Testing Precision Limits of Neural Network-Based Quality Control Metrics in High-Throughput Digital Microscopy.

OBJECTIVE: Digital microscopy is used to monitor particulates such as protein aggregates within biopharmaceutical products. The images that result encode a wealth of information that is underutilized in pharmaceutical process monitoring. For example, images of particles in protein drug products typically are analyzed only to obtain particle counts and size distributions, even though the images also reflect particle characteristics such as shape and refractive index. Multiple groups have demonstrated that convolutional neural networks (CNNs) can extract information from images of protein aggregates allowing assignment of the likely stress at the "root-cause" of aggregation. A practical limitation of previous CNN-based approaches is that the potential aggregation-inducing stresses must be known a priori, disallowing identification of particles produced by unknown stresses. METHODS: We demonstrate an expanded CNN analysis of flow imaging microscopy (FIM) images incorporating judiciously chosen particle standards within a recently proposed "fingerprinting algorithm" (Biotechnol. & Bioeng. (2020) 117:3322) that allows detection of particles formed by unknown root-causes. We focus on ethylene tetrafluoroethylene (ETFE) microparticles as standard surrogates for protein aggregates. We quantify the sensitivity of the new algorithm to experimental parameters such as microscope focus and solution refractive index changes, and explore how FIM sample noise affects statistical testing procedures. RESULTS & CONCLUSIONS: Applied to real-world microscopy images of protein aggregates, the algorithm reproducibly detects complex, distinguishing "textural features" of particles that are not easily described by standard morphological measurements. This offers promise for quality control applications and for detecting shifts in protein aggregate populations due to stresses resulting from unknown process upsets.

Development of an RTD-Based Flowsheet Modeling Framework for the Assessment of In-Process Control Strategies.

Continuous manufacturing (CM) is an emerging technology which can improve pharmaceutical manufacturing and reduce drug product quality issues. One challenge that needs to be addressed when adopting CM technology is material traceability through the entire continuous process, which constitutes one key aspect of control strategy. Residence time distribution (RTD) plays an important role in material traceability as it characterizes the material spreading through the process. The propagation of upstream disturbances could be predictively tracked through the entire process by convolution of the disturbance and the RTD. The present study sets up the RTD-based modeling framework in a commonly used process modeling environment, gPROMS, and integrates it with existing modules and built-in tools (e.g., parameter estimation). Concentration calculations based on the convolution integral requires access to historical stream property information, which is not readily available in flowsheet modeling platforms. Thus, a novel approach is taken whereby a partial differential equation is used to propagate and store historical data as the simulation marches forward in time. Other stream properties not modeled by an RTD are determined in auxiliary modules. To illustrate the application of the framework, an integrated RTD-auxiliary model for a continuous direct compression manufacturing line was developed. An excellent agreement was found between the model predictions and experiments. The validated model was subsequently used to assess in-process control strategies for feeder and material traceability through the process. Our simulation results show that the employed modeling approach facilitates risk-based assessment of the continuous line by promoting our understanding on the process.

An automated modular assembly line for drugs in a miniaturized plant.

We report here a fully automated, end-to-end, integrated continuous manufacturing process for a small-molecule generic medication with built-in quality assurance. The entire process fits into a box of 30.7 m2 modular footprint and a total residence time of less than 30 h, with a throughput up to 40.3 × 106 tablets per year.

Stability of proteins in the presence of carbohydrates; experiments and modeling using scaled particle theory.

The effects of sucrose and fructose on the free energy of unfolding, DeltaG(N-->D), and on the change in hydrodynamic radius, R(H), upon unfolding were measured for RNase A and alpha-lactalbumin. Recently we analyzed the results for RNase A and showed that the effects of the carbohydrates on the protein's thermal stability can be accurately accounted for by scaled particle theory (SPT), and are thus largely entropic in nature. In this paper we extend this analysis to alpha-lactalbumin and demonstrate the generality of this finding. We also investigate the relationship between SPT and the thermodynamic formalism of preferential interactions. The preferential binding parameters calculated using SPT are in excellent agreement with experimentally measured values available in the literature. This agreement is expected to hold as long as enthalpic interactions between the cosolute and the protein are not important, as appears to be the case here. Finally we use the experimental data and SPT to calculate the change in the number of sugar molecules excluded from the protein surface during unfolding from knowledge of the preferential binding parameter for the native and denatured state of the protein.

Emerging technology: A key enabler for modernizing pharmaceutical manufacturing and advancing product quality.

Issues in product quality have produced recalls and caused drug shortages in United States (U.S.) in the past few years. These quality issues were often due to outdated manufacturing technologies and equipment as well as lack of an effective quality management system. To ensure consistent supply of safe, effective and high-quality drug products available to the patients, the U.S. Food and Drug Administration (FDA) supports modernizing pharmaceutical manufacturing for improvements in product quality. Specifically, five new initiatives are proposed here to achieve this goal. They include: (i) advancing regulatory science for pharmaceutical manufacturing; (ii) establishing a public-private institute for pharmaceutical manufacturing innovation; (iii) creating incentives for investment in the technological upgrade of manufacturing processes and facilities; (iv) leveraging external expertise for regulatory quality assessment of emerging technologies; and (v) promoting the international harmonization of approaches for expediting the global adoption of emerging technologies.

Simultaneous determination of structural and thermodynamic effects of carbohydrate solutes on the thermal stability of ribonuclease A.

This communication describes a new technique for the study of the effects of carbohydrates on the thermal stability of proteins. This approach combines capillary electrophoresis (CE) and protein charge ladders, collections of proteins that differ incrementally in number of chemically modified charged groups, to provide information on both the thermodynamics (i.e., the free energy, DeltaGN-D, of denaturation), and structural changes (i.e., the effective hydrodynamic radius, RH, of proteins in both the native and denatured states) associated with stability. This information, obtained in a single set of electrophoresis experiments, allows a simple microscopic interpretation of the effects of carbohydrate solutes on protein stability. We use this technique to show that the stabilization of ribonuclease A at pH 8.4 by sucrose and fructose can be explained entirely by the contribution these solutes make to the entropy of formation of the protein-solution interface. There is no need, in this case, to refer to quasichemical concepts such as preferential hydration, binding, or exchange of solutes with water at specific sites on the protein to account for the stabilizing effects observed.