🧬 Aging Wiki

R44

43 min read

log/R44.md — Round 44 entries

Sub-file of log — see parent for index.

[2026-05-19] R44 close + R38–R44 CAMPAIGN CLOSE — chemical peels + hormonal + AGE-breakers + resurfacing + skin microbiome (11 pages); 7-round skin-aging coverage campaign complete

Final round of the R38–R44 skin-aging coverage campaign. 11 atomic pages seeded + verified single-day. 22 agent invocations (11 seeders + 11 verifiers) + 1 propagation pass. All 11 pages flipped to verified: true (most with partial scope for closed-access primaries).

Pages added (all 11)

Chemical peels (4): molecules/compounds/{glycolic-acid, lactic-acid, salicylic-acid}.md + interventions/pharmacological/chemical-peels.md (class MOC)

Hormonal (1): interventions/pharmacological/topical-estrogens.md

AGE biology (1): interventions/pharmacological/age-crosslink-breakers.md

Dermatologic procedures (1): interventions/pharmacological/dermatologic-resurfacing.md (class MOC; lasers + microneedling + IPL + RF + HIFU)

Skin microbiome (4): microbiome/{cutibacterium-acnes, staphylococcus-epidermidis, malassezia, skin-microbiome-aging-shifts}.md

Major verifier findings

R44 verifier yield was exceptionally high — multiple critical canonical-DB hallucinations + study-design fabrications caught despite the feedback_canonical_db_memory_unreliable discipline. Total ~85 corrections across the 11-page batch.

Critical canonical-DB hallucinations (continuing the R40/R41 pattern):

  • Alagebrium PubChem CID FABRICATED: I asserted CID 9904184 (which is sodium tungstate!); correct CID is 216305. Yet another instance of asserting canonical IDs from training memory and getting them wrong.

Wholesale study-design / quantitative fabrications:

  • Vasan 2000 PNAS (alagebrium): I asserted “aged rhesus monkeys” + “Vasan S, Foiles P, Founds H et al.”; paper is aged mongrel dogs + Asif M et al. (Sara Vasan is 3rd author). Stiffness constant figures + 1 mg/kg/day dose added. PROPAGATED ERROR: processes/advanced-glycation-end-products.md + processes/glucosepane.md (both previously verified) had inherited “rhesus monkey” language; corrected during propagation pass.
  • Streeter 2020 (glucosepane antibodies): I asserted “human retinal tissue”; paper actually aging mouse retinae + needs-human-replication. PROPAGATED ERROR: processes/glucosepane.md also had “human retinal tissue” attribution; corrected.
  • Haykal 2025 HIFU SR (dermatologic-resurfacing): I asserted “n=1,087 across 26 studies”; actual paper has 45 clinical trials/cohorts + completely different quantitative findings (laxity 18-30%; body circumference −2.5 to −4.5 cm). Wholesale fabrication.
  • Foppiani 2025 microneedling SR: I asserted “34 studies”; actual 21 articles / 723 patients.
  • Sales 2022 IPL SR: “18 studies” → 16 studies.
  • Modena 2025 HIFU meta-analysis outcome FABRICATED: I asserted “FACE-Q Appraisal of Lines and Wrinkles” outcome; actual analysis was neck skin sagging in 2 RCTs (n=475).
  • Robati 2020 microneedling vs Er:YAG DOI WRONG: 10.1111/jocd.12908 (totally unrelated lactobionic-acid pH paper, Tasić-Kostov 2019) → 10.1111/jocd.13440 (PMID 32359018). n also wrong: 30 → 24 completers / 32 enrolled.
  • de Filippi Sartori 2022 (split-face fractional CO2): year, design, n, biopsy timing all wrong. 2021 → 2022; “rct” → “prospective split-face comparative non-randomised”; n=16 confirmed; biopsy at 30 days NOT 90.
  • Manstein 2004: no animals in study (two human cohorts n=15 + n=30); prototype at 1.5 µm (NOT CO2/Er:YAG as I implied — fractional principle subsequently adopted by ablative/non-ablative platforms).
  • Smith 1996 lactic acid: I asserted “rct”; abstract describes no randomization, blinding, or control arm — corrected to “comparative study (non-randomized)”.

Microbiome cohort n + method fabrications:

  • Shibagaki 2017 n WRONG (on skin-microbiome-aging-shifts): 74 → 37 subjects (n=74 was sample count, 4 sites × subjects).
  • Shibagaki 2017 Staphylococcus DIRECTION INVERTED: I asserted ↑ forearm; paper actually says significantly DECREASED in older forearm. Direction reversed.
  • Shibagaki 2017 16S region: V1-V3 → V1-V2 (454 pyrosequencing).
  • Kim 2022 n WRONG (on both malassezia.md AND skin-microbiome-aging-shifts.md): ~40/118 → n=61 (29 young + 32 older). The same fabrication propagated to two pages.
  • Kim 2022 Malassezia restricta direction CORRECTED: I implied higher in older; paper shows higher in younger group on cheeks.
  • Jung 2024 n WRONG (cutibacterium-acnes): “n=86 Korean adults” → n=60 (30+30); method also wrong: 16S rRNA → whole-metagenome sequencing.
  • Podbielski 2024 scope OVERSTATED (cutibacterium-acnes): “shoulder prosthetic joint infections” → broader deep-seated infections (blood/bone/joint/deep tissue); key negative finding I’d missed (no virulence-gene-content differences between commensal + infection strains — distinction driven by SNPs in housekeeping/metabolic genes).
  • Trompezinski 2016 bakuchiol attribution MISLEADING (cutibacterium-acnes): clinical porphyrin-reduction data are from BGM complex (bakuchiol + Ginkgo biloba + mannitol) NOT isolated bakuchiol; isolated bakuchiol MIC 0.0005% is in-vitro only.
  • Xu 2007 M. globosa genome size WRONG: I asserted ~7.1 Mb; paper states ~9 Mb assembled (CBS 7966). Lipase count: “at least 8” → 14 lipase genes (13 predicted secreted).
  • Yu 2024 first author WRONG: Yu C → Yu T (Crossref).
  • Howard 2022 first author FABRICATED: “Howard AGA” (invented middle initials) → Howard B (Brian Howard).
  • Negari 2021 cell model FABRICATED: “human fibroblast cell line” → mouse 3T3 fibroblast cells (murine throughout).

Lin 2025 NMA author-list + framing corrections (the high-velocity citation I overused in seeder briefs):

  • “Lin L, Xu Z” → Lin L, Chen X, Liu C,…, Xu Z (12 authors total; Lin L + Chen X co-first; Xu Z corresponding/last).
  • n=3,905 added (was missing on multiple R44 pages).
  • Tazarotene worst-safety finding: significantly worse than placebo (OR 0.2093, p=0.0013) — was missing from R44 framing.
  • Glycolic acid non-significant for fine wrinkles: OR 2.36, 95% CI [0.56, 10.06], p=0.2413 — clearer than “tretinoin ranks higher” hedging.

Mechanism misattributions:

  • KLK5/KLK7 desquamation mechanism MISATTRIBUTED to Almeman 2024 (glycolic-acid): paper doesn’t contain KLK5/KLK7/kallikrein/corneodesmosin — mechanism actually from Egelrud/Rawlings literature. Tagged #gap/unsourced with attribution note.
  • deRamon 2022 pentosinane RECHARACTERIZED: I asserted “AGE crosslink”; paper describes it as post-translational modification structurally related to pentosidine/glucosepane — bifunctional inter-chain crosslink nature not established. Important pharmacological distinction.

Sumita 2018 DOI propagation fix:

  • chemical-peels MOC page had wrong DOI 10.1111/jdv.15114; corrected to 10.1111/jdv.15020 per R39-verified study page.

Propagation work completed

  • processes/advanced-glycation-end-products.md line 144 + 306: Vasan 2000 attribution corrected — “aged rhesus monkeys” → “aged mongrel dogs”; “Vasan S, Foiles P, Founds H” → “Asif M, Egan J, Vasan S,…”; stiffness-constant figures added; full PDF citation status updated.
  • processes/glucosepane.md line 123: “rats, dogs, and rhesus monkeys” → “rats and dogs (NOT rhesus monkeys — Asif/Wagle 2000 PNAS canine study was previously misattributed)”.
  • processes/glucosepane.md line 115 + 163: Streeter 2020 attribution corrected — “human retinal tissue” → “aging mouse retinae” + needs-human-replication; full footnote citation updated.
  • 9 new mechanism classes registered in frameworks/intervention-classes.md (corneocyte-desquamation/AHA-exfoliation, keratolytic/BHA-exfoliation, controlled-wounding-collagen-stimulation with fractional-photothermolysis + microneedling-controlled-injury sub-values, estrogen-receptor-agonist, age-crosslink-cleavage + glucosepane-specific-cleavage).

