Biopharmaceutical companies routinely rely on manufacturing platforms to accelerate process development. However, when an external asset is internalized, they come with inherently off-platform cell culture processes. This includes cell lines, media, production formats, and process control strategies with which the company may have limited experience. Furthermore, these off-platform processes may not fit internal manufacturing capabilities. This talk will highlight lessons learned from several recent integrations at Amgen, highlighting both the challenges faced, and solutions.

As the industry continues to develop processes for higher titers and cell densities, there is a growing need for more advanced single-use bioreactors that offer improved mixing, kLa, and CO2 stripping, while minimizing foam. When a new single-use bioreactor is purchased, it usually comes with proprietary controllers, representing a significant capital investment and complicating automation integration. A universal controller helps address these challenges by lowering initial CapEx and integration costs, maintaining user interfaces, accelerating implementation, increasing flexibility, and reducing cost.

In large-scale manufacturing of monoclonal antibodies, scale-up and facility changes can introduce variability that significantly impacts product quality. One case study investigates elevated osmolality linked to slower growth observed at large-scale production while the other examines increased afucosylation driven by elevated high mannose following the technology transfer to a new manufacturing site. Together, these case studies highlight the importance of leveraging small-scale studies to deepen process understanding and mitigate risks.

This talk summarizes the challenging dynamic to rapidly develop, scale-up, and transfer novel advanced process controls while maintaining legacy platform constraints of an existing vaccine franchise. Dictated by aggressive timelines, rapid development efforts focused on key unit operations with refined parameter optimization of a Saccharomyces cerevisiae fermentation process expressing target protein antigens. This phase-appropriate development strategy was complemented by state-of-the-art scale down systems to enable clinical manufacture at unprecedented speeds.

Recent advances in upstream processes have led to increased cell densities and productivities, resulting in more sub-micron particles that pose challenges to the clarification process. Traditional harvest strategies were inadequate to achieve scale-up needs. Alternative techniques were developed, including acid precipitation prior to harvest and pDADMAC treatment to enhance harvest process performance, along with various control strategies to mitigate product quality concerns. Therefore, a combination of centrifugation and a pre-harvest treatment show promise in meeting scale-up requirements.

The complexity of mammalian cell culture and the heterogeneity of large molecule products have historically limited the application of robotic automation platforms in production and characterization to mainly either early stages for small-quantity, stage-gate quality material, or later stages for industrializing specific task accomplishments. Here, we discuss our next-generation automation strategy for bridging the gap to prepare large-quantity of high-quality material, solving an essential need for more complex Biologics modalities.

Diamyd Medical, with our lead product candidate Diamyd (rhGAD65/alum) aimed at immune system modulation, is in phase III clinical development to treat patients with Type 1 Diabetes. We have developed a manufacturing platform for our unique protein antigen therapy and we strongly believe that this platform could be applicable for other projects. The talk will include a discussion around challenges such as establishing comparability during development.

Cell culture medium serves as a foundational element for robust processes to produce antibodies and various complex biologics. The industry has increasingly sought intensified processes that enhance productivity and simplify operations, all while aiming for lower production costs and improved product quality. To meet these demands, supplements for cell culture media have evolved significantly. In this presentation, we will demonstrate case studies that successfully enhance the CHO cell upstream processes.

Engineered producer cell lines have led to antibody yield increases driven by an improved understanding of the cellular machinery influencing cell health and protein production. In this study, I will describe the utilization of multiple culture longevity-prolonging strategies (chemical and process-related) to enable up to a three-fold increase in total antibody production as well as three-fold higher cell-specific productivity to further improve yields.

• Types of mechanistic, hybrid, or data-driven models that are most effective for optimizing upstream bioprocess parameters
• Modeling strategies that help align process understanding between sending and receiving sites
• Limitations and challenges of current modeling tools in capturing the complexity of upstream bioprocesses
• The role of digital twin technology in enabling real-time process optimization
  

The design of cell culture media and feeding strategies continues to evolve as the industry advances toward more intensified and robust bioprocesses. The growing demand to produce complex molecules and novel modalities introduces additional layers of complexity. This roundtable discussion provides a platform to openly share challenges related to media formulation and feeding strategies, as well as successful optimization approaches. It also serves as a forum for brainstorming innovative ideas and compiling a wish list for future technological or material advancements.

AI-driven technologies are revolutionizing bioprocessing, especially in cell-culture optimization and bioreactor scale-up. There is huge potential in the application of AI/ML methodologies in a variety of use cases such as deviation detection, process understanding and optimization, device control and recovery to reduce batch failures and to improve process outcomes both in development and production. The horizon for these innovations are digital twins. The latest advances in digital twins combine real-time data, physics informed simulations and AI-driven predictions to adjust and optimize bioprocess outcomes. However, there are non-trivial challenges faced in integrating AI into environments that require precision with a high bar for regulatory compliance.

In Upstream process development, reducing timelines and increasing flexibility of processes are crucial drivers in delivering medications to patients. The transformative potential of unique biologics has led to enormous diversity in development platforms and manufacturing technologies. To effectively deliver life changing substances, it has become paramount to navigate these compounding complexities using modern, data driven approaches. In this talk, we will discuss the potential of custom in silico modeling tools to simplify process development and enable more efficient knowledge transfer between manufacturing processes and scales. We will present hybrid and mechanistic case studies that demonstrate the power of building an in silico toolbox that compliments upstream development by enabling more efficient design of experiments and a more robust understanding of our processes.

The yeast-based fermentation process for recombinant protein vaccine production was improved through development of novel process controls and a design of experiments (DOE) study. To increase process robustness, novel controls were implemented for oxygen uptake rate and respiratory quotient. A DOE approach was utilized to evaluate optimal set points for these parameters and harvest timing. Through implementation of these process improvements productivity doubled with increased robustness.

Selecting optimal data-driven modeling methods and integrating batch and time-series data are key challenges in biopharmaceutical manufacturing. Accurate modeling requires capturing both inter-batch variability and intra-batch dynamics. This presentation describes the application of smart process data analytics software and tensorial methods to industrial end-to-end biomanufacturing datasets for monoclonal antibody production, demonstrating how automated DA/ML tool selection and multiway modeling improve predictive accuracy and process understanding.

The biopharmaceutical industry is increasingly turning to data science to optimize manufacturing processes and enhance product quality. This presentation explores two innovative case studies that leverage data science to improve efficiency in biologics manufacturing. The first case study focuses on a cloud-based predictive modeling application designed to enhance the predictability of mammalian cell culture processes by forecasting bioreactor potency from at-line process parameters over a multiple-day horizon. The second case study presents a novel application for assessing the out-of-specification risk associated with drug product potency. The cloud-based statistical software application streamlines the evaluation of alternate potency targets.

While the in vitro reaction for RNA synthesis is traditionally performed in a batch mode, fed-batch in vitro transcription holds promise to more efficiently use expensive catalysts and agents for co-transcriptional capping. We develop and mathematically optimize a mechanistic model for IVT to find feeding policies that maximize RNA production while maintaining substrate ratios essential minimizing formation of uncapped RNA impurities. Experimental validation demonstrates how these model-based strategies can be used to accelerate process development and optimize costly resources.