A New Peptide Beat Retinol in an RCT. Should You Care?

A peptide beat retinol. The catch is the retinol barely showed up.

A 2025 randomized controlled trial by Chang et al. tested a new ingredient called cyclized hexapeptide-9, or CHP-9, against retinol and a vehicle serum in adults ages 30 to 55 with aging skin. After 56 days, CHP-9 came out ahead on wrinkles, roughness, elasticity, and barrier measures. If you stop reading there, it sounds like skincare finally found a cleaner, gentler retinol replacement.

That is not what this study proves.

The headline result is real: in this trial, the peptide beat the comparator it was given. The problem is that the comparator was 0.002% retinol, used twice daily for 56 days. That is far below the concentrations that have shown anti-aging benefit in peer-reviewed human trials. According to a 2022 systematic review of over-the-counter vitamin A products, clinical studies reporting efficacy used retinol products in the 0.025% to 1.0% range. A 2022 review of topical retinoids describes 12-week randomized trials with 0.25%, 0.5%, and 1.0% retinol that improved fine lines, coarse wrinkles, mottled pigmentation, skin tone, and tactile roughness. The 0.002% retinol in the CHP-9 trial sits 125 to 500 times below that 0.25% to 1.0% range.

So yes, CHP-9 beat retinol here. But it beat a dose of retinol that has not been established as an effective anti-aging comparator. That matters more than any p-value in the paper.

What the trial actually found

The study randomized 51 adults, 17 per arm, to one of three twice-daily serums for 56 days: 0.002% CHP-9, 0.002% retinol, or vehicle. It was double-blinded and included both an active comparator and a vehicle control, which is better than the usual single-arm before-and-after cosmetic study.

The peptide group improved on several instrumental measures. The published abstract reports statistically significant reductions in crow's feet number, area, and roughness, with mean changes of -2.20 (95% CI, -4.38 to -0.03), -3.95 (95% CI, -5.80 to -2.11), and -1.95 (95% CI, -3.30 to -0.59), respectively. It also reports significant improvement in forehead wrinkle measures and skin barrier-related outcomes. The retinol arm, by contrast, showed little movement over the same 56-day period, and CHP-9 was also better tolerated.

Those are favorable results for the peptide. They are also early results from one small trial. Seventeen people per arm is enough to generate a signal. It is not enough to settle a category war.

Why the peptide is still scientifically interesting

The skepticism here is about the comparison, not the entire idea.

A 2025 review by Pintea et al. lays out a plausible case for peptides in skin aging, especially when chemists improve stability and penetration. Cyclization can lock a peptide into a ring structure, which may help it resist degradation and cross biological barriers better than a floppy linear version. That is standard medicinal chemistry logic, not branding copy wearing a lab coat.

The broader peptide category also has some human evidence behind it. A 2025 clinical paper on acetyl dipeptide-31 amide reported anti-inflammatory and anti-aging benefits with good tolerability. A 2026 study of a hexapeptide-containing serum also reported improvements in wrinkles and skin texture. Those studies do not validate CHP-9 specifically, and they do not prove mechanism in living human skin. They do support a narrower claim: engineered peptides are credible candidates, not fairy dust in a frosted bottle.

That makes the Chang trial worth noticing. It does not make it practice-changing.

Why the retinol comparison is the whole story

Retinol usually does not work fast, and it usually does not work at trace concentrations. That is not a minor technicality. It is the central flaw in the trial design.

A foundational review of retinoids in skin aging summarizes evidence that therapeutically dosed retinoids increase epidermal turnover, thicken the epidermis, reduce matrix metalloproteinase activity, and support collagen synthesis. Those effects are biologically plausible and clinically relevant, but they take time and dose. A 56-day study can capture some early changes, yet many retinol trials run 12 weeks or longer, and the positive studies summarized in the 2022 reviews used much higher concentrations than 0.002%.

That is why calling CHP-9 "more potent than retinol" goes too far. The study does not show superiority over clinically used retinol. It shows superiority over a very low retinol dose under a short follow-up window. Those are different claims.

If you want a fair head-to-head, compare CHP-9 with retinol at a dose people actually buy because it has evidence behind it.

What this paper can tell you, and what it cannot

What it can tell you

• A cyclized peptide can produce measurable short-term improvements in wrinkle and barrier-related skin endpoints in a small randomized trial.
• The specific CHP-9 formula in this study appeared better tolerated than the retinol comparator used here.
• The molecular design makes sense. Better stability and permeability are reasonable goals for a peptide intended for topical use.

What it cannot tell you

• That CHP-9 outperforms standard anti-aging retinol regimens used in clinical practice or supported by prior trials.
• That CHP-9 will keep improving results over 12 weeks, 6 months, or a year.
• That the observed effects generalize beyond this China-only cohort of adults ages 30 to 55.
• That the results are independently reliable. There is no published replication yet.

The paper also leaves some ordinary but important confidence questions hanging in the air. It is a single study, the sample is small, and the available report does not disclose a funding source. None of that proves bias. It does lower confidence, especially in a field where cosmetic studies often look impressive right up until someone tries to repeat them.

How this fits with the rest of the skin-aging evidence

The bigger literature on skin aging is messy. Plenty of ingredients show wrinkle or elasticity improvements in small trials: fermented bilberry extract, oleocanthal and oleacein, a probiotic ointment, oral low-molecular-weight collagen peptides, and more. Some are randomized, some are exploratory, some are placebo-controlled, and many rely on device-based surrogate endpoints over a few weeks to a few months.

That does not make the evidence useless. It means the standard for replacing a known category should be higher than "looked good in one 56-day trial." Retinol has limits: irritation, variable stability, and slower onset than people want. It also has a long evidence base at workable concentrations. CHP-9 has a clever design and one human trial with a weak active comparator. Those are not equivalent evidence positions.

The practical verdict

CHP-9 is interesting. Cyclized peptides deserve serious attention, and this study gives researchers a reason to test them properly. The next step is obvious: larger randomized trials, longer follow-up, independent replication, and a comparator that reflects retinol doses with demonstrated efficacy, not a ghost of retinol diluted into irrelevance.

For now, this paper does not justify throwing out retinol. It justifies curiosity about a new peptide class and suspicion toward a study designed to produce a dramatic headline.

The shortest honest summary is this: the chemistry looks smart, the early signal looks real, and the fight was not fair.