Schema escalations surfaced (final R38–R44 campaign tally)

The R38–R44 campaign surfaced 8 schema escalations queued for next CLAUDE.md cleanup:

  1. type: tissue — needs formal frontmatter block (3 seeders flagged in R38; skeletal-muscle.md is de facto precedent)
  2. pubchem-cid-synthetic: / pubchem-cid-l-form: / chembl-id-active-metabolite: — pattern for stereoisomer / prodrug compounds with biologically distinct forms (alpha-tocopherol R42 + lactic-acid R44 + tazarotene R41 precedents)
  3. publication-type: field for type: study (Purohit 2016 Letter format precedent, R39)
  4. mechanisms: field route-of-administration ambiguity (niacinamide topical vs systemic NAD+ precursor, R42)
  5. model-architecture: field for type: biomarker (SVM vs elastic-net vs PC clocks — R43 Bormann precedent)
  6. mendelian-randomization: enum extension to include not-applicable for structural/device biomarkers (R43 AGE Reader precedent)
  7. mode: procedural for type: intervention (device + surgery interventions don’t fit pharmacological/lifestyle/etc — R44 dermatologic-resurfacing precedent)
  8. type: microbe schema extensions: (a) add kingdom: field for Fungi-vs-Bacteria distinction; (b) gram-stain: enum should accept not-applicable-eukaryote; (c) oxygen-tolerance: enum should include aerotolerant-anaerobic; (d) literature-checked-through: field should be formally added to type: microbe schema. (R44 Malassezia + C. acnes + S. epidermidis precedents)

R38–R44 CAMPAIGN COMPLETE — 7 rounds; ~62 pages; ~210 agent invocations

RoundThemePagesMajor findings
R38 (closed 2026-05-19)Tissue + cell-type substrate8Inomata 2009 ATM direction inverted (caught + corrected); 6 schema decisions formalized
R39 (closed 2026-05-19)Foundational primary studies12Mascré 2012 construct fabrication (K10-CreERT was wrong); Solé-Boldo 2020 subset names entirely wrong
R40 (closed 2026-05-19)Effector proteins + matrix + wound healing12MMP-1 mouse-ortholog accessions resolved to wrong proteins (CXCL11 + GALM); 6 DOI fabrications + multiple n-value errors
R41 (closed 2026-05-19)Retinoid cluster + class MOC76 canonical-DB hallucinations (CHEMBL/CAS pointing to unrelated compounds — buparlisib, probenecid, 5,6-dihydroxy-8-aminoquinoline); duplicate class registration reconciled
R42 (closed 2026-05-19)Topical antioxidants + UV protection6First wiki-wide hallmark coverage of genomic-instability (UV-protection); Hughes 2013 2×2 factorial design corrected; ferulic-acid metal-chelation mechanism fabricated
R43 (closed 2026-05-19)Skin biomarkers + AGE imaging4Qi 2026 inherited author errors (Wegner J → K; Frey S → C; Feng L → Y); Meerwaldt 2004 vs 2005 r-value misattribution
R44 (closed 2026-05-19)Chemical peels + hormonal + AGE-breakers + resurfacing + microbiome11Alagebrium PubChem CID + Vasan 2000 species/author + Streeter 2020 species + Haykal 2025 cohort + multiple SR study-counts all wrong

Campaign-level lessons documented in feedback memories:

  • feedback_seeder_brief_doi_unreliable (R28 origin) — DOIs/years/journals from training memory are unreliable
  • feedback_canonical_db_memory_unreliable (R41 origin) — ChEMBL/CAS/UniProt accessions from training memory resolve to unrelated real compounds (not 404s) — only catchable by direct API lookup; provide PubChem CID + name only, let seeder pull cross-refs

Forward queue surfaced for follow-up rounds:

  • Qi 2026 study page PRIORITY VERIFICATION → COMPLETE 2026-05-19: user-provided PDF at ; 6 corrections applied (see R44 post-close entry below)
  • Aldh1a1/2/3 RALDH family proteins (R41 retinoid biology); Sod2 protein page (R39 Velarde study); Eln/Lox/Fbn1/Col1a1/Col3a1 already done R40
  • studies/falckenhayn-2024-dhm-dnmt-inhibitor.md (R42 PDF downloaded; natural seed)
  • studies/hughes-2013-sunscreen-photoaging.md (R42 anchor RCT)
  • studies/lin-2005-ce-ferulic-photoprotection.md (R42 PDF downloaded)
  • Allen 2023 + Mansouri 2025 + Bienkowska 2026 propagation to sirtuin-activators / nad-precursors class pages
  • Menendez Vazquez 2025 MitraSolo/MitraCluster — next-gen skin clock biomarker candidate
  • Boersma 2025 SAF-cancer-incidence companion study (R43 surfaced)
  • phenotypes/arterial-stiffening.md (referenced from R44 age-crosslink-breakers but not seeded)
  • log/R36.md historical reference to “rhesus monkeys” preserved as historical record (corrections documented in R44 close; don’t retroactively edit historical log entries)

Round + campaign disposition

R44 closed cleanly. 11/11 pages flipped to verified: true. ROADMAP.md updated to mark R38–R44 campaign complete. The skin-aging coverage cluster is now structurally complete: phenotype hub (skin-aging.md verified 2026-05-05) is now backed by ~62 new atomic pages spanning tissue → cell-type → effector protein → study → compound → intervention → biomarker → microbiome layers.

[2026-05-19] post-close verify — Qi 2026 campaign-initiator study page; PDF read end-to-end

Page: studies/qi-2026-dhm-epigenetic-skin-aging.md PDF: user-provided at (14 pages read) Previously: verified: false since R38; Springer OA blocked WebFetch for all prior rounds.

Corrections (6 substantive):

  1. DNAm-age reduction: −2.1 yr → −2.08 yr. Fig 2a label is “Δ pred. age = −2.08 y”. Paper text rounds this to “around 2.1 years”. Wiki now reports the precise value with the paper’s rounding noted. Propagated to TL;DR, findings table, and qi-23k-epidermis-clock-2026.md intervention-responsiveness table.

  2. Cross-validation metric label: “MAE 5.66 yr” → “cross-validation error 5.66 yr (glmnet type.measure=‘mse’). The paper states “cross-validation error (type measure=‘mse’) of 5.66 years” — glmnet was run with MSE as the loss function. The 5.66 value is not explicitly labeled as MAE in the paper; the 4.88 yr multi-ethnic figure IS explicitly called MAE. The wiki’s prior “MAE” label for 5.66 was imprecise. Updated in body, TL;DR, clock table, and qi-23k-epidermis-clock-2026.md.

  3. Correlation wrinkle volume: R=0.31, p<0.05 → r=0.314, p=0.025. Fig 5a caption gives exact values. Updated in secondary endpoints table and correlation discussion paragraph.

  4. Array platform: “probably 450k/EPIC” → confirmed Infinium EPICv2 (Methylation EPIC version 2.0 BeadChip). The Methods section explicitly states “Methylation EPIC version 2.0 BeadChip arrays (Illumina).” Corrected in Methods section.

  5. Sampling site (product-use cohort): previously had #gap/needs-pdf-verification for the bilateral/site description. PDF Methods confirms periorbital crow’s-feet area; imaging details (59-60 subjects for wrinkle/echogenicity, 51 for 3D profilometry) confirmed from Table 1. Gap tag removed; instrument-level n values added to secondary endpoints table.

  6. Sunscreen confound partial-correlation analysis added. PDF page 7 contains a formal partial-correlation analysis (R=0.25, p=0.05 serum-use correlation; R=−0.14, p=0.29 sunscreen-use correlation) that was not captured in the original extraction. Added as a new paragraph after the correlation discussion.

Author list / COI verified against PDF full text (page 14):

  • 8 Beiersdorf employees: Qi M, Pitta P, Wegner K, Kristof B, Gallinger J, Winnefeld M, Falckenhayn C, Grönniger E — confirmed.
  • Consultation fees from Beiersdorf AG: Guadanhim L, Frey C, Lyko F — confirmed.
  • Speaker fees from Beiersdorf SA: Rodríguez-Paredes M — confirmed.
  • No COI: Raddatz G, Feng Y, Wanitphakdeedecha R — confirmed.
  • Author-name corrections from R43 (Wegner K not J; Feng Y not L; Frey C not S) all confirmed correct against PDF.

Unverified (supplementary materials not in local PDF): Supp Figs S1–S10, Supp Tables S1–S7. No primary quantitative claim in the body depends on these.

Downstream propagation applied:

  • biomarkers/qi-23k-epidermis-clock-2026.md — verified-scope updated; CV metric corrected; DNAm-age change corrected to −2.08 yr; [^qi2026] footnote updated to reflect PDF verified; cross-reference note flipped to verified: true.
  • molecules/compounds/dihydromyricetin.md — [^qi2026study] footnote updated with precise values; cross-reference note flipped to verified: true.

Final state: verified: true, verified-date: 2026-05-19, verified-by: claude.

