Translation failure is normal, not a scandal
Why most promising laboratory findings do not become treatments, what that says about a field, and how to read early results without cynicism.
Most candidate treatments that look promising in the laboratory do not become treatments. This is the normal shape of therapeutic development across all of medicine, and it reflects the difficulty of the problem rather than incompetence or bad faith.
Understanding this changes how early results should be read. A promising in vitro finding is the beginning of a process with a high attrition rate, not a preview of a product. Presenting it as the latter is where the misrepresentation happens, and it usually happens in marketing rather than in the research itself.
The shape of the pipeline
Therapeutic development narrows at every stage. Many candidates show activity in cell systems. Fewer show effects in animals. Fewer still show benefit in early human studies. Fewer again survive larger trials. The drop at each stage is substantial, and the pattern is consistent across disease areas and decades.
We are not going to attach percentages to those stages, because the figures vary by field and by how attrition is counted, and a number quoted without its source is exactly the practice this publication avoids. The shape is what matters and the shape is not disputed.
Why so much is lost
Model limitations
Cell culture and animal models are simplifications chosen to be tractable. Every simplification removes something, and sometimes the removed thing is what determines the outcome in a person.
Dose and delivery
Effects in culture occur at concentrations the experimenter chooses. Achieving comparable exposure in tissue may be impossible. For preparations applied to skin, this is the central issue, as set out in can a vesicle cross the skin barrier.
Redundancy
Biological systems have overlapping mechanisms. Pushing one input often produces less change than expected because compensation absorbs it. This is a good property of organisms and an inconvenient one for therapeutics.
Endpoint mismatch
A molecular change may be real and may not produce anything a person notices. Discussed in endpoints that mean something.
Regression of early estimates
Early positive results are systematically inflated, for the statistical reasons in sample size and what it buys you. Some apparent failure at later stages is the correction of an estimate that was never right.
A promising laboratory finding indicates that a clinical treatment will follow.
- Proposed mechanism
- Demonstrated biological activity leads to therapeutic application.
- What has been shown
- Attrition through the stages of therapeutic development is high and consistent across fields and decades. Most candidates showing early activity do not reach clinical use. This is a well recognised feature of medical research rather than a property of any particular field.
- Highest level reached
- Larger human studies
- Main confounders
- Attrition rates are measured differently across sources. Fields differ. Selection of which candidates advance is itself informative and not random.
GradeSUPPORTED, CONSISTENT
What would change thisLittle. The general pattern is well established. For any specific candidate, what would change the assessment is the candidate completing the stages, which is the point of the process.
What this means for reading this field
Extracellular vesicle biology is a legitimate and active research area. Cells release vesicles, the vesicles carry material, recipient cells take them up, and the consequences are worth studying. None of that is in question here and we would not want a reader to leave with the impression that we think it is.
The claim we examine repeatedly is narrower: that a specific commercial preparation, applied by a specific route, produces a specific outcome in human skin. That claim sits several stages along a pipeline where most candidates are lost, and the evidence supporting it is generally at the early stages of that pipeline.
Saying so is not scepticism about the biology. It is placing a claim correctly on a path that everyone in research understands and that marketing tends to elide.
| Stage reached | What is reasonable to say | What is not |
|---|---|---|
| Cell culture activity | The material can affect these cells under these conditions | It works |
| Animal model effect | It produces an effect in a whole organism of another species | It will work in people |
| Small human study | Early human data exist and are unreplicated | Clinically proven |
| Replicated human trials | More than one group finds the same direction | Guaranteed for any individual |
Why this framing helps rather than deflates
Two failure modes exist in reading this field. One is credulity: treating mechanism as outcome. The other is cynicism: treating an unproven claim as a lie and a young field as a fraud. Both are wrong and the second is lazier.
The accurate position is that this is early-stage science being sold at product prices, which is a commercial problem rather than a scientific one. The research is proceeding normally. The gap is between what the research supports and what the market claims, and locating the problem there keeps both scepticism and interest proportionate.
The publication's position
We expect some claims we currently grade at the earlier levels to move up, and some to move down or to be quietly abandoned. That is what a normal field looks like from inside. Our commitment is that each panel names in advance what would change our assessment, so that when we move a grade it can be checked against what we said we were waiting for. That is the point of the format, and it is the only defence a publication has against updating for convenience.
Questions readers ask
Why do most laboratory findings not become treatments?
Because models simplify, achievable doses in tissue differ from those used in culture, biological systems compensate for pushed inputs, molecular changes may not produce noticeable outcomes, and early effect estimates are systematically inflated.
Does high attrition mean the research is bad?
No. Attrition is consistent across fields and decades and reflects the difficulty of the problem. A field where most candidates fail in translation is a normal field.
Does this mean exosome treatments do not work?
No. It means the claim that a specific product applied a specific way produces a specific outcome sits early on a pipeline with high attrition, and should be read as an early-stage claim rather than an established one.
Is it fair to be sceptical of a young field?
Scepticism about specific claims is appropriate; cynicism about the field is not. The research is proceeding normally. The gap is between what the research supports and what the market asserts.
What would move a claim up your scale?
Independent replication in adequately powered human studies with appropriate controls and pre-specified endpoints. Every evidence panel on this site names the specific study that would change that assessment.