Where is Biopharmaceutical Manufacturing Heading?

Monoclonal-antibody (MAb) manufacture presents substantial current and upcoming challenges to the biopharmaceutical industry. This is true both from an economic point of view and from a technical perspective. The goal of producing sufficient quantities of high-dose antibodies, and the concurrent pressure to reduce cost-per-treatment, is culminating in a vigorous public debate. Recent presentations and publications suggest that manufacturing costs are particularly critical for scientists and engineers. So, where is the biopharmaceutical industry heading?

GE HEALTHCARE

MULTITON MANUFACTURING: WILL WE NEED IT?

Multiton scale manufacturing is on the horizon. Or is it? There's considerable doomsday talk about future demands in downstream processing and the industry's lack of preparedness. Yet industry trends suggest a need for reduced, rather than increased, production scales.

Gunter Jagschies, PhD

In a recent case study of large-scale MAb manufacturing, Kelley investigated a single-branded product need of up to 10,000 kgs per year.¹ He concluded that such production is possible with current technology. In addition, he saw little need to turn to unproven alternatives simply for cost reasons. Indeed, in his analysis, cost was not a major driver away from current technology. The caveat with Kelley's study may be that he assumed access to a facility planned and built for the production in question, and clearly not everyone has such access. He also assumed that downstream processing technology would make some evolutionary progress (e.g., in capacity) before such a hypothetical case would become reality, yet large-scale MAb manufacturing is already taking place. Finally, Kelley concluded that it was unlikely that this scale of production would ever be required.

It is important to note that Kelley was undertaking a case study to make a point. He was not intending to create a blueprint for the biopharmaceutical industry.

A study conducted by GE Healthcare BioSciences began by examining the production scale of currently marketed biopharmaceuticals (Figure 1). Investigators identified four general trends that could reduce the average production scale for novel protein drugs. These trends are: diminished likelihood of discovering blockbuster drugs, increased competition, dosage reductions, and diagnostic improvements.

Diminished Likelihood of Finding Blockbuster Indications

The chances of finding blockbuster indications are not improving. Many major medical indications are already served by several protein drugs, and more drugs are coming through the pipeline. Rheumatoid arthritis is one example; anemia and diabetes also fall in this group.

Increased Competition

Perhaps more important than the diminished need for blockbuster drugs is the issue of competition. Competition in the biopharmaceutical industry has increased significantly over the past decade. For example, in the early days of the industry, the first competitor to a first-to-market drug appeared, on average, eight to 10 years after the drug's market introduction.² Today, the first competitor typically appears much sooner—within 1.5 years. As a result, few individual branded biopharmaceuticals are likely to enjoy high market shares.

Quick Recap

Dosage Reductions

MAb-based medicines are high-dose drugs, and intensive research is underway to reduce dosages. This is being done to minimize patient risks and reduce drug manufacturing costs. Examples of this include the targeted manipulation of glycosylation patterns and the replacement of MAbs with smaller proteins (such as antibody fragments), with the desired effect of improved efficacy (for example, an enhanced ability to penetrate solid tumors).

Diagnostic Improvements

Evolving improvements in diagnostic tools are expected to lead to disease prevention, as well as earlier treatment, with less need for medication. One aspect of this evolution is the earlier identification of patients who are not likely to respond to treatment, and of patients who may experience severe side effects. Thus, such individuals would not be treated at all with the drug in question.

LOWER-SCALE PRODUCTION APPROACHES ARE NEEDED

For MAbs specifically, the current spectrum of marketed protein drugs (shown in Figure 1) indicates that production scales for a few of these have already reached ton levels. One example is Rituximab, a MAb used for treating B-cell non-Hodgkin's lymphoma, B-cell leukemias, and some autoimmune disorders. Rituximab is marketed as Rituxan by Genentech and as MabThera by Roche. For most MAbs, the scale of production ranges from 50 and 100 kgs to several hundred kilograms. An example of this would be Adalimumab, a drug marketed by Abbott under the name Humira for treating rheumatoid, psoriatic, and juvenile idiopathic arthritis, as well as Crohn's disease and chronic psoriasis. In addition, most recombinant proteins other than antibodies are manufactured at relatively small scales—as low as a few hundred grams per year, and rarely more than 10 kgs per year. The exception to this would be the various insulins, which are individually manufactured at lower-ton scales but which, as a group, may reach 10 tons of production annually.

