Easing the Bottleneck

It's become an embarrassment of riches. New, robust cell lines now churn out therapeutic monoclonal antibodies (mAbs) in quantities so high, biopharmaceutical manufacturers are having a hard time purifying the drugs with state-of-the-art tools and techniques. The bottleneck, many claim, arises because, simply put: downstream processing technologies have not kept pace with improvements in upstream production technologies.

(ILLUSTRATION BY M. MCEVOY. IMAGES: YAMADA TARO, JOEL SARTORE, IMAGE SOURCE/GETTY IMAGES)

"Since the introduction of monoclonal antibodies, upstream productivity has increased about 1000-fold," says Uwe Gottschalk, group vice-president for purification technologies at Sartorius Stedim Biotech in Gottingen, Germany. "Downstream technologies have improved only about 10-fold when we compare the productivity of critical steps such as chromatography. There is a gap, and it's widening. If we keep the existing strategies and technologies, then there is little chance of closing the gap," says Gottschalk, who edited the recently published book, Process Scale Purification of Antibodies.

State-of-the-art purification

In the current paradigm, therapeutic mAbs are produced by mammalian cells. These cells are maintained in large bioreactors—vats that can hold tens of thousands of liters. The cells are suspended in a culture medium, essentially a nutritive broth, and genetically engineered to produce therapeutic mAbs. The cells generally secrete the antibody products into the medium, along with the molecular waste of cellular metabolism. Many cells—sometimes as many as 50%—die in the bioreactor during the course of production, and these dead cells often split, or lyse open, and spill their contents into the medium. These contents include bits of the fatty cell membrane, cellular proteins, virus particles that may have infected the cell or were suspended in the medium, and the nucleic acids DNA and RNA, which, in this context, are also considered waste products. Purification, therefore, has two goals.

The first goal is to retain product, and this is accomplished in the "capture" phase of purification. The second goal is to discard contaminants, achieved during the "polishing" phase. Before either of these phases is initiated, however, the contents of the bioreactor are "clarified," usually by centrifugation, followed by filtration. The idea is to remove particles, including the remaining and now unwanted intact cells. Centrifugation separates out components by weight, so the intact cells fall to the bottom of the centrifuge tube, leaving the lighter components—including the antibody product—suspended in the fluid on top of the centrifuge tube. The solid matter is discarded, and the supernatant is then filtered to remove any remaining particulate matter. Capture and polishing then follow. Of these two phases, capture has received the most the attention in terms of technical innovation and cost reduction.

Downstream burden

The current wisdom posits that high-yield bioreactors are producing so much material that biopharmaceutical manufacturers are having a hard time purifying it all. But not all manufacturers are experiencing the bottleneck, or at least not as severely as others. Manufacturers producing very large quantities of blockbuster drugs, at quantities that can exceed one metric ton per year, in fixed facilities may feel the problem more acutely than others. For them, the problem boils down to this: A manufacturer may have two options to deal with the bottleneck. Either it acquires additional purification equipment, or purifies the product in batches. But producers using dedicated, fixed facilities may simply not have the space to accommodate additional equipment. For them, the best option might be to purify the product in batches. Three to five batches may be necessary to purify all of the protein product now produced by today's high-yield cell-culture systems.

Figure 1: The major steps in monoclonal antibody purification. (FIGURE IS COURTESY OF THE AUTHOR, ADAPTED FROM SOURCE 4)

Gunter Jagschies, senior director of strategic customer relations at GE Healthcare BioSciences in Uppsala, Sweden, is less swayed by stories of bottlenecks. "You hear these stories, and at the same time, they seem to come down to existing installations built with too small downstream processing equipment to accommodate very much higher productivity upstream. This is not so much a problem of the technology but more of the ability to predict what would happen with cell culture," he says. Jagschies points to case studies that indicate that 10 metric tons of product—roughly 10 times today's highest production levels—can be made with existing technologies. The problem, as he sees it, has to do with the rate at which high-yield supernatants can be run through a column.

