log/R39.md — Round 39 entries

Sub-file of log — see parent for index.

[2026-05-19] R39 close — foundational primary-source studies (12 pages)

Second round of the R38–R44 skin-aging coverage campaign. 12 study pages seeded + verified in single-day round. 24 agent invocations (12 seeders + 12 verifiers) + 1 propagation pass. All 12 pages flipped to verified: true (partial scope for 3 closed-access primaries: Sumita 2018, Kovacs 2010, Purohit 2016).

Pages added (all 12)

studies/fisher-1996-photoaging-ap1-mmp.md (Nature 1996; local PDF)
studies/fisher-2009-collagen-fragmentation-mmp.md (Am J Pathol 2009; local PDF)
studies/velarde-2012-mitochondria-skin-senescence.md (Aging 2012; local PDF)
studies/demaria-2014-senescent-cells-wound-healing.md (Dev Cell 2014; local PDF)
studies/hattori-2004-scf-solar-lentigo.md (J Invest Dermatol 2004; local PDF)
studies/krutmann-2017-skin-aging-exposome.md (J Dermatol Sci 2017; local PDF; review)
studies/sumita-2018-tretinoin-photoaging.md (JEADV 2018; closed-access; partial scope)
studies/kovacs-2010-fibroblast-solar-lentigo.md (Br J Dermatol 2010; closed-access; partial scope)
studies/purohit-2016-smad3-fibroblasts.md (J Dermatol Sci 2016; closed-access; partial scope)
studies/bormann-2016-epidermis-clock.md (Aging Cell 2016; local PDF)
studies/rodriguez-paredes-2026-tapelift-clock.md (npj Precision Oncology 2026; DOI resolved during seeding from Qi 2026 reference list; downloaded mid-session)
studies/inomata-2009-melanocyte-stem-cell-atm.md (Cell 2009; bronze OA; downloaded)

Major verifier findings — fabricated/inverted/wrong claims caught

R39 verifier yield exceeded R38 in absolute corrections per page (R38: ~45 across 8 pages = ~5.6/page; R39: ~60 across 12 pages = ~5/page; many higher-impact). Continued the seeder-fabricates-outcomes + seeder-DOI-memory-unreliable patterns from prior memory.

Critical biology inversion caught:

Krutmann 2017 / Liebmann 2010 red-light claim inverted: seeder wrote red light (632 nm, 648 nm) + IRA “reduce proliferation and promote keratinocyte differentiation”; paper says they have NO impact — only blue light (412–426 nm) reduces proliferation. Direction was backwards.

Wholesale primary-source misattributions:

Fisher 2009 MMP-3/MMP-9 elevation claim fabricated: seeder claimed paper presents quantitative MMP-3 + MMP-9 elevation data; paper has NO such data in main results — MMPs 3/9 appear only in Discussion as possible downstream amplifiers of MMP-1 cleavage products. Propagated correction to tissues/dermis.md (line 99 also had this misattribution from the prior R38 batch).
Krutmann 2017 / Fisher 1996 attribution fabricated: seeder wrote “Krutmann et al. cite Fisher 1996” for sub-erythemogenic UV MMP claim; Fisher 1996 does NOT appear anywhere in Krutmann’s 92-reference list. Block-review citation (refs [13–19]) is the actual source.
Inomata 2009 primary model misattributed: seeder wrote K14-SCF transgenic mice as primary model (sourced from main agent brief’s training-memory error — K14-SCF is actually the Nishimura 2005 model). Paper actually uses Dct-lacZ transgenic mice as primary reporter in every main figure. K14-SCF NOT mentioned anywhere in Inomata 2009.

