You are on page 1of 14

Research Article

Received: 9 November 2020 Revised: 24 January 2021 Accepted article published: 31 January 2021 Published online in Wiley Online Library:

(wileyonlinelibrary.com) DOI 10.1002/jctb.6686

Contribution of protein A step towards cost


of goods for continuous production of
monoclonal antibody therapeutics
Vikrant Bansode, Paridhi Gupta, Nikhil Kateja and Anurag S Rathore*

Abstract
BACKGROUND: Monoclonal antibodies (mAbs) presently dominate the biotherapeutic market. However, biotherapeutics in
general, and mAbs in particular, are expensive therapies and outside the realm of affordability for most of the world. In an
attempt to lower their cost of manufacturing, a novel in-house non-protein A continuous platform has recently been reported
for mAb purification. In this paper, we perform an in-depth evaluation of the process economics to assess the impact of remov-
ing the protein A step on the cost of goods (COGs).
RESULTS: The non-protein A platform was found to result in significant savings in COGs at about 23% when comparing both
platforms in batch mode. Conversion of the non-protein A platform from batch to continuous mode led to a further increase
in these savings with about 39% gains in COGs over the conventional protein A platform.
CONCLUSIONS: The non-protein A purification platform offers a viable alternative to the standard protein A platform and can
be considered when setting up a new facility. The continuous non-protein A platform can especially be considered given the
flexibility of the setup and the utility at both small and large scales.
© 2021 Society of Chemical Industry

Supporting information may be found in the online version of this article.

Keywords: continuous processing; economic analysis; cost evaluation; biotherapeutics; monoclonal antibodies; non-protein a
purification

INTRODUCTION A ligands which specifically bind to the Fc domain of mAbs, thus selec-
Monoclonal antibodies (mAbs) have been dominating and will con- tively purifying the molecules from a pool of host cell proteins (HCP)/
tinue to dominate the biosimilar market. They have been success- host cell DNA (HCD).14,15 Protein A capture is generally preceded by
fully used for the treatment of cancers and autoimmune diseases. several clarification steps, followed by viral inactivation and additional
There are a growing number of mAbs being developed for infectious polishing chromatography and formulation steps.16-18 Economic anal-
and neglected diseases, including the newly discovered SARS-CoV-2 ysis for the conventional protein A-based purification platform has
virus as well as other viral diseases like Zika and Ebola.1–6 Although been performed previously by our group.19 Although the platform
there is a steady increase in the demand of mAb therapeutics, the has been time tested, the cost of protein A resin is almost two times
manufacturing cost remains high.7–9 The raw material cost for bio- the cost of other resins. This makes protein A chromatography the
pharmaceutical production is 20–100 times more than that for tradi- most expensive unit operation in the purification platform.20–22 Even
tional pharmaceutical drugs. Biopharmaceuticals are also required in though robust protein A resins have been developed for higher toler-
higher doses, which further results in higher costs of treatment.6 ance of cleaning-in-place agents like NaOH, protein A ligand leaching
Countries with some form of universal healthcare are better off as remains a problem for manufacturers.23 This also reduces the number
the costs of biopharmaceuticals are either completely or partially of times the resin can be reused and incurs an additional cost of anal-
shared by the state. Patients in countries with no such facility incur ysis for protein A leachate.24
the total cost of biopharmaceutical drugs.10 Outsourcing drug devel- Several attempts have been made to replace the conventional
opment is a strategy used by companies to reduce the development purification platform by proposing alternatives to the protein A
time but this results in a higher development cost.11 The mAb mar- capture step. Researchers have developed a platform with
ket is predicted to be around $319 billion by 2026. Even the Indian
biosimilar market is estimated to be at $6 billion by 2024.1 Costs of
* Correspondence to: AS Rathore, Department of Chemical Engineering,
cell culture media and chromatography media are two of the major
contributors to mAb production.10–13 Indian Institute of Technology, Hauz Khas, New Delhi, 110016, India.
E-mail: asrathore@biotechcmz.com
The purification regime used for mAbs is more or less same across
the industry and can be termed as the protein A platform. Protein A Department of Chemical Engineering, Indian Institute of Technology Delhi,
is a type of affinity chromatography which uses immobilized protein New Delhi, India
1

J Chem Technol Biotechnol 2021 www.soci.org © 2021 Society of Chemical Industry


www.soci.org V Bansode et al.