[2026-05-19] verify — Malassezia microbe page

Page: microbiome/malassezia.md

Sources checked:

  • NCBI Taxonomy ID 55193 — confirmed via eutils efetch: genus Malassezia, lineage Basidiomycota/Ustilaginomycotina/Malasseziomycetes/Malasseziales/Malasseziaceae; 19 named species per current NCBI count
  • Xu 2007 (doi:10.1073/pnas.0706756104; PMID 18000048) — full PDF downloaded and verified
  • Kim 2022 (doi:10.1038/s41598-022-06189-5; PMID 35149745) — full PDF downloaded and verified
  • Han 2023 (doi:10.3389/fnagi.2023.1268751; PMID 37854034) — full PDF downloaded and verified
  • Jo 2016 (doi:10.1016/j.jid.2016.05.130; PMID 27476723) — PMC5687974 full text verified (DOI lookup failed)
  • Sparber 2019 (doi:10.1016/j.chom.2019.02.002; PMID 30870621) — bronze OA, not downloaded; PubMed abstract cross-checked only; γδ T-cell specifics flagged as partial-verification
  • Dawson 2007 (doi:10.1038/sj.jidsymp.5650049) — not_oa locally; no-fulltext-access retained
  • Sowell 2022 (doi:10.1007/s40266-022-00930-5) — not_oa locally; no-fulltext-access retained
  • Supersession check: PubMed 2023–2026 “Malassezia skin aging” — 3 results; no meta-analysis or large RCT superseding existing claims found

Corrections made:

  1. Genome size: ~7.1 Mb → ~9 Mb assembled (Xu 2007 Table SI + Results: “assembled genome size of 9 Mb”)
  2. genome-size-mb frontmatter: updated to reflect ~9 Mb M. globosa with M. restricta as partial sequence
  3. Lipase count: “at least 8 lipases” → “14 lipase genes (13 predicted secreted)” per Xu 2007 Table 1
  4. ORF count: “~4,300 predicted ORFs” → “4,285 predicted protein-coding genes” (exact value from Table 1)
  5. Kim 2022 n: n=118, ages 20–66 → n=61 (Y-group n=29 ages 19–28; O-group n=32 ages 60–63) per PDF Table 1
  6. Kim 2022 M. restricta directionality: added that M. restricta was more abundant in the YOUNGER group (not merely a generic discriminating taxon); sebum/TEWL p-values added
  7. Jo 2016 n: children n=10 → n=14; adults n=18 → n=19; body-site count: 14 → 10 (9 skin + nares) per PMC
  8. Jo 2016 body text: added M. globosa vs M. restricta species dominance pattern per PMC results
  9. Han 2023: n was absent (“n unclear from abstract”) → n=95 (47 PD, 48 controls) added throughout; 95% CI for OR added (2.931–29.880); M. sympodialis prevalence added (74.5% vs 54.2%); PCR method limitation (4 species only) added
  10. Species count: “18–21 recognized species” → “19 recognized species” per current NCBI Taxonomy
  11. Intro ~50–80% mycobiome claim: replaced with accurate “large majority of fungal sequence reads in sebaceous body sites” (Xu 2007 does not quantify a percentage for mycobiome share)
  12. Sparber 2019 γδ T-cell claim: qualified as mouse-model description not fully confirmable from abstract; CCR6+ Th17 memory T-cell human finding from abstract added

Unverifiable claims (no full text):

  • Dawson 2007 claims: SD population prevalence (5% general; 34–83% immunosuppressed), pityriasis versicolor mechanism, M. furfur yeast-to-hyphal transition — all sourced to closed-access review; retained with no-fulltext-access
  • Sowell 2022 claims: SD second peak >65 yr, Malassezia folliculitis in immunocompromised — not_oa; retained with no-fulltext-access

Result: Flipped to verified: true (partial scope — Sparber 2019/Dawson 2007/Sowell 2022 unverified due to access; all verified sources corrected). literature-checked-through: 2026-05-19 set.

[2026-05-19] verify — salicylic acid compound page

Page: molecules/compounds/salicylic-acid.md

Sources checked:

  • PubChem CID 338 — REST API query confirmed all identity fields
  • Liu 2020 Cochrane (doi:10.1002/14651858.CD011368.pub2; PMC7193765) — PMC full text verified
  • Ravikumar 2022 (PMID 35309278) — PubMed efetch abstract confirmed
  • Lu 2019 (doi:10.1111/exd.13934) — abstract-only; closed-access (not_oa)
  • Zhou 2025 (doi:10.1007/s00266-024-04180-1) — abstract-only; closed-access (not_oa)
  • Alvarez 2025 (doi:10.1016/j.jaad.2025.04.021) — abstract-only; closed-access (not_oa)
  • intervention-classes.md keratolytic / BHA-exfoliation class — confirmed present and correctly attributed

Result: Flipped to verified: true (partial scope). Corrections made:

  1. Liu 2020 body wording: “minimal difference versus tretinoin” → “there may be little or no difference between salicylic acid and tretinoin” (the review’s exact language). Added RR 1.00 (95% CI 0.92–1.09), n=46, and named the outcome measure (PGA — participants’ global self-assessment of moderate to excellent improvement).
  2. Liu 2020 evidence quality: “low to very low” → “low” for the specific SA-vs-tretinoin comparison. The “low to very low” language applies across the full review for different comparisons; the SA vs tretinoin specific rating in the PMC full text is “low-quality evidence.”
  3. Liu 2020 footnote banner updated: “abstract + metadata verified; PDF verification pending” → “PMC full text verified 2026-05-19.”

Confirmed correct (no changes needed):

  • All canonical identity fields (CID 338, InChIKey, formula, weight, SMILES, IUPAC name, XLogP 2.3): exact match with PubChem REST
  • Ravikumar 2022: n=40 (20/arm), 50% GA vs 30% SA, biweekly×6 weeks, 4 follow-up points, 45% vs 0% >75% reduction, p<0.0001 — confirmed against efetch abstract
  • Ravikumar PIH discordance framing (treatment-induced PIH risk vs pre-existing PIH clearance): logically coherent; no correction needed
  • Aspirin disambiguation section: mechanistically accurate and clearly separated
  • Salicylism caveat and concomitant NSAID warning: framing appropriate and proportionate
  • Pregnancy ≤2% safe / high-concentration peels avoided: appropriately qualified with pharmacokinetic reasoning; no primary study cited (gap tag present and appropriate)
  • intervention-classes.md keratolytic / BHA-exfoliation class: confirmed present at line 1499; correctly attributes salicylic acid; no correction needed

Supersession candidates:

  • PMID 40960645 (Qiao 2025, Lasers Med Sci): SSA + 1927 nm TFL combination for photoaging; n=36 split-face; confirms SSA efficacy for photoaging but is a combination-therapy study, not a supersession of the SA-alone framing. No editorial revision needed; consistent with existing framing.
  • Five recent acne RCTs/meta-analyses (2023–2025) identified — none is a meta-analysis or large RCT (n>100) that supersedes the Liu 2020 Cochrane or the broader SA-for-acne evidence base. No supersession.
  • literature-checked-through: 2026-05-19 was already set on the page; confirmed current.

Unverifiable claims (closed-access):

  • Lu 2019 AMPK/SREBP1 mechanism: abstract title confirms topic; n not specified; gap tag present
  • Zhou 2025 SSA RCT detailed methods (double-blind design, primary endpoint definition): abstract-only
  • Alvarez 2025 Delphi detailed results: abstract-only

Downstream propagation needed:

  • frameworks/intervention-classes.md (already updated in R44 seeding — no further propagation required)
  • No other entity pages currently cite salicylic-acid.md as a primary source for corrected claims

[2026-05-19] verify — lactic acid compound page

Page: molecules/compounds/lactic-acid.md

Sources checked:

  • PubChem CID 612 (DL) and CID 107689 (L) — REST API confirmed all identity fields (MW, InChIKey, formula, XLogP, TPSA, HBD/HBA, SMILES)
  • ChEMBL CHEMBL1200559 (DL) and CHEMBL330546 (L) — ChEMBL REST API confirmed pref_name and molecule_type
  • CAS 50-21-5 (DL) and 79-33-4 (L) — confirmed via PubChem synonyms endpoint
  • Stiller 1996 (doi:10.1001/archderm.1996.03890300047009; PMID 8651713) — PubMed abstract verified (closed-access, no local PDF); Crossref DOI confirmed
  • Smith 1996 (doi:10.1016/s0190-9622(96)90602-7; PMID 8784274) — PubMed abstract verified (closed-access, no local PDF); Crossref DOI confirmed
  • Almeman 2024 (doi:10.2147/CCID.S453243) — full local PDF read (25 pages, CCID 2024:17 1661–1685)
  • Goldberg 2020 (doi:10.1111/jocd.13371) — Crossref DOI confirmed (title, year, journal)
  • Kim 2023 (doi:10.5021/ad.22.177; PMID 38086356) — Crossref and PubMed abstract confirmed first author Kim HW, corresponding Ryoo YW

Result: Flipped to verified: true (partial scope — see frontmatter). Corrections made:

  1. Smith 1996 study design: “rct (stated design)” → “comparative study (non-randomized; abstract describes no randomization or control arm)”. The PubMed abstract does not describe randomization, blinding, or a control arm. Also added exact abstract language to the footnote.
  2. Smith 1996 body framing: Added “(non-randomized design; n not reported)” caveat and needs-replication tag to the Smith 1996 body paragraph.
  3. Stiller 1996 patient description: “moderate photodamage” → “moderately severe photodamaged” per PubMed abstract exact wording.
  4. Comparison table data quality row: “Strong (landmark 1996 RCT)” (×2, implying both Smith and Stiller were RCTs) → Stiller 1996 correctly labelled as RCT; Smith 1996 correctly labelled as comparative study for concentration-response.

