Technical and Economical Evaluation of Downstream Processing Options for Monoclonal Antibody (Mab) Production
Downstream process design can increase facility output through improved overall process yield or higher batch capacity in mass and volume.
Technology to manufacture high-dose therapeutic monoclonal antibodies (MAbs) at scale has evolved in several waves over the last two decades. Major cost reductions have been reported through the use of platform technologies upstream, downstream, and with the portfolio of analytical assays used.1 This review describes a reference downstream process developed at GE Healthcare2 and discusses technical and updated economical analysis of various process design options found in the literature and conference presentations.
Process Improvements
Antibody titres in mammalian cell culture have improved one hundred fold over the last 20 years, mainly through the ability to grow higher cell numbers (~10 fold) and through increasing productivity of those cells (~10 fold)3 (Figure 1). Recent reports from industry demonstrate that antibody titres higher than 1g/L are the rule, and 5 g/L readings are already on the horizon for processes currently under development. At the same time, manufacturers observe new challenges, resulting from the increased generation of antibody aggregates and increased presence of host-cell proteins from the high-producer cells used in such upgraded processes.
Figure 1.Twenty years of development in mammalian cell culture have led to significant increases in product titres for monoclonal antibodies (MAbs), as a result of higher expression levels, faster process times, and higher cell densities. Factors contributing to these improvements have been seen in the media for growth and the production phase, feed composition and strategy, process control and design, and host-cell engineering.
During the same period, purification by chromatography has improved its productivity by a factor of 10–20. This improvement was achieved mainly through the introduction of products allowing significantly shorter residence time at similar or higher binding capacity than previous generations of chromatography resins (Figure 2). Recent Protein A affinity resins and ion exchangers enable processing of 50 kg of antibody per batch per day using process schemes with a normal degree of optimization and without going to extremes in dimensioning the columns.4 In combination with up-to-date process engineering strategies that remove non-productive time from the critical path of manufacturing, these products are likely to resolve the near- to mid-term bottlenecking problem in the downstream process.
Figure 2. Twenty years of development in agarose resins have led to increased stability, higher capacity, and better volumetric flow.
It is important to realize, however, that significant improvement in process economy is still possible upstream, primarily by shortening the very slow cell culture process (2–3 weeks for harvesting)5 and by de-bottlenecking the link between cell culture and the relatively fast downstream process (1–5 day batch time, depending on degree of process intensification). Frequently, several (up to 5–6) large, staggered upstream lines are needed to feed one downstream production line.
It is likely that the protein mass to be handled per batch will increase further with some "blockbuster" MAbs, but almost certainly not with all future MAb processes (i.e., for most promising medical indications). As a result, further improvements are desirable downstream of cell culture. End product purity and safety, as well as process robustness, are not debatable and cannot be compromised when considering alternate process design options or potential economic advantages. It is essential that alternatives under consideration be reviewed based on their practical value for today's process design work.
It should also be noted that certain efficiency improvements in one stage of development may turn out to be costly or difficult to manage in later stages at full scale and with intensified processing. For example, using disposables is attractive for operating a multi-purpose plant and handling a large number of candidate MAbs in development, but is not necessarily more economical for regular, large-scale manufacturing. For certain situations, the savings created by using disposables may be at risk due to process changes between clinical phases. It is not easy to balance straightforward process economics against the advantages of an agile pilot plant.
Priorities Realizing Economic Advantages
A number of generalizations are possible and a priority scheme for seeking economic improvements can be developed (Figure 3). Work on technical improvements should adhere to that order of priority to secure the greatest impact.
