October 15, 2011 (Vol. 31, No. 18)
Rolf Ostendorp, Ph.D.
Marcel Ottens, Ph.D.
Erica Shane, Ph.D.
Andrew Zydney, Ph.D.
An Exclusive Q&A with Our Expert Panel
From the Editor in Chief
One of the most important bioprocessing advances in recent years has been the rise in levels of manufactured therapeutic proteins. Indeed, the overall average yield reported for commercial monoclonal antibody this year was 2.18 g/L—up from 1.94 g/L in 2009, according to BioPlan Associates’ 8th annual survey of biopharmaceutical manufacturers, which also noted that mammalian cell culture commercial production yields were lucky to achieve 1 g/L not so long ago.
The downside to this positive development is the oft quoted “downstream bottleneck.” Simply put, many separation and purification technologies have not evolved to the point where they can handle the increasing amounts of protein coming their way after leaving a fermentor or bioreactor. The degree to which this is perceived as a serious problem varies by biomanfacturer.
To help our readers who may be experiencing their own bottleneck issues, we’ve assembled a panel of bioprocess experts for this issue of GEN’s Tech Tips. Based on their experience and expertise, I’m certain that their comments, insights, and advice will be extremely useful as you work at optimizing your own upstream and downstream operations.
—John Sterling [email protected]
Does downstream processing introduce bottlenecks during the manufacture of biologic drugs?
Yes. Due to major advances in upstream process, today’s biological drugs can be produced at higher concentrations than before. A few years ago, a typical antibody process had a titer of two grams per liter at best. Today, titers of ten grams per liter or higher are common. Current fixed facilities can handle high titer processes with minor upstream modifications but require major downstream and facility modifications. This can be prohibitive due to high capital costs, facility shutdowns, production losses, and space limitations.
It doesn’t for vaccine manufacture, at least at our scale. We are running at 600 liter scale and do not experience bottlenecks. I suspect they might arise as our volumes increase, but right now our process volumes are still quite small.
Our products are purified through a combination of centrifugation and depth filtration for clarification, column chromatography, membrane filtration for DNA removal, and UF for buffer exchange.
Remember that vaccine batches are usually of much smaller volume than processes for monoclonal antibodies, where manufacturers have super-high yields and use huge capture columns and buffer volumes. But even for antibodies the bottlenecks usually occur with the processing aids rather than the product stream. For example, how do you prepare columns and how do you deal with the column and buffer volumes? Our largest column today is 20 L. Even if you have to flush with 10 column volumes it’s just 200 liters, it’s still manageable. The picture changes though for a 100 liter column, particularly if you have five different buffers.
When dealing with high-titer, high-cell density production, definitely yes. During the last decade, intensive research in proteomics and metabolomics provided a much better understanding of cell physiology. Bioprocess engineering converted this knowledge into far more efficient cell culture systems, which left the downstream side quite far behind. Titers have exploded and increased five- to tenfold over the last ten years, while workhorse purification methods such as protein A capture of antibodies have reached their limits.
Yes, especially with the high titers currently being achieved during monoclonal antibody production. Existing platform processes need to be adapted to cope with higher titers. But opinions on this differ. Some experts say that by proper facility fitting and more optimized processing, existing equipment can handle increasing titers, so novel purification techniques are unnecessary. Others claim that this might be true in some cases, but shifting to new technologies might yield more intensified processes and improve processing economics substantially, and thereby create some kind of competitive edge.
Downstream processing may be perceived as causing bottlenecks when the downstream equipment at a facility is not in balance with the requirements for handling the cell culture output. If the result is having to discard significant cell culture material, then I would agree that downstream would be the bottleneck. However, if processing high-titer material simply takes a bit longer than desired, whether or not one calls it a bottleneck depends on the impact of the increased cycle number on overall plant throughput.
As cell culture productivity or titer increases, the number of downstream processing steps that require “cycling” will rise. Cycling adds processing time and slows overall facility throughput. Product volumes may also introduce a challenge to a facility and usage of available tanks, especially if dilutions are needed to operate downstream steps.
Still, the question should be asked: Does the decrease in number of lots per year offset the increased processing time per batch? Each facility will need to answer this question independently as every facility has different constraints and must strike the proper balance between time and batch output.
Although I don’t particularly like the term “bottleneck,” there are clearly a range of challenges and opportunities facing downstream processing that are driven by major advances in upstream productivity over the past 10 years. This is true for many therapeutic proteins, particularly monoclonal antibodies, as well as many vaccines.
Which downstream unit operations are most responsible for bottlenecks and why?
Chromatography operations, particularly capture, exhibit the most constraints. High resin costs, low dynamic capacity, column size limitations, and buffer volume requirements are all responsible for downstream bottlenecks.
Right now none, but we’re anticipating a tenfold or larger increase in scale in the near future. To reduce volumes and thereby put less stress on downstream operations, we’re betting on yield improvements upstream, through fermentation and molecular biology.
At that point, depth filtration could become a bottleneck, specifically the availability of filter media, being able to source filters when we need them. We produce hemagglutinin-based influenza vaccines using baculovirus-infected insect cells. As we’ve experienced in the past, shortages do occur when you need large numbers of certain products. At much larger scale I would expect the situation to be more acute.
