TLDR
- Animal studies are an important part of peptide research, but they are models of human biology, not miniature human clinical trials.
- Species can differ in receptor structure, receptor expression, metabolism, protease activity, immune responses, organ physiology, and pharmacokinetics.
- A peptide that works in mice may reach very different concentrations or tissues in humans.
- Dose cannot be converted from animals to humans simply by multiplying by body weight.
- Disease models can reproduce selected parts of a human condition without reproducing the complete disease.
- Experimental design matters. Randomization, blinding, sample size, controls, and replication can substantially affect the reliability of animal results.
- A 2024 umbrella review found that about 50% of animal-tested interventions in the included literature progressed to some human study, 40% reached randomized trials, but only about 5% ultimately received regulatory approval.
- That 5% figure does not mean animal experiments are wrong 95% of the time. Drug development can fail for many reasons after animal testing.
- FDA is increasingly encouraging human-relevant New Approach Methodologies, including organoids, organs-on-chips, computational models, and other systems that can complement or sometimes replace animal studies.
A peptide can produce striking results in mice and later show little benefit in people.
That is not unusual, and it does not necessarily mean anyone made a mistake.
Animal studies answer an important question: what does this experimental molecule do in this animal model under these conditions? Human clinical research asks a related but different question.
The distance between those questions is the problem known as translation.
Animal studies remain useful for studying pharmacology, toxicity, drug distribution, metabolism, and disease mechanisms. FDA still uses nonclinical evidence when deciding whether investigational drugs can proceed into human studies. But in 2026, FDA is also actively moving toward more human-relevant alternatives because animal biology cannot perfectly predict human biology.
For peptide research, that limitation can be particularly important.
Animal Models Are Models, Not Replicas
A laboratory mouse and a human share a remarkable amount of basic biology.
Both have receptors, enzymes, hormones, kidneys, livers, immune systems, nervous systems, and many of the same cellular signaling pathways.
That shared biology is why animal research can be useful.
But similarity is not identity.
A good animal model isolates enough of the biology under investigation to answer a useful question. It does not need to reproduce every feature of human physiology.
Problems arise when conclusions move beyond what the model actually represents.
A mouse tumor model might answer whether a peptide slows growth of a particular experimental tumor. It cannot, by itself, establish that the peptide will treat human cancer.
Peptide Receptors Can Differ Between Species
Many peptides work by binding receptors.
That makes receptor pharmacology one of the first translational questions.
A mouse and human may both express a receptor with the same name, but differences can exist in amino acid sequence, three-dimensional structure, expression level, tissue distribution, and downstream signaling.
A peptide optimized against a human receptor may bind the corresponding mouse receptor:
more strongly,
less strongly,
or sometimes hardly at all.
The opposite can happen too.
This is especially important for highly selective peptides because a small difference in receptor structure can change ligand binding substantially.
Researchers therefore often evaluate receptor activity directly across species rather than assuming that identical receptor names imply identical pharmacology.
Receptor Expression Can Matter as Much as Receptor Sequence
Even when a peptide binds the receptor similarly in two species, the receptor may not occur in the same places.
Suppose receptor X is abundant in mouse liver but expressed at a lower level in human liver.
The same circulating peptide concentration could then produce very different effects.
Differences can involve:
- tissue distribution;
- receptor density;
- cell type;
- developmental stage;
- disease state.
This is one reason a peptide’s mechanism should be examined in the species being used.
Showing activity against a purified human receptor does not prove that a mouse experiment accurately models human receptor biology.
Peptides Can Be Metabolized Differently
Peptides face an additional translational problem: proteolysis.
The body contains numerous proteases and peptidases capable of cleaving peptide bonds.
Species can differ in:
- enzyme expression;
- enzyme activity;
- tissue localization;
- plasma composition;
- renal processing.
A peptide that remains intact for hours in a mouse might have a shorter or longer lifetime in humans.
This changes exposure.
And exposure can determine whether an otherwise active peptide reaches its receptor often enough and for long enough to produce a meaningful effect.
Species differences in drug-metabolizing systems are well established more broadly, which is why pharmacokinetic translation requires more than simple body-size conversion.
Pharmacokinetics Can Completely Change the Result
Pharmacokinetics describes what the body does to the molecule.
Important parameters include:
Cmax: maximum observed concentration.
AUC: total exposure over time.
Half-life: how quickly circulating concentration declines.
Clearance: how rapidly the drug is removed.
Volume of distribution: how extensively it appears to distribute beyond the bloodstream.
Imagine a peptide that activates its receptor at 20 nanomolar in vitro.