Figure 3. Priority ranking of options to generate economic gains related to development and manufacturing
The industry is communicating the success of using "platform technology" approaches to process development routines and to the ultimate process design.1 A platform consists of verified technology for cell biology, cell culture, downstream processing, and analytical work as used for a specific category of target active drug substance such as MAbs. Platforms are regularly updated with new, verified technology. It has been reported that using such platforms has supported time savings of several months on the critical path to clinical trials. Using platforms has allowed some companies to reduce the need for process alteration prior to full manufacturing scale-up. Gaining three months on the way into the clinic can potentially save tens of millions of US dollars (MUSD) in net present value (NPV) for companies looking at an average MAb (~500 MUSD peak annual sales).6 In other words, deviating from the platform concept or not developing a platform at all is one of the most costly strategy issues MAb- producing companies can face in relation to manufacturing process development.
Many companies view the use of Protein A chromatography as a capture step essential to a successful downstream processing platform1 , and some results suggest that developing alternative capture steps, such as using small molecule ligands described as Protein A alternatives, may both take longer and result in reduced robustness.1,7,8 Just one failed batch as a result of a robustness issue would cost tens of millions of US dollars in lost sales value or a possible risk to the patient population.
A key economic driver relates to the output from the production facility, i.e., the quantity of product produced in it per time. Facility output is primarily a function of its utilization (expressed as percent of total facility time used), the productivity of the cell culture (key parameters are cell culture time and product titres), productivity of the downstream process (main parameters are capacity to process product mass and volumes), and "smart" process engineering solutions (through the ability to remove unproductive time from the critical path delivering product to the filling plant). Depending on these factors, the cost per gram of antibody produced seems to vary between 100 and 1,000 USD/g (Figure 4).
Figure 4. Shortening cell culture time and increasing facility use are two ways to reduce the cost of antibody produced. As seen in the table, reducing cell culture time from 14 to 10 days increases yearly production by 67 kg and reduces the cost per gram by 78 USD. The graph shows that increasing facility utilization from 20 to 100% can lead to a 5-fold reduction in the cost of goods.
Downstream process design can increase facility output through improved overall process yield or higher batch capacity in mass and volume. Higher yields enable the production of more drug product or a reduction in the number of batches required to satisfy the market need. Combining these changes could make room for production of additional products in the same facility. Use of recent technology as well as state-of-the-art process engineering will be necessary to achieve significant improvement in both of these categories.
The single most effective improvement of drug manufacturing costs comes from resin and filter re-use optimization, in spite of cleaning and validation costs that need to be considered as a result. This view is particularly valid for the Protein A step, but also applies to ion exchangers in large-scale production with high batch frequency. A scenario comparing a single-use membrane adsorber with re-use column chromatography was recently published.9 This paper discusses the cost comparison in great detail and includes all related costs in the evaluation, yet it uses an old ion exchange resin as the point of reference. Table 1 represents the main conclusion derived from this study as a cost comparison for 10 years of use together with a calculation based on a modern anion exchanger. Data were obtained following the same approach and using the raw data supplied in this comparison.9
Table 1. Importance of using recent technology in cost of use estimates: The most recent technology offers the best long-term cost of use at slightly higher initial raw material investment combined with re-use and improved mass and volume capacity (from ref.9).
Other incremental improvements of a process step, either through optimization of process parameters or exchange of one resin for another or for a membrane, will only produce small cost reductions unless any of the following items can be achieved:
- Overall process productivity improvements or time savings (satisfying higher product needs in an existing plant),
- Sizeable reallocation of facility capacity for other productive purposes,
- Reduced number of failed batches through improved robustness and lower production risk.
An example in line with the priority ranking shown in Figure 3 would be doubling cell culture titre and binding capacity for all steps in the purifications sequence, which could cut in half the number of batches required to meet market needs, thus releasing the remaining capacity for other manufacturing or alternatively to double the output of one product. Comparing capacities for first generation resins and recently introduced products makes it clear that such improvement is achievable when moving from old resins to new. A novel cation exchange resin has binding capacity of 80–120 g/L IgG at residence times of 2 to 6 min and typical pH and conductivity values for the feed stream. This represents twice the amount of useful capacity for SP Sepharose Fast Flow in a typical first generation process (Figures 2,5).
