Shaw 2013 (doi:10.1038/nri3547) — download failed (no OA URLs); tagged no-fulltext-access.
Rentschler 2026 (doi:10.1007/s11357-026-02100-6) — download failed; tagged no-fulltext-access.
Pang 2011 / Tierney 2014 gap investigation — PubMed-confirmed; see corrections below.

Corrections (8):

Opening paragraph: removed “satellite-cell dysfunction, and HSC aging” from the list of STAT1 aging roles in the lead sentence — these specific STAT1 roles lack dedicated primary-source support and had been conflated with STAT3/broader JAK-STAT effects.
Author correction (Au-Yeung 2013 footnote): “Au-Yeung & Bhatt” → “Au-Yeung N, Mandhana R, Horvath CM” (third author is Curt M. Horvath; “Bhatt” does not appear in the paper).
Author correction (Gonzalo 2023 footnote): “Gonzalo S et al.” → first author is “Cancado de Faria R”; Gonzalo S is corresponding/last author. Full author list added.
Gonzalo 2023 quantitative claims added: baricitinib (not ruxolitinib) was the primary JAK inhibitor in this paper; Stat1 haploinsufficiency (+/−, not full KO) extends lifespan 15%; calcitriol 22% median survival improvement (145 vs 119 days, log-rank); complete Stat1 KO was detrimental.
Hu 2021 author correction: first author “Xiaoyi Hu”; full first-author family name added.
Pharmacology section: corrected “Mannick 2018 ruxolitinib frailty trial” claim — Hu 2021 does not reference this trial; Mannick trials used rapalogs (mTOR inhibitors), not ruxolitinib. Claim corrected to accurately reflect Hu 2021 content.
HSC aging section: removed implicit citation of Pang 2011 for STAT1-in-HSC claim. Pang 2011 (PMID 22123971) exists and describes myeloid bias in aged human HSCs but contains no STAT1 data. Claim recited to use Li 2023 for general HSC-aging IFN context.
Satellite cell section: corrected “Tierney 2014 Cell Stem Cell” claim. That paper is actually Tierney MT et al. 2014 Nature Medicine (PMID 25194572; doi:10.1038/nm.3656) and is about STAT3 (not STAT1) activated by IL-6. The seeder brief conflated STAT1 with STAT3. The satellite cell section now correctly identifies STAT3 as the documented effector and flags STAT1-specific satellite cell evidence as a gap.

Unverifiable claims (2 sources):

Shaw 2013 NRI — ISG/interferon-tone accumulation claim; not_oa.
Rentschler 2026 GeroScience — STAT1/NF-kB cytokine-induced senescence claim; download failed.

Supersession candidates: None identified. No meta-analyses or large RCTs on STAT1-specific aging interventions found in recency search. literature-checked-through: 2026-05-13 confirmed.

Downstream propagation needed:

Any page citing STAT1 as directly implicated in satellite-cell aging (via STAT1/STAT3 conflation) should be reviewed. The wiki does not yet have a dedicated satellite-cell page citing this claim, but if one is seeded citing “Tierney 2014 Cell Stem Cell + STAT1,” it will need correction.

[2026-05-13] verify | molecules/proteins/jak1.md

Page verified: molecules/proteins/jak1.md

Sources checked:

Xu 2015 PNAS (10.1073/pnas.1515386112) — local PDF read end-to-end. Corrections applied (see below).
Kandhaya-Pillai 2022 Aging Cell (10.1111/acel.13646) — local PDF read end-to-end. Corrections applied (see below).
Xu 2016 Pharmacological Research (10.1016/j.phrs.2016.05.015) — review/perspective; download failed (no OA URLs). Not re-read; programmatic perspective claim preserved as seeded.
Levy 2002 Nat Rev Mol Cell Biol (10.1038/nrm754) — not_oa; tagged no-fulltext-access in footnote.
UniProt P23458 — live REST endpoint queried 2026-05-13. All canonical IDs and domain boundaries confirmed.

Corrections (6):

Treatment duration: 8 wk → 10 wk (Xu 2015 Fig. 7 legend and text state 10 weeks).
Mouse age: “24–27 months” → “24-month-old” (paper specifies 24-month-old male mice throughout).
n per group: “~20/group” → “n=8–9/group” (CLAMS: n=8; grip/RotaRod: n=9 per Fig. 8 legends).
Tofacitinib scope corrected: only ruxolitinib (INCB18424) was used in mouse frailty phenotyping; tofacitinib tested only in vitro. Wiki body and table corrected.
Kandhaya-Pillai cell model footnote: “WI-38 human fibroblasts” → “HUVECs” (paper uses HUVECs exclusively; body prose was correct, only footnote was wrong).
Limitation statement on Xu 2015 serum cytokines corrected: paper did quantify serum cytokine reductions (Fig. 7A); limitation revised to correctly identify the unproven causal link between cytokine reduction and frailty improvement.

Unverifiable claims: Levy 2002 JH2 autoinhibition (not_oa; no-fulltext-access tagged in footnote). Xu 2016 perspective details (download failed; low-priority review).

Supersession candidates (R25): No powered human RCT or meta-analysis found targeting JAK inhibitors for aging/frailty as primary endpoint (search 2016–2026). literature-checked-through: 2026-05-13 confirmed.

Downstream propagation needed:

studies/xu-2015-jak-inhibitor-sasp-frailty — verify n, mouse age, duration, tofacitinib scope.
studies/kandhaya-pillai-2022-tnf-ifng-jak-sasp — confirm HUVECs (not WI-38) as cell model.

[2026-05-13] verify | molecules/proteins/tyk2.md

Page verified: molecules/proteins/tyk2.md

Sources checked:

Dendrou 2016 (Sci Transl Med; 10.1126/scitranslmed.aag1974) — PDF read end-to-end. Corrections applied (see below).
Roskoski 2023 (Pharmacological Research; 10.1016/j.phrs.2022.106642) — PDF read end-to-end. Domain residue ranges, selectivity ratios, and FDA approval confirmed.
Armstrong 2025 (Dermatology and Therapy; 10.1007/s13555-025-01606-9) — PDF read end-to-end. Study design clarified (post-hoc non-responder analysis, not primary ITT).
UniProt P29597 — domain residue ranges confirmed via REST API.
Kreins 2015 (10.1084/jem.20140280) — not_oa; not PDF-verified; no-fulltext-access retained.
Chen 2025 (10.1007/s10067-025-07597-4) — not_oa; not PDF-verified.

Corrections applied:

“at least 7 autoimmune conditions” → “10 different autoimmune conditions” (Dendrou paper p.4 explicit count; Fig. 1C)
“~30–60% reduced risk” quantification removed — this phrasing does not appear in the paper; replaced with the actual reported measures: >50% lower pSTAT in homozygotes, ~70% less IL-12/IL-23 signaling, non-additive dosage effect
Minor allele frequency added: ~0.20% in UK general population (Dendrou p.5)
Selectivity claim upgraded: “>100-fold” → “100- to 200-fold vs JAK1/3; ~3000-fold vs JAK2” (Roskoski p.10, cellular assay data)
Armstrong 2025 misidentified as primary POETYK PSO efficacy paper; corrected to post-hoc apremilast non-responder analysis (n=165 switchers). PASI 75 results in the body now correctly attributed to the primary JAAD 2023 papers for week-16 data, and Armstrong 2025 for week-52 switcher data.
MDM2 “no interaction” bullet removed from Key Interactors (uninformative noise)
⚠️ banner removed

Supersession check (R25 adapted): 13 PubMed hits for deucravacitinib RCTs/meta-analyses 2023–2026. 4-year POETYK long-term extension (PMID 40045918; n=1,519; JEADV 2025) confirms durable efficacy and stable safety profile. No supersession of existing claims — all newer literature supports the current framing. literature-checked-through: 2026-05-13 confirmed.

[2026-05-13] verify | molecules/proteins/ifnar2.md

Page verified: molecules/proteins/ifnar2.md

Sources checked:

Bastard 2021 (Sci Immunol; 10.1126/sciimmunol.abl4340) — cross-checked against verified study page studies/bastard-2021-anti-ifn-autoantibody-age-prevalence.md; PDF not re-read (study page is verified: true). All age-band prevalence figures confirmed consistent: 0.18% (18–69y), 1.1% (70–79y), 3.4% (>80y) by 10 ng/mL assay; ~6.3% (>80y) by 100 pg/mL assay; 18.5%/13.3% of deceased COVID patients. Six pre-COVID-19 confirmed samples. All consistent.
Stark & Darnell 2012 (Immunity; 10.1016/j.immuni.2012.03.013) — local PDF read end-to-end. Confirmed: JAK1→IFNAR2 constitutive association; TYK2→IFNAR1 constitutive association; IFN-β higher affinity than IFN-α confirmed; ISGF3 assembly (p-STAT1/p-STAT2/IRF9) confirmed. Paper is a Perspective/review (confirmed; correctly labeled in footnote).
TULIP-2 Morand 2020 (NEJM; 10.1056/NEJMoa1912196) — PDF downloaded and read end-to-end. Corrections applied (see below).
Zuo 2025 (Cell Mol Immunol; 10.1038/s41423-025-01266-x) — not_oa (closed-access). Claims tagged no-fulltext-access.
Pestka 1997 (PMID 9208871) — no DOI in PubMed (confirmed via efetch). PMID-only citation correct. Title “The interferon receptors,” journal Seminars in Oncology, 1997 — confirmed.
ClinicalTrials.gov v2 API queried live 2026-05-13.

Corrections (4):

TULIP-2 p-value: P<0.001 → P=0.001 — PDF Table 2 gives adjusted P=0.001 (Cochran–Mantel–Haenszel stratified test with weighted Holm procedure). “P<0.001” was incorrect. Body and footnote corrected.
TULIP-2 raw counts and CI added — PDF confirms 86/180 anifrolumab vs 57/182 placebo; difference 16.3 pp (95% CI 6.3–26.3). Dosing interval clarified to “every 4 weeks” (not “monthly”). Bronchitis 12.2% added (present in PDF abstract but absent from wiki).
Active anifrolumab trial count: 4 → 20 — seeder’s count of 4 was substantially wrong. Live ClinicalTrials.gov v2 API 2026-05-13 returned 20 RECRUITING or ACTIVE_NOT_RECRUITING studies. Body updated with accurate count and representative trial list.
Zuo 2025 not_oa flagged — no-fulltext-access tag added to body claim and footnote; abstract-only sourcing documented.

Unverifiable claims:

Zuo 2025 (closed-access): β-hydroxybutyrylation STAT1-K592 mechanism, O-GlcNAc competition, fructose rescue in aged mice. Abstract-only; cannot confirm quantitative details or methods. Tagged no-fulltext-access.
Kalunian 2023, Órpez-Zafra 2017, Fricke-Galindo 2022, Bonafe 2020 — OA papers pending download; qualitative body claims preserved as seeded; no contradictory evidence found from available sources.
Kaewin 2026 (ARD) — 2026 publication, not yet indexed in archive; qualitative claim preserved.

Supersession candidates (R25): PubMed search returned 30 meta-analyses/RCTs post-2021 on anifrolumab. None supersede the TULIP-2 efficacy conclusion. Most relevant newer work: PMID 41342907 (indirect meta-analysis vs belimumab, n=4,332, 7 RCTs) — consistent with existing framing; confirms higher herpes zoster rate with anifrolumab. No supersession found. literature-checked-through: 2026-05-13 confirmed.

