Monoclonal antibodies production platforms: an ...

Download PDF
15 downloads 0 Views 544KB Size Report
Comment
Sep 1, 2017 - non-protein-A chromatographic platform based on process economics† ..... Both loans will fund 50% of the corresponding costs, the remainder ...
Research Article Monoclonal antibodies production platforms: an opportunity study of a non-protein-A chromatographic platform based on process economics† António Lima Grilo†, Marilia Mateus*, M. Raquel Aires-Barros, Ana M. Azevedo iBB – Institute for Bioengineering and Biosciences, Department of Bioengineering; Instituto Superior Técnico, Universidade de Lisboa. Av. Rovisco Pais, nº1 — 1049-001 Lisbon, Portugal †Current Address: Biological Systems Engineering Laboratory, Department of Chemical Engineering, Imperial College London. South Kensington Campus, Exhibition Road SW7 2AZ London, United Kingdom *To whom correspondence should be addressed: [email protected] Keywords: Monoclonal Antibodies; Multimode Chromatography; Process Economics; Process Design

†

This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: [10.1002/biot.201700260]. This article is protected by copyright. All rights reserved Received: March 30, 2017 / Revised: September 1, 2017 / Accepted: September 11, 2017

1

 

Monoclonal antibodies production platforms: an opportunity study of a non-protein-A chromatographic platform based on process economics António Lima Grilo†, M. Mateus*, M. Raquel Aires-Barros, Ana M. Azevedo

iBB – Institute for Bioengineering and Biosciences, Department of Bioengineering; Instituto Superior Técnico, Universidade de Lisboa Av. Rovisco Pais, nº1 — 1049-001 Lisbon, Portugal †Current Address: Biological Systems Engineering Laboratory, Department of Chemical Engineering, Imperial College London. South Kensington Campus, Exhibition Road SW7 2AZ London, United Kingdom *To whom correspondence should be addressed: [email protected]

Abstract Monoclonal antibodies currently dominate the biopharmaceutical market with growing sales having reached 80 billion USD in 2016. As most top-selling mAbs are approaching the end of their patent life, biopharmaceutical companies compete fiercely in the biosimilars market. These two factors present a strong motivation for alternative process strategies and process optimization. In this work a novel purification strategy for monoclonal antibodies comprising phenylboronic acid multimodal chromatography for capture followed by polishing by ion-exchange monolithic chromatography and packed bed hydrophobic interaction chromatography is presented and compared to the traditional protein-A-based process. Although the capital investment is similar for both processes, the operation cost is 20% lower for the novel strategy. This study shows that the new process is worthwhile investing in and could present a viable alternative to the platform process used by most industrial players.

List of abbreviations ATF EQ EPC FCI FDA HCCF HCP HIC HMW IRR mAb NPV 2

Alternate tangential flow Equilibration Equipment purchase costs Fixed capital investment Food and Drug Administration Harvested Cell Culture Fluid Host cell proteins Hydrophobic interaction chromatography High molecular weight Internal rate of return Monoclonal Antibody Net present value

 

QA ROI SPD

Quaternary amine Return on investment SuperPro Designer

1. Introduction The biologicals market has had an outstanding performance over the last years and it is foreseen to keep growing at similar rates in the next few years with monoclonal antibodies (mAbs) increasingly dominating its growth [1–5]. A 2009 review on the business perspective of the mAbs industry splits currently marketed products into two tiers: 8 products belong in the first tier (those with sales above 1 billion USD/yr) and a second tier comprising of the remaining 13 products with sales bellow 1 billion USD. For the second tier sales are still quite high ranging between 17 and 472 million USD/year[6]. In Figure S1 (supplementary material), we present the global sales of the monoclonal antibody market from 2012 to 2016 showing an increasing market size at just over 80 billion USD on sales last year. The mAbs portfolio is known to have several products under development especially as most top-selling drugs are running off patent in 2 to 3 years. A 2012 study by BioProcess Technology Consultants suggested that demand for existing commercial mAbs was following an increasing trend and could nearly double by the end of this year reaching 13.4 metric tons in total [Levine, H. L., Global Demand and Utilization of Mammalian Cell Culture Manufacturing Capacity, in: Informa Eighth Annual BioProcess International Europe, Prague, Czech Republic, April 18-19, 2012. https://www.bptc.com/global-demand-utilization-mammalian-cell-culture-manufacturingcapacity/]. For new mAbs, the demand may reach 5 metric tons per new product per year according to the same analysis. The foreseen growth would require a productivity increase and hence appropriate purification strategies, presenting a great opportunity for a paradigm change downstream. The most important change sought by many industrial players is to move towards continuous bioprocessing be it upstream, downstream or end-to-end. These technologies have been studied from technical, environmental and economic perspectives over the last years with promising results. Pollock et al. have compared fed-batch to spin filter and alternate tangential flow (ATF) technologies in a comprehensive study including different production scales and different configurations considering the suitable downstream processing strategies [7]. Zydney has reviewed the technologies for both upstream and downstream [8], Kultz et al. have performed cost evaluations in end-to-end continuous vs. batch processes [9] and Steinbach et al. reviewed the use of continuous chromatography for biopharmaceuticals production, namely its economic savings [10]. 2. mAbs Process Development: the Hegemony of Platforms Purification of mAbs obeys, in most cases, to a platform strategy requiring one/two harvesting step(s) followed by one capture step accomplished by protein-A chromatography, a low pH hold for virus inactivation, two polishing steps, and finally a virus filtration and ultrafiltration/diafiltration (UF/DF) into formulation buffer thus being ready for packaging and 3

 

