The opportunities for improved regulatory practice discussed above are exciting, and surely the future will bring other unforeseen opportunities. Translation of these opportunities into practical methods and approaches suitable for routine application in product development and regulation is, however, not a trivial exercise. Key elements of the necessary evaluation and validation process include:
- Demonstration of a clear understanding of the relationship between the endpoint(s) measured and the biological outcome of interest (biological validation, often referred to as "evaluation").
- Determination of the performance characteristics of the assays employed (analytical validation), including sensitivity, accuracy, and reproducibility within and among laboratories.
- Identification of interfering factors that may modify assay outcome, yielding "false" or misleading results that may under- or over-estimate the biological event of interest.
- Development of consensus among the scientific community and responsible regulatory bodies on appropriate application of methods and approaches.
Satisfactory demonstration of these elements is difficult and time-consuming even for a single-endpoint assay. Defining these elements for highly multiplexed assays capable of monitoring many hundreds or thousands of endpoints simultaneously presents a significant challenge. "Biological validation" will need to include studies in important model species, and must include demonstration of an understanding of the relationship of biomarkers employed to cell and tissue injury. For example, it is important to distinguish whether a biomarker is a measure of a rate-limiting defense process that will prevent pathology until a defined threshold is passed, whether it is a marker that indicates that a specific type of damage has already occurred, etc. It is also important to understand the differences and commonalties in such responses among well-established laboratory animal and cellular models, and the human. The principles of assay and biomarker validation have been delineated (ICCVAM, 1997, 1999), and will need to be applied to each of the new biomarkers discussed above.
Some in the field have stated that it will take decades to achieve appropriate validation, but regulatory implementation will likely be much more rapid. Though the pace of scientific change often seems slow to those engaged in its practice, reflection on the rapidity of the adoption into practice of the major advances in science and technology during the past century reveals the opposite. For example, the periods from the discovery that DNA was the genetic material to the construction of transgenic organisms with modified genetic information, from first heavier than air flight to well-established commercial aviation, from invention of the transistor to the current prevalence of microelectronic integrated circuits in our society, and from the first descriptions of intracellular enzymes to the use of these enzymes as biomarkers of cellular toxicity were all achieved in a few decades or less. Why then has the approach to toxicological assessment been so stable over a comparable period of time? This likely stems from two key factors: the excellence of the strategy devised by early toxicologists and the need for conservative change associated with the dependence of the economic viability of new product development on well-established and predictable regulatory rules and practices. However, the current intense focus on the areas discussed above suggests that the field is entering a major transition that will employ the impressive technologies of the biological revolution to improve our approaches to product development and regulation.
Among these improvements, we may look forward to reconstitution of the fundamental set of biomarkers used to identify and monitor pathological and toxicological effects, and introduction of a more sensitive and specific set of markers that allows characterization of tissue sites of damage as well as mechanisms of cellular perturbations. Indeed, there is the potential to develop a new quantitative molecular pathology approach to supplement, or in some cases replace, the present semiquantitative histopathological evaluation that is the principal endpoint upon which many safety decisions are currently based. Molecular techniques may prove to be more objective, more quantitative, and more sensitive than the current approach, which relies on human judgements about changes in morphological structures and cell population alterations. The ability to monitor these biomarkers in vivo should allow increased reliance on direct human studies, as biomarker measurements in the human become more possible. This, coupled with bridging biomarkers that allow comparison of responses in the human with those in laboratory animal models, promises to greatly reduce the present uncertainty in quantitative extrapolation of results from laboratory models to human outcomes. Together, these approaches should dramatically improve selection of lead compounds in discovery, evaluation of toxicity in animal models, linkage between animal models and humans, and human monitoring. To reach these goals, applied research will be required to establish the necessary linkage between each new biomarker and the pathologies of interest, as well as to establish the statistical performance characteristics of the system of measurement (reproducibility, robustness, etc.). This will require commitment and collaboration among all sectors involved in product development, regulation, and utilization—the public, industry, and government.