The Retinoid Receipts

The prescription everyone is suddenly asking for

For most of the last decade the retinoid conversation was about retinol: percentages on the bottle, "gentle" versions, serums with a waiting list. Lately the talk has moved up the shelf. Social media has made prescription tretinoin, usually shortened to "tret," a small status symbol, helped along by telehealth services that will write the prescription after a photo and a questionnaire. What the videos tend to skip is the irritation, the sun sensitivity and the fine print on pregnancy.

That is a lot of attention for a molecule first sold for acne in the early 1970s. It is also a useful test case, because tretinoin is one of the few topical anti-aging treatments with a trial record long enough to audit. So this issue asks two questions. Does the prescription version do what its reputation says? And how much of that reputation transfers to the retinol sold without a prescription?

What sun-damaged skin has lost

Photoaged skin is not just older skin. Chronic ultraviolet exposure degrades the collagen in the dermis, the structural layer under the surface, and the skin responds by building less of it. In a 1993 study, collagen I formation in sun-damaged forearm skin was 56% lower than in sun-protected skin from the same people.¹

Tretinoin is all-trans retinoic acid, the active form of vitamin A. It binds receptors inside skin cells and changes which genes are switched on. Retinol, the cosmetic cousin, is a precursor: skin has to convert it into retinoic acid before it acts. That extra step is the usual explanation for why retinol is gentler, and it is a reason not to assume it works equally well. In a study that applied both to human skin and took biopsies, both raised epidermal thickness and the genes for collagen I and III within four weeks, but the magnitude with retinol "tends to be smaller."² One caveat on that paper: several authors were affiliated with a company that sells retinol products.

Four decades of vehicle-controlled data

The modern record starts in 1988 with a small, tidy experiment. Each participant put tretinoin on one forearm and the cream base alone (the vehicle) on the other, and in addition faces were assigned to tretinoin or vehicle, half and half.³

The next study asked what is happening under the surface. Griffiths and colleagues treated 29 people with photodamaged skin with daily 0.1% tretinoin (15) or vehicle (14) for 10 to 12 months and took biopsies.¹

The trial that most closely resembles real-world use came in 2005. Kang and colleagues enrolled 204 people with moderate to severe facial photodamage and gave them 0.05% tretinoin emollient cream or placebo for two years.⁴

Two more findings round out the picture. A 1997 maintenance study gave 126 people who had finished 48 weeks of daily 0.05% tretinoin either three-times-weekly dosing, once-weekly dosing or nothing for 24 weeks. Gains held up at three times weekly, partly held at once weekly, and tended to fade with no treatment.⁵ It supports reduced frequency only after 48 weeks of daily use. It did not test reduced frequency as a starting plan. And a 2022 systematic review of seven randomized trials found benefit in every one, appearing as early as one month and lasting up to 24 months.⁶ It did not pool the data, so it is a consistent record, not a meta-analysis.

Nine trials, four of them null

Retinol has far less going for it. The best-known positive result is a 24-week trial in 36 elderly residents of two senior-citizen facilities (average age 87), in which study staff applied 0.4% retinol lotion to one arm and vehicle to the other. Fine-wrinkle scores moved by -1.64 points with retinol and -0.08 with vehicle (P<.001).⁷

A 52-week program of vehicle-controlled studies of a stabilized 0.1% retinol reported 44% improvement in crow's feet and 84% in mottled pigmentation, but the facial study had 62 participants in total across retinol and vehicle, and the forearm study had 12.⁸ Neither this nor the Kafi trial settles much on its own.

The sharper view comes from a 2021 systematic review of nine double-blind, vehicle-controlled trials of over-the-counter retinol. Four found no significant difference from vehicle. The other five offered what the author called weak evidence for fine lines only, with major methodological flaws, and the review suggested some positive trials "may serve as tools for advertising."⁹ It is a single-author review with a PubMed-only search, so it is not the last word. It is, though, the most complete accounting available, and it does not support the idea that any retinol at any percentage is a safe bet.

Does retinol really match tretinoin?

This is the claim doing the most work in marketing, and the direct evidence is thin. In a 12-week double-blind trial of 45 women, a retinol serum stepped up from 0.25% to 1.0% was compared with tretinoin cream stepped up from 0.025% to 0.1%. The investigators rated the two as equivalent, and skin dryness improved significantly in the retinol group (P<0.001), an improvement not seen in the tretinoin group.¹⁰ A co-author is employed by TOPIX Pharmaceuticals, a skincare company, and the study was small and short, and the step-up schedule is not how tretinoin is normally started.

Against that, a 2024 systematic review of 25 studies comparing tretinoin with other topical treatments found the comparator better in 7, equivalent in 13 and worse in 3, with most comparators less irritating. The authors described most of these comparison trials as poor quality or biased.¹¹ The fair summary is that equivalence has been claimed more often than it has been demonstrated.

Adapalene is the one alternative with a head-to-head result worth citing. A 24-week Brazilian trial found 0.3% adapalene gel non-inferior to 0.05% tretinoin cream on global photoaging and on forehead and eye-area wrinkles.¹² The trial was investigator-blinded rather than double-blind.

Irritation is real. The pregnancy and cancer questions are narrower than they sound.

The daily cost of tretinoin is irritation: dryness, peeling and redness, usually worst in the first weeks and managed by starting a few nights a week. The prescribing label also says treated skin can be more sensitive to sunlight, which makes the popular claim that retinoid photosensitivity is a myth too strong. Sun protection is part of the regimen, not an optional add-on.

Pregnancy is the question readers ask most. A 2026 Nordic register study followed 3,869,586 infants, 2,172 of them exposed to topical retinoids in the first trimester. Major malformations occurred in 3.3% of exposed infants and 3.0% of unexposed (adjusted risk ratio 1.1, 95% CI 0.87–1.38).¹³ That agrees with an earlier meta-analysis covering 654 exposed pregnancies, which also found no clear signal.¹⁴ Both sets of authors still advise avoiding topical retinoids in pregnancy: the Nordic study counted only pregnancies that ended in a birth, which could underestimate risk, and the meta-analysis lacked the statistical power to rule out a smaller effect. The data are reassuring for someone who used a cream before knowing they were pregnant. They are not a green light to start.

One more result sits in the other direction from the hype. A Veterans Affairs trial that randomized 1,131 high-risk veterans found that 0.1% tretinoin, used for 1.5 to 5.5 years, did not reduce new basal or squamous cell skin cancers.¹⁵ Tretinoin repairs some features of photodamage. It does not replace sunscreen as protection against the thing that caused the damage.

A rare strong rating, with an asterisk on retinol