LL-37 Results: The Follow-Up Trial That Didn't Confirm Them

LL-37 has exactly two randomized, placebo-controlled human trials, and both tested the same thing: a topical formulation on hard-to-heal venous leg ulcers. The first, smaller trial found a real effect. The second, larger and more rigorous trial did not confirm it in the group it was designed to detect it in. That is the entire human results record for this peptide. Anyone citing “LL-37 results” to justify skin, muscle, or general antimicrobial use is extrapolating from a wound-care signal that failed its own confirmation test.

Two trials, one wound type, no anti-aging data

LL-37 is the only cathelicidin-family antimicrobial peptide the human body makes, produced by processing a precursor protein and expressed across epithelial tissue and immune cells ^[1]. That biology gets cited constantly in forum threads about skin repair and general immune support. The clinical testing does not back that reach.

• What exists: two randomized, placebo-controlled trials, both topical, both in venous leg ulcers.
• What does not exist: any trial of systemic LL-37, injected LL-37, or LL-37 for skin aesthetics, acne, or general antimicrobial use in healthy people.

That boundary, trial design, dose range, and patient population, defines the entire scope of what counts as an established LL-37 result in humans. Everything else is mechanism and inference; none of it has been tested as an outcome in a trial.

What the first trial found

The first randomized, placebo-controlled human trial of LL-37, a 2014 first-in-man study by Grönberg and colleagues, tested a topical formulation in 34 patients with hard-to-heal venous leg ulcers ^[2]. It was not systemic or cosmetic use. It was a dose-ranging design: three concentrations (0.5, 1.6, and 3.2 mg/mL) applied directly to the wound bed twice weekly, against placebo ^[2].

The results were dose-dependent in an odd way.

6x faster healing rate versus placebo at the 0.5 mg/mL dose in the 2014 trial PMID 25041740

What the numbers showed: the 0.5 mg/mL dose produced a healing rate roughly six times that of placebo (p = 0.003), and the 1.6 mg/mL dose roughly three times placebo (p = 0.088, short of the conventional significance threshold). The highest dose tested, 3.2 mg/mL, showed no difference from placebo at all ^[2].

More peptide did not mean more effect. The dose-response curve bent backward, which is itself a flag that something about the mechanism at higher concentrations works against the drug. Forum summaries rarely mention that detail. The trial reported no safety concerns, with topical LL-37 described as safe and well tolerated at all three doses ^[2].

What the bigger trial found instead

A result this promising, in a trial this small, calls for a bigger confirmatory study, and LL-37 got one. A larger follow-up randomized, placebo-controlled trial published in 2021, called HEAL LL-37, tested the same two lower doses (0.5 and 1.6 mg/mL) against placebo in 148 patients with venous leg ulcers, combined with standard compression therapy ^[3].

LL-37's two human trials

Both in venous leg ulcers; nothing else has been tested.

In the full study population, there was no significant improvement in healing with LL-37 compared with placebo ^[3]. That is the headline result of the larger, better-powered trial, and it is the opposite of what the 2014 study would have predicted.

The 2021 paper does not end there. A post hoc subgroup analysis of patients with wounds of at least 10 square centimeters, a population known to heal worse on standard care, did find a significant benefit ^[3]. That is a real finding, but it is a post hoc one: identified after the data came in, in a trial that was not designed or powered to test that specific subgroup. It is a reason to run another trial in large wounds specifically. Treating it as an established result would be getting ahead of the data.

Bottom line: LL-37’s one genuinely positive trial result, after the field’s one attempt to confirm it, survives only as a secondary finding in a wound-size subgroup nobody pre-specified. That is a weaker evidence base than a single small positive trial by itself, because it means the original effect did not replicate where it was supposed to.

This was an early-phase, dose-ranging design in the first case, and this was still a phase IIb trial rather than a large multi-site registration study in the second. Neither is a definitive answer. But the direction of travel, bigger trial, smaller effect, matters more than either result read alone.

The vitamin D connection is a correlation, not a treatment result

Outside the two trials, most of what circulates about LL-37 is observational biology, and the distinction between an observational finding and an interventional one is the whole story here.

An observational cohort study of 121 critically ill ICU patients found that day-one plasma hCAP18, the LL-37 precursor, correlated directly with vitamin D levels (rho = 0.30, p = 0.001), and patients with the lowest hCAP18 levels had higher 90-day mortality, a finding reported by Leaf and colleagues ^[4].

121 critically ill ICU patients whose cathelicidin precursor levels tracked vitamin D status PMID 25887571

That is an association in a sick population. It is not a treatment effect, and a correlation between vitamin D status and cathelicidin levels in an observational cohort does not establish that raising LL-37 levels would improve outcomes in those patients, any more than finding that taller people earn more establishes that height causes income. Nobody has run the trial that would test it: give critically ill patients a vitamin D or LL-37 intervention and see whether mortality changes because of the peptide specifically.