Downstream propagation needed: None urgent. The p-value and CI corrections are contained to this page; no other wiki pages cite the TULIP-2 p-value directly. Main agent should note that when molecules/compounds/anifrolumab.md is seeded, the verified TULIP-2 figures should be pulled from this corrected page.

[2026-05-13] schema | R34 mouse-ortholog formalized + R35 type:pathway extensions

User-acknowledged schema-gap-fill follow-on to a prior session’s flagged gaps. Two CLAUDE.md schema edits + 2 new pages + propagation pass.

R34 schema formalization — mouse-ortholog: on type: protein:

Field was already used informally on 164 protein pages with a consistent convention (mouse gene symbol in MGI sentence-case). Now formally documented with three syntax variants captured: (1) single-symbol default (e.g., Akt1); (2) semicolon-delimited multi-subunit (e.g., Rpa1 (RPA70); Rpa2 (RPA32); Rpa3 (RPA14)); (3) slash-delimited paralogs (e.g., Pot1a / Pot1b); (4) parenthetical inline annotation when useful.
Added to CLAUDE.md type: protein block (line ~154). Tagged #gap/needs-mouse-ortholog introduced for protein pages with extrapolation discussion but no ortholog listed.
Zero existing pages need modification — the 164-page precedent is already compliant.

R35 schema escalation — type: pathway field additions (2 fields, batched as R35):

kegg-secondary: (optional list) — handles pathway pages spanning multiple KEGG IDs (common for broad signaling axes; surfaced by type-i-interferon-signaling.md which spans hsa04623 cytosolic-DNA-sensing AND hsa04622 RIG-I-like). Lint pass updated to not treat secondary KEGG IDs as conflicting when same ID appears as primary on another page (e.g., shared with cgas-sting.md).
literature-checked-through: (optional ISO date) — mirrors R28 hallmark + R33 process patterns. Populate for pathway pages with active intervention pipelines (JAK inhibitors, NRTIs, STING antagonists, rapalogs, etc.). Lint pass flags pathway pages with date >18 months old when populated; null is acceptable for slow-turnover canonical-signaling pages.

Lint-pass section updated with new pathway-staleness check, between compound/intervention (12mo) and hallmark/model-organism (24mo) cadences.

New atomic pages (2; both verified: true after wiki-verifier passes — full details in companion [2026-05-13] verify entries below):

pathways/type-i-interferon-signaling.md (178 lines; 8 primary-source citations; 7 verifier corrections including a fabricated ruxolitinib appearance in Jin 2022, wrong cell model in Kandhaya-Pillai 2022, and mis-attributed Bastard 2021 figures to the 2020 paper)
molecules/proteins/ikbke.md (215 lines; 9 primary-source citations; 7 verifier corrections including a critical reversal of the Reilly/Oral 2017 Phase 2 framing — seeder fabricated a null-primary-endpoint outcome when the trial actually hit its primary HbA1c endpoint with pre-defined responder criteria; first-author also corrected from Reilly → Oral)

Propagation check (clean):

No other wiki pages cite [^bastard2020] for the age-stratified prevalence figures
No other wiki pages cite [^jin2022] for ruxolitinib as microglial-senescence rescue
tbk1.md only cites Reilly 2013 (correctly framed; not the erroneous 2017 paper); no inherited error
amlexanox.md does not exist yet (implicit stub) — when seeded, must use the verifier-corrected Reilly/Oral 2017 framing
No downstream propagation required for the seeded errors — fabrications were contained to the two new pages

Seeder-fabrication pattern recurrence (memory reinforcement):

Existing feedback_seeder_fabricates_outcomes.md memory continues to predict the failure mode accurately. In this round the seeder fabricated:
- Jin 2022: ruxolitinib appearance (drug not in paper — only IFNAR1/2 shRNA)
- Jin 2022: claimed mechanism reversal of senescence markers
- Kandhaya-Pillai 2022: wrong cell line (HUVECs framed as IMR90/WI-38)
- Kandhaya-Pillai 2022: fabricated TMPRSS2 upregulation (paper only reports ACE2/DPP4)
- Bastard 2020 / Bastard 2021: prevalence figures mis-attributed to wrong companion paper
- Reilly/Oral 2017 (IKKε): full reversal of primary-endpoint outcome + wrong first author
The verifier brief explicitly flagged the load-bearing claims for verification — caught all of them. The existing memory + brief-with-explicit-load-bearing-claims pattern is working as intended.

Implicit stubs created across the two pages (candidates for next batch):

[[ifnar1]], [[ifnar2]], [[jak1]], [[tyk2]], [[stat1]], [[stat2]], [[irf9]], [[irf7]] — ISGF3 components and JAK kinases
[[mavs]], [[rig-i-mavs-pathway]], [[tlr3-trif-pathway]] — upstream signaling
[[amlexanox]] — compound page (TBK1/IKKε dual inhibitor; already in clinical use 1997 + Phase 2 for T2D)
[[studies/bastard-2021-anti-ifn-autoantibody-age-prevalence]] — verifier-flagged companion paper to Bastard 2020; should be seeded so the corrected 1%/4%/7% prevalence attribution lands on a proper study page

[2026-05-09] verify | molecules/proteins/camp

Pages verified: 1

molecules/proteins/camp.md (corrections: 8 factual; verified: true, partial scope — see verified-scope in frontmatter)

Sources checked:

Wang 2004 (10.4049/jimmunol.173.5.2909) — not OA (no PMC; download failed); verified against PubMed abstract (PMID:15322146)
Liu 2006 (10.1126/science.1123933) — not_oa in a local paper archive; verified against PubMed abstract (PMID:16497887)
Gombart 2005 (10.1096/fj.04-3284com) — not_oa in a local paper archive; verified against PubMed abstract (PMID:15985530)
Sørensen 2001 (10.1182/blood.v97.12.3951) — local PDF downloaded and read end-to-end; cleavage mechanism verified
Zhang 2019 (10.1016/j.immuni.2018.11.003) — green OA (PMC7191997); download failed but PMC full text read via efetch; aging mechanism verified
Dhawan 2015 (10.1002/jcp.24729) — not_oa; verified against PubMed abstract (PMID:25078430)
UniProt P49913 — REST API confirmed all identity fields + feature table
GTEx v8 — API confirmed expression (gencodeId ENSG00000164047.4); age-stratified ρ not available via API
Open Targets Platform — GraphQL confirmed druggability tier 2; all Approved Drug flags = false; zero known-drugs entries
PubMed MR search — zero results for “CAMP cathelicidin Mendelian randomization”; mr-causal-evidence: not-tested confirmed

Corrections made:

Propeptide boundary in protein structure table: 31–133 → 31–131 (per UniProt P49913 feature table; verified via REST API)
Chromosomal location: 3p21.3 → 3p21.31 (per NCBI Gene 820 maplocation field)
VDRE primate-specificity: “mice lack the corresponding element” → “mice, rats, and dogs all lack the corresponding element” (Gombart 2005 abstract: comparative genomics showed absence in mouse, rat, and canine genomes)
Gombart 2005 VDRE mechanism detail added: VDRE resides in a primate-specific SINE element; murine VDR-KO and WT bone marrow have equivalent CAMP mRNA levels (no baseline VDR-driven induction in mice)
Zhang 2019 cell type precision: “dermal fibroblast-lineage cells” → “THY1hi PDGFRA+ dFB progenitors” that differentiate into antimicrobial immature adipocytes; mechanism is adipogenesis suppression, not direct fibroblast cathelicidin loss
Zhang 2019 age specificity: “aged mice” → “1-year-old C57BL/6 male mice” (explicit from PMC methods section); TGF-β isoform also specified as TGF-β1 elevation in adulthood
Dhawan 2015 C/EBP site coordinates added: binding site at −627/−619, VDRE at −615/−600; SWI/SNF complex involvement via Brm ATPase
GTEx expression populated: null → Whole Blood 68.9 TPM, Spleen 10.1, Lung 5.2 (confirmed via API with correct gencodeId ENSG00000164047.4); age-stratified ρ still needs Portal query (#gap/needs-gtex-aging-data maintained)
Open Targets druggability tier 2 confirmed (previously assigned by reasoning; now confirmed against live API data)
Footnotes updated with PMIDs, verification source notes, and enriched mechanistic detail for all 6 footnotes

Unverifiable (closed-access, abstract-only):

Wang 2004 (10.4049/jimmunol.173.5.2909) — not OA; no PMC; VDRE mechanistic claims verified against abstract only; full PDF needed for VDRE reporter assay details
Liu 2006 (10.1126/science.1123933) — not_oa; TLR-VD chain and African American vitamin D data verified against abstract only
Gombart 2005 (10.1096/fj.04-3284com) — not_oa; VDRE primate-specificity confirmed at abstract level; full comparative genomics data not read
Dhawan 2015 (10.1002/jcp.24729) — not_oa; C/EBP coordinates confirmed via abstract; full ChIP + reporter data not read

Downstream pages to check: ll-37 (cites Zhang 2019 for cathelicidin decline; corrected cell-type specificity may require propagation); disabled-adaptive-immunity (hallmark page citing cathelicidin decline)

[2026-05-09] verify | molecules/compounds/ll-37

Pages verified: 1

molecules/compounds/ll-37.md (corrections: 5 factual; verified: true, partial scope — see verified-scope in frontmatter)

Sources checked:

Sørensen 2001 (10.1182/blood.v97.12.3951) — bronze OA; PDF downloaded and read end-to-end; all mechanistic claims verified
Sørensen 2003 (10.1074/jbc.M301608200) — CC BY OA; PDF downloaded and read end-to-end; ALL-38/gastricsin claims verified
Miranda 2023 (10.1007/s00403-023-02657-8) — hybrid OA; PDF downloaded and read end-to-end; n, p-values, and outcome data verified
Zhang 2019 (10.1016/j.immuni.2018.11.003) — green OA (PMC7191997); download failed; page numbers verified via PubMed efetch + Crossref
Singh 2013 — DOI corrected via PMID 23386607; paper not available locally
Grönberg 2014 (10.1111/wrr.12211) — closed-access (not_oa); n=34 sourced from Miranda 2023 ref 21; quantitative claims not independently verified against primary PDF
PubChem CID 16198951 — REST API confirmed formula C205H340N60O53, MW 4493 Da, all match wiki

Corrections made:

Miranda 2023 n: “n=not specified” → n=25 (13 LL-37, 12 placebo) with full p-values (days 7/14/21/28: 0.031/0.009/0.006/0.037) and trial details (NCT04098562)
Zhang 2019 page numbers: “50(1):232–245” → “50(1):121–136.e5” (wrong page range in footnote; corrected via Crossref + PubMed)
Singh 2013 DOI: “10.1074/jbc.M112.442350” → “10.1074/jbc.M112.440883” (DOI had a transposition error in the final digit; the wrong DOI resolves to a cyanophage RNA polymerase paper); journal pages corrected from 5633–5645 to 8258–8268; issue from 8 to 12
Grönberg 2014 body: “n=not specified” → n=34, 4 weeks twice-weekly application (sourced from Miranda 2023’s citation; not independently verified against closed-access primary)
Footnote archive status updates: Sørensen 2001 pending→completed; Sørensen 2003 not_oa→completed (OA); Miranda 2023 not_oa→completed