patient delivery [11–14]. This strategy, outlined in Figure 1A , is followed by most, if not all, manufacturers. 2.1 Upstream The early years and even recent studies on mAbs process development were far more focused on the upstream section leading to significant developments both in terms of new reactor operating modes (namely through the introduction of perfusion bioreactors) and much higher mAb titres. In 1992, CHO cell titres were reported to range from 90 to 550 mg/L [15]. During the 1990s and early 2000s, titres reached 1 g/L and later 5 g/L, which were considered as impressive titres for mAb production processes. A titre of 11.3 g/L was reported in 2011 [16] but some companies have even reported 10 to 13g/L in 2009 [12]. However, upstream progresses were faster than most scientists would dare to foresee. Indeed, right on the change of the decade Dutch company DSM patented the XD® technology able to achieve 240 million cells/mL and 27 g/L mAb concentration [Linz, F., High Titer Production. in: BioProcessing Network Conference 2011, Adelaide SA, Australia, October18-20, 2011. http://www.bioprocessingnetwork.com.au/PDFs/2011-present/Fritjof-Linz-bpn-2011.pdf]. Harvesting operations follow cell culture and are aimed at delivering a particle-free material, i.e., to remove cells. These operations are, therefore, gaining an increasing importance as cell densities increase. Harvesting can be accomplished by tangential flow micro-filtration (TFF), sedimentation, centrifugation or depth filtration [17,18]. Even if centrifugation may not remove all particles efficiently [19] producing no more than a “cellpaste” [17] it has been considered as the preferred technique for the harvest of mammalian cell culture broths [19,20]. Microfiltration could potentially yield a 100% cell-free permeate but at the cost of requiring high residence times due to limited control over the performance [17,19]. If centrifugation is not enough to achieve the desired clarification and the product is extra-cellular [17] a depth filter can be employed in conjunction with centrifugation. Depth filtration can remove some small particles and/or increase the average particle size by bringing some particles together. It has been reported that harvesting operations can account for as much as 25% of all downstream processing (DSP) costs [21]. 2.2 Downstream While upstream processing depends more on the experience a company has in some operations and/or on a particular host system over the others, purification strategies are very much dominated by platforms. The term platform refers to a pre-defined sequence of operations to be employed for all molecules with small adjustments. As some argue that platforms bring the advantage of dramatically decrease the cost of DSP studies in R&D, others claim that each molecule is different and therefore not all mAbs should be treated as equal. The first post-harvesting step is capture where mAbs are selected from the complex harvested cell-culture fluid mixture. The composition of such fluids includes a range of proteins, antifoams, DNA, sugars, acids, etc. and, therefore, capturing mAbs selectively is not an easy task. Protein-A affinity chromatography dominates capture technologies due to an almost unbeatable selectivity for mAbs achieving purities greater than 95% [19]. Even if it 4

 

contributed to a decrease in process costs when it was introduced [22], the use of protein-A affinity chromatography is now an economic challenge due to the high resin price motivating the development of alternative technologies as reviewed by Gagnon [23,24]. There have been some studies attempting at using non-chromatographic methods [25,26], cation-exchange chromatography (CEX) [27] or multimodal chromatography [28] for capture. Among these, phenylboronic acid multimodal chromatography has shown a good performance in capturing mAbs from cell culture supernatants [29,30]. Once mAbs have been recovered from the cell culture supernatant, further purification is required. Common remaining impurities may include host cell proteins (HCP), DNA, leached protein-A, mAb aggregates, clipped mAbs, mAb variants, viruses and a range of chemicals [19]. Usually two chromatographic polishing steps are employed which, for mAbs purification platform strategies, include ion-exchange (IEX) and hydrophobic interaction chromatography (HIC). CEX has been shown useful to clear leached protein-A, HCP and aggregates whereas anion exchange chromatography (AEX) is better for DNA and endotoxin removal [19]. HIC on its turn is usually employed to clear aggregates and, if used in bind-and-elute mode, it can also clear DNA and leached protein-A. Although these techniques are usually employed in packedbed mode, membrane chromatography and monolithic chromatography have been tested for mAbs DSP as well [31]. Monoliths are convective flow devices and therefore allow for working at higher flowrates [32] and can be produced in a variety of formats in order to fulfil bespoke requirements [33].

Although it is often claimed that protein-A is hardly replaceable because of its superior selectivity for mAbs, a direct comparison between two unit operations is hardly very insightful or useful. A fairer comparison is that between two processes: can they deliver the same product quality? At what costs? In this study, a platform method using protein-A, CEX and HIC for mAbs DSP is compared to an alternative strategy using phenylboronic acid multimodal chromatography for mAbs capture and monolithic AEX chromatography followed by packedbed HIC chromatography for polishing using technical and economic arguments. This analysis aims at showing how the alternative process compares to the traditional one and also how the alternative process could be optimized. The diagram in Figure 1B describes the newly proposed process.

3. Materials and Methods This study was performed using a computer simulation of both processes in SuperPro DesignerTM process simulator (Intelligen, USA). Data were obtained in several literature studies published over the last years. This work focuses on the DSP branch of the process as the upstream and harvesting are assumed to be the same. Costs for harvesting and cell-culture operations were estimated based on costs breakdown analyses previously published in literature. Data from process simulation and cost factors for cell-culture and harvesting processes were analysed together using Microsoft Excel® for a complete economic analysis of each process.

5

 

3.1 Process Simulation Data It is foreseen that mAbs demand grows up to ≤5 metric tons per new molecule per year for each new molecule entering the market after 2016 [Levine, H. L., Global Demand and Utilization of Mammalian Cell Culture Manufacturing Capacity, in: Informa Eighth Annual BioProcess International Europe, Prague, Czech Republic, April 18-19, 2012. https://www.bptc.com/global-demand-utilization-mammalian-cell-culture-manufacturingcapacity/]. Therefore, a demand of 2000 kg/yr is assumed. A state-of-the-art perfusion cell culture process able to deliver 25 g/L mAb was assumed. The harvested cell culture fluid (HCCF) includes mAb and HCP, high molecular weight (HMW) mAb species and DNA as main impurities. Based on data reported by DSM [Linz, F., High Titer Production. in: BioProcessing Network Conference 2011, Adelaide SA, Australia, October18-20, 2011. http://www.bioprocessingnetwork.com.au/PDFs/2011-present/Fritjof-Linz-bpn-2011.pdf], its composition is 25 g mAb/L, 300 mg HCP/L, 1.25 g HMW/L and 6000 µg DNA/L, respectively. Two capture steps were compared: protein-A affinity chromatography and phenylboronic acid multi-mode chromatography. Technical and economic data are shown in Table 1A. Data for protein-A chromatography were taken from studies by Natarajan et al. [34] and Ghose et al. [35] and also from current practice in the authors’ laboratory. Technical data for phenylboronic acid chromatography were taken from recent works [29,30] and, when required, economic data were considered similar to average values for other resins as reported by Shukla and Yigzaw [36]. After capture, a viral inactivation (VI) by holding at low pH is required. VI is accomplished with acetic acid [34] aiming at achieving pH 2 under agitation for 60 minutes. Tris is then used for neutralization [37] for 20 more minutes. The following unit-operation is UF/DF aimed at conditioning the viral inactivated capture pool via (1) reducing the conductivity of the protein solution and (2) obtaining a mAb concentration between 10 and 20g/L. This operation is necessary because the following step is an ion-exchange chromatography which requires that solution conductivity be below 5 mS/cm. After UF/DF an absolute filtration is employed and the material is pooled to avoid product quality inhomogeneity. Data for both filtration types employed in both processes are shown in Table 2A. As mentioned above, the first polishing step is an ion-exchange chromatography. For the traditional process, a CEX column is employed whereas the new proposed process includes a monolithic anion exchange chromatography step using a quaternary amine (QA) as ligand. CEX equilibration (EQ) buffer is 250 mM phosphate at pH 6.0 which requires both NaH2PO4 and Na2HPO4. The latter shows a stronger concentration vs. conductivity dependency and 0.5% is enough to achieve the imposed maximum of 5 mS/cm [Emerson Process Management, Conductance data for commonly used chemicals, Irvine, USA, 2010. http://www2.emersonprocess.com/siteadmincenter/PM Rosemount Analytical Documents/LIQ_MAN_6039_Conductance_Data_Commonly_Used_Chemicals.pdf]. Data for CEX chromatography was taken from the work of Shukla and Yigzaw [36] and for monolithic chromatography information was taken from the work of Nascimento et al. [32]. It is assumed the increased purity of monolithic chromatography outlet stream is due to HCP removal since 6