Confirmed correct (no changes needed):

  • All canonical identity fields (both CIDs, InChIKey, MW 90.08, formula C3H6O3, XLogP −0.7, TPSA 57.5, HBD 2 / HBA 3, SMILES): exact match with PubChem REST
  • DL InChIKey JVTAAEKCZFNVCJ-UHFFFAOYSA-N in frontmatter is correct for CID 612; L-form InChIKey JVTAAEKCZFNVCJ-REOHCLBHSA-N appropriately documented in identity section
  • ChEMBL IDs: CHEMBL1200559 = “LACTIC ACID” (DL), CHEMBL330546 = “L-LACTIC ACID” — both confirmed
  • Stiller 1996: n=74 women 40–70 yr; 8% GA vs 8% L-LA vs vehicle; 22-week double-blind RCT; 76% GA / 71% LA / 40% vehicle ≥1-grade improvement; p<0.05 for both vs vehicle — all confirmed against PubMed abstract
  • Smith 1996: 5% vs 12% LA twice daily × 3 months; 12% → epidermal + dermal firmness/thickness + smoothness/wrinkles; 5% → surface + epidermal only, no dermal signal — confirmed against PubMed abstract
  • Almeman 2024: narrative review (not systematic); journal CCID 2024:17 pages 1661–1685; single author Almeman AA; covers AHA mechanism (corneocyte desquamation, apoptosis), regulatory framework (FDA/EU/Australia), clinical evidence for photoaging/acne/hyperpigmentation, and AHA formulation — all consistent with wiki description
  • Glycolic acid MW 76.05 Da (from PubChem CID 757): correct; Almeman Table 1 rounds to integer 72 but PubChem is the authoritative source for molecular weight
  • PLLA FDA approvals cited (Aug 2004 HIV lipoatrophy; Jul 2009 cosmetic) — not independently re-checked against FDA records; flagged in verified-scope
  • All other footnote DOIs (Singh 2024, Fanning 2023, Razi 2022, Fluhr 2025, Kim 2023): Crossref DOI confirmed where checked

Supersession candidates (R25):

  • PubMed search for “lactic acid topical photoaging AND (meta-analysis OR RCT OR systematic review)” 2020–2026: 0 results. Broader AHA systematic review search 2020–2026: 5 results, none superseding the lactic-acid-photoaging evidence base (striae, axillary hyperpigmentation, and tretinoin-comparison topics). No supersession identified.
  • literature-checked-through: 2026-05-19 was already set on the page; confirmed current.

Unverifiable claims:

  • Stiller 1996 “no statistically significant difference between 8% GA and 8% LA on any efficacy endpoint” — abstract consistent (76% vs 71%, no significance reported between arms) but cannot confirm from abstract alone; marked as abstract-consistent
  • Smith 1996 exact n — not reported in abstract or archive
  • PLLA FDA approval exact dates — not independently re-checked against FDA records

Downstream propagation needed:

  • No other wiki pages currently cite lactic-acid.md as a primary source; no downstream propagation required
  • frameworks/intervention-classes.md references lactic acid as an AHA-class member — no quantitative claims from lactic-acid.md are present there; no update needed

[2026-05-19] verify — AGE crosslink breakers intervention page

Page: interventions/pharmacological/age-crosslink-breakers.md

Sources checked:

  • Vasan 2000 (doi:10.1073/pnas.040558497) — full PDF downloaded and read (PMID:10706607)
  • Hartog 2011 BENEFICIAL (doi:10.1093/eurjhf/hfr067) — Crossref abstract (closed-access PDF; all p-values confirmed)
  • Draghici 2015 Science (doi:10.1126/science.aac9655) — Crossref metadata confirmed (not_oa; title/authors/pages verified)
  • deRamon 2022 JACS (doi:10.1021/jacs.2c09626) — Crossref + PubMed abstract confirmed
  • Streeter 2020 ACS Chem Biol (doi:10.1021/acschembio.0c00093) — PubMed abstract confirmed
  • PubChem CID for alagebrium — REST API verified (CID 216305)
  • Yang 2003 (PMID:12646266) — closed-access; load-bearing quote retained with gap tag
  • Little 2005 (doi:10.1016/j.cardfail.2004.09.010) — closed-access; abstract-level only

Corrections made:

  • PubChem CID 9904184 → 216305 (9904184 is sodium tungstate, not alagebrium)
  • Vasan 2000 model organism: “aged rhesus monkeys” → “aged mongrel dogs” (paper is entirely a dog study; n=8 treated, n=5 untreated, n=7 young)
  • Vasan 2000 footnote first author: “Vasan S, Foiles P, Founds H” → “Asif M, Egan J, Vasan S, Jyothirmayi GN” (Asif M is first author; Vasan S is third)
  • Vasan 2000 footnote quantitative detail added: 1 mg/kg/day oral × 4 wk; stiffness constant 57.1±6.8 → 33.1±4.6 mmHg·m²/ml (P<0.001); no EF change confirmed
  • Evidence summary table: “rodents, dogs, and primates” → “rodents and dogs” (no primate data in any cited paper)
  • Streeter 2020 tissue: “human retinal tissue” → “aging mouse retinae” (abstract explicitly describes mouse IHC; human applicability inferred but not directly demonstrated); gap tag added needs-human-replication
  • deRamon 2022 pentosinane: described more carefully as a “PTM” not confirmed “crosslink” per abstract; bifunctional crosslink nature not established from published data; clarifying note added
  • Draghici 2015 footnote: added explicit closed-access (not_oa) flag and no-fulltext-access
  • Archive status updated: Vasan 2000 noted as “downloaded” (was “pending”)

Unverifiable claims (closed-access):

  • Yang 2003 (PMID:12646266) — verbatim quote retained; class-level critique confirmed at abstract level; full methods/tissue details unverifiable (#gap/no-fulltext-access implied; existing needs-replication tag sufficient)
  • Little 2005 — n=21, 420 mg/day × 16 wk, endpoint details (LV mass, E’ velocity) — abstract-level only

Supersession candidates:

  • None. Post-2011 PubMed search (n=26 papers) returned only preclinical/mechanistic papers; no meta-analysis or large RCT superseding BENEFICIAL trial found.

Final verified state: verified: true (with verified-scope noting closed-access papers Yang 2003 + Little 2005 + Draghici 2015)

Downstream propagation needed:

  • processes/advanced-glycation-end-products.md — may reference alagebrium PubChem CID or the rhesus-monkey claim; main agent should check
  • processes/glucosepane.md — Draghici 2015 and Streeter 2020 are cited there; Streeter 2020 mouse-vs-human correction should be propagated
  • frameworks/intervention-classes.md — age-crosslink-cleavage and glucosepane-specific-cleavage mechanism classes listed in this page’s frontmatter; confirm they exist in intervention-classes.md

[2026-05-19] verify — dermatologic-resurfacing intervention page

Page: interventions/pharmacological/dermatologic-resurfacing.md

Sources checked (all via Crossref + PubMed abstract; all 9 papers confirmed not_oa in a local paper archive):

  • Manstein 2004 (doi:10.1002/lsm.20048) — DOI confirmed via Crossref; DOI lookup failed (no free-text URL)
  • de Filippi Sartori 2022 (doi:10.1093/asj/sjab357) — PMID 34618888; not_oa
  • Foppiani 2025 (doi:10.1007/s00266-025-04972-z) — PMID 40542236; not_oa
  • Vassão 2022 (doi:10.1080/14764172.2022.2100914) — PMID 35856592; not_oa
  • Sales 2022 (doi:10.1007/s00403-021-02283-2) — PMID 34609598; not_oa
  • Haykal 2025 (doi:10.1093/asj/sjaf053) — PMID 40184185; not_oa
  • Modena 2025 (doi:10.1007/s10103-025-04424-9) — PMID 40167815; not_oa
  • Chao 2023 (doi:10.1055/s-0043-1772197) — Crossref confirmed; not_oa
  • Robati 2020 — wiki DOI (10.1111/jocd.12908) resolves to UNRELATED lactobionic acid paper; correct DOI found PMID 32359018 = doi:10.1111/jocd.13440

Corrections made:

  1. Robati 2020 DOI WRONG: 10.1111/jocd.12908 → 10.1111/jocd.13440
  2. Robati 2020 n: “n=30 (split-face)” → n=24 completers (32 enrolled)
  3. Foppiani 2025 study count: “34 studies” → 21 articles / 723 patients; pooled satisfaction 83% (95% CI 0.76–0.88)
  4. Vassão 2022 study count: “12 RF studies” → 21 articles / 354 patients
  5. Sales 2022 study count: “18 studies” → 16 studies (PRISMA)
  6. Haykal 2025 n and study count: “n=1,087 across 26 studies” → 45 clinical trials/cohort studies; n figure not in abstract; laxity improvement 18–30%; circumference reduction 2.5–4.5 cm; <5% adverse events
  7. Haykal 2025 maintenance interval: “repeat annually” not supported by abstract; flagged needs-replication
  8. Modena 2025 outcome measure: “FACE-Q Appraisal of Lines and Wrinkles” NOT confirmed in abstract; actual is neck skin sagging; n=475; 12 articles (3 RCT + 9 non-RCT); only 2 RCTs in meta-analysis; needs-fulltext added
  9. de Filippi Sartori year: “2021” throughout → 2022 (PubMed year; online Oct 2021, print 2022)
  10. de Filippi Sartori design: “rct (split-face randomized)” → prospective split-face comparative study (non-randomised)
  11. de Filippi Sartori n: “n not extractable” → n=16 female patients confirmed; quantitative values added (collagen I: 158.7±5.3 vs 139.2±5.0, P<0.0001)
  12. de Filippi Sartori biopsy timing: “90 days” → 30 days post-laser (biopsy obtained during upper blepharoplasty)
  13. Manstein 2004 subject description: “n=animal + human histology pilot” → n=15 (forearm) + n=30 (periorbital); no animal subjects in abstract
  14. Manstein 2004 re-epithelialization: “within 24–48 hours” → “approximately 24 hours” (abstract: “complete within 1 day”)
  15. Manstein 2004 prototype wavelength: added note: original device was 1.5 µm (not CO2 or Er:YAG); principle later adopted by other platforms

Unverifiable claims:

  • Full-text methodology and results tables for all 9 not_oa papers
  • Modena 2025 specific meta-analytic outcome measure (#gap/needs-fulltext)
  • Haykal 2025 specific maintenance/repeat intervals

Supersession candidates: None. 2025 laser-vs-other SR (PMID 40906045; n=497) consistent with current framing.

