2026-05-27

log/2026-05-27 — ad-hoc daily entries

Sub-file of log — see parent for index.

[2026-05-27] verify | HYAL2 + HA + skin-aging batch (Sun 2025 anchor)

• verified: molecules/proteins/hyal2.md → verified: true (2026-05-27, claude)
• verified: molecules/compounds/hyaluronic-acid.md Sun 2025 ADDENDUM → scope updated (already verified: true; addendum cross-checked)
• verified: phenotypes/skin-aging.md Sun 2025 ADDENDUM → scope updated (already verified: true; addendum cross-checked)
• Sun 2025 (doi:10.3389/fmed.2025.1529936) PDF
• Corrections made (2 factual):
  1. Acute UVB dose framing: “UVB 300 mJ/cm² × 5 days” → “300 mJ/cm² total over 5 days (5 MEDs at 100 µW/cm²)” — the original implied per-day dosing; paper specifies total dose over the 5-day protocol. Corrected in body text of hyal2.md and in all three [^sun2025] footnotes across all three pages.
  2. Chronic UVB protocol: “10-wk UVB” → two-phase detail added: 120 mJ/cm²/wk × 4 wk, then 180 mJ/cm²/wk × 6 wk; 130 MED total (per Figure 4A and Methods). Added to all three footnotes.
  3. Open Targets AB modality: added “GO CC high conf TRUE” to druggability section (confirmed via live API query); previously missing from page.
• No corrections needed to:
  • He 2026 AKT direction: wiki says “suppresses AKT” — confirmed correct (abstract: LMW-HA suppresses/inhibits AKT phosphorylation)
  • Sinova 2023 HYAL2 upregulation claim: confirmed. (Note: abstract also shows acute UV increases HA in human epidermis — consistent with existing gap/contradictory-evidence tag)
  • Lee 2026 directional claim: confirmed (HYAL1+HYAL2 downregulated by treatment)
  • Zych 2023 directional claim: confirmed (HYAL2 upregulated by Sphagnum extracts)
  • All UniProt canonical IDs: confirmed (Q12891, 473 aa, 1-20 signal peptide, 21-448 mature chain, Asp135 active site, 5 disulfide bonds, 3 N-glycosylation sites, EGF-like domain 361-439, GPI anchor Gly448)
  • NCBI gene 8692, HGNC 5321, Ensembl ENSG00000068001: confirmed via Open Targets API (target returned correctly)
  • n=3 per group: confirmed from all figure legends
• Supersession check (R25): PubMed search 2024-2026 “HYAL2 AND skin aging/photoaging” returned 3 hits: Sun 2025 (anchor paper), Lee 2026 (already cited), and PMID 41955380 (2026 Science — HYAL2 at blood-brain barrier, not skin aging; not a supersession). No meta-analysis or RCT found for HYAL2 + aging endpoints. literature-checked-through: 2026-05-27 confirmed on hyal2.md.