"You have to process monoclonal antibodies quickly," says Jagschies. "There are problems maintaining the antibodies' stability in liquid." The protein product is vulnerable to enzymatic degradation by proteases, with the result that the composition of a batch at the end of an 18-hour holding period may differ from the starting composition. Theoretically, product yields could become so high as to require several days to purify the contents of just one bioreactor. "But no one is there yet. Current [purification] technologies can handle a batch in 12 to 18 hours," Jagschies says.

Ironically, the purification problem may be more acute for contract manufacturers and niche manufacturers, who are producing the smallest product quantities. Current blockbusters—Genentech's anticancer drug Rituxan (rituximab), for example, or Abbott's arthritis drug Humira (adalimumab)—enjoy billion-dollar sales because they address diseases with large patient populations. But the day of the blockbuster may be over—even for the relatively new mAb drugs.

In coming years, niche drugs for relatively rare conditions, or targeted therapeutics that are effective for subpopulations of patients suffering from fairly common conditions such as arthritis, will require much smaller production batches to serve the reduced markets. But smaller batches may mean very expensive purification equipment that is not used to its maximal capacity or for its total life span, which raises the the cost per use of such equipment. Even producers of blockbusters are looking for lower-cost options as pressure increases on them to lower drug prices, and as they stare down the specter of competition from follow-on biologics. In fact, cost considerations have been as much of a driver of innovation as the high product yields.

Focus on capture

Of the steps in mAb purification, the capture step receives the most attention, possibly because it's so important in trapping the product, but also possibly because of the expense associated with it. Standard protocol relies heavily on the affinity of human antibodies for a compound called Protein A.

In the early years of antibody purification, manufacturers relied on "natural" Protein A; that is, the protein as it was isolated from the cell walls of the bacterium Staphylococcus aureus. These days, most commercially available Protein A has been genetically engineered and molecularly cloned. As used in a chromatography column, Protein A is part of a resin. Resins from different vendors differ markedly to respond to the distinctive needs of their customers, but they all share some common features. In general, this resin consists of some kind of porous bead inside of which is a gelatinous matrix to which the Protein A is chemically affixed. The antibody product is poured through the chromatography column, diffuses inside the beads and through the gelatinous matrix where it finds and binds to Protein A. Almost all of the contaminants, most of which do not have an affinity for Protein A or the resin components, exit the column. In subsequent steps, buffers are run through the column to release the antibody product from Protein A, and the now somewhat purified product is collected and shunted into subsequent "polishing" steps.

Protein A is so efficient at attracting and binding the antibody product that, says Henrik Ihre, product manager for Protein A and program manager for mAb products at GE Healthcare BioSciences, "you go from 1% to above 98% purity in one chromatographic step." And while some companies have tried to engineer more antibody-binding sites onto Protein A, nature it seems, has produced the best overall variant of the molecule.

Souped-up Cells

"Nature has engineered Protein A to be pretty efficient and robust," marvels Laura Whitehouse Pew, vice-president of market development at Massachusetts-based Repligen. Repligen produces the recombinant Protein A that other manufacturers immobilize in their resins.

So if they can't improve upon nature with regard to the affinity of Protein A for antibodies, many resin manufacturers have tried to improve the durability of the resin to increase its lifetime and, thus, lower its cost per use. Ian Sellick, director of marketing at Pall Life Sciences in New York, says that a chromatography column can cost as much as $5 million to pack and is good for up to approximately 200 purification cycles. To increase durability, commercial vendors such as GE have developed Protein A variants that can hold up for more purification cycles. Resins frequently degrade during the harsh cleaning protocol in between cycles. Cleaning makes use of very caustic chemicals such as sodium hydroxide—lye—which decouples Protein A from the matrix. The Protein A in GE's MabSelectSure resins has been genetically engineered to resist such cleansing-induced decoupling. As a result, the resins may be usable for more than 300 cycles, thus reducing the cost of use per cycle.

Other companies are offering somewhat more rigid matrixes, with the hope of reducing the compressibility of the matrix and increasing flow-through rates. For example, DSM (Heerlen, The Netherlands) developed EBA resins. These Protein-A-based resins, developed in partnership with the Danish firm Upfront, use tungsten carbide particles. According to DSM literature, the added weight allows fluids to flow through the column at increased rates, with the aim of accelerating processing times. Pall, too, offers a zirconium-based resin.

