Route · Injection

Collagenase injection: it beat placebo across 843 people, and 7.6% responded

An enzyme injected to cut the fibrous septa that tether cellulite dimples down. It did significantly better than placebo in two phase 3 trials, and on the stricter endpoint 7.6% and 5.6% responded (against 1.9% and 0.5% on placebo).

Last checked Sep 20, 2026

Start here

What it is: An enzyme that breaks down collagen, injected directly into the dimples of cellulite.
What it is used for: Cellulite on the buttocks and thighs.
How it is said to work: The premise is that dimpling comes not from more fat but from fibrous bands pulling the skin down, and the enzyme is described as cutting those bands.

There are no numbers and no citations in this box. What has been established, and how far, is below.

At a glance

Session length: 20–40 min
Pain: Moderate
Downtime: 7–21 days
Sessions: 1–3 sessionsReported in a trialIn the phase 3 trials behind the approval, treatments were given up to three times, 21 days apart.
How long it lasts: Not applicable
Relative cost: Higher
How well established: Moderate evidence

“Reported in a trial” means a number a study on this page actually reported. “Typical range” is not a measured value but roughly how it is usually done, and it varies by clinic and by person.

Cost is not an amount but a band: where this procedure falls when the ones on this site are lined up per session, or per area treated. Real prices differ several-fold by clinic and by how much is treated, so they are not given here. Surgery is measured on a different scale and cannot be compared with this band.

This is not how large the effect is but how well confirmed it is. It shows the highest grade among the findings on this page.

Korean classification Being checkedThe regulatory class has not yet been confirmed against a primary document. It will be filled in once it is.

Cellulite dimples are not about having more fat. Fibrous septa running from the skin down to the fascia beneath pull across the fat layer, and fat bulges between them, creating the contour.

Collagenase is an enzyme that breaks down collagen. This injection puts it directly into those septa to cut them. In principle it does with an enzyme what subcision does with a needle.

What makes it stand out on this site is the scale of the evidence. Two placebo-controlled phase 3 trials with 843 participants. That size is rare for a cosmetic treatment.

And it is those same trials that give this page its most important number. It is set out below.

What is established

Moderate evidence

It beat placebo significantly in two phase 3 trials

The trials are RELEASE-1 and RELEASE-2. 843 women with cellulite received at least one injection.

RELEASE-1: 210 on drug, 213 on placebo
RELEASE-2: 214 on drug, 206 on placebo

Assessment used a composite of photonumeric scales scored separately by the clinician and by the participant. Both had to see at least a one-level (or two-level) improvement for it to count as a response.

At least one level of improvement:

RELEASE-1: 37.1% vs 17.8% (P < .001)
RELEASE-2: 41.6% vs 11.2% (P < .001)

All four figures are significantly above placebo, and the result was reproduced independently twice. That is the strength of this evidence.

We still did not raise it to 'strong evidence'. This site's rule blocks it: both trials were funded by the manufacturer, and employees of the manufacturer appear on the author list. Scale does not remove that condition.

Randomised controlled trial in people · 423 participants · Compared with vehicle (same formula minus the active) · Funding manufacturer-funded · Endpoint scores from blinded assessors PMID 33840781
Randomised controlled trial in people · 420 participants · Compared with vehicle (same formula minus the active) · Funding manufacturer-funded · Endpoint scores from blinded assessors PMID 33840781

Moderate evidence

But 'significant' and 'large' are different things

Now the stricter endpoint from the same trials — the proportion where clinician and participant both saw at least a two-level improvement.

RELEASE-1: 7.6% vs 1.9% (P = .006)
RELEASE-2: 5.6% vs 0.5% (P = .002)

The difference against placebo is clear. From 1.9% to 7.6% is a fourfold gap, and it is statistically significant.

And 92 of every 100 people who received the drug did not reach that bar.

Both sentences are true. Write only the first and it is advertising; write only the second and it is disparagement of real evidence. That is why this site writes both.

On the one-level bar (37.1%, 41.6%) the numbers are far larger. Which bar matters depends on whether one level of improvement means anything to you — and that is a question the trial does not answer.

A 2022 pooled analysis of the two trials pointed the same way.

Adverse events were mostly at the injection site. Few discontinued because of them, as reported. Bruising in this class is common and lasting.

Randomised controlled trial in people (2) · 843 participants · Compared with vehicle (same formula minus the active) · Funding manufacturer-funded · Endpoint scores from blinded assessors PMID 35646496

Known risks

Bruising is common and lasts. Adverse events were mostly at the injection site, with bruising, pain, swelling and nodules reported.
Many injections in one visit. Each dimple is injected, so the number of punctures is high and the bruising is spread across them.
The enzyme breaks down collagen. Outside the intended target it can act on surrounding tissue.
Only trials in women are cited here. The trials above enrolled women.
We could not confirm its regulatory status in Korea from primary sources. This site does not state a classification it has not verified.

What is not established

We could not include data on how long it lasts. Whether the cut septa re-form, and how many years the effect holds, cannot be answered here.
We found no randomised head-to-head against another method. How it compares with subcision, which does the same thing with a needle, is unestablished.
No data for areas other than buttocks and thighs.
Both trials were funded by the same manufacturer. Independent replication has not appeared.

Whether to have it done is a decision to make with a doctor.

What this article relies on

PubMed — 미국 국립의학도서관 문헌 데이터베이스National Library of Medicine · Checked on 2026-09-16