Thymosin beta-4 eye drops for dry eye: the 72-person trial missed its main goals

In a phase 2 trial of 72 people with moderate to severe dry eye, twice-daily thymosin beta-4 eye drops for 28 days missed both outcomes the trial was designed to test: inferior corneal staining and ocular discomfort were, if anything, slightly worse on the drops than on placebo [1]. The analyses the authors ran afterwards did find improvements in other measures of corneal damage and discomfort, and the drops caused no worrying side effects over four weeks. The larger of the two published trials is a clean negative on its own endpoints — and the positive signals both trials share are the reason to keep testing these drops, not the reason to sell them.

Why this study matters

Thymosin beta-4 (Tβ4) is a 43-amino-acid peptide found in platelets and white blood cells, where it helps control the actin scaffolding inside cells [1]. In rats and mice with injured corneas, Tβ4 drops sped up healing and lowered inflammatory signals [1]. Whether that translates into people is exactly the kind of question a placebo-controlled trial exists to answer.

It's also relevant to anyone who has come across Tβ4 in the injectable peptide world, where the name “TB-500” gets attached to it. This trial tested one thing only: a sterile, preservative-free 0.1% eye-drop formulation of synthetic Tβ4, called RGN-259 [1]. It tells you nothing about injected products — and if you are weighing a product sold as TB-500, the questions that matter are the ordinary ones for any injectable: verified purity, sterility, and testing for endotoxin and heavy metals. This trial speaks to none of them.

We picked this trial over the other published phase 2 study, a nine-patient trial in severe dry eye [2], because it's eight times larger and pre-specified its main outcomes. That second trial comes up again below.

What they did

The trial used a “controlled adverse environment” chamber, which standardises the conditions that aggravate dry eye and, according to the authors, gives a treatment more room to show an effect [1]. Participants went into the chamber before starting treatment and again on day 28, with check-ups the day after and two days after.

The two primary outcomes, fixed in advance, were inferior corneal staining (a dye test that shows damaged surface cells on the lower cornea) and ocular discomfort, both measured on day 29, the day after the second chamber visit [1]. Each person's worse eye was analysed.

What they found

On the pre-specified outcomes, Tβ4 didn't win. Mean inferior corneal staining was 2.08 on Tβ4 and 1.90 on placebo (p = 0.26); ocular discomfort was 1.6 versus 1.3 (p = 0.22) [1]. Both numbers were slightly higher, meaning slightly worse, in the Tβ4 group, though neither gap was meaningful.

The authors then adjusted for baseline differences in “several ad hoc statistical analyses”, in their words, and found some significant results [1]:

The superior staining p value is reported as 0.0210 in the abstract and 0.0109 or 0.0181 in the paper's table, depending on the test; the discomfort result appears as 0.0244 in the abstract and 0.0224 in the text [1]. Small inconsistencies like these don't change the picture, but they're the kind of detail that makes us read a paper's conclusions more carefully.

The statistical caveat is real, and we will state it once. When the primary outcomes fail and several alternative analyses are run afterwards, one or two will cross p = 0.05 by chance. That is the reason these findings need a trial built around them, not a reason to ignore the direction they point in.

Safety

Here the paper's own summary and its results section don't quite agree. The abstract says no adverse events were observed. The full text reports that two people on Tβ4 (5.6%) and five on placebo (13.9%) had at least one eye-related adverse event, all mild on Tβ4: one case each of eye pain, blurred vision and a stye [1]. One placebo-group participant withdrew because of vomiting judged unrelated to treatment. Vision, eye pressure, corneal sensitivity and eye examinations showed no clinically significant changes [1].

The fair reading is that 28 days of these drops caused no safety concerns in 36 treated people — a short exposure in a small group, and a clean one.

The nine-patient trial

The other phase 2 study randomised nine patients with severe dry eye, some with graft-versus-host disease, at two US sites [2]. They used 0.1% Tβ4 or vehicle drops six times a day for 28 days, then were followed for another 28 days. At day 56, the Tβ4 group (12 eyes) had 35.1% less ocular discomfort than the vehicle group (6 eyes, p = 0.014) and 59.1% less corneal staining (p = 0.011) [2]. The authors reported the drops as safe and well tolerated [2].

Those are bigger effects than the 72-person trial found, in sicker patients, but with nine people the estimates are fragile. A single responder can move the averages a long way.

Caveats

• Primary outcomes failed. Everything positive in the larger trial comes from secondary or post hoc analyses [1].
• Short and small. 28 days, 72 people, a single site [1]. Dry eye is a chronic condition, a point the authors make themselves [1].
• Artificial setting. The chamber model is designed to provoke symptoms and may not reflect everyday dry eye.
• Who's behind it. The drug was RegeneRx's RGN-259 formulation [1]. The authors declared no conflicts of interest, and the text we read doesn't include a separate funding statement.
• Mostly White women. 75% women and 93% White [1].

What it means in practice

Tβ4 eye drops looked safe over four weeks, and several secondary measures hinted at less corneal damage and less discomfort under stress. But the one trial big enough to be taken seriously didn't meet its own targets, and everything positive in it comes from analyses chosen after the fact. As a practical matter, that leaves these drops as a promising candidate rather than a dry-eye treatment you can rely on — and neither trial says anything about Tβ4 given any other way.

What is still open

• Whether the secondary improvements hold up in a trial that makes them the primary outcomes. A PubMed search for RGN-259 in September 2026 turned up no peer-reviewed phase 3 dry-eye results.
• Whether effects last beyond four weeks, or what longer use does to the eye.
• Which patients, if any, benefit most. The nine-patient trial enrolled sicker people and saw larger effects, but nine people is far too small to say whether severity is the reason.

What this means

Our position is that thymosin beta-4 eye drops deserve a properly powered phase 3 trial, not a shrug. We say that with the primary result stated plainly, not softened: the trial was built to test two outcomes and it failed both of them, and we will not dress that up as a win for the drops [1]. But the direction of the evidence is consistent rather than random. This trial's corneal-staining and discomfort measures moved the same way in the nine-patient study, which found 35.1% less discomfort and 59.1% less staining than vehicle at day 56 [2] — and the larger trial's own positive signals came from staining and discomfort, the same two domains as its failed primaries. A compound that fails a test and a compound that does nothing tend to look different, and this one looks like the first.

For a reader weighing Tβ4 today, that distinction is the whole point, and it cuts both ways. The drops are not something to buy and rely on: the only product tested is RGN-259, a specific sterile, preservative-free 0.1% formulation that is not an approved medicine anywhere, and a 28-day trial of 72 people can no more establish that a treatment works than it can rule it out [1]. Anyone selling Tβ4 eye drops as a proven dry-eye fix is selling the post-hoc analyses and skipping the primaries.

Where we push back hardest on the hype is the leap from eye drops to injections. Nothing here tests TB-500 given by injection, and the formulation in this trial — sterile, preservative-free, and made to a pharmaceutical standard — is exactly what a grey-market vial of “TB-500” is not guaranteed to be. If the question is what is in the vial, the answer is purity, sterility, endotoxin and heavy-metal testing. We would rather a reader ask for those than treat either trial as clearance for the injectable market.

What readers report

First-hand accounts, shared with permission. They are self-reported, not evidence, and sit alongside the research above rather than in place of it.

Educational content only — not medical advice. Many peptides discussed on HPR are not approved for human use. Talk to a qualified clinician before making any decision about your health.