The Connected Process

In the early days of biomanufacturing, companies focused primarily on approvals, with process optimization being a much lower priority. This remained the case while the prices of biologics far outstripped the costs of manufacture, but in the age of biosimilars and cost-conservative payers, companies are now re-examining cost drivers in biologics manufacture. “Today, manufacturing is a strategic imperative and an area for differentiation – an opportunity to look at new technologies that can really drive down costs,” explains Michael Phillips, Director, Next Generation Bioprocessing R&D, at MilliporeSigma. “Companies are now focusing on how to get more out of their processes, how technologies, such as single use, can help, and looking to implement in-line sensors and analytics. Increasingly, companies are also looking at the potential to connect bioprocesses.”

Here, Merrilee Whitney and Jonathan Souquet described next generation bioprocessing as continuous, connected processes, but what is a connected process? In traditional biopharma manufacturing, each operation is performed in isolation from other operations in the manufacturing process. One group manages the production bioreactor, another the clarification step, another the chromatography, another the polishing, another the virus removal, and so on.

“Connected processing is about running adjacent unit operations as a single unit, which results in improved efficiencies,” says Phillips. “For example, upstream processes have been made more productive by replacing fed-batch with perfusion operation, and downstream bottlenecks can be addressed by employing multi-column Protein A chromatography. But how can you improve productivity when you have reached the limits of engineering? At this point, “adjacency” makes sense. The logical outcome is a fully continuous manufacturing system, in which all unit operations are connected to their neighbors, or
run simultaneously.”

A connected process would reduce operational footprint – by making holding tanks redundant for example – but there is still a big question about how processes should be connected. Continuous and connected processes demand communication between unit operations, so that any process can adjust for variations in preceding processes. “The industry is going to need very reliable sensors and process analytics, which will be a challenge,” says Phillips. “Ultimately, however, I believe that a shift to connected processing is inevitable and part of the evolution of biopharma.”

Regulatory disconnects

Change is rarely straightforward, and a move to connected processes in the heavily regulated biopharma industry will certainly create challenges. Janmeet Anant, Regulatory Advocate at MilliporeSigma, explains, “Right now, connected systems for large molecules are new to regulators, but I am optimistic. Regulators understand that there is a lot of variability in biomanufacturing processes and are encouraging the industry to develop innovations that can produce more consistently high quality products.”

Phillips adds, “From the presentations and discussions I’ve heard, I’d say that regulators are trying to be very accommodating. We have also seen some encouraging progress, such as regulatory approval for Janssen to switch from batch to continuous processing for an already approved small-molecule drug. Amgen has also implemented connected processing for a biologic. A continuous bioreactor expression (i.e. perfusion) has been approved by regulators to commercially manufacture 23 different molecules.”

But according to Anant, one of the biggest regulatory hurdles is that regulators are not always well-versed on innovative connected or continuous manufacturing approaches, unlike their extensive experience with batch processing. It is difficult for manufacturers to know what data will be required by regulators to prove that a connected process delivers products clinically equivalent to those manufactured by traditional approaches. To ease the process, Anant has highlighted that experts within MilliporeSigma will guide manufacturers in terms of validating and comparing processes, which is one core element of regulatory acceptance. “Typical questions that manufacturers need to answer include, how long can continuous operations be sustained before equipment wear and tear becomes an issue? How should validation testing be changed to accommodate the new process? In non-connected systems, you can validate the bacterial filter by challenging it with a pulse of micro-organisms; how should this approach be modulated in continuous processing?” says Anant. “Similarly, where validation technology has improved – for example, virus inactivation processes once required several hours but can now maybe accomplished in a few minutes – how can we persuade regulators to accept these advances?”

The industry, including its regulators, are going through a learning curve regarding continuous, connected processes, but the end result should be more consistently high quality products. For these innovative manufacturing approaches, regulatory compliance needs to move beyond checklists towards an emphasis on product and process understanding and control based on sound science and quality risk management. A compliance-based checkbox exercise will not work – evidenced by the fact that many companies still experience quality problems, despite meticulously checking off their checkboxes. “Between now and the next five years, the industry will look to make connected processes more continuous, which will involve a true understanding of processes and variability – in other words, quality by design,” says Anant. “This will also benefit the development of advanced online monitoring techniques, since you will know what to measure and how it affects the final drug product. Anticipating the future era of connected bioprocessing, regulators are now looking to expert organizations, vendors or suppliers that can develop suitable analytical process technologies.”

For Phillips, there are still some regulatory questions that need answering, such as the definition of ‘batch’ – in the context of perfusion systems. “Is the batch the entire perfusion bioreactor run of, say, 60 days? Or do you cut it into smaller units based on time?” asks Phillips. “We still need consensus about this. Another ongoing issue is the lack of harmonization between different regulators. Clearly, there is less incentive to change when process alterations will require the manufacturer to satisfy several sets of guidelines.”

Breaking barriers and building links

As well as regulations, companies also face internal challenges when they want to connect processes. “People are impeded by the traditional paradigm of process development,” says Phillips. “In current practice, process development typically involves the independent optimization of each unit operation. The implementation of connected processing will necessitate a more holistic approach to process development. Optimizing the conditions for one unit operation may make subsequent operations less efficient, so process development experts will need to think about more global optimization at the whole process level that will require an understanding of all unit operations. Good communication and collaboration between the groups responsible for different unit operations will be essential; upstream experts will need to understand the language of downstream experts, and vice versa.

“Also, don’t forget training,” adds Phillips. “For example, optimizing a multi-column chromatography operation requires a different thought process compared to the optimization of a traditional single-column approach. Process development scientists and manufacturing operators will need direct exposure to these new technologies before they would be willing to adopt them.”

The potential for significantly improved process robustness and process economics definitely justifies efforts in uncovering opportunities for connected processing and more holistic process optimization. Improvements can be significant and are often unanticipated. For example, a typical downstream process optimized at the unit operation level may require upward of 10 to 15 different buffers. For a process optimized more holistically, Phillips has seen instances where the total number of buffers has been decreased to only 4 or 5. “You may sacrifice some efficiency in one operation, but gain a more efficient overall process. As always, there is usually resistance to change, but change is usually required to perform at a different level.”