Purpose: Bioanalytical validations for PK and Immunogenicity assays are often performed on MesoScale Discovery technology and instruments. Bioanalytical guidance’s from the FDA and other regulatory agencies sets validation parameters that should be evaluated including accuracy and precision. We report here a case where there was an observed inconsistency in the PK assay performance in assessing stability. This was partially caused by an inherent variability in an instrument-dependent manner, based on sector position, read time and sequence of data acquisition.
Methods: The PK assay developed to measure a biologic based therapeutic molecule was based on an anti-drug mouse monoclonal antibody immobilized to the surface of MSD plates. The calibration curves and quality control samples made in human serum in predefined levels were diluted to the appropriate minimum required dilution and added to the MSD plate wells where the capture antibody binds the drug molecule. The wells were washed and a ruthenium tagged anti-drug antibody was added to the wells to complete the sandwich and enable the detection of the captured molecule. The plates were subsequently washed, resuspended in MSD 2X read buffer, then read in an MSD Sector Imager 600 instrument. The reported ECL units were recorded for each well and used to generate standard curves and measured concentrations of the quality control samples and unknown samples.
Short term stability experiments were done using freshly prepared quality control samples containing the drug at the assigned LLOQ levels, then incubated at various temperatures for different periods of time. The samples were subsequently stored at -80°C until ready for analysis.
Results: The initial short term stability results showed the molecule was stable and generated reproducible results (less than 25% CV and less than 25% RE) at 4-8°C. However, the short team stability failed at room temperature at 24 hours ( >30% CV%) and 4 hours ( >37% CV%), suggesting the molecule did not have stability at temperatures greater than 8°C. Analysis showed that quality control samples prepared on ice had acceptable accuracy and precision. Further analysis of calibration curves was done to compare different preparations and placed onto MSD plates by rows across the plate. Interestingly, the curves in the top 4 rows had 20% lower ECLU’s than the bottom two rows. Additional analysis was done using a small set of quality control samples repeated in every well in the plate. The Sector 600 instrument reads plates by sector, which is comprised of 16 wells (4x4) at one time. The instrument starts at sector 1 correlating to wells located in Rows E and H, columns 1 – 4. The instrument then moves to sector 2 (rows E-H, columns 5-8), then sector 3 (rows E-H, columns 9-12), sector 4 (rows A-D, columns 9-12), sector 5 (rows A-D, columns 5-8), and finally sector 6 (rows A-D, columns 1-4). These data show a drop in ECLU from the bottom left (sector 1) vs the top left (sector 6), ranging from 16 – 25% less signal.
Conclusion: There were short term stability issues observed with the molecule at room temperature and variability due to signal drop across the plate during the plate reading. This suggested that the immune complexes were unstable at room temperature even during the short time required to read the plate. Laboratories doing analysis on MSD plates that may involve low affinity interactions (ex. anti-drug antibody assays) should assess signal loss across the plate during reading to ensure the measured signals are biologically related and not instrument related. Standard curve points that are placed on the top left part of the plate will have an instrument-derived loss of signal that can also affect the shape of the curve and analyte recovery back-calculations. The impact of not evaluating this could lead to incorrect conclusions based on low signals observed in the top part of the plate vs other regions of the plate.