Why most peptide evidence is in animals, and how to read it

Most peptides sold online have been tested mainly in rats, mice or cells, and a result in a rodent is a reason to run a human trial, not a substitute for one. The best recent estimate is that about half of treatments tested in animals ever reach a human study, and only 5% end up approved [1]. This guide explains why the gap is so wide and gives you a quick way to read any peptide abstract.

Why so much of the evidence is animal work

Animal studies are cheap, quick and easy to publish compared with human trials. A university lab can inject a peptide into 40 rats, measure a tendon or a scar, and write it up within a year. A properly run human trial needs manufacturing to medicine standards, safety testing, ethics approval, a regulator's sign-off, and years of recruitment and follow-up. Somebody has to pay for all of that, and it's usually a company that expects to own and sell the result.

Many of the peptides people buy online have no such company behind them, or had one that stopped. So the literature stalls at the stage that's affordable: dozens of rodent papers, often from one or two groups, and no trial. That's a gap in the evidence, not a hidden vote of confidence.

How often animal results carry over

Meta-researchers have tried to put numbers on this. Three studies give a fair picture.

The 2024 umbrella review adds a twist worth knowing. When a therapy did reach human studies, positive animal results and positive human results agreed 86% of the time, and the authors argued translation may be better than its reputation [1]. The catch sits in the last number. It took a median of 10 years from animal study to approval, and 95% of therapies never got there [1]. A promising rodent paper is the start of a long, mostly unsuccessful road.

Drugs that make it into people still fail most of the time. An analysis of industry data from 2003 to 2011 found that only around one in ten drugs entering phase 1 trials went on to approval [10].

Why rodent results mislead

The biology differs

Mice aren't small humans. A large genomic study compared blood gene responses in people after trauma, burns and sepsis with the matching mouse models. The human conditions looked strikingly similar to one another; the mouse models barely matched the human responses at all, with correlations close to random [7]. For anything framed as “anti-inflammatory” or “immune-modulating”, that's a real limit.

Rodent wounds heal differently

This matters for the healing and recovery peptides that dominate online marketing. Mice and rats have a thin muscle layer under the skin, the panniculus carnosus, that pulls wound edges together, so their skin wounds close largely by contraction, whereas human wounds rely more on growing new tissue to fill the gap [8]. Mouse skin also differs in thickness, hair-follicle density and immune make-up. A reviewer of the field put it plainly: no animal model fully predicts clinical outcomes in skin healing [8]. Tendon, gut and muscle injury models have their own versions of this problem, because the injury is induced in a young, healthy animal on a set schedule.

Doses don't scale by body weight

Small animals burn through drugs faster, so doses in mg/kg don't transfer directly. The standard method, used by regulators to pick safe starting doses, scales by body surface area. For rats, you divide the mg/kg dose by about 6.2 to get a human-equivalent mg/kg dose; for mice, the factor is about 12.3 [9]. For first-in-human trials, regulators then divide by a further safety factor, typically 10 [9].

That conversion is a starting point for safety, not a promise of effect. It says nothing about whether a peptide survives digestion, how long it lasts in human blood, or whether it reaches the target tissue. Forum dosing charts that multiply a rat's micrograms per kilogram by a person's weight are doing the sum wrong and answering the wrong question.

Many animal studies are weakly designed

A survey of 271 animal studies from publicly funded UK and US institutions found that 87% didn't use randomisation to assign animals to groups and 86% didn't blind the people measuring the outcome [6]. Only 59% even stated the study's aim along with the number and characteristics of the animals used [6]. Without those two safeguards, results lean towards what the researchers hoped to see.

Negative results often go unpublished

In animal stroke research, only 2% of 525 publications reported no significant effect on the main outcome. Statistical adjustment for the missing studies suggested publication bias inflated the reported benefit by about a third [5]. Stroke is a well-studied case; there's no reason to think peptide research is cleaner.

Sometimes animals point the wrong way

A systematic comparison found cases where animal and human results flatly disagreed [4]. Corticosteroids helped in animal models of head injury but not in the human trial. The stroke drug tirilazad shrank brain damage by 29% in animals and was linked to worse outcomes in patients [4].

How to read a peptide abstract in two minutes

You don't need the full paper to triage a study. The abstract usually answers these questions, and the answers tell you how much weight it can carry.

Two phrases deserve extra suspicion. “Trend towards” means the result wasn't statistically significant. And “improved significantly from baseline” in a controlled trial may hide the fact that the control group improved too; look for a comparison between groups.

A reasonable way to hold animal evidence

Treat a single rodent study as a question. Treat several consistent rodent studies from independent labs as a good question. Treat a small human trial as the first real clue about people, and a replicated, blinded, adequately sized human trial as evidence. On this site we label every article with the strongest kind of evidence it rests on for exactly this reason, and we name the species in the same sentence as the result.

What we still don’t know

• Which animal models, if any, reliably predict human results for peptides specifically; the meta-research above covers drugs and therapies in general.
• How much unpublished negative peptide research exists.
• Whether newer approaches (human organoids, tissue chips) will close the gap; they're promising but largely unvalidated for predicting treatment effects.

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.