[2026-05-27] verify | escin / horse chestnut seed extract

• verified: molecules/compounds/escin.md → verified: true (2026-05-27, claude)
• Pittler 2012 Cochrane: full PDF read end-to-end (17 pages including results, discussion, appendices)
• Diehm 1996 Lancet: abstract-verified via PubMed efetch (PDF not downloadable — bronze OA, no PMC; DOI lookup attempt returned “no URLs after filtering”)
• Yang 2024 (PMID 38670962): abstract-verified; n confirmed as 87 (was “NR” on seeded page)
• Kuznetsov 2025 (PMID 41805617): abstract-verified; n confirmed as 180 (was not stated on seeded page)
• CAS numbers resolved via PubChem synonyms REST: CID 6433489 → 6805-41-0; CID 6540709 (β-aescin) → 8047-15-2 (gap/needs-canonical-id cleared)
• Supersession check (R25): 11 PubMed hits 2019–2026; no Cochrane update to Pittler 2012 found; Martínez-Zapata 2020 (69 RCTs, class-level phlebotonics) already cited on page; one potentially relevant 2021 Phlebology network meta-analysis (PMID 33983078, sulodexide vs phlebotonics, 45 RCTs) identified — addresses drug class but does not supersede escin-specific findings; no action needed
• Corrections made (6 factual):
  1. WMD CI bounds inverted: “−50.72 to −13.49” → “13.49 to 50.72” (correct direction; positive = reduction favouring HCSE)
  2. Trial count: “~17 RCTs” → “17 RCTs” (exact count confirmed)
  3. Adverse event frequency: “<3%” → “ranged 1–36% across studies; 4 studies reported no treatment-related events” (source says 14 of 17 trials reported AEs; frequency 1–36%)
  4. Yang 2024 n: “NR” (not reported) → 87 (explicitly stated in abstract)
  5. Kuznetsov 2025 n added: 180 female patients (was absent from body text)
  6. Mechanism: hyaluronidase inhibition only → hyaluronidase AND elastase inhibition (both explicitly cited in Pittler 2012 Discussion, attributed to Lorenz 1960)
• CAS number added to frontmatter (cas-number: 6805-41-0 for canonical mixture)
• verified-scope: Pittler 2012 full PDF; Diehm 1996 abstract-only; Yang 2024 and Kuznetsov 2025 abstract-only; mechanism claims review-cited (Pittler 2012 Discussion); PK parameters not independently verified against primary PK studies
• literature-checked-through confirmed: 2026-05-27

[2026-05-27] ingest | escin / horse chestnut seed extract

• added: molecules/compounds/escin.md (type: compound; verified: false ⚠️ auto-extracted)
• context: user-requested ad-hoc seed; venotonic phlebotonic compound for chronic-venous-disease; NOT added to ROADMAP per ad-hoc discipline
• canonical IDs resolved via web (not training memory):
  • PubChem CID 6433489 (aescin mixture, canonical) + CID 6540709 (β-aescin / Escin Ia) — confirmed via PubChem REST
  • ChEMBL507233 (Escin Ia) + CHEMBL505939 (Escin Ib) — confirmed via ChEMBL REST
  • CAS not confirmed by PubChem/ChEMBL REST; needs-canonical-id filed on page
  • DrugBank absent (herbal extract, not in DrugBank)
• anchor citations: Pittler 2012 Cochrane; Diehm 1996 Lancet (DOI lookup confirmed, PDF download pending)
• recency search (R25): PubMed date-filter 2019–2026, terms: escin OR aescin OR “horse chestnut seed extract” with [meta-analysis OR RCT OR systematic review] filter → 11 hits; most relevant: Yang 2024 (Vascular, post-ablation RCT, PMID 38670962) and Kuznetsov 2025 (post-RFA RCT, PMID 41805617) — both procedural-adjunct contexts, NOT primary CVI RCTs. No major new RCT or systematic review superseding Pittler 2012 found. literature-checked-through: 2026-05-27 set.
• clinical trials: ClinicalTrials.gov v2 API query (escin OR aescin OR horse chestnut OR Aesculus, RECRUITING + ACTIVE_NOT_RECRUITING) → 2 active trials; neither targets primary CVI (both procedural-adjunct edema). clinical-trials-active: 2 set.
• new mechanism class added to frameworks/intervention-classes.md: venotonic / phlebotonic (canonical values: venotonic, hyaluronidase-inhibition, venous-smooth-muscle-sensitization) — required by R16 before shipping compound page
• implicit stubs created (new wikilinks to non-existent pages): none — all linked pages exist (chronic-venous-disease, veins, frameworks/intervention-classes)
• gaps surfaced: CAS number unconfirmed; DrugBank absent; PK parameters (Tmax, t½, protein binding) sourced from secondary review, not primary PK studies; molecular target of venotonic mechanism undefined; no active primary CVI trial; no CEAP-progression endpoint RCT published or planned
• schema note: hallmarks: [] left empty intentionally — no direct hallmark-level mechanistic evidence for escin; aging relevance is via phenotype link (chronic-venous-disease hallmark mapping), not compound-level hallmark engagement
• verification priority for wiki-verifier: (1) Pittler 2012 Cochrane — local PDF available; verify WMD −32.1 mL and CI, trial count, safety verdict; (2) Diehm 1996 Lancet — PDF pending download; verify leg volume numbers (43.8 mL, 46.7 mL, 9.8 mL), n=240, p-values; (3) mechanism-level claims (hyaluronidase inhibition, PGF2α sensitization) — need primary pharmacology papers located and verified

