---
title: "Speculative & Emerging Oral Compounds for Skin Aging — Mechanism-First Tier Report (Pass 1)"
kind: "Research pass"
topic: "Health & self-care"
date: "2026-09-01"
url: "https://canopticon.com/p/speculative-emerging-oral-compounds-for-skin-aging-mechanism-first-tie"
author: "Chris Telford"
license: "CC BY 4.0"
---

# Speculative & Emerging Oral Compounds for Skin Aging — Mechanism-First Tier Report (Pass 1)

## TL;DR
- **Almost everything in this universe is "mechanism-plus-preclinical," not proven by oral skin RCTs.** Only a small handful have *any* human skin endpoint, and most of those used *topical* delivery, not oral. The best-supported genuinely-oral-systemic candidates are taurine, urolithin A, *Centella asiatica*, sea buckthorn omega-7, and (for reference) collagen peptides; the most mechanistically elegant but skin-unproven are fisetin, pterostilbene, astragaloside IV/cycloastragenol, berberine, and spermidine.
- **Brutal honesty up front:** the supplement industry is wildly over-selling telomerase activators (astragaloside IV), reishi, and "senolytic" botanicals as skin-rejuvenation agents. The strongest human senolytic skin datapoint is from the Hickson 2019 D+Q pilot, where epidermal p16⁺ cells fell ~20% — but **no oral senolytic has ever shown a wrinkle or elasticity benefit in a human RCT.**
- **Recommended action:** treat this as a research menu, not a shopping list. Tier by (1) does it reach skin orally, (2) is there a human skin endpoint, (3) safety. Reasonable to add now: taurine, creatine, and possibly a urolithin A product; *Centella*/sea buckthorn for botanical breadth. Everything else is experimental.

## Key Findings
1. **Topical ≠ oral.** Many "skin RCTs" for these compounds (urolithin A, pterostilbene, saffron, rapamycin, Cordyceps) are *topical*. Oral systemic delivery to dermis is a separate, usually unproven, question — the single biggest source of hype-confusion in the category.
2. **Fibroblast/collagen in-vitro data are cheap and abundant; they predict almost nothing about oral human efficacy.** Nearly every botanical here "increases collagen in fibroblasts" — a very low bar.
3. **The senescence/senolytic story is real biology but clinically immature for skin.** Fisetin skin data are cell-culture and mouse-graft only; D+Q human trials are tiny pilots.
4. **A few compounds have genuine, recent, oral human skin RCTs:** *Centella asiatica* (oral, Korean women, 12 weeks), Astragalus+Centella saponins (oral, 12 weeks), and sea buckthorn omega-7. Taurine has landmark systemic aging biology (though not a skin RCT, and now contested).
5. **Bioavailability is the recurring killer.** Resveratrol, fisetin, curcumin, astragaloside IV, cordycepin, and spermidine all face poor oral absorption; pterostilbene is the notable exception.

## Details — Tiered by Plausibility × Evidence

### TIER 2 — Oral human skin RCT exists, or landmark systemic aging biology (strongest in this speculative universe)

**Taurine** — (a) *Mechanism:* reduces cellular senescence, protects telomerase, suppresses mitochondrial dysfunction, decreases DNA damage, attenuates inflammaging. (b) *Evidence:* the landmark is Singh et al., *Science* 2023 (eabn9257) — a systemic aging study, **not** a skin RCT. Per Singh et al. (restated in a 2024 review, ScienceDirect S1568163724002782): "A decline in circulating taurine is a feature of aging in multiple species, including humans, with levels falling by around 80% over the human lifespan." Supplementation extended healthspan/lifespan in mice and healthspan in monkeys. (c) *Preclinical:* strong multi-species. **Contested:** Marcangeli et al., *Aging Cell* 2025 (doi:10.1111/acel.70191) found "No association between circulating taurine levels and age, muscle mass, strength, physical performance, or mitochondrial function was observed…Taurine deficiency is therefore unlikely to be a primary driver of aging in humans." (d) *Traditional:* none skin-specific. (e) *Bioavailability:* excellent, water-soluble. (f) *Dose:* studied broadly 1–6 g/day. (g) *Cost:* very cheap. (h) **Tier 2** (systemic aging biology strong but now disputed; skin endpoint inferred, not measured). (i) *Synergy:* benign; pairs with most stacks. (j) *Red flags:* human aging/skin benefit unproven; the 2025 Aging Cell paper directly disputes the premise.