Pall, in an attempt to reduce resin costs, is offering a product that uses a synthetic substitute to Protein A. According to Sellick, Pall's "gel-in-a-shell" resin is reconfigured to offer the same level of separation at a rate that's faster and which holds up better during sodium-hydroxide cleaning than traditional Protein A-based resins. To cap it all off, Pall's Hypercel product line is a quarter the price of some of the traditional Protein A-based resins. Sellick adds that Pall's new resin can withstand between 100 and 200 purification cycles.

Pall is not the only manufacturer exploring the use of synthetic capture chemistries, or ligands. BAC (Naarden, The Netherlands), for example, produces affinity ligands which are derived from Llama antibodies. One of their products is marketed via GE as IgSelect, a complementary product to Protein A-based resins.

Manufacturers stress that different resins offer biopharmaceutical manufacturers a range of options so they can match their chromatography solution to their particular equipment needs and to the exigencies of their particular monoclonal product. But as much as manufacturers may differ on their choice of resins, they all seem to have converged on the use of disposables. Every provider of chromatography resin offers disposable options. In many cases, this means that biopharmaceutical manufacturers have the choice of purchasing prepacked, disposable columns.

Vendors note that the disposable option eliminates the enormous upfront capital expenditure related to the purchase of stainless-steel chromatography equipment. Disposables also eliminate the time and expense related to packing the column initially. Most of these disposable columns can be used through several purification cycles—somewhere in the neighborhood of 10 to 20 cycles. The disposable option may be especially good for smaller batches of niche products, early-stage process development, contract manufacturers who have to produce many different products in the same facility, or for any manufacturer who—for whatever reason—does not need to run hundreds of purification cycles.

Millipore (Billerica, MA) has gone the furthest with a disposable gambit. At the Interphex 2009 show in March, the company unveiled its suite of Mobius products. The Mobius line (marketed as Mobius Flex Ready Solutions) is made up of two components. The first is a metal cart, approximately the size and shape of your average New York City hotdog vendor's stand. Into these can be inserted the entire disposable purification train, with all of the bags and hoses already attached. (Disposable trains for other aspects of biopharmaceutical processing are also available to work in concert with the Mobius carts.)

The truly intrepid biopharmaceutical manufacturers may move away from column-chromatographic solutions entirely. In an attempt to reduce the footprint of the purification apparatus as well as costs, some vendors are developing membrane-based purification devices. Lisa Crossley, president and CEO of the Canadian firm Natrix Separations, described in January at the Biomanufacturing Summit in San Diego, CA, organized by WTG, a novel capture membrane the company has developed. Unlike other membrane-based capture technologies that put the affinity chemistry on the membrane's surface only, Natrix's membrane offers a more dimensional structure. The apparatus is composed of a support membrane that houses a porous hydrogel to which is affixed either Protein-A or synthetic ligands, depending on the customer's needs. It is available in both multicycle and single-use, disposable formats.

The dimensionality of the Natrix product is intended to increase the binding surface area, diffusion rates, and flow rates, all while reducing the equipment's footprint, and significantly reducing operating costs, Crossley says. In addition, the membranes can be used for simultaneous clarification and capture, eliminating the need for separate harvest and primary purification steps. "There's a big cost savings in combining two steps in one," Crossley says.

The Natrix membranes offer such an impressive combination of attributes that one person attending Crossley's San Diego presentation exclaimed that Natrix had delivered the "Holy Grail" of mAb purification. Others in the industry are not quite as moved. The pharmaceutical industry is slow to adopt any technology that might engender regulatory scrutiny, and so may take a "wait-and-see" stance on this and other technological developments. Furthermore, skeptics point out that the Natrix technology's robustness has not yet been proved for full commercial-scale use. To date, the technology has been used for commercial-scale manufacturing of nutraceuticals and veterinary vaccines in North America and Europe, but is still in the preclinical stage for biopharmaceutical applications.

