In vitro, in vivo, in human
What each level of evidence can establish, why the gaps between them are where most claims fail, and how to place a study on the ladder.
In vitro means in a dish: cells or molecules outside a living organism. In vivo means in a living organism, usually an animal model. In human means in people, which itself ranges from a handful of participants to a large randomised trial.
Each level answers a different question and the gaps between them are where most candidate treatments fail. A result at one level is not a weaker version of a result at the next; it is an answer to a different question. Reading a cell culture finding as a preview of a clinical result is the commonest error in this field and the one most rewarded by marketing.
What each level actually asks
In vitro
Cells in a dish, in a controlled medium, at a concentration the experimenter chose. The question is whether the material can affect these cells under these conditions. It is a question about capability, not about what happens in a body.
What makes in vitro work valuable is control: you can vary one thing and observe the result. What makes it limited is the same thing. A monolayer of fibroblasts has no matrix around it, no mechanical loading, no immune cells, no blood supply, no adjacent tissue types, and no clearance. It is also exposed to whatever concentration the experimenter applies, which may exceed anything achievable in tissue by a large factor.
In vivo, in animals
A living organism restores the missing context: circulation, immune response, clearance, mechanical environment. The question becomes whether the material does something in a whole system.
The limits are species differences and model artificiality. Rodent skin differs structurally from human skin in ways that matter for wound healing, and most animal wound models involve acute, clean injury in young healthy animals, which is not the situation being modelled in aesthetics.
In human
The only level that answers whether a treatment works in people. It ranges enormously, from a case series of a few individuals with photographic assessment to a randomised controlled trial with blinded assessment and objective endpoints. Human is a level, not a standard, and the range within it is wider than the gap between the levels.
A demonstrated effect on cells in culture predicts a clinical effect in human skin.
- Proposed mechanism
- If a preparation changes cell behaviour in vitro, it will change the same cells in tissue.
- What has been shown
- Culture systems reliably demonstrate what a preparation can do to cells under chosen conditions. The general pattern across therapeutic development is that most candidates showing activity in early systems do not show clinical benefit later, which is why development pipelines have the shape they do. That pattern is well recognised across medicine and is not specific to this field.
- Highest level reached
- In vitro only
- Main confounders
- Concentration in culture is chosen and often far exceeds what tissue could encounter. Culture lacks matrix, immune context, clearance and mechanical environment. Endpoints in culture are molecular rather than clinical.
GradeNOT SUPPORTED
What would change thisNothing at the level of the general claim. Each specific candidate has to be tested at the next level. The value of in vitro work is in generating and refining hypotheses, which is a real and necessary function, not in predicting outcomes.
Why the gaps are where things fail
Moving up a level adds variables the previous level excluded, and each added variable is an opportunity for the effect to disappear. Delivery may fail. Clearance may remove the material before it acts. The dose achievable in tissue may be far below the dose used in culture. Compensatory mechanisms may absorb the change. The endpoint that mattered in the dish may not translate into anything visible.
This is not pessimism, it is the ordinary structure of therapeutic development, and it is the reason regulators require human evidence rather than accepting mechanism. A field in which most candidates fail in translation is a normal field. We develop that point in translation failure is normal.
| Study type | Level | What it can establish |
|---|---|---|
| Cells in monolayer culture | In vitro | Capability under chosen conditions |
| Three-dimensional skin equivalent | In vitro, more complex | Capability with some tissue architecture restored |
| Ex vivo human skin explant | Between levels | Behaviour in real tissue without circulation or immune response |
| Rodent wound model | In vivo, animal | Effect in a whole organism, with species differences |
| Case series in patients | In human, uncontrolled | That something was observed, with no comparison |
| Randomised controlled trial | In human, controlled | Whether the treatment causes the outcome |
The ex vivo middle ground
Human skin explants, meaning real tissue maintained outside the body, sit between levels. They retain architecture and cell diversity, which is a substantial advantage over monolayer culture, and they lack circulation, immune trafficking and clearance, which matters for anything depending on those.
Explant work is genuinely more informative than monolayer culture for questions about penetration and tissue-level response. It is also frequently presented as though it were clinical evidence, which it is not: no person experienced the treatment and no clinical outcome was measured.
How to place a study in one minute
Read the methods, not the abstract. Three questions place almost any study.
- What was treated? Cells, tissue, an animal, or people.
- What was measured? A molecular readout, a tissue measurement, or a clinical outcome.
- Compared with what? Nothing, an untreated group, a vehicle, or an active comparator.
Those three answers determine what the study can support, and they can usually be extracted in a minute. Everything else, including the discussion section, is interpretation, and the discussion section of a paper is where authors are most free to speculate.
Why we grade by level explicitly
Our evidence panels carry a level field for exactly this reason. Saying that a claim has reached in vitro only, or animal models, or small human studies, is more informative than saying the evidence is limited, because it tells you which gap has not yet been crossed and therefore what would be needed next.
It also makes our own position falsifiable. If we grade a claim at in vitro only and a competent human trial appears, we are obliged to move it. That is the intended discipline of the format, and the fixed vocabulary is published at the grading vocabulary so that our use of it can be checked against itself.
Questions readers ask
What does in vitro mean?
Literally in glass. It refers to experiments on cells, tissue fragments or molecules outside a living organism, under conditions set by the experimenter, including the concentration applied.
Why do results in cells not carry over to people?
Because moving to a living system adds variables the dish excluded: delivery, clearance, immune response, mechanical environment and achievable dose. Each of those is an opportunity for an effect to disappear.
Are animal studies good evidence?
They answer a different question from cell culture and a different question again from human studies. Species differences and the artificiality of most injury models limit how far animal results predict human outcomes.
What is an ex vivo skin study?
A study on real human skin maintained outside the body. It retains tissue architecture, which is a real advantage over monolayer culture, and lacks circulation, immune trafficking and clearance. It is not clinical evidence.
What does clinically proven mean?
In practice it is used at every level including cell culture, so it means very little without a reference. The useful response is to ask which study, at which level, in which system, and to read the methods.