**Urolithin A (Mitopure)** — (a) *Mechanism:* gut microbiome postbiotic from ellagitannins (pomegranate); induces mitophagy, improves mitochondrial function, reduces inflamm-aging. (b) *Evidence:* Amazentis ran three trials — but these were **topical** (1% cream), not oral, showing wrinkle reduction and up-regulated collagen-organization genes (medRxiv 2023.06.16.23291378). Oral urolithin A has muscle/mitochondrial RCTs but **no published oral *skin* RCT.** (c) *Preclinical:* strong mitophagy data. (d) *Traditional:* pomegranate is an ancient beauty food. (e) *Bioavailability:* the key advantage — direct urolithin A bypasses the majority of people who can't make it endogenously. Per *Trends in Molecular Medicine* 2021 (S1471-4914(21)00118-0) and Singh et al., *Eur J Clin Nutr* 2021, only ~12% of people had detectable urolithin A at baseline and only ~40% converted meaningfully even after pomegranate juice — i.e. ~60% are non/poor-producers. (f) *Dose:* 500–1000 mg/day (oral Mitopure studies). (g) *Cost:* expensive (branded). (h) **Tier 2 topical / Tier 3 oral-for-skin.** (i) *Synergy:* complements NAD⁺ precursors. (j) *Red flags:* skin trials topical and company-sponsored; some not peer-reviewed.

**Centella asiatica (gotu kola) — oral** — (a) *Mechanism:* triterpenes (asiaticoside, madecassoside, asiatic/madecassic acids) stimulate collagen, reduce MMPs, antioxidant/anti-inflammatory. (b) *RCT:* a 12-week randomized, double-blind, placebo-controlled trial in middle-aged Korean women (*Nutrients* 2024, doi:10.3390/nu18101505) evaluated oral CA extract on wrinkles, hydration, TEWL, elasticity using objective instruments. A second RCT (*J Medicinal Food*) reported elasticity/hydration improvement after 6 months oral. (c) *Preclinical:* extensive; oral CA attenuated UVB photoaging in hairless mice (PMC12785454). (d) *Traditional:* major Ayurvedic/TCM skin-and-wound herb. (e) *Bioavailability:* moderate; triterpenes absorbed. (f) *Dose:* ~1100 mg twice daily used in diabetic dry-skin RCT. (g) *Cost:* cheap. (h) **Tier 2** (genuine oral skin RCTs, modest effect). (i) *Synergy:* good with collagen peptides. (j) *Red flags:* rare contact dermatitis; effect sizes modest, not retinoid-level.

**Astragalus + Centella saponins (ACS) — oral** — (a) *Mechanism:* combined saponin collagen stimulation. (b) *RCT:* Chan et al. (*Cosmetics* 2025, MDPI 12/4/131), 150 adults, randomized double-blind placebo-controlled; 12-week **oral** capsules raised skin brightness 2.5%, elasticity 4.7%, collagen content ~13% vs placebo. (c) *Preclinical:* saponin fibroblast data. (d) *Traditional:* both are TCM/Ayurvedic staples. (e) *Bioavailability:* saponins variable. (f) *Dose:* per product. (g) *Cost:* moderate. (h) **Tier 2.** (i) *Synergy:* overlaps astragaloside IV mechanism. (j) *Red flags:* single research group; small effect sizes.