 

the analysis evaluates variations in total protein content. The two strategies aim at removing impurities, namely HCP. A comparison between the two is presented in Table 1B. The second polishing step aiming at removing HCP and remaining HMW species is HIC used in flow-through mode. The HIC strategy is the same for both processes. It consists of an equilibration step with 250 mM sodium citrate, 25 mM Tris buffer at pH 8.0 for 3 CVs. Loading takes place afterwards at 21g HCP/L-resin and the linear velocity is set at 300 cm/h. Then, the material is washed with 3 CVs of the same buffer used for equilibration. Elution takes place with 25 mM Tris and 50 mM sodium citrate at pH 6.0 for 10 CVs. Finally, 3 CVs of water are used for stripping. The lifespan of the column is assumed to be 100 cycles. The difference lies on the column diameter: the traditional process uses a 77 cm wide column whereas the new process requires a wider column (93 cm diameter) but both columns have 10 cm bed height. It is assumed that both resin costs are 1000 USD2007/L-resin [35], value which is updated to 2016 prices assuming an average annual inflation rate of 2.4% and estimated to be 1250 USD2016/L-resin. Prior to loading the HIC chromatography column, the material undergoes a new buffer exchange into 500 mM sodium citrate and 25 mM Tris. Again, the strategy is very similar in both processes, as shown in Table 2B. Finally, the protein is buffer exchanged into formulation buffer again by UF/DF. The formulation buffer is assumed to be citrate buffer at pH 5.0 containing 5% sucrose aiming at creating high stability conditions [14]. After this operation the target mAb concentration of 70g/L is achieved. This value is taken as the average of a range of products in the market according to data provided by their manufacturers (Table S1, in supplementary materials). Data for the formulation buffer exchange by UF/DF for both processes is shown in Table 2C. Please find the simulation flowsheets (Figures S2 for the traditional process and S3 for the new process) and the corresponding Gantt charts (Figure S4) as supplementary material. 3.2 Economic Analysis Structure One of the first parameters to be defined when analysing the economic performance of an industrial project is its lifetime. This is a biopharmaceuticals project and hence it will be assumed that the molecule to be produced has been patented and that the company holds all rights for the product for 25 years. It will be assumed that when the molecule is ready to enter production, it can be marketed under patent rights for 15 years. Another very important parameter to be obtained ahead is the inflation rate. For the purpose of estimating costs and other relevant economic figures in this work, it will be assumed that the year of investment is 2016 and the average annual inflation rate between 2000 and 2016 is 2.4% [InflationData.com, Long-term US inflation (2013). http://inflationdata.com/Inflation/Inflation_Rate/Long_Term_Inflation.asp (accessed June 12, 2016)]. As previously mentioned, the biologicals market has been growing and it is foreseen to keep growing with mAbs dominating it. It is foreseen that by 2020 there will be 70 mAbs in the market with global sales of 125 billion USD [38]. As previously mentioned, the demand for this molecule is set at 2000 kg per year and the selling price is 4000 USD2008/g. This leads to overall 8 billion USD2008 per year in sales which, under the same inflation rate assumption, 7

 

translates to 9.7 billion USD2016 in sales revenues, making it a market blockbuster. This figure is not unrealistic: 5 mAbs have had sales between 6 and 16 billion in 2016. A significant part of process development costs is related to the R&D required to bring a product to the market. Most companies report a value in the range of 1.3 to 1.9 billion USD spent per new drug [39,40]. However, a different analysis has been published in Forbes magazine, which shows that companies hardly spend less than 4 billion USD per new molecule. For companies with 4 to 6 approvals a year, average expenditures reach 5 billion USD which is the value assumed in this simulation [M. Herper, The cost of creating a new drug now $5 billion, pushing big pharma to change, Forbes. (2013). http://www.forbes.com/sites/matthewherper/2013/08/11/how-the-sta…g-cost-ofinventing-new-drugs-is-shaping-the-future-of-medicine/; M. Herper, The truly staggering cost of inventing new drugs, Forbes. (2012) 38 http://www.forbes.com/sites/matthewherper/2012/02/10/the-truly-staggering-cost-ofinventing-new-drugs/] which translates to about 5.4 billion USD at 2016 prices. As explained before, the simulation included the DSP part of the process only. In order to account for upstream processing and harvesting costs, it is necessary to establish how those relate to the DSP cost. First, it is considered that harvesting accounts for 25% of all DSP costs [21] and therefore the simulated costs account for no more than 75% of the total DSP costs. It is then considered that the whole DSP section accounts for 60% of the overall costs and upstream processing accounts for the remaining 40% [41]. The simplified framework described by Peters et al. [42] for economic analysis of chemical industry projects was used to estimate all major components of the economic analysis. Under this framework most costs are estimated as a percentage of the equipment purchase cost (for capital costs) or the costs of raw materials and consumables (for operating costs). The costs of the DSP section (equipment purchase costs and raw materials/consumables) were obtained from the SuperPro Designer (SPD) simulations. The overall costs for the entire processes were estimated based on those using the cost factors for harvesting and upstream sections, as indicated above. Then, a cash-flow model was built in order to allow the calculation of the relevant parameters for project comparison. In simple terms, the costs can be split into what is commonly called capex (capital expenditure) or total capital investment using the terminology employed by Peters et al. and the opex (operating expenditure) herein called total product costs or simply operating cost. Both components are further decomposed and detailed bellow, following the methodology proposed by Peters et al.

