The future of industrial bioprocessing: Batch or continuous?

Abstract

For over 75 years, continuous bioprocessing has been investigated in both academia and industry as an alternative to batch bioprocessing. Many classic textbooks have major sections on continuous bioprocessing (e.g., Aiba et al., 1973; Bailey and Ollis, 1977; Blanch and Clark, 1996; Bungay et al., 1997; Shuler and Kargi, 2002; Wang et al., 1979; Yamane, 1995). Full of mass balances, kinetic expressions, and quantitative analysis of process alternatives such as cell recycle, these textbook sections make for both rigorous and interesting curriculum material. They often include equations for comparing the volumetric productivity of continuous versus batch bioprocessing. For most scenarios, such as when the cells are the intended product, continuous bioprocessing usually wins. One may then ask why many industrial firms still largely use batch rather than continuous bioprocessing. Excellent answers to this question have been provided by many of the same textbooks cited above. The following issues have been observed for continuous operations: (i) Lack of homogeneity in the continuous reactor vessel, including nutrient shortages or regions of cell accumulation, (ii) Challenges regarding long-term operability and maintenance of sterility, (iii) Poor short-term flexibility to handle multiple products, due to long run times, (iv) High cost and long times required for process development experiments using complex continuous culture systems in labs, and (v) Genetic instability of cells These observations point to scenarios wherein continuous bioprocessing makes sense, such as when it is desirable to have ongoing, evolution of a mixed population of cells, as in wastewater treatment, thus capable of consuming large volumes of variable feedstock year around. Another scenario wherein continuous bioprocessing makes sense arises when the product consists of optimal cell strain(s) that are highly evolved to grow under extreme conditions, such as Dunaliella salina, and thus there are few issues with contaminating organisms or genetic instability over long term culture in dedicated facilities. Other scenarios where continuous bioprocessing makes sense include production of unstable products, such as certain biopharmaceuticals, that would otherwise degrade upon incubation in batch culture. If a company already has process development platforms and manufacturing plants in place based upon continuous bioprocessing technology, it often makes sense to use these same platforms and plants for future projects, whether or not the products are unstable in batch culture. However, if a company already has manufacturing plants and process development platforms in place based upon batch culture technology, does it make sense for it to switch to continuous culture technology for stable products that can be successfully made in batch culture? The answer is clearly no. When recombinant CHO and hybridoma cell culture rapidly grew in the 1980's, several vendor firms marketed novel bioreactor systems for continuous perfusion cultures. These systems were tested by many product firms, such as Genentech, and found to have many of the same problems noted in paragraph two above. One pilot plant manager complained that “my definition of hell is a plant filled with those units”. Many product firms chose batch culture in stirred tanks for their standard platform. Some firms incorporated certain semi-continuous aspects into their batch systems, such as “repeated batch” (Yamane, 1995) approaches to seed trains, inoculum trains, and/or production cultures. Some firms now perfuse medium through production cultures, using an ultrafiltration membrane that retains the product in the bioreactor. Like the more traditional fed-batch approach, this is a variant of the batch approach, as the product is typically still harvested only once at the end of the culture. Back in the 1980's and 1990's, biopharmaceutical firms with unstable products had no choice but to work through the problems with continuous perfusion cultures. Some other firms, even with stable products, simply chose to start with continuous perfusion, and also worked through the problems. Some plant start-ups took over 20 tries at scale before achieving one fully successful, complete continuous perfusion run. Many firms that started with continuous perfusion to make stable products for early clinical trials eventually switched to batch for their late trials and commercial production. Early-stage process development and clinical manufacturing are often on the critical path to get a new product into clinical trials. Clinical delays due to process development or manufacturing issues can be very costly (including career wise!). Standard