**Sea buckthorn (Hippophae) omega-7 — oral** — (a) *Mechanism:* palmitoleic acid (omega-7) supports skin/mucosal barrier lipids; SIRT1 activation + MMP-1 inhibition claimed; carotenoid/vitamin E antioxidants. (b) *RCT:* a randomized controlled trial found oral sea buckthorn oil improved skin hydration, thickness, and elasticity; Yang et al. 2008 reported hydration +49%, elasticity +26%, wrinkle −9% over 3 months; a women's (>40 yr) trial showed improved hydration and reduced TEWL at 12 weeks. (c) *Preclinical:* consistent collagen/wound data. (d) *Traditional:* Tibetan/Himalayan skin food. (e) *Bioavailability:* good (lipid). (f) *Dose:* 1–2 g/day oil; ~220 mg omega-7. (g) *Cost:* cheap-moderate. (h) **Tier 2 for hydration/barrier; weaker for wrinkles.** (i) *Synergy:* fine with most. (j) *Red flags:* small trials; bleeding-risk interaction with anticoagulants; orange skin/urine tint possible.

### TIER 3 — Strong mechanism + preclinical skin/fibroblast data; human skin data topical or absent

**Pterostilbene** — (a) *Mechanism:* dimethylated resveratrol analog; SIRT1/AMPK, antioxidant, autophagy. (b) *RCT:* a 0.1% **topical** emulsion, 28-day double-blind split-face (n=31, *JDSCT*) improved elasticity, firmness, wrinkles, thicker epidermis with increased collagen/elastic fibers. No oral skin RCT. (c) *Preclinical:* MDPI *Cosmetics* 2025 in-vitro anti-photoaging. (d) *Traditional:* blueberry/Indian kino. (e) *Bioavailability:* its key edge — per Kapetanovic et al., *Cancer Chemother Pharmacol* 2011 (PMID 21116625): "Resveratrol and pterostilbene were approximately 20 and 80% bioavailable, respectively," with half-life 105 min vs 14 min and Cmax ~36× higher for pterostilbene. (f) *Dose:* longevity stacks use 50–250 mg/day. (g) *Cost:* moderate. (h) **Tier 3** (best-bioavailable stilbene; skin data topical). (i) *Synergy:* overlaps resveratrol — don't double up. (j) *Red flags:* may raise LDL at high dose in some studies; oral skin benefit unproven.

**Cordyceps (militaris/sinensis/cicadae)** — (a) *Mechanism:* cordycepin (adenosine analog), NAD⁺/sirtuin, autophagy; polysaccharides drive hyaluronan synthesis. (b) *RCT:* a *Cordyceps sinensis* extract was tested **topically** in a 40-subject double-blind placebo trial (PMC11049886) — 28% wrinkle reduction, +10% collagen in vivo. No oral skin RCT. (c) *Preclinical:* strong — cordycepin reduced fibroblast ROS, raised collagen/elastin, 2.76-fold skin-regeneration activity; C. cicadae extract raised fibroblast hyaluronan to 1293 ng/mL. (d) *Traditional:* Tibetan/TCM tonic. (e) *Bioavailability:* cordycepin is rapidly deaminated by adenosine deaminase — a real oral problem. (f) *Dose:* not skin-established. (g) *Cost:* wild cordyceps very expensive; cultivated *militaris* cheap. (h) **Tier 3 preclinical.** (i) *Synergy:* pairs with NAD⁺ theme. (j) *Red flags:* species/adulteration issues; bioavailability.

**Reishi (Ganoderma lucidum) triterpenes** — (a) *Mechanism:* triterpenes + polysaccharides; Nrf2/antioxidant gene induction, senomorphic. (b) *RCT:* none for skin. (c) *Preclinical:* Kühnel et al. 2025 (*CIMB*, doi:10.3390/cimb47020130) — Reishi ethanol extract on etoposide-senescent human dermal fibroblasts up-regulated HO-1, γGCS, NQO1, suppressed p16INK4a, and showed senolytic-like reduction of SA-β-gal⁺ cells. (d) *Traditional:* "mushroom of immortality," 2000+ yrs. (e) *Bioavailability:* triterpene absorption poor; polysaccharides act via gut/immune. (f) *Dose:* not skin-established. (g) *Cost:* moderate. (h) **Tier 3 in-vitro.** (i) *Synergy:* adaptogen overlap. (j) *Red flags:* heavy hype; anticoagulant interaction.

