May 15, 2009 (Vol. 29, No. 10)
Gail Dutton
Firms Pursue New and More Effective Methods in Their Efforts to Reduce Risk
Cell culture productivity has improved 50-fold during the past 20 years, leading to bottlenecks in downstream processing. Increasingly high titers have further slowed processes. During the same period, new agents and contamination risks have emerged that have required companies to discover new and more effective methods to remove viruses from bioprocess lines in an ongoing effort to minimize risks. To that end, researchers are developing new methods and revisiting others, including filtration, precipitation, and fractionation.
Currently, “most processes for therapeutic antibodies use Protein A for capture chromatography, followed by one or two polishing steps,” explained Alahari Arunakumari, Ph.D., senior director of process development, Medarex. Medarex, however, is developing alternate purification methods that do not rely on Protein A for human mAb purification, which she detailed at Bioprocess International’s “European Conference and Exhibition,” held in Dusseldorf last month.
Medarex has successfully replaced a concentration and diafiltration step that preconditions the load for cation exchange capture chromatography in non-Protein A purification processes for human mAbs, with precipitation strategies that remove either the host cell proteins or the antibody from the cell culture. “Both offer high-throughput strategies,” Dr. Arunakumari said, and can be added easily to existing manufacturing facilities.
“The process of removing host-cell proteins works in three column steps,” Dr. Arunakumari continued. “In the first step, a small molecule is added to the clarified, pH-adjusted CHO cell culture supernatant (105 ng/mg), which results in the precipitation of most of the host-cell proteins, leaving about 2×102 ng/mg in the supernatant. In the second step, this supernatant is purified further by cation exchange(CEX) chromatography. CHO host-cell protein levels are reduced further, to greater than 10 ng/mg,” Dr. Arunakumari said.
Purity, therefore, is greater than that obtained through Protein A capture for the same human mAb purification, she said. This process also speeds purification by minimizing the number of capture cycles because of high-binding CEX resins.
The third step in the purification process involves anion exchange membrane chromatography to remove any residual, negatively charged contaminants and adventitious viruses. Loading capacity is as high as approximately 20 g/mL, Dr. Arunakumari added. “We’re still scaling up that method and transferring it to the manufacturing process,” she said. “It’s pretty robust.”
“Another approach under development at Medarex selectively removes the antibodies by using a specific salt that is generally used to make buffers in antibody purifications.”
When antibody precipitate is dissolved in a resuspension buffer, “the volume is reduced by one-quarter to one-third from the original,” she explained, thus shortening the time needed to load the chromatography columns. This strategy reduced host-cell protein by more than 50 ng/mg after a cation exchange separation step, which was further reduced by anion exchange membrane chromatography. This step reduced host-cell protein levels to below the detection limits. A two-step process, therefore, appears feasible. “It seems like this shows more variation in host-cell protein removal during scale-up,” she added.
In comparing the two methods, Dr. Arunakumari noted that the first method, which removes host cell proteins from the supernatant by depth filtration, is more efficient for many human mAbs, and also, offers the benefits of disposable chromatography, thus enabling a better facility fit.
The second method, in contrast, “needs a centrifuge to remove the precipitate and also requires relatively more optimization for robust scale-up. Different human immunoglobulin G isotypes are being tested by both precipitation process types. Either of the two processes could be indicated for non-Protein A purifications,” she concluded.
Medarex is developing alternative purification methods that do not rely on Protein A for human mAb purification.
Filtration
Filtration with viral removal filters is a traditional clearance step, and most manufacturers have been able to optimize their product for filtration without any significant loss of product. “However, when the process undergoes the actual virus clearance studies, and viruses are spiked into the product in a scaled-down study, the performance of the process step can sometimes be problematic,” according to Joe Hughes, Ph.D., vp biologics testing, WuXi AppTec. “Between 90 and 95 percent of the time, filters run as expected,” he reported.
For that remaining 5 to 10%, the quality of the virus spike often causes the filters to plug. To address this situation, WuXi AppTec is advancing two technologies to prepare virus for virus removal studies—ultracentrifugation for enveloped viruses and chromatography using membrane absorbers for nonenveloped viruses. Dr. Hughes will discuss WuXi’s approach in detail at Informa Life Science’s “Viral Safety for Biologica” conference in Cologne next month.
Ultracentrifugation involves putting glycerol, which acts like a filter, on the bottom of the centrifugation tube. “The virus pellets through the glycerol,” he said, “resulting in a cleaner virus preparation, but holding back much of the cellular proteins and membrane fragments.” WuXi’s version is known as Ultra 1, and routinely results in a five to seven log viral removal, without reducing flux, reported Dr. Hughes. It’s not a good fit for all viruses, however.
Therefore, WuXi also offers Ultra 2, which is designed for parvovirus and for minute virus of mice. In quality control testing, compared to Ultra 1, “Ultra 2 takes away three- to fivefold more protein and 100-fold more nucleic acid,” Dr. Hughes noted. “Performance of the Ultra 2 virus preparations on virus removal filters has demonstrated good removal of the parvoviruses, often without any changes in the product filtration performance.”
For the parvovirus purification, an anion exchange based membrane absorber is used. The virus will bind to the absorber in a low salt solution and can be eluted with 0.4 M sodium chloride solution, but the nucleic acid will be strongly bound and is essentially removed from the resulting Ultra 2 virus preparation.
