Both the industry and regulators have broadly adopted the conceptual use of biomarkers as a tool in development of pharmaceuticals. There is still, however, considerable confusion over how and when they should be used and interpreted. This is due in part to uncertainty over the degree of emphasis that can be placed on marker data in preclinical and clinical studies. This confusion is compounded by diffuse definitions applied to biomarkers, particularly when discussing concepts of qualification and validation.

In addition, there is considerable variety in the types of technologies that can be used to generate biomarker data, from complex imaging technologies through to more familiar and commonplace laboratory-based clinical diagnostic approaches. In this article we hope to clarify some of these points and identify where biomarkers may be of value to both the sponsor and regulator in getting safe and effective drugs to the waiting patient population.

DEFINITION

The most commonly-quoted definition of a biomarker is the following: a characteristic that can be measured and evaluated as an indicator of a normal biological process, pathogenic processes, or a pharmacologic response to therapeutic intervention (1). This definition is extremely wide and can encompass traditional clinical assessment and diagnostic measurements as well as imaging technologies and genomic, proteomic and metabolomic analyses. What this definition does not provide, however, is a clear classification of which markers are going to identify whether a drug is safe and efficacious.

A great deal of time has been spent considering whether a biomarker fulfils the criteria to be considered a surrogate marker of efficacy or safety clinical endpoints. ICH4 defines a surrogate endpoint as an endpoint that allows the prediction of a clinically important outcome but does not in itself measure a clinical endpoint. In order to meet this criterion a considerable amount of data from clinical experiments needs to be collected to determine whether the selected biomarker (or biomarkers) does allow prediction of the selected clinical outcome. The terminology around the process of demonstrating whether a biomarker is predictive of clinical outcome has been unhelpful. Many still refer to this as validation, leading to the confusion of this process with technical assessments of the performance of the methodologies generating assessment data.