DOI / metadata fabrications:

Velarde 2012 / Li 1995 DOI: seeder used 10.1038/ng0995-97 from memory; correct DOI is 10.1038/ng1295-376 (Li Y et al., Nat Genet 1995;11:376–381). The wrong DOI returned 404. Cross-checked wiki-wide — no further propagation needed (DOI only appeared on the velarde study page itself + log entry).
Krutmann 2017 footnote DOI/year/author chain: 6 separate citation errors caught — Rowe 2000 → Rowe/Guyuron 2010 (book chapter, not journal); Richardson 1987 (not “Rickards D”); Doshi 2007 (not 1997); Tamai 2014 (not “Yamai 2001”); Lohan 2016 Exp Dermatol (not “J Dermatol Treat 2015”); Thiele 1997 J Invest Dermatol (not J Clin Invest). Multiple authors + journals + years all misremembered.

Quantitative metadata corrections in already-verified phenotypes/skin-aging.md (verified 2026-05-05; R39 verifiers caught 3 errors that the 2026-05-05 verifier missed):

Sumita 2018: n=60 → n=24 women (48 forearms, bilateral within-person design); pages 1734-1741 → 1819-1826. Inline-corrected by Sumita verifier.
Kovacs 2010: growth factors “bFGF and SCF” → HGF, KGF (FGF-7), and SCF (bFGF/FGF-2 was conflated with KGF/FGF-7). Model description also corrected from “fibroblast conditioned media + melanocyte coculture” → “human dermal fibroblasts photoaged in vitro + IHC on human SL tissue sections”. Inline-corrected by Kovacs verifier.
Purohit 2016: J Dermatol Sci 2016;83(2):92-99 → 83(1):80-83. Inline-corrected by Purohit verifier; same fix propagated to dermal-fibroblasts.md footnote.

Other cohort/n-value/site corrections:

Demaria 2014 GCV administration timing: “at time of wounding” → 5 consecutive daily i.p. injections at 25 mg/kg from day 1 to day 6 after wounding; primarily endothelial senescent population (not co-equal with fibroblast).
Demaria 2014 author count: 14 (not ~30 as my brief misestimated).
Fisher 2009 biopsy site: “upper inner arm” → sun-protected buttock skin (Methods p.103).
Fisher 2009 in-vivo MMP-1 protein/mRNA discordance added: 2-fold protein vs 8-fold mRNA elevation in aged dermis (different n=4 mRNA P<0.02 vs n=6 protein P<0.01).
Fisher 2009 in-vivo protein-carbonyl data entirely missing from seeder draft: protein carbonyls 2-fold higher in aged vs young dermis (n=7, P<0.05; Fig 1 — paper’s first reported result).
Hattori 2004 c-KIT n=6 corrected from placeholder; GROα 1.13-fold added for prose consistency.
Inomata 2009 reagent route: “systemic” → subcutaneous injection (after dorsal depilation, 7-week-old mice).
Inomata 2009 doses added: busulfan 40 mg/kg s.c., MMC 4 mg/kg s.c., H₂O₂ 1% × 10 ml/kg s.c.
Bormann 2016 SVM probe set: hedged “not stated” → confirmed complete set of 450k probes (p.565 Methods explicit).
Rodríguez-Paredes 2026 cohort: 95 → 105 enrolled (95 passed pre-array QC; 89 passed post-array; 71 train / 18 internal val / 19 independent test).
Rodríguez-Paredes 2026 license: CC BY-NC 4.0 → CC BY 4.0.
Rodríguez-Paredes 2026 funding: “Beiersdorf-funded” → Projekt Deal OA funding (Beiersdorf’s involvement is via employee authorship + patent/consultation competing interests).
Velarde 2012 SA-βgal KO penetrance added: 9/9 KO vs 1/8 WT (Fisher Exact p<0.05).
Velarde 2012 rotenone SA-βgal: AG21837 keratinocytes ~80% vs HCA2 fibroblasts ~55% at 100 nM/9 days (Supp Fig 1B).
Krutmann 2017 CARET beta-carotene dose: 25,000 IU/day → 30 mg/day (25,000 IU was the retinyl palmitate dose; the two were swapped).
Krutmann 2017 IRA penetration: “30% of total IR reaching dermis” → IRA = 30% of total IRR; 65% of IRA reaches dermis (two distinct stats merged).
Krutmann 2017 COI: 3 L’Oréal-employee authors (AB, GS, BAB) + Krutmann (JK) + Passeron (TP) also received L’Oréal consultancy fees — seeder missed the two additional disclosures.