phenylboronic acid multimodal chromatography as the capture TCI =total plant direct cost ðTPDCÞ
step followed by ion exchange chromatography and hydrophobic
+total plant indirect cost ðTPICÞ
interaction chromatography as the polishing steps.25 The invest- ð1Þ
ment was similar to that of the conventional platform while a +contractor0 s fee and contingency + other cost
20% decrease in the annual operating expenditure (OpEX) was
estimated. Other researchers have evaluated several sets of chro-
matography steps as replacement of protein A chromatography TPDC, TPIC and OpEX have been defined similar to our previous
in an attempt to reduce the purification cost of mAbs.26,27 work.19
Another study explored the use of activated carbon in flow- All the components of CapEX were obtained by multiplying a
through mode to clear impurities and replace protein A capture.28 suitable multiplier to the equipment purchase cost, as given in
While the results showed product quality comparable to that of Table S1 (supporting information). However, when prototype
the standard protein A platform, the researchers did not evaluate devices were used in PFDs, estimated assumptions and extrapola-
process economics. tions were applied, as provided in Table S2 (supporting informa-
A non-protein A continuous processing platform for mAb purifica- tion). Working capital is estimated to cover expenses for 30 days
tion was reported recently.29 The platform employs a novel coiled of labor, 30 days of raw materials, 30 days of utilities and 30 days
flow inversion reactor (CFIR) to aid in clarification after precipitation of waste treatment.
for removal of HCP/HCD and for inactivation of enveloped viruses. For calculating various components of OpEX, suitable
Cation exchange chromatography is used to capture the product assumptions were made. Cost of raw materials was taken from
as well as control charge variants. Unlike protein A chromatography, industry-specific vendor sources. Labor and facility costs were
cation exchange utilizes the difference in charge of the product and calculated in SuperPro Designer software using Eqns (2) and
impurities at a particular pH and conductivity values for separation. (3), as given below. Consumable costs include annual con-
In this paper, we examine the impact of using this platform on the sumption of chromatography resins and filters used in depth
economics of commercial production of mAbs. filtration, clarification, ultrafiltration and diafiltration. Costs of
consumables were also taken from vendor sources (Table S3,
supporting information). Wastewater treatment/disposal cost
EXPERIMENTAL
was assumed as $0.05 kg−1. Utilities include cost of air at
Economic analysis
0.5 VVM, cost of power and cost of steam or chilled water or
The approach for economic assessment was same for both protein- glycol, depending on the use.
A-based and non-protein-A-purification strategies. SuperPro
Designer (Intelligen Inc., Scotch Plains, NJ) was used to make the pro- !
1+ benefits+ supervision
cess flow diagrams (PFDs) of batch as well continuous manufactur- Labour cost = ðbasic rateÞ×
ing mode. Figures 1 and 2 describe the PFDs for protein A and +operating supplies +administration
non-protein A purification strategies as derived from SuperPro ð2Þ
×ðlabour hoursÞ
Designer software. Figure 3 provides a breakdown of the overall
expenditure. After incorporating all the process assumptions, eco-
nomic analysis assumptions were also added to the PFDs. The eco- Facility cost =maintenance+depreciation +insurance
nomic evaluation reports hence generated gave total capital ð3Þ
investment (CapEX) and OpEX for the end-to-end PFDs, along with +local tax +factory expenses
the unit-operation-wise distribution (discussed in the following sub-
section). The obtained data were thereafter used for calculating eco- Various economic indicators, namely COGs, NPV and payback
nomic indicators such as cost of goods sold (COGs), net present time, were calculated using the SuperPro Designer software.
value (NPV) and payback time. Subsequently, sensitivity analysis with COGs ($ g−1) allows us to understand the cost incurred in
scale of manufacturing was also performed at two scales – small manufacturing per gram of the final product produced. COGs
scale and large scale of production. Small scale represents clinical incorporate overall CapEX incurred for an assumed 15 years of
material generation and large scale represents a typical manufactur- production and annual OpEX, along with annual throughput of
ing scale for a ‘blockbuster’ drug. Throughputs of 15 kg per annum final product. The equation for calculation of COGs is given below:

  Total capital investment=years of operation ð$ yr−1 Þ +annual operating cost ð$ yr−1 Þ


COGs $ g−1 = … ð4Þ
Annual production rate ðg yr−1 Þ

and 500 kg per annum were considered for small and large scale, In order to estimate the profitability of investment in protein A
respectively.19 Process parameters were optimized in house at the versus non-protein A purification strategies, NPV analysis was per-
laboratory scale to be used for this study. formed based on 15 years of plant operation in both cases. In the
NPV analysis, 3 years of manufacturing time for plant was
Economic parameters and assumptions assumed before the operation period. Discount rate of 7% per
CapEX is the same as total capital investment (TCI), as derived from annum and income tax at 10% per annum were used for the
the SuperPro Designer software. TCI is derived as per Eqn (1) below: NPV assessment. In order to calculate annual revenue generated,
2

wileyonlinelibrary.com/jctb © 2021 Society of Chemical Industry J Chem Technol Biotechnol 2021


Contribution of protein A step towards cost of goods www.soci.org

Figure 1. PFDs for protein A platform in (a) batch mode and (b) continuous mode.