Pressure toward greater efficiency and lower cost may, surprisingly, drive the industry back to the future as they explore alternative technologies, notes Sartorius's Gottschalk. "Because of the expense, many other industries don't use chromatography at all," he says. He points to technologies used to purify components from blood plasma. Typically, these manufacturers precipitate the desired products out of the plasma, a technique that Gottschalk offers might be adapted to mAb purification in the years ahead.

Jagschies, on the other hand, is dubious that chromatography will disappear. "Right now, the world leader in plasma-derived biopharmaceuticals, CSL from Australia, has put chromatography in place pretty much everywhere," he says. "This will not change back, not in my time at work for sure." Several players, notes Pall's Sellick, are also exploring fractionation and crystallization.

Putting the polish on polishing

Gottschalk doesn't stop there, though. He is looking at ways to improve purification beyond the capture step. "We need to shift our focus from product capture, where yields are increasing, to contaminant removal," he says. He suggests that it may be possible to remove more impurities—nucleic acids, for example—in the early clarification steps, thereby reducing the burden on the capture materials and possibly reducing the number of polishing steps required as well.

Millipore has also been considering ways to beef up impurity removal. For example, the company is specifically honing in on ways to improve the efficiency and lower costs of some initial clarification steps. Following centrifugation and before capture, the contents of a bioreactor go through depth-filtration.

"People argue that the current template [for this stage] is not broken, so why fix it?" asks Fred Mann, program manager for Downstream Process Solutions at Millipore. "But you can refine it, improve productivity, and make the process more robust," he says. As a result, Millipore has restructured its depth filters to incorporate a medium specifically designed for high-titer feedstock filtration, which, he says, will allow customers to run more liters of feedstock through the filters. In addition, the filters are self-contained and installed into hardware with no product contact rather than in stainless-steel housings that require hoists and cranes for assembly. The filtration system, which Millipore markets under the name MilliStak? Pod, reduces the need for cleaning and cleaning validation of the filtration assembly between runs, is safer for operators, and reduces labor costs, says Mann.

Other polishing steps include rounds—typically two—of ion-exchange chromatography. The material is put through a cation exchange column to remove positively-charged host cell proteins and high-molecular-weight impurities such as protein aggregates. Following that is an anion-exchange step, where negatively charged contaminants, including retroviruses, parvoviruses, host-cell protein, and residual DNA, are removed. In the final steps, the product is run through filters and irradiation to remove any remaining virus particles. After that, the now-pure mAb product is put into the formulation buffer.

Gottschalk is convinced that the ion-exchange steps present the greatest opportunity for improving the downstream paradigm. He notes that when run in "flow-through," it becomes possible to replace a 100-L column with a (disposable) 1-L anion-exchange membrane, thus reducing the requirement for space, among other things. As for cost? "In the worst case, the costs are equivalent for membranes. Since so much cost is in fixed assets, virtually all processes benefit significantly from disposable strategies in polishing," he says.

Overall, Gottschalk predicts that cost and efficiency considerations will force a whole new process to emerge for mAb purification. "The paradigm will shift," he says. "And whatever it will be, I predict we'll see it within the next 5 to 10 years."

Sources

1. F. Detmers, P. Hermans, and M. Ten Hafft, "Affinity Chromatography Based upon Unique Single Chain Antibodies," LCGC The Peak, Sept. 2007.

2. G. Jagschies, "Where Is Biopharmaceutical Manufacturing Heading?," BioPharm Inter., 21 (10), 72–88 (2008).

3. L. Giovannoni, M. Ventani, and U. Gottschalk, "Antibody Purification Using Membrane Adsorbers," BioPharm Inter., 21 (12), 48–52 (2008).

4. U. Gottschalk, Ed., Process Scale Purification of Antibodies (John Wiley & Sons, 2009).

5. U. Gottschalk, "Downstream Processing of Monoclonal Antibodies: From High Dilution to High Purity," BioPharm Inter., 19 (6), (2005).

6. M. Rios, "Eluting Possibilities with Mixed-Mode Chromatography," Pharm. Technol. 31 (5) 40–48 (2007).