LL-37’s direct antimicrobial activity has a similar evidence shape. It has been demonstrated in laboratory assays, including killing of the bacterium Legionella micdadei through interaction with its membrane phospholipids, by Palusińska-Szysz and colleagues, rather than in a human infection trial ^[5]. That is real mechanism, shown in cell-based systems. It is not evidence that applying or injecting LL-37 treats an infection in a person.

The same peptide drives two inflammatory skin diseases when it runs high

Here is the part the wound-healing framing skips entirely, and it is the one genuine surprise in LL-37’s literature: this peptide is not a uniformly protective molecule. It is independently implicated as a cause of disease when its signaling runs too hot, in two separate skin conditions.

• Psoriasis. LL-37 bound to self-DNA forms complexes that are taken up by plasmacytoid dendritic cells and trigger Toll-like receptor 9, a mechanism identified by Lande and colleagues as a key driver of the autoimmune activation seen in psoriasis ^[6]. In plain terms, LL-37 is what converts the body’s own DNA, normally invisible to the immune system, into something the immune system attacks.
• Rosacea. Reviews of cathelicidin biology, including one by Schauber and colleagues, describe overabundant expression of processed LL-37 peptide fragments in the skin of rosacea patients as a driver of the condition’s inflammation, the opposite problem from the decreased cathelicidin expression seen in atopic eczema ^[7].

The warning the hype skips: LL-37 is not a dial that only goes one direction. Too little of it is linked to poor antimicrobial defense. Too much of it, or the wrong processing of it, is linked to two separate inflammatory skin diseases. A peptide that drives autoimmune activation in psoriasis at excess levels is not an obvious candidate for self-administration to “support skin health.”

That double role is not unique to LL-37. Epidermal growth factor promotes healthy gut lining growth in breastfeeding infants and aggressively drives the growth of certain breast cancer cells in adults, the same molecule doing opposite things depending on context and dose. Biology is full of signals that help at the right level and harm outside it. LL-37 belongs on that list, not on the list of molecules where more is simply better.

What the trial dosing looked like, and what was never tested

The 2014 trial applied LL-37 topically at 0.5, 1.6, or 3.2 mg/mL directly onto the wound bed twice weekly, not as a systemic or subcutaneous injection ^[2]. That is the only dosing regimen with trial data behind it, and it describes a wound dressing concentration rather than anything resembling a systemic dose.

No published human trial has tested subcutaneous or intramuscular injection of LL-37. There is no established dose, interval, or pharmacokinetic profile for that route in people. Protocols circulating outside the two published trials are not drawn from any clinical data at all; they are guesses dressed up as regimens.

Where this leaves someone deciding what to do with it

The evidence supports LL-37 as a topical, clinically supervised option specifically being studied for hard-to-heal venous leg ulcers, not as a general wound-healing or longevity peptide. Even within that narrow lane, the field’s own confirmatory trial came back negative in the population it was designed to treat.

For comparison, thymosin beta 4, the peptide sold under the name TB-500, has phase 2 human trial data across pressure ulcers, venous stasis ulcers, and epidermolysis bullosa wounds, reviewed by Kleinman and Sosne, a broader interventional record than LL-37’s two trials, both confined to venous leg ulcers ^[8]. That does not make thymosin beta 4 a settled success story either; it has its own gaps, covered in our look at what its trials can and cannot explain. It does mean that if the goal is general chronic wound care rather than venous ulcers specifically, LL-37 is not the better-studied choice.

Because LL-37 is also implicated in driving inflammatory skin disease when its signaling runs high, self-administering it for skin or antimicrobial purposes outside a clinical trial carries a plausible risk that has never been tested in healthy people. Nobody has run a safety trial in people who are not being treated for an open wound. The psoriasis and rosacea literature did not appear because researchers were looking for reasons to worry; it appeared because that is what the molecule does at the wrong level.

LL-37 is, like a lot of endogenous peptides getting marketed as general-purpose tools, a molecule the body already makes for a specific job, now being proposed for jobs nobody has tested it for. Our broader look at which natural peptides have real human trial support runs into the same pattern repeatedly. For LL-37 specifically, the honest summary is short: one wound-care trial worked, the bigger one that was supposed to confirm it did not, and the same peptide has an independent paper trail showing it can make skin disease worse. That is not a reason to panic about an endogenous peptide your body produces anyway. It is a reason to stop treating “LL-37 results” as settled evidence for anything beyond the ulcer it was tested on.

This article is for research and informational purposes only and is not intended to diagnose, treat, cure, or prevent any disease. The peptides discussed here are sold for research use only and are not for human consumption. Nothing in this article constitutes medical advice. Consult a qualified clinician before making changes to a health, training, or supplementation protocol.