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

Maddux, Nathaniel R.; Daniels, Austin L.; Randolph, Theodore W.

doi:10.1016/j.xphs.2017.01.030

Cited by 19 publications

(29 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Particulate matter in therapeutic protein products is the focus of increased attention due both to industrial quality control and patient safety concerns. [1][2][3][4][5] "Subvisible" particles (defined here as objects 25 mm in size) are contained in all commercial therapeutic prote in formulations. 3,6,7 Subvisible particles may be composed of aggregated proteins or nonbiological materials (e.g., silicone oil).…”

Section: Introductionmentioning

confidence: 99%

“…The images of subvisible particles returned by FIM are believed to contain a significant amount of structural information about the particles in a given sample. 5 Characterization techniques capable of leveraging the structural information embedded in FIM images show promise as tools for evaluating therapeutic protein drugs at different stages of their life spans (from the manufacturing plant to delivery to patients) and early steps in this direction have been recently proposed. 5 Historically in FIM analysis, accurate characterization of subvisible particles from FIM images has required explicit identification of the relevant morphological features of the particles from the raw FIM images.…”

Section: Introductionmentioning

confidence: 99%

“…Most FIM image analysis methods developed to date only use the small number of morphological features reported by FIM instruments such as aspect ratio, compactness, and particle intensity to build a "representation" of the imageda numerical description of the features of the image that can be used for characterization and classification tasks. 3,5,21,22 However, this short list of features neglects a significant amount of potentially relevant structural information contained in the full RGB (or grayscale) FIM images. Furthermore, the short list of morphologic features contains features that are highly correlated, further reducing their information content.…”

Section: Introductionmentioning

confidence: 99%

“…A simple "data pooling" strategy is combined with deep ConvNets to obtain a classifier obtaining nearly perfect performance in distinguishing various protein solutions subjected to different stresses or processing conditions using as few as 20 images to predict a class. We revisit some of the more challenging data sets explored in the study by Maddux et al 5 to quantitatively and qualitatively demonstrate various advantages of our new ConvNet-based approach. Specifically, we reanalyze FIM images of aggregates of a mAb that were created as a result of freeze-thawing and mechanical agitation, 2 stresses that are relevant in therapeutic protein manufacturing, transportation, and drug administration.…”

Section: Introductionmentioning

confidence: 99%

“…Specifically, we reanalyze FIM images of aggregates of a mAb that were created as a result of freeze-thawing and mechanical agitation, 2 stresses that are relevant in therapeutic protein manufacturing, transportation, and drug administration. 5 In addition, we explore the use of deep ConvNet analysis of FIM images to detect differences in subvisible particles produced during simulated fill-finish operations using a fixed protein therapeutic formulation but 2 slightly different operating conditions. To do this, we created protein aggregates by pumping a single solution of intravenous immunoglobulin (IVIG) antibodies through 1 of 2 nominally identical reciprocating pistontype fill-finish pumps.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

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

Calderon

Daniels

Randolph

2018

Journal of Pharmaceutical Sciences

Self Cite

View full text Add to dashboard Cite

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

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

Calderon

Daniels

Randolph

2018

Journal of Pharmaceutical Sciences

Self Cite

View full text Add to dashboard Cite

Machine learning approaches to root cause analysis, characterization, and monitoring of subvisible particles in monoclonal antibody formulations

Greenblott

Zhang

Calderon

et al. 2022

Biotech & Bioengineering

Self Cite

View full text Add to dashboard Cite

Processing stresses on therapeutic proteins may cause formation of subvisible particles. Different stress mechanisms generate particle populations with characteristic morphological "fingerprints," and machine learning techniques like convolutional neural networks (CNNs) allow classification of microscopy images of these particles according to known stresses at their root cause. Using CNNs to classify novel particle types not included during network training may lead to inaccurate classification, however, using CNNs to monitor the presence of particulate matter not explicitly used in training could serve as a useful process analytical technology.We used CNNs to classify and identify the root cause of particles generated by subjecting three monoclonal antibodies (mAbs) to various common manufacturing stresses. We probed the generality of particles generated by stressing different mAbs in different formulations and showed that CNN analyses were sensitive not only to the applied stress, but also the buffer conditions and the particular mAb that generated particle populations. Thus, models trained on images of particles created with one mAb and buffer system may not provide accurate root cause analysis when applied to particles generated by other mAb and buffer systems. A lever-rule analysis of CNN-derived fingerprints was used to characterize the composition of mixtures of particle types. Finally, we monitored the temporal evolution of CNN-derived fingerprints when novel populations of particles, which were not included during training, were generated by pumping mAb solutions through a peristaltic pump.

show abstract

Supervised and unsupervised machine learning approaches for monitoring subvisible particles within an aluminum‐salt adjuvanted vaccine formulation

Greenblott,

Wood,

Zhang

et al. 2024

Biotech & Bioengineering

Self Cite

View full text Add to dashboard Cite

Suspensions of protein antigens adsorbed to aluminum‐salt adjuvants are used in many vaccines and require mixing during vial filling operations to prevent sedimentation. However, the mixing of vaccine formulations may generate undesirable particles that are difficult to detect against the background of suspended adjuvant particles. We simulated the mixing of a suspension containing a protein antigen adsorbed to an aluminum‐salt adjuvant using a recirculating peristaltic pump and used flow imaging microscopy to record images of particles within the pumped suspensions. Supervised convolutional neural networks (CNNs) were used to analyze the images and create “fingerprints” of particle morphology distributions, allowing detection of new particles generated during pumping. These results were compared to those obtained from an unsupervised machine learning algorithm relying on variational autoencoders (VAEs) that were also used to detect new particles generated during pumping. Analyses of images conducted by applying both supervised CNNs and VAEs found that rates of generation of new particles were higher in aluminum‐salt adjuvant suspensions containing protein antigen than placebo suspensions containing only adjuvant. Finally, front‐face fluorescence measurements of the vaccine suspensions indicated changes in solvent exposure of tryptophan residues in the protein that occurred concomitantly with new particle generation during pumping.

show abstract

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

Cited by 19 publications

References 31 publications

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

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

Machine learning approaches to root cause analysis, characterization, and monitoring of subvisible particles in monoclonal antibody formulations

Supervised and unsupervised machine learning approaches for monitoring subvisible particles within an aluminum‐salt adjuvanted vaccine formulation

Contact Info

Product

Resources

About