Introduction

This chapter aims to present and discuss some statistical approaches to identifying and validating the relationship between diagnostic markers and their clinical endpoints. Much of the work in this area has been developed in the context of clinical trials, although the principles extend naturally to cohort studies measuring exposure–end-point associations.

Informally, intermediate end-points or surrogates are laboratory measurements, X-ray results or other biological markers which are observed before the clinical end-point of interest occurs and which are useful in diagnosis, prognosis, monitoring patients or in the evaluation of the effects of an exposure or clinical intervention (E/I) to which patients are subjected (see Figure 3.1).

Motivation for the use of intermediate end-points or surrogates

Ideally, when studying the effects of an E/I, the appropriate outcome measure is the true clinical end-point of interest (CE), which is usually survival or the incidence of a clinical event. However, the CE may require invasive tests, be expensive, occur rarely or be temporally distant from the E/I. For example, in trials of cholesterol-lowering drugs, the fundamental end-point of interest may be death or development of coronary artery disease (CAD). Such a study may require the recruitment of thousands of patients followed up for many years. Clinicians, patients and funding bodies are keen to assess the effectiveness of promising new therapies quickly. This leads investigators to explore markers that may serve as intermediate end-points or surrogate end-points (IE/Ss) of the CE and which are more frequently occurring and temporally closer to the E/I.