If innovative therapeutics are to reach patients more quickly, at least three functions need to be fulfilled more effectively. Fortunately, progress with respect to two of these functions—target identification and drug development—is already evident, thanks to new technologies such as functional genomics, computational design, and high-throughput screening. That leaves the third function: manufacturing. Although manufacturing is also benefiting from new technologies, it has yet to realize the grand plans described by Industry 4.0 advocates. Somehow, the biopharmaceutical industry lags other industries in implementing “smart” technologies, that is, technologies that set up virtuous cycles in which improvements in the digital and physical realms reinforce each other.

And yet the pieces are in place to support a digital transformation of biomanufacturing. These pieces include tools for creating virtual twins for modeling processes and facilities; tools for streamlining both development and production; and tools for enriching, connecting, and deriving value from varied data sources. So, what’s the holdup? Biomanufacturing, like the biopharmaceutical industry as a whole, is highly regulated and steeped in a paper culture.

Nonetheless, many companies are taking on the challenge and transitioning, piece by piece, to a new process development and manufacturing paradigm. Open to sharing information on what it takes to turn piecemeal solutions into highly integrated solutions, process development and manufacturing experts met at Informa’s BioProcess International Conference last September to discuss the steps they are taking to accelerate the introduction of innovative therapeutics.

Designing (and executing) a plan

A digital transformation’s pieces can include robotics platforms, paperless batch records, enterprise-wide information systems, and even artificial intelligence. But which pieces come first? And how can they be put together? At the conference, these questions were addressed by Janani Swamy, global head of technology transfer programs, Sanofi.

“Know when to engage your organization,” Swamy said. “You want to ensure that what you deliver will make their lives easier. Be prepared. First deployments are not perfect and require refinement. Be agile during deployment and develop the right check points to course correct as needed.”

Swamy cautioned that an immediate focus on cost savings can shortchange a digital transformation’s overall value. Indeed, she recommended that early considerations should include safety, quality, and right-first-time development practices. When these elements are in place, cost savings will follow.

The pharmaceutical industry is highly regulated and reliant on GMPs. Over time, it has fashioned a culture of complexity that generates a lot of paper and resists change. Swamy said that if this culture is to change, it may be necessary to rethink assumptions that are part of the industry’s conventional wisdom. For example, she suggested that the industry should focus on adding value and manufacturing high-quality medicines, not delivering paper.

Moreover, data needs to become more actionable. “We need to go from an industry that collects a lot of data to leveraging that data to improve performance,” Swamy insisted. She also outlined how a digital transformation may progress in stages. Initially, a core system can be set up to focus on a few tasks, such as creating electronic batch records, implementing laboratory information management systems, and digitizing standard operating procedures. Over time, the core system can be enhanced to extend applications and take on additional tasks such as data mining.

One may start with systems for monitoring deviations and line losses, Swamy suggested. Eventually, these systems may become integral to digital twins. “The key,” she advised, “is to put the foundation in place before adding more sophisticated tools.”

Finally, Swamy noted that organizational change should be anticipated and understood prior to implementation because climbing the “change curve” requires investment and training. “There is a lot to be done even with the fundamentals,” Swamy remarked. “The vision is easy to describe, but practically speaking, it takes investment and commitment. Taking a complex topic and making it simple is the key to digital transformation.”

Modeling virtual facilities

In the early 1990s, a consortium of biopharmaceutical and other companies approached Purdue University for help in managing manufacturing complexity. The request resulted in the development of VirtECS, a general-purpose manufacturing modeling software. By 1993, it had been commercialized by Advanced Process Combinatorics.

VirtECS software is used to develop high-fidelity models of individual manufacturing facilities. It helps operations select the most effective engineering projects, design and retrofit facilities, and maximize throughput.

“While traditional technology gives you a path based on a rule of thumb, VirtECS takes specified performance metrics and evaluates all the possible ways to schedule a plant to home in on the best options,” explained Joseph Pekny, PhD, the CEO of Advanced Process Combinatorics, and a professor of chemical engineering at Purdue University. “If you can help people manage complexity, you can squeeze more medicine, food, or products from the process.”

VirtECS uses “augmented intelligence” to deal with NP-Complete problems, that is, planning and scheduling problems for which efficient algorithmic solutions do not exist. A famous NP-Complete problem is the Traveling Salesman Problem. Solving it involves determining how a number of tasks may be performed sequentially in the most efficient way possible. “These problems are very difficult to solve even though they sound simple,” Pekny said.

VirtECS can help users react to information about processes in real time, facilitating information exchanges in enterprise-wide systems and improving manufacturing activities such as finite capacity scheduling, production planning, process modeling, capacity analysis, and debottlenecking.

Although Advanced Process Combinatorics indicates that VirtECS can eliminate spreadsheets and simplify process management, the company recognizes the importance of human oversight. “Modeling is a frame of mind that requires acknowledgment that you have a problem,” Pekny noted. “Investment is required, but the payback time is short.”

To broaden access, VirtECS Symphony can be used to publish plans on an intranet. It contains a comment section that supports ongoing dialogue. Comments can be synchronized with the VirtECS Scheduler interface so that production planners have access to plant floor feedback when they are ready to update their schedules. “This VirtECS-enabled social media phenomenon has taken on a life of its own,” Pekny asserted.

Shortening process development timelines

Shorter timelines are driving companies to strategically implement digital tools. “Patients need medicines as soon as possible,” said Cécile Brocard, PhD, director of downstream development, biopharmaceutical development operations, at Boehringer Ingelheim’s regional center in Vienna. “That means we have to develop products faster.”

Brocard works with non-platform microbial products. Doing so is challenging because each product needs a unique process.