[2026-05-27] synthesis | escin disease-modification plausibility (hypothesis)

• user question: are there trials showing NO CEAP-progression prevention, or no trials at all? + is there a plausible disease-modification MOA?
• answer: untested, not refuted — all HCSE trials ≤16 wk (Diehm 12 wk; Cochrane modal ≤12 wk), endpoints are oedema surrogates; CVI progresses ~1.4%/yr (C2) to ~4.3%/yr overall (Edinburgh, on chronic-venous-disease) — far too slow for a 12-wk window. PubMed sweep (escin + progression/long-term/CEAP) found no long-term HCSE RCT; off-patent status removes funding incentive. Distinct from limited-negative (STACs) — escin is #gap/long-term-unknown.
• added: ## Disease-modification plausibility (hypothesis — not clinically tested) subsection to molecules/compounds/escin.md — maps escin onto 4 nodes of the CVI progression loop (venotonic / leukocyte-adhesion / NLRP3 inflammasome / MMP-9 suppression), with extrapolation table + #gap/needs-human-replication
• new footnotes (all abstract-verified via PubMed efetch, full PDFs NOT read; all non-peripheral-vein models): Sun 2023 (PMID 37716034, escin ↓MMP-9 via AMPK/Cav-1, stroke BBB, incl. 12-mo mice); Yan 2025 (PMID 40090427, escin Ia ↓MMP-9 via LOXL2, DSS colitis); Li 2025 (PMID 40093806, escin ↓MMP-9 + ↓NLRP3 pyroptosis, rat sagittal-sinus thrombosis — closest venous model); Barreca 2024 (PMID 39273200, EBGS combo ↓endothelial adhesion + MMP-2/-9); + Raffetto 2021 cross-ref footnote for human vein-wall MMP claim
• verified-scope updated with addendum documenting the abstract-verification level of the new subsection; page remains verified:true (core claims full-PDF; hypothesis subsection explicitly flagged abstract-level + non-venous extrapolation)
• mechanistic load-bearing claim: 3 independent in-vivo rodent models show escin downregulates MMP-9; MMP-9 is among proteases driving human vein-wall ECM breakdown → coherent but unproven disease-modification hypothesis; strengthens existing next-experiment field
• follow-up gap: full-PDF verification of the 4 mechanism papers + a primary-source search for escin effects specifically in peripheral venous tissue remain open needs-verification

[2026-05-27] verify | escin disease-modification mechanism footnotes (full PDF pass)

Pages updated: molecules/compounds/escin.md (verified: true maintained; verified-scope addendum updated)

PDFs obtained:

• Li 2025 (doi:10.7150/ijms.102624, PMID 40093806): gold OA — downloaded from PMC (PMCID PMC11905272); full PDF read end-to-end
• Barreca 2024 (doi:10.3390/ijms25179252, PMID 39273200): gold OA — downloaded from PMC (PMCID PMC11395133); full PDF read end-to-end
• Sun 2023 (doi:10.1016/j.phymed.2023.155071, PMID 37716034): closed-access; not_oa per a local paper archive + Unpaywall; remains abstract-verified only
• Yan 2025 (doi:10.1016/j.jep.2025.119623, PMID 40090427): closed-access; not_oa per a local paper archive + Unpaywall; remains abstract-verified only