Cell harvest and capture are the predominant biomanufacturing bottlenecks, due respectively to significantly increased cell densities of 100 million viable cells per milliliter and product concentrations in the 10–20 g/L range.
Chromatography. The capacity for product capture in the same batch scheduling procedure is insufficient.
For mammalian cell culture, viral filtration is often a challenging fit. The shear number of filters and the requirement for 100% post-use integrity testing may create a pinch-point, as processing ceases while waiting for testing results. Another challenging unit operation is formulation, especially for higher concentration products. Protein concentration at this step is often higher than at any other manufacturing stage, which means that volumes are at their lowest. Balancing the extremes of protein concentration from the start to the conclusion of formulation using the same equipment can be challenging.
Initial capture chromatography is almost always demanding regardless of the mode of chromatography employed. But it can also be viewed as an opportunity to fix the batch mass by discarding “excess” material. Doing so can increase manufacturing predictability from batch to batch within a campaign, albeit at the cost of small yield loss.
I think the greatest challenges today are probably in the initial product capture chromatography step, typically protein A for monoclonal antibodies, and parvovirus filtration. The productivity of the initial capture step is determined almost entirely by the product mass. Although the dynamic binding capacity of protein A resins have increased over the years, these improvements have not kept up with the very large increases in product mass generated by developments in cell culture technology.
Parvovirus filtration is challenging due to issues of membrane fouling, low filtrate flux, and incomplete virus retention. However, there are certainly opportunities for significant improvements in other downstream operations to take full advantage of the increased product titers in reducing the overall cost of goods.
How can companies overcome bottlenecks, if at all?
Higher capacity resins and buffer concentrates can reduce the burden on purification process, as does the ability to regenerate today’s resins more times than in the past.
Higher capacity resins allow us to process more product using the same unit operations and without the need for new buffer and intermediate storage tanks. In a typical antibody process, the new resins can increase the productivity by 30 to 40 percent but they cost more.
Use of buffer concentrates can alleviate a good deal of buffer capacity issues. We’ve found concentrates to be especially helpful for high-titer processes requiring several cycles and large diameter columns. In these instances you can reduce buffer volume requirements by 30 to 40 percent.
Another strategy involves eliminating chromatography steps with other technologies that may require less space and buffers and also may be more cost effective.
Again, speaking from the perspective of a small-volume vaccine producer, bottlenecks will probably arise more from sourcing purification tools, including filtration and chromatography resins. So, good planning is a must, and a frank discussion with vendors should be part of this.
At the point where we’re producing ten times as much vaccine as we do today, outsourcing to a reputable contract manufacturer also becomes an option.
First, they must realize that downstream bottlenecks will arise earlier in the product’s life cycle than expected. In the past, you could expect bottlenecks during later-stage clinical development, during process optimization. Today they can come during early clinical development, when companies aim for a seamless scaling of initial upstream and downstream process steps to commercially viable production trains. The idea is to minimize the risk of lengthy and risky comparability exercises later on, which is known as the “front-loading” approach.
Furthermore, price pressures on cost-of-goods have increased considerably, which causes companies to focus on capital expenditures. In the future, low-titer large-scale stainless-steel processes for blockbuster drugs will be replaced by flexible high-titer small-scale processes in disposable bioreactors for more diverse nonblockbuster product portfolios.
Install larger columns, or more columns in parallel, shift to continuous processing. For example, counter current chromatography or simulated moving bed use less resin and buffers and are better able to concentrate product than conventional chromatography.
Another strategy involves the use of higher capacity resins, which resin manufacturers are constantly striving for.
A third possibility is to work in flow-through mode, through “negative” chromatography, instead of bind-and-elute. This captures only the impurities, obviating the need for high-capacity resins.
Continuing to look for novel purification technologies or vendor-improved technologies is extremely helpful in overcoming production challenges. Companies may find that they need to have multiple systems available for certain unit operations, which will be tailored to the needs of a particular program. In addition, operational strategies such as splitting batches during recovery, continuous processing (to avoid volumetric limitations of product pool tanks), and in-line dilution of buffers to accommodate tank limitations may also be explored.
There is definitely potential for optimization of existing downstream operations. However, it is likely to be difficult if not impossible to match the order of magnitude improvements that have been achieved in product titer simply by optimizing conventional unit operations. Instead, more innovative approaches will likely be needed that focus on novel downstream technologies that have inherent performance characteristics that overcome the limitations of column chromatography.
What role might alternative or nontraditional purification technologies play, if any, in resolving bottleneck issues, for example, crystallization, precipitation, or alternatives to protein A capture?
I think biotechnology companies will implement more of these technologies in the future, especially when the production mode changes from batch to semi/continuous and there is higher demand for cost reduction. However, due to batch operation and the need for platform processing and speed to market, I do not see any changes in the near future.
With regard to specific technologies like precipitation and crystallization, I do not believe at this point that they’re robust enough to apply globally, although they may be appropriate for selected products or processes. But moving forward, biotech manufacturing will come to resemble the traditional pharmaceutical industries that are characterized by lower cost of goods.
