🧬 Aging Wiki

retinoids

Properties1
aliasestopical retinoids, retinoid class, vitamin A derivatives, retinoic acid class, RAR agonists topical

14 min read

Topical Retinoids

Topical retinoids are the best-evidenced pharmacological intervention against photoaging, with two on-label FDA-approved aging-indication compounds — tretinoin/Renova (1995) and tazarotene/Avage (2000) — and over 25 years of accumulated Phase 3 trial data. The shared class mechanism is RAR/RXR nuclear receptor agonism leading to AP-1 transcription factor transrepression, which suppresses MMP-1/3/9-mediated collagen degradation and simultaneously stimulates collagen-I synthesis in dermal fibroblasts and keratinocytes. This page covers the class as a whole. Per-compound mechanism, dose-response, and pharmacology data live on the six R41 sister compound pages listed below.

This page covers the class as a whole. Agent-specific data lives in molecules/compounds/:

  • tretinoin — all-trans retinoic acid; FDA Rx; RAR-α/β/γ pan-agonist; canonical mechanism anchor
  • retinol — OTC prodrug; ~5–20x less potent than tretinoin after two-step conversion
  • retinaldehyde — OTC; one conversion step to RA; ~3–5x more potent than retinol
  • adapalene — synthetic third-gen; RAR-β/γ-selective; FDA OTC (2016); well-tolerated
  • tazarotene — synthetic third-gen acetylenic; RAR-β/γ-selective; FDA-approved photoaging (Avage 2000)
  • bakuchiol — non-retinoid meroterpenoid; functional analog; mechanism incompletely characterised

Family Overview

CompoundStructure classFDA aging-indicationReceptor selectivityRelative potencyTolerabilityOTC / RxKey trial anchor
tretinoinNatural RARenova (1995) photoaging RxRAR-α/β/γ pan-agonistHighest (reference)Moderate–low (irritation, peeling)RxFisher 1996 1; Sumita 2018 2; Lin 2025 NMA 3
tazaroteneSynthetic 3rd-gen acetylenicAvage (2000) photoaging RxRAR-β/γ-selectiveHigh (≥ tretinoin 0.05%)Low (most irritating)RxAvage clinical programme
adapaleneSynthetic 3rd-gen naphthoic acidNone (acne only); off-label photoagingRAR-β/γ-selectiveModerate (~tretinoin 0.05%)HighOTC (0.1%); Rx (0.3%)Bagatin 2018 4
retinaldehydeNatural retinoidNoneRAR-α/β/γ (after conversion)~3–5x < tretinoinModerate–highOTCneeds-replication vs tretinoin RCT
retinolNatural retinoidNoneRAR-α/β/γ (after 2-step conversion)~5–20x < tretinoinHighOTCDhaliwal 2019 (vs bakuchiol) 5
bakuchiolNon-retinoid meroterpenoidNoneIncompletely characterised; not a classical RAR agonistEquivalent to retinol 0.5% in one RCT 5Highest (least irritating)OTC cosmeticDhaliwal 2019 5; Puyana 2022 6

Shared Mechanism

(a) RAR/RXR nuclear receptor agonism

Retinoids (tretinoin, tazarotene, adapalene, and retinol/retinaldehyde after enzymatic conversion) bind ligand-binding domains of retinoic acid receptors (RAR-α, RAR-β, RAR-γ). Activated RAR forms heterodimers with retinoid X receptors (RXR) and binds retinoic acid response elements (RAREs) in promoters of target genes, recruiting coactivators (e.g., CBP/p300, SRC family) to initiate transcription. This RARE-mediated transactivation drives collagen-I synthesis and other photoaging-reversal programs in dermal fibroblasts and keratinocytes 1.

(b) AP-1 transrepression — the canonical anti-MMP mechanism

The most important anti-photoaging mechanism is not simple RARE-mediated transactivation but rather AP-1 transrepression: ligand-bound RAR physically interacts with Jun and Fos subunits of the AP-1 transcription factor complex, blocking AP-1 from binding its cognate promoter elements on MMP genes. This protein-protein interference mechanism was definitively characterised by Fisher et al. 1996 1:

  • Tretinoin reduces AP-1 DNA-binding activity in sun-exposed human skin by ~70% (text-stated on p.338 of the original paper) 1
  • MMP-1 and MMP-9 protein levels are suppressed by 50–80% after tretinoin treatment vs. vehicle 1
  • The mechanism is AP-1 protein-protein interaction with RAR — it is NOT mediated by TIMP-1 induction and NOT by CBP/p300 competition at shared promoter elements (per the R39-verified framing on the study page)

This AP-1 transrepression mechanism explains why synthetic retinoids with RAR-β/γ selectivity (adapalene, tazarotene) retain most anti-photoaging efficacy despite lacking pan-agonist activity: RAR subtype specificity for the MMP-suppression arm (RAR-γ in keratinocytes vs RAR-β in fibroblasts) is not characterised in Fisher 1996 and is drawn from later receptor-pharmacology literature needs-replication 1.