The result, Dr. Hughes says, is that when the virus is spiked, “it doesn’t clog the filter.” This method is most effective for parvovirus and other nonenveloped viruses. Researchers are looking at this method for other nonenveloped viruses, as well as other matrices as a way to expand its applications. “All of these enhancements should allow manufacturers to effectively validate their filters, and perhaps even reduce the size and overall cost of the filters needed in the process.”
Millipore launched the Viresolve® Pro family in several phases in 2008, and the FlexReady platform in March 2009, as a virus-clearance solution to remove parvovirus from mammalian or recombinant sources. The technology, which was described at the Dusseldorf conference, features a patent-pending membrane, a newly designed device format, and more sensitive performance tests.
“The design of the membrane is based on worldwide feedback from our customers,” explained Gerd Kern, field marketing manager, bioprocess division. The result of their feedback is a membrane that helps companies ensure regulatory compliance and manage their business risks associated with virus safety, he explained.
The dual-layer PES membrane is designed to simultaneously provide high retention, high capacity, and high flux. This is important, Kern said, because there is a trend to compress the chromatography steps before this purification step from three to two, leading to a greater impurity load at this point. “It processes fast,” he added, “and is caustic stable, so it can be sanitized with everything in line. This is a big leap forward in terms of overall performance.”
The new device format scales from development through manufacturing and integrates easily into the new Mobius FlexReady platform, Kern noted. The Viresolve membrane is available in Micro, Modus®, and Magnus® formats, which are designed to fit in all processes and enable productivity gains in all production scales.
“Viresolve Pro has a fully disposable flow path, so there’s no cleaning or cleaning validation,” Kern said. More robust performance tests including binary gas tests also have been developed “to detect the smallest defect and to confirm virus retention confirm.” The Viresolve Pro has been designed to remove Ž4 logs of parvovirus and Ž5 logs of retrovirus. Versions of the Viresolve Pro are available for process development and virus validation, the Modus for pilot scale and medium processing, the Magnus for large-scale processing by using the Viresolve Pro holder.
Generic Data Use
In February, the EMEA issued a final guideline governing the use of generic data in viral reduction. “The guidance can drive up the cost of studies,” because of the seemingly contradictory aspects contained within it, noted James Berrie, Ph.D., senior group leader, purification development, Lonza Biologics. Dr. Berrie will discuss the use of generic data in more detail at the upcoming “Viral Safety for Biologica” meeting.
For example, he explained, the guidance allows the use of generic viral reduction data along with two spiking studies that show reduction. Reducing the scope of the study in this way reduces the cost by targeting process steps, “yet two studies don’t prove the mathematical principle.” Therefore, he said, “the scope of the study should be taken into account in the context of this new regulatory guideline, even though some aspects can appear contradictory.”
Generic data has been used successfully as part of the license application by companies that have a downstream process that doesn’t vary, and also antibodies of the same type, Dr. Berrie elaborated. For contract manufacturers like Lonza, however, the challenges are different because, “we don’t have our own products and so don’t own the data.”
Consequently, generic virus reduction data can be used by CMOs with repeat customers who have a manufacturing pipeline for a particular product. “That’s good for customers with a pipeline, but be very careful about variability in the process,” Dr. Berrie cautioned.
He advised performing a gap analysis focused on the process parameters to identify any variables that could affect virus reduction. That includes the downstream process variables as well as any fluctuations in the compound ingredients themselves.
With customer approval, CMOs could, in theory, pool data to derive generic virus- reduction data that could then be used to benefit members of that pool, Dr. Berrie speculated. “Customers who hadn’t contributed to that pool would still have to perform wet studies. Having said that, Lonza has a downstream process for mAbs, so we know what reduction factors you could expect to see.”
Such information, he explained, may be used as a starting point from which to base decisions to evaluate particular steps. The goal, Dr. Berrie explained, is to use generic virus-reduction data to allow researchers to focus their efforts on steps that have the greatest potential for virus reduction, effectively ignoring steps with reductions of one log or less.
Lonza has found that the capacity for virus-reduction changes, even across such robust steps as a low pH hold and virus-reduction filtration. “In some process steps we see a reduction of virus removal with more dilution, making it less good.”
The “Viral Safety for Biologica” conference will also include an update on cold ethanol fractionation from Herbert Dichtelmüller, Ph.D., director of virus validation of Biotest.
Cold fractionation is an important purification method for organizations involved with the blood supply. “However, the European guidelines leave some space for interpretation, and some, not all, regulators in Europe believe that the steps of fractionation are not reliable enough,” Dr. Dichtelmüller explained.
“Cold ethanol fractionation does not inactivate viruses, but partitions viruses into the precipitate together with the waste protein. Viruses can be recovered from the precipitate almost without loss,” Dr. Dichtelmüller said. Partitioning, however, is not equally effective for all viruses, and varies with the step and the virus. “Therefore, validation is essential and robustness is required.”
Although the details of his latest research are, as yet, not public, Dr. Dichtelmüller showed that cold ethanol fractionation is a relevant virus-removal procedure. His lab has performed and evaluated 280 studies on cold ethanol fractionation to determine its effectiveness. That method “is working reliably for the precipitation of fraction III or I/III for immunoglobulins, and for precipitation and removal of fraction II/III or I/II/III, and precipitation and removal of fraction IV/IV (both steps of the albumin fractionation process),” he says.
While cold ethanol fractionation contributes to viral safety of immunoglobulins and albumin, it can never be the only measure to remove viruses from the products. Further inactivation steps are essential.
Viresolve membrane is available in Micro, Modus, and Magnus formats, which are designed to fit in all processes and enable productivity gains in all production scales, according to Millipore.
