Exploring the Correlation between LC–MS Multi-Attribute Method and Conventional Chromatographic Product Quality Assays through Multivariate Data Analysis

Biotherapeutics are subject to inherent heterogeneity due to the complex biomanufacturing processes. Numerous analytical techniques have been employed to identify, characterize, and monitor critical quality attributes (CQAs) to ensure product safety, and efficacy. Mass spectrometry (MS)-based multi-attribute method (MAM) has become increasingly popular in biopharmaceutical industry due to its potential to replace multiple traditional analytical methods. However, the correlation between MAM and conventional methods remains to be fully understood. Additionally, the complex analytical workflow and limited throughput of MAM restricts its implementation as a quality control (QC) release assay. Herein, we present a simple, robust, and rapid MAM workflow for monitoring CQAs. Our rapid approach allowed us to create a database from ~700 samples, including site-specific post-translational modifications (PTMs) quantitation results using MAM and data from traditional charge variant and oxidation characterization methods. To gain insights from this database, we employ multivariate data analysis (MVDA) to thoroughly exploit the data. By applying partial least squares regression (PLSR) models, we demonstrate the ability to quantitatively predict charge variants in ion exchange chromatography (IEX) assay and oxidation abundances in hydrophobic-interaction chromatography (HIC) assay using MAM data, highlighting the interconnectivity between MAM and traditional product quality assays. These findings help evaluate the suitability of MAM as a replacement for conventional methods for release, and more importantly, contribute to enhanced process and product understanding.