April 15, 2009 (Vol. 29, No. 8)

Angelo DePalma Ph.D. Writer GEN

Bottlenecks Are Not as Ubiquitous as Believed

For at least the last decade, protein purification has increasingly been viewed as a process bottleneck—not just for individual unit operations, but for the entire purification train. Current wisdom holds that production titers will continue to rise, and with them, the height and number of hurdles encountered downstream. While this belief is certainly true for some products, particularly long-lived drugs manufactured at extremely high scale, it is something of a simplification. Downstream bottlenecks are neither universal nor inevitable.

Downstream bottlenecks differ according to the product’s development stage, notes Alan Hunter, Ph.D., senior principal scientist at Pfizer. Since processes are designed with equipment and facilities in mind, early-stage clinical manufacturing tends to encounter limitations in those areas, particularly when outsourcing is involved. Dr. Hunter advises that, in those situations, drug developers should “try to be more proactive in planning and designing processes to fit into the contractor’s capacity.”

Analytics represent another early-stage downstream hurdle. HPLC analysis of column fractions, for example, takes between 20 and 40 minutes each. “Throw in GMP requirements and the delay can reach a day for ten to twenty fractions.” The answer, explains Dr. Hunter, is to minimize the number of fractions, or collect them intelligently, so as to minimize the number of samples.

Later clinical stages processors should have had the time to solve facility and equipment issues, but here, capacities of chromatography resins and virus filters can become intractable—not from an engineering standpoint, but from the perspective of economics.

Virus filters are easily scaled, but their price tag (around $25,000 per run) and their single use add significantly to process costs. Protein A resins are also scalable and expensive—up to $10,000 per liter (around $1.5 million for one column’s worth). Protein A is recyclable, but the regeneration process is expensive and time-consuming.

Nearly every top mAb manufacturer is looking for alternatives to protein A capture, but these efforts are not always in harmony with process requirements at various development stages. “If you want to eliminate the use of protein A you would like to establish the alternatives relatively early, but early is precisely when you really need protein A, to crank out antibodies and get them into the clinic,” Dr. Hunter continues.

Most severe bottlenecks occur in facilities that are between 10 and 20 years old, where designers could not anticipate huge upstream productivity enhancements. “High titers affect all downstream unit operations, with the possible exception of centrifuges,” notes Günter Jagschies, Ph.D., senior director for strategic customer relations at GE Healthcare Life Sciences.

Nevertheless, Dr. Jagschies believes that downstream capacity issues may be over-hyped since very few products are manufactured at mega scales. “Most mAbs reaching the market over the next few years will require only up to about 200 kg per year.” The advent of biosimilars will further reduce the need for large-scale production of blockbuster biotherapies.

“When people talk about severe bottlenecks, they’re usually referring to first-generation processes. Resins and membranes being introduced today, which belong to the second and third generations of purification tools, offer up to five times the throughput of first-generation technologies.” For example, Enbrel is currently manufactured at three facilities, which, according to Dr. Jagschies, would not be necessary if Amgen/Wyeth employed current-generation downstream technology.

Experts also questions the degree to which higher titers are unduly stressing downstream operations. While titers are indeed rising, biopharmaceuticals are becoming more potent, which reduces dosing and in many situations batch size as well. In fact, bioprocessors may soon find themselves in the enviable situation where downstream capacities catch up. Further improvements in cell culture productivity will continue, but their impact will be to improve upstream efficiencies or perhaps to shrink expression and fermentation operations.


Sartorius Stedim Biotech’s Sartobind Q Mega membrane adsorber is used in large-scale polishing.

Volume- vs. Mass-Driven

“Upstream productivity is volume-driven, whereas downstream operations are mass-driven,” observes Uwe Gottschalk, Ph.D., group vp for purification at Sartorius Stedim Biotech. In other words, protein expression is a function of cell density, an intensive property, while purification is, by nature, an intensive operation (more protein equals more buffer, larger tankage, more filtration area, etc.).

The most glaring challenges downstream, according to Dr. Gottschalk, are biomass removal and protein capture. The former has, in fact, kept pace with rising titers, since depth filtration and centrifugation or depth filtration are scalable.

