End‐to‐end continuous bioprocessing: Impact on facility design, cost of goods, and cost of development for monoclonal antibodies

This article presents a systematic approach to evaluate the business case for continuous processing that captures trade‐offs between manufacturing and development costs for monoclonal antibodies (mAbs). A decisional tool was built that integrated cost of goods (COG) with the cost of development models and new equipment sizing equations tailored to batch, hybrid, and end‐to‐end continuous processes. The COG analysis predicted that single‐use continuous facilities (sized using a dedicated downstream processing train per bioreactor) offer more significant commercial COG savings over stainless steel batch facilities at annual demands of 100–500 kg (~35%), compared to tonnage demands of 1–3 tons (~±10%) that required multiple parallel continuous trains. Single‐use batch facilities were found to compete with continuous options on COG only at 100 kg/year. For the scenarios where batch and continuous facilities offered similar COG, the analysis identified the windows of operation required to reach different COG savings with thresholds for the perfusion rate, volumetric productivity, and media cost. When considering the project lifecycle cost, the analysis indicated that while end‐to‐end continuous facilities may struggle to compete on development costs, they become more cost‐effective than stainless steel batch facilities when considering the total out‐of‐pocket cost across both drug development and commercial activities.


| INTRODUCTION
Continuous processing has been the subject of renewed interest in recent years as a contender to traditional batch processing in the biopharmaceutical industry for products such as monoclonal antibodies (mAbs). This can be attributed to the benefits of continuous manufacture that include higher productivities and equipment utilization rates compared to traditional batch processes, which can translate into reduced facility footprints, capital expenditure, and manufacturing costs as well as the ability to rely on single-use technologies (Farid et al., 2014;Pollock et al., 2017;Schofield, 2018;Walther et al., 2015). Furthermore, technology gaps are being overcome with new solutions that make it easier to envision configurations for end-to-end continuous bioprocesses. However, there is debate on whether the uptake of continuous processes will streamline process development and validation efforts during drug development or increase them (Croughan et al., 2015;Farid et al., 2014;Kaltenbrunner, 2018;. Process development and clinical manufacturing have been estimated to contribute up to 17% of the total R&D cost, which translates into hundreds of millions of dollars per market success (Farid et al., 2020). Hence, it is important to be able to explore the balance between manufacturing cost savings and implications on the process development effort to achieve them when considering switching to new technology platforms. This article presents a decisional tool to evaluate the business case for end-toend and hybrid single-use continuous bioprocessing that captures trade-offs between manufacturing and development costs applied to mAbs.
The sector is debating and evaluating differing degrees of adoption of continuous technologies. On the upstream processing (USP) front, this has been enabled by the introduction of external retention devices (such as alternating tangential flow or tangential flow filtration technologies) for perfusion culture that overcome the limitations of earlier technologies and allow higher cell culture productivities and smaller manufacturing trains compared to batch methods (Clincke et al., 2013;Lim et al., 2006;Xu et al., 2017). On the downstream processing (DSP) front, continuous unit operations, specifically for bioprocesses, have been developed more recently. These include multicolumn chromatography, single-pass tangential flow filtration (SPTFF), and continuous virus inactivation (Casey et al., 2011;Gjoka et al., 2017;Jungbauer, 2013;Mahajan et al., 2012;Pagkaliwangan et al., 2019). With this toolbox of continuous unit operations, partially integrated continuous processes have been established to specifically improve DSP productivity and equipment utilization. These processes have commonly consisted of perfusion cell culture coupled with multicolumn capture chromatography or fed-batch cell culture with a continuous DSP train (Gjoka et al., 2017;Pollock et al., 2017;Warikoo et al., 2012;Xenopoulos, 2015). Advancements have been made towards more end-to-end continuous processes that have typically used perfusion cell culture, multicolumn capture and intermediate chromatography steps, and flow-through polishing Walther et al., 2015). Arnold et al. (2019) incorporated continuous viral inactivation (VI) and SPTFF operations to aid continuous flow within their fully integrated continuous process and minimize the size and number of surge tanks employed.
