September 1, 2009 (Vol. 29, No. 15)
Angelo DePalma Ph.D. Writer GEN
Manufacturing Inefficiencies Can Be Reduced by Bringing Analytics Closer to the Operation
Bioprocess monitoring and control, although not as widespread as it should be, is nevertheless increasingly viewed as a routine part of doing business. Significant trends include interconnectivity between sensors, instruments, bioreactors, sampling systems, and software, collaborations among vendors to promote interoperability, and compatibility with single-use equipment. Yet, biotech still lags behind other industries in terms of manufacturing efficiency, and the deficiencies can be blamed in part on gaps in real-time monitoring.
“People still come in at night and on weekends to take care of processes that run 24 hours a day,” notes Brian O’Flaherty, Ph.D., sales director at Groton Biosystems. “Automation can free their time to perform higher-level tasks.”
One aspect of the PAT initiative involves bringing analytics closer to the process. FDA defines proximity as offline (usually in another room), at-line (nearby, but not connected), and online. Unfortunately, nobody imagined 15 years ago that instrumentation and controllers would need to be so close to a bank of development-scale bioreactors—or for that matter next to production reactors.
“If you start bringing in equipment that is normally offline, in another room, and transfer it to a process-development lab, you will encounter a space problem,” observes Dr. O’Flaherty. “As it is, all the equipment in these labs is shoehorned in.”
The solution, he says, is to design analytics and control into labs and manufacturing space, as one Groton customer recently did. “They were able to wheel in our sampling system and a Shimadzu HPLC, went online with them, and they’re working wonderfully.”
Groton has collaborations with DASGIP for advanced sampling interfaces to that company’s parallel reactor system, with Agilent Technologies for bioprocess HPLC, and with YSI for interfacing with a nutrient monitor.
Groton’s ARS-M sampling system consists of three basic models: the 140 interfaces to one reactor and four instruments, the 440 to four reactors and four instruments, and the 840 to eight reactors and four instruments. ARS-M is also compatible with single-use bags from major manufacturers and all stainless steel bioreactors. Groton’s principal market is process development, but it is beginning to break into GMP markets as well.
Interoperability and Collaboration
The integration of analyzers, bioreactors, and auto-samplers, particularly during product development, has been a notable trend in bioprocess monitoring. YSI, whose 2700 and 7200 biochemistry analyzers measure lactate, glutamate/glutamine, ammonium, glucose, lactate, and other analytes, has partnered with Groton on connecting the analyzers to its ARS autosamplers for real-time process monitoring.
The 2700-based systems are used mostly for process development, says Jamie Lussier, product manager at YSI. Another collaboration, with Groton and DASGIP integrates the YSI 7100, Groton’s automated reactor sample system, and the DASGIP bioreactors and control system.
YSI has also collaborated with Flownamics by coupling YSI nutrient analyzers to Flownamics’ SEG-FLOW™ sampling system. SEG-FLOW withdraws a cell-free sample from up to eight bioreactors, and sends them to up to four analyzers or fraction collectors. The sampling system can then control feed to the reactors based on set target concentrations in Flownamics’ Flow-Web™ software. The system activates a feed pump (or an existing feed system) based on the measured result and target concentrations. Users can determine sampling frequency and size.
Michael Biksacky, Flownamics president, says that two customers use the YSI 2700-SEG-FLOW combination to monitor and control delivery of glucose for cell culture processes. SEG-FLOW works with a variety of instruments from major vendors, like the Cedex cell density analyzer from Roche Innovatis. The key to success, says Biksacky, is to provide customers with flexibility. “You don’t want to get to the point where have to tell customers ‘you must use this analyzer.’”
NIR: An Enabling Technology
Near-infrared (NIR) spectroscopy is fulfilling its promise for rugged, robust process analysis. A vibrational spectroscopy, NIR uses light between the visible and IR regions. Typically, one fiber-optic probe interrogates the sample or process while another reads the absorbance. As with IR, any molecule whose vibrational modes entail changes in dipole moment is suitable for NIR analysis. But where IR measures fundamental vibrations, NIR hones in on lower-absorbing overtone or combination bands.
“This is an advantage,” notes Todd Strother, Ph.D., an applications scientist at Thermo Fisher Scientific. Strongly absorbing IR specimens must be diluted, which makes IR a poor real-time analysis method. “NIR is different. Since materials don’t absorb as much, you can measure them directly.” With NIR it is possible to position probes directly inside a bioreactor (or, for that matter, a powdered sample) and obtain a spectrum.
NIR absorbances are not as straightforwardly identified with particular chemical groups or their concentrations in solution. “Analytes in bioprocesses don’t follow Beer’s law,” observes Dr. Strother. “They absorb differently and nonlinearly at varying concentrations.”