R38 inter-verifier contradiction resolved (Fisher 1996 ~70%/50-80% figures):

Fisher 1996 verifier read PDF end-to-end and confirmed verbatim text quote on pp.338-339: “with all-trans retinoic acid (t-RA) reduced UVB-induced AP-1 DNA binding by 70% (Fig. 4a). t-RA pretreatment also reduced UVB-induced 92K gelatinase (Fig. 4b) and interstitial collagenase (Fig. 4c) mRNAs, proteins and activities by 50–80%.”
The figures are both text-stated AND figure-derivable — the R38 keratinocytes verifier was wrong to demote these to #gap/unsourced. R38 dermis verifier was correct.
Propagation: cell-types/keratinocytes.md #gap/unsourced tag removed; percentages restored with explicit Fisher 1996 pp.338-339 attribution.

Propagation work completed

Fisher 1996 ~70%/50-80% citation restored on cell-types/keratinocytes.md (R38 erroneous gap-tag removed).
tissues/dermis.md line 99 MMP-3/9 attribution fixed: UV-driven multi-MMP induction now attributed to Fisher 1996 [^fisher1996]; Fisher 2009 attribution narrowed to MMP-1 specifically + secondary-processor framing for MMP-3/9.
Sumita 2018 metadata propagated to skin-aging.md by Sumita verifier (n=60→n=24, pages 1734-1741→1819-1826).
Kovacs 2010 growth-factor list propagated to skin-aging.md by Kovacs verifier (bFGF/SCF → HGF/KGF/SCF; model description also corrected).
Purohit 2016 metadata propagated to skin-aging.md + dermal-fibroblasts.md by Purohit verifier (83(2):92-99 → 83(1):80-83; full 8-author list expanded).
cell-types/melanocyte-stem-cells.md Inomata wikilink fixed: dangling [[studies/inomata-2009-genotoxic-stress-melanocyte-stem-cell]] → [[studies/inomata-2009-melanocyte-stem-cell-atm]] (matches actual filename).
cell-types/melanocytes.md Hattori 2004 footnote: added missing [[studies/hattori-2004-scf-solar-lentigo]] wikilink prefix to footnote line.
frameworks/biological-age-measurement.md TapeLift row updated with correct cohort + cross-ethnic framing attribution (Caucasian-only training, n=105 enrolled, 71/18/19 splits, RMSE figures; “no non-white bias” property attributed to Qi 2026 independent application not Rodríguez-Paredes 2026).

Propagation work confirmed not needed:

Velarde 2012 Li 1995 DOI fix (ng0995-97 → ng1295-376): grep wiki-wide returned no further occurrences beyond the study page itself + log entry.
Fisher 2009 YAP/TAZ wiki sweep: already cleaned during R38 propagation; no remaining instances.
Hair-greying.md Inomata model description: page does NOT cite Inomata directly; no K14-SCF reference to fix.

Schema escalations surfaced

publication-type: field for type: study — Purohit 2016 is a Letter format (distinct from research article), with significantly less methods detail + no structured abstract. Recommended values: research-article | letter | brief-communication | review | meta-analysis | case-report | editorial | preprint. Adding to next CLAUDE.md cleanup queue.
volume: + issue: + pages: as optional type: study fields — currently in use de facto (per multiple R39 seeders) following Horvath 2013 precedent but not formalized in CLAUDE.md.