(c) Collagen synthesis stimulation

Separately from MMP suppression, RAR agonism stimulates collagen-I synthesis in dermal fibroblasts via RARE-driven gene programs. This is well-documented in the tretinoin photoaging literature but the precise receptor-subtype mediating this effect (RAR-β in fibroblasts is the dominant hypothesis) is not as sharply characterised as the AP-1/MMP arm needs-replication (for RAR subtype specificity of the collagen-synthesis arm).

For the full mechanism extraction, see fisher-1996-photoaging-ap1-mmp.

Generation History

Retinoids are classified by generation based on structural class and selectivity profile:

  • 1st generation (natural): tretinoin / all-trans retinoic acid (FDA acne 1971; Renova photoaging 1995), isotretinoin (oral acne 1982; not topical photoaging). Pan-agonist profile (RAR-α/β/γ).
  • 2nd generation (mono-aromatic): etretinate (psoriasis 1986; withdrawn 1998 due to teratogenicity + long half-life), acitretin (psoriasis 1996). Primarily dermatology/psoriasis applications; not photoaging-indicated.
  • 3rd generation (poly-aromatic / synthetic): adapalene (acne 1996 Rx; OTC 2016), tazarotene (acne 1997; Avage photoaging 2000), bexarotene (oral cutaneous T-cell lymphoma 1999 — RXR-selective, not a typical RAR-agonist). RAR-β/γ-selective profile for adapalene and tazarotene.

The shift from 1st to 3rd generation was driven by two goals: (1) receptor-subtype selectivity to reduce side effects while preserving efficacy, and (2) metabolic stability (resistance to isomerisation and oxidation relative to natural all-trans RA).

Receptor-Selectivity and Tolerability

Receptor selectivityCompoundsImplication
RAR-α/β/γ pan-agonistTretinoin, retinol/retinaldehyde (after conversion)Maximal signalling breadth; highest irritation potential
RAR-β/γ-selectiveAdapalene, tazaroteneSpares RAR-α-mediated pro-inflammatory pathways; better tolerability
RXR-selectiveBexarotene (oral CTCL only)Not relevant to topical photoaging class
Incompletely characterisedBakuchiolPartial RAR agonism reported in some studies, disputed in others; may also activate β-catenin and TGF-β pathways no-mechanism

RAR-α activation in skin mediates some of the irritation (retinoid dermatitis: erythema, peeling, dryness) associated with first-generation retinoids. The β/γ-selective profile of adapalene and tazarotene is associated with better tolerability in acne trials and is presumed (not conclusively proven for photoaging) to carry the same tolerability advantage.

Tolerability Spectrum

From most-tolerated to least-tolerated (approximate; individual variation is substantial):

bakuchiol → retinol → retinaldehyde → adapalene → tretinoin → tazarotene

This ranking is roughly inverse to potency, with bakuchiol as an outlier (less potent AND less irritating than any retinoid). The Chien 2022 JAMA Dermatology trial 7 found tretinoin precursors caused erythema 6-fold less frequently than tretinoin itself (11% vs 64%), providing a direct within-person quantification of the tolerability cost of conversion-step efficiency.

Practical consequence: patients unable to tolerate tretinoin are typically stepped down to adapalene or retinaldehyde; patients with sensitive skin or seeking cosmetic (non-Rx) options use retinol or bakuchiol. The Siddiqui 2024 systematic review recommends retinaldehyde as the preferred second-line option for poor tretinoin tolerators 8.

Combination and Sequencing Strategies

Mandatory co-intervention: sunscreen

All topical retinoids increase photosensitivity (tretinoin’s photodegradation also necessitates evening application). Sunscreen with SPF ≥30 is a mandatory co-intervention in any retinoid photoaging protocol — not optional. See uv-protection (R42 forward-ref; page pending).

