Biomarker

Biomarker is a measurable biological signal used to stand in for something that cannot be observed directly — a drug exposure, a physiological state, or a change in brain function. The term covers a wide range: drug or metabolite concentrations in blood or urine, receptor occupancy measured on a brain scan, EEG power patterns, or enzyme and protein levels tied to a pathway of interest.

Biomarkers occupy a specific role in the evidence hierarchy: they are not outcomes themselves but proxies for outcomes, chosen because the thing they represent — receptor binding, subjective drug effect, recovery from a disorder — is difficult or impossible to measure directly. The relationship between a biomarker and the phenomenon it represents is always an empirical claim, and that claim varies widely in how well it has been tested.

What this design can establish

A well-chosen biomarker can establish that exposure occurred, that a biological system responded, or that a drug reached and engaged its target. Pharmacokinetic biomarkers — plasma drug levels, urinary metabolites — can confirm absorption and persistence. Pharmacodynamic biomarkers can confirm target engagement: a receptor-occupancy scan can show whether a drug bound to its intended site at a given concentration.

When a biomarker has been rigorously validated against the clinical outcome it is meant to predict, it can substitute for that outcome in earlier research phases — making trials faster and less costly. In that validated state, a change in the biomarker can serve as reasonable evidence that an intervention is producing a measurable biological effect.

What it cannot

A biomarker cannot establish that the biological change it captures produced any clinical effect, subjective experience, or harm. The association between biomarker and outcome is an assumption until it is directly tested, and many biomarkers used routinely in research have never been fully validated against the outcomes that matter. A drug can shift a biomarker substantially while producing no meaningful clinical change, or do the reverse.

The inference a reader is most likely to make — that a biomarker improvement means a patient improved, or that a biomarker elevation signals danger — is exactly what a biomarker alone cannot support. Biomarker data and outcome data answer different questions, and conflating them is one of the most common errors in reading research reports.

Absent biomarker data is also common for less-studied substances. When no biomarker studies have been published for a compound, exposure, target engagement, and pharmacodynamic effects remain uncharacterised. That absence is a stated gap in the record, not a silent one — it carries real information about how much is known.