In mice, a particular dose might maintain concentrations above 20 nM for six hours.
In humans, that same molecule might remain above 20 nM for only 30 minutes.
Even if receptor pharmacology were identical, the biological effect could differ because the exposure changed.
This is why drug development increasingly uses PK/PD modeling rather than assuming that the same milligram-per-kilogram dose should have the same effect across species.
You Cannot Simply Scale a Mouse Dose by Body Weight
A common mistake when reading animal research is treating an animal dose as though it were directly transferable to humans.
Suppose a mouse study uses:
10 mg/kg.
Multiplying that number by a person’s body weight is generally not an appropriate way to predict a human dose.
Species differ in:
- metabolism;
- circulation;
- body-surface-area relationships;
- organ function;
- receptor pharmacology;
- clearance.
Traditional approaches have included body-surface-area scaling and allometric scaling. Modern first-in-human development can also use pharmacokinetic modeling and approaches such as the minimum anticipated biological effect level, or MABEL, depending on the molecule and risk.
The correct human starting dose is therefore a drug-development question, not a multiplication problem.
Albumin Binding Can Be Species Dependent
Several modern peptide medicines are intentionally modified to bind serum albumin.
Semaglutide, tirzepatide, and retatrutide are examples of peptides engineered with lipid groups that prolong systemic exposure partly through albumin association.
That raises another translational variable.
Albumin from different species is similar but not identical.
Binding affinity can differ.
Those differences can affect:
- free peptide concentration;
- distribution;
- clearance;
- half-life.
A long-acting peptide therefore needs pharmacokinetic evaluation in relevant species rather than assuming that a modification producing a long half-life in one animal will produce exactly the same duration in humans.
The Immune System Is Another Major Difference
Immune biology is particularly difficult to translate.
Mice and humans share many immune pathways, but they differ in:
- immune-cell proportions;
- cytokine biology;
- receptor expression;
- inflammatory responses;
- microbiome;
- antigen recognition.
These differences become especially important when the experimental peptide is intended to alter inflammation, immunity, or cancer biology.
They also matter when assessing immunogenicity.
A synthetic peptide can potentially be recognized differently by immune systems from different species.
An animal showing no immune reaction therefore does not prove that humans will respond the same way.
Human Disease Is Often More Complicated Than the Animal Model
An experimental disease model is deliberately constructed.
Human disease is not.
Consider obesity.
A laboratory animal may be placed on a defined high-fat diet for a controlled period while researchers standardize:
- genetics;
- age;
- environment;
- food;
- light cycle.
Human obesity can develop over decades and interact with:
- genetics;
- diet;
- medications;
- sleep;
- socioeconomic factors;
- endocrine disorders;
- activity;
- other diseases.
A peptide reducing body weight in a standardized animal model therefore provides useful biological evidence.
It does not capture the complete human condition.
Cancer Models Show the Problem Clearly
Cancer research provides one of the clearest examples of animal-model limitations.
A human tumor cell line can be implanted into an immunodeficient mouse.
Researchers can then determine whether a peptide slows tumor growth in a living organism.
That is useful.
But the model may differ from natural human cancer in several ways.
The mouse may lack a normal immune system.
The implanted tumor may be genetically more uniform.
The tumor may grow much more quickly.
Its blood supply and microenvironment may differ.
And if the tumor is implanted under the skin rather than in the organ where the cancer normally occurs, the anatomical environment is different too.
Each model answers some questions better than others.
Animal Models Can Still Be Highly Informative
The limitations of animal studies do not make them useless.
They can answer questions impossible to answer with a simple cell culture.
For example:
Does the peptide reach the liver?
Does it cross the blood-brain barrier?
How quickly does the kidney clear it?
Does repeated administration alter blood pressure?
Does the peptide accumulate in particular organs?
Does it affect an intact endocrine system?
Does toxicity emerge only after repeated exposure?
An intact organism supplies interconnected physiology that isolated cells cannot reproduce.
The important point is to interpret animal results as animal evidence, not prematurely label them human evidence.
How Often Does Animal Research Translate?
This question is more complicated than commonly presented.
A large 2024 umbrella review in PLOS Biology evaluated 122 systematic reviews covering 54 human diseases and 367 therapeutic interventions.
The researchers reported that approximately:
50% of interventions progressed from animal studies to some form of human study.
40% progressed to a randomized controlled trial.
Only about 5% ultimately received regulatory approval.
The median times from the first animal study were approximately five years to any human study, seven years to a randomized trial, and ten years to regulatory approval.
At first glance, 5% sounds like evidence that animal research rarely predicts humans.