Unverifiable (closed-access, download failed or not_oa):

Grönberg 2014 (10.1111/wrr.12211) — not_oa; quantitative healing-rate claims not independently verified
Henzler-Wildman 2003 (10.1021/bi034520a) — not_oa; toroidal-pore mechanism claims unverified
Porcelli 2008 (10.1021/bi702036s) — download failed; carpet-model claims unverified
Liu 2006 Science (10.1126/science.1123933) — not_oa; TLR-VD axis claims unverified
Wan 2014 (10.1189/jlb.3HI0313-168R) — not_oa; FPR2/phagocytosis claims unverified
Gombart 2005 (10.1096/fj.04-3284com) — not_oa; VDRE/primate-specificity claims unverified

Downstream pages to check: camp (cross-links to this page); disabled-adaptive-immunity (hallmark link); dysbiosis (secondary hallmark link)

[2026-05-09] verify | molecules/compounds/kpv

Pages verified: 1

molecules/compounds/kpv.md (corrections: 7 substantive; verified: true, partial scope — Kannengiesser 2008 and Catania 2000 abstract-only due to no-fulltext-access)

Sources checked:

Dalmasso 2008 (10.1053/j.gastro.2007.10.026) — bronze OA; PDF downloaded and read end-to-end
Laroui 2010 (10.1053/j.gastro.2009.11.003) — bronze OA; PDF downloaded and read end-to-end
Kannengiesser 2008 (10.1002/ibd.20334) — closed-access; verified against PubMed abstract (PMID:18092346) only
Catania 2000 (10.1111/j.1749-6632.2000.tb05387.x) — not_oa; verified against PubMed abstract (PMID:11268348) only
PubChem CID 125672 — REST API confirmed: formula C16H30N4O4, MW 342.43 Da correct

Corrections made:

NF-κB mechanism: “inhibits activation of NF-κB (by preserving IκBα)” qualified to “delays IκBα degradation and accelerates IκBα recovery” — indirect mechanism confirmed by Dalmasso 2008 western blot kinetics
MAPK targets: “p38 and ERK” → “ERK1/2, JNK, and p38” (JNK suppression confirmed in Dalmasso 2008 Fig 1D)
Dalmasso 2008 in vivo dose: “Not specified per abstract” → “100 µmol/L in drinking water, n=10/group” (stated in paper Methods)
NP polymer type (Laroui 2010): “PLGA nanoparticles” → “PLA (polylactic acid) nanoparticles” (~366 nm by PCS)
Laroui 2010 route: clarified as daily oral gavage of 150 µL hydrogel, not drinking water
Laroui 2010 dose quantification: added actual figures (~25.2 ng/day colonic delivery vs ~200 µg/day free KPV in drinking water); 12,000-fold claim confirmed in paper Discussion
Footnotes updated: PMIDs added for Kannengiesser and Catania; Cutuli 2000 companion paper noted for antimicrobial mechanism detail; “PDF pending download” removed; abstract-only verification flagged for closed-access sources

Downstream pages to check: none — kpv.md is newly seeded with no inbound wikilinks found

[2026-05-09] verify | molecules/compounds/melanotan-ii

Pages verified: 1

molecules/compounds/melanotan-ii.md (corrections: 4; verified: true, partial scope — see verified-scope)

Sources checked:

Mallory 2021 (10.1016/j.esxm.2020.100298) — gold OA; PDF downloaded and read end-to-end
Dorr 1996 (10.1016/0024-3205(96)00160-9) — closed-access (not_oa); Crossref metadata confirmed (title, journal, year, volume, pages, authors)
Habbema 2017, Burian/Jemec 2019, Gilhooley 2021, Breindahl 2015 — closed-access (not_oa); PubMed/Crossref metadata only
PubChem CID 92432 — REST API confirmed: formula C50H69N15O9, MW 1024.2 Da, InChIKey JDKLPDJLXHXHNV-MFVUMRCOSA-N all correct
ChEMBL CHEMBL430239 — resolved via PubChem CID 92432 cross-reference
NCT07437560 — ClinicalTrials.gov v2 API confirmed RECRUITING, Phase 2, vitiligo

Corrections made:

ChEMBL ID null → CHEMBL430239 (frontmatter + Identity section; needs-canonical-id removed)
Breindahl 2015 author: “Kimersgaard A” → “Kimergård A” (PubMed PMID 24771717)
Mallory 2021 footnote expanded with PDF-confirmed case details (patient 55yo male, 2 mg SC, 30h priapism, phenylephrine 4500 mcg failed, PSD, 15-wk follow-up: new ED rigidity 4/10 + corpora fibrosis; two prior cases named: Dreyer 2019 + Nelson 2012)
Body priapism text updated to name all three cases rather than “at least three”

2021 melanoma review — not located after exhaustive PubMed + Europe PMC search; unsourced retained

Downstream pages updated: none (melanotan-ii.md has zero inbound links as of 2026-05-09)

[2026-05-09] verify | molecules/compounds/semax

Pages verified: 1

molecules/compounds/semax.md (corrections: 3 substantive; verified: true, partial scope)

Sources checked:

Dmitrieva 2010 (10.1007/s10571-009-9432-0) — green OA; downloaded via EuropePMC (PMC11498467); PDF read end-to-end
Gusev 1997 (PMID 11517472) — Russian-language; abstract verified via PubMed efetch
Gusev 2005 (PMID 15792140) — Russian-language; abstract verified via PubMed efetch
Gusev 2018 (doi 10.17116/jnevro20181183261-68; PMID 29798983 confirmed) — Russian-language; abstract verified via PubMed efetch
Lebedeva 2018 (PMID 30225715) — abstract verified via PubMed efetch; n=24 confirmed
Dolotov 2006 (10.1016/j.brainres.2006.07.108) — closed-access (not_oa); claims flagged no-fulltext-access
Shadrina 2010 (10.1007/s12031-009-9270-z) — closed-access (not_oa); claims flagged no-fulltext-access
Eremin 2005 (10.1007/s11064-005-8826-8) — closed-access (not_oa); claims flagged no-fulltext-access
PubChem CID 9811102 — REST API confirmed: formula C37H51N9O10S, MW 813.9, InChIKey AFEHBIGDWIGTEH-AQRCPPRCSA-N all correct
“22% infarct-volume reduction” claim — exhaustive PubMed + Crossref search; no indexed English-language primary source found; confirmed unverifiable; gap tags retained

Corrections made:

NT-4 and p75NTR removed from Dmitrieva 2010-derived ischemic-tissue claim (paper measures BDNF, NGF, NT-3, TrkA, TrkB, TrkC only — NT-4 and p75NTR are not mentioned)
Route of administration added to Dmitrieva 2010 body claim and footnote: IP injection at 100 µg/kg (not intranasal)
n per group added to Dmitrieva 2010 footnote: n=15 Semax/pMCAO, n=17 saline/pMCAO, n=15 PGP/pMCAO; organism confirmed as adult male Wistar rat (270–320 g)
Lebedeva 2018 n added to body (n=24: 14 Semax, 10 placebo; mean age 43.9±9.5 years) and footnote updated with full author list and timepoints (baseline, 5 min, 20 min post-dose)
Gusev 2018 footnote updated: PMID 29798983 added; sex breakdown (43 male, 67 female) and mean age (58±9.7 years) added; study design detail (early vs late rehabilitation groups) added
Neurotrophin table updated with per-row source citations and no-fulltext-access tags for Dolotov 2006 and Shadrina 2010 rows
⚠️ auto-extraction banner removed

Pages unverifiable (closed-access):

Dolotov 2006, Shadrina 2010, Eremin 2005 — closed OA; claims from these sources remain with no-fulltext-access
“22% infarct-volume reduction” — confirmed no primary source indexed in English; remains unsourced no-fulltext-access

Downstream pages to check (main agent task):

None: semax.md has no downstream entity pages citing it as a study footnote as of this verification pass

[2026-05-09] verify | molecules/compounds/mk-677

Pages verified: 1

molecules/compounds/mk-677.md (corrections: 9 substantive; verified: true, partial scope)

Sources checked:

Copinschi 1997 (10.1159/000127249) — closed-access; abstract verified via Europe PMC
Nass 2008 (10.7326/0003-4819-149-9-200811040-00003) — PMC OA (PMC2757071) but DOI lookup failed; verified via PubMed efetch abstract
Adunsky 2011 (10.1016/j.archger.2010.10.004) — closed-access; abstract verified via Europe PMC
Murphy 1999 (10.1359/jbmr.1999.14.7.1182) — bronze OA but PDF download failed; abstract verified via Europe PMC
Murphy 2001 (10.1210/jcem.86.3.7294) — closed-access; abstract verified via Europe PMC
Svensson 1998 (10.1210/jcem.83.2.4539) — bronze OA but PDF download failed; abstract verified via Europe PMC
Smith 2007 (10.1196/annals.1404.023) — closed-access; abstract verified via Europe PMC
Barzilai 2012 (10.2337/db11-1300) — bronze OA; local PDF downloaded (PMC3357299) and read end-to-end
PubChem CID 178024 — REST API confirmed: InChIKey, formula C27H36N4O5S, MW 528.7 Da all correct
PCAC October 2024 vote — not independently verified; FDA advisory-committee pages returning 404; gap tag retained

Corrections made:

Copinschi 1997 sleep disambiguation confirmed and expanded: stage IV ~50% increase is young-adult finding; REM >20% increase is young-adult finding; older-adult finding is ~50% REM increase (not stage IV) — the wiki conflated two different populations. Body text and footnote now precisely distinguish young-adult (slow-wave dominant) vs older-adult (REM dominant) patterns. Study design clarified: young adults = 3-period Latin-square (5 mg, 25 mg, placebo × 7 days); older adults = 2-period sequential (2 mg × 14 days, 25 mg × 14 days), not a standard crossover.
Svensson 1998 age range: “18–40 yr” → “18–50 yr” (abstract specifies 18–50; corrected in body text and footnote)
Svensson 1998 study design: “crossover” → “parallel group” (two independent groups, not crossover; corrected in body and footnote)
Svensson 1998 cortisol: wiki listed “transient increase in cortisol” as established adverse effect without qualification; Svensson found cortisol NOT elevated (p=NS at 2 and 8 weeks); safety table and footnote now distinguish Nass 2008 (cortisol +47 nmol/L, older adults) from Svensson (no cortisol increase, obese young males)
Nass 2008 fasting glucose: “~15–20% in older adults” → “mean +0.3 mmol/L (+5 mg/dL)” (abstract gives exact quantity; percentage was fabricated); corrected in safety table and body text
Nass 2008 body text and footnote: added placebo CI (−1.1 to 0.2 kg), body weight changes, LDL change, cortisol value, body cell mass finding; added full author list
Adunsky 2011 n: “n not confirmed” → n=123 (MK-677 n=62, placebo n=61) confirmed from abstract; corrected in body text and footnote
CRITICAL — Adunsky 2011 early termination: trial terminated early due to congestive heart failure safety signal — entirely absent from original wiki page; added to Hip fracture section, Safety section, and Limitations section; footnote updated
Barzilai 2012 lifespan percentages: removed fabricated “40–70% longer” and “~16% longer” quantitative claims attributed to this review — Barzilai 2012 discusses these phenomena but does not originate the specific percentages; replaced with qualitative framing with note that specific % are in primary sources cited by the review