Corrections and enrichments:

1. Li 2025 — no corrections to core claims; significant enrichment:

   • Confirmed: MMP-9 DOWN (Western blot, SSST group vs SSST+Escin, p<0.05) — correct as abstract-verified
   • Confirmed: NLRP3/Caspase-1/IL-1β/IL-18/GSDMD ALL DOWN (Western blot; p<0.05) — correct
   • Confirmed: no sinus recanalization — LSCI CBF: no significant difference between SSST and SSST+Escin (p=0.931) — correct
   • Enriched: model detail added — Sprague-Dawley rats, 200–250 g, 2-month-old male, FeCl₃ induction; dose = 3 mg/mL × 1 mL × 2/day × 1 week gavage; n=12 behavioral / n=6–8 molecular assays
   • Enriched: neuronal survival data added (NeuN+ immunohistochemistry, ↑neurons in SSST+Escin, p<0.905 power 3.322)
   • Footnote upgraded from abstract-verified to full PDF verified

2. Barreca 2024 — one CORRECTION to the abstract-verified characterisation:

   • CORRECTION: Wiki footnote said “escin within a 4-compound combination — escin-specific contribution not isolable.” This is partially wrong for the platelet adhesion endpoint: Figure 3B (adhesion assay) shows all four single compounds including escin alone significantly reduced platelet adhesion to TNFα-activated HUVEC (p<0.0001 vs TNFα). Escin-specific attribution IS established for platelet adhesion.
   • The “not isolable” caveat remains accurate for the MMP-2/-9 endpoint — only EBGS combination MMP data is reported (ELISA Figure 5B); no single-compound MMP data.
   • CLARIFICATION: Authors’ Conclusions section states “compounds singularly had no significant effects” — this appears to refer to the full multi-endpoint profile; contradicts Figure 3B for the platelet-adhesion-only endpoint. Both noted in footnote.
   • Confirmed: EBGS ↓MMP-9 in TNFα+Platelets condition (p<0.005); EBGS ↓MMP-2 (both ± platelets)
   • Confirmed: FAK ↓, VCAM-1 ↓ (Western blot); P-selectin/GP1b/integrins αV,β3 ↓ (RT-PCR); IL-6/IL-8 ↓ (ELISA)
   • Model clarified: HUVEC = human umbilical vein endothelial cells (venous endothelium, not peripheral vein wall)
   • Dose: escin 0.25 µg/mL in EBGS combination (each of 4 compounds); single-compound arms also tested at this concentration
   • Footnote upgraded from abstract-verified to full PDF verified

3. Sun 2023 — AMPK direction confirmed from structured abstract (critical for pleiotropy question):

   • AMPK direction: ACTIVATION — escin increases phospho-AMPK (↑), which activates Cav-1 (↑), which suppresses MMP-9 (↓)
   • Causal chain confirmed: AMPK inhibitor + Cav-1 siRNA experiments used to demonstrate dependency
   • The wiki did not previously state the AMPK direction explicitly — now added to both table and footnote
   • Remains abstract-verified only (closed-access); no change to verification level

4. Yan 2025 — LOXL2/MMP-9 axis confirmed from structured abstract:

   • LOXL2 overexpression rescue confirmed: “elevated LOXL2 expression diminished the compound’s protective effects”
   • Direction: escin suppresses LOXL2 → reduced LOXL2 → ↓MMP-9
   • Remains abstract-verified only (closed-access); no change to verification level

Peripheral venous tissue search (the “extra research”):