8

 

3.2.1 Capital Investment Costs The capital investment costs refer to the fixed costs associated with the process including: •

FCI (fixed capital investment) — accounts for tangible assets (such as plant or equipment) which is further divided into o

Direct costs — directly related to an investment  EPC – Equipment purchase costs (output of SPD simulation)  Installation (47% of EPC)  Equipment related costs • Piping (68% of EPC) • Instrumentation (36% of EPC) • Insulation (8% of EPC) • Electrical facilities (11% of EPC) • Buildings (18% of EPC) • Yard improvement (10% of EPC) • Auxiliary facilities (70% of EPC)

o

Indirect costs — indirectly related to an investment  Engineering (33% of EPC)  Construction (41% of EPC)  Contractors fee (22% of EPC)  Contingency fee (44% of EPC)

•

Working capital – described by Peters et al. as the term accounting for stock raw materials, supplies, finished product and semi-finished product; cash kept for monthly payments; accounts payable and taxes payable (89% of EPC)

•

Interim interest – interests to be paid in the period between start of using borrowed money and start of production, i.e., when loan payment starts. (details on estimation provided under Investment Plan

•

Up-front R&D costs and royalties (estimated at 5.4 billion USD)

SPD has an internal equipment price database. Prices are for year 2000 [Inteligen, SuperPro Designer Manual, 2001 doi:10.1201/9781420026221.fmatt]. The user can therefore input the year of analysis (2016) and an average annual inflation rate (2.4% as mentioned above) for the software to estimate the current prices.

9

 

3.2.2 Total product costs (Operating costs) Total product costs (also referred to as operating costs or opex) are the costs inherent to operating the plant. These will encompass: •

Direct production costs o Raw materials and consumables (output of SPD simulation) o Labor (15% of the total product costs) o Supervisory labor (15% of labor) o Laboratory QA/QC (15% of labor) o Utilities (10% of total product costs) o Waste treatment (output of SPD simulation) o Maintenance (5% of simulated FCI) o Variable royalties and patents (1% of total product costs)

•

Fixed charges o Depreciation (calculated using SPD equations, see below) o Insurance (0.7% of simulated FCI)

•

Plant overhead costs o Medical services, safety and protection, restaurant, recreating, laboratories, storage (50% of labor + supervisory labor + maintenance)

•

General expenses o Administrative costs (20% of labor + supervisory labor + maintenance) o Distribution and marketing (20% of total product costs) o Variable royalties (5% of total product cost)

SPD calculates the annual demand for raw-materials, consumables, etc. and uses userinput prices to calculate the annual costs of those. The SPD economic calculations assume that depreciation (d) is calculated as a function of the salvage fraction (fs), the start-up and validation cost (S) and the FCI from equation (1). 𝑑𝑑 = (1 − 𝑓𝑓𝑠𝑠 )𝐹𝐹𝐹𝐹𝐹𝐹 + 𝑆𝑆 − 𝑓𝑓𝑠𝑠 𝐹𝐹𝐹𝐹𝐹𝐹

(1)

A fraction corresponding to 1/N of this value (where N is the lifetime of the project) should be added as a negative cash-flow. In the case of this project, the start-up costs are assumed to be 5% of the FCI and the salvage fraction, which accounts for the value of the equipment after the project lifespan is 10%. This is accounted for as an earning after the project lifetime is over, i.e., during year 15. Royalties are usually treated in the basis of alliances between companies and usually, royalties fees are agreed on a sales percentage basis. Common values are in the range of 5 to 15% of sales depending on the stage and on the success of the project [43]. An average value of 5% is assumed, in line with the suggestions of Peters et al. [42]. 10

 

3.2.3 Investment Plan The investment plan is a map of how investment will be spent in time. It is considered that 12 months (year 0) are required for the plant to be able to start producing. For simplicity, it will be considered that they are all in the same year. During the investment phase the project designer will detail on the time-spans of the following investments: •

• • • • • • • • • • •

Equipment project and construction — estimated as 15% of the total equipment purchase costs; assumed to be paid in 3 main instalments of 25% each and a final instalment split evenly over the remaining 9 months Land plot — estimated as 6% of the EPC; paid in 1 instalment Buildings — spent over 6 months EPC — from SPD simulation; assumed to be spent over 6 months Installation — spent over 6 months Piping — spent over 8 months Utilities and services — spent over 5 months Instrumentation — spent over 3 months Electrical facilities — spent over 7 months Start-up and validation — spent on the last 2 months before operation starts Contingency — spent over 12 months Working capital — 1 instalment one month before operation starts

3.2.4 Financing Although the World Bank assumes an annual interest rate of 3.3% for the US market [World Bank, Lending interest rate, (2016). http://data.worldbank.org/indicator/FR.INR.LEND (accessed June 10, 2016)], these are times where interest rates have gone unusually low and hence it has been decided to use a higher rate of 5% for the two loans considered. It is considered that R&D expenses have been spent over 20 years prior to the start of production. Once in the market, the product must pay this cost off and, hence, the R&D loan (capital and interest) is considered an upfront cost. Capital costs related to working capital and start-up and validation are spent over 12 months prior to the start of production. The corresponding loan (including the interim interest as explained below) is to be paid over 10 years. Both loans will fund 50% of the corresponding costs, the remainder being self-funded. It is assumed that both loans are paid in 10 equal instalments during the first 10 years of operation. Interim interests are paid to the bank for using money in the period where the self-funds are over but production has not started yet – and hence the loan will not start to be paid. Once the investment plan has been designed, it is possible to find where the selffunding will be over. The interim interests will be paid for the remaining n months. If Iif is the interest free investment, a is the fraction of self-funding investment and j is the interest rate, the interim interests (J) are calculated by equation (2). 11

 

𝑗𝑗 × 𝑛𝑛 12 𝐽𝐽 = 𝐼𝐼𝑖𝑖𝑖𝑖 𝑗𝑗 × 𝑛𝑛 1 − �(1 − 𝑎𝑎) 12 � (1 − 𝑎𝑎)

3.2.5 Cash Flow Model

(2)

All information mentioned above is then used in a cash-flow model. All positive and negative cash-flows are used as model inputs and, from this model, it is possible to calculate project evaluation measures, namely the internal rate of return (IRR) the net present value (NPV), the payback time (PBT) and the return on investment (ROI) which will provide de the desired project comparison indicators. For this purpose, a stand-alone model is assumed, i.e., it is considered that this project is the only project the company has in its portfolio. Federal corporate income taxes in the USA are applied according to legislation [Department of the Treasury - Internal Revenue Service, US Federal Income Tax, Instr. Form 1120. (2015). https://www.irs.gov/pub/irs-pdf/i1120.pdf (accessed June 12, 2016)]. The total federal tax is calculated by adding a “fixed tax” to a percentage of the yearly income and depends on how much income is made in each year. Another component of the corporate tax is the State corporate income tax, which depends on the State the plant is built in. Having the possibility of operating in states where zero tax applies (states of Nevada, Washington or Wyoming) [Tax Foundation, State Corporate Income Tax Rates and Brackets for 2016, (2016). http://taxfoundation.org/article/statecorporate-income-tax-rates-and-brackets-2016 (accessed June 12, 2016)], this possibility is considered. It is assumed that in the first year of operation, the plant produces at 80% of its installed capacity, increasing to 90% on the second year and to full capacity on the third year and onwards.