Figure 1. Production volume and revenue map, 2006 data

The spectrum of marketed protein drugs includes the plasma proteins human albumin and intravenous immunoglobulin. Interestingly, these first-generation biopharmaceutical proteins are, as a group, manufactured at multiton scales—up to 500 tons per year for global albumin. Chromatographic purification has been used to manufacture some of these proteins since the 1970s, and it is expanding in this sector as a means of improving the manufacturing, purity, and safety of these drugs.³ This approach is feasible and economical because members of the plasma fractionation industry are among the most mature and financially pressured players. These companies are probably on a level with insulin manufacturers because insulins are typically manufactured using multiple chromatographic purification steps. Together, these two categories of biopharmaceuticals provide additional proof that today's technology can provide for patients' needs.

In the future, successful antibody manufacturing companies will have fewer ton-scale proteins (perhaps five to seven brands at the most), and will find economical ways to produce their much lower-scale antibody products (those of between 50 and 500 kgs per year, and rarely approaching 1,000 kgs per year). Although there will probably be few technical hindrances, a number of practical matters will require attention. One of these matters will be using existing facilities for product manufacture. Most of the first generation and many of the second generation facilities have been built with little flexibility to accommodate additional processes with different design or at different scale. Another will be much greater economic pressure resulting from two factors: pressure from healthcare systems to reduce reimbursement levels, and competition with biosimilars that reduce the margins of the most profitable protein drugs.

NEW PRODUCTS REQUIRE FLEXIBLE MANUFACTURING

Fortunately, there is still a very dynamic pipeline of new products in the biopharmaceutical industry. Figure 2 compares this pipeline as recorded by the Pharmaceutical Research and Manufacturers of America (PhRMA), first in 2004 and then in 2006.^4,5 Within two years, antibody-related projects grew from 76 to an impressive 160. The combination of various recombinant proteins—including growth factors, interferons, and interleukins—amounted to 94 projects in 2006, up from 57 in 2004. At an average success rate of 20%, these two parts of the pipeline will lead to more than 50 approved products in the near future. Additional products such as vaccines are likely to be added.

Figure 2. Dynamic portfolio driven by monoclonal antibodies

One conclusion to be drawn from these data is that increased numbers of companies will manufacture more than one protein drug. To achieve this, some companies will use contract manufacturing organizations (CMOs). Others will invest in new manufacturing facilities of their own. And a few will convert existing facilities originally designed for single-product manufacturing.

As more drugs enter the market, the need for multiproduct manufacturing facilities with sufficient flexibility to match the diverse production schedules of different drugs—both now, and in a dynamic, difficult-to-predict future—will increase. As described above, many of the manufactured proteins will be relatively small-scale products, so the issue of a lack-of-scale advantage will also need to be considered.

Figure 3 shows the production scenarios that may emerge when companies deal with an increasing number of diverse approved biopharmaceuticals manufactured at widely varying scales.

Figure 3. Emerging production scenarios. More companies will have more than one product in production, many of them small in scale.

TRADITIONAL FACILITIES VERSUS DISPOSABLE SOLUTIONS

The traditional "stainless-steel space" is a facility that includes fixed installations, such as several bioreactors at any scale from 100 to 25,000 L, combined with one or two purification suites with chromatography columns and membrane skids, together with the corresponding buffer and intermediate product storage tanks. For products required at ton scale, this type of manufacturing facility is currently used and will continued to be used in the future. Many mid-scale products are also manufactured in production units with fixed stainless-steel installations. Beginning with facilities owned by CMOs such as Lonza and Boehringer Ingelheim, however, some first-generation sites have been expanded to add capacity, or have been redesigned to become multiproduct facilities.

On the other hand, the "disposables space" is gaining importance for efficient manufacturing. For some time now, disposable production tools have been used in pilot plants to make material for preclinical trials or product and process characterization. In addition, current practice in clinical trials is to produce product in few batches often at 2,000-L bioreactor scale, within reach for disposable solutions. Disposables enable the manufacturer to be flexible and cost-efficient. No major upfront equipment investment is required and the company can avoid the delays associated with waiting for equipment to be installed and with validating cleaning regimes for operating this equipment in re-use mode.