Downstream pages to check (by main agent):

None identified — mk-677.md has no downstream entity pages citing these specific claims in the wiki as of 2026-05-09

[2026-05-09] verify | molecules/compounds/mots-c

Pages verified: 1

molecules/compounds/mots-c.md (corrections: 12 substantive — see below)

Sources checked:

Lee 2015 (10.1016/j.cmet.2015.02.009) — local PDF in a local paper archive, read end-to-end; verified
Miller 2022 JCI (10.1172/JCI158449) — local PDF in a local paper archive, read end-to-end; verified
Reynolds 2021 (10.1038/s41467-020-20790-0) — downloaded from PMC, read end-to-end; verified
Zempo 2021 (10.18632/aging.202529) — downloaded from PMC, read end-to-end; verified
Woodhead 2021 DOI confirmed via PubMed efetch + Crossref
Yu 2021 DOI confirmed via PubMed efetch + Crossref
Zheng 2023 DOI confirmed via Europe PMC
Yoon 2022 DOI confirmed via Europe PMC
PubChem CID 146675088 — REST API confirmed; MW 2174.6 Da verified correct for 16-aa sequence MRWQEMGYIFYPRKLR

Corrections made:

CRITICAL — Zempo 2021 DOI: 10.18632/aging.202593 → 10.18632/aging.202529 (typo in last four digits; archive had different paper at wrong DOI)
CRITICAL — Zempo 2021 centenarian claim: wiki stated variant “is enriched in centenarian cohorts in Japan”; source explicitly concludes NO enrichment — C allele frequency 7.7% centenarians vs 7.5% controls (n=736); claim reversed and corrected with direct contradiction noted; contradictory-evidence tag added
CRITICAL — Zempo 2021 cohort n: “n=Japanese cohort (sedentary/active)” → “n=27,527 (J-MICC n=11,852, MEC n=3,387, TMM n=12,288)”; pooled OR 1.34 (95% CI 1.14–1.54) added; kinesio-genomic interaction quantified (65% greater T2D in lowest-activity C-allele men, p=0.01)
PubChem CID 146675088 MW resolved: wiki had flagged a “mismatch” claiming expected ~1800–1900 Da; MW 2174.6 Da is correct for MRWQEMGYIFYPRKLR (16 aa, 2 Met); seeder’s estimate was wrong. molecular-formula and molecular-weight-da frontmatter now populated; needs-canonical-id tags removed.
Reynolds 2021 human n: “human n not reported in abstract” → n=10 (healthy young male volunteers, 24.5 ± 3.7 yr); exercise increased muscle MOTS-c 11.9-fold
Reynolds 2021 mouse strain: “C57BL/6” → “C57BL/6N” (paper specifies C57BL/6N throughout)
Reynolds 2021 dose: “3x/week” alone → “15 mg/kg/day IP; 3×/week for late-life intermittent study”; acute studies used 5 and 15 mg/kg/day
Reynolds 2021 lifespan outcome: “increased physical capacity and healthspan” clarified — lifespan trended upward (median 912 vs 970 days) but did not reach significance (P=0.05 only until 31.8 months); multiple healthspan measures (grip, gait, walking, blood glucose) did improve
Lee 2015 mouse strain: HFD study uses CD-1 (outbred) not C57BL/6; aged insulin-resistance study uses C57BL/6J (12 months) at 5 mg/kg/day × 7 days
Woodhead 2021 DOI: 10.1016/j.bbagen.2021.130024 → 10.1016/j.bbagen.2021.130011 (confirmed via PubMed PMID 34520826)
Yu 2021 DOI: 10.1016/j.mito.2021.02.003 → 10.1016/j.mito.2021.02.010 (confirmed via PubMed PMID 33639272)
Zheng 2023 DOI: 10.3389/fendo.2023.1119164 → 10.3389/fendo.2023.1120533 (confirmed via Europe PMC PMID 36761202; original DOI returns 404)
Yoon 2022 DOI: 10.4093/dmj.2021.0174 (unconfirmed) → 10.4093/dmj.2022.0092 (confirmed via Europe PMC PMID 35656563)

Downstream pages to check:

None identified (mots-c.md is not yet cited by other wiki pages per inbound-link check at time of seeding)

[2026-05-09] verify | molecules/compounds/dsip

Pages verified: 1

molecules/compounds/dsip.md (corrections: 9 substantive — see below)

Sources checked:

Schoenenberger & Monnier 1977 (10.1073/pnas.74.3.1282) — local PDF downloaded from Europe PMC (PMC430668), verified end-to-end
Schoenenberger 1978 (PMID 568769, doi:10.1007/BF00581575) — PubMed efetch abstract
Bjartell 1989 (PMID 2554357, doi:10.1016/0306-4530(89)90004-8) — PubMed efetch abstract + Crossref
Graf 1984 CSF (PMID 6549071) — PubMed efetch abstract + Crossref
Graf 1984 milk (PMID 6547144) — PubMed efetch abstract
Schneider-Helmert 1981a (PMID 6895513) — PubMed efetch abstract
Schneider-Helmert & Schoenenberger 1981b (doi:10.1007/BF01971753, PMID 7028502) — Crossref + PubMed
Monti 1987 (PMID 3583493) — PubMed efetch abstract + Crossref MEDLINE
Schneider-Helmert 1987 (doi:10.1159/000116143, PMID 3622582) — PubMed efetch abstract
Bes 1992 (doi:10.1159/000118919, PMID 1299794) — PubMed efetch abstract
PubChem CID 68816 + ChEMBL CHEMBL2104403 — REST APIs confirmed

Corrections made:

CRITICAL — Schneider-Helmert 1981 paper identity: [^schneider1981] (doi:10.1007/bf01971753, Experientia) was attributed the findings of a different 1981 paper (PMID 6895513, Int J Clin Pharmacol Ther Toxicol). The Experientia paper studied 6 middle-aged chronic insomniacs; the IJCPTT paper studied 6 healthy volunteers and reported the 59% median TST increase. Split into two footnotes: [^schneider1981a] (PMID 6895513, healthy volunteers) and [^schneider1981b] (doi:10.1007/BF01971753, chronic insomniacs). Human evidence table now has 5 rows.
Bjartell 1989 journal citation: “14(6):415-422” → “14(5):347-355” (confirmed MEDLINE)
Bjartell 1989 n: “n=small (exact n not available)” → “n=11 healthy male volunteers ages 25-39”; body text updated; plasma-cortisol-unaffected nuance added
Monti 1987 journal citation: “7(3):233-239” → “7(2):105-110” (confirmed MEDLINE)
Schneider-Helmert 1987 journal citation: “26(4):228-233” → “27(2):120-129” (confirmed MEDLINE); PMID added; daytime-performance and post-treatment-night detail added from abstract
Graf 1984 CSF: volume/issue “21(6):865-868” → “21(5):761-766” (confirmed MEDLINE); DOI added
Graf 1984 milk: DOI added; body text updated to note that most milk immunoreactivity was in a form larger than the nonapeptide (DSIP itself confirmed by HPLC only)
Schoenenberger 1977: PDF now downloaded (PMC430668); footnote now includes actual quantitative results (53.9% neocortex, 39.3% limbic delta increase; p<0.01 ~10 min post-infusion; group breakdown 11/23/11/13)
hallmarks: [“altered-intercellular-communication”] → hallmarks: [] — cleared per R36 wave-1 convention; no primary-source-grade evidence links DSIP to this hallmark; consistent with TB-500/GHK-Cu/Dihexa precedent

Pages unverifiable (closed-access):

Schneider-Helmert 1981b (Experientia) — not_oa; MEDLINE-level verified
Schneider-Helmert 1987 (Eur Neurol) — not_oa; MEDLINE-level verified
Bes 1992 (Neuropsychobiology) — not_oa; MEDLINE-level verified

Downstream pages updated: none (dsip not cited by other pages as of 2026-05-09)

[2026-05-09] verify | molecules/compounds/dihexa

Pages verified: 1

molecules/compounds/dihexa.md (corrections: 8 — see below)

Sources checked:

McCoy 2013 (10.1124/jpet.112.199497) — local PDF downloaded via PMC3533412, verified end-to-end (all 14 pages)
Benoist 2014 (10.1124/jpet.114.218735) — retraction confirmed via DOI lookup (title field prefixed “RETRACTED:”); content not downloaded (retracted; do not read for claims)
Wright/Harding 2015 review (10.1016/j.pneurobio.2014.11.004) — confirmed not_oa; cannot verify; tagged no-fulltext-access in footnote
PubChem CID 129010512 — REST API confirmed: formula C27H44N4O5, MW 504.7, InChIKey XEUVNVNAVKZSPT-JTJYXVOQSA-N, IUPAC name matches page

Corrections made:

BDNF potency claim: “UNVERIFIED pending full-text review” → “NOT IN ANY SURVIVING PRIMARY SOURCE — confirmed absent from McCoy 2013 full text”; claim documented as untraceable to any non-retracted primary source
Donepezil comparison: “UNVERIFIED, pending full-text review” → “confirmed absent from McCoy 2013 full text; McCoy 2013 discussion explicitly flags this as planned future work not yet conducted”; unsourced strengthened
Scopolamine model quantitatives added: doses (i.c.v. 0.1/1 nmol; i.p. 0.05/0.25/0.50 mg/kg; oral 1.25/2.0 mg/kg), n=8–10/group, complete dose-response per route
Aged rat model quantitatives added: n=6, 24-month-old mixed-sex Sprague-Dawley, oral 2 mg/kg/day, Mann-Whitney U P<0.03
Synaptogenesis/spinogenesis quantitatives added: 10⁻¹² M, n=200 dendrite segments, ~3-fold spine increase (41 vs 15/50 µm, P<0.001); mEPSC 1.6-fold increase (4.82 vs 3.06 Hz, n=29, P<0.001); organotypic slices 11 vs 7 spines/50 µm (P<0.01)
Pharmacokinetics: corrected from “not formally published” → documented from McCoy 2013 Table 3 (serum t½=335 min; i.v. circulating t½≈12.7 days; Vd=54.4 L/kg; CL=0.0026 L/min/kg; BBB confirmed by [³H]dihexa brain-accumulation study)
Mechanism section: clarified that McCoy 2013 explicitly presents HGF/c-Met as “unpublished data” (not an established finding in that paper); mechanism never published in a surviving non-retracted primary study
Limitations bullet: “Pharmacokinetics not formally published” → corrected; gap redirected to human PK/primate data only

Pages unverifiable (closed-access):

Wright/Harding 2015 review (10.1016/j.pneurobio.2014.11.004) — not_oa; tagged in footnote

Downstream pages updated: none (dihexa.md not cited by other pages as of 2026-05-09)

[2026-05-09] verify | molecules/compounds/epitalon

Pages verified: 1

molecules/compounds/epitalon.md — corrections: 5 substantive (see below); verified-scope: partial (closed-access Khavinson Bull Exp Biol Med papers abstract-only)

Sources checked:

