How Regulators Decide What Counts As A Surrogate Endpoint

Some treatments are approved on the basis of a laboratory measurement rather than a demonstrated effect on how patients fare. The reasoning behind that substitution has explicit conditions.

The substitution saves time

Outcomes that matter directly, including survival and disease progression, often take years to accumulate in sufficient numbers to compare groups reliably.

A marker that changes within weeks allows a trial to conclude far sooner, which matters most where the condition is serious and options are limited.

The trade is between speed and certainty. A shorter trial answers a different question, and the value depends on how tightly the marker tracks the outcome.

Correlation is insufficient

A marker correlating with outcomes shows that both reflect disease severity. It does not show that changing the marker changes the outcome.

Validating a surrogate requires evidence that the treatment's effect on the outcome is captured by its effect on the marker, across multiple treatments.

This is a demanding standard, and relatively few markers have met it convincingly. Many in use are accepted on weaker grounds, described as reasonably likely to predict benefit rather than established.

Failures have been instructive

Several historical examples exist where a treatment improved a marker while worsening outcomes, which is the failure mode the validation requirement exists to prevent.

These cases typically involved treatments affecting the marker through a route that also carried harm not reflected in the measurement.

The lesson drawn was that a marker can be a reliable indicator of disease while being an unreliable indicator of treatment effect, and those are separate properties.

Conditional approval attaches obligations

Regulators may approve on a surrogate while requiring confirmatory trials measuring the outcome directly, making availability conditional on later evidence.

Completing those trials has proved difficult in practice, partly because patients are reluctant to enter a trial for a treatment already available.

Withdrawal when confirmatory evidence does not materialise is possible but administratively slow, which weakens the mechanism intended to balance early access.

Reading approvals requires the distinction

An approval based on a surrogate means the treatment changes a measurement, which is a narrower statement than that it helps patients.

Regulatory documents state which basis was used, though summaries reported more widely frequently omit it.

Decisions about any specific treatment rest with a treating clinician who can weigh the evidence base against an individual's circumstances and alternatives.