But the same analysis found an approximately 86% concordance between positive animal and clinical-study results among the evidence it examined.
The authors specifically cautioned against the simplistic claim that animal-to-human translation is uniformly poor.
Five Percent Approval Does Not Mean 95 Percent of Animal Studies Were Wrong
There are many reasons an experimental therapy never becomes an approved medicine.
A candidate can fail because:
- toxicity emerges;
- clinical benefit is too small;
- a competitor develops a better drug;
- manufacturing is impractical;
- the commercial program is abandoned;
- the formulation fails;
- trial recruitment fails;
- the human dose cannot produce enough exposure;
- regulatory requirements are not met.
Some drugs may never even be taken into clinical development.
So regulatory approval is an important endpoint, but it is not a pure measurement of animal-model accuracy.
The more useful question is whether the preclinical evidence correctly predicted the specific human endpoint it was designed to predict.
Study Quality Can Create Translation Problems Too
Sometimes the problem is not the species.
It is the experiment.
A poorly designed mouse study can exaggerate an apparent peptide effect.
Potential problems include:
small sample sizes,
lack of randomization,
lack of blinding,
selective exclusion of animals,
selective reporting of favorable outcomes,
failure to replicate the experiment.
These issues are serious enough that ARRIVE 2.0 identifies sample size, randomization, blinding, outcome definition, statistical analysis, and inclusion/exclusion criteria among the essential information needed to evaluate an animal study.
A biologically imperfect but rigorous animal model can be useful.
A biologically relevant but poorly conducted study can still produce unreliable evidence.
Healthy Laboratory Animals Can Be Too Uniform
Laboratory research often reduces variability intentionally.
Animals may be similar in:
- genetics;
- age;
- sex;
- body weight;
- diet;
- housing.
That makes treatment effects easier to detect.
But human populations are heterogeneous.
Patients vary in:
- genetics;
- age;
- disease severity;
- other medications;
- kidney function;
- liver function;
- diet;
- comorbidities.
A peptide that performs consistently in genetically similar young mice may encounter far more variability in a human trial.
Better translational research therefore often includes several models rather than relying on one narrowly defined population.
Sex Can Affect Results
Historically, some animal research relied heavily on males to reduce variability associated with hormonal cycles.
That practice can create another translation gap.
Hormones, metabolism, body composition, immune function, and receptor expression can differ between males and females.
A peptide affecting endocrine or metabolic pathways may therefore behave differently.
Modern research guidelines increasingly emphasize considering biological sex in experimental design rather than assuming one sex represents both.
The Route of Administration Matters
A peptide injected intravenously in a mouse is not equivalent to the same peptide administered subcutaneously in a person.
Route changes:
- absorption;
- concentration-time profile;
- tissue exposure;
- degradation;
- local reactions.
Researchers should ideally use a route that is relevant to the intended clinical application or explain why another route was needed.
A powerful result after direct injection into a mouse brain, for example, provides little evidence that the peptide can reach the brain after a conventional systemic injection.
Timing Matters Too
Disease experiments often begin treatment at convenient experimental stages.
That can produce misleading comparisons with clinical medicine.
A peptide might be given to an animal:
before disease develops,
immediately after an induced injury,
or when a tumor is extremely small.
Human patients may not receive treatment until disease is well established.
A therapy that prevents an experimentally induced problem is not necessarily capable of reversing an advanced human disease.
Study timing should therefore be part of the translational assessment.
Human-Relevant Models Are Expanding
The alternative to imperfect animal experiments is not simply eliminating biological research.
Scientists are developing additional models that can complement animal data.
These New Approach Methodologies, or NAMs, include:
- human cell systems;
- organoids;
- organs-on-chips;
- computational toxicology;
- physiologically based pharmacokinetic models;
- AI-assisted prediction.
In March 2026, FDA released draft guidance describing a validation framework for using NAMs in drug development. FDA said the goal is to improve the human relevance of nonclinical evidence while reducing reliance on animal testing.
In April 2026, FDA reported completing first-year milestones from its roadmap for reducing animal testing.
This does not mean animal research suddenly has no role.
It means regulators increasingly recognize that the best predictive model may sometimes be a combination of approaches.
Humanized Models May Help, but They Are Still Models
Researchers can genetically alter mice or transplant human cells and tissues into them to reproduce selected aspects of human biology.
For example, humanized mice are being studied to improve prediction of human drug metabolism and transport.
A 2026 systematic review noted the same underlying problem: standard animal models can face translational challenges because of species differences in drug metabolism and transport.