• PubMed search (escin OR aescin) AND (saphenous OR varicose OR “venous wall” OR “venous smooth muscle” OR “venous endothel” OR phlebology OR “vein wall”): 58 results
• Europe PMC searches (escin + MMP + venous; escin + varicose/saphenous): 0 results each
• PubMed search (escin OR aescin) AND (MMP OR “matrix metalloproteinase”): 14 results, none in peripheral venous tissue
• Result: NO study exists testing escin’s MMP effects in peripheral venous tissue — confirms and strengthens the gap
• NEW PAPER FOUND: Gwozdzinski K et al., Biomedicines 2023;11(6):1702 (PMID 37371797; doi:10.3390/biomedicines11061702) — escin tested ALONE in human varicose vein endothelial cells (HVVEC); escin reduced adhesive properties of HVVEC; this is the ONLY identified study testing escin directly in peripheral varicose-vein endothelial tissue. Added as [^gwozdzinski2023] footnote; table and extrapolation rows updated accordingly.

Page changes summary:

• Blockquote intro updated: “non-venous tissues” → “primarily non-peripheral-venous tissues” + acknowledged HUVEC (umbilical vein) and Gwozdzinski 2023 (varicose-vein ECs)
• Mechanism table row “Leukocyte/platelet–endothelial adhesion”: corrected caveat (escin-alone arm exists for platelet adhesion; MMP contribution non-isolable); added Gwozdzinski 2023 reference
• Mechanism table row “MMP effector”: added AMPK activation direction explicitly; corrected “3 independent in-vivo models” → “2 independent in-vivo rodent models” (Li 2025 is the third but is NLRP3/pyroptosis, not the AMPK pathway; the MMP row correctly now cites Sun 2023 and Yan 2025 only)
• Narrative paragraph updated with confirmed AMPK activation direction and Gwozdzinski 2023
• Extrapolation table “Pathway conserved?” row updated to reflect Gwozdzinski 2023 partial evidence in venous tissue
• Closing paragraph: ex-vivo varicose-vein wall explant suggestion added as feasible intermediate step
• Footnotes: [^li2025sss] and [^barreca2024] upgraded to full-PDF-verified with detailed methods/results; [^sun2023] and [^yan2025] enriched with AMPK direction / LOXL2 rescue details; [^gwozdzinski2023] added (new)
• verified-scope addendum updated to document the full PDF pass

[2026-05-27] research + synthesis | escin systemic-mechanism collateral effects

• user question: AMPK & MMP act in many tissues — what negative side effects could accompany the venous benefit?
• full-PDF verification pass (delegated): Li 2025 + Barreca 2024 upgraded to full-PDF-verified (gold OA via PMC); Sun 2023 + Yan 2025 remain abstract-only (closed-access, no OA). Corrections: Barreca — escin ALONE ↓platelet adhesion (Fig 3B, p<0.0001); combo-only caveat applies just to MMP-2/-9 endpoint. MMP-9 narrative “3 models” → “2 models” (Li’s primary claim is NLRP3, not MMP-9). AMPK direction confirmed = ACTIVATION (↑p-AMPK→↑Cav-1→↓MMP-9; AMPK-inhibitor + Cav-1-siRNA dependency experiments).
• peripheral-venous search result: no escin+MMP study in peripheral venous tissue exists (PubMed n=14, Europe PMC n=0). One direct study found + added: Gwozdzinski 2023 (PMID 37371797, Biomedicines) — escin alone in human varicose-vein endothelial cells (HVVEC), ↓adhesive properties, no MMP data. [
• added: ### Systemic-mechanism collateral effects (disease-modification scenario) subsection — reconciliation (low oral bioavailability + ≤16-wk trials explain “mild/infrequent” safety; disease-mod regimen needs sustained/higher exposure → exits validated envelope); MMP-9 wound-healing irony + marimastat MSS precedent does NOT apply (Fields 2019: selective MMP-2/-9/-13 inhibition doesn’t cause MSS; escin is indirect signaling-level modulator, not zinc chelator); AMPK anabolic/sarcopenia caveat (proportionate); NLRP3→infection trade-off (IL-1 blockade/CANTOS); saponin hemolysis + antiplatelet (dose/route-gated, explains oral≫IV).
• new footnotes (title/abstract-verified — all 6 PMIDs confirmed via esummary to resolve to claimed papers; full PDFs not read): Wołosowicz 2025 (41304763), Fields 2019 (31461880), Isa 2025 (40833723), Ridker/CANTOS 2017 (28845751), Gwozdzinski 2025 hemolysis (41009489), Lichota 2020 (33121005)
• verified-scope addendum documents the title/abstract-verification level; page remains verified:true with the new subsection flagged hypothesis-scenario
• remaining gap: Sun 2023 + Yan 2025 full PDFs (closed-access — no-fulltext-access); full-PDF read of the 6 collateral-effects refs needs-verification

