Surrogate Endpoint

A surrogate endpoint substitutes a measurable intermediate for the outcome anyone actually cares about: a laboratory value, an image, a physiological sign. Using one validly requires more than correlation with the outcome. The marker must lie on the causal path from treatment to benefit and capture essentially the whole treatment effect, so that no separate pathway bypasses it. Very few markers have been shown to meet that condition.

Glycated haemoglobin, LDL cholesterol, blood pressure, bone mineral density and viral load are the accepted examples, each validated by drugs that moved the marker and the outcome together. The failures are instructive. In the CAST trial, antiarrhythmics that suppressed ventricular ectopy after myocardial infarction increased mortality. Torcetrapib raised HDL cholesterol substantially and increased deaths, ending the programme in 2006. The accelerated approval pathway is built entirely on surrogates judged reasonably likely to predict benefit, with confirmation deferred.

Surrogates buy speed and smaller trials. The distinction they force is between demonstrated pharmacology and demonstrated benefit: a drug that moves the marker has proved it does something, and a drug that moves a hard endpoint has proved that something is worth having. Labels, guidelines and reimbursement decisions track the second, not the first.

This is the most productive source of overstatement in the peptide literature. A rise in IGF-1, a fall in an inflammatory cytokine, a change in a collagen assay or a nitrogen balance shift are all mechanistic readouts presented as though they were clinical results. Surrogate validity is drug-specific rather than marker-specific, so a marker that worked for one class carries no warranty for a compound acting through another pathway.