**Berberine** — (a) *Mechanism:* AMPK activator (metformin-like); suppresses mTOR/S6K, prevents stress-induced senescence, restores autophagy + NAD⁺ (PMC4854918, PMC3796215). (b) *RCT:* none for skin; many metabolic RCTs. (c) *Preclinical:* note that in cardiac/renal fibroblasts berberine *reduces* collagen synthesis (anti-fibrotic, PMID 26722438) — a **double-edged mechanism** for skin, where you generally *want* dermal collagen. (d) *Traditional:* TCM (Coptis/Berberis) for skin/gut. (e) *Bioavailability:* notoriously poor (<1%); dihydroberberine better. (f) *Dose:* 500 mg 1–3×/day metabolic. (g) *Cost:* cheap. (h) **Tier 3 mechanism (ambiguous direction for dermal collagen).** (i) *Synergy:* additive with metformin — caution hypoglycemia/GI. (j) *Red flags:* major CYP/drug interactions; GI upset; pregnancy contraindicated.

**Spermidine** — (a) *Mechanism:* autophagy inducer (Eisenberg 2009, *Nat Cell Biol*). (b) *RCT:* the strong human data are for **hair** — Rinaldi et al. 2017 (*Dermatol Pract Concept*), 100-subject RCT, 90 days, prolonged anagen with effects persisting 3 months post-treatment. No skin-wrinkle RCT. (c) *Preclinical:* broad longevity/autophagy. (d) *Traditional:* dietary (wheat germ, natto, aged cheese). (e) *Bioavailability:* oral spermidine may convert to spermine; debated. (f) *Dose:* 1–6 mg/day supplements. (g) *Cost:* cheap-moderate. (h) **Tier 3 (hair RCT; skin extrapolated).** (i) *Synergy:* fits NAD⁺/autophagy theme. (j) *Red flags:* claims outpace skin data.

**Astragaloside IV / Cycloastragenol (CAG)** — (a) *Mechanism:* telomerase activation (TA-65 is encapsulated CAG); keratinocyte proliferation/migration. (b) *RCT:* none for skin appearance. (c) *Preclinical:* CAG induced human keratinocyte proliferation at 1 ng/mL and improved wound healing/fur in aged mice (Spandidos etm.2018.6501); telomerase/telomere effects in fibroblasts/MSCs (Springer 2020). (d) *Traditional:* Astragalus, TCM tonic since ~200 CE. (e) *Bioavailability:* astragaloside IV has very low solubility/poor oral absorption — a serious problem. (f) *Dose:* supplements ~50 mg. (g) *Cost:* expensive (98% extracts). (h) **Tier 3 mechanism / heavily over-marketed.** (i) *Synergy:* overlaps ACS. (j) *Red flags:* telomerase activation raises theoretical cancer concern; bioavailability; rampant supplement hype.

**Saffron (crocin/crocetin)** — (a) *Mechanism:* crocin inhibits collagenase, elastase, hyaluronidase; antioxidant; reduces UVB fibroblast senescence (β-gal), preserves collagen-1. (b) *RCT:* topical 3% saffron cream reduced redness/dark spots (8-wk trial); an oral standardized saffron-extract capsule RCT for wrinkles/pigmentation is registered (NCT07324759) — results pending. (c) *Preclinical:* good fibroblast photoaging data. (d) *Traditional:* Persian/TCM beauty spice. (e) *Bioavailability:* crocetin absorbed; crocin hydrolyzed in gut. (f) *Dose:* oral 15.5–30 mg/day saffron extract. (g) *Cost:* expensive spice; capsules moderate. (h) **Tier 3.** (i) *Synergy:* mood/HPA benefit bonus. (j) *Red flags:* doses >1.5 g toxic; pregnancy contraindicated; adulteration common.

