In stroke research, the drop is easy to measure. A 2020 systematic comparison of stroke studies gathered 209 animal studies, 75 early clinical trials and 50 phase 3 trials of neuroprotective agents, compounds meant to protect brain tissue after a stroke. In animals, 69% of studies were positive. In early human trials, 32% were. In the large phase 3 trials, just 6% were.
The same ideas, tested at each step, looked weaker the closer they came to patients. That pattern sits at the heart of preclinical vs clinical research: lab and animal work generates promising leads, and human trials test whether those leads hold.
This article explains what preclinical research is, how clinical trial phases work, why so many findings fall away between them, and where research materials fit.
Quick answer: what is the difference between preclinical and clinical research?
Preclinical research happens before any human testing: lab experiments on cells (in vitro) and studies in animals (in vivo). Clinical research tests in people, through trials that move from small phase 1 studies to large phase 3 trials. Preclinical work generates leads; clinical trials test whether they hold in humans.
What Is Preclinical Research?
Preclinical research is everything that happens before a compound is tested in people. It usually starts in vitro, with cells, tissues or enzymes in a dish, then moves in vivo, into living animals such as mice, rats or dogs.
Its job is to answer early questions:
- Does the compound do anything measurable in a biological system?
- How does it behave in a living animal: how long it lasts, where it goes, how it is broken down?
- What harms appear, and under what conditions?
A pharmacokinetic study in rats and dogs is a typical example. Researchers measured how quickly BPC-157 was cleared and found a half-life (the time for half of a substance to be removed) of under 30 minutes in both species. That kind of finding only comes from a living system, and it describes those animals, not people.
Animal studies have their own reporting standard. The ARRIVE guidelines (Animal Research: Reporting of In Vivo Experiments) list what a paper should report so others can judge and repeat the work. The 2020 update defines an “Essential 10” minimum and notes that adherence has been inconsistent.
Cell and animal models each have strengths and blind spots of their own, covered in In Vitro vs In Vivo.
Phases of Clinical Trials Explained
A clinical trial tests an intervention in people according to a pre-specified plan. Trials run in phases, each asking a different question. The US FDA describes them roughly as follows:
The FDA estimates that about 70% of drugs move on from phase 1, about 33% from phase 2 and 25 to 30% from phase 3.
What a phase 1 trial looks like
A first-in-human trial is the first time a compound is tested in people. One 2025 example, sponsored by Novo Nordisk, was randomised, double-blind and placebo-controlled, and enrolled 144 adults. Its primary endpoint, the main outcome a trial is designed to measure, was the number of adverse events. (An endpoint is any outcome a trial measures.)
Trials have a reporting standard too. CONSORT (Consolidated Standards of Reporting Trials) sets out what a randomised trial report should include, and its 2025 version is a 30-item checklist.
Preclinical vs Clinical Research: The Key Differences
The two stages differ in almost every respect, from what is studied to what counts as an answer:
The differences matter when reading claims. A preclinical result tells you a compound did something in a model, under the conditions of that experiment. A clinical result tells you what happened in a defined group of people.
The first can justify the second; it cannot replace it. A compound can look strong in every animal study and still show nothing in a trial, because the trial is asking a different, harder question. Keeping the two stages distinct, and noting which one a claim comes from, is the simplest guard against overreading.
Why Most Preclinical Findings Fail in Clinical Trials
The stroke figures are not unusual. An umbrella review of animal-to-human translation found that about 5% of therapies tested in animals reached regulatory approval. Several forces drive the drop.
Publication bias
Publication bias is the tendency for positive results to be published while negative ones go unreported. In a 2010 meta-research analysis of animal stroke studies, researchers estimated that publication bias might account for around a third of the reported efficacy, and that about 14% of experiments went unpublished. The 2020 stroke comparison also found clear publication bias, and estimated the average statistical power of the animal studies at just 17%.
Species differences
Mice are not small humans. In a 2013 comparison of mouse models with human patients with acute inflammatory conditions, gene responses in the mice correlated poorly with those in people, close to random for many genes. Even basic behaviour varies by species: the BPC-157 half-life above was measured in rats and dogs, and on its own says nothing about any other species.
Different questions
The stroke authors also point to design differences: animal studies and clinical trials often used different primary endpoints and different timing.
