In Journal of pharmaceutical sciences
A novel approach to identify five types of simulated stresses that induce protein aggregation in pre-filled syringe (PFS)-type biopharmaceuticals was developed. Principal components analyses of texture metrics extracted from flow imaging microscopy images were used to define sub-groups of particles. Supervised machine learning methods including convolutional neural networks (CNNs) were used to train classifiers to identify sub-group membership of constituent particles to generate distribution profiles. The applicability of the stress-specific signatures for distinguishing stress source types was verified. The high classification efficiencies (100%) precipitated the collection of data from more than 20 independent experiments to train support vector machines, k-nearest neighbors and ensemble classifiers. The performances of the trained classifiers were validated. High classification efficiencies for friability (80% - 100%) and heating at 90°C (85% - 100%) are indicative of high reliability of these methods for stress-stability assays while extreme variations in freeze-thawing (2% - 100%) and heating at 60°C (2.25% - 98.25%) indicate the unpredictability of particle composition profiles for these forced degradation conditions. We also developed subvisible particle classifiers using CNN to automatically identify silicone oil droplets, air bubbles and protein aggregates. The developed classifiers will contribute to mitigating aggregation in biopharmaceuticals via the identification of stress sources.
Gambe-Gilbuena Arni, Shibano Yuriko, Krayukhina Elena, Torisu Tetsuo, Uchiyama Susumu
Analysis, Antibody drugs, Biopharmaceutical characterization, Injectables, Machine learning, Neural Network, Protein aggregation