Optimization and Validation Total Binding Antibody Assays

5 OPTIMIZATION AND VALIDATION TOTAL BINDING ANTIBODY ASSAYS 

The optimization and validation of immunoassays for immunogenicity (ADA) testing has been described in detail in several publications [9,14,33,34]. In this section, we will describe the evaluation of relevant performance characteristics (validation parameters) that require the most effort. Some of these are different from the validation of traditional bioanalytical pharmacokinetic (PK) methods for macromolecules [35 37]. Precision, specificity, robustness, and ruggedness are determined similarly between ADA and PK methods. However, recovery/accuracy, sensitivity, stability, linearity, system suitability controls, and selectivity are treated differently between these two types of assays. 

For certain assay formats such as ECL, the signal of the diluent (reagent blank) samples is usually greater than the signal of the matrix blank samples. In such cases, the percent reduction in background is irrelevant. So the MRD is simply the dilution level that results in an optimal fold change between the diluted samples to matrix blank, without compromising much on variability, that is, the dilution level that optimizes the Z factor. 

Accuracy is applicable to PK assays while recovery is associated with ADA methods. The majority of ADA assays do not use a standard curve because a well-characterized reference standard is not obtainable. In addition, ADA responses are usually heterogeneous and differ from subject to subject, such that calculating accuracy is not relevant. Thus, recovery is employed to demonstrate that the matrix does not interfere with the detection of the ADA. In practice, the response for the maximum dilution (titer) of matrix that has been spiked with ADA versus matrix without ADA is evaluated and should remain within an acceptable range. The acceptance ranges for recovery are usually expressed in response values rather than concentration units (accuracy). The ADA used in recovery experiments maybe obtained from one subject, a pool of subjects, or even from a different species than the test subjects and thus does not necessarily reflect the ADAs in the study samples. 

Three types of screening assay cut point may be used depending on the characteristics of the data from these samples. They are (1) fixed cut point, (2) floating cut point, and (3) dynamic cut point [34]. 

---