Tolerability buffers

  • niacinamide (R42 forward-ref; page pending) — barrier-repair niacinamide reduces retinoid irritation in combination use; mechanisms include ceramide synthesis upregulation and potential mild AP-1 suppression in its own right.
  • Ceramide-containing moisturisers — supportive barrier repair without mechanistic crosstalk.
  • Low-frequency initiation (every-other-night → nightly) is the standard tolerability-adaptation strategy for tretinoin.

Complementary mechanisms

  • ascorbic-acid / vitamin C (R42 forward-ref; page pending) — complementary collagen-synthesis stimulus via hydroxylation-cofactor and AP-1-independent TGF-β pathway; combination with tretinoin is theoretically additive.

Escalation ladder

Adapalene (OTC 0.1% → 0.3%) → tretinoin (0.025% → 0.05% → 0.1%) → tazarotene (0.05% → 0.1%) represents the typical clinical escalation from lowest to highest potency/irritation. Switching from tretinoin back to retinol is a common “tolerability reset” strategy during initial adaptation or in older/sensitive-skin patients.

Pregnancy and Teratogenicity

All retinoids carry teratogenicity risk due to disruption of RAR-mediated embryonic patterning programs (limb, craniofacial, neural tube). Topical risk is lower than oral but non-zero (particularly for tazarotene, which has higher systemic absorption than tretinoin):

  • Tazarotene (Avage/Tazorac): FDA Pregnancy Category X (contraindicated) — systemic absorption measurable; craniofacial defects in animal models at high doses.
  • Tretinoin (Renova/Retin-A): FDA Pregnancy Category C/D depending on indication — topical systemic absorption low but detectable; clinical guidance: avoid, especially first trimester.
  • Adapalene: FDA Pregnancy Category C — lowest absorption of the Rx retinoids; Rx guidelines still recommend avoidance.
  • Retinol / retinaldehyde (OTC): Category C equivalent — systemic conversion to RA is low but possible; many clinicians advise discontinuation during pregnancy and nursing despite OTC status.
  • Bakuchiol: No formal FDA pregnancy category; marketed as the safer retinoid alternative during pregnancy, but formal teratogenicity data are limited. Psoralen contamination in bakuchiol extracts has been flagged as a photosensitization safety concern in a proportion of commercial preparations — see bakuchiol for detail long-term-unknown.

Aging-Context Evidence: Recent Syntheses

Lin 2025 — Bayesian NMA (n=3,905) 3

The largest quantitative synthesis of topical photoaging interventions as of this writing: 23 RCTs, 3,905 participants, Bayesian network meta-analysis. Key findings:

  • Fine wrinkles: isotretinoin and tretinoin ranked highest; tazarotene ranked highest for coarse wrinkles.
  • Hyperpigmentation: tretinoin and retinol superior to comparators.
  • Safety: tretinoin had the most favorable overall safety profile in this analysis; tazarotene and glycolic acid carried higher adverse-event risk.
  • Class confirmation: the NMA confirms tretinoin as the current gold standard for topical photoaging across multiple outcome domains.

This network meta-analysis supersedes earlier head-to-head comparisons in statistical power and confirms the class hierarchy established by two decades of individual RCTs. needs-replication — the NMA includes 23 RCTs but most are 12–24 weeks; long-term (>52 weeks) comparative data remain sparse 3.

Siddiqui 2024 — Systematic Review (25 studies) 8

Systematic review of 25 studies comparing tretinoin to alternative topical agents. Findings:

  • 7 studies: comparator showed greater efficacy than tretinoin.
  • 13 studies: comparator showed equivalent efficacy.
  • 3 studies: comparator showed less efficacy.
  • Most comparators were less irritating than tretinoin.
  • Recommendation: retinaldehyde preferred as second-line for poor tretinoin tolerators; bakuchiol as OTC alternative.
  • Tretinoin retains the broadest evidence base and remains the clinical benchmark 8.

Bagatin 2018 — Adapalene 0.3% vs Tretinoin 0.05% (n=58, 24 weeks) 4

24-week investigator-blinded parallel-group RCT in Brazil comparing adapalene 0.3% gel once daily vs tretinoin 0.05% cream once daily for photoaging (1:1 allocation, n=29/arm). Key findings: both treatments did not differ significantly across global photoaging, periorbital wrinkles, ephelides/melanosis, forehead wrinkles, or actinic keratosis endpoints. Adapalene demonstrated non-inferior efficacy with a similar safety profile. This is the most directly comparative published RCT for these two agents in a photoaging indication 4.

