Why Most Peptide Trials Are Small and Short

The census

Every figure below is the total study count ClinicalTrials.gov returned for that exact query term in August 2026. Counts are a property of a query, not of the world — see the caveat further down.

AOD-9604 — 0 studies.

TB-500 — 1 study.

CJC-1295 — 1 study.

BPC-157 — 3 studies.

Ipamorelin — 3 studies.

GHK-Cu — 3 studies.

Ibutamoren or MK-677 — 8 studies.

Thymosin beta 4 — 18 studies.

Tesamorelin — 24 studies.

Survodutide — 24 studies.

Retatrutide — 34 studies.

Cagrilintide — 44 studies.

Tirzepatide — 286 studies.

Semaglutide — 759 studies.

That is a spread of at least 759 to nothing, across compounds discussed in the same conversations and often on the same forums. The gap is not a measure of which molecules are interesting.

Reason one: somebody has to pay for it

A phase 3 outcomes trial is one of the most expensive things a company can do. Semaglutide's cardiovascular outcomes trial randomised 17,604 patients and followed them for a mean of 39.8 months¹. Retatrutide's registered cardiovascular and kidney outcomes trial, NCT06383390, lists an enrollment of 10,000 and a listed completion date of February 2029.

That is not funded without a period of market exclusivity to recover it from.

A peptide first described in the public literature decades ago is difficult to protect. Without protection there is no return on a trial of that size, so the trial does not get run — regardless of how promising the preclinical work is.

Reason two: who is left to run it

When there is no commercial sponsor, the trials that do happen are investigator-initiated, and academic budgets buy tens of participants, not thousands.

The pattern shows up directly in the census. On the small end, the largest listed enrollment in the BPC-157 set as of August 2026 is 120, on a phase 2 record listed as recruiting. On the large end, a single retatrutide record lists 10,000.

Both are legitimate science. They answer questions of completely different sizes.

Reason three: surrogates are cheap and outcomes are not

A trial that measures a blood marker can be short and small. A trial that measures whether people have fewer heart attacks has to be long and large, because the events are rare.

The growth-hormone secretagogue literature is the clearest case. A 563-patient trial found MK-677 raised serum IGF-1 by 72.9% at twelve months and showed no significant difference on any of its four clinical efficacy measures². A 161-patient hip-fracture trial found an 84% IGF-1 rise against 17% on placebo, and no significant difference in functional performance³.

Both trials moved their surrogate easily. Both cost far more to answer the real question, and both answered it in the negative. That asymmetry is exactly why so much of this literature stops at the marker. We take the class apart in growth-hormone secretagogues.

What "small" actually costs you

Precision. A small trial produces a wide confidence interval, so the true effect could be much larger or much smaller than the point estimate — and the headline is the point estimate.

Rare harms. A 100-person trial cannot detect a side effect that happens to one person in 500. The absence of a signal in a small trial is not evidence of its absence.

Fragility. The clearest illustration in this field: a phase 2 dry-eye trial of thymosin β4 in nine patients reported a 59.1% reduction in corneal fluorescein staining against vehicle, p = 0.0108⁴. The phase 3 that followed enrolled 601 and posted co-primary values separated by less than 0.15 of a point. Nine people to 601 is where that kind of result usually goes.

What "short" actually costs you

Weight regain, tolerance, adherence and late harms all live past the end of a short trial.

The incretin class has some of the longest data in the category and it took years to build. Semaglutide's outcomes trial ran a mean of 39.8 months of follow-up¹. Compare that with a 16-week ascending-dose phase 1, which is where several currently-discussed molecules stop.

Neither is wrong. But a 16-week number and a 39.8-month number are not the same kind of fact, and stacking them in one comparison is a very common error in this subject.

The caveat that makes these counts honest

A registry search matches text anywhere in a record, including outcome-measure descriptions. That over-counts.

Two examples from this same census. The sermorelin query returned 42 studies, but the set includes many trials of growth-hormone-releasing hormone generally, not of that specific analog. The MOTS-c query returned 9, of which most are studies measuring MOTS-c as a biomarker in some other context — one record in that set is an interventional phase 2 of the compound itself, listing 120 participants and marked recruiting.

So a count is a count of what a query returned on a date. It is not a claim that no other study exists anywhere, and any of these numbers can be checked by re-running the query.

What would change any of these rows

A commercial sponsor with a reason to fund a large trial. A regulatory pathway that makes the trial worth funding. Or a public funder deciding the question matters enough.

Until one of those happens, a compound's registry row stays where it is — and that row describes the state of its evidence, not the state of the molecule. Our research index tracks the rows as they move, and healing peptides covers the compounds sitting at the bottom of this list in detail.