Due to the promising nature of continuous processing, many studies have specifically looked at the cost savings it offers. Often, the manufacturing-related costs of batch and continuous processes at commercial manufacturing scales have been compared. Such studies have shown that continuous facilities can typically result in a reduction in the commercial cost of goods (COG) between 10% and 30% (Arnold et al., 2019;Hummel et al., 2019;Pollock et al., , 2017Walther et al., 2015;Xu et al., 2017). Even greater cost reductions in capital investment estimates have been reported at between 40% and 50% Walther et al., 2015). This is largely attributable to a reduction in the size of manufacturing trains and the ability to implement single-use technologies at these smaller scales. Pollock et al. (2017) have also been able to demonstrate these savings in the COG within clinical manufacturing facilities. In addition to manufacturing costs, some studies have gone further to examine the cash flows and demonstrated increases in net present value (NPV) (Walther et al., 2015) and savings in net present cost (NPC) (Pollard et al., 2016) in the order of hundreds of millions of dollars.
As continuous processing is still in its infancy compared to traditional batch processing (which has experienced a great deal of evolution over the past few decades) within the biopharmaceutical sector, a larger development effort and cost could be required to establish a continuous process (Croughan et al., 2015;Farid et al., 2014;Kaltenbrunner, 2018 Notebook. The tool structure with its key inputs, outputs, and calculations is shown in Figure 1. The tool's database and key process economics equations were adapted from previous UCL work (Farid et al., 2020;Pollock et al., , 2017Simaria et al., 2012). New features built into the tool included: (a) extending the repertoire of continuous technologies in the unit operation library and database from perfusion and multicolumn chromatography to include also continuous versions of VI, ultrafiltration, and diafiltration operations; (b) new design equations to capture end-to-end continuous dynamics; (c) updating the database of unit costs and default process parameters, and (d) correlations between batch and continuous CMC process development costs. Table 1 shows the equations used to calculate the main costs examined in this article: the COG and the cost of CMC to ensure a market success (C CMC-Total ). The COG includes the direct (e.g., materials and labor) and indirect costs (e.g., facility-related overheads) incurred during manufacturing. The indirect costs are derived from the fixed capital investment (FCI), which is calculated using the Lang factor method (Lang, 1948). The Lang factors used for stainless steel and single-use facilities were calculated based on the methods described in Novais et al. (2001) and . C CMC-Total covers the portfolio costs related to the process development, validation, and manufacturing activities across the drug development cycle, from preclinical trials to the submission of a licence application for regulatory review (e.g., BLA or MAA), as defined by Farid et al. (2020). As a result, C CMC-Total includes the costs spent on failed drug candidates and the development activities performed at-risk.

| Facility designs modeled
This tool was used to compare the COG and C CMC-Total of four facility types: stainless steel batch (SS-Batch), single-use batch (SU-Batch), end-to-end continuous (SU-EE), and hybrid (SU-Hybrid). The SU-Batch process adopts single-use unit operations over the stainless steel versions used in SS-Batch wherever possible, such as single-use bioreactors, pre-packed chromatography columns, and hold bags. Figure 2 shows the flowsheets used and the scheduling of each unit operation for the facilities modeled in this study. A typical mAb flowsheet was used for the batch processes (Gronke & Gilbert, 2018;Kelley, 2009).