Final verified state: verified: true (verified-scope: abstract-level cross-check; all papers not_oa)

Downstream propagation needed: None. No existing study pages for any of the 9 citations.

[2026-05-19] verify — cutibacterium-acnes microbe page

Page: microbiome/cutibacterium-acnes.md

Sources checked:

  • NCBI Taxonomy ID 1747 — eutils efetch confirmed: ScientificName = Cutibacterium acnes, Rank = species, lineage Actinomycetota → Actinomycetes → Propionibacteriales → Propionibacteriaceae → Cutibacterium. Correct.
  • doi:10.1099/ijsem.0.001367 (Scholz 2016) — Crossref confirmed: authors Scholz CFP + Kilian M; journal Int J Syst Evol Microbiol; vol 66 issue 11 pp 4422–4432; year 2016. not_oa; PDF not in archive.
  • doi:10.1016/j.chom.2024.06.002 (Yu 2024) — Crossref confirmed: first author Yu T (not Yu C as seeded); journal Cell Host & Microbe; year 2024; 38 citations. not_oa; PDF not in archive. n=1,234 isolate count not verifiable.
  • doi:10.1111/jdv.19540 (Dreno 2024) — PDF read (HAL preprint/accepted manuscript version); confirmed 6 phylotypes (IA1, IA2, IB, IC, II, III); IA1 dominance in acne; antibiotic dysbiosis concern; full author list confirmed.
  • doi:10.1007/s40257-020-00531-1 (Dreno 2020) — PDF read; confirmed as Am J Clin Dermatol 2020;21(Suppl 1):S18–S24; 4 authors (Dreno B, Dagnelie MA, Khammari A, Corvec S); content confirmed.
  • doi:10.3390/microorganisms12112179 (Jung 2024) — PDF read; n=60 (not 86): 30 younger (20–29 yr) + 30 older (60–75 yr); method is whole-metagenome sequencing (not 16S rRNA); nasolabial fold; C. acnes abundance 83.14±22.18% younger vs 60.55±35.68% older (p=0.007); diversity indices all p<0.001.
  • doi:10.3389/fragi.2025.1644012 (Swaney 2025) — PDF read; n=59 (30 YA, 29 OA) from Birmingham UK; 16S rRNA V1-V3; face and antecubital fossa; C. acnes significant decrease on face and arm; gender-driven effect (facial change significant only in female participants); paper was not yet downloaded at seeding time.
  • doi:10.1111/jdv.18794 (Mias 2023) — download failed (not_oa after retry); Crossref abstract confirmed: Th17 pathway + loss of phylotype diversity as acne inflammation driver; authors confirmed. Unverifiable against full text; no-fulltext-access flag retained.
  • doi:10.3389/fcimb.2024.1433783 (Podbielski 2024) — PDF read; n=45 isolates (18 healthy control skin + 27 clinical deep-seated infections); covers bone/joint/blood/deep-tissue infections — NOT specifically shoulder arthroplasty; clinical isolates distributed across IA1, IB, II; no significant virulence-gene-content differences; SNPs in housekeeping genes distinguish commensal from clinical isolates.
  • doi:10.1111/jocd.70105 (Shao 2025) — PDF read; strain CCSM0331 confirmed as Type II (phylotype); DPPH scavenging 76.20% (undiluted); ABTS 99.56% (undiluted); 5% CFS reduced ROS to 78.61% (p<0.001); Nrf-2 + HO-1 activation confirmed; genome 2,496,923 bp, GC 60.02%.
  • doi:10.3389/fimmu.2018.02155 (Erdei 2018) — PDF read; TNIP1 confirmed as NF-κB negative regulator downstream of TLR2/TLR4; dose-dependent, strain-independent; ATRA increases TNIP1 + reduces TLR2-driven TNFα/CCL5 while increasing TLR4 downstream; full author list confirmed.
  • doi:10.2147/CCID.S110655 (Trompezinski 2016) — PDF read; confirmed BGM complex = bakuchiol + Ginkgo biloba + mannitol; MIC of bakuchiol isolated = 0.0005% vs P. acnes CIP A 179; clinical n=17 volunteers (19–34 yr, oily skin, acne-prone); BGM cream reduced porphyrins at day 28 and 56 (p<0.05); industry-sponsored study (NAOS/Bioderma).

Corrections made:

  1. Jung 2024 n: n=86 → n=60 (30+30); critical numeric correction
  2. Jung 2024 method: 16S rRNA → whole-metagenome sequencing; method correction
  3. Swaney 2025 body text: added actual n=59, age statistics, site-specific p-values, gender effect note; updated from vague “UK cohort” placeholder
  4. Podbielski 2024 body text: “shoulder prosthetic joint infections” → “deep-seated infections (systemic bloodstream, bone/joint, deep tissue)”; the paper does not focus on shoulder arthroplasty; that characterization was an unsourced extrapolation
  5. Podbielski 2024 body text: added that no significant virulence-gene-content differences were found (SNP-in-housekeeping-genes mechanism); the wiki omitted this key finding
  6. Bakuchiol section: clarified BGM complex = bakuchiol + Ginkgo biloba + mannitol (not just “bakuchiol-containing complex”); added MIC for isolated bakuchiol and clinical n=17; re-framed attribution correctly
  7. Yu 2024 first author: Yu C → Yu T (Crossref confirmed)
  8. Yu 2024 body: n=1,234 isolate count tagged no-fulltext-access (closed-access, cannot verify)
  9. Scholz 2016 citation count: 435 → 455 (Crossref)
  10. All 7 previously-pending footnotes updated with correct metadata, PDF paths, and n/design details (Swaney, Dreno 2020, Jung, Podbielski, Erdei, Trompezinski, Dreno 2024)
  11. Banner removed; verified flag flipped to true

Unverifiable claims:

  • Yu 2024 (doi:10.1016/j.chom.2024.06.002) n=1,234 isolate count: paper is not_oa; count tagged no-fulltext-access in body
  • Mias 2023 (doi:10.1111/jdv.18794) body claims: download failed; abstract-level verification only; no-fulltext-access on footnote retained
  • Scholz 2016 internal methodology details: not_oa; DOI/author/journal/page confirmed via Crossref

Supersession candidates: None. PubMed search (2024–2026) returned no meta-analyses or large RCTs superseding the aging-skin-microbiome or phylotype evidence base. literature-checked-through: 2026-05-19 (already set).

Schema note: oxygen-tolerance: aerotolerant-anaerobic is a new enum value not listed in CLAUDE.md examples (which show only anaerobic/aerobic/facultative). Value is scientifically accurate for C. acnes (grows anaerobically, tolerates O₂ but does not require it). Should be added to CLAUDE.md type:microbe schema during next schema cleanup pass.

Downstream propagation needed:

  • molecules/compounds/bakuchiol.md — bakuchiol page should be updated to reflect that antibacterial C. acnes data is from BGM complex (not isolated bakuchiol alone); MIC data should be added
  • hallmarks/dysbiosis.md — if it references C. acnes skin compartment, check for stale Jung 2024 n=86 claim
  • processes/sasp.md or skin-aging related pages — Podbielski correction (shoulder→deep-tissue) should be checked if propagated elsewhere

Final verified state: verified: true (with verified-scope noting Yu 2024 and Scholz 2016 are not_oa with DOI-level verification; Mias 2023 download failed)

[2026-05-19] verify — staphylococcus-epidermidis microbe page

Page: microbiome/staphylococcus-epidermidis.md

Sources checked:

  • NCBI Taxonomy ID 1282 — confirmed via eutils efetch (TaxId=1282, ScientificName=Staphylococcus epidermidis, Rank=species)
  • Zheng 2022 (doi:10.1016/j.chom.2022.01.004) — local PDF, full read pp.1-10; verified
  • Keshari 2019 (doi:10.3390/ijms20184477) — local PDF pp.1-6; verified
  • Nguyen 2017 (doi:10.3389/fcimb.2017.00090) — local PDF pp.1-5; verified
  • Cogen 2010 (doi:10.1038/jid.2009.243) — local PDF pp.1-6; verified
  • Negari 2021 (doi:10.3390/ijms22031414) — local PDF pp.1-5; verified
  • Hu 2023 (doi:10.1111/srt.13432) — local PDF pp.1-5; verified
  • Swaney 2025 (doi:10.3389/fragi.2025.1644012) — local PDF pp.1-5; verified
  • Naik 2015 (doi:10.1038/nature14052) — not_oa; verified via PubMed efetch abstract (PMID 25539086)
  • Lai 2010 (doi:10.1038/jid.2010.123) — failed download; verified via PubMed efetch abstract (PMID 20463690; PMC2922455)
  • Howard 2022 (doi:10.1016/j.jid.2021.11.029) — not_oa; verified via PubMed efetch abstract (PMID 34890626)
  • Howard 2022 author identity — confirmed via Crossref (first author Brian Howard, not “Howard AGA”)

Corrections made:

