Machine Learning is believed to be a game changer for industry, especially by big Pharma reflected by significant investments. Starting from 101 in machine learning, this contribution outlines the current machine learning status for process development with examples, provides future directions towards more global high-value machine learning-centric development, concepts to engage people in the digital evolution, and ideas for machine learning-centric business models.

This talk will describe GPEx Boost technology in more detail and discuss case study comparisons. GPEx® cell line development technology is a proven technology that generates highly stable, high titer production cell lines.

In an effort to improve the performance of a manufacturing clone expressing a complex therapeutic protein, we investigated cell-line engineering, as well as directed modulations of the cell culture process. Applying a pro-productive small RNA in combination with process optimization, we successfully enhanced specific productivity and final product titer, proving the superiority of this integrated approach.

One of the most relevant limitations of CHO cells in culture is apoptosis, which can be triggered by nutrient depletion or by-product accumulation. In this work, the impact of 8 anti-apoptotic genes was tested in terms of anti-apoptotic effect, changes in metabolism and productivity in batch and fed-batch. To this end, targeted integration of those genes combined with a recombinant protein has been used to generate isogenic cell lines that only differ in the anti-apoptotic gene inserted.

Dr. Corrigall will provide a brief overview of how GlaxoSmithKline has embedded the Beacon system to efficiently identify cell lines secreting antibody-based molecules with high productivity and >99% monoclonality assurance. Holly will also describe recent validation studies of a novel assay reagent for both traditional and novel-format antibody-based molecules.

We have established 0.25 S/m as a threshold cytoplasmic conductivity below which cells cannot be recovered from apoptosis. This point occurs at a time when there are observable changes in the cell membrane that can be observed by high-resolution optics. These changes can be observed up to 20h before cells stain with trypan blue. This allows dielectrics and high-resolution optics to be offered as a means of monitoring a bioprocess and maintaining cell viability over prolonged time periods.

Insect cells with recombinant baculoviruses constitute a well-established production system for viral vectors for gene therapy and vaccines. As a lytic production system with released proteases likely to affect product quality attributes, the ability to track culture progression is especially important. This talk will explore how monitoring can be used to detect deviations, improve process understanding and support the critical time-of-harvest decision, enabling an information-based approach that aligns with PAT principles.

Research involving culture of mammalian cells depends extensively on conditions devised in the mid-twentieth century that do not well-replicate the physiological microenvironment. There remains a huge research need for cell culture to expand fundamental knowledge of cellular functions and to facilitate preclinical drug discovery. Culture conditions that more closely reproduce physiological milieus are important to enhance biological relevance. I will discuss recent research investigating the effects of extracellular matrix and soluble secreted proteins, the secretome.

The application of intensified and dynamic strategies has widely been used for the improvement of bioprocess performance. In line with this, we present case studies where the application of an intensified seed train and dynamic feeding strategies led to reduced timelines, as well as higher product yields. The implementation of these approaches in combination with online monitoring technologies enables real-time process insights and holds the opportunity to facilitate bioprocess operation through automation.

One of the bottlenecks in the development of innovative biopharmaceuticals is found in endless process development and characterization times. Within this work, we compared classical Design of Experiments (DoE) with an intensified DoE approach applied on an E. coli fed-batch process. While the screening of a classical DoE required 29 weeks, the intensified characterization could be finished within 10 weeks while still uncovering the best process conditions. The generated model will further be used for the process control showcase.

Implementation of QbD/PAT concepts in biopharmaceutical production asks for advanced process control concepts to put the idea of in-process product quality control into practice. Today, a set of PAT tools is available that can be used to develop such concepts. In this talk, one example based on use of a hybrid modeling approach and a second showcase exploiting process information directly delivered by complex analyzers will be presented.

Efficient process development in biotechnology relies on the Quality-by-Design (QbD) paradigm. QbD is a scientific, risk-based proactive approach to drug development that aims to have a full understanding of how the process and product are related. Within this talk, we provide approaches and tools to apply this principle.

We uncovered numerous cellular genes that are overexpressed in CHO cell lines that produce high levels of therapeutic proteins. While some were the consequence of heterologous protein overexpression, others appeared to limit the expression and secretion of heterologous proteins. This was assessed by overexpressing these genes, resulting in increased expression of various therapeutic proteins. Interestingly, several CHO cell metabolic activities were simultaneously limiting in pathways as diverse as cell signaling, response to cellular stress, cytoskeleton organization, and lipid metabolism.

Adoption of perfusion in N-1 bioreactors yields a reduction in process time and/or increase in manufacturing capacity without increasing volume capacity. Due to recent advances in cell retention technologies, cell line and media development, perfusion becomes a more viable alternative to expand performance beyond what traditional platforms can attain.  This presentation explores the benefits of a pre-sterilized, high-throughput perfusion filtration systems in an N-1 perfusion application and demonstrates scalability from 3 L to 50 L.