ABSTRACT
In this chapter, we explore several statistical issues which arise in the collection and interpretation of assay data. The hierarchical nonlinear modeling techniques discussed in this book, particularly those in Chapter 5, provide a useful approach to these issues. In developing drugs for therapeutic use, an important challenge is the derivation of suitable assays for measuring drug levels in serum, plasma, or other biological matrices. For instance, availability of reliable assays is central to the conduct of phar-macokinetic studies of the type discussed in the previous chapter. For low molecular weight drugs, assay techniques such as high-performance liquid chromatography (HPLC) are often adequate. The growth of genetic engineering and the consequent ability to exploit recombinant techniques to manufacture macromolecules such as proteins for therapeutic use has presented new challenges for assay development. Traditional methods such as HPLC often lack the specificity and sensitivity required to detect proteins in complex biological matrices. Instead, methods used to measure protein levels frequently exploit the specificity of immunoassay techniques such as radioimmunoassay (RIA) or enzyme-linked immunosorbent assay (ELISA). Such techniques can often achieve sensitivity in the pg/ml range. Although immunoassays are extremely useful in quantifying protein levels, a common requirement is the development of an assay that provides a direct measure of the biological activity of the protein, such as a cell-based bioassay. Such an assay is usually required for regulatory purposes, and reflects the fact that the degree of antigen binding exhibited by a polypeptide such as an antibody may not always provide a good indication of its biological activity. The relaxin data, illustrated in Figure 1.3, provide an example of this type of assay.
