Preclinical or Human: The Evidence Distinction That Decides What You Can Say

The distinction that decides what you are allowed to say

There is one line running underneath every compliance problem in research peptide ecommerce, and most operators have never had it stated plainly.

Evidence about a rodent, a cell line or an isolated tissue is evidence about a rodent, a cell line or an isolated tissue. Evidence about people is evidence about people. The first does not become the second by being careful, by being peer-reviewed, or by being summarised in a systematic review.

Once you can see that line, most of the category's regulatory difficulty resolves into a single question you can ask of any sentence before publishing it: is this a statement about the study, or about the reader?

Why the distinction is not academic pedantry

The distinction is tempting to regard as a technicality that lawyers care about and scientists find obvious. It is neither.

Biology does not translate cleanly across species. Dose scaling, metabolism, receptor distribution, immune response and route of exposure all differ. A compound producing a striking effect in a rodent injury model may produce nothing in people, or something different, or something the model was not designed to detect. The history of drug development is substantially a history of preclinical results that did not replicate in humans, which is why human trials exist at all rather than being a formality after the animal work.

So the distinction is not a legal fiction layered on top of good science. It is the reason the regulatory framework is shaped the way it is.

The evidence tiers, in practical order

Not all evidence is preclinical or human. The useful working hierarchy for reading a peptide literature set runs roughly like this.

In vitro. Cells, isolated tissue, biochemical assays. Establishes mechanism and interaction. Says nothing directly about a whole organism.

Animal in vivo. A living organism with intact physiology, but not a human one. Establishes that an effect can occur in a system, and generates hypotheses.

Human observational. People, but without controlled assignment. Establishes association and generates hypotheses. Confounding is the standing problem.

Human controlled trial. Randomised, controlled, with a defined outcome measure. The tier the substantiation standard actually asks for.

Systematic review or meta-analysis. A synthesis, whose evidence class is inherited from its inputs. A rigorous review of animal studies remains animal evidence.

That last point is the one most often missed. A review sits at the top of the familiar evidence pyramid diagram, and readers carry that image over to a literature set where the underlying studies are preclinical. The pyramid assumes the base is human trials. If it is not, the top of the pyramid is not where the diagram implies.

What the substantiation standard actually asks for

The FTC's Health Products Compliance Guidance states that claims about the benefits of health-related products require substantiation in the form of competent and reliable scientific evidence, and that, as a general matter, substantiation of health-related benefits will need to be in the form of randomized, controlled human clinical testing to meet that standard.

Read against the tiers above, that sentence draws the line at the fourth level.

The consequence for a research peptide catalogue is direct. Where the available literature is in vitro and animal work, benefit claims about people are unsubstantiated no matter how many papers are cited, because the papers are not about people. Citation density is not evidence class.

The sentence test

Here is the practical instrument. Take any sentence you are about to publish and ask what its subject is.

About the study. In a rat model, accelerated tendon-to-bone healing was observed by the authors. Subject: the authors, the model, the observation. This is reporting.

About the compound in general. The compound accelerates tendon healing. Subject: the compound, unqualified. The model has silently disappeared and the statement now spans every organism.

About the reader. A second-person construction promising a personal result, of the kind that pairs "you" with an outcome noun. Subject: the reader. This is a benefit claim requiring human evidence, and it also establishes intended use.

The first is defensible. The second is a claim about people wearing the clothes of a study description, and it is where most catalogue copy actually sits. The third is unambiguous.

Notice how little wording separates them, and that the compliance difference is total.

The intended use consequence

The evidence question and the regulatory status question are connected, and it is worth being precise about how.

The Code of Federal Regulations (CFR) provision 21 CFR 201.128 defines intended use by the objective intent of the persons responsible for labeling, shown by labeling claims, advertising matter, or written or oral statements. A sentence asserting a benefit for a person is exactly such a statement.

So a sentence that drops the model system does two things at once. It becomes an unsubstantiated claim under the advertising standard, and it contributes to establishing a therapeutic intended use under the drug framework. One editorial slip, two separate exposures, which is covered further in Research Use Only and the FTC track.

How to carry the model system without ruining the prose

The most common objection to this discipline is that it makes copy unreadable. That objection is usually a symptom of trying to keep the claim while adding a qualifier.

Qualifying a benefit claim is awkward because the sentence was built to assert something it cannot support. Reporting a finding is not awkward, because the model is part of what makes the finding interesting.

• Lead with the study, not the compound. "A 2019 paper in Cell Tissue Res examined accelerated healing in a rat model" reads naturally.

• Name the endpoint. Specific measurements are more persuasive than vague outcomes, and they carry their own context.

• Let the reader draw their own inference. You are not required to tell them what it means. You are required not to tell them something the evidence does not support.

• Link the primary source. A PubMed Identifier (PMID) is the strongest possible signal of confidence and costs nothing.

Copy written this way is frequently better than the alternative, because it is specific. Vague outcome language is not persuasive, it is merely familiar.

What this means for a catalogue in practice

Three operational consequences follow.

Product descriptions become descriptions. Compound, structure, indexed literature with model systems, specification, batch documentation. Every element checkable.

Marketing claims need a tier check before a rule check. Before asking whether a sentence is prohibited, ask what evidence tier supports it. Most sentences fail here first, which makes it the cheaper check to run.

Literature pages carry more weight than benefit pages. A page that accurately reports what eight papers examined is more useful to a serious buyer than a page asserting outcomes, and it is defensible. Worked through for one compound in what the BPC-157 research actually shows.

What this article does not establish

Nothing here describes what any compound does in a person, and nothing here is guidance on use. It describes how to classify evidence and how that classification governs what may be published.

Compounds referenced anywhere on this site are supplied for laboratory research purposes only, and this is not legal or medical advice.

The check that takes ten seconds

If everything above compresses to one habit, it is this.

Before publishing any sentence about a compound, read it and name its grammatical subject. If the subject is a study, an author or a model, you are reporting. If the subject is the compound with no model attached, or the reader, you have moved into claim territory and the evidence tier has to support it.

That is the whole instrument. It requires no legal training, it takes a few seconds, and it catches the failure earlier and more reliably than a prohibited word list does, because it operates on structure rather than vocabulary.

A word list can always be evaded by synonym. A sentence whose subject is the reader cannot be rescued by word choice, and one whose subject is the study does not need to be.

Frequently asked questions

Does a systematic review count as human evidence?

Only if its inputs are human studies. A review inherits the evidence class of what it reviews, and a rigorous synthesis of animal work remains animal evidence.

Can I cite an animal study at all?

Yes. Report what it observed with the model system stated. The problem is never the citation, it is dropping the context that makes the citation accurate.

What if there is one small human trial?

Then there is human evidence, and its size, design and outcome measures determine what it supports. One small trial does not support a broad claim, and the substantiation question is always about the claim as worded.

Is mechanism evidence useless commercially?

No. It is genuinely interesting to a technical buyer and it is defensible when reported as mechanism. It is only a problem when it is presented as an outcome.

Where does this leave a catalogue with only preclinical literature?

Reporting the literature accurately, being specific about analytical documentation, and not asserting benefits for people. That is a viable and honest position, and it is more credible to a serious buyer than the alternative.

Where this fits

The compound-level worked example is in what the BPC-157 research actually shows, the advertising standard in the FTC track, and the page-level application in anatomy of a compliant peptide product page.

All compounds referenced anywhere on this site are supplied strictly for laboratory research purposes only. Nothing here is for human consumption, and nothing here is intended to diagnose, treat, cure, or prevent any disease.