  1. Howard 2022 first author: “Howard AGA” → “Howard B” (Brian Howard; confirmed via Crossref)
  2. Negari 2021 cell model: “human fibroblast cell line” → “mouse 3T3 fibroblast cells” (paper uses 3T3 mouse fibroblasts, not human; this is a factual error in the seeder’s extraction)
  3. Hu 2023 n: “n not confirmed in abstract” → “n=48” (Methods: 48 healthy Chinese males, 4 groups: 14+7+7+20)
  4. Swaney 2025 n: “n not confirmed in abstract” → “n=59” (30 young mean 26.7±4.45 yr; 29 old mean 72.3±4.04 yr)
  5. Zheng 2022 enzyme name: Generic “sphingomyelinase” → specific “neutral sphingomyelinase Sph (gene sph)”; added phosphocholine co-product; added sph gene prevalence (47/48 SE skin isolates; 98%); added human in-vivo expression data (14/20 armpits, 10/20 face)
  6. Keshari 2019 butyric acid concentration: Added specific “~6 mM” (confirmed from Figure 2b); retained “millimolar” qualifier
  7. Cogen 2010 PSM nomenclature: Added specific PSMγ and PSMδ names and MBC quantitative values (>64 µM vs SE; 4–16 µM vs S. aureus/GAS)
  8. Extrapolation table — Zheng 2022 ceramide row: “not-tested” → “partial” for phenotype conservation; added sph human expression confirmation
  9. SE abundance quantitative values added: Hu 2023 strain-level percentages across age groups added to body text
  10. All pending/not-downloaded footnotes updated to reflect current archive status (Nguyen, Keshari downloaded; Lai verified via PMC)

Confirmed correct (no changes needed):

  • NCBI Taxonomy ID 1282 — correct
  • Naik 2015 T cell type: IL-17A+ CD8+ — confirmed
  • Naik 2015 mechanism: skin-resident DC subsets — confirmed
  • Howard 2022 n=158 Caucasian females 20–74 yr, 6 age groups — confirmed
  • Howard 2022 finding: only Lactobacillus + Cutibacterium significant pan-site declines — confirmed
  • Swaney 2025 S. hominis + C. acnes primary aging-differentiating taxa — confirmed
  • Lai 2010 molecule size “<10 kDa”, TLR2-dependence, hBD2/hBD3 induction — confirmed
  • Keshari 2019 FFAR2/GPR43 receptor, IL-6 suppression — confirmed
  • Firmicutes/Bacillota dual naming — correct and consistent with NCBI
  • Strain diversity (ica operon, ATCC 12228, RP62A) — consistent with Nguyen 2017 review

Supersession candidates (R25 check):

  • PubMed search (SE + skin microbiome + aging + meta-analysis/RCT, 2019–2026): no results. None found.
  • literature-checked-through field confirmed as 2026-05-19.

Schema note (item 9 from task brief):

  • literature-checked-through: field is present on this page under type: microbe but is NOT listed in CLAUDE.md’s type: microbe frontmatter schema block. Field is functionally appropriate (skin microbiome is fast-moving). Should be added to CLAUDE.md schema block in next schema cleanup round.

Final verified state: verified: true

Downstream propagation needed:

  • hallmarks/dysbiosis.md — if it cites SE-specific claims; main agent should check
  • hallmarks/chronic-inflammation.md — linked from SE page; Naik 2015 mechanism is referenced
  • processes/sasp.md — not directly cited but SE immune-aging content may overlap
  • No study pages exist for any of these citations; main agent may wish to seed stub study pages for Zheng 2022, Naik 2015, and Keshari 2019 given their load-bearing role

[2026-05-19] verify — lin-2005-ce-ferulic-photoprotection study page

Page: studies/lin-2005-ce-ferulic-photoprotection.md PDF: full 7-page PDF read from local archive at Source: Lin FH et al. J Invest Dermatol 2005;125(4):826–832. DOI 10.1111/j.0022-202X.2005.23768.x

Corrections (5 substantive):

  1. PMID WRONG: 16170318 → 16185284. PMID 16170318 resolves on PubMed to Dar et al. 2005 Nat Immunol (CXCR4/SDF-1 paper — unrelated). Correct PMID confirmed via DOI-linked efetch search returning idlist ['16185284'].
  2. Quote page location WRONG: p.830 → p.829 (Discussion). The sentence “The mechanism of ferulic acid’s stabilizing effect on vitamins C and E is unknown.” appears on p.829 in the Discussion section. p.830 is the start of the Materials and Methods section. Corrected in TL;DR critical caveat, blockquote, and gap table.
  3. Caspase-7 omitted throughout: Paper measures both caspase-3 AND caspase-7 (stated in abstract, demonstrated in Figure 2 a–d, n=3/group for both). All mentions of “caspase-3” alone updated to “caspase-3 and caspase-7”: frontmatter key-findings, Methods section, Key Findings §5 heading and body, Limitations section, gap table. The abstract states verbatim: “Inhibition of apoptosis was associated with reduced induction of caspase-3 and caspase-7.”
  4. Wikilink path prefix on ferulic-acid: [[molecules/compounds/ferulic-acid]] → [[ferulic-acid]] (Obsidian canonical short-form without path prefix).
  5. Wikilink path prefix on uv-protection: [[interventions/uv-protection]] → [[uv-protection]] (same convention).

Confirmed correct (no changes needed):

  • Author list: Lin FH, Lin JY, Gupta RD, Tournas JA, Burch JA, Selim MA, Monteiro-Riviere NA, Grichnik JM, Zielinski J, Pinnell SR — exact match p.826 byline
  • Formulation: 15% L-ascorbic acid + 1% α-tocopherol + 0.5% ferulic acid — confirmed p.830 Methods
  • ~8-fold photoprotection: confirmed p.827 Results (“provided approximately 8-fold protection”) and p.826 abstract (“approximately 8-fold”)
  • C+E alone ~4-fold: confirmed p.827 (“about 4-fold protection”)
  • n=6/group for erythema/sunburn cell assays: confirmed Figure 1 legend p.827 (“n=6”)
  • n=3/group for Western blot: confirmed Figure 2 legend p.828 (“n=3”)
  • Yorkshire pig model (Sus scrofa domesticus, weanling): confirmed p.830 Methods
  • CPD/thymine dimer endpoint: confirmed p.828 and Figure 4
  • Sunburn cell endpoint: confirmed Figure 1d and p.827 Results
  • Erythema (colorimetry) endpoint: confirmed Figure 1c and p.827 Results
  • Ferulic acid alone <C+E combination: confirmed p.827 Results (“statistically different from ferulic acid alone at 2×, 6×, and 8× MED”)
  • Ferulic acid UV absorption peak ~307 nm (log e=4.19): confirmed p.829 Discussion
  • Mechanism-unknown quote is verbatim: confirmed p.829 Discussion exact sentence
  • Statistics: two-tailed Student’s t-test — confirmed p.831 Statistics section
  • Journal: J Invest Dermatol 125:826–832 — confirmed masthead

Supersession candidates (R25):

  • PubMed search: “ferulic acid vitamin C vitamin E topical photoprotection AND (meta-analysis OR randomized controlled trial)” 2008–2026 — 0 results.
  • Broader: “CE ferulic acid topical antioxidant photoprotection skin” 2022–2026 — 0 results.
  • No superseding meta-analysis or large RCT found. The porcine-only evidence base is unchanged; the gap tag needs-human-replication remains appropriate.
  • literature-checked-through: 2026-05-19 set.

Unverifiable claims:

  • Citation count (340), FWCI (13.6), and citation-percentile (100th) listed in the Significance section are from DOI lookup ingest; not independently verifiable against Clarivate/Scopus from this tool.

Final verified state: verified: true (full scope; full PDF read)

Downstream propagation needed:

  • molecules/compounds/alpha-tocopherol — cites Lin 2005 for ~8-fold and for the mechanism-unknown gap; PMID in that page’s footnote should be confirmed as 16185284 (not 16170318)
  • molecules/compounds/ascorbic-acid — likely cites Lin 2005 for the C+E+F combination; PMID should be checked
  • Any entity page citing this study should now also reflect the caspase-7 endpoint (not just caspase-3)

[2026-05-19] post-R44 forward-queue batch — 7 atomic pages + 2 propagation passes (Allen/Mansouri/Bienkowska + RALDH/SOD2/arterial-stiffening/Menendez/Boersma)

Continuation of post-R44 forward-queue work (NOT a formal new round). 7 atomic pages seeded via parallel wiki-seeder batch + 2 propagation sub-agents. All 7 pages remain verified: false (banners in place; verification deferred — most primary PDFs not locally available; SOD2 + arterial-stiffening have OA PMC sources but verifier pass not yet dispatched).