Forward queue surfaced by R39

Dong et al. 2025 (Nat Cell Biol; PMID 41053225) — “seno-differentiation” concept partially updates Inomata 2009’s “differentiation NOT senescence” framing for McSC fate. Candidate for context update on cell-types/melanocyte-stem-cells.md + phenotypes/hair-greying.md. Add to R45.
molecules/proteins/sod2.md — implicit stub surfaced from Velarde 2012 study page; no SOD2 protein page exists.
pathways/hippo-yap-taz.md — surfaced from Fisher 2009 study page (paper does NOT use this pathway but the wider mechanosensing literature does); could be useful for cross-reference.
molecules/compounds/ultraviolet-radiation.md — surfaced as implicit stub on Fisher 1996 + Krutmann 2017 pages (but UV is not a “compound” — could be a processes/uv-radiation.md instead).
Additional skin-aging-cluster studies that R39 surfaced as worth seeding eventually: Nishimura 2005 (Science; closed-access; canonical McSC ectopic-differentiation paper); Liebmann 2010 (since R39 caught Krutmann’s misattribution of red-light/IRA findings, the underlying primary paper warrants a dedicated extraction).

Round disposition

R39 closed cleanly. 12/12 study pages flipped to verified: true. ROADMAP.md updated. Ready to dispatch R40 (effector proteins + wound-healing process, 12 pages).

[2026-05-19] verify | studies/krutmann-2017-skin-aging-exposome.md (R39 full-PDF verification)

Pages verified: 1 — studies/krutmann-2017-skin-aging-exposome.md.

Source: J Dermatol Sci 2017;85(2):152–161, doi:10.1016/j.jdermsci.2016.09.015. Full PDF read (10 pages). OA (CC BY-NC-ND); local PDF confirmed present.

Corrections applied (10):

CARET dose error: beta-carotene “25,000 IU/day” → 30 mg/day; 25,000 IU is the retinyl palmitate dose. PDF p.156 explicit.
Liebmann 2010 red-light claim inverted: wiki said red light “reduce proliferation and promote differentiation” → PDF (ref [25], p.155) says red light and IRA have no impact; only blue light reduces proliferation.
IRA penetration figure conflated: “30% of total IR reaching dermis” → corrected to “IRA = 30% of total IRR; 65% of IRA reaches the dermis” (two distinct figures from PDF p.154).
Fisher 1996 attribution removed: wiki claimed “Krutmann et al. cite Fisher 1996” but Fisher 1996 does not appear in Krutmann reference list. Sub-erythemogenic claim supported via review-paper block refs [13–19]. Gap tag added.
Rowe 2000 → Rowe/Guyuron 2010: twin-study book chapter is Springer Textbook of Aging Skin, 2010, pp. 441–446.
Richardson 1987 author/ref fix: seeder wrote “Rickards D, ref [46]” → correct is “Richardson D, ref [47]” (Arch Environ Health 42(1) 1987 19–25).
Doshi 1997 → Doshi 2007: actual year of Arch Dermatol twin-smoking paper is 2007; original wiki DOI belongs to a 1997 paper.
Yamai 2001 → Tamai 2014: first author is Tamai Y (not Yamai T); year is 2014 (Tob. Control); Krutmann ref [42].
Lohan 2015 journal/year fix: journal is Exp. Dermatol. 25(5) (2016) 380–385, not J Dermatol Treat 2015; Krutmann ref [28].
Thiele 1997 journal fix: journal is J. Invest. Dermatol. (not J. Clin. Invest.); DOI format 10.1172/JCI… does not match a JID paper.

Additional: COI disclosure expanded — JK and TP also received L’Oréal consultancy fees (PDF p.159). Hüls 2016b/2016c footnotes corrected from doi:10.1038/jid.2016.35 (belongs to ref [31] only) to conference-abstract citations for refs [32] and [37].

Supersession check: no superseding meta-analyses or large RCTs for skin photoaging/sunscreen found 2020–2026. literature-checked-through: 2026-05-19.

Downstream (for main agent): pages citing Fisher 1996 as “Krutmann-cited” for sub-erythemogenic MMP claim should be reviewed; Rowe 2000 citations should be updated to 2010 book chapter.

[2026-05-19] verify | studies/rodriguez-paredes-2026-tapelift-clock.md (R39 full-PDF verification)

Pages verified: 1 — studies/rodriguez-paredes-2026-tapelift-clock.md.