F I G U R E 1 Structure of the decisional tool used in this study. The cost of CMC (C CMC-Total ) includes the costs related to the process development, validation, and manufacturing activities across the drug development cycle, from preclinical trials to the submission of a licence application for regulatory review. In this study, the COG associated with commercial manufacturing scenarios is usually calculated over a period of one year unless specified otherwise. CMC, chemistry, manufacturing and controls; COG, cost of goods For SS-Batch or SU-Batch processes, each unit operation is carried out sequentially and sized using standard mass balancing equations that are based on the mass entering each unit operation per batch. The rationale behind the setup of the SU-EE facility was to convert the standard batch process into an end-toend continuous process that enables the continuous flow of material from the production bioreactor to the final DSP unit operation. This was achieved through the use of unit operations that are specifically designed for continuous bioprocesses, such as perfusion cell culture, multicolumn chromatography, and SPTFF for concentration and diafiltration.
When simulating SU-EE, the outlet from perfusion cell culture is directly loaded onto the multicolumn capture system. The discrete elution pools generated by each capture column are then collected into a VI vessel. The continuous VI system uses two alternating vessels so that while one vessel collects chromatography eluates, the other carries out inactivation and feeds material onto the following unit operation. Next, the eluates from cation exchange chromatography are pooled into a collection vessel. When enough eluates have been pooled, this vessel is drained at a constant rate so that a process stream can be continuously fed through the remaining unit operations, which are inherently run in flow-through mode. Arnold et al. (2019) have already been able to demonstrate a fully integrated continuous process using most of these unit operations. Additionally, to avoid the need for large surge vessels and hold times between unit operations, the flow rate of the process stream between the outlet of one unit operation and the inlet of the next must be as Note: For symbol definitions please refer to the nomenclature section. a The term reagents is used as an umbrella term for process reagents (e.g., media and buffer) and direct utilities (e.g., WFI and steam used for CIP/SIP).
b For consumables the number of units used per campaign account for the reuse limit. c This is a function of the number of operators scheduled to be working each day and their utilization on the manufacturing floor. For example, the operator utilization on the manufacturing floor is proportional to the number of reactors running in each facility to meet the target demands and the utilization will increase as more reactors and batches are required in each facility. d The general utility cost per unit area is assumed to be $525/m 2 . This cost accounts for the utility charges (e.g., HVAC) to run a facility. e This is the sum of the costs incurred across each phase of development from preclinical trials to the submission of a licence application for regulatory review to ensure a market success.

MAHAL ET AL.
| 3471 close as possible. Therefore, all unit operations running in continuous mode are sized based on the flow rate of the outlet of the previous unit operation. Table S1 in the Supporting Information Material shows the design equations used for this method.
The SU-Hybrid facility aims to provide a manufacturing method that does not involve changing the entire traditional batch flowsheet. This process operates in the same way as SU-EE up to VI, that is, with perfusion culture, multicolumn capture chromatography, and continuous VI. After this point, material is pooled for a set duration, generating discrete pools. Each pool is then individually processed in a batch manner by the remaining polishing chromatography and filtration unit operations.

| Case study assumptions
2.3.1 | Commercial manufacturing Table 2 shows some of the key assumptions made for each facility. A degree of automation was assumed within the continuous facilities (SU-EE and SU-Hybrid), so fewer operators were required per shift compared to the batch facility . However, when multiple perfusion bioreactors are required they must run in parallel, whereas the harvest of multiple fed-batch bioreactors can be staggered along the DSP train and processed sequentially. This is why a team of operators can handle more fed-batch bioreactors over perfusion bioreactors. When reviewing cell culture performance, fed-batch methods have previously given titers between 2 and 3 g/L, but 5 g/L is now becoming more common and there are reports stating even 10 g/L can be achieved (Kelley, 2007;Lindskog, 2018;Xu et al., 2017). Perfusion has the ability to offer higher cell densities and hence productivities that are greater than fed-batch by over sevenfold Walther et al., 2019). Therefore, a titer of 5 g/L for fed-batch cell culture and a volumetric productivity of 3 g/L/d for perfusion cell culture were used unless specified otherwise. Greater resin loading capacities were also assumed when using multicolumn chromatography due to the ability to increase the utilization of resin capacity when adopting this method (Jagschies, 2018;.