the selling price of mAbs was taken at $20 mg−1 for both protein and ultrafiltration–diafiltration. The protein A platform used in this
A and non-protein A strategies of manufacturing. study in batch operation consisted of a fed-batch cell culture pro-
The payback time in number of years was calculated using cess in upstream using CHO (Chinese hamster ovary) cell line. The
Eqn (5) as follows: purification regime started with harvest clarification using centri-
fuge and depth filters followed by a protein A capture step and
Net profit ð$Þ viral inactivation. Two polishing steps of anion and cation
Payback time = … ð5Þ
Revenue ð$ yr−1 Þ−annual operating cost ð$ yr−1 Þ exchange chromatography were then followed by ultrafiltration
and buffer exchange producing the final drug substance. The
Revenue and net profit (for any year) were calculated using PFD of the batch protein A platform is shown in Fig. 1. The batch
Eqns (6)–(8): setup was modified for continuous operation with standard chro-
    matography system being replaced by a Cadence™ BioSMB PD
Revenue $ yr−1 =Amount of product g yr−1 system (Pall Corporation, USA). The protein A capture step was
  ð6Þ operated in periodic countercurrent mode.30,31 The mixing tank
×average price realized $ g−1 …
for viral inactivation was replaced by a CFIR and ultrafiltration–
Net profit ð$Þ =Gross profit ð$Þ−income tax ð$Þ… ð7Þ diafiltration was replaced by Cadence™ inline concentrator (ILC)
Gross profit ð$Þ= Revenue ð$Þ−annual operating cost ð$Þ… ð8Þ and inline diafiltration (ILD) modules (Pall Corporation, USA). The
upstream process was modified to use a perfusion cell culture
process over the fed-batch process. Details of the protein A batch
Protein A platform and continuous processes are given in Table S4 (supporting
For the comparison of the two platforms, an IgG1 antibody was information).
considered as the reference mAb. The mAb had a molecular
weight of 150 kDa and a pI value of 8.7. A standard protein A Non-protein A platform
mAb purification platform can be broken down into: cell culture, The non-protein A platform consisted of a upstream fed-batch
clarification, protein A capture, viral inactivation, polishing steps reactor followed by centrifugation as the first stage of clarification.
3

J Chem Technol Biotechnol 2021 © 2021 Society of Chemical Industry wileyonlinelibrary.com/jctb


www.soci.org V Bansode et al.

Figure 2. PFDs for non-protein A platform in (a) batch mode and (b) continuous mode.

The harvest then underwent low-pH treatment for the precipita- that have examined the economic impact of continuous produc-
tion of HCP/HCD while inactivating enveloped viruses and was tion versus batch production.32,33 In this study we perform an eco-
depth-filtered followed by neutralization. Cation exchange chro- nomic assessment of a non-protein A platform and compare it to
matography was the capture step which also served as the pri- the traditional protein A-based continuous process platform. A
mary polishing step by controlling the charge variant profile of titer of 5 g L−1 and an annual scale of production of 500 kg yr−1
the mAb. This was followed by multimodal chromatography have been assumed.32 Optimized process parameters were used
(AEX/HIC) for further polishing. Ultrafiltration and diafiltration for both platforms. As the upstream process for both the plat-
were used to afford the final drug substance. The PFD for the forms is the same, comparison between the downstream pro-
non-protein A platform is shown in Fig. 2. The continuous non- cesses utilized in both the platforms was the prime focus of the
protein A platform utilized a perfusion reactor to provide for con- economic analysis. However, while talking about COGs and also
tinuous feed for the downstream train. Purification started with a while comparing NPV in a later section, we have included the
continuous centrifuge followed by a CFIR which was used for low- upstream spending to provide a holistic comparison of both the
pH treatment of the harvest. Depth filtration further clarified the platforms.
harvest and a Cadence™ BioSMB PD system (Pall Corporation,
USA) was employed for a countercurrent setup of CEX and MMC Protein A vs. non-protein A platform: a batch-to-batch
chromatography. The final drug substance was made using comparison
Cadence™ ILC and ILD modules (Pall Corporation, USA) to make A novel non-protein A platform for the purification of mAbs was
the process continuous. The non-protein A continuous and batch recently developed in our laboratory and in this study it is com-
process details are summarized in Table S5 (supporting pared to the conventional protein A-based platform. The non-
information). protein A platform aims to replace the protein A capture step with
precipitation and cation exchange chromatography.
From the economic analysis data it was observed that an overall
RESULTS AND DISCUSSION reduction in COGs of 23% was obtained on transformation of the
Protein A-based process platforms for mAb purification have wide batch purification process from protein A to non-protein
acceptance in the industry and there have been several studies A. Chromatography accounted for about 80% of this reduction
4

wileyonlinelibrary.com/jctb © 2021 Society of Chemical Industry J Chem Technol Biotechnol 2021


Contribution of protein A step towards cost of goods www.soci.org

Figure 3. Breakdown of capital and operating expenditure.

which is reflected in CapEX and OpEX comparison. There was a modes of interaction and thus facilitates process intensification.
34% reduction in CapEX and 45% reduction in OpEX linked with Figure 4 illustrates a comparison of the unit-wise CapEX, OpEX
the chromatography step. The reduction in CapEX can be attrib- and COGs for protein A and non-protein A batch platforms.
uted to the reduction in the number of chromatography steps Similar to CapEX, the elimination of the protein A capture step
and hence the equipment required which is responsible for 90% accounted for >77% of the total reduction in OpEX. This reduction
of the total reduction in CapEX. Also, as the viral inactivation step in OpEX is equally shared by the capture and polishing chroma-
is done in tandem with HCP/HCD precipitation during the clarifi- tography steps at about 40% and 38%, respectively. The major
cation stage, the number of tanks required in the setup is reduced cost savings expected from any non-protein A platform come
and this further contributes to the cost saving. Further breaking from elimination of protein A chromatography. The expensive
up the chromatography steps, it can be seen that the polishing nature of protein A resin and the need for a substitute is quite evi-
chromatography contributes to 73% reduction in CapEX, dent from the high share of reduction percentage of capture chro-
highlighting the advantage of eliminating a chromatography matography in OpEX at a large scale. The percentage reductions
step. Here it should be noted that although cation exchange chro- from protein A to non-protein A batch platforms are shown in
matography is the capture step in the non-protein A platform, it Fig. 5. The facility-dependent cost is a major share in OpEX being
also simultaneously acts as the polishing step by controlling the 64% and 77% for protein A and non-protein A platforms, respec-
charge variants and reducing low-molecular-weight impurities. tively. Elimination of the protein A chromatography step
Further, although the multimodal chromatography resin that fol- accounted for 74% of the facility cost reduction. Even though
lows the capture step in the non-protein A platform is more the non-protein A platform introduces an ultrafiltration step
expensive than the traditional ion exchange resins, it offers dual before the capture chromatography, the elimination of the
5

J Chem Technol Biotechnol 2021 © 2021 Society of Chemical Industry wileyonlinelibrary.com/jctb


www.soci.org V Bansode et al.

