There are problems with using such surrogate outcomes. They are often only tenuously associated with the real disease, in a very abstract theoretical model, and often developed in the very idealised world of an experimental animal, genetically inbred, kept under conditions of tight physiological control.