Al-Dulaimi 2025 (10.1007/s10522-025-10315-x) — local PDF downloaded (PMC12411320), verified end-to-end
Araj 2025 (10.3390/ijms26062691) — local PDF downloaded (PMC11943447), verified end-to-end
Yue 2022 (10.18632/aging.204007) — local PDF downloaded (PMC9037278), verified end-to-end
Khavinson 2020 (10.3390/molecules25030609) — local PDF downloaded (PMC7037223), verified end-to-end
Gatta 2025 (10.1007/s12015-025-10911-x) — local PDF downloaded (PMC12356729), downloaded but not read in full (claims secondary to other sources)
Khavinson 2003 telomerase (PMID 12937682) — abstract via PubMed efetch; not_oa
Khavinson 2004 divisions (PMID 15455129) — abstract via PubMed efetch; not_oa
Khavinson/Morozov 2003 mortality (PMID 14523363) — abstract via PubMed efetch; full PDF unavailable
Korkushko 2006 (PMID 17426848) — abstract via PubMed efetch; not_oa
PubChem CID 219042 — REST API confirmed (C14H22N4O9, MW 390.35, InChIKey HGHOBRRUMWJWCU-FXQIFTODSA-N)

Corrections made:

Al-Dulaimi 2025 cancer vs normal cell mechanism: wiki said “cancer cells showed ALT pathway activation” (correct) but missing specificity that telomerase enzyme activity was NOT significantly elevated in cancer cells (only hTERT mRNA was), while normal cells showed fourfold (IBR.3) and 26-fold (HMEC) telomerase activity increase — body and footnote updated with quantitative detail; ALT ~10-fold 21NT, ~3-fold BT474 added
Yue 2022 spindle-defect timing: wiki said “decreased spindle defects at 12–24 h” — corrected to “significant at 24h only (not at 12h per Figure 3B)”; non-monotonic dose response (1–2 mM failed to reduce ROS) added to footnote
Yue 2022 institution: “independent Chinese Academy of Sciences group” → “Shanxi Medical University + Chinese Academy of Sciences State Key Lab” (CAS is the State Key Lab affiliation, not the primary institution)
Al-Dulaimi 2025 erratum noted: doi:10.1007/s10522-025-10326-8 not previously documented
Khavinson 2003 telomerase: wiki said “human fetal pulmonary fibroblasts” — abstract confirms “human fetal fibroblast” but Araj 2025 review also identifies HeLa cells were used; footnote updated
All four closed-access Khavinson footnotes: upgraded from “abstract-only” to “abstract-verified via PubMed efetch” with confirmed PMID; tagged no-fulltext-access

Epithalamin-vs-epitalon disambiguation: CONFIRMED CORRECT by both PubMed abstracts (PMID 14523363 uses “Epithalamin”; PMID 17426848 uses “epithalamine”) and by Araj 2025 independent review which maintains the same disambiguation throughout.

Pages unverifiable (closed-access): PMID 12937682, 15455129, 14523363, 17426848 — all not_oa per a local paper archive; abstract-level verification only

Downstream pages updated: none (epitalon not yet cited by other wiki pages)

[2026-05-09] verify | molecules/compounds/ghk-cu

Pages verified: 1

molecules/compounds/ghk-cu.md (corrections: 6 — see below)

Sources checked:

Pickart 2015 (10.1155/2015/648108) — local PDF downloaded (PMC4508379), verified end-to-end
Pickart 2018 (10.3390/ijms19071987) — local PDF downloaded (PMC6073405), verified end-to-end
Maquart 1988 (10.1016/0014-5793(88)80509-x) — not_oa; concentrations verified via Crossref abstract
PubChem CID 139035031 (GHK-Cu) and 73587 (free GHK) — REST API confirmed
PMIDs 41997403, 39963574, 39795193, 37896245 — DOIs confirmed via PubMed
Pickart 1980 DOI — corrected via PubMed PMID 6246126 lookup

Corrections made:

Maquart 1988 concentration onset: “beginning at 10⁻¹² M” → “beginning between 10⁻¹² and 10⁻¹¹ M” (per Crossref abstract)
Gene modulation count: “>4000 genes” attributed to Pickart 2018 → corrected to “31.2% of human genes (~2152 at ≥50% cutoff per Table 1 in Pickart 2018); the >4000 figure is from Pickart 2014 dataset cited within both reviews”
Plasma decline source: “not locatable” → “traced to Pickart 1973 PhD thesis (UCSF) in both OA review PDFs — not a peer-reviewed analytical study”; unsourced retained
Pickart 1980 DOI: 10.1002/jcp.1041040107 (wrong paper) → 10.1002/jcp.1041020205 (confirmed PMID 6246126); volume/page corrected (104(1) → 102(2):129-139)
Four unconfirmed DOIs resolved: Hu 2026, Mortazavi 2025, Ogorek 2025, Dymek 2023; doi-unconfirmed tags removed
Dymek 2023 encapsulation efficiency: “20-32%” → “31.7% (cationic) and 20.0% (anionic)”

Pages unverifiable (closed-access):

Maquart 1988 — not_oa; abstract-only verification
Schlesinger 1977 and Pickart 1987 — not_oa; low-stakes claims not independently verified

Downstream pages updated: none

[2026-05-09] verify | molecules/compounds/tb-500.md

Pages verified: 1

molecules/compounds/tb-500.md — corrections: 4 substantive (see below)

Sources checked:

Esposito 2012 (10.1002/dta.1402) — local PDF verified end-to-end (downloaded from Ghent University OA repository)
PubChem CID 10169788 — REST API confirmed (formula C36H66N10O13, MW 847.0 Da)
NCT07487363 — ClinicalTrials.gov v2 API confirmed (Phase 1/2, RECRUITING, n=80, Hudson Biotech)
Yu 1993 (10.1016/s0021-9258(18)54179-x) — author list corrected via Crossref; PDF not downloadable

Corrections made:

Esposito 2012 author list: “Esposito S, Deventer K, Geldof L, Van Eenoo P” → “Esposito S, Deventer K, Goeman J, Van der Eycken J, Van Eenoo P” — “Geldof L” was a fabricated author; Goeman J and Van der Eycken J were omitted (confirmed from paper title page)
“Products” → single vial: Esposito 2012 studied one TB-500 vial (~10 mg, seized by Belgian Customs), not multiple “commercial products”; body and footnote corrected
Ac-LKKTETQ MW added: 888.49 Da (Esposito 2012 HRMS; theoretical 888.4911); distinct from PubChem LKKTETQ unacetylated MW 847.0 Da (correctly stated in frontmatter); identity section and footnote updated
Yu 1993 author list: last two authors “Bhatt SN, Chen LB” → “Yin HL” (Crossref: 5 authors, last is Yin HL; Bhatt/Chen not authors of this paper); archive status updated to not_oa
Rahaman 2024: all Ac-LKKTE metabolite-activity claims tagged no-fulltext-access (closed-access, no PMC); author list confirmed via Europe PMC

Pages unverifiable (closed-access): Rahaman 2024, Philp 2004, Yoon 2021, Ou 2025 — all not_oa Pages unverifiable (no OA URL): Yu 1993

Downstream pages updated: none (tb-500 not yet cited by other pages)

[2026-05-09] verify | molecules/compounds/bpc-157

Pages verified: 1

molecules/compounds/bpc-157.md (corrections: 7 — see below)

Sources checked:

Jozwiak 2025 (10.3390/ph18020185) — local PDF, verified end-to-end
Vasireddi 2025 (10.1177/15563316251355551) — local PDF, verified end-to-end
Seiwerth 2021 (10.3389/fphar.2021.627533) — local PDF, verified end-to-end
Staresinic 2022 (10.3390/biomedicines10123221) — local PDF, verified end-to-end
Lee 2025 IV (PMID 40131143) — PubMed efetch
Lee 2024 IC (PMID 39325560) — PubMed efetch
Lee 2021 knee (PMID 34324435) — PubMed efetch
NCT02637284 — ClinicalTrials.gov v2 API
NCT07437547 — ClinicalTrials.gov v2 API
WADA 2026 list — confirmed via WADA website
PubChem CID 9941957 — REST API

Corrections made:

NCT02637284 status: “Active, not recruiting” → “UNKNOWN” (confirmed via CT.gov v2 API); description updated to reflect Jozwiak’s “cancelled submission” language
WADA listing: clarified that Jozwiak 2025 erroneously stated BPC-157 “is not currently listed” — the 2026 WADA Prohibited List (effective Jan 1 2026) retains it under S0 as a Specified Substance; sports body prohibition table added from Vasireddi 2025
Krivic 2006 / Cerovecki 2010: strain corrected from unspecified to “Male Wistar rat” (per Vasireddi 2025 Table 2); n noted as unconfirmed; no-fulltext-access applied
Retrospective knee series: source clarified as Lee 2021 (PMID 34324435); 11/12 (91.6%) significant improvement vs 7/12 reporting >6-month duration distinction explained; new [^lee2021knee] footnote added
clinical-stage: phase-1 → phase-2 (NCT07437547 Phase II RCT now recruiting)
Opening paragraph: “combined n < 60” → “combined n < 30” (IV n=2, IC n=12; knee series was retrospective not a pilot); Phase I status updated
MW note: added that Jozwiak 2025 states 1419.55 Da (rounding vs PubChem 1419.5 Da)

Pages unverifiable (closed-access):

Krivic 2006 (10.1002/jor.20096) — Wiley not_oa; tagged no-fulltext-access
Cerovecki 2010 (10.1002/jor.21107) — Wiley not_oa; tagged no-fulltext-access
Seiwerth 2018 (10.2174/1381612824666180712110447) — not_oa; mechanistic claims cross-checked via secondary sources only

Downstream pages updated: none (BPC-157 compound page has no downstream entity pages citing it per current wiki state)

[2026-05-08] seed | molecules/compounds/vitamin-k2.md

Seeded by user request (ad-hoc); primary corpus was a Physionic transcript with 22 user-supplied DOIs. R25 recency search ran 2021–2026 PubMed/Europe PMC. Seeded with human-evidence-level: limited-negative and translation-gap: phase-3-rct-needed, reflecting the systematic-review/RCT picture (mostly null for arterial plaque + CAC progression at the dose-frequency-population combinations tested) vs the cohort/observational picture (favorable for K2). Verifier ran immediately on the seed and made 7 substantive corrections (see next entry); page now verified: true with partial scope.

[2026-05-08] verify | molecules/compounds/vitamin-k2.md

PDFs read in full: AVADEC (Diederichsen 2022, Circulation); Hasific 2023 (JACC Advances, AVADEC CAC substudy); de Vries 2025 (Nutrients).

Verified from abstracts only (not_oa or download unavailable): Knapen 2015, Knapen 2013, Schurgers 2007, Geleijnse 2004, Brandenburg 2017, Vermeer 2020.