Figure 4. Unit-wise comparison of protein A and non-protein A batch platforms: (a) CapEX and OpEX comparison; (b) COGs comparison.

protein A chromatography step greatly helps reduce the facility of clarification for low-pH precipitation of HCP/HCD while simulta-
footprint by reducing the number of chromatography systems neously inactivating enveloped viruses facilitates this cost reduc-
required from three to two while also reducing the number of tion. The subsequent depth filtration step resulted in improved
required columns and saves on OpEX in the long run. About clarification and as a result there was a 65.7% reduction in total
70% of the total reduction in consumable cost in the non-protein OpEX. A similar trend was observed when comparing OpEX with
A setup is due to the replacement of protein A chromatography as 79% of the cost reduction attributable to chromatography split
the capture step. into 36% due to capture and 43% due to polishing steps, as seen
in Fig. 7. The facility-dependent cost being the major contributor
How beneficial is the non-protein A platform in to OpEX (about 55% of OpEX) reflected this behavior with 37%
continuous mode? and 45% of the facility cost reduction attributed to capture and
Major economic gains were realized on transition from the con- polishing steps, respectively. Consumable cost was the next major
ventional protein A purification platform to the non-protein A share in OpEX at about 43%. There was a 23% overall reduction in
platform in the batch mode. Further, we aimed to examine if these the consumable cost with the clarification and viral inactivation
gains are further amplified by converting the mode of operation steps accounting for the major share of 58% in cost reduction fol-
from batch to continuous, as driven by the current and enormous lowed by chromatography at 42%. The savings here were primar-
interest in continuous manufacturing.30,34 ily due to better control strategies deployed in the continuous
Figure 6 summarizes the overall benefits of continuous non- setup allowing the maximum use of membrane area and thus
protein A platform over batch mode. We see a 34% reduction in reducing the number of filters used for clarification. The amount
COGs when the non-protein A platform is operated in a continu- of resin utilized also decreased due to the efficient use of resin
ous mode, of which 63% of the reduction is in the chromatogra- binding capacity in the continuous setup.
phy step. When comparing CapEX from the non-protein A batch
to continuous setup the share of chromatography in cost reduc- Is continuous non-protein A platform better than its
tion at large scale is about 82% which can be split into about protein A counterpart?
37% coming from the capture step and 45% coming from the pol- We have seen the benefits of the non-protein A platform over the
ishing chromatography. Use of the novel CFIR in the second phase protein A platform in batch mode and also the advantages of
6

wileyonlinelibrary.com/jctb © 2021 Society of Chemical Industry J Chem Technol Biotechnol 2021


Contribution of protein A step towards cost of goods www.soci.org

Figure 5. Percentage cost reduction from protein A batch to non-protein A batch platform: (a) CapEX and OpEX; (b) facility cost; (c) consumable cost.

operating the non-protein A platform in continuous mode. How- spending for both continuous platforms. There is a 64% reduction
ever this comparison would be incomplete without talking about in consumables due to the elimination of protein A resin from the
the continuous protein A platform. process. Due to the simultaneous precipitation of HCP/HCD and
As the effects of converting a batch process to a continuous pro- viral inactivation, the number of depth filters and 0.22 μm filters
cess would be similar for both platforms, we expected the gains to required is also reduced, accounting for about 35% of consum-
be less than those in batch-to-batch comparison. However, there able cost reduction.
was a 25% reduction in COGs going from continuous protein A
platform to continuous non-protein A platform. Over 75% of the
reduction in COGs was seen to be due to chromatography steps Taking the leap from protein A batch platform to non-
as seen in the unit-wise COGs comparison in Fig. 8. The share of protein A continuous platform
chromatography in CapEX reduction is about 85% and that in There are major benefits in adopting a non-protein A setup for
OpEX reduction is about 74%. The major advantage of running a manufacturing of mAbs in a batch mode. And even though many
continuous non-protein A platform is that the flow rates at which studies have compared the traditional protein A batch setup to a
the process operates are low enough to be accommodated by a protein A continuous setup, the majority of the biopharmaceuti-
BioSMB pilot-scale system, whereas the continuous protein A plat- cal industry still uses protein A platforms.35,36 The investments
form requires a BioSMB process-scale system which adds signifi- required for converting any of the platforms to continuous mode
cantly to CapEX. The same is reflected in OpEX and seen when are the same. Both the platforms require the CFIR, BioSMB and
further looking at the facility-dependent cost where about 86% ILC/ILD modules. We will therefore focus on the benefits of mov-
of the reduction is due to the smaller BioSMB system. Figure 9 ing from the traditional protein A batch setup to a non-protein
gives a comprehensive comparison of the facility and consumable A continuous setup. The comparison follows the same pattern
7

J Chem Technol Biotechnol 2021 © 2021 Society of Chemical Industry wileyonlinelibrary.com/jctb


www.soci.org V Bansode et al.