Pages seeded (7)

Proteins (4):

  • molecules/proteins/aldh1a1.md — RALDH1 (UniProt P00352; HSC self-renewal marker, dopaminergic neuron vulnerability, 4-HNE detox; druggability tier 2 probe-literature-derived, Open Targets endpoint returned 400, flagged #gap/no-opentargets-entry)
  • molecules/proteins/aldh1a2.md — RALDH2 (UniProt O94788; dominant embryonic RALDH; Niederreither 2001 E9.5 lethality canonical; adult meninges/mesentery/testis)
  • molecules/proteins/aldh1a3.md — RALDH3 (UniProt P47895; nasal/eye-development; MCOP8 anophthalmia LoF phenotype; ALDH+ CSC marker; Marcato 2011 ALDEFLUOR specificity)
  • molecules/proteins/sod2.md — MnSOD (UniProt P04179; SIRT3-K68/K122 deacetylation, FOXO3 transcription; Sod2-/- cardiomyopathy + Velarde 2012 skin-senescence; Van Remmen 2003 lifespan null result preserved as #gap/contradictory-evidence; mimetic pipeline MnTBAP/EUK-134/avasopasem-ROMAN-Phase-3; literature-checked-through: 2026-05-19)

Phenotype (1):

  • phenotypes/arterial-stiffening.md — large-artery stiffening (cfPWV >10 m/s threshold; ICD-10 I77.1 / ICD-11 BD52 approximate, flagged #gap/needs-canonical-id — no dedicated codes exist; mechanisms: elastin fragmentation, AGE crosslinks, VSMC senescence, medial calcification; Mitchell 2010 Framingham anchor; Clayton 2023 senolytic-navitoclax reversal data integrated; alagebrium failure documented; hallmark mapping: tagged [[altered-intercellular-communication]] as closest existing hallmark — flagged that a dedicated loss-of-elasticity or ecm-crosslinking hallmark may warrant creation given cross-cutting relevance)

Biomarker (1):

  • biomarkers/menendez-vazquez-2025-mitrasolo-mitracluster.md — MitraSolo + MitraCluster two-clock framework (DOI 10.1038/s41514-025-00314-0; PMID 41387742; PMC PMC12820032; n=462, ~4 yr MAE, intra-individual variation <2 yr; enzymatic methyl-seq; Yamanaka rejuvenation signal; independent of Beiersdorf/DKFZ lineage; abstract-only — CpG count + cohort demographics + exact MAE tagged #gap/no-fulltext-access; literature-checked-through: 2026-05-19)

Study (1):

  • studies/boersma-2025-saf-cancer-incidence.md — Boersma 2025 BMC Cancer (DOI 10.1186/s12885-025-14801-w; PMID 40859200; PMC12379543; gold OA; n=77,961 LifeLines; unadjusted HR 2.36 → fully-adjusted HR 1.11 95% CI 1.05-1.17 p<0.001; sex-stratified males 3.13 vs females 1.96; site-specific significance levels only — full HRs flagged for verifier pass; companion to Boersma 2024 Sci Rep CVD-mortality; queued)

Propagation pass A — R45-batch cross-link integration (21 edits across 14 files)

  • molecules/compounds/retinol + retinaldehyde — RALDH1/2/3 wikilinks in conversion-cascade paragraphs
  • molecules/proteins/sirt3 + foxo3 + bnip3 — SOD2 wikilinks in substrate/target/control tables
  • hallmarks/mitochondrial-dysfunction — SOD2 + Velarde 2012 wikilinks added to ETC-decline paragraph (page previously had no SOD2 mention)
  • processes/oxphos + mitochondrial-biogenesis — SOD2/MnSOD wikilinks at canonical mention points
  • processes/advanced-glycation-end-products + glucosepane — arterial-stiffening wikilinks in CVD-mechanism sections
  • biomarkers/skin-autofluorescence-age-reader — arterial-stiffening wikilink; new “Cancer incidence: Boersma 2025” subsection with full HR + adjustment + site-specific signals + outcome-table row + [^boersma2025] footnote
  • frameworks/biological-age-measurement — MitraSolo/MitraCluster row added to skin-clock comparison table; “third independent skin clock” framing updated to “four”
  • biomarkers/tapelift-clock-2026 + qi-23k-epidermis-clock-2026 + bormann-epidermis-clock-2016 — Menendez Vazquez 2025 wikilinks in peer-clock comparison sections (plain-text mentions already existed; this pass added wikilinks)

Skipped: tretinoin.md (downstream of RALDH conversion; no contextual RALDH mention).

Propagation pass B — Allen/Mansouri/Bienkowska to STAC class + NAD precursors (6 files)

Mansouri 2025 confirmed via PubMed efetch: Mansouri F, Feliziani G, Bordoni L, Gabbianelli R · “Impact of Resveratrol Supplementation on Human Sirtuin 1: A Grading of Recommendations Assessment, Development and Evaluation-Assessed Systematic Review and Dose-Response Meta-Analysis of Randomized Controlled Trials” · J Acad Nutr Diet 2025 Sep;125(9):1299-1314.e1 · doi:10.1016/j.jand.2025.03.011 · PMID 40158656. 11 RCTs pooled; gene-expression SMD=0.05 P=0.73; protein-expression SMD=0.3 P=0.18; serum MD=−0.04 P=0.7 — all null. ROADMAP “11 RCTs” claim confirmed. Search dates Mar–Apr 2024 predate Karim 2025 Nov publication; Karim could NOT have been included.

Allen 2023 confirmed: Allen NC,…, Damian DL (57 authors) · “Nicotinamide for Skin-Cancer Chemoprevention in Transplant Recipients” · N Engl J Med 2023 Mar 2;388(9):804-812 · doi:10.1056/NEJMoa2203086 · PMID 36856616 · n=158 (79/79); 500 mg BID × 12 mo; rate ratio 1.0 (95% CI 0.8-1.3; P=0.96); ONTRANS trial stopped early for poor recruitment (caveat preserved).

Bienkowska 2026 confirmed: Bienkowska A, Qi M, Kanta K,…, Falckenhayn C · “Unraveling the complexity of skin’s biological aging utilizing epigenetic clocks” · Clin Epigenetics 2026 Apr 9;18(1):58 · doi:10.1186/s13148-026-02101-4 · PMID 41957838 · PMCID PMC13063880 · n=851; 20 deceleration / 17 acceleration factors; DHM + aspirin among deceleration-consistent. Last author = Falckenhayn (same DHM-pipeline group as Qi 2026).

Edits:

  • interventions/pharmacological/sirtuin-activators — Mansouri 2025 row added to top of Human Clinical Evidence table (highest-tier synthesis); Karim 2025 row annotated to note Mansouri search-cutoff predates Karim; Class Assessment #4 extended with formal meta-analytic confirmation; [^mansouri2025] footnote; literature-checked-through: 2026-05-19
  • molecules/proteins/sirt1 — Pharmacology § Resveratrol bullet extended with Mansouri meta-analytic non-engagement finding
  • molecules/compounds/resveratrol — Mansouri 2025 added to Human Evidence table as target-engagement evidence; bottom-line paragraph extended
  • interventions/pharmacological/nad-precursors — Allen 2023 + Chen 2015 disambiguation upgraded to proper footnote refs ([^allen2023], [^chen2015nmsc]); compound table updated; literature-checked-through: 2026-05-19
  • molecules/compounds/niacinamide — Allen 2023 + Tosti 2023 references promoted to proper footnotes
  • molecules/compounds/dihydromyricetin — new “Bienkowska 2026 population-cohort observational support” subsection (n=851, same-group caveat); [^bienkowska2026] footnote

Deferred: Bienkowska 2026 mention on frameworks/biological-age-measurement (touched by parallel propagation agent in pass A; user can merge later if desired).

Key methodological notes

  1. Mansouri 2025 subgroup nuance preserved: the headline null was paired with a subgroup signal showing transient Sirt1 gene-expression elevation in studies <12 wk in blood tissue — surfaced on sirtuin-activators + resveratrol pages rather than reporting only the headline null. Reflects citation-discipline rule about quality signal in the footnote.

  2. Allen 2023 stopped-early caveat: ONTRANS halted for poor recruitment at n=158/group; n=79/group is half typical NEJM Phase 3 target. Interpretive weight should be tempered by underpowering even though the null is consistent with biological-mechanism expectation.

  3. Bienkowska 2026 same-group attribution: Falckenhayn is last author of both Qi 2026 and Bienkowska 2026 — observational n=851 support is NOT independent replication of Qi 2026. Critical interpretive framing preserved across all propagation targets.

  4. Lin 2005 PMID propagation check (post-R44 final): verified that the wrong PMID 16170318 did not propagate to any active wiki page; alpha-tocopherol.md footnote already had the correct PMID 16185284 (added during R42 verifier pass); ascorbic-acid.md footnote updated to PMID 16185284 + caspase-7 finding noted + verified-scope updated to reflect Lin 2005 PDF-verified status. Round propagation complete.

Forward queue remaining

  • R45 follow-on items: cell-types/cutaneous-immunosenescence-cells.md, pathways/il-34-csf1r.md, tissues/hair-follicle.md, cell-types/sebocytes.md (queued from R44 close summary; not yet seeded)
  • ~7 newly-seeded verified: false pages awaiting verifier passes (ALDH1A1/2/3 + SOD2 + arterial-stiffening + Menendez Vazquez + Boersma 2025)
  • Schema escalation candidate: dedicated loss-of-elasticity or extracellular-matrix-crosslinking hallmark page (arterial-stiffening surfaced gap)
  • Skin-methylation-clocks class MOC candidate — 4 standalone skin-clock biomarker pages now exist, framework page already covers them in comparison table but a dedicated class MOC may be warranted

Final state: All 7 new pages drafted with banner; cross-links integrated; class-page evidence base updated to current literature.