Source: npj Precision Oncology 2026;10:89, doi:10.1038/s41698-026-01302-7. Full PDF read (10 pages). Gold OA (CC BY 4.0). Local PDF confirmed at. DOI confirmed via Crossref.

Corrections applied (8):

Initial cohort n corrected: “95 donors initially enrolled” → 105 donors initially enrolled. QC was two-stage: 95/105 passed pre-array ΔCt <3 qPCR; 89/95 passed post-array p-value QC (>75% probes p<0.05).
Training/validation splits added: 80/20 random partition → N=71 training, N=18 internal validation (not “89 training/19 validation” as in frontmatter n-subjects comment). N=19 is a separate independent cohort.
Elastic-net clock CpG count confirmed: 157 CpGs (was “#gap/needs-pdf-verification”).
PC clock CpG count confirmed: 5,021 CpGs (was partially stated but unconfirmed).
Independent validation RMSE values corrected and split: elastic net RMSE 6.2y, PC clock RMSE 5.1y in independent 19-donor cohort (wiki had “~5.1–6.2 yr (range)” applied to both identically).
OA license corrected: CC BY-NC 4.0 → CC BY 4.0 (confirmed from Crossref license assertion and PDF page 10).
Funding statement corrected and expanded: “Beiersdorf AG / DKFZ collaboration” is correct for authorship but funding for OA publication was Projekt Deal; competing interests (patent applications, Beiersdorf employment/consultation/speaker fees) documented in full.
High-confidence probe set count added: N=535,384 CpGs after four-stage filtering pipeline (Fig. 1B).

Claims confirmed without change: forehead sampling site; 4 adhesive strips; ~49 ng gDNA yield; RMSE 5.8y (elastic net internal validation) and 4.3y (PC clock internal validation); 50 TSGs with age-associated hypermethylation (full list Table S7); named TSGs (ZIC1, RBM38, UCHL1, FOXA2, ID4) with Pearson r values; Caucasian-only Fitzpatrick I–IV training; cross-ethnic validation comes from Qi 2026, not this paper; 859 TSGs analyzed (from TSGene 2.0 n=1217 filtered); mitotic-age AK→cSCC progression signal.

Downstream pages updated directly (stub footnote correction only):

phenotypes/skin-aging.md — footnote [^rodriguezParedes2026] updated from “primary source not yet retrieved” placeholder to correct DOI, n, CpG counts, RMSE, and correct Caucasian-only attribution note.

Downstream pages for main agent propagation (not edited here):

phenotypes/skin-aging.md line 118: clock table row “also no non-white bias in Qi 2026 validation [^rodriguezParedes2026]” — framing is acceptable (correctly attributes to Qi 2026) but footnote was placeholder; now fixed. Body framing needs no change.
tissues/epidermis.md line 110: “No non-white bias per Qi 2026 validation” — framing correct; footnote reference not present (links to Qi 2026 footnote, which is appropriate).
frameworks/biological-age-measurement.md line 76: “Independent comparator; no non-white bias in cross-ethnic validation” — no citation on this cell; should add [^rodriguezParedes2026] pointing to study page, and clarify that cross-ethnic validation was performed by Qi 2026, not this paper itself. Flag for propagation pass.

Supersession check (R25): study is from 2026; no superseding meta-analysis or large RCT applicable to this specific TapeLift clock. Not applicable (type: study, single paper). literature-checked-through: 2026-05-19.

[2026-05-19] verify | studies/inomata-2009-melanocyte-stem-cell-atm.md (R39 full-PDF verification)

Pages verified: 1 — studies/inomata-2009-melanocyte-stem-cell-atm.md.

Source: Cell 2009;137(6):1088–1099, doi:10.1016/j.cell.2009.03.037. Full PDF read (12 pages including all figures + methods). OA bronze; local PDF confirmed present.