Viable alternatives to ion exchange or capture chromatography do not exist for us now. Membrane technology is still too expensive and inefficient to use for protein capture. The technology is not there yet. As a vaccine maker we develop purifications for our processes as they exist today; we are not experts at developing novel purification technologies. For that we look to vendors and service providers. We will certainly listen to them if they offer anything that might benefit us.
A generic problem solver among these techniques is not yet available. The big players are working on their next-gen approaches mostly in stealth mode, while the industry at large awaits commercially available systems. The hunt for alternative downstream approaches so far mostly resulted in a revival of the early days of recombinant protein production, such as protein extraction, precipitation, or crystallization.
Initial results, for example, from second-generation expanded bed adsorption technology or crystallization have been scientifically promising. However, proof of concept remains to be shown, specifically if such alternative methods can meet regulatory expectations and raw material supply and scale-up challenges, all in light of the increased price pressure.
As stated earlier, there is lively debate about this and many efforts of academia and industry are devoted to test these “new” technologies to replace protein A capture. One approach involves replacing protein A with high-capacity, mixed-mode ligands. These have higher capacity and selectivity, and are cheaper than protein A. Crystallization of mAbs is an intriguing possibility, but finding the right processing conditions has been difficult. Still, all major pharmaceutical companies have demonstrated that crystallization is possible, but it will not become a standard technique. A related approach involves precipitation as a first concentration and purification step. Genentech has done some very nice work in this area.
For the two examples I’ve provided, viral filtration and formulation, alternative technologies may not solve the problem. However, I do believe that recognizing these process challenges early on during process development, and then designing the process to alleviate pinch-points, is critical to good process design.
That being said, I believe that the alternative technologies mentioned can be useful for specific products. However, a true game-changer would be an alternative technology or a combination of technologies with broad application to multiple products.
Alternative technologies can play a role in resolving the challenges generated by advances in upstream productivity, although it is still difficult to say which of these technologies will ultimately be adopted and to what extent. However, the implementation of these technologies will require a concerted commitment by the industry to support, optimize, and implement these technologies across the manufacturing platform.
Do you agree that purification will always lag behind upstream productivity?
Upstream processing is straightforward. You can amplify cell lines, extend culture times, and obtain higher titers while still employing basically the same equipment and facilities, and given the choice between two grams per liter and five, the latter is usually better. But as titers increase, you’re adding more demands on purification. You require larger columns, or more of them, so you need different types of skids, as well as higher buffer capacity and more intermediate hold tanks.
Moving forward, improved resins with lower cost will be the key: resins with higher dynamic capacity and better selectivity that will allow reducing the number of chromatography steps, improving throughput by running at much higher linear velocity. Lower resin cost will reduce the chromatography process cost, which is, by far, the highest in downstream process.
Yes, it’s true because you can never get more out of a process than what you put in. The most substantial improvements will always come from upstream processing, and these will affect purification. Based on our upstream process improvement efforts, I can easily foresee a tenfold improvement upstream from molecular biology and cell culture. We might, at some point, be able to double our purification yields, but we could not very likely triple it, and a tenfold improvement is out of the question.
No, the downstream side will pick up as more and more research focuses on this area. Remember the impact of a single innovation such as protein A on antibody downstream processes. Maybe the solutions aren’t that far away if we can leave the beaten track and learn from experts in other fields, for example, the food industry, where cost pressures are critical.
There is no fundamental law dictating that this should be the case. If upstream productivity improves, the product will eventually be produced in a solid form, for example, inclusion bodies, precipitates, or crystals, which are easily purified and processed further.
Downstream processing has certainly been pushed hard by the significant strides made in cell culture. There have been major gains in downstream processing capacity over the past 10 or 20 years as well, such as higher capacity chromatography media and larger columns. But while these gains do not completely match the improvements in titer, the addition of a few more chromatography cycles is a small price to pay for overall improvements in upstream productivity. Nevertheless, at some point it may be worthwhile to view the process holistically, with a view toward overall productivity, rather than as an upstream versus downstream competition.
This may well be a matter of perspective. Since downstream processing always follows upstream, the requirements for downstream purification technology will always be driven by new developments upstream. However, the current challenges facing downstream processing are, to some extent, an historical artifact associated with the evolution and growth of the bioprocessing industry.
The basic technologies used for downstream processing, both column chromatography and membrane filtration, were already used at large scale in the food industry and were then adapted for use in the purification of early biotechnology products.
By contrast, cell culture technology for large-scale production of recombinant proteins had to be developed largely from scratch, including the discovery of many genetic and molecular approaches that have been employed over the last decade. Although significant advances in both chromatography and membrane processes have been achieved by re-designing these systems to meet the specific needs of the biotechnology industry, it is not surprising that these developments have lagged behind the enormous advances in upstream productivity through significant improvements in cell-line engineering, growth media, and bioreactor design.
If downstream purification processes had originally been developed with five or ten grams per liter product titers in mind, it is likely that the industry today would be focused on general issues of cost reduction and optimization across the entire manufacturing process instead of the current emphasis on downstream bottlenecks.