When is a surrogate not a surrogate?

When is a surrogate not a surrogate?
5th September 2016 Sue Carr

Most people involved in drug development have heard of surrogate endpoints and many continue to argue that they are needed to accelerate drug development in situations of high clinical need. However, there is often confusion between a simple correlate and of what a ‘surrogate’ endpoint truly means and the circumstances under which a previously accepted surrogate endpoint can be transferred to a new, but related clinical setting. An example of the former is offered in Epserion’s recent unfortunate experience. While LDL-C lowering is generally accepted as a surrogate endpoint for CV outcome for statin therapies, FDA was apparently unwilling (and not unreasonably from a statistical perspective) to transfer LDL-C surrogacy to new mechanisms of action. Read more.

The onerous depth of the challenge in demonstrating that an intermediate endpoint is a true surrogate for the effect of a treatment intervention on some clinically meaningful clinical endpoint is often underappreciated in drug development, as is the non-transferability of a surrogate to other diseases and new clinical settings. The challenge of evaluating surrogacy for a new endpoint was recently taken up for IgA nephropathy, a slowly progressing autoimmune disease whereby the antibody immunoglobulin A (IgA) lodges in the kidney ultimately leading to end stage renal disease and death. In search of true surrogacy for clinical outcome, protein urea was formally evaluated in a meta-analytic evaluation of a range of randomised controlled trials using the most contemporary of statistical methodologies. The promising results have now been published and can be seen here.