Send to printer »

Feature Articles : Oct 15, 2012 (Vol. 32, No. 18)

Confronting Scaleup Challenges Head On

  • Angelo DePalma, Ph.D.

Despite new heights of process characterization and knowledge, scaleup still can befuddle the most experienced of bioprocessors.

Wei Chen, Ph.D., chief scientist at Phage Pharmaceuticals, believes that a disconnect between early-stage R&D and manufacturing—a defect that has persisted since biotech’s earliest days—is one culprit.

“Companies have come a long way in streamlining clone selection and expansion, but this activity more closely resembles basic research than manufacturing,” he says. “Cells at this stage have almost no chance of surviving in a large culture in the manufacturing environment.”

The analogous activity in small molecule development, the handoff of an investigatory synthesis route to process development, suffers from far fewer sharp breaks.

Dr. Chen blames both development “silos” and the complexity of biological systems for the discontinuity.

“Particularly as companies grow, the various activities like early clone selection, cell-line engineering, expansion, scaleup, and production become compartmentalized. At larger scale many more variables are operational,” he explains.

While small molecule development teams may re-work a synthetic pathway for lower cost or scalability, scaleup becomes a straightforward chemical engineering problem once the manufacturing process is set. In biotech a cell’s response to media and feed, dissolved gases, waste products, and the myriad stresses of the bioreactor environment are far more complicated than heat and mass transfer in chemical reactors, particularly at significantly different scales.

Reverse Engineering

Dr. Chen suggests an approach involving reverse-engineering the larger process by getting more large-scale parameters into the picture very early in development. “But in most large biopharmaceutical companies this is next to impossible, as cell biologists lack manufacturing expertise and process development and manufacturing specialists know little about cell biology,” he points out.

The other issue Dr. Chen mentions as hampering scaleup is the lack of industrialization and standardization of bioprocesses for therapeutics. While we hear of “platform” processes for certain expression systems and products (e.g., CHO cells and monoclonal antibodies), experts still view many cell culture processes as unique protocols. Hence the intellectually suspect adage, “The process is the product.”

This philosophy seems to repeat in history. Commercial wine-making began with a similar philosophy based on hand-made, empirically individualized recipes. “But today all top wine-makers have standardized their processes, so that the quality of the product can be maintained,” says Dr. Chen. Bioprocess scaleup’s success depends on the integration of biology into engineering.

Serialization vs.Synchronicity

Introducing new processes into a busy, established facility poses significant logistical and technical challenges, according to Gregory R. Naugle, director of process development at Amgen’s West Greenwich, RI, production facility, the largest in the Amgen manufacturing network. Key issues involve ensuring the new product’s success while not rocking the boat with respect to existing processes.

Tech transfer of multiple products at differing stages of development and scaleup add to the complexity, maintains Naugle. “A high run rate facility runs continuously, and provides very little ‘white space,’ with very little free equipment. There are limited windows of opportunity to modify equipment to accommodate the new product or process.”

The top priority should be maintaining the fidelity of the currently licensed process. “When you modify shared equipment, you want to ensure you’re not altering the process that’s already there,” Naugle adds. “In essence, you want to change something without changing it. If the modification involves hardware, for example piping or the diameter of a transfer panel, you must assure there are no unintended consequences to the existing product.”

Synchronicity is what complicates tech transfers into busy plants. Under circumstances where projects enter sequentially, one is able to conduct a full “lessons learned” and still have time to prepare for the next project.

“When they’re happening simultaneously or nearly so, you don’t have the decompression time to do a lessons learned, implement those learnings, and move forward,” says Naugle. “You need to transmit that knowledge very rapidly.”

Again, the confounding variables concern the fact that cells are living organisms that are significantly affected by environmental factors.

Moreover, with high-run facilities, the operational distinctions between scaleup and tech transfer become more semantic than substantive. Both involve freeing resources for the new process and assuring the integrity of existing processes. “It’s just the nomenclature we choose to use,” Naugle, says. “If we scale up from a pilot facility in a different building we don’t consider that a tech transfer because of the co-location. But if the process were coming in from Amgen headquarters at Thousand Oaks, it would be considered a tech transfer. So yes, scaleup presents many of the same challenges as a full tech transfer, particularly when you’re talking about maintaining the fidelity of the existing process.”

Scaling Productivity at Similar Scale

At times bioprocessors seek to scale up productivity without necessarily expanding facilities or equipment.

“In the past bioprocessors would improve cells first, then work on media and feed regimens to get to where they wanted to be,” observes Peggy Lio, senior process science fellow, Gibco® PD-Direct Bioprocess Services, Life Technologies.

Scaleup involved “throwing the process over the fence” to the next link in the development chain, which employs its own media. “But now we look at this exercise more holistically, integrating both media/feed and cell improvement. You can have a cell line that works well, but without addressing media and feed early on, you won’t achieve true optimization.”

This approach is based on the idea that one can never be certain that cells have been optimized if media and feed are not.