4. Results and Discussion Traditional and new processes are compared in terms of their capital investment costs (Table S2) and total product costs (Table 3). The capital costs of the platform process are very similar to those of the newly suggested process. The most significant share of capital costs for both processes accounts for the up-front R&D costs which are two orders of magnitude greater than all other costs. If R&D costs are set aside, there is still no capital costs difference between the processes. This is surprising but very promising regarding the future application of the newly suggested process. It would be better if a reduction could be achieved and it might be achievable via process optimization. Regarding total product costs, our analysis shows that the new process provides a 44% cost reduction versus the traditional process. It is important to mention that the method proposed by Peters et al. for process economic analysis has a drawback for the type of study herein shown: as referred above, almost all contributors for the operating costs are estimated as a percentage of the costs of “Raw materials and consumables”. The traditional process employs the expensive protein-A 12

 

affinity resin which leads to significantly higher costs of consumables and hence increases all other costs if this method is used. In other words, the cost of protein-A propagates throughout the analysis. A fairer comparison can be performed using •

• •

the original figure for “Raw materials and consumables” from the simulation but keeping the other direct production costs equal to those of the newly suggested process; using the original fixed charges which depend on the capital costs of the process; using the general expenses figures from the newly proposed project because they are unlikely to be much different.

Using this alternative estimation methodology, the cost reduction of non-protein-A processing with respect to the protein-A platform processing is 20% which is still quite significant. The capital investment costs do not change and the total product costs are still 20% lower in the proposed non-protein-A processing (as seen above). The global cash-flow analysis allows for the calculation of several important economic indicators: the NPV, the IRR, the ROI and the payback time. The NPV is the sum of all project cash-flows “brought” to the year of analysis at a given discount rate. Such rate accounts for the opportunity cost of capital for the investors. In this analysis 7%, 9% and 11% discount rates are used. The IRR is the rate that zeroes the NPV. It can be interpreted as the opportunity cost of capital at which the project is no longer worth investing in. In a simple analysis, it can be said that the investors are looking for the highest possible IRR (although having the IRR is not enough reason to opt-in a project). Both the chosen discount rate and the IRR can be compared to the return rates of other possible capital applications, being also important to consider the risk of each application. The ROI is how much money is made per each unit invested, i.e., how much dollars (in this case) are obtained per each dollar invested. Finally, the payback time is the time that it takes to obtain a zero cumulative cash-flow. It is only then that operation yields a real profit. The economic measures of the traditional process when the above-mentioned assumptions are taken into consideration are slightly more favourable: the NPVs are 42.3, 36.4 and 31.5 billion USD for 7%, 9% and 11% discount rates respectively. The same trend is followed by the ROI (4.9, 4.1 and 3.4 respectively) and the payback periods (2.73, 2.77, 2.82 years respectively). The IRR is now 68%. The NPVs for the newly proposed process are 42.9, 36.9 and 32.0 billion USD for 7%, 9% and 11% discount rates a trend also followed by the ROI values (5.0, 4.3, 3.5) and the payback times (2.71, 2.75, 2.80) with a IRR of 69%. This analysis is not only more adequate for comparison purposes but also allows concluding that capital investment plays a very significant role in project economics. The cumulative cash-flow profiles for both projects are presented in Figure 2 A and B, respectively for the traditional and the new process and all economic indicators are shown in Figure 2C. For all discount rates used, the NPVs of the traditional process are lower than those of the new project which is a first indication on why to choose this process. The IRR of the newly proposed project is slightly better, yet very comparable to that of the traditional process, 13

 

reaching 69%. The payback periods are slightly shorter for the new process but still very comparable. In summary all economic figures indicate the option for the new process would be more advisable. The more expensive traditional project costs will translate into less efficient process.

5. Concluding Remarks Paradigm changes and other routes for process development and optimization are currently highly sought after by all drug manufacturers. The main driving forces for changes are the approaching end of the patent life of blockbuster drugs and consequent increase of the biosimilars market. In this study we suggest a new purification strategy using phenylboronate multimodal chromatography, monolithic AEX chromatography and packed-bed HIC as alternative to protein-A-based purification strategies and we compare the economic performance of both processes. The results of this work show that it is worth investing in finding alternatives for the purification of mAbs. The so-called platform process has been used for a long time and is very robust. However, its economic performance can be challenged by new technical options recently studied, namely multimodal chromatography and monolithic anion exchange chromatography as alternatives for capture and polishing of mAbs, respectively. Not surprisingly, chromatography-associated costs dominate the overall operational costs. Therefore, any process parameter (such as the host cell protein load from upstream) that may affect the lifespan of chromatography columns may significantly impact the economic performance of the processes). In part, the much higher product costs calculated for the traditional platform process over the newly proposed result as a consequence of using protein-A for mAb capture. In the second part of the analysis it is shown how the estimation methodology could be adapted to correct this discrepancy and obtain a faired comparison. Although the advantage goes to the new process, the sceptic advocate could argue that the process is not developed enough, as of yet, nor has it any history of industrial applications. In authors’ view, this work shows an opportunity, supporting that it is worth investing in this new process, obtaining more data and performing more detailed comparisons from both economic and technical perspectives. The capital cost investment is what controls the economic performance of both projects and, since these are almost equal for the two projects under consideration, their economic indicators are, likewise, very similar, especially when the higher consumables cost propagation is not included in the analysis. It cannot be neglected, nevertheless, that the new project is able to provide a 20% operational cost reduction. This opens good perspectives for the optimization of the new process focusing on the technical feasibility at industrial scale and cost reduction at both capital and operating levels. In conjunction with previous technical work by Santos et al. [29], Rosa et al. [30] and Nascimento et al. [32] this work aims at capturing the attention of the academic and scientific communities for alternative unit operations in mAbs process development, particularly, to the use of multimodal and monolithic chromatographies in the downstream processing of monoclonal antibodies. 14

 

Acknowledgments The authors acknowledge the support of Pedro Pereira for assistance and valuable discussions while performing SPD simulations. Funding: This work was supported by FCT− Fundação para a Ciência e a Tecnologia [project PTDC/QEQ-PRS/0286/2014, research unit iBB UID/BIO/04565/2013 and research contract of A.M. Azevedo IF/00048/2014/CP1214/CT0010), by the European Union [7th framework program agreement Nr.312004] and Programa Operacional Regional de Lisboa 2020 [N.007317]. Conflict of interest statement The authors declare no commercial or financial conflict of interest.