Translational Research Explained
Translational research is the work of moving a finding from lab to clinic: deciding which preclinical leads deserve human testing, and designing trials that test them fairly.
A central tool is the biomarker, a measurable sign, such as a level in the blood, that may track a biological process. Biomarkers change faster and are easier to measure than clinical outcomes, which makes them useful. They can also mislead.
When a biomarker moves but the outcome doesn’t
In a 2008 randomised controlled trial of 563 people with Alzheimer’s disease, MK-677, a non-peptide compound that stimulates growth hormone release, raised blood IGF-1 by about 60% at 6 weeks and 73% at 12 months. Yet on every clinical measure of disease progression, there was no significant difference from placebo. The target was engaged; the outcome did not change.
Biomarkers have real value when used carefully. In a meta-research analysis of more than 406,000 clinical trial records, trials that used biomarkers to select participants had higher overall success rates than trials without them. The lesson is to read a biomarker as a signal, not as the outcome itself.
Reading Clinical Claims Carefully
Human trials need careful reading too. Spin is the presentation of results in a more favourable light than the data support.
A real example: in a 2015 phase 2 trial of thymosin beta 4 in 72 people with dry eye, neither primary endpoint showed a significant difference from placebo. The authors still concluded that the study “confirms the efficacy” of the product, relying on secondary endpoints. Secondary endpoints can generate hypotheses, but a trial that misses its primary endpoints has not passed its main test.
Two more checks:
- Sponsor funding. Who paid for and ran the trial? The phase 1 example above was sponsored by the company developing the compound. That does not invalidate a study, but it is context worth knowing.
- The size of the evidence base. For some widely discussed peptides, human data are minimal. A 2026 systematic review found only three small human studies of BPC-157, 30 participants in all, with the certainty of evidence rated very low.
A single study, however striking, rarely settles a question. For a fuller checklist, see How to Evaluate the Quality of a Research Study.
Where Research Materials Fit
Research Use Only (RUO) materials belong at the start of this path: preclinical laboratory work such as in vitro experiments. They are not made or documented for clinical use, and nothing about their quality makes them suitable for it.
At this stage, the quality of the material shapes the quality of the result. If a peptide’s identity is wrong, or its purity and counter-ion are undocumented, an experiment can produce a signal that reflects the material rather than the biology. Given how many preclinical findings already fall away before reaching people, avoidable error at the bench only narrows the path further.
That is why researchers check two things before an experiment begins. The first is identity: confirming the molecule is what the label says. The second is documented quality: purity, net peptide content and counter-ion, recorded for the specific batch in use. Both are explained in What Is Peptide Identity and Why Does It Matter?.
For an example of batch-level documentation, see the OHR COA page. A COA describes research quality only, not suitability for any use outside the lab.
Frequently Asked Questions
What comes before a clinical trial?
Preclinical research: laboratory studies in cells and tissues, then studies in animals, to look for a measurable effect and identify risks. Only after that work does a compound move into a phase 1 trial in people. Animal studies should be reported using guidelines such as ARRIVE, so others can judge and repeat them.
What is a phase 1 trial?
Phase 1 is the first stage of human testing. It typically involves 20 to 100 healthy volunteers or people with the condition, and focuses mainly on safety and how the body handles the compound. When it is the very first time a compound is tested in people, it is called a first-in-human trial.
Why do drugs fail in phase 3?
Phase 3 trials are large enough to expose effects that smaller, earlier studies overstated. In stroke research, 69% of animal studies were positive but only 6% of phase 3 trials, with publication bias, low statistical power and design differences all contributing.
What does “first-in-human” mean?
It describes a trial in which a compound is tested in people for the first time, after preclinical work. These are usually phase 1 trials focused on safety. A 2025 example, sponsored by Novo Nordisk, enrolled 144 adults and used adverse events as its primary endpoint.
Closing
A strong preclinical result is a starting point, not an answer: it earns a compound the right to further testing, nothing more. For researchers sourcing well-documented materials for preclinical laboratory work, explore the OHRÂ research catalogue.
Read Further: How to Evaluate Study Quality: Tools Researchers Use
This content is for informational and educational purposes only. Products referenced are intended for laboratory research use only and are not for human or animal consumption.