**Black garlic / S-allylcysteine (SAC)** — (a) *Mechanism:* potent anti-glycation — inhibits AGE and carboxymethyllysine (CML) formation; antioxidant — relevant to dermal collagen cross-linking/stiffening (CML doubles in diabetic skin). (b) *RCT:* cardiovascular/BP RCTs exist (e.g., 0.25 mg/day SAC, 12 wk); no skin RCT. (c) *Preclinical:* strong in-vitro antiglycation (Ahmed/Elosta, PMID 17321518; *J Nutr*). (d) *Traditional:* East Asian food. (e) *Bioavailability:* SAC is well-absorbed and stable (its main virtue). (f) *Dose:* aged garlic extract ~250–1000 mg. (g) *Cost:* cheap. (h) **Tier 3 (anti-AGE mechanism sound, skin-specific data absent).** (i) *Synergy:* benign. (j) *Red flags:* anticoagulant interaction.

### TIER 4 — Mechanism-only or ethnobotanical; minimal/no oral skin data

**Fisetin** — (a) *Mechanism:* flavonoid senolytic; in dermal fibroblasts selectively kills senescent cells via caspase-3/8/9 apoptosis. (b) *RCT:* **No published human oral skin RCT.** Mayo trials (AFFIRM-LITE NCT03675724; AFFIRM NCT03430037) use the "Mayo protocol" of 20 mg/kg/day × 2 consecutive days, intermittently; primary endpoints are inflammation/frailty, not skin — and **as of June 2026 no peer-reviewed results have been published.** The closest human skin-tissue senolytic datapoint is the related D+Q regimen (below). (c) *Preclinical:* *Biogerontology* 2024 (PMID 37736858) — fisetin (intraperitoneal, not oral) increased collagen density and reduced SASP/MMPs in aged human skin grafts on mice. (d) *Traditional:* none. (e) *Bioavailability:* **poor** — water-insoluble, rapidly cleared; liposomal/SNEDDS/fenugreek forms attempt to fix this. (f) *Dose:* practitioner "hit-and-run" ~20 mg/kg intermittent (~1,400 mg for 70 kg) or 100–500 mg/day daily; products run ~$15–40/bottle. (g) *Cost:* cheap-moderate. (h) **Tier 4** (elegant senolytic mechanism, human skin benefit unproven). (i) *Synergy:* avoid combining with CYP3A4 substrates, quinolone antibiotics, PPIs, azole antifungals, macrolides. (j) *Red flags:* hepatic risk at very high acute doses; vendor extrapolation of mouse skin data to humans is rampant.

**Dasatinib + Quercetin (D+Q)** — (a) *Mechanism:* canonical senolytic cocktail. (b) *RCT:* Hickson et al., *EBioMedicine* 2019 (47:446–456, PMID 31542391, N=9 diabetic kidney disease, oral D 100 mg + Q 1000 mg × 3 days, NCT02848131): epidermal p16INK4A⁺ cells were "20% less abundant 11 days after completing 3 days of treatment with D+Q than at baseline" (baseline 1.95 ± 0.63 cells/mm; p=0.026). In adipose tissue the same paper reported p16INK4A, p21CIP1 and SA-β-gal "reduced by 35%, 17% and 62%" respectively, plus reduced circulating SASP factors "IL-1α, IL-6, and MMPs-9 and −12." Justice et al. 2019 IPF pilot (N=14, D 100 mg + Q 1250 mg, 3 days/wk × 3 wk, NCT02874989) improved physical function but did not measure skin. **No skin-appearance endpoint exists anywhere for D+Q.** (c) *Preclinical:* foundational senolytic literature. (d) *Traditional:* none (dasatinib is a chemo drug). (e) *Bioavailability:* dasatinib is prescription-only; quercetin poorly absorbed. (f) *Dose:* as above, intermittent. (g) *Cost:* dasatinib expensive/Rx-only. (h) **Tier 4 for skin — and dasatinib should NOT be self-administered.** (i) *Synergy:* N/A. (j) *Red flags:* dasatinib is a serious oncology drug (pleural effusions, cytopenias, QT prolongation); do not DIY.