Corrections applied (5):

Primary model corrected: K14-SCF stated as primary model → Dct-lacZ transgenic mice are the primary system throughout all main figures. K14-SCF is not the dominant model. Model Systems section rewritten; XPD TTD, Trp53−/−, Ink4a/Arf−/−, Mc1re/e genetic controls added.
Route corrected: “systemic administration” → “subcutaneous injection” (s.c. on dorsal skin, per Methods).
Genotoxin doses added to Methods section and footnote: busulfan 40 mg/kg b.w., MMC 4 mg/kg b.w., H₂O₂ 1% in PBS 10 ml/kg b.w. (confirmed from Methods and Figure 1 legend).
Per-group n limitation reworded: “not in archive abstract” → “not stated in main text” with confirmation SEM/Student’s t test used; likely in Supplemental Experimental Procedures.
Footnote updated: model descriptor corrected; doses added; MITF/TYRP1/TYR upregulation + no-TUNEL/no-caspase-3/no-SA-β-gal confirmation added.

Claims confirmed without change: ATM direction (protective; 3 Gy Atm−/− vs 5 Gy WT confirmed via Figure 6A/K text); differentiation-not-apoptosis conclusion; MITF as master differentiation driver; aged WT (22M) EPM phenocopy (Figure 1X); all 11 authors in correct order.

Supersession check (R25): PMID 41053225 (Dong et al., Nat Cell Biol 2025) — introduces “seno-differentiation” (senescence-coupled differentiation) framing. Partially updates Inomata 2009 but does not contradict ATM-protective or ectopic-differentiation axis. Flagged for downstream propagation to melanocyte-stem-cells page. PMID 38099887 (JID 2024 review) and PMID 38033857 (Front Cell Dev Biol 2023): cite Inomata 2009 as foundational; no contradiction. literature-checked-through: 2026-05-19.

Downstream pages for main agent propagation: [[melanocyte-stem-cells]] (K14-SCF model correction; 2025 seno-differentiation flag), [[hair-greying]] (model correction), [[atm]] (ATM direction confirmed correct per R38; model description may need update).

[2026-05-19] verify | studies/fisher-2009-collagen-fragmentation-mmp.md (R39 full PDF verification)

Pages verified: 1 — studies/fisher-2009-collagen-fragmentation-mmp.md.

Source: primary PDF at (14 pages; Am J Pathol 2009, Vol.174 No.1, pp.101–114). Full read end-to-end.

Corrections applied (6):

MMP-3/MMP-9 misattribution removed — paper presents NO quantitative MMP-3/MMP-9 data; mentioned in discussion only as potential downstream amplifiers.
In-vitro n range corrected: n=3–5 (not 3–6 as briefed).
Biopsy site corrected: buttock (not “upper inner arm”).
In-vivo protein oxidation finding added: carbonyls twofold higher in aged dermis, n=7, P<0.05 (Fig 1).
MMP-1 protein in-vivo data added: twofold elevated, n=6, P<0.01 (Fig 2B).
MitoQ comparator data added: MitoQ₅/MitoE₂/NAC substantially less effective (Fig 9B); n=5 rescue confirmed.

R38 priority checks confirmed: YAP/TAZ absent; “~1%/yr collagen loss” absent. Supersession check: no meta-analyses or large RCTs found (2022–2026 PubMed).

Downstream pages that may cite MMP-3/MMP-9 claims from this paper — check at next propagation pass:

cell-types/dermal-fibroblasts.md, tissues/dermis.md, phenotypes/skin-aging.md

[2026-05-19] verify | studies/velarde-2012-mitochondria-skin-senescence.md (R39 full-PDF verification)

Pages verified: 1 — studies/velarde-2012-mitochondria-skin-senescence.md.

Source: open-access PDF at. Full 12-page PDF read.

Corrections: (1) Li 1995 DOI 10.1038/ng0995-97 → 10.1038/ng1295-376; (2) SA-βgal KO penetrance 9/9 KO vs 1/8 WT added; (3) n-subjects Western blot clarified (SOD2 confirmation n=8/9; senescence Westerns n=6/6); (4) rotenone SA-βgal values quantified ~80% vs ~55% at 100 nM/9 days; (5) KO harvest age 17-20 days added.