Humanization can reduce a specific mismatch.
It does not transform the entire animal into a human biological system.
The Best Translational Research Builds a Chain of Evidence
A strong peptide-development program does not ask one mouse experiment to do everything.
It may proceed something like this:
Human receptor assays establish molecular pharmacology.
Human cell models test signaling.
In vitro stability experiments identify degradation.
Relevant animals provide PK, distribution, and toxicity information.
Human-derived organoids investigate tissue-specific effects.
Computational models connect exposure across species.
Early clinical trials finally test the predictions directly in humans.
If several different systems point in the same direction, confidence increases.
That approach is stronger than declaring a peptide effective because one mouse study produced a statistically significant result.
How to Read an Animal Peptide Study
The most useful question is not:
Did it work in mice?
Instead ask:
What species and strain were used?
Does the peptide bind that species’ receptor similarly to the human receptor?
Was human-relevant exposure measured?
Was the disease model appropriate?
Were animals randomized?
Was outcome assessment blinded?
Were both sexes studied or was the choice justified?
Was the experiment replicated?
Did researchers measure pharmacokinetics?
And, most importantly:
Which part of the result actually needs to translate to humans?
That last question prevents a large amount of overinterpretation.
Animal Evidence Is a Step, Not the Finish Line
Animal studies have helped researchers understand peptide biology and develop important medicines.
But they were never meant to replace human evidence.
Their role is to reduce uncertainty.
A strong animal study might tell us that a peptide:
reaches the target organ,
produces the expected pharmacological response,
has an acceptable preliminary safety profile,
and deserves further development.
That is valuable information.
It is also different from demonstrating that the peptide is safe and effective in patients.
Human clinical trials exist because eventually there is only one reliable way to determine how a new therapeutic peptide behaves in humans:
study it carefully in humans.
FAQs
If a peptide works in mice, does that mean it will work in humans?
No. Mouse results provide preclinical evidence. Differences in receptor biology, metabolism, pharmacokinetics, immune function, and disease biology can change the human result.
Are animal studies unreliable?
That is too broad. Animal studies can provide valuable and reproducible information when the model and design fit the scientific question. Their limitation is that no animal perfectly reproduces human biology.
Why are mice used so often?
Mice are relatively well characterized genetically and biologically, have short generation times, and can be studied using numerous established disease models and genetic tools.
Can researchers calculate a human dose from a mouse dose?
Animal data can contribute to human dose selection, but simple milligram-per-kilogram conversion is generally inadequate. Modern dose selection considers exposure, pharmacokinetics, pharmacodynamics, receptor biology, toxicity, and other factors.
What does the 5% animal-to-human approval figure mean?
A 2024 umbrella review found that about 5% of the animal-tested interventions it evaluated ultimately received regulatory approval. That does not mean animal findings were incorrect 95% of the time. Many additional clinical, commercial, manufacturing, and regulatory factors affect whether a therapy is approved.
Are organoids replacing animal research?
They are increasingly complementing and, in some contexts, replacing portions of animal testing. FDA is actively developing frameworks for using validated human-relevant New Approach Methodologies in drug development.
References
- Ineichen BV, Furrer E, Grüninger SL, Zürrer WE, Macleod MR. Analysis of animal-to-human translation shows that only 5% of animal-tested therapeutic interventions obtain regulatory approval for human applications. PLOS Biology. 2024. Read the full PLOS Biology study
- U.S. Food and Drug Administration. Step 2: Preclinical Research. FDA preclinical research overview
- U.S. Food and Drug Administration. General Considerations for the Use of New Approach Methodologies in Drug Development. Draft Guidance, March 2026. FDA NAMs draft guidance
- U.S. Food and Drug Administration. FDA Achieves Year 1 Goals in Reducing Animal Testing in Drug Development. April 2026. FDA 2026 animal-testing update
- Percie du Sert N, et al. The ARRIVE Guidelines 2.0: Updated Guidelines for Reporting Animal Research. PLOS Biology. 2020. ARRIVE Guidelines 2.0
- Nair A, Morsy MA, Jacob S. Dose translation between laboratory animals and human in preclinical and clinical phases of drug development. Drug Development Research. 2018. PubMed record
- Martignoni M, Groothuis GMM, de Kanter R. Species differences between mouse, rat, dog, monkey and human CYP-mediated drug metabolism, inhibition and induction. Expert Opinion on Drug Metabolism & Toxicology. 2006. PubMed record
- U.S. Food and Drug Administration. CDER Streamlined Nonclinical Studies and Acceptable New Approach Methodologies. FDA CDER NAMs resource