Surrogate endpoint

What is a surrogate endpoint?

A surrogate endpoint is an outcome used in place of a direct measure of clinical benefit, such as survival, improved function or symptom relief. It is expected to predict that benefit but does not measure the benefit itself. Blood pressure may be used as a surrogate for cardiovascular risk, while tumour size may be used as a surrogate for cancer progression in an appropriate disease and treatment context.

Surrogate endpoints can make a trial shorter or more practical when the clinical outcome takes a long time to occur, is uncommon or would require lengthy follow-up. They may also help a study meet regulatory requirements, but their suitability depends on the strength of evidence connecting the surrogate to the outcome that matters to patients.

Why do surrogate endpoints matter in clinical development?

In clinical development, a surrogate endpoint can provide an earlier indication of whether a treatment is working. This may accelerate development decisions and support a regulatory submission before mature evidence on survival, symptoms or long-term complications is available.

The trade-off is uncertainty. A treatment may improve the surrogate without improving the final clinical outcome, particularly if the surrogate captures only one part of the disease or misses adverse effects. Teams must therefore explain why the endpoint is relevant, how well it predicts benefit and what evidence will address remaining uncertainty.

How is a surrogate endpoint selected and validated?

Selection starts with the treatment’s mechanism, the disease pathway, the intended clinical benefit and the feasibility of measuring that benefit directly. The endpoint also needs a clear definition, an appropriate assessment schedule and a measurement method that is accurate and consistent across trial sites.

Validation examines whether changes in the surrogate reliably predict changes in the relevant clinical outcome. Association between a marker and an outcome is not enough on its own: a useful surrogate should also capture the effect of treatment on that outcome. Evidence may come from multiple trials, different interventions and longer-term follow-up. Validation is specific to the disease, patient population, treatment class and intended outcome, so acceptance in one setting does not automatically apply elsewhere.

Are surrogate endpoints accepted by regulators, payers and HTA bodies?

Acceptance varies with the decision being made and the maturity of the evidence. Regulators may accept a well-supported surrogate when direct clinical outcomes are impractical or unavailable at the time of review. They may still expect later evidence to confirm that the anticipated benefit occurs.

Payers and health technology assessment bodies often need to estimate comparative benefit, duration of effect and long-term value. If these estimates depend on a surrogate, they may challenge the relationship with the final outcome, the transferability of validation evidence and the assumptions used to extrapolate beyond the trial. Additional follow-up or real-world data may therefore be needed.

Is every biomarker or intermediate outcome a surrogate endpoint?

No. A biomarker is a measurable biological characteristic, and an intermediate outcome occurs before the final outcome of interest. Either may be proposed as a surrogate, but it becomes useful as a surrogate endpoint only when there is sufficient evidence that treatment effects on it predict treatment effects on a meaningful clinical outcome.

Teams can go wrong by treating biological plausibility, correlation or precedent as proof of validation. They may also describe an endpoint as universally validated when the evidence applies only to a particular indication, disease stage, treatment type or clinical outcome.

What evidence should teams present when using a surrogate endpoint?

Teams should state the direct clinical outcome being predicted, the evidence supporting the relationship and the setting in which that evidence was generated. They should also describe measurement limitations, uncertainty in extrapolation, planned follow-up and how the surrogate result affects benefit-risk or value conclusions.

The endpoint should be incorporated consistently into the trial strategy, statistical analysis and wider evidence generation programme. Good practice distinguishes observed effects on the surrogate from predicted effects on clinical outcomes rather than presenting the two as equivalent.

How does a surrogate endpoint differ from related endpoint terms?

A clinical endpoint directly reflects how a patient feels, functions or survives. A biomarker records a biological characteristic, while an intermediate endpoint occurs earlier in the disease or treatment pathway. A surrogate endpoint is defined by its role: it substitutes for a direct clinical outcome because it is expected to predict that outcome. It should not be confused with the clinical benefit it is intended to represent.