15

 

References [1] Aggarwal, S., What’s fueling the biotech engine — 2008. Nat. Biotechol. 2009, 27, 987– 993. doi:10.1038/nbt1109-987 [2] Aggarwal, S, What’s fueling the biotech engine — 2009-2010. Nat. Biotechol. 2010, 28, 1165–1171. doi:10.1038/nbt1110-1165 [3] Aggarwal, S., What’s fueling the biotech engine — 2010 to 2011. Nat. Biotechol. 2011, 29, 1083–1089. doi:10.1038/nbt.2060 [4] Aggarwal, S., What’s fueling the biotech engine — 2011 to 2012. Nat. Biotechol. 2012, 30, 1191–1197. doi:10.1038/nbt.2437 [5] Aggarwal, S., What’s fueling the biotech engine — 2012 to 2013. Nat. Biotechnol. 2014, 32, 32-39. doi:10.1038/nbt.2794 [6] Scolnik, P. A., mAbs: A business perspective, mAbs 2009, 1, 179–184. doi:10.4161/mabs.1.2.7736 [7] Pollock, J., Ho, S. V., Farid, S.S., Fed-batch and perfusion culture processes: Economic, environmental, and operational feasibility under uncertainty. Biotechnol. Bioeng. 2013, 110, 206–219. doi:10.1002/bit.24608 [8] Zydney, A. L., Perspectives on integrated continuous bioprocessing - opportunities and challenges. Curr. Opin. Chem. Eng. 2015, 10, 8–13. doi:10.1016/j.coche.2015.07.005 [9] Klutz, S., Holtmann, L., Lobedann, M., Schembecker, G., Cost evaluation of antibody production processes in different operation modes, Chem. Eng. Sci. 2016, 141, 63–74. doi:10.1016/j.ces.2015.10.029 [10] Steinebach, F., Müller-Späth, T., Morbidelli, M., Continuous counter-current chromatography for capture and polishing steps in biopharmaceutical production, Biotechnol. J. 2016, 11, 1126-1141. doi:10.1002/biot.201500354 [11] Liu, H. F., Ma, J., Winter, C., Bayer, R., Recovery and purification process development for monoclonal antibody production, mAbs, 2010, 2, 480–499. doi:10.4161/mabs.2.5.12645 [12] Kelley, B., Industrialization of mAb production technology: The bioprocessing industry at a crossroads. mAbs 2009, 1, 443–452. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2759494/ [13] Kelley, B., Blank, G., Lee, A., Downstream processing of monoclonal antibodies: Current practices and future opportunities, in: Gottschalk, U. (Ed.), Process Scale Purification of Antibodies, John Wiley & Sons, Inc., 2009, pp. 1–23. doi:10.1002/9780470444894.ch1 [14] Casnocha, S., Fedechko, R., Mensah, P., Mott, J., Nema, S., Process development considerations for therapeutic monoclonal antibodies in mammalian cell culture, in: Gadamasetti, K., Braish, T. (Eds.), Process Chemistry in the Pharmaceutical Industry, Vol. 2, CRC Press, Boca Raton, FL 2007, pp. 427–453. doi:10.1201/9781420008357.ch26 [15] Farid, S. S., Established bioprocesses for producing antibodies as a basis for future planning. Adv. Biochem. Eng. Biotechnol. 2006, 101, 1–42. doi:10.1007/10_014 [16] Kuystermans, D., Al-Rubeai, M., Bioreactor systems for producing antibody from mammalian cells, in: Al-Rubeai, M. (Ed.), Antibody Expression and Production. Cell Engineering, Vol. 7, Springer, Dordrecht 2011, pp. 25–52. doi:10.1007/978-94-0071257-7 16

 

[17] Seewoester, T., Cell separation and product capture, in: Ozturk, S. S., Hu, W.-S. (Eds.), Cell Culture Technology for Pharmaceutical and Cell-Based Therapies, CRC Press, Boca Raton, FL 2005, pp. 417–438. doi:10.1201/9780849351068.ch12 [18] Farid, S. S., Process economic drivers in industrial monoclonal antibody manufacture, in: Gottschalk, U. (Ed.), Process Scale Purification of Antibodies, John Wiley & Sons, Inc., 2009, pp. 239–261. doi:10.1002/9780470444894.ch12 [19] Han, X., Hewig, A., Vedantham, G., Recovery and purification of antibody, in: Al-Rubeai, M. (Ed.), Antibody Expression and Production. Cell Engineering, Vol. 7, Springer, Dordrecht 2011, pp. 305–340. doi:10.1007/978-94-007-1257-7_14 [20] Kelley, B., Very large scale monoclonal antibody purification: The case for conventional unit operations. Biotechnol. Prog. 2007, 23, 995–1008. doi:10.1021/bp070117s [21] Marichal-Gallardo, P. A., Álvarez, M. M., State-of-the-art in downstream processing of monoclonal antibodies: Process trends in design and validation. Biotechnol. Prog. 2012, 28, 899–916. doi:10.1002/btpr.1567 [22] Curling, J., The development of antibody purification technologies, in: Gottschalk, U. (Ed.), Process Scale Purification of Antibodies, John Wiley & Sons, Inc., 2009, pp. 25–52. doi:10.1002/9780470444894.ch2 [23] Gagnon, P., Technology trends in antibody purification. J. Chromatogr. A 2012, 1221, 57–70. doi:10.1016/j.chroma.2011.10.034 [24] Gagnon, P., Emerging challenges to protein A: Chromatin-directed clarification enables new purification options. BioProcess Int. 2013, 11(9), 44–52. http://www.bioprocessintl.com/upstream-processing/biochemicals-rawmaterials/emerging-challenges-to-protein-a-347353/ [25] Azevedo, A. M., Rosa, P. A. J., Ferreira, I. F., Aires-Barros, M. R., Chromatography-free recovery of biopharmaceuticals through aqueous two-phase processing. Trends Biotechnol. 2009, 27, 240–247. doi:10.1016/j.tibtech.2009.01.004 [26] Soares, R. R. G., Azevedo, A. M., Van Alstine, J. M., Aires-Barros, M. R., Partitioning in aqueous two-phase systems: Analysis of strengths, weaknesses, opportunities and threats, Biotechnol. J. 2015, 10, 1158–1169. doi:10.1002/biot.201400532 [27] Lain, B., Cacciuttolo, M. A., Zarbis-Papastoitsis, G., Development of a high-capacity mAb capture step based on cation-exchange chromatography, BioProcess Int. 2009, 7(5), 26– 34. http://www.bioprocessintl.com/downstreamprocessing/chromatography/development-of-a-high-capacity-mab-capture-step-basedon-cation-exchange-chromatography-183798/ [28] Oehme, F., Peters, J., Mixed-mode chromatography in downstream process development, Biopharm. Int. 2010. http://www.biopharminternational.com/mixedmode-chromatography-downstream-process-development [29] dos Santos, R., Rosa, S. A. S. L., Aires-Barros, M. R., Tover, A., Azevedo, A. M., Phenylboronic acid as a multi-modal ligand for the capture of monoclonal antibodies: Development and optimization of a washing step. J. Chromatogr. A. 2014, 1355, 115– 124. doi:10.1016/j.chroma.2014.06.001 [30] Rosa, S. A. S. L., dos Santos, R., Aires-Barros, M. R., Azevedo, A. M., Phenylboronic acid chromatography provides a rapid, reproducible and easy scalable multimodal process for the capture of monoclonal antibodies. Sep. Purif. Technol. 2016, 160, 43–50. doi:10.1016/j.seppur.2016.01.002 17