Figure 6. Unit-wise comparison of non-protein A batch and continuous platforms: (a) CapEX and OpEX comparison; (b) COGs comparison.

as before wherein all data are calculated based on the purification investment during production. A way to compare the investment
regime while the COGs comparison includes the upstream cost. in capture chromatography media is to look at the cost of resin
We see the highest reduction of 39% in COGs comparing the per unit of product produced. As protein A resin cost is the highest
protein A batch platform to the non-protein A continuous plat- consumable cost in the purification regime, we have normalized
form. About 70% of this reduction is due to the chromatography the cost of capture resin in each platform to the protein A resin
step. Figures 8 and 9 give a stepwise comparison moving from cost in the standard batch platform. Figure 10 shows a plot of cap-
protein A batch process to protein A continuous process to non- ture resin cost per kilogram of product normalized to the protein
protein A continuous process. The share of chromatography in A resin cost for the batch platform. The normalized resin cost per
CapEX reduction is about 83% and that in OpEX reduction is kilogram of product for protein A batch and continuous platforms
79%. Even though the BioSMB PD system costs more than a stan- is very close after about 80 kg of product is produced. In the case
dard chromatography system, the protein A batch setup requires of the non-protein A platform in batch and continuous mode, the
three systems to generate 500 kg yr−1 while the continuous non- normalized resin cost per kilogram of product overlaps after just
protein A setup can run on a single BioSMB PD system, leading to 5 kg of production.
major savings in terms of CapEX and facility-dependent cost. The Although any investment will eventually pay for itself over time,
benefits of eliminating a chromatography step along with the there is a very large difference in the initial investment required in
apparent advantage of a continuous process make the continu- the capture resin for both the platforms. The protein A continuous
ous non-protein A platform a lucrative option. process required about 72% of the resin required for the protein A
The motive behind developing a non-protein A platform was to batch setup. But the costs of capture resin for the non-protein A
reduce the costs incurred of purchasing the protein A resin. The batch and continuous setups are about 7% and 5% of the total
high cost of protein A resin translates into a heavy initial cost of protein A resin required in the batch mode. This stark
8

wileyonlinelibrary.com/jctb © 2021 Society of Chemical Industry J Chem Technol Biotechnol 2021


Contribution of protein A step towards cost of goods www.soci.org

Figure 7. Percentage cost reduction from non-protein A batch to continuous platform: (a) CapEX and OpEX; (b) facility cost; (c) consumable cost.

difference in the initial investment is what drives research towards or continuous manufacturing facilities, an investment in the con-
non-protein A-based mAb purification platforms. tinuous facility will earn $44 million extra as opposed to its batch
counterpart, over 15 years of operation.
NPV analysis A shift from protein A continuous manufacturing platform to
non-protein A continuous manufacturing platform shows a fur-
NPV is an important tool for understanding the economic profitabil-
ther advantage (ΔNPV) of $25 million. To put this gain into per-
ity of any project. It provides complete understanding of economic
spective, with this advantage three new non-protein A
aspect as it takes capital investment as well as annual operational
costs into consideration. NPV considers time value of money as it is continuous facilities can be built, each with a throughput of
the sum of present values of all the future cash flows in the estimated 500 kg yr−1, as the CapEX requirement of an end-to-end non-
time period of plant operations. Payback period is the number of protein A continuous facility is $7.4 million (upstream and down-
years required to reach the breakeven point for the initial invest- stream combined). The ΔNPV values moving from one platform
ment. NPV analysis is a better metric than payback time because it to another are summarized in Table 1.
considers annual variation in expenses and revenue over the years. Overall, the continuous manufacturing mode of the non-protein
Selling price of mAb for both the purification strategies has been A purification strategy is the most cost-effective option against
fixed at $20 mg−1. For calculating NPV, it was assumed that produc- protein A batch manufacturing, as it gives an advantage (ΔNPV)
tion began after 3 years of capital expenditure in setting up the of $81 million, enough to build ten new continuous non-protein
plant. A discount rate of 7% per annum and income tax of 10% per A-based mAb manufacturing facilities producing 500 kg yr−1.
annum were used for the NPV assessment. After setup, 15 years of The payback time for all platforms being considered is similar at
plant operation has been assumed. CapEX and OpEX values were just over 3 years. However, the reason for this is the extremely
taken from previous sections. high-value product being considered.
A shift from protein-A batch manufacturing platform to non-
protein-A batch manufacturing platform shows an advantage Effect of scales on non-protein A platform: batch to
(ΔNPV) of $36 million. NPV is a holistic economic evaluation of continuous
manufacturing platforms and a gain in NPV over 15 years of oper- For the purpose of this study two scales were assumed at a titer of
ation suggests that a non-protein A-based batch facility will give 5 g L−1. The small scale was assumed to be at 15 kg yr−1 and large
higher returns to the tune of $36 million as opposed to a protein scale was assumed at 500 kg yr−1.32 Here we compare the effect
A-based batch facility over the same time period. Hence, NPV of scale on the non-protein A platform going from batch to con-
gains aid decision-making for selection and development of bio- tinuous mode of operation (Fig. 11).
processing technology. At the small scale we see a 32% reduction in CapEX, while at the
In going from batch to continuous manufacturing mode for the large scale it was 75% reduction when comparing both modes.
non-protein A purification strategy, NPV gains (ΔNPV) of $44 mil- There was about 78% reduction in CapEX for the clarification
lion were observed. Therefore, given a choice of investing in batch and viral inactivation step at small scale which can be attributed
9

J Chem Technol Biotechnol 2021 © 2021 Society of Chemical Industry wileyonlinelibrary.com/jctb


www.soci.org V Bansode et al.