The definition of a batch is also very important for continuous processes. In this article, a "QC batch" is defined as the amount of  Abbreviations: COG, cost of goods; DSP, downstream processing; QC, quality control; USP, upstream processing; WFI, water for injection. a The media cost includes the cost of a base and feed media. For fed-batch and perfusion cell culture the proportion of feed media is 25% and 10%, respectively, of the total media consumed during cell culture. b Collected titer is measured in grams of product per litre of harvested cell culture fluid.
c For perfusion cell culture it was assumed that product collection started after the initial growth and ramp-up phase (8 days). d Volumetric productivity is measured in grams of product produced per litre of the bioreactor working volume per day. e Perfusion rate is measured as the equivalent number of bioreactor vessel working volumes (vv) exchanged per day.

MAHAL ET AL.
| 3473 material qualifying for a QC batch release test. For a continuous process, this is the amount of material produced every four days, which is equivalent to the period of time that occurs before switching out the final filter in the manufacturing train. When referring to a "batch" this includes the amount of material produced over the duration of one cell culture run, regardless of the facility type used. A "QC batch" and a "batch" for SS/SU-Batch are synonymous. Table 3 highlights the assumptions used to estimate C CMC-Total , which is a sum of the total cost of CMC process development (C CMC-PD ) and

| CMC activities during drug development
the total COG related to the manufacture of material for (pre-) clinical trials as well as PPQ batches (C CMC-MFG ) per market success.
C CMC-PD was taken to include all bulk process and formulation development as well as analytical effort for process characterization and validation studies and the technology transfer activities (Farid et al., 2020 Farid et al. (2020). As the manufacturing process is typically locked at Phase III, the size of the manufacturing facility here is based on the optimal size to meet a target market demand of 200 kg/year. The number of drug candidates modeled in this study and shown in Table 3 was calculated based on the target number of market successes for each company and the attrition rates at each phase provided by Paul et al. (2010).
As mentioned in the introduction, the cost to develop a continuous process can be higher than a standard batch process. With the additional level of parameters that need close monitoring and complex unit operations, more experimental data may be required for process characterization to demonstrate process robustness and product quality from a company adopting continuous processing for the first time. In addition, those working in regulatory support may need to be more diligent to ensure the process is compliant with guidelines from regulatory bodies.
At each phase of process development, scale-up and/or optimization occurs, which means many of the challenges noted above will be experienced up to the regulatory review stage. To prevent the extra process development activities associated with a continuous process falling on the critical path and causing delays to clinical trials, additional personnel may be necessary. As there has not been a suggestion of the actual process development costs for continuous processes, the costs used in this study have been calculated based on Reg. review 7.2 14.4 10.8 Note: The overall process yield for batch processes was 64%. The overall process yield for continuous processes (Hybrid and EE) was 68%. a This is for the production of PPQ batches that feed into the licence application submission for regulatory review. b It has been assumed that a small company launches one drug every 2 years. the increase in time and additional personnel that may be needed to develop a continuous process. It was estimated that development costs are two times greater for the SU-EE process compared to the batch processes modeled in this study (SS-Batch and SU-Batch).
When developing the SU-Hybrid process the increase in process development costs may only be 1.5 times greater than SS/SU-Batch, as more than half of the DSP unit operations are carried out in a standard batch mode.

| RESULTS AND DISCUSSION
The integrated drug development (CMC) and manufacturing (COG) economics tool was used to determine the rankings of batch, hybrid and end-to-end continuous facilities initially from a commercial COG perspective and then from a total project lifecycle out-of-pocket cost perspective that weighed up COG against CMC costs. Scenarios using different starting assumptions are presented to highlight how this impacts the rankings of the facilities in terms of COG, CMC, and total lifecycle out-of-pocket cost.