All 10 R38 priority checks confirmed correct. No supersession candidates (2022-2026 PubMed).

[2026-05-19] verify | studies/sumita-2018-tretinoin-photoaging.md (R39 n/pages metadata correction)

Pages verified: 1 — studies/sumita-2018-tretinoin-photoaging.md.

Source is permanently closed-access (not_oa confirmed). Verification scope: canonical-DB metadata verified via PubMed efetch (PMID 29704456) + Crossref (doi:10.1111/jdv.15020) on 2026-05-19. Abstract text also confirms: 1 dropout (per-protocol n=23, 46 forearms), P<0.05 for the three divergent-endpoint findings, cream vs. peel directions on corneal thickness/dermis echogenicity/Ki67 specified explicitly. Quantitative body claims beyond the abstract remain unverifiable (#gap/no-fulltext-access). Supersession check performed: two systematic reviews found (PMID 39348007, 2024; PMID 40707570, 2025) — both confirm tretinoin as gold-standard; no supersession detected.

Corrections applied (2 total across 2 pages):

studies/sumita-2018-tretinoin-photoaging.md: verified flag flipped true with closed-access partial scope; dropout (per-protocol n=23, 46 forearms) documented; HRT-exclusion criterion added; P<0.05 for three divergent endpoints added; Ki67 directionality specified per abstract. literature-checked-through: 2026-05-19.
phenotypes/skin-aging.md footnote [^sumita2018]: pages 1734-1741 → 1819-1826; n=60 → n=24 women enrolled (23 per-protocol; 1 dropout); 48 forearms bilateral within-person design; correction annotation added.

[2026-05-19] verify | studies/bormann-2016-epidermis-clock.md (R39 full PDF verification)

Pages verified: 1 — studies/bormann-2016-epidermis-clock.md.

Source: primary PDF confirmed present at local DOI lookup. Gold OA; PMID 27004597; PMC4854925.

Corrections applied (1 total):

Methods § Array platform: hedged phrase “exact post-filtering probe count is not explicitly stated in the main text (full probe list in supplement)” replaced with confirmed text: paper explicitly states “complete set of 450k probes as independent variables” (Methods, p.565); preprocessing steps (minfi, sex-chromosome removal, SNP/cross-reactive probe filtering) also documented from paper text.

All other quantitative claims confirmed exact:

Pages 563–571 (Crossref + PubMed efetch + PDF header confirmed).
13 authors (Crossref + PubMed both return 13; list matches).
n=48 punch biopsy (18–27 yr young n=24, 61–78 yr old n=24) + n=60 suction blister (20–79 yr) = 108 total. Age ranges confirmed from Methods p.569.
LOOCV ρ=0.92, p<2.2×10⁻¹⁶, MAE <5.25 yr — confirmed (p.565 body text + Fig. 3A).
Vandiver validation: Bormann ρ=0.96, p=1.66×10⁻¹⁰, R²=0.93, MAE=6.72 yr; Horvath ρ=0.89, p=7.6×10⁻⁷, R²=0.794, MAE=12.1 yr — all confirmed exact (p.565 body text).
Spatial autocorrelation P=5.4×10⁻⁷ — confirmed (p.566).
Intramethylome variance P=0.0016 — confirmed (p.566).
Vandiver 2015 complete holdout (not used in training) — confirmed.

Supersession check (R25): PubMed search 2023–2026 for epidermis DNA methylation clock aging returned 3 papers (PMIDs 42034839, 41593290, 41387742) — all are extension/application papers that build on or apply Bormann’s framework; none supersede or contradict the core findings. literature-checked-through set 2026-05-19.