Figure 8. Unit-wise comparison of protein A batch, protein A continuous and non-protein A continuous platforms: (a) CapEX and OpEX comparison;
(b) COGs comparison.

to process intensification which resulted in fewer tanks and hence development. Uncertainties like upstream titer, chromatographic
a reduced facility footprint. At large scale the chromatography yield and impurity removal have been studied by other
step showed the highest reduction with 80% and 78% in capture researchers by developing an optimization algorithm to find the
and polishing steps, respectively. This trend is somewhat reflected best column sequencing and sizing strategy.38 Other researchers
when comparing Opex at both scales. There was a 33% reduction have studied the effect of nine parameters at various levels on
in OpEX at small scale and about 65% reduction at large scale. The specific growth rate in perfusion mode of cell culture for the pro-
clarification and viral inactivation step is the major contributor to duction mAbs to help in process design.39
savings at 76% at small scale whereas the chromatography steps As the aim of the present work was to emphasize the role
save about 76% each at large scale. that protein A chromatography plays in the total cost of mAb
It is common to see increased savings at clinical scales when production, a comparative sensitivity analysis of the capture
comparing batch and continuous manufacturing.32,37 In the case steps was performed for both platforms. Depending on the
of the non-protein A process platform, however, the reason for type of product, the interaction with the resin may change
the discrepancy is the chromatography equipment used at both the yield of the step and/or dynamic binding capacity (DBC)
scales. The cost of multicolumn chromatography equipment used of the resin.
at small scale increased the overall cost at that scale. However, the The effect of capture chromatography yield was evaluated on
same equipment can also be used for large-scale production, the overall COGs. For the protein A platform, a yield variation of
unlike in batch manufacturing where larger sized equipment is 70% to 90% was considered where the optimized process yield
required for larger scale and costs considerably more than the was 90%, whereas for the non-protein A platform a range of
corresponding equipment used at small scale. 75% to 95% was considered where the optimized process yield
was 95%. The analysis was performed by keeping the facility
Sensitivity of capture step parameters size and other unit operation yields constant. As the yield of
Analyzing the effect of multiple parametric uncertainties on the the capture step decreases, the overall production rate
10

cost of production is of interest to those involved in process decreases resulting in an increase in the total COGs as shown

wileyonlinelibrary.com/jctb © 2021 Society of Chemical Industry J Chem Technol Biotechnol 2021


Contribution of protein A step towards cost of goods www.soci.org

Figure 9. Unit-wise facility and consumable cost comparison of protein A batch, protein A continuous and non-protein A continuous platforms.

Figure 10. Plot of normalized capture resin cost per kilogram of product.

Resin aging is directly dependent on the DBC as the number of


Table 1. ΔNPV analysis across platforms over 15 years of operation
cycles required to process a certain amount of product increases
Non-protein A Non-protein A or decreases with the DBC in turn affecting the total amount of
batch continuous resin used. As protein A amounts to more than 70% of the total
consumable cost for the protein A platform, the effect of DBC on
Protein A batch $36 million $81 million total consumable cost was evaluated and compared with the
Protein A NA $25 million non-protein A platform. Analysis was done for both platforms by
continuous keeping the facility size constant and only changing the DBC of
Non-protein A NA $44 million the capture step. As both the capture steps operate by exploiting
batch different modes of separation, the ranges of both platforms vary
considerably. For the protein A platform in batch mode a DBC
range of 40 to 50 g L−1 was considered. For the protein A platform
in Fig. 12. The percentage increase in COGs was the same in in continuous mode, the capture step operated in periodic counter-
each range for both platforms as the overall yield also changed current mode such that the complete binding capacity can be uti-
in the same ratio. However in both batch and continuous lized by overloading the column. The unbound product was then
modes, the highest COGs achieved in the non-protein A plat- captured by the second column in series. Hence a DBC range of
form were $37 g−1 and $24 g−1 at a yield of 75% which were 64 to 80 g L−1 was considered for this mode. A DBC range of
still less than the lowest COGs in the protein A platform which 32 to 40 g L−1 was considered for the non-protein A platform in
were $41 g−1 and $27 g−1 at 90% yield.
11

both batch and continuous modes. As cation exchange

J Chem Technol Biotechnol 2021 © 2021 Society of Chemical Industry wileyonlinelibrary.com/jctb


www.soci.org V Bansode et al.

Figure 11. Unit-wise effect of scale on non-protein A batch and continuous operation: (a) CapEX; (b) OpEX; (c) COGs.