fed-batch culture services are offered by many biopharmaceutical contract manufacturers. Perfusion culture services are far less common and, even when found, are readily available only with the particular cell retention method already in use by that vendor. The compatibility of that method for a particular process may take time to resolve. Although many firms have successfully operated continuous perfusion for years, it has certainly not proven to be a more generally reliable production method than batch. Two major biopharmaceutical shortages occurred due to problems with commercial production using continuous perfusion systems (Brown, 2002; Dimond, 2009). With longer term perfusion, one might expect more problems with amino acid variants that arise due to mutations in product genes. One might argue that the biopharmaceutical industry will eventually evolve toward continuous processing, such as happened in the oil industry. This argument ignores the huge differences in required processing capacity and value of products sold. A small oil refinery processes about 50,000 barrels (∼8 million liters) of crude oil per day, resulting in roughly $5 million worth of refinery products and only a tiny fraction of the worldwide daily demand for such products. A large biopharmaceutical plant processes less than 40,000 L (∼250 42-gal bbl.) of harvested culture fluid per day, resulting in far more than $5 million worth of product and a large fraction or even >1× multiple of the worldwide daily demand for that particular biopharmaceutical. We will never need a biopharmaceutical plant that processes anywhere near 50,000 bbl. (8 M liters) per day and thus truly needs to be continuous on a capacity basis. Matthew S. Croughan Industry Professor Amgen Bioprocessing Center Keck Graduate Institute Claremont, CA, USA References Aiba S, Humphrey AE, Millis NF. 1973. Continuous cultivation. In: Biochemical engineering. New York: Academic Press. p 128. Bailey JE, Ollis DF. 1977. Design and Analysis of Biological Reactors. In: Biochemical engineering fundamentals. New York: McGraw-Hill. p 497. Blanch HW, Clark DS. 1996. Bioreactor design and analysis. In: Biochemical engineering. New York: Marcel Dekker. p 276. Brown SF. 2002. Growing drugs is a tricky business. Fortune. November 25. http://archive.fortune.com/magazines/fortune/fortune_archive/2002/11/25/333496/index.htm Bungay HR, Humphrey AE, Tsao GT. 1997. Biochemical engineering. In: Perry RH, Green DW, Maloney JO editors. Perry's Chemical Engineers' Handbook. New York: McGraw Hill. 24s. Dimond PF. 2009. Genzyme plant shutdown could mean up to $300 M in lost sales. Gen Eng Biot News. July 2. http://www.genengnews.com/analysis-and-insight/genzyme-plant-shutdown-could-mean-up-to-300m-in-lost-sales/57506615/ Shuler ML, Kargi F. 2002. Operating considerations for bioreactors for suspension and immobilized cultures. In: Bioprocess engineering, basic concepts. Upper Saddle River New Jersey: Prentice Hall. p 245. Wang DIC, Cooney CL, Demain AL, Dunnill P, Humphrey AE, Lilly MD. 1979. Continuous culture. In: Fermentation and enzyme technology. New York: John Wiley and Sons. p 98. Yamane T. 1995. Bioreactor operation modes. In: Asenjo JA, Merchuk JC editors. Bioreactor system design. New York: Marcel Dekker. p 479. Not yet. And while the above is not out of the question, we should instead ask: If we build a new biologics manufacturing facility, what should it look like? We have options. Certainly, the facility of the future should be built for the process of the future, and not for the process of the past or even the present. There are historical, business and technical reasons suggesting that the bioprocess of the future will be continuous with an integrated upstream and downstream; such a process will be high intensity, simplified and streamlined. Among the multiple advantages of continuous bioprocessing (Warikoo, 2012; Woodcock, 2014), we focus here on three—operational flexibility, product quality and cost. These are reviewed in the context of integrated up- and down-stream continuous processing, expanding the scope far beyond the familiar perfusion bioreactor technology. It is instructive to take a brief look at lessons learned from other industries that have adopted continuous methodology. Many have transitioned from batch to continuous manufacturing as they matured, for example, petrol, steel, glass, paper, chemical and, recently, pharmaceuticals. The underlying business drivers are common: flexible operation, high productivity and quality, decreased cost, smaller facilities, process integration/simplification and, in some cases, safety. The transition leads to the emergence of a “dominant” process design, widely adopted as an industry progresses towards commoditization (Utterback, 1994). In all cases, the transition to continuous manufacturing has been highly successful, resulting in a transformative change that has sustained industries through market growth and expansion. The initial implementation, however, was slow due to technical challenges and deeply entrenched mental models rooted in the old technology. While therapeutic protein manufacturing is not yet commoditized, our industry is rapidly maturing. On the way, we are facing multiple challenges: frequent drug shortages, volatile market demands, health care cost containment, and growing needs of developing nations. As also suggested by the FDA, these challenges can be addressed by a transition to more advanced and continuous manufacturing technology (Woodcock, 2014). Biotech company success strongly depends on operational flexibility to adjust production capacity in response to rapidly changing demand forecast, unexpected failures in late stage clinical trials or faster than expected clinical adoption. One needs to accommodate pipeline diversity, small-volume (<10 kg/year) and large-volume (>100 kg/year) products, as well as stable and labile molecules. A continuous process designed around an intensified and simplified bioprocess requires smaller equipment that can be scaled based on time and parallelization rather than volumetric expansion. Flexibility can be further enhanced with single-use technology. Standardization and portability of continuous processing would enable decentralization of manufacturing to address regional regulations. Smaller equipment of uniform scale can support process development, clinical production, launch and commercial manufacturing, which reduces technology transfer risk (Daszkowski, 2013). Moreover, the low residence time in a continuous process enables the production of both stable (e.g., MAb) and unstable proteins. For example, a single 500 L perfusion bioreactor operated at a cell density of 100e6 cells/mL for 280 days/year would yield more than 500 kg/year of product in reactor harvest (assuming stable productivity of 40pg/cell-day). A continuous chromatography system of modest size using disposable Protein A columns (e.g., 45 cm diameter) can deliver several tons of product per year. When properly configured, the same up- and downstream equipment can be used to produce small-volume drugs in the range of 10–20 kg/year. Continuous (perfusion) culture emerged 25 years ago because fed-batch technology was inadequate for production of labile proteins (e.g., enzymes and blood factors). The low residence time, <24h, in a perfusion bioreactor enables rapid product removal, thus preserving the high quality of the molecule. In an integrated continuous platform, these advantages are extended downstream, as non-value added unit operations and hold steps are eliminated (Konstantinov, 2013). Total processing time can be reduced to hours, and risks associated with degradation pathways (proteolytic, glycolytic, aggregation, oxidation, etc.) are minimized. Potential quality issues with stable molecules are addressed as well (Pacis, 2011; Robinson, 1994; Reid, 2010), as continuous processes may be operated with robust control, improving product quality and minimizing heterogeneity. The objective of decreased cost is maximizing profitability and patient access in expanding markets with pricing pressure. The minimized process footprint and the high level of automation in a multi-product continuous biomanufacturing facility leads to a significant reduction in capital and operating cost with benefits to both industry and patients. Recent modeling and financial valuation supports the expectation for major cost reduction in manufacturing of non-MAb drugs (Walther, 2013). The potential of cost reduction with large market MAb's is harder to demonstrate, yet possible, in respect to optimized fed-batch processes in fully depreciated facilities operating near capacity. In comparison, the continuous MAb processes are still far from technological maturity, but rapidly evolving. Recent results demonstrate high cell density of >100e6 cells/mL and productivity of 3.5 g/l-day (Clincke, 2013; Yin, 2014). This suggests a significant upward potential, which will challenge and likely outperform the established fed-batch paradigm. Commercialization of continuous bioprocessing requires resolution of both technical and organizational issues. Existing fully depreciated production facilities counterweigh the advantages of the new platform. While the long-term benefits of continuous operation are recognized, business priorities are often driven by short-term goals. Not the least, long established mental models, hardened by impressive personal and corporate successes with the old technology, create additional barriers