Corrections made (7):

Schurgers 2007 DOI corrected: 10.1182/blood-2006-04-015123 → 10.1182/blood-2006-08-040709
MK-7 half-life specification: “~3 days” → “~68 hours (~2.8 days)” per AVADEC paper + Schurgers 2007
de Vries 2025 completely recharacterized: n=165 (not 100), dose=180 µg/d (not 360), result=positive for Young’s modulus in post-menopausal women (p=0.035), not null; paper is a post-hoc analysis of a 243-subject trial, not a standalone RCT
Naiyarakseree 2023 corrected: n=96 (not 159), 375 µg/d (not 360), 24 weeks (not 12 months), open-label
Geleijnse 2004 framing corrected: “RR ~0.43 per SD increase” → “RR 0.43 upper vs lower tertile comparison”
Hasific 2023 characterized as partially positive: primary endpoint p=0.089 (trending), CAC≥400 AU subgroup p=0.047 (significant), safety events fewer (p=0.048) — not simply “null”
Dose-response table revised: 180 µg/d row now cites de Vries 2025 as confirmatory positive; 360 µg/d row de-cited de Vries 2025 (wrong dose); 720 µg/d notes Hasific 2023 subgroup signal

Supersession check (2023–2026): Additional AVADEC substudy found — Hasific et al. 2025 (Atherosclerosis, doi:10.1016/j.atherosclerosis.2025.120540, n=388 men, 720 µg/d K2 + 25 µg D3 × 24 months): no effect on epicardial adipose tissue volume/attenuation, PCAT attenuation, or systemic inflammation — consistent with AVADEC main result. Not a supersession of existing framing; extends the null-for-hard-endpoint picture.

Downstream pages that may need updates: any studies/ pages for the above trials (if seeded).

[2026-05-08] ingest | GLP-1 RAs in osteoarthritis (Meurot 2022 + Qin 2026)

Ad-hoc ingest of two user-supplied papers extending the GLP-1-agonist class evidence base into a new aging-relevant indication (osteoarthritis):

Meurot et al. 2022 (Sci Rep 12:1567, doi:10.1038/s41598-022-05323-7) — IA liraglutide in MIA mouse OA; first GLP-1R protein-confirmation in human OA cartilage + synovium; full PDF read end-to-end (locally archived; download_status=completed); page seeded verified=true.
Qin et al. 2026 (Cell Metab 38:582–597, doi:10.1016/j.cmet.2026.01.008, PMID 41666927) — systemic semaglutide in obesity-OA mice; weight-loss-independent chondroprotection via GLP-1R-AMPK-PFKFB3 axis (chondrocyte glycolysis→OXPHOS); pilot RCT ChiCTR2200066291. PDF NOT yet in a local paper archive (DOI not ingested by OpenAlex pipeline); page seeded verified=false with abstract-only sourcing flagged in the banner. Retry DOI lookup in 2–4 weeks.

Pages added

studies/meurot-2022-liraglutide-oa.md — verified=true (PDF cross-checked: IC50/EC50 values, mouse n’s, p-values, exendin-9-39 reversal, COI disclosure)
studies/qin-2026-semaglutide-oa.md — verified=false (abstract + MeSH/keywords from PubMed efetch only)
phenotypes/osteoarthritis.md — verified=false (banner-flagged); minimal seed page to support cross-link target — was previously referenced from senolytics, fisetin, anakinra, but had no canonical home

Pages updated

molecules/compounds/semaglutide.md — new “Cartilage protection in osteoarthritis” section under Geroprotective angle; added mitochondrial-dysfunction hallmark + AMPK/PFKFB3 targets; cross-reference to osteoarthritis; verified-scope updated to flag Qin 2026 abstract-only sourcing
interventions/pharmacological/glp1-agonists.md — new “Joint and cartilage” section consolidating Meurot 2022 + Qin 2026 + Gudbergsen 2021 negative-systemic counterpoint; new hallmark mapping table for OA application; cross-references to osteoarthritis and both new study pages

Gaps surfaced

contradictory-evidence — systemic liraglutide failed in OA pain (Gudbergsen 2021) but systemic semaglutide reported chondroprotective in mice (Qin 2026); resolution requires powered systemic-semaglutide OA RCT
needs-pdf-verification — Qin 2026 mouse n’s, dose, AMPK/PFKFB3 mechanism specifics (genetic vs pharmacologic), OXPHOS measurement modality, ChiCTR2200066291 design
needs-replication — Meurot 2022 has significant commercial COI (4P-Pharma + 4Moving Biotech CEOs hold OA-GLP1 patents); independent IA-liraglutide work essential
liraglutide compound page is still a stub — should be expanded with the Meurot 2022 IA-OA findings as the joint/cartilage anchor section (deferred — not in scope for this ad-hoc ingest)
Gudbergsen 2021 primary source not yet retrieved; n=156 + null-result figures derived from Meurot 2022 narrative — surface for verification pass

Verification artifact summary

1 study page verified=true (Meurot 2022 — full PDF cross-check)
1 study page verified=false (Qin 2026 — abstract-only, PDF pending archive ingest)
1 phenotype page verified=false (osteoarthritis — banner-flagged minimal seed)
2 atomic-content pages updated with verified-scope extended to document the abstract-only sourcing for Qin 2026 (semaglutide, glp1-agonists)

[2026-05-08] resolution | tier-1 → tier-2 re-rating (6 protein pages)

User directed re-rating of the 6 tier-1 protein pages flagged in the R34 batch-1 tier-1 audit. All 6 flipped from druggability-tier: 1 → druggability-tier: 2:

alpha-synuclein.md — prasinezumab/cinpanemab Phase 2 immunotherapy failures; no FDA-approved α-syn-targeted agent
caspase-1.md — VX-740/VX-765 discontinued; no direct caspase-1 inhibitor in clinical development
hsc70.md — VER-155008/MAL3-101 are tool compounds; oncology-stage HSP70-family inhibitors are not aging-validated
nampt.md — FK866/APO866 oncology-only Phase 1/2; dose-limited by thrombocytopenia/GI; no NAMPT activator clinical-stage
pdk1.md — OSU-03012/BX795/GSK2334470 are preclinical/early-clinical oncology probes
s6k1.md — no S6K1-selective drug; clinical access entirely upstream via rapamycin/mTORC1

For each page: (1) frontmatter druggability-tier: field flipped; (2) “Aging-context tier-1 rationale” body section rewritten as “Druggability — tier-2 (re-rated 2026-05-08)” — the prior justification is preserved as historical context but reframed as the reason tier-1 was not defensible; (3) #gap/contradictory-evidence on tier assignment tag removed; (4) “Reviewer note: this rating could be reconsidered as tier-2…” sentence removed.

The aging-relevance of each protein is unchanged — pathway centrality (S6K1 in the mTORC1 axis, NAMPT in the NAD+ salvage pathway, HSC70 in CMA, etc.) is preserved on each page; the change is strictly to druggability-tier: from “clinical drug exists for an aging-context-relevant indication” (1) to “high-quality probe” (2).

Tier-1 protein pages now: 38 (was 44; 6 demoted). Re-rating cascades to Dataview druggability queries on frameworks/interventions-by-hallmark.md and any other tier-1-filtered framework pages.

[2026-05-08] R34 backfill | bucket cleared (59/59 pages stamped)

The R34 literature-checked-through: backfill bucket — 59 pre-R25-schema pages of type: compound / type: intervention / type: biomarker / type: hypothesis with NULL on the schema-required field — is now fully cleared. Verification: 0 pages remaining in the null bucket; 72 total pages of these types now carry the field (counts grew slightly from new-page seeding during the work).

Round structure

Sub-batch 0 (user-driven): rapamycin RAPA-EX-01 incorporation
Batch 1 (5 parallel agents): UniProt drift sweep + tier-1 rationale + caloric-restriction + metformin + NAD-precursor cluster — 5 pages cleared, 1 UniProt drift fixed (p21.hgnc), 15 tier-1 rationale notes added, 6 pages flagged for re-rating
Batch 2 (5 parallel agents): D+Q senolytics + BCL-xL senolytics class + mitophagy/autophagy mimetics + reprogramming + canakinumab — 12 pages cleared
Batch 3 (5 parallel agents): lifestyle cluster + pharmacological classes + stem-cell + biomarkers + hypotheses — 31 pages cleared
Batch 4 (2 parallel agents): mitochondrial/CR-mimetic compounds + GLP-1 niche cluster — 9 pages cleared
Three propagation passes interleaved between batches

Final tally (59/59)

Compounds (20): rapamycin, metformin, nmn, nr, dasatinib, quercetin, fisetin, navitoclax, a1331852, urolithin-a, spermidine, canakinumab, creatine, taurine, egcg, elamipretide, semaglutide, empagliflozin, lonafarnib, paroxetine ✅
Interventions (24): caloric-restriction, nad-precursors, senolytics, aav-osk, aav-tert, in-vivo-partial-reprogramming-therapy, ketogenic-diet, methionine-restriction, time-restricted-eating, heat-exposure, exercise, sleep, ampk-activators, sirtuin-activators, senomorphics, HSCT, iPSC-derived-cell-therapy, mesenchymal-stem-cell-therapy, AAV-follistatin, AAV-klotho, CRISPR-base-editing-pcsk9, plasma-exchange, mtor-inhibitors, glp1-agonists ✅
Hypotheses (7): hyperfunction-theory, antagonistic-pleiotropy, disposable-soma-theory, free-radical-theory-of-aging, information-theory-of-aging, mitohormesis, negligible-senescence ✅
Biomarkers (8): horvath-clock-2013, hannum-clock-2013, phenoage-2018, grimage-2019, dunedinpace-2022, lehallier-proteomic-clock-2019, telomere-length-leukocyte, frailty-index ✅

Major findings landing across the round (selected)

Rapamycin axis (positive + negative both consolidated)

Stanfield 2026 RAPA-EX-01 (n=40): cycling hypothesis null/negative on chair-stand (PP Cohen’s d −0.90, p=0.007 favoring placebo); second consecutive null aging-rejuvenation rapamycin RCT
Ivimey-Cook 2025 Aging Cell vertebrate meta-analysis: rapamycin replicates DR-magnitude lifespan extension across 911 effect sizes / 167 papers / 8 species; metformin does not

Senolytic axis (mixed)

Klier 2025 NEJM Evidence BEHOLD: UBX1325 / foselutoclax intravitreal +5.6 ETDRS letters in DME — first positive Phase 2 primary endpoint in BCL-xL-axis senolytic class
Farr 2024 Nat Med: first placebo-controlled D+Q RCT (n=60) primary endpoint missed; exploratory subgroup signals robust
Lombardo 2026 PNAS: D+Q corpus-callosum demyelination in mice — emerging safety signal

Inflammaging axis (mixed)

CANOPY-A/-1/-2/-N: all four canakinumab NSCLC trials negative; CANTOS exploratory cancer signal does NOT replicate in therapeutic-setting NSCLC
Cardiovascular MACE finding intact + replicated in He 2025 25-RCT meta-analysis (RR 0.86)

Reprogramming/gene-therapy axis (positive)

AAV-TERT modTERT paradigm shift: NCT05837143 MERCURY-DCM Phase 1 ACTIVE since 2023 with catalytically-inactive nuclear-retained modTERT — first regulated human AAV-TERT trial
Tabar 2025 Nature: bemdaneprocel iPSC PD Phase 1 (n=12) MDS-UPDRS-III −23 at 18 mo
Berdugo-Vega 2026 Neuron (Gräff lab EPFL): non-Sinclair OSK partial-reprogramming of engram cells — partially mitigates single-lab-dominance concern

GLP-1 axis (negative AD signal)

Cummings 2026 Lancet EVOKE/EVOKE+ (n=3,808): both AD trials primary endpoint failed (CDR-SB Δ −0.08 / 0.10), discontinued. Largest aging-rejuvenation Phase 3 GLP-1 endpoint failure to date.