Downstream propagation needed: none for this correction (the SVM probe-set hedge correction affects body text only; no downstream entity pages carry that specific phrasing). The Vandiver and LOOCV metric values cited downstream in biomarkers/bormann-epidermis-clock-2016.md (if/when seeded) and studies/qi-2026-dhm-epigenetic-skin-aging.md are all confirmed correct.

[2026-05-19] verify | studies/kovacs-2010-fibroblast-solar-lentigo.md (R39 growth-factor correction)

Pages verified: 1 — studies/kovacs-2010-fibroblast-solar-lentigo.md.

Source is permanently closed-access (not_oa confirmed). Verification scope: growth-factor identity + canonical-DB metadata verified via PubMed abstract (PMID 20662835) and Crossref (doi:10.1111/j.1365-2133.2010.09946.x). Quantitative claims (sample sizes, fold-changes, p-values) remain unverifiable (#gap/no-fulltext-access). Supersession check performed — no superseding meta-analysis or RCT found; 2020 review (PMID 32243635) corroborates HGF/KGF/SCF framing.

Corrections applied (2 total across 2 pages):

studies/kovacs-2010-fibroblast-solar-lentigo.md: verified flag flipped to true with partial-scope note. Internal note on bFGF/KGF terminology discrepancy updated from hedged “should be corrected pending PDF” to confirmed correction. Relationship-to-wiki-pages entry updated accordingly.
phenotypes/skin-aging.md footnote [^kovacs2010]: “bFGF and SCF” → “HGF, KGF (FGF-7), and SCF”; description updated to reflect IHC + photoaged-fibroblast model (not “conditioned media + coculture”); KGF→keratinocyte→SCF relay added; correction note appended.

Downstream: no further propagation needed — skin-aging.md was the only downstream page citing this study footnote with the bFGF attribution. No other entity pages carry the bFGF/Kovacs conflation.

[2026-05-19] verify | studies/purohit-2016-smad3-fibroblasts.md (R39 metadata correction)

Pages verified: 1 — studies/purohit-2016-smad3-fibroblasts.md.

Source is permanently closed-access (not_oa confirmed). Verification scope: canonical-DB metadata only (PubMed efetch PMID 27132061 + Crossref doi:10.1016/j.jdermsci.2016.04.004). Quantitative body claims remain unverifiable (#gap/no-fulltext-access).

Corrections applied (3 total across 3 pages):

studies/purohit-2016-smad3-fibroblasts.md: two ⚠️ banners replaced with single permanent closed-access notice; verified flipped true with partial-scope note; verified-scope documents PDF-unverifiable status.
phenotypes/skin-aging.md footnote [^purohit2016]: 83(2):92-99 → 83(1):80-83; added study page wikilink + Letter tag.
cell-types/dermal-fibroblasts.md footnote [^purohit2016]: 83(2):92-99 → 83(1):80-83; expanded to full 8-author list; added study page wikilink + Letter tag.

Both downstream corrections propagated inline (same session; minor citation-metadata fix, not synthesis change).

[2026-05-19] verify | studies/fisher-1996-photoaging-ap1-mmp.md (R39 canonical resolution)

Pages verified: 1 — studies/fisher-1996-photoaging-ap1-mmp.md. Full PDF read (all 5 pages). Local copy confirmed.

R38 inter-verifier contradiction resolved: both “~70% AP-1 reduction” and “50–80% MMP reduction” figures are explicitly text-stated on pp.338–339, not solely figure-derivable. Exact text: “with all-trans retinoic acid (t-RA) reduced UVB-induced AP-1 DNA binding by 70% (Fig. 4a). t-RA pretreatment also reduced UVB-induced 92K gelatinase (Fig. 4b) and interstitial collagenase (Fig. 4c) mRNAs, proteins and activities by 50–80%.”

Corrections applied (2): (1) MMP-2/t-RA “not altered” page attribution p.338 → p.339; (2) removed misleading parenthetical on Fig. 4c n=7 activity — it IS in the figure legend (paper text). All 7 priority checks confirmed. Page flipped to verified: true.

🧬 Aging Wiki

Explorer

R39