However, disposables have some drawbacks. Neither disposable bioreactors nor single-use bags for buffer and product storage are regularly available, and many companies do not find them useful above 1,000 liters. In addition, because of the cost of replacing disposable devices after every production run, many companies limit the use of disposables in regular production to processes that run only a few batches per year. For large batch numbers—that is, for regular manufacturing that uses most of the available facility time—traditional re-use is still considered economically favorable. It is still common practice to make even the few clinical batches for Phase 1 and 2 trials in, for example, a 2,000-L stainless-steel bioreactor, mimicking the full-scale designs envisioned for later development stages. This practice is established because clinical manufacturing serves not only to formulate products for patients, but also to study the processes involved in manufacturing under commercial conditions.

IMPACT OF INCREASED MAMMALIAN CULTURE PRODUCT TITERS

One development that will impact production scenarios and may also change common practice is the increasing product titers achieved in mammalian cell culture, the dominant method of biopharmaceutical manufacturing. Because of low product titers of one gram per liter or even less, some protein drugs require large bioreactors and many batches to yield the required quantities. However, many projects currently in the clinic have reached product titers of three to five grams per liter.

This development opens up two options, both with potentially significant effects on the facility of the future. These options are: producing only a few batches with very large cell-culture volumes, or producing many batches with a reduced scale of operation.

Few Batches, Very Large Cell-Culture Volumes

A company may decide to manufacture the same product quantities in fewer batches using large cell-culture volumes, producing between 50 and 125 kgs of protein in one batch. This set-up could create facility time in the plant for manufacturing additional products or, in a less positive situation, it could create a problem with unused facility time. In addition, this scenario challenges downstream processing because the batch sizes envisioned here require high-capacity resins for all steps. Although such products are either under development or have only recently become available, the biopharmaceutical industry is concerned about downstream processing becoming a bottleneck.

Many Batches, Reduced Scale of Operation

As the second option, a company may reduce the scale of operation and manufacture the product at a 500- to 1,000-L bioreactor scale in many small batches. In principle, this option would allow much of the production to be run on disposable equipment—albeit at the cost of replacing such equipment for every batch. However, such a small-scale operation could be advantageous by requiring relatively low upfront investments and smaller plants, and by allowing more flexibility to add capacity or move production to other sites.

FIXED FACILITIES ARE UNDER PRESSURE

The classic fixed facility is coming under pressure. No longer is it sufficient for companies to demand that processes be adapted to a facility design that is "cast in steel." To get full value from its facility investment, a company should realize that rethinking its facility design is just as important as carefully creating a process design. In the changing business landscape, technical process improvements are providing new ways of operation.

Whether homemade or contracted, the custom engineering of production plants suffers from a lack of standardization in both hardware and automation. Wheels are re-invented, and available standard solutions are ignored. This situation often exists because, in making short-term decisions related to costs, companies ignore the long-term savings of standardization. The results are facilities that are adequate but too specific; thus, cost issues arise when a new process must be accommodated or a site transfer takes place between two facilities. Moreover, in a maturing biopharmaceutical industry that faces significant economic challenges, both the cost and time for completing an entirely new plant or a major reconstruction are viewed as heavy burdens.

What, therefore, has the biopharmaceutical industry achieved in performance and in manufacturing? Figure 4 summarizes a "best-in-class performance" study conducted at GE Healthcare BioSciences during the last two years. When this information is compared with GE Healthcare BioSciences' own performance-related data, the conclusion is that most companies can further improve their technical frameworks. Few companies have achieved full success. Those that have succeeded have multiple products in their facilities already and can look back on a decade of successful large-scale antibody manufacturing.

Figure 4. What is best-in-class performance?

THE BIGGEST DRIVER OF MANUFACTURING COST

Facility Use

Facility use is the most important factor in a good manufacturing economy. Using 80% of available facility time can be considered excellent, but new facilities rarely enjoy such high utilization from the beginning. For companies making a single product (and sometimes for other companies, as well), this percentage can be achieved only indirectly when assigning manufacturing to a CMO.