Applying NIR in bioprocess monitoring, therefore, requires that operators first construct spectral training sets. Using independent primary analysis methods, with the aid of mathematical algorithms, operators can then assign qualitative and quantitative attributes to spectral peaks, connecting them to analytes of interest, e.g., glucose, lactate, ammonia, and others. Once the training set is constructed for a particular fermentation or cell culture the primary method need not be used again.
NIR analyzers can provide virtually instantaneous readouts of all analytes of interest, which makes them ideal for process analytic technology. Most upstream NIR probes operate inside bioreactors and can undergo CIP/SIP. It is also possible to construct an NIR system for disposable processing by taking advantage of the penetration of NIR radiation through thin plastic.
NIR was popular in the early 1990s for analyzing microbial fermentations. “That died down somewhat but is picking up again,” says Denise Root, marketing manager at FOSS NIRSystems. “Now, for every two mammalian cell fermentation projects we get, we hear from one microbial cell-based manufacturer.”
One NIR benefit is its flexibility with respect to reactor interface. Root says her company’s probes fit through all standard ports, but dissolved oxygen ports are a favorite.
FOSS’s principal bioprocess product is the XDS Process Analytics™ MicroBundle Analyzer, which uses 80 optical fibers—40 for illumination and 40 for collection. The device can monitor up to nine channels (analytes) in a single reactor, or nine separate reactors/fermentors at a time. The company has recently completed a special design for a single-use reactor as well.
Finesse Solutions, which specializes in measurement and control products for upstream bioprocessing, manufactures a line of sterilizable and single-use sensors. On the sterilizable side there is the TruDO Optical dissolved oxygen sensor with a built-in transmitter, TrupH (pH), TruDO (dissolved oxygen), and a NIR-based TruCell2 (cell density).
The single-use TruFluor™ dissolved oxygen and temperature sensor consists of a disposable sheath, optical reader, and transmitter. The sheath is inserted in a disposable bioreactor bag port and irradiated with the bag to preserve a sterile barrier. All finesse sensors are precalibrated, RFID-tagged, and meet USP Class 6 standards for cGMP operation.
Bioreactor pressure has become a critical measurement since the advent of polymer-based, single-use process bags. “People spend $5,000 to $10,000 on a disposable bioreactor and $100,000 on media, and if you don’t keep an eye on pressure the bag can explode,” explains Mark Selker, Ph.D., CSO. At “Interphex,” Finesse introduced TruTorr™, a single-use pressure sensor for measuring headspace pressure in single-use containers to prevent overpressure damage.
In late June, Seahorse Biosystems held a meeting with biotech experts to ascertain their needs in bioprocessing and process development. The convened experts agreed that more robust process monitoring and control would help industrialize biomanufacturing, reduce cost, accelerate scale-up, and improve product quality. Furthermore, implementing monitoring and control earlier in the development process would permit evaluation of each input paramenter. “The effects of these factors may be different for each clone and, therefore, many experiments are needed to obtain this information,” explains product manager Peter Russo, Ph.D.
“Our design team experts cited scientific need, as opposed to regulation, as the primary driver for this trend. There is a need to obtain the data required to define those process parameters that produce high product quality in the early stages of development. As a result, more robust processes that produce better quality products can be developed quickly and efficiently. There is also the advantage that this information can then be applied to quality-by- design initiatives.”
Seahorse has popularized SimCell™ Micro Bioreactors and the SimCell™ 200 Workstation. SimCell is not a process monitoring system per se, but rather, allows creation of thousands of highly controlled sets of reaction conditions for subsequent use.
SimCell bioreactors are microplate-sized disposable plastic cards containing six 700 uL cell culture chambers that mimic conditions in a bench-scale bioreactor. Oxygen and carbon dioxide diffuse easily through the SimCell’s highly permeable chamber walls, and the large surface area supports high cell densities. Gas bubble stirring matches the stirring shear force and frequency of a larger reactor. The SimCell 200 automated cell culture manager agitates, feeds, and pH balances up to 200 microbioreactor cards (1,200 chambers) for six weeks or more, with unattended operation for up to one week, Dr. Russo notes.
Simcell conducts complex fed-batch protocols with automated monitoring and control of temperature, pH, dissolved oxygen, and glucose across many clone and process variations. The correlation between the SimCell and benchtop bioreactors is very good, with correlation coefficients as high as 0.97, Dr. Russo reports. This technology can be used to displace uncontrolled platforms, such as shake flasks and well plates, during the early stages of development, to design robust, scalable processes, he adds.
Because they faithfully track conditions in much larger reactors, SimCell allows investigators to measure the impact of each process variable quickly, with tight correlation to bench-scale bioreactors, and in as many replicates as desired.