Dhaliwal 2019 — Bakuchiol 0.5% vs Retinol 0.5% (n=44, 12 weeks) 5

Prospective, randomized, double-blind 12-week trial: bakuchiol 0.5% cream twice daily vs retinol 0.5% cream once daily for facial photoaging (n=22/arm). Both significantly decreased wrinkle surface area and hyperpigmentation with no statistically significant difference between compounds. Retinol users reported significantly more scaling and stinging. Interpretation: bakuchiol is non-inferior to retinol 0.5% for standard photoaging endpoints at 12 weeks, with better tolerability. Because retinol 0.5% is considerably weaker than tretinoin 0.05%, this finding does not imply bakuchiol is equivalently efficacious to tretinoin 5.

Chien 2022 — Tretinoin Precursors vs Tretinoin (JAMA Derm) 7

Randomized trial (n=20) comparing tretinoin precursors (retinaldehyde and retinol formulation) vs tretinoin for moderate-to-severe facial photodamage. No significant efficacy difference, but erythema occurred 6-fold less frequently with precursors (11% vs 64%). MMP-2 reduction correlated with fine wrinkle improvement, providing mechanistic support for the MMP-suppression axis in clinical photoaging regression needs-replication (small n) 7.

Evidence Gaps and Limitations

  • No epigenetic-age clock endpoints. No published RCT has used DunedinPACE, GrimAge, or any other biological-age clock as a photoaging treatment endpoint. This is the single highest-value “next experiment” for the class — the field’s 25+ years of clinical data establish wrinkle/pigmentation reversal, but whether this reflects true age-related collagen/proteostasis restoration or merely cosmetic endpoint improvement remains unanswered. needs-human-replication
  • Head-to-head data in elderly populations sparse. Most comparative RCTs recruited middle-aged adults (mean ~50); trials in adults ≥70 with frailty or comorbidity are lacking. dose-response-unclear
  • Bakuchiol long-term data limited. Only the 12-week Dhaliwal 2019 trial exists for direct retinol comparison; no long-term efficacy data or mechanistic pathway characterization comparable to Fisher 1996 5. no-mechanism (for bakuchiol’s precise receptor binding)
  • Psoralen contamination in bakuchiol. Some commercial bakuchiol preparations contain psoralen-class phototoxic contaminants from Psoralea corylifolia seed extraction; this is a safety qualifier not always disclosed in clinical trials. See bakuchiol for detail.
  • Biomarker-guided dosing not established. Optimal tretinoin dose for a given patient’s baseline MMP-1 activity, collagen density, or receptor expression profile is not characterised. dose-response-unclear
  • Systemic aging benefit unproven. Retinoids act on skin; no evidence that topical retinoid use affects systemic hallmarks of aging (e.g., circulating inflammatory markers, epigenetic clocks in blood). The intervention is a skin-aging intervention, not a systemic anti-aging intervention. The loss-of-proteostasis and genomic-instability hallmark links are limited to dermal/epidermal tissue.

SENS / Hallmark Mapping

  • loss-of-proteostasis (primary target) — MMP-driven extracellular matrix collagen fragmentation is the prototypical skin-aging proteo-structural lesion; retinoids suppress MMP-1/3/9 and stimulate collagen-I synthesis, partially restoring ECM integrity.
  • chronic-inflammation (AP-1/NF-κB transrepression) — UV-induced AP-1/NF-κB activation drives a pro-inflammatory MMP-expression cascade; retinoid-mediated AP-1 transrepression suppresses this inflammatory signaling axis.
  • genomic-instability (indirect) — RAR agonism modulates cell-cycle checkpoint and differentiation programs in keratinocytes; relevance to photocarcinogenesis prevention is documented for tretinoin and tazarotene in actinic keratosis, but this is weaker evidence than the ECM/proteostasis arm.

SENS category: loosely fits extracellular crosslinks (ECM damage) and wasting / loss of cells (fibrocyte depletion) but SENS framing is not the primary lens for this class.

Class Registration Note (R41 Propagation Pass)

The mechanism class topical retinoid / RAR agonist has been registered in intervention-classes with canonical mechanism values RAR-agonist, AP-1-transrepression, MMP-1-suppression, collagen-synthesis-stimulation, and functional-retinoid-analog (for bakuchiol). The R41 propagation pass should apply these values to all six sister compound pages.