Cost of Sales

Cost of sales (CoS) represents the ratio of manufacturing costs to revenue. Some companies with portfolios of protein drugs report CoS as low as 15–16% of revenue. The lowest level of production costs per gram of antibody is allegedly approximately $100. However, such numbers invite skepticism. To maintain confidentiality, companies tend to give incomplete explanations for economic information in presentations or papers, or they use model calculations. Nevertheless, the models used at GE Healthcare BioSciences are based on a large variety of independent and publicly available input data points, and this information suggests that $100 per gram (including both fixed and variable costs) can be considered a top value.

Process Yield

After CoS, process yield is the next most important economic driver, as long as all products can be sold. Overall process yield should not be lower than 70%. Ideally, it should approach 80%.

Analytical Costs

Analytical costs are another significant cost driver for this industry. Successful companies have created analytical platforms that serve their entire product research and manufacturing ranges. More indirectly, hold times and batch times have been reduced to minimum levels in these companies. But analytical efforts result in more hold times or longer batch times. Proper risk management allows continued processing of intermediates without first waiting for this information.

Batch time is decisive in determining how many batches can be produced in a given facility time for a given product. The best values for upstream batch times seem to be 10 to 12 days in mammalian cell culture, and two days for the corresponding downstream process. These values allow parallel operation with six bioreactors feeding product to a single downstream production line (i.e., one that produces a new batch every second day).

Yield

While many approved processes still show product titers of approximately one gram per liter or even less, indications are that novel processes now reach three to five grams per liter. Even higher product titers are envisioned by some mammalian cell-culture scientists. To achieve basic economic competitiveness in the future, titers should probably be at least three grams per liter.

Protein A Media

For downstream processing of monoclonal antibodies, media capacity and lifetime are among the key cost influencers. For a Protein A–based capture step, the best capacities currently achieved are approximately 40 g/L. For subsequent ion exchange steps, the best values range from 80 to 100 g/L. If the antibody product is not bound to the medium, and if capacity can be reserved for impurities (flow-through mode), several hundred grams to several kilograms of antibody can be processed in one step, depending on where in the process sequence the step is located.

Even a relatively sensitive affinity medium with Protein A as ligand can tolerate several hundred process runs, if the best combination of medium and cleaning method is chosen. This is the key to the cost of the downstream process. Cleaning is best performed with solutions containing sodium hydroxide; a Protein A medium resistant to NaOH cleaning was introduced a few years ago specifically to achieve this goal. Cleaning with sodium hydroxide solutions is also desirable in terms of cleaning costs. The best published values for buffer costs (assuming large-scale buffer production) range from 30 to 50 cents per liter.

Finally, a number of best-in-class practices are closely related to avoiding unnecessary steps or hold times throughout the process. These include in-line buffer preparation or exchange.

CONCLUSION

Some industry experts see multi-ton MAb manufacturing looming on the horizon, but several important industry trends could actually reduce production scales. Therefore, lower-scale production approaches are needed. In the future, successful companies will have fewer ton-scale proteins and will look for economical ways to produce low-scale products. In addition, the pipeline of new products will require increased flexibility in manufacturing. The classic facility may become inadequate, and although companies will need to re-think their facility designs, they'll be reluctant to do so because of the time and money involved. A willingness to change will be essential, however, for everyone in the industry, including the manufacturers and suppliers.

Günter Jagschies, PhD, is senior director of strategic customer relations at GE Healthcare BioSciences, Uppsala, Sweden, +46.18.612.0000, guenter.jagschies@ge.com

REFERENCES

1. Kelley B. Very large scale monoclonal antibody purification: The case for conventional unit operations. Biotechnol Prog. 2007 Sep–Oct;23(5):995–1008.

2. Tufts Center for the Study of Drug Development. Outlook 2006. Boston: Tufts University; 2006. Available from: csdd.tufts.edu/InfoServices/OutlookPDFs/Outlook2006.pdf

3. Curling JM, editor. Methods of plasma protein fractionation. New York: Academic Press; 1980.

4. Pharmaceutical Research and Manufacturers of America. 2004 Survey: Medicines in development: Biotechnology. Washington, DC: PhRMA; 2004. Available from: www.phrma.org/files/Biotech%20Survey.pdf

5. Pharmaceutical Research and Manufacturers of America. 2006 Report: Medicines in development: Biotechnology. Washington, DC: PhRMA; 2006. Available from: www.bioimpact.org/userfiles/PhRMA%20Survey%202006%20-%20Biotech%20drugs%20in%20devpt.pdf