Figure 12. Comparison of overall COGs for both platforms in batch and continuous mode at different capture step yields.

chromatography is the capture as well as polishing step in the non- change in the consumable cost as seen in Fig. 13. In batch mode
protein A platform, it cannot be operated in periodic countercur- at a DBC of 40 g L−1, the protein A resin requirement resulted in
rent mode as it would result in loss of charge variant resolution, a 28% increase in the total consumable cost. There was no change
and charge variant profile is a critical quality attribute for mAbs. in protein A resin requirement at DBC values of 45 and 50 g L−1.
Hence CEX in continuous mode was loaded the same as in As the continuous mode operates by always utilizing the maxi-
batch mode. mum binding capacity of the resin, the resin utilized always
The decrease in DBC resulted in an increase in OpEX for all plat- increases with a decrease in DBC. The consumable cost for the
forms. However, the increase was not significant, due to which the protein A platform in continuous mode increased by 7% and
COGs did not increase significantly. There was however a notable 16% at DBC of 72 and 64 g L−1, respectively. For the non-protein
12

wileyonlinelibrary.com/jctb © 2021 Society of Chemical Industry J Chem Technol Biotechnol 2021


Contribution of protein A step towards cost of goods www.soci.org

Figure 13. Comparison of total consumable cost for both platforms in batch and continuous mode at different capture step DBC values.

A platform in batch mode, there was no change in the resin REFERENCES


requirement within the DBC range considered. In continuous 1 Expanding access to monoclonal products A global call to action. IAVI/-
mode, the non-protein A consumable cost increased by 1% and Wellcome (2020).
2% at DBC values of 36 and 32 g L−1, respectively. 2 Wec AZ et al., Broad neutralization of SARS-related viruses by human
Protein A resin amounts to more than 70% of the total consum- monoclonal antibodies. Science 369:731–736 (2020).
3 Marston HD, Paules CI and Fauci AS, Monoclonal antibodies for emerg-
able cost for the protein A platform. It has a lower lifetime com- ing infectious diseases: borrowing from history. N Engl J Med 378:
pared to CEX resins and hence has to be replaced more often.10 1469–1472 (2018).
This is reflected in the overall consumable cost change as the resin 4 Chen X et al., Human monoclonal antibodies block the binding of
requirement changes. SARS-CoV-2 spike protein to angiotensin converting enzyme
2 receptor. Cell Mol Immunol 17:1647–3649 (2020).
5 Singh S, Kumar NK, Dwiwedi P, Charan J, Kaur R, Sidhu P et al., Mono-
clonal antibodies: a review. Curr Clin Pharmacol 13:85–99 (2018).
CONCLUSIONS 6 Werner RG, Economic aspects of commercial manufacture of biophar-
The presented non-protein A continuous processing platform maceuticals. J Biotechnol 113:171–182 (2004).
provides a lucrative alternative to the standard protein A continu- 7 Drucker A et al., The cost burden of trastuzumab and bevacizumab
therapy for solid tumours in Canada. Curr Oncol 15:136 (2008).
ous processing platforms. Although transforming an existing pro-
8 Nandi S, Kwong AT, Holtz BR, Erwin RL, Marcel S and McDonald KA,
tein A batch facility to a non-protein A facility could result in extra Techno-economic analysis of a transient plant-based platform for
investment, one could consider the novel platform when building monoclonal antibody production. MAbs 8:1456–1466 (2016).
a new facility. One could also consider investing in a new contin- 9 Rattanapisit K, Chao Z, Siriwattananon K, Huang Z and
uous non-protein A facility as one of the key advantages of the Phoolcharoen W, Plant-produced anti-enterovirus 71 (EV71) mono-
clonal antibody efficiently protects mice against EV71 infection. Plan
platform is that the chromatography equipment at small scale Theory 8:560 (2019).
can be employed for large-scale production thus eliminating the 10 Rathore AS, Pathak M, Ma GJ and Bracewell DG, Re-use of protein a
need for investing in two setups. This is also evident in the NPV resin: fouling and economics. BioPharm Int 28:28–33 (2015).
comparison where the continuous non-protein A platform 11 DiMasi JA, Smith Z and Getz KA, Assessing the financial benefits of fas-
ter development times: the case of single-source versus multi-
showed the highest NPV gains. The non-protein A platform is flex-
vendor outsourced biopharmaceutical manufacturing. Clin Ther
ible and can be extended to a wide range of products when 2018 40:963–972 (2018).
coupled with single-use equipment. 12 Kelley B, Kiss R and Laird M, A different perspective: how much innova-
tion is really needed for monoclonal antibody production using
mammalian cell technology? Adv Biochem Eng Biotechnol 165:
ACKNOWLEDGEMENTS 443–462 (2018).
13 Klutz S, Holtmann L, Lobedann M and Schembecker G, Cost evaluation
The authors acknowledge funding support from the Department of antibody production processes in different operation modes.
of Biotechnology, Ministry of Science and Technology Chem Eng Sci 141:63–74 (2016).
(no. BT/COE/34/SP15097/2015). 14 Choe W, Durgannavar TA and Chung SJ, Fc-binding ligands of immu-
noglobulin G: an overview of high affinity proteins and peptides.
Materials 9:994 (2016).
15 Nosrati M, Solbak S, Nordesjö O, Nissbeck M, Dourado DFAR,
CONFLICT OF INTEREST Andersson KG et al., Insights from engineering the Affibody-fc inter-
The authors declare that they do not have any conflict of interest. action with a computational-experimental method. Protein Eng Des
Sel 30:593–601 (2017).
16 Kelley B, Very large scale monoclonal antibody purification: the case for
conventional unit operations. Biotechnol Prog 23:995–1008 (2007).
SUPPORTING INFORMATION 17 Liu HF, Ma J, Winter C and Bayer R, Recovery and purification process
Supporting information may be found in the online version of this development for monoclonal antibody production. MAbs 2:
13

article. 480–499 (2010).

