September 1, 2016 (Vol. 36, No. 15)

In Bioprocessing, You Can Move Out of Your Comfort Zone and Push the Limits of Productivity

In 2004, the U.S. Food and Drug Administration issued an official guidance on process analytic technology (PAT), to encourage its voluntary adoption as a means of achieving quality by design (QbD). In theory, PAT and related instrumentation facilitate the design and development of well-understood and -controlled manufacturing processes resulting in predefined product quality. 

“PAT methods may form part of a QbD approach to the development of bioprocess control strategies,” says Nick Hutchinson, Ph.D., technical content manager for bioprocess solutions at Sartorius Stedim Biotech.

Moreover, the adoption of disposable processing technologies by contractors and innovator firms spurred development of sensors in single-use formats. “Initially this presented something of a challenge,” Dr. Hutchinson continues. “However, novel single-use sensors are becoming increasingly available for both upstream and downstream applications.”

For example, the integration of in-line biomass sensors into single-use bioreactors leads to enhanced monitoring and control strategies while in-line disposable UV sensors can monitor protein concentration to support control of downstream ultrafiltration steps.

“By linking process measurements to advanced automation strategies,” explains Dr. Hutchinson, “we can not only increase consistency and process robustness, but also ensure we control processes within design spaces that deliver product with the required quality attributes.”

Scientists at Amgen have recently shown how the collection of chromatography pools based on purity measurements via online HPLC helps prevent lot rejection. Similarly, monitoring of nutrients and metabolites in cell cultures allows for the fully automated control of glucose feeds in real time by connecting the analysis system to local automation such as programmable logic control (PLC) and supervisory control and data acquisition (SCADA) systems.

“Integrating sensors with automation platforms to ensure process consistency will greatly expedite the adoption of continuous processing by the biopharmaceutical industry,” Dr. Hutchinson predicts. “The drive to accomplish this is likely to herald a new wave of innovation in sensor technology.”

More sensors equals greater complexity and, inevitably, more information streaming from processes to analysis and control systems. Multivariate data analysis techniques help make sense of bioprocess data generated across scales, and converts that information into process knowledge and understanding. Bioprocess engineers must be able to retrieve data reliably from both miniaturized experiments (for example, from the Sartorius ambr® systems and others) that today’s developers use to define design space and to predict performance of process runs at intermediate to commercial scales.

“This is a significant ‘big data’ challenge,” notes Dr. Hutchinson. “Currently, it is being addressed by the industry through knowledge management tools.”

Single-use sensors are built into disposable process containers, which together are gamma-irradiated and delivered to the customer as a sterile system. Challenges for single-use sensors include reliability, standardization, and consistency. According to Thermo Fisher Scientific, a producer of single-use sensors, research into quality gaps could help to deliver an ideal product to the market.

Biomass Monitoring at Scale

At-scale operation, however, has been an issue, although some advanced sensing modalities are breaking through. Capacitive biomass sensors from Aber Instruments have been adopted or evaluated by most of the large biomanufacturers, from development through cGMP production. But Aber’s commercial capacitive biomass system, Countstar, is marketed only to breweries in South America, according to Aber’s marketing director John Carvell, Ph.D.

Sensors that reliably operate across scales, including “design of experiment” (DOE) volumes of parallel micro- and mini-bioreactors, will become increasingly relevant. For example, Aquila Biolabs’ Cell Growth Quantifier for microbial fermentations measures biomass in specially designed shake flasks. The system uses an optical sensor that reads cell density through the bottom of the flask. Conventional biomass measurements employ lengthy manual sampling and photometric optical density measurements.

Even simpler is a handheld biomass sensor from BugLab, which according to company literature measures cell density in six seconds at up to 30 optical density units.

For cell culture biomass determinations, options include the Countstar system from Aber (and sold by Applikon Biotechnology). The company has published data showing a greater than 99% correlation between Countstar measurements and conventional determinations of CHO biomass through hemocytometry. Countstar is also available for microbial cultures.

Other companies, including Hamilton (through its acquisition of Fogale Nanotech), provides sensors suitable for measuring viability of small-scale cell cultures used in biomedical research.

Online Sensing and the Bottom Line

The integration and application of sensing technologies is becoming critical to bioprocessors. “Companies are driven by QbD,” says Ken Clapp, senior manager for applications, integration, and technology, GE Healthcare Life Sciences. “As a consequence, they apply sensors to help achieve that quality, lower costs, and improve their bottom lines and the availability of drugs.”

Ultimately, it matters little whether a clean-in-place or plastic platform is used. “The need,” Clapp insists, “is the same—to get the most out of bioprocesses.”

Measuring intermediate metabolites or substrates on which cells grow are gaining importance as well, since controlling these parameters lead to higher quality and greater yields. Many sensing technologies have emerged but are of little use without demonstrable robustness at production scale. In that regard sterilization remains a barrier for in-process sensors, and noninvasive sensors are highly desirable.

Sensing and monitoring will gain even more prominence as continuous bioprocessing becomes more commonplace. “With perfusion cell culture lasting weeks, even months, as many as one to one-and-a-half bioreactor volumes are run through the bioreactor every day,” Clapp observes. “You get a lot of protein, but productivity must be measured against media costs. If sensors could show the way to reducing media consumption, bioprocessors will save money and obtain more and higher-quality protein.”