Hypothesis layer (one major, multiple modest)

Sanchez 2026 Nature: cardiac Foxo1/Trim63 protective-in-young → pathology-in-aged: cleanest single-gene antagonistic-pleiotropy demonstration since Tyner 2002 p53
Hukkanen 2026 Nat Commun Finnish Twin Cohort: direct human population-level disposable-soma support
Chiavacci 2026 Aging Cell Greenland shark: cardiac aging hallmarks present despite ~300-yr lifespan (supports “resilience-not-absence” reframing for negligible senescence)

Biomarker axis (mostly null)

METFORAGING (first metformin × epigenetic-clocks RCT, n=40-60): all 11 clocks NS — biomarker-level confirmation of Ivimey-Cook 2025 metformin null
Argentieri 2024 Nat Med ProtAge (UK Biobank 204-protein Olink): effectively supplants Lehallier 2019 as canonical proteomic clock
RAPA-EX-01 PCGrimAge + DunedinPACE NS as secondary outcomes (cross-linked)

Frontmatter rating changes during the round

caloric-restriction: human-evidence-level: limited → strong (CALERIE-2 ancillary analyses convergent)
spermidine: human-evidence-level: limited → limited-negative (SmartAge 2022 null + CATIS 2025 contradictory)
elamipretide: clinical-stage: phase-3 → fda-approved (Forzinity Sept 2025 Barth syndrome)
aav-tert: human-evidence-level: none → limited; clinical-stage: preclinical → phase-1 (MERCURY-DCM)
in-vivo-partial-reprogramming-therapy: clinical-stage: preclinical → phase-1

Final propagation pass (this entry)

phenotypes/alzheimers-disease.md — therapeutic table EVOKE/EVOKE+ row updated Phase 3 ongoing → Phase 3 NEGATIVE with Cummings 2026 numerics
hallmarks/mitochondrial-dysfunction.md — therapeutic-angles section gained “Mitochondria-targeted peptide (elamipretide / SS-31)” subsection covering Forzinity FDA approval + Schauer 2026 respiration-vs-function dissociation

R34 backlog status update

R34-surfaced backlog	Status
59-page `literature-checked-through` backfill	CLEARED ✅
Tier-1 protein aging-context rationale (16 pages)	CLEARED ✅ — 15 edits + 1 no-op (parp1) + 6 flagged for re-rating
Full UniProt drift sweep (177 proteins)	CLEARED ✅ — 13 drifts caught + fixed

The three R34-surfaced backlogs are now all cleared. Six tier-1 protein pages remain flagged for re-rating decision (alpha-synuclein, caspase-1, hsc70, nampt, pdk1, s6k1) — held for user.

Forward queue surfaced during the round (next-rounds candidates)

Study pages for: Stanfield 2026 RAPA-EX-01 (created), Cummings 2026 EVOKE/EVOKE+, Klier 2025 BEHOLD, Farr 2024 D+Q-bone, Tabar 2025 bemdaneprocel, Sanchez 2026 cardiac AP, Berdugo-Vega 2026 OSK-engram, Schloesser 2026 NA-conversion mechanism, Mitchell 2026 chemical-reprogramming-negative, Argentieri 2024 ProtAge → biomarker-page promotion candidate
Standing watchlists: Altintas/MacArthur 2026 MK-8722 preprint → peer-reviewed publication; AMBAR Phase 3 confirmatory trial activation; VERVE-101 heart-1 Phase 1 full publication

[2026-05-08] propagation | post-batch-2 propagation pass (R34 backfills)

Cross-page propagation after the second parallel batch of 5 R34-backfill agents. Five major findings ripened:

CANTOS cancer signal does NOT replicate (canakinumab agent finding)

All four prospective NSCLC trials (CANOPY-A/-1/-2/-N) reported negative primary endpoints. Wiki claim moved from “needs replication” → “does not replicate in therapeutic-setting NSCLC.” Cardiovascular MACE finding intact + replicated in He 2025 25-RCT meta-analysis. Propagated to hallmarks/chronic-inflammation.md.

AAV-TERT modTERT paradigm shift (reprogramming agent finding)

NCT05837143 (MERCURY-DCM, Juvensis JV001 = AAV9-modhTERT^Y707F,D868A^) ACTIVE since 2023 in DCM (n=12). Catalytically-inactive nuclear-retained modTERT engineers around cancer-permissivity. Mechanism papers Zhao 2026 + Xie 2026. Propagated to molecules/proteins/tert.md and hallmarks/telomere-attrition.md.

UBX1325 / foselutoclax — first positive BCL-xL Phase 2 (BCL-xL agent finding)

Klier 2025 NEJM Evidence BEHOLD Phase 2 (n=65, intravitreal DME): +5.6 ETDRS letters. First positive primary-endpoint Phase 2 in the BCL-xL-axis senolytic class. Local delivery sidesteps thrombocytopenia. Propagated to hallmarks/cellular-senescence.md and molecules/proteins/bcl-xl.md.

D+Q Farr 2024 placebo-controlled null + Lombardo 2026 demyelination

First placebo-controlled D+Q RCT (Farr 2024 Nat Med n=60) primary missed (p=0.611). Lombardo 2026 PNAS — corpus callosum demyelination in mice via OPC UPR. Propagated to hallmarks/cellular-senescence.md.

Spermidine `limited → limited-negative`

SmartAge 2022 null cognition primary; CATIS 2025 contradictory direction. Propagated to interventions/lifestyle/caloric-restriction.md.

R34 backfill bucket progress

Batch 1: 5 cleared (rapamycin, caloric-restriction, metformin, nmn, nr)
Batch 2: 12 cleared (dasatinib, quercetin, fisetin, navitoclax, a1331852, senolytics, urolithin-a, spermidine, aav-osk, aav-tert, in-vivo-partial-reprogramming-therapy, canakinumab)
17/59 cleared total. 42 remaining.

[2026-05-08] propagation | post-batch-1 propagation pass (R34 backfills)

Cross-page propagation after the first parallel batch of 5 R34-backfill agents (UniProt drift sweep + tier-1 rationale + caloric-restriction recency + metformin recency+R26c + NAD-precursor cluster). Agents wrote their own logs separately. This entry captures the cross-cluster propagation that the main agent did after batch completion.

TAME / NCT correction propagation

The metformin agent discovered NCT02432287 = MILES (n=16, COMPLETED 2017), NOT TAME — and that the TAME 3,000-person trial does not currently have a ClinicalTrials.gov registration. Spot-checked two other NCT numbers cited as TAME elsewhere in the wiki: NCT03138915 = unrelated radiomics study (CHESS1701), NCT04977829 = invalid (no CT.gov record). All three were independently fabricated/mis-cited.

Propagation edits:

interventions/pharmacological/ampk-activators.md — next-experiment: field corrected
pathways/ampk.md — TAME paragraph rewritten; speculative timelines flagged
phenotypes/frailty.md — NCT03138915 mis-cite removed
model-organisms/homo-sapiens.md — NCT04977829 mis-cite removed (table row + footnote)

Lower-priority generic “TAME trial in progress” wording on 12 other pages left as-is — not factually wrong, just imprecise about registration status. Metformin page is the canonical detailed status reference.

Caloric-restriction `human-evidence-level: limited → strong` upgrade

Three downstream pages mention “limited human CR data” but they refer to specific mechanism-level subdomains (liver lipophagy CR data, mtDNA/TFAM preservation under CR) where CR data is independently limited. The CALERIE-2 ancillary analyses upgrade the global aging-biomarker evidence base, not these mechanism-level claims. No edits made — both states co-exist correctly.

Ivimey-Cook 2025 meta-analysis propagation

Vertebrate meta-analysis (911 effect sizes / 167 papers / 8 species): rapamycin replicates DR-magnitude lifespan extension; metformin does not. Propagated to:

molecules/compounds/rapamycin.md — new paragraph in Mammalian lifespan data section + footnote
hypotheses/hyperfunction-theory.md — new subsection + footnote

The Ivimey-Cook + RAPA-EX-01 combination on rapamycin’s page is informative: cross-vertebrate lifespan signal robust and matches DR; human aging-rejuvenation RCTs at 5–10 mg/week intermittent dose null/negative.

Six tier-1 protein pages flagged for re-rating

Tier-1 rationale agent flagged 6/16 audited pages where tier-1 designation may not be defensible: alpha-synuclein, caspase-1, hsc70, nampt, pdk1, s6k1. Each page has an in-body “Reviewer note” + #gap/contradictory-evidence on tier assignment. No re-rating applied — held for user decision.

R34 backfill bucket progress

1/59 in first sub-batch (rapamycin, user-driven)
4 more in parallel batch-1 (caloric-restriction, metformin, nmn, nr) + 1 already-clear (nad-precursors as class page checked)
5/59 cleared total. 54 remaining.

[2026-05-08] R34 | lint cleanup + framework wiring (housekeeping)

Small-batch infra round per ROADMAP plan. Items 1–6 (concrete fixes) completed; items 7–9 (broad sweeps) executed at survey/scaffolding scope with deeper backlogs surfaced.

Concrete fixes:

Wired 6 framework MOCs into index.md under three headings (Co-equal MOCs, Cross-cutting intervention slicers, Biomarker layer): hallmark-causality-graph, causal-graph-data, interventions-by-modality, intervention-classes, biological-age-measurement. (interventions-by-hallmark and intervention-by-target-immunogenicity were already wired pre-R34.)
telomere-attrition.causes self-loop — already fixed in an earlier round per log entry 1756; no action needed.
Wired orphan biomarker pages: hallmarks/telomere-attrition.md now points to [[biomarkers/telomere-length-leukocyte]] at the top of its biomarker section; phenotypes/frailty.md now points to [[biomarkers/frailty-index]] from its assessment-instruments table.
Cross-referenced empagliflozin from hallmarks/deregulated-nutrient-sensing.md (intervention table SGLT2 row expanded with mechanism + trial inventory + verified-page link), hallmarks/chronic-inflammation.md (new section “SGLT2 inhibitors — caloric-restriction-mimetic + anti-inflammatory dual track”), and interventions/pharmacological/glp1-agonists.md cross-references (mechanistic-comparison paragraph). One in-progress edit on chronic-inflammation accidentally inserted an unsourced “NIA ITP +6.4–10% male mouse lifespan” claim attributed to empagliflozin — caught and removed before final state. Ship-discipline reminder: when expanding hallmark tables with cross-refs, only assert claims that exist on the linked atomic page; don’t invent supporting numerics.
Cross-referenced lamp2 from processes/chaperone-mediated-autophagy.md: replaced 2 wikilinks of the form [[lamp-2a]] (which resolved by alias only) with the canonical-filename form [[lamp2|LAMP-2A]]; updated cross-references list to remove the outdated (stub) qualifier.
Updated lint script in sops/lint-pass.md:
- Replaced bash broken-link checker with an alias-aware Python script that builds a slug → file map from filenames AND frontmatter aliases: arrays before checking each [[link]]. Resolves the long-standing R10e+R26e false-positive bucket ([[inflammaging]] → chronic-inflammation, [[bnip3l]] → nix, etc.).
- Updated the orphan-pages bash check to also match [[<dir>/basename]] form, so studies/cuervo-2000-cma-aging.md is no longer false-flagged as orphan when other pages link to it via the path-prefixed form.
- Added a UniProt drift-sweep section with a working Python script (gene-level ENSG extraction via first_ensg helper). Sample run on 10 high-traffic proteins surfaced one real drift (p21.hgnc was CDKN1A symbol instead of numeric 1784); fixed in-line.