“The real problems begin after you have a clarified harvest containing fifty or a hundred kilograms of an antibody—a process that might have produced five or ten kilos just a few years ago,” he says. “Bind and elute steps, in particular, run into binding capacity limitations, and if protein increases by a factor of ten, then processors need approximately ten times the column size, ten times as much resin, and ten times the buffer.”

There are alternatives, such as splitting batches or cycling the capture column. Dr. Gottschalk refers to these measures as “inelegant, inconvenient, and costly,” since they consume multiples of the time, effort, and cost of a single batch. The benefits of scale—so compelling upstream, are lost during purification.

New facilities or major renovations can address these resource shortages, for example, by specifying greater buffer prep and hold volumes, larger chromatography columns, and utilities and floor space to match. But for existing facilities, adjusting for rising titers is not always possible.

“It is here we face bottlenecks that are very difficult to overcome,” Dr. Gottschalk admits. Bioprocessors are, consequently, thinking beyond protein A capture, for example, by replacing bind-and-elute with continuous chromatography, or with higher-capacity cation exchange resins. Others are investigating precipitation and extraction.

“Continuous chromatography requires a substantial capital investment,” observes Nihal Tugcu, Ph.D., a research fellow at Merck. “It will be difficult to implement using existing technology, but looking forward, we obviously will need to make such investments to meet the demands of high-cell culture titers.”

Dr. Tugcu’s group is investigating mixed-mode cation or anion exchange resins as a way to replace protein A and perhaps to reduce purification steps. “Mixed-mode resins provide selectivities we haven’t seen with regular cation exchangers,” she says. “We use multimodal purification not just on mAbs but on other therapeutic proteins, as well.”

Precipitation is an idea one hears from time to time. One interesting twist is to precipitate the impurities rather than the antibody, but this requires exquisite selectivity of precipitation conditions, “which is why people use chromatography,” continues Dr. Tugcu.

Polishing is another downstream operation that has kept pace with demands from upstream. Traditionally, bioprocessors relied on large ion exchange columns operating in flow-through mode. Today, they are adopting membrane adsorbers/chromatography for polishing whenever feasible. Disposable membrane capsules of one to two liters provide the flow rate and capacity of some 100 liter columns with only 5% of the buffer requirement, according to Dr. Gottschalk.

Most membrane adsorbers for polishing employ anion and cation exchange, but second-generation products are now utilizing hydrophobic interaction chromatography or salt-tolerant chemistries for removal of aggregates, DNA, host-cell proteins, and viruses.

Buffer filtration and buffer prep are among the most significant filtration bottlenecks, according to Stefan Egli, global product manager for filtration at Pall “With more transfer containers than prep tanks, processors prefer high-speed filtration from prep tanks into transfer containments.”

Pall addresses this need with filter cartridges incorporating Ultipleat® pleating technology combined with an optimized narrow cartridge core that speeds filtration or reduces filtration units, depending on customer preferences. This, in turn, frees capacity for buffer prep, so customers can quickly move on to preparing the next buffer.

Similarly, Pall has introduced a 0.2 micron Supor® UEAV cartridge, a polyethersulfone bioburden control filter offering high flow (up to 20 L/min), which allows smaller filter sizing or faster filtration. Pall is developing another PES high-speed filter for processors requiring sterilizing-grade filtration.

The company is also developing new tangential flow filtration technologies that it expects will allow linkage of downstream process steps to reduce production time. When available, this product line “will potentially get users a step closer to continuous processing,” says Jon Petrone, global technical director.

Pall has also introduced two products that reduce bottlenecking associated with column packing. The first is Resolute® columns, which incorporate “pack in place” technology that facilitates column maintenance. The other is a slurry transfer system through which a single operator can slurry sorbent from multiple containers with buffer and easily transfer it, through hydraulics, to a slurry vessel for column packing with minimal exposure of the sorbent to the environment. Pall claims this technology reduces transfer time by 80%.


GE Healthcare technicians pack an AxiChrom low-pressure chromatography system with MabSelect Sure, one of the company’s capture resins.

Previous articleResearchers Pinpoint Prognostic Markers for Prostate Cancer
Next articleGSK and Pfizer Create HIV-Focused Company