3.1 | COG at commercial scales 3.1.1 | Base case analysis across demands of 100-3000 kg/year Figure 3 shows the COG/g modeled across a range of commercial scales of production, from 100 to 3000 kg/year, as well as the key features of each facility type using the base case assumptions highlighted in Table 2.
This figure demonstrates that single-use facilities (continuous or batch) are able to offer a COG advantage of~35% over SS-Batch at the smallest scale of 100 kg/year. Beyond 500 kg/year, SU-Batch starts to become less favorable and from 1000 kg/year the continuous facilities start to offer COG values similar to or greater than SS-Batch (±9%).
The COG breakdowns and facility features highlighted in Figure 3 demonstrate how changes in the importance of each cost category and the attributes of the manufacturing trains can influence the cost rankings relative to SS-Batch. For example, at smaller scales of production, investment-driven indirect costs dominate the total COG. As SU-Batch is able to reduce the fixed capital investment by shifting some of the equipment costs to consumable-related costs, notable savings in the indirect costs and the total COG are achieved. This COG saving is also driven by a reduction in the reagent costs by 70% as the requirement of clean-in-place (CIP) and sterilization-in-place (SIP) is eliminated when using single-use technologies. Figure 4a shows that CIP and SIP contribute to the majority of the total reagent costs for SS-Batch at 100 kg/ year. As the scale increases, SU-Batch loses its COG advantage over SS-Batch due to the requirement for multiple bioreactors given the 2000 L size limitation typically assumed for single-use bioreactors (see embedded table in Figure 3a). At the ton scales, the need for parallel production lines (each with multiple staggered bioreactors sharing a DSP train) increases the indirect, labor and quality control (QC) costs incurred, and makes SU-Batch the least favorable facility type in terms of COG/g.
In contrast to SU-Batch, the single-use continuous facilities are able to offer COG reductions up to 2000 kg/year and only require multiple bioreactors at the tonnage demands modeled. This is driven largely by the reductions in indirect costs. Figure 3a demonstrates that the total bioreactor volume is smaller when using continuous facilities over batch facilities by sixfold to sevenfold, due to the higher productivities reached when using perfusion cell culture. Additionally, the generation of small harvested culture volumes from perfusion also permits the use of a  Figure 4c,d). This is also reflected in Figure 3b, which shows the increasing contribution of USP to the total COG value. Even if the constraint of single reactor trains was lifted and the same level of indirect cost savings of~50% could potentially be achieved at all scales of production, a decline in the total COG savings that continuous facilities provide will still be seen at the ton scales due to the reduction in the importance of indirect costs at the ton scales of production. The ability to see savings in the COG by pooling multiple bioreactors should also be weighed up against the need for larger DSP trains and the higher risk of discarding more material if the harvest from one of the pooled reactors fails to meet quality criteria.

| COG sensitivity analysis
The base case demonstrates that the continuous facilities modeled In this section, only SU-EE is compared to the batch facilities.
SU-EE and SU-Hybrid give a similar COG breakdown at every scale modeled and their key cost drivers are relatively similar. This is because the majority of the COG (~70%) are associated with the unit operations before the polishing steps. In addition to this, SU-Hybrid is still able to provide a smaller DSP train from the polishing steps onwards, compared to the batch facility, due to the smaller scale of the continuous unit operations. Figure 5a demonstrates that the ability to reduce the number of QC tests or the number of operators for SU-EE can improve the percent difference in COG between SU-EE and SU-Batch at 100 kg/ year. However, the impact of these parameters is not as significant when looking at the percent difference in COG between SU-EE and SS-Batch at 3000 kg/year (Figure 5b). Figure 4 highlights that fixed costs (e.g., QC and labor) do not contribute as much to the total COG values at larger scales of production as the scale-dependent costs do (e.g., reagents).