[2026-05-29] verify | SPARC (molecules/proteins/sparc.md)

• verified: molecules/proteins/sparc.md → verified: true (2026-05-29, claude; partial scope)

Sources checked:

• Tyshkovskiy 2026 aging claims: cross-checked against pre-verified study page studies/tyshkovskiy-2026-universal-transcriptomic-hallmarks; no corrections needed.
• Canonical-DB identity (UniProt P09486, NCBI Gene 6678, HGNC 11219, Ensembl ENSG00000113140): confirmed via UniProt REST API — all correct.
• GenAge absence confirmed: SPARC not in GenAge Build 21 (307 human genes); genage-id: null + gap note updated.
• Druggability tier 3 confirmed via Open Targets: med-quality pocket (SM only); no approved drug, no Phase 1 clinical — consistent with stated rationale.
• Mouse ortholog MGI:98373 resolved via UniProt P07214 cross-reference — gap #gap/needs-mouse-ortholog-mgi-confirm closed.

Corrections made:

1. mouse-ortholog frontmatter: Sparc (unconfirmed gap) → Sparc (MGI:98373) — resolved from UniProt P07214 cross-reference.
2. Identity table: mouse ortholog row updated to include confirmed MGI:98373.
3. GenAge row in identity table: updated to “Not in GenAge Build 21 (307 human genes; confirmed 2026-05-29)” — gap tag removed.
4. GenAge frontmatter comment: updated to “absent confirmed”.
5. Banner: reduced from auto-extraction warning to partial-verification note.

New content added (recency check, protein page — optional check run anyway):

• Ham et al. 2023 (Int J Mol Sci, doi:10.3390/ijms241512179): SPARC expression higher in young vs aged human skin; recombinant SPARC activates TGF-β pathway and upregulates collagen/downregulates MMP-1 in fibroblasts. Added to Expression section and footnotes.
• Toba & Takai 2024 (Am J Physiol Heart Circ Physiol, doi:10.1152/ajpheart.00565.2024): SPARC proinflammatory in CVD contexts; SPARC deletion attenuates cardiac aging injury. Added context-dependence nuance to Druggability section and Gaps section with #gap/contradictory-evidence.
• Human skin aging “Replicated in humans?” row updated to note Ham 2023 confirms reduced SPARC in aged human skin.
• literature-checked-through: 2026-05-29 set.

Unverifiable at this pass:

• Knockout phenotype papers (Bradshaw 2002/2003 JID/PNAS, Rentz 2007 JBC, Gehwolf 2016 Sci Rep, Bradshaw 2016 YJMCC, Seet 2018, Atorrasagasti 2019, others) — OA papers triggered for DOI lookup but remained pending at verification time; closed-access (YJMCC 2016) flagged not_oa.

Supersession candidates: None. Recent reviews (Cesur 2026, Matrix Biol; Toba 2024, AJPHEART) are consistent with existing framing; the cardiac context-dependence was added as a nuance note, not a supersession.

Downstream propagation needed: None urgent — SPARC page has no downstream entity pages citing it yet; inbound links are 0 beyond the seeded pages (Tyshkovskiy study page already verified).