Drawing a line in the sand: impact of analytical choices on anti-drug antibody cut-points and testing outcomes

The identification of anti-drug antibodies (ADA) raised against biologic drugs is important for understanding and ensuring efficacy and safety. Because the immunogenicity of novel biologics is unknown before clinical trials, ADA assays are often developed with positivity thresholds assigned based on statistically determined cut-points defined by testing of samples from treatment-naïve donors.

While the standard approaches are based on reasonable theoretical models, we aimed to identify the impact of alternative analytical pipelines on assignment of ADA positivity by analysis of Tier 1 screening and Tier 2 confirmatory ADA bridging assay data for up to 138 mAbs across up to hundreds of serum samples. We evaluate the utility of data augmentation through bootstrapping and the impact of outlier removal approaches on the consistency of ADA status determinations.

We find that bootstrapping supports assay development efficiency by improving confidence in threshold setting in the context of limited numbers of test samples and that while outlier exclusion approach led to different apparent levels of ADA positivity, immunogenic drug products were identified by differences in the distribution of sample profiles from naïve and treated participants by each method evaluated.

NCT02716675 https://clinicaltrials.gov/study/NCT02716675NCT02568215NCT02568215 https://clinicaltrials.gov/study/NCT02568215NCT03875209NCT03875209 https://clinicaltrials.gov/study/NCT03875209NCT04173819NCT04173819 https://clinicaltrials.gov/study/NCT04173819.