Stress Temperature Studies in Small Scale Hastelloy® Drug Substance Containers Lead to Increased Extent of and Increased Variability in Antibody-Drug Conjugate and Monoclonal Antibody Aggregation: Evidence for Novel Oxidation-Induced Crosslinking in Monoclonal Antibodies.

Health authorities require that suitable stability of the drug substance be shown in relevant materials of construction. ICH Q1A(R2) explicitly states that "stability studies should be conducted on drug substance packaged in a container closure system that is the same as or simulates the packaging proposed for storage and distribution". Stainless steel containers are commonly used for the long-term storage of frozen bulk drug substances (DSs). Hastelloy®-based metal containers are sometimes used due to their higher corrosion resistance and significantly lower iron content to mitigate the potential corrosion-related risks associated with high salt formulations. Despite their benefits, we have found that elevated temperature stability studies in small scale Hastelloy® containers can lead to degradation that is not representative of degradation under typical storage conditions relevant to the manufacturing process. We provide evidence for an oxidation-induced aggregation mechanism that is based on Fenton chemistry with peroxide being supplied by the autoxidation of polysorbate at stress temperatures. Further, variation in the rates of iron leaching between individual small scale containers is shown to be the cause of the variable rates of degradation through strong correlations between leached iron levels and the extents of oxidation and aggregation. The addition of a metal chelator or the removal of polysorbate from the formulation mitigates the oxidation and the non-representative behavior. Extended characterization by LC-MS and 18O labeled peptide mapping shows that a significant portion of the aggregate formed under these conditions is covalently crosslinked and that the predominant covalent species is either a dityrosine or tyrosine-tryptophan crosslink between an Fc peptide and a Fab peptide. This report is the first time either of these two crosslinks have been reported for antibodies with detailed analytical characterization. Because the behavior observed in these studies is not representative of degradation under typical storage conditions relevant to the manufacturing process, this study demonstrates that small scale stress studies in metal containers should be performed with caution and that extended incubation times can lead to non-representative degradation mechanisms.

A Random Forest Approach for Counting Silicone Oil Droplets and Protein Particles in Antibody Formulations Using Flow Microscopy.

PURPOSE: To evaluate a random forest model that counts silicone oil droplets and non-silicone oil particles in protein formulations with large class imbalance. METHODS: In this work, we present a novel approach for automated image analysis of flow microscopy data based on random forest classification enabling rapid analysis of large data sets. The random forest approach overcomes many of the limitations of traditional classification schemes derived from simple filters or regression models. In particular, the approach does not require a priori selection of important morphology parameters. RESULTS: We analyzed silicone oil droplets and non-silicone oil particles observed in four model systems with protein concentrations of 20, 50 and 125 mg/mL. Filters based on random forests achieve higher classification accuracies when compared to regression based filters. Additionally, we showcase a procedure that allows for accurate counting of particles ≥1 µm. CONCLUSIONS: Our method is generally applicable for classification and counting of different classes of particles as long as class morphologies are differentially expressed.

A Small-scale Model to Assess the Risk of Leachables from Single-use Bioprocess Containers through Protein Quality Characterization.

