Biomarkers and Surrogate Endpoints in Clinical Trials, Explained
In clinical trials, we cannot always wait to see if patients live longer or feel better. Biomarkers and surrogate endpoints let researchers measure success earlier, using biological signals instead of clinical outcomes. They speed up drug development and have saved millions of lives. They have also, at times, caused real harm. This article explains what they are, when they work, and when they fail.
Biomarkers: a quick taxonomy
A biomarker is a defined characteristic measured as an indicator of normal biological processes, disease processes, or response to treatment. The FDA's BEST resource, the agency's official glossary of biomarkers, endpoints, and other tools, uses exactly that framing.
Biomarkers come in several flavors:
- Diagnostic biomarkers confirm that a disease is present.
- Prognostic biomarkers predict how a disease will progress.
- Predictive biomarkers predict whether a patient will respond to a treatment.
- Pharmacodynamic biomarkers show that a drug hit its intended target.
- Safety biomarkers flag toxicity early.
What makes a biomarker a surrogate endpoint
Not every biomarker qualifies as a surrogate endpoint. A surrogate endpoint is a biomarker intended to substitute for a clinical endpoint, the direct measure of how a patient feels, functions, or survives. A biomarker only becomes a surrogate when there is strong evidence that changes in it reliably predict changes in the clinical outcome patients actually care about.
That evidence standard matters because surrogates carry real regulatory weight. The FDA's accelerated approval pathway, created in 1992, lets drugs be approved on surrogate endpoints that are reasonably likely to predict clinical benefit, with confirmatory trials required afterward.
Why surrogates are appealing — and risky
Three reasons drive their use. Speed: instead of waiting five years for heart attacks, a trial can measure cholesterol at six months. Size: continuous surrogate measurements need hundreds of patients, while clinical events need thousands. Feasibility: some outcomes are too rare or too distant to observe at all.
The risk is that the surrogate can lie. A drug that lowers cholesterol might still cause heart attacks through other mechanisms. The surrogate captures one thread of the drug's effects; the body has many others.
Case studies: when surrogates worked
Blood pressure is the classic validated surrogate for stroke risk. Decades of data show a consistent relationship between blood pressure reduction and cardiovascular risk reduction, and the FDA allows antihypertensives to be approved on blood pressure reduction alone. It is listed among the surrogate endpoints that have supported drug approvals.
The HIV story is bigger still. In the 1990s, the FDA accepted CD4 cell count and viral load as surrogates for accelerated approval of antiretroviral drugs. Treatments reached patients years earlier than they otherwise would have, and later trials confirmed the surrogates tracked real survival gains. HbA1c works the same way in diabetes: it reflects average blood sugar over two to three months, landmark trials tied reductions in it to fewer complications, and it is accepted as a primary endpoint for diabetes drug approval.
When surrogates lied
The failures are instructive. In the CAST trial, drugs that suppressed arrhythmias after a heart attack actually increased mortality. The surrogate of "fewer arrhythmias" looked good on paper and made patients worse. Vioxx reduced pain and gastrointestinal bleeding while increasing heart attacks and strokes, with an estimated 88,000 excess cardiovascular events in the US (the video cites this figure). The common thread: drugs have multiple effects, and a surrogate only captures some of them.
More recently, during the COVID-19 pandemic, many trials used reduction in viral RNA as a surrogate for clinical improvement. The correlation with real clinical benefit turned out weaker than expected. Pandemics create pressure to accept surrogates fast, but validating them takes time that pandemics do not give.
Oncology shows the mixed record in one number: the video reports that roughly 65% of FDA oncology approvals between 2020 and 2023 rested on surrogate endpoints rather than overall survival. For some cancers, tumor shrinkage strongly predicts survival. For others, drugs that shrink tumors do not extend life.
How the FDA qualifies biomarkers
The FDA runs a formal biomarker qualification program to evaluate biomarkers for regulatory use. Qualification demands two things: analytical validation, can the biomarker be measured reliably, and clinical validation, does it predict the outcome it claims to. Qualified biomarkers include serum creatinine for kidney injury and troponin for cardiac injury. The program accepts submissions from industry, academia, and patient groups.
Researchers also lean on statistical discipline. Prentice's criteria for surrogate validation require that the treatment affects the surrogate, the surrogate affects the clinical outcome, and the treatment's effect on the clinical outcome is fully captured by the surrogate. The third criterion is almost never met, which is why the strongest validation comes from meta-analyses across many trials where both the surrogate and the real outcome were measured.
The future
New tools are arriving. Liquid biopsies detect cancer DNA in a simple blood test, catching recurrence months before scans. Digital biomarkers from wearables measure gait, heart rate variability, and sleep. Multi-omics panels combine genomics, proteomics, and metabolomics into composite signatures. Each one will face the same question: does it predict what patients actually experience?
The bottom line
Validated surrogates like blood pressure, HbA1c, and CD4 count have saved millions of lives. Unvalidated ones have harmed patients. Surrogate endpoints are not shortcuts around evidence. Used wisely, they accelerate medicine. Used carelessly, they cause harm.
This article is based on the Clarity Clinical Solutions video "Biomarkers and Surrogate Endpoints in Clinical Trials Explained." Watch it here: Biomarkers and Surrogate Endpoints in Clinical Trials Explained
References
- Clarity Clinical Solutions — "Biomarkers and Surrogate Endpoints in Clinical Trials Explained" (video). https://www.youtube.com/watch?v=gp0wD5K-fhg
- FDA/NIH BEST Resource — official definitions of biomarkers, clinical endpoints, and surrogate endpoints. https://www.ncbi.nlm.nih.gov/books/NBK326791/
- FDA Biomarker Qualification Program — how biomarkers are formally qualified for drug development. https://www.fda.gov/drugs/drug-development-tool-qualification-programs/biomarker-qualification-program
- FDA table of surrogate endpoints — surrogate endpoints that have supported drug approval or licensure. https://www.fda.gov/drugs/development-resources/table-surrogate-endpoints-were-basis-drug-approval-or-licensure
- FDA accelerated approval — the pathway that allows approval based on surrogates with confirmatory trials. https://www.fda.gov/patients/fast-track-breakthrough-therapy-accelerated-approval-priority-review/accelerated-approval