Plastic/Steel Integration

Interoperability between disposable sensors and fixed-tank bioreactors is still limited, even though it would offer several advantages. Besides lowering overall costs and enhancing flexibility in operations, such interoperability could facilitate the cross-platform transfer of PAT-related practices.

An advocate of such interoperability is Surendra Balekai, senior global product manager, Thermo Fisher Scientific. He has argued in favor of standardizing single-use sensors and developing the means to apply them to stainless-steel bioreactors and fermenters.

In the August 2016 issue of GEN, in an article entitled “Plastic or Steel, Bioreactors Aim to Put Quality First,” Balekai pointed out that single-use probes could, in principal, be used in fixed-tank systems. In most cases, however, they lack the appropriate physical dimensions. Making them fit would require reworking the bioreactors’ steam-in-place (SIP)-tolerant plastic ports, and demonstrating equivalence for the sensors themselves.

Nonetheless, as Balekai notes here, interoperability remains an attractive prospect: “Manufacturers with large fixed-tank installations would readily use the same analytics in steel as in plastic, which would enable easier method transfer in production, as well as for ancillary operations like mixing.”

This kind of plastic/steel integration will likely remain out of reach until standardization efforts mature. Analyte sensors (that is, sensors for media components such as glucose and lactate) and sensors for osmolarity are generally unavailable for stainless-steel bioreactors. In any case, such sensors are of the offline type and hence unsuitable for PAT. “Some companies claim to have them, but we’ve seen reliability issues,” comments Balekai. “They are not in single-use format.”

This image shows the installation of probes and tubing on the Xcellerex bioreactor, a single-use stirred-tank platform manufactured by GE Healthcare Life Sciences. The company notes that sensing technologies should demonstrate robustness at production scale to ensure they contribute to higher quality and greater yields.

Trickle-Down Analytics

Underdeveloped standards, limited interoperability, and questionable reliability—these adverse factors tend to dampen enthusiasm for PAT and QbD. Nonetheless, PAT and QbD are often featured in conference programs and trade publications. For example, Chantal Cazeault, the executive director of product quality at Amgen, has spoken on the need for PAT in continuous processing (itself a worthy topic), noting the lack of “guidance specific to continuous bioprocessing.” Many similar presentations have been delivered by experts affiliated with best-in-class companies.

One wonders, however, about the true depth of interest among rank-and-file companies, whether innovation has actually trickled down to them from leading companies. In theory, these initiatives improve process understanding. In practice, their investigation and deployment may impose resource costs that are unlikely to be recouped. Although some savings may be realized, they may not even matter due to “perfection pricing” for biopharmaceuticals.

Bioprocess understanding would at least be more approachable if sensors were more reliable and their readings more reproducible across scales. “Scalability and consistency at different scales remain issues,” Balekai observes.

Single-use technologies are already enjoying significant adoption in downstream processes. Presumably, these technologies are helping to satisfy production demands, which include reliability and the implementation of PAT. Still, agreeable single-use trends could be reinforced with the development and implementation of better sensors.  Ideally, manufacturers would strengthen their regulatory status by clarifying production hurdles and identifying process efficiency and quality gaps. “But you still have to work at it,” Balekai insists.

What’s Next?

In bioprocess sensing, two of the more critical parameters are pH and oxygen. “Online measurements of these two parameters allow control of culture conditions, and provide an excellent overview of culture status and performance,” says Gernot John, Ph.D., director of marketing and innovation, PreSens.

Online bioprocess sensing, however, does not stop here. Determinations of parameters such as carbon dioxide are highly desirable. “In cell culture, carbon dioxide is so important that it could even surpass pH as an indicative parameter,” notes Dr. John. “This has been discussed for a while and can be verified in real processes only after reliable sensors become be available.”

Optical sensors are gaining greater significance in bioprocess monitoring. For example, in the monitoring of oxygen levels, bioprocessors are replacing Clark-type electrodes with optical sensors. Optical sensors are also being used more frequently in the monitoring of pH, if only in selected applications. “Optical pH measurement,” observes Dr. John, is “primarily utilized in single-use systems where it offers contactless pH monitoring.” In general, however, optical sensors are unlikely to overtake classic pH electrodes, which provide a much broader measurement range.

Optical sensors are also becoming more attractive through systems integration efforts. For example, it has recently become feasible to connect optical sensors to standard controllers through conversion devices such as the Optrode Dual from PreSens.

Expanding the pH measurement range is a current development goal. “PAT encourages the application of more online sensing technologies, but it also stipulates the measurement of chemical parameters,” Dr. John notes. “The online determination of product titer is what users dream about—so there is plenty of room for further research and development.”

PAT and QbD have provided great impetus for bioprocess sensor development. Industry’s commitment to PAT, in particular, is stronger than ever. Although PAT has taken more than a decade to reach its current state, there’s no doubt that manufacturers understand the relationship between sensors, PAT, and quality, asserts Ken Clapp: “The prospects for greater competitiveness and lower costs are huge drivers.”

Previous articleNovartis Eliminates Cell & Gene Therapies Unit, Restates CAR T-Cell Commitment
Next articleTheranos Withdraws FDA Application for miniLab Zika Test