Lio’s group incorporates improved media and feed from the earliest stages, using a platform that is appropriate for cell-line development. The challenge as the process moves forward is adopting media and feed that are appropriate for dissimilar development stages.

“Media used in production has certain characteristics that are not always suitable during cell-line development. The trick is finding a basic strategy that works with both,” she says.

Lio’s notion of “flexibility in scaleup” refers to the ability to improve the titer of established manufacturing processes in situations where changing unit operations, scale, or facilities is impractical or impossible. Processors often desire to obtain the most out of a process without implementing extensive changes that may result in downtime or regulatory scrutiny.

Lio says that in some instances this may be achieved through “functional additives.” She was unable to elaborate on what these consisted of due to the proprietary nature of the cell culture media business. She would say, however, that they are “not exotic components” that might add to cost of goods. “They may be components that are already in the medium.”

More “Interesting” than Cell Culture

According to David Greene of New York-based biopharm design consultancy Greene Associates, scaling up microbial fermentation is “more interesting” than mammalian cell culture.

“Yields have become so high in cell culture that scaling up the culture is no longer the problem. The bottlenecks now occur in recovery and purification, which have not kept up with upstream processing.”

Peggy Lio concurs. “We’re not really having problems achieving high titers any more. Many well-developed processes already exist for that. The question now becomes how to fine-tune older platforms and processes.”

Microbial cultures are more problem-prone because of their large size. As rising titers have made capacity issues in cell culture processes almost moot, microbial fermentations are still routine at the 20,000 gallon (80,000 liter) scale.

The sheer size and densities of these cultures create issues in mass and heat transfer that do not exist with mammalian cells. Besides smaller scale operation, mammalian cultures tend to be less biomass-dense than microbial cultures.

“Mammalian cells take in a lot of oxygen and don’t release a lot of heat,” says Greene.

Removing heat from huge microbial systems, by contrast, presents significant problems. At some scale, water jackets cannot remove all the heat. Cooling coils can help, but introduce a cleaning issue. And with both techniques, cells may not withstand the temperature differential between the fermentation and cooling system. Microbial systems also use prodigious quantities of compressed air for oxygenation and agitation. Large fermentations must deal with treating the off-gases and high agitation rates as well.

The comparison between mammalian and microbial cultures comes down to size. The former efficiently produce materials of extremely high value such as monoclonal antibodies; the latter manufacture commodity products such as industrial chemicals, antibiotics, enzymes, and amino acids, hence the larger culture volumes.

New Process, Old Issues

Scaleup issues for influenza vaccine produced through the egg inoculation method were resolved decades ago. The introduction of vaccines manufactured through cell culture introduces scaleup issues that are new to this industry.

To meet these new challenges, Protein Sciences relies on a QbD approach to large-scale vaccine manufacturing that claims to produce vaccines at large scale in fewer than 50 days from identifying the target virus.

“Manufacturing a recombinant subunit protein for influenza prevention poses unique challenges, as regular modifications to the antigen are required to protect against circulating influenza viruses,” explains Robert R. Boulanger, Ph.D., manager of USP development at the company. Dr. Boulanger has developed a series of standard operating procedures based on his experience with the production and purification of dozens of recombinant human agglutinins (rHAs) through the baculovirus expression vector system (BEVS).

“This approach expedites and streamlines our response to an influenza strain change announcement,” he says.

It involves utilizing processes employed for a similar rHA, typically from the same influenza virus subtype H1, H3, or B. If the process produces an acceptable antigen, the new antigen is produced at manufacturing scale. If not, the standard operating procedure defines the process design space.

“This guides the development team through a series of systematic experiments to evaluate specific conditions and identify required process changes for manufacturing batch records. Once the new process is established, the antigen produced is evaluated to verify it meets all acceptance criteria limits,” Dr. Boulanger explains.

Implemented properly, QbD improves manufacturing efficiencies, reduces costs, and facilitates the implementation of process changes and improvements.

“The universal process we have developed for our seasonal influenza vaccine, FluBlok®, is a good example of how QbD reduced our timeline from identification of a new rHA sequence to manufacturing at the full cGMP scale,” continues Dr. Boulanger.

Protein Sciences follows this strategy for PanBlok®, the vaccine supported by the Biomedical Advanced Research and Development Authority to achieve pandemic flu preparedness. Like the FluBok program, PanBlok is based on the company’s prior experience with rHA antigens and flexibility in accommodating the physicochemical properties of new rHA antigens.

The company has also implemented a process monitoring and a continuous improvement program.

To illustrate the success of this approach, on February 28, 2012, an FDA advisory panel announced two strain changes for the 2012–2013 seasonal influenza vaccine.

“We utilized a QbD-based influenza vaccine strain change approach and completed the process from cloning of the new rHA for the A/Victoria/361/2011 sequence to completion of the first 450 L cGMP lot in 50 days,” says Dr. Boulanger.