Cross-References

Footnotes

Footnotes

  1. fisher-1996-photoaging-ap1-mmp · doi:10.1038/379335a0 · n=6–17 (varies per experiment; e.g., n=9 for AP-1 binding; n=7–10 for MMP mRNA/protein/activity) · in-vivo + mechanistic · AP-1 DNA-binding ~70% reduction with tretinoin (text-stated p.338); MMP-1/9 suppression 50–80% (text-stated p.338); AP-1 transrepression mechanism (not TIMP-1 or CBP/p300 competition) · model: adult human buttock skin in vivo · Nature 379:335-339 · 1996 · Fisher GJ et al. · PMID: 8552187 · local PDF confirmed in a local paper archive (download_status: completed); numerics per R39 verified study page ↩ ↩2 ↩3 ↩4 ↩5 ↩6

  2. sumita-2018-tretinoin-photoaging · doi:10.1111/jdv.15020 · n=24 women (48 forearms, bilateral within-person design; 1 dropout, per-protocol n=23) · rct · evaluator-blinded · ~20% photoaging reduction and ~60% AK reduction (both arms comparable); three endpoints diverged at P<0.05 (corneal thickness, dermis echogenicity, Ki67 — cream superior for ultrasonographic markers, peel superior for field cancerization) · model: human forearm skin (women >60 yr, Fitzpatrick II-III) · J Eur Acad Dermatol Venereol 32(10):1819-1826 · 2018 · Sumita JM et al. · PMID: 29704456 · closed access; body claims are abstract-only no-fulltext-access ↩

  3. doi:10.1038/s41598-025-12597-0 · meta-analysis · 23 RCTs, n=3,905 participants · Bayesian network meta-analysis of topical photoaging interventions · tretinoin/isotretinoin ranked highest for fine wrinkles; tazarotene for coarse wrinkles; tretinoin best overall safety profile · Scientific Reports 2025 · Lin L et al. (PMID 40707570) · archive: in index, OA gold, download pending ↩ ↩2 ↩3

  4. doi:10.1684/ejd.2018.3320 · rct · n=58 (29/arm) · 24-week investigator-blinded parallel-group · adapalene 0.3% gel vs tretinoin 0.05% cream once daily · no significant difference across global photoaging, periorbital wrinkles, melanosis, actinic keratosis; non-inferior efficacy for adapalene · European Journal of Dermatology 2018 · Bagatin E et al. (PMID 30105991) · archive: in index, bronze OA, download pending ↩ ↩2 ↩3

  5. doi:10.1111/bjd.16918 · rct · n=44 (22/arm) · 12-week double-blind · bakuchiol 0.5% bid vs retinol 0.5% qd · equivalent wrinkle + hyperpigmentation outcomes; retinol more scaling/stinging · British Journal of Dermatology 2019 · Dhaliwal S et al. (PMID 29947134) · archive: in index, bronze OA, download pending; citation percentile 100th (118 citations) ↩ ↩2 ↩3 ↩4 ↩5 ↩6

  6. doi:10.1111/jocd.15748 · systematic-review · bakuchiol dermatology applications; variable evidence quality across included studies; authors note psoralen contamination safety concern · Journal of Cosmetic Dermatology 2022 · Puyana C et al. (PMID 36176207) · archive: not confirmed in index ↩

  7. doi:10.1001/jamadermatol.2022.1891 · rct · n=20 · bilateral within-person randomized trial · tretinoin precursors vs tretinoin; equivalent efficacy; erythema 6× less frequent with precursors (11% vs 64%); MMP-2 reduction correlated with wrinkle improvement · JAMA Dermatology 2022 · Chien AL et al. (PMID 35675051) · archive: in index, green OA via PMC, download pending; citation percentile 100th ↩ ↩2 ↩3

  8. doi:10.1007/s40257-024-00893-w · systematic-review · 25 studies · tretinoin vs alternative topical agents in photoaging · 13 equivalent, 7 comparator-superior, 3 tretinoin-superior for efficacy; most comparators less irritating; retinaldehyde recommended as 2nd-line · American Journal of Clinical Dermatology 2024 · Siddiqui Z et al. (PMID 39348007) · archive: in index, closed OA, not downloaded; citation percentile 99th ↩ ↩2 ↩3

Graph View

Backlinks