J Chem Technol Biotechnol 2021 © 2021 Society of Chemical Industry wileyonlinelibrary.com/jctb


www.soci.org V Bansode et al.

18 Kateja N, Kumar D, Godara A, Kumar V and Rathore AS, Integrated chro- 29 Kateja N, Kumar D, Sethi S and Rathore AS, Non-protein A purification
matographic platform for simultaneous separation of charge vari- platform for continuous processing of monoclonal antibody thera-
ants and aggregates from monoclonal antibody therapeutic peutics. J Chromatogr A 1579:60–72 (2018).
products. Biotechnol J 12:1700133 (2017). 30 Gerstweiler L, Bi J and Middelberg APJ, Continuous downstream bio-
19 Gupta P, Kateja N, Mishra S, Kaur H and Rathore AS, Economic assess- processing for intensified manufacture of biopharmaceuticals and
ment of continuous processing for manufacturing of biotherapeu- antibodies. Chem Eng Sci 231:116272 (2021).
tics. Biotechnol Prog:e3108 (2020). 31 Cataldo AL, Burgstaller D, Hribar G, Jungbauer A and Satzer P, Econom-
20 Nweke MC, Rathore AS and Bracewell DG, Lifetime and aging of chro- ics and ecology: modelling of continuous primary recovery and cap-
matography resins during biopharmaceutical manufacture. Trends ture scenarios for recombinant antibody production. J Biotechnol
Biotechnol 36:992–995 (2018). 308:87–95 (2020).
21 Grom M, Kozorog M, Caserman S, Pohar A and Likozar B, Protein A 32 Yang O, Prabhu S and Ierapetritou M, Comparison between batch and
affinity chromatography of Chinese hamster ovary (CHO) cell culture continuous monoclonal antibody production and economic analy-
broths containing biopharmaceutical monoclonal antibody (mAb): sis. Ind Eng Chem Res 58:5851–5863 (2019).
experiments and mechanistic transport, binding and equilibrium 33 Schofield M, Cost modeling of an integrated, continuous downstream
modeling. J Chromatogr B 1083:44–56 (2018). mAb platform, in Integrated Continuous Biomanufacturing III, ECI
22 Feidl F et al., Model based strategies towards protein A resin lifetime Symposium Series. ECI, New York (2017).
optimization and supervision. J Chromatogr A 1625:461261 (2020). 34 Mahmuda A, Bande F, Kadhim al-Zihiry KJ, Abdulhaleem N, Majid RA,
23 Hober S, Nord K and Linhult M, Protein A chromatography for antibody Hamat RA et al., Monoclonal antibodies: a review of therapeutic
purification. J Chromatogr B 848:40–47 (2007). applications and future prospects. Trop J Pharm Res 16:713–722
24 Pathak M, Rathore AS, Lintern K and Bracewell DG, Protein a chroma- (2017).
tography resin lifetime-impact of feed composition. Biotechnol Prog 35 Langer ES, Downstream processing: promising technologies are on the
34:412–419 (2018). horizon. BioPharm Int 23:24–27 (2010).
25 Grilo AL, Mateus M, Aires-Barros MR and Azevedo AM, Monoclonal 36 Shukla AA and Gottschalk U, Single-use disposable technologies for
antibodies production platforms: an opportunity study of a non- biopharmaceutical manufacturing. Trends Biotechnol 31:147–154
protein-A chromatographic platform based on process economics. (2013).
Biotechnol J 12:1700260 (2017). 37 Xenopoulos A, A new, integrated, continuous purification process tem-
26 Follman DK and Fahrner RL, Factorial screening of antibody purifica- plate for monoclonal antibodies: process modeling and cost of
tion processes using three chromatography steps without protein goods studies. J Biotechnol 213:42–53 (2015).
A. J Chromatogr A 1024:79–85 (2004). 38 Liu S and Papageorgiou LG, Multi-objective optimisation for biophar-
27 Blank GS, Zapata G, Fahrner R, Milton M, Yedinak C, Knudsen H et al., maceutical manufacturing under uncertainty. Comput Chem Eng
Expanded bed adsorption in the purification of monoclonal antibodies: 119:383–393 (2018).
a comparison of process alternatives. Bioseparation 10:65–71 (2001). 39 Shirahata H, Diab S, Sugiyama H and Gerogiorgis DI, Dynamic modelling,
28 Ishihara T, Miyahara M and Yamamoto K, Monoclonal antibody purifi- simulation and economic evaluation of two CHO cell-based production
cation using activated carbon as a replacement for protein A affinity modes towards developing biopharmaceutical manufacturing pro-
chromatography. J Chromatogr B 1102–1103:1–7 (2018). cesses. Chem Eng Res Des 2150:218–233 (2019).
14

wileyonlinelibrary.com/jctb © 2021 Society of Chemical Industry J Chem Technol Biotechnol 2021

You might also like