Improving parameters that impact the total cost of media per campaign for perfusion cell culture can make SU-EE more costeffective at any scale. In the base case, the consumption and cost of media per campaign for perfusion is~2.8-and 1.6-fold greater, respectively, than fed-batch at all scales. When the best media unit cost is used for SU-EE, the total cost of media per campaign is only 1.2-fold greater for fed-batch cell culture. As the total cost of media contributes more to the total COG at 3000 kg/year, for all facility types, its impact on the percent difference in COG between SU-EE and the batch facilities is greater than at 100 kg/year. For these reasons, improving any parameter that impacts the total cost of media for perfusion (such as perfusion rate or volumetric productivity) can make SU-EE more cost-effective than SS-Batch at 3000 kg/year by~10%-15%. Alternatively, when the worst values of these parameters are used, SU-EE can give significantly higher COG values compared to the batch facilities. As seen in Figure 5a,b, when the perfusion media cost is similar to the fed-batch media cost, the COG value for SU-EE becomes the same as SU-Batch at 100 kg/year or significantly worse than SS-Batch at 3000 kg/year by~25%. At both scales, this higher unit cost of perfusion media increases the media cost per campaign (~2.4-fold), but the impact of this is greater at 3000 kg/year given that material costs dominate the COG/g. F I G U R E 3 COG breakdown on (a) a category basis and (b) a process stage basis for the base cases of the four batch and continuous facility types across commercial scales of 100-3000 kg/year. Here, maximum facility utilization was assumed so each fed-batch and perfusion reactor employed produces~20 and 10 batches per year, respectively. The embedded table in (a) indicates the key facility features for each batch and continuous facility and can be applied to (b) as well. In (a) the category breakdown covers labor, QC, consumables, reagents and indirect costs. In (b) the process stage breakdown covers the total USP, DSP, and QC costs. One "parallel train" is classified as a group of USP reactors supported by one DSP train. For the continuous facilities, it was assumed that each perfusion bioreactor requires a dedicated DSP train and that parallel trains are installed when the maximum single-use bioreactor size available is exceeded. In this figure: fed-batch cell culture titer = 5 g/L and perfusion volumetric productivity = 3 g/L/d. The term reagents is used as an umbrella term for process reagents (e.g., media and buffer) and direct utilities (e.g. WFI and steam used for CIP/SIP). CIP, clean-in-place; DSP, downstream processing; QC, quality control; SIP, sterilization-inplace; SS-Batch, a stainless steel batch facility; SU-Batch, a single-use batch facility; SU-EE, a single-use and end-to-end continuous facility; SU-Hybrid, a single-use and hybrid facility (with batch and continuous unit operations); USP, upstream processing; WFI, water for injection F I G U R E 4 Contribution of each cost category to the total COG value modeled in the base case for: (a) SS-Batch at 100 kg/ year, (b) SU-EE at 100 kg/year, (c) SS-Batch at 3000 kg/year, and (d) SU-EE at 3000 kg/year. CIP/SIP refers to acid, caustic, WFI and steam that are required during the cleaning cycles of unit operations and hold vessels within the process. The cost of CIP/ SIP is less than 1% for SU-EE so is not visible. The term reagents is used as an umbrella term for process reagents (e.g., media and buffer) and direct utilities (e.g. WFI and steam used for CIP/ SIP). CIP, clean-in-place; SIP, sterilization-in-place; WFI, water for injection The ability to improve certain parameters for continuous processes may also mean similar improvements can be achieved within batch facilities. For instance, the unit cost of a buffer can be the same regardless of the facility type modeled. Figure 5c,d show the impact when the most sensitive parameters are changed for both SU-EE and batch facilities. At 100 kg/year the impact that these parameters have on the percent difference in the COG between SU-EE and SU-Batch is minimal. The only parameter that causes a significant change at 100 kg/year is the worst fed-batch titer (of 3 g/L). A lower titer for fed-batch means more bioreactors are required to meet the target demand, which results in an increase in costs that make up a fairly large proportion of the COG at smaller scales such as QC and labor.