Figure 5. Large-scale Chromaflow columns are typically used in the downstream processing of recombinant proteins.
The Protein A step (in good company with virus removal filtration) is often considered the most cost-intensive step in current platform downstream processes. Even this step, however, represents only about 3% of total manufacturing costs. Replacing Protein A with a less costly resin will reduce the raw material cost, but will have little impact on the total product cost because of resulting productivity losses. There is a risk that such a small cost improvement will be accompanied by lower process performance of the new step; the benchmark would be tough: 95–97% yield and >99% purity. Similar constraints of improvement options would hold true for subsequent downstream steps.
Thus, efforts to improve downstream technologies should focus more on the ability of the downstream process to handle an ever increasing mass of drug product rather than on the raw material or step costs. We believe that this interpretation will remain relevant until there is a paradigm shift to entirely different unit operations that offer similar overall purification performance, safety, and robustness without constraints in mass and volume handling and at a cost of ownership below current costs for membranes and resins.
Emerging Reference Process for MAbs
Figure 6. Schematic overview of consensus process design (left), resins for the generation 2 process (middle), and other options that have been tested by the industry (right).Initial data about the design of a reference process were recently published and are used for research purposes of GE Healthcare Life Sciences (Figure 6).2,4 This is a second generation process using similar purification principles and step sequences to the current, most widely used platform process used to manufacture today's MAbs as described in the literature and at recent scientific conferences. The current most popular process may use resin and membrane products from different suppliers, but almost all are from the same generation as Sepharose Fast Flow. Consequently, for the most part, the latest generation of downstream purification technology is not being used.
The purpose of the development project is to address challenges such as the increase in upstream productivity of monoclonal antibodies in an integrated fashion (not solely in selected step improvements), and to remove downstream process bottlenecks now and in the near future. Notwithstanding other development projects and their possible outcome, it is our goal that the resin technology selected for this process will remain relevant for most processes developed in the next 5–10 years and be able to remain in full scale production for the next two decades with no need to alter the fundamental design. This goal coincides with the need to avoid major process changes that delay development and increase regulatory costs in the industry. Improvements offered by the new reference process compared with current popular practices can be described by two categories: process robustness, and productivity and costs.
Process Robustness
In any downstream process sequence, the capture step dominates robustness in many ways. Capture transfers the drug product into a stable environment, concentrates the dilute feed ~10 fold, and ideally removes most impurities that could compromise the stability of the drug throughout the remainder of the process time. It also minimizes the level of impurities to a low level handled reproducibly by subsequent polishing steps, independent of feed stream quality variations. The robustness period can be extended to include robust performance between various development projects, covering a wide range of monoclonal antibodies (i.e., a"platform").
Figure 7. The stability of a novel Protein A resin ("MAbSelect SuRe," GE Healthcare) following several hundred clean-in-place cycles with 0.1 M NaOH..
A novel Protein A resin, MabSelect SuRe for second generation MAb capture, has been studied from various aspects directly related to robustness: stability of the matrix under cleaning in place (CIP) conditions (NaOH), stability of the ligand on exposure to feed, and interaction with various antibodies. The resin exhibits long-term stability to CIP regimes using sodium hydroxide (Figure 7) and the ligand offers protease stability significantly higher than the level of classical Protein A (Figure 8).10 While Protein A consists of five different subunits with different binding specificities, the novel ligand is comprised of four identical subunits. This aspect has been used to explain the observation of more homogeneous elution behavior for a wide selection of antibodies and Fc fusion proteins compared with a classic Protein A resin.11 The subsequent two chromatography steps have been successfully tested for their ability to remove remaining impurities to current acceptance levels.10
Figure 8. The stability of the "SuRe" ligand (GE Healthcare, shown in red) and classical Protein A ligand (rPrA, shown in green) following incubation in CHO cell lysate for 18h at pH 5.0.
