Machine Learning and Accelerated Stress Approaches to Differentiate Potential Causes of Aggregation in Polyclonal Antibody Formulations During Shipping.

Therapeutic proteins are among the most widely prescribed medications, with wide distribution and complex supply chains. Shipping exposes protein formulations to stresses that can trigger aggregation, although the exact mechanism(s) responsible for aggregation are unknown. To better understand how shipping causes aggregation, we compared populations of aggregates that were formed in a polyclonal antibody formulation during live shipping studies to populations observed in accelerated stability studies designed to mimic both the sporadic high g-force and continuous low g-force stresses encountered during shipping. Additionally, we compared the effects on aggregation levels generated in two types of secondary packaging, one of which was designed to mitigate the effects of large g-force stresses. Aggregation was quantified using fluorescence intensity of 4,4'-dianilino-1,1'-binaphthyl-5,5'-disulfonic acid (bis-ANS) dye, size exclusion high performance liquid chromatography (SECHPLC), and flow imaging microscopy (FIM). FIM was also combined with machine learning methods to analyze particle morphology distributions. These comparisons revealed that the morphology distributions of aggregates formed during live shipping resemble distributions that result from low g-force events, but not those observed following high g-force events, suggesting that low g-force stresses play a predominant role in shipping-induced aggregation.

Hydrogel Coatings on Container Surfaces Reduce Protein Aggregation Caused by Mechanical Stress and Cavitation.

This work reports the ability of hydrogel coatings to protect therapeutic proteins from cavitation-induced aggregation caused by mechanical stress. Here, we show that polyacrylamide hydrogel coatings on container surfaces suppress mechanical shock-induced cavitation and the associated aggregation of intravenous immunoglobulin (IVIg). First, crosslinked polyacrylamide hydrogels were grown on the surfaces of borosilicate glass vials. Treatment with ultrasound showed that these hydrogel surfaces suppressed cavitation events to levels below those found for unfunctionalized borosilicate glass. Next, IVIg solutions were loaded into these vials and subjected to tumbling, horizontal shaking, and drop testing. Aggregation was quantified by bisANS fluorescence staining and particle counting by flow imaging microscopy (FIM). In all cases, the presence of polyacrylamide hydrogels on the vial surfaces reduced the amount of IVIg aggregation and the number of particulates. In addition, the polyacrylamide appeared to have a protective effect that prevented additional aggregates from forming at extended tumbling times. Finally, drop test studies showed that the polyacrylamide coatings suppressed detectable cavitation. This work reveals how even a simple hydrogel vial coating can have a profound effect on stabilizing protein therapeutics.

Machine learning and statistical analyses for extracting and characterizing "fingerprints" of antibody aggregation at container interfaces from flow microscopy images.

Therapeutic proteins are exposed to numerous stresses during their manufacture, shipping, storage and administration to patients, causing them to aggregate and form particles through a variety of different mechanisms. These varied mechanisms generate particle populations with characteristic morphologies, creating "fingerprints" that are reflected in images recorded using flow imaging microscopy. Particle population fingerprints in test samples can be extracted and compared against those of particles produced under baseline conditions using an algorithm that combines machine learning tools such as convolutional neural networks with statistical tools such as nonparametric density estimation and Rosenblatt transform-based goodness-of-fit hypothesis testing. This analysis provides a quantitative method with user-specified type 1 error rates to determine whether the mechanisms that produce particles in test samples differ from particle formation mechanisms operative under baseline conditions. As a demonstration, this algorithm was used to compare particles within intravenous immunoglobulin formulations that were exposed to freeze-thawing and shaking stresses within a variety of different containers. This analysis revealed that seemingly subtle differences in containers (e.g., glass vials from different manufacturers) generated distinguishable particle populations after the stresses were applied. This algorithm can be used to assess the impact of process and formulation changes on aggregation-related product instabilities.

Flow Microscopy Imaging Is Sensitive to Characteristics of Subvisible Particles in Peginesatide Formulations Associated With Severe Adverse Reactions.

The presence of subvisible particles in formulations of therapeutic proteins is a risk factor for adverse immune responses. Although the immunogenic potential of particulate contaminants likely depends on particle structural characteristics (e.g., composition, size, and shape), exact structure-immunogenicity relationships are unknown. Images recorded by flow imaging microscopy reflect information about particle morphology, but flow microscopy is typically used to determine only particle size distributions, neglecting information on particle morphological features that may be immunologically relevant. We recently developed computational techniques that utilize the Kullback-Leibler divergence and multidimensional scaling to compare the morphological properties of particles in sets of flow microscopy images. In the current work, we combined these techniques with expectation maximization cluster analyses and used them to compare flow imaging microscopy data sets that had been collected by the U.S. Food and Drug Administration after severe adverse drug reactions (including 7 fatalities) were observed in patients who had been administered some lots of peginesatide formulations. Flow microscopy images of particle populations found in the peginesatide lots associated with severe adverse reactions in patients were readily distinguishable from images of particles in lots where severe adverse reactions did not occur.

Deep Convolutional Neural Network Analysis of Flow Imaging Microscopy Data to Classify Subvisible Particles in Protein Formulations.

Flow-imaging microscopy (FIM) is commonly used to characterize subvisible particles in therapeutic protein formulations. Although pharmaceutical companies often collect large repositories of FIM images of protein therapeutic products, current state-of-the-art methods for analyzing these images rely on low-dimensional lists of "morphological features" to characterize particles that ignore much of the information encoded in the existing image databases. Deep convolutional neural networks (sometimes referred to as "CNNs or ConvNets") have demonstrated the ability to extract predictive information from raw macroscopic image data without requiring the selection or specification of "morphological features" in a variety of tasks. However, the inherent heterogeneity of protein therapeutics and optical phenomena associated with subvisible FIM particle measurements introduces new challenges regarding the application of ConvNets to FIM image analysis. We demonstrate a supervised learning technique leveraging ConvNets to extract information from raw images in order to predict the process conditions or stress states (freeze-thawing, mechanical shaking, etc.) that produced a variety of different protein particles. We demonstrate that our new classifier, in combination with a "data pooling" strategy, can nearly perfectly differentiate between protein formulations in a variety of scenarios of relevance to protein therapeutics quality control and process monitoring using as few as 20 particles imaged via FIM.

Microflow Imaging Analyses Reflect Mechanisms of Aggregate Formation: Comparing Protein Particle Data Sets Using the Kullback-Leibler Divergence.

Subvisible particles in therapeutic protein formulations are an increasing manufacturing and regulatory concern because of their potential to cause adverse immune responses. Flow imaging microscopy is used extensively to detect subvisible particles and investigate product deviations, typically by comparing imaging data using histograms of particle descriptors. Such an approach discards much information and requires effort to interpret differences, which is problematic when comparing many data sets. We propose to compare imaging data using the Kullback-Leibler divergence, an information theoretic measure of the difference of distributions (Kullback S, Leibler RA. 1951. Ann Math Stat. 22:79-86). We use the divergence to generate scatter plots representing the similarity between data sets and to classify new data into previously determined categories. Our approach is multidimensional, automated, and less biased than traditional techniques. We demonstrate the method with FlowCAM® imagery of protein aggregates acquired from monoclonal antibody samples subjected to different stresses. The method succeeds in classifying aggregated samples by stress condition and, once trained, is able to identify the stress that caused aggregate formation in new samples. In addition to potentially detecting subtle incipient manufacturing faults, the method may have applications to verification of product uniformity after manufacturing changes, identification of counterfeit products, and development of closely matching bio-similar products.