Biomarkers vs Clinical Outcomes
The difference between a biomarker, a surrogate endpoint and a clinical outcome, why a lab number can improve while health does not, and how to read claims, with no universal winner.
Content last updated . This page has not been medically reviewed.
Two kinds of result
A study can report how a lab number changed or how people felt, functioned or survived. These are different questions. This page compares them as information only. Labels used: established, promising but limited, evidence of harm, marketing claim (see how we label evidence).Definitions
| Term | Plain meaning |
|---|---|
| Biomarker | A defined characteristic that is objectively measured as an indicator of normal biological processes, disease processes or responses to an intervention |
| Clinical outcome | How a person feels, functions or how long they live |
| Surrogate endpoint | A biomarker used in a trial as a stand-in for a clinical outcome |
These definitions follow the FDA, which says clinical outcomes directly measure whether people in a trial feel or function better, or live longer. [1] The FDA also says a surrogate endpoint does not measure clinical benefit itself but is expected to predict it. [1]
When a biomarker can stand in
The FDA says surrogate endpoints are sometimes used because outcomes such as strokes may take a very long time to study. [1] Before one is accepted, extensive evidence must accumulate, usually from clinical trials showing it predicts clinical benefit in a particular context. [1] Blood pressure and stroke is the FDA's example: many trials showed that lowering systolic blood pressure reduced stroke risk. [1] That is established for that use, within a stated context of use.The FDA describes three levels: candidate, reasonably likely and validated surrogate endpoints. [1] A biomarker named in marketing is worth checking against these levels.
When the number improves but people do not
The ACCORD trial randomized 10,251 adults with type 2 diabetes to intensive therapy aimed at a glycated hemoglobin below 6.0% or standard therapy. Intensive therapy reached lower levels (6.4% versus 7.5% at one year), yet intensive therapy was stopped early because of higher mortality (257 versus 203 deaths; hazard ratio 1.22, 95% confidence interval 1.01 to 1.46). [2] The main combined cardiovascular outcome was not significantly different. [2] This is one patient group and one strategy, so it does not mean lowering glucose is harmful in general. It shows why the surrogate needs testing in the intervention being used.Why this matters for emerging science
Many new technologies report biomarkers: a gene expression pattern, an inflammation marker, a brain-scan signal, a biological-age score. A change there is a lab or imaging result, not proof of benefit. See biomarker changes vs meaningful health outcomes, surrogate vs patient-important outcomes and biomarkers and biological age tests. For how a test's accuracy differs from its usefulness, see clinical validity and clinical utility.Questions to ask of a claim
- Was the outcome how people felt or lived, or a marker?
- Has this marker been shown to predict benefit for this specific intervention?
- Was the trial large and long enough to see harms?
- Does the report mention side effects and deaths, not only the marker?
No universal winner
Clinical outcomes answer the question people care about most but can take years. Biomarkers are faster and often essential, but only as reliable as the evidence linking them to outcomes. Both have a role; neither is better in every setting.Frequently asked questions
What is a surrogate endpoint?
A trial endpoint used as a substitute for a direct measure of how a patient feels, functions or survives. [1]
Does a better blood test mean better health?
Not necessarily. In ACCORD, lower glycated hemoglobin with intensive therapy came with higher mortality. [2]
Does the FDA ever accept surrogate endpoints?
Yes. Validated surrogates are accepted as evidence of benefit, and the FDA says it approved 45 percent of new drugs between 2010 and 2012 based on a surrogate. [1]
What does "reasonably likely" mean?
Supported by strong rationale but not enough clinical data to be validated, and used for accelerated approval. [1]