At 3000 kg/year, reagent costs (particularly media and buffers) are a significant proportion of the total COG, so have the ability to have a greater influence on the percent difference in the COG between SU-EE and SS-Batch. Figure 4 shows that media alone is 22% and 33% of the total COG for SS-Batch and SU-EE, respectively.
Therefore, reductions in the total media cost through better unit prices, titers, and volumetric productivities for both facility types can narrow the difference in the COG between SU-EE and SS-Batch.
When a smaller unit cost of media is used for both facilities, the total cost of media per campaign for perfusion is still higher than fedbatch by~1.6-fold, but the proportion of the COG attributed to media is now 13% and 20% for SS-Batch and SU-EE, respectively. For this reason, even if the media cost and consumption is improved for both SS-Batch and SU-EE, the COG of SU-EE becomes more comparable with SS-Batch.
For companies that wish to make the switch from batch to continuous processing, a COG saving greater than a certain threshold such as 20% may need to be achieved to justify such a change. Even when the best values of the parameters shown in Figure 5 are used, this threshold is not reached at the scales presented. To reach a target COG reduction of 20% or above, multiple parameters need to be improved in parallel. Figure 5 shows that the volumetric productivity, perfusion rate and media cost have a large impact on the difference in the COG between the SU-EE and batch facilities at 100 and 3000 kg/year. These parameters can also be interdependent. For example, it has been found that media with expensive feed additions or higher perfusion rates can improve the volumetric productivity of perfusion cell culture (Clincke et al., 2013;Xu et al., 2017). even if an expensive perfusion media, of $29/L, is used to achieve high volumetric productivities (greater than 3.6 g/L/d), as long as the perfusion rate is minimized to 0.5 or 1 vv/d. If a cheaper perfusion media is used, of $10/L, a COG advantage of at least 20% can be seen across a wider range of perfusion rates up to 2-3 vv/d, as long as the volumetric productivity is greater than 3.6 g/L/d. However, the target COG saving of 20% cannot be achieved at the 100 kg/year scale in any scenario presented in Figure 6. This is due to the smaller influence that the total media cost has on the total COG value at this scale. At this scale, improvements in fixed costs such as QC or labor could also be considered to help achieve greater COG savings. year, the ability to improve perfusion's volumetric productivity from 3 to 4.9 g/L/d can lead to a~40% reduction in COG between SU-EE and SU-Batch. This is a large difference from the 9% saving SU-EE offers in the base case. Improving the volumetric productivity here not only reduces the media cost, which is 27% of the total COG, but also reduces the indirect cost as the number of parallel trains halves to one. Overall, the COG savings seen when adopting continuous manufacturing will be largely dependent on the company scenario.

| Cost of process development versus COG
The

| CMC cost
Initially, the impact on the CMC cost (C CMC-Total ) was considered.
This refers to the total cost of process development (including process characterization/validation) and manufacturing activities (for trials and PPQ batches) across the drug development cycle to ensure a market success. When assessing the costs related to drug development activities, such as the CMC cost, it is important to consider a company's portfolio size. As Table 1 highlights, the number of drug  (d) show the regions where a target cost saving greater than 40% compared to the SS-Batch facility is achieved. All percentage differences shown in this figure are relative to SS-Batch. C CMC-MFG , the total cost of goods related to the CMC activities, to manufacture material for (pre-) clinical trials and PPQ batches; C CMC-PD , the total cost of CMC process development from preclinical trials to the submission of a licence application for regulatory review; C CMC-Total , the costs related to the process development, validation and manufacturing activities across the drug development cycle, from preclinical trials to the submission of a licence application for regulatory review;C COMM-MFG , the total cost of goods over a 10 year commercial manufacturing period, when producing 200 kg/year; C Lifecycle , a project's total out-of-pocket cost to ensure a successfully launched product and supply the market for 10 years; CMC, chemistry, manufacturing and controls;COG, cost of goods; PPQ, process performance qualification MAHAL ET AL.
