Purpose: As part of a sound pharmaceutical quality system (ICH Q10), an increasing trend of MS-based techniques in lab settings traditionally associated with optical-based assays has been observed in an effort to improve the body of knowledge surrounding a drug product. This has facilitated the introduction of LCMS-based methods for semi-targeted monitoring of biotherapeutic protein attributes or “Multi-Attribute Monitoring” in an effort to reduce redundant assays and increase productivity. Challenges associated with these approaches include data-mining expansive data sets, determining screening parameters, and transferring methods. In this study, we have generated a data set representative of a process development setting to demonstrate how these challenges can be addressed with a single CDS platform with MS control and data processing capabilities.
Methods: To facilitate a data-based discussion in addressing the challenges associated with the deployment of a MAM-based method in process development, a mAb peptide map data set that contains 48 injections was acquired using a fit-for-purpose LCMS platform incorporating GXP practices. System suitability was performed following USP 621 guidelines using the integrated functionality of the CDS software to assess system readiness based on a reference standard. 19 product quality attributes (PQAs) including glycosylation, oxidation, deamidation, and isomerization were monitored over a 40 min RPLC-based separation using selected ion recording (SIR) for MS acquisition. Visualization of data was accomplished using integrated tools to streamline data review and selection of critical quality attributes (CQA) for deployment into regulated development and QC/lot release roles.
Results: System suitability results were processed in an automated fashion with pass/fail criteria based on USP 621 guidance (N=6). System readiness for both LC and MS components were found to pass within specification criteria based on % RSD of retention time, peak area, and mass accuracy. From the 48-well injection series, 1,142 unique results were generated in relationship to the attributes being monitored. This comprehensive data set was refined in an automated fashion using screening parameters based on % RSD and % modification to generate a sub-set of data (N < 10) as potential CQA for QC/lot release. Integrated reporting functionality including control charts and threshold plots were used to aid in the visualization of data as well as streamline the identification process. Attributes of interest were transferred to a manufacturing environment using the same LCMS-based CDS platform. Using targeted monitoring, CQA abundance or % modification of a biological replicate (N=3) representing a release assay was quickly assessed using pass/fail criteria integrated into the reporting functionality of the CDS for efficient review of data and lot release.
Conclusion: Appropriate screening parameters and visualization tools streamline the identification process of CQAs from expansive data sets prior to manufacturing.