**Ashwagandha (withanolides)** — (a) *Mechanism:* HPA-axis/cortisol reduction → psychoneuroimmunology skin pathway; chronic cortisol drives barrier impairment and aging. (b) *RCT:* robust *stress/cortisol* RCTs (meta-analysis PMC12242034, 15 trials, n=873; morning cortisol significantly reduced) but **no skin-appearance RCT.** (c) *Preclinical:* fibroblast-resilience claims, thin. (d) *Traditional:* Ayurvedic rasayana. (e) *Bioavailability:* standardized extracts (KSM-66, Shoden) decent. (f) *Dose:* 240–600 mg/day standardized root extract. (g) *Cost:* cheap. (h) **Tier 4 (indirect stress→skin pathway).** (i) *Synergy:* good adaptogen base; may raise testosterone, lower cortisol. (j) *Red flags:* leaf extracts (withaferin A, withanone) are cytotoxic/hepatotoxic; rare liver-injury reports; thyroid/immune interactions — prefer root-only.

**Creatine monohydrate** — (a) *Mechanism:* ATP buffering; pre-treatment protected human dermal fibroblasts from H₂O₂-induced senescence and preserved IGF-1 (PMID 34836359). (b) *RCT:* no oral skin RCT. (c) *Preclinical:* the fibroblast study is encouraging but pre-treatment-only (post-treatment didn't work). (d) *Traditional:* none. (e) *Bioavailability:* excellent. (f) *Dose:* 3–5 g/day. (g) *Cost:* very cheap. (h) **Tier 4 (intriguing, safe, cheap; skin unproven).** (i) *Synergy:* benign, near-universal. (j) *Red flags:* essentially none at standard dose.

**Boron** — (a) *Mechanism:* collagen cross-linking cofactor; up-regulates type-I collagen and osteopontin in osteoblasts; anti-inflammatory. (b) *RCT:* bone-health RCTs; no skin RCT. (c) *Preclinical:* in-vitro collagen/bone data; a skin-rejuvenation patent exists (boron complexes increasing proline incorporation). (d) *Traditional:* none specific. (e) *Bioavailability:* good. (f) *Dose:* 3 mg/day common. (g) *Cost:* very cheap. (h) **Tier 4.** (i) *Synergy:* pairs with collagen/silica for connective tissue. (j) *Red flags:* upper limit ~20 mg/day; high doses reproductive toxicity.

**Rapamycin (topical — included for contrast)** — (a) *Mechanism:* mTOR inhibition reduces senescence. (b) *RCT:* the Drexel **topical** trial (*GeroScience* 2019, NCT03103893, n=36→17) significantly reduced p16INK4A (p=0.008) and raised collagen VII in dorsal-hand skin — **one of the few positive human senescence-targeting skin trials, but topical, not oral.** Oral rapamycin for skin aging has no RCT and carries immunosuppression risk. (c) *Preclinical:* strong. (d) *Bioavailability:* oral systemic = immunosuppressive. (e) *Dose:* topical 10 µM cream. (f) *Cost:* moderate (compounded). (g) **Tier 2 topical / NOT recommended oral.** (h) *Red flags:* oral = infections, metabolic effects; do not use orally for cosmesis.

