for assessing them—for example, sensitivity, specificity, reliability, and discrimination. Since biomarkers typically quantify physiological states or therapeutic responses, choosing the values in decision rules—for example, “cutoff points”—becomes very important and difficult, as different values can yield quite different perspectives. In the familiar examples of creatinine for kidney injury, troponin for cardiac injury, and alanine aminotransferase (ALT) for liver injury, the higher is the value, the higher is the probability of true injury, yet low values may signal the early phase of damage.
The use of biomarkers often involves a trade-off between sensitivity, or the proportion of positive responses that a biomarker correctly identifies as positive, and specificity, or the proportion of negative responses that a biomarker correctly identifies as negative. Different degrees of sensitivity and specificity are needed in different circumstances, and will be dependent upon the intended use of the biomarker.
Individual biomarkers differ in the extent to which they reflect a known biological mechanism. Greater understanding of mechanism can be extremely helpful in such tasks as comparing the action of related drugs or gauging the relevance of animal findings to humans. However, biomarkers can provide useful information even when a detailed understanding of mechanism is lacking.
No one biomarker is likely to have all of the characteristics necessary to provide a robust understanding of response As a result, future use of combinations of multiple biomarkers to enable improved prediction of drug efficacy and safety is likely. Yet the use of such combinations of biomarkers may introduce its own challenges, including technical issues of how to combine results, how to control quality, and how to interpret results in different clinical contexts.
The improper use or interpretation of biomarkers can be detrimental in both clinical and research settings by misdirecting therapy or research activities. If biomarkers are to be used properly, there needs to be an understanding of their sensitivity and specificity, how and in what contexts to use them, how to interpret them in those various contexts, and how to properly validate them.
To better understand the current state of the art in the development of biomarkers, consider the issues involved in their development and use, and discuss their future development, the Institute of Medicine’s (IOM’s) Forum on Drug Discovery, Development, and Translation held a 1-day workshop on October 24, 2008, on “Assessing and Accelerating the Development of Biomarkers for Drug Safety.” Participants included experts from academia, government, and industry. To ensure a manageable range of content, the