Leachables from single-use bioprocess containers (BPCs) are a source of process-related impurities that have the potential to alter product quality of biotherapeutics and affect patient health. Leachables often exist at very low concentrations, making it difficult to detect their presence and challenging to assess their impact on protein quality. A small-scale stress model based on assessing protein stability was developed to evaluate the potential risks associated with storing biotherapeutics in disposable bags caused by the presence of leachables. Small-scale BPCs were filled with protein solution at high surface area-to-volume ratios (≥3× the surface area-to-volume ratio of manufacturing-scale BPCs) and incubated at stress temperatures (e.g., 25 °C or 30 °C for up to 12 weeks) along with an appropriate storage vessel (e.g., glass vial or stainless steel) as a control for side-by-side comparison. Changes in protein size variants measured by size exclusion chromatography, capillary electrophoresis, and particle formation for two monoclonal antibodies using both the small-scale stress model and a control revealed a detrimental effect of gamma-irradiated BPCs on protein aggregation and significant BPC difference between earlier and later batches. It was found that preincubation of the empty BPCs prior to protein storage improved protein stability, suggesting the presence of volatile or heat-sensitive leachables (heat-labile or thermally degraded). In addition, increasing the polysorbate 20 concentration lowered, but did not completely mitigate, the leachable-protein interactions, indicating the presence of a hydrophobic leachable. Overall, this model can inform the risk of BPC leachables on biotherapeutics during routine manufacturing and assist in making decisions on the selection of a suitable BPC for the manufacturing process by assessing changes in product quality. LAY ABSTRACT: Leachables from single-use systems often exist in small quantities and are difficult to detect with existing analytical methods. The presence of relevant detrimental leachables from single-use bioprocess containers (BPCs) can be indirectly detected by studying the stability of monoclonal antibodies via changes by size exclusion chromatography, capillary electrophoresis sodium dodecyl sulfate, and visible/sub-visible particles using a small-scale stress model containing high surface area-to-volume ratio at elevated temperature alongside with an appropriate control (e.g., glass vials or stainless steel containers). These changes in protein quality attributes allowed the evaluation of potential risks associated with adopting single-use bioprocess containers for storage as well as bag quality and bag differences between earlier and later batches. These leachables appear to be generated during the bag sterilization process by gamma irradiation. Improvements in protein stability after storage in "preheated" bags indicated that these leachables may be thermally unstable or volatile. The effect of surfactant levels, storage temperatures, surface area-to-volume ratios, filtration, and buffer exchange on leachables and protein stability were also assessed.

Predicting the Agitation-Induced Aggregation of Monoclonal Antibodies Using Surface Tensiometry.

Adsorption of antibody therapeutics to air-liquid interfaces can enhance aggregation, particularly when the solution does not contain protective surfactant or when the surfactant is diluted as occurs during preparation of intravenous infusion bags. The ability to predict an antibody's propensity for interfacially mediated aggregation is particularly useful during product development to ensure the quality, potency, and safety of the therapeutic. To develop a predictive tool, we investigated the surface pressure and surface excess of a panel of 16 antibodies as well as determined their aggregation propensity at the air-liquid interface in an agitation stress model. Our data demonstrated that the initial rate of surface pressure increase upon antibody adsorption to the air-liquid interface strongly predicted the extent of agitation-induced aggregation. Other factors, including the hydrophobicity, equilibrium surface pressure, and interfacial concentration of an antibody, were not adequate predictors of its susceptibility to aggregation. In addition to developing a predictive tool, we extended the interfacial characterization to better understand the mechanisms of antibody aggregation at an air-liquid interface during agitation stress. We believe that the kinetics of antibody rearrangement and conformational change after adsorbing to the interface, leading to the development of attractive antibody-antibody interactions, dictated the extent of aggregation. Overall, our results demonstrate how surface pressure measurements can be implemented as a rapid screening tool for the identification of antibodies with a high propensity to aggregate upon adsorption to an air-liquid interface while also furthering our understanding of interfacially mediated protein aggregation.

Quantification and characterization of micrometer and submicrometer subvisible particles in protein therapeutics by use of a suspended microchannel resonator.

The ability to characterize micrometer and submicrometer particles in solution is of fundamental importance to understanding the relationship between protein particles in biotherapeutics and concerns raised regarding immunogenicity. While a number of characterization methods are available for analyzing subvisible particle content in protein pharmaceuticals, counting and characterizing particles within the entire subvisible size range remains a significant challenge due to the properties of the proteinaceous particles themselves and to the limitations of the available techniques. Additionally, as silicone oil-lubricated prefilled syringes become a favored primary packaging for biotherapeutic products, proteinaceous subvisible particle characterization is further complicated by the presence of silicone oil droplets in solution. Here, we critically evaluate and apply a novel method for particle characterization that relies on differences in particle buoyant mass to characterize particle content in the range of ca. 0.5-5 µm. A model particle system was specifically designed to evaluate the ability of the suspended microchannel resonator (SMR) to distinguish between buoyant particles (e.g., silicone oil) and dense particles (e.g., protein particles) in aqueous solution. In addition, this emerging technique was successfully applied to high-concentration monoclonal antibody solutions stored in prefilled syringes in stressed stability studies. It is shown that the SMR system can potentially distinguish between silicone oil droplets and protein particles in a size range that is challenging for many subvisible particle characterization methods. Limitations of the SMR method are also discussed.

Antibody binding to a tethered vesicle assembly using QCM-D.