To model new processes and optimize existing ones, Boehringer Ingelheim developed the AI-assisted Smart Process Design platform. The software is used to predict the behavior of the process steps and ensure a fit into the production facility. “In the production plant, you have to deal with limitations in space, tank volumes, and other parameters that you may not have considered during development,” Brocard noted.

The software can run simulations, perform analyses of multivariate data, identify bottlenecks, and indicate process modifications that could improve production. Brocard’s team uses Smart Process Design to identify critical parameters (such as flow rates) and to assess how readily laboratory processes may be turned into production processes.

According to Brocard, the Smart Process Design platform implements a toolbox approach. “We have plasmid and expression toolboxes,” she details. “We have also robotized a high-throughput approach.” Pipetting robots are used to run experiments in parallel. For example, 32 miniature fermenters can be run in parallel to screen for conditions or strains with specific characteristics.

The toolboxes and robots are part of an automation strategy that is employed to define methods for downstream purification. Binding and elution properties of resins are tested in parallel miniaturized form in multiwell plates, and then in small RoboColumns, before verifying the conditions on a bench-scale column.

In Escherichia coli platforms, cytoplasmic or periplasmic proteins may be extracted from the cell lysate or inclusion bodies. “We are highly experienced with inclusion bodies,” Brocard pointed out. “But these aggregated proteins are only partially active and require a chemical reaction to refold and assume their native state.” The pipetting robots facilitate monitoring. The resultant data are put into the Smart Process Design software to model processes and determine optimal conditions.

Digitalization provides access to more data, enables more data analysis, and reduces the number of wet lab experiments. “To speed up development timelines, ultimately, we will convince the regulatory authorities that in silico data are sufficient for validation to demonstrate that you can control your process at each critical step,” Brocard stated.

Boehringer Ingelheim operator using a rotary tablet press for clinical trial manufacturing in pharmaceutical developmen
Shorter timelines mean companies have to develop products faster. Boehringer Ingelheim has implemented many transformative tools such as modeling software, plasmid preparation and gene expression toolboxes, and laboratory automation systems to accelerate process development and facilitate the transfer of products between processing units. This image shows an operator using a rotary tablet press for clinical trial manufacturing in pharmaceutical development in Biberach, Germany. [© Peter Ginter]

Making the transition

The digital transformation of biomanufacturing demands a new mindset, one that countenances compliance with FAIR (findable, accessible, interoperable, and reusable) data practices and the integration of various enabling technologies, suggests Nandkishor Nere, PhD, director/senior principal research engineer, AbbVie. (Nere is also a research fellow at the Center of Excellence for Isolation and Separation Technologies.) According to Nere, integration efforts should encompass classical information technology, automation, data science–based predictive modeling, and process development.

With respect to downstream processing, Nere discussed the importance of technologies such as chromatography and tangential flow filtration (TFF). For example, he noted that reductions in solvent usage and increases in process yield can be achieved if continuous chromatography is implemented in combination with first principle–based predictive modeling. He also observed that a model-based process understanding of TFF in combination with the implementation of custom automated platforms not only increases the efficiency of process development but also imparts robustness to process design and operation.

“For more than a decade, AbbVie has developed and deployed the infrastructure for digital technologies through the Cross-functional Modeling Forum, or CMF,” Nere pointed out. He suggested that the CMF has encouraged a mindset change and supported a gradual buildup and usage of FAIR data. “This approach,” he added, “has addressed classical challenges faced by many due to step-change transformation.”

Finally, Nere stated that the full potential of digitalization could be realized if digital technologies were developed and deployed holistically and integrated seamlessly with traditional process development and manufacturing practices. He also noted that the use of digital technology to replace routine and often time-consuming activities will increase in the near future, along with the implementation of broader process innovations.

Overcoming challenges

Digital and automation technologies can accelerate the product development life cycle by removing the reliance on paper as a primary storage location for information and manual knowledge transfers, said Kristen Manchester, director of integrated process development, Cytiva. These tools can provide other benefits. For example, simulation software can test a range of process conditions without increasing the number of experiments performed, helping with risk reduction. Also, relying on automation can increase productivity, and improving access to data can support better decision-making.

The first-generation digital and automation tools that are being deployed offer various capabilities. According to Manchester, hardware for laboratory and manufacturing facilities can be programmed to execute recipes; electronic laboratory notebooks and manufacturing execution systems can capture details about experiments and production runs; and laboratory information management systems can analyze results. Tracking inventory through enterprise-level software and handling critical documentation through electronic quality management systems can extend the paybacks.

“But there are challenges for end users and vendors as the industry focuses on the broader implementation of digital technology,” Manchester remarked. “Leading cultural change to accept new ways of working is crucial—as is driving technology efforts without missing a beat on advancing pipeline therapies.”

The deployment of new technologies and tools may necessitate staff development or hiring to meet evolving skillset needs. “You need to time your investments, and you need to navigate regulatory expectations,” Manchester noted. “You also need to ensure data integrity and security across integrated equipment and software from multiple vendors.”

Vendors must also navigate obstacles. Often there are unclear user requirements and prioritization of use cases to identify high-priority problems. The lack of standards/ontology and variations across customers/sites/products/modalities make it difficult to develop universal solutions. Plus, changing systems is complex, and historical information captured on paper or in legacy systems must be digitized and rationalized.

Technology roll-out to a network of customer sites is complicated, especially when it may involve external contract research organizations and contract development and manufacturing organizations. “The overarching goal is to deliver safe, effective, affordable, and sustainable therapies that improve human health,” Manchester pointed out. Although she acknowledged that areas of risk and potential new requirements must be addressed, she expressed optimism that digital technologies are converging to form a “navigation system” that will guide and accelerate future drug development and manufacturing.

Previous articleAccelerating Recombinant Protein Vaccine Development
Next articleThe Juice from J.P. Morgan: Takeaways from San Francisco