Animal vs human peptide studies often disagree, and the gap is large. In a JAMA analysis of 76 highly cited animal studies, only 37% were later confirmed by human randomized trials, while 18% were contradicted and 45% were never tested.[1] Across all drugs, about 90% of candidates that enter human clinical trials fail.[2] So a peptide that heals tendons in rats has not been shown to help people.
This guide explains why results don’t carry over and how to weigh animal data. New to peptides? Start at the peptide basics hub.
Animal vs human evidence at a glance
| Animal studies | Human trials | |
|---|---|---|
| Purpose | Explore how a drug works, screen for toxicity | Test safety and benefit in people |
| Typical size | Often small groups of animals | Dozens to thousands of people |
| Disease | Artificially induced (cut tendon, chemical injury) | Real disease as it occurs |
| Dose | Scaled by body surface area, not weight[4] | Set and tested directly |
| Can support approval alone? | No | Yes, with adequate trials |
The key studies on translation
Hackam and Redelmeier, JAMA 2006. The authors screened animal studies from leading journals with more than 500 citations. Of 76, 28 (37%) were replicated in human randomized trials. The median time to replication was 7 years. Few animal studies used random allocation or blinded outcome assessment.[1]
Seok et al., PNAS 2013. Researchers compared gene activity after trauma, burns and sepsis in people and in standard mouse models. Human responses were highly similar across causes. Mouse responses correlated poorly with the human ones, close to random for many genes.[3]
Sun et al., 2022. This review traced the 90% clinical failure rate to issues including lack of effect and unmanageable toxicity in people, and argued that drug selection overlooks how drugs distribute into human tissues.[2]
Why peptide results don’t transfer
- Different biology. Receptors, enzymes and immune systems differ between species, so a peptide can bind or break down differently.[3]
- Dose scaling. Small animals clear drugs faster. FDA guidance converts animal doses to human-equivalent doses using body surface area, not body weight, and adds safety margins.[4] A dose that is effective in a rat says little about what is safe or effective in a person.
- Artificial injuries. A surgically cut rat tendon heals differently from a chronic human tendon problem.
- Weak study design. Missing randomization and blinding inflate effects.[1] The ARRIVE 2.0 guidelines were written to improve how animal research is reported.[8]
- Publication bias. Positive results are more likely to be published, so the animal literature can look more consistent than reality.
A dose-scaling example
FDA’s guidance converts an animal dose in mg/kg to a human-equivalent dose by body surface area. For a rat, that means dividing the mg/kg dose by about 6.2, and for a mouse by about 12.3.[4] A rat dose of 10 mcg/kg therefore corresponds to roughly 1.6 mcg/kg in a person, not 10. Even then, the guidance applies a further safety factor, usually dividing by 10, to set a cautious starting dose for a first human trial.[4] These conversions are designed for planning supervised trials. They are not a way to pick a personal dose of an unapproved peptide.
How to weigh an animal study you read about
When an article or seller cites research on a peptide, ask:
- What species? Mice and rats are standard but differ from people in important ways.[3]
- How was the disease made? An injury created in a healthy young animal is not the same as a long-standing human condition.
- Was it randomized and blinded? Many animal studies are not, which inflates effects.[1]
- Has anyone repeated it? One lab’s results, however many papers, are weaker than independent replication.
- Is there any human trial? If not, the honest summary is “unknown in people.”
It works both ways: animal risks may not apply
Animal studies can also flag risks that may or may not affect people. Semaglutide causes thyroid C-cell tumors in rodents at clinically relevant exposures. The Wegovy label carries a boxed warning but states that human relevance “has not been determined.”[5] See our semaglutide profile.
Case study: BPC-157
BPC-157 is a clear example of a peptide promoted on animal data. A 2025 systematic review screened 544 articles and included 36 studies: 35 preclinical and 1 clinical.[6] A 2026 systematic review of human evidence found three small, uncontrolled studies with 30 participants in total: knee injections (16), bladder instillation for interstitial cystitis (12) and an IV safety pilot (2). It rated the certainty of evidence very low and concluded BPC-157 should not be used clinically outside regulated trials.[7]
TB-500 is in a similar position. See BPC-157 vs TB-500 for a head-to-head look at the evidence.
What’s established vs preliminary
Established:
- Most promising animal findings are not confirmed in human trials.[1][2]
- Mouse models can mimic human disease poorly, at least for acute inflammation.[3]
- Human-equivalent dosing needs body-surface-area scaling and safety margins.[4]
Preliminary:
- Whether lab-grown human tissues, organoids and computer models can replace some animal testing. FDA announced a plan in April 2025 to reduce, refine or potentially replace animal testing for monoclonal antibodies and other drugs.[9]
Limitations of this evidence
- The JAMA analysis covered highly cited studies from top journals, which may translate better than average work.[1]
- Some animal models work well for specific questions, such as basic toxicity screening.
- Translation rates vary by disease area, and peptide-specific data are limited.
What’s next
Expect more human-based testing methods alongside, not instead of, clinical trials. To check whether a claim rests on rats or people, use our guide on how to read peptide research. For which peptide results did hold up in people, see peptide research highlights 2026.
Questions to discuss with a clinician
- Has this peptide been tested in randomized human trials, or only in animals?
- How was the dose I’m being offered chosen?
- Do any animal safety findings, like tumor risks, apply to me?
- Are there regulated clinical trials I could join instead?