The bilayer-tethered vesicle assembly has recently been proposed as a biomimetic model membrane platform for the analysis of integral membrane proteins. Here, we explore the binding of antibodies to membrane components of the vesicle assembly through the use of quartz crystal microbalance with dissipation monitoring (QCM-D). The technique provides a quantitative, label-free avenue to study binding processes at membrane surfaces. However, converting the signal generated upon binding to the actual amount of antibody bound has been a challenge for a viscoelastic system such as the tethered vesicle assembly. In this work, we first established an empirical relationship between the amount of bound antibody and the corresponding QCM-D response. Then, the results were examined in the context of an existing model describing the QCM-D response under a variety of theoretical loading conditions. As a model system, we investigated the binding of monoclonal antidinitrophenyl (DNP) IgG(1) to tethered vesicles displaying DNP hapten groups. The measured frequency and dissipation responses upon binding were compared to an independent measure of the amount of bound antibody obtained through the use of an in situ ELISA assay. At saturation, the surface mass density of bound antibody was approximately 900 ng/cm(2). Further, through the application of QCM-D models that describe the response of the quartz when loaded by either a single homogeneous viscoelastic film or by a two-layered viscoelastic film, we found that a homogeneous, one-layer model accurately predicts the amount of antibody bound to the tethered vesicles near antibody surface saturation, but a two-layer model must be invoked to accurately describe the kinetic response of the dissipation factor, which suggests that the binding of the antibody results in a stiffening of the top layer of the film.

Viscoelastic characterization of high concentration antibody formulations using quartz crystal microbalance with dissipation monitoring.

With increasing protein concentrations, therapeutic protein formulations are increasingly demonstrating significant deviations from ideal dilute solution behavior due to protein-protein interactions. These interactions lead to unique biophysical challenges in the administration of biopharmaceuticals including high apparent viscosity and viscoelasticity as well as challenges in maintaining the physical stability of proteins in solution. Here, we describe a straightforward analytical method to calculate the complex modulus and viscosity of high concentration protein solutions from measurements made using quartz crystal microbalance with dissipation monitoring (QCM-D). Further, this methodology was used to investigate the dependence of the storage and loss moduli (G' and G'', respectively) of a humanized monoclonal antibody solution on solution pH. Unlike recent reports, the effect of protein deposition onto the surface of the quartz sensor crystal was measured and explicitly accounted for during analysis when determining the solution's complex modulus. It was found that the ratio G''/G' was significantly greater than unity for all solutions investigated, but demonstrated a distinct maximum at pH 5.5 indicating that the solution exhibited the greatest liquid-like behavior at this pH. In addition, measurements were made at higher frequencies, which were found to be more sensitive to the changes in pH than those made at lower frequencies. It was also found that the viscoelastic ratio was relatively insensitive to the frequency of measurement at lower pH, but showed greater dependence on frequency as pH increased. The characterization of the rheological properties of high concentration antibody solutions provides insight into protein-protein interactions, and the methodology presented here demonstrates a straightforward way to determine the viscoelastic properties using ultrasonic rheology without the drawbacks of numerical fitting.

Quantitative analysis of tethered vesicle assemblies by quartz crystal microbalance with dissipation monitoring: Binding dynamics and bound water content.

To implement the molecular recognition properties of membrane proteins for applications including biosensors and diagnostic arrays, the construction of a biomimetic platform capable of maintaining protein structure and function is required. In this paper, we describe a tethered phospholipid vesicle assembly that overcomes the major limitations of planar supported lipid bilayers and alternative biomimetic membrane platforms and characterize it using quartz crystal microbalance with dissipation monitoring (QCM-D) and fluorescence microscopy. We provide evidence of a one-step mechanism for bilayer formation and monitor the subsequent adsorption and binding of streptavidin, vesicles, and streptavidin-coated microspheres. For all three species, we identify a critical surface density above which a significant amount of coupled interstitial water contributes to the response of the quartz resonator in a phenomenon similar to dynamic coupling due to surface roughness. A Sauerbrey-type analysis is sufficient to accurately interpret the QCM-D results for streptavidin binding if water is treated as an additional inertial mass, but viscoelastic models must be invoked for vesicle and microsphere binding. Additionally, we present evidence of vesicle flattening, possibly enhanced by a biotin-mediated membrane-membrane interaction.