(T1130-02-07) Mitigation of the Presence of Crosstalk in an ADA Bridging Assay Using the MSD Platform for the Detection of Antibodies against an Antibody-Drug Conjugate

Purpose: A bridging assay for the detection of anti-drug antibodies (ADAs) against an antibody-drug conjugate (ADC) was developed and validated successfully using the widely used Meso Scale Diagnostics (MSD) platform. This method was then used for analysis of clinical study samples during which unexpected results due to crosstalk were obtained.

Methods: The method developed and validated for detection and characterization of antibodies against this ADC drug product was the industry standard ADA bridging multi-tiered assay using the MSD platform which included screening, titer and domain characterization tiers. A confirmatory tier was not included due to detectable levels of pre-existing antibodies in ~90% of individuals.

Results: The validated assay was used to analyze clinical study samples for the detection and characterization of ADA against the ADC drug product. During study sample analyses, the negative control samples (tested as 3 sets per plate) occasionally generated higher than expected signals (above acceptance threshold), leading to multiple run failures. Initial investigations into the causes for this unexpected run failure rate ruled-out variability due to reagent lots, incubation steps and minimum required dilution schemes. Further investigations focused on the presence of pre-existing ADAs observed in most human samples, with some reaching extremely high levels not observed during development or validation; this was due to the antibody-bacterial toxin conjugate (ADC) used in this study that included the toxin from a relatively common bacteria to which many people had already been exposed. Based on data review and high signal sample localization, a hypothesis was that very high signals in certain plate wells could affect readings in neighboring wells. This hypothesis was consistent with the fact that higher NC signals were not observed in the titer assay tier plates where very high ADA signals were not observed (samples are always tested diluted in the titer assay). Knowing that the MSD sector imaging system takes readings of 96‑well plates in 6 sectors of 4 by 4 wells, NC sample signals were correlated with the total signal of adjacent wells (R² = 0.35) and with the total signal of the MSD plate sector (R² = 0.68). This crosstalk phenomenon was observed if the total signal in an MSD plate sector was very high, increasing the signal of all wells in an affected sector by up to ~100 ECL counts. Crosstalk in this ADA assay could increase run failure rate (NC signal above threshold), generate false-positive results in the screening tier and generate false-negative/positive results in the characterization tiers by affecting signal of tested samples with and/or without competitor. Since most sample analysis results had already been generated when this crosstalk phenomenon was identified, study samples with low signals were retested in the screening and/or characterization tiers in MSD plate sectors with low signals. Results from retested samples revealed false-positive results in the screening assay and false-negative results in the characterization assays.

Conclusion: Enormous amounts of ADAs in pre-clinical/clinical settings could lead to crosstalk issues when using the MSD platform. This phenomenon was not observed during method development and validation as these very high signals were only obtained in some study samples from treated subjects. In such cases, carefully designed assay plate maps should be considered to avoid control sample failures and to ensure accurate results in the different tiers.