**Other Tier-4 ethnobotanicals (mechanism-only / no oral human skin RCT):** Reishi/Cordyceps polysaccharides (gut-immune), Lion's Mane (rat wound-healing only — topical HEFB raised collagen/angiogenesis, no human skin data; hericenones/erinacines act mainly via NGF/BDNF), Schisandra, Rhodiola/salidroside, Magnolia (honokiol/magnolol), Scutellaria (baicalin/wogonin Nrf2), Salvia miltiorrhiza (salvianolic acids, microcirculation), Rehmannia, Angelica sinensis, kudzu/puerarin, mulberry DNJ (anti-AGE), Polygonum/emodin, licorice (11β-HSD1/cortisol), shilajit/fulvic acid (mineral carrier, no skin RCT), Lycium/goji polysaccharides, maqui/aronia/lingonberry anthocyanins, Kakadu plum/camu camu (vitamin C source), Opuntia betalains (anti-AGE), black cumin/thymoquinone, ornithine alpha-ketoglutarate (wound-healing data), betaine/TMG (methyl donor), oral glucosamine (HA substrate). All share the same profile: a plausible antioxidant/anti-inflammatory/anti-glycation or HPA mechanism, in-vitro or animal support at best, and **zero oral human skin RCTs.**

## Recommendations (staged)

**Stage 1 — Add now (favorable safety, real systemic biology, cheap):**
- **Taurine 1–3 g/day** — best-evidenced systemic anti-senescence biology, trivially safe and cheap (acknowledging the 2025 *Aging Cell* dispute).
- **Creatine 3–5 g/day** — safe, cheap, fibroblast-protective in vitro; low-risk bet with broad ancillary benefits.
- Consider **oral urolithin A (Mitopure) 500 mg/day** if budget allows — best mitophagy mechanism; note skin data are topical, and ~60% of people are poor endogenous producers (so the direct compound is the rational choice over pomegranate).

**Stage 2 — Add for botanical breadth (modest oral skin RCT support):**
- **Centella asiatica oral extract** (~500–1000 mg standardized) — has the genuine oral skin RCTs.
- **Sea buckthorn oil 1–2 g/day** — for barrier/hydration, with real (if small) oral RCTs.
- **Pterostilbene 50–100 mg/day** in preference to resveratrol (bioavailability ~80% vs ~20%) — but recognize skin data are topical.

**Stage 3 — Experimental, only with self-aware risk acceptance / monitoring:**
- **Spermidine 1–6 mg/day** (autophagy; hair RCT, skin extrapolated).
- **Fisetin** intermittent — only as a deliberate experiment; no human skin proof, watch CYP3A4/antibiotic/PPI interactions and use a bioavailable (liposomal/fenugreek) form.
- **Black garlic/SAC** for anti-glycation if AGE load is a concern (higher-sugar diet, dysglycemia).

**Do NOT do:** oral rapamycin or self-administered dasatinib for cosmetic skin aging (immunosuppression / oncology-drug toxicity); high-dose astragaloside IV chasing telomerase (theoretical cancer risk + poor absorption + heavy hype).

**Benchmarks that would change these tiers:** a published *oral* RCT with objective skin endpoints (cutometer elasticity, biopsy collagen density, p16 immunohistochemistry) would promote fisetin, pterostilbene, or urolithin A to Tier 2-oral. Publication of Mayo's AFFIRM-LITE fisetin results is the single most important pending readout in this space.

## Caveats
- **Topical-vs-oral conflation** is pervasive; I have flagged each instance. Treat any "skin RCT" claim as topical unless explicitly stated oral.
- **In-vitro collagen stimulation is nearly universal and nearly meaningless** for predicting oral human benefit.
- **Industry/sourcing bias:** several positive results (urolithin A topical, ACS, Cordyceps topical) come from manufacturer-sponsored or single-group studies, some not peer-reviewed.
- **Telomerase activators and senolytics carry theoretical oncogenic or serious-drug risks** that justify conservatism.
- The **taurine** aging premise is now actively contested (Marcangeli et al., *Aging Cell* 2025), so do not treat it as settled.
- This is **Pass 1** (~40 compounds). Deeper synergy-stack modeling (Pass 2) and a regulatory/safety/contamination deep-dive on top candidates (Pass 3) are available on request.