 

[31] Barroso, T., Hussain, A., Roque, A. C. A., Aguiar-Ricardo, A., Functional monolithic platforms: Chromatographic tools for antibody purification. Biotechnol. J. 2013, 8, 671– 681. doi:10.1002/biot.201200328 [32] Nascimento, A., Rosa, S. A. S. L., Mateus, M., Azevedo, A. M., Polishing of monoclonal antibodies streams through convective flow devices. Sep. Purif. Technol. 2014, 132, 593–600. doi:10.1016/j.seppur.2014.06.005 [33] Pfaunmiller, E. L., Paulemond, M. L., Dupper, C. M., Hage, D. S., Affinity monolith chromatography: A review of principles and recent analytical applications. Anal. Bioanal. Chem. 2013, 405, 2133–2145. doi:10.1007/s00216-012-6568-4 [34] Natarajan, V., Zydney, A. L., Protein A chromatography at high titers. Biotechnol. Bioeng. 2013, 110, 2445–2451. doi:10.1002/bit.24902 [35] Ghose, S., McNerney, T., Hubbard, B., Protein A affinity chromatography for capture and purification of monoclonal antibodies and Fc-fusion proteins: Practical considerations for process development, in: Shukla, A. A., Etzel, M. R., Gadam, S. (Eds.), Process Scale Bioseparations for the Biopharmaceutical Industry, CRC Press, Boca Raton, FL 2007, pp. 463–489. doi:10.1201/9781420016024.ch16 [36] Shukla, A. A., Yigzaw, Y., Modes of preparative chromatography, in: Shukla, A. A., Etzel, M. R., Gadam, S. (Eds.), Process Scale Bioseparations for the Biopharmaceutical Industry, CRC Press, Boca Raton, FL 2007, pp. 179–225. doi:10.1201/9781420016024.ch6 [37] Brodsky, Y., Zhang, C., Yigzaw, Y., Vedantham, G., Caprylic acid precipitation method for impurity reduction: An alternative to conventional chromatography for monoclonal antibody purification. Biotechnol. Bioeng. 2012, 109, 2589–2598. doi:10.1002/bit.24539 [38] Ecker, D. M., Jones, S. D., Levine, H. L., The therapeutic monoclonal antibody market. mAbs 2015, 7, 9–14. doi:10.4161/19420862.2015.989042 [39] DiMasi, J. A., Grabowski, H. G., The cost of biopharmaceutical R&D: Is biotech different? Manage. Decis. Econ. 2007, 28, 469–479. doi:10.1002/mde.1360 [40] Jagschies, G., Bright future outlook and huge challenges to overcome: An attempt to write the short story of the biopharma industry with current status, selected issues, and potential solutions in discovery, R&D, and manufacturing, in: Knäblein, J. (Ed.), Modern Biopharmaceuticals: Recent Success Stories, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 2013, pp. 487–526. doi:10.1002/9783527669417.ch22 [41] Sommerfeld, S., Strube, J., Challenges in biotechnology production—generic processes and process optimization for monoclonal antibodies. Chem. Eng. Proc. 2005, 44, 1123– 1137. doi:10.1016/j.cep.2005.03.006 [42] Peters, M. S., Timmerhaus, K. D., West, R.E., Plant Design and Economics for Chemical Engineers, 5th ed., McGraw-Hill, 2003. ISBN-13: 978-0072392661 [43] Edwards, M.G., Biotechnology and pharmaceutical commercialization alliances: their structure and implications for university technology transfer offices, in: Krattinger, A., Mahoney, R. T., Nelsen, L., Thomson, J. A., Bennett, A. B., Satyanarayana, K., Graff, G. D., Fernandez, C., Kowalski, S. P. (Eds.), Intellectual Property Management in Health and Agriculture Innovation: A Handbook of Best Practices. MIHR and PIPRA, Oxford, UK, and Davis, CA 2007, pp. 1227–1245. http://www.iphandbook.org/

18

 

Table 1 | Cromatographic data for processes comparison: Process and economic data for protein-A and phenylboronic acid chromatographies (A) and first polishing steps by CEX for and AEX monoliths chromatografies (B). (A)

Protein-A

Phenylboronic Acid

30 g mAb/L-resin 100 cycles

22 g mAb/L-resin 400 cycles

Load linear velocity (LV) LV other steps

200 cm/h

306 cm/h

300 cm/h

918 cm/h (equilibration and wash) 1530 cm/h (elution)

Bed size (D x H)

68 x 15 cm

72 x 15 cm

15 000 USD/L-resin

1 250 USD/L-resin

25 mM tris, pH 7.0 (5CV)**

20 mM HEPES, pH 7.5 (5CV)

DBC* Lifespan

Resin price Equilibration Wash Elution Stripping (B) Mode of operation Lifespan DBC Bed size Resin price LV Pre-equilibration Equilibration Loading Wash Elution Stripping

Same as equilibration 150 mM acetic acid (4CV) 150 mM phosphoric acid (3CV)

100 mM D-sorbitol, 10 mM tris-HCl, pH 7.5 (5CV) 500 mM D-sorbitol, 150 mM NaCl, 10 mM tris-HCl (5CV) —

CEX chromatography

AEX monolithic chromatography

Bind and elute 100 cycles 50 g mAb/L-resin 91.3 x 15 cm 1250 USD2016/L-resin 150 cm/h