for the acceptance of the new concepts. As a consequence, significant work is needed to convert the appeal of continuous processing into a formalized corporate strategy. The need for stable, high performance cell lines and media is not new, but perhaps more acute within the framework of high intensity continuous processing. Many continuous unit operations are commercially available today. Others, particularly downstream, need to be further developed to meet commercial expectations. Regulatory guidances reflecting continuous processing exist, but experience in regulatory process review and inspection needs to be established in collaboration with government agencies. The path forward is not dissimilar to the pharmaceutical industry, where various technical challenges have been addressed through the joint efforts and collaboration of industry, academia and government (Konstantinov and Cooney, 2015). Figure 1 illustrates the trend in bioprocessing over the last 20 years. The dominant design for stable protein production was established in the 1990s based on 10–20 kL stainless steel bioreactors and large volume purification columns. A decade later, biotech companies began to move towards smaller bioreactors (e.g., 2 kL) and columns operating at higher frequency with smaller processing lots. In this context, integrated continuous operation is seen as a transformative step on the process evolution trajectory, moving towards high intensification, smaller equipment, maximum capacity utilization and smaller processing lots. Returning to the question “should we close existing batch operations?” the answer, at least for now is no, but the factory of the future embodied in new facility design is likely to evolve around integrated continuous bioprocessing. Thus, the development of this new platform should be given serious consideration. Konstantin B. Konstantinov Late Stage Process Development Bio Realization, Sanofi R&D Framingham, MA, USA Charles L. Cooney Department of Chemical Engineering Massachusetts Institute of Technology Cambridge, MA, USA References Warikoo V, Godawat R, Brower K, Jain S, Cummings D, Simons S, Johnson T, Walther J, Yu M, Wright B, McLarty J, Karey K, Hwang C, Zhou W, Riske F, Konstantinov K. 2012. Integrated continuous production of recombinant therapeutic proteins. Biotechnol Bioeng 109(12):3018-3029. Daszkowski T. 2013. Continuous Processing in Biotech Production: An alternative to a modern single use, batch, facility? Integrated Continuous Biomanufacturing Conference, Oct 20-24, 2013, Barcelona, Spain. http://www.engconf.org/staging/wp-content/uploads/2013/12/7.-Daszkowski.pdf Konstantinov K. 2013. The Promise of Continuous Biomanufacturing (keynote presentation) Integrated Continuous Biomanufacturing Conference, Oct 20-24, 2013, Barcelona, Spain. http://www.engconf.org/staging/wp-content/uploads/2013/12/The-Promise-of-Continuous-Biomanufacturing-Barcelona-Oct-20-2013-final.pdf Woodcock J. 2014. Modernizing Pharmaceutical Manufacturing—Continuous Manufacturing as a Key Enabler (keynote presentation). International Symposium on Continuous Manufacturing of Pharmaceuticals, May 20-21, MIT, Cambridge, USA. http://iscmp.mit.edu/sites/default/files/documents/ISCMP2014_Keynote_Slides.pdf Konstantinov K, Cooney C. 2014. White Paper on Continuous Bioprocessing. J Pharm Sci 104: 813–820. Utterback JM. 1994. Mastering the Dynamics of Innovation. Harvard Business School Press. Clincke MF, Mölleryd C, Zhang Y, Lindskog E, Walsh K, Chotteau V. 2013. Very high density of CHO cells in perfusion by ATF or TFF in WAVE bioreactor™. Part I. Effect of the cell density on the process. Biotechnol Prog 29:754-767. Walther J. 2013. Building a business case for fully integrated continuous biomanufacturing. platform. Integrated Continuous Biomanufacturing Conference, Oct 20-24, 2013, Barcelona, Spain. http://www.engconf.org/staging/wp-content/uploads/2013/12/ICBM-2013-Jason-Walther-2.pdf Yin J. 2014. Advances in Integrated Continuous Bioprocessing. Cell Culture Engineering XIV, Quebec City, May 4-9, 2014. Robinson D, Chan C, Yu C, Tsai P, Tung J, Seamans T, Lenny A, Lee D, Irwin J, Silberklang M. 1994. Characterization of a recombinant antibody produced in the course of a high yield fed-batch process. Biotechnol Bioeng 44(6):727-35. Reid C, Tait A, Baldascini H, Mohindra A, Richer A, Bilsborough S, Smales C, Hoare M. 2010. Rapid whole monoclonal antibody analysis by mass spectrometry: An ultra scale-down study of the effect of harvesting by centrifugation on the post translational modification profile. Biotechnol Bioeng 107:85-95. Pacis E, Yu M, Autsen J, Bayer R, Li F. 2011. Effect of Cell Culture Conditions on Antibody N-Linked Glycosylation—What Affects High Mannose 5 Glycoform. Biotechnol Bioeng 108:2348-2358.