Statistical methods of screening cut point determination in immunogenicity studies

Area of Science:

• Pharmacokinetics and Drug Metabolism
• Immunology
• Biostatistics

Background:

• Current methods for calculating screening cut points (CP) in immunogenicity studies often use assay validation data with replicates applied to nonreplicate immunogenicity study data.
• Independent and identically distributed (IID) methods treat replicates as independent samples, while average (AVE) methods reduce inter-assay variability but may not fully capture uncertainty.
• This mismatch can impact the accuracy of determining antidrug antibody rates.

Purpose of the Study:

• To propose and evaluate a random effect model (REM) for calculating screening cut points (CP).
• To investigate the impact of design incompatibilities between assay validation and immunogenicity studies on CP.
• To compare the performance of REM against existing IID and AVE methods.

Main Methods:

• Development of a random effect model (REM) to incorporate the covariance structure of repeated measurements.
• Investigation of the impact of applying replicate-based validation data to nonreplicate immunogenicity data.
• Comparative analysis of CP values generated by REM, IID, and AVE methods.

Main Results:

• The independent and identically distributed (IID) method may be unsuitable when replicate variability is a dominant source of uncertainty.
• The random effect model (REM) effectively considers the covariance structure of repeated measurements.
• Screening cut points calculated by REM are lower than those from IID but higher than those from AVE.

Conclusions:

• A random effect model (REM) offers a more appropriate approach for calculating screening cut points (CP) in immunogenicity studies.
• REM provides a more accurate CP by accounting for the inherent variability in repeated measurements.
• The proposed REM method offers a balanced approach compared to IID and AVE, improving the reliability of antidrug antibody assessments.