Retinyl palmitate: a notified active, and we found no trial of it on its own

A randomised, double-blind trial in JAMA Dermatology (2022), carried out at Johns Hopkins.

Twenty-four patients with moderate to severe facial photodamage were randomised into two arms. Single centre, parallel groups.

• one arm — retinoic acid (tretinoin) 0.02%
• the other — a 1.1% formulation containing retinol, retinyl acetate and retinyl palmitate
• applied daily for 24 weeks
• the final analysis covers 20 White women (9 on the blend, 11 on tretinoin)

The results.

• no significant difference in Griffiths photoaging scores at week 24 — median 4 versus 5, difference −1 (95% CI −2 to 1), P = .27
• erythema was six times less frequent on the blend — 11% versus 64%, difference −0.53 (95% CI −0.88 to −0.17), P = .01
• gene analysis showed CRABP2 induction, confirming retinoic acid receptor signalling, but no significant change in procollagen I or MMP1/3/9
• instead MMP2 fell (P = .02), and that change correlated with the change in fine wrinkles (r = 0.54)
• people with worse wrinkles at baseline improved more

How to read it.

First, what was pitted against tretinoin was not this ingredient alone. The formulation contained retinol, retinyl acetate and retinyl palmitate. So 'retinyl palmitate tied with tretinoin' is not a sentence this trial supports. The sentence it supports is 'this 1.1% blend of the three tied with 0.02% tretinoin'.

Second, we do not turn 'no significant difference' into 'the same'. The final analysis is 20 people and the interval runs −2 to 1. That width means the blend could be better, and tretinoin could be better. It was not designed as a non-inferiority trial.

Third, the erythema figure is clear. 11% versus 64%, P = .01, with an interval that does not contain zero. What this trial showed most firmly is not equal efficacy but a difference in irritation. It points the same way as the bakuchiol page.

Fourth, look at who was measured. Twenty White women. Photoageing and erythema both vary with skin colour, which makes these numbers hard to carry across.

Fifth, the funding. Three of the authors declared SkinMedica grants during the conduct of the study. This site records that as industry funding.

Sixth, we carry the mechanistic result as published. Procollagen I did not rise. The marker most often used to explain how retinoids work did not move here; MMP2 fell instead, and the authors offered MMP2 as a possible mediator. A possible one.

So for this ingredient it sits at 'not established' — not because the trial is weak, but because this design cannot tell what this ingredient contributed.

A randomised, controlled, grader-blinded study in J Cosmet Dermatol (2013).

An oil-based moisturiser containing retinyl palmitate and antioxidants was used for 12 weeks, while the skin condition of a no-treatment group was recorded over the same period. An expert graded all body sites live, and the face and neck were also graded from photographs.

The results.

• compared with the no-treatment group and with baseline, the face and neck improved on fine lines, coarse wrinkles, mottled pigmentation, tone, roughness, firmness and clarity (12 weeks, P < 0.001)
• the décolletage, arms and lower legs improved on crepey texture, dryness, scaling and roughness
• some parameters improved as early as 2 weeks
• the degree of improvement ran legs > arms > décolletage > face > neck

How to read it. This entry is a textbook case of the problem this site writes about most often.

First, the control was no treatment. It was compared against applying nothing. This site's grading conditions include 'no no-treatment control unless the endpoint is clinical'. The reason is visible in this very trial — dryness, scaling and roughness improve from putting oil on them.

Second, the vehicle is an oil. And the improvement was largest on the legs and smaller on the face. Legs are usually the driest site. This design cannot separate the moisturising share from the retinoid share.

Third, it is mixed again. The formulation contained antioxidants as well.

Fourth, the number of people is not in the abstract. It is left as not reported. We do not invent it.

So it sits at 'not established'. What this trial showed is 'the side that applied this oil did better than the side that applied nothing', and retinyl palmitate's share cannot be taken out of that sentence.

Korea's functional-cosmetic review rule, Annex 4, names four notified wrinkle actives — retinol (2,500 IU/g), retinyl palmitate (10,000 IU/g), adenosine (0.04%) and polyethoxylated retinamide (0.05–0.2%).

We do not convert that figure into a percentage. The rule sets it in IU/g, and the factor converting IU to mass differs between esters. Not computing a number we do not have is a rule of this site.

What the amount means, again. Not 'this concentration works' but 'a product with this amount may be filed for the claim without submitting its own efficacy data'.

And a 'wrinkle-improving functional cosmetic' label on the back of a product does not tell you how much retinyl palmitate is in it. The claim may have been granted on a different active from the same list. The ingredient list usually settles it.