| 3481 candidates entering each phase of drug development will determine the total process development and manufacturing costs and as a result, their weight on the total CMC cost value. In this study, C CMC-Total was calculated for each facility type (SS-Batch, SU-Batch, SU-EE, and SU-Hybrid) and two company sizes, a large and small company. A larger company is assumed to have more drug candidates in the pipeline, which can lead to higher CMC costs, compared to a smaller company as they may have more frequent product launches. Table 3 shows the number of drug candidates entering each phase of development for these company sizes.   (Farid et al., 2014;. This could potentially reduce the development activities required at the later phases and bring down C CMC-PD and C CMC-Total . However, it was found that the trends originally seen in the CMC and project lifecycle costs in Figure 7a,b do not change significantly. Even though development activities and costs for the continuous facilities are reduced from Phase III onwards, the COG savings are also reduced as the process optimization activities that would usually occur at Phase III do not happen as the process cannot be changed after Phase I. As shown in  (Table S2 and Figure S2).
Considering the above, companies with larger portfolios may be more inclined to adopt single-use batch facilities over continuous facilities. For large companies that wish to adopt continuous processing, similar to what is required for a batch process. This reduction in C CMC-PD will need to be met with a reduction in C CMC-MFG from SS-Batch by at least 60%, which is significantly greater than the current C CMC-MFG saving of~20%. If the C CMC-PD can become less than SS-Batch by 25%, this can reduce the target C CMC-MFG saving to 50%; such a scenario could correspond to having an optimized platform continuous process that is locked at Phase I and hence reduces the late phase process development costs.
From the project lifecycle cost perspective (Figure 7d), a 40% saving in C Lifecycle could be met more easily as this cost is relatively insensitive to C CMC-PD , as C COMM-MFG dominates the total lifecycle cost. For example, the target C Lifecycle saving threshold can be met even at the current cost of process development (C CMC-PD ) for the continuous facilities (×1.5-2fold), as long as a 60% saving in the total COG across development and commercial manufacture can be achieved. Currently, the total COG saving is~30% for both SU-EE and SU-Hybrid, so work needs to be carried out to reduce the COG of continuous manufacturing further. If C CMC-PD for continuous is +25% or below compared to the cost for SS-Batch, the target total COG saving can be lowered to 50%, which could potentially be achieved with some of the improvements suggested in Section 3.1.
Overall, the analysis has presented an array of scenarios that can be mapped onto different company scenarios to help prioritize development efforts with continuous manufacture.

| CONCLUSION
This article demonstrated the development of an integrated drug development (CMC) and manufacturing (COG) economics tool to support a systematic analysis of the business case for the implementation of continuous processing for the production of mAbs.
The tool incorporated new design equations that were specifically developed to aid comparison of the COG between single-use continuous processes (hybrid and end-to-end) and batch processes (stainless steel and single-use) across a wide range of company scenarios. Additionally, correlations to compare the cost of CMC development and manufacturing activities between continuous and batch processes were generated. This made it possible to determine if any savings seen in the COG at commercial and clinical scales would be outweighed by the large process development costs that continuous processes may incur. The analysis highlighted that continuous facilities offer COG savings when at smaller commercial demands where single USP and DSP trains are required. At larger scales, when parallel continuous trains are implemented and the cost of media dominates the COG, the cost effectiveness of continuous manufacture is reduced. The business case for continuous processing will depend on whether only a CMC perspective is considered or the total project lifecycle out-of-pocket cost across drug development and commercial activities. The tool predicted that although CMC costs with continuous processes are likely to be higher than batch processes when considered as a new technology, it is possible for continuous processes to be as or more competitive than the best batch process when the total lifecycle out-of-pocket costs are considered. Once future platform continuous processes are established that simplify process development, scaling, and optimization, it was predicted that this would make continuous facilities more attractive than batch processes in scenarios where they currently cannot compete in terms of cost.