[2026-05-19] verify — molecules/proteins/sod2.md

  • Verified: molecules/proteins/sod2.md → verified: true
  • Sources checked:
    • doi:10.1038/ng1295-376 (Li 1995) — full PDF read; claims confirmed
    • doi:10.18632/aging.100423 (Velarde 2012) — cross-checked against verified study page (not re-read from PDF)
    • doi:10.1073/pnas.93.18.9782 (Lebovitz 1996) — PubMed abstract
    • doi:10.1152/physiolgenomics.00122.2003 (Van Remmen 2003) — PubMed abstract only (not_oa)
    • doi:10.1016/j.cmet.2010.11.015 (Qiu 2010) — PubMed abstract
    • doi:10.1016/j.eclinm.2025.103539 (ROMAN Phase 3) — PubMed abstract
    • doi:10.1200/JCO.19.01507 (Anderson 2019 Phase IIb) — PubMed abstract
    • UniProt P04179 — REST API (length, subcellular location, PTMs, subunit confirmed)
    • PMIDs 41943851 and 42042474 — PubMed efetch (recency integrations confirmed faithful)
  • Corrections:
    1. ROMAN trial primary/secondary endpoint framing wrong → corrected: primary endpoint (SOM incidence p=0.045) WAS met; grade-4 OM was a secondary endpoint that failed (p=0.052). Body text, footnote, and frontmatter druggability-tier comment all updated.
    2. K122 “primary” / K68 “secondary” SIRT3 deacetylation sites — distinction not stated in Qiu 2010 (“two critical lysine residues” only); removed primary/secondary hierarchy in frontmatter key-ptms, body, and cross-reference section.
    3. Van Remmen 2003 n=~100/group — not stated in abstract; closed-access full text; tagged no-fulltext-access and added abstract-sourced 8-oxodG elevation quantification (15–60% across tissues at 26 months).
    4. Added ROMAN Phase 3 n to footnote: 407 primary analysis (241/166).
  • Unverifiable: Van Remmen 2003 n per group; Velarde 2015 PNAS details (PDF still pending download); Qiu 2010 specific K68/K122 hierarchy vs. magnitude data (abstract only).
  • Supersession: none found for the aging-biology core claims; avasopasem/ROMAN clinical programme is actively evolving but no new aging-indication evidence supersedes current framing.

[2026-05-19] verify — phenotypes/arterial-stiffening.md

Page verified: phenotypes/arterial-stiffening.md Sources checked: 7 DOIs. Reference Values 2010 (PDF), Mitchell 2010 (PMC full text), Sun 2015 (PDF), Jennings 2019 (PDF), Clayton 2023 (PMC abstract + PMID). Laurent 2006 (archive failed; abstract-only). Boutouyrie 2021 (archive failed; abstract-only).

Corrections applied (6):

  1. Mitchell 2010 HR: “14% higher risk per 1 m/s” → “48% higher risk per 1 SD (HR 1.48, 95% CI 1.16–1.91, p=0.002)”. The paper reports no per-1-m/s estimate; uses inverse-transformed standardized units.
  2. Jennings 2019 primary outcome: “cfPWV reduced ~0.8 m/s (95% CI −1.3 to −0.3)” → no significant change in cfPWV (P=0.60). Significant finding was AIx@75 improvement (−12.4, 95% CI −24.4 to −0.5, p=0.04) in UK sub-cohort only (n=225). Trial was 1 year, not 3 months.
  3. Jennings 2019 trial n: “not extracted” → n=1,294 randomized (NU-AGE trial; 5 European centres).
  4. Jennings 2019 citation: pages 1021-1031 issue 5 → pages 578-586 issue 3.
  5. Clayton 2023 cfPWV reduction: “~15%” → ~20% (446→356 cm/s; paper states ~20%).
  6. Clayton 2023 citation: pages 2198-2211 → 2072-2087; strain C57BL/6J → C57BL/6N; added n=14-16/group, dosing schedule, EDD values, p16-3MR parallel arm details.
  7. Reference Values 2010: clarified n=11,092 = reference-value population (not all-comers); total database n=16,867; normal-value population n=1,455; added age-stratified cfPWV table values; confirmed 8 countries (not stated as count in original wiki).
  8. Sun 2015: corrected page range 252-261 → 252-256; clarified it is a 5-page review.

Unverifiable: Laurent 2006 full text (paywalled + archive failed) — 10 m/s threshold and AIx ≠ cfPWV statements not independently confirmed beyond abstract; Boutouyrie 2021 full text (blocked); sveen2015 carried over from glucosepane page.

Supersession: Venkatasubramanian/Clayton 2025 (Hypertension 82:1767-1777; PMID 40747550) extends Clayton 2023 to doxorubicin-induced stiffening model — consistent with and reinforces the Clayton 2023 framing; does not supersede aging-stiffening claims. No Mediterranean diet cfPWV-specific RCT supersession found.

[2026-05-19] seed — molecules/compounds/ferulic-acid.md

Closure of long-standing orphan-link risk — ferulic acid was referenced repeatedly across R42 (Lin 2005 C+E+F formulation; alpha-tocopherol + ascorbic-acid + dihydromyricetin pages) but never seeded. Lin 2005 PDF-verified anchor now in place.

Canonical IDs confirmed (REST API, 2026-05-19):

  • PubChem CID 445858 (InChIKey KSEBMYQBYZTDHS-HWKANZROSA-N; C10H10O4; 194.18 Da)
  • ChEMBL: CHEMBL32749 (NOT CHEMBL66653 from training memory — wrong; CHEMBL66653 resolves to an unrelated synthetic). Corrected via InChIKey lookup. Continuation of canonical-DB-memory-unreliable pattern (R41-R44 precedents).
  • CAS 1135-24-6 (confirmed); DrugBank null (not a regulated drug)

Recency literature search (R25): 3 RCTs integrated — Qin 2025 (n=50 split-face Fraxel; doi:10.1111/jocd.70251); Kim 2025 (n=31 split-neck microneedle RF; doi:10.1080/09546634.2025.2504655); Shi 2026 (n=64 split-face CO2 laser; doi:10.1111/jocd.70634). All three test the C+E+F COMBINATION, not isolated ferulic acid — all have SkinCeuticals/L’Oréal sponsorship. Extends Lin 2005 porcine evidence into post-procedure human contexts; no discordance found. literature-checked-through: 2026-05-19.

Foundational claim integration: Lin 2005 stabilisation mechanism “unknown” preserved as #gap/no-mechanism (faithful to source paper’s explicit statement). Isolated-FA-vs-combination evidence ambiguity preserved as #gap/dose-response-unclear. No isolated-ferulic-acid aging-endpoint RCT exists — #gap/needs-replication.

Final state: verified: false (banner in place); ChEMBL ID + canonical-DB correction documented in body. Suggested verifier priority: (1) confirm ChEMBL CHEMBL32749 pref_name; (2) verify 3 abstract-only RCTs against PDFs; (3) confirm CAS authority.

[2026-05-19] seed — biomarkers/skin-methylation-clocks.md (class MOC)

Consolidates 4 atomic skin-clock biomarker pages (Bormann 2016, Qi 23k 2026, TapeLift 2026, Menendez Vazquez 2025 MitraSolo/MitraCluster). type: framework navigational overlay; per CLAUDE.md does NOT carry verified fields.

Placement decision: biomarkers/ (not frameworks/). Precedent established R35 (log/R35.md line 725 explicitly anticipated “biomarkers/skin-methylation-clocks.md as an aggregate”) and reinforced by frameworks/biological-age-measurement.md line 80 cross-link to the same path. Class MOCs scoped to a single subdirectory’s children follow proximity-discoverability convention.

Structure:

  • Comparison table of all 4 clocks (architecture, training n, MAE, tissue, independence)
  • Three independent lineages framing (Beiersdorf-DKFZ: Bormann + Qi + Bienkowska 2026; TapeLift: Rodríguez-Paredes; Mitra Bio: Menendez Vazquez)
  • Methodological-dimensions section (tissue source, array vs methyl-seq, architecture, intervention-responsive evidence)
  • Open questions (cross-validation, intervention-responsiveness, sun-exposure confounding, tissue-region specificity)

No primary-source claims originate here — all quantitative claims link via wikilinks to atomic pages where footnoted (framework discipline). Currently DHM (Falckenhayn 2024 + Qi 2026) is the only well-validated intervention with positive skin-clock RCT signal.

Schema gap flagged by seeder: CLAUDE.md does not explicitly govern where narrow-scope type: framework MOCs (scoped to a single subdirectory’s children) should live. Resolved here by following R35 precedent: biomarkers/. If user prefers frameworks/ consistency across all type: framework pages, escalate as a CLAUDE.md schema clarification.

[2026-05-19] verify — molecules/proteins/aldh1a3.md

  • verified: molecules/proteins/aldh1a3.md → flipped to verified: true
  • corrections: (1) Codony 2019 DOI wrong (10.1016/j.cllc.2019.01.009 → 10.1016/j.cllc.2019.02.005); (2) Codony 2019 body claim overstated ALDH1A3 isoform specificity — paper studies pan-ALDH1, not ALDH1A3 specifically; qualified + gap-tagged; (3) Marcato 2011 Cell Cycle paper is a commentary — functional data (knockdown → ALDEFLUOR+ loss, sphere-forming, xenograft) attributed to companion Stem Cells paper (doi:10.1002/stem.563); footnote and body corrected; (4) Kesim 2023 footnote n corrected from “multiple families” to “9 individuals, 7 families”; brief-communication type noted; (5) Thomas 2016 footnote enriched with confirmed in-vivo model details (NOD/SCID, n=6/group, nanoparticle-encapsulated citral, orthotopic); (6) literature-checked-through set to 2026-05-19; banner removed
  • PDFs not available (download failed): Marcato 2011 Cell Cycle (bronze OA, paywalled URL), Duester 2009, Kumar 2017 (Exp Eye Res) — these footnotes partially unverifiable; gap documented in verified-scope

Graph View

Backlinks