Flow through 50 cycles 21 g HCP/L-monolith 4 units of 4 L each 1250 USD2016/L-monolith 852 cm/h

250 mM NaPhosphate, pH 6.0 (3CV)

—

25 mM NaPhosphate, pH 6.0 (3CV) At 90% DBC; 200 cm/h 25 mM NaPhosphate, 50 mM NaCl, pH 6.0 (3CV) 25 mM NaPhosphate, 200 mM NaCl, pH 6.0 (3CV) Same as Pre-EQ buffer

20mM NaPhosphate, pH 8.0 (10CV) Load at 21 g HCP/L-resin — — 20 mM NaPhosphate, 1 M NaCl, pH 7.5 (10CV)

*DBC is the dynamic binding capacity taken as the maximum load a certain resin/chromatographic medium can take under dynamic conditions, i.e., operating flow conditions before a significant unbound protein breakthrough occurs. **CV = column volumes. 5CV, for instance, means that a volume equivalent to five times the column volume is passed through the column on each chromatography cycle.

19

 

Table 2 | Comparison of membrane filtration operations in analysed processes: Post-VI UF/DF and viral filtration (A), pre-HIC UF/DF (B) and technical data for buffer exchange by UF/DF (C). (A)

UF/DF

Absolute Filtration

Traditional process Diafiltration volumes Post-diafiltration volumetric concentration Filtration time Filtrate flux Membrane price (2016)* Membrane area Filtration time Filtrate flux Membrane area Membrane price (2016)**

(B)

UF/DF

4 2.5x (by volume) 4h 55 L/m2/h 600 USD/m2 72 m2

70 m2 4h 40 L/m2/h 10 m2 1500 USD/m2

Traditional process Diafiltration volumes Post-diafiltration volumetric concentration Filtration time Filtrate flux Membrane price (2016)* Membrane area

(C)

UF/DF

New process

2 1.2x

1.3x 4h 55 L/m2/h 600 USD/m2

12.2 m2 Traditional process

Diafiltration volumes Post-diafiltration stages Filtration time Filtrate flux Membrane area

New process

11.4 m2 New process

3 1 at 3x volumetric concentration 4h 55 L/m2/h 2 5.1 m 4.4 m2

* The unit price was reported as 400 USD/m2 in year 2000 and is converted to 2016 prices assuming an average annual inflation rate of 2.4%.

** This price is assumed to be 2.5 times higher than that for an UF/DF membrane.

20

 

Table 3 | Total product costs in 2016 for mAb production using traditional and new processes at full capacity of designed plants. [million USD2016]

Calculation

Traditional

New

SPD simulation

97

51

15% of total product costs

48

27

Direct Production Costs 1

Raw Materials and Consumables

2

Labor

3

Supervisory labor

15% of 2

7.2

4.1

4

Laboratory QA/QC

15% of 2

7.2

4.1

5

Utilities

10% of total product costs

32.0

18.0

6

Waste Treatment

SPD simulation

0.0

0.0

7

Maintenance

5% of purification FCI

1.49

1.49

8

Variable Royalties and Patents

1% of total product costs

3.20

1.80

9

Total Direct Costs

1+...+8

196

107

SPD method

5.63

5.63

0.7% of purification FCI

0.21

0.21

10 + 11

6

6

50% of (2 + 3 + 7)

28

16

9 + 12 + 13

230

130

20% of (2 + 3 + 7)

11

7

20% of total product costs

64

36

5% of total product cost

16

9

15+ 16 + 17

91

52

14+18

321

181

19 x 0.25/0.75

107

60

19 + 20

428

241

21 x 0.4/0.6

286

161

21+ 22

714

402

Fixed Charges 10

Depreciation

11

Insurance

12

Total fixed charges

Plant Overhead Costs Medical services, safety and protection, restaurant, 13 recreation, laboratories, storage 14

TOTAL Manufacturing Costs of Purification only General Expenses

15

Administrative Costs

16

Distribution and Marketing Expenses

17

Running R&D

18

Total General Expenses

19

Total Product Costs for Purification

20

Product Costs for Harvesting

21 Total DSP Product Costs 22 USP Product Costs 23

21

TOTAL PRODUCT COSTS

 

List of Figure Captions Figure 1 | Comparison of downstream processing strategies for recovery, purification and polishing of monoclonal antibodies from mammalian cell culture supernatants − traditional strategy (top) versus the new one suggested in this paper (bottom). The order and type of chromatographic step will be determined in the process development stages and vary from molecule to molecule. Shading: partial=not simulated in this study; shaded=simulated and equal between traditional and new processes; not shaded=simulated in this work and where differences exist between the two processes under analysis. Figure 2 | Comparative economic evaluation of mAb production processes under analysis – with traditional and non-protein-A based purification (new, this study). Discounted cumulative cashflows of traditional (A) and new (B) projects and respective economic indicators (C), at discount rates of 7% (diamonds), 9% (squares) and 11% (circles).

22

 

a protein-A-based process

Cell Culture

Harvesting

Viral Filtration

IEX/CEX Chromatography

Protein A Chromatography

UF/DF

UF/DF

Viral Inactivation

HIC Chromatography

UF/DF for Formulation

alternative, non-protein-A process

Cell Culture

Harvesting

Viral Filtration

AEX Monolithic Chromatography

Phenylboronate Chromatography

UF/DF

UF/DF

Viral Inactivation

HIC Chromatography

UF/DF for Formulation

Figure 1 | Comparison of downstream processing strategies for recovery, purification and polishing of monoclonal antibodies from mammalian cell culture supernatants − traditional strategy (top) versus the new one suggested in this paper (bottom). The order and type of chromatographic step will be determined in the process development stages and vary from molecule to molecule. Shading: partial=not simulated in this study; shaded=simulated and equal between traditional and new processes; not shaded=simulated in this work and where differences exist between the two processes under analysis.

Discounted Cumulative Cashflow (in 106 USD)

40

A

40

30

30

20

20

10

10

0

0

-10

-10 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Year

C Discount rate NPV (billion USD) ROI Payback period (years) IRR

7% 42.3 4.9 2.73

Traditional 9% 11% 36.4 31.5 4.1 3.4 2.77 68%

2.82

B

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Year

7% 42.9 5.0

New 9% 36.9 4.2

11% 32.0 3.5

2.71

2.75

2.80

69%

Figure 2 | Comparative economic evaluation of mAb production processes under analysis – with traditional and non-proteinA based purification (new, this study). Discounted cumulative cashflows of traditional (A) and new (B) projects and respective economic indicators (C), at discount rates of 7% (diamonds), 9% (squares) and 11% (circles).

23