Statistical considerations for calculation of immunogenicity screening assay cut points

Most therapeutic proteins induce an unwanted immune response. Antibodies elicited by these therapeutic proteins may significantly alter drug safety and efficacy, highlighting the need for the strategic assessment of immunogenicity at various stages of clinical development. Immunogenicity testing is generally conducted by a multi-tiered approach whereby patient samples are initially screened for the presence of anti-drug antibodies in a screening assay. The screening assay cut point is statistically determined by evaluation of drug-naïve samples and is typically chosen to correspond to a false positive rate of 5%. While various statistical approaches for determination of this screening cut point have been commonly adopted and described in the immunogenicity literature, the performance of these approaches has not been fully evaluated. This paper reviews various statistical approaches for cut point calculation, evaluates the impact of sampling design and variability on the performance of each statistical approach, and highlights the difference between an 'average' or 'confidence-level' cut point in order to develop more specific recommendations regarding the statistical calculation of immunogenicity screening cut points.