Productivity and Costs
Productivity and costs will vary between individual process designs developed by different groups testing the second generation process, depending on the specific antibody for which it is applied. Therefore, we have calculated a scenario with a 10,000-L fermentation volume containing 50 kg MAb (titre 5 g/L) using conservative assumptions for capacity and volumetric flow values comparing the first and second generation processes.4 The new reference process is designed to produce up to 50 kg of antibody per batch per day of processing time. The comparison indicates that it offers 45% lower operating costs calculated for downstream steps (harvesting to the final UF/DF step), and a reduction of time needed to process the MAb by 80% compared to the first generation process.9
Process Options
Process platforms described by different companies vary somewhat in the two post-Protein A steps. Compared with the design shown in Figure 7, some processes have a flow-through anion exchange step, usually with a Q ligand in position two. Others use hdrophobic interaction chromatography (HIC) or ceramic hydroxyapatite in one of the steps. In recent years, membrane adsorbers have been reported to be useful for the flow-through anion exchange step as an alternative to Q resin chromatography. Process step comparisons published for this step variant to date all use the first generation resins and are challenged by the second generation reference process described here (Table 1).9
Conclusion
Modern downstream process technology allows processing of batch sizes equivalent to ton scale production per year. There is no reason to assume that future antibody drugs cannot be produced with technology that has been introduced in recent years. This finding is not a contradiction to the permanent need for improvements of process performance and economy. Solutions presented here may provide development teams with efficient options to manufacture MAbs and sufficient time to develop a third generation technology adaptable to manufacturing needs as they evolve in the future.
Quick Recap
Günter Jagschies, Anna Grönberg, Tomas Björkman, Karol Lacki, and Hans J. Johansson work at GE Healthcare Bio-Sciences AB, Uppsala Sweden, +46.18.6120880, fax +46.70.6121247, guenter.jagschies@ge.com
References
1. Slaff G. Application of Technology Platforms to the Purification of Monoclonal Antibodies; presentation at BioProcess International European Conference, 2005 April 11–14, Berlin, Germany.
2. Grönberg A, Monié E, Murby M, Rodrigo G,Wallby E, Johansson HJ. A strategy for development of a mAb purification platform; poster presentation at BioProcess International European Conference, 2006 February 20–23, Prague, Czech Republic.
3. Wurm F. Production of recombinant protein therapeutics in cultivated mammalian cells. Nature Biotechnol. 2004; 22(11): 1393–1398.
4. Johansson, HJ. Advances in the Purification of Monoclonal Antibody-Based Therapeutics, presented at IBC Life Sciences Conference "Antibody Development & Production," 2006 March 1–3, Carlsbad, CA.
5. Werner R. presentation at 4th Annual Biological Production Forum, 2005 April, Edinburgh, UK.
6. Bezy P. Efficient Pilot Plant Utilization, presented at BioProcess International European Conference, 2005 April 11–14, Berlin, Germany.
7. Ambrosius D. Assessment of Purification Technologies, presented at BioProcess International European Conference, 2005 April 11–14, Berlin, Germany.
8. Ghose S, Hubbard B, Cramer SM. Biotechnol. Prog. 2005; 21: 498–508.
9. Zhou JX, Tressel T. Basic Concepts in Q Membrane Chromatography for Large-Scale Antibody Production, Biotechnol. Prog. 2006; 22: 341–349.
10. Johansson HJ, Lacki K, Brekkan E, Antti M, Bj�an T, Grönberg A. Multi-modal chromatography for purification of monoclonal antibodies, poster presentation at Recovery of Biological Products XII, 2006 April 2–7, Phoenix, AZ.
11. Ghose S, Allen M, Hubbard B, Brooks C, Cramer. SM. Antibody variable region interactions with Protein A: Implications for the Development of Generic Purification Processes. Biotechnol and Bioeng.2005; 92(6): 665–673.