Broad sweeps (surveys + backlogs surfaced):

UniProt drift sweep — full ~177-protein sweep deferred to a dedicated round; sample of 10 found 1 fixable drift (p21.hgnc) and confirmed no accession-level drift. Drift script now lives in sops/lint-pass.md for future use.
Recency refresh per R25 schema — 0 pages currently exceed the type-specific cadence (compound/intervention >12mo, biomarker >18mo, hypothesis/hallmark/model-organism >24mo, process >18mo). However, 59 pages have NULL on a literature-checked-through field where the schema requires populating it (8 biomarkers, 7 hypotheses, 24 interventions, 20 compounds). This is a schema-backfill backlog from the R25-pre-existing pages. Surfaced; populating all 59 via PubMed searches per page is a multi-round effort that should be batched separately (probably 8–12 per round).
Open Targets tier-1 audit — 44 protein pages currently carry druggability-tier: 1. Spot-checked the most-cited candidates (p53, parp1, alpha-synuclein, mdm2, igf-1, p21): only PARP1 has explicit aging-context rationale documented in body prose. The other tier-1 protein pages tend to be defensible (most have an FDA-approved or clinical-stage drug for an age-related indication: cancer for p53/MDM2/PARP1; PD for α-synuclein; thyroid eye disease / acromegaly for IGF-1) but the rationale isn’t always written down. Backlog surfaced: per-page rationale documentation for the 16 tier-1 protein pages without an obvious aging-context note. Defer to a dedicated round.

Files touched (concrete fixes only; surveys did not touch pages):

index.md (framework MOC wiring)
hallmarks/telomere-attrition.md (biomarker pointer)
hallmarks/deregulated-nutrient-sensing.md (empagliflozin xref)
hallmarks/chronic-inflammation.md (empagliflozin xref)
phenotypes/frailty.md (frailty-index pointer)
interventions/pharmacological/glp1-agonists.md (empagliflozin xref)
processes/chaperone-mediated-autophagy.md (lamp2 canonical link)
molecules/proteins/p21.md (hgnc field: CDKN1A → 1784)
sops/lint-pass.md (alias-aware broken-link script + orphan path-form fix + UniProt drift script)

Backlogs surfaced (added to ROADMAP):

59-page literature-checked-through backfill bucket (compound/intervention/biomarker/hypothesis types only; sorted by inbound count for prioritization).
16-page tier-1 protein aging-context rationale backlog (the not-OK column from the spot-check).
Full 177-protein UniProt drift sweep (script ready; runtime ~3 min + per-drift verification cost).

[2026-05-08] propagation | tier-1 aging-context rationale (R34 backlog)

Scope: R34 spot-check identified 16 type: protein pages with druggability-tier: 1 that lacked explicit aging-context rationale prose per the IRAK4 (molecules/proteins/irak4.md) + PARP1 (molecules/proteins/parp1.md) precedent (R27 schema). Documentation-only audit — no new claims, no PubMed searches, no PDF reads.

Result summary (16 pages):

parp1.md — no-op (R34 grep was a false positive; “cancer approvals” rationale already documented at line 111)
gpr109a.md — edited; pre-existing brief rationale prose at line 113 was upgraded to the standard “Aging-context tier-1 rationale” block format with niacin (FDA-approved for dyslipidemia, AIM-HIGH/HPS2-THRIVE failed) framing
alpha-synuclein.md — edited; rationale block added; flagged for re-rating consideration (prasinezumab/cinpanemab Phase 2 failures; tier-1 borderline)
asm.md — edited; olipudase alfa (FDA 2022 for ASMD/Niemann-Pick) + FIASMA-class indirect inhibitors (paroxetine et al.) framing
bcl-xl.md — edited; navitoclax Phase 1/2 (oncology) + venetoclax FDA-approved (CLL/AML) BCL-2-selective congener; senolytic-tractable rationale
caspase-1.md — edited; flagged for re-rating consideration (VX-740/VX-765 discontinued; rating defended via canakinumab/CANTOS upstream-axis validation; contradictory-evidence on tier assignment tagged)
hsc70.md — edited; flagged for re-rating consideration (no clinical-stage HSC70 selective drug; rating defended via Open Targets oncology-class probes + CMA-pathway centrality; tag added)
igf-1.md — edited; mecasermin (FDA-approved Laron syndrome) + teprotumumab (anti-IGF1R, FDA-approved thyroid eye disease 2020) + ganitumab (oncology) framing; lifespan-genetic robustness anchors
il-18.md — edited; tadekinig alfa (Phase 3 NLRC4-MAS, Phase 2 AOSD) framing; AMD/CV-inflammaging anchors
il-1b.md — edited; canakinumab (FDA-approved CAPS/SJIA/gout) + CANTOS Phase 3 RCT framing — strongest human-evidence anchor of any tier-1 protein on the wiki
mdm2.md — edited; nutlin-class clinical-stage (idasanutlin, AMG-232) for AML/liposarcoma; antagonistic-pleiotropy framing
myostatin.md — edited; bimagrumab/trevogrumab/domagrozumab/apitegromab Phase 2/3 across DMD/SMA/disuse/T2D-obesity (no FDA approvals); sarcopenia mechanism centrality
nampt.md — edited; flagged for re-rating consideration (FK866/APO866 oncology Phase 1/2 only, dose-limited; no NAMPT activator clinical; rating defended via NAD+-axis rate-limiting role + adjacent NMN/NR clinical activity; contradictory-evidence tag added)
p53.md — edited; indirect via MDM2 inhibitors (nutlins) framing; antagonistic-pleiotropy aging trade-off centrality
pdk1.md — edited; flagged for re-rating consideration (no FDA-approved or late-stage PDK1 inhibitor; rating defended via biochemical/structural tractability + IIS/mTOR axis position; tag added)
s6k1.md — edited; flagged for re-rating consideration (no S6K1-selective drug; rating defended via Selman 2009 KO lifespan + rapamycin/rapalogs upstream FDA-approved; tag added)

Total edits: 14 pages (parp1 was a confirmed no-op; the 15 remaining all received new aging-context rationale blocks following the IRAK4/PARP1 prose precedent).

Pages flagged for re-rating to tier-2 (6): alpha-synuclein, caspase-1, hsc70, nampt, pdk1, s6k1. Each carries an in-body “Reviewer note” sentence and a #gap/contradictory-evidence on tier assignment tag. The user should decide whether to re-rate; the tier-1 rating is defensible via the documented aging-relevant biology even where strict Open Targets Approved Drug = true is absent, but five of these have NO FDA-approved or late-stage clinical drug at all (vs the cleaner cases like BCL-xL where navitoclax + venetoclax are clinical-stage and FDA-approved respectively for non-aging indications).

Pages where the rationale was straightforward (8): asm, bcl-xl, igf-1, il-18, il-1b, mdm2, myostatin, p53, gpr109a — all have FDA-approved drugs or active Phase 2/3 programs for non-aging indications that engage aging-relevant biology.

Cross-link audit: no new wikilinks created; the rationale blocks reference existing pages (loss-of-proteostasis, stem-cell-exhaustion, chronic-inflammation, caloric-restriction, etc.). contradictory-evidence tag count increased by 6 (one per re-rate-flagged page) — the lint pass should pick these up under the standing #gap/contradictory-evidence aggregation.

No new claims, no new citations. All edits are documentation-only and carry no new biological claims; no new footnotes were added. The prose strictly references drugs already cited in the body of each page.

🧬 Aging Wiki

Explorer

R34

log/R34.md — Round 34 entries

[2026-05-13] verify | molecules/proteins/irf9.md

[2026-05-13] verify | molecules/proteins/stat2.md

[2026-05-13] verify | molecules/proteins/ifnar1.md

[2026-05-13] verify | molecules/proteins/irf7.md

[2026-05-13] verify | pathways/tlr3-trif-pathway.md

[2026-05-13] verify | pathways/rig-i-mavs-pathway.md

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

[2026-05-13] verify | molecules/proteins/stat1.md

[2026-05-13] verify | molecules/proteins/jak1.md

[2026-05-13] verify | molecules/proteins/tyk2.md

[2026-05-13] verify | molecules/proteins/ifnar2.md

[2026-05-13] schema | R34 mouse-ortholog formalized + R35 type:pathway extensions

[2026-05-09] verify | molecules/proteins/camp

[2026-05-09] verify | molecules/compounds/ll-37

[2026-05-09] verify | molecules/compounds/kpv

[2026-05-09] verify | molecules/compounds/melanotan-ii

[2026-05-09] verify | molecules/compounds/semax

[2026-05-09] verify | molecules/compounds/mk-677

[2026-05-09] verify | molecules/compounds/mots-c

[2026-05-09] verify | molecules/compounds/dsip

[2026-05-09] verify | molecules/compounds/dihexa

[2026-05-09] verify | molecules/compounds/epitalon

[2026-05-09] verify | molecules/compounds/ghk-cu

[2026-05-09] verify | molecules/compounds/tb-500.md

[2026-05-09] verify | molecules/compounds/bpc-157

[2026-05-08] seed | molecules/compounds/vitamin-k2.md

[2026-05-08] verify | molecules/compounds/vitamin-k2.md

[2026-05-08] ingest | GLP-1 RAs in osteoarthritis (Meurot 2022 + Qin 2026)

Pages added

Pages updated

Gaps surfaced

Verification artifact summary

[2026-05-08] resolution | tier-1 → tier-2 re-rating (6 protein pages)

[2026-05-08] R34 backfill | bucket cleared (59/59 pages stamped)

Round structure

Final tally (59/59)

Major findings landing across the round (selected)

Final propagation pass (this entry)

R34 backlog status update

Forward queue surfaced during the round (next-rounds candidates)

[2026-05-08] propagation | post-batch-2 propagation pass (R34 backfills)

CANTOS cancer signal does NOT replicate (canakinumab agent finding)

AAV-TERT modTERT paradigm shift (reprogramming agent finding)

UBX1325 / foselutoclax — first positive BCL-xL Phase 2 (BCL-xL agent finding)

D+Q Farr 2024 placebo-controlled null + Lombardo 2026 demyelination

Spermidine limited → limited-negative

R34 backfill bucket progress

[2026-05-08] propagation | post-batch-1 propagation pass (R34 backfills)

TAME / NCT correction propagation

Caloric-restriction human-evidence-level: limited → strong upgrade

Ivimey-Cook 2025 meta-analysis propagation

Six tier-1 protein pages flagged for re-rating

R34 backfill bucket progress

[2026-05-08] R34 | lint cleanup + framework wiring (housekeeping)

[2026-05-08] propagation | tier-1 aging-context rationale (R34 backlog)

Graph View

Table of Contents

Backlinks

Spermidine `limited → limited-negative`

Caloric-restriction `human-evidence-level: limited → strong` upgrade